Effect of LPS Remodeling and LPS-Binding Proteins on Outer Membrane Vesicle

Effect of LPS Remodeling and LPS-binding Proteins on Outer Membrane Vesicle

Production and Composition

Katherine Elizabeth Bonnington

Department of Biochemistry Duke University

Date:______Approved:

______Meta J. Kuehn, Supervisor

______Michael Boyce

______Richard Brennan

______Kenneth Kreuzer

______Thomas McIntosh

Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Biochemistry in the Graduate School of Duke University

2017

ABSTRACT

Effect of LPS Remodeling and LPS-binding Proteins on Outer Membrane Vesicle

Production and Composition

Katherine Elizabeth Bonnington

Department of Biochemistry Duke University

Date:______Approved:

______Meta J. Kuehn, Supervisor

______Michael Boyce

______Richard G. Brennan

______Kenneth Kreuzer

______Thomas McIntosh

An abstract of a dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Biochemistry in the Graduate School of Duke University

2017

Abstract

Environmental transitions cause bacteria to simultaneously alter lipopolysaccharide

(LPS) structure, release LPS-binding proteins, and increase outer membrane vesicle

(OMV) production. However, any relationship between these events is unknown. The research described here aims to fill this gap in knowledge by examining the effect of environmental membrane remodeling factors and LPS-binding proteins on the regulation of OMV production and composition.

The ability of Gram-negative bacteria to carefully modulate outer membrane

(OM) composition is essential to their survival. However, the asymmetric and heterogeneous structure of the Gram-negative OM poses unique challenges to the cell’s successful adaption to rapid environmental transitions. Although mechanisms to recycle and degrade OM phospholipid material exist, there is no known mechanism to remove unfavorable lipopolysaccharide (LPS) glycoforms except by slow dilution through cell growth. As all Gram-negative bacteria constitutively shed outer membrane vesicles

(OMVs), we propose that cells may utilize OMV formation as a way to selectively remove environmentally-disadvantageous LPS species. We examined the native kinetics of OM composition during physiologically relevant environmental changes in Salmonella enterica¸a well-characterized model system for activation of PhoP/Q and PmrA/B two component systems (TCS). In response to acidic pH, toxic metals, antimicrobial peptides,

and lack of divalent cations, these TCS modify the LPS lipid A and core, lengthen the O- antigen, and up-regulate specific OM proteins. An environmental change to PhoP/Q- and PmrA/B-activating conditions simultaneously induced the addition of modified species of LPS to the OM, down-regulation of previously dominant species of LPS, greater OMV production, and increased OMV diameter. Comparison of the relative abundance of lipid A species present in the OM and the newly-budded OMVs following two sets of rapid environmental shifts revealed the retention of lipid A species with modified phosphate moieties in the OM concomitant with the selective loss of palmitoylated species via vesiculation following exposure to moderately acidic environmental conditions. A polymorphic model for regulation of OM LPS composition may explain the propensity of different LPS structures, with varied geometries and biophysical properties, to be retained in the OM or to be shed via OMVs.

Many pathogenic bacteria secrete toxins which are also lectins, and an increasing number of these proteins have been found to bind to LPS in addition to the host receptor. To investigate the effect of LPS-binding proteins on OM dynamics, we first defined the binding affinity and specificity of heat-labile enterotoxin-LPS binding. We then characterized the ability of the toxin to modulate membrane properties and stimulate OMV production in enterotoxigenic E. coli. Finally, we present a model wherein external toxin binds to LPS and crowds upon the OM to stimulate OMV formation.

Contents

Abstract ...... iv

List of Tables ...... xi

List of Figures ...... xii

Acknowledgements ...... xv

1. Introduction ...... 1

1.1 Model pathogens: Salmonella and Escherichia ...... 2

1.1.1 Salmonella enterica serovar Typhimurium ...... 2

1.1.2 Enterotoxigenic E. coli ...... 5

1.2 Architecture of the Gram-negative cell envelope ...... 7

1.2.1 Inner membrane ...... 9

1.2.2 Periplasmic space ...... 10

1.2.3 Outer membrane ...... 12

1.2.4 Transmembrane protein complexes ...... 17

1.3 Outer membrane vesicles ...... 19

1.3.1 Prevalence of OMVs ...... 19

1.3.2 Methods for quantitative OMV analysis ...... 21

1.3.3 Functional roles of OMVs...... 23

1.3.3.1 Invasion ...... 24

1.3.3.2 Adherence ...... 24

1.3.3.3. Antibiotic resistance ...... 25

1.3.3.4. Damage to host cells ...... 25

1.3.3.5. Modulation of host immune response ...... 26

1.3.3.6 Biofilm formation and nutrient acquisition...... 27

1.3.3.7 Promotion of virulence ...... 28

1.3.3.8 Interspecies cooperation ...... 29

1.3.3.9 Response to cellular stress ...... 29

1.3.4 Protein selection and export via OMVs ...... 30

1.3.4.1 The OMV pathway in comparison to other methods of protein export .... 30

1.3.4.2 Mechanistic models for protein export via OMVs ...... 33

1.3.4.3 OMV cargo enrichment or exclusion ...... 37

1.3.4.4 OMV-associated versus OMV-independent export ...... 41

1.3.4.5 OMV subpopulations ...... 44

1.3.4.5 Timing of OMV production ...... 45

1.3.4.6 Perspectives ...... 47

1.4 Membrane bending and curvature formation ...... 49

1.4.1 Lipid polymorphism model for membrane lipid composition ...... 49

1.4.2 Curvature generation in eukaryotic membranes ...... 51

1.4.3 Curvature formation in Gram-negative outer membranes ...... 55

2. Outer membrane vesicle production facilitates lipopolysaccharide remodeling and maintenance in Salmonella during environmental transitions ...... 57

2.1 Summary ...... 57

2.2 Introduction ...... 57

vii

2.3 Development of a protocol for measuring environmentally-controlled OM remodeling and maintenance ...... 60

2.4 Monitoring the lipid A composition of the OM ...... 71

2.5 Quantification of OM and OMV-localized lipid A species ...... 79

2.6 OMV-localized lipid A species are somewhat depleted for modified species during the shift from 7.6H to 5.8L conditions ...... 82

2.7 OMV-localized lipid A species are substantially depleted for modified species during the shift from 5.8L to 7.6H conditions ...... 84

2.8 OMV production increases during the shift to 5.8L media ...... 92

2.9 OMV production initially remains high during the shift from 5.8L to 7.6H but later reverts to basal levels ...... 95

2.10 Growth in acidic conditions results in increased OMV size ...... 96

2.11 The observed OMV composition does not support a stochastic model of OMV production in all environmental conditions ...... 99

2.12 Discussion ...... 103

3. Lipopolysaccharide binding by heat-labile enterotoxin influences outer membrane dynamics in enterotoxigenic Escherichia coli ...... 111

3.1 Introduction ...... 111

3.2 LTB binds to galactose residues in the R1 type LPS core...... 114

3.3 LTB binds to ΔwaaV LPS with low micromolar affinity ...... 119

3.4 LTB binding decreases the ΔH of soluble ΔwaaV LPS gel-to-liquid crystalline phase transition ...... 124

3.5 LT secretion and LPS binding are necessary for wildtype-levels of OMV production in ETEC ...... 129

3.6 LTB expression increases OMV production in ETEC ...... 132

viii

3.7 Discussion ...... 133

4. Concluding remarks ...... 138

5. Materials and methods ...... 148

5.1 Bacterial strains and media ...... 148

5.2 Environmental shift protocol ...... 150

5.3 Polymyxin B sensitivity assay...... 150

5.4 Sytox green cell integrity assay ...... 151

5.5 OMV purification ...... 152

5.5.1 OMV production quantification ...... 152

5.5.2 Opti prep density gradient fractionation of OMVs ...... 153

5.6 SDS-PAGE and immunoblotting ...... 153

5.6.1 Protein staining ...... 153

5.6.2 Silver staining...... 154

5.6.3 Immunoblotting ...... 154

5.7 Cell fractionation ...... 155

5.7.1 Sucrose density gradient fractionation of IM/OM ...... 155

5.7.2 Peptidoglycan isolation ...... 156

5.7.3 Braun’s lipoprotein quantification: WC versus PG-associated ...... 156

5.8 Lipid A isolation ...... 156

5.9 Glycerophospholipid isolation ...... 156

5.10 Electrospray ionization-mass spectrometry (ESI-MS) of lipid A samples ...... 157

5.11 Thin-layer chromatography (TLC) ...... 157

5.11.1 TLC densitometry ...... 157

5.11.2 2D TLC of glycerophospholipid samples ...... 158

5.11.3 TLC staining methods ...... 158

5.12 Transmission electron microscopy ...... 161

5.13 Quantitative mass spectrometry ...... 161

5.14 Lipopolysaccharide purification ...... 164

5.15 Toxin purification ...... 164

5.16 LPS Sedimentation/LT depletion binding assay ...... 164

5.17 Isothermal titration calorimetry ...... 165

5.18 Differential scanning calorimetry...... 165

5.19 Fourier-transform infrared spectroscopy ...... 166

Appendix 1 ...... 167

Appendix 2 ...... 169

References ...... 176

Biography ...... 199

List of Tables

Table 1. Lipid A structures present in each environmental shift...... 77

Table 2. Comparison of OM and OMV lipid A compositions during the 7.6H-5.8L and 5.8L-5.8L conditions...... 90

Table 3. Plasmids used in Chapter 3 ...... 148

Table 4. Strains used in Chapter 3 ...... 149

List of Figures

Figure 1. Organization of the Gram-negative bacterial cell envelope...... 8

Figure 2. Models of envelope organization prior to outer membrane vesicle (OMV) budding that could lead to cargo enrichment and exclusion...... 35

Figure 3. Shape and spontaneous structures of lamellar and nonlamellar lipids...... 50

Figure 4. Mechanisms of curvature formation in phospholipid bilayers...... 53

Figure 5. Salmonella lipid A structures present under different environmental conditions...... 64

Figure 6. Previous exposure to mildly acidic low-magnesium conditions increases survival during polymyxin B challenge...... 66

Figure 7. Environmental shift conditions slightly affect cell growth, but not membrane integrity...... 68

Figure 8. The environmental shift conditions do not affect cell morphology...... 69

Figure 9. Comparison of OM and OMV lipid A compositions during the 7.6H-7.6H condition...... 72

Figure 10. Comparison of OM and OMV lipid A compositions during the 7.6H-5.8L condition...... 73

Figure 11. Lipid A structures present during the 7.6H-5.8L environmental shift with Group numbers and TLC densitometry labeling...... 76

Figure 12. TLC densitometry band delineation and charring reproducibility...... 81

Figure 13. Comparison of OM and OMV lipid A compositions during the 5.8L-5.8L condition...... 87

Figure 14. Comparison of OM and OMV lipid A compositions during the 5.8L-7.6H condition...... 88

Figure 15. Comparison of OM and OMV lipid A compositions during the 7.6H-7.6H, 7.6H-5.8L, 5.8L-5.8L, and 5.8L-7.6H conditions...... 89

xii

Figure 16. OMV production during the environmental shift conditions...... 93

Figure 17. Evaluation of OMV density, OMV protein content, bacterial Lpp-PG crosslinking levels, and OMV production levels for growth in various environmental conditions...... 94

Figure 18. OMV morphology during the environmental shift conditions...... 98

Figure 19. Comparison of experimentally-determined OMV composition with theoretical stochastic models...... 101

Figure 20. Comparison of experimentally-determined OMV composition with theoretical stochastic models...... 102

Figure 21. LTB binding to mutants of the E. coli R1 core...... 117

Figure 22. Analysis of LTB-ΔwaaV LPS binding by isothermal titration calorimetry...... 121

Figure 23. Infrared spectra in the range of amide I (predominantly C=O stretch) vibrational band for WT LTB and LTB Y18H alone and in the presence of hydrated ΔwaaV LPS...... 122

Figure 24. Heat-labile enterotoxin concentration-dependent changes in LPS gel-to-liquid crystalline phase transition...... 126

Figure 25. Infrared spectra in the range of the antisymmetric stretching vibration of the negatively charged phosphate (PO2-) and the range of the C-O and C-O-C stretching of sugar residues at 23°C for different [LPS]:[LTB] molar ratios...... 127

Figure 26. External localization and LPS-binding ability of LTB are necessary for promotion of OMV production...... 131

Figure 27. Model illustrating the diversity of lipid A structures and their preference for non-lamellar phase to lamellar phase in acidic conditions...... 141

Figure 28. Constitutive activation of phoP increases OMV production in 7.6H and 5.8L conditions...... 143

Figure 29. A comparison of charring with malachite green staining from whole-cell lipid A isolated from 5.8L conditions...... 160

xiii

Figure 30. Clustering of proteins that were differentially expressed between OM and OMV groups...... 172

Figure 31. Abundance of proteins identified in the 5.8L OM samples...... 173

Figure 32. Abundance of proteins identified in the 5.8L OMV samples...... 174

Figure 33. Relative abundance of OM-localized proteins in 5.8L-grown OM and OMV fractions...... 175

xiv

Acknowledgements

This work was supported by NIH grants R01AI079068, R01GM099471, LIPID

MAPS glue grant GM-069338, and Duke University Medical Center. This work was performed in part at the Duke University SMIF, a member of the North Carolina

Research Triangle Nanotechnology Network (RTNN), which is supported by the

National Science Foundation (NSF grant ECCS-1542015) as part of the National

Nanotechnology Coordinated Infrastructure (NNCI). Parts of this work were performed at the Research Triangle Materials Research Science and Engineering Center (MRSEC)

Soft Matter Lab (NSF grant DMR-11-21107).

We thank Ziqiang Guan for discussion and running ESI-MS samples, Michelle

Gignac Plue of SMIF for training on the transmission electron microscope, Daniel

Rodriguez for discussions and ITC and preliminary cosedimentation data, Allison

Stelling for training and discussions on the Al-Hashimi lab FT-IR spectrophotometer,

Parisa Yousefpour for training on the MRSEC differential scanning calorimeter, and

Ellene Mashalidis for training and discussion on the Brennan/Schumacher isothermal calorimeter. We would like to thank the contributions of Matthew W. Foster and M.

Arthur Moseley of the Duke Proteomics Core Facility for their contributions to the scientific oversight, study design, sample processing, and collection and analysis of data for the quantitative mass spectroscopy portion of this work.

1. Introduction

Bacteria are one of the most numerous organisms on the earth. There are 100 million fold more bacteria in the oceans than there are stars in the universe and over three-fold more microorganisms in just a single teaspoon of soil than there are people living the United States of America (2011). The vast majority of these bacteria live and replicate in the environment or reside as commensals in a host organism. Of the inestimable number of microbial species in the world, less than 1% have evolved the ability to cause adverse health effects in the organisms they utilize as hosts (2011).

Millions of years of co-evolution with host organisms have equipped pathogenic bacteria with a complex array of survival and virulence strategies. Effective pathogenic strategies require a combination of robust defensive and offensive tactics; the pathogen must be able to grow and replicate within a host as well as out-compete rivals, including the host itself, to valuable resources. Bacteria possess the ability to dynamically change the physical characteristics they present to the environment in response to host conditions and immune system defenses (Kato et al., 2012). Bacteria can also secrete factors which result in the procurement of more resources from host cells or which solidify their specific replicative niche (Casadevall and Pirofski, 1999).

In recent years the rise of antibiotic resistance amongst many bacterial species has established itself as a global public health concern. Species of non-typhoid

Salmonella, including serovars Typhimurium and Dublin, are of especial concern for the

development of resistance to certain kinds of antibiotics due to their prevalence in poultry and the food industry at large (Crump et al., 2011). A better understanding of the virulence and survival mechanisms utilized by pathogenic bacteria would aid in the development of novel antibiotics and therapies for treatment of multi-drug-resistant pathogens. Accordingly, this chapter presents details on the pathogenesis and cellular physiology of two Gram-negative enteric pathogens, Salmonella enterica serovar typhimurium and enterotoxigenic E. coli (ETEC).

1.1 Model pathogens: Salmonella and Escherichia

Salmonella and Escherichia are both members of the Enterobacteriaceae family.

The genera of Salmonella and Escherichia diverged from a common ancestor approximately 120 to 160 million years ago, based upon estimates from 5S and 16S rRNA sequence analysis (Ochman et al., 1999). Both Salmonella typhimurium and ETEC are leading causes of diarrheoeal disease.

1.1.1 Salmonella enterica serovar Typhimurium

Salmonella enterica serovars are rod-shaped Gram-negative intracellular pathogens which infect warm-blooded vertebrates and reptiles (Rubin and Weinstein,

1977). Salmonella enterica serovar Typhimurium infects humans following the ingestion of contaminated food (including eggs, milk, poultry, and green vegetables) or contact with infected animals (2005). Infection with Salmonella typhimurium typically results in a mild cases of gastroenteritis, though the severity of the disease is intensified in children,

the elderly, and those with weakened immune systems (Coburn et al., 2007). In rare cases, Salmonella typhimurium infiltrates the gut epithelial layer to cause a more serious systemic infection or bacteraemia (LaRock et al., 2015). Estimates of the yearly number of cases of human intestinal disease vary between 200 million and 1.3 billion, with 3 million deaths due to non-typhoidal Salmonella strains recorded worldwide (2005).

In the human gut Salmonella obtains a growth advantage over the resident microbiota by eliciting an inflammatory reaction in the intestinal epithelium and utilizing certain released factors as nutritional sources (Stecher et al., 2007; Winter et al.,

2010). However, these inflammatory actions lead to the migration of macrophages and neutrophils. While Salmonella typhimurium is able to invade host intestinal epithelial cells and macrophages, salmonellae are rapidly cleared by neutrophils (Martinez-Moya et al.,

1998; Niedergang et al., 2000). Therefore, the intracellular environments of both epithelial and macrophage cells offer protection and opportunities for systemic dissemination of the bacteria.

Upon contact with host cells, Salmonella transports effector proteins across the eukaryotic plasma membrane using its needle-like type III secretion system (T3SS). In epithelial cells these effectors induce the actin cytoskeletal ruffling and micropinocytosis processes prerequisite to bacterial uptake. Phagocytic cells (i.e. macrophages) take up

Salmonella through micropinocytosis, but may do so both in an effector-independent or – dependent manner. Once inside the cell, the large phagosomal compartment undergoes

structural remodeling and maturation to become a unique compartment known as the

Salmonella-containing vacuole, or SCV (Drecktrah et al., 2007). The environment of the

SCV is acidic (Drecktrah et al., 2007), nutrient-poor, and limiting in magnesium, phosphate, and ferric iron (Eriksson et al., 2003). Salmonellae modify the morphology and membrane protein and lipid content of the SCV to avoid degradation by the host cell (Garner et al., 2002; Steele-Mortimer et al., 1999). SCV-secreted effectors combine with those secreted upon bacterial uptake to alter many host cell processes, including inflammatory responses, autophagy processes, and vacuolar trafficking (LaRock et al.,

2015; Miao et al., 2003). Failure of the host cell to detect intracellular salmonellae can result in the release of the SCV contents to the intestinal submucosa and further spread of the infection.

Recent work has shed light upon differences in the intracellular life of Salmonella between host cell types (Knodler, 2015). In epithelial cells, a small proportion of

Salmonella escape the SCV to live and replicate in the cytoplasm (Knodler et al., 2014;

Malik-Kale et al., 2012). The epithelial cytoplasm is a nutrient-rich, neutral pH environment which allows for speedier replication than the acidic SCV (Eisenreich et al.,

2010). In macrophages, autophagy machinery clears cytoplasmic salmonellae and damaged SCVs (LaRock et al., 2015). Conversely, in epithelial cells two populations of

Salmonella, vacuolar and cytoplasmic, possess different gene expression profiles, which

may explain how these cytoplasmic bacteria avoid detection and killing by the autophagy machinery (Knodler, 2015; Malik-Kale et al., 2012).

1.1.2 Enterotoxigenic E. coli

The majority of E. coli strains exist as facultative anaerobes in the human colon

(Chang et al., 2004). The acquisition of different repertoires of virulence factors by ancestral commensal strains expanded the range of E. coli diversity to include six different pathotypes of E. coli that cause intestinal infections: enteroaggregative E. coli, diffuse adherent E. coli, enteroinvasive E. coli, enteropathogenic E. coli, enterohemorragic

E. coli, and enterotoxigenic E. coli (Crossman et al., 2010).

Worldwide, enterotoxigenic E. coli (ETEC) is the causative agent of the majority of E. coli-mediated cases of diarrhea. ETEC infections are associated with acute watery diarrhea, ranging from moderately symptomatic to severe cholera-like diarrhea (Sack et al., 1971). The acute diarrhea may last for over a week and feature additional symptoms, including nausea, fever, headache, and vomiting (Yoder et al., 2006). To cause these disease symptoms, strains of ETEC must produce and deliver heat-labile (LT) and/or heat-stabile (ST) enterotoxin to host cells (Fleckenstein et al., 2010).

LT is the major virulence factor of ETEC H10407. LT is part of the AB5 family of toxins, which are characterized by a single catalytic A subunit covalently connected to a homopentamer of B (binding) subunits (Evans et al., 1977; Finkelstein et al., 1976;

Kunkel and Robertson, 1979; Mudrak and Kuehn, 2010a). In LT, the A subunit is made

up of two domains, A1 and A2, linked by a disulfide bridge (Fleckenstein et al., 2010).

A1 is the enzymatically active portion of the toxin, possessing ADP-ribosylase activity

(Moss and Vaughan, 1981; Tsuji et al., 1991). A2 is α-helical in structure and secures the

A subunit to the B pentamer (Sixma et al., 1993). The B pentamer binds to the host cell ganglioside GM1 with 1:1 stoichiometry, though the binding sites recognizing the terminal galactose and surrounding sugars of GM1 are located interstitially between subunits (Merritt et al., 1994b). Each AB5 toxin B-pentamer is characterized by the same structural fold, but this fold adopts different overall topology and possesses different numbers, affinities, and specificities of binding sites for each toxin (Beddoe et al., 2010;

Mudrak and Kuehn, 2010a). The B pentamer of LT binds to two other carbohydrate ligands: type A blood sugar and LPS (Holmner et al., 2007; Mudrak and Kuehn, 2010a;

Mudrak et al., 2009). Similarly, the B pentamer of Stx2 also binds LPS in addition to host cell glycolipid Gb3 (Gamage et al., 2004).

ETEC secretes LT through the general secretion pathway, a type II secretion system (T2SS) apparatus (Pugsley, 1993; Tauschek et al., 2002). First, an N-terminal signal sequence is recognized by the signal recognition particle which directs the LT- translating ribosome to the IM Sec translocon (Hirst and Holmgren, 1987; Hirst et al.,

1984). The signal is cleaved and the nascent LT peptide chain is translated through the

Sec channel and into the periplasm, where LT monomers spontaneously fold and assemble into a holotoxin. The mature holotoxin is recognized by the T2SS before being

launched through the general secretory pathway pore to the extracellular space

(Sandkvist, 2001).

Once secreted, LT binds to GM1 gangliosides present in caveolae on the intestinal epithelial cell surface, triggering endocytosis of the toxin (Lencer et al., 1999; Wolf et al.,

1998). Once inside the host cell, the A1 portion of the toxin interacts with ADP- ribosylating factors to ADP-ribosylate Gsα, inhibiting its GTPase activity (Moss and

Vaughan, 1981). This inhibition leads to the constitutive activation of adenylate cyclase, increasing levels of cAMP. Higher levels of intracellular cAMP activate the chloride channel cystic fibrosis transmembrane regulator (CFTR), resulting in the secretion of electrolytes and water preceding diarrhea (Fleckenstein et al., 2010; Hamilton et al., 1978;

Moss and Vaughan, 1981; Shimizu et al., 1984).

1.2 Architecture of the Gram-negative cell envelope

Gram-negative bacteria are characterized by a diderm cell envelope which distinguishes them from the singly membraned Gram-positive bacteria (Fig.1). The following discussion of the cell envelope will start from the innermost component and work outwards.

Figure 1. Organization of the Gram-negative bacterial cell envelope.

The external milieu is at the top, with the cell interior at the bottom. Cell compartments are labeled according to color on the left, with individual components labeled in the diagram and on the right as necessary. A structural diagram of the lipopolysaccharide molecule is displayed on the far right.

1.2.1 Inner membrane

The cytoplasmic membrane of the Gram-negative cell envelope is a phospholipid bilayer which encloses and regulates the flow of nutrients and metabolic products into and out of the cell cytoplasm. The membrane acts as a hydrophobic barrier to polar molecules, with the action of protein channels and transporters allowing for the selective accumulation and retention of metabolites and proteins in the cytoplasmic space.

As bacteria do not possess the intracellular compartments of eukaryotes, all membrane-associated processes of various eukaryotic organelles are performed at the bacterial inner membrane (IM). Enzymatic activity involved in energy production, lipid biosynthesis, protein secretion, and transport are all localized to the IM (Silhavy et al.,

2010).

The inner membrane (IM) is composed of a roughly equal amount of phospholipid and protein, with a ratio of approximately 30-40 phospholipids per protein present in each leaflet (Kadner, 1996). In E. coli and S. enterica, the major membrane phospholipids, in order of prevalence, are phosphatidyl ethanolamine (PE, 70-80%), phosphatidyl glycerol (PG, 15-25%), and cardiolipin (CL, 5-10%) (Silhavy et al., 2010).

Minor amounts of phosphatidyl serine, phosphatidic acid, lysophospholipids, and diacylglycerol exist due to their role as biosynthetic intermediates or turnover products

(Kadner, 1996). The lipid composition of the IM determines not only the permeability

and barrier properties of the membrane, but also the activity and interactions of IM proteins (Lee, 1991).

1.2.2 Periplasmic space

The periplasmic space extends outwards 13 to 15 nm from the cell’s IM to its outer membrane boundary (Oliver, 1996). Although it accounts for merely 8-16% of the cell’s total volume, many proteins and important cell processes are localized to the periplasm. Functions of periplasmic localized proteins include: transmembrane transport, catabolism, detoxification, and envelope biogenesis (Oliver, 1996).

Importantly, the compartmentalization of the periplasmic space allows the cell to sequester proteins which may be otherwise toxic (Silhavy et al., 2010).

In examination of its physical characteristics, the periplasm is more viscous than the cytoplasm (Silhavy et al., 2010). In complete medium, the measured mean diffusion coefficient for GFP in the E. coli periplasm (2.6-3.4 µm2s-1) is approximately 3-fold lower than the mean diffusion for GFP in the cytoplasm (7.7-9 µm2s-1) (Mullineaux et al., 2006;

Sochacki et al., 2011). This reduced rate of diffusion could be due to a few factors: high viscosity or reduced aqueous space following from the higher protein/unpolymerized peptidoglycan content of the periplasmic space, and/or slowed diffusion through the hydrated, net-like peptidoglycan layer (discussed below) (Oliver, 1996).

Located within the periplasmic space, the peptidoglycan layer gives structural integrity to the cell. The thin peptidoglycan layer is composed of a network of linear

glycan strands, made up of alternating N-acetylmuramic acid and N-acetylglucosamine sugars, crosslinked by short peptides of L-alanine, D-alanine, D-glutamic acid, and meso- diaminopimelic acid (Cloud-Hansen et al., 2006; Vollmer and Bertsche, 2008). The peptidoglycan sacculus is connected to the Gram-negative OM through covalent attachment of lipoproteins (i.e. Braun’s lipoprotein, Lpp), binding by major OM porin

OmpA (Park et al., 2012), and interactions with other proteins and lipoproteins (i.e. peptidoglycan-associated lipoprotein, Pal). Bacterial cells globally modify the structure, composition, and OM-attachment level of peptidoglycan during changes in cellular growth (i.e. during the transition from exponential growth to stationary phase) (Glauner et al., 1988; Pisabarro et al., 1985). These kinds of changes in peptidoglycan fine structure have been associated with the pathogenesis of H. pylori and S. typhimurium (Costa et al.,

1999; Quintela et al., 1997).

Often-overlooked, membrane-derived oligosaccharides (MDOs) are an important component of the periplasm. MDOs, also known as osmoregulated periplasmic glucans, are widely distributed in Gram-negative bacteria (Bohin, 2000; Kennedy, 1982).

Compositionally, MDOs are a closely related but heterogeneous family of highly branched glucose oligosaccharides (Kennedy, 1982). Additions of phosphoglycerol, phosphoethanolamine, and succinyl ester residues give each molecule of MDO three to six negative charges (Bohin, 2000). The presence of MDOs is thought to aid in bacterial resistance to anionic detergents by a mechanism of electrostatic repulsion (Rajagopal et

al., 2003). Furthermore, the synthesis and accumulation of MDOs is inversely proportional to the osmotic strength of the bacterial cell’s environment (Bohin, 2000). In

E. coli cells grown in conditions of low osmolarity, MDOs can encompass 3.5-5% of the cellular dry weight (Bohin, 2000).

1.2.3 Outer membrane

The OMs of Gram-negative bacteria are unique structures due to their compositional asymmetry. Unlike the IM, the OM is not a phospholipid bilayer. Though it contains phospholipids, these are restricted to the inner leaflet, whereas the outer leaflet is populated chiefly by the glycolipid lipopolysaccharide (LPS). LPS, also known as endotoxin, is the causative agent of septic shock.

The cell envelope, including both the inner and outer membranes, of E. coli is composed of approximately 107 glycerophospholipids and 106 LPS molecules (Galloway and Raetz, 1990). The precise structure of LPS greatly diverges between bacterial species and, furthermore, deviates even between individual molecules on a single bacterial cell.

Lipid A, the membrane anchor of lipopolysaccharide, is generally a glucosamine disaccharide modified with varying numbers and lengths of acyl chains; six to seven acyl chains are characteristic of E. coli and S. typhimurium, but only five to six in

Pseudomonas aeruginosa, and four to six in Helicobacter pylori (Ernst et al., 2007; Raetz and

Whitfield, 2002). The lengths of the acyl chains varies between 10, 12, and 16 carbons in

P. aeruginosa, 12, 14, and 16 carbons in E. coli and Salmonella, and 12, 16, or 18 carbons in

H. pylori (Raetz et al., 2007; Stead et al., 2008).

The 1 and 4’ carbon hydroxyl groups of the glucosamine disaccharide may be capped by a variety of modifications (Raetz et al., 2007). In most rich media, the predominate E. coli and Salmonella lipid A structure is hexa-acylated with 1 and 4’ phosphate modifications. In environments featuring neutral-basic pH and non-limiting divalent cations, a small proportion of the lipid A molecules may also present a diphosphate moiety, an addition catalyzed by IM enzyme LpxT, at the 1-position.

Gram-negative bacteria dynamically sense and alter the modification state of its lipid A moieties, as these directly influence the OM permeability and barrier functionalities (Kato et al., 2012). In response to changes in growth conditions, including altered pH, lowered divalent cation concentration, and sensing of cationic antimicrobial peptides, many Gammaproteobacteria, including Escherichia, Salmonella, and

Pseudomonas species, mask the negatively-charged phosphates of lipid A (Miller et al.,

2011; Murata et al., 2007). Activation of the PmrA/B two-component system can result in the addition of phosphoethanolamine (pEtN) and/or L-aminoarabinose (L-Ara4N) to the

1 and/or 4’ lipid A phosphates (Chen and Groisman, 2013). Similar conditions activate the PhoP/Q system, resulting in the formation of hepta-acylated lipid A by though the palmitoylation activity of OM enzyme PagP (Bishop et al., 2000).

As an alternative strategy to masking the phosphate groups, some bacteria, including Francisella tularensis, Porphyromonas gingivalis, Bacteroides fragilis, and H. pylori, remove the lipid A phosphates altogether (Kumada et al., 1995; Moran et al., 1997; Suda et al., 2001; Tran et al., 2006; Vinogradov et al., 2002; Weintraub et al., 1989). In H. pylori, a pEtN group is added at the C-1 hydroxyl, the position of the removed phosphate group (Tran et al., 2006). In F. tularensis, the phosphate may be removed, but it may also be masked by L-Ara4N or modified with galactosamine (Gunn et al., 1998; Phillips et al.,

2004).

Outwards from the lipid A backbone, two Kdo (3-deoxy-D-manno-oct-2-ulosonic acid) sugars link to the oligosaccharide core. Of particular note, Kdo2-lipid A is the minimum structure required for E. coli viability (Raetz and Whitfield, 2002; Schnaitman and Klena, 1993). The Kdo sugar is unique and invariably present in LPS, making it a useful indicator for assaying LPS concentration (Lerouge and Vanderleyden, 2002). The oligosaccharide core and attached O-antigen sugars are not generally required for growth in laboratory conditions but contribute to bacterial survival in the presence of antibiotics, the complement system, and various other environmental stressors (Raetz and Whitfield, 2002).

The general structure of the oligosaccharide core is strain-specific, and may vary between even closely related bacteria. In E. coli, five core types predominate: K-12, R1,

R2, R3, and R4 (Heinrichs et al., 1998b). The core sugars may be modified by phosphate groups, sulfate groups, or ethanolamine (Wilkinson, 1983).

The O-antigen portion of LPS is composed of a variable number of repeating sugars units, usually 3 to 5 sugars in length (Lerouge and Vanderleyden, 2002).

Variations in the O-antigen sugar content contribute to the many antigenic types between species and strains of Gram-negative bacteria. In Salmonella typhimurium, the O- antigen repeat region varies greatly in length, containing any number from zero to greater than 80 subunits (Peterson and McGroarty, 1985).

In the OM, the rate of diffusion of LPS molecules (~ 3 x 10-6 nm2s-1/10-19 cm2s-1) is extremely low when compared to lipid diffusion in a prototypical eukaryotic plasma membrane (10-9 cm2s-1) or the bacterial inner membrane (1.3 x 10-9 cm2s-1) (Fahey and

Webb, 1978; Schindler et al., 1980). However, LPS molecules themselves differ drastically from the typical phospholipid. To begin with, LPS molecules possess fully saturated acyl chains, and as such have a very small area per hydrocarbon chain (Snyder et al., 1999). This feature allows the molecules to pack closely together. Furthermore, LPS molecules are tightly knit together in the OM through bridges of divalent cations between the negatively charged phosphate groups that flank the lipid A head groups

(Nikaido, 2003).

As a permeability barrier, the carbohydrate chains of LPS act as a hydrophilic shield, preventing the passage of hydrophobic molecules across the OM (Hiruma et al.,

1984; Nikaido and Vaara, 1985) . The permeability barrier function of the outer membrane is also related to the gel-to-liquid crystalline phase transition temperature, or melting temperature (Tm), of the membrane (Brandenburg and Seydel, 1984). In general, the melting temperatures of various LPS species are higher than that of eukaryotic membrane lipids. The Tm of LPS structures lowers with decreased core sugar length.

Conversely, the addition of divalent cations increases the Tm of deep-rough (displaying truncated oligosaccharide cores) LPS structures (Snyder et al., 1999).

The S. typhimurium OM GPL composition has been estimated to be 78% phosphatidylethanolamine (PE) and 22% phosphatidylglycerol (PG), acylphosphatidylglycerol (acyl-PG), and cardiolipin (CL) (Dalebroux et al., 2014). The

OM GPL composition changes upon PhoP/Q activation, wherein the membrane increases the concentration of acyl-PG and CL.

The proteins of the outer membrane primarily function in the active and passive transport of nutrients and ions into the cell. Structurally, the proteins of the OM can be divided into three classes: beta-barrel proteins, lipoproteins, and large alpha-helical complexes (i.e. Wza in capsule secretion and the type II secretion system secretin GspD)

(Jeeves and Knowles, 2015). Outer membrane porins (OMPs) all possess beta barrel structures, varying in size from 8 to 26 strands and existing as monomers or trimers

(Nikaido, 2003). Notably, some OM proteins require the presence of LPS to fold properly

(de Cock et al., 1999).

In general, there are very few enzymes localized to the OM. The PhoP/Q- activated enzymes PagP and PagL are notable exceptions to this rule. Both PagP and

PagL act on LPS; PagP catalyzes the addition of a palmitate group to lipid A from phospholipids (Bishop et al., 2000), whereas PagL catalyzes the removal of a 3-O-linked lipid A acyl chain (Kawasaki et al., 2004). E. coli OmpT and S. typhimurium PgtE are

PhoP/Q-activated proteases which primarily act on antimicrobial peptides (Guina et al.,

2000). OM phospholipase A (OMPLA, or PldA) is a constitutively-expressed calcium- activated serine hydrolase, hydrolyzing membrane phospholipids to maintain OM structural activity and stability (Scandella and Kornberg, 1971; Wang et al., 2016).

The OM performs the important task of interfacing with the external environment, maintaining barrier and containment functionalities throughout a range of disparate environmental conditions. Nonetheless, the outer membrane is likely the

Gram-negative organism’s most vulnerable compartment, as it receives the brunt of external assaults yet, as the outermost compartment, is the last to receive aid from cellular services.

1.2.4 Transmembrane protein complexes

The presence of two concentric membranes poses a challenge for any cellular process which must shuttle components across these membranes. To complicate matters, the periplasmic space is devoid of any traditional energy source (i.e. ATP) (Silhavy et al.,

2010). Gram-negative bacteria overcome this obstacle by being well-equipped with an

arsenal of machineries which either permanently or transiently span the two membranes, deriving energy from activity at the cytoplasmic membrane to aid in functions related to membrane biogenesis, secretion, efflux, and motility.

The Lpt pathway transports lipopolysaccharide molecules from the site of biogenesis to their final destination in OM outer leaflet (Okuda et al., 2016). The Lpt complex forms a bridge between the inner and outer membranes featuring a hydrophobic groove to shield LPS acyl chains during transit. As a new molecule is loaded onto the IM side of the bridge, each preceding molecule slides upwards, pushing the final molecule into the OM (Okuda et al., 2016)

The Lol system catalyzes the transports of OM-destined lipoproteins to the OM inner leaflet. The Lol system is composed of five proteins: three IM proteins, one periplasmic binding protein, and one OM lipoprotein (Tokuda and Matsuyama, 2004).

Currently, the mechanistic details of how the hydrophobic lipoproteins cross the hydrophilic periplasm are unknown.

The flagellar apparatus and type III secretion system both span the inner and outer membranes to function in mobility and effector secretion, respectively (Cornelis,

2006). The type III injectisome shares a significant similarity with eight proteins of the flagellar apparatus.

1.3 Outer membrane vesicles

Portions of this section appear in the journal Biochimica et Biophysica Acta

(Bonnington and Kuehn, 2014).

1.3.1 Prevalence of OMVs

All Gram-negative bacteria studied to date release outer membrane vesicles

(OMVs) in every stage of growth and in a range of different environmental conditions

(Beveridge, 1999; Kadurugamuwa and Beveridge, 1997; Li et al., 1998; Mayrand and

Grenier, 1989). Originating from the cell envelope, OMVs are spheroid particles, approximately 10 to 300 nm in diameter, composed of a membrane bilayer enclosing a proteinaceous lumen (Beveridge, 1999). Components of OMVs include lipopolysaccharide (LPS), phospholipids, DNA, RNA, as well as proteins localized to the cytoplasm, inner membrane (IM), periplasm, and outer membrane (OM).

Formation of vesicles is a ubiquitous process, occurring in liquid culture, solid culture, and in biofilms (Kuehn and Kesty, 2005). Notably, OMVs contain newly synthesized proteins and form in the absence of cell death or bacterial lysis (McBroom et al., 2006; Mug-Opstelten and Witholt, 1978; Yaganza et al., 2004; Zhou et al., 1998).

OMVs are observed in the process of budding and pinching off from multiple sites on the bacterial OM by electron microscopy (Chatterjee and Das, 1967; Kadurugamuwa and

Beveridge, 1995; Li et al., 1998; Mayrand and Grenier, 1989; Pettit and Judd, 1992; Zhou et al., 1998).

Much research has delved into the complex cellular regulation of OMV production. The regulation of OMV production seems to be dependent upon a number of factors including environmental conditions, pathogenicity, and the overall cellular metabolic state. Neither the rates of OMV production between various bacterial species nor the production rates for a single species between varying environmental conditions are uniform.

Quantitative analyses have demonstrated that Escherichia coli packages an estimated 0.2–0.5% of OM and periplasmic proteins into OMVs (Hoekstra et al., 1976;

Kesty and Kuehn, 2004; Mug-Opstelten and Witholt, 1978). Approximately 1% of OM material is incorporated into vesicles for typical lab strains of E. coli and Pseudomonas aeruginosa, whereas log-phase cultures of Neisseria meningitidis incorporate 8–12% of total protein and endotoxin into vesicles (Bauman and Kuehn, 2006; Devoe and Gilchrist,

1973; Gankema et al., 1980; Wensink et al., 1978). The utilization of OMVs for protein export is suggested by experimental evidence of enrichment and exclusion of membranous and soluble cargo in vesicles as compared with their respective concentrations in whole bacteria, the periplasm, or the OM. As formation and release of

OMVs results in the export of more cellular material than other methods of secretion, vesicle production is likely highly regulated and optimized for maximum functionality.

1.3.2 Methods for quantitative OMV analysis

In order to study OMVs, proper quantitation of OMV production and content is vital. However, quantitation is nontrivial as OMVs are nonhomogenous mixtures of protein and lipid material varying in both density and size. Extracellular material (e.g. flagellar machinery) may cosediment with OMVs, rendering further purification steps necessary to obtain pure samples for content analysis.

To measure the relative rates of OMV production between two strains, environmental conditions, or antibiotic treatments, the OMV material can be quantified by protein or lipid content (Chutkan et al., 2013). Measurement of relative protein content by Bradford assay is imprecise unless performed on gradient-purified OMVs, free of cosedimented material. However, measurement of relative protein content by

OMP densitometry does not suffer from this constraint. Similarly, measurement of relative lipid content by FM4-64 or a similar lipophyllic dye which fluoresces only when intercalated into a membrane work well on samples prior to gradient purification

(Chutkan et al., 2013).

To count the number and determine the size distribution of all OMVs present in a sample, a number of experimental methodologies have been employed. Traditionally, transmission electron microscopy and/or scanning electron microscopy have been utilized to measure the size distribution of OMVs in a sample (Beveridge, 1999). With these techniques come the critical caveats of the effect of sample dehydration on

measurements and the concern that not all components of a sample will stick with equal readiness to the sample grid. Recently, instruments equipped with the technology of nanoparticle tracking analysis and tunable resistive pulse sensing have been utilized to give more exact counts and size distributions of aqueous samples (Maas et al., 2014;

Roier et al., 2016; Rompikuntal et al., 2015)

Numerous quantitative and qualitative proteomic studies have defined the proteinaceous content of OMVs from numerous Gram-negative bacteria grown under an assortment of environmental conditions. However, features of the study design often complicate proteomic data interpretation. The use of rich media to grow bacterial cultures introduces greater levels of error due to high level of batch-to-batch variation.

The use of a well-defined minimal media results in better reproducibility and lower standard error. The isolation of OMVs from cells grown from the initial lag to stationary phase growth will result in an OMV population with mixed biogenesis timing. If the cell’s metabolic state and physical properties of the OM matter for the mechanistic processes of OMV production, then sampling from a wide array of cellular states, including variable divalent cation concentration, nutrient availability, and pH, will likely mask temporally-restrictive trends in proteomic content. To combat these effects, analysis should be performed on OMVs isolated from logarithmically growing cells and/or cellular growth should ideally be conducted in apparatuses which monitor and maintain pH, dissolved oxygen, and nutrient availability.

In general, qualitative proteomic studies are useful to determine the potential functional qualities OMVs secreted from a particular bacterial species, but often shed little light on OMV biogenesis mechanisms. The data from these studies may also be used to compare the content of OMVs as it varies between strains or species, with consideration for differing media, growth phase, and purification methods between studies.

Comparative proteomic studies, whether qualitative or quantitative, are more useful to examine potential mechanisms of OMV formation. The ideal design of a quantitative proteomic study would compare OMV content against purified OM, purified periplasm, and whole cell protein. Comparison of each subcellular fraction to the whole protein could provide information on the efficacy of the purification methods.

Quantitative comparison of the OMV protein content to that of the periplasmic and OM fractions could yield useful information for the design of follow-up study on OMV biogenesis mechanisms.

1.3.3 Functional roles of OMVs

Bacteria export an assortment of proteins via OMVs. Functionally, many vesicle- associated proteins are virulence factors, playing diverse roles in invasion, adherence, antibiotic resistance, damage to host cells, modulation of the host immune response, biofilm formation, and promotion of virulence. Other vesicle-associated proteins may play roles in interspecies cooperation and intercellular communication. For concision,

only bacterial products which have been functionally analyzed in association with

OMVs will be discussed.

1.3.3.1 Invasion

Proteins which enhance the invasive abilities of bacteria may associate with

OMVs to increase the invasive abilities of cells or may simply promote more efficient

OMV internalization. Vesicles containing Ail, a protein known to confer an invasive phenotype to laboratory E. coli strains, exhibit 10-fold higher association with host cells than vesicles without Ail (Kadurugamuwa and Beveridge, 1998; Kesty and Kuehn,

2004). In Shigella flexneri, host uptake of vesicles is predicted to be catalyzed by OM- localized invasins IpaB, IpaC, and IpaD (Kadurugamuwa and Beveridge, 1998).

1.3.3.2 Adherence

Components present in Actinobacillus vesicles enhance the ability of Actinobacillus cells to adhere to oral KB epithelial cells (Meyer and Fives-Taylor, 1994).

Exactly how OMV-localized material could influence the adherence of parent bacteria is unknown. Potentially, the OMVs could bind epithelial cells and bridge the interaction between bacterial and mammalian cell. Alternatively, internalized OMVs could initiate changes in epithelial cell surface receptor expression to enhance the ability of bacterial cell binding.

However, not all OMV-based adhesins positively impact bacterial adherence. For instance, the OM proteins OspA and OspB bind bacterial cell receptors on human

umbilical vein epithelial, dendritic, and epithelial cells. As these proteins are present on the surface of both OM and vesicles, the vesicles of B. burgforferi compete for host cell binding with bacterial cells (Shoberg and Thomas, 1993).

1.3.3.3. Antibiotic resistance

For most Gram-negative bacteria, β-lactamase activity plays a key role in raising the intrinsic level of resistance to penicillins and cephalosporins (Nikaido, 2003). P. aeruginosa packages β-lactamase into the lumen of vesicles as a means of promoting resistance to β-lactam antibiotics (Ciofu et al., 2000). Interestingly, the resistance gained from this type of OMV export becomes a community good, as extracellular β-lactamase benefits the entire bacterial population and not just the cells which produce the resource.

Additional components present in P. aeruginosa vesicles which are able to kill other species may serve to reduce instances of social ‘cheating,’ or use of the public good without equal contribution (Kadurugamuwa and Beveridge, 1996; Li et al., 1996).

1.3.3.4. Damage to host cells

Enterohemorrhagic E. coli O157:H7 and S. dysenteriae both package active Shiga toxin, Stx1 and Stx2, into OMVs (Dutta et al., 2004; Yokoyama et al., 2000). Shiga toxin is an AB5 family toxin with RNA-N-glycosidase activity which works to inhibit eukaryotic protein synthesis (Beddoe et al., 2010). Shiga toxin appears to localize to both the OMV lumen and surface, as suggested by its partial protease sensitivity and partial protease protection (Dutta et al., 2004; Kolling and Matthews, 1999; Yokoyama et al., 2000). Its

surface localization may be related to the ability of some Shiga toxins to bind to certain structures of LPS (Gamage et al., 2004).

The OMVs of Enterotoxigenic E. coli (ETEC) contain a substantial portion of physiologically active heat-labile toxin (LT) (Horstman et al., 2004). A member of the

AB5 toxin family, LT uses the GM1 ganglioside binding capacity of its B subunit to enter intestinal epithelial cells before the enzymatic A subunit ADP-ribosylates the alpha subunit of the Gs protein, leading to the activation of adenlyate cyclase, increased cAMP levels, and eventual chloride efflux. V. cholerae has also been reported to secrete a portion of the highly related cholera toxin (CT) via OMVs (Lindmark et al., 2009). Like LT, CT also seems to mediate host cell internalization of the pathogen-derived vesicles (Kesty and Kuehn, 2004; Lindmark et al., 2009).

Campylobacter jejuni exports biologically active cytolethal distending toxin (CDT), a genotoxin responsible for eukaryotic cell cycle arrest at the G2/M stage, in association with vesicles (Lindmark et al., 2009). Unlike most other Gram-negative pathogens, the C. jejuni genome gives no evidence of identifiable secretion systems for delivery of virulence factors into host cells (Elmi et al., 2012). Accordingly, export of CDT via OMVs may be the major mechanism for coordinated release of the tripartite toxin.

1.3.3.5. Modulation of host immune response

Vesicle-associated proteins possess the capability to enact either pro- or anti- inflammatory effects on host immune systems. Porphyromonas gingivalis preferentially

packages its major virulence factors, a group of proteases named gingipains, into OMVs

(Haurat et al., 2011). Gingipains degrade cytokines in order to down-regulate host inflammation and immune response. In contrast, C. jejuni OMVs work to provoke immune response. Export of sixteen immunogenic N-linked glycoproteins in OMVs delivers proteins normally localized to the periplasm to host cells (Elmi et al., 2012).

Furthermore, OMVs isolated from H. pylori, P. aeruginosa, and Neisseria gonorrhea were shown to deliver peptidoglycan to host cells, activating cytosolic NOD1 and promoting inflammation (Kaparakis et al., 2010).

Soluble ETEC LT seems to elicit ultimately similar cytokine responses as OMV- associated LT but through different activation pathways (Chutkan and Kuehn, 2011).

However, comparison between the two export forms shows kinetic differences in CREB

(cyclic-AMP response element binding protein) activation, possibly due to the manner in which the different presentations of the toxin are trafficked in the cell. In contrast, macrophage cells sense both lipid and protein components of P. aeruginosa OMVs to produce unique, strain-specific cytokine responses (Ellis et al., 2010).

1.3.3.6 Biofilm formation and nutrient acquisition

OMVs have been observed as components of the biofilm matrix (Schooling and

Beveridge, 2006). However, a general role for OMV production in biofilm formation is unknown. The supplementation of purified OMVs to H. pylori grown in conditions

which did not normally produce robust biofilm formation increased biofilm thickness

(Yonezawa et al., 2009).

P. gingivalis OMVs contain the surface-exposed heme-utilization protein (HmuY), a putative heme-binding lipoprotein associated with the OM (Smalley et al., 2011).

HmuY aids in biofilm formation and cell survival during periods of starvation characteristic of plaque conditions (Olczak et al., 2010).

Nutrient-binding proteins, siderophores, proteases, and other degradative enzymes are often packaged into OMVs. As nutrient availability affects OMV production, it is tempting to speculate that OMVs may be involved in the long-distance

(centimeter scale) acquisition or degradation of nutrient sources (Kulp and Kuehn, 2010).

1.3.3.7 Promotion of virulence

OMV production levels increase during conditions that stress the bacterial envelope, such as host conditions during infection (Ellis and Kuehn, 2010). In Salmonella typhimurium, the two-component system PhoP/Q directly and indirectly regulates changes in LPS structure and expression of virulence-related proteins necessary for the bacteria to transition from an external environment to the host environment. A set of

PhoP/Q-activated genes, including pagC, pagK1/K2, and pagJ, encode proteins which are associated with OMVs (Kitagawa et al., 2010; Yoon et al., 2011). Although the function for PagC is currently unknown, a PagC-deficient strain displays decreased resistance to complement-mediated killing (Nishio et al., 2005). PagK and PagJ play unknown roles in

virulence but as both proteins are only 39 amino acids long, it is unlikely that they possess enzymatic activity (Yoon et al., 2011).

1.3.3.8 Interspecies cooperation

Moraxella catarrhalis OMVs export proteins that bind and inactivate the complement protein C3. This secretion, acting as a sort of ‘community good’, benefits M. catarrhalis and a co-inhabitant, Haemophilus influenza, by promoting survival (Tan et al.,

2007).

Moreover, the presence of OMVs in the extracellular milieu offers binding sites for OM-specific stressors, such as cationic antimicrobial peptides and bacteriophages. In this manner, the OMVs act as cellular decoys for antimicrobial peptides and phages, increasing the survival rate of cells in the presence of these external threats (Manning and Kuehn, 2011, 2013).

1.3.3.9 Response to cellular stress

The production of OMVs can be useful in alleviating stress on the cell envelope

(Kulp and Kuehn, 2010). Strains of E. coli which produce higher levels of OMVs than WT were able to better survive after ethanol treatment, antimicrobial peptide assault, overexpression of a toxic periplasmic protein, and overexpression of protein mimicking a misfolded OMP (McBroom and Kuehn, 2007).

1.3.4 Protein selection and export via OMVs

1.3.4.1 The OMV pathway in comparison to other methods of protein export

Protein export via OMVs offers several distinct advantages over other secretory mechanisms. Firstly, the protein cargo exits the cell in a medium inaccessible to extracellular degradative enzymes. Proteins located in the vesicle lumen are insensitive to protease treatment (Horstman and Kuehn, 2000a; Kolling and Matthews, 1999).

Accordingly, vesicles are thought to be capable of long-distance transport. OMV antigens disperse to sites far from initial colonization, such as the urine, blood, and a number of organs, as demonstrated in Borrelia burgdorferi-infected mice, dogs, and humans (Dorward et al., 1991).

Protein cargo bereft of a mechanism to self-direct to target cells may utilize adhesins or other surface-associated virulence factors present on the OMV surface for transport to an appropriate destination. For example, Actinobacillus actinomycetemcomitans leukotoxin or other unknown factors can target OMVs to host cells. Leukotoxin binds nucleic acid on bacterial membrane surfaces and interacts with the host cell surface receptor β-2 integrin (Kato et al., 2002; Ohta et al., 1993; Ohta et al.,

1991; Prasadarao et al., 1996). However, the association of A. actinomycetemcomitans OMVs with the host cytoplasmic membrane may occur independently of leukotoxin (Ohta et al., 1993). A. actinomycetemcomitans vesicles are enriched not only in leukotoxin, but also in an OmpA homologue, an OM lipoprotein component of an ABC transporter, as well as a minor lipid species not detectable in the 30

OM, and these factors may be critical in directing OMVs to designated host cells (Ohta et al., 1991). In E. coli K1, OmpA interacts with surface receptor Ecgp on brain microvascular endothelial cells to mediate invasion (Bomberger et al., 2009). It is possible that OmpA in A. actinomycetemcomitans OMVs plays a similar role to mediate

OMV invasion.

In regard to cargo, OMV-mediated secretion uniquely enables export of membrane-embedded proteins, membrane-associated proteins, and other proteins which lack canonical secretion signal sequences. Acinetobacter baumannii utilizes OMVs for secretion of AbOmpA, a porin which allows for the passage of small solutes across the OM (Kato et al., 2002). This protein is abundant in culture supernatants and functions as a virulence factor by directly contributing to host cell death. Proteomic analysis of A. baumannii OMVs reveals the presence of several other virulence factors besides AbOmpA, including active AmpC β-lactamase, a putative hemolysin, and a

Resistance-Nodulation-Cell Division (RND) superfamily transporter (which catalyzes substrate efflux via an H+ antiport mechanism) (Jin et al., 2011). In the absence of OMV formation, membrane proteins like OmpA must rely on bacterial-host cell contact for delivery. Vesicles, however, may adeptly accomplish interactions with target host cells, reaching even those cells residing in deep tissue often inaccessible to colonizing bacteria

(Kuehn and Kesty, 2005).

Additional advantages to packaging cargo into OMVs are that cargo molecules may reach their destination in a concentrated manner and/or simultaneously with other bacterial factors. P. aeruginosa vesicles concurrently deliver multiple virulence factors, including β-lactamase, Cif, hemolytic phospholipase, and alkaline phosphatase, to host cells in their fully-folded and enzymatically active forms (Bomberger et al., 2009). The delivery of multiple virulence factors in a single concentrated bolus may allow pathogens to impact host cells in a more complex, nuanced manner.

Furthermore, the specific environment offered by a membranous compartment may further allow for the establishment of a more potent or biologically active enzyme. A. actinomycetemcomitans utilizes OMVs for export of a Type-I secreted RTX family-member, leukotoxin (Ohta et al., 1993; Ohta et al., 1991). Once inside human polymorphonuclear leukocytes and monocytes, toxin activity induces cell lysis.

Leukotoxin is not only an enriched component of OMVs, it is also more potent when associated with vesicles than with the OM (Ohta et al., 1991).

Similarly, OMVs could facilitate the formation of higher-order complexes from periplasmic or OM precursors. The pore-forming cytotoxin ClyA exemplifies this advantage as its oligomeric OMV-associated form exhibits higher cytotoxic activity than its monomeric periplasmic form (Wai et al., 2003).

1.3.4.2 Mechanistic models for protein export via OMVs

Electron microscopy data suggests a simplistic vesicle formation model: the OM blebs outwards and pinches off, encapsulating lumenal content derived from the periplasm. Yet the composition of the OMV membrane often differs from the cell OM; likewise, the OMV lumen differs from the cell periplasm, suggesting a specific and/or selective mechanism for OMV cargo inclusion.

Preferential packaging of specific OM lipid and protein components into vesicles may originate during the cell envelope biogenesis processes. Theoretically, compositionally distinct microdomains could result from the insertion of new membrane material by LPS and OM protein biogenesis mechanisms (Ursell et al., 2012).

Addition of new membrane material is believed to occur in distinct patches which subsequently remain relatively stable due to the unique nature of the outer leaflet of the

OM. Specifically, the outer bilayer LPS molecules and many OMPs are not observed to substantially diffuse (except by cell growth) (Ursell et al., 2012). Thus, OM microdomains formed during envelope biogenesis could possess compositional differences rendering certain areas more or less prone to form OMVs.

Sequestered into domains, particular LPS and OMP compositions (both dependent on environmental conditions) could function as ‘signposts’ for sites of vesicle initiation (green symbols, Fig. 2A). The ‘signpost’ properties may pertain to physical membrane attributes such as distinct curvature, fluidity, charge, and/or affinity.

Appropriate inner leaflet composition could organize to complement these features, as phospholipids and lipoproteins within the inner leaflet likely experience greater freedom to diffuse. Potentially, interleaflet coupling with LPS acyl chains and the locations of pre-existing covalent linkages to peptidoglycan could influence this process.

Ultimately, the establishment of these OMV formation-competent domains may trigger lumenal cargo enrichment, detachment from the underlying peptidoglycan, and recruitment of any vesicle-formation machinery.

In the following sections, multiple models have been proposed which attempt to explain how the enrichment and exclusion of proteins and lipids occurs and ultimately results in OMV formation (Fig. 2). More mechanistic details of how envelope architecture impacts OMV formation have been comprehensively discussed in other reviews (Kulp and Kuehn, 2010; Schwechheimer et al., 2013).

Figure 2. Models of envelope organization prior to outer membrane vesicle (OMV) budding that could lead to cargo enrichment and exclusion.

In (A), Outer membrane proteins (OMPs) and lipopolysaccharide (LPS) are found in pre- existing microdomains in the outer membrane, likely resulting directly from the membrane biogenesis processes. The content clustered in one area may be inherently more prone to budding. This organization can lead to the enrichment of particular OMPs into OMVs. Subsequent partitioning of envelope components can lead to enrichment of other envelope factors (B-F). Envelope features for parts B-F are as indicated in part A: outer membrane (OM); periplasm (PP); inner membrane (IM); and individual envelope components coded by shape (indicated in the side legends) and color (OMV excluded cargo: red symbols; enriched cargo: green symbols; neither enriched/excluded cargo: blue symbols). (B) The localization of periplasmic proteins (PP proteins) to the external face of the inner membrane (IM) may lead to exclusion from OMVs while localization to the OM may enhance incorporation into OMVs. (C) Recruitment factors, potentially OM lipoproteins (OMLPs) or OMPs (green symbols), may bind to or interact with OMV cargo to enhance its localization to future budding 35

sites. (D) Exclusion factors, such as OMPs or OMLPs excluded from OMVs (red symbols) may interact with or bind to PP proteins in order to hinder their incorporation into OMVs. (E) Protein secretion apparatuses or other transmembrane complexes may exist near potential sites of budding. If their secretion substrate binds an OM component or OM-associated component, then these secreted proteins may cluster on the external surface. These areas of high local protein concentration could cause crowding and thereby provide the energy for establishment of positive curvature. (F) A combination of all of the previously-described mechanisms most likely is at work to promote budding and OMV formation. In each of the models, lipoprotein (LPP) – peptidoglycan (PG) crosslinks are notably absent from potential OMV bud sites, either by an exclusion mechanism or by conformational changes to the LPP.

1.3.4.3 OMV cargo enrichment or exclusion

Notably, the mechanism by which certain protein cargo is enriched or excluded from vesicles remains elusive. Enrichment is achieved when the concentration of a certain protein cargo, normalized to major OMP concentration, lipid concentration, or total protein, is higher relative to that protein’s concentration in the whole cell or a particular cellular compartment. Negative enrichment, where relative concentration of a protein is higher in the cell than in OMVs will be referred to as exclusion.

Evidence suggests that the cellular localization of a protein greatly affects its potential for inclusion into OMVs. For instance, in Helicobacter pylori the majority of OM proteins (77%) can be found in OMVs (Olofsson et al., 2010). Even so, some highly abundant OM proteins are actively excluded from OMV export. For instance, the Serratia marcescens OMPs TolC, LptD, maltoporin, and YaeT are present in the OM yet are absent or undetectable in OMVs (McMahon et al., 2012). YaeT, part of the Omp85 family, is the OM portion of the beta-barrel assembly machinery (Bam), protein complexes consisting of a single OMP and four OM lipoproteins (OMLPs) which catalyze the assembly of OMPs by an unknown mechanism (Selkrig et al., 2014; Silhavy et al., 2010).

TolC forms a large, extended pore which connects to the pump AcrA/B to export a diversity of molecules and proteins from the cell cytoplasm (Andersen, 2003). LptD shuttles LPS into the OM outer leaflet from the bridge-like Lpt machinery (Okuda et al.,

2016). As TolC, LptD, and YaeT are all components of transmembrane complexes, these

structural connections may contribute to their absence in OMVs. Unsurprisingly, this type of exclusion observed by OMV proteomic studies is a theme throughout a wide range of Gram-negative species. Excluded OMPs, lacking the ability to diffuse, could act to mark a region of OM for retention (red OMP symbols, Fig. 2A), and recruit other excluded membrane and periplasmic factors (red periplasmic proteins, OMLPs, and

LPS, Fig. 2B-F).

In an intriguing juxtaposition, the S. marcescens OMP MipA is found exclusively in OMVs and is undetectable in OM (McMahon et al., 2012). In the case of proteins like

MipA, absence from the OM suggests a mechanism whereby specific newly synthesized proteins are targeted for incorporation into the OM before immediate export via OMVs

(green OMP symbol, Fig. 2A). In this scenario, the localization of active Bam complexes would be proximal to locations of future budding OMVs. Bulk flow of envelope content into these sites by random diffusion events would lead to neutral (neither enriched nor excluded) incorporation into OMVs.

A protein localized to the periplasmic space but tightly associated to the IM (red periplasmic proteins, Fig. 2B) will likely experience lesser incorporation into OMVs than an OM-associated protein (green periplasmic proteins, Fig. 2B). For instance, removal of

11 C-terminal residues from ClyA abolishes its OMV-association as ClyA becomes associated with the IM (Wai et al., 2003). In contrast, periplasmic proteins associated with the inner leaflet of the OM may display higher rates of incorporation into OMV.

Localization of this sort plays a key role in the export of CDT. In A. actinomycetemcomitans and C. jejuni, the three subunits of CDT (CdtA, B, and C) are translocated to the periplasm. The “lipobox” contained within the N-terminal signal sequence of CdtA targets this subunit for lipid modification. This modification targets

CdtA to the OM as a lipoprotein (Ueno et al., 2006). With CdtA acting as a spatial anchor, the full CDT complex forms in association with the OM prior to additional processing and eventual export via OMVs. It is possible that lipid-modifications and other interactions with OMLPs play a significant, but as of yet unidentified, role in the mechanism of cargo selectivity and vesicle formation. Study of OMLPs may reveal novel ties to OMV formation or production. Approximately 100 different OMLPs exist in the E. coli OM, yet very few have known functions (Nikaido, 2003).

OMV secretion or cellular retention decisions for envelope proteins may be governed by tags (Fig. 2C, D). In S. typhimurium, the N-terminal signal sequence from

PagK is necessary for its periplasmic localization but is not sufficient for incorporation into OMVs (Yoon et al., 2011). However, the cytolysin ClyA of E. coli accumulates in the periplasmic space prior to surface exposure and OMV-export without the cleavage of any N-terminal signal sequence (Wai et al., 2003). No common primary structure has been identified as a OMV sorting or exclusion signal, however it is possible that a more complex tag, utilizing properties from a protein’s three-dimensional structure, could direct specific cargo into vesicles.

Cargo proteins could also directly promote their own export through localized, physical enhancement of OMV production (Fig. 2E). In Shigella dysenteriae, treatment with mitomycin C induces higher Shiga toxin expression. The increased production of

Shiga toxin, a known OMV cargo protein, leads to increased OMV production, toxicity, and size (Yokoyama et al., 2000). Directed localization of secretion apparatuses and/or

Bam complexes could promote high local concentrations of cargo protein within a region of periplasmic space, within the OM, or attached to the cell surface. A high local periplasmic or OM concentration of a cargo protein could accelerate OM budding. In the case of surface-associated proteins destined for OMV export, the high concentration of proteins bound to the membrane may impart a protein-protein crowding effect that accelerates formation of OM curvature appropriate for OMV formation (green external symbols, Fig. 2E).

Alternatively, sites of budding vesicles may contain protein and/or lipid recruitment factors which lead to the accumulation of appropriate OMV cargo. In P. gingivalis, mutations which alter LPS structure disrupt normal protein sorting into

OMVs (Haurat et al., 2011). In P. aeruginosa, two classes of LPS, one negatively-charged and one neutral, exist in the OM, yet primarily negatively-charged LPS is found in vesicles (Kadurugamuwa and Beveridge, 1995). This negatively-charged LPS is required for the formation of peroxide stress-induced OMVs (Macdonald and Kuehn, 2013).

In most cases, the mechanistic cause for differential degrees of enrichment or exclusion for various protein cargo in OMVs is unclear, and a combination of factors is likely important in promoting envelope distension and OMV formation (Fig. 2F).

1.3.4.4 OMV-associated versus OMV-independent export

For some OMV-associated proteins, extracellular release is only possible as OMV cargo. Most of these proteins are either not substrates for other secretion systems or are embedded in the OM. Alternatively, some proteins which utilize other secretion systems bind LPS with such a high affinity that the vast majority of extracellular protein observed is in an OMV-associated state. Such is the case for the heat labile enterotoxin

(LT), with the vast majority (95%) of LT activity in ETEC culture supernatant detected in the OMV fraction (Horstman et al., 2004). The ability of LT to bind LPS allows it to remain associated with the bacterial OM. Accordingly, extracellular LT is found both on the surface and interior of OMVs. The toxin’s ability to bind LPS as well as intestinal epithelial cell receptor GM1 ganglioside allows it to target LT-laden OMVs to host cells and initiate internalization (Kesty and Kuehn, 2004).

The strength of a certain protein’s association with OMVs may be related to the affinity of the protein for LPS. Many gingipains and RTX toxins (including leukotoxins and hemolysins) also bind or form high-molecular weight complexes with LPS

(Czuprynski and Welch, 1995; Takii et al., 2005). However, not all of these proteins are found as highly-associated with OMVs as LT. Instead, both OMV-associated and OMV-

independent forms coexist extracellularly. An in-depth study of the association of α- hemolysin with vesicles from different strains of E. coli reveals an important relationship between LPS structure and toxin enrichment. Extraintestinal E. coli exports α-hemolysin

(HlyC), a fatty acid acyltransferase and RTX toxin family member, to the extracellular space through a contiguous protein transmembrane channel. Despite being a known

Type I secretion substrate, a portion of extracellular HlyC is vesicle-associated. Four extraintestinal E. coli isolates (likely R1 core type) display from 1.7% to 31.4% OMV- associated HlyC in culture supernatants. Two E. coli K-12 derivatives (K-12 core type) expressing HlyC show 11% and 17% of toxin as OMV-associated, while in a previously- identified high OMV-producing mutant of MC1061 (galE15 galU galK16, conferring a truncated “rough” LPS core), 66% of the α-hemolysin is vesicle-associated. The differences in LPS structure between these strains may influence the association of α- hemolysin with OMVs and hint at a greater, but as of yet undiscovered, role for OM lipids and lipopolysaccharide in OMV formation (Balsalobre et al., 2006).

However, both OMV-associated export and OMV-independent export also occur for many OMV-associated proteins which are not known to bind or associate with LPS.

Functionally, multiple methods of release for a single protein into the external milieu could enhance different aspects of protein activity. In the case of two virulence factors of H. pylori, VacA and CagA, OMV-association affects function, targeting, and likelihood of delivery with other proteins. In highly virulent H. pylori strains, the oncoprotein CagA

and vacuolating toxin VacA are involved in a functional relationship with one another thought to fine-tune the bacterial-host interaction. Both VacA and CagA are OMV- associated and found on the surface of vesicles (Boquet and Ricci, 2012; Fiocca et al.,

1999; Keenan et al., 2000; Olofsson et al., 2010). The presence of VacA on H. pylori OMVs increases the rate of vesicle association with host cells, as VacA is known to bind sphingomyelin, glycosphingolipids, GPI-anchored proteins, and receptor-like protein tyrosine phosphatase β (Parker et al., 2010). However, VacA is not found exclusively associated with vesicles in culture supernatants. In two strains of H. pylori, only 25% of extracellular VacA is OMV-associated (Fiocca et al., 1999). Likely the presence or absence of CagA in VacA-containing vesicles adds further complexity to the interaction of H. pylori with the host. Additionally, the OMV-associated toxin displays poor vacuolating activity, indicating that the two export methods of VacA, OMV-associated and OMV- independent, may play different pathobiological roles. This difference between OMV- associated and OMV-independent VacA may arise in part due to the necessity for urease, an enzyme secreted independently of OMVs by H. pylori, to induce VacA toxicity

(Keenan et al., 2000). Internalized VacA remains stable in eukaryotic cells and its toxicity may be induced days post-internalization (Sommi et al., 1998).

1.3.4.5 OMV subpopulations

As vesicles serve many diverse functions for bacteria, a population of OMVs is likely varied in composition and size to more optimally accomplish a smaller number of these roles.

In A. actinomycetemcomitans, active CDT (as well as the previously-discussed leukotoxin) associates tightly with OMVs (Rompikuntal et al., 2011). After density- gradient centrifugation, analysis of fractions for CdtB, LtxA, and OmpA reveals potential vesicle subpopulations. Higher density vesicles contain CDT while the vesicles from lower density fractions contain LtxA. Intermediate fractions display the presence of both CDT and LtxA, representing either a third population containing both toxins or an overlap between two content-disparate populations in size-profiles and lipid-to-protein ratios.

In H. pylori OMVs, analysis of density-gradient centrifugation fractions reveals an interesting relationship between OMV subpopulations and cargo enrichment. VacA, known to be enriched in vesicles, is present in all fractions in equivalent concentrations from fractions 5–11. CagA, which is neither enriched nor excluded from vesicles (by protein content) in comparison with its concentration in the OM, can only be observed in the higher density fractions (6–11) and its concentration appears to increase with increasing vesicle density. Two moderately-excluded adhesins, BabA and SabA, display similar profiles in the gradient, being present throughout fractions 3–11 with peaks in

concentration around fraction 7 (Olofsson et al., 2010). As both VacA and CagA also display adhesive properties, the lower concentration of BabA and SabA in the higher density vesicles which contain the most VacA and CagA may indicate that different populations of OMVs could be targeted to different host cells and contain finely-tuned ratios of virulence factors in order to play specialized roles in pathogenicity.

Further analysis of OMVs may be necessary in other Gram-negative species to ascertain the existence, relationship with protein enrichment, and function of multiple vesicle subpopulations.

1.3.4.5 Timing of OMV production

Proteomic analyses and many other studies have yet to fully account for the effect of the many cellular and environmental factors which feed into the complex process of OMV formation and regulation on sample preparation. Vesicles originate from a dynamic cell envelope with structure and composition constantly changing in response to cellular and environmental signals. As such, OMV samples collected from cultures grown overnight are beset with several problems which make it difficult to identify potential vesicle subpopulations and complicate the discovery of a single conserved “vesicle formation machinery.”

Firstly, a culture grown overnight from initial lag phase to the eventual stasis in stationary phase will have undergone dramatic transitions in both peptidoglycan structure and OM composition. The peptidoglycan sacculus of exponentially growing

and stationary-phase E. coli cells differs in glycan length, crosslinking, and covalent attachment of Lpp. Specifically, in stationary phase, the glycan length of peptidoglycan decreases and both crosslinking and covalent attachment of Lpp increase (Vollmer and

Bertsche, 2008).

Furthermore, OM protein composition may not remain constant during a cell’s progression through growth phases. Fernando and Kumar demonstrate the correlation of bacterial growth phase with the varying expression levels of two A. baumannii OM porins (Fernando and Kumar, 2012). Analysis of expression of carO and oprD genes via qPCR revealed a maximum for carO at higher optical densities whereas oprD expression peaked at early exponential phase (Fernando and Kumar, 2012). In V. cholerae, the RTX toxin is found in both an OM-independent and OMV-associated form, the concentrations of which are dependent on complex regulatory mechanisms (Boardman et al., 2007). RTX toxin activity is influenced by the growth-phase-dependent regulation of the Type I secretion system which secretes the toxin, by decreased toxin expression in stationary phase, and by increased protease expression in stationary phase cells.

Accordingly, while RTX toxin may be found abundantly in OMV-independent form during exponential growth, all remaining toxin is observed to be vesicle- or OM- associated in stationary phase (Boardman et al., 2007).

1.3.4.6 Perspectives

The determination of whether a specific protein utilizes the OMV pathway for export in a substantial manner is a complex issue. Proteomic data generate insight on protein presence in a particular vesicle sample but can be misleading, as these do not reveal abundance (unless the analysis is painstakingly undertaken using a quantitative assay) and thus fail to expose whether a component is actively partitioned into vesicles.

However, determination of abundance and enrichment in comparison to other cellular compartments is only meaningful if the amount of an exported protein necessary to elicit a functional consequence is taken into consideration. Whether or not the OMV-associated protein is active, potent, in complex with other requisite factors, or found in OMV-independent forms should also be determined.

Additionally, if distinct subpopulations of vesicles are isolated and analyzed, greater insight into OMV formation mechanisms could be gained. Comparisons between functionally different vesicles subpopulations of a single species could reveal differences in lipid composition, as well as protein enrichment and exclusion. These features could then be related to the state of the envelope under different growth or inducing conditions to reveal how timing and environmental factors contribute to OMV production and function.

Accordingly, an exhaustive study on a specific protein exported by OMVs should address the following points: number of export forms (how much protein is

present in an OMV-independent form versus an OMV-associated state), enrichment/exclusion (how much protein is present in OMVs in comparison to a relevant subcellular compartment), subpopulations (whether the exported protein is only present in a subset of the total OMV population, what other proteins are co- exported within the overall population or subset), timing (whether protein expression correlates with export or whether OMV production and cargo incorporation change during different cell growth phases), activity (measurement of whether protein activity/potency is variable in cell-associated, OMV-associated, and OMV-independent forms, or assessment of whether OMV-association enhances protein complex formation), and location (determination of protein presence in the lumen and/or on the surface of

OMVs).

Future studies of protein secretion via bacterial OMVs will likely benefit from the continually advancing technology in microscopy. As achievable optic resolution improves, along with the availability of new, more visible compartment markers and molecular tags, studies to determine protein location coincident with the dynamics of

OM biogenesis and the cell envelope will soon be feasible.

Discovery of the mechanisms involved in the selectivity of protein export via

OMVs may be species-specific or rely on structural similarities unable to be identified by simple homology studies. Potential vesiculation machinery must be able to integrate and respond to multiple and diverse signals: OM homeostasis, growth stage, metabolic state,

pathogenicity requirements, communication needs, and environmental conditions.

Comparative studies of multiple species using multiple conditions will be important to elucidate the process.

1.4 Membrane bending and curvature formation

All living organisms are composed of one or more cells and all cells are enclosed by a membrane. Transient deformations in membrane curvature are necessary for many cellular processes: vesicular trafficking, cell division, and cell movement (Prinz and

Hinshaw, 2009).

1.4.1 Lipid polymorphism model for membrane lipid composition

Although biological membranes are composed of an overwhelming diversity of lipids, phospholipids are prevalent in both eukaryotic and bacterial membranes.

Phospholipids (PL) may vary in head group identity, acyl chain lengths, and level of acyl chain saturation (Raetz and Dowhan, 1990).

The different PL head groups possess different net charges in physiological conditions. Phosphatidylethanolamine (PE) is a zwitterionic lipid, whereas phosphatidylglycerol (PG) and cardiolipin (CL) are anionic phospholipids.

The nature of the phospholipid acyl chains determines lipid-phase behavior

(Cullis and de Kruijff, 1978). Acyl chain length and saturation level both affect the phospholipid melting temperature and the ability of the lipid to pack within a bilayer. In eukaryotic membranes, the presence of polyunsaturated lipids helps to drive the

formation of membrane rafts due to their poor ability to pack with cholesterol (Bakht et al., 2007; Wassall et al., 2004).

The relative cross-sectional area of the lipid head-group to its acyl chains determines the ability of a lipid to promote positive or negative curvature in a membrane bilayer (Fig. 3). Cylindrical phospholipids like PG promote the formation of lamellar structures (Dowhan, 2013). Conversely, conical-shaped phospholipids PE and

CL spontaneously form hexagonal-phase structures and therefore induce or prefer to localize to areas of negative curvature when contained within a bilayer (Bishop, 2014).

For instance, the poles and septa of many species of bacterial cells are enriched for CL content (Mileykovskaya and Dowhan, 2009).

Figure 3. Shape and spontaneous structures of lamellar and nonlamellar lipids.

(A) A nonlamellar lipid of conical shape is diagrammed. These lipids self-assemble into the hexagonal (HII) phase shown on the right. (B) A lamellar lipid of cylindrical shape is shown assembling into a bilayer of Lα phase.

Bacteria closely regulate the ratio of lamellar to nonlamellar lipids contained within their bilayer. In Acholeplasma laidlawii and Clostridium butyricum, the ratio of bilayer to non-bilayer lipids respond to environmental changes (Goldfine et al., 1987;

Wieslander et al., 1980). An E. coli strain unable to synthesize PE compensates for this deficiency by increasing levels of negatively charged phospholipids (Rietveld et al.,

1993).

It is possible that precise membrane lipid compositions and lipid packing are required for important membrane processes and/or membrane protein functions

(Rietveld et al., 1993). An increasing number of membrane proteins have been discovered to require nonbilayer lipids to stimulate their activity: protein kinase C, phospholipase A2, bacterial translocase secYEG, rhodopsin, and bacterial Mg2+-ATPase, etc. (van den Brink-van der Laan et al., 2004). It follows that the sensing of lipid composition could be performed by proteins which sense lateral membrane pressure or require specific lipid compositions for their function (Rietveld et al., 1993; van den

Brink-van der Laan et al., 2004).

1.4.2 Curvature generation in eukaryotic membranes

The asymmetric distribution of lipid species between membrane leaflets can lead to the formation of membrane curvature (Fig. 4A). Proteins play an important role in the development of lipid asymmetry in this model, but the intrinsic properties and geometries of the lipid species themselves act as the driving force of curvature

generation. Phospholipid flippases, lysophospholipid acyltransferases, phospholipase A, and sphingomyelinases have all been shown to induce local membrane curvature

(McMahon and Boucrot, 2015). Additionally, the clustering of eukaryotic plasma membrane lipids by multivalent bacterial toxins (i.e. cholera toxin, Shiga toxin) concentrates similarly shaped lipids on the outer leaflet to ultimately generate negative curvature (Romer et al., 2007).

Some peripheral membrane proteins can insert wedge-like domains into the bilayer, preferentially increasing the surface area in a single leaflet (Fig. 4B). To accommodate these protein motifs, which are typically amphipathic helices, the membrane bends outwards, with more pronounced effects occurring with numerous closely-spaced protein insertions. Proteins containing wedge-like amphipathic helices involved in membrane curvature formation include: epsin, endophilin, amphiphysin, α- synuclein, and Arf proteins (McMahon and Boucrot, 2015; Prinz and Hinshaw, 2009).

Transmembrane protein clustering (Fig. 4C) may lead to membrane bending if the integral membrane proteins possess a conical shape or a large extracellular domain on only one side of the membrane (Aimon et al., 2014; Copic et al., 2012). Additionally, the oligomerization of integral membrane proteins can serve to scaffold the membrane and may stabilize membrane curvature. For example, the clustering of low-density lipoprotein receptors during endocytosis stabilizes the formation of clathrin-coated pits

(Ehrlich et al., 2004).

Similarly, a high local concentration of membrane-binding or membrane- associated proteins on one side of a membrane has been demonstrated to induce positive membrane curvature (Stachowiak et al., 2012). In this mechanism protein- protein crowding creates steric pressure that is relieved by outward membrane bending

(Fig. 4D). A high enough membrane protein coverage of any protein, regardless of shape or possession of amphipathic helix, will lead to outward bending in response to crowding (Derganc and Copic, 2016; Stachowiak et al., 2012).

Figure 4. Mechanisms of curvature formation in phospholipid bilayers.

(A) Curvature is generated through the asymmetric distribution of phospholipid species. Phospholipids with round head groups represent non-lamellar lipids. A phospholipid flippase (purple) aids in the establishment of the unequal lipid disruption. (B) Asymmetric insertion of wedge-like protein or small molecule components (shown in green) leads to the formation of positive curvature. (C) Clustering of inverted conical- shaped transmembrane channels (in yellow) is influencing curvature. The asymmetric presence of a large proteinaceous domain (pink) also contributes to curvature formation. (D) The crowding of membrane-bound proteins (orange and pink) leads to positive curvature.

1.4.3 Curvature formation in Gram-negative outer membranes

The well-studied methods of curvature generation in eukaryotic model membranes do not translate seamlessly into the study of the curvature generation in the

Gram-negative OM.

For instance, the natural asymmetry between bilayer leaflets complicates the applicability of the model of curvature formation through an asymmetric distribution of lipid species between leaflets. Flippases in the bacterial OM maintain the asymmetry rather than instigate local disruptions. Instead, to examine how this curvature model would play out in the OM, the geometry and intrinsic preference of each lipid species for lamellar or non-lamellar structures must be considered. In E. coli and S. enterica, the major phospholipid of the OM inner leaflet is PE, estimated to makeup 70-80% of the

OM phospholipid content. As the cross-sectional area of the PE acyl chains is larger than that of the head-group, PE displays a slight preference for the non-lamellar hexagonal

HII phase. The minor phospholipids of the OM inner leaflet, CL and PG display preference for the hexagonal and lamellar phases, respectively.

Overall, the phospholipid component of the OM inner leaflet is poised for the adoption of negative curvature. The competing curvature preference of the outer leaflet

LPS, interactions with OM proteins, and lipoprotein stapled to the underlying peptidoglycan sacculus all work to keep the OM in a stable lamellar state.

The insertion of wedge-like domains preferentially to one bilayer leaflet has been proven to be an effective generator of curvature in the OM. The Gram-negative opportunistic pathogen Pseudomonas aeruginosa exemplifies this mechanism though the production of the 2-heptyl-3-hydroxyl-4-quinoline, or Pseudomonas quinolone signal

(PQS). PQS is involved in Pseudomonas quorum sensing and virulence (Diggle et al.,

2006). Once secreted from the cell it intercalates between LPS molecules, increasing the bilayer cross-sectional area preferentially in the OM outer leaflet. This leads to curvature formation and increased instances of OM blebbing (Mashburn-Warren et al., 2008).

Instances of transmembrane protein clustering or protein-protein crowding on the Gram-negative OM have yet to be observed. Much of the study of interactions of proteins or small molecules with OM components focuses largely on those which increase the permeability or otherwise disrupt the OM for antibiotic purposes.

The following chapters outline how LPS structure and LPS-protein interactions influence OM curvature and subsequent OMV formation. In Chapter 2, we utilized changing environmental conditions to modulate the diversity of lipid A structures in wild-type Salmonella in order to study how different lipid A modifications influenced

OMV formation and composition. In Chapter 3, we analyzed the binding of heat-labile enterotoxin to ETEC LPS and analyzed the effect of toxin binding on LPS melting temperature and ETEC OMV production.

2. Outer membrane vesicle production facilitates lipopolysaccharide remodeling and maintenance in Salmonella during environmental transitions

Much of this chapter appears in the journal mBio (Bonnington and Kuehn, 2016).

2.1 Summary

All Gram-negative bacteria alter the structural composition of LPS present in their OM in response to various environmental stimuli. We developed a system to track the native dynamics of lipid A change in Salmonella typhimurium following environmental shift to PhoP/Q- and PmrA/B-inducing conditions. We show that growth conditions influence OMV production, size, and lipid A content. We further demonstrate that the lipid A content of OMVs does not fit with a stochastic model of content selection, revealing the significant retention of lipid A species containing covalent modifications which mask their 1- and 4’- phosphate moieties during host-like conditions. Furthermore, palmitoylation of the lipid A to form hepta-acylated species substantially increases the likelihood of its incorporation into OMVs. These results highlight a role for the OMV response in OM remodeling and maintenance processes in

Gram-negative bacteria.

2.2 Introduction

Gram-negative bacteria alter the composition of their outermost membrane in order to adapt to the disparate conditions encountered in various environmental niches

(Nikaido, 2003; Raetz et al., 2007). Differences in the pH, osmolarity, and abundance of 57

nutrients require constant fine-tuning of the bacterial envelope for maximal fitness. The abilities to respond to environmental changes and survive in stressful conditions are vital to bacterial proliferation and pathogenesis. In the hostile environment of a host organism additional hazards, such as variable concentrations of reactive oxygen and nitrogen species, the presence of bile acids, and assault via antimicrobial peptides, necessitate appropriate and timely membrane changes by cells in order to retain proper barrier functionality (Dalebroux and Miller, 2014).

Vast networks of two-component systems (TCSs) detect external conditions via inner membrane (IM) sensor kinases which can activate cognate transcription factors to promote and/or repress target gene expression (Capra and Laub, 2012). Two of the most extensively studied TCSs regulating outer membrane (OM) composition are PhoP/Q and

PmrA/B. PhoP/Q responds to low pH, low magnesium concentration, and the presence of cationic antimicrobial peptides to activate genes encoding OM proteins, enzymes that covalently modify OM components, and the connector protein PmrD which indirectly activates PmrA-regulated genes (Dalebroux and Miller, 2014). In response to mildly acidic pH, high ferric iron, and toxic aluminum concentrations, the PmrA/B TCS activates genes responsible for modulation of O-antigen length (wzzfepE, wzzST), LPS core decorations (pmrG, cptA), and lipid A modifications (pmrC, pmrR, pbgP) (Chen and

Groisman, 2013). These modifications are known to be critical for Salmonella survival and replication, particularly during the periods of its pathogenic lifestyle when it resides

within mammalian host cell lysosomes or has been engulfed by a phagocytic cell (Ernst et al., 1999).

Certain characteristics central to the Gram-negative OM architecture and biogenesis complicate the process of rapid OM remodeling. For instance, the rate of lipopolysaccharide (LPS) diffusion in the OM is extremely low (Schindler et al., 1980).

Phospholipids may be degraded by phospholipases (e.g. PldA (Scandella and Kornberg,

1971)) and recycled, but there is no known mechanism for LPS or OMP degradation or recycling present in the OM (Rassam et al., 2015). Though LPS molecules may be modified through the acylation activity of PagP (Bishop et al., 2000), a very specific and unique OM-localized enzyme, other covalent modifications of LPS which are up- regulated by environmental signals can only occur during the cytoplasmic process of

LPS biosynthesis. Post-synthesis, these modified LPS species must then be incorporated into the OM through the stochastic process of OM biogenesis (Ursell et al., 2012) to replace the “old” LPS in quantities sufficient to generate the beneficial phenotypic outcome (e.g. resistance to an otherwise deleterious environmental condition).

Furthermore, once generated, sufficient levels of modified LPS species must be retained in order for the cell to derive maximal benefit from the energy-consuming biosynthetic process.

All Gram-negative bacteria studied to date produce OM vesicles (OMVs) (Kulp and Kuehn, 2010). OMVs are the result of a constitutive process whereby sections of the

OM bud outwards from the peptidoglycan layer and pinch off to form spherical structures of 10-200 nm in diameter (Schwechheimer and Kuehn, 2015). We considered that modulation of OMV production as well as content could affect the pace and maintenance of environmental change-induced OM remodeling. Additional selective enrichment of OMV LPS content for species detrimental to bacterial survival in current environmental conditions would further enhance the cell’s ability to change and/or maintain an environmentally ideal membrane composition.

In this study we established a protocol in which bacteria change or maintain lipid

A composition in a defined, reproducible native system. Utilizing this system, we examined OM and OMV composition and measured OMV production levels. We compared experimentally determined relative ratios of LPS species to stochastic model- based theoretical values in order to define the contribution of OMV production to the

OM remodeling and maintenance processes of wild-type bacteria undergoing physiologically-relevant changes in environmental conditions.

2.3 Development of a protocol for measuring environmentally- controlled OM remodeling and maintenance

In order to investigate the role of OMVs to mediate and maintain OM remodeling, it was necessary to develop a protocol that would allow us to examine wild- type cells which are induced to accomplish a significant, measureable OM change in response to a physiologically-relevant environmental shift but have minimal opportunity to remodel their membrane via dilution upon cell growth and division. We 60

further decided to focus on LPS since there are few known OM-localized enzymes

(beyond the limited effect of PagP, PagL, and LpxR) that are known to act on this component after it has reached the OM.

We utilized a well-established model system for PmrA/B- and PhoP/Q- repressing and inducing conditions for Salmonella enterica serovar typhimurium 14028s

(Gibbons et al., 2005). In this model, two growth mediums differing in pH and divalent cation content simulate conditions present in the external environment and the lysosomal compartment. Substantial remodeling of the LPS, particularly its lipid A moiety, occurs upon the shift in environmental conditions (Zhou et al., 2001). At pH 7.6 and 10 mM Mg2+ (7.6H), the lipid A subtypes present are hexa-acylated lipid A species:

1,4’-bisphosphate, 1-diphosphate, and 1-diphosphate,4’-phosphate (Fig. 5A). The high concentration of divalent cations tightly bridges together the negatively-charged phosphate moieties of lipid A molecules, creating a formidable permeability barrier

(Murata et al., 2007). After a shift to pH 5.8 and 10 µM Mg2+ (5.8L), the cell’s membrane composition must rapidly adapt to the loss of magnesium ions, as the inability to bridge the acidic phosphate and pyrophosphate moieties of lipid A with divalent cations creates structural instability. Relief comes through the addition of positively charged 4- amino-4-deoxy-L-arabinose (L-Ara4N) and zwitterionic phosphoethanolamine (pEtN) groups to the phosphates of lipid A, which decrease the electrostatic repulsion between neighboring LPS molecules to stabilize OM bilayer structure (Murata et al., 2007).

Similarly, palmitoylation of lipid A is believed to confer a stabilizing effect on bilayer structure due by increasing hydrophobic interactions with neighboring LPS molecules

(Murata et al., 2007). Accordingly, the number of lipid A subtypes present during the

7.6H-5.8L shift increased due to the gain of both covalent modifications and acyl chain derivatizations (Fig. 5B). The covalent modifications observed include one or two pEtN and/or 4-L-Ara4N groups that are added to the 1 and/or 4’ phosphates (Zhou et al.,

2001). Although wild-type S. typhimurium primarily modifies the 1-position of lipid A with pEtN when L-Ara4N synthesis is occurring, the enzymes which carry out these modifications possess the ability to modify both the 1- and 4’-phosphates with L-Ara4N and/or pEtN groups (Zhou et al., 1999; Zhou et al., 2001). Derivatizations that co-occur with these modifications include the addition of a palmitoyl group to the acyl chain at position 2 (catalyzed by PagP) and/or the replacement of the secondary myristoyl chain at position 3’ with a 2-hydroxymyristoyl by LpxO (Fig. 5B). Modifications catalyzed by

LpxR are not observed in these conditions due to the absence of calcium in the minimal media; likewise, modifications catalyzed by PagL do not occur due to the inhibitory L-

Ara4N modifications decorating the lipid A present in PhoP/Q-inducing conditions

(Manabe and Kawasaki, 2008; Reynolds et al., 2006).

For our quantitative in vitro experiments, we simulated the organism’s pathogenic lifestyle by shifting growth conditions to resemble the environmental challenges faced during epithelial cell invasion, replication, and dissemination (Fig 5C).

A defined minimal medium was used, as this has been demonstrated to result in the most reproducible rates of cell growth (Nikaido, 2003). Salmonella cultures were grown to mid-logarithmic phase (OD600 0.6) in neutral, high Mg2+ media “7.6H” (pH 7.6, 10 mM

MgSO4), lightly centrifuged, and gently resuspended in fresh media of either the same condition (7.6H) or shifted to fresh mildly acidic, low Mg2+ media “5.8L” (pH 5.8, 10 µM

MgSO4) (Fig. 5D). Cells exposed to the reverse order of environmental conditions were grown first to mid-logarithmic phase (OD600 0.6) in 5.8L media, lightly centrifuged, and gently resuspended in fresh media of either the same composition (5.8L) or different composition (7.6H) (Fig. 5D). The 7.6H-5.8L shift in conditions simulates changes encountered upon entry into the acidified Salmonella-containing vacuole (SCV), the 5.8L-

5.8L mock shift in condition mimics survival and replication in an acidified SCV, the

5.8L-7.6H shift represents the changes upon release from an SCV and dissemination into the submucosa, and the 7.6H-7.6H mock shift mimics the environment present during growth in the intestinal lumen or submucosa (Fig. 5C).

Figure 5. Salmonella lipid A structures present under different environmental conditions.

The colored additions (blue: additional phosphate, red: phosphoethanolamine (PEtN), 4- amino-4-deoxy-L-arabinose (L-Ara4N), green: palmitoyl chain) to the base structure are representative of variations present at (A) pH 7.6 10 mM Mg2+ and (B) pH 5.8 10 µM Mg2+. (C) A diagram indicates the infection events mimicked by the environmental shift conditions. Salmonella invasion of epithelial cells to become contained within an acidifying vacuole in step 1. In step 2, Salmonella replication within this acidic Salmonella- containing vacuole (SCV). The third step indicates Salmonella release into the submucosa while the fourth step shows survival and dissemination in this location. (D) The various environmental shift protocols are listed, with high Mg2+ concentration corresponding to 10 mM MgSO4 and low corresponding to 10 µM MgSO4.

Cationic antimicrobial peptide resistance arises due to the OM remodeling that occurs in response to the stimuli present in the 5.8L condition (Zhou et al., 2001). In order to determine what length of exposure to the 5.8L condition would result in a physiologically significant change in the amount of modified LPS in the OM, we assessed survival against a challenge with polymyxin B (PMB). We chose to examine low concentrations of PMB, between 2 and 20 µg/mL, known to increase the permeability of the OM and cause cell death without depolarizing the inner membrane

(Daugelavicius et al., 2000). Cultures were shifted from 7.6H to 5.8L conditions for 0, 45, or 90 minutes prior to treatment with 0, 4, or 8 µg/mL PMB (Fig. 6A-C). Notably, the cultures which experienced greater growth periods in the 5.8L condition displayed higher percentage survival at both 2 and 4 hours after PMB addition. These data demonstrated that the OM changes occurring in the 90 min protocol reflect physiologically relevant and beneficial cellular events. We utilized this 90 min duration in all further analysis.

Figure 6. Previous exposure to mildly acidic low-magnesium conditions increases survival during polymyxin B challenge.

0, 4, and 8 µg/mL PMB was administered to cultures which underwent (A) 0, (B) 45, or (C) 90 minutes in the 7.6H-5.8L environmental shift protocol. These cultures were allowed to grow for 0, 2, and 4 hours before CFU was determined in order to obtain the percent survival of the treated cultures. Representative results are shown.

In order to perform direct comparisons on cells grown in each shift protocol, the effect of the media change on cell proliferation needed to be equivalent. The growth rate, cell morphology, and membrane integrity were characterized. Measurement of OD600 and CFU over time yielded very similar growth curves for each experimental condition

(Fig. 7A, B). The cultures doubled approximately 2 times over the course of the 90 minute shift conditions. Notably, the morphology of the 7.6H-grown cells did not significantly vary in length, diameter, or general appearance from cells grown or shifted into the 5.8L condition (Fig. 8A-D). Further, membrane integrity was not perturbed, as measured by Sytox Green uptake at multiple time points post-shift (Fig. 7C). Therefore, the data revealed no significant differences in cell viability or integrity dependent on environmental conditions that would confound the interpretation of results.

Figure 7. Environmental shift conditions slightly affect cell growth, but not membrane integrity.

(A) Growth curves of OD600 cell density and CFU measurements from 90 minutes of the 7.6H-7.6H and 7.6H-5.8L conditions shown in log scale. (B) Growth curves of OD600 cell density and CFU measurements from 90 minutes of the 5.8L-5.8L and 5.8L-7.6H conditions shown in log scale. (C) The relative florescence cells treated with Sytox Green during time points of environmental shift protocols as compared to heat-killed cells of the same condition.

Figure 8. The environmental shift conditions do not affect cell morphology.

(A) TEM micrograph of cells grown until directly prior to the environmental shift (t=0). (B) TEM micrograph of cells having undergone 90 min of 7.6H-7.6H treatment. (C) TEM micrograph of cells having undergone 90 min of 7.6H-5.8L treatment. (D) ImageJ quantification of cell length and diameter from at least 25 cells chosen at random from 3 representative micrographs at each condition. 2D TLC of glycerophospholipids isolated from the OM of cells having undergone 90 min of (E) 7.6H-7.6H, (F) 7.6H-5.8L, (G) 5.8L- 5.8L, and (H) 5.8L-7.6H treatment. (I) Ruby stained SDS-PAGE analysis of the OM of cells after 90 minutes of environmental shift. Lane 1, unstained protein ladder (Biorad); Lane 2, 7.6H-7.6H OM; Lane 3, 7.6H-5.8L OM; Lane 4, 5.8L-5.8L OM; and Lane 5, 5.8L- 7.6H OM.

To further ensure comparability, the glycerophospholipid (GPL) and protein components in the OM of cells grown in each shift protocol were characterized in order to confirm that minimal changes occur during the initial 90 minutes of each environmental change. The S. typhimurium OM GPL composition has been estimated to be 78% phosphatidylethanolamine (PE) and 22% phosphatidylglycerol (PG), acylphosphatidylglycerol (acyl-PG), and cardiolipin (CL) (Dalebroux et al., 2014). GPLs were isolated from the purified OM of cells which had undergone 90 minutes of 7.6H-

7.6H, 7.6H-5.8L, 5.8L, or 5.8L-7.6H shift conditions before separation via 2D thin-layer chromatography (TLC). A similar gross abundance of two primary GPLs, PE and PG, was observed in the OM for all conditions (Fig. 8E-H). In agreement with prior work demonstrating the regulatory role of PhoP/Q in increasing the levels of CL and palmitoylated acyl-PG within the OM, the OM isolated from the 7.6H-5.8L (shift to

PhoP/Q-inducing) condition displayed a marked rise in acyl-PG and a slight increase in

CL levels in comparison to the 7.6H-7.6H condition (Fig. 8E,F) (Dalebroux et al., 2014).

Exposure to mildly acidic low-Mg2+ media (as in: 7.6H-5.8L, 5.8L-5.8L, 5.8L-7.6H) resulted in roughly comparable levels of both CL and acyl-PG between conditions.

Although minor changes observed in OM-localized GPL head-group composition must be considered in the comparison of cells undergoing the 7.6H-7.6H and 7.6H-5.8L shifts,

GPL composition differences between 5.8L-5.8L and 5.8L-7.6H conditions were largely indistinguishable.

Analogously, the protein composition of OM purified from each condition show extensive similarity (Fig. 8I). However, the end-point of the 7.6H-5.8L shift condition exhibited greater abundance of a ~37 kDa band (likely due to acidic pH-induced OmpC expression) and the presence of a ~23 kDa and ~17 kDa band absent in the 7.6H-7.6H condition OM. The differences between the 5.8L-5.8L and 5.8L-7.6H were minor; the intensity of ~37 kDa band (OmpF/C) decreased and the ~17 kDa band disappeared due to the shift in environment. Accordingly, alterations in OMP composition minimally impact comparison between the two sets of environmental shift conditions.

2.4 Monitoring the lipid A composition of the OM

OM lipid A structures of growing cultures were assessed at 15-30 minute intervals over the 90 minutes following the media change (Fig. 9A,10A,13A,14A,11A).

Lipid A was purified from whole cell samples via the standard chloroform/methanol extraction and hydrolysis method prior to separation by TLC and visualization by charring. In this TLC system, analogous to those used prior (Gibbons et al., 2005), three bands represent the primary lipid A species present during the 7.6H condition (Fig

9A,10A,11A). By contrast, more numerous lipid A species were present in cultures shifted from or shifted to 5.8L media (Fig 10A,13A,14A). Notably, the number of lipid A bands increased fairly rapidly in number (to approximately seven) by 60 min after the shift from 7.6H to 5.8L media (Fig 10A,11A), whereas upon a reverse 5.8L-7.6H shift in conditions, a modest and slow change in lipid A composition occurred, with only the

gradual intensification of upper bands apparent over the 90 minute time period (Fig.

14A).

Figure 9. Comparison of OM and OMV lipid A compositions during the 7.6H-7.6H condition.

(A) Representative TLC separation of lipid A from WC time points (OM) and OMV from the 7.6H-7.6H condition. (B) Negative ion ESI mass spectrometry showing the doubly charged lipid A species from 7.6H-7.6H OM (upper) and OMV (lower) 90 min time points. (C) Values of the ratios derived from densitometric quantification of the bands present in TLC separations. All ratios are in reference to band A. (D) Representation of the densitometric ratios as percentage composition. The ratios presented in (C) are shown are percentages of the total sum of all ratios present per time point. The SE in the composition measurements for each time point are shown in Fig. 15A.

Figure 10. Comparison of OM and OMV lipid A compositions during the 7.6H-5.8L condition.

(A) Representative TLC separation of lipid A from WC time points (OM) and OMV from the 7.6H-5.8L condition. Some lanes have been adjusted in overall contrast for this presentation so the images can be more easily compared, but unaltered images were used during quantification. (B) Negative ion ESI mass spectrometry showing the doubly charged lipid A species from the 7.6H-5.8L OM (upper) and OMV (lower) at the 90 minute time points. (C) Values of the ratios derived from densitometric quantification of the bands present in TLC separations. All ratios are in reference to band A. (D) Representation of the densitometric ratios as percentage composition. The ratios presented in (C) are shown are percentages of the total sum of all ratios present per time point. The SE in the composition measurements for each time point are shown in Fig. 15B.

In previous work, the various lipid A species were designated with group numbers (Gibbons et al., 2005) to indicate both structure and the separation pattern these species take after analysis by TLC (Fig. 11A, C). These group numbers (1-4) are indicated in the results where appropriate.

Mass spectrometry was used to identify the lipid A species present after the 90 min shift in each condition. Three hexa-acylated lipid A species, 1,4’-bisphosphate

(1798/1814 Da; Group 2), 1-diphosphate (1718/1734 Da; Group 1), and 1-diphosphate, 4’- phosphate (1878 Da; Group 3) were confirmed to be present during the 7.6H -7.6H mock shift (Fig. 11C, Table 1). At 90 minutes post 7.6H -5.8L shift, in addition to lipid A species seen in the 7.6H -7.6H mock shift (with the exception of the 1-diphosphate, 4’-phosphate species), the hexa-acylated lipid A species detected were covalently modified with pEtN and/or L-Ara4N, or a hepta-acylated derivative of the base lipid A 1,4’-bisphosphate with a palmitoyl group attached to the position 2 acyl chain (2037 Da; Group 2) (Fig.

11C, Table 1). The covalent modifications observed were as follows: a single pEtN, attached to either the 1- or 4’- phosphate (1841/1857 Da, 1921/1937 Da; Group 3), a single

L-Ara4N, attached to the 1- or 4’- phosphate (1930/1946 Da; Group 3), and one pEtN and one L-Ara4N, attached at the 1- and 4’- phosphates (2053/2069 Da; Group 4) (Fig. 11C,

Table 1). The major species were all hexa-acylated: the base lipid A structure, lipid A modified with 1 or 4’- P-PEtN, and a 1-P-PEtN, 4’-P-L-Ara4N lipid A species (Table 1).

Unlike the culture analyzed 90 min after shift to 5.8L media (7.6H-5.8L), in cultures starting from the 5.8L conditions (either undergoing a mock shift (5.8L-5.8L) or a shift (5.8L-7.6H) in conditions), the cells have had additional time to incorporate further-modified species into the OM. Consequently these featured many singly and doubly PEtN-containing species (2161/2180, 2287/2303 Da), a few L-Ara4N-containing species (1850/1865, 1930, 2103 Da), and a plethora of palmitoylated species (1957, 2103,

2117/2133, 2161/2180, 2287/2303 Da) (Fig. 11C, Table 1). Interestingly, 1-diphosphate, 4’- phosphate species were in moderate abundance (1978/1894, 2117/2133 Da, Group 3) (Fig.

11C, Table 1). The major species in the 5.8L cultures were all hexa-acylated lipid A: 1,4’- bisphosphate, 1-diphosphate, 4’-phosphate, and a singly P-L-Ara4N modified structure.

Figure 11. Lipid A structures present during the 7.6H-5.8L environmental shift with Group numbers and TLC densitometry labeling.

(A) Thin layer chromatography (TLC) separation of whole-cell (WC) lipid A from cells which have undergone the 7.6H-5.8L environmental shift condition. The bands are individually labelled on the right with their predicted Group number on the far right. (B) Three WC lipid A samples are shown charred and stained with ninhydrin to identify which bands represent nitrogen-modified species (pEtN or L-4-AraN). Lane 1, 7.6H-5.8L t=75 min; 2, 7.6H-7.6H t=90 min; 3, 5.8L-5.8L t=90 min. (C) Structures of lipid A species shown with the Group numbers assigned to them by H.S. Gibbons (Gibbons et al., 2005).

Table 1. Lipid A structures present in each environmental shift.

Acyl chain Groups at 1- and/or 4'-positions additions [M-H]- P-L- Media Mra Pc di-Pd P-PEtNe C16:0g OHh pb (m/z) Ara4Nf 7.6H- 7.6H 1718 1715 1 1798 1775 2 1798 1796 2 1813 1811 2 1 1878 1876 1 1

7.6H- 5.8L 1798 1775 2 1798 1796 2 1813 1811 2 1 1841 1838 1 1856 1854 1 1 1921 1918 1 1 1929 1926 1 1 2036 2033 2 1 2053 2049 1 1 2068 2065 1 1 1

5.8L- 5.8L 1718 1716 1 1734 1732 1 1 1798 1751 2 1798 1767 2 1798 1796 2 1850 1847 1 1865 1863 1 1 1878 1882 1 1 1894 1898 1 1 1 1957 1954 1 2088 2086 1 1 2103 2101 1 1 1 2117 2120 1 1 1 2133 2136 1 1 1 1 2160 2161 1 1 1 77

2175 2180 1 1 1 1 2283 2287 2 1 2298 2303 2 1 1 aThe Mr is the calculated molecular weight for the proposed structure bThe [M-H]-p is the predicted value for singly charged MS ions c P, unsubstituted monophosphate ddi-P, unsubstituted diphosphate; At the 1-position eP-PEtN, diphosphoethanolamine; At either the 1- or 4’-position, but likely at the 1- position when L-Ara4N is present fP-L-Ara4N, phospho-L-Ara4N gC16:0, palmitate addition mediated by pagP hOH, in 2-hydroxymyristate mediated by lpxO

2.5 Quantification of OM and OMV-localized lipid A species

In order to investigate the possibility that OM vesiculation could enhance OM remodeling, we needed to analyze the lipid A composition of OMVs produced explicitly over the duration of the environmental shifts. To accomplish this, large volumes of bacterial culture were necessary to obtain sufficient amounts of OMV material to yield the quantities of purified lipid A needed to perform compositional analysis of lipid A subspecies. Due to this restraint, standard methods of 32P labeling and subsequent quantitative TLC analysis could not be feasibly employed. Instead, we performed densitometric analysis of the lipid A species present in the TLC bands observed using the charring method.

For ease of quantitation, each band was assigned a label A-F, with partially- overlapping bands grouped into a single letter category (Fig. 11A). We were aware that the intensities of each band do not directly correspond to the amount of material per sample as different species may char at slightly different rates and inconsistent yields of lipid A may occur per set amount of starting material in purification. Therefore, we controlled internally for these factors by calculating the ratios between an individual band in reference to band A, a band present in all samples analyzed. The reproducibility of this method was confirmed for dilutions of both simple and complex samples using biological and technical replicates (Fig. 12A, B). Additionally, since all TLC plates

analyzed contained an OM sample prepared from the initial (t=0 minute) culture, the consistency of the charring method between different plates could be further verified.

Correlations between the separation pattern and intensity of bands present by our TLC system and the relative abundance lipid A species present by mass spectrometry in a series of samples allowed us to assign the bands into the Groups 1-4 developed by H.S. Gibbons et al (Gibbons et al., 2005) (Fig. 11A,C, Table 2). Accordingly, we strongly suspect that the high intensity of band A on TLC plates (Fig. 9A) corresponds to the abundance of the unmodified 1,4’-bisphosphate lipid A species

(Group 2) observed by ESI-MS (Fig. 9B, Table 1), making it an appropriate denominator for comparisons. Likewise, the comparison between the intensity of bands E and F in the

90 minute 7.6H-5.8L OMV and OM samples (Fig. 10A), coupled with the respective abundance/absence of the doubly-modified lipid A species (Fig. 10B, Table 2) lead us to assign these bands to Group 4. Confirmation of these assignments was obtained by staining for nitrogen-containing groups with ninhydrin (Fig. 11B). As expected, we observed the presence of nitrogenous modifications in the lower bands of the 90 minute time point of the 7.6H-5.8L and 5.8L-5.8L conditions, but not in the 7.6H-7.6H condition.

As bands C and D appear in both the 7.6H-7.6H and 7.6H-5.8L conditions after charring but only in the 7.6H-5.8L condition after ninhydrin staining, these were placed into the diverse Group 3, containing lipid A species modified with either diphosphate, a single

P-PEtN or P-L-Ara4N (Fig. 11C). Group 1 species, which lose phosphate moieties during

the purification process, were confirmed through the lesser intensity of these higher bands by malachite green staining than by charring (see Chapter 5).

Figure 12. TLC densitometry band delineation and charring reproducibility.

Thin layer chromatography (TLC) separation of whole-cell (WC) lipid A from cells which have undergone 90 min of the (A) 7.6H-7.6H or (B) 7.6H-5.8L environmental shift condition. In each dilution, the bands are individually labelled on the left (A-D or A-F). In (B), two overlapping peaks are represented in the labels B and C for ease and reproducibility of quantitation. In the dotted rectangle, the Gel function of ImageJ is demonstrated, with lines drawn from the troughs between the peak(s) of each letter grouping. Easily distinguishable noise is also excluded with lines. The area delineated from this process is measured via the Wand (tracing) tool. Values are then made proportional to band A before conversion to percent composition, as shown on the right. Dark colored columns show the values presented in Figure 15 of this paper while light colored columns show the data obtained from experimental replicates run at a later date.

2.6 OMV-localized lipid A species are somewhat depleted for modified species during the shift from 7.6H to 5.8L conditions

As expected, over the 90 minute duration of the 7.6H-7.6H mock shift, the lipid A composition of the OM remained static (Fig. 9A,C). The B:A ratio hovered around

71.00+/-9.35%, C:A at 19.79+/-7.15%, and D:A at approximately 9.22+/-2.88% (Fig. 9D,

15A). The composition of the OMVs produced over 90 minutes of the 7.6H-7.6H mock shift in conditions closely mirrored the state of the OM over this time: 77.64+/-33.02 %

B:A, 11.03+/-0.29 % C:A, 11.33+/-0.21 % D:A (Fig. 9D). By ESI-MS, the 1,4’-bisphosphate

(1798 Da, m/z 898 [M-2H]2-) and 1-diphosphate hexa-acylated lipid A structures (1878

Da, m/z 939 [M-2H]2-) displayed identical relative intensities in the 7.6H-7.6H OM and

OMVs (Fig. 9B).

In contrast, over the 90 minutes of the 7.6H-5.8L shift in conditions, the lipid A composition of the OM changed markedly as additional Group 3 and 4 species were synthesized (Fig. 10A,B, Table 1). Accordingly, B:A decreased to 18.59+/-1.66% of the total and C:A peaked at 30 minutes post-shift before leveling out at 33.34+/-4.48% after

90 minutes (Fig. 10D, 15B). The D:A ratio, representing Group 3 lipid A species, increased before plateauing around 45 minutes, but due to fluctuations in the abundance of other species this particular ratio made up a relatively stable percentage of the total species, staying in the range of 8.09+/-2.74 to 13.61+/-3.30%, over the 90 minute time period (Fig. 10C,D). The ratios of bands making up Group 4, E:A and F:A, both increased over the course of the 7.6H-5.8L shift. Band E began to appear at only 15 82

minutes post-shift and by 90 minutes E:A makes up 34.78+/-8.52% of the total. Band F initially appeared later, at 30 minutes post-shift, and F:A only increased up to 4.66+/-

0.87% of the total species by 90 minutes (Fig. 10D, 15B).

Notably, the composition of the OMVs collected during the first 30 minutes of the shift from 7.6H-5.8L conditions resembled the state of the OM at 15 minutes. The 30 minute OMVs were composed of 52.40+/-4.37% B:A, 36.96 +/- 15.45% C:A, 5.87+/-2.79%

D;A, and 4.78+/-0.61% E:A (Fig. 10D). The lipid A content of the OMVs produced over 90 minutes of the 7.6H-5.8L shift condition best resembled the OM at a time between 30 and 45 minutes. The 90 minute OMVs were composed of 28.2+/-4.6% B:A, 50.40+/-9.67%

C:A, 9.79+/-0.40% D:A, and 10.98+/-1.69% E:A (Fig. 10D). The percentage of E:A nearly tripled in the OM from the 30 to 90 minute time point. For the OMV composition, the abundance of E:A only doubled when comparing the OMVs collected over 90 minutes to those collected over 30 minutes. Notably, band F was undetectable after 30 and 90 minutes of OMV production in the 7.6H-5.8L shift condition.

Comparison of the lipid A composition of the final OM time point with the

OMVs produced over the 90 minutes of the 7.6H-5.8L shift revealed major differences in the relative intensity of singly- and doubly- modified lipid A species (Fig. 10B, Table 2).

Lipid A structures modified with a single P-PEtN (1841/1856, 1921 Da; Group 3) were present in the OM at approximately two-fold the intensity that they were found in the

OMVs. Similarly, lipid A species modified with a single P-L-Ara4N (1929 Da; Group 3)

were observed in low quantities in the OM, but were not detected in the OMVs. Most strikingly, lipid A species decorated with both P-L-Ara4N and P-PEtN (2053/2068 Da;

Group 4) were seen in high quantities in the 90 minute 7.6H-5.8L OM but were not detected in the OMVs produced over the duration of the environmental shift. These results suggest that OMVs were depleted for LPS species induced upon the shift from

7.6H-5.8L conditions, which could indicate a positive role for OMVs in enabling OM remodeling towards an environmentally-adapted LPS composition.

In regard to acylation changes, the only lipid A-targeted PagP activity detected within the 90 minutes of the 7.6H-5.8L environmental shift was the palmitoylation of

1,4’-bisphosphophate lipid A (Fig. 10B, Table 2). Though the hepta-acylated 1,4’- bisphosphate lipid A species (2036 Da; Group 2) was not observed in the OM of the

7.6H-5.8L shift until after the 60 minute time point, this palmitoylated lipid A displays similar intensity in both the 90 minute OM and OMV samples (Table 2).

2.7 OMV-localized lipid A species are substantially depleted for modified species during the shift from 5.8L to 7.6H conditions

Over the 90 minute time period of the 5.8L-5.8L mock shift, the lipid A composition of the OM remained relatively static, as expected (Fig. 13A, C, Fig. 15C).

The B:A ratio hovered around 11.57 +/- 1.80%, C:A at 7.88 +/- 0.74%, D:A at 1.25 +/-

0.16%, E:A at 29.49 +/- 3.44%, F:A at 12.71 +/- 6.17%, G:A at 17.40 +/- 4.11%, and H:A at

19.70 +/- 2.74% (Fig. 13D, 15C). For the cultures undergoing a shift in conditions from

5.8L-7.6H, we observed a decreased abundance of modified lipid A species over time, 84

although the onset of change was delayed compared to the 7.6H-5.8L protocol (Fig. 14A,

B). Significantly, the late Group 4 ratios G:A and H:A fell to 11.75 +/- 5.64% and 11.90 +/-

6.03% of the total species, with the majority of the decrease occurring after 60 minutes of growth in the new 7.6H environment (Fig. 14C, 15D). The early Group 4 ratios experience a similar trend; E:A levels decreased to 17.00 +/- 0.98% and F:A hovered around 11.18% over the entirety of the 5.8L-7.6H shift time period. Concurrently, the

Group 1-populated B:A ratio swelled from approximately 10.41% to 22.00 +/- 3.48% with the majority of this increase occurring in the last 30 minutes of the 5.8L-7.6H shift. The

Group 3 C:A and D:A levels both increase gradually by 2-3 fold over the 90 minutes to become 20.44 +/- 12.55% and 6.11 +/- 4.18% of the total species, respectively (Fig. 14C,

15D).

Surprisingly, the composition of the OMVs collected from both the 5.8L-7.6H shift and 5.8L-5.8L mock shifted cells did not closely resemble the state of the OM of the cells from either condition. The 90 minute 5.8L-5.8L OMVs were composed of 48.00+/-

1.44% B:A, 13.46+/-0.30% C:A, 8.13+/-7.05% D;A, 11.46+/-4.64% E:A, 8.36+/-4.57% F:A,

4.09+/-0.50% G:A, and 6.50+/-0.92% H:A and the 5.8L-7.6H OMVs were composed of

34.28 +/- 7.11% B:A, 13.70 +/- 2.11% C:A, 13.21 +/- 10.3% D;A, 16.67 +/-6.82% E:A, 10.61 +/-

6.73% F:A, 6.82 +/- 2.15% G:A, and 4.72 +/- 3.89% H:A (Fig. 13D, 14C). The Group 3 lipid

A species present in bands G and H were significantly less abundant in the 5.8L-5.8L and 5.8L-7.6H OMVs than in the respective OM time points. Depletion of these species

in the OMVs is at the expense of the less modified Group 1-2 species present in band B, as these lipid A species make up a significantly higher percentage of the OMV composition than seen in the OM at all time points (0-90 minutes) following the 5.8L-

5.8L and 5.8L-7.6H shifts. These results indicate that OMVs produced in the 5.8L conditions were substantially depleted for the specifically modified LPS species, even after a shift back to 7.6H.

To determine the exact species experiencing enriched incorporation into OMVs during this environmental condition, ESI-MS analysis revealed the relative intensity of lipid A species isolated from both the OM and OMVs (Table 2). Notably, 1-diphosphate,

4’-phosphate hepta-acylated lipid A derivatives (2117/2133 Da), present in low amounts in the 5.8L-5.8L OM, was the most abundant species of 5.8L-5.8L OMVs. Conversely,

1,4’-bisphosphate (1734-1798 Da) and 1,4’-bisphosphoethanolamine hexa-acylated lipid

A species (2283/2298 Da), the predominate species of the 5.8L-5.8L OM, experienced undetectable incorporation into OMVs. Hepta- and hexa-acylated P-L-Ara4N-modified lipid A species (1850/1865, 2088/2103 Da) displayed comparable levels of intensity in

OMVs and OM (Table 2).

Figure 13. Comparison of OM and OMV lipid A compositions during the 5.8L-5.8L condition.

(A) Representative TLC separation of lipid A from WC time points (OM) and OMV from the 5.8L-5.8L condition. (B) Negative ion ESI mass spectrometry showing the doubly (m/z 800-1200) and singly (m/z 1700-2200) charged lipid A species from the 5.8L-5.8L OM (upper) and OMV (lower) at the 90 minute time points. (C) Values of the ratios derived from densitometric quantification of the bands present in TLC separations. All ratios are in reference to band A. (D) Representation of the densitometric ratios as percentage composition. The ratios presented in (C) are shown are percentages of the total sum of all ratios present per time point. The SE in the composition measurements for each time point are shown in Fig 15C.

Figure 14. Comparison of OM and OMV lipid A compositions during the 5.8L-7.6H condition.

(A) Representative TLC separation of lipid A from WC time points (OM) and OMV from the 5.8L-7.6H condition. (B) Values of the ratios derived from densitometric quantification of the bands present in TLC separations. All ratios are in reference to band A. (C) Representation of the densitometric ratios as percentage composition. The ratios presented in (B) are shown are percentages of the total sum of all ratios present per time point. The SE in the composition measurements for each time point are shown in Fig 15D.

Figure 15. Comparison of OM and OMV lipid A compositions during the 7.6H- 7.6H, 7.6H-5.8L, 5.8L-5.8L, and 5.8L-7.6H conditions.

The data from (A) Figure 9C, (B) 10C, (C) 13C and (D) 14C, presented as percent composition with SE. The number of replicates averaged for each time point along with the number of TLC plates these replicates were gathered from are shown below each time point.

Table 2. Comparison of OM and OMV lipid A compositions during the 7.6H-5.8L and 5.8L-5.8L conditions.

Groups at 1- and/or 4'- Acyl chain Relative positions additions intensity [M- H]-pb P- P-L- OMi OMVj P c di-Pd C16:0g OHh e f Media Mra (m/z) PEtN Ara4N

7.6H- 1798 1775 2 100 100

5.8L 1798 1791 2 30 42

1798 1796 2 57 62 1813 1811 2 1 30 37 1841 1838 1 38 22 1856 1854 1 1 16 ND 1921 1918 1 1 84 37 1929 1926 1 1 15 ND 2036 2033 2 1 20 18 2053 2049 1 1 97 ND 2068 2065 1 1 1 36 ND

5.8L- 1718 1716 1 42 ND 5.8L 1734 1732 1 1 100 ND 1798 1751 2 47 ND 1798 1767 2 38 ND 1798 1795 2 <15 ND 1850 1847 1 26 ND 1865 1863 1 1 <15 20 1878 1882 1 1 34 38 1894 1898 1 1 1 36 39 1957 1954 1 25 <15 2088 2086 1 1 <15 23 2103 2101 1 1 1 26 37 2117 2120 1 1 1 <15 90 2133 2136 1 1 1 1 32 100 2160 2161 1 1 1 40 22 2175 2180 1 1 1 1 45 ND 2283 2287 2 1 63 ND 2298 2303 2 1 1 73 ND

aThe Mr is the calculated molecular weight for the proposed structure bThe [M-H]-p is the predicted value for singly charged MS ions cP, unsubstituted monophosphate ddi-P, unsubstituted diphosphate; At the 1-position eP-PEtN, diphosphoethanolamine; At either the 1- or 4’-position, but likely at the 1- position when L-Ara4N is present fP-L-Ara4N, phospho-L-Ara4N gC16:0, palmitate addition mediated by pagP hOH, in 2-hydroxymyristate mediated by lpxO iAbundance of <15 indicates very low abundance. jAbundance of ND was not detected above the noise.

2.8 OMV production increases during the shift to 5.8L media

The relative concentrations of OMVs present in the cell-free supernatants of cultures from each condition were determined by measuring protein content via OMP densitometry and lipid content via fluorescence from the lipophilic dye FM4-64, as described in prior studies (McBroom et al., 2006). Both methods found OMV production to be significantly higher, 2.5-fold by protein measurement and 6-fold by lipid measurement, in cultures undergoing the 7.6H-5.8L shift versus the 7.6H-7.6H mock shift in conditions (Fig. 16A, B). The protein composition of the OMVs collected in both shift conditions was highly similar (Fig. 17B).t

The disparity between FM4-64 and OMP densitometry measurements in the fold increase of 7.6H-5.8L shift over 7.6H-7.6H mock shift OMV production lead us to believe that the OMVs produced in the shift condition may have more lipid material per protein content than those produced in the mock shift. In order to test this prediction, the density profiles of the two OMV populations were determined by equilibrium gradient centrifugation (Fig. 17A). The density of the 7.6H-7.6H OMVs resembles a sharp, symmetric distribution, with the bulk of the OMV material (>60%) residing in fractions 6 and 7. The distribution of the 7.6H-5.8L OMVs is both more broad and shifted towards the lighter side of the gradient, with a peak around fraction 6 and greater than 90% of the material in fractions 4-7.

Figure 16. OMV production during the environmental shift conditions.

The relative OMV production, normalized to CFU, during the 90 minute period of the 7.6H-7.6H, 7.6H-5.8L (A,B) and 5.8L-5.8L, 5.8L-7.6H (C,D) conditions, as calculated for (A,C) protein content by OMP densitometry and (B,D) lipid content by FM4-64 fluorescence. For all conditions, n >3; *, statistically significant differences between production values for the indicated pairs of conditions, p<0.05.

Figure 17. Evaluation of OMV density, OMV protein content, bacterial Lpp-PG crosslinking levels, and OMV production levels for growth in various environmental conditions.

(A) OptiPrep density gradient fractionation of OMVs collected from 7.6H-7.6H and 7.6H-5.8L conditions was performed. Fractions were collected (1= lowest density – 12= highest density) and run on SDS-PAGE, stained with SYPRO Ruby Red (Molecular Probes) and OMV production levels were quantified by OMP densitometry. The level of fluorescence was normalized to total OMP fluorescence of all gradient fractions. (B) Ruby stained SDS-PAGE analysis of OMVs collected after 90 minute environmental shift. Lane 1, unstained protein ladder (Biorad); Lane 2, 7.6H-7.6H OMVs; Lane 3, 7.6H- 5.8L OMVs; Lane 4, 5.8L-5.8L OMVs; and Lane 5, 5.8L-7.6H OMVs. (C) Cross-linked Lpp levels in environmental shift cells. Cells which underwent 90 minutes of the 7.6H-7.6H, 7.6H-5.8L, and 5.8L-5.8L environmental shifts were treated with lysozyme, PG was isolated, and the amount of Lpp co-purified with the PG analyzed by quantitative western blotting using anti-Lpp antibody (Silhavy Lab). The amount of crosslinked Lpp was normalized to cell pellet weights, and relative amounts are shown. (D,E) Overnight OMV production in conditions varying in pH and Mg2+ concentration. Cells were grown in four environmental conditions (7.6H, 7.6L, 5.8H, and 5.8L) overnight before collection of OMVs. The OMV production per CFU relative to the first condition calculated by (D) protein content by OMP densitometry and (E) lipid content by FM4-64 fluorescence.

2.9 OMV production initially remains high during the shift from 5.8L to 7.6H but later reverts to basal levels

Unlike the upregulation of OMV production in the shift from 7.6H-5.8L as compared to the 7.6H -7.6H mock shift, OMV production by 5.8L cultures shifted to

7.6H or mock shifted to 5.8L conditions did not differ greatly over the 90 minute time period (Fig. 16C, D). OMV production during the shift from 5.8L-7.6H was 0.64-fold that of the 5.8L-5.8L mock shift in condition, as measured via OMP densitometry, but was not significantly different from the 5.8L-5.8L mock shift in condition as measured by

FM4-64 fluorescence.

Over this short time period the protein composition of the secreted OMVs from both conditions did not visibly differ (Fig. 17B). However, we did note that upon overnight growth post-shift for both 5.8L-7.6H and 7.6H-5.8L cultures, the abundance of acidic pH-induced OMPs present in the OMVs changed and OMV production of 5.8L-

7.6H cultures returned to the lower levels observed in the 7.6H-7.6H mock shift and

7.6H overnight growth condition (data not shown).

The separate effects of changes in Mg2+ concentration and pH on OMV protein and lipid were also studied. OMV production upon overnight growth in pH 7.6 conditions, regardless of the high or low concentrations of Mg2+, was significantly lower than OMV production in pH 5.8 media as determined by OMPs (Fig. 17D). However, the production of OMVs as determined by FM4-64 for 5.8L-grown overnight cultures was as low as that for 7.6H/L-grown cultures (Fig. 17E). These results suggest that induction of 95

increased levels of OMVs is a long-term effect initiated by the mildly acidic pH and that

Mg2+ concentration may influence the ratio of lipid to protein material packaged into the

OMVs and/or the accessibility of the OMV surface to FM4-64 intercalation.

2.10 Growth in acidic conditions results in increased OMV size

In order to further assess the disparity between protein and lipid measurements of OMVs produced upon shift, we examined negative-stained TEM images of the OMVs.

OMVs collected from 7.6H-7.6H mock shift and 7.6H-5.8L shift cultures (Fig. 18A, B) revealed gross morphological similarities with slight differences in size. The mean diameter of OMVs produced during the 7.6H-5.8L shift protocol was over 20 nm larger:

68 +/- 5.67 nm (n=55) for 7.6H-5.8L OMVs versus 43 +/- 2.63 nm (n=133) for 7.6H-7.6H

OMVs (Fig. 7E). Indeed, the size distribution of the OMVs produced during the 90 min

7.6H-7.6H mock shift skewed towards the smaller sizes (Fig. 18F).

To determine whether the size distribution of the OMVs produced in the 5.8L-

7.6H shift or 5.8L-5.8L mock-shifted cultures differed in response to environmental change or simply possess different properties in varying environmental conditions, we assessed the morphology and size of the 5.8L-5.8L mock shift and 5.8L-7.6H shift OMVs via TEM (Fig. 18C, D). The morphology of these OMVs was similar to the 7.6H-7.6H and

7.6H-5.8L OMVs, however the size distribution for both conditions was shifted to the right, with the average diameters of the 5.8L-5.8L mock shift and 5.8L-7.6H shift OMVs measured as 100.43 +/- 5.7 nm (n=112) and 84.16 +/- 2.7 nm (n=184), respectively (Fig.

18E,G). Therefore, greater time spent in the mildly acidic media correlates with a larger diameter of the OMVs produced in each growth condition (in the order of 5.8L-5.8L (5.5 hr), 5.8L-7.6H (4 hr), 7.6H-5.8L (1.5 hr), and 7.6H-7.6H (0 hr)). Additionally, we observed a greater variation between the mean, median, and mode of the measured diameter of

OMVs produced during an environmental change (the 7.6H-5.8L or 5.8L-7.6H shift conditions) as compared to the corresponding mock shift conditions, where the three parameters did not significantly differ (Fig 18E).

Figure 18. OMV morphology during the environmental shift conditions.

Representative TEM micrographs of negatively stained OMVs collected from 90 minutes of the (A) 7.6H-7.6H, (B) 7.6H-5.8L, (C) 5.8L-5.8L, and (D) 5.8L-7.6H conditions. Size bars indicate 200 nm (A,B) and 500 nm (C,D). (E) Measurement of the mean, median, and mode diameter of OMVs from each environmental shift condition (over three different micrographs, n≥55). The distribution of the measured diameters for the (F) 7.6H-7.6H and 7.6H-5.8L conditions, as well as the (G) 5.8L-5.8L and 5.8L-7.6H conditions.

2.11 The observed OMV composition does not support a stochastic model of OMV production in all environmental conditions

If packaging of lipid A species into OMVs was a completely stochastic process, then the composition of the OMVs produced over the 90 minutes of the environmental shift protocols would be expected to exactly mirror the composition of the OM material over time. For this model we must assume that vesiculation occurs at a constant rate fixed to cell growth, that LPS can mix ideally, and that neither protein content nor the physical properties of LPS molecules affect OMV formation.

To obtain values for the total abundance of each band of lipid A subtypes created over time, we fitted each relevant lipid A ratio (B-H:A) to curves which modeled the progression of membrane lipid A composition. Integration of the curves representing the continuous values of each lipid A component allowed for the formulation of the theoretical composition of OMVs displayed in Fig. 19. We compared the values we derived from this model to our experimental results in order to test if this stochastic model could account for the lipid A composition we observed in the OMVs produced during each condition.

As the predicted lipid A composition of the 7.6H-7.6H mock shift OMVs closely resembled the OM compositions displayed in Fig. 9C, the experimentally determined lipid A composition of these mock shifted OMVs matched closely with the stochastic model (Fig 19A). However, a slightly higher percentage of B:A was observed in the

experimentally determined composition (Fig. 20A). The observed and theoretical lipid A compositions of the 7.6H-5.8L OMVs exhibited substantial differences only in the levels of Group 4 species, whereas the other species were mostly in agreement (Fig 19A). For both E:A and F:A, abundance was less than predicted by the progression of the OM change (Fig. 20A).

Intriguingly, for the 5.8L-5.8L condition, the theoretical model of OMV lipid A composition was vastly different from the experimentally observed composition (Fig.

19B). The abundance of B:A observed in the 5.8L-5.8L mock shift OMVs was approximately 4-fold higher than expected, whereas the amounts of the late Group 4 species, G:A and H:A, were roughly 3-fold lower than anticipated (Fig 19B). An identical trend was observed in the OMVs from the 5.8L-7.6H shift experiment (Fig. 19B, 20B).

These results indicate that the stochastic model for lipid A incorporation into

OMVs seems to only hold for neutral pH/high Mg2+ condition, an environment very favorable for cell growth and not generally associated with other stressors in vivo.

Changing environmental conditions seems to result in delayed incorporation of newly- synthesized species into OMVs. Additionally, exposure to the mildly acidic pH seems to disfavor the incorporation of the highly-modified Group 4 lipid A species into OMVs, regardless of whether the cultures experienced a long-term exposure to mildly acidic pH or a recent shift from mildly acidic pH. These deviations from the stochastic model indicate that other factors influence the regulation of OMV lipid A composition.

100

Figure 19. Comparison of experimentally-determined OMV composition with theoretical stochastic models.

The percent composition of each lipid A species ratio. The observed experimental data is shown next to theoretical values based upon the composition of the OM over time in each condition. (A) Experimental 7.6H-7.6H OMV compositions display the least difference from the predicted composition, whereas the 7.6H-5.8L values deviate somewhat from the respective predicted compositions. (B) Experimental 5.8L-5.8L and 5.8L-7.6H OMV compositions substantially deviate from the predicted respective compositions. (C) The seven main lipid A structures present during the 5.8L-5.8L shift are shown here as 3D space-filling models (in typical CPK coloring, with the majority of Hs not shown for simplicity). The placement of each molecule on the x-axis represents its propensity to be retained in the OM (left, blue), expelled in OMVs (right, red), or neither (center, purple), as estimated from the ESI-MS data presented in Table 2. A calculation combining estimations for the van der Waals volume of each molecule’s head group (HG V) and acyl chains (hydrocarbon V), along with the net charge (absolute value used in calculation) of each molecule was used to place the molecules along the y- axis. The values for the HGV*|charge|/HCV ratio are shown from low (red) to high (yellow).

101

Figure 20. Comparison of experimentally-determined OMV composition with theoretical stochastic models.

Data of Fig. 19 shown in bar graph form with SE.

102

2.12 Discussion

The ability of Gram-negative bacteria to rapidly alter the outermost membrane in response to changing environmental conditions is vital to their survival. In this article, we established a protocol for producing reproducible, measurable OM change in wild- type bacteria undergoing physiologically-relevant changes in environmental conditions.

We performed quantitative comparison of the structural heterogeneity of the lipid A portion of LPS between the OM and OMV content produced over time.

Considering the relatively low levels of LPS diffusion (Muhlradt et al., 1974;

Schindler et al., 1980) present in the OM, coupled with the stochastic, patchwork nature of OM biogenesis (Ursell et al., 2012), the Gram-negative OM could theoretically suffer from widespread and/or localized areas of vulnerability during periods of rapid environmental change if the only means of altering LPS structural variants is through the gradual dilution concomitant with normal growth and cell division. Hostile environmental conditions which restrict or slow cell growth rate could be very detrimental to bacteria in transition to a new ideal membrane composition.

In the neutral pH, high magnesium media, extensive crosslinking between lipid

A phosphate and pyrophosphate residues creates a strong permeability barrier (Murata et al., 2007). The relatively low production and small average size of OMVs observed during the 7.6H-7.6H condition may reflect the robustness of the membrane structure.

103

As the OMV lipid A composition directly mirrored the OM, vesiculation in this condition may be representative of the action of steady-state maintenance processes.

The 7.6H-5.8L environmental shift destabilizes the highly-crosslinked membrane structure via the loss of divalent cations. As such, the progression of the OM lipid A change occurred rapidly in the environmental 7.6H-5.8L shift in conditions, with covalently-modified species detected at the OM after only 15 minutes and the percentage of modified species, reaching approximately 40% of the total composition after 90 minutes. Conversely, the modified species decreased from 80% to 50% of the total composition throughout the 90 minutes of the 5.8L-7.6H reverse shift in conditions.

OMVs produced during both changing (7.6H-5.8L, 5.8L-7.6H) and steady-state

(5.8L-5.8L) acidic, magnesium-limiting conditions were found to contain lesser amounts of the more highly covalently-modified lipid A structures than would be predicted by modeling stochastic production based upon the OM lipid A composition over time.

Likewise, the hepta-acylated 1-diphosphate, 4’-phosphate lipid A structure was found to be enriched in the OMVs secreted during the 5.8L-5.8L shift, while the hepta-acylated

1,4’-bisphosphate species was enriched in the OMV lipid A content during the 7.6H-5.8L shift. The appearance of these trends across disparate environmental shifts (7.6H-5.8L,

5.8L-5.8L, 5.8L-7.6H) points to the importance of the intrinsic physical properties of different lipid A molecules in the vesiculation process (Figure 19C, Table 2). Therefore, we considered whether the well-established model for polymorphic regulation of lipid

104

composition could explain some facets of the observed non-stochastic incorporation of lipid A molecules into OMVs during 5.8L conditions.

In the model for polymorphic regulation of membrane lipid composition, balance between bilayer and non-bilayer-promoting lipids is closely regulated, maintained throughout disparate environments, and essential for cell viability (Tate et al., 1991; van den Brink-van der Laan et al., 2004). Non-lamellar types of lipids, which have an overall conical shape and a preference for the hexagonal phase (HII), are thought to aid in the formation of non-bilayer structures. Accordingly, both the geometry and associated propensity of individual membrane components to form non-bilayer lipid phases in the membrane are important to vesiculation processes (Kates and Manson, 1984).

Considering that 70-80% of the inner leaflet PLs of the OM is comprised of derivatives of PE, a nonbilayer lipid, the intrinsic curvature preference of this leaflet is highly negative (Bishop, 2014). This property is likely mirrored in the outer leaflet, where the lipid A molecules must exert an equally powerful negative spontaneous curvature to maintain proper bilayer structure (Frolov et al., 2011). The opposing spontaneous curvatures of the OM leaflets results in substantial stored membrane stress, the driving force of transformation to nonbilayer lipid structures (Frolov et al., 2011).

Prior to vesicle formation, differences in the elastic properties between the inner and outer leaflets (due to trans-membrane asymmetry) initiate curvature generation (Frolov et al., 2011).

105

As the estimated bulk (68%) of the van der Waals volume of the base 1,4’- bisphosphate hexa-acylated lipid A lies in the hydrocarbon chains (Figure 19C), we have categorized it as a slightly nonlamellar lipid. Increased preference for the HII phase may occur when the effective head group size decreases, the splay of hydrocarbon chains increases, or hydrocarbon chain length increases (Tate et al., 1991). Accordingly, palmitoylation by PagP may serve to increase the nonbilayer propensity of lipid A molecules by increasing both acyl chain length and hydrocarbon splay. The addition of covalent head group modifications could induce the opposite effect through increasing the effective head group size.

Given the estimated rate of diffusion for an LPS molecule (3 x 10-6 nm2/s

(Muhlradt et al., 1974)) is much slower than that of phospholipids, the polymorphic model for regulation of lipid composition must be modified to fully explain the behavior of the much larger LPS molecules. If we assume non-ideal mixing of the lipid A molecules due to interactions between individual molecules, then the importance of the covalent modifications becomes two-fold: geometric and electrostatic. It has been previously proposed that the ability of L-Ara4N and pEtN groups to mask anionic phosphate residues may be important in maintaining the stability and bulk electrostatic properties of the membrane in the absence of divalent cations (Murata et al., 2007).

Supporting this assertion, the hexa- and hepta-acylated 1-diphosphate, 4’-phosphate lipid A species possess greater negative charge to volume ratios than other species

106

present in the 5.8L conditions and both structures are also the most prone to secretion via OMV (Figure 19C). Taken together, both physical and geometric considerations serve to partially explain the increased release of palmitoylated lipid A species via

OMVs and the decreased release of more highly covalently-modified lipid A species

(Figure 19C).

The principles of a polymorphic regulation model combined with considerations for non-ideal mixing of lipid components can also explain the results of a recent study which observed an accumulation of deacylated lipid A species in the OMVs of LB- grown S. typhimurium overexpressing pagL (Elhenawy et al., 2016). The 3-O-deacylase activity of PagL reduces the area of the hydrophobic cross-section of lipid A, giving it a cylindrical molecular shape (Seydel et al., 2000). Although this shape should be lamellar- promoting, in the low concentration of divalent cations present in LB this modification may primarily serve to increase the negative charge to total volume ratio of deacylated species, making them more prone to secretion via OMVs in order to maintain the proper electrostatic properties of the OM surface.

As vesicle production is a complex process involving numerous cell envelope components, additional mechanistic factors may be involved in establishing the lipid A content of released OMVs. Notably, the OMP composition of many bacteria has been demonstrated to change in response to pH. In S. enteritidis, the abundance of major

OMPs C, F, and A varied dependent upon media pH; OmpF and OmpA were both more

107

abundant than OmpC at pH 7, whereas at pH 5-6, a new OMP appeared at 26 kDa while

OmpF experienced lesser expression than OmpC and OmpA (Chart et al., 1994). We postulated that LPS-association with OMPs, as observed for OmpF (Arunmanee et al.,

2014), could possibly influence the rate at which certain species are retained in the OM if the OMPs themselves experience varying rates of incorporation into OMVs. Differences in the distribution of charged residues present on lipid A head group-proximal regions of OMPs may partially account for the observed retention of more highly modified (and thus less negatively-charged) lipid A structures. Despite the small differences in OM- localized protein content which develops during the environmental shifts (seen in Fig.

8I), the observed protein composition of OMVs derived from the 90 minutes of either the

7.6H-7.6H and 7.6H-5.8L shifts or the 5.8L-5.8L and 5.8L-7.6H shifts did not significantly differ (Fig. 17B). Therefore, differential OMP incorporation into OMVs was not likely a major source of the lipid A selectivity or production level differences we observed.

However, it is entirely possible that differences in protein content not visible by the methods used in this work may influence the lipid A content retained in the OM and/or released via OMVs (see Appendix 2 for quantitative comparison of 5.8L OM and OMV protein content by quantitative mass spectrometry).

We next considered that the acidic pH could affect regulatory processes localized in the periplasmic space, resulting in an overall increase in OMV production levels. In previous mechanistic studies, the activity of enzymes which modulate the structure of

108

the peptidoglycan (PG) layer and the activity of proteins which modulate the levels of covalent crosslinking of the PG to the OM via the lipoprotein Lpp have been shown to influence OMV production (Schwechheimer et al., 2015). We found that the total cellular amount of Lpp-PG crosslinking observed during the 7.6H-5.8L shift, 7.6H-7.6H mock shift, and 5.8L-5.8L mock shift conditions did not differ significantly (Fig. 17C).

However, this experiment gave no information on whether the Lpp-PG crosslinks may be altered in their lateral distribution, allowing for the targeted retention of certain OM areas over others.

The OMV lipid A composition of the 5.8L-5.8L condition may have given us further insight into the nature of OM biogenesis. While 50-100 nm bursts of protein material being added to the OM have been observed using electron and fluorescent microscopy, the LPS content and protein heterogeneity of these additions remains entirely unknown (Smit and Nikaido, 1978; Ursell et al., 2012). Due to the slow diffusion rate of LPS molecules (3 x 10-6 nm2/s (Muhlradt et al., 1974)), our data suggests that certain groups of similarly-modified lipid A molecules may be incorporated into the OM in a single burst event.

The relative dearth of covalently-modified LPS species in OMVs produced during the 5.8L-7.6H shift, the 5.8L-5.8L mock shift, and (to a lesser extent) the 7.6H-5.8L shift in conditions may reflect the ability of bacteria to maintain a remodeled OM composition despite an environmentally-induced increase in OMV production. A

109

multitude of organismal benefits could result from this retention of modified lipid A species in the OM. The resultant masking of the negatively-charged phosphate moieties may allow for a more robust permeability barrier in the lower concentration of divalent cations. Unmodified lipid A species, which take less energy to synthesize, may be more energetically favorable for the cell to lose to the OMV functions of OMP maintenance and membrane damage alleviation. Furthermore, the release of less modified and more highly-acylated lipid A structures could act to manipulate the host immune system response, whereby the structure most encountered by immune cells would not be representative of the invading cell’s actual membrane composition.

A model based upon the polymorphic regulation of different lipid A structures combined with the stochastic, patchwork-nature of OM biogenesis may account for the unexpected retention of more highly-modified lipid A species in the OM during the mildly acidic environmental conditions. Sensing of these local differences through association with OMPs, periplasmic regulatory protein activity and/or by differential

Lpp tethering of the OM to the underlying PG could account for the deviation of observed OMV lipid A content with the stochastic OMV formation model.

110

3. Lipopolysaccharide binding by heat-labile enterotoxin influences outer membrane dynamics in enterotoxigenic Escherichia coli

3.1 Introduction

Carbohydrates are the most abundant class of organic molecule in nature (Wade,

1999). These molecules function not simply as energy storage, but also as the surface covering for most types and species of eukaryotic and bacterial cells. Due to their high potential for structural variability, carbohydrates serve as molecules for cell recognition, cell-cell signaling, and adhesion (Ghazarian et al., 2011). Considering the biological importance and ubiquity of these molecules, microbes inevitably exploit eukaryotic carbohydrate receptors for pathogenic infection.

Enterotoxigenic E. coli (ETEC), the primary causative agent of traveler’s diarrhea, is responsible for as many as 1.5 million deaths per year (Qadri et al., 2005). Heat labile enterotoxin (LT) is the main virulence factor of ETEC. Without LT expression, ETEC fails to colonize mouse intestines or instigate disease symptoms in gnotobiotic piglets (Allen et al., 2006; Berberov et al., 2004).

LT is a member of the AB5 toxin family and, as such, possesses five binding (B) subunits which assemble into a donut-shaped homopentamer topped with the catalytic

A subunit. The B subunits ensure GM1-mediated uptake of the holotoxin by small intestinal epithelial cells, while the ADP-ribosylase activity of the A subunit is responsible for initiating a cascade of intracellular events leading to host cell chloride

111

efflux. The pentamer of the toxin’s B subunit, LTB, can bind three different carbohydrate- containing ligands: ganglioside GM1, group A blood antigen, and lipopolysaccharide

(LPS) (Mudrak and Kuehn, 2010a). The binding sites for each of these sugars exist interstitially between subunits in different regions of the LTB pentamer: GM1 binds to the pentamer base distal to the A subunit, blood sugar binds to the pentamer’s periphery, and LPS binds to areas on the side and base of the pentamer, both distinct from and partially overlapping with both GM1 and blood sugar binding sites.

LTB shares 83% amino acid identity with the B subunit of the homologous cholera toxin (CT) (Holmner et al., 2007). CT and LT share catalytic activity of their A- subunits and both utilize the type II secretion system to reach the extracellular milieu

(Sandkvist et al., 1997; Tauschek et al., 2002). However, the majority of extracellular LT is found in complex with secreted ETEC outer membrane vesicles (OMVs) whereas CT is not observed in complex with Vibrio cholera OMVs in cellular supernatants (Horstman and Kuehn, 2000b, 2002). Like LT, CT is able to bind to the ETEC cell surface (Horstman et al., 2004; Horstman and Kuehn, 2002).

A more distantly-related AB5 toxin, Shiga toxin (ST, specifically Stx2), was shown to bind to E. coli O157:H7 LPS (Gamage et al., 2004) along with its host receptor, glycosphingolipid globotriaosylceramide Gb3. In mouse cells and model phospholipid membranes, ST (Stx1) induces narrow tubular membrane invaginations (Romer et al.,

2007). Both a precise molecular architecture of the binding sites and a fluid bilayer are

112

necessary to form these concave invaginations through membrane-mediated clustering

(Pezeshkian et al., 2016). Conversely, protein-protein crowding forms convex membrane bending through the buildup of lateral pressure generated by collisions between membrane-bound proteins (Stachowiak et al., 2012). Protein-protein crowding is a well- established model for curvature formation in eukaryotic membranes, however, classical models of curvature formation applicable in phospholipid membranes have yet to be studied in the context of the unique composition of the asymmetric Gram-negative outer membrane (OM).

To further investigate the interaction of LTB with LPS, we mapped which sugars of the oligosaccharide core (OC) were vital for LTB binding using a set of truncated structures of the R1 OC in a sedimentation/depletion assay. We determined the binding affinity and stoichiometry of LTB-LPS binding by isothermal titration calorimetry (ITC).

Utilizing Fourier-transform infrared (FT-IR) spectroscopy to investigate secondary structural changes of the toxin B-pentamer, we found that LTB secondary structure does not substantially change upon binding R1 OC LPS. In order to investigate the effect of this lectin’s ability to bind LPS on membrane fluidity we measured the effect of WT and cell surface-binding deficient LTB concentrations upon the gel-to-liquid crystalline phase transition of R1 OC LPS using differential scanning calorimetry (DSC). We then assessed the physiological effect of LT binding to native LPS membranes by measuring outer membrane vesicle production of a set of ETEC strains with tunable LT expression.

113

Finally, we present a model whereby external LT binds LPS and the low diffusion rate of

LPS in the Gram-negative OM intensifies protein crowding effects via lateral confinement (Derganc and Copic, 2016), deforming localized areas of membrane to augment subsequent vesicle formation processes.

3.2 LTB binds to galactose residues in the R1 type LPS core

Prior work in the Kuehn and Fleckenstein labs uncovered the sequence-specific ability of LT to bind to the cell surface of ETEC and other E. coli strains of the R1 and R2

LPS OC types (Johnson et al., 2009; Mudrak et al., 2009). Accordingly, we ascribed this cell surface binding ability to an affinity of the holotoxin for LPS. A 2010 study posited that LT did not bind LPS, utilizing LPS of an R3 core type to demonstrate this claim

(Jansson et al., 2010). Additionally, our lab demonstrated that both LT and homologous

CT cannot bind the Vibrio cell surface or its LPS (Horstman et al., 2004). As different strains of Gram-negative bacteria possess different LPS core structures (Fig. 21A) and LT is primarily secreted by cells displaying the R1 type OC, we hypothesized that the cell surface-binding ability of LT may be determined by both the presence and placement of certain sugars and/or phosphate moieties within a specific carbohydrate structure.

To determine the minimal structure necessary for LTB binding, we performed

LTB sedimentation/depletion experiments with a variety of R1 OC type truncations purified by the Darveau and Hancock method. The LPS sedimentation/LTB depletion experiments measured the ability of each LPS type to deplete the concentration of

114

soluble LTB through cosedimentation with LPS molecules upon ultracentrifugation. For these experiments, the same concentration of LTB (7 µM, to ensure consistent detection via Ruby-stained SDS-PAGE) was utilized throughout, paired with a series of concentrations of LPS isolated from full-length (ΔwaaV) and five truncations of the R1 core (ΔwaaT, ΔwaaG, ΔwaaP, ΔwaaQ, and ΔwaaY) in order to determine any differences in binding affinity (Fig. 21 B,C).

LTB bound significantly to ΔwaaV, ΔwaaP, ΔwaaY, and ΔwaaQ core structures, but poorly to ΔwaaT and ΔwaaG cores (Fig. 21 D,E). In analyzing the significant differences in binding affinity between ΔwaaV/P/Q/Y and ΔwaaT/G structures, we noticed that these two groups of structures differed in galactose content. The presence of galactose I and II significantly increased the likelihood that an R1 core truncation would display higher affinity binding to LTB. Corroborating this finding, the Vibrio cholerae OC contains no galactose within its structure and is unable to bind toxin (Horstman et al.,

2004).

In each of the OC types which significantly bound to LTB, R1 and R2, one to two

β-galactose moieties branch out from the main length of the core structure, that which is in-line with the O-antigen repeats. We propose that the toxin recognizes this accessible, terminal galactose residue. Notably, in the region where the R1 core displays galactose residues the R3 core type presents glucose residues. These may obscure the singular galactose sugar of the R3 core, making it inaccessible for toxin binding.

115

The presence of phosphates on heptose I and heptose II was found to be dispensable for toxin binding. We observe only a minimal difference in estimated Kd between ΔwaaV and ΔwaaP (Fig. 21 D,E) with the non-equilibrium LPS sedimentation/LTB depletion assay.

116

Figure 21. LTB binding to mutants of the E. coli R1 core.

(A) All of the structures of the LPS cores previously assessed for LTB binding ability are represented. The structures of each of the R1 core mutants utilized in this study are shown in (B). (C) The different OC structures are separated on silver stained 12% SDS- 117

PAGE gel. Data from the LPS cosedimentation/LTB depletion assay were fit to one-site binding curves (E) in order to obtain the estimated non-equilibrium Kd values shown in (D). The standard error (SE) of the calculated Kd and the r2 of the fit are also shown in (D).

118

3.3 LTB binds to ΔwaaV LPS with low micromolar affinity

We next sought to quantitatively determine the affinity and stoichiometry of the

LTB-LPS binding interaction. We utilized isothermal titration calorimetry (ITC) to determine the binding thermodynamics of LTB to purified ΔwaaV LPS in the triethylamine salt form. The purified ΔwaaV LPS was used in lieu of ETEC LPS as the truncation of the O-antigen gave these samples a higher level of homogeneity between molecules and superior solubility in aqueous buffers.

We found a 1:1 stoichiometry between each LTB monomer and LPS molecule.

Using a one site binding model, the Kd was measured as 4.4 ± 0.2 µM (Fig. 22).

Comparison of the binding thermodynamics between LT-LPS and previously published values for LT-GM1 binding (0.57 nM) reveals a four orders of magnitude difference in the binding affinity.

Although superimposed crystal structures of sialic acid-bound and free LTB revealed little-to-no structural differences (PDB: 1qb5, 5g3l; RMSD =0.285 Å for all 2995 atoms), binding to small molecules like sugars may elicit more dynamic structural changes that may be better captured by solution-state methods. Therefore, we decided to utilize Fourier-transform infrared spectroscopy (FTIR) to probe for possible secondary structural changes of the LTB pentamer induced upon binding to ΔwaaV LPS. The amide

I region (primarily C=O stretching vibrations) of the infrared spectra presents bands characteristic of certain protein secondary structures. The two main bands at 1634 cm-1

119

and 1665 cm-1 are indicative of antiparallel β-sheets and β-turns, respectively (Barth,

2007). The band at 1652 cm-1 is indicative of α-helical structure (Fig. 23 A-D). We utilized curve fitting to determine the relative composition of secondary structural features in the amide I region. Pure LTB displays majority β-sheet and β-turns (Fig. 23 A). After incubation with ΔwaaV LPS, only slight changes in the secondary structure occur (Fig. 3

B). Similarly, minimal differences in percentages of α-helical character, β-sheet character, and turns were observed between pure cell surface-binding-deficient LTB Y18H (Mudrak et al., 2009) and LTB Y18H incubated with ΔwaaV LPS (Fig. 23 C,D).

120

Figure 22. Analysis of LTB-ΔwaaV LPS binding by isothermal titration calorimetry.

A representative scan is shown with the binding curve obtained from fitting to a one-site binding model. The N is calculated as 0.96 ± 0.02 sites, Kd as 4.4 ± 0.2 µM, ΔH as -1.469E4 ± 390.7 cal/mol, and ΔS as -24.8 cal/mol/°C. These prelimary calculated Kd, N, ΔH, and ΔS values with standard error are obtained from one experiment. Data collected by Daniel L. Rodriguez.

121

Figure 23. Infrared spectra in the range of amide I (predominantly C=O stretch) vibrational band for WT LTB and LTB Y18H alone and in the presence of hydrated ΔwaaV LPS.

122

For each spectra (A-D), the original data is displayed in red, the fitted trace in blue, and the component bands of the fitted trace in green. A representative spectra for the B- pentamer alone is shown in (A) for WT LTB and in (C) for LTB Y18H. A representative spectra from molar ratio of 1:0.125 ΔwaaV LPS:LTB monomer is displayed in (B) and (D) for WT LTB and LTB Y18H, respectively. The two main bands at 1634 cm-1 and 1665 cm-1 are indicative of β-sheets and β-turns, respectively. The band at 1652 cm-1 is indicative of α-helical structure. For each fitted trace, R2 ≥ 0.99 and RMS noise ≤ 9.77E-6.

123

3.4 LTB binding decreases the ΔH of soluble ΔwaaV LPS gel-to- liquid crystalline phase transition

We sought to determine the effect of LTB binding on the biophysical characteristics of the ΔwaaV LPS molecules. With differential scanning calorimetry

(DSC) we were able to observe changes in the gel to liquid crystalline phase transition of

ΔwaaV and ΔwaaT LPS molecules upon incubation with LTB pentamers in solution. In

Fig. 24A the specific excess heat is plotted against temperature for three LTB: ΔwaaV LPS molar ratios. The pure ΔwaaV LPS phase transition calorimetric enthalpy (ΔHcal) was

53.9 kJ/mol, with a heat capacity maximum at 32.3°C. With the addition of increasing ratios of LTB per ΔwaaV LPS molecule, the melting temperature (Tm) remains constant whereas the ΔHcal of the phase transition experiences a linear increase. This effect on the phase transition is consistent with a hydrophilic protein interacting with bilayer surface through electrostatic forces (Canadas and Casals, 2013). Notably, an increase in ΔHcal does not occur in the phase transition of ΔwaaT LPS upon LTB addition (Fig. 24 B), indicating that sugar binding is necessary for wild-type levels of ΔHcal increase.

To further investigate the electrostatic nature of the LT-LPS binding interaction, we examined the phase transition of ΔwaaV LPS in the absence of divalent cations (Fig.

24 C). As before, the Tm of ΔwaaV LPS remains constant with increasing LTB concentrations (Fig. 24 C). However, the ΔHcal of the phase transition experiences a linear decrease with increasing molar ratios of LTB in this non-physiological condition.

Importantly, the addition of cell surface binding-deficient mutant LTB Y18H did not 124

change the Tm or the ΔHcal of the phase transition (Fig. 24 D); therefore the binding of LT to LPS is necessary to affect changes in the LPS phase behavior in the presence and absence of magnesium.

The infrared spectra were also analyzed with respect to an influence of LTB on the antisymmetric stretching vibration of the negatively charged phosphate vas (PO2-) of

LPS in the spectral range 1300-1180 cm-1 (Parikh and Chorover, 2007). The band contour of this vibration usually consists of two to three single components resulting from different degrees of hydration of the phosphate groups of the LPS. Bands at a higher wave number correspond to lower water binding and vice versa (Brandenburg et al.,

2003). Pure ΔwaaV LPS exhibits a broad range of hydrations with overlapping bands

(Fig. 25 A,B). The band around 1230 cm-1 is usually attributed to the 1-phosphate of the lipid A portion of LPS, whereas a band at approximately 1260cm-1 is characteristic of the

4’-phosphate (Brandenburg et al., 1997). The 1-phosphate is thought to project outwards into the hydrophilic environment due to the strong inclination of the lipid A molecule with respect to the membrane plane (Mashburn-Warren et al., 2008; Seydel et al., 2000).

With an increase in the molar ratio of toxin to LPS, the band at 1233 cm-1 experiences a reduction in intensity (Fig. 25 A), indicative of loss of dipole moment via partial phosphate immobilization (Brandenburg et al., 1998). This phosphate immobilization is also observed for the interaction of cell surface-binding-deficient mutant LTB Y18H with

ΔwaaV LPS (Fig. 25 B), though not as strongly as with the WT B-pentamer (Fig. 26 C).

125

This trend is replicated in the vibrations arising from sugar C-O stretching, with significant reduction in the band at 1077 cm-1 upon toxin B pentamer addition. Notably, the C-O stretching band around 1050 cm-1 (Parikh and Chorover, 2007) is further depressed only in the interaction of WT LTB with ΔwaaV LPS (Fig. 25 C). These data support the conclusion that the interaction of LTB with LPS is that of a hydrophilic protein binding to the surface sugars electrostatically.

Figure 24. Heat-labile enterotoxin concentration-dependent changes in LPS gel-to-liquid crystalline phase transition.

Molar excess heat capacity versus temperature determined with DSC for 100 µM of either ΔwaaV LPS (A) ΔwaaT LPS (B) in buffer containing equimolar MgSO4 or ΔwaaV LPS in buffer devoid of divalent cations (C, D) (representative curves shown).

126

To compensate for the influence of neighboring vibrations, a baseline between 3000 cm-1 and 900 cm-1 was subtracted. In A and B, the spectra were normalized to the most intense

127

band, 1077cm-1, with an abscissa set to 1822 cm-1, an area which consisted of low levels of baseline noise. In order to observe the C-O stretching vibrations in C, the spectra were normalized to a common band at 1546 cm-1, with an abscissa set to 1822 cm-1. The inset in C is normalized to 1077cm-1 in order to accentuate the difference at 1050 cm-1.

128

3.5 LT secretion and LPS binding are necessary for wildtype- levels of OMV production in ETEC

In the physiological context of the bacterial OM, lipopolysaccharides are thought to exist in the gel-state, due to their high-level of crosslinking by divalent cations and rigid, impermeable nature. In purely phospholipid bilayers, temperatures above the Tm lead to the liquid crystalline state, where the creation of membrane area and fluidity may result in the spontaneous shedding of vesicles (Dobereiner et al., 1993). Considering the effect of LTB binding upon the pure LPS phase transition in vitro, the binding of the toxin to the rigid LPS outer leaflet in vivo may act to lower the energy barrier necessary to transition to a more fluid and vesicle-promoting state.

To test this hypothesis and assess the physiological impact of LT-LPS binding on native membrane dynamics, we examined the ability of LT+ ETEC cells to form OMVs.

The relative OMV content existing in the cell-free supernatants of different strains grown in CFA media were determined through measurement of protein concentration via OMP densitometry and/or lipid concentration via fluorescence from the lipophilic dye FM4-64 as described in a prior report (McBroom et al., 2006). We first compared the vesiculation phenotypes of WT ETEC and ETEC ΔeltA, a strain that cannot express the

LT holotoxin, finding that the OMV production level of the ETEC ΔeltA strain was approximately 2.5-fold lower than WT (Fig. 26A). Accordingly, the expression of LT seems to be necessary for WT-levels of OMV production in ETEC.

129

Prior studies of mechanistic factors which influence OMV production levels have found that the build-up of proteins in the periplasm correlates with higher OMV production (Schwechheimer and Kuehn, 2013). As LT expression results in both LT localization to the periplasmic space prior to secretion and localization to the cell surface following secretion, we sought to determine whether the secreted or periplasmic population of LT promotes wildtype-levels of OMV production. To this end, we examined the vesiculation phenotype of ETEC ΔgspD. In the ΔgspD strain, holotoxin localization remains periplasmic due to the deletion of the T2SS OM secretin. Compared to WT, ΔgspD displayed 2.5-fold lower OMV production (Fig. 26 A). Therefore, the proper secretion of LT is necessary for its influence on OMV production.

130

Figure 26. External localization and LPS-binding ability of LTB are necessary for promotion of OMV production.

The relative fold OMV production, as measured by OMP densitometry, of three ETEC strains (WT, ΔeltA, ΔgspD) is shown in (A). In (B), the relative fold OMV production, measured by both OMP densitometry and FM4-64 fluorescence, for ΔeltA empty vector, ΔeltA pILT, ΔeltA pILT T47A at 0, 5, and 25 µM IPTG is displayed. LTB production in ΔeltA empty vector, ΔeltA pILT, and ΔeltA pILT T47A at 0, 5, 25 µM IPTG is shown by Western blot in (C). Comparisons between strains and treatments were performed using the Student’s t test. All statistical analyses were performed using JMP (SAS Institute). Results are presented as means ± standard error of the means, and all experiments were performed in triplicate.

131

3.6 LTB expression increases OMV production in ETEC

To confirm that increased production of LT is correlated with increased OMV production, we compared the OMV production of ETEC ΔeltA strains containing a plasmid for IPTG-inducible WT LTB expression to ΔeltA carrying an empty vector control (Fig. 26B). As measured by both relative lipid and protein content, the OMV production of ΔeltA with inducible LTB expression increased with increasing concentrations of IPTG relative to the empty vector control. Therefore, we confirmed a link between higher LT expression and higher levels of OMV production.

To determine if the LPS-binding ability of LT is necessary for its role in promotion of OMV production, we compared the vesiculation phenotypes of ΔeltA containing a plasmid for either inducible WT LTB or LTB T47A expression (Fig. 26B). For both LT-expressing strains, the concentration of WC LT per WC total protein increased with increasing IPTG concentrations (Fig. 26C). However, for the point mutant deficient in cell-surface binding ability (Mudrak et al., 2009), increased concentrations of WC

T47A LTB did not correlate with the increased OMV production we observed in ΔeltA expressing WT LTB. Relative to whole cell (WC) protein content, the periplasmic protein content did not significantly differ between ΔeltA with empty vector, inducible WT LTB, and inducible LTB T47A at varying concentrations of IPTG (data not shown). We therefore conclude that the external binding of LTB to LPS is necessary for the promotion of OMV production.

132

3.7 Discussion

We thoroughly demonstrated the ability of LT to bind lipopolysaccharides using

LPS cosedimentation/LTB depletion assays, ITC, and FT-IR. Through FT-IR, we observed the secondary structure of LTB shift slightly upon LPS binding to destabilize steady-state levels of β-sheet structure and increase α-helical character. Additionally, the bands corresponding to antisymmetric phosphate stretching and sugar C-O stretching of LPS were depressed upon LTB addition, indicating substantial immobilization. This suggests that both phosphate and sugar moieties contribute to LTB binding specificity.

Through LPS cosedimentation/LTB depletion assays, we determined that an accessible terminal galactose moiety is required for LTB to be able to bind an LPS OC structure. In ganglioside GM1 the terminal galactose residue forms the most contacts to

LTB and the majority of the galactose sugar becomes buried within the binding pocket

(Merritt et al., 1994a; Mudrak and Kuehn, 2010a). As the residues of LTB important for

LPS binding partially overlap with those of the GM1 galactose binding pocket, it is possible that both substrates utilize the same site for galactose recognition (Mudrak et al., 2009). However, since LT bound to LPS on the surface of OMVs has been shown to bind to both GM1 and LPS simultaneously (Horstman and Kuehn, 2002), it is also possible that two separate galactose binding sites, varying in binding affinity and substrate specificity, exist on the LTB pentamer. In this model the different residues and

133

secondary structural features surrounding two distinct sites would result in varied preferences for substrate glycolipid saccharides leading up to the galactose residue.

Possessing multiple binding sites for similar ligands would not be unprecedented in the AB5 toxin family; the B-pentamer of Shiga toxin (Stx1 and 2) displays 15 binding sites of varying affinity for the digalactose motif of glycolipid Gb3

(Soltyk et al., 2002). For Stx1 and Stx2, differential affinities of the B-subunits for host glycolipid Gb3 variants correlate with different toxicity and in vivo potency (Shimizu et al., 2007). Differences in toxicity between CT and LT have been mapped to a portion of the A subunit responsible for holotoxin stability during uptake and transport into epithelial cells (Rodighiero et al., 1999). Furthermore, LT internalized in conjunction with OMVs exert functions that LT alone cannot (Chutkan and Kuehn, 2011).

By ITC we measured the Kd of LTB-LPS binding as 4.44 µM. However, a 4-fold lower Kd was measured via the LPS cosedimentation/LTB depletion assay. Since the ITC measurements were made in the presence of the solubilizing agent triethylamine, this method might be less physiologically accurate than the cosedimentation/depletion assay, which utilized LPS in solution without any solubilizing agents. Nonetheless, the lowest observed Kd for LTB binding to LPS binding is still ~300-fold higher than the Kd measured by surface plasmon resonance for GM1 binding, ~57-fold higher than the Kd measured for GD1b, and ~11-fold higher than the Kd measured for asialo-GM1 (MacKenzie et al., 1997). Physiologically, this lower strength of binding would not hinder LT from

134

binding LPS; following secretion via the type II secretion system the LT encounters high concentrations of LPS on the bacterial cell exterior and the multivalent B-pentamer can bind multiple LPS molecules at once.

Binding of proteins to LPS can cause a variety of different effects on the biophysical characteristics of the membrane based upon where on the LPS molecule the protein binds. Proteins which bind to and partially penetrate the bilayer cause a decrease in the Tm and ΔH of the gel to liquid crystalline phase transition due to disruption of van der Waals interactions between LPS acyl chains (Canadas and Casals,

2013). The interaction of the lectin pulmonary surfactant protein A (SP-A) with model

LPS membranes occurs in this manner (Canadas et al., 2008). Proteins (e.g. lysozyme) which bind to the lipid A phosphates and acyl chains of lipid A generally cause both a decrease in ΔH and an increase in Tm of the phase transition, effectively rigidifying the acyl chains through binding (Brandenburg et al., 1998). Similarly, the interaction of hemoglobin with LPS stabilizes the gel phase through increasing the Tm with no effect on the enthalpy (Brandenburg et al., 2003). In stark contrast, purely electrostatic interactions (e.g. poly(L-lysine) binding to phosphatidylglycerol) lead to substantial increases in the ΔH of the gel to liquid crystalline transition in a dose-dependent manner

(McElhaney, 1986). We observed this type of membrane effect in the interaction of LT with LPS. Data from the DSC and FT-IR experiments in this work confirmed that LT binds to LPS via electrostatic interactions with OC sugar and phosphate moieties. We

135

suspect that the more distant a protein’s binding site is to the lipid A portion of the LPS molecule, the lesser the effect this binding will have on the biophysical properties of the membrane.

Examination of the effects of LT expression on in vitro OMV production indicated that both the amount of extracellular LT and its ability to bind LPS are mechanistically important for OMV formation processes. Generally, the effect of protein binding on a membrane depends on the shape of the protein, the density of the protein binding to the membrane, and the composition of the membrane. In eukaryotic model membranes, peripheral membrane proteins generally induce membrane curvature via one of three mechanisms: using their intrinsically curved structures to scaffold the membrane (e.g.

Bin/amphiphysin/Rsv161 domain proteins), inserting amphipathic segments into a membrane leaflet, or creating steric pressure from the crowding of membrane-bound proteins (Rossy et al., 2014).

However, the composition and character of the Gram-negative OM differs greatly from eukaryotic plasma membranes. The OM is asymmetric in its lipid content, displaying phospholipids on its inner leaflet and lipopolysaccharide on its outer face.

Rates of LPS diffusion are low (Schindler et al., 1980); LPS molecules are knit together through divalent cation bridges between the negatively charged phosphate groups which flank the lipid A head groups and their fully saturated hydrocarbon chains allow for tight molecular packing (Nikaido, 2003).

136

For LT binding the ETEC OM, the clustering of the type II secretion system apparatus (Buddelmeijer et al., 2009) could lead to the creation of high local concentrations of membrane-bound toxin. Furthermore, the multivalency of LTB and low-diffusion rate of LPS could contribute the development of protein-protein crowding interactions by spatially confining areas of high protein coverage (Derganc and Copic,

2016).

The mechanics and regulation of OMV production are topics of great interest to the study of bacterial pathogenesis as OMVs carry virulence determinants and immunomodulatory molecules that are important to infection. OMV production has been shown to depend on periplasmic “pressure” (Schwechheimer and Kuehn, 2013), modulation of Lpp-PG crosslinks (Schwechheimer et al., 2015), as well as small molecule interaction with LPS (Mashburn-Warren et al., 2008). The data shown in this work reveal another mechanism that can regulate production of vesicles from the bacterial outer membrane: OMV production can be governed via biophysical influences from an external proteinaceous stimulus.

137

4. Concluding remarks

Gram-negative bacteria maintain the barrier properties of the outer membrane

(OM) in a wide array of physiological conditions despite their inability to degrade lipopolysaccharide (LPS) and protein material present in the outer leaflet of the OM.

Through characterization of the native dynamics of outer membrane LPS change we recently described a mechanism in which these diderm organisms overcome this design flaw. In response to different environmental stimuli Salmonella enterica modulates the export of specific structural variants of lipid A via outer membrane vesicles (OMVs). We proposed that the polymorphic model for regulation of membrane lipid content could largely account for the structural differences between secreted and retained lipid A species. However, differences in OMV production levels and size observed between environmental conditions remain unexplained. Further exploration into the relationship between OMV production level and content specificity may shed light onto the enigmatic mechanisms of OMV formation.

The lipopolysaccharide content of the asymmetric Gram-negative outer membrane is highly heterogeneous. Each molecule may differ in the length of its O- antigen chain, modifications to its core region, covalent additions to the lipid A head group and/or differential acylation of the lipid A anchor. Proper balance of the bulk properties of this mixture of LPS molecules allows the membrane to retain its barrier functionality.

138

During logarithmic growth in the neutral pH (7.6H, pH 7.6 10 mM MgSO4) condition, we observed that the lipid A composition of the secreted OMVs directly paralleled the OM composition over time. The neutral pH and high levels of magnesium cations in the 7.6H-to-7.6H environment allow for the 1-diphosphate and 1,4’- bisphosphate hexacylated lipid A species to create a highly-crosslinked stable membrane structure. The OMVs produced in this condition were smaller and more protein-dense on average than OMVs produced in acidic media.

Two-component systems PmrA/B and PhoP/Q upregulate specific envelope modifications upon sensing acidic pH, toxic metals, cationic antimicrobial peptides, and low divalent cation concentrations. Consequently, lipid A structures modified with phosphoethanolamine (pEtN) and aminoarabinose (L-Ara4N) at the 1 and/or 4’ phosphates were added to the OM during cell growth in acidic media (5.8L, pH 5.8 10

µM MgSO4). These modifications neutralize the negatively charged phosphates which flank the lipid A head group, allowing the molecules to compensate for the loss of the divalent cation bridges. Additionally, the PhoP/Q-activated OM enzyme PagP was activated in the 5.8L conditions, resulting in palmitoylation of a variety of lipid A species in order to maintain the membrane barrier. In general, covalent modification of the lipid A head groups with pEtN and L-Ara4N decreased the likelihood of that molecule’s incorporation into OMVs. In contrast, hepta-acylated species created through palmitoylation were more likely to be found in secreted OMVs. Concomitantly, OMV

139

production levels and size increased in the 7.6H-to-5.8L, 5.8L-to-5.8L, and 5.8L-to-7.6H conditions.

These observations are in agreement with the well-established model of polymorphic regulation of lipid composition. In this model, a regulated balance between bilayer-promoting and non-bilayer-promoting lipids is maintained throughout different environmental conditions. The geometry of each lipid molecule contributes to its intrinsic preference for lamellar or non-lamellar phases. Lipids with a more conical shape prefer non-lamellar phases (hexagonal, HII) and aid in the transition to nonbilayer structures, as would occur during OMV formation. As increased preference for the HII phase occurs when hydrocarbon chain length or splay increases and decreases with effective head group size increase, the higher levels of palmitoylated lipid A species and lower levels of covalently modified species we observe in OMVs produced in 5.8L conditions fit well with the polymorphic regulation model (Figure 27).

140

Figure 27. Model illustrating the diversity of lipid A structures and their preference for non-lamellar phase to lamellar phase in acidic conditions.

Lipid A molecules are shown as three-dimensional space-filling models in typical Corey- Pauling-Koltun coloring (with the majority of hydrogen atoms omitted for simplicity). The proposed relative propensity of each lipid A structure towards the hexagonal phase (top, represented by a blue cone) or towards the lamellar phase (bottom, represented by an orange cylinder) is shown here. From top to bottom these lipid A structures are shown from most to least likely to be found secreted in OMVs during 5.8L conditions according to ESI-MS data.

141

To probe the relationship of non-lamellar-promoting lipids with OMV production levels, we quantified the OMV production levels of phoPc (a mutant in which phoP is constitutively active) and phoPcpagP- (pagP deletion in the phoPc background) strains (generously provided by SI Miller). We find that the constitutive expression of the phoP operon results in higher levels of OMV production in both the 7.6H and 5.8L conditions (Figure 28). Deletion of pagP in the phoPc background lowered this production back to near WT levels in both environmental conditions. As the PagP-catalyzed palmitoylation of lipid A may serve to stabilize non-lamellar phases, its responsibility for higher production levels in the phoPc strain adds further support for the polymorphic regulation model.

142

Figure 28. Constitutive activation of phoP increases OMV production in 7.6H and 5.8L conditions.

The relative OMV production, normalized to CFU, during overnight growth of WT (14028s), phoPc (14028s pho-24), and phoPcpagP- (14028s pho-24 pagP::TnPhoA) strains in 7.6H and 5.8L media, as calculated for protein content by OMP densitometry. For all conditions, n >3; *, statistically significant difference for the indicated pair, p < 0.005.

143

However, it is important to note that PagP is a bifunctional enzyme; PagP catalyzes the both the palmitoylation of lipid A and phosphatidylglycerol (PG) in the

OM (Dalebroux et al., 2014). While the lysophospholipid byproduct of the PagP enzymatic reaction is swiftly transported to the IM to be reacylated (Hsu et al., 1989), palmitoylation of PG produces acyl-phosphatidylglycerol (acyl-PG) at the expense of

OM PG concentrations. The creation of acyl-PG thus depletes the OM of a key lamellar- promoting phospholipid (Bishop, 2014). Accordingly, enzymatic action of PagP may tip the balance of the curvature-preference in the OM. With the OM inner leaflet more prone to adopt negative curvature due to PG depletion, the formation of OMVs may be intrinsically promoted.

Furthermore, changes to the OM lipid environment in both leaflets, may influence the conformation of OMPs and the ease of which proteins fold or fit into the membrane. As cells prefer different suites of OMPs in different environmental conditions, environmentally-triggered changes to the permeability of an OMP could favor its removal from the OM. In order to incentivize this result, bacterial cells could build a fail-safe into their system by relying on lipid changes to execute the necessary protein removal. If shifting lipid content results in non-ideal folding conditions for certain OMPs, OMVs could provide a convenient means of disposal for these unfolded proteins. Additionally, if the new lipid environment results in a mismatch between the hydrophobic regions of proteins and the hydrophobic membrane space, then strain from

144

the membrane bending in order to accommodate the inappropriate protein will tip the scales towards membrane bending and vesicle formation.

We found that extended exposure to mildly acidic, low-magnesium conditions correlated with increased average diameter of secreted OMVs. One membrane feature which may influence OMV size may be the steric factors relating to lipid A head group size and propensity for intermolecular crosslinking. If differently modified structures contribute differently to the membrane bending modulus, then these intrinsic structural properties could lead to a preferred OMV size distribution per characteristic lipid mix.

We also considered that reduction of the level of OM-peptidoglycan (PG) connectivity might lead cells to shed larger vesicles. Although we did not find significant differences in the levels of the lipoprotein Lpp covalently attached to PG between the 7.6H-to-7.6H, 7.6H-to-5.8L, and 5.8L-to-5.8L conditions, changes in other mechanisms of attachment between the OM and PG layer may still occur or develop after further exposure to acidic conditions. Environmentally-regulated alterations to peptidoglycan metabolism and changes in OMP composition, especially in OmpA, could affect the level of OM-PG crosslinking.

In the phospholipid membranes, presence of nonbilayer lipids can influence peripheral and transmembrane protein function. Accordingly, the increased levels of

OM cardiolipin observed under PhoP/Q-activating conditions could play additional role in OMV formation through the modulation of protein functionality. The activation or

145

inhibition of protein activities in the inner leaflet of the OM could affect the membrane’s connectivity to the PG layer, remodeling of the PG sacculus, and other periplasmic processes.

OMV production levels could be influenced by these same factors, along with the afore-mentioned degree of OM attachment to the PG layer. Effects of the environment on periplasmic-localized processes could also influence OMV production levels in as-of- yet unknown ways.

Content selection in OMV formation is likely principally influenced by innate bilayer-forming propensity of areas of membrane. Therefore the processes involved in

OM biogenesis would play a major role in dictating the potential of certain areas of membrane to bleb. Content selection could also be influenced by the interactions between individual lipid molecules with each other and the lateral surface of OMPs.

However, since we do not observe any large differences between the protein composition of the OM and OMV during any of our environmental shifts utilizing SDS-

PAGE, we cannot yet discern if these smaller intermolecular interactions to play a significant role in content selection.

While many questions about OMV formation and the relationship between production levels and content selection remain, further probing into well-characterized systems may yield insights in the future. Additionally, the disparity in LPS composition

146

between the two membrane structures could have interesting implications for the processes of host immune activation and pathogen immune evasion.

147

5. Materials and methods

Portions of the following methods have been published previously by the author in the journal mBio (Bonnington and Kuehn, 2016).

5.1 Bacterial strains and media

The strain used in Chapter 2 of this work is Salmonella typhimurium 14028s

(virulent wild type). Bacteria were grown at 37°C in N minimal media (5 mM KCl, 7.5 mM (NH4)2SO4, 0.5 mM K2SO4, 1 mM KH2PO4, 0.1 M Tris-HCl, 0.1% Casamino acids

(Nelson and Kennedy, 1971) with 0.4% glucose (Gibbons et al., 2005), adjusted to either pH 5.8 or pH 7.6 with HCl, with or without 10 mM MgSO4 as indicated (Hmiel et al.,

1986).

The strains and plasmids used in Chapter 3 of this work are listed in Tables 3 and

4 below. Bacteria were grown at 37°C in CFA medium (1% casamino acids, 0.15% yeast extract, 0.005% MgSO4, 0.005% MnCl2) with the addition of antibiotics as appropriate.

Table 3. Plasmids used in Chapter 3

Plasmid Description Source or reference pILT IPTG-inducible LT; Ampr (Mudrak et al., 2009) pILTB IPTG-inducible LT-B (Mudrak and Kuehn, subunit; Ampr 2010b) pILTB Y18H IPTG-inducible LT B- This work subunit with Y18H point mutation; Ampr

148

Table 4. Strains used in Chapter 3

Strain or Description Relevant characteristics Source or plasmid reference ATCC H10407 wild type ETEC strain ATCC 35401 jf570 H10407 ΔeltA ETEC strain with polar (Johnson et insertion in eltA (LT al., 2009) deficient) MK1053 H10407 ΔeltA pTrc99a jf570 with empty plasmid; (Mudrak Ampr and Kuehn, 2010b) MK1052 H10407 ΔeltA pILT jf570 with inducible LT (Mudrak et plasmid; Ampr al., 2009) MK1200 H10407 ΔeltA pILT[T47A] jf570 with inducible LT (Mudrak et T47A plasmid; Ampr al., 2009) MT13 H10407 ΔgspD ETEC strain with polar (Tauschek et insertion in gspD; Kanr al., 2002) MK741 DH5α degP::Tn5/pDsbA/pILT E. coli K-12, degP (McBroom knockout, carrying a et al., 2006) plasmid copy of dsbA and an inducible LT plasmid; Kanr Cmr Ampr MK1207 DH5α As described for MK741, (Mudrak et degP::Tn5/pDsbA/pILT[T47A] with inducible T47A; Kanr al., 2009) Cmr Ampr MK1284 DH5α As described for MK741, This work degP::Tn5/pDsbA/pILTB[Y18H] with inducible Y18H; Kanr Cmr Ampr CWG311 F470 waaV::aaC1 R1 core, lacks O antigen; (Heinrichs Gmr et al., 1998a) CWG309 F470 waaT::aaC1 R1 core, two terminal Gal (Heinrichs residues removed; Gmr et al., 1998a) CWG303 F470 waaG::aaC1 R1 core, entire outer core (Heinrichs removed; Gmr et al., 1998a) CWG296 F470 waaP::aaC1 R1 core, phosphate on (Yethon et heptose I and II are al., 1998) absent, terminal gluocose uncertain; Gmr CWG297 F470 waaQ::aaC1 R1 core, phosphate on (Yethon et 149

heptose II absent, heptose al., 1998) III removed; Gmr CWG312 F470 waaY::aaC1 R1 core, phosphate on (Yethon et heptose II is absent ; Gmr al., 1998)

5.2 Environmental shift protocol

N-minimal medium (either pH 7.6 10 mM MgSO4; or pH 5.8 10 µM MgSO4, as indicated) was inoculated (1:100 dilution) from bacterial cultures grown overnight

(37°C, shaking 200 rpm). Cultures were grown to OD600 0.6 and pelleted with a Beckman

Avanti J-25 centrifuge (JLA-10.500 rotor; 3000 x g, 7 min, 25°C) and the resulting supernatant decanted. Cells were thoroughly resuspended in fresh N-minimal media

(either pH 7.6 10 mM MgSO4 or pH 5.8 10 µM MgSO4, as indicated) and grown at 37°C for 90 minutes.

5.3 Polymyxin B sensitivity assay

Cells exposed to either 0, 45, or 90 minutes of an environmental shift protocol

(7.6H-to-5.8L) were lightly centrifuged and resuspended in fresh N-minimal media

(7.6H) containing either 0, 4, or 8 µg/mL of the cationic antimicrobial peptide polymyxin

B (PMB). The OD600 of these cultures were recorded after 0, 2, and 4 hours of treatment.

This data was reported as percent survival relative to the untreated culture of each length of environmental shift.

While high concentrations of PMB are capable of cytotoxic effects in as little as 30 minutes, low concentrations of polymyxin B, such as those used in our experiment, lead

150

to increased permeability of the OM and eventual cell death without depolarization of the IM. This mechanism of action takes longer to effect cell death yet more directly assesses the effects of PMB dependent upon association with the OM. Additionally, the presence of high levels of Mg2+ has been shown to temper/slow the cytotoxic effects of

PMB. As our experiments were carried out using a range of low PMB concentrations in the presence of 10 mM Mg2+, it is not surprising that we continued to observe cell death at 4 hours post-treatment. Additionally, as our experiments were carried out in 7.6H media and normalized to the untreated culture in order to obtain percent survival, we could observe effects of recovery.

5.4 Sytox green cell integrity assay

As in Schwechheimer et al (Schwechheimer et al., 2015), Sytox Green (Invitrogen) was used to assess membrane integrity. Cultures (1 L) were grown and 100 µL was placed in a white 96-well plate (in duplicate). To prepare heat-killed cells for a positive control, 5 mL of bacterial culture was pelleted (10,000 × g, 5 min, room temperature), resuspended in 1 mL sterile N-minimal media, and boiled for 3 min, followed by sterile filtering (0.45-µm Ultra-free spin column filters; Millipore); lysates were diluted 10- to

1,000-fold and 100 µL was placed in a 96-well plate (in duplicate). To each sample, 3 µM

Sytox Green was added and the mixture was incubated at room temperature in the dark for 10 min. Measurement of fluorescence (excitation: 500 nm, emission: 550 nm) was performed on a Molecular Devices SpectraMAX GeminiXS spectrometer. The average of

151

the duplicate sample measurements was multiplied by the dilution factor and divided by the OD600 of the original culture.

5.5 OMV purification

OMVs were isolated from broth cultures as follows. The 1 L cultures were pelleted with a Beckman Avanti J-25 centrifuge (JLA-10.500 rotor, 10000 x g, 10 min, 4°C) and the resulting supernatant was filtered through 0.45 µm cellulose membrane. The filtered supernatant was then subject to centrifugation (JLA 16.250 rotor, 38400 x g, 3 hrs,

4°C) before pellets were resuspended in either DPBSS (for SDS-PAGE or other analysis) or 50 mM HEPES, pH 7.4 (for lipid A purification) and frozen at -20°C until utilization.

For a large-scale OMV preparation, the cell-free supernatant was first concentrated using a tangential flow apparatus to reduce supernatant volume to 750 mL before centrifugation (JLA 16.250 rotor, 38400 x g, 3 hrs, 4°C).

5.5.1 OMV production quantification

To quantitate OMV yield, OMV preparations were boiled for 10 min in Laemmli buffer, separated by 15% SDS-PAGE, and stained with SYPRO Ruby Red (Molecular

Probes) overnight. The gel was fixed for 1 hr in a solution of 10% methanol and 7% acetic acid prior to and post-staining. Ruby-stained proteins were visualized under UV light. OMPs F/C and A were quantified via densitometry (NIH Image J software)

(Chutkan et al., 2013). A second portion of the OMV preparations was incubated with

FM4-64 (Molecular Probes) (3.3 µg/mL in phosphate-buffered saline for 10 min at 37°C).

152

OMVs alone and FM4-64 probe alone were negative controls. After excitation at 506 nm, the emission at 750 nm was measured with a Molecular Devices SpectraMAX GeminiXS fluorometer (McBroom et al., 2006). OMV production was normalized by dividing by the CFU/mL for each culture, and these values were further divided by the OMV production of the control condition to give rise to relative fold OMV production.

5.5.2 Opti prep density gradient fractionation of OMVs

A step gradient made from OptiPrep 60% stock solution, diluted in OptiPrep diluent buffer was layered into a 15 mL tube (Chutkan et al., 2013). The series of volumes and Opti concentrations (final % Optiprep, v:v) of the gradient steps, from bottom to top are as follows: 2 mL of sample in 45%, 1 mL of 40%, 1 mL of 35%, 3 mL of

30%, 3 mL of 25%, and filled to the top with 20%. This was centrifuged (Beckman

Optima LE-80K, SW 41 Ti rotor) for 18 hr, 41000xg at 5°C. Fractions of equal volumes were collected by pipetting from the top (Fraction 1= lowest density, Fraction 12=highest density).

5.6 SDS-PAGE and immunoblotting

5.6.1 Protein staining

Protein samples were run on 15% or 12% polyacrylamide gels before staining with SYPRO Ruby Redstain (Molecular Probes) overnight, according to the manufacturer’s instructions. Ruby fix/wash solution (10% methanol, 7% acetic acid [v/v])

153

was used both to fix the gel prior to staining and wash the gel post-staining. Ruby- stained proteins were visualized under UV light.

5.6.2 Silver staining

Gels were stained using the BioRad Silver Stain Plus Kit. After staining, the gels were imaged on the Azure c300 (Azure Biosystems).

5.6.3 Immunoblotting

Protein samples were run on 15% or 12% polyacrylamide gels. Proteins were transferred to nitrocellulose membranes, which had been soaked in transfer buffer at approximately 10V for 30 minutes. Membranes were blocked with Blotto (Tris-buffered saline with 0.1% Tween-20 [v/v], 30% skim milk [w/v], and 0.01% NaN3) for at least 1 hour at room temperature on an orbital shaker before incubation with primary antibody diluted in Blotto overnight at 4°C. Anti-CT antibody (Sigma) was used at 1:15,000. The next day membranes were washed thrice with TBS-T (Tris-buffered saline with 0.1%

Tween-20) for 15 minutes each and then incubated with goat anti-rabbit fluorescently conjugated secondary antibody (Odyssey), diluted 1:20,000 in TBS-T, for 1 hour at room temperature. Following three 10 minute washes with TBS-T, the membranes were washed once with TBS for 10 minutes before imaging on the Odyssey imager.

154

5.7 Cell fractionation

5.7.1 Sucrose density gradient fractionation of IM/OM

The cell pellet from 2L of culture (grown in pH 5.8, 10 µM MgSO4 N-minimal media) was resuspended in PBS + lysozyme (1 mg/ mL) before lysis using a French

Pressure cell. Cell debris was centrifuged out of the sample (10000xg, 10 min) at 25°C before the lysate was separated by Opti prep density gradient fractionation.

Pure OM fractions were isolated via the method of Dalebroux et al (Dalebroux et al., 2014). For proteomic analysis, a modified method utilized an Opti gradient instead of the classic sucrose gradient. A step gradient made from OptiPrep 60% stock solution, diluted in OptiPrep diluent buffer was layered into a 15 mL tube. The series of volumes and Opti concentrations (final % Optiprep, v:v) of the gradient steps, from bottom to top are as follows: 2 mL of sample in 45%, 2 mL of 40%, 2 mL of 35%, 4 mL of 30%, 1 mL of

25%, and filled to the top with 20%. This was centrifuged (Beckman Optima LE-80K,

SW 41 Ti rotor) for 18 hr, 41000xg at 5°C. Fractions of equal volumes (1 mL) were collected by pipetting from the top (Fraction 1 = lowest density, 12+=highest). The fractions were diluted into PBS and centrifuged to remove Opti (Beckman Optima LE-

80K, SW 41 Ti rotor) for 2 hr, 41000xg at 5°C. The sample was then run on SDS-PAGE,

Ruby-stained for protein, and Emerald Green-stained for LPS. The peak OM fractions

(9-10) were pooled, washed, and resuspended in <100 uL of 50 mM ammonium bicarbonate.

155

5.7.2 Peptidoglycan isolation

Peptidoglycan was isolated as described before (Lam et al., 2009; Schwechheimer et al., 2015).

5.7.3 Braun’s lipoprotein quantification: WC versus PG-associated

Lysozyme-digested peptidoglycan samples were analyzed by western blotting, probed with anti-Lpp primary antibody (generously provided by the Silhavy laboratory)

(Cowles et al., 2011), and the amount of Lpp associated with the purified PG quantitated by densitometry.

5.8 Lipid A isolation

For OMV and whole cell preparations, salt removal was performed by washing three times with 50 mM HEPES pH 7.4. We performed a variant of the Zhou et al protocol (Zhou et al., 2001), the samples (1 mL) were lysed with 2:1 methanol:dichloromethane (3 mL) for 1 hr at room temperature. Centrifugation (2000 rpm, 30 min, 25°C) of the samples was followed by a wash with 2:1 methanol:dichloromethane, and another centrifugation step. The washed, dried (under

N2) pellet was boiled for 30 minutes in 12.5 mM sodium acetate (adjusted to pH 4.5 with acetic acid). Lipid A was then isolated via a two-phase Bligh/Dyer separation and subsequent washes with fresh neutral upper phase before final drying under N2.

5.9 Glycerophospholipid isolation

Glycerophospholipids were isolated via standard Bligh & Dyer protocol.

156

5.10 Electrospray ionization-mass spectrometry (ESI-MS) of lipid A samples

Mass spectra were acquired on a QSTAR-XL quadrupole time-of-flight tandem mass spectrometer (ABI/MDS-Sciex, Foster City, CA) equipped with an ESI source. The spectra were acquired in the negative-ion mode and largely were the accumulation of 60 scans over the range of 500–2500 atomic mass units (Raetz et al., 2006). For MS analysis, the extracted lipids were dissolved in 50 µL of dichloromethane/methanol (2:1, v/v) and infused into the ion source at 5–10 µL/min. The negative-ion ESI was performed at

−4200V. Data acquisition and analysis were carried out using AnalystQS software and displayed using mMass tools (Strohalm et al., 2008).

5.11 Thin-layer chromatography (TLC)

Lipid A samples were spotted on glass-backed silica gel 60 thin-layer chromatography plates and allowed to dry before development in a solvent system consisting of 25:15:4:4 (chloroform: methanol: acetic acid: water).

5.11.1 TLC densitometry

The lanes of each sample run on a TLC plate were set to a common width (.29) across all plates. After selection of the lanes, the intensity of the bands was plotted by the

Gel Analyzer function, with peaks inverted. The area of each peak was defined by a straight line at the base, stretching from the lowest point of the surrounding troughs, before measurement with the Wand (tracing) tool.

157

As loading levels across a TLC plate can be variable, we used an internal standard, detailed in Chapter 2. We calculated the intensity of each band (A-H) via densitometry. Further detail on how the beginning and end of each band were objectively selected can be seen in Figure 12. In Figure 12, the Gel Analyzer function is performed on each lane, giving rise to peaks (upon inversion). The line drawn to connect the troughs surrounding each peak delineates the area to be measured inside the peak.

Once collected, the peak areas were transformed into ratios, all in reference to band A.

Band A was chosen as the denominator as it is present in all samples and is not complicated by the arrival of new species in the same area of the plate.

5.11.2 2D TLC of glycerophospholipid samples

Purified phospholipids were separated via the method of Dalebroux et al

(Dalebroux et al., 2014).

5.11.3 TLC staining methods

A charring method was utilized to visualize all organic material on the TLC plate. Once dried, TLC plates were sprayed with 10% H2SO4 in ethanol for detection of all species after careful, even heating at 300°C. This method irreversibly degrades the samples. During charring, different lipid A species may char at slightly different rates due to their different chemical makeup. Across a single plate, the charring may be inconsistent if the spraying of the plate is not conducted laboriously, a lower-quality hot plate (such as one which also can stir, or does not measure temperature on its calibration

158

knob) is used and/or the TLC plate is not shifted continuously (sautéed) during charring and halted prior to the development of substantial background charring. Using careful technique, even lower dilutions of a sample can give densitometry reads comparable to lanes which are loaded higher or charred darker as long as each peak is readily distinguishable and the intensity of band A is above threshold.

We cannot confirm our analysis via other standard lipid stains. We could not use

Bial’s reagent (orcinol), as this stain detects pentoses (not present in lipid A). We could not use iodine, as the lipid A acyl chains are fully saturated, making it unreactive to the stain.

We could confirm the charring using malachite green staining (stains for phosphate, thus all bands are stained), but we found the lesser contrast between the green bands on the inconsistently orange background to complicate analysis by densitometry by dampening the height of each peak.

In order to visualize localization of primary amines, ninhydrin staining was used. Staining solution was prepared by dissolving 100 mg ninhydrin in 100 mL water- saturated n-butanol and 3 mL acetic acid. Pink spots were visualized after careful, even heating on a 150°C hotplate.

159

Figure 29. A comparison of charring with malachite green staining from whole-cell lipid A isolated from 5.8L conditions.

The same 5.8L whole cell lipid A sample is shown visualized by malachite green (top right) and by charring (top left) with corresponding densitometry analysis. Comparison of the area under each labeled peak, relative to peak A, is shown below.

160

5.12 Transmission electron microscopy

Bacterial cells or OMVs were washed and reconstituted in 50 mM HEPES pH 7.5.

Cells were fixed with 4% formaldehyde. Samples were placed on 300 mesh copper grids

(Electron Microscopy Sciences) before staining with 1% uranyl acetate (Manning and

Kuehn, 2011).

The OMVs and cells observed on the micrographs were measured from the outermost points of the membrane using the measurement function in the NIH ImageJ software.

5.13 Quantitative mass spectrometry

All proteomic analyses were performed by the Duke Proteomics Core, and the following methods were written by Matthew W. Foster.

OM and OMV samples in PBS buffer were solubilized by probe sonication in the presence of 0.2% (v/v) acid labile surfactant (ALS-1). After centrifugation, solubilized proteins were quantified in supernatants by Bradford assay. 1.5 micrograms of each sample were normalized to equal volumes with 50 mM ammonium bicarbonate (AmBic) containing 0.2% ALS-1, DTT was added to a final concertation of 10 mM, and samples were reduced and denatured by heating at 80 ºC for 10 min. Next, samples were alkylated with 20 mM iodoacetamide in the dark for 30 min. Sequencing grade modified trypsin (Promega; 1:50 w/w trypsin:protein) was added, and proteins were digested at

37 ºC overnight. Samples were then acidified by addition of a final concentration of 1%

161

(v/v) trifluoroacetic acid (TFA) and 2% (v/v) acetonitrile (MeCN), followed by heating at

60 ºC for 2 h, to inactivate trypsin and degrade ALS-1. After centrifugation, supernatants were lyophilized to dryness. Following reconstitution, peptides were purified using C18

StateTips and eluents were lyophilized. Finally, reconstituted peptides were resuspended in 10 µl of 1% TFA/2% MeCN and transferred to Maximum Recovery LC vials (Waters). QC pools were prepared by mixing equal volumes of all samples.

Quantitative one-dimensional liquid chromatography, tandem mass spectrometry (1D-LC-MS/MS) was performed on 250 ng of the peptide digests per sample in singlicate, with additional analyses of conditioning runs and QC pools, as described in Table S1. Samples were analyzed using a nanoACQUITY UPLC system

(Waters) coupled to a QExactive Plus high resolution accurate mass tandem mass spectrometer (Thermo) via a nanoelectrospray ionization source. Briefly the sample was first trapped on a Symmetry C18 180 µm × 20 mm trapping column (5 µl/min at 99.9/0.1 v/v H2O/MeCN) followed by an analytical separation using a 1.7 µm Acquity HSS T3

C18 75 µm × 250 mm column (Waters) with a 90 min gradient of 5 to 40% MeCN with

0.1% formic acid at a flow rate of 400 nl/min and column temperature of 55 °C. Data collection on the QExactive Plus MS was performed in data-dependent acquisition

(DDA) mode with a 70,000 resolution (@ m/z 200) full MS scan from m/z 375 to 1600 with a target AGC value of 1e6 ions followed by 10 MS/MS scans at 17,500 resolution (@ m/z 200) at a target AGC value of 5e4 ions. A 20 s dynamic exclusion was employed. The

162

total analysis cycle time for each sample injection was approximately 2 h. Following 10 total UPLC-MS/MS analyses (including 1 conditioning runs and 3 replicate QC injections; Table S1), data was imported into Rosetta Elucidator v.4 (Rosetta Biosoftware,

Inc.), and analyses were separately aligned based on the accurate mass and retention time of detected ions (“features”) using PeakTeller algorithm in Elucidator. Relative peptide abundance was calculated based on area-under-the-curve (AUC) of the selected ion chromatograms of the aligned features across all runs. The MS/MS data was searched against a custom Swissprot/Trembl database with Salmonella typhimurium taxonomy (downloaded on 05/31/16), as well as an equal number of reversed-sequence

“decoys” for false discovery rate determination (10,752 total entries). Mascot Distiller and Mascot Server (v 2.5, Matrix Sciences) were utilized to produce fragment ion spectra and to perform the database searches. Database search parameters included fixed modification on Cys (carbamidomethyl) and variable modifications on Asn and Gln

(deamidation). An additional database search was performed using semitrypsin specificity on unmatched MS/MS spectra. After individual peptide scoring using the

PeptideProphet algorithm in Elucidator, the data was annotated at a 1.0% peptide false discovery rate. For quantitative processing, the data was first curated to contain only high quality peptides with appropriate chromatographic peak shape and the dataset was intensity scaled to the robust mean.

163

5.14 Lipopolysaccharide purification

LPS from was purified from CWG311, CWG309, CWG303, and CWG308 via the method of Darveau and Hancock (Darveau and Hancock, 1983). The lipid samples were suspended directly in buffer to create aqueous dispersions at high buffer content, i.e. above 80% using 20 mM Tris-HCl, before temperature cycling between 4 and 70°C in 100 uL aliquots in the Multigene thermocycler (Labnet International, Inc.) to achieve full hydration. LPS was incubated at 4°C for at least 12 hrs prior to analysis. Concentration of LPS was determined by the periodate-thiobarbituric acid assay for 3-deoxy-D-manno-

2-octulosonic acid (KDO) concentration (Waravdekar and Saslaw, 1957).

5.15 Toxin purification

Toxin was prepared from MK741 or a derivative thereof expressing mutant toxin as described previously (Mudrak et al., 2009). Cells were diluted 1:100 in CFA medium from an overnight culture and grown to an optical density at 600 nm (OD600) of 0.6 before induction with 200 µM IPTG for 5 hrs. Purification using galactose resin (Sigma) followed by gel-filtration chromatography (GE S200 column). The concentration of LT was quantified via the Bradford method with bovine serum albumin as the standard.

5.16 LPS Sedimentation/LT depletion binding assay

LPS was water-bath sonicated at the indicated concentration in Tris-HCl pH 8.0.

WT LTB was co-incubated for 10 min with the LPS, and then centrifuged at 410000 x g for

60 min at 20 °C. The supernatants were collected and a portion (20 µl) applied to an

164

SDS-PAGE gel. The 11 kDa LTB band was detected by Sypro Ruby stain, and the protein bands quantitated by densitometry (NIH ImageJ software) and background was subtracted before normalization to samples with the highest level of LTB depletion per

LPS type. Kd was calculated by non-linear curve fitting (Michaelis Menten function) using Origin software (OriginLab Corp).

5.17 Isothermal titration calorimetry

LPS and LTB samples were dialyzed in a buffer containing 50 mM Tris-HCl, pH

7.6, 200 mM NaCl, 1 mM EDTA, 3 mM NaN3, 1% triethylamine (v/v). 178 µM of LTB was titrated into 17 µM ΔwaaV LPS. This titration was performed in triplicate (technical replicates) at 25°C using a MicroiTC200 (GE Healthcare). The total heat released during each injection was fit to a single-site binding isotherm with Kd and ΔH set as independent parameters. Data were analyzed and figures generated using Origin software (OriginLab Corp).

5.18 Differential scanning calorimetry

A stock solution of ΔwaaV or ΔwaaT LPS (fully hydrated) was dispersed in TN buffer (20 mM Tris-Hcl pH 8.0, 100 mM NaCl) with or without 100 µM MgSO4. DSC measurements were performed on a Nano-DSC (TA Instruments, Duke MRSEC) at a heating/cooling rate of 1 °C/min. Heating curves were measured in the temperature interval from 5 to 60°C or 5 to 100°C. Due to the dissociation of the LTB pentamers above

165

50°C, only the first heating scan could be utilized for analyzing pentamer-LPS mixtures.

Data was analyzed and figures were generated in NanoAnalyze (TA Instruments).

5.19 Fourier-transform infrared spectroscopy

The IR measurements were performed on a PerkinElmer Frontier FTIR with a

Pike Technologies MIRacle ATR (attenuated total reflectance) KSR5 coated diamond optic and a narrow band MCT detector. Measurements on buffer exchanged LPS and protein were performed by taking the background scans with the dialysis buffer on the

-1 crystal. Typically, 32 scans and 4 cm resolution using approx. 5 µL of 50 to 70 µM protein sample and/or 250, 400, or 750 µM LPS sample was used; and measurements

were performed at ambient temperature. CO2 and water vapor was subtracted from the spectrum using the provided automatic background removal function. Measurements were performed at the intrinsic instrument temperature at a room temperature of 23°C.

For overlapping bands in the analysis of the amide-I vibration, curve fitting was applied using GRAMS/AI (Thermo Fisher Scientific). An estimated number and positioning of band components was obtained from analysis of the second derivative of the original spectra. The band shapes on the single components are superpositions of

Gaussian and Lorentzian. Best fits were obtained by assuming a Gauss fraction of 0.55-

0.60.

166

Appendix 1

CDT: cytolethal distending toxin

CFA media: colonization factor antigen media

CL: cardiolipin

CT: cholera toxin

DSC: differential scanning calorimetry

ETEC: Enterotoxigenic E. coli

FT-IR: Fourier-transform infrared spectroscopy

GPL: glycerophospholipid

IM: inner membrane

Kdo: 3-deoxy-D-manno-oct-2-ulosonic acid

LB: lysogeny borth

Lol: system for lipoprotein transport

Lpp: Braun’s lipoprotein

LPS: lipopolysaccharide

Lpt: Lipopolysaccharide transport

LT: heat-labile enterotoxin

MDO: membrane-derived oligosaccharide

OC: oligosaccharide core

OM: outer membrane

167

OMV: outer membrane vesicle

Pal: peptidoglycan-associated lipoprotein

PE: phosphatidyl ethanolamine

PG: phosphatidyl glycerol

PL: phospholipids

SCV: Salmonella-containing vacuole

ST: Shiga toxin

T2SS: type II secretion system

TCS: two-component system

WC: whole-cell

168

Appendix 2

The comparative proteomic profiles of the OM and OMVs produced under acidic, low-divalent cation conditions in a defined, minimal media reveals slight differences in the incorporation of a suite of OM-localized proteins into OMVs. Close analysis of the differential secretion of OM- and periplasmic-localized protein components may provide insight into the complex process of OMV formation during specific environmental conditions.

250 ng of peptide digests from solubilized membranes were analyzed by ultraperformance liquid chromatography, tandem mass spectrometry (LC-MS/MS). A quality control (QC) pool containing an equal mixture of samples was analyzed at the beginning, middle, and end of the study. Individual samples were run in a random order (Table S1). Next, data was imported into Rosetta Elucidator followed by accurate mass and retention time alignment, and database searching and quantitation of area- under-the-curve of identified features. The dataset had 106,811 quantified features and

140,217 high collision energy (peptide fragment) spectra that were subjected to database searching (see Methods in Chapter 5).

Following database searching and peptide scoring using the PeptideProphet algorithm, the data was annotated at a 1% peptide false discovery rate, resulting in identification of 4,334 peptides and 937 proteins. Next the data to the robust mean across all samples. After these steps, expression values for total of 4,298 peptides and 934

169

proteins. Of these, 581 proteins were quantified by 2 or more peptides, a criterion for higher confidence of identification and quantification.

In order to assess technical reproducibility, we calculated the percent coefficient of variation (%CV) for each protein across the three injections of a QC pool that were interspersed throughout the study. To assess biological variability, %CVs were measured for each protein across the individual analyses. The mean %CV of the QC pools was 9.3% for all proteins and 5.9% for proteins quantified by two or more peptides. Biological variability was 99% for all proteins and 95% for proteins quantified by two or more peptides. Thus, the technical reproducibility was excellent and was much higher than the biological variability.

We also used two-dimensional hierarchical clustering to visualize the expression across the six biological samples (Fig. 30). By this method, OM and OMV samples clustered into separate branches, with the OM C and OMV 3 samples being the most different. Further analysis of these proteins revealed their subcellular localization to be either periplasmic or cytoplasm (Fig. 31, 32). Since the only rational basis of comparison for the OM and OMV samples is examination of the OM-localized proteins, we focused our analysis on those 50 proteins which were both quantified by 2 or more peptides and could be annotated with definitive OM-localization.

Examination of the OM-localized differences between the OM and OMV samples revealed few dramatic differences. Notably, the OM lipoprotein STM14_0095

170

experienced 64-fold enrichment in the OMV samples as compared to the OM. The TonB dependent ferric-rhodotorulic acid outer membrane transporter FhuE saw 36-fold enrichment. Modest OMV enrichment of approximately 10- to 15-fold was experienced by rare lipoprotein A (RlpA), putative outer membrane lipoprotein YcfM, TonB- dependent ferric enterobactin receptor IroN, outer membrane lipoprotein Lol machinery component LolB, and TonB-dependent ferrioxamine receptor FoxA.

Direct comparison of the relative abundance of the 50 OM-localized proteins in the OM samples as compared to the OMV samples revealed a linear trend (Fig. 33). The higher levels of OMV abundance overall can be observed from this comparison. This data may indicate that the OMV purification method (see Chapter 5) is more reliable than the OM purification method. Overall, the OM-localized 5.8L OM and 5.8L proteomes are very similar, supporting conclusions made in the Chapter 2 discussion.

171

Figure 30. Clustering of proteins that were differentially expressed between OM and OMV groups.

There were 91 proteins that had an FDR-corrected p < 0.05. There were also visualized by 2D hierarchical clustering. This visualization showed that a larger proportion of the differentially-expressed proteins were higher in the OMV versus the OM group.

172

Figure 31. Abundance of proteins identified in the 5.8L OM samples.

The inner pie chart is representative of the entirety of the OM proteome (identified by 2 or more peptides), divided into colored fractions based upon subcellular localization. The size of each fraction indicates the abundance of proteins annotated to each subcellular location: cell, cytoplasm, extracellular, inner membrane, outer membrane, periplasm, pilus, unknown, and unknown with a signal peptide. The outer sunburst represents the subset of OM-localized proteins in the 5.8L OM sample. The size of each piece corresponds to the abundance of the protein encoded by the labelled gene name.

173

Figure 32. Abundance of proteins identified in the 5.8L OMV samples.

The inner pie chart is representative of the entirety of the OMV proteome (identified by 2 or more peptides), divided into colored fractions based upon subcellular localization. The size of each fraction indicates the abundance of proteins annotated to each subcellular location: cell, cytoplasm, extracellular, inner membrane, outer membrane, periplasm, pilus, unknown, and unknown with a signal peptide. The outer sunburst represents the subset of OM-localized proteins in the 5.8L OMV sample. The size of each piece corresponds to the abundance of the protein encoded by the labelled gene name.

174

Figure 33. Relative abundance of OM-localized proteins in 5.8L-grown OM and OMV fractions.

Values for the abundance (normalized to the abundance of PhoE, a protein present in equal abundance in the OM and OMV fractions) of the OM-localized proteins displayed in Figure 31 are plotted against those for OMV-localized proteins shown in Figure 32.

175

References

(2005). World Health Organization Drug-Resistant Salmonella. In WHO website.

(2011). Microbiology by numbers. Nat Rev Microbiol 9, 628.

Aimon, S., Callan-Jones, A., Berthaud, A., Pinot, M., Toombes, G.E., and Bassereau, P. (2014). Membrane shape modulates transmembrane protein distribution. Dev Cell 28, 212-218.

Allen, K.P., Randolph, M.M., and Fleckenstein, J.M. (2006). Importance of Heat-Labile Enterotoxin in Colonization of the Adult Mouse Small Intestine by Human Enterotoxigenic Escherichia coli Strains. Infect Immun 74, 869-875.

Andersen, C. (2003). Channel-tunnels: outer membrane components of type I secretion systems and multidrug efflux pumps of Gram-negative bacteria. Rev Physiol Biochem Pharmacol 147, 122-165.

Arunmanee, W., Harris, J.R., and Lakey, J.H. (2014). Outer membrane protein F stabilised with minimal amphipol forms linear arrays and LPS-dependent 2D crystals. J Membr Biol 247, 949-956.

Bakht, O., Pathak, P., and London, E. (2007). Effect of the structure of lipids favoring disordered domain formation on the stability of cholesterol-containing ordered domains (lipid rafts): identification of multiple raft-stabilization mechanisms. Biophys J 93, 4307- 4318.

Balsalobre, C., Silvan, J.M., Berglund, S., Mizunoe, Y., Uhlin, B.E., and Wai, S.N. (2006). Release of the type I secreted alpha-haemolysin via outer membrane vesicles from Escherichia coli. Mol Microbiol 59, 99-112.

Barth, A. (2007). Infrared spectroscopy of proteins. Biochim Biophys Acta 1767, 1073- 1101.

Bauman, S.J., and Kuehn, M.J. (2006). Purification of outer membrane vesicles from Pseudomonas aeruginosa and their activation of an IL-8 response. Microbes Infect 8, 2400-2408.

Beddoe, T., Paton, A.W., Le Nours, J., Rossjohn, J., and Paton, J.C. (2010). Structure, biological functions and applications of the AB5 toxins. Trends Biochem Sci 35, 411-418.

Berberov, E.M., Zhou, Y., Francis, D.H., Scott, M.A., Kachman, S.D., and Moxley, R.A. (2004). Relative importance of heat-labile enterotoxin in the causation of severe diarrheal

176

disease in the gnotobiotic piglet model by a strain of enterotoxigenic Escherichia coli that produces multiple enterotoxins. Infect Immun 72, 3914-3924.

Beveridge, T.J. (1999). Structures of gram-negative cell walls and their derived membrane vesicles. J Bacteriol 181, 4725-4733.

Bishop, R.E. (2014). Emerging roles for anionic non-bilayer phospholipids in fortifying the outer membrane permeability barrier. J Bacteriol 196, 3209-3213.

Bishop, R.E., Gibbons, H.S., Guina, T., Trent, M.S., Miller, S.I., and Raetz, C.R. (2000). Transfer of palmitate from phospholipids to lipid A in outer membranes of gram- negative bacteria. Embo J 19, 5071-5080.

Boardman, B.K., Meehan, B.M., and Fullner Satchell, K.J. (2007). Growth phase regulation of Vibrio cholerae RTX toxin export. J Bacteriol 189, 1827-1835.

Bohin, J.P. (2000). Osmoregulated periplasmic glucans in Proteobacteria. FEMS Microbiol Lett 186, 11-19.

Bomberger, J.M., Maceachran, D.P., Coutermarsh, B.A., Ye, S., O'Toole, G.A., and Stanton, B.A. (2009). Long-distance delivery of bacterial virulence factors by Pseudomonas aeruginosa outer membrane vesicles. PLoS Pathog 5, e1000382.

Bonnington, K.E., and Kuehn, M.J. (2014). Protein selection and export via outer membrane vesicles. Biochim Biophys Acta 1843, 1612-1619.

Bonnington, K.E., and Kuehn, M.J. (2016). Outer Membrane Vesicle Production Facilitates LPS Remodeling and Outer Membrane Maintenance in Salmonella during Environmental Transitions. MBio 7.

Boquet, P., and Ricci, V. (2012). Intoxication strategy of Helicobacter pylori VacA toxin. Trends Microbiol 20, 165-174.

Brandenburg, K., Garidel, P., Andra, J., Jurgens, G., Muller, M., Blume, A., Koch, M.H., and Levin, J. (2003). Cross-linked hemoglobin converts endotoxically inactive pentaacyl endotoxins into a physiologically active conformation. J Biol Chem 278, 47660-47669.

Brandenburg, K., Koch, M.H., and Seydel, U. (1998). Biophysical characterisation of lysozyme binding to LPS Re and lipid A. Eur J Biochem 258, 686-695.

Brandenburg, K., Kusumoto, S., and Seydel, U. (1997). Conformational studies of synthetic lipid A analogues and partial structures by infrared spectroscopy. Biochim Biophys Acta 1329, 183-201.

177

Brandenburg, K., and Seydel, U. (1984). Physical Aspects of Structure and Function of Membranes Made from Lipopolysaccharides and Free Lipid-A. Biochimica Et Biophysica Acta 775, 225-238.

Buddelmeijer, N., Krehenbrink, M., Pecorari, F., and Pugsley, A.P. (2009). Type II secretion system secretin PulD localizes in clusters in the Escherichia coli outer membrane. J Bacteriol 191, 161-168.

Canadas, O., and Casals, C. (2013). Differential scanning calorimetry of protein-lipid interactions. Methods Mol Biol 974, 55-71.

Canadas, O., Garcia-Verdugo, I., Keough, K.M., and Casals, C. (2008). SP-A permeabilizes lipopolysaccharide membranes by forming protein aggregates that extract lipids from the membrane. Biophys J 95, 3287-3294.

Capra, E.J., and Laub, M.T. (2012). Evolution of two-component signal transduction systems. Annu Rev Microbiol 66, 325-347.

Casadevall, A., and Pirofski, L.A. (1999). Host-pathogen interactions: redefining the basic concepts of virulence and pathogenicity. Infect Immun 67, 3703-3713.

Chang, D.E., Smalley, D.J., Tucker, D.L., Leatham, M.P., Norris, W.E., Stevenson, S.J., Anderson, A.B., Grissom, J.E., Laux, D.C., Cohen, P.S., et al. (2004). Carbon nutrition of Escherichia coli in the mouse intestine. Proc Natl Acad Sci U S A 101, 7427-7432.

Chart, H., Frost, J.A., and Rowe, B. (1994). Expression of outer membrane proteins by Salmonella enteritidis relating to pH. FEMS Microbiol Lett 123, 311-314.

Chatterjee, S.N., and Das, J. (1967). Electron microscopic observations on the excretion of cell-wall material by Vibrio cholerae. J Gen Microbiol 49, 1-11.

Chen, H.D., and Groisman, E.A. (2013). The biology of the PmrA/PmrB two-component system: the major regulator of lipopolysaccharide modifications. Annu Rev Microbiol 67, 83-112.

Chutkan, H., and Kuehn, M.J. (2011). Context-dependent activation kinetics elicited by soluble versus outer membrane vesicle-associated heat-labile enterotoxin. Infect Immun 79, 3760-3769.

Chutkan, H., Macdonald, I., Manning, A., and Kuehn, M.J. (2013). Quantitative and qualitative preparations of bacterial outer membrane vesicles. Methods Mol Biol 966, 259-272.

178

Ciofu, O., Beveridge, T.J., Kadurugamuwa, J., Walther-Rasmussen, J., and Hoiby, N. (2000). Chromosomal beta-lactamase is packaged into membrane vesicles and secreted from Pseudomonas aeruginosa. J Antimicrob Chemother 45, 9-13.

Cloud-Hansen, K.A., Peterson, S.B., Stabb, E.V., Goldman, W.E., McFall-Ngai, M.J., and Handelsman, J. (2006). Breaching the great wall: peptidoglycan and microbial interactions. Nat Rev Microbiol 4, 710-716.

Coburn, B., Grassl, G.A., and Finlay, B.B. (2007). Salmonella, the host and disease: a brief review. Immunol Cell Biol 85, 112-118.

Copic, A., Latham, C.F., Horlbeck, M.A., D'Arcangelo, J.G., and Miller, E.A. (2012). ER cargo properties specify a requirement for COPII coat rigidity mediated by Sec13p. Science 335, 1359-1362.

Cornelis, G.R. (2006). The type III secretion injectisome. Nat Rev Microbiol 4, 811-825.

Costa, K., Bacher, G., Allmaier, G., Dominguez-Bello, M.G., Engstrand, L., Falk, P., de Pedro, M.A., and Garcia-del Portillo, F. (1999). The morphological transition of Helicobacter pylori cells from spiral to coccoid is preceded by a substantial modification of the cell wall. J Bacteriol 181, 3710-3715.

Cowles, C.E., Li, Y., Semmelhack, M.F., Cristea, I.M., and Silhavy, T.J. (2011). The free and bound forms of Lpp occupy distinct subcellular locations in Escherichia coli. Mol Microbiol 79, 1168-1181.

Crossman, L.C., Chaudhuri, R.R., Beatson, S.A., Wells, T.J., Desvaux, M., Cunningham, A.F., Petty, N.K., Mahon, V., Brinkley, C., Hobman, J.L., et al. (2010). A commensal gone bad: complete genome sequence of the prototypical enterotoxigenic Escherichia coli strain H10407. J Bacteriol 192, 5822-5831.

Crump, J.A., Medalla, F.M., Joyce, K.W., Krueger, A.L., Hoekstra, R.M., Whichard, J.M., Barzilay, E.J., and Emerging Infections Program, N.W.G. (2011). Antimicrobial resistance among invasive nontyphoidal Salmonella enterica isolates in the United States: National Antimicrobial Resistance Monitoring System, 1996 to 2007. Antimicrob Agents Chemother 55, 1148-1154.

Cullis, P.R., and de Kruijff, B. (1978). The polymorphic phase behaviour of phosphatidylethanolamines of natural and synthetic origin. A 31P NMR study. Biochim Biophys Acta 513, 31-42.

Czuprynski, C.J., and Welch, R.A. (1995). Biological effects of RTX toxins: the possible role of lipopolysaccharide. Trends Microbiol 3, 480-483.

179

Dalebroux, Z.D., Matamouros, S., Whittington, D., Bishop, R.E., and Miller, S.I. (2014). PhoPQ regulates acidic glycerophospholipid content of the Salmonella Typhimurium outer membrane. Proc Natl Acad Sci U S A 111, 1963-1968.

Dalebroux, Z.D., and Miller, S.I. (2014). Salmonellae PhoPQ regulation of the outer membrane to resist innate immunity. Curr Opin Microbiol 17, 106-113.

Darveau, R.P., and Hancock, R.E. (1983). Procedure for isolation of bacterial lipopolysaccharides from both smooth and rough Pseudomonas aeruginosa and Salmonella typhimurium strains. J Bacteriol 155, 831-838.

Daugelavicius, R., Bakiene, E., and Bamford, D.H. (2000). Stages of polymyxin B interaction with the Escherichia coli cell envelope. Antimicrob Agents Chemother 44, 2969-2978. de Cock, H., Brandenburg, K., Wiese, A., Holst, O., and Seydel, U. (1999). Non-lamellar structure and negative charges of lipopolysaccharides required for efficient folding of outer membrane protein PhoE of Escherichia coli. J Biol Chem 274, 5114-5119.

Derganc, J., and Copic, A. (2016). Membrane bending by protein crowding is affected by protein lateral confinement. Biochim Biophys Acta 1858, 1152-1159.

Devoe, I.W., and Gilchrist, J.E. (1973). Release of endotoxin in the form of cell wall blebs during in vitro growth of Neisseria meningitidis. J Exp Med 138, 1156-1167.

Diggle, S.P., Cornelis, P., Williams, P., and Camara, M. (2006). 4-quinolone signalling in Pseudomonas aeruginosa: old molecules, new perspectives. Int J Med Microbiol 296, 83- 91.

Dobereiner, H.G., Kas, J., Noppl, D., Sprenger, I., and Sackmann, E. (1993). Budding and fission of vesicles. Biophys J 65, 1396-1403.

Dorward, D.W., Schwan, T.G., and Garon, C.F. (1991). Immune capture and detection of Borrelia burgdorferi antigens in urine, blood, or tissues from infected ticks, mice, dogs, and humans. J Clin Microbiol 29, 1162-1170.

Dowhan, W. (2013). A retrospective: use of Escherichia coli as a vehicle to study phospholipid synthesis and function. Biochim Biophys Acta 1831, 471-494.

Drecktrah, D., Knodler, L.A., Howe, D., and Steele-Mortimer, O. (2007). Salmonella trafficking is defined by continuous dynamic interactions with the endolysosomal system. Traffic 8, 212-225.

180

Dutta, S., Iida, K., Takade, A., Meno, Y., Nair, G.B., and Yoshida, S. (2004). Release of Shiga Toxin by Membrane Vesicles in Shigella dysenteriae Serotype 1 Strains and In Vitro Effects of Antimicrobials on Toxin Production and Release. Microbiol Immunol 48, 965-969.

Ehrlich, M., Boll, W., Van Oijen, A., Hariharan, R., Chandran, K., Nibert, M.L., and Kirchhausen, T. (2004). Endocytosis by random initiation and stabilization of clathrin- coated pits. Cell 118, 591-605.

Eisenreich, W., Dandekar, T., Heesemann, J., and Goebel, W. (2010). Carbon metabolism of intracellular bacterial pathogens and possible links to virulence. Nat Rev Microbiol 8, 401-412.

Elhenawy, W., Bording-Jorgensen, M., Valguarnera, E., Haurat, M.F., Wine, E., and Feldman, M.F. (2016). LPS Remodeling Triggers Formation of Outer Membrane Vesicles in Salmonella. MBio 7.

Ellis, T.N., and Kuehn, M.J. (2010). Virulence and immunomodulatory roles of bacterial outer membrane vesicles. Microbiol Mol Biol Rev 74, 81-94.

Ellis, T.N., Leiman, S.A., and Kuehn, M.J. (2010). Naturally produced outer membrane vesicles from Pseudomonas aeruginosa elicit a potent innate immune response via combined sensing of both lipopolysaccharide and protein components. Infect Immun 78, 3822-3831.

Elmi, A., Watson, E., Sandu, P., Gundogdu, O., Mills, D.C., Inglis, N.F., Manson, E., Imrie, L., Bajaj-Elliott, M., Wren, B.W., et al. (2012). Campylobacter jejuni outer membrane vesicles play an important role in bacterial interactions with human intestinal epithelial cells. Infect Immun 80, 4089-4098.

Eriksson, S., Lucchini, S., Thompson, A., Rhen, M., and Hinton, J.C. (2003). Unravelling the biology of macrophage infection by gene expression profiling of intracellular Salmonella enterica. Mol Microbiol 47, 103-118.

Ernst, R.K., Guina, T., and Miller, S.I. (1999). How intracellular bacteria survive: surface modifications that promote resistance to host innate immune responses. J Infect Dis 179 Suppl 2, S326-330.

Ernst, R.K., Moskowitz, S.M., Emerson, J.C., Kraig, G.M., Adams, K.N., Harvey, M.D., Ramsey, B., Speert, D.P., Burns, J.L., and Miller, S.I. (2007). Unique lipid a modifications in Pseudomonas aeruginosa isolated from the airways of patients with cystic fibrosis. J Infect Dis 196, 1088-1092.

181

Evans, D.G., Evans, D.J., Jr., and DuPont, H.L. (1977). Virulence factors of enterotoxigenic Escherichia coli. J Infect Dis 136 Suppl, S118-123.

Fahey, P.F., and Webb, W.W. (1978). Lateral diffusion in phospholipid bilayer membranes and multilamellar liquid crystals. Biochemistry 17, 3046-3053.

Fernando, D., and Kumar, A. (2012). Growth phase-dependent expression of RND efflux pump- and outer membrane porin-encoding genes in Acinetobacter baumannii ATCC 19606. J Antimicrob Chemother 67, 569-572.

Finkelstein, R.A., LaRue, M.K., Johnston, D.W., Vasil, M.L., Cho, G.J., and Jones, J.R. (1976). Isolation and properties of heat-labile enterotoxin(s) from enterotoxigenic Escherichia coli. J Infect Dis 133 Suppl, 120-137.

Fiocca, R., Necchi, V., Sommi, P., Ricci, V., Telford, J., Cover, T.L., and Solcia, E. (1999). Release of Helicobacter pylori vacuolating cytotoxin by both a specific secretion pathway and budding of outer membrane vesicles. Uptake of released toxin and vesicles by gastric epithelium. J Pathol 188, 220-226.

Fleckenstein, J.M., Hardwidge, P.R., Munson, G.P., Rasko, D.A., Sommerfelt, H., and Steinsland, H. (2010). Molecular mechanisms of enterotoxigenic Escherichia coli infection. Microbes Infect 12, 89-98.

Frolov, V.A., Shnyrova, A.V., and Zimmerberg, J. (2011). Lipid polymorphisms and membrane shape. Cold Spring Harb Perspect Biol 3, a004747.

Galloway, S.M., and Raetz, C.R. (1990). A mutant of Escherichia coli defective in the first step of endotoxin biosynthesis. J Biol Chem 265, 6394-6402.

Gamage, S.D., McGannon, C.M., and Weiss, A.A. (2004). Escherichia coli serogroup O107/O117 lipopolysaccharide binds and neutralizes Shiga toxin 2. J Bacteriol 186, 5506- 5512.

Gankema, H., Wensink, J., Guinee, P.A., Jansen, W.H., and Witholt, B. (1980). Some characteristics of the outer membrane material released by growing enterotoxigenic Escherichia coli. Infect Immun 29, 704-713.

Garner, M.J., Hayward, R.D., and Koronakis, V. (2002). The Salmonella pathogenicity island 1 secretion system directs cellular cholesterol redistribution during mammalian cell entry and intracellular trafficking. Cell Microbiol 4, 153-165.

Ghazarian, H., Idoni, B., and Oppenheimer, S.B. (2011). A glycobiology review: carbohydrates, lectins and implications in cancer therapeutics. Acta Histochem 113, 236- 247. 182

Gibbons, H.S., Kalb, S.R., Cotter, R.J., and Raetz, C.R. (2005). Role of Mg2+ and pH in the modification of Salmonella lipid A after endocytosis by macrophage tumour cells. Mol Microbiol 55, 425-440.

Glauner, B., Holtje, J.V., and Schwarz, U. (1988). The composition of the murein of Escherichia coli. J Biol Chem 263, 10088-10095.

Goldfine, H., Rosenthal, J.J., and Johnston, N.C. (1987). Lipid shape as a determinant of lipid composition in Clostridium butyricum. The effects of incorporation of various fatty acids on the ratios of the major ether lipids. Biochim Biophys Acta 904, 283-289.

Guina, T., Yi, E.C., Wang, H., Hackett, M., and Miller, S.I. (2000). A PhoP-regulated outer membrane protease of Salmonella enterica serovar typhimurium promotes resistance to alpha-helical antimicrobial peptides. J Bacteriol 182, 4077-4086.

Gunn, J.S., Lim, K.B., Krueger, J., Kim, K., Guo, L., Hackett, M., and Miller, S.I. (1998). PmrA-PmrB-regulated genes necessary for 4-aminoarabinose lipid A modification and polymyxin resistance. Mol Microbiol 27, 1171-1182.

Hamilton, D.L., Johnson, M.R., Forsyth, G.W., Roe, W.E., and Nielsen, N.O. (1978). The effect of cholera toxin and heat labile and heat stable Escherichia coli enterotoxin on cyclic AMP concentrations in small intestinal mucosa of pig and rabbit. Can J Comp Med 42, 327-331.

Haurat, M.F., Aduse-Opoku, J., Rangarajan, M., Dorobantu, L., Gray, M.R., Curtis, M.A., and Feldman, M.F. (2011). Selective sorting of cargo proteins into bacterial membrane vesicles. J Biol Chem 286, 1269-1276.

Heinrichs, D.E., Yethon, J.A., Amor, P.A., and Whitfield, C. (1998a). The assembly system for the outer core portion of R1- and R4-type lipopolysaccharides of Escherichia coli. The R1 core-specific beta-glucosyltransferase provides a novel attachment site for O-polysaccharides. J Biol Chem 273, 29497-29505.

Heinrichs, D.E., Yethon, J.A., and Whitfield, C. (1998b). Molecular basis for structural diversity in the core regions of the lipopolysaccharides of Escherichia coli and Salmonella enterica. Mol Microbiol 30, 221-232.

Hirst, T.R., and Holmgren, J. (1987). Transient entry of enterotoxin subunits into the periplasm occurs during their secretion from Vibrio cholerae. J Bacteriol 169, 1037-1045.

Hirst, T.R., Sanchez, J., Kaper, J.B., Hardy, S.J., and Holmgren, J. (1984). Mechanism of toxin secretion by Vibrio cholerae investigated in strains harboring plasmids that encode heat-labile enterotoxins of Escherichia coli. Proc Natl Acad Sci U S A 81, 7752-7756.

183

Hiruma, R., Yamaguchi, A., and Sawai, T. (1984). The effect of lipopolysaccharide on lipid bilayer permeability of beta-lactam antibiotics. FEBS Lett 170, 268-272.

Hmiel, S.P., Snavely, M.D., Miller, C.G., and Maguire, M.E. (1986). Magnesium transport in Salmonella typhimurium: characterization of magnesium influx and cloning of a transport gene. J Bacteriol 168, 1444-1450.

Hoekstra, D., van der Laan, J.W., de Leij, L., and Witholt, B. (1976). Release of outer membrane fragments from normally growing Escherichia coli. Biochim Biophys Acta 455, 889-899.

Holmner, A., Askarieh, G., Okvist, M., and Krengel, U. (2007). Blood group antigen recognition by Escherichia coli heat-labile enterotoxin. J Mol Biol 371, 754-764.

Horstman, A.L., Bauman, S.J., and Kuehn, M.J. (2004). Lipopolysaccharide 3-deoxy-D- manno-octulosonic acid (Kdo) core determines bacterial association of secreted toxins. J Biol Chem 279, 8070-8075.

Horstman, A.L., and Kuehn, M.J. (2000a). Enterotoxigenic Escherichia coli secretes active heat-labile enterotoxin via outer membrane vesicles. J Biol Chem 275, 12489-12496.

Horstman, A.L., and Kuehn, M.J. (2000b). Enterotoxigenic Escherichia coli secretes active heat-labile enterotoxin via outer membrane vesicles. J Biol Chem 275, 12489-12496.

Horstman, A.L., and Kuehn, M.J. (2002). Bacterial surface association of heat-labile enterotoxin through lipopolysaccharide after secretion via the general secretory pathway. J Biol Chem 277, 32538-32545.

Hsu, L., Jackowski, S., and Rock, C.O. (1989). Uptake and acylation of 2-acyl- lysophospholipids by Escherichia coli. J Bacteriol 171, 1203-1205.

Jansson, L., Angstrom, J., Lebens, M., and Teneberg, S. (2010). No direct binding of the heat-labile enterotoxin of Escherichia coli to E. coli lipopolysaccharides. Glycoconj J 27, 171-179.

Jeeves, M., and Knowles, T.J. (2015). A novel pathway for outer membrane protein biogenesis in Gram-negative bacteria. Mol Microbiol 97, 607-611.

Jin, J.S., Kwon, S.O., Moon, D.C., Gurung, M., Lee, J.H., Kim, S.I., and Lee, J.C. (2011). Acinetobacter baumannii secretes cytotoxic outer membrane protein A via outer membrane vesicles. PLoS One 6, e17027.

184

Johnson, A.M., Kaushik, R.S., Francis, D.H., Fleckenstein, J.M., and Hardwidge, P.R. (2009). Heat-labile enterotoxin promotes Escherichia coli adherence to intestinal epithelial cells. J Bacteriol 191, 178-186.

Kadner, R.J. (1996). Cytoplasmic Membrane. In Escherichia coli and Salmonella: Cellular and Molecular Biology, F.C.N.e. al, ed. (Washington, DC: ASM Press), pp. 58-87.

Kadurugamuwa, J.L., and Beveridge, T.J. (1995). Virulence factors are released from Pseudomonas aeruginosa in association with membrane vesicles during normal growth and exposure to gentamicin: a novel mechanism of enzyme secretion. J Bacteriol 177, 3998-4008.

Kadurugamuwa, J.L., and Beveridge, T.J. (1996). Bacteriolytic effect of membrane vesicles from Pseudomonas aeruginosa on other bacteria including pathogens: conceptually new antibiotics. J Bacteriol 178, 2767-2774.

Kadurugamuwa, J.L., and Beveridge, T.J. (1997). Natural release of virulence factors in membrane vesicles by Pseudomonas aeruginosa and the effect of aminoglycoside antibiotics on their release. J Antimicrob Chemother 40, 615-621.

Kadurugamuwa, J.L., and Beveridge, T.J. (1998). Delivery of the non-membrane- permeative antibiotic gentamicin into mammalian cells by using Shigella flexneri membrane vesicles. Antimicrob Agents Chemother 42, 1476-1483.

Kaparakis, M., Turnbull, L., Carneiro, L., Firth, S., Coleman, H.A., Parkington, H.C., Le Bourhis, L., Karrar, A., Viala, J., Mak, J., et al. (2010). Bacterial membrane vesicles deliver peptidoglycan to NOD1 in epithelial cells. Cell Microbiol 12, 372- 385.

Kates, M., and Manson, L.A. (1984). Membrane fluidity (New York: Plenum Press).

Kato, A., Chen, H.D., Latifi, T., and Groisman, E.A. (2012). Reciprocal control between a bacterium's regulatory system and the modification status of its lipopolysaccharide. Mol Cell 47, 897-908.

Kato, S., Kowashi, Y., and Demuth, D.R. (2002). Outer membrane-like vesicles secreted by Actinobacillus actinomycetemcomitans are enriched in leukotoxin. Microb Pathog 32, 1-13.

Kawasaki, K., Ernst, R.K., and Miller, S.I. (2004). 3-O-deacylation of lipid A by PagL, a PhoP/PhoQ-regulated deacylase of Salmonella typhimurium, modulates signaling through Toll-like receptor 4. J Biol Chem 279, 20044-20048.

185

Keenan, J., Day, T., Neal, S., Cook, B., Perez-Perez, G., Allardyce, R., and Bagshaw, P. (2000). A role for the bacterial outer membrane in the pathogenesis of Helicobacter pylori infection. FEMS Microbiol Lett 182, 259-264.

Kennedy, E.P. (1982). Osmotic regulation and the biosynthesis of membrane-derived oligosaccharides in Escherichia coli. Proc Natl Acad Sci U S A 79, 1092-1095.

Kesty, N.C., and Kuehn, M.J. (2004). Incorporation of heterologous outer membrane and periplasmic proteins into Escherichia coli outer membrane vesicles. J Biol Chem 279, 2069-2076.

Kitagawa, R., Takaya, A., Ohya, M., Mizunoe, Y., Takade, A., Yoshida, S., Isogai, E., and Yamamoto, T. (2010). Biogenesis of Salmonella enterica serovar typhimurium membrane vesicles provoked by induction of PagC. J Bacteriol 192, 5645-5656.

Knodler, L.A. (2015). Salmonella enterica: living a double life in epithelial cells. Curr Opin Microbiol 23, 23-31.

Knodler, L.A., Crowley, S.M., Sham, H.P., Yang, H., Wrande, M., Ma, C., Ernst, R.K., Steele-Mortimer, O., Celli, J., and Vallance, B.A. (2014). Noncanonical inflammasome activation of caspase-4/caspase-11 mediates epithelial defenses against enteric bacterial pathogens. Cell Host Microbe 16, 249-256.

Kolling, G.L., and Matthews, K.R. (1999). Export of virulence genes and shiga toxin by membrane vesicles of escherichia coli O157:H7. Appl Environ Microbiol 65, 1843-1848.

Kuehn, M.J., and Kesty, N.C. (2005). Bacterial outer membrane vesicles and the host- pathogen interaction. Genes Dev 19, 2645-2655.

Kulp, A., and Kuehn, M.J. (2010). Biological functions and biogenesis of secreted bacterial outer membrane vesicles. Annu Rev Microbiol 64, 163-184.

Kumada, H., Haishima, Y., Umemoto, T., and Tanamoto, K. (1995). Structural study on the free lipid A isolated from lipopolysaccharide of Porphyromonas gingivalis. J Bacteriol 177, 2098-2106.

Kunkel, S.L., and Robertson, D.C. (1979). Purification and chemical characterization of the heat-labile enterotoxin produced by enterotoxigenic Escherichia coli. Infect Immun 25, 586-596.

Lam, H., Oh, D.C., Cava, F., Takacs, C.N., Clardy, J., de Pedro, M.A., and Waldor, M.K. (2009). D-amino acids govern stationary phase cell wall remodeling in bacteria. Science 325, 1552-1555.

186

LaRock, D.L., Chaudhary, A., and Miller, S.I. (2015). Salmonellae interactions with host processes. Nat Rev Microbiol 13, 191-205.

Lee, A.G. (1991). Lipids and their effects on membrane proteins: evidence against a role for fluidity. Prog Lipid Res 30, 323-348.

Lencer, W.I., Hirst, T.R., and Holmes, R.K. (1999). Membrane traffic and the cellular uptake of cholera toxin. Biochim Biophys Acta 1450, 177-190.

Lerouge, I., and Vanderleyden, J. (2002). O-antigen structural variation: mechanisms and possible roles in animal/plant-microbe interactions. FEMS Microbiol Rev 26, 17-47.

Li, Z., Clarke, A.J., and Beveridge, T.J. (1996). A major autolysin of Pseudomonas aeruginosa: subcellular distribution, potential role in cell growth and division and secretion in surface membrane vesicles. J Bacteriol 178, 2479-2488.

Li, Z., Clarke, A.J., and Beveridge, T.J. (1998). Gram-negative bacteria produce membrane vesicles which are capable of killing other bacteria. J Bacteriol 180, 5478-5483.

Lindmark, B., Rompikuntal, P.K., Vaitkevicius, K., Song, T., Mizunoe, Y., Uhlin, B.E., Guerry, P., and Wai, S.N. (2009). Outer membrane vesicle-mediated release of cytolethal distending toxin (CDT) from Campylobacter jejuni. BMC Microbiol 9, 220.

Maas, S.L., De Vrij, J., and Broekman, M.L. (2014). Quantification and size-profiling of extracellular vesicles using tunable resistive pulse sensing. J Vis Exp, e51623.

Macdonald, I.A., and Kuehn, M.J. (2013). Stress-induced outer membrane vesicle production by Pseudomonas aeruginosa. J Bacteriol 195, 2971-2981.

MacKenzie, C.R., Hirama, T., Lee, K.K., Altman, E., and Young, N.M. (1997). Quantitative analysis of bacterial toxin affinity and specificity for glycolipid receptors by surface plasmon resonance. J Biol Chem 272, 5533-5538.

Malik-Kale, P., Winfree, S., and Steele-Mortimer, O. (2012). The bimodal lifestyle of intracellular Salmonella in epithelial cells: replication in the cytosol obscures defects in vacuolar replication. PLoS One 7, e38732.

Manabe, T., and Kawasaki, K. (2008). Extracellular loops of lipid A 3-O-deacylase PagL are involved in recognition of aminoarabinose-based membrane modifications in Salmonella enterica serovar typhimurium. J Bacteriol 190, 5597-5606.

Manning, A.J., and Kuehn, M.J. (2011). Contribution of bacterial outer membrane vesicles to innate bacterial defense. BMC Microbiol 11, 258.

187

Manning, A.J., and Kuehn, M.J. (2013). Functional advantages conferred by extracellular prokaryotic membrane vesicles. J Mol Microbiol Biotechnol 23, 131-141.

Martinez-Moya, M., de Pedro, M.A., Schwarz, H., and Garcia-del Portillo, F. (1998). Inhibition of Salmonella intracellular proliferation by non-phagocytic eucaryotic cells. Res Microbiol 149, 309-318.

Mashburn-Warren, L., Howe, J., Garidel, P., Richter, W., Steiniger, F., Roessle, M., Brandenburg, K., and Whiteley, M. (2008). Interaction of quorum signals with outer membrane lipids: insights into prokaryotic membrane vesicle formation. Mol Microbiol 69, 491-502.

Mayrand, D., and Grenier, D. (1989). Biological activities of outer membrane vesicles. Can J Microbiol 35, 607-613.

McBroom, A.J., Johnson, A.P., Vemulapalli, S., and Kuehn, M.J. (2006). Outer membrane vesicle production by Escherichia coli is independent of membrane instability. J Bacteriol 188, 5385-5392.

McBroom, A.J., and Kuehn, M.J. (2007). Release of outer membrane vesicles by Gram- negative bacteria is a novel envelope stress response. Mol Microbiol 63, 545-558.

McElhaney, R.N. (1986). Differential scanning calorimetric studies of lipid-protein interactions in model membrane systems. Biochim Biophys Acta 864, 361-421.

McMahon, H.T., and Boucrot, E. (2015). Membrane curvature at a glance. J Cell Sci 128, 1065-1070.

McMahon, K.J., Castelli, M.E., Vescovi, E.G., and Feldman, M.F. (2012). Biogenesis of Outer Membrane Vesicles in Serratia marcescens Is Thermoregulated and Can Be Induced by Activation of the Rcs Phosphorelay System. J Bacteriol 194, 3241- 3249.

Merritt, E.A., Sixma, T.K., Kalk, K.H., van Zanten, B.A., and Hol, W.G. (1994a). Galactose-binding site in Escherichia coli heat-labile enterotoxin (LT) and cholera toxin (CT). Mol Microbiol 13, 745-753.

Merritt, E.A., Sixma, T.K., Kalk, K.H., van Zanten, B.A., and Hol, W.G. (1994b). Galactose-binding site in Escherichia coli heat-labile enterotoxin (LT) and cholera toxin (CT). Mol Microbiol 13, 745-753.

Meyer, D.H., and Fives-Taylor, P.M. (1994). Characteristics of adherence of Actinobacillus actinomycetemcomitans to epithelial cells. Infect Immun 62, 928-935.

188

Miao, E.A., Brittnacher, M., Haraga, A., Jeng, R.L., Welch, M.D., and Miller, S.I. (2003). Salmonella effectors translocated across the vacuolar membrane interact with the actin cytoskeleton. Mol Microbiol 48, 401-415.

Mileykovskaya, E., and Dowhan, W. (2009). Cardiolipin membrane domains in prokaryotes and eukaryotes. Biochim Biophys Acta 1788, 2084-2091.

Miller, A.K., Brannon, M.K., Stevens, L., Johansen, H.K., Selgrade, S.E., Miller, S.I., Hoiby, N., and Moskowitz, S.M. (2011). PhoQ mutations promote lipid A modification and polymyxin resistance of Pseudomonas aeruginosa found in colistin-treated cystic fibrosis patients. Antimicrob Agents Chemother 55, 5761-5769.

Moran, A.P., Lindner, B., and Walsh, E.J. (1997). Structural characterization of the lipid A component of Helicobacter pylori rough- and smooth-form lipopolysaccharides. J Bacteriol 179, 6453-6463.

Moss, J., and Vaughan, M. (1981). Mechanism of action of choleragen and E. coli heat- labile enterotoxin: activation of adenylate cyclase by ADP-ribosylation. Mol Cell Biochem 37, 75-90.

Mudrak, B., and Kuehn, M.J. (2010a). Heat-labile enterotoxin: beyond G(m1) binding. Toxins (Basel) 2, 1445-1470.

Mudrak, B., and Kuehn, M.J. (2010b). Specificity of the type II secretion systems of enterotoxigenic Escherichia coli and Vibrio cholerae for heat-labile enterotoxin and cholera toxin. J Bacteriol 192, 1902-1911.

Mudrak, B., Rodriguez, D.L., and Kuehn, M.J. (2009). Residues of Heat-Labile Enterotoxin Involved in Bacterial Cell Surface Binding. J Bacteriol.

Mug-Opstelten, D., and Witholt, B. (1978). Preferential release of new outer membrane fragments by exponentially growing Escherichia coli. Biochim Biophys Acta 508, 287- 295.

Muhlradt, P.F., Menzel, J., Golecki, J.R., and Speth, V. (1974). Lateral mobility and surface density of lipopolysaccharide in the outer membrane of Salmonella typhimurium. Eur J Biochem 43, 533-539.

Mullineaux, C.W., Nenninger, A., Ray, N., and Robinson, C. (2006). Diffusion of green fluorescent protein in three cell environments in Escherichia coli. J Bacteriol 188, 3442- 3448.

189

Murata, T., Tseng, W., Guina, T., Miller, S.I., and Nikaido, H. (2007). PhoPQ-mediated regulation produces a more robust permeability barrier in the outer membrane of Salmonella enterica serovar typhimurium. J Bacteriol 189, 7213-7222.

Nelson, D.L., and Kennedy, E.P. (1971). Magnesium transport in Escherichia coli. Inhibition by cobaltous ion. J Biol Chem 246, 3042-3049.

Niedergang, F., Sirard, J.C., Blanc, C.T., and Kraehenbuhl, J.P. (2000). Entry and survival of Salmonella typhimurium in dendritic cells and presentation of recombinant antigens do not require macrophage-specific virulence factors. Proc Natl Acad Sci U S A 97, 14650-14655.

Nikaido, H. (2003). Molecular basis of bacterial outer membrane permeability revisited. Microbiol Mol Biol Rev 67, 593-656.

Nikaido, H., and Vaara, M. (1985). Molecular basis of bacterial outer membrane permeability. Microbiol Rev 49, 1-32.

Nishio, M., Okada, N., Miki, T., Haneda, T., and Danbara, H. (2005). Identification of the outer-membrane protein PagC required for the serum resistance phenotype in Salmonella enterica serovar Choleraesuis. Microbiology 151, 863-873.

Ochman, H., Elwyn, S., and Moran, N.A. (1999). Calibrating bacterial evolution. Proc Natl Acad Sci U S A 96, 12638-12643.

Ohta, H., Hara, H., Fukui, K., Kurihara, H., Murayama, Y., and Kato, K. (1993). Association of Actinobacillus actinomycetemcomitans leukotoxin with nucleic acids on the bacterial cell surface. Infect Immun 61, 4878-4884.

Ohta, H., Kato, K., Kokeguchi, S., Hara, H., Fukui, K., and Murayama, Y. (1991). Nuclease-sensitive binding of an Actinobacillus actinomycetemcomitans leukotoxin to the bacterial cell surface. Infect Immun 59, 4599-4605.

Okuda, S., Sherman, D.J., Silhavy, T.J., Ruiz, N., and Kahne, D. (2016). Lipopolysaccharide transport and assembly at the outer membrane: the PEZ model. Nat Rev Microbiol 14, 337-345.

Olczak, T., Wojtowicz, H., Ciuraszkiewicz, J., and Olczak, M. (2010). Species specificity, surface exposure, protein expression, immunogenicity, and participation in biofilm formation of Porphyromonas gingivalis HmuY. BMC Microbiol 10, 134.

Oliver, D.B. (1996). Periplasm. In Escherichia coli and Salmonella: Cellular and Molecular Biology, F.C. Neidhardt, ed. (Washington, DC: ASM Press), pp. 88-103.

190

Olofsson, A., Vallstrom, A., Petzold, K., Tegtmeyer, N., Schleucher, J., Carlsson, S., Haas, R., Backert, S., Wai, S.N., Grobner, G., et al. (2010). Biochemical and functional characterization of Helicobacter pylori vesicles. Mol Microbiol 77, 1539-1555.

Parikh, S.J., and Chorover, J. (2007). Infrared spectroscopy studies of cation effects on lipopolysaccharides in aqueous solution. Colloids Surf B Biointerfaces 55, 241-250.

Park, J.S., Lee, W.C., Yeo, K.J., Ryu, K.S., Kumarasiri, M., Hesek, D., Lee, M., Mobashery, S., Song, J.H., Kim, S.I., et al. (2012). Mechanism of anchoring of OmpA protein to the cell wall peptidoglycan of the gram-negative bacterial outer membrane. FASEB J 26, 219-228.

Parker, H., Chitcholtan, K., Hampton, M.B., and Keenan, J.I. (2010). Uptake of Helicobacter pylori outer membrane vesicles by gastric epithelial cells. Infect Immun 78, 5054-5061.

Peterson, A.A., and McGroarty, E.J. (1985). High-molecular-weight components in lipopolysaccharides of Salmonella typhimurium, Salmonella minnesota, and Escherichia coli. J Bacteriol 162, 738-745.

Pettit, R.K., and Judd, R.C. (1992). The interaction of naturally elaborated blebs from serum-susceptible and serum-resistant strains of Neisseria gonorrhoeae with normal human serum. Mol Microbiol 6, 729-734.

Pezeshkian, W., Hansen, A.G., Johannes, L., Khandelia, H., Shillcock, J.C., Kumar, P.B., and Ipsen, J.H. (2016). Membrane invagination induced by Shiga toxin B-subunit: from molecular structure to tube formation. Soft Matter 12, 5164-5171.

Phillips, N.J., Schilling, B., McLendon, M.K., Apicella, M.A., and Gibson, B.W. (2004). Novel modification of lipid A of Francisella tularensis. Infect Immun 72, 5340-5348.

Pisabarro, A.G., de Pedro, M.A., and Vazquez, D. (1985). Structural modifications in the peptidoglycan of Escherichia coli associated with changes in the state of growth of the culture. J Bacteriol 161, 238-242.

Prasadarao, N.V., Wass, C.A., Weiser, J.N., Stins, M.F., Huang, S.H., and Kim, K.S. (1996). Outer membrane protein A of Escherichia coli contributes to invasion of brain microvascular endothelial cells. Infect Immun 64, 146-153.

Prinz, W.A., and Hinshaw, J.E. (2009). Membrane-bending proteins. Crit Rev Biochem Mol Biol 44, 278-291.

Pugsley, A.P. (1993). The complete general secretory pathway in gram-negative bacteria. Microbiol Rev 57, 50-108.

191

Qadri, F., Svennerholm, A.M., Faruque, A.S., and Sack, R.B. (2005). Enterotoxigenic Escherichia coli in developing countries: epidemiology, microbiology, clinical features, treatment, and prevention. Clin Microbiol Rev 18, 465-483.

Quintela, J.C., de Pedro, M.A., Zollner, P., Allmaier, G., and Garcia-del Portillo, F. (1997). Peptidoglycan structure of Salmonella typhimurium growing within cultured mammalian cells. Mol Microbiol 23, 693-704.

Raetz, C.R., and Dowhan, W. (1990). Biosynthesis and function of phospholipids in Escherichia coli. J Biol Chem 265, 1235-1238.

Raetz, C.R., Garrett, T.A., Reynolds, C.M., Shaw, W.A., Moore, J.D., Smith, D.C., Jr., Ribeiro, A.A., Murphy, R.C., Ulevitch, R.J., Fearns, C., et al. (2006). Kdo2-Lipid A of Escherichia coli, a defined endotoxin that activates macrophages via TLR-4. J Lipid Res 47, 1097-1111.

Raetz, C.R., Reynolds, C.M., Trent, M.S., and Bishop, R.E. (2007). Lipid A modification systems in gram-negative bacteria. Annu Rev Biochem 76, 295-329.

Raetz, C.R., and Whitfield, C. (2002). Lipopolysaccharide endotoxins. Annu Rev Biochem 71, 635-700.

Rajagopal, S., Eis, N., Bhattacharya, M., and Nickerson, K.W. (2003). Membrane-derived oligosaccharides (MDOs) are essential for sodium dodecyl sulfate resistance in Escherichia coli. FEMS Microbiol Lett 223, 25-31.

Rassam, P., Copeland, N.A., Birkholz, O., Toth, C., Chavent, M., Duncan, A.L., Cross, S.J., Housden, N.G., Kaminska, R., Seger, U., et al. (2015). Supramolecular assemblies underpin turnover of outer membrane proteins in bacteria. Nature 523, 333-336.

Reynolds, C.M., Ribeiro, A.A., McGrath, S.C., Cotter, R.J., Raetz, C.R., and Trent, M.S. (2006). An outer membrane enzyme encoded by Salmonella typhimurium lpxR that removes the 3'-acyloxyacyl moiety of lipid A. J Biol Chem 281, 21974-21987.

Rietveld, A.G., Killian, J.A., Dowhan, W., and de Kruijff, B. (1993). Polymorphic regulation of membrane phospholipid composition in Escherichia coli. J Biol Chem 268, 12427-12433.

Rodighiero, C., Aman, A.T., Kenny, M.J., Moss, J., Lencer, W.I., and Hirst, T.R. (1999). Structural basis for the differential toxicity of cholera toxin and Escherichia coli heat- labile enterotoxin. Construction of hybrid toxins identifies the A2-domain as the determinant of differential toxicity. J Biol Chem 274, 3962-3969.

192

Roier, S., Zingl, F.G., Cakar, F., Durakovic, S., Kohl, P., Eichmann, T.O., Klug, L., Gadermaier, B., Weinzerl, K., Prassl, R., et al. (2016). A novel mechanism for the biogenesis of outer membrane vesicles in Gram-negative bacteria. Nat Commun 7, 10515.

Romer, W., Berland, L., Chambon, V., Gaus, K., Windschiegl, B., Tenza, D., Aly, M.R., Fraisier, V., Florent, J.C., Perrais, D., et al. (2007). Shiga toxin induces tubular membrane invaginations for its uptake into cells. Nature 450, 670-675.

Rompikuntal, P.K., Thay, B., Khan, M.K., Alanko, J., Penttinen, A.M., Asikainen, S., Wai, S.N., and Oscarsson, J. (2011). Perinuclear localization of internalized outer membrane vesicles carrying active cytolethal distending toxin from Aggregatibacter actinomycetemcomitans. Infect Immun 80, 31-42.

Rompikuntal, P.K., Vdovikova, S., Duperthuy, M., Johnson, T.L., Ahlund, M., Lundmark, R., Oscarsson, J., Sandkvist, M., Uhlin, B.E., and Wai, S.N. (2015). Outer Membrane Vesicle-Mediated Export of Processed PrtV Protease from Vibrio cholerae. PLoS One 10, e0134098.

Rossy, J., Ma, Y., and Gaus, K. (2014). The organisation of the cell membrane: do proteins rule lipids? Curr Opin Chem Biol 20, 54-59.

Rubin, R.H., and Weinstein, L. (1977). Salmonellosis : microbiologic, pathologic, and clinical features (New York: Stratton Intercontinental Medical Book).

Sack, R.B., Gorbach, S.L., Banwell, J.G., Jacobs, B., Chatterjee, B.D., and Mitra, R.C. (1971). Enterotoxigenic Escherichia coli isolated from patients with severe cholera-like disease. J Infect Dis 123, 378-385.

Sandkvist, M. (2001). Biology of type II secretion. Mol Microbiol 40, 271-283.

Sandkvist, M., Michel, L.O., Hough, L.P., Morales, V.M., Bagdasarian, M., Koomey, M., and DiRita, V.J. (1997). General secretion pathway (eps) genes required for toxin secretion and outer membrane biogenesis in Vibrio cholerae. J Bacteriol 179, 6994-7003.

Scandella, C.J., and Kornberg, A. (1971). A membrane-bound phospholipase A1 purified from Escherichia coli. Biochemistry 10, 4447-4456.

Schindler, M., Osborn, M.J., and Koppel, D.E. (1980). Lateral diffusion of lipopolysaccharide in the outer membrane of Salmonella typhimurium. Nature 285, 261- 263.

Schnaitman, C.A., and Klena, J.D. (1993). Genetics of lipopolysaccharide biosynthesis in enteric bacteria. Microbiol Rev 57, 655-682. 193

Schooling, S.R., and Beveridge, T.J. (2006). Membrane vesicles: an overlooked component of the matrices of biofilms. J Bacteriol 188, 5945-5957.

Schwechheimer, C., and Kuehn, M.J. (2013). Synthetic Effect between Envelope Stress and Lack of Outer Membrane Vesicle Production in Escherichia coli. J Bacteriol 195, 4161-4173.

Schwechheimer, C., and Kuehn, M.J. (2015). Outer-membrane vesicles from Gram- negative bacteria: biogenesis and functions. Nat Rev Microbiol 13, 605-619.

Schwechheimer, C., Rodriguez, D.L., and Kuehn, M.J. (2015). NlpI-mediated modulation of outer membrane vesicle production through peptidoglycan dynamics in Escherichia coli. Microbiologyopen.

Schwechheimer, C., Sullivan, C.J., and Kuehn, M.J. (2013). Envelope control of outer membrane vesicle production in Gram-negative bacteria. Biochemistry 52, 3031-3040.

Selkrig, J., Leyton, D.L., Webb, C.T., and Lithgow, T. (2014). Assembly of beta-barrel proteins into bacterial outer membranes. Biochim Biophys Acta 1843, 1542-1550.

Seydel, U., Oikawa, M., Fukase, K., Kusumoto, S., and Brandenburg, K. (2000). Intrinsic conformation of lipid A is responsible for agonistic and antagonistic activity. Eur J Biochem 267, 3032-3039.

Shimizu, M., Sakano, T., Yamamoto, J., and Sakurai, K. (1984). Diarrhea in neonatal piglets caused by K99+ST+ enterotoxigenic Escherichia coli. Microbiol Immunol 28, 645- 649.

Shimizu, T., Sato, T., Kawakami, S., Ohta, T., Noda, M., and Hamabata, T. (2007). Receptor affinity, stability and binding mode of Shiga toxins are determinants of toxicity. Microb Pathog 43, 88-95.

Shoberg, R.J., and Thomas, D.D. (1993). Specific adherence of Borrelia burgdorferi extracellular vesicles to human endothelial cells in culture. Infect Immun 61, 3892-3900.

Silhavy, T.J., Kahne, D., and Walker, S. (2010). The bacterial cell envelope. Cold Spring Harb Perspect Biol 2, a000414.

Sixma, T.K., Kalk, K.H., van Zanten, B.A., Dauter, Z., Kingma, J., Witholt, B., and Hol, W.G. (1993). Refined structure of Escherichia coli heat-labile enterotoxin, a close relative of cholera toxin. J Mol Biol 230, 890-918.

194

Smalley, J.W., Byrne, D.P., Birss, A.J., Wojtowicz, H., Sroka, A., Potempa, J., and Olczak, T. (2011). HmuY haemophore and gingipain proteases constitute a unique syntrophic system of haem acquisition by Porphyromonas gingivalis. PLoS One 6, e17182.

Smit, J., and Nikaido, H. (1978). Outer membrane of gram-negative bacteria. XVIII. Electron microscopic studies on porin insertion sites and growth of cell surface of Salmonella typhimurium. J Bacteriol 135, 687-702.

Snyder, S., Kim, D., and McIntosh, T.J. (1999). Lipopolysaccharide bilayer structure: effect of chemotype, core mutations, divalent cations, and temperature. Biochemistry 38, 10758-10767.

Sochacki, K.A., Shkel, I.A., Record, M.T., and Weisshaar, J.C. (2011). Protein diffusion in the periplasm of E. coli under osmotic stress. Biophys J 100, 22-31.

Soltyk, A.M., MacKenzie, C.R., Wolski, V.M., Hirama, T., Kitov, P.I., Bundle, D.R., and Brunton, J.L. (2002). A mutational analysis of the globotriaosylceramide-binding sites of verotoxin VT1. J Biol Chem 277, 5351-5359.

Sommi, P., Ricci, V., Fiocca, R., Necchi, V., Romano, M., Telford, J.L., Solcia, E., and Ventura, U. (1998). Persistence of Helicobacter pylori VacA toxin and vacuolating potential in cultured gastric epithelial cells. Am J Physiol 275, G681-688.

Stachowiak, J.C., Schmid, E.M., Ryan, C.J., Ann, H.S., Sasaki, D.Y., Sherman, M.B., Geissler, P.L., Fletcher, D.A., and Hayden, C.C. (2012). Membrane bending by protein- protein crowding. Nat Cell Biol 14, 944-949.

Stead, C.M., Beasley, A., Cotter, R.J., and Trent, M.S. (2008). Deciphering the unusual acylation pattern of Helicobacter pylori lipid A. J Bacteriol 190, 7012-7021.

Stecher, B., Robbiani, R., Walker, A.W., Westendorf, A.M., Barthel, M., Kremer, M., Chaffron, S., Macpherson, A.J., Buer, J., Parkhill, J., et al. (2007). Salmonella enterica serovar typhimurium exploits inflammation to compete with the intestinal microbiota. PLoS Biol 5, 2177-2189.

Steele-Mortimer, O., Meresse, S., Gorvel, J.P., Toh, B.H., and Finlay, B.B. (1999). Biogenesis of Salmonella typhimurium-containing vacuoles in epithelial cells involves interactions with the early endocytic pathway. Cell Microbiol 1, 33-49.

Strohalm, M., Hassman, M., Kosata, B., and Kodicek, M. (2008). mMass data miner: an open source alternative for mass spectrometric data analysis. Rapid Commun Mass Spectrom 22, 905-908.

195

Suda, Y., Kim, Y.M., Ogawa, T., Yasui, N., Hasegawa, Y., Kashihara, W., Shimoyama, T., Aoyama, K., Nagata, K., Tamura, T., et al. (2001). Chemical structure and biological activity of a lipid A component from Helicobacter pylori strain 206. J Endotoxin Res 7, 95-104.

Takii, R., Kadowaki, T., Baba, A., Tsukuba, T., and Yamamoto, K. (2005). A functional virulence complex composed of gingipains, adhesins, and lipopolysaccharide shows high affinity to host cells and matrix proteins and escapes recognition by host immune systems. Infect Immun 73, 883-893.

Tan, T.T., Morgelin, M., Forsgren, A., and Riesbeck, K. (2007). Haemophilus influenzae survival during complement-mediated attacks is promoted by Moraxella catarrhalis outer membrane vesicles. J Infect Dis 195, 1661-1670.

Tate, M.W., Eikenberry, E.F., Turner, D.C., Shyamsunder, E., and Gruner, S.M. (1991). Nonbilayer phases of membrane lipids. Chem Phys Lipids 57, 147-164.

Tauschek, M., Gorrell, R.J., Strugnell, R.A., and Robins-Browne, R.M. (2002). Identification of a protein secretory pathway for the secretion of heat- labile enterotoxin by an enterotoxigenic strain of Escherichia coli. Proc Natl Acad Sci U S A 99, 7066-7071.

Tokuda, H., and Matsuyama, S. (2004). Sorting of lipoproteins to the outer membrane in E. coli. Biochim Biophys Acta 1693, 5-13.

Tran, A.X., Whittimore, J.D., Wyrick, P.B., McGrath, S.C., Cotter, R.J., and Trent, M.S. (2006). The lipid A 1-phosphatase of Helicobacter pylori is required for resistance to the antimicrobial peptide polymyxin. J Bacteriol 188, 4531-4541.

Tsuji, T., Inoue, T., Miyama, A., and Noda, M. (1991). Glutamic acid-112 of the A subunit of heat-labile enterotoxin from enterotoxigenic Escherichia coli is important for ADP- ribosyltransferase activity. FEBS Lett 291, 319-321.

Ueno, Y., Ohara, M., Kawamoto, T., Fujiwara, T., Komatsuzawa, H., Oswald, E., and Sugai, M. (2006). Biogenesis of the Actinobacillus actinomycetemcomitans cytolethal distending toxin holotoxin. Infect Immun 74, 3480-3487.

Ursell, T.S., Trepagnier, E.H., Huang, K.C., and Theriot, J.A. (2012). Analysis of surface protein expression reveals the growth pattern of the gram-negative outer membrane. PLoS Comput Biol 8, e1002680. van den Brink-van der Laan, E., Killian, J.A., and de Kruijff, B. (2004). Nonbilayer lipids affect peripheral and integral membrane proteins via changes in the lateral pressure profile. Biochim Biophys Acta 1666, 275-288.

196

Vinogradov, E., Perry, M.B., and Conlan, J.W. (2002). Structural analysis of Francisella tularensis lipopolysaccharide. Eur J Biochem 269, 6112-6118.

Vollmer, W., and Bertsche, U. (2008). Murein (peptidoglycan) structure, architecture and biosynthesis in Escherichia coli. Biochim Biophys Acta 1778, 1714-1734.

Wade, L.G. (1999). Organic chemistry, 4th edn (Upper Saddle River, N.J.: Prentice Hall).

Wai, S.N., Lindmark, B., Soderblom, T., Takade, A., Westermark, M., Oscarsson, J., Jass, J., Richter-Dahlfors, A., Mizunoe, Y., and Uhlin, B.E. (2003). Vesicle-mediated export and assembly of pore-forming oligomers of the enterobacterial ClyA cytotoxin. Cell 115, 25- 35.

Wang, X., Jiang, F., Zheng, J., Chen, L., Dong, J., Sun, L., Zhu, Y., Liu, B., Yang, J., Yang, G., et al. (2016). The outer membrane phospholipase A is essential for membrane integrity and type III secretion in Shigella flexneri. Open Biol 6.

Waravdekar, V.S., and Saslaw, L.D. (1957). A method of estimation of 2-deoxyribose. Biochim Biophys Acta 24, 439.

Wassall, S.R., Brzustowicz, M.R., Shaikh, S.R., Cherezov, V., Caffrey, M., and Stillwell, W. (2004). Order from disorder, corralling cholesterol with chaotic lipids. The role of polyunsaturated lipids in membrane raft formation. Chem Phys Lipids 132, 79-88.

Weintraub, A., Zahringer, U., Wollenweber, H.W., Seydel, U., and Rietschel, E.T. (1989). Structural characterization of the lipid A component of Bacteroides fragilis strain NCTC 9343 lipopolysaccharide. Eur J Biochem 183, 425-431.

Wensink, J., Gankema, H., Jansen, W.H., Guinee, P.A., and Witholt, B. (1978). Isolation of the membranes of an enterotoxigenic strain of Escherichia coli and distribution of enterotoxin activity in different subcellular fractions. Biochim Biophys Acta 514, 128-136.

Wieslander, A., Christiansson, A., Rilfors, L., and Lindblom, G. (1980). Lipid bilayer stability in membranes. Regulation of lipid composition in Acholeplasma laidlawii as governed by molecular shape. Biochemistry 19, 3650-3655.

Wilkinson, S.G. (1983). Composition and structure of lipopolysaccharides from Pseudomonas aeruginosa. Rev Infect Dis 5 Suppl 5, S941-949.

Winter, S.E., Thiennimitr, P., Winter, M.G., Butler, B.P., Huseby, D.L., Crawford, R.W., Russell, J.M., Bevins, C.L., Adams, L.G., Tsolis, R.M., et al. (2010). Gut inflammation provides a respiratory electron acceptor for Salmonella. Nature 467, 426-429.

197

Wolf, A.A., Jobling, M.G., Wimer-Mackin, S., Ferguson-Maltzman, M., Madara, J.L., Holmes, R.K., and Lencer, W.I. (1998). Ganglioside structure dictates signal transduction by cholera toxin and association with caveolae-like membrane domains in polarized epithelia. J Cell Biol 141, 917-927.

Yaganza, E.S., Rioux, D., Simard, M., Arul, J., and Tweddell, R.J. (2004). Ultrastructural alterations of Erwinia carotovora subsp. atroseptica caused by treatment with aluminum chloride and sodium metabisulfite. Appl Environ Microbiol 70, 6800-6808.

Yethon, J.A., Heinrichs, D.E., Monteiro, M.A., Perry, M.B., and Whitfield, C. (1998). Involvement of waaY, waaQ, and waaP in the modification of Escherichia coli lipopolysaccharide and their role in the formation of a stable outer membrane. J Biol Chem 273, 26310-26316.

Yoder, J.S., Cesario, S., Plotkin, V., Ma, X., Kelly-Shannon, K., and Dworkin, M.S. (2006). Outbreak of enterotoxigenic Escherichia coli infection with an unusually long duration of illness. Clin Infect Dis 42, 1513-1517.

Yokoyama, K., Horii, T., Yamashino, T., Hashikawa, S., Barua, S., Hasegawa, T., Watanabe, H., and Ohta, M. (2000). Production of shiga toxin by Escherichia coli measured with reference to the membrane vesicle-associated toxins. FEMS Microbiol Lett 192, 139-144.

Yonezawa, H., Osaki, T., Kurata, S., Fukuda, M., Kawakami, H., Ochiai, K., Hanawa, T., and Kamiya, S. (2009). Outer membrane vesicles of Helicobacter pylori TK1402 are involved in biofilm formation. BMC Microbiol 9, 197.

Yoon, H., Ansong, C., Adkins, J.N., and Heffron, F. (2011). Discovery of Salmonella virulence factors translocated via outer membrane vesicles to murine macrophages. Infect Immun 79, 2182-2192.

Zhou, L., Srisatjaluk, R., Justus, D.E., and Doyle, R.J. (1998). On the origin of membrane vesicles in gram-negative bacteria. FEMS Microbiol Lett 163, 223-228.

Zhou, Z., Lin, S., Cotter, R.J., and Raetz, C.R. (1999). Lipid A modifications characteristic of Salmonella typhimurium are induced by NH4VO3 in Escherichia coli K12. Detection of 4-amino-4-deoxy-L-arabinose, phosphoethanolamine and palmitate. J Biol Chem 274, 18503-18514.

Zhou, Z., Ribeiro, A.A., Lin, S., Cotter, R.J., Miller, S.I., and Raetz, C.R. (2001). Lipid A modifications in polymyxin-resistant Salmonella typhimurium: PMRA-dependent 4- amino-4-deoxy-L-arabinose, and phosphoethanolamine incorporation. J Biol Chem 276, 43111-43121.

198

Biography

Katherine Bonnington was born on May 5, 1989 in Houston, TX. She graduated from Memorial High School in May 2007 and began her studies at Carnegie Mellon

University that fall. A Science and Humanities Scholar at CMU, Katherine graduated with honors, receiving a B.S. in Biological Sciences in 2011. She entered graduate school at Duke University through the Biochemistry Department and joined the Kuehn lab in

April of 2012. Katherine has presented her research at the 2013 Mid-Atlantic Microbial

Pathogenesis meeting, the 2015 Glycobiology Gordon Research Conference, and the 2016

Bacterial Cell Surfaces Gordon Research Conference. She has authored one review titled

“Protein selection and export via outer membrane vesicles,” one article titled “Outer membrane vesicle production facilitates LPS remodeling and outer membrane maintenance in Salmonella during environmental transitions,” one microreview titled

“Breaking the bilayer: OMV formation during environmental transitions,” and is currently preparing two manuscripts titled “Quantitative comparison of the outer membrane and secreted OMV proteomes of Salmonella typhimurium grown in SCV-like conditions” and “Lipopolysaccharide binding by heat-labile enterotoxin influences outer membrane dynamics in enterotoxigenic E. coli.” In March of 2017 she plans to defend her thesis prior to graduation in May.

199