Using Natural Products to Treat Resistant and Persistent Bacterial Infections

Antimicrobial resistance is a growing threat to human health both worldwide and in the United States. Most concerning is the emergence of multi-drug resistant (MDR) bacterial pathogens, especially the ‘ESKAPE’ pathogens for which treatment options are dwindling. To complicate the problem, approvals of antibiotic drugs are extremely low and many research and development efforts in the pharmaceutical industry have ceased, leaving little certainty that critical new antibiotics are nearing the clinic. New antibiotics are needed to continue treating these evolving infections. In addition to antibiotics, approaches that aim to inhibit or prevent antimicrobial resistance could be useful. Also, studies that improve our understanding of bacterial pathophysiology could lead to new therapies for infectious disease. Natural products, especially those from the microbial world, have been invaluable as resources for new antibacterial compounds and as insights into bacterial physiology. The goal of this dissertation is to find new ways to treat resistant bacterial infections and learn more about the pathophysiology of these bacteria. Investigations of natural products to find molecules able to be used as new antibiotics or to modulate resistance and other parts of bacterial physiology are crucial aspects of the

leaving little certainty that critical new antibiotics are nearing the clinic. New antibiotics are needed to continue treating these evolving infections. In addition to antibiotics, approaches that aim to inhibit or prevent antimicrobial resistance could be useful. Also, studies that improve our understanding of bacterial pathophysiology could lead to new therapies for infectious disease. Natural products, especially those from the microbial world, have been invaluable as resources for new antibacterial compounds and as insights into bacterial physiology. The goal of this dissertation is to find new ways to treat resistant bacterial infections and learn more about the pathophysiology of these bacteria.
Investigations of natural products to find molecules able to be used as new antibiotics or to modulate resistance and other parts of bacterial physiology are crucial aspects of the included studies.
The first included study, which is reported in chapter two, details a chemical investigation of a marine Pseudoalteromonas sp. Purification efforts of the microbial metabolites were guided by testing against a resistance nodulation of cell division model of efflux pumps expressed in E. coli. These pumps play an important role in the resistance of MDR Gram negative pathogens such as Pseudomonas aeruginosa and Enterobacteriaceae. Through this process, 3,4-dibromopyrrole-2,5-dione was identified as a potent inhibitor of the RND efflux pumps and showed synergistic effects against the E. coli strain with common antibiotics including fluoroquinolones, beta-lactams, tetracyclines, aminoglycosides, and chloramphenicol. The efflux pump inhibitory mechanism was further proved through an accumulation assay with the Hoechst dye 33342.
In chapter three, we report the discovery of a 1,2-benzisoxazole with new antibacterial activity against MDR A. baumannii, a pathogen with a critical need of new treatments. This compound was produced by bacterial fermentation and synthetic preparation and shows minimum inhibitory concentrations as low as 6.25 μg/mL against a panel of four clinically relevant A. baumannii strains. Key structure activity relationships were demonstrated using synthetic analogs of the lead 1,2-benzisoxazole.
We advocate for further studies to advance the development of this compound.
The third study, describes an in vitro quiescent state of uropathogenic E. coli (UPEC) and bacteria-produced signals that can prevent this state. Quiescence was seen in the classic UPEC strain CFT073 only when grown on glucose M9 minimal medium agar plates seeded with ≤10 6 CFU. Interestingly, this quiescent state is seen in ~80% of E. coli phylogenetic group B2 multilocus sequence type 73 strains, as well as 22.5% of randomly selected UPEC strains isolated from community acquired urinary tract infections in Denmark. Furthermore, it was determined that CFT073 forms a high persister cell fraction under these growth conditions. Both the persistent and quiescent states were inhibited significantly by a cocktail of lysine, tyrosine, and methionine at concentrations relevant to those in human urine. The use of CFT073 mini-Tn5 metabolic mutants (gnd, gdhA, pykF, sdhA, and zwf) showed that both quiescence and persistence require a complete TCA cycle, but that the dormant states differ in that persistence requires a non-functional rpoS gene and quiescence does not. These results suggest that interference with these central metabolic pathways may be able to mitigate UPEC infections.
In the fifth chapter, cranberry oligosaccharides and related compounds were determined to be able to reduce the quiescent and persistent phenotypes of UPEC CFT073. This is the first report describing components of cranberry juice with the ability to modulate these important physiological aspects of UPEC and further suggests that cranberry oligosaccharides may be vital to the effectiveness of cranberry juice products in urinary tract infections.
v ACKNOWLEDGEMENTS This dissertation is the culmination of the work I have completed but it is really an achievement of many people who I need to acknowledge for their important contributions. Every one of you has made a lasting impact on my life and career and your knowledge, generosity, and companionship are things that I will always treasure, thank you.
I need to firstly acknowledge my advisor, mentor, and friend, Dr. David C.
Rowley. Dave, it has been a long but exciting journey growing into a scientist with you.
It is certainly amazing to think how much I have learned working with you: starting out as a media-maker and extract-prepping machine and now designing complex experiments and presenting research in China. Your passion for teaching and research is obvious, but your passion for mentoring and developing your graduate students might be even stronger. I am now realizing just how much of yourself you have given towards my development in the form of countless hours spent reading, editing, researching, and thinking. You once told me that it is really hard to find people that are really enjoyable to work with. I was lucky enough to find one of those people on my first try when I sat in the basement of Fogarty and asked if you had any room in your lab for another student. I only hope that one day I can make an impact on someone's career in the way you have made on mine. Thanks Dave.
Dr. Navindra P. Seeram has made an extremely positive impact on my time here at URI. While only a committee member in title, your lessons to me have gone far vi beyond that. I thank you for your encouragement, support, and wisdom throughout graduate school. You taught me to especially focus on the details and subtleties when going about my work and that these little things like committing person's name to memory after one meeting can make a huge impact. I will try my best to build myself into a professional with style and recognition that people enjoy being around, something you have mastered. Thank you Dr. David C. Smith was also part of my graduate committee but has contributed much more than that to my development. David, I especially want to thank you for your genuine interest in me as a person beyond the classroom and lab. You have taught me to always pursue what I love and that the road before me is not paved in stone, but rather something that I will build as I go. You have given me confidence that will help me pursue my next challenge, thank you.
I also need to acknowledge Dr. Paul S. Cohen. Paul, your unbridled enthusiasm for research permeates everyone around you and I am very lucky to have worked with you. Our talks always left me feeling excited and energized to complete the next task and I strive to bring that energy and commitment to my future endeavors. Thank you for your passion and your thoughtfulness and your strong spirit, I am a better person for having worked with you. Your friendship is something I will cherish forever. I will truly miss working with you.
Additionally, I would like to acknowledge all of the students past and present with whom I was able to work with and mentor. For those who taught me, your knowledge has made me a better researcher but it was your willingness to help that I am most thankful for. For those who I mentored, you have kept me humble and taught me that teaching others is the only way to truly master a skill. There are too many names to list here, but you all have contributed, thank you.
Finally, I want to thank my family. To my parents, thank you for always believing in and supporting me no matter the endeavor. While I have had to work extremely hard to viii accomplish all that I have, it was your hard work before mine that afforded these amazing opportunities for me. Every day I try to represent you in the best way that I can and I know that everything I do reflects on you as parents. This dissertation would absolutely not be possible without you and I hope you can feel a great sense of accomplishment when I finally finish my degree. Thank you so much for enriching my life, and letting me follow the path that was right for me.
To my wife: Lilli, thank you for your unwavering support, patience, love, and faith in me. Every day you inspire me to be better in my career and as a person. I know that my extended tenure in school wasn't an easy thing, but you never showed any impatience. Thank you for being my best friend. I am so lucky to have had you by my side during this journey, and I am so excited for our next one.  Tables   Chapter 2  Table 1  Infectious disease remains a significant threat to human health in the United States and worldwide. In 2010, 117,716 deaths were attributed to infections of bacterial, viral, or parasitic origins in the US. 1 As a category, this ranks infectious disease as the fourth most deadly form of illness behind only cardiovascular disease, cancer, and respiratory disease, and ahead of diabetes and Alzheimer's disease. Worldwide, infectious disease has an even larger impact, accounting for 10 million deaths in 2011. 2 This ranks it as the second most lethal disease category behind only cardiovascular disease. 2 A growing concern in the treatment of infectious disease is the drastic increase in antimicrobial resistance. This is present in bacteria, viruses, and fungi, and is complicating the treatment of infectious disease considerably. In fact, certain pathogenic bacteria now exist that are resistant to nearly all treatment options, forecasting the socalled 'post-antibiotic era.' [3][4][5][6] The problem of antimicrobial resistance is immense with an estimated 700,000 deaths worldwide each year, and these numbers are increasing. 4 It is estimated that in 2050 there will be 10 million deaths due to antimicrobial resistance globally in a single year. 4,7 These deaths would lead to an estimated net loss of 100 trillion dollars of economic output. 4,7 Deaths due to infections are not the only clinical consequence of a post-antibiotic era. Surgery, chemotherapy, and organ transplantation rely heavily on antibiotics to facilitate their success. 4,7 Continuing on the trend of increasing antimicrobial resistance would lead to a catastrophic change in lifestyle both economically and medically.

List of Figures
Among the most threatening problems in antimicrobial resistance are bacterial infections involving multi-drug-resistant (MDR) pathogens. These are especially a problem in the United States, where 23,000 people die each year due to MDR bacterial infections. 3 This is an especially common problem in the hospital setting where antimicrobial resistance is difficult to eradicate and is seen in common nosocomial pathogens. 8  Enterobacter species). 9 These organisms are in general difficult to treat and can cause significant morbidity and mortality. 9,10 With the exclusion of E. faecium and S. aureus, the ESKAPE pathogens are Gram-negative bacteria. This is important because there are in general fewer treatment options for Gram-negative bacteria and developing antibiotics against these bacteria has been historically less successful. 11 The World Health Organization recently released a ranking of highest priority bacteria for antibiotic development, and all three critically important organisms are MDR Gram negative bacteria. 12 A major reason that the problem of antimicrobial resistance is increasing is the dearth of new antibiotics being approved or researched. Since 1998, only four systemic antibiotics (linezolid, daptomycin, tigecycline, and telavancin) have been approved that have novel mechanisms of action. 10 To combat this issue, antibiotic stewardship programs can help, but long term solutions exist in the drug development arena. In 2010, the Infectious Disease Society of America (IDSA) launched a campaign known as the "10 x '20 Initiative" that called for the development 10 new antibiotics approved by 2020. 10 Currently we are falling well short of the goal, with only four new antibiotics being approved since the initiative and none of these with activity against the important Gram-negative bacteria. 13 Additionally, the current pipeline of antibiotics contains only five drugs in phase II or III development with target indications of MDR Gram negative infections. 13 There is a clear void in the future treatment of infectious disease especially for MDR bacterial infections, and without changes to the current course, a post-antibiotic era is likely.

Bacterial mechanisms of resistance
Bacteria have developed mechanisms to avoid the toxic effects of antibiotics, causing resistance. Inhibiting or circumventing these biological functions can restore the activity of the antibacterial drug. Understanding the mechanisms of resistance is crucial to developing agents that can restore the potency of antibiotics, especially in ESKAPE pathogens and Gram-negative bacteria. Resistance mechanisms fall into these general categories: drug alteration/inactivation, reduced drug intracellular accumulation, binding site modification, and biofilm formation. 8 Drug altering or inactivating enzymes are well studied and therapies have been developed that can inhibit these enzymes. A well-studied example is the β-lactamase type enzymes which are capable of inactivating β-lactam antibiotics such as penicillins, cephalosporins, monobactams, and carbapenems. 14 These enzymes have been the targets of β-lactamase inhibitor drugs such as clavulanic acid, tazobactam, and sulbactam, all of which are used in combination with β-lactam antibiotics to successfully treat bacteria that produce β-lactamases. 15,16 While these approaches have been successful, the emergence of extended spectrum β-lactamases (ESBLs) such as New Delhi β-lactamase (NDM-1) and oxacillin hydrolyzing enzymes (OXA) that are resistant to β-lactamase inhibitors and confer resistance to most β-lactams is concerning. 15,17,18 When bacteria are able to reduce the intracellular concentration of antibiotics, this is frequently due to reduced production of channels that transport antibiotics into the cell, or the up-regulation of efflux pumps able to eject antibiotics. 8 Efflux pumps expel antibiotics from the bacterial cytoplasm before they can reach their site of action. 19,20 These pumps can expel a broad range of substrates by coupling the efflux with a proton or other ion. 21 Certain efflux pumps, such as the resistance nodulation cell division (RND) family pump AcrAB, have been suggested as biomarkers for the MDR phenotype since they confer resistance to ≥ 5 antibiotics. 22 Efflux pumps also contribute to the resistance present in the ESKAPE pathogens. 23,24 The ability to inhibit efflux pumps with a drug adjuvant to antibacterial therapy could help overcome resistance and treatment failures. There has been progress in identifying efflux pump inhibitors (EPIs) of Grampositive pathogens such as S. aureus, but there is a sizable knowledge gap for Gramnegative efflux pumps. 24 Discovery of EPIs for Gram-negative bacteria is greatly needed to help supplement the already bare pipeline of new therapies for these problematic pathogens.

Persistence and quiescence of bacterial pathogens
The phenomenon of bacterial persistence is a rapidly expanding research area in the field of antibiotic development. Similar to resistance, bacterial persistence makes the treatment of infections much more difficult. Persistence is characterized by dormant bacterial cells (persister cells) that are highly tolerant to antibiotics. 25 When these cells are isolated, re-cultured, and tested for antibiotic susceptibility, they show full susceptibility, indicating these are not resistant mutants. In contrast, antibiotic resistant bacteria will show a decrease in antibiotic susceptibility (require higher concentrations of antibiotics to inhibit the growth of the bacteria). Persister cells are implicated as the causative phenotype in chronic infections such as recurrent urinary tract infections (UTIs), infections in cystic fibrosis, and tuberculosis. [26][27][28] In addition to chronic infections, persister cells are commonly associated with biofilms, which can allow them to evade the immune system as well as exhibit antibiotic tolerance, causing antibiotic treatment failures. 25,26,28 Uropathogenic E. coli (UPEC) is an ideal organism to study persistence. 85% of all UTIs are caused by UPEC and reinfection is common. 29 Studies have shown that UPEC are able to invade bladder and kidney epithelial cells during a UTI. 30,31 Following invasion, the UPEC can either establish an intracellular bacterial community that evades host immune defenses (in acute infection) or a quiescent intracellular reservoir (for recurrent infections). 31 The quiescent intracellular reservoir can remain viable for months and cause reinfection following exfoliation of UPEC containing epithelial cells. Drugs capable of inhibiting persistence and thereby inhibiting reinfection would be valuable tools in the treatment of UTIs caused by UPEC. Additionally, compounds that reduce persistence in UPEC could help identify targets and mechanisms necessary for persister cell formation in other important bacterial pathogens. This could lead to new drugs and improved treatment outcomes in challenging bacterial infections.
Natural products as drug leads and antibiotics Natural products represent both a classical source of drug molecules and a promising future resource of novel medicines. 32,33 From the years of 1981-2014, 1211 small molecule drugs were approved for use, and of those 791 (64%) were either natural products or inspired by natural products. 32 More importantly, 59% of all drugs approved to treat bacterial infections were naturally derived or inspired. 32 The fraction increases to 74% when you exclude vaccines and look at only small molecule drugs (e.g., antibiotics), indicating that natural products are by far the most significant source of antibiotics available. Of the 12 major classes of antibiotics, nine of these, including β-lactams, macrolides, tetracyclines, aminoglycosides, glycopeptides, and lipopeptides, contain a natural product derived structure. 34 Microorganisms have also been prolific producers of antibacterial compounds. It is estimated that 16,500 antibiotic secondary metabolites have been discovered from microorganisms and roughly 12,000 of those have antibacterial effects. 35 While this seems like a vast amount of compounds, genome sequencing has revealed that many of the biosynthetic gene clusters are dormant under laboratory conditions, potentially indicating that we have only scratched the surface of antibiotic compounds from microorganisms. 35,36 An additional resource is the many 'unculturable' species. It is estimated that upwards of 90% of all terrestrial microorganisms, and >99% of all aquatic microorganisms do not grow under normal laboratory conditions. 35,37 The amount of species diversity yet unstudied is vast, and we are now developing techniques to culture these 'unculturable' microorganisms. 37,38 It is clear that the most prolific producers of antibiotics, microorganisms, are still largely unstudied and could yield thousands more antibiotics as yet undiscovered. With the current shortcomings in the development of new anti-infective agents, additional investigations of natural products, especially those from microorganisms should be pursued.

Marine microorganisms as sources of bioactive natural products
There is significant data to suggest the promise of marine organisms as potential sources of bioactive compounds, including antibiotics. In a marine environment, an average of 10 6 bacterial cells and 10 3 fungal cells exist per milliliter of seawater which creates a constant and mobile threat of infection to macroorganisms, as well as competition between microorganisms for survival. 34 Investigations of marine organisms have uncovered compounds in diverse classes of secondary metabolites (ribosomal peptides, non-ribosomal peptides, polyketides, alkaloids, and terpenes) that possess potent bioactivities. 34 Marine sediments also harbor vast microbial diversity (10 8 cells in a gram of wet sediment) that has been largely unstudied. 34,39 Compared to terrestrial studies, the search for antibiotics from the marine environment is still in its infancy. 34 The marine environment comprises a vital resource of biodiversity that should be further explored for new antibiotics.
Natural products discovered from the marine environment span a large variety of organisms, but microorganisms show significant potential. Because terrestrial microorganisms have been such a significant source of antibacterial compounds, it stands to reason that marine microorganisms would also be prolific producers of bioactive molecules. 491 (42%) of the new compounds isolated from the marine environment in 2013 were isolated from microorganisms indicating they are already a major source of marine chemistry. 40 These compounds exhibit diverse bioactivities including antibacterial, anticancer/cytotoxic, anti-quorum sensing, antiviral, anti-inflammatory, immunosuppressant, antiprotozoal, and antifungal. [40][41][42] Some antibacterials discovered from marine microorganisms include non-ribosomal peptides (bogorol A, emericellamide A, and thiocoraline), polyketides (abyssomycin C and pestalone), and alkaloids (marinopyrrrole A). 34 Additionally, while the macroorganisms such as sponges, algae, and tunicates represent a significant source of chemistry, it is proposed that microorganisms acting in symbiotic relationships are responsible for at least a portion of the identified novel chemistry. 43,44 There are currently 20 drugs (molecules in clinical trials or approved by the FDA) either isolated from a marine source or inspired by marine chemistry. 45 It has been proposed that 17 are produced by microorganisms associated with the macroorganisms from which those drugs were originally isolated. 45 3 leaving few antibiotics in the therapeutic repertoire to treat these infections. [6][7][8] In the search for new antimicrobials, random screening of libraries of synthetic or natural products are estimated to have a primary hit-rate of up to 1,000-fold lower against Gram-negative than for Gram-positive bacteria. 9 Of the antibiotics approved by the FDA from 1998 to 2005, including those in clinical trials, and various stages of preclinical development, most lack appropriate activity against any Gramnegative bacteria. 6,7 Making incremental improvements to the chemical scaffolds of existing antibiotics is at best a short-term strategy for the impelling need for both new drugs and novel approaches to combat multidrug resistant pathogens. 10 Consequently, developing compounds targeting the resistance mechanisms themselves is warranted (e.g. the clinically proven beta-lactam/beta-lactamase inhibitor cocktail, amoxicillin/clavulanic acid), 4 thereby i) obviating the emergence of resistance, and ii) regaining antibiotic potency.
The rapid spread of resistance is, in part, a result of constitutive over-expression of transmembrane efflux pumps that expel antibiotics before they can reach their intracellular target. 11 Members of the Resistance Nodulation Cell Division (RND) superfamily of MDR pumps have been implicated in the high intrinsic resistance of Gram-negative species, 12, 13 whose tripartite RND pumps recognize and expel a broad range of substrates (including antibiotics, charged and neutral molecules, organic solvents, lipids, bile salts, and quorum signal molecules), 9 via a coupled exchange with protons or ions. 14 Permanent overexpression of RND pumps leads to multidrug resistance in bacteria, 15 while their deletion restores antibiotic susceptibility 16 further confirming this transporter is an important therapeutic target. Polyspecificity of RND pumps is central to the emergence and spreading of efflux-mediated resistance, as these pumps subsequently allow for acquisition of additional resistance mechanisms, 9,17 and have a significant role in bacterial pathogenicity/virulence, invasion, adherence and host colonization. 14 19 making these efflux pumps "key" targets for the development of an efflux pump inhibitor (EPI) as an adjuvant to existing antibiotics. 20 Our search for small molecule EPIs from the microbial realm has been aided by the fact that natural products have often been selected precisely for their ability to penetrate both outer and inner membranes of Gram-negative bacteria. 21 A countermeasure by antibioticproducing microbes is to co-evolve inhibitors of their competitor's resistance mechanisms to enhance the efficacy of their own antibiotics, 21 exemplified by the Streptomyces spp. producing both beta-lactam antibiotics and the beta-lactamase inhibitor clavulanic acid. 22 There is substantial evidence that marine bacteria produce cocktails of both antibiotics to control surface colonization, 23,24 and nontoxic secondary metabolites capable of quenching quorum sensing-controlled activities in other species. 25 However, to our knowledge no systematic study has screened for EPIs from marine microbial exudates (i.e., compounds excreted into the extracellular medium) against RND pumps.
Regardless, our approach is validated by previous screening of terrestrially-derived microbial fermentations which resulted in two new natural product EPIs targeting MexAB-OprM from Streptomyces (EA-371α and EA-371δ) potentiating levofloxacin minimum inhibitory concentrations (MICs) 4-fold and 8-fold, respectively. 26 The microbial EPI, MP-601,205 is currently used to treat P. aeruginosa respiratory infections in cystic fibrosis patients. 14 Thus far, a diverse set of natural product chemical scaffolds (including polyphenols, flavones, flavonols, flavonolignans, flavonoids, diterpenes, triterpenoids, oligosaccharide-glycosides, and pyridines) have been validated as EPIs in Gram-positive bacteria such as Staphylococcus aureus. 27 A remaining challenge is the discovery of EPIs targeted toward Gram-negative efflux pumps. 27 We hypothesized that microorganisms obtained from the marine environment produce EPIs as regulators of diverse ecological interactions, and as such present a unique bioprospecting opportunity. In the present paper our objectives were: (i) to screen our inhouse chemical library to identify marine microbial isolates capable of reducing  Information). To our knowledge, this paper represents the first description of 3,4dibromopyrrole-2,5-dione from a microbial source, although the original report of its isolation from nature was from the marine sponge Axinella brevistyla collected in western Japan, where it was reported to exert modest antifungal activity and cytotoxicity against murine lymphocytic leukemia cells. 28 However, since over 50% of a sponge's biomass can be attributed to microorganisms, it is conceivable that the origin of 3,4dibromopyrrole-2,5-dione may be from sponge-associated microorganisms. and >1 is antagonistic. The results, shown in Table 1 Figure 1B, show there is no effect of 3,4dibromopyrrole-2,5-dione on H33342 fluorescence in AG100A, which is consistent with the hypothesis that RND pumps are the target of 3,4-dibromopyrrole-2,5-dione and indicates MDR reversal is limited to efflux pump expressing E. coli strains. In addition, if the mechanism of action of 3,4-dibromopyrrole-2,5-dione was via membrane permeabilization, we would have seen a dose-dependent increase in H33342 accumulation regardless of test bacterial strain, which we did not observe.
Demonstration of efflux activity in AG100 requires that accumulation of H33342 has taken place. Following H33342 "loading", monitoring of efflux was initiated in the  3,4-Dibromopyrrole-2,5-dione also is known to react with disulfide linkages in the presence of a strong reducing agent such as (tris(2-carboxyethyl)phosphine). 34 It is conceivable, but unlikely, that 3,4-dibromopyrrole-2,5-dione is displacing and coordinating certain accessible, reduced disulfide linkages in the RND pumps and permanently modifying their conformation by this crosslinking type displacement; however, the lack of antibiotic activity and range of pharmacokinetics argue against this type of action. Although the subject of another study, experiments employing halogenated and non-halogenated maleimides will be informative in terms of mode of action.

Chemophylogenetic Analysis of Extracts from Pseudoalteromonas Isolates.
Analysis of our untargeted -HRESI metabolomics comparison indicated the presence of many halogenated features, the distributions of which differed greatly across Pseudoalteromonas clades ( Figure 2B and 2C, Figure S7, In order to discern which chemical features were unique to Clade IV (and subsequently strain A757), we examined chemical feature loadings on the first principal component. It is important to note that great care must be taken to avoid the assumption that any of these features alone would be significantly differentially produced by Pseudoalteromonas strains, 35  and were therefore likely candidates to distinguish Clade IV from the remaining clades ( Figure 2B). Further analysis of these 221 features indicated that i) several shared retention times, and ii) had masses ~2 amu apart, suggesting that many chemical features were isotopes of halogenated metabolites. We determined that of these 221 features, 129 were isotopes of 46 individual brominated metabolites based upon both isotopic distribution and shared retention time (Table S2, Figures S8-30, Supporting Information).
Of the 46 halogenated metabolites found, all were exclusively produced by members of Clade IV ( Figure 2C), demonstrating that there is a distinct Pseudoalteromonas chemotype characterized by halogenation of the exo-metabolome. Moreover, this "halogenome" appears to be dominated by brominated metabolites, with only some metabolites (<10) that are additionally chlorinated (Table S2, Supporting Information). In addition, we observed several other brominated compounds (~30-40) based upon isotopic signatures produced by A757 that were not accounted for in the initial 221 chemical features (screened based on their loadings), suggesting that a large percentage of A757 exo-metabolome may be subject to halogenation (Tables S2 and S3, Supporting Information). Relative concentrations of halogenated compounds from Clade IV appeared different even within members of the same species ( Figure 2C), indicative of intraspecies chemical diversity and further sub-clustering among isolates of the same species (A757, A754, A746 and B149) a phenomenon previously described for P. luteoviolacea strains. 36 Taken together, these results indicate that the production of halogenated compounds could be a biomarker for marine isolates with enhanced biosynthetic potential. Of the four strains in our culture collection designated as belonging to Clade IV, A757 and A754 produced all 46 metabolites, whereas strains A746 and B149 produced only select halogenated metabolites, including 2,3,4,5-tetrabromopyrrole, a known weak antibiotic from Pseudoalteromonas spp., 37 ( Figure 2C). 2,3,4,5-Tetrabromopyrrole appears to be the most abundant brominated compound we observed (retention time 20.8 min, major ion m/z 381.6722, Figure 2C, Table S2, Supporting Information) and present in all members of Clade IV. Because of the dominance of 2,3,4,5-tetrabromopyrrole in our samples and its shared carbon skeleton with 3,4dibromopyrrole-2,5-dione, we tested an authentic standard of 2,3,4,5-tetrabromopyrrole in EPI functionality assays and determined 2,3,4,5-tetrabromopyrrole is not responsible for the EPI activity of strain A757 (data not shown). We also determined the presence of 2,3,4-tribromopyrrole (retention time 20.1 min, major ion m/z 301.7639, Figure 2C, Table   S2, Supporting Information) present in A757, A754 and A746, but not in B149). 2,3,4-Tribromopyrrole is reported to be produced by P. luteoviolacea, 38 found within Clade IV, and is a known feeding deterrent in marine systems. 39 The limit of detection of 3,4-dibromopyrrole-2,5-dione with an authentic standard was established to be 11.1 ng/mL. Our yield from 16.5L of A757 culture was on the order of 1 mg/L of 3,4-dibromopyrrole-2,5-dione. 3,4-Dibromopyrrole-2,5-dione should have been detected in our crude that was initially screened for antibiotic activity, however, we did not see this ion until culture scale-up and further purification, suggesting ionization masking effects from the presence of a complex mixture including 2,3,4,5tetrabromopyrrole, which was observed to co-elute with 3,4-dibromopyrrole-2,5-dione under these chromatographic conditions. In addition to numerous halogenated metabolites, we also were able to observe other unique metabolites produced by members  Figure S31 and S33, Table S5, Supporting Information). Although we did not seek to fully annotate chemical features detected in +HRESI, these data also indicate (along with -HRESI results) that members of Clade IV do possess a characteristic exometabolome.
Screening features detected with our untargeted metabolomics approach (both -HRES and +HRES) against an in-house database containing previously reported Pseudoalteromonas metabolites indicated isolates in our collection potentially produce molecules previously described for Pseudoalteromonas spp. Molecular features annotated in +/-HRESI mode data with isotopic distribution patterns and predicted formulae matching these dereplicated compounds are shown in Figure 3 and Tables S8 and S9, Supporting Information. A potential annotation for the antibiotic, 2-n-heptyl-4quinolinol, 40 known to also influence bacterium-phytoplankton interactions, 23 was exclusively found within members of Clade IV in both +HRESI and -HRESI data sets  49 appeared to be expressed in the majority of our isolates, with a few exceptions ( Figure 3B). Future work will be able to discern the true identities of these chemical features, as several isomers are possible for each of the detected ions in our analysis. Regardless, chemophylogenetic analysis indicated that some compounds were i) characteristic of a particular isolate, ii) clade specific (e.g., halogenated species, 2-n-heptyl-4-quinolinol, and 2-n-pentyl-quinolinol) or iii) ubiquitous in the majority of isolates.
Increasingly, non-photosynthetic Gram-negative bacteria (NPGNB) are being recognized as the true source of pharmaceutically-relevant molecules from marine macroorganisms; 29 however, the difficulty in culturing marine-derived strains to sufficient quantities has likely hampered intense bio-prospecting efforts. 29 Indeed, the majority (86%) of our marine isolates found to have MDR reversal activity in our initial screening efforts fall within the NPGNB group, and have been isolated from both abiotic and biotic surfaces (Table S1, Supporting Information). The cosmopolitan marine genus Pseudoalteromonas (class Gammaproteobacteria), which constitute 0.5-6% of bacterial species globally, 50 has been found in seawater, marine sediments and epiphytically associated with marine eukaryotes, and has been a prolific source of brominated compounds, 37,51,52 including pentabromopseudilin, 36 the first marine microbial natural product to be described. 53,54 Compounds isolated from this genus function in multiple ecological roles including their involvement in chemical protection, settlement, germination and metamorphosis of marine invertebrate and algal species, as well as more commercial uses as antifoulants, antibacterial, antifungal and cytotoxic agents. 37,52,55 Recent genome mining work has uncovered the biosynthetic pathways responsible for brominated pyrrole/phenol biosynthesis (bmp) 38 indole derivatives, siderophores, polyketides, homoserine lactones, peptides (both ribosomal and non-ribosomal origin) and hybrid molecules, 56 which likely represent just the tip of the iceberg, as the number of pathways encoded in Pseudoalteromonas genomes eclipses the fraction of molecules identified thus far. 55 For marine bacteria, including many antibiotic producing Pseudoaltermonas spp., 37,57 a viable strategy by these organisms may be to secrete an EPI to enhance their own antibiotic effectiveness. Previous research suggests the dominance and enriched diversity 51 of Pseudoalteromonas spp. in biofilms could be attributed to their ability to rapidly form microcolonies and produce extracellular antibacterial compounds. 36 (2)

Culture Production Scale-up, Bioassay Guided Fractionation, and Chemical
Analysis. Marine isolate A757 (GenBank KM596702), determined to be most closely related to P. piscicida by 16S rRNA sequence comparison, was targeted for scaled-up regrowth to obtain additional material for chemical identification of the putative EPI compound. A starter culture of A757 was used to inoculate eleven, 1.5 L Fernbach flasks for a total of 16.5 L of culture medium that was processed as described in "Bacteria Culture and Chemical Library Production", yielding a total of 1729.7 mg of crude organic extract. At each subsequent fractionation step an aliquot of material was resuspended in DMSO and assayed in the INT assay against MG1655 ΔBC/pXYM to confirm retention of MDR reversal activity. A total of 1.7 g of crude extract was applied to a silica gel column and eluted with a step-gradient of 100% isooctane 1-508-289-3640. Email: tmincer@whoi.edu

Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Notes
The authors declare no competing financial interest.      Keywords: multidrug resistant bacteria, Acinetobacter baumannii, benzisoxazole, antibacterial

Introduction
Acinetobacter baumannii is a non-fermenting, Gram-negative bacterium commonly implicated in nosocomial infections such as wound infections, sepsis, and ventilator associated pneumonia. [1] The World Health Organization recognizes antibiotic resistance as one of the three greatest threats to human health, and recently classified carbapenem-resistant A. baumannii as one of the most urgent threats, calling for a renewed investment in antibiotic development. [2] A. baumannii tenacity in hospitals is enhanced both by its ability to develop antibiotic resistance and capacity to survive on surfaces, including skin, for several days. [3] A. baumannii can also spread via aerosolization, promoting this pathogen's ability to easily colonize new environments. [4][5][6] A. baumannii is a member of the 'ESKAPE' pathogens [7] and has garnered significant attention due to its naturally occurring resistance to the most commonly used antibiotics. [8] Alarmingly, intensive care unit patients contracting A. baumannii infections have a high mortality rate (>50%), complicated by the fact that new pan-drug resistant (PR) strains of A. baumannii with resistance to all treatment options have now emerged. [1,[8][9][10] Carbapenems, typically in combination with other antimicrobial agents (e.g. colistin) are a commonly utilized treatment option for MDR A. baumannii but it is estimated that more than half of MDR strains are now carbapenem-resistant. [1,11,12] Currently, the polypeptide antibiotic colistin is the primary method of treatment for patients with carbapenem-resistant A. baumannii; however, this therapy has significant toxicological and dosing concerns. [1,12] With the lack of new antibiotics in the drug discovery pipeline to treat Gram-negative infections, coupled with accelerated evolution of antibiotic resistance, it is imperative that new antibacterial drugs for treating A.
baumannii infections be developed. [13] Most antibacterial drugs are products of microbial biosynthesis and derivatives thereof. [14] Herein we report the identification of a 1,2-benzisoxazole antibiotic produced by a marine bacterium identified as a Bradyrhizobium denitrificans. A series of synthetically prepared analogs define key structural features for the antibacterial effects against A. baumannii. This molecule represents a new lead structure for antibacterial drug development against clinical isolates of A. baumannii that possess MDR phenotypes.

Results
Exudate extract from Bradyrhizobium denitrificans (Isolate B158) was initially screened in the p-iodonitrotetrazolium chloride (INT) assay to assess MDR reversal potential. [15] Phylogenetic analysis by 16S rRNA gene sequence comparison indicated that isolate B158 was most closely related to Bradyrhizobium denitrificans. Bioassayguided fractionation of extract generated from B158 microbial culture resulted in the isolation of 1 and two other known compounds (lumichrome and (1H-indol-3-yl) oxoacetamide; see 11 and 12 in Supplementary Materials) [16,17] with varying degrees of bioactivity.  All minimum bactericidal concentrations (MBCs) for 1 were >100 μg/mL.
To further understand the structural requirements necessary for the antibacterial effects of 1 against A. baumannii, as well as to potentially improve upon the potency of 1, we synthesized and purchased analogs to perform a structure activity relationship (SAR) study (Figure 1). None of the analogs (2-10) demonstrated increased potency against the panel of A. baumannii strains ( Table 2). However, structural requirements for optimal potency were revealed. Replacement of the hydroxyl substituent at C6 (as in 2-6) displayed marked decreases in potency. Based on these data, a hydrogen bond donor substituent at C6 appears to be required for antibacterial effects, with hydroxyl (1) as the preferred substituent over amino (5). Though the C6 methoxyl group (6) did not completely abolish activity, the dramatic decrease in potency concludes that alkyl modifications to the hydroxyl group are unlikely to improve potency in this pharmacophore. Because of the preference of the hydroxyl substituent at the C6 position, it was also investigated whether an additional hydroxyl group might improve the activity.
Surprisingly, a second hydroxyl group at C4 (7) abolished the antibacterial activity.  In addition to substituents on the benzene ring, the necessity of the isoxazole ring to the pharmacophore was investigated. Methylation of the nitrogen atom (8 and 9) led to inactive products; it is noted that this modification leads to loss of conjugation in the hydroxyl-isoxazole ring. Similarly, antibacterial activity was also completely absent for the oxazolone analog (10). 1,2-benzisoxazoles are bicyclic, heteroaromatic structures found in many approved drugs. [21,22] In general, 1,2-benzisoxazoles are able to cross the blood brain barrier (BBB) and have been classified as privileged structures for drug design for CNS disorders and other therapeutic areas. [21,22] The heterocyclic ring structure of 1 shares similarities to two clinically used antibiotics: cycloserine and linezolid (and other oxazolidinone antibiotics). Interestingly, cycloserine was inactive against our panel of A.

Current therapies for MDR
baumannii strains (data not shown). Oxazolidinone antibiotics are only effective against Gram-positive pathogens, and match the ring structure of the inactive 10, further supporting the highly specific nature of the 1,2-benzisoxazole pharmacophore.
In the search for antibacterial compounds for MDR pathogens, it has been shown that 'old' antibiotics that have been excluded from the clinic for decades can be useful as alternative treatment options. [23,24] In addition to older clinical antibiotics, there are many antibacterial compounds discovered decades ago that have not been developed, including compound 1 that may be useful against modern MDR pathogens. Compound 1 was previously reported to possess potent growth inhibition of Gram-negative bacteria including E. coli, Proteus spp., Salmonella spp., Klebsiella pneumoniae, Enterobacter spp., and Serratia marcescens, as well as low toxicity with an LD 50 of over 1500 mg/kg in mice. [18,19] No inhibition was observed with the Gram-positive bacteria Staphylococcus aureus, Bacillus spp., Micrococcus sp., Corynebacterium sp., or the Gram-negative Pseudomonas spp. [18,19] Narrow spectrum of antibacterial drugs are becoming more desirable due to advances in diagnostic tests and changes to the regulatory environment. [25] Resistance is expected to develop more slowly to narrow spectrum antibiotics and collateral damage to the gut microbiome, such as with antibiotic-associated colitis, is less likely. [26] We suggest that further studies advancing the development of 1,2-benzisoxazoles, including 1, as anti-A. baumannii antibiotics are warranted. Investigations into its cellular target, in vivo efficacy, and further structure activity relationships are vital next steps to improving the outlook for treatment of infections with MDR A. baumannii. Louis, MO). MDRO was defined as organism with non-susceptibility to one agent in three antimicrobial classes, according to previously described definitions. [20] These strains are available in the investigators' laboratory and stored frozen at -80 ºC in 20% glycerol.  Once isolate B158 was determined to contain active compound(s), a total of 118.5 L of B158 culture was grown in 1.5 L Fernbach flasks to obtain enough crude material for structural elucidation. Briefly, 2 mL of "starter" culture in TSW was used to inoculate

Introduction
Uncomplicated urinary tract infections (UTIs) affect about 25% of women in their lifetime and at least 80% of those infections are caused by uropathogenic E. coli (UPEC) (1). Recurrent UTIs affect between 10% and 40% of women (2) and in up to 77% of those cases the recurrent infections are caused by the same UPEC strain that caused the initial infection (3,4). UPEC infections generate annual costs in excess of two billion dollars in the United States alone, placing a significant burden on the health care system (5). Although the causes of recurrent UTI are complex (6), it appears that UPEC can bind to, enter, and replicate within superficial facet cells in the human and mouse bladder epithelium, resulting in biofilm-like intracellular communities (IBCs) (6,7). IBCs escape from infected superficial facet cells within hours of development (6). The superficial facet cells then exfoliate, exposing underlying transitional epithelial cells, which can be infected with IBC-derived UPEC progeny (6,8). Upon infection of urothelial transitional cells, UPEC appear to enter a non-growing quiescent intracellular state (6,8). These

quiescent UPEC cells have been called QIRs (Quiescent Intracellular Reservoirs) (6) and
it is thought that QIRs are a major cause of recurrent UTIs (6,8). QIRs also help to explain why antibiotics have failed to eradicate UPEC reservoirs in the bladders of mice, since quiescent UPEC may not be readily affected by antibiotics (6,8).
The quiescence of QIRs and their insensitivity to antibiotics is reminiscent of the persister state (8,9,10,11). Persister cells are dormant cells formed in normal microbial populations as small subpopulations that are highly tolerant to antibiotics, but upon regrowth in the absence of antibiotics regain full sensitivity (11). Persisters appear to play a major role in the ability of chronic infections to withstand antibiotic treatment (11). In the present study, we report that when ≤ 10 6

Media. LB broth (Lennox) (Difco Laboratories) and LB agar (Lennox) (Difco
Laboratories) were used for routine cultivation. SOC medium was prepared as described by Datsenko and Wanner (14). Liquid M9 minimal medium (15)    Colonies that were ampicillin sensitive, signifying loss of the pUT suicide plasmid, were toothpicked to sterile 16 mm diameter culture tubes containing 250 µl of 0.2% glucose M9 minimal medium. Culture tubes were incubated overnight at 37°C with shaking, 5 ml of M9 minimal medium lacking a carbon source was then added to each tube, and 3 µl from each tube was spotted on a 0.2% glucose M9 minimal medium agar plate. Spots were allowed to dry (with lids slightly ajar) and plates were incubated overnight at 37°C.
Each spotted mini-Tn5 Km mutant that grew was retested by seeding 10 5 CFU of an overnight 10 ml liquid 0.4% glucose M9 minimal medium culture on a 0.2% glucose M9 minimal medium agar plate and incubating at 37°C for 24 hours. The gene inactivated in each of the mini-Tn5 Km mutants that grew as a lawn was determined by arbitrary PCR (18), as described below. In addition, to be sure that the mini-Tn5 Km insertion was the cause of the ability of the individual mutants to grow on glucose plates, the insertion in each mutant was transferred into a fresh E. coli CFT073 Str R background by the method of Wanner and Datsenko (14). Each mutant thus obtained was confirmed for the ability to grow as a lawn on glucose plates and for the position of the insertion within the E. coli CFT073 chromosome by both PCR and sequencing (see Table 2 for primers). Five confirmed mutants were isolated from approximately 2000 mini-Tn5 Km mutants tested.
Arbitrary PCR. Arbitrary PCR was performed as described previously (18). Genomic Amino acids in 50-fold concentrated culture supernatants were identified and quantified using a slightly modified version of a method described by Yuan et al. (19). Photography. Images of agar plates were made using a BIO-RAD Molecular Imager® Gel Doc™ XR+ System with Image Lab™ Software. Table 4 and in Figures 4,5,7,8, and 9 were compared by a two-tailed Student t test. P values of ≤ 0.05 were interpreted as indicating a significant difference.

E. coli CFT073, a UPEC strain, and E. coli Nissle 1917, a closely related probiotic
strain, grow in liquid glucose M9 minimal medium, but fail to grow on glucose M9 minimal medium agar plates. We were attempting to determine which colicins and microcins are active on the sequenced E. coli CFT073, a phylogenetic group B2 multilocus sequence type 73 (ST73) UPEC strain (20,21), and the closely related ST73 probiotic strain, E. coli Nissle 1917 (22). As expected, we found that after overnight incubation at 37°C, both strains formed lawns of growth on 0.2% glucose M9 minimal medium agar plates (hereafter called glucose plates) when either 10 7 or 10 8 CFU were plated from an overnight 0.4 % glucose M9 minimal medium liquid culture.
Unexpectedly, however, when 10 6 CFU or fewer were plated on glucose plates they failed to form lawns or colonies after 24 h and 48 h of incubation at 37°C. This result was unusual in that a number of human commensal E. coli strains, including MG1655, HS, F-18, EFC1, and EFC2 (Table 1), and the O157:H7 strain E. coli EDL933 (Table 1) tested at 10 5 CFU grew as lawns on the glucose plates after 24 hours at 37°C and viable counts could be determined for each strain on glucose plates (not shown). That all the E.
coli strains tested for growth on glucose plates are spontaneous streptomycin resistant mutants ( quiescence is not dependent on using 0.2% glucose M9 minimal medium agar plates, identical results were obtained using 0.4% glucose M9 minimal medium agar plates (not shown).
Testing additional E. coli strains for quiescence on glucose plates. E. coli can be separated into four major phylogenetic groups (A, B1, B2 and D) and two additional phylogenetic groups that have recently been defined: phylogenetic group AxB1, containing strains that derive most of their ancestry from A and B1, and phylogenetic group ABD, containing a heterogeneous set of strains with multiple sources of ancestry (23). Thirty E. coli strains representing various multilocus sequence types (ST) of the 6 phylogenetic groups, i.e. two ST10 and two ST453 strains from phylogenetic group A; two ST58, two ST410, and two ST101 strains from phylogenetic group B1; two ST73, two ST95, and two ST131 strains from phylogenetic group B2; two ST69, two ST354, and two ST648 strains from phylogenetic group D; two ST90 and two ST642 strains from phylogenetic group AxB1; two ST62 and two ST117 strains from phylogenetic group ABD were grown in liquid glucose M9 minimal medium and the 30 strains were tested for the ability to grow on glucose plates seeded with 10 5 CFU and to respond to the MG1655 stimulus. Of the 30 strains, 2 failed to grow on glucose plates and those strains responded to the MG1655 stimulus. The 2 strains that failed to grow on glucose plates were ST73 strains. ST73 is a very common UPEC lineage, accounting for 11% and 16.6% of UPEC isolated from patients in 2 recent studies (24,25). Importantly, E. coli CFT073 and E. coli Nissle 1917 are also ST73 strains.
Testing additional ST73 strains for quiescence on glucose plates. Since it appeared that quiescence on glucose agar plates may be characteristic of the ST73 lineage, 40 additional ST73 strains were tested for the ability to grow on glucose plates seeded with 10 5 CFU and to respond to the MG1655 stimulus . Two of the strains failed to grow overnight in liquid glucose M9 minimal medium, but of the 38 strains that grew, 30 (78.9%) failed to grow on glucose plates, but responded to the MG1655 stimulus.

Testing 40 UPEC strains isolated from community acquired UTIs in Denmark for
quiescence on glucose plates. Forty randomly selected UPEC strains isolated from community acquired UTIs in Denmark were tested for the ability to grow on glucose plates seeded with 10 5 CFU and to respond to the MG1655 stimulus. Of the 40 UPEC strains tested, all grew overnight in liquid glucose M9 minimal medium, but 9 failed to grow on glucose plates (22.5%) unless stimulated to do so by the MG1655 stimulus.
Three of the 9 strains that failed to grow on glucose plates were ST73 strains (5 of the 40 UPEC strains tested [12.5%] were ST73) and 3 of the 9 strains that failed to grow on glucose plates were ST141 strains (3 of the 40 UPEC strains tested were ST141 strains . It therefore appears that the inability to grow on glucose plates, yet respond to the MG1655 stimulus, is not limited to the ST73 group. The inability of E. coli CFT073 to grow on minimal agar plates is not limited to glucose as sole carbon source. E. coli CFT073 was tested for the ability of 10 5 CFU to grow on M9 minimal agar plates containing 0.2% acetate, arabinose, fructose, fucose, galactose, gluconate, glycerol, N-acetylglucosamine, maltose, mannose, ribose, and xylose as sole carbon sources. E. coli CFT073 grew overnight in liquid M9 minimal medium containing each carbon source, but 10 5 CFU only grew as a lawn on agar plates containing glycerol, ribose, and xylose as sole carbon sources (Fig. 2) (Fig. 3A). Analysis of one such E. coli MG1655 cell-free supernatant revealed the presence of a number of unknown small molecules and 14 amino acids (Table 3). Importantly, 5 µl of an amino acid cocktail identical in composition to the amino acids in the 50-fold concentrated E. coli MG1655 supernatant displayed similar stimulus activity to 5 µl of the 50-fold concentrated supernatant (Fig. 3B). A cell-free supernatant derived from a 50-fold concentrated E. coli CFT073 culture was nearly identical to the E. coli MG1655 supernatant in amino acid composition, but in addition contained aspartic acid ( Table 3). As expected, 5 µl of the E. coli CFT073 cell-free supernatant also displayed stimulus activity on glucose plates seeded with 10 5 CFU of E. coli CFT073. Perhaps even more importantly, 5 µl of sterile filtered human urine collected from one of us and 5 µl of a cocktail mimicking the amino acid composition of human urine ( [26] and Table 3) both displayed stimulus activity on glucose plates seeded with 10 5 CFU of E. coli CFT073 (Fig. 3C and 3D).
Lysine, methionine, and tyrosine involvement in preventing quiescence. 1.0 mM solutions of each of the 20 standard L-amino acids were prepared and 5 µl of each was tested on glucose plates seeded with 10 5 CFU of E. coli CFT073. None displayed stimulus activity. However, testing 5 µl of single amino acid deletions from the amino acid cocktail mimicking the concentrations of amino acids in urine (Table 3) and in the 50-fold concentrated E. coli MG1655 supernatant (Table 3)  of E. coli CFT073 growth by the mixture of lysine and tyrosine was minimal (Fig. 3G).
A mixture of 1.0 mM each of methionine and tyrosine failed to stimulate E. coli CFT073 growth (Fig. 3H). In summary, E. coli MG1655 is not required to provide the stimulus that prevents E. coli CFT073 quiescence, a mixture of lysine, methionine, and tyrosine, found in concentrated 50-fold concentrated E. coli MG1655 supernatants, is just as effective. are highly tolerant to antibiotics, but upon regrowth in the absence of antibiotics regain full sensitivity (11). Persister cells appear to play a role in the ability of bacteria causing chronic infections to withstand antibiotic treatment (11). Because E. coli CFT073 becomes quiescent on glucose plates and quiescence is reminiscent of persistence, we wondered whether E. coli CFT073 generates a high level of persister cells in liquid glucose M9 minimal medium. 0.4% Glucose M9 minimal medium grown overnight cultures of E. coli CFT073 were diluted 20-fold into fresh 0.2% glucose M9 minimal medium (A 600 =0.1, ~10 8 CFU/ml) containing or lacking ampicillin (100 µg/ml) and

E. coli CFT073 and E. coli
viable counts were followed for 24 hours at 37°C (Fig. 4). During the first 4 h of incubation, viable counts in the E. coli CFT073 cultures decreased 10-fold in both the presence and absence of ampicillin, i.e. from 10 8 CFU/ml to 10 7 CFU/ml. No further death occurred between 4 h and 6 h in the absence of ampicillin, and by 24 h E. coli CFT073 had grown to almost 10 9 CFU/ml (Fig. 4). In contrast, between 4 h and 6 h in the presence of ampicillin, viable counts decreased almost an additional 10-fold to about 10 6 CFU/ml in the E. coli CFT073 cultures (Fig. 4). However, there was little further E.
coli CFT073 death in the presence of ampicillin between 6 h and 24 h (Fig. 4) increased immediately in the absence of ampicillin and in the presence of ampicillin decreased continuously for 6 h to a level of about 10 3 CFU/ml (10 -3 %) and remained at that level at 24 h (Fig. 4).  a level of about 5 x 10 3 CFU/ml (Fig. 5). That the survivors at 24 were persister cells was shown by the fact that when regrown in LB broth without ampicillin, they regained sensitivity, as described above. In the absence of amino acids and in the presence of ampicillin, E. coli CFT073 persister cells were again generated at a level of about 10 6 CFU/ml (Fig. 5). Therefore, in the presence of amino acids, about 100-fold fewer E. coli CFT073 persister cells were generated than in their absence (Fig. 5).

E. coli CFT073 generates a low level of persister cells in liquid glucose
Isolation and characterization of E. coli CFT073 mini-Tn5 mutants that grow on glucose plates. Since E. coli CFT073 grows overnight in liquid glucose M9 minimal medium, but not on glucose plates, we thought it possible that expression of one or more genes on glucose plates, but not in liquid glucose cultures might be responsible. If so, knockout of the responsible gene(s) would result in growth on glucose plates. Therefore, E. coli CFT073 mini-Tn5 Km (kanamycin) mutants were generated by random insertional mutagenesis (see Materials and Methods) and any mutant that grew as a lawn on glucose plates was confirmed by transferring the insertion into a fresh E. coli CFT073 background and retesting it for growth on glucose plates (see Materials and Methods).
Five confirmed non-quiescent mini-Tn5 Km mutants were isolated from approximately 2000 mutants tested. E. coli CFT073 and the mini-Tn5 Km mutants were grown overnight in liquid glucose M9 minimal medium and viable counts were made on both glucose plates and LB agar plates. As expected, E. coli CFT073 assayed from the overnight cultures failed to grow on the glucose plates when ≤ 10 6 CFU were plated, but when assayed on LB agar viable counts showed that the cultures contained ~10 9 CFU/ml.
In contrast, when assayed on either glucose plates or LB agar plates the 5 mini-Tn5 Km mutant cultures each contained about ~10 9 CFU/ml. It therefore appears that the mini-Tn5 Km insertion in each of the 5 genes completely prevented quiescence on glucose plates.
The mini-Tn5 Km insertions that resulted in mutants able to grow on glucose plates after transfer into a fresh E. coli CFT073 background were in the following 5 genes (Fig.   6): (i) sdhA: encodes the succinate-binding flavoprotein subunit of succinate dehydrogenase (27). As a consequence, the E. coli CFT073 sdhA mutant fails to grow on succinate as sole carbon source, but grows normally on glucose. (ii) gnd: encodes 6phosphogluconate dehydrogenase which functions in the oxidative branch of the pentose phosphate pathway to synthesize ribulose-5-phosphate from 6-phosphogluconate (28,29). Ribulose-5-phosphate is an essential precursor in the synthesis of FAD, nucleotides, and LPS. (iii) zwf: encodes glucose-6-phosphate dehydrogenase, which functions in the oxidative branch of the pentose phosphate pathway and the Entner-Doudoroff pathway when E. coli is grown on glucose (28,29). (iv) pykF: encodes pyruvate kinase, which converts phosphoenolpyruvate to pyruvate in the Embden-Meyerhof-Parnas pathway (30). (v) gdhA: encodes glutamate dehydrogenase, which catalyzes the amination of αketoglutarate to glutamate (31).
It might be argued that the mini-Tn5 Km insertions in the identified genes are not the cause of non-quiescence, but that non-quiescence is caused by downstream polarity effects. Indeed, the mini-Tn5 Km transposon used in the present study has strong transcription termination sequences flanking both ends of the kanamycin resistance gene (17). However, the intergenic number of nucleotides and nucleotide sequences between Moreover, in E. coli MG1655 and therefore in E. coli CFT073, there is a strong presumptive promoter between gnd and ugd (33) and in E. coli MG1655 and therefore in E. coli CFT073 both lpp and edd have their own experimentally identified promoters (33). It is therefore highly likely that non-quiescence is caused by interrupting gnd, pykF, and zwf, and not by downstream polarity effects. Also, in E. coli CFT073, gdhA is immediately upstream of c2163, which is transcribed in the opposite direction to gdhA  (21,32,33). It is therefore highly unlikely that non-quiescence caused by the insertion in sdhA is due to a downstream polarity effect on the sucABCD operon in E. coli CFT073. on glucose plates and responds to lysine, methionine, and tyrosine like the rpoS mutant used here (Fig. 11), but generates about 2000-fold fewer persister cells in liquid glucose M9 minimal medium than the rpoS mutant used in the present study (Table 4).

Of the 5 mini-Tn5
Therefore, it appears that the mutant rpoS gene is necessary for the high level of persistence observed, but is not necessary for E. coli CFT073 quiescence on glucose plates.

Discussion
The data presented here show that the uropathogen E. coli CFT073 and the probiotic  (38).
While it is unclear as to why E. coli CFT073 gnd and zwf mutants are non-quiescent on glucose plates, the reason E. coli K-12 gnd and zwf mutants grow on glucose as sole carbon source is known (29,30). When E. coli K-12 is grown on glucose as sole carbon source, glucose-6-phosphate dehydrogenase, encoded by zwf, and 6-phosphogluconate dehydrogenase, encoded by gnd, are used for the synthesis of ribulose-5-phosphate via the oxidative branch of the pentose phosphate pathway ( Fig. 6 and [28,29]). Ribulose-5phosphate is an essential precursor in the synthesis of FAD, nucleotides, and LPS.
Furthermore, both enzymes generate NADPH for biosynthesis. It might therefore seem surprising that mutations in gnd and zwf allow growth on glucose as sole carbon source (Fig. 6). However, null mutations in gnd and zwf in E. coli K-12 do not significantly affect their growth on glucose because the non-oxidative branch of the pentose phosphate pathway runs backwards in these mutants, generating ribulose-5-phosphate from fructose-6-phosphate and glyceraldehyde-3-phosphate ( Fig. 6 and [28,29]) and the necessary NADPH for biosynthesis by increased flux through the TCA cycle (28,29). It is therefore possible that for unknown mechanistic reasons the oxidative branch of the pentose phosphate pathway operates minimally during E. coli CFT073 quiescence, generating little ribulose-5-phosphate, but that the gnd and zwf mutations cause the nonoxidative branch to run backwards and flux through the TCA cycle to increase, generating a sufficient level of ribulose-5-phosphate and NADPH to prevent quiescence.
This scenario is consistent with the observation that ribose and xylose, both of which are metabolized in the non-oxidative branch of the pentose phosphate pathway, prevent quiescence when used as sole carbon sources (Fig. 2).
The E. coli CFT073 pykF mutant also grows on glucose as sole carbon source. The pykF gene encodes pyruvate kinase which converts PEP to pyruvate (Fig. 6). There is a second pyruvate kinase, encoded by pykA, but during E. coli K-12 growth in glucose minimal medium, it contributes little to total pyruvate kinase activity (39). Moreover, in the complete absence of pyruvate kinase, pyruvate can still be generated in E. coli K-12 both through glucose transport via the phosphotransferase transport system (PTS) (40) and the Entner-Doudoroff pathway (31,38). Therefore, it is not surprising that the nonquiescent E. coli CFT073 pykF mutant can grow on glucose as sole carbon source, but why is it non-quiescent? It has been suggested that pykF mutants contain increased intracellular levels of PEP (30). If PEP, a common precursor in lysine, methionine, and tyrosine biosynthesis (Fig. 6), is increased in the E. coli CFT073 pykF mutant growing on glucose plates, sufficient intracellular levels of the 3 amino acids might be generated to stimulate growth and prevent quiescence.
The gdhA gene encodes glutamate dehydrogenase, which converts α-ketoglutarate to glutamate (Fig. 6). It is known that E. coli K-12 gdhA mutants grow on glucose as sole carbon source in the absence of glutamate dehydrogenase because glutamate can also be synthesized via the sequential action of glutamine synthetase and glutamate synthase (31,41). Therefore, it appears that the switch from producing glutamate via glutamate dehydrogenase to producing it via glutamine synthetase and glutamate synthase in E. coli CFT073 somehow prevents quiescence. It has been estimated that 88% of assimilated nitrogen in E. coli, including the nitrogen in amino acids, originates from glutamate and the remaining 12% from glutamine (42). Perhaps glutamate and glutamine are in higher concentrations in gdhA mutants growing on glucose, which could contribute to higher intracellular levels of lysine, methionine, and tyrosine and as a consequence, nonquiescence.
E. coli K-12 succinate dehydrogenase mutants grow on glucose as sole carbon source because it is not necessary for the TCA cycle to function as a complete oxidative cycle to achieve growth (43). In fact, when growing on excess glucose, the E. coli K-12 TCA cycle operates in branched mode, i.e. an oxidative branch which runs from citrate to α-ketoglutarate and a reductive branch which runs backwards from oxaloacetate to succinyl-CoA ( Fig. 6 and [43]). Succinate dehydrogenase is not necessary under these conditions and both branches serve biosynthetic functions. As a result, neither branch is used for ATP generation (42). ATP is generated from glycolysis and via the phosphotransacetylase (pta)-acetate kinase (ackA) pathway producing acetate in the process ( Fig. 6 and [43]). Therefore, it appears likely that forcing the TCA cycle to operate in branched mode somehow prevents E. coli CFT073 quiescence. In branched mode, the TCA cycle is unable to regenerate oxaloacetate and therefore gets oxaloacetate from PEP via PpC (Fig. 6). Perhaps under these conditions sufficient oxaloacetate is generated via Ppc to increase intracellular levels of lysine and methionine (Fig. 6)  shown to be severely attenuated in an ascending mouse UTI model (44), indicating a possible link between in vitro non-quiescence and reduced pathogenesis in vivo.
While it is clear why the E. coli CFT073 mini-Tn5 Km non-quiescent mutants are capable of growth using glucose as sole carbon source, it is not mechanistically clear as to why the mutations prevent quiescence on glucose plates. Possibly, a gene encoding a regulator whose synthesis or activity is inhibited by various combinations of lysine, methionine, and tyrosine, promotes quiescence. The regulator would presumably be expressed or active in the vast majority of E. coli CFT073 cells on glucose plates, but in relatively few cells of the E. coli CFT073 mini-Tn5 Km non-quiescent mutants due to the complex metabolic changes that occur in such mutants (28,29,30,39,45,46). If expression and activity of a specific E. coli CFT073 regulator is critical for generation of quiescence, much as toxin/antitoxin systems appear to be critical in generating persister cells (11,47), screening more mini-Tn5 Km mutants may lead to its identification.
Furthermore, it is unclear as to why the E. coli CFT073 gdhA and sdhA mutants generate far fewer persister cells than the gnd, zwf, and pykF mutants (Table 4). Metabolic flux analysis (28,29,30) and RNA-seq (48) may prove useful in this regard.
How might the findings reported here be relevant to recurrent UTI infections and how might its relevance be tested? It is known that E. coli CFT073 utilizes amino acids and small peptides as carbon sources and a complete oxidative TCA cycle to infect the mouse urinary tract (44,49) and appears to import small peptides to grow in mouse urine in vitro (44). UPEC may also use peptides for growth in vivo in urine during human UTI (50) glycine  11  86  21200  histidine  --9470  isoleucine  90  170  478  leucine  56  149  382  lysine  7472  4059  4480  methionine  59  37  171  phenylalanine  99  187  626  proline  144  116  serine  64  70  4000  threonine  201  461  2430  tryptophan  146  312  tyrosine  13  52  1060  valine  229  748       (D), methionine and tyrosine. Plates were incubated at 37°C for 24h. intracellularly. 6,9,11 Importantly, it is also thought that QIRs may not be affected by antibiotics, contributing to treatment failures and recurrence. 8,10 Upon exfoliation of the bladder epithelial cells containing the QIRs, the UPEC would be re-exposed to the urine and could cause recurrent infection. 6 It was also shown in vitro that the amino acid content of the urine is able to reverse quiescence in UPEC, and could serve as the necessary signal for reinfection. 12 A similar physiologic state to quiescence in UPEC is bacterial persistence. [13][14][15][16] Persister cells are a small subpopulation of bacteria in a dormant, non-dividing state characterized by high antibiotic tolerance. 13,14,16 Persistence differs from antibiotic resistance in that upon regrowth of the persister cell fraction after initial antibiotic treatment, the bacteria retains sensitivity to the antibiotic and can consistently form the same persister cell fraction. 17  This fraction was further tested in our previously published CFT073 persister cell assay, 12 and showed a greater than 3-log reduction in persister cells at 24 h at 1 mg/mL (0.1% wt/vol) in the presence of ampicillin (0.1 mg/mL) when compared with ampicillin treatment alone (Figure 1). Using CFT073 persister cell reduction to guide purification, we further purified Cranf1b using porous graphitized carbon (PGC) cartridges and determined that the second fraction (entitled 10% PGC) was similarly able to reduce the CFT073 persister cells more than 3 logs at a final concentration of 0.6 mg/mL (0.06%) (Figure 2). The 10% PGC fraction was further purified by semipreparative HPLC and analyzed using NMR and LC/MS to determine the chemical components of this fraction. This eliminative cleavage generally occurs by the free carboxylate group because of its strong electron withdrawing effect. [33][34][35] Along with these unsaturated oligosaccharides, three previously reported iridoid glucosides were also isolated and identified to be 6,7dihydromonotropein (1), deacetylasperulosidic acid (2) and monotropein (3) ( Figure   S2). [36][37][38][39] To the knowledge of the authors, this is the first identification of 2 from V.

macrocarpon.
Purified cranberry juice constituents and CFT073 quiescence: Using the compounds purified from cranberry juice, we investigated whether or not the inhibition on CFT073 persister cells could also be demonstrated with quiescent CFT073. 100 μg spots of each 4, the 10% PGC fraction, mixed methyl galacturonic acid oligosaccharides fraction (Am-F6), and 1-3 were tested in the previously published quiescence lawn overlay assay. 12 The spot from the 4, the 10% PGC fraction, and the mixed methyl galacturonic acid oligosaccharides fraction (Am-F6) all showed regrowth, thus demonstrating inhibition of quiescence ( Figure 5). It was notable that the growth with the pure 4 was not as dense as with the other active samples. Each of the tested iridoids (1-3) did not show an inhibition of quiescence.

Discussion
Despite years of practice, clinical trials, and bench research, the narrative surrounding the benefits of cranberry juice in UTIs is not fully understood. There are many explanations attempting to justify its use-from the increased acidity of the juice to the anti-adhesive compounds such as PACs-but clinical evidence fails to fully support the scientificallyderived benefits. 3,23 The under-studied oligosaccharide components of cranberry juice discussed in this manuscript comprise a previously undiscovered benefit, though we have seen anti-biofilm effects with cranberry oligosaccharides previously. 32 Pectic oligosaccharides formed in the presence of a juice-making pectinase are able to reduce persistence in UPEC. Over a 1000-fold decrease in persistent UPEC could have a tremendous effect on treatment outcomes as there would be less cells 'avoiding' the antibiotic treatment through persistence. It is important to consider that any positive effects of cranberry oligosaccharides on persister cells could only be appreciated in the presence of antibiotics, so this effect could be defined as a 'treatment' benefit. However, the inhibition of quiescence in UPEC seen with the cranberry oligosaccharides could directly hinder the ability of the UPEC to establish a QIR, thereby reducing the likelihood of a recurrent UTI. 6 Inhibition of quiescence does not require antibiotics and could therefore be loosely interpreted as a 'preventative' effect on recurrent UTIs. While the in vitro effects described in this manuscript are distinct from one another, it is important to remember that persistence and quiescence are similar biological states. It is beyond the scope of this manuscript to determine if the physiologic change in UPEC associated with cranberry oligosaccharides is consistent between the anti-persistence and anti-quiescent effects. Investigations into the biological manifestations of reducing persistence and quiescence in UPEC are necessary next experiments to further understand the beneficial role of these oligosaccharides.
While the completed experiments are preliminary, the results strongly suggest that cranberry oligosaccharides formed during the juice-making process have the potential to benefit sufferers of UTIs caused by E. coli and especially in those suffering from recurrent E. coli UTIs. It is rational to surmise that the full benefit of cranberry juice may not be realized with one class of chemical compounds, but multiple. From our previous chemical purification of oligosaccharides, we know that enrichment of the oligosaccharides is nearly synonymous with removal of phenolic compounds, including the important type A PACs. 32 One could assume that a phenolic-enriched (or PACs enriched) fraction might be similarly depleted of oligosaccharides, including those identified in this study. If the ultimate goal is to understand what the full benefit of cranberry juice components are in UTIs, more careful chemistry is needed to assure that the preparations tested include the right compounds, in sufficient quantity. Exhaustive reviews of the clinical data show clearly that inconsistencies in the chemical standardization are rampant in the literature. 23 Using the juice as the supplement is the obvious answer to including all the components, but there would still need to be clarity on how much of the juice is necessary to drink, especially in the case of the oligosaccharide content.
There is still much work left to do to determine if cranberry oligosaccharides can have a meaningful effect on the treatment or prevention of UPEC UTIs.
Pharmacokinetic, in vivo modeling, and mechanistic experiments are necessary to elucidate the answer. While these gaps in the research are significant, the current data invites a new class of compounds to be considered as part of the 'active' component of cranberry juice. We strongly feel that chemical preparations of cranberry meant to yield benefits in UTIs should not exclude oligosaccharides or they may be missing a crucial component to cranberry's anti-infective cocktail. Purification of cranberry compounds: Cranberry hulls were degraded with pectinase (Klerzyme 150, DSM Food Specialties) and fractionated as previously described with modifications. 32 Briefly, 2 g of the cranberry pectinase treated powder was dissolved in 20 ml distilled water and purified using flash chromatography. The column was eluted sequentially with 500 ml DI water, 500 ml 15% methanol/water and 500 ml methanol.

Methods
Fractions from each gradient were individually pooled and lyophilized to obtain three  absorbance was set at 235 nm for optimal monitoring of dehydro-poly-galacturonic acids.
(200 rpm) at 37°C and viable counts were measured at 0, 4, and 24 hours by plating on LB media.

Quiescence lawn assay:
The procedure for this assay follows closely follows a published protocol. 12 In general, overnight cultures of CFT073 were prepared in 0.4% glucose M9 minimal media as described in "Persister cell viability assay." Bacteria from this culture were diluted to a final concentration of 10 5 CFU in 4 mL of liquid overlay media (0.2% glucose M9 minimal media with 0.9% noble agar at 45°C). Each 4 mL overlay inoculum was poured over a prewarmed (37°C) 0.2% glucose M9 minimal media agar plate immediately after inoculation. These plates were allowed to solidify at room temperature with lids slightly ajar. Test solutions or bacteria were added to the overlay media and allowed to dry before incubating the plate upside down at 37°C for 24 or 48 hours as indicated. No growth of E. coli in the overlay was considered to be quiescence if it was able to be reversed with the positive control (3 co-spots of 5 μL of each tyrosine, lysine, and methionine at 0.1 mg/mL).

Statistics:
Persister assays were completed in triplicate and compared using a two-tailed student's t test. P-values < 0.05 were considered statistically significant.

CHAPTER 6 Perspective
The current paradigm for treating bacterial infections is about identifying the causative pathogen, and choosing an antibiotic that is capable of killing or inhibiting the growth of that particular pathogen. This is the perfect system for developing antimicrobial resistance because these antibiotics that kill or inhibit growth are the exact types of environmental stressors that a bacterium would need to develop a mechanism for circumventing in order to survive. As such, bacterial resistance to antibiotics is growing at an alarming rate and actions are urgently needed to avert this impending pandemic.
The identification of new antibacterial molecules and the development of these chemical compounds into medicines remains a crucial component in our arms race against bacteria. Declining participation in antibiotic research from the pharmaceutical industry does not help this problem. With fewer resources and time at the disposal of researchers, we must choose wisely where we search for these new antibacterial agents that we desperately need. Investigating natural products (especially those from microorganisms) has been by far the most successful method for finding molecules that become antibiotic drugs, and this remains a viable resource going forward. Importantly, while natural products research is thought of as a 'classical' research method, it has been decades since the development of natural antibacterial compounds was commonplace in industry and there have been thousands of antibacterial compounds isolated and published since. In addition, with new frontiers such as microorganisms from the marine environment to explore, the microbial natural resource of antibacterial agents remains largely undiscovered.
Continuing to fight fire with fire-that is killing bacteria with antibiotics until they develop resistance to them-is a never-ending battle. In order to fully sustain the monumental achievement of antibiotics, we must target the very mechanisms of resistance and the characteristics of bacteria that make them pathogenic to find new ways to treat infections. Efflux pumps and the subsequent efflux pump inhibitors are one good example of a way to reclaim antibiotics that are losing their potency. Other molecules have shown utility as drugs to re-sensitize resistant pathogens to antibiotics.
Additionally, targeting mechanisms of pathogenicity such as persistence and quiescence in this manuscript, or other characteristics like biofilms and quorum sensing are viable avenues towards stopping infections without stimulating resistance. Also beyond antibacterial compounds, we can learn new ways to treat infections from simple things like cranberry juice. Oligosaccharides with no antibacterial effect whatsoever through a different biological lens look particularly important to the juice's effects in UTIs. We must increase our understanding of how and why the pathogens are causing infections in order to find the best ways of treating them.
I will end this dissertation with a few broadly reaching statements and some thought provoking questions from working in this particular area of research: Preventing an infection is far better than having to treat it, but bacteria are such a huge part of human life, it seems no amount of standardized precautions would stop infections from happening all together, that being said, we must have viable ways to treat them or we face losing vital aspects of medicine that are taken for granted. Currently, research into developing truly new ways to treat bacterial infections is a bare industrial landscape. Who is going to pay for clinical trials of non-antibiotic agents for infections if the industry will not? Or better yet, what do we have to change in our health care system to incentivize developing antibiotics and similar agents? Perhaps we as a society are not yet ready to make these changes and commit to whatever it takes to preserve our antibacterial luxury. Or maybe the data is not compelling enough to force change in this developmental landscape. What will be the cost of failing to develop new antibiotics for 5 more years? What about 10 or 20? I shudder to think about pan-drug resistant Acinetobacter baumannii colonizing the majority of our hospitals. Do we really know how long it will take for something like that to happen? If we only lose 23,000 lives in this country to MDR bacteria per year now, how many lives must be lost per year for the general public to become aware of this issue? I sincerely hope that the problem of antimicrobial resistance will be solved quietly, but I fear that the noise must become louder to draw the attention necessary to prevent the impending post-antibiotic era.
then resolubilized in 33% glacial acetic acid and the optical density (OD) of stained adherent bacterial films was read at 570 nm using a spectrophotometer (ELX800, Biotek,           Table S2. 13 C and 1 H NMR chemical shifts of 1, 2 and 3.