Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Regulation of intrinsic antibiotic resistance systems in mycobacteria and identification of synergistic… Lim, Leah Elizabeth 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2012_spring_lim_leah.pdf [ 12.84MB ]
Metadata
JSON: 24-1.0072732.json
JSON-LD: 24-1.0072732-ld.json
RDF/XML (Pretty): 24-1.0072732-rdf.xml
RDF/JSON: 24-1.0072732-rdf.json
Turtle: 24-1.0072732-turtle.txt
N-Triples: 24-1.0072732-rdf-ntriples.txt
Original Record: 24-1.0072732-source.json
Full Text
24-1.0072732-fulltext.txt
Citation
24-1.0072732.ris

Full Text

REGULATION OF INTRINSIC ANTIBIOTIC RESISTANCE SYSTEMS IN MYCOBACTERIA AND IDENTIFICATION OF SYNERGISTIC DRUG COMBINATIONS TO OVERCOME INTRINSIC ANTIBIOTIC RESISTANCE by LEAH ELIZABETH LIM B.Sc., Simon Fraser University, 2009 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Microbiology & Immunology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2012 © Leah Elizabeth Lim, 2012 Abstract Mycobacterium tuberculosis, the causative agent of tuberculosis, is an adept pathogen partly due to its intrinsic antibiotic resistance systems which make an infection difficult to treat with antibiotics. Parts of these intrinsic antibi- otic resistance systems are regulated by the product of the gene whiB7 . A yeast two-hybrid experiment suggested that WhiB7 might interact with the anti-sigma factor RsbW, indirectly implying involvement with the alternate sigma factor SigF. Co-expression of WhiB7 and RsbW in Escherichia coli followed by pull-down experiments did not support this hypothesis. In order to understand mycobacterial instrinsic antibiotic resistance, a new screening method was designed and implemented to identify pairs of compounds that inhibited mycobacteria by synergistic interactions. My- cobacterium bovis BCG(lux) cells were exposed to pairwise compound com- binations and assessed for viability. A computer program was created to identify synergistic combination leads from the screen. The synergistic in- teractions might provide us with a greater understanding of the intrinsic antibtiotic resistance systems in mycobacteria. A sample of the leads from the screen tested further confirmed that 38% of domperidone combinations and 25% of rifampicin combinations were synergistic. The data from the screen has also identified new potential anti-mycobacterial compounds, specifically the avermectins. The avermectins, as a family, have never been reported to demonstrate antibiotic activity, but they have been used to treat people and other mammals for parasitic infections. The an- tibiotic activity of the avermectins was investigated in this work and found to be specific to mycobacteria. The potential of this compound family as a possible future treatment for tuberculosis therapy warrants further study. ii Preface The protocols in this work were approved by the UBC Biosafety Committee. The certificate number is B07-0169. iii Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . iv List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . x Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . xi 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 The tuberculosis threat . . . . . . . . . . . . . . . . . . . . . 1 1.2 The intrinsic antibiotic resistance of Mycobacterium tuberculosis 1 1.3 WhiB7 regulates intrinsic antibiotic resistance . . . . . . . . . 3 1.4 Screening for synergistic combinations that target intrinsic antibiotic resistance and inhibit growth . . . . . . . . . . . . 5 I Regulation and interactions of WhiB7 . . . . . . . . . . 7 2 Materials and methods . . . . . . . . . . . . . . . . . . . . . 8 2.1 Bacterial strains and culture conditions . . . . . . . . . . . . 8 2.2 Primers and plasmids . . . . . . . . . . . . . . . . . . . . . . 10 2.2.1 Primers . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2.2 Plasmids . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3 Cloning methods . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3.1 Cloning GST-rsbW . . . . . . . . . . . . . . . . . . . 13 iv 2.3.2 Cloning His-G-WhiB7 . . . . . . . . . . . . . . . . . . 15 2.3.3 Cloning M. smegmatis rsbW for overexpression . . . . 15 2.4 Protein binding experiments in E. coli . . . . . . . . . . . . . 16 2.4.1 Protein co-expression and pull-down in E. coli . . . . 16 2.4.2 SDS-PAGE and western blotting . . . . . . . . . . . . 17 2.5 Antibiotic sensitivity assays . . . . . . . . . . . . . . . . . . . 17 2.5.1 Resazurin and MTT assays . . . . . . . . . . . . . . . 17 2.5.2 Disc diffusion assays . . . . . . . . . . . . . . . . . . . 18 3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.1 Protein binding experiments in E. coli . . . . . . . . . . . . . 19 3.2 Antibiotic sensitivity profiles of M. smegmatis sigF and rsbW strains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.2.1 Generating M. smegmatis vector control and rsbW -OV 22 3.2.2 Comparing antibiotic susceptibilities in liquid and agar cultures . . . . . . . . . . . . . . . . . . . . . . . . . . 22 4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 II Drug-drug interaction screening . . . . . . . . . . . . . 33 5 Materials and methods . . . . . . . . . . . . . . . . . . . . . 34 5.1 Bacterial strains and culture conditions . . . . . . . . . . . . 34 5.2 Antibiotic combinations used in drug-drug interaction screening 34 5.3 Screening for drug-drug interactions against mycobacteria . . 36 5.3.1 Screen for drug-drug interactions against M. smegma- tis on NE agar . . . . . . . . . . . . . . . . . . . . . . 36 5.3.2 Screening for drug-drug interactions against M. bovis BCG on 7H9-agar . . . . . . . . . . . . . . . . . . . . 38 5.3.3 Screening for drug-drug interactions against M. bovis BCG in liquid 7H9 microtiter cultures . . . . . . . . . 39 5.4 Validating the results of the screens using the checkerboard assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 5.4.1 Agar lawn checkerboard assay . . . . . . . . . . . . . . 46 5.4.2 Disc diffusion assays . . . . . . . . . . . . . . . . . . . 46 5.5 Antimycobacterial activity of ivermectin . . . . . . . . . . . . 50 5.5.1 Determining the MICs of four avermectins . . . . . . . 50 5.5.2 Evidence for bacteriostatic versus bactericidal activity 50 v 6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 6.1 Screen for drug-drug interactions against M. smegmatis and M. bovis BCG on agar . . . . . . . . . . . . . . . . . . . . . 51 6.1.1 Validation of hits in a liquid culture assay . . . . . . . 51 6.1.2 M. smegmatis NE agar lawn validation of synergy be- tween ethambutol and compound Sweet-1F4 . . . . . 54 6.2 Drug-drug interactions screening in liquid 7H9 microtiter cul- ture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 6.3 Ivermectin is bacteriostatic against some Mycobacterium species 65 7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 7.1 The agar drug-drug interactions screen identified six potenti- ating combinations . . . . . . . . . . . . . . . . . . . . . . . 68 7.2 Four combinations from the liquid 7H9 drug-drug interactions screen were validated for synergy . . . . . . . . . . . . . . . . 70 7.2.1 Evaluation of the screen as a means of identifying syn- ergistic partners . . . . . . . . . . . . . . . . . . . . . 71 7.3 Data from screens can be mined for new anti-mycobacterials . 80 7.3.1 The avermectins may be a future TB treatment . . . . 80 7.4 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 A Ethambutol and compound Sweet-1F4 liquid NE checker- board results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 B “Screen Analysis” program code . . . . . . . . . . . . . . . . 91 C Disc diffusion interaction assay . . . . . . . . . . . . . . . . . 99 vi List of Tables 2.1 Mycobacterial strains used in these studies . . . . . . . . . . 9 2.2 E. coli strains used in these studies . . . . . . . . . . . . . . . 11 2.3 Primers used in these studies . . . . . . . . . . . . . . . . . . 12 2.4 Plasmids used in these studies . . . . . . . . . . . . . . . . . . 14 2.5 Antibiotics tested in these studies . . . . . . . . . . . . . . . . 18 3.1 M. smegmatis sensitivity to antibiotics was not impacted by sigF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2 Comparison of halos of inhibition of various M. smegmatis strains in response to antibiotic treatment. . . . . . . . . . . . 28 5.1 Mycobacterial strains used in these studies . . . . . . . . . . 35 5.2 Other bacterial species tested for susceptibility . . . . . . . . 35 5.3 Primary antibiotic compounds used in the drug-drug interac- tion screens . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 6.1 Liquid culture validation results of interaction hits . . . . . . 53 6.2 Checkerboard interactions of ivermectin leads from the liquid microtiter culture synergy . . . . . . . . . . . . . . . . . . . . 59 6.3 Checkerboard interactions of rifampicin leads from the liquid microtiter culture synergy . . . . . . . . . . . . . . . . . . . . 62 6.4 MIC of the avermectins for various Mycobacterium spp. . . . 66 A.1 Photometric values of the ethambutol and compound Sweet- 1F4 liquid NE liquid checkerboard assay . . . . . . . . . . . . 90 vii List of Figures 3.1 WhiB7 and RsbW were co-expressed in E. coli . . . . . . . . 20 3.2 RsbW does not co-purify with WhiB7 . . . . . . . . . . . . . 21 3.3 Does WhiB7 co-purify with RsbW? . . . . . . . . . . . . . . . 21 3.4 Transformation and integration of rsbW into M. smegmatis . 22 3.5 Antibiotic susceptibility of M. smegmatis vector control ver- sus rsbW -OV strains in liquid culture. . . . . . . . . . . . . . 25 3.6 Antibiotic susceptibility of M. smegmatis ∆whiB7 and ∆whiB7 +rsbW -OV strains in liquid culture. . . . . . . . . . . . . . . 26 3.7 Antibiotic susceptibility of M.smegmatis strains by the disc diffusion assay . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.1 Proposed interaction of WhiB7, RsbW and SigF . . . . . . . 30 5.1 Drug-drug interactions screen scheme in 7H9 liquid . . . . . . 41 5.2 General plate setup for drug-drug interaction screen in 7H9 microtiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.3 Flow of program designed to find synergy hits . . . . . . . . . 44 5.4 Pictorial representation of the liquid culture checkerboard assay 47 5.5 Pictorial representation of an potentiation assay for validating the type of interaction of two compounds . . . . . . . . . . . 49 6.1 An example of the M. smegmatis NE agar drug-drug interac- tion screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 6.2 FICI of ethambutol versus compound Sweet-1F4 in an agar checkerboard . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 6.3 Ethambutol disc diffusion interaction assay . . . . . . . . . . 56 6.4 Illustration of the four categories of curves seen in isobolograms 58 6.5 Isobolograms (ivermectin combinations) . . . . . . . . . . . . 61 6.6 Isobolograms (rifampicin combinations) . . . . . . . . . . . . 64 6.7 Time-course activity of ivermectin against M. bovis BCG . . 67 viii 7.1 Concept of generating false positives by using 14 dilutions in the screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 7.2 Graphical comparison of a linear versus non-linear relation- ship between concentration and inhibitory activity . . . . . . 74 7.3 Average absorbance versus concentration for domperidone, rifampicin and ivermectin . . . . . . . . . . . . . . . . . . . . 77 7.4 Activity kinetics curves of some 2◦ compounds across a range of concentrations . . . . . . . . . . . . . . . . . . . . . . . . . 79 C.1 Disc diffusion interaction of ethambutol and compound Sweet- 1F4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 ix List of Abbreviations Abbreviation Expanded ADC albumin, dextrose, catalase ADS albumin, dextrose, saline ATCC— American Type Culture Collection bp base pair BCG Bacillus Calmette-Guérin BLAST basic local alignment search tool dNTP deoxyribonucleotide triphosphate FIC(I) fractional inhibitory concentration (index) GST glutathione S-transferase IPTG isopropyl β-D-1-thiogalactopyranoside LB Lysogeny broth MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide NAD+/NADH nicotinamide adenine dinucleotide OADC oleic acid, albumin, dextrose, catalase OADS oleic acid, albumin, dextrose, saline OD optical density PCR polymerase chain reaction SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis TB Tuberculosis X-gal bromo-chloro-indolyl-galactopyranoside x Acknowledgements I would like to express my gratitude to my Supervisor, Charles Thompson, for his enthusiasm and support of my work both in the lab and outside the lab. Thank you to the members of my Committee, Dr. Yossef Av-Gay and Dr. Rachel Fernandez, for lively and supportive meetings regarding my work. The work described in Part I was supported with a Postgraduate Scholar- ship from the Natural Sciences and Engineering Research Council of Canada. My thanks and best wishes to my lab mates, Carol, Gaye, Jan and Santi, for the years of collaboration - I think our interaction might be synergistic! I would especially like to acknowledge Keerat Sidhu, who performed the ex- periments in section 5.4.2 under my supervision. I give tribute to Santi and Carol for performing the M. tuberculosis avermectin sensitivity assay exper- iments and to Santi for providing me with domperidone checkerboard data. I would also like to acknowledge the mentorship of Dr. Takeshi Murakami, who provided the initial inspiration for the avermectin experiments. A final thank you to my parents, my siblings and my fiancé for their unfailing support in my pursuit of this degree. The path was easier and more joyful to travel with your good company. xi 1Introduction 1.1 The tuberculosis threat Tuberculosis (TB) is one of the world’s greatest health challenges. One- third of the world’s population is infected with the etiological agent of TB, Mycobacterium tuberculosis. One in ten of these individuals will develop active TB at some point in their lifetime and, if untreated, they will infect an additional 10 to 15 people per year. In 2009 alone, 1.7 million people died of TB [30]. The emergence of multi drug-resistant, extensively drug-resistant and totally drug-resistant M. tuberculosis (MDR-TB, XDR-TB and TDR- TB respectively) has increased the threat of TB as a major cause of loss of human life [19,42]. M. tuberculosis is an adept pathogen that is easily spread through the air. Once it has infected a host it is capable of limiting and manipulating the host immune system in order to persist and it exhibits intrinsic (natural) resistance to a wide range of chemotherapeutic agents [28,45]. 1.2 The intrinsic antibiotic resistance of Mycobac- terium tuberculosis Unlike acquired antibiotic resistance, which can be gained through random mutation or horizontal gene transfer, intrinsic antibiotic resistance results from the amalgamation of a variety of natural characteristics. M. tuberculo- sis has a unique cell wall that is different from that of most Gram-positive and Gram-negative bacteria. In addition to peptidoglycan, mycobacterial cell wall components include arabinogalactan, mycolic acids and glycolipids. There is also a mycobacterial outer membrane, or polysaccharide capsule 1 outside the plasma membrane and this structure includes approximately 140 putative membrane proteins [13,29]. The mycobacterial cell wall is relatively impermeable to hydrophobic and hydrophilic compounds and blocks many common antibiotics from entering the cell [27]. Porin-like protein channels may allow the passage of some antibiotic compounds into the M. tuberculo- sis cell [20], but a variety of internal metabolic systems are activated that prevent many of these antibiotics from acting on their targets. These sys- tems appear to be multi-functional, in that they are primarily responsible for generalized physiological roles, but they are also capable of providing general resistance mechanisms [28]. The concept of a “general” resistance mechanism is counter-intuitive when held up against the traditional understanding that antibiotics cause cell death by specific, not general, modes-of-action based on their chemical structure. Yet despite diversity of antibiotics, mycobacteria have a high level of intrinsic antibiotic resistance seemingly based on these general resistance mechanisms. An alternative hypothesis regarding antibiotic-mediated cell death states that, regardless of class, bactericidal antibiotics stimulate the production of hydroxyl radicals which ultimately result in cell death [21]. Hydroxyl radical formation can be an end product of a cascade of events that take place as a general response to antibiotic challenge. In this model, the stress of bactericidal antibiotics stimulates hyperactivation of the electron transport chain, which results in superoxide formation. The superoxides damage iron-sulfur clusters which frees ferrous iron for oxidation by the Fenton reaction and this leads to hydroxyl radical formation. The hydroxyl radicals damage DNA, proteins and lipids which results in bacterial cell death [21]. The general antibiotic resistance that frustrates clearance of TB infec- tion [28] presents an attractive target for new chemotherapies. If the under- lying general resistance mechanism (against antibiotics or against the toxic metabolites they induce) can be impaired or undermined, then there may be many compounds that could then be used to treat TB that are currently considered ineffective. This manuscript describes two approaches I took to learn more about how these resistance systems operate. In section 1.3 and Part I I analyze a single component of M. tuberculosis, the WhiB7 protein, and its role as a regulator of intrinsic antibiotic resistance at the gene regu- latory level. In section 1.4 and Part II I describe a broad screening method to identify compounds that interfere with intrinsic antibiotic resistance sys- tems and may lead us to the cellular target or pathway of intrinsic antibiotic resistance in future studies. 2 1.3 WhiB7 regulates intrinsic antibiotic resistance In order for general physiological systems to act effectively in a multi- functional way to confer antibiotic resistance in mycobacteria, they should be appropriately regulated in response to the presence of antibiotics. Within the Actinomycetes phylum, expression of regulatory proteins from the WhiB family is induced in response to adverse environmental conditions. The WhiB proteins then regulate the expression of many other genes to initiate events such as cell division and sporulation (in Streptomyces spp.) and to control expression of a set of genes known as the resistome, which determines antibiotic resistance profiles of soil-dwelling bacteria [10, 28]. Antibiotic- induced changes in gene expression may also be mediated by alternate sigma factors of the RNA polymerase holoenzyme. These alternate sigma factors are themselves regulated by specific, antagonistic binding of anti-sigma fac- tors [35]. In actinomycetes, the combination of WhiB proteins, alternate sigma factors, anti-sigma factors and other regulatory mechanisms may re- sult in a general resistance response when a given antibiotic challenges the bacteria. I endeavoured to explore the relationship and interactions between the M. tuberculosis WhiB proteins, sigma factors and anti-sigma factors to gain a more complete understanding of how intrinsic antibiotic resistance against a wide-range of antibiotic compounds is regulated. In the context of M. tuberculosis multi-functional resistance systems, three genes are of particular interest: whiB7 , of the WhiB family, the alter- nate sigma factor F, sigF , and the sigF-specific anti-sigma factor rsbW . Previous studies showed that whiB7 is intimately linked to intrinsic antibiotic resistance [6, 26]. The expression of whiB7 is induced by sub- inhibitory concentrations of diverse antibiotics and a whiB7 -null mutant exhibits increased sensitivity to a number of antibiotics [26]. Microarray transcription profiles suggest whiB7 is a primary regulatory gene. Its reg- ulon consists of at least eight whiB7 -dependent transcripts, four of which are directly correlated with antibiotic resistance: tapC (Rv1258c), an efflux pump for aminoglycosides and tetracycline; MT1519 (Rv1473), a putative macrolide transporter; erm (Rv1988), a ribosomal methyltransferase; and eis (Rv2416c), an aminoglycoside acetyltransferase that confers kanamycin resistance [44]. It has been proposed that whiB7 also autoregulates its own expression [26]. This hypothesis is supported by evidence that WhiB pro- teins have DNA-binding capability [36], and by the recent identification of the putative mycobacterial whiB7 promoter target sequence [6]. The related protein WhiB3 has been shown to interact with sigma factor A (SigA), and is proposed to regulate genes via SigA [38]. It is possible that WhiB7 works 3 by interacting with a sigma factor or other direct transcriptional regulator, as proposed for WhiB3. The M. tuberculosis sigF gene is homologous to Bacillus subtilis sigF and sigB which are the sporulation and stress response sigma factors, re- spectively [32]. Stress conditions also seem to induce the expression of sigF in mycobacteria [35]. When M. tuberculosis sigF is introduced into M. bovis BCG, sigF expression is induced by the presence of some antibi- otics [25]. Additionally, a sigF -null mutant of M. smegmatis demonstrates higher transformation efficiency, which suggests it plays a role in the regula- tion of genes responsible for the building and maintenance of the mycobac- terial envelope [32]. SigF is itself regulated by the anti-sigma factor RsbW which has been proposed to regulate SigF through partner-switching between SigF and the anti-anti sigma factors RsfA and RsfB. RsbW binds RsfA under reducing conditions, as the result of a cysteine-based motif, possibly a disulfide bridge, that appears to be necessary for RsbW-RsfA interaction [21]. The other anti-anti sigma factor, RsfB, may bind RsbW in an ATP/ADP dependent manner due to a conserved serine residue at position 65 [21]. The factors that contribute to RsbW-SigF binding are currently unknown [21, 24]. Through this alternate binding, RsbW is likely a useful mechanism for M. tuberculosis to regulate gene expression in response to varying internal environmental conditions. RsbW may be directly involved with the function of WhiB7. Recently, a yeast two-hybrid experiment showed that WhiB7 and RsbW bind each other [R. Morris, personal correspondence]. It is known that RsbW specifically binds SigF [24], but the binding of RsbW to WhiB7 was an unexpected finding and it begged the question whether there is a link, physical or reg- ulatory, between WhiB7 and SigF. Part I of this thesis was based on the hypothesis that WhiB7 activity is negatively regulated by RsbW and that RsbW alternately binds WhiB7 and SigF under different stress conditions. The general aim of this project was to investigate the expression, interac- tions and functions of whiB7 , sigF and rsbW in relation to each other. I did not see evidence of any significant interaction between these genes, or their products, with regards to mycobacterial antibiotic susceptibility. 4 1.4 Screening for synergistic combinations that tar- get intrinsic antibiotic resistance and inhibit growth Targeting intrinsic antibiotic resistance systems of M. tuberculosis will likely require new drug compounds. The traditional development of a new ther- apeutic compound has been estimated to cost over $800 million US and often takes as long as 15 years to be approved for human use [2, 9]. It would be ideal to use a more economical and timely approach for developing new tuberculosis treatments. To this end, the Thompson Lab has developed drug-drug interaction screens [34]. Our approach combines the two powerful therapeutic principles of repurposing and synergy. Repurposing is the finding of new applications for currently approved drugs known to be effective in other therapies. We seek to repurpose ex- isting drugs for effective TB treatment. The synergy we are interested in occurs when combinations of drugs have a greater antimycobacterial activity together than the sum of their individual antimycobacterial activities. The screen we have developed tests a library of approved compounds for synergis- tic antimycobacterial activity. We apply this method to identify compounds or combinations of compounds that target the systems of intrinsic antibiotic resistance and inhibit mycobacterial growth. One challenge with performing a screen of this nature is the compromise between throughput-volume, efficiency, accuracy and reproducibility of the findings. The methodology of the procedure largely impacts the outcome of the experiment. This was pointed out in the paper by Pethe et al., where a chemical genetic screen performed in media containing glycerol gave disap- pointing results for drug efficacy in vivo. In this case, the lead compounds that were identified and optimized in vitro, in media containing glycerol, were not efficacious in vivo because the antibacterial properties of the lead compounds were linked with glycerol metabolism and M. tuberculosis does not metabolize glyercol during an infection [31]. In this work I used the model organisms M. smegmatis and M. bovis BCG. I tested a variety of primary compounds in pairwise combination with a library of 594 compounds (the Sweet Library) in three different screening methods. The analysis of synergy was based on the FICI analysis, which is a popular method that has been shown to work well to identify synergy [14] but it does rely on assumptions that may limit its scope [23], and appear to have impacted the results of this screen, as will be addressed in the Discussion (Part II). 5 Through these screens and follow up experiments I have identified five combinations of compounds that act in synergy in vitro against M. bovis BCG, which gives a validation rate of 11% for the best optimized screen. The combinations include: ethambutol + compound Sweet-1F4; rifampicin + compound Sweet-5H10; rifampicin + compound Sweet-5H11; rifampicin + compound Sweet-2E9; and rifampicin + compound Sweet-8A2. This work also shows that ivermectin, an anthelmentic, demonstrates antimycobacterial activity. This is surprising because no avermectin has previously been reported to demonstrate significant antibacterial activity [5]. Ivermectin is particularly promising as a potential TB treatment because it is mass-produced in oral tablet form and has practically no contra-indications in humans [18]. Overall, this work has provided a strong, multi-faceted basis for future investigations into specific targets, mechanisms-of-action and perhaps inspi- ration for in vivo testing for new TB drug treatment options. 6 Part I Regulation and interactions of WhiB7 7 2Materials and methods 2.1 Bacterial strains and culture conditions The mycobacterial species and strains used in this work are described in Table 2.1. All Mycobacteria spp. were grown in Difco Middlebrook 7H9 Broth (BD; per 180 mL: 0.94 g), with 2% glycerol (Sigma) and supplemented with 20 mL Albumin Dextrose Catalase enrichment (per 1000 mL: 8.5 g NaCl [Sigma], 50 g Bovine Serum Albumin [Sigma], 20 g D-Glucose [Bio Basic Inc.], 0.03 g catalase [Sigma]) or on Difco Middlebrook 7H10 agar (BD; 3.8 g per 180 mL: ) supplemented with 20 mL Difco BBL Middlebrook OADC Enrichment (BD). 8 Table 2.1: Mycobacterial strains used in these studies Species or strain Purpose Source Mycobacterium tuberculosis H37Rv genomic DNA as template ATCC 25618— Mycobacterium smegmatis mc26 wildtype ATCC 607— M. smegmatis mc26 ∆sigF sigF mutant Dr. Jean-Marc Reyrat M. smegmatis mc26 ∆whiB7 null-whiB7 mutant Thompson Lab (Carol Ng) M. smegmatis mc26+pMV361 vector control this work M. smegmatis mc26 rsbW -OV approximate 4x overexpression of rsbW this work M. smegmatis mc26 ∆whiB7+rsbW -OV null-whiB7 mutant and approximate 4x overexpression of rsbW this work 9 The Escherichia coli strains used for cloning and protein expression in these studies are described in Table 2.2. All E. coli strains were routinely grown in LB broth (Sigma; per 1000 mL: 20 g LB Broth) and on LB 1.5% agar (Sigma; per 1000 mL: 20 g LB Broth and 15 g Agar). For protein expression, strains were cultured in 2x YT medium (per 1000 mL: 16 g Bacto-Tryptone [BD], 10 g Bacto-Yeast extract [BD], 5 g NaCl [Sigma], pH 7.0). 2.2 Primers and plasmids 2.2.1 Primers Primers were designed using the M. tuberculosis H37Rv rsbW and whiB7 gene sequences obtained from the TubercuList Server (http://genolist.pasteur.- fr/TubercuList/). All primers were synthesized at the Nucleic Acid Protein Service unit-Integrated DNA Technologies (University of British Columbia). Table 2.3 describes primers used in cloning for protein expression in E. coli and for gene over-expression in M. smegmatis. 10 Table 2.2: E. coli strains used in these studies Strain Purpose Source E. coli JM107 cloning with pGEX4T-1 plasmid gift from Dr. John Smit E. coli TOP10 cloning with the pET29a plasmid Invitrogen E. coli BL21 (DE3) protein expression Invitrogen E. coli LIM-1 carrying pLNG1 this work E. coli LIM-2 carrying pGEX4T-1 this work E. coli LIM-3 carrying pCR1 this work E. coli LIM-4 carrying pET29a this work E. coli LIM-5 carrying pLNG3 this work E. coli LIM-6 carrying pLNG2 this work E. coli LIM-7 carrying pLNG5 this work E. coli LIM-8 carrying pLNG6 this work E. coli LIM-10 carrying pLNG4 this work 11 Table 2.3: Primers used in these studies Name Direction Sequence Note Primers: Protein expression in E. coli LEL-1 forward ATAAGGATCCATGGCCGACTCG- GATTTACC upstream of M. tuberculosis H37Rv rsbW , inserted BamHI site LEL-2 reverse ATAACTCGAGTCACCTGCTG- GATGCCGC downstream of M. tuberculosis H37Rv rsbW , inserted XhoI site Primers: Over-expression in M. smegmatis LEL-5 forward ATCAGAATTCGGTCAGATGGC upstream of M. smegmatis mc2155 rsbW , inserted EcoRI site LEL-6 reverse ATCCAAGCTTTATTCCGACGTCAC downstream of M. smegmatis mc2155 rsbW , inserted HinDIII site 361fwd forward GGAATCACTTCGCAATGG sequencing primer, upstream of pMV361 multiple cloning site 361rev reverse CCCGTTGAATATGGCTCATAAC sequencing primer, downstream of pMV361 multiple cloning site 12 2.2.2 Plasmids All plasmids were constructed using standard restriction enzyme digestion and ligation techniques (described in section 2.3). The plasmids used in these studies are listed in Table 2.4. 2.3 Cloning methods 2.3.1 Cloning GST-rsbW The M. tuberculosis rsbW gene was amplified by PCR from M. tuberculosis H37Rv genomic DNA (prepared by Carol Ng, Thompson Lab). The PCR mixture included 5 µL DyNAzyme— II Hot Start Reaction Buffer, 1 µL 10mM dNTPs, 1 µL of forward primer LEL-1, 1 µL of reverse primer LEL-2 (see Table 2.3), 0.5 µL M. tuberculosis genomic DNA (40 ng/ µL), 0.6 µL DyNAzyme— II Hot Start DNA Polymerase (New England BioLabs) and 40.9 µL H2O. The PCR conditions were: 94◦C for 10 minutes followed by thirty repeats of 94◦C for 20 seconds, 65◦C for 20 seconds and 72◦C for 30 seconds, after which there was a final incubation of 72◦C for 7 minutes 30 seconds. The PCR products were purified using the illustra GFX— PCR DNA and Gel Band Purification Kit, according to the manufacturer’s instructions (GE Healthcare) and ligated into pGEM®-T Easy. The ligation mixture included 5 µL 2X Rapid Ligation Buffer, 1 µL pGEM®-T Easy Vector, 1 µL PCR Product (24.2ng/ µL), 1 µL T4 DNA Ligase and 2 µL H2O. The mixture was incubated overnight at 4◦C. The ligation products were run on a 0.7% agarose gel electrophoresis (stained with SYBR Safe), the 3473 bp band was excised and purified (using the illustra GFX— PCR DNA and Gel Band Purification Kit). One µL of the plasmid prep was transformed into 50 µL electrocompetent E. coli JM107 cells by electroporation (using the GenePulse XCell—(Bio-Rad) at 1800 V, 25 µF, 200 Ω, 1 mm cuvette), plated on LB agar + 100 µg/mL ampicillin + X-gal and incubated at 37◦C overnight. White colonies were inoculated into 5 mL LB + 100 µg/mL ampicillan and incubated at 37◦C for 6 hours. A plasmid miniprep was performed using the PureYield— Plasmid Miniprep System, according to the manufacturer’s instructions. The plasmids were sequenced by GENEWIZ, Inc. and one clone with the correct insertion was named pLNG1 (Table 2.4), the E. coli strain carrying pLNG1 was named LIM-1 (Table 2.2). The plasmids pLNG1 and pGEX4T-1 were digested with restriction en- 13 Table 2.4: Plasmids used in these studies Plasmid Purpose Source pGEM®-T Easy cloning PCR products Promega pLNG1 amplify PCR product of M. tuberculosis rsbW this work Protein expression in E. coli pGEX4T-1 parental plasmid for GST-fusion GE Healthcare pLNG2 GST-fused RsbW protein expression this work pCR1 template of 6x his-tagged, GB1-fused [17] WhiB7 sequence Carol Ng (Thompson Lab) pET29a parental plasmid with kanamycin resistance Invitrogen pLNG4 His-G-WhiB7 protein expression, kanamycin resistance marker Jan Burian (Thompson Lab) Over-expression in M. smegmatis pMV361 parental plasmid for mycobacterial cloning [39] pLNG6 M. smegmatis RsbW gene overexpression this work 14 donucleases (New England BioLabs) in the following mixtures: 25 µL 50 ng/ µL pLNG1 or pGEX4T-1 DNA, 0.8 µL BamHI 20,000 Units/mL and 0.8 µL 20,000 Units/mL XhoI, 5 µL NE Buffer 2 and 21.4 µL H2O. The digestion mixture was incubated for 1 hour at 37◦C. The pLNG1 digest was run through a 1% agarose gel electrophoresis (stained with SYBR Safe) and the 438 bp band of rsbW was excised. Both this rsbW DNA fragment and the linearized pGEX4T-1 were purified, then put in a ligation mixture (3 µL rsbW fragment, 1 µL linearized pGEX4T-1, 4 µL 5X buffer, 1 µL T4 DNA Ligase [USB Corp.]) and incubated overnight at room temperature. The ligation products were transformed directly into E. coli JM107 by electroporation (as described previously) then plated on LB agar + 100 µg/mL ampicillin and incubated at 37◦C overnight. The remaining cultur- ing, miniprep and sequencing was performed as described. The plasmid that was confirmed for the correct sequence was named pLNG2 (Table 2.4), the E. coli strain carrying pLNG2 was named LIM-6 (Table 2.2). 2.3.2 Cloning His-G-WhiB7 The plasmid pCR1 contains a 6x His-tagged, GB1-fusion whiB7 sequence (His-G-WhiB7), however the selection of this plasmid is ampicillin resis- tance, the same as pLNG2. For selection of co-transformants using two selec- tion criteria, the His-G-WhiB7 sequence was incorporated into the pET29a plasmid backbone which offers kanamycin resistance. Jan Burian (Thomp- son Lab) generated this plasmid, named pLNG4. The E. coli TOP10 strain carrying pLNG4 (Table 2.4) was named LIM-10 (Table 2.2). 2.3.3 Cloning M. smegmatis rsbW for overexpression The rsbW gene from M. smegmatis genomic DNA was PCR amplified using the primers LEL-5 and LEL-6 (Table 2.3). The cloning steps followed the same procedure as described in subsection 2.3.1, except the restriction en- donucleases used were EcoRI and HinDIII (20,000 Units/mL each, New Eng- land BioLabs), the plasmid backbone used was the integrative pMV361 [39], conferring kanamycin resistance (Table 2.3) and the host strain was E. coli TOP10 (carrier was renamed LIM-8 [Table 2.2]). The plasmid was later isolated from LIM-8 by plasmid miniprep and transformed by electroporation into M. smegmatis mc26, then renamed M. smegmatis mc26 rsbW -OV (Table 2.1). Similarly, an empty pMV361 was cloned for the vector control strain. 15 2.4 Protein binding experiments in E. coli 2.4.1 Protein co-expression and pull-down in E. coli The plasmids pLNG2 and pLNG4 were electroporated into E. coli BL21 (DE3) cells (as described previously), plated on LB agar + ampicillin (100 µg/mL) + kanamycin (30 µg/mL) and incubated overnight at 37◦C. Colonies were then inoculated into 5 mL LB + ampicillin (100 µg/mL) + kanamycin (30 µg/mL) and incubated overnight at 37◦C with shaking. Protein expres- sion was induced by adding IPTG to a final concentration of 0.1 mM and the culture was incubated overnight at 23◦C with shaking. Purification by 6xHis-tag For purification by the 6xHis-tag, a protein miniprep was performed using a modified protocol from the QIAexpressionist—( [33]). Briefly, cells were harvested by centrifuging 3 mL of culture at 14,000 × g and resuspended in 300 µL lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0), with lysozyme added to 1 mg/mL and DTT added to 10 mM, the mixture was incubated on ice for 30 minutes. After incubation cells were lysed by vortexing, the lysate was centrifuged for 10 minutes at 14,000 × g and the supernatent isolated. To this supernatent, 20 µL of 50 % slurry of Ni-NTA resin (Qiagen) was added and the mixture incubated for 30 minutes at 4◦C with gentle rotation. After 30 minutes, the resin was washed twice with wash buffer (50 mM NaH2PO4, 300 mM NaCl and 20 mM imidazole, pH 8.0), with brief centrifugations between washes. Finally, the protein was eluted from the resin by 3 elutions with 20 µL elution buffer (50 mM NaH2PO4, 300 mM NaCl and 250 mM imidazole, pH 8.0). Purification by GST-tag For purification by the GST-tag, a protein miniprep was performed as fol- lows: Cells were harvested by centrifuging 3 mL of culture at 14,000 × g and resuspended in 100 µL of BugBuster— 10x Protein Extraction Reagent (Novagen) and mixed gently at room temperature for 15 minutes. Next, 200 µL cold PBS was added, then the lysate was centrifuged for 10 minutes at 14,000 × g and the supernatent was transferred to a fresh microfuge tube where 20 µL of a 50% slurry of Glutathione Sepharose 4B (GE) was added. This was mixed gently at room temperature for 5 minutes. The beads were washed 3 times by adding 100 µL PBS, centrifuging briefly and discarding the supernatent. The protein was eluted by adding 10 µL elution buffer 16 (Sigma; 50 mM Tris-HCl, 10 mM reduced glutathione, pH 8.0) and incubat- ing at room temperature for 5 minutes, then the mixture was centrifuged briefly and the supernatent collected for SDS-PAGE and western blotting. 2.4.2 SDS-PAGE and western blotting Whole cell lysates, supernatent and aliquots of the purified protein preps were mixed with equal volumes SDS Loading Buffer (10 mL: 2.5 mL 0.5 M Tris-HCL pH 6.8, 2 mL glycerol, 4 mL 10% w/v SDS, 0.5 mL 0.1% w/v bromophenol blue, 1 mL H2O and add 2-Mercaptoethanol to 2.5%), incu- bated in boiling water for 5 minutes and 12 µL loaded on a 10% tricine gel (prepared as described in [37]). The SDS-PAGE was run using the Power- Pac HC (Bio-Rad) for .03 A at 45 minutes, then 185 V for 1 hour immersed in anode and cathode buffers [37]. The products were transferred to nitro- cellulose using the Trans-Blot® SD Semi-Dry Electrophoretic Transfer Cell (Bio-Rad) at 15 V for 30 minutes, soaked in electrode buffer (300 mM Tris, 100 mM acetic acid, pH 8.6). The nitrocellulose was incubated with Blocking Buffer for Near InfraRed Fluorescent Western Blotting (Rockland) for 1 hour at room temperature with rocking and incubated with Blocking Buffer + 0.5 µg/mL Anti-GST (Rabbit) Antibody IRDye 700DX® Conjugated (Rockland) + Anti 6X HIS EPITOPE TAG (MOUSE) Monoclonal Antibody IRDye 700DX® Conju- gated (Rockland) for 1 hour. The membrane was washed six times with PBS-T (Sigma; per 1000 mL: 8 g NaCl, 0.2 g KCl, 2.68 g Na2HPO4-7H2O, 0.24 g KH2PO4 and 1% v/v Tween 20 adjusted to pH 7.4) before visualiza- tion using the Odyssey® Imager (LI-COR Biosciences, settings used were Resolution: 169 µm, Quality: medium, Focus Offset: zero, Intensity: 5.0), made available for use by Dr. Ninan Abraham. 2.5 Antibiotic sensitivity assays 2.5.1 Resazurin and MTT assays Microtiter 96-well clear Flat Bottom Assay Plates (Costar) were prepared with eight serial two-fold dilutions of antibiotics in 7H9. M. smegmatis strains (Table 2.1) were grown to exponential phase (OD600=0.4-0.8) then inoculated to a final OD600=0.0025 to the prepared microtiter wells. The antibiotics tested are listed in Table 2.5. The assay plates were wrapped in aluminum foil to prevent vapour loss and incubated for 3 days at 37◦C then 100 µL of 10 mg/mL resazurin 17 Table 2.5: Antibiotics tested in these studies Antibiotic Highest concentration (µg/mL) Clindamycin 512 Chloramphenicol 128 Fusidic Acid 1024 Spectinomycin 512 Tetracyclin 2 Imipenem 128 Azithromycin 128 Clarithromycin 128 Roxythromycin 128 Capreomycin 8 Clofazimine 8 Isoniazid 512 Rifampicin 256 Streptomycin 8 Ethambutol 8 Triclosan 32 Erythromycin 32 (Sigma) or 50 µL of 5 mg/mL MTT (Sigma) was added to each well. The plates were incubated overnight at 37◦C. Plates with resazurin were read by photometric measurement of wavelength 570 nm, Bandwidth 5 nm and Measurement time 100 ms Varioskan Flash (Thermo Scientific), made available by Dr. Steven Hallam. The plates with MTT had an additional 50 µL of 10% SDS (w/v) added, another overnight incubation at 37◦C followed by a photometric measurement of wavelength 560 nm, bandwidth 5 nm and measurement time 100 ms Varioskan Flash. 2.5.2 Disc diffusion assays M. smegmatis liquid culture was diluted to an OD600=0.0125 and spread evenly on 7H10 + ADC agar plates. A paper disc (BBL—, 6 mm diameter) was placed on the surface of the agar and 2.5 µL antibiotic solution was applied to each disc. Plates were incubated at 37◦C for 4 days, and the halo of inhibition was measured (in mm). 18 3Results 3.1 Protein binding experiments in E. coli The protein co-expression and purification experiments were carried out as described in section 2.4. Both GST-RsbW and His-G-WhiB7 were success- fully expressed in E. coli BL21(DE3) cells along with empty vector controls (see Figure 3.1). The GST-RsbW was predicted to be approximately 44 kDa, the His-G-WhiB7 band was predicted to be approximately 16 kDa (al- though it typically runs >20 kDa, [unpublished data]) and the GST-tag is reported to be 26 kDa [4]. The His-purification procedure was capable of pulling down the His- G-WhiB7 but the GST-RSbW did not co-elute with it (Figure 3.2). The GST-purification was successful for purifying GST-RsbW, and one out of four trials did pull-down the His-G-WhiB7 as well (Figure 3.3). This result could not be reproduced and may be due to a technical error. 19 Figure 3.1: WhiB7 and RsbW were co-expressed in E. coli. A western blot of various E. coli BL21(DE3) whole cell lysates probed with anti-His and anti-GST antibodies show successful co-expression. L is the protein standard ladder and the numbers indicate the kDa weight of the standard bands and + is the positive His- control (His-tagged Caulobacter spp. S-layer [provided by Jan Mertens, Smit Lab]). Lanes A-C are whole cell lysates from E. coli cultures where A expresses His-G-WhiB7 (WhiB7) and GST-RsbW (RsbW), B expresses GST-RsbW and 6xHis peptide and C expresses GST and His-G-WhiB7. The strain expressing both fusion proteins GST-RsbW and His-G-WhiB7 is the co-expression strain to test binding between RsbW and WhiB7 (A) and the other strains are the purification controls (B and C). The red arrow indicates the GST-RsbW band, the green arrow indicates the His-G-WhiB7 band and the blue arrow indicates the GST band. 20 Figure 3.2: RsbW does not co-purify along with WhiB7. A western blot of E. coli BL21(DE3) cells expressing GST-RsbW and the His-G-WhiB7 fusion proteins: Lane 1 was loaded with whole cell lysate and lane 2 was loaded with the His-purified fraction. The membrane was probed with anti-His and anti-GST antibodies. The r indicates the expected GST-RsbW band and the w indicates the expected His-G-WhiB7 band. The GST-RsbW fusion protein is present in the whole cell lysate but not in the His-purified fraction as predicted. Figure 3.3: Does WhiB7 co-purify with RsbW? A western blot of E. coli BL21(DE3) cells expressing GST-RsbW and the His-G-WhiB7 fusion pro- teins: Lane 1 was loaded with whole cell lysate and lane 2 was loaded with the GST-purified fraction. The membrane was probed with anti-His and anti-GST antibodies. The r indicates the expected GST-RsbW band and the w indicates the expected His-G-WhiB7 band. 21 3.2 Antibiotic sensitivity profiles of M. smegmatis sigF and rsbW strains 3.2.1 Generating M. smegmatis vector control and rsbW - OV The cloning method for generating the vector control and rsbW -OV strains was described in subsection 2.3.3. The parent vector, pMV361, is an integra- tive vector that inserts one copy of the gene into the genome. The inserted gene is overexpressed via the heat shock promoter, hsp60. The success of the transformation was measured by performing a PCR on a genomic prep from each clone to amplify the insertion site of pMV361. Figure 3.4 shows the agarose gel with the expected bands present in Lane 1 at 329 bp and Lane 2 at 761 bp, confirming the transformations were successful. Figure 3.4: Transformation and integration of rsbW into M. smegmatis was successful. Two clones, one transformed with rsbW (under control of the hsp60 promoter) and one with an empty-vector control were grown. A genomic prep from each strain was subjected to PCR and the products were run on a 1% agarose gel, shown here. L is the ladder, 1 is the PCR from the clone transformed with pMV361 (empty-vector control) and 2 is PCR of the clone expressing rsbW . 22 3.2.2 Comparing antibiotic susceptibilities in liquid and agar cultures The antibiotic sensitivities of the M. smegmatis mc26 wildtype strain and M. smegmatis mc26 ∆sigF were determined using the resazurin or MTT assay in liquid microtiter culture. MICs were identical between M. smegmatis mc26 wildtype strain and M. smegmatis mc26 ∆sigF for different antibiotics (see Table 3.1). The antibiotic sensitivities of M. smegmatis vector control and rsbW - OV strains were determined using the resazurin or MTT assay in liquid microtiter culture and the disc diffusion assay on agar lawns. In almost all cases, there was no significant difference in susceptibility of the two strains to the antibiotics tested (see Figure 3.5, Figure 3.7 and Table 3.2). Similarly, the antibiotic sensitivities of M. smegmatis ∆whiB7 and M. smegmatis ∆whiB7+rsbW -OV were determined using the resazurin or MTT assay in liquid microtiter culture and the disc diffusion assay on agar lawns. There was no significant difference between the control strain and the overex- pression strain in the wildtype or whiB7 -null background, with the exception of an eight-fold greater sensitivity of the M. smegmatis ∆whiB7 +rsbW -OV strain to azithromycin compared to M. smegmatis ∆whiB7 (see Figure 3.6, Figure 3.7 and Table 3.2). 23 Table 3.1: M. smegmatis sensitivity to antibiotics was not impacted by sigF. MIC (µg/mL) Antibiotic M. smegmatis mc26 wildtype strain M. smegmatis mc26 ∆sigF Clindamycin 64 64 Chloramphenicol 16 16 Fusidic Acid 128 128 Spectinomycin 32 32 Tetracycline 2 2 Imipenem 64 64 Triclosan 8 8 Azithromycin 64 64 Clarithromycin 64 64 Roxythromycin 32 32 Capreomycin 1 1 Clofazimine 1 1 Isoniazid 8 8 Rifampicin 128 128 Streptomycin 0.5 0.5 Ethambutol 0.5 0.5 24 Figure 3.5: Antibiotic susceptibility of control M. smegmatis mc26+pMV361 (empty-vector) versus M. smegmatis mc26 rsbW -OV (rsbW -OV) in liquid microtiter cultures. Values were determined using the resazurin susceptibility or MTT susceptibility assay (as indicated). 25 Figure 3.6: Antibiotic susceptibility of M. smegmatis mc26 ∆whiB7 (whiB7 -null) versus M. smegmatis mc26 ∆whiB7+rsbW -OV (whiB7 -null + rsbW -OV) in liquid microtiter M. smegmatis ∆whiB7) (control) strain and the M. smegmatis mc26 ∆whiB7+rsbW -OV strain. 26 Figure 3.7: Antibiotic susceptibility of M. smegmatis mc26 rsbW -OV (A.), empty-vector control M. smegmatis mc26+pMV361 (B.), M. smegmatis ∆whiB7 (C.) and M. smegmatis mc26 ∆whiB7+rsbW -OV (D.) Antibiotics tested were tetracycline (TET), erythromycin (ERY) and spectinomycin (SPEC) with final µg applied indicated centrally. Halos of inhibition for each assay are indicated (in mm). 27 Table 3.2: Comparison of halos of inhibition of various M. smegmatis strains in response to antibiotic treatment. Diameters of halos of inhibition (mm) M. smegmatis mc26 strains Antibiotic (total applied) empty- vector control rsbW -OV ∆whiB7 ∆whiB7+ rsbW -OV Tetracycline (1.6 µg) 30 31 50 51 Erythromycin (5 µg) 9.5 10 40 45 Spectinomycin (80 µg) 11 10 38 39 28 4Discussion The premise for exploring the relationship between whiB7 , sigF and rsbW and their gene products was four-fold: whiB7 is strongly implicated in regu- lating intrinsic antibiotic resistance [6,26]; an M. tuberculosis sigF -null mu- tant was reported to be hypersensitive to the rifamycin drugs, rifapentine and rifampicin, exclusive of changes in cell wall permeability [8]; RsbW is known to interact directly with and negatively regulate SigF [3]; and WhiB7 and RsbW demonstrated direct interaction in a yeast two-hybrid assay (R. Morris, personal communication). Based on this cumulative evidence, I hypothesized that RsbW alternately and negatively regulated WhiB7 or SigF, depending on the cellular stress conditions. This hypothesis is depicted in Figure 4.1. 29 Figure 4.1: Hypothesized relationship between WhiB7, SigF and RsbW. Arrows indicate known (black) and likely (grey) protein-protein binding in- teractions. It is unknown what kind of relationship, if any, exists between WhiB7 and SigF (white arrow). 30 To test this hypothesis, I looked for evidence direct protein-protein bind- ing in another system. I chose to clone and express M. tuberculosis genes in E. coli in order to take advantage of the expression plasmids available for E. coli systems. I expressed the rsbW with an N-terminal fusion of the GST tag and the whiB7 with an N-terminal fusion of 6xHis-GB1 on two different plasmids carrying different resistance genes. The dual resistance of success- ful co-transformations allowed me to efficiently select for co-transformants. I was successfully able to co-express the GST-RsbW and His-G-WhiB7 fusion proteins (see Figure 3.1). Because both proteins were tagged, I was able to perform a pull-down experiment by purifying either the GST tag or the 6xHis tag (Figure 3.2, Figure 3.3). These purification experiments gave conflicting results. The GST purification, expected to purify GST- RsbW and hypothesized to pull-down the His-G-WhiB7 did appear, in one instance, to do so (Figure 3.3). This result is questionable because it was seen once among eight trials (two purifications of four cultures each). It is more likely that the positive result was contamination of the whole cell lysate (containing both proteins) than a true pull-down. Consistent with the majority of the GST-RsbW pull-down results, the His-purification, expected to purify the His-G-WhiB7 and hypothesized to pull-down the GST-RsbW, never showed the GST-RsbW present in the purified fraction. In parallel to these pull-down experiments, I compared the antibiotic sensitivity spectrums of the various M. smegmatis strains: wildtype (mc26), ∆sigF, ∆whiB7, M. smegmatis mc26+pMV361 (vector control), rsbW -OV and ∆whiB7+rsbW -OV. Each strain was treated with 16 different antibi- otics representing a variety of antibiotic classes and targeting diverse cellular processes. The ∆whiB7 strains showed a much higher sensitivity to antibi- otics, as expected (see Figure 3.7). The MIC values measured in the resazurin or MTT studies are not pre- dicted to be absolute. Generally, the actual MIC is expected to fall within a 2-fold range of any given experimental result. By this standard, the ∆sigF did not demonstrate any aberrant sensitivity (see Table 3.1), even to ri- fampicin, which was surprising because the ∆sigF strain had been reported to be hypersensitive to rifamycin drugs ( [8]). According to the hypothesis that rsbW negatively regulates whiB7 , I expected to see a sensitivity profile that mimicked the ∆whiB7 strain in the rsbW -OV strain. Instead, over-expression of RsbW in both the wildtype- or ∆whiB7-background failed to show any difference in sensitivity as compared to control strains (Figure 3.5, Figure 3.6). The negative results of the protein pull-down experiments suggest that 31 RsbW and WhiB7 do not directly interact with each other on the protein level. The negative results of the antibiotic sensitivity assay further suggests that the genes or gene products do not interact with each other in a way that affects antibiotic sensitivity in M. smegmatis. Initially, the failure of the M. smegmatis ∆sigF strain to show hyper- sensitivity to rifamycin was disturbing because it conflicted with previously published results. However, in August 2010, Hartkoorn et al. reported that overexpression or deletion of sigF did not alter susceptibility to rifampin, contrary to a previous report, consistent with my findings [15]. The Hartkoorn et al. manuscript called into question the premise that sigF played a role in antibiotic resistance. Additionally, the susceptibility studies in this work suggest that RsbW does not play a large role in antibiotic resistance. The overexpression of rsbW did not affect the susceptibility of M. smegmatis, although the overexpression was only about four-fold above basal levels (C. Schumacher, qRT-PCR data not shown). The hypothesis is further voided with the findings that GST-RsbW was unable to consistently pull-down His-G-WhiB7 and that His-G-WhiB7 was unable to pull-down GST-RsbW. These data suggest that these two proteins do not directly bind one another. While each individual experiment could be optimized further, the del- uge of negative results strongly suggests that the hypothesis that RsbW alternately and negatively regulates WhiB7 or SigF is incorrect. 32 Part II Drug-drug interaction screening 33 5Materials and methods 5.1 Bacterial strains and culture conditions The Mycobacterial species and strains used in this work are described in Table 5.1. All Mycobacterium spp. were grown in Difco Middlebrook 7H9 Broth (BD; per 180 mL: 0.94 g), with 2% glycerol (Sigma) supplemented with 20 mL Albumin Dextrose Catalase (per 1000 mL: 8.5 g NaCl [Sigma], 50 g Bovine Serum Albumin [Sigma], 20 g D-Glucose [Bio Basic Inc.], 0.03 g catalase [Sigma]) or on Difco Middlebrook 7H10 agar (BD; 3.8 g per 180 mL: ) supplemented with 20 mL Difco BBL Middlebrook OADC Enrichment (BD). The drug-drug interaction screens were performed on NE agar, 7H9-agar and in liquid 7H9, as described below. For all screens, single cultures of M. smegmatis, M. bovis BCG and M. bovis BCG(lux) were grown to stationary phase, dispensed into 1 mL aliquots, frozen at -20◦C overnight and stored at -80◦C until use. Other bacterial species tested for susceptibility are described in Ta- ble 5.2. All species were grown in LB Broth media (Sigma; per 1000 mL: 20 g LB Broth) and on LB 1.5% agar (Sigma; per 1000 mL: 20 g LB Broth and 15 g Agar). 5.2 Antibiotic combinations used in drug-drug in- teraction screening This methodology was designed to test for drug-drug interactions between pairs of compounds. Each pair consisted of a primary compound (1◦) and 34 Table 5.1: Mycobacterial strains used in these studies Species or strain Purpose Source M. tuberculosis H37Rv susceptibility studies Dr. Richard Stokes’ Lab M. tuberculosis CDC1551 susceptibility studies Dr. Richard Stokes’ Lab M. tuberculosis Erdman susceptibility studies Dr. Richard Stokes’ Lab M. smegmatis mc2155 NE agar screen, susceptibility studies ATCC 700084 Mycobacterium bovis BCG 7H9-agar screen Pasteur 1173P2 M. bovis BCG(lux) [34] liquid 7H9 screen Dr. Yossef Av-Gay’s Lab Mycobacterium avium susceptibility studies ATCC 25291 Table 5.2: Other bacterial species tested for susceptibility Species or strain Gram-positive or -negative Pathogenic Escherichia coli O157:H7 Gram-negative Pathogenic Streptomyces lividans 1326 Gram-positive Non-pathogenic Rhodococcus jostii RHA1 Gram-positive Non-pathogenic Acinetobacter baumannii ATCC 1960 Gram-negative Opportunistic Pseudomonas aeruginosa PA01(H103) Gram-negative Opportunistic Kocuria rhizophila Gram-positive Non-pathogenic Staphylococcus aureus ATCC 25923 Gram-positive Pathogenic 35 a secondary partner compound (2◦). The 1◦ compounds were selected be- cause they are currently used in tuberculosis therapy or because they have particularly interesting properties that make them candidates for possible future tuberculosis therapy (see Table 5.3). The MIC of each 1◦ compound was determined experimentally for each method. The 2◦ compounds came from the Sweet Library which was compiled by Dr. Gaye Sweet (Thompson Lab). This library was assembled because results from a previous screen showed that synergistic hits were enriched for antibiotics partnered with other antibiotics [34]. The Sweet Library con- tains 594 compounds, the majority selected because they have antimicrobial activity. Each compound was dissolved in its optimal solvent to 5 mM, aliquoted into microtiter plates and stored at -80◦C. 5.3 Screening for drug-drug interactions against mycobacteria There were three screens carried out to identify compound combinations that were effective against mycobacteria. 1. NE agar screen: combinations of 1◦ and 2◦ compounds applied to M. smegmatis grown on NE agar. 2. 7H9-agar screen: combinations of 1◦ and 2◦ compounds applied to M. bovis BCG(lux) grown on 7H9-agar. 3. Liquid 7H9 screen: combinations of 1◦ and 2◦ compounds applied to M. bovis BCG(lux) grown in liquid 7H9. 5.3.1 Screen for drug-drug interactions against M. smegma- tis on NE agar Screening was carried out by transferring microliter volumes of the Sweet Library compounds to a lawn of bacteria growing on NE agar with sub- inhibitory concentrations of the 1◦ compounds. For this screen, the 1◦ com- pounds were clarithromycin, domperidone, ethambutol, moxifloxacin and rifampicin. First, the MIC, on agar, of each 1◦ compound was determined by the following procedure: 60 mm × 15 mm petri dish plates (Fisherbrand) were poured with two layers of agar. The bottom layer was 10 mL NE 1.5% agar (per 1000 mL: 10 g D-glucose, 2 g yeast extract [Sigma], 2 g casaminoacids 36 Table 5.3: The primary compounds used in the drug-drug interaction screens and the rationale for including each. Compound Media, bacteria (MIC in µg/mL) Rationale Clarithromycin NE agar, M. smegmatis (8) treat TB and M. avium infections [1] 7H9 liquid , M. bovis BCG(lux) (0.125) Clofazimine 7H9 liquid, M. bovis BCG(lux) (0.125) last-resort TB drug [1] Domperidone NE agar, M.smegmatis (264) butyrophenone, repurpos- ing [34] 7H9 liquid, M. bovis BCG(lux) (128) Ethambutol NE agar, M. smegmatis (0.5) current front-line TB drug [1] 7H9-agar, M. bovis BCG(lux) (4) 7H9 liquid, M. bovis BCG(lux) (2) Isoniazid NE agar, M. smegmatis (0.5) current front-line TB drug [1] Ivermectin NE agar, M. smegmatis (4) repurposing 7H9 liquid, M. bovis BCG(lux) (4) Moxifloxacin NE agar, M. smegmatis (0.125) last-resort TB drug [1] Rifampicin NE agar, M. smegmatis (64) current front-line TB drug [1] 7H9-agar, M. bovis BCG(lux) (64) 7H9 liquid, M. bovis BCG(lux) (0.03125) [Sigma], 1 g Lab Lemco powder [Sigma] and 15 g agar, pH 7.0), the top layer was 5 mL NE 0.5% agar (same as NE 1.5% agar, except this includes only 5 g of agar). The 1◦ antibiotics were added to both layers of agar at a range of concentrations, one concentration per plate. The top layer of agar also included M. smegmatis diluted to OD600=0.06 from frozen stock (stationary phase). The 1◦ compounds tested in these studies were clarithromycin, domperidone, ethambutol, ivermectin and rifampicin. The screen was performed with the 1◦ compounds at 8-fold and 4-fold below the previously determined MIC (18 MIC and 1 4 MIC) in combination with the Sweet Library compounds. The MIC for each compound are listed 37 in Table 5.3. First, bottom agar (NE 1.5% agar) was freshly prepared and 1◦ antibiotic added directly to the molten agar at either 18 MIC or 1 4 MIC. Twenty-two mL of bottom agar were pipetted into a sterile, single-well omni tray (Nunc, from VWR). This layer was allowed to solidify. Next, top agar (NE 0.5% agar) was freshly prepared, 1◦ antibiotic added to the molten agar at either 1 8 MIC or 1 4 MIC and M. smegmatis diluted to OD600=0.06 in the molten agar. Twenty-two mL of the top agar layer were pipetted evenly on top of the solidified bottom agar, then allowed to set. Control plates were prepared without antibiotics in the agar. The Sweet Library compounds (thawed) were shaken on a Titer Plate Shaker (Banstead Lab-Line) for 1 minute, then centrifuged at 1000 × g for 5 minutes (Sorvall LegendRT). Using the Boekel 96 pin replicator, the Sweet Library compounds were transferred to the solidified top layer of agar. The replicator was rinsed in 70% ethanol and flame sterilized. The omnitrays were covered in foil and incubated at 37◦C for 3 days. All pipetting and pinning steps were performed in a biological safety cabinet. After incubating for 4 days, the plates were scanned with a Canon CanoScan 5600F machine, and analysis was performed using the scanned images. 5.3.2 Screening for drug-drug interactions against M. bovis BCG on 7H9-agar The preliminary set-up and the screening was carried out similarly to the M. smegmatis drug-drug interaction screen, as described in subsection 5.3.1, with a few exceptions: bottom agar was 7H9-1.5% agar (per 180 mL: 0.94 g 7H9 [BD], 3 g agar [Sigma]) supplemented with OADS (20 mL OADS per 200 mL 7H9-agar. OADS per 1000 mL: 9.5 g NaCl [Sigma], 50 g Bovine Serum Albumin [Sigma], 20 g D-glucose [Bio Basic Inc.], 268 mg oleic acid [Sigma]) and top agar was 7H9-0.5% agar (same as 7H9-1.5% agar, except with 1 g agar) supplemented with 20 mL OADS. There was no glycerol in the media. M. bovis BCG was diluted in top agar to OD600=0.015 from frozen stock. The 1◦ compounds tested in these studies were ethambutol and rifampicin. Omnitrays were incubated for 3 weeks at 37◦C with 5% CO2. 38 5.3.3 Screening for drug-drug interactions against M. bovis BCG in liquid 7H9 microtiter cultures The MIC of the 1◦ compounds used in this screen (clarithromycin, clo- fazimine, domperidone, ethambutol, ivermectin and rifampicin) were deter- mined in microtiter cultures by the MTT assay. Cultures were grown in 7H9 media supplemented with ADS (20 mL ADS per 200 mL 7H9. ADS per 1000 mL: 9.5 NaCl [Sigma], 50 g Bovine Serum Albumin [Sigma], 20 g D-glucose [Bio Basic Inc.] and 2% glycerol) in 96-well microtiter plates (Costar) at 100 µL total volume, with M. bovis BCG(lux) diluted to OD600= 0.00625. The 1◦ compounds were prepared fresh and applied in two-fold serial dilutions. The cultures were incubated for 7 days at 37◦C with 5% CO2. After incu- bation, 50 µL of 5 mg/mL MTT was added to each well. The plates were incubated overnight at 37◦C, then 50 µL of 10% SDS (w/v) was added and cultures were incubated overnight at 37◦C. Finally, absorbance was read on the Varioskan Flash (made available by Dr. Steven Hallam) at wavelength 560 nm, bandwidth 5 nm and measurement time 100 ms (per well). The values obtained for each well were compared to control wells. A cul- ture was determined to be viable if the absorbance indicated growth greater than 10% (or there was less than 90% growth inhibition) of the non-treated culture. The lowest concentration that can achieve this growth inhibition is called the MIC90. In this experiment, growth was evaluated using the MTT cell viability assay, where viable cultures reduce the yellow tetrazolium MTT to purple formazan. A spectrophotometer quantifies this conversion by mea- suring the absorbance at 560 nm. The MIC90 is the lowest concentration of compound (in a series of dilutions) required to inhibit growth of 90% of the culture. To calculate the absorbance value that delineates the threshold be- tween a growing culture (growth > 10%) from an inhibited culture (growth < 10%), the following formula is used: Threshold = [(Absorbanceviable−Absorbancebkgrd)×0.1)]+Absorbancebkgrd The threshold is calculated by the absorbance of the control cultures (viable, no compounds applied) adjusted for background absorbance (no cells, only media). Once the MICs for each of clarithromycin, clofazimine, domperidone, ethambutol, ivermectin and rifampicin were established, the drug-drug in- teraction screen was performed. For every pairwise combination of 1◦ and 2◦, three concentrations of 1◦ compound (0 µg/mL or no compound, 18 MIC and 14 MIC) were combined with eight concentrations of 2 ◦ compound (four- fold dilutions starting from 100 µM and decreasing to 25 µM, 6.25 µM,..., 39 6.1 nM). This combination scheme is presented diagramatically in Figure 5.1 for a single combination of 1◦ compound + 2◦ compound. The screen was carried out using 96-well microtiter plates. For every test plate a single 2◦ compound was applied in every column in 14 dilutions from A through H, starting at 100 µM. Three plates were set-up for every 2◦ compound in order to combine the 2◦ compound dilution series with the three concentrations of the 1◦ compound at 0 µg/mL 18 MIC and 1 4 MIC. The layout of one set of 96-well plates is depicted in Figure 5.2. This set-up was performed in batches for all combinations of six 1◦ com- pounds and 594 2◦ compounds. After set-up, the plates were incubated for 7 days at 37◦C with 5% CO2. After incubation, 50 µL of 5 mg/mL MTT was added to each well. The plates were incubated overnight at 37◦C, then 50 µL of 10% SDS (w/v) was added and cultures were incubated overnight at 37◦C. Absorbance was read on the Varioskan Flash at wavelength 560 nm, bandwidth 5 nm and measurement time 100 ms (per well). The output of the absorbance readings was one Microsoft Excel® spreadsheet per 96- well plate. To compare the MIC90(2 ◦) of the pairwise combinations for the changing 1◦ compound (0 µg/mL, 18 MIC and 1 4 MIC), a custom software program was developed (described in section 5.3.3). 40 Figure 5.1: A representation of how each pair of 1◦ and 2◦ compound were combined in the 7H9 liquid culture drug- drug interaction screen. The black drops (solid and lined) represent the 1◦ compound (clarithromycin, clofazimine, domperidone, ethambutol, ivermectin or rifampicin) and the blue drops represent the 2◦ compounds of the Sweet Library . The 1◦ compounds were held constant at one of two different concentrations (18 MIC and 1 4 MIC) and the 2◦ compounds were used in a series of eight four-fold dilutions with concentrations ranging from 6.1 nM to 100 µM (size of the drop implies concentration). Theoretically, if the drugs interact in synergy the concentration of the 2◦ compound required to inhibit bacterial growth will drop when comparing its use alone with its use in combination with the 1◦ compound as illustrated by the inhibition/no inhibition labels. 41 Figure 5.2: Schematic representation of the drug-drug interaction screen against M. bovis BCG(lux) in liquid 7H9 microtiter culture. The wells are coloured to represent the presence of compounds and the arrows indicate the direction of the serial dilutions. The control plate has cultures of the three 1◦ concentrations used - 0 µg/mL or no compound (white), 18 MIC (pale orange) and 1 4 MIC (orange). The control plate also has two columns of cells treated with serial 12 dilutions of isoniazid as a control for inhibition (yellow). Finally, the control plate has a positive control for synergy: in column 9 spectinomycin is applied in serial 12 dilutions (green), in column 10 bromperidol is applied in serial 12 dilutions (red) and in columns 11-12 spectinomycin is applied at 1 µg/mL and bromperidol is applied in serial 12 dilutions (green and red). The other three plates represent the replicate plates for a single 1◦ compound combined with twelve 2◦ compounds. Each plate contains either 0 µg/mL, 18 MIC or 1 4 MIC in all wells (white, pale orange and orange respectively). Each column in each plate contains serial 14 dilutions, where each column (1-12) across these three plates contains the same 2◦ compound and dilutions (indicated by the replicate colours). 42 Hit analysis by custom software program The output data from the interactions screen in liquid 7H9 microtiter cul- ture, described in subsection 5.3.3, existed as Microsoft Excel® spreadsheets (one for every 96-well plate). To evaluate this data efficiently and consis- tently, I developed a custom computer program, “Screen Analysis” using the Python— programming language, with teaching and support by Shane Mottishaw. The program took the input (the Excel spreadsheets) and user param- eters (such as what MIC endpoint to use, the default was 90) to process the data and provide the analyses as a single Microsoft Excel® spreadsheet containing a list of every combination of compounds and the MIC of the 2◦ compound alone and when combined with each concentration of each 1◦ compound (18 MIC and 1 4 MIC). The process of analysing the data is shown, simplified, in Figure 5.3. The code is printed in Appendix B. Using Microsoft Excel® tools, the data was mined for 2◦ compounds that partnered synergistically with multiple 1◦ compounds and for combinations that were extremely potent together. 43 Figure 5.3: A simplified representation of the algorithm used to determine which combinations were synergistic. This algorithm works with Microsoft Excel® spreadsheets and the cells described here refer to spreadsheet cells. The input is all the raw data (in Microsoft Excel® spreadsheet form) and the output is a single Microsoft Excel® spreadsheet which lists all combinations categorized as positive/negative for a hit and the MIC values of the 2◦ compounds for every given combination. 44 5.4 Validating the results of the screens using the checkerboard assay The checkerboard assay is a two-dimensional array of combinations of two compounds [23, p. 367]. Each compound is applied in a range of concentra- tions. Every concentration of one compound is combined with every concen- tration of the second compound. Every combination of the two compounds (within their respective ranges) exists within the checkerboard. To evaluate the results of a checkerboard, the fractional inhibitory con- centration index (FICI) is calculated. First, the fractional inhibitory con- centration (FIC) if each compound was determined by: FICdrug A = MIC(drug A when combined with drug B) MIC(drug A alone) The FICI is determined by addition of the two FICs: FICI = FICdrug A + FICdrug B Combinations are categorized, according to their FICI, as synergistic when FICI ≤ 0.5, indifferent when 0.5 < FICI ≤ 4.0 and antagonistic when FICI ≥ 4.0 [23, p.372]. In order to identify combinations that had a mildly synergistic interaction, this work included a category called potentiation when 0.5 < FICI ≤ 1.0. Checkerboard assay in 7H9 liquid culture Checkerboard assays were set up as one pairwise combination per 96-well plate. The 1◦ compound was diluted vertically, from high to low (rows G-A) and the 2◦ compound was diluted horizontally, from high to low (columns 2- 11). Column 12 contained control wells. The set up is depicted in Figure 5.4. Checkerboard assays for mycobacteria were performed in 7H9 + ADS + glycerol, where checkerboard assays for other bacterial species (listed in Table 5.2) were performed in LB media. Bacteria were inoculated at a final OD600=0.00625. All M. tuberculosis and M. bovis BCG strains were incubated for 7 days at 37◦C + 5% CO2 and all other bacterial strains were incubated for 3 days at 37◦C. After incubation, 50 µL of 5 mg/mL MTT was added to each well. The plates were incubated overnight at 37◦C, then 50 µL of 10% SDS (w/v) was added and cultures were incubated overnight at 37◦C. 45 To determine the MIC of each compound, the absorbance of each well was read using the Varioskan Flash (made available by Dr. Steven Hallam) at wavelength 560 nm, bandwidth 5 nm and measurement time 100 ms. The MIC was determined as the MIC90, using the formula described in subsection 5.3.3. These MICs were used to calculate the FICs and FICIs. 5.4.1 Agar lawn checkerboard assay An agar lawn checkerboard for the combination of ethambutol + compound Sweet-1F4 was performed by growing M. smegmatis lawns in individual 60 mm × 15 mm petri dish plates, prepared one-by-one with the various combi- nations of both compounds to make up the checkerboard array. Briefly, M. smegmatis was diluted to OD600=0.0125 directly into molten NE 1.5% agar. The molten agar was divided into smaller aliquots and various combinations of concentrations of ethambutol and compound Sweet-1F4 were added to each aliquot. Then 10 mL aliquots were pipetted into the petri dishes and allowed to set. Plates were incubated at 37◦C for 3 days. The MIC of ethambutol (alone), compound Sweet-1F4 (alone) and etham- butol + compound Sweet-1F4 was determined by eye, based on visual as- sessment of growth of an M. smegmatis lawn. These MIC values were used to calculate the FICs and FICI. 5.4.2 Disc diffusion assays M. smegmatis was diluted to OD600=0.0125 directly into molten NE 1.5% agar with and without sub-inhibitory concentrations of one drug. Ten milliliters of molten agar was pipetted into a 60 mm × 15 mm petri dish plate and allowed to set. A paper disc (BBL—, 6 mm diameter) was satu- rated with 28.5 µL of solution containing the second antibiotic. The disc was placed on the surface of the agar and plates were incubated at 37◦C for 4 days. The halo of inhibition was measured (diameter, in mm). 46 Figure 5.4: Pictorial representation of the liquid culture checkerboard assay. The 1◦ compound is serially diluted two-fold through rows G to A (orange colour). The 2◦ compound is serially diluted two-fold through columns 2-11 (green colour). Column 12 and well H1 are control wells: M. bovis BCG(lux) with no treatment (red lines) and media-only (grey hatching). 47 Alternative liquid microtiter drug-drug interaction assay (poten- tiation assay) The liquid checkerboard assay is labour intensive to set-up. To achieve efficient throughput of validating the drug-drug interactions of hits, an al- ternative assay was developed (potentiation assay). It uses the same culture conditions as the liquid checkerboard, but a different plate set-up. The prin- ciple of the FICI was used as the basis for this higher throughput protocol: the 1◦ compounds were used at 0 µg/mL, 18 MIC and 1 4 MIC and the 2 ◦ compound was used in combination as a dilution series. On a microtiter plate, dilution of the 2◦ compound was performed vertically from A through H, while the concentration of the 1◦ was varied column by column. This set-up is represented in Figure 5.5. Using this method the 1◦ compound is used in the combination at either 1 4 MIC or 1 8 MIC. Therefore, at the point of growth inhibition with the combination of compounds, the 1◦ compound will have an individual FIC of 0.25 or 0.125. The 2◦ compound can be screened for a four-fold reduction of MIC across the different concentrations of the 1◦ compound (which gives an individual FIC of 0.25 or less, and a combined FICI ≤ 0.5, indicating synergy). 48 Figure 5.5: Pictorial representation of an alternative assay (to the checkerboard assay), called a potentiation assay, used for validating the type of interaction of two compounds. The 1◦ compound (red) is used in serial 12 dilutions in column 2, and the theoretical 14 MIC and 1 8 MIC in columns 4-5 and 6-7, respectively. The 2 ◦ compound (green) is used in serial 12 dilutions in column 1. Column 3 contains no compound, only cells (grey hatched). The 1 ◦ and 2’◦ compounds are combined (mixed red and green), holding the 1◦ compound constant at either the theoretical 1 4 MIC or 1 8 MIC (8-9 and 10-11, respectively), while varying the 2 ◦ compound in the 12 dilution, as in column 1. Column 12 is a control with no cells (white). 49 5.5 Antimycobacterial activity of ivermectin Ivermectin and other avermectins, doramectin, moxidectin and selamectin, were tested for activity against mycobacteria. 5.5.1 Determining the MICs of four avermectins The MIC of the avermectins (doramectin, ivermectin, moxidectin and se- lamectin [Sigma, Alpha Diagnostics Intl. Inc., Fluka Analytical and Eu- ropean Pharmacopeia,]) were determined for M. smegmatis, M. avium, M. bovis BCG and M. tuberculosis, as well as the bacterial species listed in Table 5.2, in liquid culture using the MTT assay as described in subsec- tion 2.5.1. The MIC of ivermectin was determined for M. smegmatis in NE agar by inoculating serial dilutions of the compound directly into molten NE agar. M. smegmatis was spread on the surface of the solid agar and incubated at 37◦C for 4 days. Growth of an M. smegmatis lawn was determined visually. 5.5.2 Evidence for bacteriostatic versus bactericidal activity To explore whether ivermectin acted as a bacteriostatic or bactericidal agent M. bovis BCG was treated with 64 µg/mL of ivermectin (8x MIC) and the OD600 was measured over time. Briefly, 100 mL 7H9 was inoculated with M. bovis BCG at OD600 = 0.0625. Cultures were treated with 0 µg/mL ivermectin (control) or 64 µg/mL ivermectin and grown at 37◦C with shak- ing. At Day 0, 4, 10 and 14, the OD600 of the culture was measured using a BioPhotometer (Eppendorf). Additionally, following the final OD600 mea- surement, an aliquot of each culture was inoculated into fresh 7H9 media and incubated at 37◦C with rotation for over 10 generation periods to determine whether cells could grow when ivermectin was removed. 50 6Results 6.1 Screen for drug-drug interactions against M. smegmatis and M. bovis BCG on agar The interactions screening against M. smegmatis was carried out on NE agar as described in subsection 5.3.1. Pairs of compounds were combined by putting the 1◦ compound in the agar with the 2◦ compounds applied by pinning. The outcome of those experiments were plates such as those shown in Figure 6.1. To look for interesting combinations, the halo of inhibition around each 2◦ compound was compared under three conditions: 0 µg/mL 1◦ antibi- otic in the agar, 18 MIC 1 ◦ antibiotic in the agar and 14 MIC 1 ◦ antibiotic in the agar. The 1◦ compounds were clarithromycin, domperidone, etham- butol, moxifloxacin and rifampicin. Potentiation or synergy between two compounds was identified when the halo of inhibition of a single 2◦ com- pound increased from the plate with 0 µg/mL to either or both the 18 MIC and 14 MIC plates for a given 1 ◦ antibiotic. Similarly, the drug-drug interactions screen against M. bovis BCG on agar was carried out on 7H9-agar + OADS using the 1◦ compounds etham- butol and rifampicin, as described in subsection 5.3.2. A hit analysis was performed by comparing halo sizes (as described above). 6.1.1 Validation of hits in a liquid culture assay Lead compound combinations from the 7H9-agar interactions screen (for M. bovis BCG) were further analyzed by Keerat Sidhu in liquid microtiter culture (7H9 + ADS + glycerol) using a potentiation assay as described 51 Figure 6.1: An example of the M.smegmatis NE agar lawn drug-drug inter- action screen. Halos of inhibition appear dark. On all three plates shown here 2◦ compounds of the Sweet Library have been transferred onto the agar by pinning. Plate A, 0 µg/mL 1◦ antibiotic in the agar; B, 18 MIC 1 ◦ an- tibiotic in the agar; and C, 14 MIC 1 ◦ antibiotic in the agar (C). The red arrow indicates a halo of inhibition that increases as the concentration of the 1◦ increases, which is a positive hit for potentiation or synergy. 52 in section 5.4.2. These experiments hold the 1◦ compound concentration constant at pre-established 18 MIC and 1 4 MIC and vary the 2 ◦ compound concentration by serial dilution. This was performed in liquid microtiter culture and relied on the MTT assay to test for viability. The results of this validation showed some combinations that acted as potentiators and some combinations that did not have any confirmed inter- action (see Table 6.1). 53 Table 6.1: Validation results of selected interaction hits from the M. bovis BCG interactions screen in 7H9-agar. This assay was performed in liquid 7H9 microtiter culture holding the concentration of the 1◦ compound con- stant at 18 MIC and 1 4 MIC and combining the 2 ◦ compound in serial dilution fashion. Viability was determined by the MTT assay. The compound in- teraction of each combination was determined to be synergy (FICI ≤ 0.5), potentiation (0.5 < FICI ≤ 1.0) or no interaction (1.0 < FICI < 4.0). 1◦ compound 2◦ compound Synergy, potentiation or no interaction Ethambutol Amikacin no interaction Ethambutol Apramycin no interaction Ethambutol Chetomin no interaction Ethambutol compound Sweet-1C11 potentiation Ethambutol compound Sweet-1F4 potentiation Ethambutol Moxidectin no interaction Ethambutol compound Sweet-2C2 potentiation Ethambutol Isepamicin no interaction Ethambutol Kanamycin no interaction Isoniazid compound Sweet-4A6 potentiation Isoniazid Apramycin no interaction Isoniazid BAY 11-7085 no interaction Isoniazid Beauvericin no interaction Isoniazid Capreomycin no interaction Isoniazid Carbenicillin no interaction Isoniazid compound Sweet-1C1 potentiation Isoniazid Chetomin no interaction Isoniazid Clopidol no interaction Moxifloxacin Amikacin no interaction Moxifloxacin Apramycin no interaction Moxifloxacin Capreomycin no interaction Moxifloxacin Clopidol no interaction Moxifloxacin compound Sweet-4H6 potentiation 54 6.1.2 M. smegmatis NE agar lawn validation of synergy be- tween ethambutol and compound Sweet-1F4 The validation of only 6/23 combinations for potentiation and none for syn- ergy, was unexpected (see Table 6.1). To determine whether culture con- ditions (agar versus liquid) may have contributed to a lack of validation, further experiments were performed on agar. I tested the combination of ethambutol (1◦) and compound Sweet-1F4 (2◦) for synergy, although this combination showed only slight potentiation in the liquid 7H9 experiment. I used this combination of compounds in an NE agar checkerboard and in parallel with a liquid NE checkerboard. The results of the agar checkerboard validated synergy between ethambu- tol and compound Sweet-1F4. For ethambutol, MICalone= 0.5 µg/mL but MICcombination= 0.125 µg/mL. For compound Sweet-1F4, MICalone= 128 µg/mL but MICcombination= 16 µg/mL. This is a classical example of syn- ergy and gives an FICI of 0.375 (see Figure 6.2). Interestingly, the liquid NE checkerboard did not validate synergy between ethambutol and compound Sweet-1F4, although it did suggest potentiation (see Appendix Table A.1) The disc diffusion assay (on NE agar, described in subsection 5.4.2) was performed to better understand whether the method of application of the compounds affected the outcome of a synergy test. Ethambutol was applied by paper disc to an agar lawn with and without sub-inhibitory con- centrations of compound Sweet-1F4. Visual comparison (see Figure 6.3) shows halos of inhibition are much greater in the presence of sub-inhibitory concentrations of compound Sweet-1F4 which indicates potentiation or syn- ergy (this qualitative analysis was inconclusive for synergy). Interestingly, compound Sweet-1F4 alone or in combination with sub-inhibitory concen- trations of ethambutol had no significant effect (see Appendix Figure C.1), which suggests that the method of application of the compounds may play a role in the strength of a compound’s effectiveness and whether synergy occurs under given conditions. 55 Figure 6.2: FIC ethambutol versus FIC compound Sweet-1F4 in an NE agar checkerboard. The optimal FICI (indicated by the red point) is 0.375, positive for synergy. 56 Figure 6.3: Disc diffusion interaction assay for ethambutol and compound Sweet-1F4. M. smegmatis was grown as a lawn on NE agar with and without a sub-inhibitory 2 µg/mL concentration of compound Sweet-1F4 (left and right columns, respectively). Paper discs saturated with ethambutol (total ng as indicated) were applied to the lawn. 57 6.2 Drug-drug interactions screening in liquid 7H9 microtiter culture Another round of screening was performed in liquid 7H9 microtiter culture using M. bovis BCG(lux), as described in subsection 5.3.3. A computer program called “Screen Analysis” was designed to process the raw data of 786 Microsoft Excel® files. This program analysed the entire data set and produced a list of positive leads. Positive leads were those 2◦ compounds that showed an MIC90 decrease of at least four-fold when combined with 1 8 MIC and 14 MIC of one of the 1 ◦ compound (clarithromycin, clofazimine, domperidone, ethambutol, ivermectin, rifampicin), as compared to the MIC of the 2◦ compound used alone. Table 6.2 shows the list of positive leads produced by the computer pro- gram for combinations with the 1◦ compound ivermectin. For the majority of the ivermectin leads, individual checkerboard assays were performed to determine the exact interaction of the compounds. The interaction of any two compounds, such as the data from a checkerboard assay, can be rep- resented graphically in an isobologram. An isobologram plots the FIC of each compound across a range of concentrations for the combinations of two compounds. Figure 6.4 illustrates the types of curves seen in the isobolo- grams. By this analysis, none of the ivermectin combinations were positive for synergy although some combinations showed potentiation (Figure 6.5). In parallel with the ivermectin combinations, I also assayed some strongly interacting rifampicin combinations identified by the computer. Individ- ual checkerboard assays were performed to determine the FICIs, plot the isobolograms and determine the exact interaction of the compound combi- nations. Four out of 16 combinations were positive for synergy and seven out of the remaining 12 combinations showed potentiation. The compounds tested and their interactions are listed in Table 6.3. The isobologram graphs for the rifampicin combinations tested by the checkerboard assay are shown in Figure 6.6. 58 Figure 6.4: Isobolograms are graphs that plot the FICs of two compounds used in a combination. An FIC indicates the relative change in concentration when using a compound alone versus using a compound in combination. In the isobologram, every given combination of concentrations yields an FIC for each of the partner compounds. These FICs are plotted to show an FICI curve. This is an illustration of three categories of curves seen in isobolograms, with the FIC of compound A along the y-axis and the FIC of compound B increasing along the x-axis. U-shaped lines, hugging the x-axis and y-axis, represent synergistic combinations (A); mildly U-shaped lines, where 0.5 < FICI value ≤ 1.0 represent potentiation (B); and linear lines, or slightly humped lines (1.0 < FICI ≤ 4.0), represent no interaction (C). 59 Table 6.2: Positive leads with ivermectin by “Screen Analysis” were tested in a checkerboard assay for synergy. The FICI of each combination was calculated and the interaction was labelled as synergy, potentiation, or no interaction. 2◦ compound FICI Synergy, potentiation, or no interaction 2-Mercaptopyridine 1.0125 no interaction 4-Aminosalicylic acid 1.5 no interaction Acivicin n/a no data compound Sweet-2E7 0.75 potentiation compound Sweet-4B3 0.75 potentiation Aripiprazole 1.5 no interaction compound Sweet-1B6 0.75 potentiation Cannabidiol n/a no data compound Sweet-1C1 0.75 potentiation Danofloxacin 1.0 no interaction Diacylglycerol Kinase Inhibitor II 1.25 no interaction compound Sweet-1F3 0.75 potentiation Ketoconazole 1.0 no interaction compound Sweet-2G5 0.75 potentiation Mezerein 1.5 no interaction compound Sweet-2H11 0.75 potentiation Streptomycin 1.0 no interaction Thioridazine 1.0 no interaction compound Sweet-8D3 0.625 potentiation compound Sweet-2D12 0.75 potentiation 60 61 Figure 6.5: Isobolograms, based on MIC90, of FICs ivermectin and 2 ◦ com- pounds in liquid 7H9 checkerboard assays. The positive control is the known synergistic combination of spectinomycin and bromperidol [34]. The FIC of ivermectin is on the vertical axes and the FIC of the 2◦ compounds (as in- dicated) are on the horizontal axes, except for the control, where the FIC of spectinomycin is on the vertical axis and the FIC of bromperidol is on the horizontal axis. The FICI of each pair is indicated on each graph. The points in red indicate an FICI (FICA + FICB) < 0.5, which defines synergy. The control was positive but no ivermectin combinations were validated for synergy. 62 Table 6.3: Combinations of compounds with rifampicin identified by “Screen Analysis” were tested in a liquid 7H9 checkerboard assay. The FICI of each combination was calculated and the interaction was labelled as synergy, potentiation, or no interaction. 2◦ compound FICI Synergy, potentiation, or no interaction Acivicin 1.0 no interaction compound Sweet-4B3 0.625 potentiation Balofloxacin 1.0 no interaction Bismuth (III) subsalicylate 1.0 no interaction compound Sweet-5H10 0.15625 synergy compound Sweet-5H11 0.14 synergy compound Sweet-2E9 0.1875 synergy Cryptotanshinone 1.25 no interaction Doxycline 1.5 no interaction compound Sweet-4G9 0.625 potentiation compound Sweet-5A6 0.625 potentiation compound Sweet-5B8 0.75 potentiation compound Sweet-7F5 0.75 potentiation compound Sweet-8A2 0.1875 synergy compound Sweet-5G1 0.75 potentiation compound Sweet-2D12 0.75 potentiation 63 64 Figure 6.6: Isobolograms, based on MIC90, of FICs for rifampicin and 2 ◦ compounds in liquid 7H9 checkerboard assays. The FIC of rifampicin is on the vertical axes and the FIC of the 2◦ compounds are on the horizon- tal axes (as indicated). The FICI of each pair is indicated on each graph. The points in red indicate an FICI (FICA + FICB) < 0.5, which defines synergy. Four compounds, compound Sweet-5H10, compound Sweet-5H11, compound Sweet-2E9 and compound Sweet-8A2 showed synergy with ri- fampicin. 65 6.3 Ivermectin is bacteriostatic against some My- cobacterium species Since ivermectin has never, to my knowledge, been described to be effec- tive against any bacteria, I designed experiments to better understand its activity. Firstly, I was interested in determining whether ivermectin was effective against many bacterial species, only mycobacterial species or only M. smegmatis. Secondly, I wanted to know whether ivermectin was uniquely effective, or whether closely related compounds within the avermectin family might also be effective. To explore these questions, I tested the effective- ness of four avermectins: doramectin, ivermectin, moxidectin and selamectin against a variety of bacterial species (gram-positive, gram-negative, my- cobacteria, pathogenic and environmental). All mycobacterial species except M. avium were susceptible to iver- mectin, as shown in Table 6.4. The other avermectins tested also showed varying degrees of effectiveness at inhibiting the Mycobacterium spp. tested. Interestingly, none of the other non-mycobacterial bacterial species tested were susceptible to any avermectin, up to 128 µg/mL (Table 6.4). Another question about ivermectin’s activity was whether it was bacte- riostatic or bacteriocidal against mycobacteria. A time-course measurement of OD600 during treatment with ivermectin was performed for M. bovis BCG and M. smegmatis, and the results were that the OD600 of the cultures re- mained constant over multiple generation times. Furthermore, there was obvious growth in the culture that had been treated with ivermectin, im- plying that ivermectin was bacteriostatic; upon reduction of the ivermectin concentration (to approximately 14 MIC), the M. bovis BCG began to grow. Taken together, these data support that ivermectin is bacteriostatic against M. bovis BCG (shown in Figure 6.7) and M. smegmatis (data not shown). 66 Table 6.4: The MIC of the avermectins against mycobacteria and other bacteria in liquid microtiter culture. All strains or species were tested up to 128 µg/mL MIC90 ( µg/mL): Strain or species Doramectin Ivermectin Moxidectin Selamectin M. smegmatis 128 8 4 4 M. avium >128 >128 >128 16 M. bovis BCG 8 4 4 4 M. tuberculosis H37Rv 8 8 2 1 M. tuberculosis CDC1551 8 8 2 1 M. tuberculosis Erdman 16 16 8 2 E. coli >128 >128 >128 >128 S. lividans >128 >128 >128 >128 R. jostii >128 >128 >128 >128 A. baumannii >128 >128 >128 >128 P. aeruginosa >128 >128 >128 >128 K. rhizophila >128 >128 >128 >128 S. aureus >128 >128 >128 >128 67 Figure 6.7: Ivermectin appears to be bacteriostatic. The graph in (A) shows the OD600 measurements of a 100 mL culture of M. bovis BCG treated with super-inhibitory concentrations of ivermectin (64 µg/mL) or untreated. The treated culture maintained a steady OD600 throughout the experiment. Following that experiment, aliquots from both the treated and untreated culture were inoculated into fresh media ( 150 dilution) and grown for one week, the results are shown in (B). 68 7Discussion 7.1 The agar drug-drug interactions screen iden- tified six potentiating combinations The NE agar drug-drug interactions screen (described in section 6.1) yielded pairs of lead compound combinations that appeared, by qualitative analysis, to interact in potentiation or synergy against M. smegmatis. Twenty-three pairs of compounds were tested in liquid 7H9 microtiter culture using an as- say developed in the Thompson Lab to quantitatively define their interaction (see section 5.4.2). Although none of the 23 pairs demonstrated synergistic activity under these conditions, there were six combinations that demon- strated potentiation (see Table 6.1). Initially, it was surprising that the quantitative assay validated none of the lead combinations as synergistic, but later drug-drug interaction experiments, using the same experimental method, showed that cleaning of the replicator pins between applications was inadequate, resulting in carryover of 2◦ compounds onto subsequent plates which confounded the results. This was likely true in the agar screens in the present study and may have led to some of the apparent false positives identified through the quantitative assay. Another factor that may have affected the results is the differences in media used between the screen and subsequent quantitative assay. In the screen, the media was NE (for M. smegmatis) or 7H9-agar without glycerol, supplemented with OADS (for M. bovis BCG). In the quantitative assay, the media was 7H9 with glycerol, supplemented with ADS. The changing presence of glycerol and oleic acid (the “O” in OADS) may have contributed to different activity levels of the compounds. The presence of glycerol in the media changes the metabolic activity of the cell and can contribute to or 69 detract from the effectiveness of certain compounds [31]. Oleic acid may be used by mycobacteria as a nutrient, and may have a similar influence. In future screens, it is imperative to use the same media formulation from experiment to experiment to maximize the true hit rate. Of the six combinations that showed potentiation using the quantitative assay, ethambutol + compound Sweet-1F4 were tested a second time in NE agar using an agar checkerboard. The agar checkerboard results gave a positive validation for synergy, with an FICI = 0.375. Repeats of both the agar checkerboard and a liquid NE microtiter checkerboard confirmed that the synergy between ethambutol and compound Sweet-1F4 is only seen on solid media. In liquid media, the interaction appears to be potentiation. The results from the disc diffusion assays also imply that the media affects the nature of the interaction of ethambutol + compound Sweet-1F4. In these experiments, one of the partner compounds was present in the agar, and one was applied to the bacterial lawn via a paper disc saturated with the compound. Synergy was qualitatively apparent (visually) when sub- inhibitory concentrations of compound Sweet-1F4 were present in the agar and ethambutol was applied to the M. smegmatis lawn by the paper disc (Figure 6.3). Interestingly, there was no apparent synergy when ethambutol was present in the agar, and compound Sweet-1F4 applied by the paper disc (see Figure C.1). This may be a trivial artefact of chemical diffusion from the paper disc, as compound Sweet-1F4 has very limited effectiveness applied alone via the paper disc (by qualitative analysis). The strikingly different results between liquid media and agar media, and again between applying the compounds in the agar versus via the paper disc, suggest that the interaction between two compounds, and whether the interaction is synergy or potentiation, is not a simple matter of combining drugs. This disparity in results may be an effect of the change in media condi- tions. It is likely that the metabolism or growth patterns of the culture are different in agar and in liquid, which may result in the bacteria being more or less susceptible to the compound(s). It may be that the mechanism of interaction, currently unknown, between ethambutol and compound Sweet- 1F4 is synergistic only under certain physiological conditions and specific compound exposures. Alternatively, the differences between experiments may be due to length of pre-exposure to one of the compounds. In the NE agar screen, the 1◦ com- pound is added to the molten agar and the bacteria are exposed to that 1◦ compound for a period of time (up to 45 minutes) before the 2◦ compound is applied. This lag time is recognized in disc diffusion methods as the “critical 70 time” . The “critical time” is a component of how bacterial growth com- petes with diffusion of the compound through the agar to ultimately define the edge of the zone of inhibition [23, p. 31]. By performing the compound application in this stepwise fashion, the effect (if any) of the 1◦ compound has time to be realized before the 2◦ compound is applied to the agar. This lag between application of the two compounds may allow for time-sensitive changes in the metabolism of the bacteria and these metabolic changes may be different between replicate and alternate experiments. Considering the physiological response systems of mycobacteria to antibiotics (see Part I), it is possible that preliminary exposure to the 1◦ compounds may have affected the way the bacteria responded to the 2◦ compounds. The goal of the agar screens was to find combinations of compounds that worked in synergy in liquid culture: this successfully identified ethambutol + compound Sweet-1F4 as synergistic partners against M. smegmatis (under NE agar media conditions). We wanted to extract more leads from the Sweet Library, containing hundreds of antibiotics, to find more combinations. To more predictably screen for pairs of compounds that interacted in liquid 7H9 growth conditions used for validation, a new screen in liquid 7H9 microtiter culture was performed. 7.2 Four combinations from the liquid 7H9 drug- drug interactions screen were validated for syn- ergy The screen in liquid 7H9 microtiter culture was performed with six 1◦ com- pounds: clarithromycin, clofazimine, domperidone, ethambutol, ivermectin and rifampicin. This generated an enormous body of data to be analyzed. There were a total of 75,000 unique combinations and 786 Microsoft Excel® files generated by this screen. Subsequent data analysis (by the computer program “Screen Analysis”) yielded 266 out of 3564 pairs of compounds that appeared to be synergistic. For follow-up, approximately ten percent of total lead combinations were selected for further experiments. Twenty lead combinations (positive for synergy) with ivermectin and 16 lead combinations with rifampicin were carried forward into checkerboard validation experiments. The advantage of the liquid 7H9 microtiter interaction screen, as op- posed to the agar interactions screen, was that fewer variables were changed between the original screen and the subsequent validation test. The me- dia, culture conditions, volumes and bacteria tested were identical and it 71 was expected that this methodology would be more successful in confirming synergistic pairs of compounds that are effective in liquid 7H9 culture. Four rifampicin combinations were validated: compound Sweet-5H10, compound Sweet-5H11, compound Sweet-2E9 and compound Sweet-8A2. These four combinations represent a validation rate of 11% of all combinations tested, or 25% of rifampicin combinations tested. 7.2.1 Evaluation of the screen as a means of identifying syn- ergistic partners It had been our hope to have greater improvement in lead combination vali- dation. In retrospect, we saw some potential limitations in this methodology that could be addressed in the future to achieve further improvement. Besides outright experimental or analytical error, there were some dif- ferences between the drug-drug interactions screen and the checkerboard validation that may account for the discrepancy of the results. In the screen the M. bovis BCG(lux) was diluted from stationary phase frozen stock di- rectly into media containing the 1◦ compounds. The bacteria were exposed to these sub-inhibitory compound concentrations for at least 30 minutes be- fore addition of the 2◦ compound. As described previously (section 7.1), this lag between compound application may have affected the bacterial re- sponse to the 2◦ antibiotic. In contrast, the checkerboard assay used fresh stationary-phase M. bovis BCG(lux) cells and exposed them to the 1◦ and 2◦ compounds simultaneously. Beyond differences in the experimental set-up, the design of the screen experiment and checkerboard validation may have been in conflict. The interactions screen was performed by holding steady the 1◦ compound at an assumed 18 MIC and 1 4 MIC in combination with serial four-fold serial dilutions of 2◦ compounds. The difficulty with this set-up is twofold: first, the use of four-fold serial dilutions of 2◦ compounds may have provided false positive hits due to general variance in MIC values between different experimental plates. This concept is illustrated in Figure 7.1. Secondly, the assumption that the concentrations used for the 1◦ compounds were 18 MIC and 14 MIC may be incorrect due to general variance in MIC values from experiment to experiment. Future experiments should always include a serial dilution of the 1◦ compound to confirm that the concentrations used are correct. The checkerboard validation used two-fold serial dilutions for both the 1◦ and 2◦ compounds which increased the resolution of the combinatorial inhibitory activity and could differentiate between potentiation and synergy 72 more clearly than four-fold serial dilutions could. Judging by the success of the checkerboard validation experiments, where 11% of combinations tested were confirmed to be synergistic, the false pos- itive rate appears to be close to 90%, however this assumes that the combi- nations selected for checkerboard validation are a fair representative of the leads from the screen and relies on the FICI analysis as the final and cor- rect verdict of drug-drug interactions. The FICI analysis assumes that in a given combination, both compounds’ kinetics are concentration-dependent and linear within the ranges tested [23], however, the activities of many com- pounds are believed to inhibit bacterial growth in a concentration-dependent, exponential way [22]. The kinetics curve of the compounds could preferen- tially favour the FICI classification of combinations as synergistic or as no interaction, depending on the shape of the curve. To demonstrate the variability in compound kinetic curves, Figure 7.2 illustrates linear, linear (out-of-range) and exponential concentration to ac- tivity relationship curves. In this example, an unknown compound is tested in two-fold serial dilutions, using a concentration of 3 and 6 (concentration units). The supposed effectiveness of the compound from one dilution to the next is extrapolated (indicated by the length of the double-headed arrow on each graph). In (A), if there is a linear relationship between concentration and inhibitory activity, the data points at 3 and 6 can be extrapolated to accurately predict the effect of the compound for other concentrations. In (B), the relationship between concentration and inhibitory activity is linear, but not within the concentration range tested, so the data points at 3 and 6 do not extrapolate to accurately predict the effect of the compound for other concentrations. In (C), the relationship between concentration and inhibitory activity is exponential which makes the extrapolation from the data points at 3 and 6 even less accurate in predicting the effect of the compound at other concentrations. 73 Figure 7.1: Concept of how 14 dilutions of 2 ◦ compound may have led to false positive results. The liquid 7H9 microtiter screen used four-fold dilutions of compounds (A), where the checkerboard validation used two-fold dilutions (B). This example illustrates how normal, expected fluctuation of an MIC from plate to plate, in this example from 26 to 20 (arbitrary units), would have been reported as a positive hit in the screen, but a negative validation in the checkerboard assay. In (A), it appears that the MIC is reduced four-fold upon addition of sub-inhibitory concentrations of the 1◦ compound (reported MIC of first plate compared to the reported MIC of the second plate), which appears to be a synergistic interaction, even though the MIC is only fluctuating between 26 and 20. In (B), the resolution is finer and although it appears that the MIC is reduced two-fold upon addition of sub- inhibitory concentrations of the 1◦ compound, this is not considered synergy. 74 Figure 7.2: Graphical comparison of a linear versus non-linear relationship between compound concentration and its inhibitory activity, and the effect of the range of concentrations tested. A linear relationship has increasing inhibitory activity with increasing concentration (A). If the range of concentrations tested contains the concen- trations where a linear relationship plateaus, the relationship between concentration and inhibitory activity may be de-emphasized (B). Non-linear relationships, such as exponential relationships, between concentration and inhibitory activity also exist (C). 75 The following hypothetical example illustrates how the shape of the in- hibition curve is likely to affect the reliability of the screen and therefore the rate of validation: consider 2 drugs whose concentration to inhibitory activity profiles are exponential, such as in Figure 7.2 (C), and whose MIC is 6 concentration units for both compounds when used alone (ie. the cul- ture requires an inhibition power of 6.4 inhibition units to cause no growth, which is achieved by 6 concentration units). If you combine the compounds at half the effective concentration, ie. 3 concentration units, and see an inhibition power of 6.4 inhibition units and no growth, it would appear to be no interaction, but in fact, it would be synergy. It is synergistic because the additive inhibitory activity of 3 concentrations units combined with 3 concentration units would be 0.8 + 0.8 = 1.6 inhibition units, but if the actual combinatorial effectiveness is 6.4 inhibition units then this is synergy because the combination is more effective than the sum of the individual compounds’ effectiveness. Using the FICI analysis, this type of combination would never be identified as synergistic because the assumption of linear concentration to inhibitory activity profiles would calculate the FICI as 1.0 which is interpreted as no interaction. This combination would be negative by checkerboard analysis. Perhaps some of the combinations I tested in a checkerboard assay that were found to be negative for synergy by the FICI did have some interaction between them. To explore the possibility that the checkerboard FICI anal- ysis was biased toward negative results by the compound’s kinetics curves, I plotted the effect of domperidone, ivermectin and rifampicin across the range of concentrations tested, as shown in Figure 7.3. The effect of each compound over a range of two-fold concentrations can be represented by a curve, and the unique shape of this curve may bias the likelihood of identifying synergistic combinations by FICI. Standard FICI analyses consider the effect of using one compound at a concentration four- fold below the MIC90 ( 1 4 MIC) in combination with another compound at 1 4 MIC. In Figure 7.3 the 14 MIC for domperidone, rifampicin and ivermectin is indicated by the green point on each graph. Comparing the three graphs, the effect of the 14 MIC for each compound is different, depending on the curve of the compound. Used alone, the effect of applying 14 MIC dom- peridone reduces growth to 22%, the effect of applying 14 MIC rifampicin reduces growth to 46% and the effect of applying 14 MIC ivermectin reduces growth to 87% (where absorbance represents growth). Therefore, a shallow curve, such as domperidone’s, is the most favourable for identifying syner- gistic combinations by FICI and a steep curve, such as ivermectin’s, is less favourable for identifying synergistic combinations by FICI. This correla- 76 tion was evident in the checkerboard analysis, where 38% of domperidone checkerboards were positive for synergy by FICI (Dr. Ramón-Garćıa, data not shown), 25% of rifampicin checkerboards were positive for synergy by FICI and 0% of ivermectin checkerboards were positive for synergy by FICI. 77 Figure 7.3: Absorbance versus concentration for domperidone (A), rifampicin (B) and ivermectin (C). The points are mean absorbance percentages (compared to absorbance of untreated microtiter culture) over 15 trials. Error bars are standard deviation. The red point on each graph indicates the MIC90 concentration, which is the lowest concentration that inhibits 90% absorbance (or absorbance falls below 10%) where absorbance measures growth. The green points are four-fold below the MIC. . 78 These curves may have contributed to generating false positives in the “Screen Analysis” of the original drug-drug interactions screen by mistaking combinations that are additive as synergistic. Based on their kinetics curves, domperidone and rifampicin would be expected to be less stringent and yield a greater number of leads because the effect of using 14 MIC of each reduces growth to 22% and 46% of absolute growth (growth with no treatment), so additive combinations with weakly inhibiting 2◦ compounds are more likely to be identified as leads. To help address this issue the “Screen Analysis” computer program identified partner 2◦ compounds that reduced the relative growth (at 14 MIC or 1 8 MIC) of the microtiter culture by an additional 90%. So under these circumstances, the 2◦ partner compound must reduce growth to 2.2% absolute growth for domperidone or 4.6% absolute growth for rifampicin. This means, in absolute terms, a 2◦ compound that, at sub-MIC levels, further reduces growth by 19.8% would be a lead with domperidone (22.0%−19.8% = 2.2%). Similarly, a 2◦ partner compound that, at sub-MIC levels, reduces growth by 41.4%, would be a lead with rifampicin (46.0%− 41.4% = 4.6%). Additionally, the accuracy of absorbance readings at such low values is compromised, generating significant variability in the data and resulting in false positive hits. In contrast, ivermectin is expected to be more stringent and yield fewer leads because the effect of using 14 MIC of ivermectin only reduces growth to 87% before the addition of a 2◦ compound. Here, for a 2◦ compound to meet the criteria as a lead, at sub-MIC levels it must reduce growth by 78.3% (87.0%− 78.3% = 8.7%). This prediction proves true in what we see from the “Screen Analysis” results: domperidone has 41 leads, rifampicin has 70 leads and ivermectin has 20 leads. To precisely determine the interaction between two compounds, we would first need to establish the concentration to inhibitory activity profile across a linear range of concentrations for each compound alone and then establish the effectiveness of the combination across a linear range of concentrations for both compounds. Then we would need to set a threshold for how much each compound must be reduced (in combination, compared to sole applica- tion) to be considered synergistic. These steps are necessary because every compound has a unique kinetic curve, as illustrated by the comparison of domperidone, rifampicin and ivermectin and additionally with the curves of some 2◦ compounds shown in Figure 7.4. Unfortunately, while precise, this type of experiment is not particularly efficient or practical for testing a large number of combinations. Fundamentally, the FICI analysis has long been established as a useful tool [14], and it is the current standard for establishing synergy of compound 79 Figure 7.4: Activity kinetics curves of some 2◦ compounds across a range of concentrations. The shape of the curve is unique to each compound. The MIC90 of each compound is identified as the first point that falls beneath the 10% line. combinations. However, the FICI is also a limited tool because it relies on the evaluation of the MIC, which is variable, and because it assumes a linear relationship between concentration of a compound and its inhibitory activ- ity. Overall, the FICI is an efficient, standardized method for confirming synergistic combinations but it has limitations which may discount some interesting combinations. Many of these interesting combinations are the combinations we iden- tified to interact by potentiation. The difference between potentiation and synergy (by FICI) is that in potentiation one of the compounds is used at 1 2 MIC, rather than 1 4 MIC (FICI = 0.75). While none of the ivermectin combinations validated as synergy by checkerboard and FICI, there were eight compounds that showed potentiation with ivermectin (see Table 6.2). Considering the steep slope of the kinetics activity curve of ivermectin be- tween growth and no growth (Figure 7.3), any 2◦ compound that can be used at sub-inhibitory concentrations to make ivermectin more effective at 1 2 MIC is an interesting partner. 80 7.3 Data from screens can be mined for new anti- mycobacterials The results of the drug-drug interactions screening experiments have so far identified four combinations that are effective in synergy against mycobac- teria in liquid 7H9 microtiter culture and one combination that is effective against mycobacteria in an agar lawn. It is exciting to not only have a short-list of combinations that may be effective against M. tuberculosis, but to have a large data set that indicates single compounds that are effective against mycobacteria, independent of their effectiveness within combina- tions. It is especially interesting to find families of compounds that show inhibitory activity against M. bovis BCG(lux), such as the avermectins (dis- cussed below). 7.3.1 The avermectins may be a future TB treatment The avermectins, including ivermectin, while known to be potent anthel- mintics, have traditionally been regarded as being ineffective against bac- teria [5, 12, 16]. This was true for all bacterial species I tested (E. coli, S. lividans, R. jostii, A. baumannii, P. aeuruginosa, K. rhizophila and S. aureus) with the exception of mycobacteria. The apparent specificity of ivermectin for mycobacteria implies that its target is unique to this taxon of bacteria. The remarkable complexity and uniqueness of the mycobacterial cell wall, not common to other bacteria that are resistant to avermectins (in- cluding S. lividans and R. jostii RHA1, both related actinobacteria), points towards this system as a potential target [11]. There is very little evidence as to what structure or pathway the aver- mectins may be targeting in mycobacteria. In invertebrate nematodes, iver- mectin specifically binds to glutamate-gated chloride channels present in nerve and muscle cells, causing paralysis of movement, paralysis of the phar- ynx and reduced ability to reproduce [43]. There are no homologues of these channels present in the mycobacterial cell wall, according to BLAST search. In considering ivermectin as a potential future treatment for TB, cur- rent understanding of the drug is promising. Ivermectin is a lipophilic com- pound that is distributed to most parts of the body [7, 18]. We expect ivermectin would successfully reach the lungs of humans because of its ef- fectiveness against the subcutaneous nematode Onchocerca volvulus larvae in humans [40] and because in other mammals such as foals, orally admin- istered ivermectin has been effective in treating Strongylus vulgaris larval infection of the lungs [41]. Ivermectin does not bind human ligand-gated 81 chloride channels in vivo, possibly due to the blood-brain barrier and non- specific pumps that prevent ivermectin from reaching the central nervous system [18,43]. These data suggest that ivermectin is likely to reach the site of TB infection in the lungs and it is unlikely to have toxic effects on the human nervous system. To my knowledge this is the first report demonstrating the antimycobac- terial activity of avermectins. The data provided here recommend future investigation of ivermectin for TB treatment. Its apparent specificity for mycobacteria and its safety profile in humans make it an attractive thera- peutic option. 7.4 Future work To carry this work forward, one direction to explore is the cellular target and the specific mode of action of the avermectins which is seemingly unique to mycobacteria. This is likely to point to or elucidate unique essential functions of mycobacteria as compared to other bacteria. I propose that an M. smegmatis mutant library would begin to address these questions and could be a first-step in furthering this work. To continue exploring the potential therapeutic usefulness of ivermectin, future studies should include ex vivo infection studies and in vivo infection studies. It is imperative to determine whether ivermectin is, as hypothesized, able to reach the site of M. tuberculosis infection and inhibit bacterial growth under those conditions and whether the dose is tolerated by mammalian systems and host cells. It is also of great interest to consider ivermectin as a potential MDR-TB treatment. The first steps toward exploring this possibility have already been taken: experiments in collaboration with Dr. Bill Jacobs’ Lab are underway to test whether ivermectin is effective against strains of MDR- and XDR-TB, in vitro. In another vein of future work, the large data bank from the screens can be mined for other interesting compounds or classes of compounds that may show promise for therapeutic use in TB treatment. With the information presented here and the rich data bank that these studies have provided, there are many options to carry forward. An inves- tigation into the target and interaction of the synergistic compounds may be a study that is similar to the pure science, finely-focused approach that was adopted for the work described in Part I. Further data mining of the screen results may be a study that is similar to the clinically-relevant, widely- 82 focused approach that was adopted to do the drug-drug interaction screens in the first place, as described in Part II. Ultimately, any work that launches from the information presented here has a great capacity to further our understanding of antibiotic susceptibility and intrinsic antibiotic resistance in Mycobacterium. It is our aim that this information will lead to a timely introduction of new TB treatments for the world. 83 References [1] Handbook of anti-tuberculosis agents. Tuberculosis (Edinburgh, Scot- land), 88(2), Mar 2008. [2] C. P. Adams and V. V. Brantner. Estimating the cost of new drug development: is it really 802 million dollars? Health affairs (Project Hope), 25(2):420–428, Mar-Apr 2006. [3] J. Beaucher, S. Rodrigue, P. E. Jacques, I. Smith, R. Brzezinski, and L. Gaudreau. Novel mycobacterium tuberculosis anti-sigma factor an- tagonists control sigmaf activity by distinct mechanisms. Molecular microbiology, 45(6):1527–1540, Sep 2002. [4] Amersham Biosciences. GST Gene Fusion System Handbook. Number 18-1157-58. AA edition, 2002. [5] R. W. Burg, B. M. Miller, E. E. Baker, J. Birnbaum, S. A. Currie, R. Hartman, Y. L. Kong, R. L. Monaghan, G. Olson, I. Putter, J. B. Tunac, H. Wallick, E. O. Stapley, R. Oiwa, and S. Omura. Avermectins, new family of potent anthelmintic agents: producing organism and fer- mentation. Antimicrobial Agents and Chemotherapy, 15(3):361–367, Mar 1979. [6] J. Burian, S. Ramón-Garćıa, G. Sweet, A. Gomez-Velasco, Y. Av- Gay, and C. J. Thompson. The mycobacterial transcriptional regulator whib7 links redox homeostasis and intrinsic antibiotic resistance. The Journal of biological chemistry, Nov 8 2011. [7] A. Gonzalez Canga, A. M. Sahagun Prieto, M. J. Diez Liebana, N. Fer- nandez Martinez, M. Sierra Vega, and J. J. Garcia Vieitez. The phar- macokinetics and interactions of ivermectin in humans–a mini-review. The AAPS journal, 10(1):42–46, 2008. 84 [8] P. Chen, R. E. Ruiz, Q. Li, R. F. Silver, and W. R. Bishai. Con- struction and characterization of a mycobacterium tuberculosis mutant lacking the alternate sigma factor gene, sigF. Infection and immunity, 68(10):5575–5580, Oct 2000. [9] C. R. Chong and D. J. Sullivan Jr. New uses for old drugs. Nature, 448(7154):645–646, Aug 9 2007. [10] V. M. D’Costa, K. M. McGrann, D. W. Hughes, and G. D. Wright. Sampling the antibiotic resistome. Science (New York, N.Y.), 311(5759):374–377, Jan 20 2006. [11] Chatterjee Delphi. The mycobacterial cell wall: structure, biosynthesis and sites of drug action. Current opinion in chemical biology, 1(4):579– 588, 12 1997. [12] T. G. Geary. Ivermectin 20 years on: maturation of a wonder drug. Trends in parasitology, 21(11):530–532, Nov 2005. [13] R. Goude and T. Parish. The genetics of cell wall biosynthesis in my- cobacterium tuberculosis. Future microbiology, 3(3):299–313, Jun 2008. [14] M. J. Hall, R. F. Middleton, and D. Westmacott. The fractional in- hibitory concentration (fic) index as a measure of synergy. The Journal of antimicrobial chemotherapy, 11(5):427–433, May 1983. [15] R. C. Hartkoorn, C. Sala, S. J. Magnet, J. M. Chen, F. Pojer, and S. T. Cole. Sigma factor f does not prevent rifampin inhibition of rna polymerase or cause rifampin tolerance in mycobacterium tuberculosis. Journal of Bacteriology, Aug 20 2010. [16] I. K. Hotson. The avermectins: A new family of antiparasitic agents. Journal of the South African Veterinary Association, 53(2):87–90, Jun 1982. [17] J. R. Huth, C. A. Bewley, B. M. Jackson, A. G. Hinnebusch, G. M. Clore, and A. M. Gronenborn. Design of an expression system for detecting folded protein domains and mapping macromolecular inter- actions by nmr. Protein science : a publication of the Protein Society, 6(11):2359–2364, Nov 1997. [18] Merck & Co. Inc. Stromectol (ivermectin), 2007. 85 [19] Mandeep Jassal and William R. Bishai. Extensively drug-resistant tu- berculosis. The Lancet Infectious Diseases, 9(1):19–30, 1 2009. [20] B. Kartmann, S. Stenger, and M. Niederweis. Porins in the cell wall of mycobacterium tuberculosis. Journal of Bacteriology, 181(20):6543– 6546, Oct 1999. [21] M. A. Kohanski, D. J. Dwyer, B. Hayete, C. A. Lawrence, and J. J. Collins. A common mechanism of cellular death induced by bactericidal antibiotics. Cell, 130(5):797–810, Sep 7 2007. [22] B. R. Levin and K. I. Udekwu. Population dynamics of antibiotic treatment: a mathematical model and hypotheses for time-kill and continuous-culture experiments. Antimicrobial Agents and Chemother- apy, 54(8):3414–3426, Aug 2010. [23] Victor Lorian ed. Antibiotics in laboratory medicine - 5th edition. Lip- pincott, Williams & Wilkins, Philadelphia, PA, 2005. [24] S. S. Malik, A. Luthra, S. K. Srivastava, and R. Ramachandran. My- cobacterium tuberculosis usfx (rv3287c) exhibits novel nucleotide bind- ing and hydrolysis properties. Biochemical and biophysical research communications, 375(3):465–470, Oct 24 2008. [25] T. M. Michele, C. Ko, and W. R. Bishai. Exposure to antibiotics in- duces expression of the mycobacterium tuberculosis sigF gene: implica- tions for chemotherapy against mycobacterial persistors. Antimicrobial Agents and Chemotherapy, 43(2):218–225, Feb 1999. [26] R. P. Morris, L. Nguyen, J. Gatfield, K. Visconti, K. Nguyen, D. Schnappinger, S. Ehrt, Y. Liu, L. Heifets, J. Pieters, G. Schoolnik, and C. J. Thompson. Ancestral antibiotic resistance in mycobacterium tuberculosis. Proceedings of the National Academy of Sciences of the United States of America, 102(34):12200–12205, Aug 23 2005. [27] L. Nguyen, S. Chinnapapagari, and C. J. Thompson. Fbpa-dependent biosynthesis of trehalose dimycolate is required for the intrinsic mul- tidrug resistance, cell wall structure, and colonial morphology of my- cobacterium smegmatis. Journal of Bacteriology, 187(19):6603–6611, Oct 2005. [28] L. Nguyen and C. J. Thompson. Foundations of antibiotic resistance in bacterial physiology: the mycobacterial paradigm. Trends in micro- biology, 14(7):304–312, Jul 2006. 86 [29] M. Niederweis, O. Danilchanka, J. Huff, C. Hoffmann, and H. Engel- hardt. Mycobacterial outer membranes: in search of proteins. Trends in microbiology, 18(3):109–116, Mar 2010. [30] World Health Organization. Fact sheet no 104: Tuberculosis, 2010. [31] K. Pethe, P. C. Sequeira, S. Agarwalla, K. Rhee, K. Kuhen, W. Y. Phong, V. Patel, D. Beer, J. R. Walker, J. Duraiswamy, J. Jiricek, T. H. Keller, A. Chatterjee, M. P. Tan, M. Ujjini, S. P. Rao, L. Cama- cho, P. Bifani, P. A. Mak, I. Ma, S. W. Barnes, Z. Chen, D. Plouffe, P. Thayalan, S. H. Ng, M. Au, B. H. Lee, B. H. Tan, S. Ravindran, M. Nanjundappa, X. Lin, A. Goh, S. B. Lakshminarayana, C. Shoen, M. Cynamon, B. Kreiswirth, V. Dartois, E. C. Peters, R. Glynne, S. Brenner, and T. Dick. A chemical genetic screen in mycobacterium tuberculosis identifies carbon-source-dependent growth inhibitors de- void of in vivo efficacy. Nature communications, 1:57, Aug 24 2010. [32] R. Provvedi, D. Kocincova, V. Dona, D. Euphrasie, M. Daffe, G. Eti- enne, R. Manganelli, and J. M. Reyrat. SigF controls carotenoid pig- ment production and affects transformation efficiency and hydrogen peroxide sensitivity in mycobacterium smegmatis. Journal of Bacteri- ology, 190(23):7859–7863, Dec 2008. [33] Qiagen. The QIAexpressionist. Fifth edition, 2003. [34] S. Ramón-Garćıa, C. Ng, H. Anderson, J. D. Chao, X. Zheng, T. Pfeifer, Y. Av-Gay, M. Roberge, and C. J. Thompson. Synergistic drug combi- nations for tuberculosis therapy identified by a novel high-throughput screen. Antimicrobial Agents and Chemotherapy, 55(8):3861–3869, Aug 2011. [35] S. Rodrigue, R. Provvedi, P. E. Jacques, L. Gaudreau, and R. Man- ganelli. The sigma factors of mycobacterium tuberculosis. FEMS mi- crobiology reviews, 30(6):926–941, Nov 2006. [36] J. Rybniker, A. Nowag, E. van Gumpel, N. Nissen, N. Robinson, G. Plum, and P. Hartmann. Insights into the function of the whib- like protein of mycobacteriophage tm4–a transcriptional inhibitor of whib2. Molecular microbiology, 77(3):642–657, Aug 2010. [37] H. Schagger. Tricine-sds-page. Nature protocols, 1(1):16–22, 2006. 87 [38] A. J. Steyn, D. M. Collins, M. K. Hondalus, W. R. Jacobs Jr, R. P. Kawakami, and B. R. Bloom. Mycobacterium tuberculosis whib3 in- teracts with rpov to affect host survival but is dispensable for in vivo growth. Proceedings of the National Academy of Sciences of the United States of America, 99(5):3147–3152, Mar 5 2002. [39] C. K. Stover, V. F. de la Cruz, T. R. Fuerst, J. E. Burlein, L. A. Benson, L. T. Bennett, G. P. Bansal, J. F. Young, M. H. Lee, and G. F. Hatfull. New use of BCG for recombinant vaccines. Nature, 351(6326):456–460, Jun 6 1991. [40] M. J. Taylor, A. Hoerauf, and M. Bockarie. Lymphatic filariasis and onchocerciasis. Lancet, 376(9747):1175–1185, Oct 2 2010. [41] M. A. Turk and T. R. Klei. Effect of ivermectin treatment on eosinophilic pneumonia and other extravascular lesions of late strongy- lus vulgaris larval migration in foals. Veterinary pathology, 21(1):87–92, Jan 1984. [42] A. A. Velayati, M. R. Masjedi, P. Farnia, P. Tabarsi, J. Ghanavi, A. H. Ziazarifi, and S. E. Hoffner. Emergence of new forms of totally drug- resistant tuberculosis bacilli: super extensively drug-resistant tubercu- losis or totally drug-resistant strains in iran. Chest, 136(2):420–425, Aug 2009. [43] A. J. Wolstenholme and A. T. Rogers. Glutamate-gated chloride channels and the mode of action of the avermectin/milbemycin an- thelmintics. Parasitology, 131 Suppl:S85–95, 2005. [44] M. A. Zaunbrecher, R. D. Sikes Jr, B. Metchock, T. M. Shinnick, and J. E. Posey. Overexpression of the chromosomally encoded aminogly- coside acetyltransferase eis confers kanamycin resistance in mycobac- terium tuberculosis. Proceedings of the National Academy of Sciences of the United States of America, 106(47):20004–20009, Nov 24 2009. [45] K. Honer zu Bentrup and D. G. Russell. Mycobacterial persis- tence: adaptation to a changing environment. Trends in microbiology, 9(12):597–605, Dec 2001. 88 Appendices 89 Appendix A Ethambutol and compound Sweet-1F4 liquid NE checkerboard results Ethambutol and compound Sweet-1F4 were tested for synergy using a liquid NE checkerboard in microtiter culture (to compare to an NE agar checker- board). Viability was determined using the MTT assay. The absorbance values of the checkerboard assay are shown in Table A.1. 90 Table A.1: Photometric values for NE liquid checkerboard of ethambutol (serial 12 dilutions high to low from rows G through H) and compound Sweet-1F4 (serial 12 dilutions high to low from columns 2 through 11). The highest concentration of ethambutol was 2 µg/mL and the highest concentration of compound Sweet-1F4 was 32 µg/mL. The MICethambutol = D1 = 0.25 µg/mL and the MICcompoundSweet−1F4 = H5 = 4 µg/mL. Strictly from growth versus no-growth, the combination of ethambutol + compound Sweet-1F4 is no more effective than either compound alone, however, the photometric value at C6 is lower than either ethambutol or compound Sweet-1F4 alone, which suggests the compounds may interact with slight potentiation. 1 2 3 4 5 6 7 8 9 10 11 12 A 3.636 0.072 0.070 0.072 0.081 2.091 2.444 1.824 1.955 3.220 2.229 2.691 B 3.082 0.065 0.063 0.061 0.067 1.622 2.183 3.072 2.929 2.669 2.646 2.194 C 1.565 0.063 0.062 0.060 0.062 1.006 1.292 1.359 1.012 1.620 1.770 1.406 D 0.264 0.064 0.059 0.056 0.058 0.058 0.079 0.116 0.119 0.146 0.158 0.084 E 0.094 0.062 0.060 0.059 0.058 0.057 0.062 0.065 0.070 0.067 0.074 0.075 F 0.080 0.066 0.059 0.058 0.059 0.065 0.061 0.062 0.065 0.068 0.072 0.075 G 0.078 0.067 0.066 0.060 0.061 0.060 0.067 0.078 0.067 0.082 0.075 0.077 H 2.609 0.074 0.076 0.083 0.208 3.155 2.640 3.129 2.592 3.234 2.498 3.753 91 Appendix B “Screen Analysis” program code This is the raw code for the computer program “Screen Analysis”, which was developed and used to analyse the interactions screen results. \font import xlrd import glob import codecs #compound 1 is the primary compound, compound 2 is the secondary compound, from the Sweet Library def read_control(sheet, startrow, endrow, startcolumn, endcolumn): values = [] #list of values to be used later for column in range(startcolumn, endcolumn+1): for row in range (startrow, endrow+1): values.append(sheet.cell(row,column).value) return values def read_assay(sheet, startrow, endrow, startcolumn, endcolumn, start_concn2, diln_factor2, ID_prefix): values = {} #table of tables of values. ID keys, indexed to tables of concn2 keys, indexed to photometric values current_concn2=start_concn2 for row in range(startrow, endrow+1): assay_col_counter = 1 92 for column in range(startcolumn, endcolumn+1): ID_current = ID_prefix + str(assay_col_counter) assay_col_counter = assay_col_counter + 1 if ID_current not in values: values[ID_current] = {} values[ID_current][current_concn2] = sheet.cell(row,column).value current_concn2 = current_concn2 * (1/float(diln_factor2)) return values def MICvalue(alive, dead, MICxx): a_average = sum(alive, 0.0)/len(alive) d_average = sum(dead, 0.0)/len(dead) return (a_average - d_average) * (1-(float(MICxx)/100.0)) + d_average def make_da_allkill_tables(MICvalue, assay_directory): datable = {} allkill_table = {} cmpd1_diln_list = assay_directory.keys() cmpd1_diln_list.sort() for cmpd1_diln in cmpd1_diln_list: table = assay_directory[cmpd1_diln] for ID_current in table.keys(): concn2_alive = 400.0 #400 is a placeholder for an MIC "off the charts", or higher than the highest concentration2 tested allkill_flag = True current_concn2_list = table[ID_current].keys() current_concn2_list.sort(reverse=True) for current_concn2 in current_concn2_list: if table[ID_current][current_concn2] > MICvalue: allkill_flag = False break concn2_alive = current_concn2 if allkill_flag: if ID_current not in allkill_table: allkill_table[ID_current] = [] allkill_table[ID_current].append(cmpd1_diln) if ID_current not in datable: datable[ID_current] = [] datable[ID_current].append(concn2_alive) return (datable, allkill_table) 93 def find_synergyhits(datable): hits_table = {} for ID_current in datable.keys(): concns2 = datable[ID_current] ctrl_concn2 = concns2[0] hit = False for current_concn2 in concns2[1:]: hit = hit or (ctrl_concn2 > current_concn2) if hit: hits_table[ID_current] = concns2 return hits_table def check_synergy(datable, screening_table, cmpd1): #screening table has all pairs listed for ID_current in datable.keys(): if ID_current not in screening_table: screening_table[ID_current] = {} concns2 = datable[ID_current] prev_concn2 = concns2[0] all_equal_flag = True hit_flag = True for current_concn2 in concns2[1:]: if current_concn2 < prev_concn2: all_equal_flag = False if current_concn2 >= prev_concn2: # Both the 1/8 and 1/4 value must show synergy all_equal_flag = False hit_flag = False screening_table[ID_current][cmpd1] = (False, concns2) break if all_equal_flag: screening_table[ID_current][cmpd1] = (False, concns2) elif hit_flag: screening_table[ID_current][cmpd1] = (True, concns2) def read_librarydb(sheet): #link ID to cmpd2 name and properties librarydb = {} column = 0 94 row = 0 begin = False cellvalue = sheet.cell(row, column).value while cellvalue != ’END’: if cellvalue == ’BEGIN’: begin = True if begin == True: if cellvalue != ’’: cmpd2_name = sheet.cell(row, column+1).value cmpd2_class = sheet.cell(row, column+13).value cmpd2_target = sheet.cell(row, column+14).value cmpd2_mechanism = sheet.cell(row, column+15).value librarydb[cellvalue] = (cmpd2_name, cmpd2_class, cmpd2_target, cmpd2_mechanism) row = row + 1 cellvalue = sheet.cell(row, column).value return librarydb batch_list = [’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 1’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 2’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 3’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 4’,’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 5’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 6’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 7’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ Library 8’, ’Y:\Leah\Synergy Screening\BCG luc liquid assay\Python Program\ IVR with Library’] cmpd1_list = [’ETH’, ’RIF’, ’CLA’, ’CLOF’, ’DPD’, ’IVR’] library_list = [’1’, ’2’, ’3’, ’4’, ’5’, ’6’, ’7’, ’8’] plate_list = [’A’, ’B’, ’C’, ’D’, ’E’, ’F’, ’G’, ’H’] concn1_list = [’18’, ’14’] #represents 1/8 and 1/4 sheet_to_read = 1 #excel sheet where data is MIC_percent_list = [90] hits_output_fname = ’hits.csv’ effective_cmpd_output_fname = ’effective.csv’ screening_output_fname = ’screen.csv’ librarydb_fname = ’Library Functions.xls’ 95 alldata_fname = ’All Combinations.xls’ hits_output_file = codecs.open(hits_output_fname,’w’,’utf_16’) effective_cmpd_output_file = codecs.open(effective_cmpd_output_fname,’w’,’utf_16’) screening_output_file = codecs.open(screening_output_fname, ’w’,’utf_16’) #hits output file headers: MICxx, CMPD1, ID, 0.0, 0.125, 0.25 where 0.125 is 1/8 and 0.25 is 1/4 hits_output_file.write(’MICxx\tCMPD1\tID\t0.0\t0.125\t0.25’) hits_output_file.write(’\n’) #effective cmpd output file headers: MICxx, ID, MIC(cmpd2) effective_cmpd_output_file.write(’MICxx\tID\tMIC(CMPD2)’) effective_cmpd_output_file.write(’\n’) #screening output file headers: ID, NAME, COUNT, CMPD1, HIT?, 0.0, 0.125, 0.25 where 0.125 is 1/8 and 0.25 is 1/4 screening_output_file.write(’ID\tNAME\tCOUNT’) for cmpd1 in cmpd1_list: screening_output_file.write(’\tCMPD1\tHIT\t0.0\t0.125\t0.25’) screening_output_file.write(’\n’) wb_librarydb = xlrd.open_workbook(librarydb_fname) sh_librarydb = wb_librarydb.sheet_by_index(0) librarydb = read_librarydb(sh_librarydb) screening_table = {} hits_histogram = {} for folder in batch_list: print folder for MIC_percent in MIC_percent_list: for cmpd1 in cmpd1_list: cmpd1_fpat = folder + ’\\’ + ’*’ + cmpd1 + ’ controls*’ cmpd1_fnames = glob.glob(cmpd1_fpat) if len(cmpd1_fnames) > 1 : print ’Error opening control file for compound ’ + cmpd1 exit() elif len(cmpd1_fnames) == 0: 96 continue wb_cmpd1_control = xlrd.open_workbook(cmpd1_fnames[0]) sh_cmpd1_control = wb_cmpd1_control.sheet_by_index(sheet_to_read) alive = read_control(sh_cmpd1_control, 16, 23, 1, 2) dead = read_control(sh_cmpd1_control, 16, 17, 7, 8) MIC_cmpd1 = MICvalue(alive, dead, MIC_percent) for library in library_list: for plate in plate_list: # get library/plate control check_plate_cmpd1_pat = folder + ’\\’ + library + plate + ’ ’ + cmpd1 + ’*’ check_plate_cmpd1_list = glob.glob(check_plate_cmpd1_pat) if len(check_plate_cmpd1_list) == 0: continue plate_control_path = folder + ’\\’ + library + plate + ’ control.xls’ wb_plate_control = xlrd.open_workbook(plate_control_path) sh_plate_control = wb_plate_control.sheet_by_index(sheet_to_read) # add control to assay_list assay_list = [] assay_list.append(read_assay(sh_plate_control, 16, 23, 1, 12, 100, 4, library + plate)) for concn1 in concn1_list: # build assay_list plate_cmpd1_concn1_path = folder + ’\\’ + library + plate + ’ ’ + cmpd1 + ’ ’ + concn1 + ’.xls’ wb_plate_cmpd1_concn1 = xlrd.open_workbook (plate_cmpd1_concn1_path) sh_plate_cmpd1_concn1 = wb_plate_cmpd1_concn1. sheet_by_index(sheet_to_read) assay_list.append(read_assay(sh_plate_cmpd1_concn1, 97 16, 23, 1, 12, 100, 4, library + plate)) assay_directory = {0:assay_list[0]} start_concn1 = 0.125 current_concn1 = start_concn1 diln_factor1 = 2.0 for assay in assay_list[1:]: assay_directory[current_concn1] = assay current_concn1 = current_concn1 * diln_factor1 (datable, allkill_table) = make_da_allkill_tables(MIC_cmpd1, assay_directory) check_synergy(datable, screening_table, cmpd1) hits_table = find_synergyhits(datable) for hit_id in hits_table.keys(): #column headings: #’MICxx,CMPD1,ID,CMPD1=0,CMPD1=1/8,CMPD1=1/4\n’ cmpd2_concn_list = hits_table[hit_id] hits_output_file.write(str(MIC_percent)+’\t’+ cmpd1+’\t’+’[’+hit_id+’]’) for curr_cmpd2_concn in cmpd2_concn_list: hits_output_file.write(’\t’+str(curr_cmpd2_concn)) hits_output_file.write(’\n’) # column headings: #’MICxx,ID,CMPD1=0\n’ effective_cmpd_output_file.write(str(MIC_percent)+ ’\t’+’[’+hit_id+’]’+’\t’+str(cmpd2_concn_list[0])+’\n’) # count the times a cmpd2 is considered a hit if hit_id not in hits_histogram.keys(): hits_histogram[hit_id] = [] hits_histogram[hit_id].append(cmpd1) hits_output_file.close() effective_cmpd_output_file.close() 98 for cmpd2 in screening_table.keys(): cmpd2_hit_count = 0 for cmpd1 in screening_table[cmpd2].keys(): (hit, cmpd2_concns) = screening_table[cmpd2][cmpd1] if hit: cmpd2_hit_count = cmpd2_hit_count + 1 hits_histogram[cmpd2] = cmpd2_hit_count for cmpd2 in screening_table.keys(): cmpd2_name = ’’ if cmpd2 not in librarydb: print ’no library information for cmpd2 ID: ’ + cmpd2 else: #screening output file headers: ID, NAME, COUNT, CMPD1, HIT?, 0.0, 0.125, 0.25 (name, cmpd2_class, cmpd2_target, cmpd2_mechanism) = librarydb[cmpd2] cmpd2_name = name screening_output_file.write(’[’+cmpd2+’]’ + ’\t’ + cmpd2_name + ’\t’ + str(hits_histogram[cmpd2])) for cmpd1 in screening_table[cmpd2].keys(): (hit, cmpd2_concns) = screening_table[cmpd2][cmpd1] hit_string = ’No’ if hit: hit_string = ’Yes’ screening_output_file.write(’\t’ + cmpd1 + ’\t’ + hit_string + ’\t’ + str(cmpd2_concns[0]) + ’\t’ + str(cmpd2_concns[1]) + ’\t’ + str(cmpd2_concns[2])) screening_output_file.write(’\n’) screening_output_file.close() print ’Done!’ 99 Appendix C Disc diffusion interaction assay Figure C.1: Disc diffusion interaction assay of ethambutol and compound Sweet-1F4. 100

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072732/manifest

Comment

Related Items