The O-Antigen Capsule of Salmonella Typhimurium in Acute and Chronic Infection. Dissertation

Transcription

1 The O-Antigen Capsule of Salmonella Typhimurium in Acute and Chronic Infection Dissertation Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy In the Graduate School of The Ohio State University By Joanna M. Marshall, B.S, B.A. Graduate Program in Integrated Biomedical Sciences The Ohio State University 2013 Dissertation Committee: John Gunn, Advisor Jesse Kwiek Robert Munson Daniel Wozniak

2 Copyright by Joanna Marie Marshall 2013

3 Abstract: Salmonella Typhimurium, similarly to other enteric pathogens, produce a group IV O-antigen (O-ag) capsule exhibiting structural resemblance to the lipopolysaccharide (LPS). Polysaccharide capsules are known virulence factors of many bacterial pathogens, facilitating evasion of immune recognition and systemic dissemination during acute infection. Capsular polysaccharides can also aid establishment of chronic infection and environmental spread of disease by increasing bacterial surface adherence, selfaggregation and resistance to stress, all of which are critical steps in the continued cycle of infection. In this work, we sought to further characterize the O-ag capsule, a recently described surface polysaccharide of Salmonella Typhimurium. We have established that functional surface assembly of the O-ag capsule shares several common enzymatic pathways with lipopolysaccharide (LPS) biosynthesis and that disruption of the O-ag capsule does not interfere with production of LPS. Whole cell imaging has revealed heterogeneous capsular expression within sessile and biofilm bacterial populations and indicates that capsular expression may shield recognition of the LPS O-ag by specific antibodies. Additionally, we have observed that absence of the O-ag capsule results in dysregulated surface expression of highly immunostimulatory phase I flagellin (FliC) and increases bacterial susceptibility to killing by human serum. Due to their high level of attenuation, strains lacking the alternative RNA polymerase sigma factor RpoS (ΔrpoS) have been extensively investigated for possible future inclusion in a live vaccine strain and the currently licensed live-attenuated vaccine strain for Salmonella Typhi TY21A is RpoS-deficient. Efforts to produce attenuated vaccines with RpoS-deficient phenotypes have focused on achieving a balance of attenuation and immunogenicity, reporting that a lack of systemic replication of rpos mutants necessitates multiple doses of >10 10 CFU to achieve moderate protection against subsequent infection. We have observed in a murine model of acute infection that ΔyihO ii

4 mutants carrying an inactivating mutation in rpos exhibit efficient systemic dissemination without restoration of virulence and that inoculation with this live attenuated strain is able to provide high-level, long lasting protection against fatal infection after a single oral dose. Additional studies were conducted to better understand the role of the O-ag capsule and Vi-ag capsule in chronic disease and gallstone biofilm formation, which has been shown to facilitate carriage. In order to better characterize the extracellular matrix, we have developed a method of sectioning Salmonella biofilms grown ex vivo on the surface of human gallstones, permitting direct visualization of individual matrix components. We have established that although neither O-ag capsule nor Vi-ag capsule are required for systemic dissemination or gallstone/cholesterol biofilm formation, both exhibit characteristics in vitro that may facilitate these infectious stages. Furthermore, both capsules are abundantly detected within the extracellular matrix of gallstone biofilms, indicating a novel non-structural role for these polysaccharides in chronic infection. Collectively these data indicate that the O-ag capsule is a distinct accessory surface polysaccharide expressed by a subset of cells, potentially facilitating acute and chronic infection by mediating evasion of detection and killing by the host immune system. iii

5 Dedication: I dedicate this work to my family: to my grandparents, for teaching me the value of hard work, determination and kindness; to my father and brother, for teaching me how to experiment, explore, adventure and create; above all, to my mother, for giving me the love, confidence and means to do all I ve ever hoped for. iv

6 Acknowledgments: I would like to thank my advisor, Dr. Gunn, for the many hours of listening, teaching, editing, discussion and kind direction that have allowed me to reach this point. I also thank my dissertation committee for their excellent guidance, as well as the entire CMIB/MI&I for their collective friendship and assistance throughout my research. I am especially thankful to Jake, Kyle, Dawn, Jesse and Mia for making work a fun place to be. I give my genuine and heartfelt gratitude to Anice, Shilpa and Nrusingh for leading by example and always providing reliable, intelligent, insightful advice on all matters, scientific and otherwise. To Simon, Helyn, Emily, Andy, Mairin, Mary, Marlis, Tim, Nash, Hamilton, Leah, Eowyn and everyone who has shared my laughter, smiles, happiness and tears over the past 5 years (or much longer). You have made my world a bright, vibrant and beautiful place to be and I thank you all from the bottom of my heart. Finally, I d like to thank Forrest for being my truest friend, confidant, sounding board, editor, scientific advisor and general pal. I most certainly would not be here without you. v

7 Vita: B.S. Microbiology, The Ohio State University B.A. French, The Ohio State University Teaching Assistant and Public Health Intern 2005 to Undergraduate Research Assistant 2008 to present... Graduate Research Associate Publications 1. Gonzalez-Escobedo, G., Marshall, J.M., Gunn, J.S., 2011, Chronic and acute infection of the gallbladder by Salmonella Typhi: understanding the carrier state. Nat. Rev. Microbiol. 9, Marshall, J.M. and Narayan, R.J., 2009, Two Photon Polymerization: An Emerging Method for Rapid Prototyping of Ceramic Polymer Hybrid Materials for Medical Applications. JACerS.bul. vol 88, n 5, pp Major Field: Integrated Biomedical Sciences Specialization: Microbial Pathogenesis Fields of Study vi

8 Table of Contents: Abstract... ii Dedication... iv Acknowledgments... v Vita... vi Table of Contents... vii List of Tables... viii List of Figures... ix Chapter 1: Introduction... 1 Chapter 2: The O-ag Capsule of Salmonella Typhimurium facilitates serum resistance and regulated surface expression of phase I flagella Chapter 3: The O-Ag Capsule: Biofilms and Chronic Disease Chapter 4: Discussion of Dissertation Research References vii

9 List of Tables: Table 1. Salmonella enterica serovars and plasmids used in this study Table 2. Antibodies and sources Table 3. Oligonucleotide primers used in this study viii

10 List of Figures: Figure 1. Modeling of LPS O-ag and the O-ag capsule Figure 2. Predicted functions and localization of O-ag capsule operon-encoded proteins Figure 3. Detection of O-ag capsule using adsorbed polyclonal antisera generated against purified S. Enteritidis capsular polysaccharide Figure 4. Visualization of LPS and O-ag capsule Figure 5. Preliminary screening of O-ag operon mutants for production of LPS and O-ag capsule Figure 6. Detection and visualization of LPS and O-ag capsule on whole S. Typhimurium bacterial cells Figure 7. Visualization of capsular localization to bacterial cell surface and observation of heterogeneous expression within culture population Figure 8. Association of H 2 O 2 resistance, RDAR morphotype and RpoS status Figure 9. Resistance of O-ag capsule deficient mutants to killing by human serum.. 58 Figure 10. Analysis of secreted and cell associated FliC in S. Typhimurium Figure 11. Detection of cell-associated FliC and FljB in S. Typhimurium ix

11 Figure 12. Oral virulence of S. Typhimurium strains and protection following inoculation with JSG Figure 13. Systemic colonization and gallstone biofilm forming abilities of S. Typhimurium in murine chronic infection Figure 14. S. Typhimurium O-ag capsule ΔyihO mutants do not exhibit altered biofilm formation on cholesterol or polystyrene in vitro Figure 15. Photomicrographs of sectioned human gallstones with surface associated biofilms of S. Typhimurium Figure 16. IF staining controls demonstrating minimal non-specific fluorescence Figure 17. Visualization of polysaccharide antigens in DAPI counterstained Salmonella biofilms grown on human gallstones Figure 18. Flagella and CsgA detection in bacterial microcolonies and whole cell lysates Figure 19. Visualization of flagella in DAPI counterstained gallstone biofilm sections x

12 Chapter 1: Introduction Bacterial surface-associated polysaccharides are abundant and highly diverse in structure and function. They can serve as potent danger signals to activate the host immune system or provide a stealthy cover from recognition and attack. In human tissues and in the environment, polysaccharides often facilitate the process of surface adherence, self-aggregation and resistance to environmental stressors; all of which are critical steps in the continued cycle of infection. In this work, we sought to further characterize the O- antigen (O-ag) capsule, a recently described surface polysaccharide of S. Typhimurium. We have established that functional surface assembly of the O-ag capsule shares several common enzymatic pathways with lipopolysaccharide (LPS) biosynthesis and that disruption of the O-ag capsule does not interfere with production of LPS. Whole cell imaging has revealed heterogeneous capsular expression within sessile and biofilm bacterial populations and indicates that capsular expression may shield recognition of the LPS O-ag by specific antibodies. Additionally, we have observed that absence of the O- ag capsule results in dysregulated surface expression of highly immunostimulatory phase I flagellin (FliC) and increases bacterial susceptibility to killing by human serum. In order to better characterize the extracellular matrix, we have developed a method of sectioning Salmonella biofilms grown ex vivo on the surface of human gallstones, permitting direct visualization of individual matrix components. We have established that although neither 1

13 O-ag capsule nor Vi-ag capsule are required for gallstone/cholesterol biofilm formation, both are abundantly detected within the extracellular matrix, indicating a novel role for these polysaccharides in chronic infection. Using in vivo animal modeling of acute and chronic disease we show that the O-ag capsule, like the Vi-ag capsule of S. Typhi, is not required for systemic dissemination or cholesterol/gallstone biofilm formation, but exhibits several characteristics in vitro, which may greatly facilitate these stages of infection. Furthermore, we demonstrated that O-ag capsule deficient mutants carrying an inactivating mutation in rpos exhibit improved systemic dissemination without restoration of virulence, and that inoculation with this live attenuated strain is able to provide high-level, long lasting protection against fatal infection after a single oral dose. Salmonella: There are currently 2,579 known serovars within the genus Salmonella, a comprehensive list of which is maintained by the World Health Organization Collaborating Centre for Reference and Research on Salmonella (WHOCC-Salm) 1. The Kauffmann-White scheme differentiates between serovars on the basis of antigenic variability of LPS O-antigens (O) and flagellar (H) antigens and defines serovars based upon this antigenic formula. Genetically and phenotypically the genus Salmonella are a highly variable group of organisms consisting of 2 distinct species S. enterica and S. bongori. Within S. enterica, there are 6 subspecies (S. enterica subsp. enterica, S. enterica subsp. salamae, S. enterica subsp. arizonae, S. enterica subsp. diarizonae, S. enterica subsp. indica, and S. enterica subsp. houtenae) however S. enterica subsp. enterica accounts for over 2,300 of the known Salmonella serovars and is the only subspecies associated with disease in mammals 2. Salmonella enterica and Escherichia coli are estimated to have diverged from a common ancestor over 130 2

14 million years ago 3 and yet over 80% of the coding regions within their genomes remain identical at the nucleotide level. There are 4,489 genes annotated within the S. Typhimurium chromosome, 71% of which have a close ortholog in E. coli 4. Many of the genetic loci which differentiate Salmonella from E. coli appear to have been acquired through horizontal gene transfer and are termed Salmonella Pathogenicity islands (SPI s) due to their encoded virulence functions 5. It is believed that subsequent to this divergence, S. Typhi underwent extensive genomic degradation and reductive evolution (losing approximately 11% of the genes encoded by S. Typhimurium), which resulted in its highly host-specific nature 6,7. The recent publication of the genome of epidemic invasive isolate DT313 of S. Typhimurium has demonstrated genomic degradation and host adaptation resembling that of S. Typhi. Many of the virulence related pseudogenes, which seem to have contributed to the host-adaptation and increased systemic fitness of S. Typhi, have undergone similar transformations in DT313 8, potentially indicating the emergence of a new host adapted and highly virulent subspecies of Salmonella 8. Epidemiology: Although the vast majority of S. enterica subsp. enterica serovars have the capacity to cause disease, 90% of human infections are caused by fewer than 30 serovars and 50% of human infections are attributable specifically to serovars Typhimurium, Enteritidis, and Newport 9. The US Centers for Disease control compiles a record of laboratory confirmed human cases of Salmonella which indicates that in 2009 (the most recent year for which data are available) there were 7,144 cases of Salmonella Enteritidis, 6,120 cases of S. Typhimurium and 432 cases of S. Typhi accounting for 17.5%, 15% and 1.1% respectively of the 40,828 confirmed cases of Salmonellosis 10. It is believed that fewer that 1 in 30 cases of Salmonella are clinically confirmed; therefore 3

15 the annual estimate for incidence of Salmonelloses in the U.S. is 1.2 million new infections per year Globally, Salmonella enterica are estimated to be responsible for over 93 million new infections annually 17, 21 million of which are caused by Salmonella Typhi 18. In spite of decades of research on pathogenesis, Salmonella enterica continues pose a major threat to global public health 18 and diagnosis, treatment and prevention have proven particularly difficult in developing countries. Clinical isolates of both typhoidal and non-typhoidal Salmonella have now been reported with high-level resistance to virtually every available first line antibiotic 19. While there have been major public health efforts aimed at vaccination, the most widely distributed form of the vaccine elicits protective immunity in only 50-70% of patients and is not approved for use in young children, leaving a significant portion of the population unprotected from infection 20,21. Pathogenesis and virulence mechanisms of Salmonella enterica: Over 60 genes are required for virulence of Salmonella Typhimurium 22. Many of these genes relate to fundamental steps in the pathogenic process such as environmental survival, passage through the acidic environment of the stomach and invasion into gastric epithelial cells. Subsequent systemic dissemination relies upon efficient epithelial invasion and the ability to survive and replicate in the bloodstream and phagocytic cells. Ingestion and initial infection: Infection with Salmonella enterica begins most commonly through the ingestion of food or water contaminated with fecal matter from an infected individual. Upon ingestion, the bacteria is able to withstand the highly acidic ph of the stomach and once in the small intestine, employ adhesive fimbriae to attach to host 4

16 cells and subsequently gain access to the intestinal epithelium and the underlying lymphoid organs through a variety of mechanisms. Entry into host cells: Salmonella employ the SPI-1 encoded type III secretion system (T3SS1) to induce their own uptake by the typically non-phagocytic cells of the GI epithelium. Translocation of bacterial effectors into the host cytosol results in transient induction of a rearrangement in the host cytoskeleton resulting in bacterial uptake 23. It is believed that the primary method of invasion is through endocytic or SPI-1-mediated entry into microfold (M) cells 24, a specialized subset of intestinal lymphoid follicleassociated epithelial cells which normally function to sample the contents of the intestinal lumen 25. Groupings of M cells, termed Peyer s patches, act as sentinels facilitating immune tolerance to food antigens and commensal flora and localizing pathogen uptake to the underlying follicle dome which is rich in mucosal immune cells. Many enteropathogens are able to exploit this function to traverse the mucosal epithelium and it has recently been demonstrated that Salmonella not only target these M-cells, but also induce their differentiation within the host 26. The subepithelial dome of Peyer s patches is occupied by macrophages, T-cells, B-cells and resident dendritic cells, which provide an additional mechanism for bacterial translocation across the epithelium by extending dendrites into the lumen and directly phagocytosing bacteria into the subepithelial dome Salmonella may also gain direct access to the bloodstream if the host intestinal barrier function is compromised. Initial response to infection: The host inflammatory response to Salmonella infection largely dictates the clinical course of infection (localized gastroenteritis vs. disseminated infection) and the bacterium has evolved many mechanisms to evade or manipulate this 5

17 response 28. Early control of infection relies largely on innate-immune mediated recognition of pathogen associated molecular patterns (PAMPS) by host cell pattern recognition receptors (PRRs). PRRs such as TLR4 and TLR5 and TLR9 detect microbial ligands (TLR4/LPS and TLR5/Flagellin, TLR9/DNA) 29. TLR/NLR-mediated detection of PAMPS causes recruitment of neutrophils and macrophages and production of proinflammatory cytokines 30,31 leading to increased production of antimicrobial peptides and reactive oxygen and nitrogen species. TLR4 and complement receptor 3 expressed by recruited neutrophils interact with bacterial ligands leading to phagocytosis and an oxidative burst 32. Although pro-inflammatory cytokines are also produced in response to the activity of secreted bacterial effectors, it is thought that the ability to interfere with the host detection of surface associated PAMPS- specifically LPS and flagellin- is key to causing invasive disease. One of the defining characteristics of gastroenteritis caused by nontyphoidal Salmonella is the presence of neutrophils in the feces. Termed inflammatory diarrhea this condition is indicative of a rapid and efficient innate immune detection of Salmonella invading the GI mucosae 33 resulting in neutrophil recruitment (IL-8) 34 and macrophage activation (INF-γ). This process, which is mediated predominantly by host detection of flagellin 35 and LPS 36, results in clinical symptoms of gastroenteritis and ultimately restricts the infection from disseminating systemically. Patients infected with S. Typhi rarely exhibit evidence of neutrophil extravasation to the GI lumen, indicating a dampened host response, a pathogenic effect which is attributed in part to production of the Vi-ag capsule 37. Intracellular lifestyle: Upon entry into host cells, Salmonella employ SPI-II T3SS (T3SS2) secreted effectors to avoid killing by delaying vacuolar acidification 38, 6

18 inhibiting NF-κB mediated inflammatory signaling, delaying host cell apoptosis 39 and redirecting the Salmonella containing vacuole (SCV) to a perinuclear location to gain nutrients 40. Bacteria then replicate within the SCV 41 or in epithelial cells, they may escape and replicate in the host cytosol 41. In this manner, Salmonella profoundly interfere with host defense and generation of an efficient cell-mediated immune response, instead disseminating in phagocytic cells via the lymphatics and circulatory system to the liver, bone marrow and spleen resulting in acute disease Clinical features of Infection: In otherwise healthy hosts, the severity of symptoms of Salmonella infection are largely a function of the infecting serovar. Infection with nontyphoidal salmonellae such as S. Typhimurium, S. Enteritidis or S. Newport typically cause self-limiting enterocolitis while typhoidal serovars S. Typhi and S. Paratyphi are more adept at systemic dissemination. However, manifestations of disease are a result of a complex interplay between host and bacterial factors- and exceptions to the rules occur. In certain host environments S. Typhi, normally associated with more aggressive disease, can result in decades-long asymptomatic carriage. Conversely, in a permissive host environment infection with S. Typhimurium, can result in overwhelming and rapidly fatal bacteremia. As discussed above, in healthy individuals, infection with nontyphoidal Salmonella is associated with rapid host detection of bacterial PAMPS leading to IL-8 mediated neutrophil recruitment, ultimately resulting in the symptoms of inflammatory diarrhea. It is thought that production of highly bactericidal antimicrobial peptides and reactive oxygen/nitrogen species by recruited neutrophils is primarily responsible for restricting nontyphoidal Salmonella to the GI lumen 46. 7

19 Typhoid: Endemic throughout much of Southeast Asia and sub-saharan Africa, enteric fever (or typhoid fever) is a severe febrile illness caused by infection with S. Typhi or S. Paratyphi 47. Onset of symptoms occurs within 6-30 days with clinical presentation most commonly including nonspecific symptoms such as high fever, headache and malaise. The most common sites of acute infection are the ileum, liver, spleen, bone marrow and gallbladder 45. In endemic regions, diagnosis is often made based upon clinical presentation alone and is frequently confused with Malaria. Laboratory diagnostic methods have variable efficacy depending on the phase of disease. The Widal serum agglutination assay has low sensitivity and specificity while bone marrow culture and nested PCR are the most sensitive (80%) diagnostic methodologies 48,49 but are not suitable for many clinical settings. With effective diagnosis and appropriate antibiotic treatment, recovery from typhoid generally occurs within 21 days 50. The most common severe complications of infection is intestinal perforation leading to overwhelming septicemia, the incidence of which (2.5%) is not affected by antibiotic treatment. Without effective medical care, illness is severe, recovery is slow and case fatality rates (in otherwise healthy individuals) are as high as 20-50% Chronic Typhoid: Following resolution of clinical disease, 3-5% of patients progress to become asymptomatic chronic carriers and continue to intermittently shed viable bacteria in their feces for over 1 year Epidemiologically, these carriers are of critical importance as S. Typhi is a human-restricted pathogen, making infected individuals the sole reservoir for continued spread of the disease. The asymptomatic nature of carriage and lack of efficient means to resolve infection makes identification, and treatment of carriers within a population extremely difficult 59. As a result, typhoid carriers are 8

20 frequently unaware of the risk they may pose to others. Furthermore, in the absence of clinical symptoms of disease, justification of costly, painful or dangerous interventions is difficult. Gallbladder Carriage: Chronic infections occur most frequently in the gallbladder 58 and are highly associated with the presence of cholesterol gallstones 60. A study conducted in a typhoid endemic region of Nepal using PCR-based detection of S. Typhi in tissue samples from over 400 randomly selected cadavers demonstrated that overall, 8.2% were positive for S. Typhi and of these, hepatobiliary tissues were positive in 85% of samples 57. As the site of bile concentration and storage, the gallbladder is an environment that is habitable exclusively by organisms that are resistant to bile's detergent-like properties such as Klebsiella spp., Escherichia coli, Enterobacter spp. and Salmonella spp. 61,62. Salmonellae, along with other enteric pathogens, are not only highly resistant to bile, but also respond to the presence of bile by alternately regulating the expression of many resistance-related genes 63. Salmonella spp. are able to colonize the gallbladder during the acute phase of disease, most likely through vasculature or ducts emanating from the liver 64. Once established, chronic infections cause periodic shedding of bacteria in feces 65, can persist for decades 66 and are notoriously difficult to treat with antibiotics 20,67. Surgical resection of the gallbladder is the most successful method to resolve chronic infection 68. Although effective, this procedure carries a high rate of complications in developing countries 69 and is often not appropriate, particularly in light of the absence of clinical symptoms and unreliable diagnosis associated with chronic carriage. 9

21 A modern Typhoid Mary : Mary Mallon ( ), otherwise known as Typhoid Mary, was an asymptomatic carrier of S. Typhi who worked as a cook in New York during the early 20 th century 70. Although a number of outbreaks were associated with her food, she believed that reports of her typhoid carriage were mistaken, as she did not experience symptoms of illness. Ms. Mallon repeatedly defied the orders of public health officials by returning to her work as a cook and in this manner it is thought that she spread disease to as many as 50 people 71. Ms. Mallon was ultimately arrested and held for over two decades in involuntary isolation on North Brother Island, NY until her death in Typhoid Mary became a notorious symbol of the potential health threat posed by asymptomatic typhoid carriers as well as a commonly cited example of the conflict between individual patient rights and the protection of public health. Although public education and improved sanitation has done much to reduce the global burden of typhoid fever, a 2010 study conducted by the National Institute of Communicable Diseases (NICD) in South Africa highlighted the ongoing role of carriers by tracking repeated outbreaks of the same strain of S. Typhi, which occurred in 1993, 2005, 2007 and Over 1600 people were sickened and infections were ultimately attributed to a single chronic carrier who either had not been treated or for whom treatment had repeatedly failed 73,74. The NICD employed a variety of techniques from mapping sewer drainage to mandating aggressive treatment for any potential carrier. Eventually after much expenditure of resources and little success, financial constraints halted the project and despite continued outbreaks, no further public health efforts are currently anticipated 74. Invasive nontyphoidal salmonellosis (ints): As we have seen, typhoidal serovars such as S. Typhi, and S. Paratyphi are highly adept at directly and indirectly subverting the 10

22 host immune response to avoid clearance from the GI tract and replicate within host macrophages, making them capable of causing systemic disease regardless of host immune status. Nontyphoidal salmonellae (NTS) elicit a robust inflammatory response upon initial interaction with the gastrointestinal (GI) mucosae in immunocompetent hosts, resulting in massive neutrophil recruitment to the GI lumen. This response manifests clinically as acute gastroenteritis and is sufficient to restrict the infection from disseminating systemically. Although invasive nontyphoidal salmonellosis has long been identified as an HIV-associated opportunistic infection 75, in recent years there has been a dramatic increase in reported cases in sub-saharan Africa 8, with ints now accounting for nearly 1/3 of clinically diagnosed bacterial bloodstream infections 76. These infections are highly associated with immune dysfunction and advanced HIV and are caused almost exclusively by S. Typhimurium or S. Enteritidis 77. Treatment of ints infections is complicated by a non-specific clinical presentation, high-level antibiotic-resistance and lack of adequate medical facilities in many affected regions, resulting in estimated case fatality rates as high as 80%. Even with adequate medical facilities, proper diagnosis and effective antibiotics, the case fatality rate in adults is 25% 78. Although advanced HIV increases the overall likelihood of developing any bacteremia by 5.3 fold, the incidence of ints is increased 48.5 fold relative to the general population 77, implying not simply an increase in host susceptibility but additional bacterial factors which may be particularly advantageous for dissemination in the HIV + host. Epidemiological studies have subsequently indicated that many ints infections are caused by a genomically degraded, host adapted strain of S. Typhimurium which may be transmitted from humanto-human, as opposed to the zoonotic transmission commonly reported with NTS

23 The primary mechanisms of control of extracellular bloodstream infection by nontyphoidal Salmonella are oxidative, phagocytic and complement-mediated killing; all of which rely on the generation of specific antibodies directed at targets on the bacterial surface 46. While O-antigens, flagella and outer membrane proteins are common targets of host immunity, it has been observed that binding of specific antibodies to outer membrane proteins is required for optimal complement mediated killing, phagocytosis and oxidative burst 83. Studies investigating the mechanism of deficient killing of Salmonella by sera of patients with ints has demonstrated an overwhelming production of low specificity IgG directed at galactose-containing surface antigens of Salmonella spp. It has been shown that antibodies directed at outer membrane proteins are present in these sera and able to kill Salmonella but O-ag specific IgG must first be removed from the serum. The deficient bactericidal activity is thought to be a result of IgG and complement deposition on high molecular weight polysaccharides located on the cell surface, resulting in assembly of the membrane attack complex at locations too distal from the membrane to result in efficient lysis. Furthermore, this process appears to form a barrier which sterically hinders membrane access of bactericidal OMP-specific antibodies 84 and hinders these antibodies from inducing efficient FcγR-mediated phagocytosis 85. Although the above mechanism certainly contributes to the susceptibility of HIV-infected patients to ints disease, numerous additional factors are likely to be involved as well. It is known that gut CD4 + T-cells, specifically T h 17 cells, are a critical contributor to protection against invasive salmonellosis 8,86,87. Intestinal CD4 + T-cells are a preferential target of HIV-1 and are disproportionately depleted throughout the course of infection 88. HIV infection is also commonly associated with intestinal enteropathy, specifically 12

24 disruption of tight junctions and loss of GI epithelial integrity 89. This results in reduced barrier function against invading pathogens, ongoing translocation of microbial products and elevated levels of circulating LPS; all thought to contribute to chronic immune stimulation and hypergammaglobulinemia 90,91. Even in healthy individuals, specificity of LPS and O-antigen directed IgG is low and persists in serum for extended periods of time 92. Among otherwise healthy adults with a single culture positive incident of S. Typhimurium or S. Enteritidis gastroenteritis, serum IgG cross reactivity with O-antigens of the other serovar is observed in 93% and 78% of samples respectively, which persist and become decreasingly specific for 12 months following infection 93. In a separate study, it was shown that anti-s. Typhimurium LPS antibodies are induced in 54% of individuals following infections with non-salmonella enteropathogens such as E. coli, Y. enterocolitica, Helicobacter and Campylobacter In short, even in healthy individuals, antibodies targeting LPS can exhibit low levels of specificity but are working in tandem with supremely efficient innate and adaptive mechanisms of infection control in the gut. However, in the context of advanced HIV, invading bacteria also benefit from compromised gut barrier function, severe T-cell depletion, B-cell dysfunction, chronic immune activation and hypergammaglobulinemia, which collectively result in lifethreatening inability to control systemic salmonellosis. Polysaccharides and Salmonella infection: Bacterial polysaccharides are a major component of the cell surface and are often the first structure to interact with the host immune system. Their highly exposed nature and importance in pathogenesis has made them key targets of host immunity and many therapeutic drugs 96. Although polysaccharides are generally poor immunogens, bacterial glycoepitopes are most 13

25 commonly associated with capsules and LPS O-antigens 97 and this fact has been successfully exploited to develop vaccines against Neisseria meningitides, Haemophilus influenzae, Streptococcus pneumonia and Salmonella Typhi Control of extracellular S. Typhimurium is mediated primarily by surface deposition of OMP and polysaccharide-specific antibodies leading to improved complement-mediated killing, phagocytosis and oxidative burst 8. Salmonella enterica produce a number of surface associated polysaccharides. In addition to LPS and O-ag capsule discussed below, additional operons encode for the production of enterobacterial common antigen (wec operon), colanic acid (wca operon), and cellulose (bcs operon) The importance of multiple bacterial surface structures in resistance to killing by antimicrobial peptides and human serum has been recognized for decades although their relative necessity is continuously revised as our understanding develops 103,104. In both Salmonella and E. coli, it is believed that specifically the longer chain LPS (>40 O-antigen repeating units) impart the majority of the serum resistance to a particular strain and that capsular polysaccharides (such as O-ag capsule, Vi-ag capsule, colanic acid and enterobacterial common antigen) assist in this function but are of secondary importance 104. Lipopolysaccharide is the primary constituent of the outer leaflet of the outer membrane in Gram-negative bacteria. This molecule consists of 4 distinct regions: the lipid A (endotoxin), inner and outer core and the O-ag which is composed of a variable number of tetrasaccharide repeating units 105. While the lipid A and core region are highly structurally conserved, the O-ag exhibits enormous interserovar variability in terms of its monosaccharide composition and linkage 105. The process for biosynthesis of LPS O-ag and O-ag capsules involves many homologous enzymatic functions. In some bacterial 14

26 species, biosynthetic pathways for production of LPS O-ag and O-ag capsule differ by only one or two genes, which are able to modify the LPS O-ag subunits sufficiently enough to create a distinct capsular structure 106. There have been variable reports as to whether S. Typhimurium and S. Enteritidis produce very long LPS O-antigen, an O-ag capsule or both and similarly conflicting reports have arisen in other bacteria producing group IV capsules, likely due to their extreme structural similarity to LPS 107. Biosynthesis of group IV O-ag capsules, like O-ag of the LPS, is dependent on the activity of Wzy polymerase. Assembly begins in the cytoplasm and involves glycosyltransferase linkage of sugar-phosphate precursors to a membrane lipid carrier (UDP). In S. Typhimurium, subsequent glycosyltransferases link 3 additional monosaccharides on to create the O-ag tetrasaccharide repeat unit which is then flipped out of the cytoplasm into the periplasm via the activity of the inner-membrane bound Wzx. Wzy polymerase then links the individual repeat units into the complete O-ag. In the case of LPS, this molecule would then be ligated to the outer core and displayed on the cell surface through the activities of Waa ligase and the Wza translocon complex 108. A variety of additional accessory molecules can participate in and potentially modify the biosynthetic assembly process. In the case of LPS biogenesis, it is known that a copolymerases (Wzz and Wzz fepe ) act as a chain-length determinants and that mutations in various Wzz genes result in a shift in to increase the proportion of short O-ag 109, while loss of Wzx can result in lethal cytoplasmic accumulations of O-ag precursors 110. Capsular Polysaccharides: One major strategy employed by bacteria in order to evade or counteract host immunity is by altering surface-exposed antigens through modification of structure, expression or exposure of that antigen 111 and capsules often greatly facilitate 15

27 this process. Bacterial capsules are known to aid in virulence of many organisms through mechanisms including inhibition of phagocytosis, increasing resistance to host antimicrobial peptides and oxidative burst and inhibition of innate and adaptive immune recognition 112. Bacterial capsules are organized into 4 types on the basis of their structure and the genetic loci required for their assembly. Group IV or O-ag capsules are very long chain polysaccharides composed of sugar repeating units which are highly similar to the O-ag of the LPS 113. The general model for group IV capsular polysaccharide structure and assembly (Fig. 1) involves re-arrangement of LPS O-ag precursors into a much larger structure termed the O-ag capsule. It is thought that the ability of bacteria to produce a large, surface structure providing additional O-antigenic structural diversity while lacking highly immunostimulatory moieties such as lipid-a and core polysaccharides provides bacteria with additional resistance to host killing and recognition. Although the term capsule implies the presence of an anchor to the cell membrane, in the case of O-ag capsules, the identity of this anchor and the precise mechanism of attachment and display at the cell surface is unclear and likely variable. Possible capsule anchors are phospholipids other than lipid A, lipid A (in which case they the capsule is termed K LPS ), outer membrane proteins or through stabilizing interaction with LPS 108, Other organisms capable of producing O-ag capsules include E. coli and Vibrio as well as Shigella sonnei, Francisella tularensis and Campylobacter jejuni 107, Functional analyses of the role of O-ag capsules during infection with these organisms demonstrates that they can facilitate serum resistance, aid in systemic dissemination, exhibit shielding of LPS, inhibit immune-recognition of the T3SS apparatus and delay apoptosis of infected macrophages 120,122. Although the work 16

28 described herein represents the first report of a role for the O-ag capsule of S. Typhimurium in acute disease, these previous studies on the O-ag capsules of other organisms clearly imply a role for this polysaccharide in acute disease. Vi-ag capsule of S. Typhi: Salmonella Typhi produces a surface polysaccharide known as the virulence antigen or Vi-ag capsule, which is composed of α-1,4(2-deoxy)-2-nacetylgalacturonic acid. Genes responsible for the production of this capsule are carried on a region of the S. Typhi chromosome known as SPI Although the Vi-ag capsule is not required for infection in vivo, it is known to facilitate systemic virulence and is thought to be particularly advantageous for survival in the blood, as over 98% of S. Typhi bloodstream isolates produce the Vi-ag capsule 124 and antibodies generated against this structure are able to provide protective immunity against typhoid infection 125. A single modification on this structure, variable O-acetylation, is responsible for its antigenicity and Vi-ag which has had this O-acetyl group removed is unable to illicit an immune response. This acetylation is added by a gene which is located outside of SPI-7 and in spite of its obvious importance, the O-acetyltransferase responsible for this modification has not yet been identified In vivo, the Vi-ag capsule increases resistance to killing by human serum by interfering with deposition of complement C3 on the bacterial surface. This facilitates bacterial survival by decreasing assembly of the membrane attack complex as well as reducing complement receptor-mediated phagocytosis 127. Vi-ag is also capable of directly binding to intestinal epithelial cells and interfering with inflammatory signaling, ultimately resulting in reduced IL-8 mediated neutrophil recruitment to the site of infection

29 Figure 1. Modeling of LPS O-ag and the O-ag capsule: Left) Lipopolysaccharide (LPS) is a major component of the Gram negative outer membrane, composed of several distinct regions; Lipid A, core and the O-ag repeating unit. The term LPS O-ag refers to the distal end of the LPS molecule, which carries 5-50 tetrasaccharide O-ag repeating units. The ability to produce LPS with a variable number of O-ag repeating units (particularly longer length O-ag) is a wellestablished virulence requirement. Right) O-ag Capsule: More recently, it has been discovered that Salmonella may also produce a capsular polysaccharide composed of the same monosaccharides which comprise the LPS O-ag, termed O-ag capsules. Typically, O-ag capsules are composed of a very long chain of O-ag repeating units (>2,000) anchored into the outer membrane. The much greater length of O-ag capsule molecules can provide protection from bacterial killing and although the carbohydrate composition is highly similar to the LPS O-ag, O- ag capsules frequently exhibit alterations in glycosidic linkages or substitutions, which confer immunologic distinction from the O-ag of the LPS. 18

30 The O-ag capsule: Studies aimed at characterizing the extracellular polymeric substance (EPS) produced by S. Enteritidis revealed the presence of a previously uncharacterized high molecular weight polysaccharide 128. Further studies determined this material to be highly structurally similar to the O-ag of the LPS but immunologically and biosynthetically distinct 129. The polysaccharide was determined to be a group IV O- antigen capsule. The O-ag gene cluster is composed of two divergent operons designated ysha-yihu and yihvw 130. The sequence of these operons is over 98% conserved among some of the most frequently isolated strains of pathogenic salmonellae including serovars Typhimurium, Typhi, Paratyphi A, B and C, Enteritidis, Dublin, Choleraesuis, Newport, Agona, Gallinarum, and Heidelburg. The encoded genes are homologous to enzymes involved in transport and biosynthesis of oligosaccharides (Fig. 2). The first gene in the O-ag operon cluster encodes YihU, which exhibits similarity to an NAD-binding oxidoreductase and may activating O-ag sugar residues for subsequent modification by YihQ- YihT isomerase and glucosyltransferase activity. YihO and YihP are predicted inner membrane proteins, exhibiting predicted structural similarity to Wzx translocases and Wzy polymerases, possibly transferring the O-ag capsule repeats into the periplasm and polymerizing them prior to export via outer membrane β-barrel lipoprotein YshA (OmpL). Characterization of the purified polysaccharide indicated it to be a homopolymer of approximately 200kDa composed primarily of galactose, mannose and rhamnose with minor amounts of tyvelose and glucose 130. The O-ag capsule was determined to be at least 3000 units in size, in contrast to the LPS, which rarely displays more than 70 O-ag repeating units 131. Further studies conducted with S. Enteritidis demonstrated that yiho and yihq are required for production and translocation of this 19

31 capsule and that capsule production is co-regulated by central regulator AgfD in conjunction with other extracellular matrix components 129. Further work conducted in our lab using transcriptional analysis of the putative promoter region upstream of yihu indicated that in S. Typhimurium, bile highly upregulated O-ag operon expression even in the absence of AgfD. It was reported that although S. Typhimurium readily forms biofilms on the surface of gallstones and cholesterol-coated surfaces even in the absence of other known biofilm components such as cellulose (ΔbcsE) and colanic acid (ΔwcaA), strains with deletions in yiho were deficient in biofilm formation 132. Although the genomes of serovars Typhimurium and Enteritidis are not identical, the O-ag capsule operon (ysha-yihu) is highly conserved throughout Salmonella species and the LPS O-ag of the two serovars share a common trisaccharide backbone of mannose, rhamnose and galactose

32 Figure 2. Predicted functions and localization of O-ag capsule operon-encoded proteins. In silico prediction of O-ag genes is complicated by low conservation of structure-function relationship. Analysis of homologues and conserved domains reveal putative functions for O-ag capsule operon genes broadly related to carbohydrate assembly and transport. O-ag assembly may be initiated by the activity of YihU oxidoreductase. Subsequent activity of YihT, YihS, YihR, YihQ isomerases and glucosyltransferases may modify monosaccharide linkages or substitutions. YihO and YihP are predicted inner membrane proteins with 12 transmembrane domains characteristic of Wzx translocase/wzy polymerase which transfer nascent O-ag repeats into the periplasm and polymerize them prior to export to the outer membrane, a function which may be facilitated by outer membrane β-barrel lipoprotein YshA (OmpL). 21

33 Polysaccharides in chronic carriage: It has been observed in Salmonella and other species that the composition of the bacterial biofilm matrix alters significantly in response to varying environmental conditions such as ph, temperature, growth medium, mechanical stress and substratum composition 133. Various types of EPS have been identified in biofilms of Salmonella spp. including cellulose 134, colanic acid 135, curli fimbriae 136 biofilm associated proteins 137 and Vi-antigen 135 (in S. Typhi and S. Paratyphi). Microscopic analysis of Salmonella biofilms grown in vitro indicated that the bacteria are surrounded by a thick matrix, which stains strongly with ruthenium red, characteristic of an acidic polysaccharide. However, strains deficient in production of any of the known polysaccharides are still able to produce matrix-encased biofilms on human gallstones, pointing to the contribution of an alternative polysaccharide EPS component 61. Salmonella Biofilms: In addition to the serious health impact of acute typhoid disease, the chronic state also represents a public health problem. This is likely because chronically infected individuals can intermittently shed the bacteria in their feces and thus contribute to the transmission of the pathogen. In contrast to acute infection, antibiotic treatment has proven poorly effective in the resolution of chronic S. Typhi colonization of the gallbladder. Even prolonged, high-dose antibiotic therapy resolves less than 2/3 of chronic infections and treatment with ampicillin has been demonstrated to be effective only in patients without gallstones 60,138. Complete resection of the gallbladder (cholecystectomy) is able to increase this cure rate but it does not guarantee elimination of the carrier state 139. Additional bacteria foci could persist in the biliary tree or liver 97,140. Although not available to many patients, the most effective treatment available is a 22

34 combination of surgery and antibiotics 68. However, as studies emerge demonstrating the rapid acquisition of multi-drug resistance; it is becoming increasingly difficult to apply previous findings to current strains. The clinical observations of carriers with regard to resistance to antibiotic treatment, confinement to the gallbladder, gallbladder removal as the most successful therapy and evidence for the long-term evasion of the immune response, are observations consistent with biofilm-related disease 141. Biofilms are communities of microorganisms that adhere to each other and to inert or live substrates and are encased in an extracellular matrix 142. Typically considered as a response to stress, biofilms have been implicated in many chronic and acute infections 143. Salmonella spp. are known to form matrix-encased biofilms on abiotic and biotic surfaces 61,135. Early studies to investigate the ability of Salmonella spp. to form biofilms on human gallstones and cholesterol-coated surfaces (see below) indicated that formation of a robust biofilm on cholesterol is dependent upon the presence of bile 61. Bile has also been demonstrated to downregulate Salmonella pathogenicity island-1-encoded (host cell invasion). However, this is likely to be a spatio-temporal response that does not interfere with invasion when bacteria reach the mucous layer of the epithelia where a decreased bile concentration is apparent 63. Bile also slightly downregulates motility genes but this transcriptional regulation does not affect the number of flagella per bacterium 144. In vitro biofilm formation models: Various static and dynamic systems have been employed to examine Salmonella spp. biofilms. Early studies of biofilms in chronic Salmonella spp. carriage used gallstones removed from patients during cholecystectomy. Gallstones incubated with bacteria over the course of 7-14 days formed dense matrixencased biofilms; however, in controls using an alternate substrate of similar size and 23

35 shape, no such biofilm was formed 61. As an in vitro surrogate of gallstones, the tube biofilm assay (TBA) was developed for the study of biofilms on cholesterol 132. This method involves the coating of siliconized microcentrifuge tubes with cholesterol. Bacteria are incubated in these tubes for a period of 24 h, after which culture is aspirated and non-adherent bacteria are removed by washing with PBS. Using this method, the role of bile and cholesterol in the enhancement of Salmonella spp. biofilm formation was confirmed 132. Bilirubin, a major component of pigment stones was also evaluated in the TBA. Salmonella spp. formed biofilms poorly on calcium billirubinate compared with cholesterol, further indicating the specificity of Salmonella spp. binding to, and subsequent biofilm formation on cholesterol-coated tubes. The use of the TBA eliminates the dependence on human gallstones and allows for assay standardization, as the cholesterol composition of gallstones is variable 132. However, while the TBA is both economic and reproducible, the flow through system is probably more representative of the gallbladder environment. This flow through system consists of media with bile flowing through chambers at a specific flow rate. The chambers contain glass or cholesterol-coated glass coverslips. This method has been recently employed to study biofilms on cholesterol-coated surfaces, corroborating the results observed in the TBA 145. Biofilm initiation on gallstones: For successful biofilm formation on gallstones, Salmonella spp. must first access and colonize the gallbladder or biliary tract. The bacteria must then attach to the surface of gallstones as well as persist in the presence of natural host defenses 146. It is thought that bile stasis that can occur in the presence or absence of gallstones contributes to successful colonization 141. Several known bacterial 24

36 biofilm-associated factors have been investigated to determine which are critical for the formation of mature biofilms on the surface of gallstones and on cholesterol-coated surfaces 61,147. Flagella and fimbriae are two bacterial appendages that have been implicated in various microorganisms to be important for the initial stages of biofilm formation on a variety of surfaces 148,149. In S. Typhimurium, the presence of flagellar filaments but not motility (verified by a mutation in the gene mota that lacks the ability to rotate the flagellum) was necessary for biofilm formation on gallstones. In contrast, motility was required for biofilm formation on glass 147. To build upon this work and examine cholesterol-specific biofilm-required factors, a pool of transposon mutants was examined in the TBA with daily selection for planktonic bacteria 144. Using this method, 49 mutants deficient in cholesterol binding and subsequent formation of biofilms were obtained. Many of the non-adherent mutants represented transposon insertions in flagellar biosynthetic genes. Specifically, the FliC subunit was demonstrated to be necessary for initial binding to cholesterol-coated surfaces. Loss of OmpC (an outer membrane protein) also negatively affected binding to cholesterol as well as subsequent biofilm formation. Additionally, 18 of the transposon-library mutants showed insertions in the fimw gene. FimW is a negative regulator of the Type 1 fimbriae operon and its deletion confers a constitutively expressed Type 1 fimbrial phenotype. Further analysis demonstrated that a hyper-fimbriate phenotype negatively affected the initial stages of biofilm formation on cholesterol. Thus, the initial attachment phase of biofilm formation may involve a combination of flagella and outer membrane proteins that can be masked by overexpression of surface fimbriae

37 Biofilm maturation on gallstones: After the initial attachment phase, biofilm development typically involves the formation of microcolonies and a subsequent mature biofilm. Both stages are characterized by the presence of extracellular polymeric substances (EPS) that aid in biofilm structure and cell-cell interaction 150. The components of EPS that have been identified in Salmonella spp. biofilms include cellulose, colanic acid, Vi-antigen, curli fimbriae, O-antigen capsule and biofilm-associated proteins 129,135,137. Deletion of the genes encoding the S. Typhi Vi-antigen does not affect biofilm formation 61 and Vi-antigen is not present in S. Enteritidis or S. Typhimuriumhowever, these strain still are able to form robust biofilms in its absence. Cellulose and colanic acid are important for biofilm formation on abiotic and biotic surfaces including plastic, Hep-2 cells and chicken intestinal tissue 135. However, while a double mutant of cellulose and colanic acid negatively affected biofilm formation on plastic and glass, biofilm formation on gallstones was unaffected. These results demonstrated that various components of EPS are required for Salmonella spp. surface-dependent biofilm development 147. Gibson and colleagues 53 identified the polysaccharide O-antigen capsule in Salmonella spp.. This capsule has been shown to be important for environmental persistence and attachment to and colonization of plants 129,151. Mutation of the gene responsible for the synthesis of galactose, (gale) which is utilized in construction of the LPS outer core and O-antigen and is putatively involved in synthesis of the O-antigen capsule, has been shown to yield mutants that are unable to form biofilms on gallstones 61. Conversely, mutations in a gene involved in synthesis of the outer core and LPS O- antigen alone (rfad) exhibit no such defect 147. In addition, mutation of the genes putatively associated with O-antigen capsule synthesis negatively affected Salmonella 26

38 spp. biofilm formation on cholesterol-coated surfaces and gallstones. Furthermore, bile has been shown to up-regulate O-antigen capsule genes and enhance capsule expression, further suggesting an important role for the O-antigen capsule in Salmonella spp. biofilms on gallstone surfaces 132. Laboratory Study of Salmonella Enterica: Although many Salmonella serovars are able to cause disease in a broad range of hosts, S. Typhi is a human specific pathogen which does not replicate well in murine tissues 152,153. Conversely, S. Typhimurium, which causes self-limiting gastroenteritis in humans, causes a disease with pathologic features very similar to typhoid fever in infected mice. For this reason S. Typhimurium infection of susceptible mice has been widely used as an experimental model for typhoid fever 154. Animal models of acute typhoid fever typically employ BALB/c or C57BL/6 mice, termed susceptible due to a homozygous mutation of the Nramp1 allele encoding the natural resistance-associated macrophage protein (NRAMP) 155. NRAMP is expressed on the phagosomal membrane of host macrophages and dendritic cells and facilitates host iron sequestration to limit nutrient availability to intracellular bacteria 156. In animals lacking a functional Nramp1, there is a decreased ability to sequester iron from the bacterial vacuole resulting in poor host control of intracellular bacterial replication and ultimately leading to development of acute systemic disease. Oral infection of Nramp1 - / - mice with S. Typhimurium typically results in death due to overwhelming septicemia within 7-10 days 157. Animal models of chronic S. Typhi infection utilize the 129x1/SvJ mouse harboring a wild type Nramp1 allele 158. These mice do not succumb to infection and have been shown to maintain a persistent infection with S. Typhimurium for up to one year with the primary site of infection being macrophages of the reticuloendothelial 27

39 system 158. The animal model of chronic gallbladder carriage of S. Typhi employs 129x1/SvJ Nramp + / + animals fed a high-cholesterol diet for 8-10 weeks and intraperitonally infected with S. Typhimurium. Our lab has previously demonstrated that these animals become colonized chronically in the gallbladder and that carriage is mediated both by bacterial invasion into gallbladder epithelium 159 and through the development of bacterial biofilms on the surface of gallstones 160, which mimics results from human gallbladder infections with S. Typhi (unpublished data, Dr. Steven Baker). RpoS and Salmonella live vaccine strains: The rpos gene encodes the σ S (RpoS) alternative sigma factor which plays an important role in bacterial survival during starvation or stress 161. At the time of ingestion, it is believed that bacteria in are in stationary phase 162, therefore the genes involved in this regulon are potentially important during early phases of infection. RpoS is regulated at the transcriptional and post-transcriptional level 163, is activated upon entry into stationary phase or exposure to certain stress conditions and induces expression of a regulon containing over 30 genes 161. In vitro, ΔrpoS mutants are more susceptible to increased osmolarity or acidity, oxidative stress and extreme temperatures. In vivo, ΔrpoS mutants of S. Typhimurium are highly attenuated in BALB/c mice. Although there are no reported effects of RpoS mutations on production of PAMPS such as LPS or Flagella 164, RpoS plays an important role in virulence through directly and indirectly enhancing expression of the Salmonella virulence plasmid (Spv) genes 165 which control the growth rate of the bacteria in the deep lymphoid organs such as spleen and liver. RpoS mutants are not deficient in macrophage colonization but are unable to colonize the GALT, are severely impaired in terms of their ability to disseminate through the bloodstream and replicate in the spleen 28

40 and liver (-5 log reduction in infectivity ratio in spleen and -4 log reduction in liver) 166 and are unable to persist chronically in mice 167. Although loss of Spv virulence plasmid expression is believed to account for much of the observed attenuation in RpoS-deficient S. Typhimurium, strains, which are cured of their virulence plasmid, exhibit less attenuation than ΔrpoS mutants, indicating that RpoS-mediated regulation of chromosomal genes is also playing a key role in virulence 168. RpoS and live vaccine candidates: Due to their high level of attenuation, ΔrpoS mutants have been extensively investigated for possible future inclusion in a live vaccine strain and the currently licensed live-attenuated vaccine strain for S. Typhi Ty21a is RpoSdeficient 169. As a method of inducing protective immunity, orally administered, live attenuated vaccines are particularly appealing because they do not require injection or medical administration, are less expensive to produce and are able to generate high-level long-lasting immunity. Current attempts to create genetically defined live vaccine strains are aimed at creating mutations which attenuate virulence through at least 2 distinct mechanisms and which do not have the potential to revert back to wild type virulence levels 170. Curtiss et al. have investigated the immunogenic properties of ΔrpoS mutants and reported that in spite of their high level of attenuation, ΔrpoS mutants have several shortcomings including poor dissemination beyond the GI tract, high required oral dose (>10 10 CFU) and failure to induce protective levels of IgG and IgA to LPS 170. In spite of these disadvantages associated with hyperattenuation, rpos mutations exhibit many promising characteristics and efforts are ongoing to determine a mechanism by which the RpoS + phenotype may be conditionally complemented and used in conjunction with other inactivating mutations to balance immunogenicity and attenuation 171,

41 Hypothesis Goals and Specific Aims of this study: Previous work from our lab and others has indicated the presence of a high-molecular weight surface polysaccharide in nontyphoidal Salmonella. In other organisms including S. Typhi, capsules are known to play an important role in acute infections and improved understanding of this function has yielded numerous therapies and preventative strategies. Capsules have also been shown to play a role in multicellular behavior and chronic infection, but previous studies had yielded conflicting results as to whether the O-ag capsule is important in multicellular behavior and biofilm formation. For that reason we sought to investigate the role of the O-ag capsule in chronic and acute infection with S. Typhimurium. Because LPS is known to be crucial for virulence, we sought first to determine the effects of capsular defects on production of LPS. We hypothesized that capsular production employs precursors which are common to both LPS and capsule biosynthesis and that many LPS deficient mutants would similarly lack O-ag capsule. We further hypothesized that all O-ag operon genes would be required for production of the O-ag capsule. Previous results had indicated the O-ag capsule to be required for biofilm formation and biofilm formation to be a critical mediator of chronic gallbladder colonization. Therefore we hypothesized that O-ag capsule deficient mutants would be unable to establish persistent gallbladder infection in a murine model of chronic disease. Alternatively, chronic colonization likely relies on establishment of initial systemic infection and capsules are known to facilitate systemic infection with many other encapsulated organisms- therefore we sought to similarly determine whether O-ag capsule deficient mutants are deficient for causation of systemic disease. The general methodology employed involved generation of individual mutations 30

42 in O-ag capsule operon genes and development of a reliable methodology to confirm the presence or absence of O-ag capsule. A mutation in yiho was selected based on previous results and initial screening of mutants indicating this mutation as having the greatest decrease in capsular production. Animal modeling of systemic disease revealed no decreased ability of this mutant to colonize the gallbladder or form gallstone biofilms. Further experiments revealed that the yiho mutation did not result in decreased biofilm formation on cholesterol-coated surfaces. These experiments did reveal a generalized defect in systemic colonization associated with loss of the O-ag capsule. Further studies employing animal modeling of acute systemic infection indicated that O-ag capsule deficient mutants are not attenuated during acute infection. In vitro assays revealed that the O-ag capsule is not expressed simultaneously by all cells, but may exhibit characteristics that would impair host control of infection in vivo. Studies of virulence and protection using yiho mutation in combination with a second attenuating mutation in rpos demonstrated that this strain exhibits severe attenuation and provides high level of protection against subsequent infection after a single oral dose. The results described herein serve to better characterize the O-ag capsule of S. Typhimurium and show that this structure may provide an additional mechanism by which the bacteria may overcome the host immune system. 31

43 Chapter 2: The O-ag Capsule of Salmonella Typhimurium facilitates serum resistance and regulated surface expression of phase I flagella Abstract: Salmonella Typhimurium, similarly to other enteric pathogens, produce a group IV O-antigen (O-ag) capsule exhibiting structural resemblance to the lipopolysaccharide (LPS). Polysaccharide capsules are known virulence factors of many bacterial pathogens, facilitating evasion of immune recognition and systemic dissemination within the host. Previous work has focused primarily on the role of the O- ag capsule in chronic infection and bacterial attachment to environmental surfaces, however, its role in acute infection remains to be described. Systemic infections by Salmonella Typhi are facilitated by the production of the Vi-antigen (Vi-ag) capsule, however S. Typhimurium are also capable of causing acute systemic infection despite lacking Vi-ag. We hypothesized therefore that the O-ag capsule in S. Typhimurium may function in a similar manner to the Vi-ag capsule of S. Typhi, by diminishing surface expression of pathogen associated molecular patterns (PAMPs) such as flagella, and increasing resistance to host immune molecules. In order to test this hypothesis, O-ag capsule deficient mutants were constructed using lambda red-mediated homologous recombination and loss of capsular surface expression was confirmed through microscopy and immunoblotting using capsule-specific antisera. We verified that loss of O-ag capsule production did not alter bacterial growth or production of LPS. Western blot analysis and confocal microscopy revealed that O-ag capsule-deficient mutants 32

44 demonstrate increased production of phase I flagellin (FliC) and were more susceptible to killing by human serum complement. Surprisingly, O-ag capsule deficient mutants demonstrated no reduction in oral virulence in BALB/c mice, but in combination with a mutation in rpos, are severely attenuated and able to provide high-level protection against subsequent infection after a single oral dose. These results indicate that expression of the O-ag capsule may provide an additional mechanism by which S. Typhimurium may evade detection and clearance by the host immune system. Introduction: Salmonella enterica are believed to cause over 93 million new infections annually 17, making it one of the most commonly acquired causes of bacterial disease. Upon oral infection with non-typhoidal Salmonella (NTS) (most commonly Salmonella Typhimurium or Salmonella Enteritidis) bacterial interaction with the intestinal epithelial cells of the GI mucosae provokes a robust inflammatory response. This response, mediated primarily by innate immune recognition of pathogen associated molecular patterns (PAMPS) such as TLR4 activation by lipopolysaccharide (LPS) and TLR5 activation by flagella results in neutrophil recruitment and efficient clearance of the organism. Although the clinical picture of NTS infection in healthy adults is typically considered to be self-limiting acute gastroenteritis, NTS are capable of extra-intestinal dissemination and bloodstream infection (BSI). In Sub-Saharan Africa, invasive nontyphoidal salmonellosis (ints) infections have increased dramatically in the past decade and are now a leading cause of BSI in many locations. In these patients, disease advances rapidly, recurs frequently and is associated with a high fatality rate. Although ints is highly associated with immune dysfunction, particularly with advanced HIV, a 33

45 number of studies have reported extraintestinal dissemination of nontyphoidal salmonellae in healthy adults A recent study of HIV-negative adults in the UK with culture-confirmed salmonellosis demonstrated extraintestinal dissemination to the bloodstream in 10% of all infections 177. Many organisms capable of causing bacteremia produce capsular polysaccharides, which can enhance bacterial virulence through a variety of mechanisms. In Salmonella Typhi, systemic dissemination is aided by the production of the Vi-ag capsule. Although not required for infection, the Vi-ag capsule facilitates virulence by masking LPS, downregulating production of immunostimulatory flagellin and directly interfering with host inflammatory signaling cascades. More recently, S. Typhimurium and S. Enteritidis have been observed to produce a group IV O-antigen (O-ag) capsule. This capsule, which is times longer than the typical O-ag of the LPS, is so-called due to the high degree of structural similarity to the O-ag of the LPS. Previous studies have reported it to be critical for environmental persistence, bacterial aggregation and surface adherence 151. Although regulated by AgfD in concert with extracellular matrix components and thus maximally expressed in stationary phase at lower temperatures (<30 C), the O-ag capsule operon (ysha-yihu) is also highly expressed in vivo, specifically in the small intestine, liver and spleen 178, indicating a possible role in systemic virulence. In the present study, we have demonstrated that the O-ag capsule in S. Typhimurium is immunologically distinct from the LPS, modulates surface expression and phase variation of flagella and facilitates serum resistance. The results described herein imply that the O-ag capsule may function in a somewhat analogous role to the Vi-ag capsule of S. Typhi thereby potentially facilitating systemic disease. 34

46 Materials and Methods: Bacterial strains, culture conditions, and reagents: Bacterial strains used in this study are shown in Table 1. Cultures were grown overnight at 37 C with aeration in Luria- Bertani (LB) broth or on LB agar plates as described below. Capsule detection and purification was conducted on cultures in stationary phase. For experiments using mid logarithmic phase bacteria, overnight cultures were diluted 1:100 in fresh LB broth and grown to an optical density at 600nM (A 600 /OD 600 ) of When necessary, media were supplemented with chloramphenicol (25 µg/ml), ampicillin (50 µg/ml), or kanamycin (45 µg/ml). Construction of Mutants: Unmarked, non-polar deletions were created according to the λ red-mediated homologous recombination method as described by Datsenko and Wanner 179. In brief; Strain JSG1727 was cultured for 5hr at 30 C to induce expression of pkd46 λ red recombinase plasmid. Cells were pelleted and washed in 10% glycerol. Inner primers with homology to target gene were used to PCR amplify the Cam r cassette of pkd3. The PCR product was gel extracted and transformed into JSG1727 and plated on LB Cam at 37 C. Transformants were patched to LB Cam and LB amp at 40 c and to remove pkd46. Cam R, Amp S colonies were selected, screened via colony PCR with outer checking primers and sequenced (OSU Plant Microbe Genomics Facility) to confirm CamR insertion (intermediate strains in which target gene is replaced by camr are referred to using the number of the final, unmarked strain followed by b ). CamR intermediate strains were grown at 37 C and transformed with purified pcp20 encoding flip recombinase, to resolve the Cam R cassette. Amp R Cam S transformants were selected, 35

47 grown at 37 C to remove pcp20 and Cam S Amp S colonies were screened via colony PCR. Deletion of target gene(s) was confirmed by sequencing and the resultant strain was given a unique JSG number and stored at -80 C in 20% glycerol. For complementation of yiho, plasmids pjm1 and pjm2 were constructed as follows. The yiho gene was amplified by PCR and digested with EcoRI and HindIII. High and low copy number cloning vectors puc18 and pwsk129 were similarly digested and linearized DNA fragments were run on 0.8% agarose gel. Insert and vectors were gel purified and ligated using T4 DNA ligase. Ligation reactions were transformed into One-Shot Top10 chemically competent E. coli and transformants were plated on LB amp containing 0.1mM IPTG, 40µG/mL Xgal to allow blue-white screening. White colonies were selected and screened by colony PCR using M13F and JG2491 primer internal to yiho. Correct clones carrying plasmid pjm1 (yiho in puc18) and pjm2 (yiho on pwsk129) were miniprepped and transformed into JSG3453 (ΔyihO/ΔrpoS 455 ) and JSG3672 yiho. All clones containing high-copy plasmid pjm1 (PyihO in puc18) exhibited minor growth defects and variable colony morphology but complementation on low-copy plasmid pjm2 (PyihO in pwsk129, 6-8 copies per cell) was unable to provide functional complementation. Purification of Bacterial Polysaccharides: Overnight bacterial cultures were normalized to OD and centrifuged 30 at 5,000XG. Pellets were homogenously resuspended in TriZol and incubated at room temperature for 30. Chloroform was added at a ratio of 1:5, the mixture was vigorously vortexed and incubated for 15 at room temperature to create a phase separation. Samples were centrifuged for 10 at 11,000 X G, the aqueous phase was removed and the organic phase was subjected to a secondary 36

48 aqueous extraction. Aqueous phases were combined, frozen at -80 C and lyophilized. Lyophilized samples were resuspended and dialyzed exhaustively against water in a 7,000 MWCO membrane for 18hrs. Dialyzed samples were lyophilized in pre-weighed tubes and resuspended at a concentration of 25µG/mL. Generation of Polyclonal Antisera: Rabbit polyclonal antisera directed at roughly purified extracellular polysaccharides of Salmonella Enteritidis was generated as previously described (generously provided by D.L. Gibson and A.P. White). The LPS O- antigen of S. Enteritidis and S. Typhimurium are structurally very similar, both sharing an identical repeating trisaccharide of α-d-manp-(1-4)-α-l-rhap-(1-3)-α-d-galp-(1-2), with the mannose C3 bearing tyvelose in the former and abequose in the latter. The primary constituents of the S. Enteritidis capsular material were the three monosaccharides which are common to both serovars (mannose, rhamnose and galactose) therefore we sought to detect the S. Typhimurium capsule using the polyclonal antisera generated against S. Enteritidis EPS (Fig. 3). Although the previous study adsorbed the antisera against a gale mutant, this strain would be deficient in numerous EPS types and therefore may not be the most appropriate strain for adsorbtion. In this work antisera was adsorbed against an O-ag capsule operon mutant JSG3675 (ΔyshA-yihW) of S. Typhimurium. In order to remove potential non-specific reactivity with other immunogenic components of the cell surface, 100mL of stationary phase culture of JSG3675 (ΔyshA-yihW) was pelleted at 4800xG for 20min at 4 C. Cells were washed in 10mL of adsorbtion buffer (10mM NaCl, 10mM KCl) and resuspended in 50mL ice-cold acetone for 30 minutes. Acetone was allowed to evaporate in fume hood until pellets were completely dry. Dried pellets were then homogenized into a fine powder. Antibody 37

49 was diluted 1:50 in PBS and incubated with 100mG acetone powder for 4h at 4 C with constant agitation. Samples were centrifuged at 14,000xG for 15 min. This procedure was repeated two more times, antisera was passed through a sterile filter, aliquoted and stored at -20 C. Western blot analysis of isolated polysaccharides: 6µG of isolated polysaccharide were boiled in 6X Laemelli sample + β-mercaptoethanol for 10 minutes and separated by electrophoresis on a 16cm 15% SDS-PAGE gel. Gels were wet-transferred to methanolactivated PVDF membranes which were subsequently blocked at 4ºC overnight in 5% Bovine Serum Albumin (BSA) in TBS (ph 7.6) (BSA/TBS). LPS was detected using murine monoclonal IgG to S. Typhimurium group B LPS (1:2,000 in 5% BSA/TBST for 2h, 22 C). Membranes were washed washed in TBST (3 x 15min) and incubated in horseradish peroxidase-conjugated goat α -mouse IgG (1:4000 in 5% BSA/TBST for 2h, 22 C). For analysis of IgG binding, 15µG of purified bacterial polysaccharide was run and transferred as described above. Membranes were blocked with 5% BSA/TBS overnight and incubated with pooled donor sera diluted 1:5000 in BSA/TBST for 2h, 22 C. Membranes were then washed 3x in TBST and incubated with horseradish peroxidase conjugated rabbit α -human IgG at 1:5,000 in BSA/TBST. Western blots were visualized using the Bio-Rad ChemiDoc system. 38

50 Figure 3. Detection of O-ag capsule using adsorbed polyclonal antisera generated against purified S. Enteritidis capsular polysaccharide. A) Pro-Q Emerald Glycoprotein stain of highly purified capsular polysaccharide (CPS) from S. Enteritidis demonstrates capsular localization within stacking gel. B) Pro-Q Emerald Glycoprotein stain of roughly purified CPS from S. Typhimurium demonstrates capsular localization within the stacking gel. C) Western blot of roughly purified CPS from S. Typhimurium using unadsorbed polyclonal antisera demonstrates two regions of reactivity prior to antibody adsorbtion. D) Western blot of roughly purified CPS from S. Typhimurium using adsorbed polyclonal antisera demonstrates a single region of reactivity corresponding to localization of purified capsular polysaccharide. 39

51 In-gel staining of EPS: Bacterial polysaccharide extracts separated on a 15% SDS- PAGE gel (16cM 2 x 1.5mM) were immersed in 100 ml of Alcian blue solution (0.005% Alcian blue, 40% ethanol, 5% acetic acid in dh 2 O) for 30 min. The gel was rinsed for 1 min in ddh 2 O and silver staining was carried out according to the method of Tsai and French 180. Visualization of EPS in Fig. 3 was conducted using ProQ Emerald glycoprotein staining as previously described 181. Production of LPS was compared to negative control strains of S. Typhimurium with defects in biosynthesis or assembly of the LPS O-ag (ΔrfaD JSG1676, ΔgalE JSG1221, ΔLPS JSG880) and are referred to here as LPS deficient for clarity. Confocal Microscopy: Capsule expression was detected using a polyclonal antisera generated against purified EPS from S. Enteritidis (gift of D.L. Gibson and A.P. White) which had been triple adsorbed against S. Typhimurium JSG3675 ( ysha-yihw). Overnight bacterial cultures were normalized in LB to O.D x 1mL. 100µL was removed and centrifuged (5 min, 5000 x G) and resuspended in 50µL Hepes buffer (0.1M hepes, 1x PBS, ph to 7.0, filter sterilized). Cells were fixed for at room temperature in 2% paraformaldehyde (PFA, ph 7.4, sterile filtered in 1xPBS), washed 3 times in HEPES buffer and blocked in filtered 5% BSA/TBS. Antibody incubations were conducted in 5% BSA/TBST at room temperature for 2h or 4 C overnight. Secondary antibody incubation was conducted simultaneously with DAPI counterstain (1uM) in the dark. Antibody incubations in experiments employing multiple fluorophores were conducted simultaneously if unique host and target animal species for primary and secondary antibody were compatible. Otherwise, incubations were conducted serially with 1hr blocking between subsequent incubation. Multiple fluorophore experiments 40

52 were conducted first with labeling controls including labeling individually and with reversed order of antibody addition, primary only, secondary only and both together if applicable. Alexafluor 488 was typically employed for targets which stained less intensely as this fluorophore tends to be more visible, while Alexafluor 594 is visually less bright. Following staining, cells were mounted on glass slides in Prolong Gold antifade, coverslipped and left to dry in the dark overnight or stored for up to 2 weeks at 4 C. Transmission Electron Microscopy: Individual colonies grown at 37 C overnight were gently resuspended in TEM sample buffer with 2.5% glutaraldehyde, (0.1M hepes, 6% sucrose, 1xPBS, ph to 7.0, filter sterilized). 25µL of sample was pipetted onto Formvarcoated 200 mesh nickel grids and allowed to settle for 25 minutes. Grids were washed with 1xTEM buffer, blocked (30 mins., 0.5% BSA, 4 C) incubated with anti-capsule antibody (1:100, 60 mins., 0.5% BSA, 4 C) followed by overnight incubation at 4 C with 10nM gold-conjugated goat anti-rabbit IgG. All staining steps were performed on a silicon pad in a humidified chamber. Serum Bactericidal Assay: Blood was collected from healthy adults by venipuncture according to the protocol #2009H0014 approved by The Ohio State University Institutional Review Board. Blood was left to clot at room temperature for 1 hour followed by 1 hour at 4 C. Serum was separated by centrifugation at 4 C and frozen in aliquots at 80 C in sterile polypropylene tubes. Serum was thawed on ice immediately prior to use. Heat-inactivation of complement was performed, when needed, by incubating serum in a water bath at 56 C for 30 minutes. Bacteria from stationary phase cultures were normalized to OD in LB serially diluted in PCMH buffer (1xPBS, 1mM HEPES, 0.5 mm MgCl 2, and 0.15 mm CaCl 2 ; ph 7.3) and 100µL of 2x

53 CFU/mL was added to 100µL of 50% normal human serum (NHS) to yield a 200µL sample containing 2x10 5 CFU in 25% serum. 10µL aliquots were removed immediately (T 0 ) and following 30, 60 and 90 minutes of incubation at 37 C, serially diluted in ice cold 1x PBS, placed on ice for 5 and plated for viability. Serum killing was calculated as compared to CFU recovered at T 0 and CFU recovered from samples in heat inactivated serum and buffer alone at various timepoints. Complement Binding: Overnight bacterial cultures were normalized to an OD 600 of 0.8 in PCM/HEPES and 100µL of 2x10 6 CFU/mL of salmonellae were added to 100µL 20% human serum (final concentration 10% NHS) for 30 minutes. 10µL aliquots were removed for viability plating at T 0 and T 30. Following 30 incubation, samples were placed on ice, washed 3X in HEPES/PCM and fixed in 4% PFA for 15 at room temperature. Fixed samples were blocked and incubated with goat α-human C3 followed by donkey α-goat alexa-fluor 596. Blocking and antibody incubations were performed in sterile-filtered 5% BSA in TBST for 1h at room temperature, samples were washed 3x following incubations with cold TBST. Samples were mounted with Pro-Long Gold antifade and allowed to dry in the dark at room temperature overnight prior to viewing on an Olympus FV1000 spectral confocal scanning laser microscope. Flagellin analysis: Bacteria were grown to mid log (to visualize both FliC and FljB) or stationary phase (as in other experiments), normalized to OD , centrifuged for 1 h at 221,000 g and 4 C to separate whole cells from the supernatant so that both could be assayed. Supernatant was filtered through 0.22-µm-pore-size low-protein-binding filters (Millex GV; Millipore) concentrated at 4 C using trichloroacetic acid (TCA) precipitation and normalized for total protein content using BCA assay. Normalized 42

54 pellet and supernatant samples were diluted in Laemmli sample buffer (50 mm Tris-Cl [ph 6.8], 100 mm dithiothreitol, 2% sodium dodecyl sulfate, 0.1% bromphenol blue, 10% glycerol) and loaded into 8% or 15% polyacrylamide gel for separation of proteins. Coomassie brilliant blue was used to visualize proteins prior to extraction and submission for mass spectrometry analysis. Western blotting for flagellin: Electrophoresed proteins were wet-transferred to methanol-activated PVDF membranes which were subsequently blocked at 4ºC overnight in 5% Bovine Serum Albumin (BSA) in TBS (ph 7.6). Membranes were probed with monoclonal antibodies (described in Table 2) to FliC (1:10,000 in BSA/TBST) or flagellin (1:8,000 in BSA/TBST) followed by horseradish peroxidase conjugated goatanti-mouse IgG (FliC-1:12,000, Flagellin-1:10,000). Motility assay: Overnight or mid log cultures were normalized to OD , stabinoculated into motility plates (LB media containing 0.3% agar) and incubated for 6hrs at 37ºC. Motility was calculated by measuring colony diameter every 30 minutes and after 18hr incubation. Ethics Statement: Animals were housed and used in accordance with established guidelines set forth by the Ohio State University Institutional Animal Care and Use Committee (IACUC) in accordance with U.S. National Research Council standards and Research Animal Welfare Act. All experimental protocols and standard operating procedures were subject to prior ethical and procedural approval (IACUC Protocol # 2009A0057-R1) and researchers were subject to ongoing oversight to ensure strict adherence to study protocol. Animals were closely monitored throughout the course of infection and every effort was made to minimize discomfort. Infected animals were 43

55 evaluated a minimum of twice daily and were humanely euthanized by CO 2 asphyxiation if moribund or at predetermined study endpoints. Murine infection: 6-10 wk. old female Balb/C mice were orally inoculated with x10 5 CFU of an overnight culture of S. Typhimurium JSG210 or JSG3453 (ΔyihO/ΔrpoS 455, n=65) JSG3672 (ΔyihO, n=26), washed and suspended in 100µL PBS. Food and water were withheld 4hrs prior to infection and returned to cages after. Cultures were diluted and enumerated on LB agar to determine CFUs inoculated. Animal mortality was observed and recorded. Animals were observed closely throughout infections and humanely euthanized at study endpoint or when observed to be moribund. To evaluate the protective efficacy, animals were orally challenged (n= 54) with a lethal dose of wild-type S. Typhimurium (2 x 10 6 CFU) at 8 weeks after initial infection, and the survival rate was evaluated for the following 21 days. Euthanized animals were dissected and organs were removed, weighed, homogenized in sterile PBS and enumerated on LB agar to calculate bacterial burden in organs. Challenge Experiments: 60 days following infection with yiho mutant, surviving animals (14/15, 93%) were orally challenged with 9x10 5 CFU of wild-type JSG210 S. Typhimurium as described above. A randomly selected animal was sacrificed prior to the challenge and organs were plated to check for colonization with yiho mutant. Challenged animals were sacrificed at day 11 and day 21 and bacterial burden in spleen and liver was determined as described above. In separate experiments, animals were challenged with wild-type JSG210 at 18, 25 and 40 weeks post-infection, sacrificed at days 11 and 21, enumerated on LB and recovered bacteria were analyzed by colony PCR to verify that animals were not colonized by the initial yiho mutant strain. Statistical 44

56 analysis All statistical analysis was performed with Prism 5 software (GraphPad Software, Inc.). Survival curves were generated by use of Kaplan-Meier estimators. Survival distributions of each treatment group vs. control mice were compared via the log-rank (Mantel-Cox) test. In bar graphs, indicated error lines represent standard deviations and unless otherwise stated, values reported represent the mean of three independent experiments carried out in triplicate. To determine significance of observed differences between groups, Student s t-tests or two way ANOVA were performed. Observed differences were considered statistically significant compared to wild type at a P value of <0.05. Asterisks indicate **P<0.01, ***P< respectively. 45

57 Figure 4. Visualization of LPS and O-ag capsule. Purified bacterial polysaccharides were separated on 15% SDS-PAGE gels for immunoblotting or silver-stain detection of LPS or O-ag capsule. A) Anti-LPS western blot. B) Anti O-ag capsule C) Silver staining of LPS. Collectively, images demonstrate that yiho mutations do not result in defective LPS production and demonstrate high molecular weight of O-ag capsule. 46

58 Results: O-ag capsule deficient mutants do not exhibit defective synthesis of LPS. Production of full length LPS with a variable number of O-ag repeats, in particular the production of long LPS O-ag, is critical for systemic virulence and resistance to host immune molecules 182. Biosynthesis of surface polysaccharides is a complex and highly sequential process and group IV O-ag capsules often share common monosaccharide precursors and selected biosynthetic machinery with LPS O-antigens 108. Therefore we sought to determine whether O-ag capsule deficient mutants exhibit reduced or truncated LPS production. LPS production was analyzed by separation of 5µG purified bacterial polysaccharide on 15% SDS-PAGE gels followed by silver staining or western blotting with α-lps. LPS production in mutants was compared to wild type JSG210 or to a previously described deep rough, LPS O-ag deficient strain (JSG880, referred to as LPS- for clarity) which demonstrated no O-ag banding pattern on silver stained gels (Fig. 4C). Results indicated that no mutations in O-ag capsule operon genes resulted in production of truncated LPS (Fig. 5A). Although O-ag capsule deficient mutants JSG3453 (ΔyihO/ΔrpoS 455 ), JSG3672 (ΔyihO) and JSG3675 (ΔyshA-yihW) all exhibited production of full-length LPS (Fig. 4A, 4C), densitometry of banding intensities indicated that O-ag capsule deficient mutants exhibited 35-50% more LPS with 1-8 O-ag repeating units than wild type, indicating that loss of O-ag capsule may have resulted in slightly increased production of shorter LPS. Confocal microscopy (Fig. 6C) and dot blots of normalized, fixed whole cells (Fig. 6A) incubated with α-lps indicated wild-type production of LPS on the bacterial surface in O-ag capsule deficient mutants. The presence of a complete LPS O-ag was further confirmed by phage P22 infectivity assay. 47

59 Figure 5. Preliminary screening of O-ag operon mutants for production of LPS and O-ag capsule. Whole cell lysates normalized for cell volume and electrophoresed on 15% SDS-PAGE gels were transferred to PVDF membrane and probed for either A) LPS or B) O-ag capsule. All mutants were observed to produce O-ag capsule at levels below that of wild type S. Typhimurium. Both LPS - mutants (ΔgalE and ΔrfaD) exhibited abrogated production of LPS and O-ag capsule. Of the O-ag capsule operon gene mutants assayed for capsular production, Δ yiho appeared to exhibit the greatest reduction in capsular production relative to the wild type. 48

60 O-ag capsule expression in O-ag operon and LPS - mutants: Defined mutations in each of the genes putatively involved in O-ag capsule synthesis were constructed, growth rates were confirmed to be unaffected and mutants were screened for O-ag capsule production using capsule-specific antisera 129. Western blotting of whole cell lysates demonstrated that deletion of yiho, encoding an protein homologous to an inner membrane permease putatively involved in transport of O-ag capsule to the surface of bacterial cells, resulted in the greatest loss of capsular expression (Fig. 4B, Fig. 5B). Immunogold TEM of whole cells labeled with anti-capsule antibody and 10nM gold-conjugated goat anti-rabbit secondary antibody, demonstrated capsular localization with the bacterial cell surface in wild type cells (Fig. 7A I) which was absent in JSG3453 (ΔyihO/ΔrpoS 455 ) (Fig. 7A II). O-ag capsule was not detected in any of the LPS-deficient mutants tested (ΔrfaD JSG1676, ΔgalE JSG1221, ΔLPS JSG880) (Fig. 5B, Fig. 6B). Whole cell dot blots and confocal microscopy demonstrated decreased capsular expression in JSG3672 (ΔyihO), JSG3453 (ΔyihO/ΔrpoS 455 ), JSG3675 ( ysha-yihw) and LPS deficient strain JSG880 (Fig. 6A, 6B). Complementation of yiho on pjm1 (yiho in high copy number plasmid puc18) partially restored O-ag capsule production (Fig. 6A). Confocal micrographs revealed that O-ag capsule appears to be heterogeneously expressed among wild type cultures (Fig. 6B, 7B) with more intense staining associated with cellular aggregates. In cells labeled with both LPS and O-ag capsule antibodies, reactivity in individual cells was observed with either LPS or capsule (Fig. 7B), but not both and was not dependent on of the order of antibody incubation. 49

61 Figure 6. Detection and visualization of LPS and O-ag capsule on whole S. Typhimurium bacterial cells. A) Quantitation of capsular expression relative to LPS in dot-blot of fixed bacterial cells. Confocal fluorescence micrograph of whole DAPI-stained bacterial cells labeled for B) capsule or C) LPS. 50

62 Figure 7. Visualization of capsular localization to bacterial cell surface and observation of heterogeneous expression throughout the culture population. A) O-ag capsule of S. Typhimurium detected with α-capsule antibody, 10nM immunogold conjugated secondary antibody and visualized with TEM. I) In wild type JSG210, the region of capsular reactivity is clearly visible and localized to the bacterial surface. II) In strain JSG3453 (ΔyihO/ΔrpoS 455 ), no capsular material or antibody reactivity is associated with the bacterial surface. B) Whole cell confocal micrograph of DAPI-counterstained wild type JSG210 S. Typhimurium labeled for LPS (green), O-ag capsule (red), DNA (blue). I) LPS and capsule staining merged with DAPI demonstrates that all LPS or capsular reactivity is localized to bacterial cells II) LPS and capsule staining alone clearly indicating heterogeneous production of O-ag capsule within the bacterial population. White arrows indicate bacterial cells which are strongly reactive with the O-ag capsular antibody and do not exhibit reactivity with the LPS antibody. Strong capsular reactivity was most commonly associated with aggregated cells but was also observed on individual cells as indicated by the arrow near the bottom of the image. 51

63 Identification of a second-site rpos mutation in JSG3453: O-ag capsule deletion mutants were screened for production of AgfD-regulated ECM components by growth on LBNS at 22 C supplemented with Congo Red or calcofluor. With the exception of JSG3453 (ΔyihO/ΔrpoS 455 ), all O-ag operon mutants demonstrated production of red, dry and rough (RDAR) colony morphology and calcofluor binding characteristic of robust extracellular matrix (ECM) production. This assay revealed spontaneous occurrence of non-rdar mutations arising in intermediate steps of the mutation process to generate JSG3453 (ΔyihO/ΔrpoS 455 ), following replacement of yiho with camr. Spontaneous non- RDAR mutants in S. Typhimurium have been reported to arise following laboratory passage as a result of rpos mutations. Non-RDAR strain JSG3453(ΔyihO/ΔrpoS 455 ) was screened for H 2 O 2 dismutation by visual inspection of gas production in the presence of H 2 O 2. Lack of bubbling indicated a reduction in RpoS-regulated catalase production. Sequencing of rpos in JSG3453 (ΔyihO/ΔrpoS 455 ) revealed that in addition to the yiho deletion, a C(G) to T(A) mutation had occurred in base pair 455 of rpos, producing a UAG amber stop codon at the 152 nd amino acid of RpoS. This point mutation results in a truncated 17kD protein putatively lacking domains 3 and 4 but still possessing AAs , which interact with the RNA polymerase core domain, indicating that this mutation may result in an RpoS protein which is able to interact with the core RNA polymerase but unable to function to initiate transcription of RpoS-specific genes. Assays to determine H 2 O 2 sensitivity (Fig. 8A) revealed that JSG3453 (ΔyihO/ΔrpoS 455 ) exhibited equal sensitivity to ΔrpoS insertional inactivation mutants (JSG1577 ΔrpoS 237 has the first 237 base pairs of rpos, while JSG1747 ΔrpoS has the entire gene replaced with kanr 168 ). Strain JSG3453 (ΔyihO/ΔrpoS 455 ) exhibited no detectible O-ag capsule 52

64 production, while other mutations in rpos alone resulted in a slight decrease (JSG1577 ΔrpoS 237 ) or an increase (JSG1747 ΔrpoS) in O-ag capsule production. Previous work has indicated that RpoS positively controls the O-ag operon through AgfD and that O-ag capsule is coordinately regulated with other extracellular matrix components. Our observation was that loss of RpoS function resulted in total loss of cellular aggregates in liquid culture and an greater decrease in O-ag capsule production than ΔyihO mutation alone, this may have been due to decreased production of cellular aggregative components with which O-ag capsule was highly associated on the cell surface. RpoS is thought to increase production of the O-ag capsule via AgfD. It is possible that this mutation was selected for in order to prevent lethal intracellular accumulation of partially assembled capsular precursors, however the final ΔyihO JSG3672 mutant did not exhibit spontaneous rpos mutations, indicating that this mutation alone is not sufficient to result in spontaneous acquisition of rpos mutations. We hypothesize therefore, that the intermediate steps in the mutagenesis technique (which involve growth at sub-optimal temperatures, multiple antibiotic selection and serial passage) may have induced high levels of stress in the bacterial cells resulting in induction of multiple stress-response pathways. In such a scenario, competition may occur between multiple stress response sigma factors for interaction with the core RNA polymerase 183, resulting in selection against whichever sigma factor is induced but dispensable for survival in that environment. Creation of ΔyihO JSG3672 was achieved by altering the mutagenesis technique in a manner intended to minimize the amount and duration of stress placed on bacterial cells during this intermediate step of mutagenesis; By decreasing the temperature used to remove temperature-sensitive plasmid pkd46 from 42 C to 37 C, 53

65 reducing the concentration of chloramphenicol from 25 µg/ml to 10 µg/ml and patching colonies for 2 days instead of 5 to ensure removal of pkd46. In this manner, ΔyihO JSG3672 was created without the acquisition of spontaneous secondary mutations in rpos. If this hypothesis is accurate, loss of yiho is likely to be stressful on the cells but under normal laboratory conditions, is not sufficiently stressful to result in spontaneous rpos mutations, however this strain should be more susceptible to these mutations if exposed to the proposed stress conditions encountered during intermediate steps of the λ- red mediated mutagenesis technique. Further testing is necessary in order to determine the validity of this hypothesis regarding the origin and functional effects of the observed rpos mutation acquired in strain J JSG3453 (ΔyihO/ΔrpoS 455 ). 54

66 Figure 8. Association of H 2 O 2 resistance, RDAR morphotype and RpoS status. Production of red, dry and rough colonies on Congo Red agar (RDAR) is indicative of extracellular matrix components, the production of which is regulated by RpoS via AgfD in S. Typhimurium. Similarly, catalase production resulting in H 2 O 2 resistance is largely mediated by functional activity of RpoS. Mutation in yiho resulted in non-rdar colonies (B-III). This morphology was not exhibited by other O-ag operon mutations (B-VII) and was not restored by complementation of yiho with either pjm1 or pjm2 (B-IV) leading to investigation of a possible second site mutation. Sequencing revealed a premature stop codon in rpos of JSG3453 (ΔyihO/ΔrpoS 455 ). JSG3672 carrying a mutation in yiho alone exhibited RDAR morphology and H 2 O 2 resistance indistinguishable from that of the wild type. Non-RDAR morphology corresponds to the introduction of yiho:camr mutation (B-II). RpoS mutations exhibit non-rdar morphotype and increased sensitivity to H 2 O 2, although differing insertional mutations appear to yield slightly different phenotypes (B-V, B-VI). 55

67 Loss of O-ag capsule results in increased serum sensitivity. Production of high molecular weight surface polysaccharides such as long chain LPS and capsular polysaccharide is frequently associated with increased resistance to killing by human serum complement 184. To determine the impact of O-ag capsule production on bacterial survival in normal human serum (NHS), O-ag capsule deletion mutants were tested for viability following incubation in 25% NHS for 180. As previously reported, wild type S. Typhimurium was highly resistant to killing by NHS while LPS deficient strains were killed within 30 minutes. No bacterial killing was observed in any strain incubated in 56 C heat inactivated serum, demonstrating that observed bactericidal activity was was due to the activity of functional complement (data not shown). O-ag capsule deficient mutants JSG3453 (ΔyihO/ΔrpoS 455 ), JSG3672 ΔyihO and JSG3675 ΔyshA-yihW exhibited increased sensitivity relative to wild type JSG210 which roughly corresponded to observed amount of capsular polysaccharide observed on these strains (as seen in Fig. 6B). The increased serum sensitivity observed in JSG3672 (ΔyihO) was restored in JSG3691 by complementation with pjm1 (yiho on puc18) (Fig. 9C). Confocal micrographs of bacterial cells incubated in 10% serum, fixed and labeled for α-human complement C3 (goat antisera generated against full length human C3, Quidel Corp.) revealed that increased serum sensitivity in JSG3453 (ΔyihO/ΔrpoS 455 ) was accompanied by a qualitative increase in deposition of complement component C3 on the bacterial surface (Fig. 9A, 9B representative images from 2 experiments). A modified far western blot of bacterial polysaccharides electrophoresed and transferred to PVDF, incubated in pooled donor sera and probed for human IgG or C3 (Fig. 9D) indicated that the target of IgG in these sera corresponded to the location of complement C3 deposition. This data 56

68 implies that complement C3 deposition is localized to the region of IgG deposition (characteristic of the classical complement pathway). Reactivity with C3 and IgG was observed in the regions of the gel corresponding to molecular weight 25kD. No C3 or IgG reactivity was observed in the stacking gel or upper portion of the separating gel corresponding to the regions where O-ag capsular material is detected. These data indicate that complement C3 deposition may be occurring through activation of the classical complement pathway and that the target of IgG antibodies present in tested sera was not the O-ag capsule but low molecular weight LPS, which had previously been observed to be more highly produced in O-ag capsule-deficient mutants. 57

69 Figure 9. Resistance of O-ag capsule deficient mutants to killing by human serum. A, B) Visualization of complement C3 bound to JSG3453 (ΔyihO/ΔrpoS 455 ) or wild type JSG210 S. Typhimurium following incubation in 10% human serum (2 serum samples shown). C) Bacterial viability following incubation in 25% human serum, graph is representative of 3 experiments with separate donor sera conducted in triplicate. *** = P > relative to JSG210, Student s t-test. D) Modified far western immunoblot of IgG (left) and C3 (right) from pooled donor sera bound to electrophoresed bacterial polysaccharides. Deposition of C3 and IgG are co-localized, possibly indicating that complement deposition is occurring primarily through activation of the classical pathway. C3 and IgG deposition appear to be occurring on a target of molecular weigh >25kD, with no C3 deposition occurring in regions where O-ag capsule reactivity is observed (molecular weight >150kD). No bactericidal activity was observed for any strain following incubation in heat-inactivated serum (data not shown). 58

70 O-ag capsule deficient mutants exhibit increased surface production of FliC. Analysis of total protein profiles of JSG210 or JSG3453 (ΔyihO/ΔrpoS 455 ) revealed a clear alteration in production of proteins at approximately 50kDa (Fig. 10B), which were determined by mass spectrometry to be flagellin. Synthesis of surface polysaccharides has been linked to flagellar regulation in a number of bacterial species S. Typhimurium express two antigenically distinct forms of flagellin encoded by divergent operons with a single Hin-invertible promoter. Alternate expression results in individual cells expressing flagella composed of either FliC (phase 1) or FljB (phase 2) flagellin on their surface 187,188. In vivo, it is thought that phase variation permits alternate expression of highly immunogenic epitopes 189, potentially facilitating immune evasion. S. Typhi is monophasic, and only expresses FliC, regulating transcription of flic production through TviA-mediated repression of the master flagellar regulator, flhd/c. In this manner, Vi-ag capsule and FliC are alternately expressed. In order to determine whether loss of O-ag capsule results in increased surface expression of FliC, whole bacterial cells were fixed and labeled with anti-flic and anti O-ag capsule antibodies (Fig. 10C). Confocal micrographs reveal an absence of O-ag capsule and qualitatively more FliC on the surface of JSG3453 (ΔyihO/ΔrpoS 455 ) and JSG3672 (ΔyihO) than on wild type cells. Micrographs reveal that wild type S. Typhimurium JSG210 and yiho complemented strain JSG3691 (ΔyihO JSG3672 complemented with yiho in pjm1) exhibited similar levels of O-ag capsule and FliC. Levels of cell-associated and secreted FliC were determined by western blotting of whole cell bacterial lysates or TCA precipitated supernatants (Fig. 10A) and supported microscopic observations of increased FliC in the absence of the O-ag capsule. Densitometric quantitation of relative band intensities in 59

71 western blots (Fig. 11A, 11B) indicated that capsule deficient mutants expressed approximately 8-10 fold more cell-associated FliC than wild-type cells. Western blotting using an antibody capable of detecting FliC or FljB (Fig. 11C) revealed that stationary phase cultures of wild-type S. Typhimurium and JSG3691 (ΔyihOpUC18:yihO) expressed primarily FljB, while capsule deletion mutants JSG3453 (ΔyihO/ΔrpoS 455 ), JSG3672 (ΔyihO) and JSG3675 (whole operon deletion ΔyshA-yihW) expressed primarily FliC with no detectible expression of FljB. Increased expression of FliC is not due to a general defect associated with loss of capsular expression as capsule-negative, LPS deficient mutant JSG880 had no associated increase in FliC expression. Densitometry of total flagella detected by western blotting in (Fig. 10C) indicates that while the phase of flagellin monomer composition is shifted from phase 2 to phase 1 in O-ag capsule deficient mutants, the overall amount of flagella produced is similar. Phase variable S. Typhimurium express both FliC and FljB in mid log phase but exhibit primarily FljB expression during stationary phase 190,191, mid log cultures of wild type S. Typhimurium and yiho complemented strain JSG3691 (ΔyihO JSG3672 complemented with yiho in pjm1) exhibited both FliC and FljB, while O-ag capsule deficient mutants expressed FliC alone. 60

72 Figure 10. Analysis of secreted and cell associated FliC in S. Typhimurium. A) Western blot analysis of secreted (TCA precipitated) and cell-associated FliC produced by wild type JSG210 or O-ag capsule deficient mutants indicating an increase in both secreted and cell-associated FliC by JSG3453 (ΔyihO/ΔrpoS 455 ) and JSG3675 (whole operon deletion ΔyshA-yihW). The increase in FliC production was not observed in strains lacking LPS (JSG880). B) Coomassie-stained gel of total protein profile of wild type or JSG3453 (ΔyihO/ΔrpoS 455 ) C) Confocal micrograph of whole cells labeled for FliC (yellow) O-ag capsule (red) and counterstained with DAPI (blue), Upper panels depicting (DAPI, FliC, O-ag capsule, merged images) demonstrate that both capsular and flagellar expression is heterogeneous throughout both wild type and O-ag capsuledeficient cultures. These images also show increased production of phase I (FliC) flagella in O-ag capsule-deficient mutants and restoration of both capsular and flagella production to wild type levels in PyihO JSG3691 (ΔyihO JSG3672 complemented with yiho in pjm1). 61

73 Figure 11. Detection of cell-associated FliC and FljB in S. Typhimurium. A) Western blot analysis and B) Densitometric analysis of average fold change in FliC expression relative to wild type in western blots of O-ag capsule mutants. C) Western blot detection of FljB. Western blotting indicates that JSG3453 (ΔyihO/ΔrpoS 455 ), JSG3672 ΔyihO and JSG3675 ΔyshA-yihW exhibit >9 fold increase in cell-associated FliC relative to wild type. Strain JSG3691 PyihO (ΔyihO JSG3672 complemented with yiho in pjm1) exhibits FliC expression levels approaching wild type. 62

74 In vivo virulence attenuation: Strains JSG3453 (ΔyihO/ΔrpoS 455 ) and JSG3672 (ΔyihO) were compared to wild type S. Typhimurium for oral virulence in a murine model of acute systemic infection. Animals were infected by oral gavage with a lethal inoculum of 2x10 6 CFU (2 log 10 above LD 50 ) of JSG3453 (ΔyihO/ΔrpoS 455 ), JSG3672 (ΔyihO) or wild type JSG210. All animals infected with wild type JSG210 or JSG3672 (ΔyihO) succumb to infection within 10 days while animals infected with JSG3453 (ΔyihO/ΔrpoS 455 ) survived to the endpoint of the experiment (Fig. 12A). 60% of animals infected with >10 10 CFU survived infection; although meningitis was observed to develop in several animals infected with this high inoculum. As discussed previously, JSG3453 (ΔyihO/ΔrpoS 455 ) carries a mutation in rpos in addition to its yiho mutation. Of note was the finding that although rpos mutations have previously been reported to exhibit severe defects for systemic colonization; specifically in Peyer s patches, spleen and liver, CFU recovered from spleens of animals infected with JSG3453 (ΔyihO/ΔrpoS 455 ) in both BALB/C and 129x1/SvJ animals indicated approximately 1 log 10 decrease in bacterial burden relative to wild type. This is in contrast to previous work indicating a 5-log 10 reduction in systemic colonization by rpos mutants 166 and an inability to establish infection at doses below 10 7 CFU. At 8 weeks postinfection (n=35) and 15 weeks post infection (n=3) animals were challenged with a lethal dose of JSG210. In contrast to naive animals, 84% of animals previously inoculated with JSG3453 (ΔyihO/ΔrpoS 455 ) survived the challenge (Fig. 12B). Animals were sacrificed at 14 and 21 days post-challenge and exhibited complete clearance of challenge inoculum by day 21 post-infection. Slide agglutination assays of normalized mid-log bacterial cultures demonstrated that sera of naïve animals were unable to agglutinate any tested 63

75 strain of S. Typhimurium. Serum from animals inoculated with JSG3675 were unable to agglutinate LPS and flagellin-deficient S. Typhimurium (JSG880) and exhibited minimal agglutination of flagellin-deficient strain JSG1190 FliC - /FljB -. Strong agglutination of JSG210 was observed by these sera and maximal agglutination occurred with FliC expressing JSG1145 and JSG3675, indicating that infection with JSG3675 may have resulted in antibody development directed at FliC. 64

76 Figure 12. Oral virulence of S. Typhimurium strains and protection following inoculation with JSG3453. A) BALB/c mice were administered 2x10 6 CFU of wild type JSG210 (black), JSG3453(ΔyihO/ΔrpoS 455, red) or JSG3672 (blue) via oral gavage and monitored for survival for 15 days. B) 8 weeks post infection, animals previous inoculated with avirulent strain JSG3453 (ΔyihO/ΔrpoS 455, blue) or naive animals (black) were administered 2x10 6 CFU of wild type S. Typhimurium and monitored for survival. Survival curves were generated by use of Kaplan- Meier estimator and distributions of challenge group vs. control mice were compared via the logrank (Mantel-Cox) test. P >

77 Discussion: In the present study we have described a number of mechanisms by which the O- ag capsule may facilitate bacterial evasion of immune detection and clearance within the host. Western blotting of purified bacterial polysaccharides indicated the presence of a high molecular weight polysaccharide (>150kDa) exhibiting reactivity with α-o-ag capsule antibody, which was absent in bacterial strains carrying isogenic mutations in O- ag capsule operon genes. O-ag capsule production was not detected in LPS deficient mutants (ΔgalE JSG1221, ΔrfaD JSG1676, ΔLPS JSG880). Conversely, all O-ag capsule mutants exhibited wild-type LPS implying that synthesis of O-ag capsule occurs downstream of, and dependent upon LPS biosynthesis or surface assembly. Preliminary screening of capsular production revealed that of tested O-ag operon mutations, ΔyihO (putative transmembrane permease) exhibited the greatest decrease in capsular production. Dot blotting of fixed whole cells demonstrated a decrease in capsular polysaccharide on the bacterial surface in JSG3672 ΔyihO, JSG3453 (ΔyihO/ΔrpoS 455 ) and JSG3675 ΔyshA-yihW mutants, with the greatest observed decrease in capsular production occurring in JSG3453 (ΔyihO/ΔrpoS 455 ). O-ag capsule production was restored in JSG3691 (PyihO, complementation of JSG3672 ΔyihO with yiho in pjm1). In contrast to our results, Gibson et al. reported that in S. Enteritidis, O-ag capsule could be detected in whole cell lysates of ΔyihO, and thus hypothesized YihO to be involved in translocation of the completed capsule to the surface of the bacterial cell 129. In this study, the capsule was not detected on the surface or in whole cell lysates of ΔyihO, possibly indicating that S. Typhimurium relies on YihO for capsular assembly as well as 66

78 translocation. Alternatively, this result may also be due to differing antibody adsorption using a ΔgalE mutant in the previous study in contrast to the ΔyshA-yihW adsorbtion employed here, which may have resulted in detection of epitopes present only in the fully assembled capsule in our study or cross reactivity with other galactose-containing surface structures in the previous work. The O-ag capsule was observed to be heterogeneously expressed on cells grown in LB at 37 C and appeared to be most highly associated with aggregated cells, which supports previous studies demonstrating O-ag capsule production to be co-regulated with extracellular matrix components. Although Gibson et al. reported minimal expression of O-ag capsule operon genes during growth at 37 C, they also observed that transcription was highly active during in vivo infection and RT-PCR has indicated expression in stationary phase cultures grown at 37 C. Initial experiments were conducted with JSG3453 (ΔyihO/ΔrpoS 455 ), which was observed to be non-rdar on Congo Red plates. Complementation of this mutation with yiho on low copy number pjm2 (yiho in pwsk129) and high copy number pmj1 (yiho in puc18) plasmids failed to restore RDAR morphology. We hypothesized due to failure of complementation that the non-rdar morphology might have been a result of a second site mutation, possibly due to a mutation in rpos. Sequencing confirmed the presence of a premature stop codon in rpos, which was not present in background strains, but appeared to have occurred following the introduction of cam R in place of yiho. This rpos mutation in JSG3453 (ΔyihO/ΔrpoS 455 ) resulted in similar sensitivity to hydrogen peroxide as that displayed by RpoS- strains JSG1747 ΔrpoS and JSG1577 (ΔrpoS 237 ), although these strains exhibited slightly differing ECM production phenotypes from each 67

79 other and from JSG3453 (ΔyihO/ΔrpoS 455 ). The observed rpos mutation in JSG3453 (ΔyihO/ΔrpoS 455 ) may have occurred during the multiple laboratory passages involved in the Datsenko and Wanner method of homologous recombination, in which mutants are struck on various antibiotics and grown at different temperatures over the course of several days to remove plasmids. The mutation appears to have first arisen in the initial steps of the mutagenesis to produce strain JSG3453 (ΔyihO/ΔrpoS 455 ). Growth of the wild type JSG210 background strain carrying pkd46 lambda red plasmid revealed no smooth colonies, however the intermediate strain ΔyihO::cam R, in which yiho was replaced with cam R, was analyzed for non-rdar morphology and observed to exhibit both RDAR and non-rdar colonies. Patching of a single RDAR colony consistently yielded a mixture of both morphotypes, while patching of non-rdar colonies yielded only non-rdar, indicating that this mutation is stable and unidirectional. Colony PCR was used to screen variants and demonstrated that both smooth and rough variants still carried the yiho mutation. RpoS mutations have been reported to arise spontaneously in laboratory passage, but have also been postulated to arise paradoxically, as a result of intracellular stress 192. RpoS is a positive regulator of AgfD and in E. coli, rpos mutants have decreased expression of yiho and other O-ag capsule operon genes 193. In S. Typhi, RpoS has been shown to fine-tune expression of the Vi-ag capsule 172. It is possible that the observed spontaneous mutations in rpos arose as compensatory second-site mutations to avoid lethal intracellular accumulation of capsular precursors as spontaneous non- RDAR mutations have been previously reported for mutants involved in LPS O-ag assembly 194. No such mutations were observed to arise during laboratory passage of JSG3675 ΔyshA-yihW, supporting this hypothesis. 68

80 Although the precise origin of this mutation is unknown, in JSG3453 (ΔyihO/ΔrpoS 455 ) the addition of an inactivating mutation in rpos had the effect of exaggerating phenotypes observed in ΔyihO mutations individually, with the exception of in vivo virulence. JSG3453 (ΔyihO/ΔrpoS 455 ) exhibited greater abrogation of capsular expression than JSG3672 (ΔyihO), which corresponded to an increase in serum sensitivity relative to ΔyihO. Laboratory strains carrying mutations in rpos were analyzed for alterations in O-ag capsular production and while JSG1577 (ΔrpoS 237 ) displayed slightly decreased capsular expression relative to wild type, JSG1747 ΔrpoS exhibited increased capsular expression. Sequencing results indicate the observed rpos mutation to occur at a location approximately half way through the RpoS open reading frame. Although JSG1577 (ΔrpoS 237 ) JSG1747 (ΔrpoS) and JSG3453 (ΔyihO/ΔrpoS 455 ) exhibit similar sensitivity to hydrogen peroxide, which is frequently used as a functional measurement of RpoS activity, assays of biofilm formation, ECM production and O-ag capsular production demonstrate clear phenotypic differences among these mutants. This may indicate that even if truncated, RpoS could still retain some functional activity. At present we do not have a strain carrying an identical mutation to JSG3453 (ΔyihO/ΔrpoS 455 ) in rpos alone, complicating interpretation of the effects of ΔrpoS 455 on the observed ΔyihO phenotypes. Neither of the single rpos mutations (JSG1747 ΔrpoS, JSG1577 ΔrpoS 237 ) resulted in alterations in LPS, flagella or serum sensitivity nor has this been previously reported as an effect of rpos mutation in S. Typhimurium. Although complementation of yiho was not able to restore the RDAR phenotype in strain JSG3453 (ΔyihO/ΔrpoS 455 ), it did restore flagellar phase and production to levels approaching that observed in wild type cells, indicating that this phenotype is due to the loss of YihO and 69

81 not RpoS. The role of RpoS in capsular assembly and production is less clear, as complementation of yiho was sufficient to restore capsular production in JSG3672 (ΔyihO) but was unable to restore capsular production in JSG3453 (ΔyihO/ΔrpoS 455 ). Ability to survive incubation in human sera was increased in the presence of the O-ag capsule. Cells deficient in LPS and O-ag capsule were much more sensitive to serum killing than those lacking only O-ag capsule, indicating that the majority of serum resistance come from LPS production. However, among strains with wild-type LPS production of O-ag capsule afforded an increase in serum resistance. This resistance was proportionate to the amount of O-ag capsule observed on the bacterial surface, with JSG3453 (ΔyihO/ΔrpoS 455 ) mutants (which produced normal LPS but no O-ag capsule) exhibiting the greatest increase in serum sensitivity relative to the wild type. This is consistent with the observed heterogeneous expression of O-ag capsule, which indicates that only a portion of the culture population would be afforded the increased serum sensitivity associated with capsular production. Future studies employing constitutive overexpression of O-ag capsule may result in greater serum resistance. Antibody-mediated complement deposition is known to be critical for control of invasive Salmonella by facilitating deposition of the membrane attack complex and enhancing phagocytosis and the oxidative burst. A modified far western blot of bacterial polysaccharides electrophoresed and transferred to PVDF, incubated in pooled donor sera and probed for human IgG demonstrated that serum IgG in these donors appears to be primarily responsive to targets of much lower molecular weight than O-ag capsule. This is consistent with previous reports of LPS-directed antibodies reacting primarily with shorter-length LPS Similar results were obtained when probing blots for C3, 70

82 showing that the region of C3 deposition corresponds to observed locations of the IgG binding. The proportion of LPS molecules with shorter O-ag repeats appeared to be increased in the absence of the O-ag capsule. This may support a model in which O-ag capsule production shifts the population of bacterial surface polysaccharides to a length and/or structure that is less recognizable by serum antibodies while simultaneously shielding and/or decreasing the production of those that are. Although bacterial surface polysaccharides are quite stable, is possible that the process of electrophoreses may have eliminated any antigenic targets or alternative complement activating moieties present in the O-ag capsule, which limits the interpretation of these data. However, is also possible that complement fixation occurring through the antibody mediated classical pathway may primarily occur in response to lower molecular weight polysaccharides and membrane targets, as opposed to the O-ag capsule. Complement deposition may also be initiated through the alternative pathway, in which spontaneous cleavage of complement C3 occurs through interaction with hydroxyl moieties on terminal sugars Bacterial polysaccharides may facilitate evasion of this mechanism of clearance by incorporation of sugars that lack complement-activating moieties, by shedding of complement-bound polysaccharides from the cell surface or by producing structures of sufficient length to allow complement deposition and subsequent MAC formation far enough from the bacterial membrane to avoid lysis. In human blood, it is believed that the classical complement pathway is critically important for control of Salmonella; therefore production of an immunologically distinct structure capable of shielding LPS recognition by specific antibodies could be highly advantageous 200,201. Conversely, the O-ag capsule might also provide a survival 71

83 advantage in the sera of HIV + patients, in which there is an overabundance of nonspecific or inhibitory IgG; in these patients, production of O-ag capsule could be a virulence advantage by providing an additional binding target, thereby recruiting more inhibitory IgG to the bacterial surface. O-ag capsule deficient mutants exhibited a dramatic increase in production of surface-associated FliC and no detectible production of FljB flagellin. This phenotype has not been previously reported to be associated with any other EPS mutations (ECA, LPS, K1 capsule in E. coli). This indicates that FliC overproduction is not a generalized pleiotropic effect associated with dysregulation of surface polysaccharides but may be indicative of inversely regulated production of these two molecules, as has been reported in Pseudomonas and S. Typhi Over 50 genes are involved in flagellar synthesis and although it is important for motility and invasion 203, bacterial flagellin is highly a highly immunostimulatory PAMP 204 therefore its expression is tightly regulated. The presence of multiple redundant mechanisms by which Salmonella control surface exposure of flagella in differing compartments reflect the importance of this molecule in host-pathogen interactions. Mechanisms of regulation of flagellar synthesis are numerous and positive regulatory mechanisms include cyclic AMP (camp) and camp receptor protein (CRP) complex and the nucleoid protein H-NS 164,205,206 while negative regulatory mechanisms include FliA, OmpR and ClpXP protease 207,208. We assessed the protein profiles of O-ag capsule deficient mutants relative to wild type S. Typhimurium which revealed a clear difference in the levels of FliC and FljB flagellins, with FljB being predominant in wild type cells and FliC in O-ag capsule 72

84 mutants, as identified by mass spectrometry. O-ag capsule deficient mutants exhibited an increase in both secreted and cell-associated FliC. Whole cell imaging indicated that in O-ag capsule deficient mutants, FliC was qualitatively increased and was associated with the cell surface in what appeared to be assembled flagella while western blot analysis similarly indicated a several fold increase in FliC production relative to wild type cells. It might be expected that increased production of a functional motility apparatus would result in an increase in cellular motility, however no significant increase in motility was observed in O-ag capsule mutants, possibly indicating that FliC was not being assembled into functional flagella, was primarily secreted, or that FljB flagellin provided similar levels of motility to wild type cells. Although we have not established a mechanism for this increased production of FliC, reports of abnormally regulated flagellar production are numerous and indicate that interference with flagellar regulation can occur at multiple levels. Perhaps loss of capsular expression alters intracellular dynamics thereby affecting activity of Hin recombinase, accumulation of ClpX, camp/crp activity or FliAmediated repression of flic. Because overproduction of FliC is observed in single yiho deletions as well as a whole operon deletion, further testing is necessary to determine the mechanism of altered flagellar regulation in O-ag capsule mutants and if FliC overproduction is mediated by yiho specifically or more generally as a secondary effect due to the loss of O-ag capsular expression. While previous studies have shown that regulation of flagellar expression and phase variation provides a survival advantage in vivo, the effects of monomeric composition and the mechanisms by which this advantage are conferred are less clear. Surprisingly, expression of flagella is not required for murine infection by Salmonella

85 and although it greatly enhances virulence, several pieces of evidence indicate that regulated expression of FliC (phase 1 flagella) may be more important for in vivo infection than FljB (Phase 2 flagella). Phase variation provides bacteria with the ability to express multiple, immunologically distinct forms of major surface antigens and is known to enhance virulence, as locking S. Typhimurium in either phase reduces virulence relative to wild type. Interestingly, although both forms of flagella appear equal in terms of their mechanical function, in vivo infections revealed that bacteria capable of expressing only FljB are avirulent, while those expressing only FliC retain virulence 189. There also appears to be an evolutionary selection bias for retention of flic, as monophasic isolates and serovars (such as S. Typhi), which express only one flagellin subtype, overwhelmingly express FliC, having lost FljB 90% of the time 210. In light of this possible functional preference for FliC, it seems reasonable to speculate that phase I flagella may be more likely to be expressed in vivo and possibly be the more immunostimulatory form of the two monomers. Indeed, animal studies using flic::gfp reporter fusions reveal that FliC is the primary flagellar subunit expressed in vivo 211 but that expression is highly restricted to specific host compartments and upon host cell invasion, intracellular bacteria rapidly repress expression below the threshold of T-cell activation and INF-γ secretion 212. This repression is coordinately regulated by PhoP in concert with other cell surface modifications which reduce inflammatory properties 213. Although bacteria preferentially expressing FliC exhibit a 5-fold competitive index relative to wild type 214, it has also been reported that S. Typhimurium constitutively overexpressing FliC exhibit increased macrophage cytotoxicity 215 and 74

86 attenuated virulence in Balb/C mice 216 indicating that properly regulated surface expression of FliC is important to bacterial virulence. Polymerized flagellin is less immunostimulatory than flagellin monomers but both FliC and FljB generate a rapid host inflammatory response mediated by TLR5 and/or NLRC4, which is crucial for coordination of an effective innate and adaptive immune response 217. Extracellular flagella activate toll-like receptor 5 (TLR-5) leading to a pro-inflammatory response and cytokine induction 218 while detection of cytosolic flagellin in a caspase-1-dependent manner via NLRC4 results in apoptotic or pyroptotic cell death 219. Although some have reported that FliC and FljB are equally capable of inducing inflammation 220, several groups have indicated that FliC is the more inflammatory of the two monomers and is the main Salmonella flagellin leading to IL-8, IL-18, IL-1β secretion and early death of macrophages Studies examining adaptive immunity to Salmonella have shown that FliC is the dominant epitope recognized by CD4 + T-cells following natural infection 222. Immunization of mice with purified FliC protein can provide cross-serovar protection against lethal challenge with Salmonella 223 and live Salmonella vaccine strains are benefitted by the presence of cellassociated flagella 224 specifically through the induction of antibodies generated against FliC. In vivo however, the high level of tissue-specific restriction of FliC expression by wild-type cells interferes with generation of a protective CD4 + T-cell response 212. Our data are consistent with previous reports indicating that S. Typhimurium 14028S which is capable of phase variation express both FliC and FljB in laboratory culture but tend to express more FljB 190. O-ag capsule deficient mutants exhibited an increase in cell-associated and secreted FliC and no detectible FljB. Complementation of 75

87 yiho restored the expression of FliC and FljB to levels approaching wild type. These data indicate that O-ag capsule mutants experience altered flagellar regulation relative to wild type cells and may be impaired for their ability to undergo phase variation. Previous reports indicate that the infecting population of S. Typhimurium in the gut lumen exhibit phenotypic variants in invasive and inflammatory properties 225 and even in strains which are phase locked to express only FliC, 20-40% of bacteria express no flagellin in vivo 211. It has been proposed that heterogeneous sub-populations of bacteria permit rapid response to host environmental fluctuations, with FliC + bacteria stimulating inflammation and cellular infiltration to the Peyer s patches, while systemic dissemination requires restriction of FliC expression. It is possible that expression of O-ag capsule within a subset of cells is advantageous at key steps of infection by facilitating decreased surface display of FliC, shielding LPS and enhancing serum resistance. 76

88 Chapter 3: The O-Ag Capsule: Biofilms and Chronic Disease Abstract: Chronic carriage of Salmonella Typhi is mediated primarily through the formation of bacterial biofilms on the surface of cholesterol gallstones. Biofilms, by definition, involve the formation of a bacterial community encased within a protective macromolecular matrix. Previous work has demonstrated the composition of the biofilm matrix to be complex and highly variable in response to altered environmental conditions. Although known to play an important role in bacterial persistence in a variety of contexts, the Salmonella biofilm matrix remains largely uncharacterized under physiological conditions. Initial attempts to study matrix components and architecture of the biofilm matrix on gallstone surfaces were hindered by the auto-fluorescence of cholesterol. In this work we describe a method for sectioning and direct visualization of extracellular matrix components of the Salmonella biofilm on the surface of human cholesterol gallstones and provide a description of the major matrix components observed therein. Confocal micrographs revealed robust biofilm formation, characterized by abundant but highly heterogeneous expression of polysaccharides such as LPS, Vi and O-antigen capsule. CsgA was not observed in the biofilm matrix and flagellar expression was tightly restricted to the biofilm-cholesterol interface. Images also revealed the presence of preexisting Enterobacteriaceae encased within the structure of the gallstone. These 77

89 results demonstrate the use and feasibility of this method while highlighting the importance of studying the native architecture of the gallstone biofilm. A better understanding of the contribution of individual matrix components to the overall biofilm structure will facilitate the development of more effective and specific methods to disrupt these bacterial communities. Introduction: Salmonella enterica serovar Typhi is the causative agent of typhoid fever, a severe systemic illness responsible for over 21 million new infections annually 18. Although rare in western countries, typhoid fever continues to pose a major threat to public health in regions of Asia, Africa and South America 226. S. Typhi is a highly virulent, human restricted pathogen, which is typically transmitted through the fecal-oral route by ingestion of contaminated food or water. Characterized by high fever and malaise, the illness usually subsides within 2-4 weeks with effective treatment 227. However, 3-5% of infected patients go on to become chronic asymptomatic carriers 55,56, continuing to shed viable bacteria in their feces for over a year and serving as a critical reservoir for the continued spread of disease 228. A primary mechanism of chronic S. Typhi carriage is believed to be formation of bacterial biofilms on the surface of cholesterol gallstones in the gallbladder 160. Unfortunately, there remains no effective non-surgical treatment to resolve chronic S. Typhi carriage in the presence of gallstones 60,139. Biofilm formation is a widely conserved mechanism of bacterial persistence characterized by adherence to a surface and subsequent formation of a protective extracellular matrix which composes up to 90% of the biofilm mass 229. Biofilm matrices are complex aggregations of hydrated macromolecules typically composed of a combination of extracellular polymeric 78

90 substances (EPS) such as proteins, lipids, polysaccharides and nucleic acids 229,230. Encased within the biofilm matrix, bacteria become highly resistant to mechanical disruption, clearance by the host immune system and many otherwise effective antimicrobial therapies 142. Resolution of biofilm-mediated chronic infections is notoriously difficult and poses a significant problem to the medical community. Recent attempts at development of therapeutics to resolve recalcitrant biofilm-mediated infections by organisms such as Staphylococcus, Pseudomonas, Haemophilus, Escherichia and Bacillus have focused on targeting key structural components of the biofilm extracellular matrix (ECM) The composition of the Salmonella spp. biofilm matrix has been demonstrated to be highly variable depending upon the environment and substratum upon which it is formed 147,151,235. Biofilms forming on cholesterol gallstones in the bile-rich gallbladder are exposed to a unique environment and in these conditions, the composition of the ECM as well as the necessity of known matrix components likely differs from other environments. Proteinaceous factors proposed to be involved include flagella and curli fimbriae composed of polymerized CsgA subunits 236. Matrix polysaccharides may include O-ag capsule or Vi-ag capsule, LPS, colanic acid and cellulose. Although previous work from our lab and others have found variable roles for O-ag and Vi-ag capsule in multicellular behavior 61,129,132, these protective surface polysaccharides are closely associated with the outer membrane of S. Typhimurium and S. Typhi respectively, making them a likely component of the biofilm matrix. Many previous attempts to study the ECM in salmonellae have characterized overall biomass or apparent biofilm quality using mutants deficient in specific ECM components or following 79

91 enzymatic treatment 61,147,237. These studies have greatly furthered our understanding of the composition and function of the biofilm matrix, but have also revealed it to be a complex and highly variable structure. Possible drawbacks to this method of examining the overall biofilm structure instead of individual components include lack of enzymatic or quantitative specificity, potential compensatory production of alternative ECM components. Furthermore, studies of the ECM in nontyphoidal salmonellae are frequently conducted at low temperatures (<28 C) but regulation of many genes involved in biofilm formation is highly temperature dependent, limiting the applicability of these findings to environmental and not in vivo conditions. In this work we sought to characterize the gallstone biofilm by sectioning and specifically labeling individual ECM constituents within the native structure of biofilms grown on the surface of patient gallstones at 37 C. We have optimized the sectioning procedure and demonstrated its potential for use in both human and animal studies through preliminary analysis of S. Typhi and S. Typhimurium biofilm composition and architecture. Our results indicate that the biofilm matrix is largely composed of polysaccharides under the conditions tested and that proteins and nucleic acids represent relatively minimal contributors to the overall ECM biomass. This work complements previous studies and highlights the importance of understanding the spatial organization of ECM constituents within the architecture of the biofilm matrix. Materials and methods: Bacterial strains and growth conditions: Wild-type reference strains for S. Typhi Ty2 (gift of Dr. Renato Morona, University of Adelaide) or S. Typhimurium 14028S (ATCC) were struck on LB agar, cultures were inoculated from an isolated colony and grown 80

92 overnight in LB liquid medium with aeration at 37 C. For matrix inducing conditions used in western blotting and microcolony staining of CsgA/FliC, cells were grown on LB no salt agar for 5 days at 22 C. Statistics: All statistical analysis was performed with Prism 5 software (GraphPad Software, Inc.). In bar graphs, indicated error lines represent standard deviations and unless otherwise stated, values reported represent the mean of three independent experiments carried out in triplicate. To determine significance of observed differences between groups, Student s t-tests or two way ANOVA were performed. Observed differences were considered statistically significant compared to wild type at a P value of <0.05. Asterisks indicate *P<0.1, **P<0.01, ***P< respectively. Gallstone biofilm formation: Patient gallstones were placed in 15mL borosilicate glass tubes using sterile forceps. To determine pre-existence of viable organisms, stones were incubated overnight in 5mL of sterile LB, which was subsequently plated on LB and XLD agar. For biofilm cultivation, overnight bacterial cultures were normalized to OD , diluted 1:100 ( 2x10 6 CFU/ml) and inoculated into LB +/- 3% ox bile. Biofilm cultures were maintained at 37 C with aeration and media was replaced every 24 hours for 6 days to allow biofilm growth. On days 2 and 6, planktonic culture removed prior to washing was enumerated by plating on LB and XLD agar. Ethics Statement: This project did not involve interaction with human subjects. Laboratory experiments involved only existing, previously collected pathological specimens obtained from Dr. Wayne Schwesinger 238 Department of General Surgery, University of Texas Health Science Center at San Antonio, San Antonio, Texas. Specimens were not collected specifically for the current research and were anonymously 81

93 analyzed having been assigned de-identified numbers, which could not be traced back to the patients and with no protected health information available to the researcher. There is no link between patient identifiers and the data. As such, this research does not meet the definition of research involving human subjects per the Ohio State University Human Research Protection Program Policies and Procedures, IRB Policy Committee revision 02/02/13 and is not subject to approval or exemption. Animals were housed and used in accordance with established guidelines set forth by the Ohio State University Institutional Animal Care and Use Committee (IACUC) in accordance with U.S. National Research Council standards and Research Animal Welfare Act. All experimental protocols and standard operating procedures were subject to prior ethical and procedural approval (IACUC Protocol # 2009A0057-R1) and researchers were subject to ongoing oversight to ensure strict adherence to study protocol. Animals were closely monitored throughout the course of infection and every effort was made to minimize discomfort. Infected animals were evaluated a minimum of twice daily and were humanely euthanized by CO 2 asphyxiation if moribund or at predetermined study endpoints. Murine model of chronic carriage: Female 129X1/SvJ mice (Nramp + / + ) were fed a normal or lithogenic diet (supplemented with 1% cholesterol) for 8-10 weeks prior to infection. The majority of animals fed this high-cholesterol diet developed cholesterol gallstones in their gallbladders. Mice were then inoculated intraperitoneally with 1x10 4 CFU of washed stationary phase S. Typhimurium or PBS in negative controls. Animals were sacrificed at 15 and 21 days post-infection by CO 2 asphyxiation. Animals were dissected immediately following sacrifice and feces, liver, gallbladder, gallstones, bile and spleen were weighed, homogenized in sterile 1x PBS, serially diluted and 82

94 enumerated on Salmonella/Shigella agar and/or LB. Initial experiments demonstrated that animals fed the high cholesterol diet occasionally exhibited non-salmonella bacteria in the gallbladder and feces, therefore subsequent experiments included an additional plating of these samples on Salmonella Shigella agar to verify the identity of recovered bacteria. Colonies of Salmonella were randomly selected from 10 plates per group, per timepoint and screened by colony PCR for presence or absence of desired mutation. All screened colonies matched the respective infecting strain. Gallstones were removed and processed for SEM. Scanning electron microscopy of gallstones and cholesterol biofilms: Samples were fixed for 24 hours at 4 C in 2.5% glutaraldehyde in SEM buffer (0.1M sucrose, 0.1M phosphate, ph 7.4). Following fixation, samples were washed 2x with buffer, dehydrated in increasing concentrations of EtOH and chemically dried with hexamethyldisilazane and dried overnight. Dried samples were mounted on aluminum pin-stub mounts, and gold sputter-coated. SEM microscopic observations were conducted using a FEI Nova Nano SEM electron microscope at the OSU Campus Microscopy and Imaging Facility (CMIF). Flow-through system biofilm assay: The Stovall flow cell chamber was employed to compare biofilm formation of bacterial strains. Glass coverslips in flow-cell chambers were coated with 8mG of cholesterol dissolved in anhydrous ether and flow-cell apparatus was assembled as previously described 239. A peristaltic pump controlled the rate of media (either LB or LB with 3% bile) flow through the cell to 270µL/min. Media flow was started to fill chamber and halted at which time 2x10 8 CFU of mid-logarithmic bacterial cultures were inoculated via syringe into the flow-cell gasket and permitted to 83

95 adhere prior for 2h prior to re-starting media flow. Biofilms were grown for 96h at 37 C prior to removal of slide from chamber for crystal violet quantitation or preparation for SEM. Microtiter biofilm assay: High-throughput quantification of biofilm formation was conducted using sterile 96-well polystyrene or cholesterol-coated glass microtiter assay plates at 37 C. Cholesterol coating was achieved by dissolving cholesterol in anhydrous ether (40mG/mL) and immediately pipetting 100µL into each well; plates were left to dry completely in fume hoods overnight. Overnight bacterial cultures were normalized to A , back diluted 1:100 (2x10 6 CFU/mL) and 100µL was added to each well. Biofilm formation was compared to wells incubated with cholesterol and sterile media alone. Crystal Violet quantitation of biofilm formation: On the final day of biofilm growth, culture media was removed, biofilms were gently washed 2x with sterile 1x PBS and heat fixed for 1h at 60 C. Biofilms were stained with 33% crystal violet staining solution (10mL gentian violet, 18 ml 1X PBS, 1 ml methanol, 1 ml isopropanol) for 5 minutes. Volume of crystal violet stain added was 110% of incubated culture volume used for biofilm cultivation to ensure complete coverage and staining of all biomass. Biofilms were washed 4x with 1X PBS to remove excess dye and stained biofilms were released with 33% acetic acid solution (33mL glacial acetic acid, 66mL 1X PBS), transferred to a clean microtiter plate and optical density was read at 570nm and recorded. Background staining was calculated as A 570 from media alone control and subtracted from biofilm readings. Screening for extracellular matrix production: Bacterial colony morphology was assessed after 5 days at 22 C on Luria-Bertani agar plates without salt (LBNS), 84

96 supplemented with Congo Red (40 µg/ml) and coomassie brilliant blue dyes (20 µg/ml) (36). Cellulose production was assessed on LBNS agar containing 10 µg/ml calcofluor (Fluorescent Brightener 28, Sigma, USA) and visualized under UV light at 366 nm following 5 days of growth at 22 C. Wild type S. Typhimurium and JSG1748 (ΔbcsA::kan) and JSG2956 (ΔagfA) were used as a positive and negative controls for cellulose and curli production respectively. Western blotting. The CsgA monomeric subunit of curli fimbriae was detected in whole cell lysates using a polyclonal antisera generated against E. coli CsgA, (gift of Dr. Matthew Chapman, University of Michigan). Western blots were conducted as previously described 240. In brief, cultures were grown as outlined above, cells were scraped from plates or pelleted from culture media, resuspended in 1mL 1xPBS ph 7.4 and normalized to A Cells were pelleted again, and the supernatant was removed and pellets were resuspended in 80µL of 88% formic acid to depolymerize CsgA. Samples were dried using a Speed-Vac and resuspended in 35 µl H 2 O. Lamelli loading buffer was added to a final volume of 70 µl and 7.5 µl was loaded onto a 15% SDS-PAGE gel along with Biorad Western C chemiluminescent molecular weight standard. Proteins were electrophoresed at 60V until the dye front reached the bottom of the gel. Proteins in the gel were transferred for 1h at 60V to MeOH-activated PVDF membrane and blocked overnight in 5% BSA in TBS. The membrane was incubated with Anti CsgA antibody (1:10,000 in 5% BSA/TBST for 2h, 22 C) washed in TBST (3 x 15min) and incubated with HRP conjugated goat α-rabbit (1:12,500 in 5% BSA/TBST for 2h, 22 C). The membrane was washed in TBST (3 x 15min) and visualized using the Bio-Rad ChemiDoc system. 85

97 Gallstone preparation, fixation, embedding and sectioning. Gallstones were prepared by the Comparative Pathology and Mouse Phenotyping Shared Resource at the Ohio State University. Following 6 days of biofilm growth, media was removed and plated as described above. Stones were washed twice in sterile PBS prior to immersion fixation in 10% neutral buffered formalin (NBF) for 24 hours at room temperature. Although several alternate methods were attempted (as discussed in results section) optimal results were achieved with the protocol outlined herein. Fixed samples were rinsed in running tap water for 30 minutes and transferred into a 1:1 solution of 50% aqueous formic acid and 20% aqueous sodium citrate overnight (approximately 16 hours). After gently rinsing in running tap water for 30 minutes, gallstones were transferred back into 10% NBF for standard overnight tissue processing in a Tissue Tek V.I.P vacuum infiltration processor (Sakura Finetek USA, Inc., Torrance, CA). In brief, samples were incubated in 10% NBF (2 x 30min), 70% ethanol (1 x 2hrs), 95% ethanol (3 x 60min), 100% ethanol (1 x 60min, 1 x 90min), toluene (1 x 60min, 1 x 90min) and submerged in paraffin (3 x 60min) prior to carefully embedding in fresh molten paraffin (Paraplast Plus, Sigma- Aldrich, St. Louis, MO) and then allowed to cool. Four micron sections were obtained using a RM2255 automated microtome (Leica Biosystems, Buffalo Grove, IL). Tissue sections were floated in a 48 C water bath and placed on Surgipath Superior Adhesive Slides (Leica Biosystems). The slides were air dried overnight and then transferred into a 60 C oven for 30 minutes. Slides were deparaffinized and hydrated to distilled water by washing in xylene (3 x 5min), 100% ethanol (2 x 1min), and 95% ethanol (2 x 1min), and then briefly rinsed in and transferred to distilled water for immunofluorescent (IF) or Immunohistochemical (IHC) staining. 86

98 Immunohistochemical (IHC) staining: Slides were treated with DakoCytomation Target Retrieval Solution (Dako, Carpinteria, CA) in a Decloaking Chamber (Biocare Medical, Concord, CA) heated to 125 C and then cooled to 90 C for 10sec, before cooling with the lid removed for 10 min to unmask epitopes for Salmonella detection. Slides were then transferred to a Dako Universal Training Center automatic immunostainer for all subsequent steps at RT. Endogenous peroxidase was inhibited in 3% H 2 O 2 for 5min, followed by serum-free protein block (Dako) for 10min. Sections were incubated with rabbit polyclonal anti-salmonella antibody (1:100) for 30min (Novus Biologicals), followed by a biotinylated anti-rabbit secondary antibody (1:200, Vector Laboratories, Burlingame, CA) for 30min, and lastly avidin-biotin complex (Vector Laboratories) for 30min. Signal was developed with 3,3' diaminobenzidine tetrahydrochloride, counterstained with hematoxylin, and coverslipped prior to viewing with a light microscope. Immunofluorescent (IF) staining: Deparaffinized slides were blocked in sterile filtered 5% bovine serum albumin (BSA) in Tris-buffered saline (TBS) ph 7.4 overnight at 4 C. All antibodies were diluted in 5% BSA/TBS+ Tween 20 (TBST). Secondary antibodies were goat or donkey anti rabbit or mouse. Staining controls included gallstones with no bacteria stained with primary and secondary antibodies and gallstones with bacterial biofilms unstained or stained with either primary antibodies or secondary antibodies alone. Following antibody incubations (Table 1), slides were washed in TBST, rinsed briefly in sterile ddh 2 O and mounted using ProLong Gold anti-fade mounting media. Slides were allowed to dry in the dark at room temperature overnight prior to viewing on an Olympus FV1000 spectral confocal scanning laser microscope. 87

99 Results: Previous work has demonstrated that both O-ag capsule and FliC are important for biofilm formation 132,144. We sought to determine whether O-ag operon mutants, which overexpressed flagella but lacked the O-ag capsule would form biofilms in vitro or in vivo. We employed the murine model of chronic gallstone carriage to determine the necessity of O-ag capsule in establishment of chronic infection. Murine chronic infection assays: O-ag capsule deficient mutants were tested in vivo to determine their ability to persistently colonize in the gallbladder of infected mice and form biofilms on the surface of cholesterol gallstones (Fig. 13). Female 129X1/SvJ mice (Nramp + / + ) were fed a high cholesterol diet for 10 weeks to induce gallstone formation and intraperitoneally injected with either wild type S. Typhimurium, O-ag capsule deficient mutants (JSG3453 ΔyihO/ΔrpoS or JSG3675 ΔyshA-yihW) or PBS alone. Although the majority of animals fed a high-cholesterol diet developed gallstones in their gallbladders we observed that extending feeding time from 8 to 10 weeks increased incidence and size of gallstones. Animals were sacrificed at 15 and 21 days post-infection and organs were homogenized and plated to calculate bacterial burden. Neither of the O- ag capsule mutants (JSG3453 ΔyihO/ΔrpoS 455 or JSG3675 ΔyshA-yihW) exhibited a significant reduction in bacterial CFU recovered from gallstones (Fig. 13C) and the O-ag capsule operon deletion mutant JSG3675 (ΔyshA-yihW) appeared to exhibit improved colonization of gallstones. Although no specific defect in gallbladder colonization was observed in O-ag capsule deficient mutants, bacterial burdens were consistently lower in the spleen for JSG3453 (ΔyihO/ΔrpoS 455 ) at days 7 and 21 and for JSG3675 (ΔyshA-yihW) at day 21 (Fig. 13A, 13B) indicating a generalized reduction in systemic fitness and 88

100 persistence which did not correspond to a decrease in gallstone biofilm formation. Scanning Electron Microscopy: Gallstones from animals were further processed for SEM imaging analysis (Fig. 13D.I-VI). Micrographs indicated that both wild type and O- ag capsule deficient mutants exhibited robust biofilm formation on the surface of recovered murine gallstones, while the gallstones of uninfected control animals did not exhibit biofilm formation. In order to minimize the number of animals used in this study, biofilm formation was subsequently analyzed through SEM examination of ex vivo biofilms grown on the surface of patient gallstones. Biofilms of ΔyihO, ΔyihV and ΔyihW mutants were compared to wild type S. Typhimurium or gallstones incubated in PBS alone (Fig. 13D.I-VI). There was no visually appreciable difference in biofilm formation by any of the employed O-ag capsule mutants. Further assays were conducted in vitro to quantitatively compare cholesterol biofilm-formation of JSG3453 (ΔyihO/ΔrpoS 455 ) and JSG3672 (ΔyihO) to wild type S. Typhimurium. Results from the previously described tube biofilm assay (TBA) 132 were highly variable, so biofilm formation was quantified using a cholesterol-coated 96-well microtiter assay and Stovall continuous flow-cell chamber at 37 C (Fig 14). Results indicated that neither ΔyihO mutant exhibited a significant alteration in biofilm formation relative to wild type S. Typhimurium. Although no difference was observed on cholesterol or polystyrene coated surfaces at 37 C, JSG3453 (ΔyihO/ΔrpoS 455 ) exhibited decreased biofilm formation on glass at room temperature

101 Figure 13. Systemic colonization and gallstone biofilm forming abilities of S. Typhimruium in murine chronic infection. A) CFU enumeration of wild type S. Typhimurium JSG210, JSG3453 (ΔyihO/ΔrpoS 455 ) or B) JSG3675 (ΔyshA-yihW) mutants recovered from the spleen at 15 and 21 days post-infection demonstrating a decrease in systemic colonization by JSG3453 (Δ yiho/δrpos 455 ) at days 15 and 21 and a decrease in strain JSG3675 (ΔyshA-yihW) at day 21. C) CFU recovered from gallstones of mice infected with either wild type S. Typhimurium JSG210 or O-ag operon deletion JSG3675 (ΔyshA-yihW). Data represent 2 independent experiments with > 3 mice per group. Student s t test was used to determine significant differences (***= P <0.0001). B) Scanning electron microscopy of in vivo and in vitro gallstone biofilms formed by O-ag operon deletion mutants. I) Wild type JSG210, II) no bacteria control, III) JSG3453 (ΔyihO/ΔrpoS 455 ), IV) ΔyihV, V) ΔyihW, VI) JSG3675 (ΔyshA-yihW whole operon deletion). All mutants were able to form robust biofilms on the surface of cholesterol/gallstones. 90

102 Figure 14. S. Typhimurium O-ag capsule ΔyihO mutants do not exhibit altered biofilm formation on cholesterol or polystyrene in vitro. A) 24hr microtiter assay demonstrating relative biofilm forming capacities by S. Typhimurium and ΔyihO JSG3672 (ΔyihO) on polystyrene or cholesterol coated surface by B) Average 96hr biofilm formation relative to wild type S. Typhimurium formed by ΔyihO mutants in vitro on cholesterol coated surfaces. Biofilm formation was determined by the crystal violet staining method. 91

103 Figure 15. Photomicrographs of sectioned human gallstones with surface associated biofilms of S. Typhimurium. Gallstone surface (pale purple band) with associated biofilm stained with anti-salmonella polyclonal antibody (brown) visualized at A) 20x B) 100x magnification and C) DIC micrographs reveal well-preserved surface-associated bacterial communities. D) Congo Red staining indicates abundance of β-amyloid fibrils and/or acidic polysaccharides. 92

104 Detection of individual matrix components within the biofilm structure: To further investigate the Salmonella biofilm matrix for the detectible presence of O-ag capsule, flagellin or other putative matrix components, we developed a method for histologic slide preparation of human gallstones, a technique that has not previously been reported in the literature. The method described here permits visualization and specific labeling of wellpreserved bacterial biofilms grown ex vivo on the surface of patient cholesterol gallstones. As previously reported, gallstone composition was variable within a single stone and among patients but was quite consistent among stones retrieved from a single patient, reflecting the complex process of lithogenesis 241. Four different stone types were tested using this method. White (cholesterol) stones responded optimally to the processing method while brown-pigmented (calcium bilirubinate) 242,243 stones did not withstand the decalcification procedure. A number of unsuccessful methods were attempted which included directly bisecting the stones, cryosectioning, and omission of overnight processing and/or decalcification, all of which resulted in gallstone dissolution during subsequent processing steps. Optimal results were achieved using immersion fixation and mild formic acid/sodium citrate decalcification followed by overnight processing, paraffin embedding and sectioning. Bacterial biofilms remained attached to the gallstone surface throughout processing (Fig. 15A-D). The best-preserved portions of the biofilm appeared to be associated with crevices and indentations in the stone surface. This may be an artifact of the rotation of the culture tubes to facilitate aeration or could reflect a preference for the increased surface contact provided by these topographical features. 93

105 Figure 16. IF staining controls demonstrating minimal non-specific fluorescence. Sectioned patient gallstones with A-C) S. Typhimurium biofilm, D-F) S. Typhi biofilm or G-I) no bacteria control. All sections were stained with primary and secondary antibodies and counterstained with DAPI. Micrographs of control gallstones (G-I ) incubated in sterile media, exhibit regions of DAPI staining resembling pre-existing bacterial aggregates within the gallstone. 94

106 Figure 17. Visualization of polysaccharide antigens in DAPI counterstained Salmonella biofilms grown on human gallstones. Upper panel A) Biofilms of S. Typhi (I-IV) and S. Typhimurium (V-VIII) grown in LB without bile. Sections were stained for LPS (I, V) Vi-ag (III) or O-ag capsule (VII). Merged images of LPS and Vi-ag/O-ag capsule (IV, VIII) illustrate the heterogeneous detection of these polysaccharides within the biofilm. Lower panel B) Representative image of S. Typhi biofilm section grown in LB with 3% bile, counterstained with DAPI and fluorescently labeled for LPS and Vi-antigen. Confocal (I) and DIC (II) images demonstrate the intimate association of the biofilm with gallstone surfaces and the abundance of polysaccharides in the extracellular matrix. 95

107 Figure 18. Flagella and CsgA detection in bacterial microcolonies and whole cell lysates. A) Although minimally detected in the gallstone biofilm extracellular matrix, confocal scanning micrographs demonstrate that flagella (red) and CsgA (green) are readily detectable in bacterial microcolonies grown in matrix inducing conditions (LBNS, 22 C, 5days). B) Western blot detecting CsgA ( 15kDa) in whole cell lysates of S. Typhimurium wild type JSG210 or ΔcsgA reveal CsgA production at 22 C but not at 37 C. 96

108 Gallstone sectioning procedure Immunofluorescent (IF) staining controls (Fig. 16) revealed minimal, low-level background fluorescence, likely due to nonspecific staining as well as background autofluorescence from cholesterol and formalin fixation (Fig. 16B, 16E, 2H) but this was nominal as compared to reactivity of specifically labeled components. Examination of control gallstones incubated in sterile LB alone and similarly stained for surface antigens, revealed the presence of pre-existing bacterial aggregates both on the surface and several layers into the interior of the stone (Fig. 16G- 16I), although no bacteria was observed at the center of the gallstone as would be expected of an object that had acted as a nucleating factor for initial cholesterol precipitation. The suspected bacterial layer in the stone was unreactive with monoclonal S. Typhi and S. Typhimurium antibodies (Fig. 16H) but strong reactivity was observed with polyclonal anti-salmonella antibody (Fig. 15A-B) reflecting reactivity with other salmonellae or Enterobacteriaceae. Heterogeneous detection of LPS and Capsule: Sectioning and labeling of ex vivo biofilms revealed that polysaccharide and proteinaceous components could be readily detected and visualized in the Salmonella gallstone biofilm matrix (Fig. 17). LPS (Fig. 17A-I, 17A-V, 17B-I) and capsular polysaccharides (Fig. 17A-III, 17A-VII, 17B-I) were clearly visible within the biofilm structures of both S. Typhi and S. Typhimurium. Interestingly, within the biofilm, heterogeneous detection of both LPS and capsule was observed, with many DAPI-positive cells staining for capsule alone, LPS alone or both (Fig. 17A-iv, 17A-vii, 17B). While Vi-ag was readily detected within the S. Typhi biofilm, LPS staining occurred on fewer than 20% of the cells staining with DAPI (Fig. 97

109 17B). Similar results were observed in free bacterial microcolonies and biofilm sections stained for LPS or capsule alone (data not shown). CsgA is not detectible in the gallstone biofilm matrix at 37 C. Although previous reports have indicated proteinaceous components such as curli fimbriae (CsgA) and flagella to be important constituents of the biofilm matrix 236, neither molecule was abundantly detected within the gallstone biofilm grown at the physiologically relevant temperature of 37 C (data not shown). To ensure that IF staining techniques employed were suitable for detection of ECM components within densely packed bacterial communities, free bacterial microcolonies grown in matrix-inducing conditions (LBNS, 22 C x 5d) displaying dry and rough morphology 244, were fixed and stained, demonstrating that when present, these putative matrix components could be specifically labeled and visualized within the microbial community (Fig. 18A). Western blotting of whole cell lysates revealed that CsgA was undetectable in cells grown at 37 C (Fig. 18B). Flagella, although minimally expressed, were tightly localized to the surface of the gallstone at the interface between the biofilm and cholesterol surface (Fig. 19). This observation is consistent with previous findings implicating flagellar filaments in cholesterol attachment

110 Figure 19. Visualization of flagella in DAPI counterstained gallstone biofilm sections. A) Confocal micrograph of S. Typhimurium biofilm demonstrating localization of flagella. B) Magnified CSLM and C) DIC/CSLM merge indicate that flagella is restricted to the interface of the biofilm and cholesterol gallstone surface. 99