A Dissertation

entitled

Characterization of the Vibrio cholerae Phage Shock Protein Response

by

Cara Marie DeAngelis

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

Doctor of Philosophy Degree in

Biomedical Science

______Jyl Matson, Ph.D., Major Advisor

______Robert Blumenthal Ph.D, Committee Member

______Mark Wooten, Ph.D, Committee Member

______Jason Huntley, Ph.D, Committee Member

______David Giovannucci, Ph.D, Committee Member

______Cyndee Grudgen, Ph.D., Dean College of Graduate Studies

The University of Toledo

May 2019

© 2019 Cara Marie DeAngelis

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

An Abstract of

Characterization of the Vibrio cholerae Phage Shock Protein Response

by

Cara Marie DeAngelis

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Biomedical Science

The University of Toledo May 2019

Cholera is a severe intestinal infection characterized by voluminous, watery diarrhea that can be fatal within hours. It is caused by the marine bacterium Vibrio cholerae of serogroups O1 and O139. While rare in the United States and other industrialized nations, cholera is endemic in more than 50 countries. In both its aquatic and intestinal life cycles, V. cholerae encounters various stressful conditions, such as fluctuating pH, bacteriophage predation, and exposure to antimicrobial peptides that may disrupt the cell envelope. The phage shock protein (Psp) system is a stress response pathway that senses and responds to inner membrane damage. The genetic components of the Psp system are present in several clinically relevant Gram-negative bacteria, including V. cholerae.

However, most of the current knowledge about the Psp response stems from in vitro studies in Escherichia coli and Yersinia enterocolitica. In fact, the Psp response in V. cholerae has remained completely uncharacterized. In this dissertation, it is demonstrated that V. cholerae does have a functional Psp response system. The overexpression of GspD (EpsD), the type II secretion system secretin, induces the Psp response, whereas other V. cholerae secretins do not. In addition, several environmental conditions were identified as inducers of this stress response. Experiments on the genetic

iii regulation and induction of the Psp system in V. cholerae suggest that the key regulatory elements are conserved with those of other Gram-negative bacteria. While a psp null strain is fully capable of colonizing the infant mouse intestine, it exhibits a colonization defect in a zebrafish model, indicating that this response may be important for disease transmission in the environment. Overall, these studies provide an initial understanding of a stress response pathway that has not been previously investigated in V. cholerae.

iv

To my parents

v

Acknowledgements

There are a vast number of individuals that supported me throughout my predoctoral studies. First, I would like to acknowledge my mentor, Dr. Jyl Matson.

While I still have a lot to learn, I will leave her laboratory with the ability to proudly call myself a scientist.

I would like to thank my committee members for counseling me on my research project for the last three years. I would also like to thank Dr. Rande Worth for his mentorship and accepting the position of serving as my faculty representative at my defense. In addition, I would like to thank our collaborators, Dr. Jeff Withey and Dr.

Dhrubajyoti Nag at Wayne State University.

I am thankful for the continued patience, guidance and support I have received from my previous lab members, Jess Saul-McBeth, Sarah Plecha and Laura Stanbery-

Nejedlik. In addition, I am grateful to past and present MMI students and the bonds of solidarity we formed over the years and monthly munches. I am also deeply grateful for the Glass City Readers and my fairy godmothers for cultivating my personal growth outside of science.

Lastly, I would like to thank my family. They supported me through all the highs and lows. In addition, Nitro has been a loving companion every single day.

vi

Table of Contents

Abstract ...... iii

Acknowledgements ...... vi

Table of Contents ...... vii

List of Tables ...... ix

List of Figures ...... x

List of Abbreviations ...... xii

Chapter 1 General Introduction ...... 1

1.1 Cholera History and Global Impact ...... 1

1.2 V. cholerae and its environment ...... 5

1.3 V. cholerae Pathogenesis ...... 8

1.4 Extracytoplasmic stress response systems ...... 11

1.5 Summary and Aims…...... 23

Chapter 2 Vibrio responses to extracytoplasmic stress ...... 24

2.1 Summary…...... 24

2.2 Introduction…...... 25

2.3 The σE response …...... 27

2.4 The Cpx response..…...... 34

2.5 Stress relief via OmpU…...... 38

2.6 Concluding remarks…...... 44

vii 2.7 Acknowledgments…...... 44

Chapter 3 Characterization of the Vibrio cholerae phage shock protein response ....45

3.1 Abstract…...... 46

3.2 Importance…...... 46

3.3 Introduction…...... 47

3.4 Results..…...... 50

3.5 Discussion…...... 65

3.6 Materials and Methods…...... 72

3.7 Acknowledgments…...... 82

3.8 Supplemental Figures…...... 83

Chapter 4 Additional Studies on the V. cholerae Psp response ...... 86

4.1 Introduction…...... 86

4.2 Growth phenotypes of psp mutants...... 86

4.3 PspC requires PspB for stability ...... 88

4.4 Investigating the relationship between the Psp and σE

response…………………………………………………………..91

4.5 Discussion……………………………………………………92

Chapter 5 Discussion ...... 97

5.1 Overview…...... 97

5.2 Discussion and future directions...... 99

5.3 Concluding Remarks…...... 108

References ...... 109

viii

List of Tables

1.1 RNA sequencing reveals psp transcripts are upregulated in O395ΔrpoE in

comparison to O395 ...... 14

2.1 Stress-associated phenotypes of the RpoE response in Vibrio species ...... 31

2.2 Role of RpoE in host-induced response to stress ...... 32

3.1 Strains and Plasmids ...... 73

4.1 Comparison of RNA-seq and qRT-PCR results of psp levels in O395 versus

O395ΔrpoE...... 91

ix

List of Figures

E 2 – 1 Model of σ , Cpx, and OmpU involvement in extracytoplasmic stress relief ...... 43

3 – 1 Genetic organization of Psp systems and PspA sequence similarity ...... 51

3 – 2 The T2SS secretin, GspD, induces the Psp response in V. cholerae ...... 55

3 – 3 GspD forms heat resistant multimers that mislocalize to the inner membrane,

where the other V. cholerae secretins do not ...... 57

3 – 4 GspD overexpression increases psp transcript levels and leads to PspA membrane

association...... 59

3 – 5 PspA is a negative regulator of the Psp response, and PspB, -C and -F are positive

regulators...... 62

3 – 6 The Psp response can be induced by specific environmental conditions ...... 64

3 – 7 The psp null strain shows reduced colonization in the zebrafish model of cholera

infection, but no defect in infant mice ...... 67

3 – 8 Model of Psp response in V. cholerae ...... 71

3 – S1 pspABC are cotranscribed...... 83

3 – S2 PspA is primarily localized in the soluble fraction under non-inducing

conditions...... 84

3 – S3 Complementation with pspBC and pspF restore psp activity to their respective

deletion strains ...... 85

4 – 1 Growth phenotypes of psp mutants in neutral, acidic, and alkaline pH ...... 89

x 4 – 2 PspC expression is unstable in the absence of PspB...... 90

4 – 3 PspA expression is not increased in the absence of rpoE conditions...... 93

xi

List of Abbreviations

AMP ...... Antimicrobial peptide ATR...... Acid tolerance response BPI ...... Bactericidal/permeability increasing CCCP ...... Carbonyl cyanide m-chlorophenyl hydrazine CT ...... Cholera toxin ECF ...... Extracytoplasmic function EMSA ...... Electrophoretic mobility shift assay ESR ...... Extracytoplasmic (Envelope) stress response GTFCC ...... Global Task Force on Cholera Control IM ...... Inner membrane KEGG ...... Kyoto Encyclopedia of Genes and Genomes LPS ...... Lipopolysaccharide MSHA ...... Mannose-sensitive hemagglutinin NHS...... Normal human sera OM ...... Outer membrane OMPs ...... Outer membrane proteins PMB ...... Polymyxin B PMF...... Proton motive force PSP ...... Phage shock protein RIP ...... Regulated intra-membrane proteolysis RNAP ...... RNA polymerase RND ...... Resistance-nodulation-division SDS ...... Sodium dodecyl sulfate T2SS… ...... Type II secretion system T3SS...... Type III secretion system T4PS… ...... Type IV pilus system TCP ...... Toxin coregulated pilus TCS ...... Two component system VBNC ...... Viable but non-culturable VNSS ...... Vaantanen nine-salt solution VPI ...... Vibrio cholerae specific pathogenicity island WASH ...... Water, sanitation and hygiene

xii Chapter 1

General Introduction

1.1 Cholera History and Global Impact

Cholera is a severe diarrheal infection caused by the Gram-negative bacterium

Vibrio cholerae. The bacterium is acquired via the fecal-oral route, when feces from an infected person contaminate drinking water or food. While cholera is currently endemic in over 50 countries, it is a disease that has been plaguing humanity for thousands of years. The first accounts of cholera are described in the Sushruta Samhita, an ancient

Sanskrit medical text that dates to around the time of the Buddha in 500-400BC [1]. For over 2000 years, China has called cholera fok lun, and other countries all over the world have their own terminology for cholera, making it apparent that it was a common historical affliction. However, the main symptoms of cholera are also shared among other diseases, making it difficult to discern the authenticity of these historical descriptions. Even with these uncertainties, it is widely accepted that cholera originated in Asia, primarily in India [1]. The disease began to spread however, due to increasing population densities, war, and industrialization. The modern era of cholera was born with the first pandemic in 1817 and the seventh, current pandemic, is still ongoing.[2].

1

The seven pandemics

The first cholera pandemic originated in India and spanned from 1817 to 1823, spreading to China, Japan, and the Middle East. In fact, the first six pandemics all originated in India before spreading along trade routes or with passaging army troops.

The second pandemic lasted from 1829 to 1851, and reached Europe and the United

States. During this pandemic, a Scottish physician known as Latta was the first to treat

15 cholera patients with intravenous saline in 1832. His treatment was criticized when only 5 patients survived, and other physicians continued treating patients with leeching, phlebotomy, and inducing vomiting until the late 19th century [1]. During this time, physicians believed the theory of miasma, or polluted air, was what caused cholera. In

1854, during the third pandemic (1852-1859), London suffered a raging cholera epidemic, with about 500 people dying within the first 10 days. A physician in London,

John Snow, believed that cholera was a waterborne disease. He investigated the location of cholera cases in the Soho area and mapped them along with the public water pumps.

This led him to believe that water from one pump in particular was responsible for spreading cholera. He famously was able to get the pump handle removed, and the cholera cases decreased. John Snow’s cholera mapping study established him as the founding father of epidemiology [3]. Concurrently, Filippo Pancini was studying the stool and intestinal samples from cholera victims in Italy, and identified comma shaped bacilli he named “vibrios.” Years later, in this midst of the fifth pandemic (1881-1896), the German physician Robert Koch found in postmortem studies that the Vibrio bacillus was only present in patients with cholera [4]. He was the first to isolate V. cholerae in pure culture, but was unable to reproduce the symptoms of cholera in animals. 2

Unfortunately, the miasma theory was still pervasive within the scientific community, and even with Snow, Pancini, and Koch’s discoveries, it was still believed that cholera was transmitted through the air. Eventually germ theory prevailed, but it took until the identification of cholera toxin in 1959 to confirm the mechanism of action of the bacteria

[5].

After the sixth pandemic (1899-1925), there was period of time in which cholera cases were confined to Asia. Unfortunately, the lull in cholera across the globe ended in

1961 and the seventh pandemic began. Two new differences occurred with this pandemic. First, the pandemic originated in Indonesia, and second, the strain infecting individuals was of a different biotype, known as El Tor (discussed below).

Presently, cholera is still a public health threat, as an estimated 2.9 million people are infected annually, with 107,000 deaths [6, 7]. The most recent epidemics have occurred in Haiti (2010), Sierra Leone (2012), South Sudan (2014), Ghana (2014),

Yemen (2017), and Zimbabwe (2018) [5, 8, 9]. Cholera is a disease of inequity as it typically afflicts people within impoverished countries. The United Nations has declared that access to safe water, sanitation and hygiene (WASH) is a basic human right, but two billion people still lack those resources. Furthermore, with global climate change, increased urbanization, and growing population density, the risk of contracting cholera will likely increase [10, 11]. In response, the Global Task Force on Cholera Control

(GTFCC) has set the goal of reducing the number of cholera cases worldwide by 90% by the year 2030 [6]. They aim to reach this goal by improving WASH in cholera hotspots, provide oral vaccines, and try to contain outbreaks more quickly [6, 7].

3

Cholera Clinical Disease, Diagnosis, and Treatments

In order for V. cholerae to cause infection, between 103 to 1011 bacteria need to be ingested by a human. The incubation period can range from 2 hours to 5 days.

Interestingly, only about 20% of individuals who ingest the bacterium experience severe symptoms, whereas the other 80% remain asymptomatic [12]. Regardless of manifestation of the disease, all carriers of V. cholerae will shed bacteria in their stool for up to 2 weeks after infection. In severe cases, the hallmark symptom for the first three days is acute watery diarrhea, also known as ‘rice water-stool’ since it resembles water that rice has been rinsed in. The secretory diarrhea is so severe that patients can lose up to a liter of fluid per hour, rapidly leading to dehydration and electrolyte imbalance.

Such extreme fluid loss is why small children, in particular, can quickly succumb to the disease. Patients may also experience vomiting as another symptom, and other secondary symptoms such as tachycardia, stroke, hypocalcemia, hypokalemia, hypothermia, and renal tubular necrosis. If patients are left untreated, 70% will die, but the mortality can be remarkably reduced to 3% with oral hydration therapy, and lower than 0.5% with intravenous rehydration therapy [3]. The oral rehydration solution consists of water, salt, and sugar, making it relatively low cost, and it does not have to be administered by a health professional like the intravenous method. Oral rehydration therapy can also be used to treat diarrheal cases caused by other pathogens, and its global use has decreased deaths from diarrheal disease to less than half of what it was before 1970 [3].

4

Vaccines

While the end goal is for all of humanity to have access to clean water and sanitation, that is not possible to achieve in the short term. Therefore, vaccination is a more immediate, cost-effective option for the prevention of cholera. There are currently six different cholera vaccines, none of which provide lasting or complete protection.

Dukoral is composed of killed whole bacteria, with ORC-Vax, Shancol, Euvichol and

Cholvax containing killed whole bacteria plus the recombinant B-subunit of cholera toxin. ORC-Vax and Cholvax are only available in Vietnam and Bangladesh, respectively. In 2016, Vaxchora was the first approved live-attenuated cholera vaccine and only requires a single dose, whereas the other 5 vaccines all require 2 doses. In addition, none of the current vaccines can be given to children under the age of 1. The current challenges with vaccination are distribution, storage, and uptake [13]. .

1.2 V. cholerae and its environment

Vibrio cholerae is a Gram-negative, curved rod with a single polar flagellum.

The cell wall of Gram-negative bacteria contains an inner membrane (IM), periplasmic space containing peptidoglycan, and an outer membrane (OM). The OM inner leaflet is primarily composed of phospholipids, whereas 75% of the outer leaflet is composed of lipopolysaccharide (LPS). LPS has three main regions, the lipid A, core polysaccharide, and O-polysaccharide chain (or O-chain) [14]. The O-chain is an extremely variable region that can be used to identify different bacterial serotypes through agglutination assays [15-17]. V. cholerae has over 200 serotypes with only the O1 and O139 serotypes causing epidemic and pandemic cholera [18]. All of the other serotypes are referred to as 5

non-O1/non-O139 serotypes, and they can sometimes cause sporadic gastroenteritis and extra-intestinal infections, such as ear and skin infections [19, 20].

O1 strains can be further broken down into two biotypes: classical and El Tor.

The El Tor biotype is named after the quarantine station, El Tor, were it was first isolated in 1905 in the Sinai in Egypt [2]. Up until the 7th pandemic, the previous pandemics were caused by the classical biotype of V. cholerae. The El Tor biotype began eclipsing the classical biotype starting with the 7th pandemic, and since 1992 has been the predominant biotype isolated from cholera outbreaks [21]. While both classical and El Tor strains belong to the serogroup O1, they evolved independently and are distinguishable by several phenotypic tests. In particular, the most useful tests for biotyping are sensitivity to the antimicrobial peptide polymyxin B (PMB), the Voges-Proskauer reaction, and chicken erythrocyte hemagglutination. The El Tor biotype is less sensitive to PMB, is positive for the Voges-Proskauer reaction, and causes agglutination of chicken erythrocytes [2]. In addition, the El Tor biotype survives better in environmental niches and causes a larger number of mild or non-symptomatic cases of cholera than classical strains. These traits of El Tor, along with modern transportation, are why the 7th pandemic has had the longest and most expansive geographic dissemination [2].

V. cholerae belongs to the family Vibrionaceae, a diverse group of pathogenic and nonpathogenic aquatic bacteria. The members of Vibrionaceae can be regularly found in oceans, from the surface to deep waters. While they can also be isolated from fresh water environments, V. cholerae tends to flourish in warm brackish waters [22, 23].

Within the aquatic environment, Vibrios can be motile and free-living, or sessile and associated with a number of different biotic and abiotic surfaces. V. cholerae has been 6

detected in the intestines of fish, soft-shelled turtles, and waterfowl. In addition, V. cholerae can attach to invertebrate hosts such as algae, shellfish, copepods, and chironomid eggs [24-30]. In fact, up to 104 V. cholerae have been found on a single copepod [31]. In the aquatic ecosystem, chitin is the predominant polysaccharide as it is a major constituent of zooplankton exoskeletons. Interestingly, V. cholerae can utilize chitin as a carbon source. Furthermore, V. cholerae can form biofilms on exoskeletons, which aids in its survival and persistence [32]. Recently, V. cholerae was found replicating within the amoeba, Acanthamoeba castellanii, where they were sheltered from external stresses and phage predation [33]. All of these diverse hosts could serve as vectors for the disease. Therefore, understanding more about the ecology and environmental hosts of V. cholerae may lead to strategies to prevent its global transmission.

Countries with endemic cholera may face seasonal outbreak patterns. On a global level, outbreaks tend to be relatively constant near the equator due to the stable, warmer temperatures. In endemic areas further from the equator, where climate is less constant, there tends to be seasonality to the cholera outbreaks [34]. In a number of countries the cases increase after rainy seasons, when water levels rise. During interepidemic periods when temperatures decrease, V. cholerae is more difficult to culture as it often enters a state of dormancy termed “viable but non-culturable” (VBNC). Interestingly, Vibrio bacteriophages have also been detected during these interepidemic phases, suggesting that bacteriophages can also contribute to cholera seasonality [35].

7

1.3 V. cholerae Pathogenesis

The hallmark symptom of cholera, the voluminous watery diarrhea, is induced by the production of the cholera toxin (CT). CT is a classic enterotoxin that is composed of a single A subunit and five identical B subunits [36]. After ingestion of V. cholerae, the majority of the bacteria are killed by the stomach gastric acid. Once V. cholerae reaches the small intestine, CT and other virulence factors are produced. The CT B pentamer subunits bind ganglioside GM1 on intestinal epithelial cells. After binding, the entire toxin is endocytosed and transported through the retrograde transport pathway to the endoplasmic reticulum. The A subunit dissociates from the B subunits in the cytosol and catalyzes ADP-ribosylation of the Gsα subunit of adenylate cyclase. This causes adenylate cyclase to be locked in a GTP-bound state, leading to continuous production of cyclic AMP (cAMP). The excessive buildup of intracellular cAMP leads to a shift in electrolyte transport. More chloride and bicarbonate are transported across the epithelial cell membrane, whereas less sodium is transported in. Water follows the ion gradient leading to an overall net fluid loss from the epithelium into the lumen of the intestine

[37].

Volunteer studies have shown that ingestion of V. cholerae strains with the CT genes deleted can still cause mild or moderate diarrhea, indicating that there are additional virulence factors [38]. Through animal models and human volunteer studies, it has been shown that toxin-coregulated pilus (TCP) and the toxR regulon are necessary for colonization and pathogenicity [2]. TCP is a type IV pilus system (T4PS) that produces a bundle of 7nm filaments which adhere to intestinal epithelial cells and form matrices that protect and clump V. cholerae [39]. TCP is also expressed under the same conditions as 8

CT, hence its name [2]. In fact, TCP, CT and a number of other factors are tightly controlled by the ToxR regulon [40]. The ToxR regulon spans over 20 different genes all involved in host survival, intestinal attachment, and toxin production. ToxR is a unique transcription factor due to the fact that it is an inner membrane protein with cytosolic

DNA binding domain. The periplasmic domain of ToxR interacts with another protein,

ToxS, leading to an optimal transcriptional activation state in ToxR. Then ToxRS activates transcription of ToxT, a secondary transcription factor, and it induces expression of CT and TCP genes [41]. A large number of other regulatory proteins and environmental factors such as temperature, pH, bile and nutrient availability control the

ToxR regulon [40]. The ToxR regulon and its relationship with the outer membrane proteins OmpT and OmpU are further described in Chapter 2. A number of other minor virulence factors have been identified in V. cholerae in addition to the two main virulence factors, CT and TCP. These minor virulence factors include Zot, soluble HA/protease, and accessory cholera enterotoxin (Ace), though their functions are less well understood.

Furthermore, motility and biofilm production also contribute to the pathogenicity of V. cholerae [2].

Animal models of cholera

In an effort to understand cholera pathogenesis in humans, a number of animal models have been used over the years. Replicating human disease is difficult, as V. cholerae does not colonize any adult mammals other than humans. However, this complication can be surpassed using surgical methods or infecting infant rabbits or mice.

It is not known why suckling mammals are susceptible to V. cholerae colonization, but it 9

is hypothesized that it is due to their immature immune systems [42]. Even so, all current cholera animal models have limitations and fail to completely capture all aspects of human infection. The most commonly used animals in cholera research are mice and rabbits. However dogs, guinea pigs, rats, fruit flies, and nonhuman primates have also been inoculated with V. cholerae [42, 43]. Notably, studies have even been carried out with human volunteers, since cholera can be treated with appropriate rehydration.

The suckling infant mouse model of cholera was first described in 1968, and for the past 30 years has been the most common animal used to study V. cholerae pathogenesis [42]. As with all animals modeling human disease, the infant mouse model has advantages and disadvantages. For example, it is useful for examining colonization of V. cholerae in the intestines, but it is not useful for vaccination studies as the mice expire before they acquire protective immunity [44]. Furthermore, the infant mice do not exhibit cholera-like diarrhea. The infant rabbit does develop cholera-like secretory diarrhea and can also be used to measure colonization, but it has only been used by a few research labs to date [45]. Another commonly used animal system is the ligated adult rabbit ileal loop model. Fluid accumulation caused by CT can be measured in the ligated intestine, making this model the best for assessing disease severity, but it requires significant surgical manipulation [42]. More recently, experts are working to establish models using mouse fetal intestinal organoids [46].

All of the mammals used for cholera research are not natural hosts of V. cholerae, therefore finding an organism that serves an actual host would be ideal. Over the years, researchers have identified V. cholerae within the intestines of fish and have concluded that fish can serve as a reservoir for V. cholerae [24, 47]. While the first experiments 10

with fish, specifically sardines, and V. cholerae were performed in 1963 [48], it was over another half a century before a zebrafish model using V. cholerae was established [49].

Danio rerio, the zebrafish, has been gaining popularity as a model organism since the

1960s. Of particular interest for cholera research, it is a species native to southeast Asia, where cholera is endemic [50]. V. cholerae colonizes the intestines and causes diarrhea in zebrafish. Importantly, infected zebrafish can disseminate V. cholerae to naïve fish, indicating that this model can also be used to study transmission within the aquatic environment. A notable difference of this model is that none of the human virulence factors (CT, TCP, and ToxR) are required for colonization. Overall, the zebrafish is a natural host for V. cholerae and can be used to study the transmission and colonization of

V. cholerae within the aquatic ecosystem [49].

1.4 Extracytoplasmic Stress Response Systems

As mentioned above, V. cholerae can exist in a number of environments, colonizing protozoans to metazoans, or existing in a free-living state. In order for bacteria to survive such diverse and often harsh niches, they need to be able to sense and respond to changes within the environment. The cell envelope is a surprisingly complex structure that requires constant maintenance and energy to remain functional [51, 52]. If cell envelope homeostasis is disrupted, the bacterium needs to be able to repair the damage or risk losing viability. Therefore, bacteria have evolved ‘danger-sensing’ systems in their membranes to alert the bacterium of fluctuating external conditions and envelope irregularities. These systems are termed extracytoplasmic stress response

(ESR) systems since they are concerned with stress localized outside of the cytoplasm: in 11

the cell envelope. Such systems include two-component systems and sigma factors [53-

55]. The phage shock protein (Psp) response and the σE response are two of those ESR systems. They are both regulated through the sequestration and release of a transcription factor when membrane homeostasis disruption is detected [56]. Although there are other

ESR systems, only the σE and Psp systems will be reviewed further due to their relevance to my thesis project.

σE Stress Response System

Sigma factors are a class of dissociable subunits of RNA polymerase. They bind to the core RNA polymerase and guide transcription of particular genes through promoter recognition. Sigma factors can be broken into two classes, σ70 and σ54, based on structure. Surprisingly, the classes of sigma factors are completely unrelated and share no sequence conservation, even though they both bind RNA polymerase [57, 58]. The

σ70 class can be further broken into two subclasses: the sigma 70-like and the extracytoplasmic function (ECF) sigma factors.

The best-studied ECF sigma factor is σE from Escherichia coli, encoded by rpoE

[59, 60]. In the absence of membrane stress, σE is bound to the cytoplasmic side of the

IM by RseA, an IM spanning protein. RseA is known as an anti-sigma factor since it restricts σE from transcription initiation. However, when a stress response is sensed, a proteolytic cascade results in the release of σE from the membrane. The stress signal is misfolded outer membrane proteins (OMPs) within the periplasmic space. A majority of

OMPs have a C-terminal sequence that, when folded properly, is disguised. When these

OMPs become misfolded, the sequence is exposed and detected by the IM protein, DegS 12

[61]. DegS is a protease that then cleaves the periplasmic side of RseA. Subsequently, another IM protease, RseP, cleaves the cytoplasmic side of RseA, releasing σE from its membrane-bound state [62]. With σE free in the cytoplasm, it guides RNA polymerase to

σE-dependent promoters. The genes within the regulon include those within its own operon, rpoE rseABC, as well as other genes whose products are involved in protein degradation and LPS remodeling [55, 63, 64].

Activation of the σE pathway through two-site cleavage of RseA is a characteristic mechanism of signaling pathways used by prokaryotes and eukaryotes alike. The specific process is known as regulated intra-membrane proteolysis (RIP) [65]. In fact, through bioinformatic surveying, a number of other bacteria have been found to contain orthologues of the σE operon within their genomes [64]. σE and its homologues often play a role in bacterial pathogenesis [57, 59, 60]. This is also true for V. cholerae, where rpoE mutants are less able to colonize infant mouse intestines [66]. The σE response is more thoroughly discussed in Chapter 2, a review of Vibrio responses to extracytoplasmic stress [67].

While the σE regulon has been defined in E. coli, in V. cholerae it is not as well characterized. Therefore, to determine which genes were σE-regulated, Dr. Jyl Matson compared the transcriptome of classical V. cholerae (strain O395) and an rpoE null mutant by RNA sequencing (unpublished). In order to induce the σE response, a sub- lethal quantity of polymyxin B was added to the cultures to create membrane stress.

Overall, 218 genes were differentially regulated, with 96 of them downregulated and 122 upregulated in O395ΔrpoE compared to wild type. Interestingly, a set of genes, pspA, pspB, and pspC were upregulated in the absence of σE (Table 1.1). These genes are a part 13

of the Psp extracytoplasmic stress response. This inverse relationship hinted that the Psp response could potentially serve in a compensatory capacity for the loss of σE. In fact, such a relationship has been observed in Salmonella enterica serovar Typhimurium [68].

However, the Psp response had never been studied in V. cholerae.

Phage Shock Protein Response

The phage shock protein response (Psp) is a system that functions to protect the integrity of the inner membrane. The Psp response was first discovered in 1990 by Peter

Model and colleagues. They noticed that a substantial amount of a 25kDa protein was produced after E. coli was infected with filamentous phage f1, and subsequently named the protein phage shock protein A (PspA) [69]. After further studies, the group realized that PspA was produced under other environmental conditions, such as exposure to ethanol, heat shock, alkaline growth conditions, and protonophores, therefore the phage shock system name is completely historical [70, 71]. A decade after the Psp response was discovered in E. coli, it was also identified in Yersinia enterocolitica during a transposon screen for mutants attenuated for virulence in a mouse model [72]. Currently,

14

the majority of knowledge about the Psp response stems from observations made in E. coli and Y. enterocolitica.

Psp Genetic Organization and Conservation

The Psp system in E. coli is organized as a multigene operon (pspFABCDE ) and an uncoupled gene, pspG. Of note, pspF is adjacent to pspABCDE, but is divergently transcribed. The Y. enterocolitica psp operon is organized in a very similar manner with a few differences: it lacks pspE and contains two extra genes, ycjXF. However, it does share the rest of the psp genes including the unlinked pspG, resulting in the Y. enterocolitica psp operon of pspFABCDycjXF, pspG. In fact, pspFABC are conserved amongst most Gram-negative bacteria and are therefore considered the minimal functional unit of the Gram-negative Psp response [73]. On the other hand, some Gram- positive bacteria only contain orthologs of PspA, indicating that perhaps PspA is the core fundamental protein. Moreover, Psp-like proteins can be found spread across all three domains of life: Eukarya, Archaea, and Bacteria [73-75]. This broad conservation of the

Psp response is not only fascinating, but leads to questions about its origins.

Transcription of the pspABCE operon in E. coli requires the coupling of RNA polymerase (RNAP) and the sigma factor, σ54. Transcription with σ54 requires the presence of an activator protein that will fuel the transition of the RNAP-σ54-promoter from a closed complex to an open complex that is necessary for initiation [76, 77]. The transcriptional activator of the Psp response is PspF, a member of the enhancer-binding protein (EBP) in the AAA+ ATPase family. Generally, EBPs have three domains: an N- terminal regulatory domain, a central ATP binding motif, and a C-terminal DNA-binding 15

domain [77, 78]. Usually the N-terminal domain will regulate the activity of the ATPase domain through environmental changes such as phosphorylation, ligand binding, or protein –protein interactions [77]. Intriguingly, PspF completely lacks this characteristic

N-terminal domain. Instead, PspA forms a complex with PspF and functions as the missing domain to negatively regulate PspF activity [79, 80].

Many in vitro and in vivo studies have demonstrated physical interaction between

PspA and PspF. Two-hybrid analysis confirmed the E. coli PspA-PspF complex and also demonstrated that a single amino acid substitution in PspF was sufficient to prevent association of the complex [81]. Furthermore, in Y. enterocolitica, subcellular fractionation paired with coimmunoprecipitation and fluorescence microscopy showed that PspA was only in the soluble fraction when IM stress was absent. However, when the Psp system is activated, PspA no longer precipitated with PspF and remained mostly membrane bound [82-84]. As a result, PspF remains unchecked in a pspA null mutant and is constitutively active [76, 85].

An interesting feature of PspA is that the PspF inhibitory function is not its only role. PspA is a bifunctional peripheral inner membrane protein that (i) under noninducing conditions, negatively regulates PspF and (ii) under inducing conditions, assumes the role of an effector by binding to the IM and the IM proteins PspB and PspC.

PspA is predicted to have a helical coiled-coil structure comprised of 4 helical domains

(HD1-HD4) [76, 81]. The first N-terminal domain, HD1, has two amphipathic helices

(AHa, and AHb) in which AHa is necessary for binding anionic lipids, and the last domain, HD4, is responsible for higher oligomerization [86, 87]. PspA forms as a hexamer and binds hexameric PspF (6PspF:6PspA) to inhibit transcription of 16

the psp genes [80]. Furthermore, when PspA dissociates from PspF, it has the ability to transition into a 36-mer when the IM is stressed. These large complexes have been visualized in E. coli with cryoelectron microscopy, and they are proposed to directly bind the IM and suppress leakage of protons in vitro [88, 89]. The 36-mer assembles in to a single ring with 9-fold rotational symmetry, leading to the hypothesis that it perhaps forms scaffolds to stabilize the IM [90, 91].

How precisely then do cytosolic PspA-PspF complexes dissociate when stress is occurring in the IM, a completely different compartment within the bacterial cell? As mentioned earlier, the Psp and σE system flip from on-off states through intramembrane cascades that result in the freeing of a transcription factor. For the Psp system, the integral IM proteins PspB and PspC form a complex (PspB-PspC) that may sense most inducers and cause the dissociation of PspA from PspF through sequestration of PspA to the cytosolic side of the IM [93, 94]. However, in E. coli, induction of the Psp response through heat shock is PspB-PspC independent, suggesting that PspA may be a sensor all on its own [71]. Furthermore, a number of PspA homologs have been found in other bacteria including Bacillus subtilis and Burkholderia pseudomallei, but the rest of the psp genes are absent, implying that perhaps PspB and PspC are not signal sensors, but may actually amplify the response or be an additional mechanism to sequester PspA away from PspF [95, 96]. It is possible that the inducing signal of the Psp system may not be relayed from protein to protein, but from membrane to protein, as there is now a considerable amount of evidence that the overall biophysical properties of membranes can regulate protein function [97].

17

The other members of the Psp response are less well-characterized. In E. coli and

Y. enterocolitica, even though physically unlinked from the rest of the psp locus, pspG is positively regulated by PspF under secretin stress and negatively regulated by PspA [92-

94]. PspG is an inner membrane protein that may affect motility, as overexpression leads to decreased motility in E. coli [92, 94]. The E. coli PspE protein is a periplasmic rhodanese, having thiosulfate sulfurtransferase activity [95]. The relationship between the rhodanese activity of PspE and the rest of the Psp response remains poorly understood. Furthermore, PspE does not appear to be essential to the Psp response, since it is absent in Y. enterocolitica. PspD, however, is present in both Y. enterocolitica and

E. coli. Similar to PspA, E. coli PspD is a peripherally membrane-bound protein, but it does not interact with PspA, -B, or -C. Otherwise, PspD has remained mostly uninvestigated, much like the gene products of ycjX and ycjY in Y. enterocolitica [96].

Inducers of the Psp response

As mentioned previously, the Psp system was originally discovered in E. coli during filamentous phage f1 infection, resulting in its historical namesake. Subsequently, the signal that caused the robust production of PspA was narrowed down to a single gene product, pIV, a pore-forming protein also known as a secretin [69]. Secretins are large cylindrical channels that are a major component of four different multiprotein secretion systems. In bacteria, the type II secretion system (T2SS), type IV pilus system (T4PS) and the type 3 secretion system (T3SS) span from the IM to the OM to transport molecules out of the cell. The protein that forms the pore across the outer membrane is the secretin. Furthermore, filamentous bacteriophages have similar secretion machinery 18

used to extrude out of cells that also relies on a secretin, including the Psp inducer, pIV

[97]. These secretins are composed of 12-15 subunits and sometimes need the aid of accessory proteins or pilotins to escort them to their OM destination [98].

pIV is not always efficiently localized to the OM, and a fraction of pIV secretin complexes mislocalize to the IM [99]. This mislocalization event is a part of what induces the Psp response. Another factor is that the secretin must have mislocalized to the IM in its multimeric state in order to trigger the Psp system [100, 101]. In Y. enterocolitica, the T3SS secretins YscC and YsaC were found to be extremely effective and reliable inducers of the Psp system [102, 103]. Even more interesting is that mislocalized secretins were found to be incredibly specific inducers of the Psp system through transcriptomic analysis in both E. coli and Y. enterocolitica [93, 104].

In addition to secretins, a number of other general environmental conditions were found to induce the Psp response. In E. coli, heat, ethanol, hyperosmotic shock, and survival during stationary phase growth triggered the Psp response [69]. The protonophore carbonyl cyanide m-chlorophenyl hydrazine (CCCP) induces the Psp system in E. coli and S. Typhimurium [68, 71]. CCCP is a lipid-soluble chemical that transports protons across the membrane, thereby disrupting the proton gradient and cell metabolism [105]. The membrane-perturbing nature of all of these stressors ultimately led to the classification of the Psp response as an extracytoplasmic stress response system.

Inducing Signal Hypotheses

19

The fact that the Psp system is induced by a diverse number of proteins and conditions leads to the question: how does the Psp system sense the stress? Do all of these factors lead to a single inducing signal or does the Psp response have multiple induction pathways? At this point in time, this important question remains unanswered within the field, but there are several hypotheses.

Initially it was thought that a decrease in proton motive force (PMF) was the global signal that induces the Psp system. The PMF is the energy generated by the flux of protons and electrons across the IM, also known as the proton gradient (ΔpH) and membrane potential (ΔΨ) respectively. There were multiple findings that supported this hypothesis. First, treatment of E. coli with CCCP leads to the induction of PspA [71].

Later it was found that protein export via the PMF-dependent Tat pathway was aided with PspA expression [106]. Moreover, rpoE and psp mutants in S. Typhimurium were more susceptible to CCCP-mediated killing [68]. Furthermore, studies in E. coli and S.

Typhimurium showed that after psp null strains were subjected to Psp inducing events, one component of the PMF, the membrane potential, was decreased. None of the data showed that a Psp inducing event in a psp+ strain showed a decrease in PMF. Therefore, the drop in PMF could not be the unifying signal of the Psp response, unless it is below the limit of detection. In 2011, the plausibility of PMF disruption serving as the common inducing signal was dispelled when Engl and colleagues demonstrated that dissipation of either component of the PMF was not sufficient to induce the Psp response in E. coli

[107].

A promising current hypothesis for the overall inducing signal is membrane- stored curvature elastic (SCE) stress. Membrane SCE stress, also known as curvature 20

frustration, occurs when phospholipids in individual leaflets of the membrane are forced into highly unfavorable conformations [114]. It has been shown that PspA from E. coli can bind vesicles with higher SCE stress in vitro independently from the potential sensors

PspB and PspC [87]. In addition, mislocalized multimeric secretins are likely to cause membrane frustration, supporting the SCE hypothesis. However, the ability to directly quantify the curvature frustration within membranes in vivo remains untestable, as there are no techniques available to do so at this time.

Virulence

Another important aspect of the Psp response is its role in virulence. Currently, out of all the investigated bacterial Psp systems, the Y. enterocolitica Psp system has the most well-characterized role in bacterial virulence. In fact, a transposon screen for avirulent mutants in a mouse model of infection is what led to the initial discovery of the

Y. enterocolitica Psp response. A pspC transposon mutant was severely attenuated in this model system [72]. Y. enterocolitica secretes effectors into host cells using a T3SS, and expression of the Psp system is required during production and assembly of the T3SS

[102]. The theory is that T3SS secretin mislocalization occurs and causes increased membrane permeability, resulting in cell death [108]. Surprisingly, only subtle virulence phenotypes have been found for the E. coli Psp system, with mutants showing small defects in motility, biofilm production, and persister cell formation [93, 109, 110].

The Psp system has ties to virulence in other bacterial pathogens as well. In S.

Typhimurium, Mycobacterium tuberculosis, and Burkholderia pseudomallei, the Psp system is required for intracellular survival within macrophages [111-113]. Similarly, 21

microarray analysis revealed that psp expression is upregulated in Shigella flexneri during macrophage infection [114].

Other Psp systems

Psp systems have been most well-studied in Gram-negative species, as is evident from the aforementioned studies focusing on E. coli and Y. enterocolitica. However, the

Psp system is widely conserved and can be found in all domains of life, Archaea,

Bacteria, and Eukarya. In the Bacterial domain, investigations into the Psp homologs of

Gram-positive bacteria have begun. In the model organism Bacillus subtilis, there are two PspA homologs, PspA and LiaH [115]. Not much is known about the PspA homolog, except that it is involved in envelope stress. However liaH is a member of a three gene system that is highly similar to Gram-negative Psp systems [116]. In fact,

LiaH forms oligomeric rings and associates with the IM under membrane stress, just like in Gram-negative bacteria [117]. Similarly, another PspA homolog has been identified in

Streptomyces lividans, where it demonstrates key characteristics of the Psp response

[118, 119].

The first PspA homolog in an archaeon was identified in the extremophile

Haloferax volcanii through proteomic analysis of salt stress [120]. H. volcanii is a halophile that can be isolated from locations like the Dead Sea, therefore the role of a

PspA-like protein in hypersaline adaptation is interesting, especially given that the rest of the core psp elements are not conserved [120].

In 2001, deletion of a gene in a model organism of plant biology, Arabidopsis thaliana, resulted in disrupted vesicle formation and thylakoid membrane degradation, 22

and therefore the inability to carry out photosynthesis. The protein of interest was named

“vesicle inducing protein in plastids 1” (Vipp1) [121]. Vipp1 is present in plant chloroplasts as well as cyanobacteria, and while similar to the E. coli PspA, it contains an extra C-terminal tail that enables it to repair photosynthetic membranes [90].

Remarkably, while E. coli PspA can form oligomeric rings composed of up to 36 subunits, Vipp1 can form 48 to 68-mer oligomeric rings [90]. It is theorized through phylogenetic analyses that vipp1 was obtained during a possible gene duplication event of pspA [122]. Overall, the conservation of the Psp system across phyla emphasizes the importance of maintaining membrane function in diverse organisms.

1.5 Summary and Aims

The starting point of this project was to determine if there was a relationship between the σE and Psp responses in V. cholerae, but it became apparent that the Psp response needed to be characterized first. Therefore, the overarching goal of this dissertation project was to characterize the phage shock protein response system in V. cholerae through two aims: first, to determine if the psp genes encoded functional products and define the regulatory pathways; second, to understand the physiological relevance of the Psp system. Accomplishment of these aims was intended to provide deeper understanding of the role of the Psp system in V. cholerae stress responses and environmental survival.

23

Chapter 2

Vibrio responses to extracytoplasmic stress

(Published in Environmental Microbiology Reports, 2018. 10(5): p. 511-521. https://doi.org/10.1111/1758-2229.12693)

Cara M. DeAngelis#, Jessica Saul-McBeth#, and Jyl S. Matson*

# Authors contributed equally to this work, with Cara DeAngelis writing the σE response,

Jessica Saul-McBeth writing the Cpx response, and both contributing to the stress relief via OmpU section.

* Corresponding author: Jyl Matson, Ph.D.

2.1 Summary

A critical factor for bacterial survival in any environment is the ability to sense and respond appropriately to any stresses encountered. This is especially important for bacteria that inhabit environments that are constantly changing, or for those that inhabit more than one biological niche. Vibrio species are unique in that they are aquatic

24

organisms, and must adapt to ever-changing temperatures, salinity levels, and nutrient concentrations. In addition, many species of Vibrio colonize other organisms, and must also deal with components of the host immune response. Vibrio infections of humans and other organisms have become more common in recent years, due to increasing water temperatures in many parts of the world. Therefore, understanding how these ubiquitous marine bacteria adapt to their changing environments is of importance. In this review, we discuss some of the ways that Vibrios sense and respond to the variety of stresses that negatively affect the bacterial cell envelope. Specifically, we will focus on what is currently known about the σE response, the Cpx response, and the contributions of OmpU to extracytoplasmic stress relief.

2.2 Introduction

Members of the Vibrio genus are Gram-negative, motile bacteria, typically found in marine or estuarine environments. The majority of species are halophilic, although some can also be found in freshwater environments. The genus contains more than 100 species, and most species possess two circular chromosomes [123]. The bacteria are able to swim freely in the water column or may form biofilms on surfaces. Many species of

Vibrio interact in some way with other organisms, and these interactions can range from being beneficial symbionts to deadly pathogens. Members of the Vibrio genus have some of the fastest replication rates of known bacteria, and therefore are very adept in reacting and adapting to changing environments [124].

Even though there are a large number of Vibrio species, the majority of research to date focuses on those that cause disease in humans. Of these species, the most studied, 25

and the most impactful on human health is V. cholerae. When water containing V. cholerae is ingested by a human, the bacteria colonize the small intestine and produce a toxin that results in voluminous watery diarrhea, leading to severe dehydration and death if untreated [125]. Due to the fact that there are an estimated 3 million cases of cholera each year, V. cholerae is the model Vibrio on which many research labs focus their studies. Two other Vibrio species that also cause significant human disease are V. parahaemolyticus and V. vulnificus [126]. Both bacteria are typically associated with contaminated seafood, with V. parahaemolyticus causing gastroenteritis and V. vulnificus causing more severe infections, including necrotizing fasciitis. Yet other Vibrio species are pathogens of fish and shellfish [47]. For example, V. harveyi is a bioluminescent species that causes disease in a number of aquatic organisms, but is of particular interest in aquaculture due to its detrimental effect on commercially farmed penaeid prawns

[127]. In addition, V. anguillarum is an important pathogen of fish [128]. V. anguillarum is problematic for aquaculture and the fishing industry, and reported to be of particular concern for the European and Asian aquaculture markets. V. coralliilyticus is a globally distributed species that causes fatal infections to a broad range of coral species, as well as other marine organisms [129]. Coral reefs are currently suffering from an unprecedented degradation due to disease, with V. coralliilyticus being an important cause. On the other end of the spectrum is V. fischeri, which is a bioluminescent symbiont of the squid,

Euprymna scolopes. These Vibrio colonize the light organ of the squid, produce light at high densities, and provide protection from predation [130] While there are many other species of Vibrio that do or do not colonize other organisms, this is clearly a diverse genus with global impacts on the marine environment. 26

Living in an aquatic environment subjects bacteria to numerous stresses that must be appropriately responded to for survival. Clearly, the bacteria will encounter changes in temperature, salinity, nutrient availability, and other insults such as phage infection and predation. In addition, those Vibrio species that colonize humans and other organisms must be able to overcome stresses produced by the host, which may include the production of acid, bile, and antimicrobial agents. Most of these insults first effect the outer surface of the bacterial cell, are sensed in some way, and defense measures are enacted in response. In this review, we will discuss a few of the ways that Vibrio species sense and respond to extracytoplasmic stress. Specifically, we will focus on what is currently known about the Vibrio σE response, the Cpx response, and the contributions of

OmpU to stress relief (Figure 2-1).

2.3 The σE response

The extracytoplasmic function (ECF) sigma factor σE, encoded by the gene rpoE, regulates genes involved in stress to the bacterial envelope. The σE response has been most extensively studied in E. coli, where it was first discovered [131]. In the absence of an inducing stimulus, σE is bound by its anti-sigma factor, the inner membrane protein

RseA, to the cytoplasmic face of the inner membrane [62]. When outer membrane proteins misfold and accumulate in the periplasmic space due to stress, RseA is degraded by two proteases, DegS and RseP (YaeL), through a process called RIP (regulated intramembrane proteolysis) [132]. Unbound σE is then released into the cytoplasm where it binds RNA polymerase and guides transcription from σE-dependent promoters. In fact, rpoE has two promoters, a distal P2 promoter regulated by σ70, and a proximal promoter, 27

P2, regulated by σE, therefore creating a regulatory loop in which σE autoregulates its own expression and also the rest of its operon (rseABC) [66]. RseB is a periplasmic protein that negatively regulates σE by binding RseA and blocking RseA degradation by

DegS [133]. RseC, an inner membrane protein, positively regulates σE by functioning as a potential anti-anti-sigma factor and relieving the inhibitory effect of RseA [62, 134].

The other known genes in the σE regulon are involved in the organization and maintenance of lipopolysaccharides and insertion of outer membrane porins (OMPs) [64,

135]. Lipopolysaccharides and OMPs are two key components of Gram-negative outer membranes exposed to the external milieu and σE expresses them when faced with extracytoplasmic stress, in order to adapt and survive in changing environments.

The σE response is conserved in numerous Vibrios and has been characterized to some degree in several species, including: V. alginolyticus, V. angustum, V. cholerae, V. harveyi, V. parahaemolyticus, and V. vulnificus [66, 136-141]. While rpoE is an essential gene in E. coli [142], it is not essential in all bacterial species. In the Vibrio species where the response has been studied, rpoE has been found to be essential in some cases, and not in others. In fact, there appears to be differences in the requirement for rpoE within different strains of the V. cholerae species. In V. cholerae El Tor strains, deletion of rpoE results in suppressor mutations accumulating in the promoter region of ompU

[137]. However, in classical strains, there is no evidence for rpoE being essential [66,

143, 144]. In V. harveyi, attempts to delete rpoE were unsuccessful, suggesting that rpoE is an essential gene in that species [139]. In the other Vibrio species in which rpoE has been studied, including V. parahaemolyticus, V. vulnificus, V. alginolyticus, and V. angustum, it does not appear to be essential [136, 138, 141, 145]]. These reported 28

differences most likely suggest that there are some differences in the σE regulon between species, and even strains, of Vibrio.

Multiple stressors have been shown to induce the σE response, and all of these stressful conditions are presumed to result in the accumulation of misfolded proteins in the periplasm. However, some of the identified σE-inducing conditions are not shared among all Vibrio species where they have been tested (Table 3.1). A general outcome of stress exposure in mutants lacking rpoE is decreased survival in its presence. For example, when V. cholerae (Classical and El Tor), V. harveyi, and V. parahaemolyticus are exposed to the antimicrobial peptide (AMP) polymyxin B (PMB), the V. parahaemolyticus and V. cholerae El Tor rpoE mutants show decreased survival where

Classical V. cholerae and V. harveyi do not [66, 139, 145, 146]. Likewise, V. alginolyticus and V. vulnificus rpoE mutants were sensitive to hydrogen peroxide stress

[138, 141] whereas V. harveyi rseBC and V. cholerae rpoE mutants were not [66, 138,

139]. While there are some differences in inducing stressors, overall these results indicate that σE is a key regulator in responding to extracellular stress.

The E. coli σE regulon may contain over 80 genes that, when induced, combat envelope stress [64]. In V. cholerae Classical and El Tor biotypes, σE regulates the type

II secretion system (T2S) [135, 147-149]. In addition, in V. alginolyticus, σE regulates

LuxR, a quorum sensing regulator, in a temperature dependent manner [141]. A number of other potential members of the rpoE regulon were found in a microarray comparing transcripts from an rpoE mutant to wild type strain in El Tor V. cholerae [135]. Some of the genes positively regulated by σE, whether indirectly or directly, include those encoding σ32, protease VC0554, protease DO, σ70, σE, TolC, and a peptide ABC 29

transporter (VCA0588, VCA0590, and VC0591). In concordance with these results, multiple periplasmic and outer membrane proteins were altered in an rpoE mutant of V. angustum [136]. In addition, a transposon mutagenesis screen was performed to examine regulators of hemolytic activity, and rpoE mutants had reduced hemolysis on sheep blood agar in V. harveyi [139]. In conclusion, all of these genes are linked to σE and therefore likely to be involved in the response to extracytoplasmic stress.

A number of factors, seemingly unrelated, have been identified as upstream regulators of rpoE in Vibrio. OmpU was found to be necessary for inducing the rpoE response in V. cholerae [146], however in contrast, OmpU was found to have no role in the rpoE response of V. parahaemolyticus [145]. In addition, when Hfq, an RNA- binding protein, is deleted σE levels are significantly increased, suggesting that it negatively regulates rpoE expression [135]. Furthermore, H-NS, a histone-like nucleoid structuring protein that binds to bacterial promoters to regulate transcription, negatively regulates the σE response. An hns mutant has an elevated σE response, however, H-NS does not directly bind the rpoE promoter [150]. The relationship between these upstream regulators and rpoE expression require more study to be fully understood in the Vibrios.

In some Vibrio species, strains lacking rpoE or rseB show different colony morphologies. In V. angustum, rpoE mutant cells form microcolonies on Vaatanen nine- salt solution (VNSS) agar, even at optimal growth temperatures (22-30°C), indicating that the cells are in a stressed state [151]. In addition, at 30°C, V. angustum rpoE mutant colonies are significantly larger in area in comparison to WT cells, but do not exhibit this

30

Table 2.1. Stress-associated phenotypes of the RpoE response in Vibrio species

Species Mutation σE Phenotypes Other stresses tested

V. alginolyticus rpoE Decreased survival in N/A [141] sucrose, ethanol, heat, H2O2 and NaCl V. angustum rpoE Decreased survival after No survival defects in [136] heat shock and oxidative exponential phase stress combined with growth, stationary carbon starvation phase, or carbon starvation V. cholerae rpoE Decreased survival in 3% Not sensitive to bile [66, 146] ethanol (Classical), 1000 salts, H202, PMB, pH, fold more sensitive to P2 or changes in antimicrobial peptide and osmolarity (Classical) PMB (El Tor)

V. parahaemolyticus rpoE Decreased survival in Not sensitive to acid [145] PMB, ethanol, growth stress, bile salts, SDS, sensitivity with high or high salinity temperature and low salinity V. vulnificus rpoE Decreased survival in Not sensitive to cold [138] ethanol, H202, SDS, and stress high temperature V. vulnificus rseB N/A Not sensitive to [138] ethanol, H202, SDS, or high temperature

31

Table 2.2. Role of RpoE in host-induced response to stress Species Infection Model Phenotype V. alginolyticus Zebrafish,  rpoE mutants have a higher LD50 [141] intramuscular vs WT injection  rpoE mutants have a competitive defect  WT infected fish die from hemorrhagic septicemia earlier than if treated with an rpoE mutant V. cholerae (Classical Infant mouse,  rpoE mutants have a higher LD50 and El Tor) [66] orogastric vs WT inoculation  rpoE mutants have a competitive defect V. harveyi Shrimp,  rseBC mutants have a competitive [139] intramuscular defect injection V. parahaemolyticus Streptomycin-treated  rpoE mutant has decreased [145] adult mouse, colonization in competition with orogastric WT inoculation  Intestinal localization showed similar amounts of both strains in the small intestine but fewer rpoE mutants in the cecum and large intestine with individual strain inoculation V. vulnificus Iron-dextran-treated  rpoE and degP mutations do not [138] mouse, affect virulence subcutaneous  rseB translucent mutants are infection attenuated for virulence

32

phenotype at 25°C [136]. V. vulnificus rseB mutants also display a deviation from WT morphology: they tend to be less opaque than WT colonies on LB agar with 0.85% salt.[138]. However, after V. vulnificus subcutaneous infection of iron-dextran treated mice, some of the recovered rseB mutants underwent phase variation to a more opaque phenotype (20-50%) while the rest retained the translucent morphology. Bacteria showing this opaque morphotype were not consistently avirulent in mice in comparison to the translucent variant. This led the authors to hypothesize that the rseB mutants may have altered exopolysaccharide expression [138]. However, due to the demonstrated accumulation of suppressor mutations in rpoE mutant V. cholerae [137], it is possible that the stress of host colonization selected for mutations to occur in this species.

In addition to existing in aquatic environments, Vibrios are capable of colonizing a number of hosts that range from protozoans to metazoans [152]. Therefore, Vibrios need to be able to adapt to the different stressors encountered in these diverse niches. In the laboratory, a number of models have been used to examine Vibrio colonization and infection of hosts. The animal models where the σE response has been examined include orogastric inoculation of infant mice or streptomycin treated adult mice, intramuscular injection of zebrafish or shrimp, and subcutaneous infection of iron-dextran-treated mice

[66, 138, 139, 141, 145]. σE is required for virulence in a number of bacteria and Vibrio species are no exception (Table 3.2). Notably, rpoE and degP mutants of V. vulnificus did not show reduced subcutaneous colonization in mice as with other Vibrio species.

Instead, the rseB mutant was attenuated for virulence in the iron-dextran-treated mouse model, which is the first report of a rseB mutant causing attenuated virulence [138]. This attenuation is speculated to have been caused by decreased exopolysaccharide expression 33

either directly or indirectly through the deletion of rseB, or the resulting overexpression of σE was deleterious to the cells. However, it is possible that σE and DegP may play a more important role in an oral route of infection [138]. In conclusion, a functioning σE response is critical for Vibrio survival within hosts.

2.4 The Cpx response

One critical way that bacteria sense and respond to stress is through the activation of two-component systems (TCS) [153]. These signal-transducing networks are able to efficiently sense a signal, transduce information across the cell envelope to the cytoplasm, and activate genes that respond to the stress. CpxRA is an example of a TCS that responds to perturbations in the cell envelope, resulting in the cell returning to homeostasis [153, 154]. Interestingly, while the signals that induce activation of the Cpx response in E. coli have been extensively studied, the signals that induce this system in

Vibrio appear to be quite different and less known [154, 155].

The cpxRA two-component system is one of several stress response pathways found in Gammaproteobacteria, including Enterobacteriaceae and Vibrionaceae [155].

The Cpx response, first described in E. coli, is mediated by a TCS composed of a sensor kinase (CpxA) and a response regulator (CpxR) [156]. CpxA responds to external stimuli through autophosphorylation of a conserved histidine residue [157]. The subsequent phosphotransfer from CpxA to the response regulator CpxR occurs on a conserved aspartic acid residue [153]. Once activated, the response regulator functions as a transcriptional activator of genes necessary to adapt to the signaling stress [158]. The

Cpx system is tightly regulated through multiple mechanisms. In the absence of a signal, 34

CpxA acts on CpxR as a phosphatase, rendering the response regulator inactive [159].

CpxR can act as its own repressor [160], and is indirectly regulated by the periplasmic protein CpxP, which acts to inhibit the autophosphorylation of CpxA. However, the mechanism of this inhibition is not fully understood [161-163].

The Cpx pathway is activated by distinct signals from the σE response [164] and in E. coli, is primarily responsible for sensing misfolded and aggregated proteins, among many diverse activating cues [164-166]. While the Cpx system in E. coli and Vibrio were thought to have the same physiological role, recent work demonstrated that there are different signals that induce the systems between E. coli and Vibrio [155, 165, 166]. For example, Cpx activation in E. coli occurs with increasing osmolarity while in V. cholerae, Cpx stimulation occurs with salinity and not osmolarity [155, 166]. These differences are thought to be due to low amino acid similarity of CpxA between these two species [165]. Although disparities exist between V. cholerae and E. coli, within the

Vibrios, the physiological function of the Cpx system appears to be conserved. This may be due to the high sequence homology between the CpxA periplasmic domains (V. parahaemolyticus and V. cholerae) [167]. Of note, V. fischeri shares the least similarity compared to the rest of the Vibrio genus members [168]. CpxA in V. fischeri has only

54% similarity to V. cholerae and 60% similarity to V. parahaemolyticus, however, no differences in the physiological function of the CpxRA TCS have been reported in the

Vibrios [168].

In the estuarine environment, a stress that Vibrios encounter frequently is changes in salt concentrations. The Cpx system enables V. cholerae to appropriately respond to these changes in salinity [154, 155]. This contrasts from E. coli, in which Cpx does not 35

have a role in adaptation to salinity [155, 160]. The importance of CpxR is exemplified in a cpxR mutant in which there is a reduction of survival when the bacteria are shifted from low to high salt concentrations [155]. Additionally, an increase in cpxP transcription is observed when increasing concentrations of sodium chloride are added to LB media

[155].

In alkaline seawater, the concentration of bioavailable iron is well below the amount required for bacterial survival, suggesting that Vibrio has mechanisms to overcome this stress [154, 169, 170]. In fact, iron levels in open surface waters typically fall below 0.2 nM [171] indicating that some Vibrios, including V. vulnificus, use low iron as a cue to turn on a range of genes related to virulence and pathogenesis [172, 173], transferrin receptors [174], and transport systems that scavenge iron from the environment [175]. Transcriptomic studies of V. cholerae El Tor overexpressing cpxR showed that CpxRA regulates genes directly and indirectly involved in low iron sensing.

These findings were not completely consistent with other studies, which may be due to strain and growth conditions [154, 166]. Genes showing increased expression in the microarray include those involved in iron acquisition, iron transport systems, and iron receptors [154]. This study suggests that the Cpx pathway is activated due to iron chelation, and repressed when iron levels are restored. This is supported by the finding that low cpxP-lux expression was found in normal LB or when iron was supplemented back into the media after iron depletion. Additionally, high promoter activity was observed when iron was absent or low. Further evidence of Cpx involvement in iron scavenging has been illustrated by the role and importance of vibriobactin in an RND

(Resistance-Nodulation-Division) efflux mutant. Vibriobactin, a siderophore, is 36

produced by V. cholerae and is secreted out of the cell to bind iron and bring it back to the cytoplasm through the RND system [175]. In the absence of RND efflux, vibriobactin is unable to exit the cell and chelates iron within the cell to generate an activation signal for Cpx [176]. While the role of Cpx in iron scavenging has only been examined under virulence inducing conditions [166], it would be interesting to determine if the Cpx response is induced when Vibrio colonizes aquatic hosts and/or in seawater.

Another important role of CpxR is responding to AMP stress. In a microarray using overexpressed CpxR by inhibiting CpxA phosphatase activity, a series of genes important to antimicrobial responses were found in V. cholerae O1 El Tor. These genes include ompT (repressed), tolC, vexRAB, and vexGH (induced), which are responsible for influx and efflux of low-molecular weight AMPs [166]. Although CpxR is nonessential for antimicrobial resistance, it does enhance antimicrobial efflux, indicating that it might act to enhance expression of the RND efflux pumps VexRAB and VexGH [166].

VexRAB and VexGH are proton-substrate antiporters that contribute to the development of antimicrobial resistance through efflux of AMPs. Inhibition of the efflux pumps results in antimicrobial susceptibility [177]. Interestingly, Cpx promotes the expression of RND efflux systems, while the Cpx response is induced if RND is inactivated showing reciprocal regulation of the system [166]. The mechanism is unknown, however, it is thought that Cpx induces an alternative pathway to rid the cell of toxic metabolites if the

RNA efflux pumps are not functional. The Cpx response is also expressed during AMP insults in E. coli [178] and Salmonella [179] indicating that this function might be conserved across multiple bacterial species.

37

2.5 Stress relief via OmpU

The outer membrane functions as a selectively permeable barrier that can allow entry of nutrients and prevent toxic compounds from entering the cell [180]. While part of the protection offered by the outer membrane is from the phospholipids [181], 2% of the outer membrane is comprised of outer membrane proteins (OMPs) or porins, providing protection through selective permeability into the cell [182]. These water- filled pores or channels extend across the outer membrane and facilitate uptake of hydrophilic compounds up to certain size and reduce the diffusion rate of toxic molecules, providing resistance to many stressors. In addition, porins in Vibrio are used as an offensive measure, allowing attachment to host tissues for colonization [183-185].

While there are at least 10 major porins currently identified in V. cholerae [182], OmpU is both a major constituent of the outer membrane, comprising up to 60% of the outer membrane proteins [181], and conserved throughout Vibrio [183], indicating its importance to cell physiology.

The regulation of OmpU expression has been well studied in V. cholerae. The

ToxR regulon, most known for regulating virulence, branches into two pathways, one that is ToxT-dependent and one that is not. In the ToxT-dependent pathway, ToxR, along with another DNA-binding protein, TcpP, activate expression of toxT, the protein product of which increases transcription of cholera toxin, toxin-coregulated pilus and other virulence factors. In the ToxT-independent pathway, ToxR induces expression of

OmpU, while decreasing expression of the OmpT porin [186]. Interestingly, other

Vibrionaceae show ToxR-dependent porin regulation, but do not contain tcpP and toxT, which are located on the V. cholerae-specific pathogenicity island (VPI). This suggests 38

that the original role of ToxR was to regulate outer membrane porins, and that V. cholerae evolved to have ToxR also regulate the virulence cascade, possibly leading to its transformation into a human pathogen. [187, 188]

While the presence of OmpU across Vibrio is conserved, the physiological function and expression levels are quite different. For example, V. fischeri does not require ToxR for expression of ompU and does not contain a homolog of OmpT. This illustrates the divergent evolution of OmpU based on the different physiological niches of

Vibrios [185]. Additionally, while the role of OmpU in virulence is well established, particularly in V. cholerae [189], the function of OmpU in adaptation to the natural environment and colonization of aquatic hosts is equally significant and varied across the

Vibrios.

Since Vibrios colonize a wide range of hosts, they must be able to successfully contend with the stress dealt by hosts’ immune systems. One of major hosts of V. splendidus is the oyster Crassostrea gigas. Because invertebrates lack an adaptive immune response, foreign pathogens are fought off exclusively with innate immunity, including AMPs such as defensins [190] and proline rich peptides [191] [192]. Similarly, in the human small intestine, V. cholerae encounters bile and AMPs. BPI

(bactericidal/permeability-increasing) is an AMP found in neutrophils and also expressed on the surface of gastrointestinal epithelial cells [186]. In both V. cholerae and V. splendidus, OmpU plays a role in survival from stress caused by AMPs. A V. cholerae ompU mutant is more susceptible to BPI, P2 (the bactericidal peptides of BPI), and PMB.

However, it was also reported that OmpU is not necessary for resistance to PMB and normal human sera (NHS) [193]. This discrepancy may be due to differences in PMB 39

concentrations used in these studies. In addition, V. splendidus is resistant to high doses of the oyster AMPs, Cg-BPI and Cg-Def due to OmpU expression [192]. Although the role of OmpU in AMP resistance is not fully understood, it is hypothesized that the anionic isoelectric point of OmpU makes it a possible sensor for cationic AMPs. After sensing AMPs, OmpU activates the envelope stress response through DegS, making it a key component of the σE response as well [61, 137]. It is likely that OmpU is involved in

AMP resistance in many Vibrio species [194].

The innate defenses of the human gut include, but are not limited to, gastric acid, bile, and AMPs [186]. V. cholerae must survive the low pH of the stomach before colonizing the small intestine. V. cholerae has the ability to respond to acid stress through the acid tolerance response (ATR), mediated through ToxR. As expected, a toxR mutant is susceptible to organic acids, however, ectopic expression of OmpU is sufficient to restore survival kinetics to wild type levels and is therefore involved in protection of V. cholerae from organic acids [195]. Bile acids are also an important immune constituent as they are detergent-like and bactericidal [196]. The mechanism by which V. cholerae resists bile is through the ToxT-independent branch of ToxR. A V. cholerae toxR mutant has a growth defect when grown with the anionic detergents deoxycholate (a component of bile) and SDS, but not with the nonionic detergent Triton X-100, likely due to the loss of ompU expression. Furthermore, OmpU transcription and expression is increased in the presence of bile [188]. Since toxR is an ancestral gene among Vibrionaceae, this toxR- mediated bile resistance was also examined in V. mimicus, V. fluvialis, and V. parahaemolyticus. All toxR mutant strains in these species were susceptible to bile, deoxycholate and SDS, like V. cholerae. Moreover, OmpU expression was increased in 40

V. fluvialis and V. mimicus, but not V. parahaemolyticus [188]. Bile is also produced by fish, which are colonized by several Vibrio species. V. anguillarum causes hemorrhagic septicemia in marine fish, leading to devastating losses in aquaculture, but may also be a part of normal fish microflora. A V. anguillarum ompU mutant has decreased survival in the presence of bile, similar to other Vibrio species [188, 197]. A toxR mutant exhibits a similar phenotype, and overexpression of OmpU is sufficient to restore survival of the toxR mutant, indicating the relative importance of OmpU in bile resistance.

OmpU also plays a role in the Vibrio response to starvation. To counteract nutrient limitations in the environment, Vibrio produces stress proteins and activates pathways that enable long term survival [198]. After 8 months of starvation in seawater, a microarray comparing OMP profiles of V. parahaemolyticus and V. alginolyticus was performed [198]. OmpU was highly expressed in V. parahaemolyticus, but not V. alginolyticus, suggesting that OmpU plays different roles in the starvation response in these two species. In contrast to V. parahaemolyticus, V. cholerae expresses OmpU in rich medium, while OmpT is predominately expressed in nutrient limiting minimal media and in the aquatic environment [199, 200]. Nutrient limitation also destabilizes or reduces expression of ToxR, indicating that ToxR regulates OmpU during nutrient stress

[199, 201]. This starvation response was then reversed by the addition of NRES

(asparagine, arginine, glutamic acid, and serine) into minimal media, resulting in increased expression of ToxR and OmpU, and decreased expression of OmpT [199].

Another way bacteria overcome nutrient stress is by entering into a VBNC (viable but not-culturable) state. Recently, it was found that a protein important for recovery from

41

VBNC in V. parahaemolyticus, YeaZ, associates with OmpU [202]. Therefore, OmpU may play a role in yet another method of adapting to nutrient stress.

Temperature fluctuations due to seasonal changes can cause dramatic change to the salinity of water. For example, V. vulnificus colonizes oysters found at lower salinities (8.0 ppt, 0.8% salt) and higher temperatures exceeding 20°C, but can also survive during colder months and near the bottom of estuaries at a much lower temperature of 6.4°C and 18.5 ppt (1.85%) salinity [203]. To adapt to these changes,

Vibrios undergo dramatic remodeling of their outer membrane components, in particular,

OmpU. In V. cholerae, cultures grown in 2% standard NaCl had 30% of the outer membrane composed of OmpU. When the same strain was grown in medium without salt, OmpU composed 60% of the outer membrane [181]. This has been further demonstrated in V. parahaemolyticus, which has the ability to grow in salt concentrations from 0.086 M to 1.5 M [204]. Microarray data comparing low salt (0.66%) and high salt

(2%) growth conditions showed that OmpU was repressed in high salt conditions, reflecting its role in low salt outer membrane remodeling and integrity [205]. These findings are supported by the fact that ToxR expression is repressed in high osmolarity

[206] and indicate that OmpU is tightly regulated by toxR in response to changing salt concentrations. Clearly, OmpU is an important component of the Vibrio outer membrane and its defense against a variety of stresses.

42

E Figure 2-1. Model of σ , Cpx, and OmpU involvement in extracytoplasmic stress relief. A variety of envelope stresses including misfolded or accumulated OMPs, fluctuating salinity, and antimicrobial peptides (AMPs) are sensed by the RpoE (σE) and Cpx extracytoplasmic stress response pathways. In the σE pathway, an inducing stress causes RseA to be degraded by two proteases, DegS and RseP, freeing σE. Then σE directs transcription of the members of its regulon to reduce the sensed stress. The Cpx pathway is a two-component system in which CpxA, the sensor kinase, senses envelope stresses and becomes autophosphorylated. The phosphate group is then transferred to CpxR, the response regulator, and CpxR directs transcription of genes whose products are involved in alleviating extracytoplasmic stress. ompU expression is modulated by ToxR in response to AMPs, bile and other fluctuating environmental nutrients. If misfolded OMPs, including OmpU accumulate in the periplasm, the σE pathway is induced.

43

2.6 Concluding remarks

While the majority of Vibrio studies to date focus on those species that are pathogenic to humans, there are scores of different species that inhabit the aquatic environment worldwide. Many of these species have a significant impact on organisms other than humans, ranging from beneficial symbionts to harmful aquatic pathogens.

Unfortunately, studies on many of these other Vibrios are generally less advanced, and the nuances of their abilities to sense and respond appropriately to stress are far less well understood. Because the growth of Vibrios in the environment is largely dictated by temperature, we are seeing increased Vibrio populations in many regions of the world due to climate change [11]. This is leading to increasing infections of both humans and other marine organisms. Therefore, studies on diverse Vibrio species are becoming even more necessary, with the overall aim of preventing diseases that impact human health and aquaculture. A better understanding of stress response mechanisms may be useful for developing new strategies to combat these ubiquitous marine organisms.

2.7 Acknowledgements

Work in the Matson lab is supported by startup funds from the University of Toledo. The authors declare no conflict of interest.

44

Chapter 3

Characterization of the Vibrio cholerae phage shock protein response

Published in Journal of Bacteriology, Mar 2019, DOI: 10.1128/JB.00761-18

Cara M. DeAngelisa, Dhrubajyoti Nagb, Jeffrey H. Witheyb, and Jyl S. Matsona#

aDepartment of Microbiology and Immunology, University of Toledo College of

Medicine and Life Sciences, Toledo, OH, USA bDepartment of Biochemistry, Microbiology, and Immunology, Wayne State University

School of Medicine, Detroit, MI, USA

Running head: Vibrio cholerae phage shock protein response

# Address correspondence to Jyl S. Matson, [email protected]

45

3.1 Abstract

The phage shock protein (Psp) system is a stress response pathway that senses and responds to inner membrane damage. The genetic components of the Psp system are present in several clinically relevant Gram-negative bacteria, including Vibrio cholerae.

However, most of the current knowledge about the Psp response stems from in vitro studies in Escherichia coli and Yersinia enterocolitica. In fact, the Psp response in V. cholerae has remained completely uncharacterized. In this study, we demonstrate that V. cholerae does have a functional Psp response system. We found that overexpression of

GspD (EpsD), the type II secretion system secretin, induces the Psp response, whereas other V. cholerae secretins do not. In addition, we have identified several environmental conditions that induce this stress response. Our studies on the genetic regulation and induction of the Psp system in V. cholerae suggest that the key regulatory elements are conserved with those of other Gram-negative bacteria. While a psp null strain is fully capable of colonizing the infant mouse intestine, it exhibits a colonization defect in a zebrafish model, indicating that this response may be important for disease transmission in the environment. Overall, these studies provide an initial understanding of a stress response pathway that has not been previously investigated in V. cholerae.

3.2 Importance

Vibrio cholerae leads a dual life cycle, as it can exist in the aquatic environment and colonize the human small intestine. In both life cycles, V. cholerae encounters a variety of stressful conditions, including fluctuating pH and temperature, and exposure to

46

other agents that may negatively affect cell envelope homeostasis. The phage shock protein (Psp) response is required to sense and respond to such insults in other bacteria, but has remained unstudied in V. cholerae. Interestingly, the Psp system has protein homologs, principally PspA, in a number of bacterial clades, as well as in archaea and plants. Therefore, our findings not only fill a gap in knowledge about an unstudied extracytoplasmic stress response in V. cholerae, but may have far-reaching implications.

3.3 Introduction

Bacteria have evolved to survive in an astounding number of habitats by monitoring their internal and external environments and modifying their genetic regulation accordingly. The only barriers to separate the interior of the cell from the outside environment are its membranes and periplasmic space, or the cell envelope.

Bacteria have complex membranes that define cellular shape, generate energy, and provide protection, while simultaneously maintaining permeability to nutrients, and are the site of numerous other essential cellular processes [207]. Gram-negative bacteria are surrounded by two membranes, the inner membrane (IM) and the outer membrane (OM).

Any threat to the stability of either membrane could lead to loss of viability. To guard against damage to the cell envelope, bacteria have signal transduction systems called extracytoplasmic stress responses (ESRs) that monitor the integrity of the membrane compartments. Once an ESR has been initiated, proteins are produced that function to restore cell homeostasis [55].

Gram-negative bacteria have a number of characterized ESRs including the σE,

Cpx, Bae, Rcs, and phage shock protein response (Psp). The σE, Cpx, Bae, and Rcs 47

pathways involve regulation of a wide array of genes, whereas the Psp response is tightly regulated [55, 92, 104]. The Psp response was initially discovered by Peter Model and his colleagues when they found that f1 filamentous phage infection of Escherichia coli resulted in production of a 25 kDa protein [69]. They subsequently named the protein phage shock protein A (PspA) and determined that its production was induced by the phage gene product pIV [69]. However, they later discovered that this response was not limited to phage infection, but that a number of other stressors, such as ethanol, heat, osmotic shock, and stationary phase growth were also inducers [69, 71]. The Psp system includes several genes, with the core set, pspFABC, considered to be the minimal functional unit in most Gram-negative bacteria [73]. Oftentimes the systems include an uncoupled gene, pspG, and may also include additional, less understood genes, that play a role in the Psp response. These systems are typically inactive when PspF, the transcriptional activator of the system, is bound in a complex with PspA. However, when membrane disruption occurs, it is sensed by PspB and PspC. PspA then releases PspF and pspABC transcription occurs, resulting in the production of a protective response [83,

208, 209]. The exact mechanism by which the individual components of the Psp system ameliorate membrane stress is not yet fully understood [210].

Secretins are homomultimeric pores that facilitate the transfer of macromolecules across the outer membrane of bacteria [211]. These cylindrical multimers are formed from 12-15 monomers and are a major component of 4 different classes of secretion systems: the type II secretion system (T2SS), type III secretion system (T3SS), type IV pili system (T4PS), and phage extrusion [97]. A common theme in the study of Psp systems has been that secretins often induce the response. The phage gene product pIV, 48

produced for phage extrusion out of the cell, is a secretin that localizes to the bacterial outer membrane. However, during phage infection some amount of this protein often mislocalizes to the inner membrane, stimulating the Psp response [99, 212]. When the

Psp system was characterized in Yersinia enterocolitica, the T3SS secretins, YsaC and

YscC, were also discovered to be inducers of the Psp response [102, 103]. Historically, the Psp response has been most thoroughly studied in E. coli and Y. enterocolitica, with limited characterization in a number of other bacteria [213].

Vibrio cholerae is a Gram-negative bacterium that is naturally found in aquatic ecosystems, but can also colonize humans, causing the severe diarrheal disease cholera

[4]. Although there are hundreds of V. cholerae serogroups, only the O1 and O139 serogroups are capable of causing pandemic cholera. The O1 serogroup is subdivided into the classical biotype, which caused the first six cholera pandemics, and the El Tor biotype, which has been the predominant cause of cholera since 1961 [18]. Regardless of location, V. cholerae is exposed to numerous stressors that negatively impact membrane integrity [151]. While other ESRs have been characterized in V. cholerae, the Psp response has yet to be studied in V. cholerae. Therefore, the aim of this work was to determine if V. cholerae has a functional Psp response and identify its inducers. In the results presented here, we show that V. cholerae does contain a functional Psp system.

As anticipated, PspA functions as a negative regulator of the system, whereas PspF,

PspB, and PspC are critical for initiation of the response. In addition, we found that the

V. cholerae Psp system is highly induced by overexpression of the T2SS secretin, GspD

(EpsD), but not by any of the other encoded secretins. We have also identified several environmental conditions that activate the response, including stationary phase growth, 49

osmotic shock, SDS treatment, heat, and ethanol stress. Furthermore, the Psp system is important for colonization in the zebrafish model of cholera infection, suggesting that it may be required for environmental transmission of disease.

3.4 Results

Genetic organization of the psp genes in V. cholerae

Previous transcriptomic studies suggested that the V. cholerae Psp system may compensate for the loss of another ESR, the σE response. ([135], unpublished data). In fact, a similar observation has been made in Salmonella enterica serovar Typhimurium

[68]. However, before examining the regulatory relationship between the two ESRs, we first wanted to determine whether or not V. cholerae has a functional Psp response. The

V. cholerae psp locus was previously predicted to contain the pspFABC and pspG genes during a TBLASTN search examining conservation of Psp systems [73]. The amino acid identities between the E. coli and V. cholerae Psp genes from PspF, -A, -B, -C, and –G are as follows: 62%, 58%, 50%, 36% and 45%, respectively. The N-terminal region of

PspA tends to be highly similar between species, and V. cholerae is no exception (Fig. 3-

1B). In PspA, the first 60 out of 222 amino acids has 75% identity and 96% similarity to the E. coli PspA. The N-terminal region of PspA is important for membrane binding, and this conservation suggests that the V. cholerae PspA may also retain those characteristics

[87]. The unlinked pspG is found between 2,750 and 2,390bp downstream of pspF in E. coli, S. enterica, and Y. enterocolitica, however pspG is 1,373bp upstream of pspF in V. cholerae. Subsequent TBLASTN analysis has revealed that V. cholerae also

50

Figure 3-1. Genetic organization of Psp systems and PspA sequence similarity. (A) Comparison of the genetic organization of the psp genes in V. cholerae with that of three well-studied Gram-negative species, E. coli, S. enterica, and Y. enterocolitica. (B) Alignment of the amino acid sequences of PspA for the four species. Multiple sequence alignment created with ExPASy 3.21 BOXSHADE server (https://sourceforge.net/projects/boxshade/). Black shading or * = fully conserved residues, grey shading or . = semi conserved residues, white = no conserved residues.

51

encodes a gene homologous to pspE (Fig.3-1A). In E. coli and S. enterica, pspE remains coupled to pspFABCD. Interestingly, in V. cholerae pspE is unlinked from the core set of psp genes, residing on the second chromosome, and has 42% similarity to E. coli PspE

(Fig. 3-1A). The E. coli PspE is a rhodanese (thiosulfate sulfurtransferase) and structural predictions of V. cholerae PspE shows it is also likely to be a rhodanese (data not shown)

[95].

The GspD secretin induces the Psp response

While the genetic components for a functional Psp system are found in the V. cholerae genome, whether the system is active under any conditions is unknown.

Therefore, we first wanted to determine if the Psp response was functional in V. cholerae.

To that end, we needed to identify an inducer that activates the system. Previous studies characterizing Psp response systems in other bacteria have shown that secretin proteins are remarkably specific inducers of these systems. This is thought to be a result of secretin mislocalization to the bacterial inner membrane (IM) when these proteins are highly expressed [92, 104]. Therefore, we wanted to determine if overexpression of any of the V. cholerae secretins would induce the response, similar to that observed in Y. enterocolitica.

The V. cholerae serogroup O1 classical biotype contains multiple secretion systems: general secretion via the Tat and Sec pathways, T2SS, T4PS, and T6SS, but only the T2SS and the T4PSs contain a secretin protein as a part of their outer membrane machinery. Furthermore, T4PS systems can be divided into two subclasses based on their mode of assembly, type of pilins, and function. T4aPS generally functions in 52

twitching motility and transfer of DNA, whereas T4bPS is usually involved in host colonization in enteric pathogens [214-216]. Classical V. cholerae strain O395 encodes three T4PS systems, though only one is fully functional [217, 218] O395 has one T4bPS, the toxin-coregulated pilus system (TCP), which is involved in attachment to mammalian intestines [219]. O395 also encodes the mannose-sensitive hemagglutinin complex

(MSHA) and Pilus (Pil) systems, which have associated secretin proteins. Therefore, we wanted to determine whether all or any of these systems contained secretins that were capable of activating the Psp system when overexpressed. The four secretins identified are as follows: GspD (aka EpsD; general secretory pathway, T2SS), TcpC (TCP, T4bPS),

PilQ (Pil, T4aPS) and MshL (MSHA, T4aPS).

In order to determine if the system was being induced, we constructed a reporter of pspA promoter activity, as in other studies [102]. We generated a chromosomal reporter fusion where the promoter of pspABC was inserted upstream of the V. cholerae endogenous lacZ gene, denoted as lacZPpspA (Table 3.1). We also confirmed that pspABC are cotranscribed, and therefore the reporter fusion represents not just pspA transcription, but also pspB and pspC (Fig. 3-S1). Each of the four identified V. cholerae secretins were then overexpressed in this background and pspA promoter activity was assessed using a -galactosidase assay. We found that only the T2SS secretin, GspD, is capable of inducing the Psp response when overexpressed (Fig. 3-2A). Additionally, each secretin overexpression construct was designed to contain a C-terminal 6xHistidine epitope to measure protein expression levels and stability. To determine if the lack of induction was due to decreased protein levels, we used immunoblotting to detect relative

53

amounts of the secretins in the cultures. The inability of the T4PS secretins to induce the

Psp response does not appear to be due to a defect in expression or stability, as they are expressed at high levels, especially in comparison to the low expression of GspD (Fig. 3-

2B). We also constructed a V. cholerae strain where pspA was tagged with a chromosomal C-terminal 6xHistidine epitope for detection. This strain was used to measure native PspA protein levels in the presence of the overexpressed secretins via immunoblotting. Further supporting the transcriptional fusion data, only GspD overexpression was capable of inducing expression of PspA (Fig. 3-2B).

Multimeric stability of secretins determines whether it initiates the Psp response

Intriguingly, levels of the monomeric form of the GspD protein are remarkably low in comparison to that of the other T4PS secretins (Fig. 3-2B). In Y. enterocolitica, secretin mislocalization and multimerization are necessary to induce the Psp response

[100]. Some secretins, like TcpC, form stable multimers in SDS sample buffer unless heated above 65°C, whereas phage protein pIV forms stable multimers even after boiling at 100°C in SDS sample buffer [219, 220]. It has been reported that, if a secretin fails to form SDS and heat-stable multimers, the Psp system is not induced [100, 220].

Therefore, we examined the ability of all four secretins to form high molecular weight,

SDS and heat-stable multimers by separating them by electrophoresis using a gradient gel. All samples were suspended in sample buffer containing SDS and boiled for 5 minutes before loading onto the gel. Only GspD forms SDS and heat-resistant multimers, while the other secretins dissociate into monomers during this treatment (Fig.

3-3A). 54

Figure 3-2. The T2SS secretin, GspD, induces the Psp response in V. cholerae. The four identified secretins of V. cholerae (GspD, TcpC, MshL, and PilQ) were overexpressed from pTTQ18, containing C-terminal His6 epitopes for detection. (A) Activity of the pspA promoter as detected by β-galactosidase production from a lacZΩPpspA reporter. (B) Chromosomal expression of PspA in response to overexpressed secretins (B). Statistical significance was determined using the Student’s t- test (***, p < 0.001; Error bars represent +/- standard deviation).

55

This helps to explain our previous observation that GspD protein levels appear to be lower than that of the other overexpressed secretins (Fig. 3-2B). In fact, the GspD present in the samples was not initially detected since the vast majority was not dissociated into monomers and was running at a very high molecular weight. This result supports Dr. Andrew Darwin and colleagues’ hypothesis that only multimeric secretins are sensed to be a threat by the Psp system and result in induction of the response [100].

Another part of this theory is that a significant amount of the multimeric secretin complex mislocalizes to the IM. Therefore, we wanted to determine the localization of multimeric

GspD in V. cholerae. Total membranes were isolated from V. cholerae overexpressing

GspD and subsequently separated into inner membrane (IM) and outer membrane (OM) fractions. OmpT and ToxR were used as OM and IM fractionation controls, respectively.

While we observed predominately GspD monomers in the OM fraction, the vast majority of the GspD multimers were found in the IM fraction (Fig. 3-3B). This supports the hypothesis that overexpressed GspD is forming mislocalized multimers in the IM, which are likely sensed by the Psp system, initiating the response.

GspD overexpression increases expression of other psp genes and causes PspA membrane association

In order to gain a more comprehensive understanding of the effect of GspD overexpression on psp gene expression, we examined pspA, pspB, pspC, pspF, pspG and pspE transcript profiles. V. cholerae was grown with and without overexpressed GspD and RNA was isolated. After cDNA generation, transcript levels for each of the

56

Figure 3-3. GspD forms heat resistant multimers that mislocalize to the inner membrane, where the other V. cholerae secretins do not. A) Cultures were grown to mid-log when protein production was induced by the addition of arabinose for one hour. Samples were resuspended in sample buffer containing SDS and boiled for 5 minutes before separation by electrophoresis. B) Induced cultures were fractionated into total membrane (M), inner membrane (IM) and outer membrane (OM) fractions and GspD multimer and monomers were detected using anti-His antibody. OmpT-FLAG and ToxR were used as outer membrane and inner membrane controls, respectively. Figure is representative of 3 experiments. 57

predicted psp genes was measured using qRT-PCR. Transcript levels were normalized to the housekeeping gene, recA. As expected from the transcriptional reporter results, pspA, pspB and pspC transcript levels were highly elevated in response to GspD overexpression

(Fig. 3-4A). pspG was also highly expressed in response to secretin overexpression.

This is the first evidence that pspG is involved in the Psp response in V. cholerae. In addition, we also found that pspE and pspF expression remained completely unaltered by secretin overexpression (Fig. 3-4A). PspF expression is negatively autogenously controlled in E. coli through blockage of the pspF promoter by RNA polymerase and

PspF itself during pspABCE transcription [79, 221]. Therefore, we did not expect to observe elevated levels of this transcript under these conditions. The lack of pspE transcript elevation with secretin overexpression suggests that this gene may not encode a functional member of the Psp response in V. cholerae. However, these results do not rule out the possibility that PspE may be induced and functional under other conditions.

In a number of bacteria, PspA has been shown to be localized to both the cytoplasm and membranes, leading to its designation as a membrane-associated protein

[69, 94, 118]. However, in Y. enterocolitica, secretin overexpression resulted in increased PspA association with the membrane [83, 94]. Therefore, we analyzed the effect of GspD overexpression on PspA localization. Without GspD induction, PspA expression is low, and it is primarily localized to the soluble fraction (Fig. 3-S2).

However, upon GspD induction, PspA is highly expressed and predominantly associated with the membrane fraction (Fig 3-4B). These results suggest that PspA spatial localization changes with the presence of an inducing stimulus, similar to what was observed in Y. enterocolitica [83]. 58

Figure 3-4. GspD overexpression increases psp transcript levels and leads to PspA membrane association. Cultures were grown to mid-log phase when protein production was induced by the addition of arabinose for one hour. (A) RNA was harvested and reverse transcribed. Transcript levels were normalized to the housekeeping gene recA (B) Induced and uninduced cultures were fractionated into soluble and membrane fractions. Crp-FLAG and ToxR were used as cytoplasmic and membrane controls, respectively. Figure is representative of 3 experiments.

59

Regulation of the Psp response

In other characterized Gram-negative Psp systems, PspA functions as a negative regulator, binding PspF and preventing it from activating transcription from the pspA promoter [102, 210, 222]. To determine if Psp regulation is similar in V. cholerae, we made deletions of each psp gene in the strain containing the lacZPpspA reporter fusion.

Deletion of any of the psp genes did not result in observable growth defects (data not shown). In the absence of induction by secretin overexpression, lacZPpspA reporter activity is generally 40-50 Miller units. However, when pspA is deleted, there is a 50- fold increase in reporter activity in spite of the lack of inducer, which is consistent with other systems (Fig. 3-5A) [102, 222]. Overexpression of pspA from a plasmid in the deletion strain dramatically reduces the activity of the reporter. This suggests that PspA is a negative regulator of the Psp response in V. cholerae.

In the other bacteria where the Psp response has been characterized, PspF, PspB, and PspC are all positive regulators of the system [70, 102]. In E. coli, PspF has been shown to be the transcriptional activator of the psp operon [223]. PspB and C are inner membrane proteins that are required to sense membrane damage and bind PspA after Psp induction in Y. enterocolitica [85, 102]. In order to determine whether PspF, PspB and

PspC have conserved positive regulatory roles, we induced the response by GspD overexpression and tested for loss of Psp induction in the mutant strains. When GspD was overexpressed, lacZPpspA activity increased 7-fold in the absence of any psp deletion (Fig. 2-5B). However, in the pspB, pspC, pspBC and pspF mutants, the secretin was no longer capable of inducing increased activity from the reporter. pspB and pspC were tested individually and in combination, as their sequences partially overlap. 60

Complementation studies resulted in restoration of activity for each of the mutants, however, these studies were hampered by the necessity of using multiple plasmids and antibiotics which negatively impacted baseline expression levels. Also, addition of

PspBC or PspF without GspD overexpression results in Psp activation, providing further evidence that they are positive regulators (Fig. 3-S3 and data not shown). Overall, these data suggest that PspB, PspC and PspF play positive regulatory roles in the Psp response in V. cholerae.

Environmental inducers of the Psp response

A number of environmental conditions including heat shock, ethanol, osmotic shock, stationary phase growth, and treatment with the protonophore CCCP, were found to increase PspA expression in E. coli [69, 71]. In addition, increasing the alkalinity of cultures using NaOH, subjecting bacteria to the detergent SDS, and hyperosmotic shock were shown to be toxic to pspA mutants in Streptomyces lividans, suggesting that the Psp system was needed to survive in the presence of those stressors [118]. Due to the varied environmental inducers of the Psp response in other bacteria, we tested a range of possible conditions to determine if any induced the response in V. cholerae. Similar to what was found for E. coli and S. enterica, stationary phase growth increased PspA expression (Fig. 3-6A) [68, 71]. Exposure to ethanol stress and SDS consistently induced

PspA expression in comparison to untreated cells (Fig. 3-6B). Heat shock induces PspA expression over a short period of time, before cellular death begins after 30 minutes of exposure. Osmotic shock, mediated by the addition of salt, and increased alkalinity, through NaOH treatment, modestly induce the PspA expression, however these 61

Figure 3-5. PspA is a negative regulator of the Psp response, and PspB, -C and -F are positive regulators. Cultures were grown to mid-log phase (4 hours). β-galactosidase activity was produced from a chromosomal promoter fusion of lacZΩPpspA. A) Loss of pspA leads to a large increase in PpspA activity, which can be reduced by complementation. B) Loss of pspB, pspC, or pspF results in the inability to induce the Psp response. Statistical significance was determined using the Student’s t-test (***, p < 0.001; Error bars represent +/- standard deviation).

62

conditions were less reliable inducers of the response (Fig. 3-6B). Unlike in E. coli, S. lividans, and S. enterica, treatment with the protonophore CCCP did not reproducibly produce an increase in PspA expression (Fig. 3-6B). Despite identifying environmental conditions that induced the Psp response, we have not observed significant survival defects when psp mutant strains are exposed to these inducing stressors (data not shown).

These results demonstrate that there are specific environmental inducers of the Psp response in V. cholerae, and illustrate potential differences in the response between bacterial species.

A psp null strain shows reduced colonization in the zebrafish model

The Psp system has been implicated in bacterial virulence in multiple species, and especially well-characterized in Y. enterocolitica [72, 111, 112]. In Y. enterocolitica, a pspC mutant is completely attenuated in a mouse model of infection [72]. Therefore, we wanted to determine if the Psp response was required for successful colonization in two different models of V. cholerae infection, the infant mouse and the zebrafish. The infant mouse is the most commonly used model for the study of factors required for V. cholerae colonization. In this model, infant mice are orally infected with a 1:1 mixture of a wild type and mutant strain in order to determine whether the mutant has a competitive disadvantage in colonization of the intestine [224]. When we examined the psp null mutant (pspFABC) in this model, we did not observe any significant defect in colonization (Fig. 3-7A). Additionally, we examined the ability of the psp null strain to colonize the zebrafish intestine [49]. Fish are natural hosts for V. cholerae and may play

63

Figure 3-6. The Psp response can be induced by specific environmental conditions. A) Cultures were grown for 24 hours and 1mL aliquots were removed at the indicated time points. Cultures were normalized by OD600 and chromosomal PspA-6xHis expression was detected by immunoblotting. B) The indicated stressors were added after 3 hours of growth. After 1 hour (or the indicated time), cultures were normalized by OD600 and total protein concentration and chromosomal PspA-6xHis expression was detected by immunoblotting. Loading control is a cross-reactive protein to ToxR antisera (59). Both figures are representative of three experiments.

64

a role in cholera transmission in the environment. Adult zebrafish were incubated in water containing wild type or mutant bacteria and colonization was allowed to occur for 6 hours. At the 6 hour time point, fish were washed and moved to clean, sterile water.

Both strains survived equivalently in the water over the 6 hour interval (wild type: 2.55 x

106 CFU/mL; pspFABC: 2.97 x 106 CFU/mL). After 18 more hours, the zebrafish intestines were harvested and colonizing bacteria were enumerated. In comparison to the wild type strain (O395∆lacZ), the psp null mutant was significantly less efficient at colonizing the zebrafish intestine (Fig. 3-7B). In addition, to examine the severity of the disease, mucin production and bacterial excretion were measured. While the reduction in mucin and bacterial levels were not statistically significant, there was an overall trend suggesting that the psp mutant causes less severe disease in zebrafish (Fig. 3-7C, D).

Overall, these observations indicate that the psp system in V. cholerae may not be required for colonization in mammals, but may play a role in the environmental transmission of disease.

3.5 Discussion

In this study, we report for the first time that V. cholerae encodes a functional Psp extracytoplasmic stress response system. We found that overexpression of the secretin,

GspD, is a specific inducer of the V. cholerae Psp response. Interestingly, the three other secretins produced by V. cholerae, TcpC, PilQ, and MshL, fail to induce the Psp response. In addition, we found that specific environmental conditions, including ethanol stress and stationary phase growth, cause an increase in PspA expression, similar to the

E. coli Psp response [69, 71]. Furthermore, the core set of Psp proteins, PspF, -A, -B, -C, 65

appear to possess identical regulatory roles as observed in other bacterial systems [102,

222]. Deletion of pspA causes the Psp system to become constitutively active, however, deletion of pspF, pspB, or pspC result in the inability to induce the response. Finally, we found that the Psp system plays a role in zebrafish intestinal colonization, providing a connection to environmental transmission of the organism.

When the Psp system was first discovered, it was theorized that the agents found to induce the Psp response functioned to dissipate the proton motive force (PMF) in the bacterial cell [68, 71, 103]. The proton ionophore, CCCP, disrupts the PMF and has been shown to induce the Psp response in E. coli and S. enterica [68, 71]. Additionally, the membrane potential component of the PMF is decreased in E. coli psp null strains [71,

209]. However, this PMF theory was brought into question by Engl and colleagues in

2011 [107]. They showed that dissipation of either the membrane potential or the proton gradient did not induce the Psp response in E. coli. To complicate matters further, some secretins are very specific inducers of the Psp response [69, 102, 225]. It has been established that the secretins must mislocalize and multimerize in the inner membrane to activate the Psp cascade [100]. The exact damage generated from secretin mislocalization is unknown. It may lead to leakage across the membrane or perhaps it destabilizes the inner membrane and increases membrane-stored curvature elastic stress

[86, 210]. We also found that secretin multimers of GspD mislocalize to the inner membrane upon overexpression, and we hypothesize that the damage generated is capable of signaling induction of the response in V. cholerae (Fig. 3-8).

While V. cholerae is predicted to encode four different secretin proteins, we found

66

Figure 3-7. The psp null strain shows reduced colonization in the zebrafish model of cholera infection, but no defect in infant mice. A) Infant mice were orally inoculated with a 1:1 ratio of 106 O395ΔlacZ and O395ΔpspFABC. B) Zebrafish were incubated in water containing either 108 O395ΔlacZ or O395ΔpspFABC. A and B) After overnight infection, intestines were harvested and bacteria were plated for enumeration. Each data point represents data from one fish or mouse. C) The water was tested for mucin concentration post-infection as a measure of fish diarrhea. The bar diagrams show the mucin level in excreted water after 24 h. D) Bacterial numbers in the water post- infection were quantified to determine the levels of excreted bacteria. (***, p <0 .001; Error bars represent +/- standard error of the mean (B) and standard deviation (C,D)).

67

that only one, GspD can induce the Psp response when overexpressed. Despite secretin similarity in structure, there are differences in how they target and insert into the OM.

Koo et al. separated secretins into five classes based on localization, stability, and requirement of aide in assembly [98]. The classes are as follows: 1) auto-assemble and capable of localizing to OM without assistance, 2) auto-assemble but need aide to reach the OM, 3) auto-assemble and can reach OM without assistance, but do so inefficiently,

4) cannot auto-assemble, but can localize to the OM, 5) cannot auto-assemble or reach the OM. GspD in V. cholerae has been classified as a type 3 secretin, which is consistent with our results. In the case of TcpC in V. cholerae, it has been classified as a type 4 secretin, meaning that it cannot assemble by itself and form stable multimers, again, consistent with our findings. The T4aPS secretins, MshL and PilQ, have not been classified to date. Based on our results suggesting that they do not appear to auto- assemble or induce the Psp system, we hypothesize that they are also class 4 secretins, though it is plausible that they could also be categorized as type 5. The T3SS secretin,

YscC in Y. enterocolitica that causes Psp induction is classified as a type 2 secretin.

Interestingly, the class of the first secretin found to induce the Psp response, the filamentous pIV secretin, has not been determined [98]. The fact that the Psp system has the ability to differentiate between monomeric and multimeric secretins in the IM highlights its sophistication.

In addition to secretins, a range of other environmental stressors are known to induce the Psp response in the bacterial species where it has been studied. In E. coli,

CCCP, heat shock, ethanol, osmotic shock, and stationary phase growth induced the response. In S. lividans, NaOH and SDS were detrimental to pspA mutant growth and 68

survival [69, 118]. Therefore, we tested many of the known inducers to determine if they induced the response in V. cholerae. Ethanol, heat, hyperosmotic shock, and detergent exposure all induced expression of PspA. However, unlike in E. coli, S. enterica, and S. lividans, the proton ionophore CCCP does not reliably increase PspA expression. Again, this brings into question the exact inducing signal that stimulates the Psp response in different bacteria.

Due to the high conservation of the Psp system and the amino acid similarity of the proteins in V. cholerae and E. coli, we anticipated that the individual Psps would maintain similar regulatory roles. We observed that deletion of pspA led to an unchecked

Psp response that was constitutively active. In addition, the ability to activate the Psp response was lost in pspF, pspB, and pspC mutants. These results suggest that PspA is a negative regulator of the system, whereas the other proteins are positive regulators.

While this is consistent with previous reports, the specific roles of these Psps still need further examination. Based on previous studies and bioinformatic analysis, we hypothesize that PspF is the transcriptional activator of the V. cholerae Psp system, binds sigma-54, and guides RNA polymerase to the pspABC promoter. In addition, we can predict that PspA binds PspF to inhibit psp transcription, as it has been shown to do in other bacteria. Furthermore, we can postulate that PspB and PspC are inner membrane proteins necessary to sense the inducing signal and subsequently sequester PspA to the inner membrane. Of note, we discovered that PspC expression is unstable without coexpression of PspB, similar to Y. enterocolitica [82]. Interestingly, expression of PspC in E. coli does not depend on PspB coexpression, accentuating differences in PspB and

PspC between species. In regards to pspG, we can only conclude at this time that its 69

expression is regulated with the rest of the core psp operon. Further studies are needed to determine if there is a PspF binding site in the pspG promoter. Finally, our studies have not revealed a role for PspE in the V. cholerae Psp response. Further investigation is required to determine if it is connected to the response under different growth conditions.

Since the Psp system has been associated with virulence in other bacteria, we wanted to determine if it played a role in V. cholerae pathogenesis [72, 111, 112]. We examined this possibility using two different animal models: the infant mouse and zebrafish. With the infant mice, we performed the classical competition between wild type and a psp null mutant, and did not find that the mutant had any competitive defect.

In the zebrafish, individual infections with both strains were performed. In this model, we found that the psp mutant had a defect in colonizing the zebrafish intestine in comparison to the wild type strain. This difference in colonization leads us to hypothesize that the V. cholerae Psp response may play a greater role in surviving environmental stress and may also contribute to transmission of disease in the aquatic environment.

In summary, we have made the first steps in characterizing the Psp response in V. cholerae, including identifying some of the signals that induce its activity. We have identified both specific protein and environmental stressors that initiate the response.

Future work will continue to investigate the role of the V. cholerae Psp stress response in environmental survival and disease transmission.

70

Figure 3-8. Model of Psp response in V. cholerae. In the absence of stress, the transcriptional activator, PspF, is bound by PspA, inhibiting transcription of psp genes. In the presence of stressors, such as mislocalized secretins from the type II secretion system (T2SS), PspA is sequestered to the inner membrane and PspF is free to initiate transcription of the psp genes. The inner membrane proteins PspB and PspC are predicted to aid in sensing damage and binding PspA to the inner membrane.

71

3.6 Materials and Methods

Strains, media, and growth conditions

All V. cholerae strains were derived from classical biotype O395.

The Escherichia coli strains JM101 and DH5αλpir were used for generating constructs, and SM10λpir was used for conjugation with V. cholerae. All bacterial strains were grown and maintained at 37°C in Luria-Bertani (LB) media or on LB agar plates supplemented with appropriate antibiotics. Plasmids used in this study include the suicide vector pKAS32 [226], the arabinose-inducible expression vectors pBAD33, pBAD18-Kan, pBAD30, and pBAD18 [227], and the tac-promoter driven, IPTG- inducible expression vector pTTQ18 [228]. E. coli strains were transformed by standard methods [229], plasmid DNA was electroporated into V. cholerae, and pKAS32was introduced into V. cholerae by conjugation with SM10λpir. Antibiotics were used at the following concentrations unless otherwise indicated: ampicillin, 100 μg/ml; kanamycin,

50 μg/ml; streptomycin, 100 μg/ml; and chloramphenicol, 30 μg/ml (E. coli), 5 μg/ml

(V. cholerae). Expression from pBAD vectors and plasmid pTTQ18 were achieved through the addition of 0.2% L-arabinose and 0.5mM IPTG respectively.

Plasmid and strain constructions

Plasmids and strains used in this study are listed in Table S1. Primer sequences are available upon request. All constructs were verified by sequencing.

Chromosomal fusions and deletions were created using SOE-ing PCR (splicing by overlap extension) [230]. The lacZPpspA chromosomal fusion was generated by

72

Table 3.1. Strains and Plasmids

Genotype/Features Reference or Source

Strains

O395 Laboratory strain

O395-pspA-His6 This study

O395ΔlacZ Laboratory strain

O395ΔpspFABC This study

O395ΔpspA This study

O395ΔpspB This study

O395ΔpspBC This study

O395ΔpspF This study

O395lacZΩPpspA This study

O395lacZΩPpspA, pspA-His6 This study

DH5α Laboratory strain

JM101 Laboratory strain

DH5αλpir Laboratory strain

SM10λpir Laboratory strain

Plasmids pTTQ18 [228] pKAS32 [226] pBAD18-kan [227] pBAD33 [227] pBAD30 [227]

73

pKAS32-pspA This study pKAS32-pspB This study pKAS32-pspC This study pKAS32-pspBC This study pKAS32-pspF This study pKAS32-pspFABC This study pKAS32-PpspA::lacZ This study pKAS32-pspA-6xHis This study pTTQ18-gspD-6xHis This study pTTQ18-tcpC-6xHis This study pTTQ18-pilQ-6xHis This study pTTQ18-mshL-6xHis This study pBAD18-kan-gspD-6xHis This study pBAD18-kan-tcpC-6xHis This study pBAD18-kan-pilQ-6xHis This study pBAD18-kan-mshL-6xHis This study pBAD33-pspA-6xHis This study pBAD33-ompT-FLAG This study pBAD30-pspA-6xHis This study pBAD30-pspBC-6xHis This study pBAD30-6xHis-pspF This study pBAD30-crp-FLAG This study

74

amplifying 500bp upstream and downstream of the start site of lacZ (VC395_2453,

KEGG) and the promoter region of pspA (VC395_1796) from +210 upstream and -110 downstream of the start site of pspA. The primers encoded 20bp of homology to the

500bp segments so that overlap could occur. The SOE-ing construct was isolated through gel excision, ligated into pKAS32, and transformed into SM10λpir. SM10λpir containing the plasmid was mated with O395, and integration and resolution of the cointegrate was selected for as previously described [231]. The strain containing a chromosomally 6xHis-tagged pspA was created in a similar manner, with the 6xHis added to the C-terminus of pspA prior to the stop codon. After the construct was validated by sequencing, it was introduced into O395 and the recombinant was selected for as above.

β-galactosidase Assays

Overnight cultures of V. cholerae were subcultured 1:100 into LB broth. The strains were grown for 3 hours at 37°C, followed by the addition of either IPTG or arabinose as indicated. Duplicate 100uL aliquots of culture were used to determine - galactosidase activity using the Miller protocol [232].

Immunoblotting

Aliquots of cultures were harvested and normalized based on optical density at

600nm. Cell pellets were either stored overnight at -20°C or placed on ice before resuspension in 1:1 water and 2X Laemmli Sample Buffer (4% sodium dodecyl sulfate

(SDS), 20% glycerol, 120mM Tris-HCl pH6.8, 5% 2-mercacptoethanol, 0.02% 75

bromophenol blue). Samples were boiled for 5min, centrifuged for 1 min at 15,000 rpm, and separated by SDS-PAGE. To detect secretin multimers, 3-10% polyacrylamide gradient gels were used, otherwise all other gels contained 10%, 12.5%, or 15% polyacrylamide. Following electrophoresis, proteins were transferred to nitrocellulose using semi-dry electroblotting. Membranes were blocked in TBST (20mM Tris-HCL,

0.5M HCl, 0.01% Tween) buffer containing 5% milk before incubation with HRP-labeled mouse monoclonal IgG1 anti-His (R and D systems), OctA-Probe (H-5) HRP (Santa-Cruz

Biotechnology), or rabbit ToxR antisera (generously provided by K. Skorupski), followed by secondary goat-anti-rabbit antibody (Thermo Scientific). The protein fractionation controls used were: CRP-FLAG (cytoplasm) and ToxR (membrane) for soluble and insoluble fractionation and ToxR (inner membrane) and OmpT-FLAG (outer membrane) for membrane subcellular fractionation. ToxR antisera also cross-reacts to another low molecular weight protein that can be utilized as a loading control [199]. Proteins were visualized by chemiluminescent detection (Clarity Western ECL Substrate, Bio-Rad) using film or SynGene imager.

Subcellular fractionation

Fractionation into soluble and insoluble fractions was performed as detailed with the following modifications [233]. O395pspA-6xHis containing pBAD30-crp-FLAG and pBAD18-Kan, or pBAD30-crp-FLAG and pBAD18-Kan-gspD-6xHis was grown for 4 or

5.5 hours. The OD600 was measured for normalization and a whole cell lysate fraction was removed and resuspended in 600uL ¼ TES (200 mM Tris-HCl, pH 8.0, 0.5mM

EDTA, 0.5 M sucrose). The remaining culture was pelleted via centrifugation, and 76

resuspended in ¼ TES buffer containing 2 Mm PMSF. The cells were lysed with 4 freeze-thaw cycles, centrifuged, and 700uL was removed for centrifugation at 45,000rpm

(125,649xg) in a TLA-45 Beckman rotor for 45 minutes. The supernatant was retained as the cytoplasmic fraction. The remaining membrane pellet was washed and resuspended in ¼ TES. All samples were stored overnight at -20°C before proteins were precipitated by treatment with 23% trichloroacetic acid (TCA) for 30 min on ice. Samples were pelleted and washed twice with cold acetone. Pellets were air dried, resuspended in SDS- sample buffer, and boiled for 10 minutes before immunoblotting.

Subcellular fractionation to isolate whole membrane, inner membrane, and outer membrane compartments was performed as described with the following modifications

[234]. All steps were carried out on ice or at 4°C unless otherwise specified. O395pspA-

6xHis containing pBAD33-ompT-FLAG and pTTQ18- gspD-6xHis was grown for 5.5 hours with induction by 0.5mM IPTG and 0.2% arabinose for the last hour. The OD600 was measured for normalization and cells were pelleted by centrifugation. The cell pellet was washed with 25mL Tris-NaCl buffer (10mM Tris-base pH 7.5, 100mM NaCl). The pellet was resuspended in 6mL Tris-NaCl buffer containing 10µg/mL polymyxin B and incubated for 10 minutes. The cell suspension was centrifuged, resuspended in 6mL

Tris-NaCl buffer, and stored overnight at -80°C. After thawing on ice, cells were lysed by sonication. The lysed sample was centrifuged for 10 minutes and the supernatant was subsequently centrifuged at 42,000rpm (109,650xg) in a TLA-45 Beckman rotor for 10 minutes. Samples were washed with 500µL Tris buffer (10mM Tris-base pH 7.5) two times to remove the cytoplasmic fraction. The samples were rocked for 30 minutes at room temperature in 360 µL TT buffer (10mM Tris-base pH 7.5, 2% Triton X-100). 77

120µL was removed for the total membrane fraction and SDS sample buffer was added.

The remaining sample was centrifuged at 42,000rpm for 20 minutes to pellet outer membrane fraction. 120µL of supernatant was removed for the inner membrane fraction and SDS-sample buffer was added. The outer membrane pellet was resuspended in TT and centrifuged for 42,000rpm for 10 minutes. The pellet was then washed with Tris buffer and resuspended in 240µL Tris buffer and SDS-sample buffer. Samples were boiled for 5 minutes before immunoblotting.

RNA isolation and qRT-PCR

The indicated strains were grown and normalized to an OD600 of 1.75 and pelleted by centrifugation. RNA was extracted using Trizol (Invitrogen) reagent. Genomic DNA was digested by incubation with DNAse at 37°C for 1 h. RNA was purified by ethanol precipitation and resuspended in RNase-free water and TE buffer. Total RNA was measured using a NanoDrop and 5ug was reverse transcribed with M-MLV reverse transcriptase (Invitrogen). cDNA production (and lack of genomic DNA contamination) was validated using PCR with Taq DNA polymerase (NEB).

qRT-PCR was performed with SYBR green (FastStart Essential DNA Green

Master Version 04, Roche) using a LightCycler 96 and the following PCR conditions: preincubation for 10 min at 95°C, 3 step amplification with 95°C for 10 sec, 51°C for 10 sec, 72°C for 10 sec for 45 cycles, and a final melting phase of 95°C for 10 sec, 65°C for

60 sec, 97°C for 1 sec. Relative quantification was performed using recA as the reference gene and data analyzed using the threshold cycle (2−ΔΔCT) method [235].

78

Zebrafish colonization assay

The zebrafish colonization assay was done following the process described previously [236]. Briefly, adult, wild-type ZDR zebrafish were housed in an automated recirculating tank system (Aquaneering, CA, USA) using water filtered by reverse osmosis and maintained at pH 7.0 to 7.5. Tank water was conditioned with Instant Ocean salt (Aquarium Systems, OH, USA) to a conductivity of 600 to 700 μS. The fish were fasted for at least 12 h prior to each experiment.

For infection, four to five zebrafish were placed into a 400 ml beaker with a perforated lid containing 200 ml of sterile infection water (autoclaved system water). V. cholerae culture was grown with aeration in LB broth at 37°C for 16 to 18 h. Then, cells were centrifuged at 10,000 g for 10 min. The resulting pellet was washed twice with 1X

PBS (pH ~7.4) and resuspended in 1X PBS to an estimated concentration of 109 cfu/ml, determined by measuring optical density at 600 nm. One ml of bacterial inoculum was added to the beaker containing fish in 400 ml infection water. Final V. cholerae cell density used was ∼5 × 106 CFU/ml for this study and was verified by plating serial dilutions of the inoculated infection water. The fish were infected for 6 hours, then washed twice for removal of surface bacteria and kept in fresh, sterile water for 18 more hours. The control group included fish that were exposed to 1 ml of 1× PBS only in place of bacterial culture. Each beaker containing fish was placed into a glass-front incubator set at 28°C with a timed light/dark cycle for the duration of the experiment.

Fish were euthanized in 100 ml of 320 μg/ml Tricaine-S (tricaine methane sulfonate; MS-222; Western Chemical, WA, USA) for a minimum of 25 min, and the intestine of each fish was aseptically dissected and placed into homogenization tubes 79

(2.0-ml screw-cap tubes; Sarstedt, Nümbrecht, Germany) with 1.5 g of 1.0-mm glass beads (BioSpec Products, Inc., Bartlesville, OK) and 1 ml of 1× PBS and held on ice.

Homogenization tubes were loaded into a Mini-Beadbeater-24 (BioSpec Products, Inc.) and shaken at maximum speed for two 1-min cycles, with the samples being incubated for 1 min on ice after both cycles. Intestinal homogenates from each fish were diluted and plated for enumeration on LB agar plates with appropriate antibiotics. Plates were incubated overnight at 37°C and CFU were counted. All animal protocols were approved by the Wayne State University IACUC.

Bacterial count and mucin assay from fish excretory water

50 ml of fish infection water was removed before the fish colonization assay as a control, in duplicate. For all assays, 50ml conical tubes were centrifuged at 10,000 rpm for 15 min at 4°C and supernatant was decanted, being careful not to disturb the pellet.

Each pellet was resuspended in 2 ml of 1× PBS. Unprocessed water samples were stored at 4°C for up to 1 week before analysis.

Excreted water (after infection) was collected as above and serially diluted and plated for enumeration on LB agar plates with appropriate antibiotics. Plates were incubated overnight at 37°C and CFU counted. The mucin content in excreted water was measured as described previously [236, 237]. Briefly, Prior to the procedure, 1 ml of a

50% (wt/vol) periodic acid (Sigma-Aldrich) stock solution was made. A 96-well plate

(Corning Costar; Corning, NY, USA) was loaded with 100 μl/well of the blank (1× PBS), mucin standards, and samples were loaded in triplicate. A volume of 50 μl/well of fresh

0.1% periodic acid solution (10 μl of the 50% periodic acid stock added to 5 ml of 7% 80

acetic acid, used immediately after making) was added and mixed by pipetting. The plate was covered in plastic wrap and incubated at 37°C for 1 to 1.5 h. After incubation, the plate was cooled to room temperature before adding 100 μl/well Schiff's reagent (Sigma-

Aldrich) and mixed with a pipette. The plate was again covered in plastic wrap and placed on a rocker or shaker for 15 to 40 min or until sufficient color developed.

Absorbance was read at 560 nm using a plate reader (Tecan Spectra Fluor plus;

Männedorf, Switzerland). The effective OD of test samples were calculated by subtraction of the PBS control (uninfected fish) water OD from the test (infected) fish excreted water OD.

Infant mouse colonization assay

Four- to five-day-old CD1 mice were inoculated intragastrically with approximately 106 bacteria as previously described [224]. Inoculated mice were incubated at 30°C for 16 h, at which time they were sacrificed and their intestines were removed and homogenized. Serial dilutions of the intestinal homogenates were plated for enumeration. The competitive index was calculated as the ratio of the wild type to the mutant in the input divided by the ratio of the wild type to the mutant in the output.

Statistics

Zebrafish intestinal colonization data was analyzed using Randomized Block

ANOVA, where the blocks are designated as experiment. Data analyses were performed in R version 3.5.2 (www.R-project.org/). All other statistical analyses, t-test and 2-way

81

ANOVA, were performed with Prism version number 5.03 for Windows, GraphPad

Software, La Jolla California USA, www.graphpad.com.

3.7 Acknowledgements

The authors would like to thank Karen Skorupski for generously providing us with the ToxR antibody. The authors would also like to thank Sadik Khuder for assistance with statistical analyses. CMD and JSM were supported by startup funds from the University of Toledo. DN and JHW were supported by U.S. Public Health Service grant R01AI27390 from the National Institute of Allergy and Infectious Diseases.

82

3.8 Supplemental Figures

Figure 3-S1. pspABC are cotranscribed. Cultures were grown to mid-log phase when GspD expression was induced for one hour. RNA was harvested for RT-PCR analysis. Primers spanning from pspA (P1) to pspC (P2) were used for amplification from cDNA made by reverse transcription (+RT), reverse transcription negative (-RT) samples, or from genomic DNA (gDNA).

83

Figure 3-S2. PspA is primarily localized in the soluble fraction under non-inducing conditions. Cultures were grown to mid-log phase before fractionation into soluble (S) and membrane (M) fractions. Crp-FLAG and ToxR were used as cytoplasmic and membrane controls, respectively. WC = whole cell lysate, S = soluble, M = membrane fractions.

84

pspBC

A 600

400 O395

pspA

P O395pspBC

 200

lacZ

0

GspD + pBAD30GspD + PspBC pspF pBAD18-kan + PspBC pBAD18-kan + pBAD30

B 600

400 O395

pspA O395pspF

P

 200

lacZ

0

GspD + PspF GspD + pBAD30 pBAD18-kan + PspF pBAD18-kan + pBAD30

Figure 3-S3. Complementation with pspBC and pspF restore psp activity to their respective deletion strains. Cultures were grown to mid-log phase when GspD and psp expression were induced for one hour. β-galactosidase activity was detected from the chromosomal reporter lacZΩPpspA.

85

Chapter 4

Additional Studies on the V. cholerae Psp response

4.1 Introduction

The overall goal of this thesis was to characterize the V. cholerae Psp stress response system. While we have previously described our major findings on the regulation, induction, and potential physiological outcomes of the Psp response, some minor observations remain undiscussed. This chapter will include some of the unpublished data mentioned in Chapter 3 and expand upon the relationship between the

σE and Psp responses. This additional data adds to our current understanding of the Psp response in V. cholerae.

4.2 Growth phenotypes of psp mutants

A genetic approach was taken in order to examine the regulatory and functional roles of the members of the Psp system. Clean, in-frame deletions of each gene and combinations of the core set of genes, pspFABC, were generated (Table 3.1). Additional strains that were examined but not listed in Table 3.1 are: O395ΔpspABC, O395ΔrpoE, and O395ΔrpoEΔpspA. Additional plasmids used in these studies were: pTL61T-PpspA,

86

pBAD33-pspB-6xHis, pBAD33-pspBC-6xHis, and pBAD33-pspC-6xHis. After the deletion strains were constructed, growth curves of each mutant were performed under regular laboratory conditions, in rich Luria Broth (LB) media at a neutral pH of 7, in order to determine whether deletion of any of the genes or gene combinations resulted in overall growth defects (Fig. 4-1A). Compared to the wild type strain, O395, there are no discernable defects in growth when single and multiple psp genes are deleted. Therefore, we were comfortable in comparing the strains in survival and β-galactosidase assays.

In S. Typhimurium, Psp expression increases under alkaline growth conditions

(pH 9) [68, 71]. Furthermore, in Y. enterocolitica, the absence of pspC causes a severe growth defect in virulence inducing conditions, due to T3SS expression and likely secretin mislocalization [102]. In fact, secretin overexpression in psp null cells is toxic in

Y. enterocolitica, potentially due to increased cytoplasmic permeability from mislocalized secretins [108]. The toxicity of secretin expression in Y. enterocolitica prompted us to explore if a similar phenotype is observable in V. cholerae. Therefore, additional growth curves were performed with all psp null mutants under a range of acidic and alkaline conditions (from pH 6-9) with (+IPTG) and without (-IPTG) expression of the Psp inducer, GspD. In the absence of GspD overexpression, we did not see any growth defects for any of the mutants as compared to the wild type strain.

However, when GspD was overexpressed we observed decreased growth for O395ΔpspA and O395ΔpspFABC under both acidic (pH 6) and alkaline conditions (pH 9) (Fig. 4-1B,

C). All of the other psp mutants examined, ΔpspB, ΔpspC, ΔpspBC, ΔpspABC, and

ΔpspF, showed similar growth patterns, regardless of which genes were deleted (data not shown). Similar growth defects were observed for all the mutants when GspD was 87

overexpressed at pH 7 and 8, but were less pronounced (data not shown). Therefore, it appears that the Psp system is necessary for normal growth when secretins are overexpressed, especially at relatively high and low pH.

4.3 PspC requires PspB for stability

As described in Chapter 3, complementation vectors for the psp null strains were constructed for use in many of the assays to examine Psp function. We first wanted to verify that each of the Psp proteins were expressed from the constructs before proceeding with complementation experiments. During these studies, it was discovered that PspC-

6xHis could not be detected when expressed on its own from a plasmid. PspB-6xHis was readily detected when expressed from the same plasmid (Fig. 4-2). A similar phenomenon was observed in Y. enterocolitica, where stable expression of PspC was only obtained when PspB was coexpressed [238]. Interestingly, no such requirement has been mentioned for E. coli. Therefore, an expression vector was constructed to produce both PspB and PspC from the same plasmid. In this construct, the 6xHis tag was added to PspC. The expression of PspB in tandem with PspC-6xHis resulted in detectable

PspC-6xHis from the construct (Fig. 4-2). As a result, complementation studies were then performed using PspBC-6xHis (Chapter 3, Fig. 3-S3A).

Figure 4-1. Growth phenotypes of psp mutants in neutral, acidic, and alkaline pH. Cultures of the indicated strains were grown in 1mL volumes, shaking, at 37°C for 8 hours under the indicated conditions ((A) pH7, (B) pH6, (C) pH9). (B and C) Expression of the GspD secretin was induced (solid data markers) or not induced (open data markers) by the addition of 0.5mM IPTG. Growth was monitored in a CLARIOstar microplate reader. Data points represent the mean and standard deviation of three biological replicates.

88

1 A O395 ΔpspA 0.1 ΔpspB

ΔpspC OD600 ΔpspBC 0.01 ΔpspABC ΔpspF 0.001 ΔpspFABC 0 2 4 6 8 Growth (Hours)

1 B

WT - IPTG 0.1 WT + IPTG ΔpspFABC - IPTG OD600 ΔpspFABC + IPTG 0.01 ΔpspA - IPTG ΔpspA + IPTG

0.001 0 2 4 6 8 Growth (Hours)

10 C

1 WT - IPTG WT + IPTG 0.1

ΔpspFABC - IPTG OD600 ΔpspFABC + IPTG 0.01 ΔpspA - IPTG ΔpspA + IPTG 0.001 0 2 4 6 8 Growth (Hours) Figure 4-1. Growth phenotypes of psp mutants in neutral, acidic, and alkaline pH 89

Figure 4-2. PspC expression is unstable in the absence of PspB. O395ΔpspB, O395ΔpspBC and O395ΔpspC containing their corresponding complementation vectors (pBAD33) were grown to mid-log phase (4 hours). Cultures were uninduced or induced at the start of growth or after 3 hours of growth with 0.2% arabinose to induce protein expression. Samples were normalized to OD600 and resuspended in sample buffer containing SDS and boiled for 5 minutes before separation by electrophoresis and immunoblotting. PspB-6xHis and PspC-6xHis were detected using anti-His antibody. Figure is representative of 3 blots.

90

4.4 Investigating the relationship between the Psp and σE response

As mentioned in Chapter 1, we first became interested in the Psp system in V. cholerae after comparison of the transcriptome of O395 and O395ΔrpoE (Table 1.1). We saw that expression of pspA, pspB, and pspC were significantly elevated in the absence of rpoE, suggesting that the Psp response might play a compensatory role for the E response. Therefore, we attempted to validate the RNA-seq results using quantitative reverse transcription PCR (qRT-PCR). RNA was isolated from O395 and O395ΔrpoE grown in the presence of sublethal polymyxin B, to reproduce the conditions used for the

RNA-seq analysis. After cDNA generation, transcript levels for pspA, pspB and pspC were measured using qRT-PCR and normalized to the reference gene, recA. While we did observe somewhat higher expression of the psp genes in the absence of rpoE, the qRT-PCR results were not as significantly high as the RNA-sequencing levels (Table

4.1).

Fold Change Average 2-∆∆Ct (Fold Change) Gene RNA-seq qRT-PCR pspA 3.6 1.5 ± 0.4 pspB 2.8 1.6 ± 0.4 pspC 2.7 1.5 ± 0.2

Table 4.1. Comparison of RNA-seq and qRT-PCR results of psp levels in O395 versus O395ΔrpoE. Cultures were grown to mid-log phase when polymyxin B was added for one hour. RNA was harvested and reverse transcribed. Transcript levels were normalized to the housekeeping gene recA. Data represents the mean and standard deviations of three biological replicates.

91

This confounding data led us to use other methodology to examine the relationship between the Psp system and the σE response. We constructed a plasmid- based transcriptional reporter fusion where the promoter of pspABC was fused to lacZ as an alternative way to measure Psp expression. This construct was introduced into O395,

O395ΔrpoE, O395ΔpspA and O395ΔrpoEΔpspA. After growing each strain for 3 hours, a sublethal concentration of polymyxin B was added, to mimic the RNA-seq studies. psp promoter activity was then assessed using a β-galactosidase assay as described in Chapter

3. We expected to see an increase in Psp expression in the rpoE mutant in comparison to the wild type strain. In addition, we expected to observe increased Psp expression in the pspA mutant as it is a negative regulator. We also wanted to determine if the deletion of rpoE in the ΔpspA strain would have a synergistic effect on Psp expression. However, similar to what we observed using qRT-PCR, we did not find Psp expression to be elevated in an rpoE mutant (Fig. 4-3). In addition, Psp expression was not elevated further in an rpoEpspA mutant (Fig. 4-3). Therefore, we cannot confirm that there is a compensatory role for the Psp response in the absence of a functional σE response in V. cholerae at this time.

4.3 Discussion

In this chapter, we present additional studies on the Psp response in V. cholerae.

First, we show that none of the psp null mutants used in Chapter 3 or this chapter exhibit

92

12000

10000 )

8000 pspA::lacZ 6000

4000

Miller Miller Units (P 2000

0 O395 ΔrpoE ΔpspA ΔrpoEΔpspA

Figure 4-3. PspA expression is not increased in the absence of rpoE. Cultures were grown to mid-log phase and treated with PMB for 1 hour (4µg/mL). -galactosidase activity was produced from a plasmid-based transcriptional promoter fusion of PpspA::lacZ . Data represent the mean and standard deviation of three biological samples.

93

growth defects under normal laboratory growth conditions in LB at pH 7 (Fig. 4-1A).

However, under more alkaline and acidic growth conditions with the added stress of secretin expression, a decrease in exponential growth rate is observed. A possible reason for this defect could lie in the disruption of the proton motive force (PMF). In order to maintain a neutral internal pH with an increasing external pH, the membrane potential must increase in order to balance out the pH gradient and maintain the PMF [239]. If a secretin is disrupting the inner membrane (IM) and also potentially destabilizing the membrane potential, it may be more difficult for the bacteria to grow quickly [71]. We performed preliminary experiments to examine the membrane potential in psp mutants using the dye, JC-1. JC-1 serves as an indicator of membrane potential by shifting from red to green as membrane potential increases [68]. Unfortunately, we were unable to detect decreased membrane potential in CCCP treated control samples. In addition, preliminary experiments examining pspA mutant survival with CCCP were inconclusive as a result of inconsistent data. Due to our lack of evidence for membrane potential disruption as an inducer of the response in V. cholerae, it is not clear if that would play any role in the growth phenotypes observed at high/low pH or if there are other possible contributing factors.

In our effort to create complementation plasmids for these studies, we learned more about PspB and PspC. PspC in V. cholerae and Y. enterocolitica requires coexpression of PspB for stability, whereas PspC can be successfully produced alone in

E. coli [70, 238]. This finding was surprising, especially since PspC in Y. enterocolitica is distinct from E. coli and V. cholerae in that it has an extra 20 amino acids at its N- 94

terminus [102]. Therefore, we would have expected the V. cholerae PspC to have more in common with the E. coli PspC. However, there are slight differences in genetic arrangement within each species. In V. cholerae O395, pspB and pspC overlap, with the start codon of pspC extending 35 base pairs into the C-terminal coding region of pspB.

In Y. enterocolitica the pspBC overlap is 20 base pairs in length, whereas the overlap in

E. coli is only a single base pair (KEGG organism codes yew and eco). These lengths may also differ depending on the genome library used and annotations for start codons.

In fact, V. cholerae O395 has two duplicate genome libraries in KEGG, one of which reports PspC as 129 amino acids instead of 138 (KEGG organism codes vco and vcr).

Therefore, open reading frame assumptions must be used with a small amount of caution due to misannotation frequency [240].

Darwin et al speculate that the protein-protein interaction between PspB and PspC may be what stabilizes PspC. Furthermore, PspB and PspC may need to be in equimolar concentrations for stability and a functional Psp response [238]. With prolonged immunoblot exposure, PspC is detectable, though at a much lower level than PspB.

Therefore, we see that PspC can be expressed alone, but we have not explored the relationship between PspC and PspB stability. Future experiments could evaluate the effect of replacing specific domains within the PspC protein with those in E. coli to see if stability is obtained.

The theory that the V. cholerae Psp response may serve in a compensatory manner in the absence of the σE response seems plausible as a similar observation was made in S. Typhimurium [68]. However, no such relationship was seen in Y. enterocolitica [103]. The discrepancies found during the course of our studies regarding 95

the relationship between these two extracytoplasmic stress responses need further investigation. Microarray analyses performed in another laboratory found elevated psp transcripts in an rpoE mutant similar to that of our RNA-sequencing data [135].

However, their samples were prepared from cultures grown in minimal media supplemented with glucose, whereas all of our cultures were grown in LB. Perhaps the inconsistences within our own studies may be due to differences in RNA handling, use of recA as a reference gene, and/or normalization strategies. It is possible that Psp may assume compensatory mechanism under conditions we did not test, such as in minimal media or under other environmental conditions.

96

Chapter 5

Discussion

5.1 Overview

Over a quarter of the world’s population drinks contaminated water and over a third lack access to sanitation facilities. The time spent seeking clean water and locating safe sanitation leads to economic losses and a continued cycle of poverty. On top of this economic crisis, many communities are also burdened with preventable diseases, many of which are diarrheal in nature [241]. One such disease is cholera, caused by the bacterium

V. cholerae. The infection is contracted through the oral-fecal route when contaminated water or food is ingested. Cholera is distinguished by its severe, watery diarrhea that can result in extreme fluid loss. Fortunately, if cholera patients are rehydrated with oral or intravenous fluids, the fatality rate can drop from around 70% to below 1% [12]. With water levels rising due to global climate change, cholera may become an even more widespread [10]. Therefore, it is important to understand colonization, transmission, and survival of V. cholerae within humans and the environment.

V. cholerae, like other Vibrio species, is a marine bacterium capable of inhabiting diverse niches. Within the aquatic environment, Vibrio species must be able to withstand

97

fluctuations in temperature, salinity, and nutrient availability. Moreover, several of these

Vibrio species are capable of colonizing humans, as well as other organisms, and must adapt to host defenses [22]. Bacterial survival hinges on the ability to sense and response appropriately to changes within these diverse environments. Some stresses target the cell envelope, which is comprised of the inner membrane (IM), periplasm, outer membrane

(OM) and proteins localized to those compartments. The set of signal transduction systems that bacteria utilize to monitor and respond to such stresses are collectively known as extracytoplasmic (or envelope) stress response (ESR) systems [55]. This dissertation has focused on ESR systems in Vibrios, and more specifically in V. cholerae.

Chapter 2 reviews the ESR systems, RpoE and Cpx, along with OmpU-mediated stress relief, in Vibrio species. Chapter 3 and 4 contain the first experiments performed to characterize the phage shock protein response (Psp) response in V. cholerae, which will be the primary focus of this discussion.

The Psp response is an ESR system that specifically responds to IM damage and has ties to virulence in some bacteria [242]. This stress response system has been predominantly studied in Y. enterocolitica and E. coli, but is present in many Gram- negative bacteria, including V. cholerae. However, the Psp response had remained completely unstudied in V. cholerae prior to this work, leaving a gap in knowledge about a stress response system that could be involved in disease transmission and virulence

[73]. Therefore, the primary goal of my project was to begin addressing this lack of knowledge by characterizing the V. cholerae Psp response.

98

5.2 Discussion and Future Directions

The studies presented in this thesis are the first to demonstrate that V. cholerae has a functional Psp system. We characterized the basic regulatory mechanisms of this system and identified some of the protein and environmental inducing conditions. In addition, we found that the V. cholerae Psp system may play a role in environmental transmission of disease. These studies provide an initial characterization of this extracytoplasmic stress response system in a new organism, while also establishing a foundation upon which future experiments can be built.

Secretins

The first step we took in characterizing the Psp response was to determine if it was functional in V. cholerae. A common theme that has arisen during the years studying the Psp response in other bacteria is that some secretins serve as remarkably specific inducers of the Psp response [210]. Secretin subunits are composed of two main parts: a highly conserved canonical secretin domain and a variable N-terminal region

[98]. The secretins in V. cholerae were identified by searching the Kyoto Encyclopedia of Genes and Genomes for this shared secretin protein family motif. V. cholerae has multiple secretion systems, but only four feature a secretin: a T2SS secretin, GspD, and three T4PS secretins, TcpC, PilQ and MshL. We observed that, out of the four secretins, only GspD was capable of inducing the Psp response (Figure 2-2). The discovery that

GspD is an inducer was the first proof that V. cholerae Psp response is functional. In addition, we showed that only GspD forms stable multimers within the IM, whereas the other secretins remain monomeric (Figure 2-3). These results were tentatively 99

anticipated based on a recently proposed hypothesis within the field, but needed to be experimentally tested due to some conflicting published data.

Nickerson et al. demonstrated that secretins can be classified into two subfamilies based on their ability to autoassemble as multimers into liposomes. The T2SS and bacteriophage secretins can autoassemble, whereas the T4PS and T3SS secretins cannot

[101]. Secretins share a common ancestry and they hypothesized through phylogenetic analysis that, at some point in time, the T2SS and phage secretins gained the ability to spontaneously multimerize, or the T4PS and T3SS secretins lost the ability. In fact, a single substitution of a conserved proline (with leucine) in a T4PS or T3SS secretin can restore its ability to autoassemble in vitro. Paradoxically, T2SS and phage secretins also contain a conserved proline that, when mutated, hinders their ability to autoassemble

[100, 101]. Nickerson et al. also showed that the Klebsiella oxytoca T2SS secretin, PulD, autoassembles and induces the Psp response. They demonstrated that the mutation,

PulDP443L resulted in the loss of multimer formation. However, PulDP443L still induced the Psp response, which is in contrast to the current Psp model of secretin induction [100,

101].

The current hypothesis proposed by Dr. Andrew Darwin, who studies the Psp system in Y. enterocolitica, is that a secretin only activates the Psp system if it has mislocalized to the IM as a multimer [100]. A number of observations over the years provide evidence for this hypothesis. The historical inducer of the Psp response, pIV, was shown to frequently mislocalized to the IM as a multimers as early as 1994 [99, 212].

In addition, Darwin’s group showed that the T4PS secretin, PilQ, from P. aeruginosa, forms monomers in the IM and does not induce the Psp response in Y. enterocolitica 100

[100]. They then showed that the proline substitution mutant PilQP562L mislocalized as multimers to the IM and induced the Psp response. They also showed the reverse situation with pIV; a proline substitution mutant resulted in the inability to multimerize and interact with Psp proteins. Interestingly, they did not show this phenotype with YsaC or YscC, the T3SS secretins that are the best characterized Psp inducers in Y. enterocolitica. According to this classification system, YsaC and YscC should not be capable of autoassembling in vitro, yet they clearly induce the Psp response. However,

Koo et al. proposed a different secretin classification system that categorizes secretins into 5 different classes based on their localization, stability, and assembly (See Chapter

3)[98].

At this time, the ability of the four V. cholerae secretins to induce (or not induce) the Psp response aligns with Dr. Darwin’s hypothesis and both methods of secretin classification. In the future, it would be intriguing to create the proline-to-leucine mutants in TcpC, PilQ and MshL and determine if the substitution transforms the V. cholerae T4PS secretins into Psp inducers. In addition, the same mutation could be made in GspD with the expectation of losing Psp induction. Overall, it has yet to been seen if secretins from different secretion machinery and species will conform to Darwin’s hypothesis.

The discrepancies between Darwin and Nickerson et al. could be due to several reasons, including the preparation of the SDS-PAGE sample containing the secretin of interest. Some secretins form highly stable multimers that are detergent and heat resistant [243]. In fact, we found that SDS-PAGE samples with T4PS secretins are only present in monomeric form after boiling and exposure to SDS, whereas GspD retains 101

multimer formation after treatment. In contrast, in Darwin’s studies, to detect secretin multimers, samples were not boiled while monomer samples were. Furthermore,

Nickerson et al. tested many conditions, including boiling, not boiling, urea, and phenol to determine their impact on secretin multimer complex stability. They found that only phenol is capable of dissociating secretin multimers into monomers [101, 243].

However, it is unclear how they prepared their secretin samples that showed Psp induction. Overall, consistent and well documented methodologies for preparing SDS-

PAGE secretin samples for oligomeric analysis via immunoblot may help clear up any discrepancies on observed multimerization differences.

In a large portion of our experiments, we used the overexpression of GspD as a method to activate the Psp response. This is a somewhat artificial system, since GspD is likely never that highly expressed in natural physiological conditions. Therefore, it would be significant, not to mention satisfying, to find a relationship between endogenous GspD expression and the Psp response. It has been shown that the T2SS expression is upregulated in response to cyclic-di-GMP [244]. Future experiments could examine if the Psp response is also induced in response to cyclic-di-GMP. However, increased overall T2SS operon expression would cause GspD and its accessory proteins,

GspAB, to be produced in the same stoichiometric concentration [245]. It is likely that overexpression of GspD alone increases mislocalization events due a skewed accessory protein:secretin ratio. Future experiments could also address whether overexpression of

GspAB in combination with GspD leads to a decreased Psp response.

One of the biggest questions in the Psp field is the exact nature of the inducing signal. Is there one unifying signal or is the Psp system capable of identifying a 102

multitude of signals? One hypothesis is that secretins could be the main inducing signal of the Psp response. It is theoretically possible that the other environmental inducing conditions cause secretins to fall short of their trajectory to the OM. These other inducing conditions include, but may not be limited to heat shock, ethanol exposure, hyperosmotic shock, alkaline pH, survival in stationary phase, and SDS treatment. All of these Psp inducing conditions may cause membrane disruption, and secretin mislocalization could be a secondary outcome [246]. Another theory is that stored curvature elastic (SCE) stress could function as the sole inducer of the Psp response, especially because it could be impacted by all of these stressors, including secretin mislocalization [86]. Overall, the current, most favorable theories for the Psp induction signal(s) are (i) SCE stress as the sole inducer, (ii) secretins as the sole inducer, or (iii) there are multiple SCE stress and secretin inducers.

At this point in time it is not feasible to rule out SCE stress as an inducer due to the inability to measure individual lipid movements. However, it might be possible to rule out secretins as the sole inducing agent. An experiment that may address this is to repeat secretin localization experiments with endogenous secretins under the aforementioned inducing conditions. If the secretins do not show increased mislocalization events, then secretin mislocalization cannot be the sole inducing signal.

However, if endogenous secretins do exhibit a higher mislocalization rate to the IM under those conditions, then it is possible that theories (i) and (iii) are valid. Some pitfalls for these studies could be that the conditions used may make membrane separation difficult due to their damage or disruption.

103

Other environmental inducers

We also examined non-secretin based induction of the Psp response. Similar to E. coli and S. lividans, we found that exposure to ethanol, heat, increased salinity, stationary phase growth, and SDS exposure induced expression of PspA. In contrast to E. coli, S. enterica and S. lividans, the proton ionophore CCCP was not a reliable inducer of PspA expression. Our studies using CCCP, whether it was for examining Psp induction, survival, or for use as a membrane potential uncoupling agent, tended to be problematic, resulting in results that were highly variable in nature. Future studies could examine the effects of alternate ionophores, such as carbonyl cyanide-4-phenylhydrazone (FCCP) or

2,4-Dinitrophenol (DNP). We also tested whether polymyxin B or bile could act as Psp inducers, but neither elicited a response (data not shown). In addition, the stressors we examined were by no means exhaustive. Future experiments could investigate the antibiotics novobiocin, nalidixic acid, and mitomycin c, as well as the lipid synthesis inhibitors dizaborine, globomycin, cerulenin for their ability to induce the response. The antibiotics listed, inducers of heat shock proteins, did not induce the E. coli Psp response but the lipid inhibitors did [247]. Additionally, it may be of interest to test other antibiotics that target the bacterial membrane.

Psp functions

As hypothesized, based on protein similarity, PspA, PspB, PspC and PspF all show similar regulatory roles in V. cholerae as in other Gram-negative species; PspA is a negative regulator with PspF, PspB and PspC fulfilling positive regulatory roles. We did not characterize these proteins further, however there is much that could be done in future 104

studies. Characterization of the E. coli Psp system has been underway for almost three decades, whereas characterization of the V. cholerae Psp response is only in its infancy.

While it may seem redundant to investigate the V. cholerae Psp response given the similarity to E. coli response, there are documented differences observed in all of the Psp systems studied to date.

First, we would like to validate that PspF binds to the promoter of pspABC and pspG. In addition, there could be internal PspF promoters within the pspABC operon.

We have also not determined if PspE is regulated by PspF in V. cholerae. To address these questions, we could perform electrophoretic mobility shift assays (EMSAs) to examine these PspF-promoter binding interactions. In addition, based on protein conservation, our results that PspA localization depends on activity of the Psp system, and the dependency of PspC on PspB for stability, we predict potential Psp protein interactions. Under noninducing conditions, PspA would likely form complexes with

PspF in the cytoplasm and PspB and PspC would likely interact within the IM. Under inducing conditions, we predict a partner swap would occur with PspA now forming a complex with PspB and PspC. However, it is quite possible that amino acids involved in the V. cholerae Psp-Psp interactions differ from those in E. coli and Y. enterocolitica, making these studies important.

One area of interest that we did not investigate in depth is the role of PspA as an effector protein. We did not find any survival defects in a pspA null mutant, though further conditions could be tested (data not shown). In addition, we never showed that

PspA is capable of forming higher oligomeric rings, as has been observed in E. coli, B. pseudomallei, and A. thaliana [88, 90, 111]. It is hypothesized that these large 36-mers 105

may stabilize and block proton leakage from the IM [91]. It is also theorized that PspA itself may function as a sensor for the Psp system, since it is the sole Psp protein in some bacteria, like B. pseudomallei [111, 210]. All of these questions merit further investigation in V. cholerae.

We would also like to gain a better understanding of the roles of the less-studied members of the Psp response, PspG and PspE. V. cholerae PspE is a predicted periplasmic rhodanese, an enzyme known for its ability to detoxify cyanide. However, the fact that the E. coli PspE does not have a high affinity for cyanide suggests that the primary role of PspE is likely not cyanide detoxification. Instead, it is proposed that it may be involved in catalyzing the formation of iron-sulfur clusters. Future studies should test whether or not the V. cholerae PspE has thiosulfate sulfurtransferase activity, if it plays a role in sulfur-iron cluster formation, and if it is linked to the Psp response. PspG has a role in motility in E. coli and may form complexes with PspA, -B and -C. It has been shown in Y. enterocolitica that PspB, -C and –A form complexes with mislocalized secretins, but it would be interesting to investigate if PspG is also a part of that complex

[93, 94, 100]. Studies examining the Psp-secretin interactions may reveal more functional information on the core Psp proteins, in addition to the less-studied PspG.

Zebrafish and Environment

One of our more interesting findings in this study was the discovery that a psp null strain exhibits an intestinal colonization defect in zebrafish (Figure 3-7). We observed that the amount of mucin secretion and bacteria excreted was decreased in zebrafish infected with the psp null mutant. While not statistically significant, this data 106

suggests that strains lacking the psp system may cause less severe disease in fish. In this model, the traditional virulence factors, CT and TCP, do not play a role in intestinal colonization. Due to these differences it has been difficult to predict essential colonization factors in this model, and the psp mutant is the only strain identified to have a colonization defect to date [49]. This is the first report of the Psp system potentially playing a role in pathogenesis outside of a mammalian model. The Psp system is required for virulence in the Y. enterocolitica mouse model of infection and may be involved in macrophage infection and survival in a number of other bacterial pathogens

[72, 242]. In contrast, we did not observe a colonization defect with the psp null strain in the infant mouse model. These findings suggest that the Psp system in V. cholerae may play a more important role in environmental transmission and survival.

In order to analyze these phenotypes further, the zebrafish colonization experiments should be repeated. First, this would increase the current sample size and the trends seen in mucin secretion and V. cholerae excretion may show more significant differences. Second, intestinal and expelled V. cholerae could be collected for RNA isolation and analyzed with qRT-PCR or RNA-seq to determine if the Psp response is upregulated during the zebrafish infection. In addition, transmission experiments could be performed. When V. cholerae is excreted from the fish, it exists in a hyperinfectious state [248]. An infected zebrafish could be placed with naïve zebrafish to determine if the psp mutant shows reduced overall transmission to a new host. The finding that the psp mutant is important for transmission in the zebrafish model not only provides clues to its importance in the environment, but provides a new tool for studying the requirements for colonization of this host. 107

In light of its namesake, it would also be interesting to determine if the V. cholerae Psp system is induced with filamentous vibriophage infection. A well-known filamentous phage of V. cholerae is CTXΦ, the phage that encodes for cholera toxin, ctxAB [249]. CTXΦ is very similar to the filamentous phage f1 originally found to induce the Psp response in E. coli via its pIV secretin. Intriguingly, CTXΦ does not encode for its own secretin and instead commandeers the V. cholerae secretin GspD for extrusion out of the cell. If GspD is mutated, less virions are secreted [250]. This leads one to wonder if expression of the Psp system in V. cholerae would enhance virion secretion or if the loss of the Psp system would result in decreased CTXΦ extrusion. In addition, it would be interesting to see if lytic vibriophages can induce the Psp response in V. cholerae. Perhaps the Psp response may serve as a phage defense system during lytic phage infection. While all of this is speculation, it would be exciting to discover if the

Psp response is more aptly named than previously thought.

5.3 Concluding remarks

In conclusion, these studies have begun filling in the gap in knowledge about the V. cholerae Psp response. In this dissertation, we have presented data examining the conservation and regulation of the core set of Psp proteins, the inducing signals, and potential role in environmental transmission. Importantly, the possibility that the V. cholerae Psp response may impact V. cholerae persistence and transmission within the aquatic environment warrants further investigation as we may be seeing a rise in cholera outbreaks due to global climate change.

108

References

1. Barua, D., History of Cholera, in Cholera, D. Barua and W.B. Greenough, Editors. 1992, Springer US: Boston, MA. p. 1-36.

2. Kaper, J.B., J.G. Morris, Jr., and M.M. Levine, Cholera. Clinical microbiology reviews, 1995. 8(1): p. 48-86.

3. Greenough, W.B., 3rd, The human, societal, and scientific legacy of cholera. The Journal of clinical investigation, 2004. 113(3): p. 334-339.

4. Reidl, J. and K.E. Klose, Vibrio cholerae and cholera: out of the water and into the host. FEMS Microbiol Rev, 2002. 26(2): p. 125-39.

5. Lippi, D., E. Gotuzzo, and S. Caini, Cholera. Microbiology Spectrum, 2016. 4(4).

6. The, L., Cholera: ending a 50-year pandemic. The Lancet, 2017. 390(10103): p. 1623.

7. The Lancet Infectious, D., Ending cholera for all. The Lancet Infectious Diseases, 2018. 18(10): p. 1047.

8. Qadri, F., T. Islam, and J.D. Clemens, Cholera in Yemen — An Old Foe Rearing Its Ugly Head. New England Journal of Medicine, 2017. 377(21): p. 2005-2007.

9. Makoni, M., Inside Zimbabwe's efforts to tame cholera. Lancet, 2018. 392(10153): p. e8.

10. Vezzulli, L., et al., Effects of Global Warming on Vibrio Ecology. Microbiology Spectrum, 2015. 3(3).

11. Baker-Austin, C., et al., Emerging Vibrio risk at high latitudes in response to ocean warming. Nature Climate Change, 2012. 3: p. 73.

109

12. Harris, J.B., et al., Cholera. Lancet, 2012. 379(9835): p. 2466-76.

13. Hsiao, A., et al., Lessons learnt from 12 oral cholera vaccine campaigns in resource-poor settings. Bulletin of the World Health Organization, 2017. 95(4): p. 303-312.

14. Silipo, A. and A. Molinaro, The Diversity of the Core Oligosaccharide in Lipopolysaccharides, in Endotoxins: Structure, Function and Recognition, X. Wang and P.J. Quinn, Editors. 2010, Springer Netherlands: Dordrecht. p. 69-99.

15. Wang, L., Q. Wang, and P.R. Reeves, The Variation of O Antigens in Gram- Negative Bacteria, in Endotoxins: Structure, Function and Recognition, X. Wang and P.J. Quinn, Editors. 2010, Springer Netherlands: Dordrecht. p. 123-152.

16. Gardner, A.D. and K.V. Venkatraman, The Antigens of the Cholera Group of Vibrios. The Journal of hygiene, 1935. 35(2): p. 262-282.

17. Aydanian, A., et al., Genetic diversity of O-antigen biosynthesis regions in Vibrio cholerae. Applied and environmental microbiology, 2011. 77(7): p. 2247-2253.

18. Faruque, S.M., M.J. Albert, and J.J. Mekalanos, Epidemiology, genetics, and ecology of toxigenic Vibrio cholerae. Microbiol Mol Biol Rev, 1998. 62(4): p. 1301-14.

19. Dutta, D., et al., Vibrio cholerae non-O1, non-O139 serogroups and cholera-like diarrhea, Kolkata, India. Emerging infectious diseases, 2013. 19(3): p. 464-467.

20. Chowdhury, G., et al., Extraintestinal Infections Caused by Non-toxigenic Vibrio cholerae non-O1/non-O139. Frontiers in microbiology, 2016. 7: p. 144-144.

21. Son, M.S., et al., Characterization of Vibrio cholerae O1 El Tor biotype variant clinical isolates from Bangladesh and Haiti, including a molecular genetic analysis of virulence genes. J Clin Microbiol, 2011. 49(11): p. 3739-49.

22. Takemura, A.F., D.M. Chien, and M.F. Polz, Associations and dynamics of Vibrionaceae in the environment, from the genus to the population level. Frontiers in Microbiology, 2014. 5.

110

23. Almagro-Moreno, S. and R.K. Taylor, Cholera: Environmental Reservoirs and Impact on Disease Transmission. Microbiology spectrum, 2013. 1(2): p. OH- 0003-2012.

24. Senderovich, Y., I. Izhaki, and M. Halpern, Fish as Reservoirs and Vectors of Vibrio cholerae. PLOS ONE, 2010. 5(1): p. e8607.

25. Halpern, M., Y. Senderovich, and I. Izhaki, Waterfowl: the missing link in epidemic and pandemic cholera dissemination? PLoS pathogens, 2008. 4(10): p. e1000173-e1000173.

26. Laviad-Shitrit, S., et al., Wild waterfowl as potential vectors of Vibrio cholerae and Aeromonas species. Tropical Medicine & International Health, 2018. 23(7): p. 758-764.

27. Huq, A., et al., Ecological relationships between Vibrio cholerae and planktonic crustacean copepods. Applied and environmental microbiology, 1983. 45(1): p. 275-283.

28. Wang, J., et al., Vibrio cholerae Colonization of Soft-Shelled Turtles. Applied and environmental microbiology, 2017. 83(14): p. e00713-17.

29. Broza, M. and M. Halpern, Chironomid egg masses and Vibrio cholerae. Nature, 2001. 412(6842): p. 40-40.

30. Islam, M.S., B.S. Drasar, and R.B. Sack, The aquatic flora and fauna as reservoirs of Vibrio cholerae: a review. J Diarrhoeal Dis Res, 1994. 12(2): p. 87- 96.

31. Colwell, R.R., A voyage of discovery: cholera, climate and complexity. Environmental Microbiology, 2002. 4(2): p. 67-69.

32. Huq, A., et al., Influence of water temperature, salinity, and pH on survival and growth of toxigenic Vibrio cholerae serovar 01 associated with live copepods in laboratory microcosms. Applied and environmental microbiology, 1984. 48(2): p. 420-424.

111

33. Van der Henst, C., et al., An intracellular replication niche for Vibrio cholerae in the amoeba Acanthamoeba castellanii. ISME J, 2016. 10(4): p. 897-910.

34. Emch, M., et al., Seasonality of cholera from 1974 to 2005: a review of global patterns. International journal of health geographics, 2008. 7: p. 31-31.

35. Faruque, S.M., et al., Seasonal epidemics of cholera inversely correlate with the prevalence of environmental cholera phages. Proceedings of the National Academy of Sciences of the United States of America, 2005. 102(5): p. 1702- 1707.

36. Gill, D.M., The arrangement of subunits in chlorea toxin. Biochemistry, 1976. 15(6): p. 1242-1248.

37. Clemens, J.D., et al., Cholera. The Lancet, 2017. 390(10101): p. 1539-1549.

38. Levine, M.M., et al., Volunteer studies of deletion mutants of Vibrio cholerae O1 prepared by recombinant techniques. Infection and immunity, 1988. 56(1): p. 161-167.

39. Krebs, S.J. and R.K. Taylor, Protection and attachment of Vibrio cholerae mediated by the toxin-coregulated pilus in the infant mouse model. J Bacteriol, 2011. 193(19): p. 5260-70.

40. Childers, B.M. and K.E. Klose, Regulation of virulence in Vibrio cholerae: the ToxR regulon. Future Microbiol, 2007. 2(3): p. 335-44.

41. Skorupski, K. and R.K. Taylor, Control of the ToxR virulence regulon in Vibrio cholerae by environmental stimuli. Mol Microbiol, 1997. 25(6): p. 1003-9.

42. Richardson, S.H., Animal Models in Cholera Research, in Vibrio cholerae and Cholera, P.A.B. I. Kaye Wachsmuth, Orjan Olsvik, Editor. 1994, American Society for Microbiology: Washington DC. p. 203-226.

43. Blow, N.S., et al., Vibrio cholerae infection of Drosophila melanogaster mimics the human disease cholera. PLoS pathogens, 2005. 1(1): p. e8-e8.

112

44. Klose, K.E., The suckling mouse model of cholera. Trends in Microbiology, 2000. 8(4): p. 189-191.

45. Abel, S. and M.K. Waldor, Infant Rabbit Model for Diarrheal Diseases. Current protocols in microbiology, 2015. 38: p. 6A.6.1-6A.6.15.

46. Navis, M., et al., Mouse fetal intestinal organoids: new model to study epithelial maturation from suckling to weaning. EMBO Rep, 2019. 20(2).

47. Austin, B., Bacterial Pathogens of Marine Fish, in Oceans and Health: Pathogens in the Marine Environment, S. Belkin and R.R. Colwell, Editors. 2005, Springer US: Boston, MA. p. 391-413.

48. Felsenfeld, O., Some observations on the cholera (E1 Tor) epidemic in 1961-62. Bulletin of the World Health Organization, 1963. 28(3): p. 289-296.

49. Runft, D.L., et al., Zebrafish as a Natural Host Model for Vibrio cholerae Colonization and Transmission. Applied and Environmental Microbiology, 2014. 80(5): p. 1710-1717.

50. Engeszer, R.E., et al., Zebrafish in The Wild: A Review of Natural History And New Notes from The Field. Zebrafish, 2007. 4(1): p. 21-40.

51. Strahl, H. and J. Errington, Bacterial Membranes: Structure, Domains, and Function. Annual Review of Microbiology, 2017. 71(1): p. 519-538.

52. Silhavy, T.J., D. Kahne, and S. Walker, The bacterial cell envelope. Cold Spring Harbor perspectives in biology, 2010. 2(5): p. a000414-a000414.

53. Manganelli, R. and M.L. Gennaro, Protecting from Envelope Stress: Variations on the Phage-Shock-Protein Theme. Trends in Microbiology.

54. Boor, K.J., Bacterial stress responses: what doesn't kill them can make then stronger. PLoS biology, 2006. 4(1): p. e23-e23.

55. Rowley, G., et al., Pushing the envelope: extracytoplasmic stress responses in bacterial pathogens. Nat Rev Microbiol, 2006. 4(5): p. 383-94. 113

56. Bury-Moné, S., et al., Global Analysis of Extracytoplasmic Stress Signaling in Escherichia coli. PLOS Genetics, 2009. 5(9): p. e1000651.

57. Kazmierczak, M.J., M. Wiedmann, and K.J. Boor, Alternative sigma factors and their roles in bacterial virulence. Microbiology and molecular biology reviews : MMBR, 2005. 69(4): p. 527-543.

58. Helmann, J.D., The extracytoplasmic function (ECF) sigma factors. Adv Microb Physiol, 2002. 46: p. 47-110.

59. Raivio, T.L., Envelope stress responses and Gram-negative bacterial pathogenesis. Mol Microbiol, 2005. 56(5): p. 1119-28.

60. Bashyam, M.D. and S.E. Hasnain, The extracytoplasmic function sigma factors: role in bacterial pathogenesis. Infection, Genetics and Evolution, 2004. 4(4): p. 301-308.

61. Walsh, N.P., et al., OMP peptide signals initiate the envelope-stress response by activating DegS protease via relief of inhibition mediated by its PDZ domain. Cell, 2003. 113(1): p. 61-71.

62. Dominique, M., et al., Modulation of the Escherichia coliσE (RpoE) heat-shock transcription-factor activity by the RseA, RseB and RseC proteins. Molecular Microbiology, 1997. 24(2): p. 355-371.

63. Raina, S., D. Missiakas, and C. Georgopoulos, The rpoE gene encoding the sigma E (sigma 24) heat shock sigma factor of Escherichia coli. The EMBO Journal, 1995. 14(5): p. 1043-1055.

64. Rhodius, V.A., et al., Conserved and Variable Functions of the σ(E) Stress Response in Related Genomes. PLoS Biology, 2006. 4(1): p. e2.

65. Ehrmann, M. and T. Clausen, Proteolysis as a Regulatory Mechanism. Annual Review of Genetics, 2004. 38(1): p. 709-724.

66. Kovacikova, G. and K. Skorupski, The alternative sigma factor sigma(E) plays an important role in intestinal survival and virulence in Vibrio cholerae. Infect Immun, 2002. 70(10): p. 5355-62. 114

67. DeAngelis, C.M., J. Saul-McBeth, and J.S. Matson, Vibrio responses to extracytoplasmic stress. Environmental Microbiology Reports, 2018. 10(5): p. 511-521.

68. Becker, L.A., et al., Compensatory role of PspA, a member of the phage shock protein operon, in rpoE mutant Salmonella enterica serovar Typhimurium. Mol Microbiol, 2005. 56(4): p. 1004-16.

69. Brissette, J.L., et al., Phage shock protein, a stress protein of Escherichia coli. Proc Natl Acad Sci U S A, 1990. 87(3): p. 862-6.

70. Brissette, J.L., et al., Characterization and sequence of the Escherichia coli stress-induced psp operon. Journal of Molecular Biology, 1991. 220(1): p. 35-48.

71. Weiner, L. and P. Model, Role of an Escherichia coli stress-response operon in stationary-phase survival. Proc Natl Acad Sci U S A, 1994. 91(6): p. 2191-5.

72. Darwin, A.J. and V.L. Miller, Identification of Yersinia enterocolitica genes affecting survival in an animal host using signature-tagged transposon mutagenesis. Molecular Microbiology, 1999. 32(1): p. 51-62.

73. Darwin, A.J., The phage-shock-protein response. Mol Microbiol, 2005. 57(3): p. 621-8.

74. Huvet, M., et al., The evolution of the phage shock protein response system: interplay between protein function, genomic organization, and system function. Mol Biol Evol, 2011. 28(3): p. 1141-55.

75. Joly, N., et al., Managing membrane stress: the phage shock protein (Psp) response, from molecular mechanisms to physiology. FEMS Microbiol Rev, 2010. 34(5): p. 797-827.

76. Dworkin, J., G. Jovanovic, and P. Model, The PspA Protein of Escherichia coli Is a Negative Regulator of ς(54)-Dependent Transcription. Journal of Bacteriology, 2000. 182(2): p. 311-319.

115

77. Bush, M. and R. Dixon, The role of bacterial enhancer binding proteins as specialized activators of sigma54-dependent transcription. Microbiol Mol Biol Rev, 2012. 76(3): p. 497-529.

78. Dworkin, J., G. Jovanovic, and P. Model, Role of upstream activation sequences and integration host factor in transcriptional activation by the constitutively active prokaryotic enhancer-binding protein PspF1. Journal of Molecular Biology, 1997. 273(2): p. 377-388.

79. Jovanovic, G., J. Dworkin, and P. Model, Autogenous control of PspF, a constitutively active enhancer-binding protein of Escherichia coli. Journal of Bacteriology, 1997. 179(16): p. 5232-5237.

80. Joly, N., et al., A Lower-Order Oligomer Form of Phage Shock Protein A (PspA) Stably Associates with the Hexameric AAA(+) Transcription Activator Protein PspF for Negative Regulation. Journal of Molecular Biology, 2009. 394(4): p. 764-775.

81. Elderkin, S., et al., Molecular Determinants for PspA-Mediated Repression of the AAA Transcriptional Activator PspF. Journal of Bacteriology, 2005. 187(9): p. 3238-3248.

82. Yamaguchi, S., et al., Membrane association of PspA depends on activation of the phage-shock-protein response in Yersinia enterocolitica. Mol Microbiol, 2010. 78(2): p. 429-43.

83. Yamaguchi, S., et al., Changes in Psp protein binding partners, localization and behaviour upon activation of the Yersinia enterocolitica phage shock protein response. Mol Microbiol, 2013. 87(3): p. 656-71.

84. Mehta, P., et al., Dynamics and stoichiometry of a regulated enhancer-binding protein in live Escherichia coli cells. Nature Communications, 2013. 4: p. 1997.

85. Maxson, M.E. and A.J. Darwin, PspB and PspC of Yersinia enterocolitica are dual function proteins: regulators and effectors of the phage-shock-protein response. Molecular Microbiology, 2006. 59(5): p. 1610-1623.

116

86. McDonald, C., et al., Membrane Stored Curvature Elastic Stress Modulates Recruitment of Maintenance Proteins PspA and Vipp1. mBio, 2015. 6(5).

87. Jovanovic, G., et al., The N-Terminal Amphipathic Helices Determine Regulatory and Effector Functions of Phage Shock Protein A (PspA) in Escherichia coli. Journal of Molecular Biology, 2014. 426(7): p. 1498-1511.

88. Hankamer, B.D., et al., Organization of the AAA+ Adaptor Protein PspA Is an Oligomeric Ring. Journal of Biological Chemistry, 2004. 279(10): p. 8862-8866.

89. Kobayashi, R., T. Suzuki, and M. Yoshida, Escherichia coli phage-shock protein A (PspA) binds to membrane phospholipids and repairs proton leakage of the damaged membranes. Molecular Microbiology, 2007. 66(1): p. 100-109.

90. Bultema, J.B., et al., Vipp1 and PspA: Related but not twins. Communicative & Integrative Biology, 2010. 3(2): p. 162-165.

91. Standar, K., et al., PspA can form large scaffolds in Escherichia coli. FEBS Lett, 2008. 582(25-26): p. 3585-9.

92. Lloyd, L.J., et al., Identification of a New Member of the Phage Shock Protein Response in Escherichia coli, the Phage Shock Protein G (PspG). Journal of Biological Chemistry, 2004. 279(53): p. 55707-55714.

93. Green, R.C. and A.J. Darwin, PspG, a New Member of the Yersinia enterocolitica Phage Shock Protein Regulon. Journal of Bacteriology, 2004. 186(15): p. 4910- 4920.

94. Engl, C., et al., In vivo localizations of membrane stress controllers PspA and PspG in Escherichia coli. Mol Microbiol, 2009. 73(3): p. 382-96.

95. Adams, H., et al., PspE (phage-shock protein E) of Escherichia coli is a rhodanese. FEBS Letters, 2002. 518(1–3): p. 173-176.

96. Adams, H., et al., Interactions between Phage-Shock Proteins in Escherichia coli. Journal of Bacteriology, 2003. 185(4): p. 1174-1180.

117

97. Korotkov, K.V., T. Gonen, and W.G. Hol, Secretins: dynamic channels for protein transport across membranes. Trends Biochem Sci, 2011. 36(8): p. 433- 43.

98. Koo, J., L.L. Burrows, and P. Lynne Howell, Decoding the roles of pilotins and accessory proteins in secretin escort services. FEMS Microbiology Letters, 2012. 328(1): p. 1-12.

99. Russel, M. and B. Kaźmierczak, Analysis of the structure and subcellular location of filamentous phage pIV. Journal of Bacteriology, 1993. 175(13): p. 3998-4007.

100. Srivastava, D., et al., Psp Stress Response Proteins Form a Complex with Mislocalized Secretins in the Yersinia enterocolitica Cytoplasmic Membrane. mBio, 2017. 8(5).

101. Nickerson, N.N., et al., A Single Amino Acid Substitution Changes the Self- Assembly Status of a Type IV Piliation Secretin. Journal of Bacteriology, 2012. 194(18): p. 4951-4958.

102. Darwin, A.J. and V.L. Miller, The psp locus of Yersinia enterocolitica is required for virulence and for growth in vitro when the Ysc type III secretion system is produced. Molecular Microbiology, 2001. 39(2): p. 429-445.

103. Maxson, M.E. and A.J. Darwin, Identification of inducers of the Yersinia enterocolitica phage shock protein system and comparison to the regulation of the RpoE and Cpx extracytoplasmic stress responses. J Bacteriol, 2004. 186(13): p. 4199-208.

104. Seo, J., et al., Global analysis of tolerance to secretin-induced stress in Yersinia enterocolitica suggests that the phage-shock-protein system may be a remarkably self-contained stress response. Molecular Microbiology, 2007. 65(3): p. 714-727.

105. Woronowicz, K., et al., Effects of the protonophore carbonyl-cyanide m- chlorophenylhydrazone on intracytoplasmic membrane assembly in Rhodobacter sphaeroides. Biochimica et Biophysica Acta (BBA) - Bioenergetics, 2015. 1847(10): p. 1119-1128.

118

106. DeLisa, M.P., et al., Phage Shock Protein PspA of Escherichia coli Relieves Saturation of Protein Export via the Tat Pathway. Journal of Bacteriology, 2004. 186(2): p. 366-373.

107. Engl, C., et al., Dissipation of Proton Motive Force is not Sufficient to Induce the Phage Shock Protein Response in Escherichia coli. Current Microbiology, 2011. 62(5): p. 1374-1385.

108. Horstman, N.K. and A.J. Darwin, Phage shock proteins B and C prevent lethal cytoplasmic membrane permeability in Yersinia enterocolitica. Molecular Microbiology, 2012. 85(3): p. 445-460.

109. Vega, N.M., et al., Signaling-mediated bacterial persister formation. Nature Chemical Biology, 2012. 8: p. 431.

110. Beloin, C., et al., Global impact of mature biofilm lifestyle on Escherichia coli K- 12 gene expression. Molecular Microbiology, 2004. 51(3): p. 659-674.

111. Southern, S.J., et al., Evaluating the role of phage-shock protein A in Burkholderia pseudomallei. Microbiology, 2015. 161(11): p. 2192-2203.

112. Karlinsey, J.E., et al., The Phage Shock Protein PspA Facilitates Divalent Metal Transport and is Required for Virulence of Salmonella enterica sv. Typhimurium. Molecular microbiology, 2010. 78(3): p. 669-685.

113. Datta, P., et al., The Psp system of Mycobacterium tuberculosis integrates envelope stress sensing and envelope preserving functions. Molecular microbiology, 2015. 97(3): p. 408-422.

114. Lucchini, S., et al., Transcriptional adaptation of Shigella flexneri during infection of macrophages and epithelial cells: insights into the strategies of a cytosolic bacterial pathogen. Infect Immun, 2005. 73(1): p. 88-102.

115. Wolf, D., et al., In-Depth Profiling of the LiaR Response of Bacillus subtilis. Journal of Bacteriology, 2010. 192(18): p. 4680-4693.

119

116. Wenzel, M., et al., Proteomic response of Bacillus subtilis to lantibiotics reflects differences in interaction with the cytoplasmic membrane. Antimicrob Agents Chemother, 2012. 56(11): p. 5749-57.

117. Dominguez-Escobar, J., et al., Subcellular localization, interactions and dynamics of the phage-shock protein-like Lia response in Bacillus subtilis. Mol Microbiol, 2014. 92(4): p. 716-32.

118. Vrancken, K., L. Van Mellaert, and J. Anné, Characterization of the Streptomyces lividans PspA Response. Journal of Bacteriology, 2008. 190(10): p. 3475-3481.

119. Vrancken, K., et al., pspA overexpression in Streptomyces lividans improves both Sec- and Tat-dependent protein secretion. Applied Microbiology and Biotechnology, 2007. 73(5): p. 1150-1157.

120. Bidle, K.A., et al., Proteomic analysis of Haloferax volcanii reveals salinity- mediated regulation of the stress response protein PspA. Microbiology (Reading, England), 2008. 154(Pt 5): p. 1436-1443.

121. Kroll, D., et al., VIPP1, a nuclear gene of Arabidopsis thaliana essential for thylakoid membrane formation. Proceedings of the National Academy of Sciences of the United States of America, 2001. 98(7): p. 4238-4242.

122. Westphal, S., et al., Vipp1 deletion mutant of Synechocystis: A connection between bacterial phage shock and thylakoid biogenesis? Proceedings of the National Academy of Sciences of the United States of America, 2001. 98(7): p. 4243-4248.

123. Heidelberg, J.F., et al., DNA sequence of both chromosomes of the cholera pathogen Vibrio cholerae. Nature, 2000. 406(6795): p. 477-83.

124. Aiyar, S.E., T. Gaal, and R.L. Gourse, rRNA promoter activity in the fast-growing bacterium Vibrio natriegens. J Bacteriol, 2002. 184(5): p. 1349-58.

125. Kaper, J.B., J.G. Morris, Jr., and M.M. Levine, Cholera. Clin Microbiol Rev, 1995. 8(1): p. 48-86.

120

126. Baker-Austin, C., et al., Environmental occurrence and clinical impact of Vibrio vulnificus and Vibrio parahaemolyticus: a European perspective. Environ Microbiol Rep, 2010. 2(1): p. 7-18.

127. Jayabalan, N. and D. Pillai, Penaeid prawns and associated luminous bacteria. Acta Microbiol Immunol Hung, 1994. 41(4): p. 415-21.

128. Frans, I., et al., Vibrio anguillarum as a fish pathogen: virulence factors, diagnosis and prevention. J Fish Dis, 2011. 34(9): p. 643-61.

129. Munn, C.B., The Role of Vibrios in Diseases of Corals. Microbiol Spectr, 2015. 3(4).

130. Visick, K.L. and M.J. McFall-Ngai, An Exclusive Contract: Specificity in the Vibrio fischeri-Euprymna scolopes Partnership. Journal of Bacteriology, 2000. 182(7): p. 1779-1787.

131. Erickson, J.W. and C.A. Gross, Identification of the sigma E subunit of Escherichia coli RNA polymerase: a second alternate sigma factor involved in high-temperature gene expression. Genes & Development, 1989. 3(9): p. 1462- 1471.

132. Kanehara, K., K. Ito, and Y. Akiyama, YaeL (EcfE) activates the σ(E) pathway of stress response through a site-2 cleavage of anti-σ(E), RseA. Genes & Development, 2002. 16(16): p. 2147-2155.

133. Cezairliyan, B.O. and R.T. Sauer, Inhibition of regulated proteolysis by RseB. Proc Natl Acad Sci U S A, 2007. 104(10): p. 3771-6.

134. Alejandro, D.L.P., C. Lynn, and G.C. A., The σE-mediated response to extracytoplasmic stress in Escherichia coli is transduced by RseA and RseB, two negative regulators of σE. Molecular Microbiology, 1997. 24(2): p. 373-385.

135. Ding, Y., B.M. Davis, and M.K. Waldor, Hfq is essential for Vibrio cholerae virulence and downregulates σE expression. Molecular Microbiology, 2004. 53(1): p. 345-354.

121

136. Hild, E., et al., Evidence for a Role of rpoE in Stressed and Unstressed Cells of Marine Vibrio angustumStrain S14. Journal of Bacteriology, 2000. 182(24): p. 6964-6974.

137. Davis, B.M. and M.K. Waldor, High-throughput sequencing reveals suppressors of Vibrio cholerae rpoE mutations: one fewer porin is enough. Nucleic Acids Res, 2009. 37(17): p. 5757-67.

138. Brown, R.N. and P.A. Gulig, Roles of RseB, σ(E), and DegP in Virulence and Phase Variation of Colony Morphotype of Vibrio vulnificus. Infection and Immunity, 2009. 77(9): p. 3768-3781.

139. Rattanama, P., et al., Sigma E Regulators Control Hemolytic Activity and Virulence in a Shrimp Pathogenic Vibrio harveyi. PLoS ONE, 2012. 7(2): p. e32523.

140. Haines-Menges, B., W.B. Whitaker, and E.F. Boyd, Alternative sigma factor RpoE is important for Vibrio parahaemolyticus cell envelope stress response and intestinal colonization. Infect Immun, 2014. 82(9): p. 3667-77.

141. Gu, D., et al., A σ(E)-Mediated Temperature Gauge Controls a Switch from LuxR- Mediated Virulence Gene Expression to Thermal Stress Adaptation in Vibrio alginolyticus. PLoS Pathogens, 2016. 12(6): p. e1005645.

142. De Las Peñas, A., L. Connolly, and C.A. Gross, SigmaE is an essential sigma factor in Escherichia coli. Journal of Bacteriology, 1997. 179(21): p. 6862-6864.

143. Matson, J.S. and V.J. DiRita, Degradation of the membrane-localized virulence activator TcpP by the YaeL protease in Vibrio cholerae. Proceedings of the National Academy of Sciences of the United States of America, 2005. 102(45): p. 16403-16408.

144. Stanbery, L. and J.S. Matson, Assay development and high-throughput screening for small molecule inhibitors of a Vibrio cholerae stress response pathway. Drug Design, Development and Therapy, 2017. 11: p. 2777-2785.

122

145. Haines-Menges, B., W.B. Whitaker, and E.F. Boyd, Alternative Sigma Factor RpoE Is Important for Vibrio parahaemolyticus Cell Envelope Stress Response and Intestinal Colonization. Infection and Immunity, 2014. 82(9): p. 3667-3677.

146. Jyoti, M., D.B. M., and W.M. K., Antimicrobial peptides activate the Vibrio choleraeσE regulon through an OmpU‐dependent signalling pathway. Molecular Microbiology, 2007. 63(3): p. 848-858.

147. Sikora, A.E., S.R. Lybarger, and M. Sandkvist, Compromised Outer Membrane Integrity in Vibrio cholerae Type II Secretion Mutants. Journal of Bacteriology, 2007. 189(23): p. 8484-8495.

148. Zielke, R.A., et al., The type II secretion pathway in Vibrio cholerae is characterized by growth phase-dependent expression of exoprotein genes and is positively regulated by sigmaE. Infect Immun, 2014. 82(7): p. 2788-801.

149. Sikora, A.E., et al., Cell Envelope Perturbation Induces Oxidative Stress and Changes in Iron Homeostasis in Vibrio cholerae. Journal of Bacteriology, 2009. 191(17): p. 5398-5408.

150. Wang, H., et al., RNA-Seq Analysis Identifies New Genes Regulated by the Histone-Like Nucleoid Structuring Protein (H-NS) Affecting Vibrio cholerae Virulence, Stress Response and Chemotaxis. PLoS ONE, 2015. 10(2): p. e0118295.

151. Srinivasan, S. and S. Kjelleberg, Cycles of famine and feast: the starvation and outgrowth strategies of a marineVibrio. Journal of Biosciences, 1998. 23(4): p. 501-511.

152. Destoumieux-Garzón, D., et al., Resistance to Antimicrobial Peptides in Vibrios. Antibiotics, 2014. 3(4): p. 540.

153. Wolfe, A.J., et al., Signal integration by the two-component signal transduction response regulator CpxR. J Bacteriol, 2008. 190(7): p. 2314-22.

154. Acosta, N., S. Pukatzki, and T.L. Raivio, The Vibrio cholerae Cpx envelope stress response senses and mediates adaptation to low iron. J Bacteriol, 2015. 197(2): p. 262-76.

123

155. Slamti, L. and M.K. Waldor, Genetic analysis of activation of the Vibrio cholerae Cpx pathway. J Bacteriol, 2009. 191(16): p. 5044-56.

156. Bernal-Cabas, M., J.A. Ayala, and T.L. Raivio, The Cpx envelope stress response modifies peptidoglycan cross-linking via the L,D-transpeptidase LdtD and the novel protein YgaU. J Bacteriol, 2015. 197(3): p. 603-14.

157. Shimohata, N., et al., The Cpx stress response system of Escherichia coli senses plasma membrane proteins and controls HtpX, a membrane protease with a cytosolic active site. Genes Cells, 2002. 7(7): p. 653-62.

158. Vogt S, A.N., Wong J, Wang JS, Raivio T., The CpxAR two-component system regulates a complex envelope stress response in Gram-negative bacteria, in wo- component systems in bacteria, B.D. Gross R, Editor. 2012, Caister Academic Press: Norfolk, United Kingdom.

159. DiGiuseppe, P.A. and T.J. Silhavy, Signal detection and target gene induction by the CpxRA two-component system. J Bacteriol, 2003. 185(8): p. 2432-40.

160. Bina, J., et al., ToxR regulon of Vibrio cholerae and its expression in vibrios shed by cholera patients. Proc Natl Acad Sci U S A, 2003. 100(5): p. 2801-6.

161. Vogt, S.L., et al., The Cpx envelope stress response regulates and is regulated by small noncoding RNAs. J Bacteriol, 2014. 196(24): p. 4229-38.

162. Acosta, N., S. Pukatzki, and T.L. Raivio, The Cpx system regulates virulence gene expression in Vibrio cholerae. Infect Immun, 2015. 83(6): p. 2396-408.

163. Raivio, T.L., Everything old is new again: an update on current research on the Cpx envelope stress response. Biochim Biophys Acta, 2014. 1843(8): p. 1529-41.

164. Raivio, T.L., S.K. Leblanc, and N.L. Price, The Escherichia coli Cpx envelope stress response regulates genes of diverse function that impact antibiotic resistance and membrane integrity. J Bacteriol, 2013. 195(12): p. 2755-67.

165. Amador, P.N.A., Characterization of the Cpx Response in Vibrio cholerae, in Microbiology and Biotechnology. 2015, University of Alberta.

124

166. Taylor, D.L., et al., Reciprocal regulation of resistance-nodulation-division efflux systems and the Cpx two-component system in Vibrio cholerae. Infect Immun, 2014. 82(7): p. 2980-91.

167. Kwon, E., et al., The crystal structure of the periplasmic domain of Vibrio parahaemolyticus CpxA. Protein Sci, 2012. 21(9): p. 1334-43.

168. T, M., The BLAST Sequence Analysis Tool, in The NCBI Handbook [Internet], M.J.a.O. J, Editor. 2002, National Center for Biotechnology Information (US): Bethesda (MD).

169. Payne, S.M., A.R. Mey, and E.E. Wyckoff, Vibrio Iron Transport: Evolutionary Adaptation to Life in Multiple Environments. Microbiol Mol Biol Rev, 2016. 80(1): p. 69-90.

170. Peng, E.D., et al., Nonredundant Roles of Iron Acquisition Systems in Vibrio cholerae. Infect Immun, 2016. 84(2): p. 511-23.

171. de Baar, H.J.a.J., J.T., Distributions, sources and sinks of iron in seawater. IUPAC Series on Analytical and Physical Chemistry of Environmental Systems. 7, ed. D.H. Turner, K. A. 2001, New York: Wiley.

172. Jayalakshmi, S. and V.K. Venugopalan, Role of iron in the virulence of Vibrio vulnificus isolated from Cuddalore coastal waters (India). Indian J Med Res, 1992. 95: p. 294-6.

173. Wright, A.C., L.M. Simpson, and J.D. Oliver, Role of iron in the pathogenesis of Vibrio vulnificus infections. Infect Immun, 1981. 34(2): p. 503-7.

174. Goldberg, M.B., S.A. Boyko, and S.B. Calderwood, Transcriptional regulation by iron of a Vibrio cholerae virulence gene and homology of the gene to the Escherichia coli fur system. J Bacteriol, 1990. 172(12): p. 6863-70.

175. Griffiths, G.L., et al., Vibriobactin, a siderophore from Vibrio cholerae. J Biol Chem, 1984. 259(1): p. 383-5.

125

176. Kunkle, D.E., X.R. Bina, and J.E. Bina, The Vibrio cholerae VexGH RND Efflux System Maintains Cellular Homeostasis by Effluxing Vibriobactin. MBio, 2017. 8(3).

177. Taylor, D.L., et al., Substrate-dependent activation of the Vibrio cholerae vexAB RND efflux system requires vexR. PLoS One, 2015. 10(2): p. e0117890.

178. Audrain, B., et al., Induction of the Cpx envelope stress pathway contributes to Escherichia coli tolerance to antimicrobial peptides. Appl Environ Microbiol, 2013. 79(24): p. 7770-9.

179. Weatherspoon-Griffin, N., et al., The CpxR/CpxA two-component system up- regulates two Tat-dependent peptidoglycan amidases to confer bacterial resistance to antimicrobial peptide. J Biol Chem, 2011. 286(7): p. 5529-39.

180. Yen, M.R., et al., Protein-translocating outer membrane porins of Gram-negative bacteria. Biochim Biophys Acta, 2002. 1562(1-2): p. 6-31.

181. Chakrabarti, S.R., et al., Porins of Vibrio cholerae: purification and characterization of OmpU. J Bacteriol, 1996. 178(2): p. 524-30.

182. Provenzano, D., C.M. Lauriano, and K.E. Klose, Characterization of the role of the ToxR-modulated outer membrane porins OmpU and OmpT in Vibrio cholerae virulence. J Bacteriol, 2001. 183(12): p. 3652-62.

183. Goo, S.Y., et al., Identification of OmpU of Vibrio vulnificus as a fibronectin- binding protein and its role in bacterial pathogenesis. Infect Immun, 2006. 74(10): p. 5586-94.

184. Duperthuy, M., et al., Use of OmpU porins for attachment and invasion of Crassostrea gigas immune cells by the oyster pathogen Vibrio splendidus. Proc Natl Acad Sci U S A, 2011. 108(7): p. 2993-8.

185. Aeckersberg, F., et al., Vibrio fischeri outer membrane protein OmpU plays a role in normal symbiotic colonization. J Bacteriol, 2001. 183(22): p. 6590-7.

126

186. Mathur, J. and M.K. Waldor, The Vibrio cholerae ToxR-regulated porin OmpU confers resistance to antimicrobial peptides. Infect Immun, 2004. 72(6): p. 3577- 83.

187. Provenzano, D. and K.E. Klose, Altered expression of the ToxR-regulated porins OmpU and OmpT diminishes Vibrio cholerae bile resistance, virulence factor expression, and intestinal colonization. Proceedings of the National Academy of Sciences, 2000. 97(18): p. 10220.

188. Provenzano, D., et al., The Virulence Regulatory Protein ToxR Mediates Enhanced Bile Resistance in Vibrio cholerae and Other Pathogenic Vibrio Species. Infection and Immunity, 2000. 68(3): p. 1491-1497.

189. Li, C.C., et al., Molecular cloning and transcriptional regulation of ompT, a ToxR-repressed gene in Vibrio cholerae. Mol Microbiol, 2000. 35(1): p. 189-203.

190. Gueguen, Y., et al., Characterization of a defensin from the oyster Crassostrea gigas. Recombinant production, folding, solution structure, antimicrobial activities, and gene expression. J Biol Chem, 2006. 281(1): p. 313-23.

191. Gueguen, Y., et al., Oyster hemocytes express a proline-rich peptide displaying synergistic antimicrobial activity with a defensin. Mol Immunol, 2009. 46(4): p. 516-22.

192. Duperthuy, M., et al., The major outer membrane protein OmpU of Vibrio splendidus contributes to host antimicrobial peptide resistance and is required for virulence in the oyster Crassostrea gigas. Environ Microbiol, 2010. 12(4): p. 951- 63.

193. Provenzano, D., C.M. Lauriano, and K.E. Klose, Characterization of the Role of the ToxR-Modulated Outer Membrane Porins OmpU and OmpT in Vibrio cholerae Virulence. Journal of Bacteriology, 2001. 183(12): p. 3652-3662.

194. Mathur, J. and M.K. Waldor, The Vibrio cholerae ToxR-Regulated Porin OmpU Confers Resistance to Antimicrobial Peptides. Infection and Immunity, 2004. 72(6): p. 3577-3583.

127

195. Merrell, D.S., et al., The ToxR-Mediated Organic Acid Tolerance Response of Vibrio cholerae Requires OmpU. Journal of Bacteriology, 2001. 183(9): p. 2746- 2754.

196. T., H.D., et al., Bile acids stimulate biofilm formation in Vibrio cholerae. Molecular Microbiology, 2006. 59(1): p. 193-201.

197. Wang, S.-Y., et al., Role for the major outer-membrane protein from Vibrio anguillarum in bile resistance and biofilm formation. Microbiology, 2003. 149(4): p. 1061-1071.

198. Abdallah, F.B., et al., Identification of Outer Membrane Proteins Altered in Response to UVC-Radiation in Vibrio parahaemolyticus and Vibrio alginolyticus. Indian J Microbiol, 2012. 52(4): p. 660-5.

199. Mey, A.R., S.A. Craig, and S.M. Payne, Effects of amino acid supplementation on porin expression and ToxR levels in Vibrio cholerae. Infect Immun, 2012. 80(2): p. 518-28.

200. Mey, A.R., H.A. Butz, and S.M. Payne, Vibrio cholerae CsrA Regulates ToxR Levels in Response to Amino Acids and Is Essential for Virulence. MBio, 2015. 6(4): p. e01064.

201. Almagro-Moreno, S., et al., Proteolysis of virulence regulator ToxR is associated with entry of Vibrio cholerae into a dormant state. PLoS Genet, 2015. 11(4): p. e1005145.

202. Chen, S.-H., The role of YeaZ in the VBNC state of Vibrio parahaemolyticus NCTC 10884, in School of Molecular and Biomedical Science. 2012, University of Adelaide, School of Molecular and Biomedical Science.

203. Wright, A.C., et al., Distribution of Vibrio vulnificus in the Chesapeake Bay. Appl Environ Microbiol, 1996. 62(2): p. 717-24.

204. Whitaker, W.B., et al., Modulation of responses of Vibrio parahaemolyticus O3:K6 to pH and temperature stresses by growth at different salt concentrations. Appl Environ Microbiol, 2010. 76(14): p. 4720-9.

128

205. Yang, L., et al., The low-salt stimulon in Vibrio parahaemolyticus. Int J Food Microbiol, 2010. 137(1): p. 49-54.

206. Miller, V.L., R.K. Taylor, and J.J. Mekalanos, Cholera toxin transcriptional activator toxR is a transmembrane DNA binding protein. Cell, 1987. 48(2): p. 271-9.

207. Silhavy, T.J., D. Kahne, and S. Walker, The Bacterial Cell Envelope. Cold Spring Harbor Perspectives in Biology, 2010. 2(5): p. a000414.

208. Weiner, L., et al., Analysis of the proteins and cis-acting elements regulating the stress-induced phage shock protein operon. Nucleic Acids Research, 1995. 23(11): p. 2030-2036.

209. Jovanovic, G., et al., Induction and Function of the Phage Shock Protein Extracytoplasmic Stress Response in Escherichia coli. Journal of Biological Chemistry, 2006. 281(30): p. 21147-21161.

210. Flores-Kim, J. and A.J. Darwin, The Phage Shock Protein Response. Annual Review of Microbiology, 2016. 70(1): p. null.

211. Disconzi, E., et al., Bacterial Secretins Form Constitutively Open Pores Akin to General Porins. Journal of Bacteriology, 2014. 196(1): p. 121-128.

212. Kazmierczak, B.I., et al., pIV, a Filamentous Phage Protein that Mediates Phage Export Across the Bacterial Cell Envelope, Forms a Multimer. Journal of Molecular Biology, 1994. 238(2): p. 187-198.

213. Manganelli, R. and M.L. Gennaro, Protecting from Envelope Stress: Variations on the Phage-Shock-Protein Theme. Trends in microbiology, 2017. 25(3): p. 205- 216.

214. Ayers, M., P. Lynne Howell, and L. Burrows, Architecture of the Type II secretion and Type 4 pilus machineries. Vol. 5. 2010. 1203-18.

215. Sharma, G., L.L. Burrows, and M. Singer, Diversity and Evolution of Myxobacterial Type IV Pilus Systems. Frontiers in Microbiology, 2018. 9(1630).

129

216. Chang, Y.-W., et al., Architecture of the Vibrio cholerae toxin-coregulated pilus machine revealed by electron cryotomography. Nature microbiology, 2017. 2: p. 16269-16269.

217. Meibom, K.L., et al., Chitin induces natural competence in Vibrio cholerae. Science, 2005. 310(5755): p. 1824-7.

218. Jouravleva, E.A., et al., The Vibrio cholerae Mannose-Sensitive Hemagglutinin Is the Receptor for a Filamentous Bacteriophage from V. cholerae O139. Infection and Immunity, 1998. 66(6): p. 2535-2539.

219. Bose, N. and R.K. Taylor, Identification of a TcpC-TcpQ Outer Membrane Complex Involved in the Biogenesis of the Toxin-Coregulated Pilus of Vibrio cholerae. Journal of Bacteriology, 2005. 187(7): p. 2225-2232.

220. Linderoth, N.A., P. Model, and M. Russel, Essential role of a sodium dodecyl sulfate-resistant protein IV multimer in assembly-export of filamentous phage. Journal of bacteriology, 1996. 178(7): p. 1962-1970.

221. Jovanovic, G. and P. Model, PspF and IHF bind co-operatively in the psp promoter-regulatory region of Escherichia coli. Molecular Microbiology, 1997. 25(3): p. 473-481.

222. Weiner, L., J.L. Brissette, and P. Model, Stress-induced expression of the Escherichia coli phage shock protein operon is dependent on sigma 54 and modulated by positive and negative feedback mechanisms. Genes Dev, 1991. 5(10): p. 1912-23.

223. Jovanovic, G., L. Weiner, and P. Model, Identification, nucleotide sequence, and characterization of PspF, the transcriptional activator of the Escherichia coli stress-induced psp operon. Journal of Bacteriology, 1996. 178(7): p. 1936-45.

224. Matson, J.S., Infant Mouse Model of Vibrio cholerae Infection and Colonization, in Vibrio Cholerae: Methods and Protocols, A.E. Sikora, Editor. 2018, Springer New York: New York, NY. p. 147-152.

130

225. Guilvout, I., et al., Bacterial outer membrane secretin PulD assembles and inserts into the inner membrane in the absence of its pilotin. The EMBO Journal, 2006. 25(22): p. 5241-5249.

226. Skorupski, K. and R.K. Taylor, Positive selection vectors for allelic exchange. Gene, 1996. 169(1): p. 47-52.

227. Guzman, L.M., et al., Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol, 1995. 177(14): p. 4121-30.

228. Stark, M.J.R., Multicopy expression vectors carrying the lac represser gene for regulated high-level expression of genes in Escherichia coli. Gene, 1987. 51(2– 3): p. 255-267.

229. Sambrook, J., E.F. Fritsch, and T. Maniatis, Molecular cloning: a laboratory manual. 1989, Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. xxxviii + 1546 pp.

230. McPherson, M.J. and S.G. Moller, PCR (THE BASICS (Garland Science)). 2006: Taylor & Francis. 292.

231. Teoh, W.P., J.S. Matson, and V.J. DiRita, Regulated intramembrane proteolysis of the virulence activator TcpP in Vibrio cholerae is initiated by the tail-specific protease (Tsp). Molecular Microbiology, 2015. 97(5): p. 822-831.

232. Miller, J.H., Experiments in Molecular Genetics. 1972, Cold Spring Harbor Laboratory, NY.

233. Petiti, M., L. Houot, and D. Duché, Cell Fractionation, in Bacterial Protein Secretion Systems: Methods and Protocols, L. Journet and E. Cascales, Editors. 2017, Springer New York: New York, NY. p. 59-64.

234. Sandrini, S.M., R. Haigh, and P.P.E. Freestone, Fractionation by Ultracentrifugation of Gram Negative Cytoplasmic and Membrane Proteins. Bio- protocol, 2014. 4(21): p. e1287.

131

235. Livak, K.J. and T.D. Schmittgen, Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the 2−ΔΔCT Method. Methods, 2001. 25(4): p. 402-408.

236. Nag, D., et al., Glucose metabolism by Escherichia coli inhibits Vibrio cholerae intestinal colonization of zebrafish. Infect Immun, 2018.

237. Nag, D., et al., Quantifying Vibrio cholerae Colonization and Diarrhea in the Adult Zebrafish Model. J Vis Exp, 2018(137).

238. Gueguen, E., D.C. Savitzky, and A.J. Darwin, Analysis of the Yersinia enterocolitica PspBC proteins defines functional domains, essential amino acids and new roles within the phage-shock-protein response. Molecular microbiology, 2009. 74(3): p. 619-633.

239. Slonczewski, J.L., et al., Cytoplasmic pH measurement and homeostasis in bacteria and archaea. Adv Microb Physiol, 2009. 55: p. 1-79, 317.

240. Poptsova, M.S. and J.P. Gogarten, Using comparative genome analysis to identify problems in annotated microbial genomes. Microbiology, 2010. 156(Pt 7): p. 1909-17.

241. Progress on drinking water, sanitation and hygiene: 2017 update and SDG baselines. 2018, Geneva: World Health Organization (WHO) and the United Nations Children's Fund (UNICEF). p. 6.

242. Darwin, A.J., Stress relief during host infection: The phage shock protein response supports bacterial virulence in various ways. PLoS Pathog, 2013. 9(7): p. e1003388.

243. Hardie, K.R., S. Lory, and A.P. Pugsley, Insertion of an outer membrane protein in Escherichia coli requires a chaperone-like protein. The EMBO Journal, 1996. 15(5): p. 978-988.

244. Sloup, R.E., et al., Cyclic Di-GMP and VpsR Induce the Expression of Type II Secretion in Vibrio cholerae. Journal of Bacteriology, 2017. 199(19).

132

245. Strozen, T.G., et al., Involvement of the GspAB complex in assembly of the type II secretion system secretin of Aeromonas and Vibrio species. J Bacteriol, 2011. 193(9): p. 2322-31.

246. Scarlato, V. and D. Roncarati, Regulation of heat-shock genes in bacteria: from signal sensing to gene expression output. FEMS Microbiology Reviews, 2017. 41(4): p. 549-574.

247. Bergler, H., et al., Inhibition of lipid biosynthesis induces the expression of the pspA gene. 1994.

248. Merrell, D.S., et al., Host-induced epidemic spread of the cholera bacterium. Nature, 2002. 417(6889): p. 642-645.

249. Faruque, S.M. and J.J. Mekalanos, Phage-bacterial interactions in the evolution of toxigenic Vibrio cholerae. Virulence, 2012. 3(7): p. 556-565.

250. Davis, B.M., et al., Convergence of the Secretory Pathways for Cholera Toxin and the Filamentous Phage, CTXϕ. Science, 2000. 288(5464): p. 333-335.

133