1 Histoplasma Capsulatum Infection

Home , Abafungin, Aquamarina, Aspergillus aculeatinus, Aspergillus fijiensis, Aspergillus ibericus, Aspergillus neoniger

Histoplasma capsulatum: Drugs and Sugars

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Kristie Goughenour

Graduate Program in Microbiology

The Ohio State University

2020

Dissertation Committee

Dr. Chad A. Rappleye, Advisor

Dr. Stephanie Seveau

Dr. Jesse Kwiek

Dr. Jason Slot

Copyrighted by

Kristie Goughenour

2020

Abstract

Histoplasma capsulatum is a thermally dimorphic fungal pathogen capable of causing clinical symptoms in immunocompetent individuals. It exists as hyphae in the environment but transitions to the yeast phase upon encountering the human host, with exposure to mammalian body temperature triggering this phase change. It is endemic to the Ohio and Mississippi River Valleys in the United States, Latin America (specifically

Brazil, Venezuela, Argentina, and Columbia), and parts of Africa, with limited reports in

China and India and demonstrates a clear medical relevance and healthcare burden.

Antifungal options for Histoplasma are limited due to a lack of fungal-specific targets.

Additionally, most studies do not take into account clinically relevant testing of fungal morphotypes and assume a one-size-fits-all approach to fungal drug testing, contributing to false leads and inaccurate frequencies of resistance. In this thesis, we develop and standardize an appropriate method for Histoplasma antifungal susceptibility testing. We show that current CLSI testing methodology is insufficient for testing of Histoplasma and that antifungal susceptibility is often phase-dependent in Histoplasma.

In addition to a need for novel antifungals, there is a real need for novel, fungal-specific drug targets. As a result, target identification is an important stage in antifungal ii development, particularly for large-scale screens of compound libraries where the target is completely unknown. In this thesis, we combine a traditional genetics approach and a novel co-fractionation mass spectrometry approach to identify target genes for a novel antifungal compound, 41F5. We generated 41F5-resistant Cryptococcus neoformans strains and identified 11 potential target genes containing SNPs. In addition, we developed a biochemical approach using size co-fractionation of proteins with unmodified drug via size co-fractionation and compound detection by LC-MS/MS. Drug co-fractionating proteins were identified by proteomics, generating 34 candidate targets.

We are currently validating candidates through overexpression of the candidate genes.

This thesis also investigates virulence factors that could serve as novel fungal-specific drug targets. We work to characterize the O-linked mannosylated proteome in order to identify proteins that may contribute to growth at elevated temperatures (38°C). Using a combined bioinformatics approach and a novel lectin-free biotinylation-based O-linked mannosylated purification we are working to identify O-linked mannosylated proteins and determine which contribute to Histoplasma thermotolerance.

We were additionally interested in other fungal-specific characteristics such as chitin as a main structural component of the fungal cell wall. As chitinases have not been comprehensively studied in fungi as a whole, this thesis generates a comprehensive phylogenetic survey of fungal chitinases in addition to characterization of enzymatic activities and gene expression of Histoplasma chitinases. We determined that the

iii previous phylogenetically determined A clade of chitinases was not monophyletic and that the A and C clades need to be recategorized. In Histoplasma, we identify a hyphal phase-specific chitinase, Cts3. We also identify a recent gene duplication event in

Histoplasma (Cts2 and Cts4) with differences in expression and enzymatic activity.

Finally, we show that enzymatic activity of chitinases does not follow previous phylogenetic predictions, indicating that more chitinases need to be studied to identify their roles in individual fungi.

Dedication

This work is dedicated to my mom and dad for all their support over the years. I couldn’t have done it without them. It’s also dedicated to my friends both from grad school and before for all the good times and awesome science conversations over the years.

Acknowledgments

I want to begin by thanking and acknowledging by advisor Dr. Chad Rappleye. Your support and willingness to let me try new projects and techniques have been really fundamental to my development as a scientist. I’ve learned so much on how to do research well from being in his lab. In addition, I’d like to thank Jason Slot, for all his invaluable assistance with the phylogenetic work for the chitinases project. I’ve learned so much from it and his mentorship was critical for the project and helping me to develop new skills. I’d also like to thank my committee for their support and advice throughout the years.

I want to acknowledge my amazing colleagues (past and present). Andrew, Qian,

Stephanie, Peter, Amita and Janice. You have provided great scientific advice and amazing companionship throughout the years. I’d also like to acknowledge the other

OSU Microbiology graduate students and SAM. You all have provided a welcoming, helpful and collaborative environment.

Vita

2014…………………………………………………………………….B.A. Microbiology

Ohio Wesleyan University

2014-2015…………………………………………………………...University Fellowship

Ohio State University

2015-2018………………………………………………… Graduate Teaching Associate

Department of Microbiology,

The Ohio State University

2018, 2019……………………………………………...NIH/NIAID T32 Graduate Fellow

CMIB/IDI

The Ohio State University

2020-Pesent………………………………………………… Graduate Research Associate

Department of Microbiology,

The Ohio State University

vii

Publications

Tan CY, Wang F, Anaya-Eugenio G, Gallucci J, Goughenour KD, Rappleye CA, Spjut

R, Carcache de Blanco E, Kinghorn AD, Rakotondraibe LH. 2019. α-Pyrone and Sterol

Constituents of Penicillium aurantiacobrunneum, a Fungal Associate of the Lichen,

Niebla homalea. Journal of Natural Products. 82 (9), 2529-2536 DOI:

10.1021/acs.jnatprod.9b00340

Garfoot AL, Goughenour KD, Wüthrich M, Rajaram MVS, Schlesinger LS, Klein BS,

Rappleye CA. 2018. O-mannosylation of proteins enables Histoplasma yeast survival at mammalian body temperatures. mBio doi: 10.1128.

Goughenour KD, Rappleye CA. 2017. Antifungal therapeutics for dimorphic fungal pathogens. Virulence. 8(2):211-221. doi:10.1080/21505594.2016.1235653.

Goughenour KD, Balada-Llasat J-M, Rappleye CA. 2015. Quantitative microplate- based growth assay for determination of antifungal susceptibility of Histoplasma capsulatum yeasts. J Clin Microbiol 53:3286–3295. doi:10.1128/JCM.00795-15.

viii

Fields of Study

Major Field: Microbiology

Table of Contents

Abstract ...... ii Dedication ...... v Acknowledgments ...... vi Vita ...... vii List of Tables ...... xiii List of Figures ...... xiv Chapter 1. Introduction ...... 1 1.1 Histoplasma capsulatum infection ...... 1 1.2 Antifungal treatment for Histoplasma ...... 3 1.2.1 Current Antifungals avaibile for Clinical use ...... 3 1.2.2 New and Alternative antifungal developments ...... 7 1.3 O-Linked Mannosylation and Chitinases in Histoplasma ...... 10 1.4 Conclusions ...... 11 Chapter 2. Quantitative Microplate-Based Growth Assay for Determination of Antifungal Susceptibility of Histoplasma capsulatum Yeasts ...... 14 2.1 Introduction ...... 14 2.2 Materials and Methods ...... 17 2.2.1 Fungal strains and culture ...... 17 2.2.2 Preparation of Histoplasma inocula ...... 18 2.2.3 Microtiter plate growth ...... 19 2.2.4 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide–based yeast vitality assay...... 19 2.2.5 Resazurin-based yeast vitality assay ...... 20 2.2.6 Antifungal drug susceptibility tests ...... 20 2.3 Results ...... 21 2.3.1 Microtiter-plate based growth of Histoplasma yeasts ...... 21 2.3.2 Optimization of microtiter-based assays with end-point indicators of yeast metabolism ...... 30 2.3.3 Application of the microtiter assay for drug sensitivity tests ...... 33 2.4 Discussion ...... 40

Chapter 3. Development of a novel method for target identification of the novel antifungal 41F5 ...... 45 3.1 Introduction ...... 45 3.2. Materials and Methods ...... 47 3.2.1. Yeast strains and growth ...... 47 Table 3.1 Fungal Strains used in this study ...... 48 3.2.2. Generation of Cryptococcus 41F5 resistant mutants ...... 49 3.2.3. Antifungal susceptibility testing ...... 49 3.2.4. Whole genome sequencing of Cryptococcus mutants ...... 50 3.2.5. Detection of 41F5 in cell lysate by HPLC-MS/MS ...... 51 3.2.6. Proteomics analysis ...... 53 3.2.7. Generation of Histoplasma strains overexpressing His-tagged potential target proteins ...... 56 3.3 Results ...... 57 3.3.1 Identification of SNPs from Cryptococcus 41F5 resistant mutants ...... 57 3.3.2 Co-fractionation approach ...... 60 3.3.3 Validation and Optimization of the Co-fractionation approach ...... 63 3.3.4 Identification of 41F5 Co-fractionating Proteins ...... 66 3.3.5 Identification of the target of 41F5 ...... 70 3.4 Discussion ...... 71 Chapter 4. Investigation into the O-linked mannosylated proteome of Histoplasma capsulatum ...... 75 4.1 Introduction ...... 75 4.2 Materials and Methods ...... 76 4.2.1 Histoplasma growth ...... 76 4.2.2. Proteomics analysis ...... 78 4.2.3. Bioinformatics ...... 80 4.2.4 Generation of RNAi Knockdowns ...... 80 4.2.5. Temperature sensitivity assay ...... 83 4.2.6. Determination of the O-linked proteome ...... 83 4.3 Results ...... 84 4.3.1 Mass spectrometry fails to adequately detect O-linked proteins ...... 84 4.3.2 Bioinformatics approach to identification of O-linked mannosylated proteins ...... 87 4.3.3 Biotinylation approach to identification of O-linked mannosylated proteins ...... 91 4.6 Discussion ...... 94 Chapter 5. Diversification of fungal chitinases and their functional differentiation in Histoplasma capsulatum ...... 97 5.1 Introduction ...... 97 5.2 Materials and Methods ...... 100 5.2.1 Phylogenetic Analyses ...... 100 5.2.2 Chitinase gene expression ...... 101 5.2.3 Purification of H. capsulatum chitinases ...... 104 5.2.4 Chitinase activity and specificity determination ...... 106 5.3 Results ...... 107 xi

5.3.1 Diversification of fungal GH18 domains ...... 107 5.3.2 Taxonomic distribution of GH18 chitinases ...... 111 5.3.3 Phylogeny of GH18 ...... 112 5.3.4 Evolution of domain architecture ...... 114 5.3.5 H. capsulatum chitinases ...... 120 5.4 Discussion ...... 128 5.4.1 Previous chitinase ontologies are largely robust to increased sampling, but the A clade may be polyphyletic...... 128 5.4.2 Some chitinase sub-clades have specific taxonomical biases with potential ecological and evolutionary ramifications...... 129 5.4.3 Chitinases containing LysM domains are distributed among plant pathogenic, insect parasitic and saprotrophic fungi ...... 132 5.4.4 Histoplasma capsulatum chitinase expression differences suggest specific functional roles ...... 133 5.4.5 Histoplasma capsulatum chitinase activities do not follow phylogenetically predicable trends in specificity ...... 135 5.4.6 Evolution of the Histoplasma chitinases indicates a degree of differentiation and multiplicity that is reflected in fungal chitinases in general reflecting a potential division of work for chitinases ...... 136 Chapter 6. Conclusions ...... 139 Bibliography ...... 145

xii

List of Tables

Table 1.1 In vitro antifungal MICs for dimorphic fungal pathogens ...... 5

Table 2.1 Antifungal susceptibility of clinical isolates ...... 37

Table 3.1 Fungal Strains used in this study ...... 48

Table 3.2 Primers used in this study ...... 56

Table 3.3 Antifungal susceptibility of Cryptococcus lines to 41F5 and Fluconazole ...... 58

Table 3.4 SNPs Identified in 41F5-resistant Cryptococcus most likely to contribute to resistance ...... 59

Table 3.5 Presence of 41F5 in different size fractions ...... 66

Table 3.6 Proteins identified as co-fractionating with 41F5 ...... 69

Table 3.7 IC50 of Candidate 41F5 Target Overexpressions ...... 71

Table 4.1 Histoplasma strains used in this study ...... 77

Table 4.2 Primers used in this Study ...... 82

Table 4.3 Predicted O-linked Mannosylated Proteins ...... 89

Table 5.1 Primers used in this study ...... 103

Table 5.2 Histoplasma strains ...... 105

Table 5.3 Bootstrap support values for chitinase sub-clades ...... 110

Table 5.4 Clade distribution of Chitin Binding Domains and Enzymatically Characterized

Members ...... 118 xiii

List of Figures

Figure 2.1 Growth of Histoplasma yeast in microtiter plates ...... 23

Figure 2.2 Correlation of OD595 with McFarland turbidity standards and yeast culture density...... 26

Figure 2.3 Optimization of inocula sizes for maximization of the dynamic range of the microplate-based assay...... 29

Figure 2.4 Optimization of parameters for metabolic reduction of MTT (A, B) or resazurin (C, D) for relative quantitation of Histoplasma yeast cell density...... 32

Figure 2.5 Antifungal dose-response curves for Histoplasma cells tested with 96-well microtiter plate microdilution assays...... 35

Figure 2.6 Antifungal susceptibility profile of Fluconazole-resistant isolates ...... 39

Figure 2.7 Summary of the optimized growth and assay parameters for microdilution testing of Histoplasma yeasts...... 41

Figure 3.1 Schematic for the Co-fractionation approach to 41F5 detection...... 62

Figure 3.2 Validation and optimization of 41F5 detection by UPLC-MS2...... 65

Figure 3.3 Detection of 41F5 in FPLC fractions...... 68

Figure 4.1 Proteomic approach fails to identify O-linked mannosylated proteins...... 86

xiv

Figure 4.2 Bioinformatic determination of candidates and initial temperature sensitivity assay...... 88

Figure 4.3 Biotinylation for purification of O-linked mannosylated proteins...... 93

Figure 5.1 Phylogenetic analysis of fungal GH18 domains...... 109

Figure 5.2 LYSM IQ Tree...... 115

Figure 5.3 Phylogenetic analysis of Hisch88_09649...... 116

Figure 5.4 Diagram of Histoplasma capsulatum chitinase genes...... 122

Figure 5.5 Histoplasma capsulatum chitinase expression...... 125

Figure 5.6 Histoplasma capsulatum chitinases enzymatic activity...... 127

Chapter 1. Introduction

This chapter contains work adapted from Goughenour KD, Rappleye CA. Antifungal therapeutics for dimorphic fungal pathogens. Virulence. 2017;8(2):211–221. doi:10.1080/21505594.2016.1235653

1.1 Histoplasma capsulatum infection

Histoplasma capsulatum is a thermally dimorphic primary pathogen capable of infecting both immunocompromised and immunocompetent individuals. It exists in a hyphal form in the environment but transitions to yeast upon encountering the human host, and this morphological transition is required for virulence (Medoff et al. 1986; Nguyen and Sil

2008). Temperature appears to be the central factor that determines the form/lifestyle of the dimorphic fungi, with exposure to mammalian body temperature triggering adoption of the pathogenic mode (Horwath, Fecher, and Deepe 2015). It is endemic to the Ohio and Mississippi River valleys in the United States, Latin America (specifically Brazil,

Venezuela, Argentina, and Columbia), and parts of Africa, with limited reports in China and India (Edwards et al. 1969; Bahr et al. 2015; Zhao et al. 2001). These areas have been largely defined by clinical case prevalence (Chu et al. 2006) or skin-reactivity tests to Histoplasma antigens (Edwards et al. 1969; Kauffman 2007). Ecological aspects of soils that favor growth of the hyphal forms and/or animal or bird patterns for dispersal are thought to underlie endemicity. Inhalation of conidia initiates infection (Kauffman 2009).

The conidia are produced solely by the environmental mycelia. The small size of these

1 propagules facilitates aerosolization upon environmental disturbance and deposition into alveoli once inhaled. Thus, occupational and recreational activities that contribute to soil and environmental disruption are a primary risk factor for infection (Wheat et al. 2016).

Most infection outbreaks can be traced back to specific events that led to conidia release from soils and their subsequent inhalation (Kauffman 2009).

Within the lungs, conidia transition into yeast cells, the pathogenic phase which survive innate immune defenses. Histoplasma yeasts are internalized by host alveolar macrophages and reside almost exclusively within these host cells. Unlike opportunistic fungal pathogens, Histoplasma is not controlled by the innate axis of the immune system.

Control of the infection requires activation of CD4+ T cells, and consequently individuals lacking aspects of cellular immunity (e.g., HIV, immunosuppression due to tissue or organ transplantation, TNFα blockade, etc.) typically progress to severe and disseminated disease. Elimination of symptoms has been assumed to indicate clearance of the infection, but evidence is now suggesting that at least in some individuals, the infection can enter a latent state which can re-emerge later when the balance between pathogen and host immunity is altered (e.g., immunosuppression of the host) (Ankobiah et al. 1990;

Jain, Evans, and Peterson 2006; van Koeveringe and Brouwer 2010; Nakelchik and

Mangino 2002; Vergidis et al. 2015).

Since inhalation is the route of exposure, histoplasmosis is initially a pulmonary disease.

In immunocompetent individuals, mild disease is mostly subclinical, often going undiagnosed. Infection causes varying degrees of pneumonia and influenza-like symptoms. For the majority of individuals, symptoms typically resolve without requiring

2 intervention. Roughly 5% of Histoplasma infections are estimated to require clinical management (Hage et al. 2015). For individuals inhaling a larger inoculum, or those that have some deficiency in cellular immune response, disease is more severe and the infection typically disseminates to extrapulmonary sites via the hemolymphatic system.

The annual incidence of infections by Histoplsama is likely inaccurate as under-diagnosis and under-reporting of infections is common. Furthermore, most infections are self- limiting without requiring clinical intervention. Nevertheless, reports of 6.1 cases per

100,000 person-years in the Midwest of the United States are seen (Vallabhaneni et al.

2016). The number of clinical cases typically range from 5-30% of infections, although in some outbreaks report 68% with symptomatic infections (Chamany et al. 2004). One survey of hospital records in the United States tallied over 3360 cases of histoplasmosis hospitalizations in one year (Chu et al. 2006). Less than half of these were in immunocompromised individuals.

1.2 Antifungal treatment for Histoplasma

1.2.1 Current Antifungals available for Clinical use

The three major classes of currently avaibile antifungals inclues the polyenes, the azoles and the echinocandins. The minimum inhibitory concentrations of the currently availvble antifungals for Histoplasma are shown in Table 1.1. Other endemic thermally dimorphic pathogenic fungi are included for comparison. Of important note is the higher MICs to

3 echinocandins, the least cytotoxic antifungal, for the pathogenic phases of endemic thermally dimorphic pathogenic fungi (Table 1.1 and Chapter 2). The reasons underlying this natural resistance of yeasts to the β-glucan synthase inhibitors is currently unknown.The in vitro antifungal susceptibilities for amphotericin B and azole-class drugs have been validated in murine models of dimorphic fungal disease and/or in clinical trials and as a consequence, the Infectious Disease Society of America has released treatment guidelines for infections with Histoplasma (Wheat et al. 2007). Despite its potential for host toxicity, amphotericin B is recommended for severe disseminated disease, with one of the liposomal formulations preferred. For less severe situations, for follow up after amphotericin B, or for possible prophylaxis of highly susceptible individuals (e.g. AIDS),

IDSA guidelines recommend treatment with an oral azole (i.e., itraconazole) with monitoring of serum concentrations to ensure sufficient absorption and bioavailability.

To ensure sufficient clearance of the Histoplasma infection, treatments typically involve protracted regimens. For mild disease, treatment durations can range from several months to a year depending on the specific dimorphic fungus. Treatment of disseminated disease and disease in immunocompromised hosts can be a year or longer depending on the antifungal used. Beyond the prolonged treatment times, the potential for latency, rather than clearance of the infection, further complicates antifungal management of

Histoplasma.

Table 1.1 In vitro antifungal MICs for dimorphic fungal pathogens

MIC range (µg/mL) Drug class Antifungal Histoplasma Blastomyce Paracoccidioide Coccidioides References s s Polyenes Amphoterici Y: <0.03-2.0 Y:<0.03- Y: 0.06-2.0 Y: 0.25-2.0 (Andreu et al. 2003; Cordeiro et al. 2006; n B M: 0.26-2.5 2.0 M: No data M: 0.03-0.50 Cruz et al. 2013; González et al. 2001; M:No data Goughenour, Balada-Llasat, and Rappleye 2015; Kathuria et al. 2014; Kohler et al. 2000; A M LeMonte et al. 2000; Ren-Kai Li et al. 2000; Nakai et al. 2003; Ramani and Chaturvedi 2007; Sugar and Liu 1996; Zhanel et al. 1997) Imidazoles Ketoconazole Y: No data Y: <0.01- Y: <0.01-0.03 Y: (de Aguiar Cordeiro et al. 2006; Andreu et M: 0.17 0.25 M: No data M: 0.03-0.16 al. 2003; Chapman et al. 1998; Cruz et al. M: 0.1-0.4 2013; Ramani and Chaturvedi 2007; Zhanel et al. 1997) Triazoles Fluconazole Y: 0.25-8.0 Y: 0.06-32 Y: 0.13-0.50 Y: No data (Chapman et al. 1998; Cordeiro et al. 2006; M: 2.0-32 M: 0.06-32 M: No data M: 2.0-64 Cruz et al. 2013; van Duin, Casadevall, and Nosanchuk 2002; González et al. 2001; González et al. 2005; Goughenour, Balada- Llasat, and Rappleye 2015; Kathuria et al. 2014; A M LeMonte et al. 2000; Nakai et al. 2003; Ramani and Chaturvedi 2007; Sugar and Liu 1996; Wheat et al. 2006; Zhanel et al. 1997) Itraconazole Y: <0.01-0.5 Y:<0.01- Y: <0.01-0.06 Y: <0.03-0.50 (Andreu et al. 2003; Chapman et al. 1998; M: 0.03-1.0 0.13 M: No data M: 0.03-1.0 Cordeiro et al. 2006; Cruz et al. 2013; van M: 0.03-4 Duin, Casadevall, and Nosanchuk 2002; González et al. 2005; Kathuria et al. 2014; Ren-Kai Li et al. 2000; Nakai et al. 2003;

5 Ramani and Chaturvedi 2007; Sugar and Liu 1996) Voriconazole Y: 0.03-0.50 Y:<0.03- Y: No data Y: <0.03-2.0 (Cordeiro et al. 2006; González et al. 2005; M: <0.01-2.0 0.25 M: No data M: 0.03-1.0 Kathuria et al. 2014; Ren-Kai Li et al. 2000; M: 0.06-2.0 Ramani and Chaturvedi 2007; Wheat et al. 2006) Posaconazole Y: <0.01- Y: <0.02- Y: No data Y: No data (Espinel-Ingroff 1998; González et al. 2005; 0.50 0.06 M: No data M: 0.06-1.0 Kathuria et al. 2014; Sugar and Liu 1996; M: 0.02-2.0 M: <0.02- Wheat et al. 2006) 2.0 Echinocandin Micafungin Y: >64 Y: 32-64 Y: >64 Y: No data (Nakai et al. 2003) s M: 0.03-0.06 M: <0.01- M: 4-16 M: 0.02 0.03 Caspofungin Y: 8-32a Y: No data Y: No data Y: No data (Cordeiro et al. 2006; Espinel-Ingroff 1998; M: 0.02-4.0 M: 0.5-8.0 M: No data M: 8-64 González et al. 2001; Goughenour, Balada- Llasat, and Rappleye 2015; Kohler et al. 2000) a two studies reported low caspofungin MICs for yeast which disagree with the majority of studies based on clinical isolates in India (0.03-1.0 µg/mL) (Kathuria et al. 2014) and for a single laboratory Histoplasma strain (MIC <0.125 µg/mL) (van Duin et al. 2002)

6 1.2.2 New and Alternative antifungal developments

With the known host toxicity risks of current antifungals (Bates et al. 2001; Cheng et al. 1982;

Nett and Andes 2016), and the endogenous resistance of dimorphic fungi to the lower toxicity echinocandins, new and alternative antifungal drug options have been explored. The eukaryotic nature shared by both the dimorphic fungal pathogens and the mammalian host is recognized as a significant obstacle to antifungal drug development. Consequently, many molecules with promising antifungal activity fail due to inadequate selectivity.

Following the goal of reduced host toxicity, some currently clinically used drugs are being evaluated for repurposing as antifungals. This includes antibiotics as their safety profiles are well-established. For example, drugs targeting the folate pathway have been effective against microbes from bacteria to parasites (Capasso and Supuran 2014), are orally available, and have a moderately safe host toxicity profile. These drugs have already been a means of prophylaxis against Pneumocystis fungal infections in HIV+ individuals. Sulfonamides in combination with dihydrofolate reductase inhibitors (e.g. cotrimoxazole) have activity against Histoplasma yeasts

(8-14 µg/mL and 2-3 µg/mL, respectively) (Brilhante et al. 2010). Using a similar rationale, ciprofloxacin, a fluoroquinolone antibacterial with very low host toxicity, has been tested for antifungal potential. The fluoroquinolones inhibit bacterial type II topoisomerases (e.g., DNA gyrase) and were suggested to potentially also inhibit fungal topoisomerase. While not effective against mycelial forms of Histoplasma, tests against Histoplasma yeasts showed ciprofloxacin has an in vitro MIC of 62.5-250 µg/mL (Brilhante et al. 2013). While not considered a low MIC, the lack of significant host toxicity for ciprofloxacin creates a good selectivity index and raises

7 its potential, at least as an adjunct therapy for fungal infections. Addition of ciprofloxacin to amphotericin B or itraconazole indicated some additive effects could be gained through drug combinations. Anti-tuberculosis drugs show some, albeit low potency, antifungal activity for dimorphic fungal pathogens. When tested in vitro against Histoplasma yeasts, pyrazinamide, isoniazid, and ethambutol had very high MICs (540 µg/mL, 126 µg/mL, and 1240 µg/mL, respectively (Cordeiro et al. 2011)) arguing these antibacterial compounds lack any significant antifungal utility. However, using isoniazid as the core, chemical structure modification improved its antifungal potential; three of nine isoniazid derivatives had significantly lower

MICs against Histoplasma yeasts with one having inhibitory activity at concentrations as low as

7.8 µg/mL (de Aguiar Cordeiro et al. 2014). Antiretroviral protease inhibitors have inhibitory activity against Histoplasma. Saquinavir is active against both filamentous and yeast forms with an MIC around 0.4 µg/mL (Brilhante et al. 2016). Another protease inhibitor, ritonavir, has an

MIC of 1.0 µg/mL against Histoplasma mycelia but is 7-fold more potent against yeast cells. As

HIV+ individuals comprise a significant portion of clinical Histoplasma cases, this raises the possibility that management of HIV through protease inhibitor cocktails could simultaneously provide prophylaxis against infection by dimorphic fungal pathogens. Repurposing of anti- cancer drugs has also been fruitful in the search for new antifungal prospects. AR-12 is a celecoxib-derivative that has been shown to have antimicrobial properties against a wide range microbes including bacteria, parasites, fungi, and even viruses (Baxter et al. 2011; Chiu et al.

2009; Collier et al. 2016; Hoang et al. 2016; Mohr et al. 2015). Early mechanistic studies in cancer cells suggested AR-12 inhibited cellular phosphoinositide-dependent kinase-1 (Pdk1), however this does not appear to be the case in fungi. Recently, it was shown that AR-12 inhibits acetyl-CoA synthetase, an essential enzyme in fungi (Koselny et al. 2016). AR-12 has antifungal

8 activity against a broad collection of fungi, including the dimorphic fungal pathogens at concentrations around 4-8 µg/mL (Koselny et al. 2016). In vitro, AR-12 is fungicidal against

Histoplasma yeasts (Rappleye CA, personal communication). These antifungal characteristics combined with AR-12’s safety as established in Phase I trials make AR-12 an attractive antifungal candidate for further development.

There are also some novel drugs in development as specifically antifungal compounds. Farnesol, an alcohol present in some essential oils but is also produced by C. albicans as a quorum sensing molecule, inhibits Histoplasma yeast growth with an average MIC of 0.02 µM (Brilhante et al.

2015) As a caution, tests with Histoplasma yeasts followed the CLSI methodology (M27-A3), which has been shown to be inferior for reliable testing of Histoplasma yeast susceptibility (see chapter 2 (Goughenour, Balada-Llasat, and Rappleye 2015)). The most advanced new antifungal option for clinical development to manage dimorphic fungal pathogens is nikkomycin Z. The polyoxins (from which the nikkomycins are derived) are peptide modified-nuceloside analogs which were originally identified by screening Streptomyces products for antifungal and insecticidal activities (Isono, Asahi, and Suzuki 1969). These compounds were later shown to be inhibitors of chitin synthesis (Endo and Misato 1969). Since chitin synthase is absent from mammalian cells, these compounds have very high selectivity for fungi. Following polyoxin studies on C. albicans, the nikkomycins were demonstrated to have good potency in vitro against dimorphic fungal pathogens, including Histoplasma (Garfoot et al. 2016; Goldberg et al. 2000;

Hector, Zimmer, and Pappagianis 1990). The nikkomycins are orally available, simplifying extension of studies to animal models of mycoses caused by dimorphic fungal pathogens.

Administration of nikkomycin Z in murine models of lethal histoplasmosis resulted in decreased

9 organ fungal burdens (3- to 28-fold) and a concomitant enhancement of mouse survival at doses from 20 to 100 mg/kg per day (Goldberg et al. 2000; Graybill et al. 1998; Hector, Zimmer, and

Pappagianis 1990). However, significant protection from lethal infection was only realized at moderate inocula suggesting nikkomycin antifungal activity may not be sufficiently rapid to curb severe infections. Still the ability to safely administer such doses is a testament to the high selectivity gained by targeting the fungi-specific molecule chitin synthase.

1.3 O-Linked Mannosylation and Chitinases in Histoplasma

Many proteins secreted by the pathogenic yeast phase of Histoplasma but not the avirulent mycelia phase are heavily glycosylated indicating a probable role in virulence (Holbrook et al.

2011). Recently our lab demonstrated that the post-tanslational modification of O-linked mannosylation is essential for Histoplasma virulence (Garfoot et al. 2018). O-linked mannosylation in Histoplasma generates linear mannose chains attached to serines and theonines and involves an initial mannose attachment by the protein mannosyltransferase (Pmt) family of proteins (Pmt1, 2, and 4), with subsequent extension carried out by a single Mnt1 (Garfoot et al.

2018). We demonstrated that strains lacking Pmt1, Pmt2, and Mnt1 were severely attenuated in a mouse model of infection (Garfoot et al. 2018). Cells lacking Pmt2 had reduced detection by the

β-glucan receptor Dectin-1 although cells were still able to enter and survive within phagocytes

(Garfoot et al. 2018). The virulence defect in o-linked deficient cells was linked to a lack of thermotolerance by these mutants leading to yeast death at 38°C (Garfoot et al. 2018).

Thermotolerance in fungi is considered essential for virulence during systemic infections

(Bhabhra and Askew 2005). While previous studies in Histoplasma have shown that blocking the temperature-dependent transition to yeast is essential for virulence, the factors needed to 10 proliferate at these higher temperatures are not well understood (Shen and Rappleye 2017).

Identifying which proteins are o-linked mannosylated (and therefore may be non-functional in the mannosylation-deficient strains) could pinpoint factors Histoplasma needs to survive at high temperatures. Once these features are identified they may serve as fungal-specific drug targets allowing for the inhibition of Histoplasma thermotolerance and therefore death of the fungus.

As previously mentioned, one of the most promising novel antifungal compounds is nikkomycin

Z, which is an inhibitor of chitin synthesis (Endo and Misato 1969). Chitin is a major structral component of the fungal cell wall, ranging from 0.5%-5% in yeasts to ≥20% in filamentous fungi

(Hartl, Zach, and Seidl-Seiboth 2012). Chitinases are glycosyl hydrolases that break the glycosidic bonds between the N-acetylglucosamine monomers which make up chitin. The roles of chitinase are highly varible in fungi and are not particularly well-characterized into patterns.

In Histoplasma, the chitinases have not been invesigated. In other organisms, some chitinases have shown roles in cell wall remodeling during various phases of life and morphologies (Duo-

Chuan 2006; Seidl 2008b). Chitinases and the remodeling of the fungal cell wall may play important roles in Histoplasma survival and morphology and may serve as potiential drug targets.

1.4 Conclusions

Current clinical management of infections by Histoplasma is limited to azole-class antifungal drugs and amphotericin B. While orally available, the azoles are not without host toxicity issues and the treatment course is lengthy for infections by dimorphic fungi. Development of resistance to azoles is not widespread, although treatment failures due to azole resistance have occurred in 11

Histoplsama and other dimorphs (Kriesel et al. 2008; Wang et al. 2005; Wheat et al. 2001).

Unfortunately, the better-tolerated echinocandin antifungals lack efficacy against the pathogenic- phase of the dimorphic fungal pathogens raising the need for alternative or second-line treatment options (Table 1 and Chapter 2).

While a number of strides have been made in repurposing existing drugs and development of new inhibitors of fungal growth, careful attention must be paid to the unique challenges posed by

Histoplasma. As only the yeast state of Histoplasma is present within the mammalian host, antimicrobial susceptibilities need to be performed with these pathogenic-phase cells, not the mycelia which has led to erroneous conclusions (See Chapter 2). Standard CLSI methodology does not address concerns for specific testing of dimorphic fungi and needs to be optomized for

Histoplasma (See Chapter 2). Additionally, since Histoplasma yeast cells reside within host phagocytes, it is also advisable for in vitro tests to be followed by tests on drug effectiveness on intracellular yeasts, at least during initial drug development stages. Our lab has attempted to take this approach with the indtification of novel anti-Histoplasma compounds (Edwards, Kemski, and Rappleye 2013).

The overall selectivity of antifungal drug candidates is critical for progression of drugs through the development pipeline. Many of the repurposed drugs have relatively high MICs (greater than

100 ug/mL), bringing into question their therapeutic utility. However, if their selectivity is sufficiently high, formulations may be developed to facilitate the high serum and tissue levels required for drug efficacy. Lower MICs have been found for drugs targeting the folate pathway: an isoniazid-hydrazone derivative, antiretroviral protease inhibitors, and the anti-cancer drug

AR-12, all of which are expected to be reasonably well-tolerated by the mammalian host. Novel drugs with good in vitro MICs and good selectivity include thioredoxin-reductase inhibitors, an aminothiazole compound, and nikkomycin Z. Since nikkomycin Z targets an enzyme absent from the host, nikkomycin has an excellent basis for high selectivity against fungi. In addition, nikkomycin Z has maintained antifungal effectiveness against multiple dimorphic fungal pathogens in animal models of disease. The identification of additional novel fungal-specific drug targets would be of great benefit to furthering antigungal development. To this end, the identification of novel virulence factors, such as O-linked mannosylated proteins which may confer thermotolerance to Histoplasma during infection, and a better understanding of the basic biology of the organism, such as the role of chitinases, are required to generate new fungal- specific drug targets.

Chapter 2: Quantitative Microplate-Based Growth Assay for Determination of Antifungal

Susceptibility of Histoplasma capsulatum Yeasts

This chapter was previously published as Goughenour KD, Balada-Llasat JM, Rappleye CA.

Quantitative Microplate-Based Growth Assay for Determination of Antifungal Susceptibility of

Histoplasma capsulatum Yeasts. J Clin Microbiol. 2015;53(10):3286–3295. doi:10.1128/JCM.00795-15

2.1 Introduction

The increasing incidence of fungal disease necessitates adequate and timely assessment of antifungal susceptibility to guide the selection and implementation of antifungal therapies.

Consequently, the Clinical and Laboratory Standards Institute (CLSI) and the European

Committee on Antimicrobial Susceptibility Testing (EUCAST) have established reference test methods for susceptibility to the major classes of antifungals, which include the polyenes, azoles, and the recently developed echinocandins (Cuenca-Estrella and Rodriguez-Tudela 2010; Pfaller and Diekema 2012). These standards utilize broth macrodilution or microdilution assays for testing yeasts (M27-A3 (Rex and Clinical and Laboratory Standards Institute 2008) and E.DEF

7.1 (Subcommittee on Antifungal Susceptibility Testing (AFST) of the ESCMID European

Committee for Antimicrobial Susceptibility Testing (EUCAST) 2008)) and filamentous fungi

(M38-A2 (Reference Method for Broth Dilution Antifungal Susceptibility Testing of

Filamentous Fungi. 2008) and E.DEF 9.1(Subcommittee on Antifungal Susceptibility Testing of the ESCMID European Committee for Antimicrobial Susceptibility Testing 2008)). Procedures, including inoculum preparation, media, assay duration, and endpoint definitions, have now been established for testing some of the more commonly encountered yeast-form pathogenic fungi

(i.e., Candida species and Cryptococcus). However, notably missing from the M27-A3 and

E.DEF 7.1 standards for testing of yeasts are methods for testing the yeast forms of the dimorphic fungi.

The dimorphic fungi that cause systemic disease are characterized by distinct yeast and filamentous morphological states (Klein and Tebbets 2007; Maresca and Kobayashi 2000).

Temperature is the principal morphology-determining factor, with yeast forms characterizing infections in mammals (Kauffman 2007). Histoplasma capsulatum is the most common clinically encountered dimorphic fungal pathogen in the United States, with an estimated 10,000 to 20,000 hospitalizations annually (Chu et al. 2006). Histoplasma is acquired by inhalation of mycelia-produced conidia from the environment. In the lung, mammalian body temperature triggers differentiation of the conidia into yeasts instead of filamentous hyphae (Kauffman

2007). Histoplasma yeasts parasitize phagocytic cells, and the yeast-infected phagocytes can facilitate extrapulmonary dissemination of the fungus to cause life-threatening disseminated histoplasmosis, particularly in immunocompromised individuals. In the majority of otherwise healthy individuals, low-dose inoculation results in a self-limiting disease with the onset of cell- mediated immunity. With higher doses, Histoplasma causes acute disease, even in

15 immunocompetent individuals (Chu et al. 2006; Hage, Knox, and Wheat 2012).

Despite the fact that the yeast form is the state most relevant to disease, some in vitro antifungal testing for Histoplasma has been performed only on the filamentous form (Borelli et al. 2008;

Espinel-Ingroff 1998; González 2009). Both yeast- and filamentous-form Histoplasma cells are generally susceptible to polyenes and triazole-class antifungals in vitro (Cheung et al. 1975;

Hage et al. 2011; Nakai et al. 2003). Early studies on the echinocandins suggested that this antifungal class, with very low host toxicity, could be effective against Histoplasma infections

(Espinel-Ingroff 1998; Johnson and Perfect 2003; van Duin, Casadevall, and Nosanchuk 2002).

However, some of these early reports were based on tests with the filamentous form, and subsequent in vitro susceptibility tests employing the yeast form demonstrated that the

Histoplasma yeasts are comparatively resistant to the echinocandins (Hage et al. 2011; Nakai et al. 2003; Kohler et al. 2000). This morphology-based discrepancy in antifungal susceptibility is not limited to echinocandins and highlights the need to use the most appropriate cell type during antifungal testing (Cordeiro et al. 2016). With the increasing importance of monitoring antifungal drug resistance and the acceleration of new antifungal drug discovery efforts, optimized methods for high-throughput antifungal tests on Histoplasma yeasts are essential.

In this study, we standardized a broth microdilution method for assaying Histoplasma yeast growth. This study was motivated by the inadequacy of CLSI-defined yeast culture procedures for the growth of Histoplasma yeasts, the clinical form of Histoplasma. In some previous studies, the CLSI M27-A2 and -A3 standards have been adapted for Histoplasma yeast (Edwards,

Kemski, and Rappleye 2013; Wheat et al. 2001; Wheat et al. 2006; Andreu et al. 2003), but no

16 consensus methods or optimization of assay parameters have been detailed. Here, we optimize the inoculum size and assay duration to maximize the dynamic range of a microdilution assay for

Histoplasma. The assay quantifies yeast growth through spectrophotometric measurement of culture turbidity to provide a more objective conclusion of endpoints than visual inspection allows. To complement culture density measurements, we also determined optimal parameters for the use of colorimetric and fluorescent indicators of yeast cell density and vitality. These methodologies were validated for use in a clinical microbiology laboratory setting by establishing antifungal susceptibilities of multiple clinical isolates from both North American phylogenetic groups of Histoplasma and a Latin American isolate.

2.2 Materials and Methods

2.2.1 Fungal strains and culture

Histoplasma capsulatum strains used were the wild-type North American type 2 (NAm2) strain

G217B (ATCC 26032), the wild-type Panama strain G186A (ATCC 26029), 7 clinical sample isolates from the Ohio State Medical Center, and 3 clinical isolates from MiraVista Labs (Wheat et al. 2006; Wheat, MaWhinney, et al. 1997; Wheat, Marichal, et al. 1997). Isolates were assigned to the North America NAm1 and NAm2 phylogenetic groups (Kasuga et al. 2003) based on PCR-restriction fragment length polymorphism markers (Bohse and Woods 2007).

Histoplasma cells were maintained as yeasts by growth at 37°C on a medium optimized for

Histoplasma yeast growth (Histoplasma macrophage medium [HMM]) supplemented with 25

µM FeSO4 and solidified with 0.6% agarose (30) or on a brain heart infusion (BHI) medium

(Becton, Dickenson and Co.) solidified with 0.6% agarose instead of agar. For liquid culture, basal cell culture media were used as the rich growth media. F-12 (Gibco) and RPMI 1604

(Gibco) media were compared, as F-12 is the base composition of HMM medium, and RPMI is more commonly used in clinical laboratories for microdilution assays. Both broth media were buffered to a pH of 7.0 with 25 mM HEPES and were optionally supplemented with 1.5% glucose and/or 0.7 mM cystine for nutritional tests. For growth in microtiter plates (see below), yeasts were added to wells in a total volume of 100 µL. For quantitative platings, serial dilutions of yeast suspensions were plated on solid HMM medium, as this medium supports the robust growth of individual colonies, and incubated at 37°C until colonies developed (8 to 10 days).

2.2.2 Preparation of Histoplasma inocula

Microtiter plate inocula of Histoplasma yeasts were prepared either by suspending yeast colonies in F-12 medium or from broth cultures of Histoplasma yeasts that had been pregrown in culture tubes for 48 h. For pregrowth of Histoplasma yeasts before dilution into microtiter plates, yeast colonies were inoculated into 2.5 ml liquid medium in 16 mm × 100 mm test tubes. Tubes were placed at a 30-degree angle and incubated for 2 days at 37°C with shaking (200 rpm) in a

Multitron orbital shaker (Infors, Inc.). Yeast density was determined directly using hemacytometer counts, estimated by optical density at 595 nm (OD595), or by a turbidity comparison to McFarland standards. For estimation of yeast density by OD595, a standard curve of OD595 versus the yeast cell number was used. For comparison to McFarland standards,

McFarland standards (of 0.5, 1.0, or 1.5) were prepared by mixing defined volumes of 1% BaCl2

(1.175% BaCl2 dihydrate) with 1% H2SO4 (50 µl, 100 µl, and 200 µl of BaCl2 in a final volume of 1 mL). Yeast suspensions were adjusted with growth medium to a density equivalent to 1.0

McFarland standard (representing approximately 8 × 106 yeasts/ml) and further diluted to the desired inoculum.

2.2.3 Microtiter plate growth

A total of 50 µL of yeast culture at twice the desired final yeast density was added to wells of a

96-well flat-bottomed microtiter plate. Subsequently, 50 µl of growth medium or medium containing antifungal drugs at twice the desired concentration (typically 2-fold dilutions from 32

µg/mL to 0.03 µg/mL) was added to each well for a final volume of 100 µL. Plates were incubated at 37°C for 5 days with twice-daily agitation (30 s at 1,000 rpm) on a microplate mixer

(Eppendorf) to improve aeration. Culture turbidity was determined by measurement of the absorbance of each well at 595 nm with a Synergy 2 microplate reader (Biotek).

2.2.4 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide–based yeast vitality assay.

Histoplasma yeasts in 100 µL of growth medium were added to 96-well plates and grown as described above for approximately 96 h. A 5× solution of 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide (MTT) was prepared in growth medium containing 0.5 mM menadione as an electron shuttle. Then, 25 µL of the MTT solution was added to each well that

19 contained Histoplasma yeasts as well as to wells that contained growth medium without

Histoplasma yeasts. Plates were incubated at 37°C to allow for reduction of MTT and sufficient color development. The assay was terminated by the addition to each well of 50 µL of 20% SDS with 10 mM HCl. After at least 2 h of incubation at ambient temperature for reduced MTT extraction, the microtiter plates were centrifuged for 5 min at 2,000 × g to collect yeast cell debris, and 50 µl of the supernatant was transferred to a microtiter plate containing 50 µl of distilled H2O. The extracted MTT was quantified by absorbance readings at 570 nm after correction with absorbance at 690 nm using a Synergy 2 plate reader (BioTek).

2.2.5 Resazurin-based yeast vitality assay

Histoplasma yeasts in 100 µL of growth medium were added to 96-well plates and grown as described above for approximately 96 h. 10 µL of a 10X concentrated solution of resazurin (7- hydroxy-10-oxidophenoxazin-10-ium-3-one) were added to each well, including control wells containing growth media without yeast cells and the microtiter plates incubated at 37°C. Kinetics of resorufin (reduced resazurin; 7-hydroxyphenoxazin-3-one) production was quantified by measurement of fluorescence (530/25 excitation, 590/35 nm emission wavelengths) every 20 min using a Synergy2 plate reader (BioTek).

2.2.6 Antifungal drug susceptibility tests

Antifungals used in clinical practice were applied to Histoplasma yeast and included

20 amphotericin B, fluconazole, and caspofungin. Antifungals were prepared according to the CLSI reference standard M27-A3 (Rex and Clinical and Laboratory Standards Institute 2008). For the

50% inhibitory concentration (IC50) and MIC determinations for Histoplasma yeasts, a 2-fold dilution series of antifungal drugs was prepared, and 50 µL of drug-containing solution was added to the wells of a 96-well microtiter plate containing 50 µL of Histoplasma yeasts at a density of 4×106 to 8×106 yeasts/mL (producing a final yeast density of 2 × 106 to 4 × 106 yeasts/mL). Yeast growth was assessed by measurement of turbidity at 595 nm after 96 to 144 h and the growth was normalized to wells containing no antifungal drug. Endpoint resazurin reduction assays were subsequently employed to confirm OD595 readings. Relative growth at each antifungal drug concentration was used to generate dose-response curves, and the IC50 was determined by nonlinear regression. MIC values were based on the minimum concentration of antifungal drug that prevented development of visual turbidity or lack of resazurin fluorescence compared to growth in the absence of antifungal drug. The morphology of Histoplasma cells following drug treatment was evaluated by phase-contrast microscopy at 100X magnification.

2.3 Results

2.3.1 Microtiter-plate based growth of Histoplasma yeasts

To determine the maximal yeast cell density that could be achieved in microtiter plates, replicate plates (n = 3) were inoculated with 1 × 107, 1 × 106, 1 × 105, and 1 × 104 yeasts/mL of the

G217B strain, a clinical strain representative of North American isolates. Plates were incubated at 37°C for 5 days. Yeast growth and viability were monitored daily by measurement of optical

21 density at 595 nm (OD595) and by plating serial dilutions of the yeast suspension to enumerate

CFU, respectively. Maximal growth in 96-well microtiter plates saturated at approximately 2.5 ×

107 CFU/mL (Fig. 2.1A). Yeast growth reached saturation at 2 to 3 days, 4 days, and 5 days for 1

× 107, 1 × 106, and 1 × 105 yeasts/mL of starting culture density, respectively. Yeast cultures started at 1 × 104 yeasts/mL significantly lagged in growth and reached saturation only after 6 to

7 days of incubation (data not shown). Yeast viability as determined by CFU closely paralleled the optical density of each well at 595 nm (Fig. 2.1B), indicating that optical density was a reliable indicator of yeast cell growth. Importantly, Histoplasma microdilution assays started at both the high and low inoculum levels recommended for yeasts by the CLSI procedures (2.5 ×

103 and 0.5 × 103 yeasts/mL, respectively (Rex and Clinical and Laboratory Standards Institute

2008) resulted in a complete failure of Histoplasma yeast growth (data not shown).

Figure 2.1 Growth of Histoplasma yeast in microtiter plates

Histoplasma yeasts were grown in 96-well microtiter plates, and the yeast cell density was measured by CFU (A) or optical density at 595 nm (OD595) (B). Growth media were inoculated with 1 × 107 (circles), 1 × 106 (squares), 1 × 105 (diamonds), and 1 × 104 (triangles) yeasts/mL, and microtiter plates were incubated at 37°C. Plates were shaken (60 s at 1,000 rpm) twice daily to improve aeration. Yeast growth was assessed at 24-hour intervals. (A) For quantitative platings, dilutions of yeast cultures from 96- well microtiter plates were plated on solid HMM medium to enumerate viable CFU. (B) Culture turbidity was measured by OD595 in a plate reader. (C) Effect of supplementation of growth media with glucose (Glc) or cysteine (Cys) on yeast cell 6 density (OD595). Data shown represent growth at 5 days after inoculation with 4 × 10 yeasts/ml in Ham's F-12 (black bars) or RPMI (gray bars) buffered to a pH of 7.0. Data points represent the average ± standard deviation of biological replicate cultures (n = 3).

As maximal growth obtained in microtiter plates was less dense than when Histoplasma yeasts were grown in culture tubes, we investigated whether glucose or cysteine was limiting in the

RPMI-based medium. CLSI procedures for growth of yeast-form organisms (i.e., Candida and

Cryptococcus) typically utilize a defined rich medium (e.g., RPMI) buffered to a pH of 7.0 (Rex and Clinical and Laboratory Standards Institute 2008). HMM, a medium optimized for growth of

Histoplasma and macrophages, is based on Ham's F-12 cell culture medium and is supplemented with 1.8% glucose and 0.7 mM cystine (Worsham and Goldman 1988). The increased cystine provides additional organic sulfur, since Histoplasma yeast cells and mycelia differ in their requirements for organic sulfur (Boguslawski and Stetler 1979; McVeigh I, Houston WE 1972;

Pine 1954; Salvin 1949; Stetler and Boguslawski 1979). Although both F-12 and RPMI have cysteine (0.2 mM and 0.4 mM, respectively), we tested if supplementation of F-12 or RPMI with cystine increased the yield of Histoplasma yeasts in microtiter plates. Neither added cystine nor added glucose significantly improved the overall rate or total growth of Histoplasma yeasts; no significant differences in yield were detected by one-way analysis of variance (ANOVA) (n = 3;

Fig. 2.1C). With some strains of Histoplasma, F-12 supported 20% to 80% higher yeast cell densities than RPMI (data not shown). We suspect that trace metals (i.e., Fe2+, Cu2+, Zn2+) present in F-12 but absent from the RPMI formulation may account for the improved growth in the F-12 medium. Thus, F-12 (lacking any supplements) buffered to a pH of 7.0 was selected as the growth medium for microtiter plate-based growth of Histoplasma.

Enumeration of yeasts by hemacytometer counts was used to establish a standard curve of optical density versus yeast cell density and to correlate Histoplasma yeast cell number to McFarland standards. Since spectrophotometers vary in their readings for light scattering due to sample

24 turbidity, the McFarland standards provide a defined optical density reference. McFarland standards of 0.5 to 4.0 were prepared in triplicate, and their optical densities at 595 nm were determined using a plate reader (Fig. 2.2A). In parallel, different yeast culture densities

(determined by hemacytometer counts) were correlated to optical density readings (Fig. 2.2B).

Both linear relationships were subsequently combined to produce a standard curve of the turbidity reference (McFarland standards) and the Histoplasma yeast cell density (Fig. 2.2C). A

McFarland standard of 1.0 was equivalent to 8.4 × 106 Histoplasma yeasts/mL.

Figure 2.2 Correlation of OD595 with McFarland turbidity standards and yeast culture density.

Optical density readings (OD595) were correlated to McFarland turbidity standards (A) or yeast cell density (B). (A) McFarland standards were constructed by mixing BaCl2 with 1% H2SO4 and then measuring the optical density. Data points represent individual readings for replicate cultures (n = 3), and the dashed line indicates the computed linear relationship (B) For determination of yeast cell density, dilutions of Histoplasma yeast cultures were counted by hemacytometer to enumerate yeast cells. Data points represent individual Histoplasma cultures. Dashed line indicates the relationship between OD595 and yeast cell density by linear regression. (C) The corresponding OD595 for each McFarland standard was matched to the OD595 standard curve for Histoplasma yeast culture density to yield a linear relationship for McFarland standards and the titer of the yeast culture.

To optimize the dynamic range of the growth of Histoplasma yeasts in 96-well microtiter plates, we investigated the parameters of starting yeast cell density and incubation time using cultures of

G217B in triplicate. Starting inocula of less than 1 × 105 yeasts/mL displayed significant lags in growth (Fig. 2.1A) or no growth at all. While inocula of around 1 × 107 yeasts/mL produced maximal growth in a short time (2 to 3 days), these inocula already showed significant optical density in the absence of any growth, thereby decreasing the dynamic range (Fig. 2.3A). Starting concentrations of around 5 × 105 yeasts/mL reached maximal growth at 4 days. Comparison of the OD595 readings at 3 to 4 days (72 to 96 h) with the OD595 readings at the start of the assay showed that initial inocula at 2 × 106 to 4 × 106 yeasts/mL had the largest dynamic range (Fig.

2.3B) and still produced near-maximal growth by 4 days of incubation. A total of 96 h of incubation provided the maximum difference, with an overall dynamic range of 35-fold and 77- fold for 4 × 106 and 2 × 106 yeasts/mL inocula, respectively. As a further test for optimal starting yeast cell density, different inocula of Histoplasma yeasts were grown in microtiter plates in the presence or absence of the antifungal drug amphotericin B. Calculation of the Z′-factor, a statistical measure of the discriminatory power of an assay for comparing growth in the presence or absence of drug (Zhang, Chung, and Oldenburg 1999), showed that, while all starting inocula could discriminate between growth and no growth (Z′ > 0.5), an inoculum of 2 × 106 to 4 × 106 yeasts/mL was optimal. From these results, we established 4 × 106 yeasts/mL as the starting yeast cell density and 96 h for the growth time for the microtiter plate-based OD595 growth assay.

Greater consistency in growth kinetics was found when the initial inoculum was prepared from a liquid culture of Histoplasma yeast pregrown in culture tubes for 2 days than when yeast

27 suspensions were made from colonies collected directly from solid medium. This was not due to transfer of any spent medium, since washing the pregrown cells before inoculation of microtiter plates yielded identical results as simply diluting pregrown cells (data not shown). The initial inoculum for the microdilution assay in 96-well plates may be determined by hemacytometer counts or by the adjustment of a yeast suspension to a McFarland standard of 1.0 followed by a

2-fold dilution of the adjusted suspension.

Figure 2.3 Optimization of inocula sizes for maximization of the dynamic range of the microplate-based assay.

(A) Representative growth curves of Histoplasma yeasts initiated at starting densities of 3.2 × 107 (circles), 8.0 × 106 (squares), 2 × 106 (diamonds), or 5 × 105 (triangles) yeast cells/mL. Data points represent average optical density readings ± standard deviation of replicate cultures (n = 3). (B) Determination of the dynamic range of OD595 readings for different inocula after growth at 37°C. The dynamic range was determined as the difference between the initial OD595 reading at the time of 0 h and then at the time of 72 h (diamonds), 87 h (squares), or 96 h (circles). (C) Z′-factor statistic indicating the discriminatory power of the microtiter plate-based assay of Histoplasma yeast growth. Cultures were initiated at 5 × 105 yeasts/mL to 3.2 × 107 yeasts/mL. For computation of the Z′-factor, wells containing cultures grown in the absence of antifungal drug were compared to wells cultured in the presence of 5 µg/ml of amphotericin B to suppress growth. Data points represent the average ± standard deviation of replicate cultures (n = 3).

2.3.2 Optimization of microtiter-based assays with end-point indicators of yeast metabolism

Measurement of optical density provides an easy and efficient way to quantify yeast growth over time. Metabolic indicators of cell density, based on metabolism-dependent reduction of colorimetric and fluorescent dyes, provide an alternative measure of yeast cell density as well as provide an indicator of yeast vitality. Two accepted endpoint assays are cellular reduction of the tetrazolium dye MTT to its formazan, which can be monitored by absorption at 590 nm, and reduction of resazurin to fluorescent resorufin. The colorimetric change in both reactions can also be used as a more objective qualitative indicator than visual determination of turbidity. To determine the suitability of these assays for measuring Histoplasma yeast cell growth, we tested the ability of Histoplasma yeasts to reduce these dyes and optimized the dye concentration used and the timing of maximal color/fluorescence development.

To determine the optimal concentration of MTT and resazurin substrates to be used, MTT was added to suspensions of Histoplasma yeasts at final concentrations of 0 to 1.0 mg/mL, and resazurin, at final concentrations of 0 to 200 µM (n = 3 for each). We used yeast cell densities at or above the maximum titer obtained in 96-well microtiter plate growth to ensure that the substrates would not be limiting for yeast grown to saturation density in microtiter plates. A dose-response relationship between the yeast cell number and formazan formation was observed

(Fig. 2.4A). Saturating levels of the MTT substrate for the highest yeast cell densities occurred around 0.4 mg/mL. Thus, 0.5 mg/mL MTT was chosen as the substrate concentration that would not be limiting for the assay with Histoplasma yeasts. To determine the timing for maximal

30 formazan formation, we grew Histoplasma yeasts in microtiter plates for 96 h, added 0.5 mg/mL

MTT, and incubated the plates at 37°C for different time intervals from 0 to 240 min before stopping the reaction. Maximal MTT reduction to its formazan was achieved at 180 min after

MTT addition (Fig. 2.4B). As with the MTT assay, a dose-response relationship between the yeast cell number and resorufin formation from resazurin was observed (Fig. 2.4C). Levels of the resazurin substrate became saturating at 100 µM. Concentrations of resazurin above 100 µM partially inhibited the reduction of resazurin to resorufin (Fig. 2.4C and data not shown).

Maximum fluorescence was achieved by 100 to 120 min of incubation with 100 µM resazurin

(Fig. 2.4D). At 90 min, the reaction was near but not yet at saturation levels; thus, 90 min was designated the optimal time for the resazurin reduction assay.

To determine if the MTT- and resazurin-based assays were suitable to detect inhibition of

Histoplasma growth, the assays were applied to Histoplasma yeasts grown in microtiter plates in the presence of the antifungal drug fluconazole (Fig. 2.4B and 2.4D). This test showed that the

MTT assay could efficiently discriminate between growth and inhibition of growth of

Histoplasma yeasts, with a dynamic range of 7-fold. No additional formazan was formed by incubation times longer than 180 min; however, the background of nongrowing yeasts continued to increase, thereby falsely reporting the relative difference between inhibited and uninhibited yeasts. Comparison of the resazurin assay between Histoplasma yeasts grown in the presence or absence of the antifungal drug fluconazole showed that the resazurin assay also had excellent discrimination between yeast growth and growth inhibition, with a Z′-factor of 0.85 and a dynamic range of 12-fold.

Figure 2.4 Optimization of parameters for metabolic reduction of MTT (A, B) or resazurin (C, D) for relative quantitation of Histoplasma yeast cell density.

For optimization of substrate concentration (A, C) Histoplasma yeast suspensions at 1 × 108 (circles), 7 × 107 (squares), 4 × 107 (diamonds), and 1 × 107 (triangles) yeasts/mL were incubated with various amounts of substrate. For optimization of the timing of the development reactions (B, D), Histoplasma yeasts in wells of a 96-well microtiter plate were grown for 96 h in the absence (circles) or presence (squares) of 4 µg/mL of fluconazole (Flc) to suppress yeast growth and then incubated with MTT or resazurin substrates. MTT reduction was monitored by formazan product formation, which was extracted from cells with SDS and quantified by absorbance at 570 nm. Resazurin reduction was monitored by fluorescence of the resorufin product. Histoplasma yeasts were incubated with MTT at final concentrations ranging from 0 to 1.0 mg/mL (A), with 0.5 mg/ml MTT and the formazan product extracted at 30 min intervals for quantification (B), with resazurin at final concentrations ranging from 0 to 200 µM (C), or with 100 µM resazurin and the fluorescence of the resorufin product measured at 20 min intervals (D). Data points represent the average ± standard deviation of replicate cultures (n = 3).

2.3.3 Application of the microtiter assay for drug sensitivity tests

To validate the optimized 96-well microtiter plate assay for drug sensitivity studies, we determined the concentrations for 50% inhibition (IC50) and the MICs of antifungal drugs against the G217B strain of Histoplasma capsulatum (Fig. 2.5). Using the OD595 parameters optimized above, yeast growth was determined over a range of antifungal drug concentrations in triplicate cultures. As optical density can quantify partial yeast growth, dose-response curves could be constructed, and the IC50 could be determined from linear regression. MICs were determined by visual inspection of the plate for lack of turbidity. Treatment of Histoplasma yeasts with amphotericin B showed an IC50 of 1.0 ± 0.03 µg/mL (mean ± standard deviation [SD]) and an

MIC of 2.0 µg/mL (Fig. 2.5A). Fluconazole had an IC50 of 0.24 ± 0.01 µg/mL (mean ± SD) and an MIC of 0.5 µg/mL (Fig. 2.5B). For the echinocandin caspofungin, Histoplasma yeasts were characterized by an IC50 of 10.4 ± 0.5 µg/mL (mean ± SD) and an MIC of 32 µg/mL (Fig. 2.5C).

These results are in agreement with established drug sensitivities for Histoplasma by macrodilution assay and modified microdilution assays (Kohler et al. 2000; Wheat et al. 2006; A.

M. LeMonte et al. 2000; R. K. Li et al. 2000). Examination of Histoplasma cells by microscopy at inhibitory and subinhibitory drug concentrations showed no drug-induced morphological conversion to mycelia from the yeast-form inocula (data not shown). Using the microdilution assay, we confirmed that the antifungal susceptibility profile of yeasts differed from mycelia; 20- fold more caspofungin was required to inhibit yeasts than mycelia (Fig. 2.5C). In contrast mycelia were more resistant to fluconazole than were the Histoplasma yeasts (Fig. 2.5B). No significant difference in the susceptibility to amphotericin B was found between the two

Histoplasma morphologies (Fig. 2.5A), consistent with previous observations of amphotericin B

33 effects on yeasts and mycelia (Hage et al. 2011). The variation in antifungal drug effects on the different morphologies underscores the importance of using the clinically relevant form (i.e., yeasts) for antifungal susceptibility profiling of Histoplasma.

Figure 2.5 Antifungal dose-response curves for Histoplasma cells tested with 96-well microtiter plate microdilution assays.

Charts depict representative dose-response curves of yeast to the antifungal compounds amphotericin B (A), fluconazole (B), and caspofungin (C) as determined by quantitative growth assay (turbidity at 595 nm). Histoplasma yeasts were inoculated into buffered F-12 media at a density of 2 × 106 yeasts/mL and grown for 96 hours, at which point the OD595 was measured. Yeast cell density was normalized to wells with no antifungal drugs added, and the relative growth was plotted. Data points represent the average ± standard deviation of replicate cultures (n = 3) for each antifungal drug concentration tested. Trend lines indicate the regression line for the yeast-phase dose response which was used to derive the IC50 for each antifungal drug. MICs were determined by visual inspection of wells and identification of wells without turbidity.

We applied the optimized microtiter plate-based assay to test the drug sensitivity profile of clinical isolates of Histoplasma using yeast-phase cells. The isolates represent clinical strains from the North America phylogenetic group 1 (NAm1), the more clinically prevalent North

America group 2 (NAm2), and a Latin American isolate (LAm), as well as from a laboratory strain originally from Panama (G186A; Pan). The antifungal susceptibility results of these clinical isolates are presented in Table 2.1. In general, Histoplasma clinical isolates were similar in their susceptibility to amphotericin B but NAm1, LAm, and the Panama isolates had slightly higher natural resistance to fluconazole (2- to 3-fold higher IC50). All isolates were relatively resistant to caspofungin, with MICs above 8 µg/ml. Of note, the NAm1, LAm, and Panama isolates were more resistant than NAm2 clinical isolates to the β-glucan synthase inhibitor caspofungin.

Table 2.1 Antifungal susceptibility of clinical isolates Susceptibility datab (µg/mL) for: Amphotericin B Fluconazole Caspofungin Phylogenetic a Isolate group IC50 MIC IC50 MIC IC50 MIC 0.11 ± 10– HC17 NAm1 0.02 0.24 1.88 ± 0.52 2.5 8.9 ± 2.4 20 0.08 ± 0.06– 0.64– 10– HC01 NAm2 0.03 0.16 0.56 ± 0.03 1.28 8.1 ± 2.0 16 0.06 ± 0.04– 0.64– HC10 NAm2 0.04 0.16 0.48 ± 0.11 1.28 5.3 ± 0.9 8–10 0.05 ± 0.04– HC14 NAm2 0.04 0.16 0.59 ± 0.22 1.28 5.5 ± 0.8 8–10 0.07 ± 0.06– 0.32– HC16 NAm2 0.04 0.32 0.25 ± 0.07 0.64 5.6 ± 3.6 8–16 0.06 ± 0.06– 0.64– HC22 NAm2 0.03 0.16 0.81 ± 0.16 1.28 6.4 ± 2.9 8–16 0.06 ± 1.28– 11.6 ± HC30 Lam 0.01 0.12 1.12 ± 0.43 2.56 0.6 20 G186 0.15 ± 0.24– 2.56– 23.4 ± 32– A Pan 0.06 0.64 1.82 ± 0.45 5.12 7.4 40 0.04 ± 12.3 ± IN2 NAm2 0.01 0.12 12.2 ± 0.5 32 0.8 20 0.07 ± 0.06– 145.3 ± 34.4 ± 40– IN14 NAm2 0.01 0.13 14.8 256–512 6.9 80 0.15 ± 0.25– 10.3 ± NY17 RFLP-VI 0.02 0.50 52.1 ± 29.7 128–256 3.5 40 aPhylogenetic group classification as described by Kasuga et al. (Kasuga et al. 2003): NAm1, North American type 1; NAm2, North American type 2; LAm Latin American isolates or, in the case of NY17, based on the RFLP classification scheme of Keath et al. (Keath, Kobayashi, and Medoff 1992). b IC50 expressed as average ± standard deviation (n = 3); MIC expressed as range (low to high; n = 3).

As further confirmation of the parameters for the Histoplasma microdilution assay for clinical and laboratory application, we tested three isolates with established resistance to fluconazole.

Clinical isolates IN2, IN14, and NY17 originated from AIDS patients who had failed or relapsed following fluconazole therapy (Wheat et al. 2006; Kasuga et al. 2003; Bohse and Woods 2007).

Compared to typical inhibitory levels of fluconazole for NAm2-class Histoplasma yeasts, we observed 22- to 200-fold increases in the IC50s and MICs of fluconazole for isolates IN2 and

IN14, respectively (Table 2.1 and Fig. 2.5). The NY17 strain also had decreased susceptibility to fluconazole (MICs, 128 to 256 µg/mL). Interestingly, the IN14 isolate also showed approximately 3-fold higher resistance to caspofungin (Fig. 2.5), even though this antifungal was not part of the treatment regimen for the patient from whom it was isolated. Further investigation of the very high fluconazole resistance of the IN14 isolate identified a mutation in the ERG11B gene that resulted in substitution at amino acid 165 (L165R; data not shown).

Figure 2.6 Antifungal susceptibility profile of Fluconazole-resistant isolates

Microdilution assays were used to test the susceptibility of fluconazole-resistant isolates IN2 (blue symbols), IN14 (red symbols), and NY14 (green symbols) to fluconazole (A), caspofungin (B), and amphotericin B (C). G217B (black symbols) susceptibility is provided as a wild-type NAm2 strain for comparison. Dose-response curves were computed by non-linear regression and error bars represent standard- deviations among tests (n=3). 39

2.4 Discussion

In this study, we optimized parameters for growing and measuring Histoplasma yeast growth in

96-well microtiter plates. Differences in antifungal susceptibilities of Histoplasma yeasts cells compared to mycelia (Hage et al. 2011; Nakai et al. 2003; Kohler et al. 2000) raise the importance of standardizing methodologies for Histoplasma yeasts, the clinically relevant morphological form. Although CLSI has developed procedures for yeast-form fungal pathogens, these methods do not work for the culture of Histoplasma yeasts. Some studies on Histoplasma have adapted the CLSI macrobroth procedures for dimorphic fungi (Wheat et al. 2001; Wheat et al. 2006; Connolly et al. 1999; R. K. Li et al. 2000), but no optimized methodology for a microdilution assay has been standardized. Our optimized methods are summarized in Fig. 2.7 and differ from the CLSI procedures for yeast by the following: higher inoculum density (2 × 106 to 4 × 106 yeasts/mL compared to 500 to 2,500 yeasts/mL), longer growth time (4 to 6 days compared to 1 to 2 days) due to the much slower growth of Histoplasma, and growth in lower volume of media (100 µL compared to 200 µL) and in flat-bottomed wells (instead of U-bottom wells) to improve oxygen diffusion. Using these parameters, saturated growth densities of approximately 2.5 × 107 Histoplasma yeasts/mL were achieved. We suspect that reduced oxygen availability caused by growth in microtiter plate wells accounts for the lower densities than are typically achieved in culture tubes. Nonetheless, sufficient growth was present to reliably distinguish growth of Histoplasma over the inoculum density. We observed that improved consistency of growth of Histoplasma yeasts in microtiter plates was achieved if the initial inoculum was prepared from pregrown liquid cultures. Nevertheless, preparation of the initial inoculum by suspension of yeast colonies taken directly from solid BHI medium was sufficient for the microdilution assay if the yeasts were obtained from freshly streaked plates (4 to 8 days

40 old).

Figure 2.7 Summary of the optimized growth and assay parameters for microdilution testing of Histoplasma yeasts.

Inoculum size, growth medium, and growth conditions are listed for initiation of the tests in 96-well microtiter plates. Assays that can be used for determination of Histoplasma growth (culture turbidity, MTT reduction, and resazurin assays) provide both visual (qualitative) and spectrophotometric (quantitative) results. Further details of the endpoint assay methodologies are found in the text.

Measurement of culture turbidity by optical density at 595 nm provides an easy and efficient means of determining the Histoplasma yeast number, which facilitates high-throughput assays and antifungal susceptibility profiling. Daily OD595 measurements provide a kinetic view of

Histoplasma yeast growth. OD595 readings are less subjective than visual inspection of wells, and, furthermore, spectrophotometric quantitation provides data for partially, but not fully, inhibited growth. Measurement of light scatter due to turbidity differs between spectrophotometers; thus, the creation of standard curves for OD595 versus turbidity using

McFarland standards should be performed for individual plate reader machines to calibrate readings. Starting yeast cell densities of 2 × 106 to 4 × 106 yeasts/mL provide an optimal balance between length of growth period in microtiter plates and dynamic range for growth measurement. An inoculum of 2 × 106 to 4 × 106 yeasts/mL yields initial plate readings (or readings of fully inhibited yeast growth) near zero, and the yeast typically reach saturation in 4 to

5 days (Fig. 2.4), although the growth period is extended to 6 days for some slower growing strains. This results in a dynamic range over 30-fold between measurements of the inoculum and saturated growth.

Metabolic activity indicator dyes can also be used to quantify Histoplasma yeasts grown in microtiter plates. We show that both MTT- and resazurin-based endpoint assays are applicable to the Histoplasma microdilution assay. With both assays, color can be used as a qualitative indicator of yeast cell number, which is less subjective than the visual determination of turbidity.

Spectrophotometric measurement of formazan formation or fluorescence due to the conversion of resazurin to resorufin also provides more objective, quantitative results. The accurate determination of relative growth requires that the measurements for maximum numbers of

Histoplasma yeasts do not saturate the assay color or fluorescence development. For MTT, we determined that 0.5 mg/mL MTT was not limiting and that a reaction time of 3 h at 37°C maximized the dynamic range of the assay. For resazurin, 100 µM and a time of 90 min at 37°C were the nonlimiting substrate concentration and optimal assay reaction time before saturation, respectively. The MTT and resazurin assays both are particularly useful for measuring the growth of Histoplasma strains which exhibit clumped rather than dispersed growth in a liquid medium (e.g., strains of the NAm1 class or Latin American isolates), since these assays are based on cellular metabolism rather than on yeast particles which, when aggregated, can lead to variable turbidity readings.

Testing of multiple clinical Histoplasma isolates from diverse geographic locations demonstrates that the optimized microdilution assay is broadly applicable to Histoplasma isolates. Antifungal susceptibility testing of the isolates showed similar susceptibility profiles for fungicidal amphotericin B. NAm1, LAm, and the Panama isolates had slightly increased natural resistance to fluconazole compared with the NAm2 isolates, but all were still below the fluconazole breakpoints defined for Candida yeasts (Clinical and Laboratory Standards Institute 2012;

Fothergill et al. 2014). Yeasts from the LAm and Panama, but not the NAm2, phylogenetic groups are characterized by cell walls that have α-glucan polysaccharides in addition to the β- glucan polysaccharide (Domer 1971; Reiss 1977; Reiss et al. 1977; Edwards, Alore, and

Rappleye 2011; Edwards and Rappleye 2011), and these isolates showed increased resistance to caspofungin compared to the resistance of NAm2 isolates. This correlation between α-glucan and less sensitivity to β-glucan synthase inhibitors is intriguing, yet the mechanism remains undefined at this point. Nonetheless, it indicates that Histoplasma strains should not be

43 considered a homogeneous group with regard to the natural profiles of antifungal susceptibility.

In addition, testing of isolates from patients that had failed or relapsed following fluconazole treatment confirms the acquisition of fluconazole resistance in these isolates and shows the applicability of the microdilution assay to provide accurate information on antifungal susceptibilities to improve clinical management of histoplasmosis.

In conclusion, the optimized parameters for measurement of Histoplasma yeast growth in 96- well microtiter plates, either through optical density measurement or through endpoint colorimetric/fluorescence assays, fill a gap in the CLSI methodology for antifungal susceptibility testing of Histoplasma yeasts. It is hoped that these standardized methods will provide for more reliable profiling of clinical isolates of Histoplasma as well as accelerate high-throughput screening of compounds for new antifungals active against the pathogenic phase of Histoplasma

(Edwards, Kemski, and Rappleye 2013).

Chapter 3. Development of a novel method for target identification of the novel antifungal 41F5

3.1 Introduction

41F5 was initially identified as a potent antifungal through the screening of a 3,600-compound library (Edwards, Kemski, and Rappleye 2013). The library contained compounds with structural similarity to purines or purine analog scaffolds and was screened using a high throughput method of monitoring intramacrophage growth, the native environment of Histoplasma throughout the course of infection, by fluorescence of RFP-expressing Histoplasma yeast (Edwards, Kemski, and Rappleye 2013). After initial high-throughput screening at 10µM, the IC50 of compounds that reduced yeast growth by at least 75% were verified and those that demonstrated a dose dependent inhibition and IC50 of less than 5µM underwent secondary host cell toxicity screening

(Edwards, Kemski, and Rappleye 2013). Host cell toxicity screening was performed using LacZ- transgenic P388D1 macrophages with β-galactosidase. Production of β-galactosidase was directly proportional to the number of surviving macrophages and therefore used as a measure of macrophage survival (Edwards, Kemski, and Rappleye 2013). 41F5 was selected as the most promising compound as it has an MIC of 2 µM, an IC50 of 0.87 µM and low host cell toxicity

(Edwards, Kemski, and Rappleye 2013). 41F5 was also tested against several other pathogenic fungi, and was found to be effective against C. neoformans with a MIC of 1.25 µM, but not effective against Blastomyces (MIC>40 µM), Histoplasma’s closest pathogenic relative

(Edwards, Kemski, and Rappleye 2013).

Once determined that a compound is a promising novel drug candidate, these compounds are modified to increase commercial and medical benefits. This includes modifying or making derivatives of the original compound in order to increase drug-like characteristics such as potency, solubility, stability, and bioavailability. One common way to determine how to modify drug candidates is Structure-activity relationship (SAR) studies. These are used to determine key structures and functional groups required for activity and how changes to other functional groups on the compound alter chemical properties of that compound. A SAR analysis was determined for 41F5 but was not able to significantly improve overall upon the original structure (Khalil et al. 2015). Thus another way to examine how to modify a novel candidate drug was required. One method used in the development of novel antibiotics is called Rational Drug Design (RDD). This method can be used in combination with screens to identify how candidate drugs interact with target molecules in order to mathematically model how modifications to the compound will change the interaction with the target. This allows for targeted modifications to the compound that increase desired drug-like characteristics without disrupting drug-target interactions.

However, this method requires a target molecule to be known.

41F5 is a prime candidate for RDD once the target is known. As antifungal targets are severally limited (Goughenour and Rappleye 2016), and 41F5 is not thought to act via known antifungal targets (Edwards, Kemski, and Rappleye 2013) identifying a novel target would be a large step forward for the field. RDD would then be able to be used to modify 41F5 to increase its drug- like characteristics and potentially design novel compounds against the same target. We have utilized a dual genetics and novel co-fractionation biochemical approach to identification of the

46 target of 41F5. We generated and whole-genome sequenced 6 41F5-resistant C. neoformans lines to identify single nucleotide polymorphisms (SNPs) in potential gene targets. While no consistent individual target was determined in all strains, some candidates were identified. In addition, a novel co-fractionation approach was developed to identify proteins that co-fractionate with 41F5 as potential targets. Current work is ongoing to validate these potential candidates by determination if increased 41F5 is required to inhibit fungi when the candidate target molecule is overexpressed (Titov and Liu 2012).

3.2. Materials and Methods

3.2.1. Yeast strains and growth

H. capsulatum strains used in this study are listed in the Table 1 and were derived from the

WU15 strain, a uracil auxotroph derivative of G217B. H. capsulatum yeasts were grown in H. capsulatum-macrophage medium (HMM) For growth of uracil auxotrophs, HMM was supplemented with 100 µg/ml uracil. Yeasts were grown with continuous shaking (200 rpm) at

37°C. For growth on solid medium, HMM was solidified with 0.6% agarose and supplemented with 25 µM FeSO4. C. neoformans strains used in this study are listed in the Table 1 and were derived from the H99 wild-type strain. Saccharomyces cerevisiae strain S288C was used in this study. Yeasts were grown on YPD plates or RPMI liquid cultures. C. neoformans and S. cerevisiae were grown in liquid at with continuous shaking (200 rpm) at 30°C.

Table 3.1 Fungal Strains used in this study

Strain Genotype Other Designation1 Wild type Cryptococcus neoformans H99 serotype A Wild type Cryptococcus neoformans B3501 serotype D S288C Wild type Saccharomyces cerevisiae WT (ATCC204508) G217B Wild type NAm1 isolate WT (ATCC26032) WU15 G217B ura5-42Δ G217B ura5-42Δ zzz::pAM05 (URA5, OSU415 ACS2:6xHIS) ACS2 G217B ura5-42Δ zzz::pAM08 (URA5, OSU416 SAH1:6xHIS) SAH1 G217B ura5-42Δ zzz::pAM01 (URA5, OSU417 WRS1:6xHIS) WRS1 G217B ura5-42Δ zzz::pAM02 (URA5, OSU418 DED81:6xHIS) DED81 G217B ura5-42Δ zzz::pAM03 (URA5, OSU419 ERG10:6xHIS) ERG10 Generated through serial CNG1 41F5 resistant H99 isolate passaging in this study Generated through serial CNK2 41F5 resistant H99 isolate passaging in this study Generated through serial CNL3 41F5 resistant H99 isolate passaging in this study Generated through serile CN12 41F5 resistant H99 isolate passaging in this study Generated through serial CN15 41F5 resistant H99 isolate passaging in this study Generated through serial CN22 41F5 resistant H99 isolate passaging in this study Generated through serile CN Beta Fluconazole and 41F5 resistant H99 isolate passaging in this study 1gene designations: zzz::T-DNA: T-DNA integration at an undetermined chromosomal location URA5: orotate phosphoribosyltransferase ACS2: Acetyl-CoA synthetase SAH1: S-adenosyl-L-homocysteine hydrolase WRS1: Cytoplasmic tryptophanyl-tRNA synthetase DED81: Aspartyl/asparaginyl-tRNA synthetase ERG10: Ergosterol biosynthesis pathway, Acetyl-CoA C-acetyltransferase

3.2.2. Generation of Cryptococcus 41F5 resistant mutants

Independent lines of 41F5-resistant Cryptococcus yeasts were generated by serial passage of

Cryptococcus in increasing concentrations of 41F5. 1x105 cells/mL were grown in RPMI for 3 days at 30°C for each passage. As isolates became capable of proliferating in the starting concentrations of 41F5 (0.4 µM), the concentration of the drug was doubled. After 25 passages, isolates were challenged to determine a preliminary MIC for 41F5 (see 3.2.3). Passaged lines capable of growing in >16 µM 41F5 from passaging were selected for secondary testing. Single colonies were isolated and subsequently passaged without 41F5 selection to determine the stability of the resistance. Stably resistant lines were identified and re-tested to determine the

MIC to 41F5. Isolate lines with an MIC >64µM (>50-fold increase in resistance) were also tested for fluconazole resistance to identify those with specific resistance to 41F5 (see 3.2.3). Isolates with 41F5 resistance but fluconazole susceptibility were utilized for whole genome sequencing

(Table 3.3 see 3.2.4).

3.2.3. Antifungal susceptibility testing

In order to identify potential general antifungal resistance mechanisms like drug-efflux pumps, all 41F5 resistant mutants were screened for fluconazole reduced susceptibility to another antifungal (i.e., fluconazole). Fluconazole was prepared according to the CLSI reference standard M27-A3 (Rex and Clinical and Laboratory Standards Institute 2008). For IC50 and MIC determinations for Histoplasma yeasts, in a 96 well plate, a 2-fold dilution series of antifungal

49 drugs was added to Histoplasma yeasts at a density of 2-4x106 yeasts/mL and incubated at 37°C for 96-144 hours. Yeast growth was then assessed via turbidity at 595nm and/or endpoint resazurin reduction assays (Goughenour, Balada-Llasat, and Rappleye 2015). Briefly, 100 µM resazurin (7-hydroxy-10-oxidophenoxazin-10-ium-3-one) was added to each well, including control wells containing growth media without yeast cells, and the microtiter plates were incubated at 37°C. Resorufin (reduced resazurin; 7-hydroxyphenoxazin-3-one) production was quantified by measurement of fluorescence (530/25 excitation, 590/35 nm emission wavelengths) after 90 min using a Synergy2 plate reader (BioTek). Relative growth at each antifungal drug concentration was used to generate dose-response curves, and the IC50 value was determined by non-linear regression. MIC values were based on the minimum concentration of antifungal drug that prevented development of measurable turbidity or lack of resazurin fluorescence compared to growth in the absence of antifungal drug. For IC50 and MIC determinations for Cryptococcus yeasts the method was as above with the following modifications: yeast cells were inoculated at 1x104 cells/mL in RPMI and grown for 48hrs at

30°C, and resorufin was measured 3hrs after adding resazurin.

3.2.4. Whole genome sequencing of Cryptococcus mutants

Cryptococcus DNA from 41F5-specific resistant mutants was extracted as follows. 3mL cultures were grown overnight at 30°C, spun down, and the cell pellet frozen at -80°C overnight. The cell pellet was lyophilized then mechanical breakage with 0.5 µm diameter glass beads was preformed. CTAB buffer was added to samples and thoroughly mixed by vortexing, followed by incubation at 65°C for 30 minutes then cooling in an ice bath for 10min (Pitkin, Panaccione, and

Walton 1996). Phenol chloroform extraction was then preformed with the aqueous layer removed for isopropanol DNA precipitation. DNA was washed with 70% ethanol then resuspended in TE buffer. Samples were treated with RNase for 30 min at 37°C, then samples were cleaned using the NucleoSpin gDNA Clean-up kit (Macherey Nagel). Samples were then sent for whole genome sequencing. Multiplex sequencing using the Illumina HiSeq2500 platform was used to obtain genome coverage of 20-25x. SNPs were then identified for each isolate sequenced via comparison to the 41F5-sensitive parental line. SNPs were prioritized as those that result in single amino acid residue changes in coding regions of proteins (known proteins will be prioritized over hypothetical), followed by promoter SNPs.

3.2.5. Detection of 41F5 in cell lysate by HPLC-MS/MS

Histoplasma yeasts were grown to exponential phase (as described above) in 160 mL. 32µL of

25mM 41F5 was added to Histoplasma cultures for 8-12hrs to allow target proteins to bind the antifungal compound. Fungal cell lysates were prepared from the yeast cells by suspension in

PBS with 1X Protease Inhibitor (ingle Use Cocktail EDTA-Free; ThermoScientific) added and mechanical disruption with 0.5 um diameter glass beads. Lysates were kept on ice to prevent degradation. Debris was removed from the cellular lysate by centrifugation (10 minutes at

14,000 x g). Additional lysate preparations were as follows. To determine if native protein was required for 41F5 interaction Histoplasma lysates were heat denatured or degraded with proteinase K. Lysates were heat denatured by incubation at 95°C for 10 min. Lysates were degraded by incubation in 30mM TrisHCl (pH 8) with proteinase K treatment at 37°C overnight.

Additionally, Saccharomyces cerevisiae cells were grown with 41F5 and lysate prepared as previously described.

Separation of the lysates and 41F5 for subsequent target identification was performed by ultrafiltration filter co-fractionation followed by size-exclusion chromatography co-fractionation and detection of extracted 41F5 using LC-MS/MS by the procedure that follows (modified based on: Becher et al., 2017; Chan et al., 2012). Initial size fractionation of the Histoplasma capsulatum lysates was first performed using a series of ultrafiltration filter fractionations with molecular weight cut-off (MWCO) filters (VIVASPIN 2 PES Membrane Sartorius). The series of MWCO filters used were 300 kDa, 100kDa, 50kDa, 30kDa, then 10kDa. Subsequent fractionation was preformed by size exclusion chromatography (SEC) on an ÄKTA UPC-900 fast protein liquid chromatography (FPLC) system with P-920 pump and Frac-950 automated fraction collector (Amersham Biosciences) equipped with a Superdex 200 Increase 10/300 GL column (GE Lifesciences) where 40 fractions (500 µl each) were collected at 2 minute intervals in a mobile phase of 10mM phosphate buffered saline (PBS) at pH 7. The fractions were then chloroform extracted and the chloroform was evaporated using a CentriVap centrifugal concentrator (Labconco). Samples were resuspended in 60% acetonitrile:water (HPLC-grade;

Sigma Aldrich) and transferred to autosampler vials for LC-MS/MS detection of 41F5.

For detection of 41F5, Ultra performance liquid chromatography (UPLC) separation was perfomed with a 50 µl injection volume of the reconstituted fractions onto a BEC C18 column

(1.7µm 2.1 x100 mm; Waters) with VanGuard BEC C18 pre-column (1.7µm 2.1 x 5 mm;

Waters) thermostatted at 50 °C using a 5 min isocratic elution with a mobile phase consisting of

70% acetonitrile:water with 0.1% formic acid on a Vanquish UPLC equipped with autosampler and column compartment (Thermo). The UPLC was coupled to a TSQ Quantiva mass spectrometer (Thermo) through a heated electrospray ionization source (HESI) with a vaporizer and ion transfer tube temperature of 325 °C, sheath gas flow of 42 arbitrary units, auxiliary gas flow of 12 arbitrary units, sweep gas flow of 1 arbitrary unit, and spray voltage of 3.5 kV.

Detection was performed using multiple reaction monitoring (MRM) mode with a collision gas pressure of 1.5 mTorr following the transitions of 351 m/z to 241, 141, and 83 m/z using collision energies of 23.3 V, 43.6 V and 33.5 V respectively.

For proteomics work the aqueous phase from the chloroform extraction was split and both cold acetone precipitated and methanol precipitated to separate out proteins. The protein pellets were spun down (10 minutes at 14000 x g) and resuspended overnight in water. These were then run on a 10% SDS PAGE gel and slices were submitted for proteomic analysis. In addition, FPLC analysis on lysate was repeated and the same fractions that were directly identified as containing

41F5 were submitted for proteomics analysis without further modification to prevent complications from protein precipitation and re-solubilization as different methodologies can impact the protein composition seen in proteomic analysis and therefore bias results (Zellner et al. 2005).

3.2.6. Proteomics analysis

In gel digestion took place with sequencing grade-modified trypsin (Promega, Madison WI) added to SDS-PAGE gel slices for 6 hours at 37ºC. The peptides were then extracted from the

53 polyacrylamide with 50% acetonitrile and 5% formic acid. Non-gel samples were also trypsin digested overnight at 37 °C. Peptide concentration was determined by nanodrop (A280nm).

Orbitrap Fusion Capillary-liquid chromatography-nanospray tandem mass spectrometry

(Capillary-LC/MS/MS) of protein identification was performed on a Thermo Scientific orbitrap

Fusion mass spectrometer equipped with an EASY-Spray™ Sources operated in positive ion mode. Samples were separated on an easy spray nano column (PepmapTM RSLC, C18 3µ

100A, 75µm X150mm Thermo Scientific) using a 2D RSLC HPLC system from Thermo

Scientific. Each sample was injected into the µ-Precolumn Cartridge (Thermo Scientific) and desalted with 0.1% Formic Acid in water for 5 minutes. Mobile phase A was 0.1% Formic Acid in water and acetonitrile (with 0.1% formic acid) was used as mobile phase B. Flow rate was set at 300nL/min. MS/MS data was acquired with a spray voltage of 1.7 KV and a capillary temperature of 275 °C is used.

The scan sequence of the mass spectrometer was based on the preview mode data dependent

TopSpeed™ method: the analysis was programmed for a full scan recorded between m/z 400 –

1600 and a MS/MS scan to generate product ion spectra to determine amino acid sequence in consecutive scans starting from the most abundant peaks in the spectrum in the next 3 seconds.

To achieve high mass accuracy MS determination, the full scan was performed at FT mode and the resolution was set at 120,000. The AGC Target ion number for FT full scan was set at 2 x

105 ions, maximum ion injection time was set at 50 ms and micro scan number was set at 1.

MSn was performed using ion trap mode to ensure the highest signal intensity of MSn spectra using both CID (for 2+ and 3+ charges) and ETD (for 4+-6+ charges) methods. The AGC Target

54 ion number for ion trap MSn scan was set at 1000 ions, maximum ion injection time was set at

100 ms and micro scan number was set at 1. The CID fragmentation energy was set to 35%.

Dynamic exclusion is enabled with a repeat count of 1 within 60s and a low mass width and high mass width of 10ppm.

Sequence information from the MS/MS data was processed by converting the .raw files into a merged file (.mgf) using MSConvert (ProteoWizard) and were searched using Mascot Daemon by Matrix Science version 2.5.1 (Boston, MA) and the database searched against an in-house generated Histoplasma capsulatum genome database fragmented into 10 kb contigs which contained 5196 total database entries. This database was constructed from the draft assembly of the Histoplasma G186A genome (December 2004 assembly, Washington University Genome

Sequencing Center). The mass accuracy of the precursor ions were set to 10ppm, accidental pick of 1 13C peaks was also included into the search. The fragment mass tolerance was set to 0.5 Da.

Four missed cleavages for the enzyme were permitted. A decoy database was also searched to determine the false discovery rate (FDR) and peptides were filtered according to the FDR. The significance threshold was set at p<0.05 and bold red peptides is required for valid peptide identification. A decoy database was also searched to determine the false discovery rate (FDR) and peptides were filtered at a FDR of 1%. Proteins identified with at least 2 unique peptides were considered and reliable identification. Any modified peptides are manually checked for validation.

3.2.7. Generation of Histoplasma strains overexpressing His-tagged potential target proteins

Candidate target genes identified from the co-fractionation studies were amplifed by PCR (Table

3.2) from wild-type H. capsulatum G217B genomic DNA and cloned into the H. capsulatum expression vector pAG38, placing the gene under transcriptional control of the H2B constitutive promoter and fusing the potential target gene to a C-terminal hexahistidine tag. Overexpression plasmids were transformed into H. capsulatum WU15 yeasts by Agrobacterium-mediated transformation (Zemska and Rappleye 2012) and transformants selected by uracil prototrophy.

Transformants were screened by immunoblotting of culture filtrates and cellular lysates for the hexahistidine tag (GnScrpt A00186 to 6xhis).

Table 3.2 Primers used in this study

Primer Primer Sequence 5' to 3' Direction1 ACS2-1 ATGTCTGACGGTCCAGTTACGC Forward ACS2-2 CTTCTTTGAAGCGTGGTAAACT Reverse SAH1-1 ATGTCAACTCACCAAACATTAACC Forward SAH1-2 CATTTCAGCCTTGAAAGGGCCA Reverse DED81-1 ATGGCCGCCACCACTTCAAT Forward DED81-2 TGGTGTACACCGTCCAGTAAAG Reverse WRS1-1 ATGACTACCCCGCCTCCTGG Forward WRS1-2 ATTTTCATCATGACGTTGTTTCG Reverse ERG10-1 ATGCAGTGCTCTCTCAGAACAG Forward ERG10-2 GTCGACCTTTTCCACTCTCTG Reverse 1Direction relative to gene transcription

3.3 Results

3.3.1 Identification of SNPs from Cryptococcus 41F5 resistant mutants

Serial passaging of Cryptococcus yeasts was used to generate 41F5-resistant mutants as

Histoplasma yeasts quickly start converting from the yeast phase to hyphae even in subinhibitory concentrations of 41F5. The conversion of yeast to mycelia alters the antifungal susceptibility of Histoplasma, preventing it from being a good candidate for serial passaging

(Goughenour, Balada-Llasat, and Rappleye 2015). Using this approach 6 independent lines were generated that were resistant to 41F5 but susceptible to fluconazole (Table 3.3). Fluconazole was chosen to monitor for potential non-specific drug efflux pump-mediated resistance mechanisms because they are a well-described mechanism of azole resistance in Candida (Kanafani and

Perfect 2008). Following isolation of a clonal line, the genomes of these lines were sequenced and single nucleotide polymorphisms (SNPs) were identified. Of the SNPs, particular attention was focused on missense, but not non-sense, mutations in the coding sequence (CDS) (as inhibition of fungal growth by 41F5 suggests the target is essential) and SNPs in regions upstream of genes, which might result in increased gene expression (e.g., increased drug-target production). 12 high priority candidate SNPs were identified (Table 3.4) with 2 of immediate interest (Erg20 in line L3 and a serine/threonine kinase in line K2).

Table 3.3 Antifungal susceptibility of Cryptococcus lines to 41F5 and Fluconazole

41F5 Fluc Fluc Fluc Strain (MIC) (MIC) (IC50) (logIC50) WT 2 2.5 1.2 0.1 CNG1 ≥64 5 3.2 0.5 CNK2 ≥64 5 2.6 0.4 CNL3 ≥64 10 4.4 0.6 CN12 ≥64 5 3.0 0.5 CN15 ≥64 2.5 1.7 0.2 CN22 ≥64 5 1.3 0.1 CN Beta ≥64 20 17.0 1.2

41F5 results are reported as Minimum Inhibitory Concentrations (MICs) as resistant lines were not inhibited at any concentration tested and therefore inhibitory concentrations of 50% (IC50) were not able to be calculated. Fluconazole results are reported as MICs and IC50. IC50s are calculated by linear regression from dose response curves. logIC50 is used to measure asymmetrical confidence intervals. Each strain was tested in biological triplicate.

Table 3.4 SNPs Identified in 41F5-resistant Cryptococcus most likely to contribute to resistance

Line Locus Encoded Function Mutation CNAG_07784 Promoter CNG1 (-23) hypothetical mutation CNAG_07784 Promoter CNG1 (-30) hypothetical mutation 4 amino acid CNK2 CNAG_03376 hypothetical insertion CNAG_08026 Promoter CNK2 (-404) hypothetical mutation CNK2 CNAG_01905 Ser/Thr kinase K>T CNL3 CNAG_02084 Erg20: farnesyl diphosphate synthase A154V homolog of RPN7 (proteasome CN12 CNAG_06899 regulatory subunit) Y>D hypothetical cytoplasmic protein (non- CN12 CNAG_02079 essential gene) H>Y CN12 CNAG_01699 Histone deacetylase 1/2 V53A 8 amino acid CN15 CNAG_05333 Hypothetical protein insertion Frameshift CN22 CNAG_07938 Hypothetical protein deletion

3.3.2 Co-fractionation approach

As a complementary approach to genetics, candidate drug targets were identified through a biochemical approach for proteins that could interact with 41F5. Initial attempts to generate a bioactive, tagged 41F5 compound that could be used for column purification were unsuccessful as the molecule could not be modified at any substituent positions without losing activity (Ishita et al., 2018). We therefore developed an approach to identify cellular proteins that co-fractionate with 41F5 as potential drug targets and UPLC-MS/MS (Ultra-high performance liquid chromatography-mass spectrometry/mass spectrometry) strategy for target identification (Chan et al. 2012; Becher et al. 2017). As 41F5 and target-interacting 41F5 should size fractionate differently (Fig. 3.1) we should be able to use size fractionation to identify which proteins co- fractionate with 41F5. To track 41F5 in cellular fractions, we exploited a mass-spectrometry based procedure to enable high sensitivity and specificity in detection. Briefly, to identify 41F5 interacting proteins, fungi are treated with 41F5 to allow target proteins to bind the antifungal compound and then cellular lysates are prepared. Lysates were fractionated by size (molecular weight cutoff filters then fast protein liquid chromatography (FPLC)) then cholorform extraction of 41F5 was preformed. UPLC-MS/MS was preformed on the chloroform fractions to identify

41F5-positive fractions are identified. The corresponding aqueous fractions along with non- chloroform extracted fractions identified as 41F5 positive are used for proteomic analysis by mass spectrometry to identify proteins that co-fractionate with 41F5. The aqueous fraction from the chloroform extraction provides a matched sample to the 41F5-detected fraction, which is the most accurate way to show 41F5-interacting proteins. However, the level of manipulation of the sample can lead to protein loss and subsequent gaps in the proteomics analysis. Therefore

60 proteomics were also preformed on fractions without further processing to ensure no proteins are lost. These two resulting lists are combined into a single proteomics list for candidate targets.

Figure 3.1 Schematic for the Co-fractionation approach to 41F5 detection.

Free 41F5 and target bound 41F5 will size fractionate differently. Some of every fraction is held for proteomic analysis. The rest of the fraction undergoes chloroform extraction of 41F5. The aqueous layer contains precipitated proteins and is submitted for proteomic analysis as well. The chloroform layer is dried and resuspended for HPLC-MS/MS detection of 41F5 in each fraction. Fractions identified as 41F5 positive are used for proteomic analysis.

3.3.3 Validation and Optimization of the Co-fractionation approach

To track the 41F5 compound, the UPLC retention time and MS/MS pattern for 41F5 was used.

Using only 41F5, a characteristic MS/MS pattern was then determined with a primary 351m/z

MS1 peak fracturing into three diagnostic MS2 peaks at 241m/z, 141m/z and 83m/z (Fig 3.2 A).

By UPLC (C18 column), the retention time of 41F5 was determined to be 1.90 min. This retention time and MS/MS peak profile was used to identify the 41F5 molecule in lysate fractions.

For initial size fractionation, we used ultrafiltration of lysates through different molecular weight cutoff filters (MWCO). This process separated the lysate into broad size ranges of >300kDa,

300-100kDa, 100-50kDa, 50-30kDa, 30-10kDa, and <10kDa. In the Histoplasma lysate from yeasts incubated without 41F5, we do not detect any 41F5 (Table 3.5). Passage of the 41F5 drug through the size filters in the absence of cells revealed it only in the high molecular weight sample (>300kDa range; Table 3.5). We hypothesize this to be the result of the largely hydrophoblic compound forming large aggregates that cannot pass through the 300 kDa MWCO filter. Lysates prepared from Histoplasma yeasts incubated with 41F5 overnight had 41F5 present in the >300 kDa fraction, but also some 41F5 in the 300-100kDa fraction (Table 3.5).

We interpret this as the 41F5 in the >300 kDa fraction representing free drug that did not bind any target within the yeast cell while the 41F5 present in the 100-300 kDa size was bound to a cellular component that fractionates in that size, bring the drug with it. The molecule to which the 41F5 molecule binds is likely a protein or protein complex between 100-300 kDa as the 41F5 molecule is not carried into the 300-100kDa range if the lysate is treated with heat or Proteinase

K before running through the MWCO filters (Table 3.5). The interation is specific to a

Histoplasma protein as incubation of Sachromyces cerevisiae (an organismnot inhibited by

41F5) with the drug and fractionation of cellular lysates prepared from these yeasts did not result in 41F5 being found in the 100-300 kDa fraction.

Figure 3.2 Validation and optimization of 41F5 detection by UPLC-MS2.

Relative diagnostic peaks at 241m/z, 141m/z and 83m/z of 41F5 generated from further fractionation of the initial 351m/z compound (A). UPLC-MS2 detection of 41F5 showing a retention time from the UPLC (x-axsis) at 1.90 minutes and the relative detection of 241m/z, 141m/z and 83m/z size molecules (B). Acetonitrile only negative control (left) showing background noise detection and 41F5 (right) detection is a clean strong peak. 65

Table 3.5 Presence of 41F5 in different size fractions

300- Sample <100kDa 100kDa >300kDa

Drug only - - ++

Histoplasma lysate only - - -

Drug + Histoplasma lysate - + ++

Drug + Heat-denatured Histoplasma lysate - - ++

Drug + Proteinase K-Treated Histoplasma lysate - - ++

Drug + S. cerevisiae lysate - - ++ ++ indicates a large peak of 41F5 detected; + indicates a peak detected; - indicates no peak detected

3.3.4 Identification of 41F5 Co-fractionating Proteins

The cellular lysate fraction from the 300-100kDa fraction was further size fractionated by FPLC to reduce the number of potential 41F5-interacting proteins. Each of the resulting fractions was tested for the presence of 41F5 (Fig. 3.3A). 41F5 was seen in the fractions 5-11 as part of the solvent front where molecules that do not separate in the range of the column (molecular weights

(Mr ) from 10,000 to 600,000) can be seen. This is likely free drug that has dissociated from the target over the course of the experiment forming aggregates too large for the column to separate.

In the fractions within separation range of the column, one clear 41F5 peak is seen in fraction 15

(Fig. 3.3A and 3.3B). Fraction 15, along with a non 41F5-containg control fraction (fraction 26) to rule out proteins that do not interact or cleanly fractionate, were sent for proteomic analysis by

LC-MS/MS. 194 proteins were identified in the 41F5-containing fraction that had a Quantitative

Value (Normalized Total Spectra) of greater than 10 (Supplemental File 1). The co- fractionating proteins were further prioritized as potential 41F5-interacting proteins as those that would be predicted to be required for viability (Table 3.6). This criterion was used as proteins in very small amounts were unlikely to be able to bind enough drug for detection. In addition as the drug is inhibitory the target is likely to be essential. All potential target genes were annotated by

BLAST analysis to identify homologs in three common fungal organisms S. cerevisiae, C. albicans and A. nidulans and the protein’s predicted essentially determined from analysis in these model genetic organisms.

Figure 3.3 Detection of 41F5 in FPLC fractions.

(A) Detection of the characteristic 241 m/z (blue) and 141 m/z (red) MS2 peaks in each collected fraction from UPLC analysis. (B) Protein concentration as determined by absorbance (280nm) in each fraction, with the 41F5 containing peak (15) and negative control peak (26) selected for proteomics indicated by red arrows.

Table 3.6 Proteins identified as co-fractionating with 41F5

Histoplasma Gene Homolog ID Annotation Histo217_09243 ACS2 Acetyl-CoA synthetase Histo217_09046 SAH1 S-adenosyl-L-homocysteine hydrolase ATP domain Histo217_03917 WRS1 Cytoplasmic tryptophanyl-tRNA synthetase Histo217_00804 DED81 Aspartyl/asparaginyl-tRNA synthetase Histo217_05335 ERG10 Ergosterol biosynthesis pathway Histo217_00586 GUS1 Glutamyl-tRNA synthetase Histo217_07910 CDC60 Leucyl-tRNA synthetase Histo217_08619 SES1 Seryl-tRNA synthetase Histo217_02253 ALA1 Alanine-tRNA ligase Histo217_03699 KRS1 Lysyl-tRNA synthetase Histo217_07818 THS1 Threonyl-tRNA synthetase Histo217_06011 YNL247W Cysteinyl-tRNA synthetase Histo217_03385 VAS1 Valyl-tRNA synthetase Histo217_07051 ILS1 Isoleucyl-tRNA synthetase Histo217_05386 PYK1 Pyruvate kinase Histo217_08561 GRS1 Glycyl-tRNA synthetase Histo217_05806 PKC1 Protein kinase C Histo217_07968 RSE1 MMS1_N and CPSF_A domain-containing Vesicle coat complex COPII, subunit SEC23 Histo217_02917 SEC23 [Intracellular trafficking and secretion] Histo217_05835 RPN1 26S proteasome regulatory complex component Histo217_01025 GFA1 Glucosamine:fructose-6-phosphate aminotransferase Protein similar to ribosomal protein Histo217_01193 RPL30 L7Ae/L30e/S12e/Gadd45 family protein Histo217_02798 BMH1/2 14-3-3 domain-containing Histo217_04092 ADE13 Adenylosuccinate lyase Histo217_06798 ALD Betaine aldehyde dehydrogenase Histo217_08045 KRE5 UDP-glucose:glycoprotein glucosyltransferase Protein containing domains CDC37_N, CDC37_M, Histo217_02640 CDC37 and CDC37_C Protein containing domains Ube1_repeat1, E1_4HB, Histo217_00835 UBA1 Ube1_repeat2, and GST_C_Omega_like Histo217_04309 RFA1 Replication factor-a protein 1 Histo217_07128 BMH1/2 14-3-3 domain-containing ATPase; HSPA5-like_NBD domain-containing Histo217_03829 KAR2 protein Histo217_04542 RPS9B/A 40S ribosomal protein S9 Histo217_07720 RPS31 Ubiquitin-like (Ubl) domain; Ribosomal protein S27a

3.3.5 Identification of the target of 41F5

To determine the target of 41F5 from the candidate proteins from Histoplasma and

Cryptococcus, we will use correlation of decreased susceptibility to the drug upon overexpression of the target protein; overexpression of the target should increase the amount of drug needed to inhibit all of the target protein which will increase the observed MIC (Titov and

Liu 2012). For the Cryptococcus resistant strains we will use Agrobacterium mediated transformation of C. neoformans (B3501) (Fu et al. 2011; Walton, Idnurm, and Heitman 2005).

This work is on-going with plasmid optimization to ensure high levels of expression of the target gene via promoter selection taking place. In Histoplasma Agrobacterium-mediated transformation is being used to introduce transgenes in which the candidate 41F5-interacting proteins are fused to a C-terminal hexahistidine-tag (6xHis) and expression controlled by the

H2B promoter, which drives high level, constitutive expression. The inclusion of the C-terminal tag facilitates confirmation of transgene overexpression by immunoblotting of lysates and affinity purification (Garfoot et al. 2016) of the protein for future biochemical characterization.

Preliminary susceptibility testing of 5 candidate proteins (ACS2, SAH1, WRS1, DED81, and

ERG10) has been preformed using two independent transformants for each gene (Table 3.7).

None show an increased IC50. Work is continuing on the others in the list (Table 3.6). Once a target is identified by an increase in resistance, final validation can occur. For final validation, candidate proteins will be purified from cellular lysates using metal-affinity chromatography (via the 6xHis tag) and testing for direct interaction (via the mass-spectrometry co-fractionation with purified protein) and inhibition of any predicted biochemical function by 41F5. 41F5 derivative

70 compounds from the SAR study that have lost antifungal activity will be used as negative control.

Table 3.7 IC50 of Candidate 41F5 Target Overexpressions

IC50 logIC50 Strain (µM) WT 0.9 -0.02 ACS2 0.7 -0.1 SAH1 1.0 -0.00002 WRS1 0.9 -0.03 DED81 0.8 -0.1 ERG10 0.6 -0.2

Results are calculated by linear regression from dose response curves of two independent transformations of each overexpression strain. logIC50 is used to measure asymmetrical confidence intervals.

3.4 Discussion

Antifungal options for treating fungal disease remain limited, particularly for Histoplasma and the thermal dimorphs (Goughenour and Rappleye 2016; Rauseo et al. n.d.). As eukaryotes, fungi have very similar proteins to those of the mammalian host, thereby limiting fungal-specific targets. In this study we combined a standard genetic approach with a co-fractionation approach to attempt to identify the molecular target of 41F5. We have developed a highly sensitive and specific assay to monitor the presence of 41F5 in biological samples that enabled the biochemical fractionation approach. One advantage of his is we were able to use the 41F5 drug

71 without having to modify the compound in any way, which could alter its activity. A second advantage it holds over radiolabeling the drug to track it is that the approach does not have to be modified for each drug (i.e., by labeling individual atoms during synthesis).

Our study identified several potential target proteins through our genetic approach.

Unfortunately, due to our small sample size, no consistent SNP pattern was found among multiple resistant lines. We initially investigated SNPs as the method of mutation as amino acid changes in the target protein are a direct way to determine if the protein interacts with a drug.

Our candidate list is limited by the identification of several hypothetical proteins, which makes speculation on their essentiality difficult. Of the few point mutations we identified, the ergosterol biosynthesis protein (Erg20) and the serine/threonine kinase are of greatest interest. Ergosterol biosynthesis is the target of the azole class of antifungals, although the azoles target the Erg11 gene (Cowen et al. 2015). Nonetheless, as ergosterol is essential for fungal viability and relatively specific, inhibition of the ergosterol pathway is an attractive target. The serine/threonine kinase is a attractive candidate as the original screen for 41F5 was comprised of purines or purine analog scaffolds therefore it would be reasonable that enzymes such as kinases that bind ATP would be potential targets. However, this specific serine/threonine kinase is not predicted to be essential in Cryptococcus questioning whether it is a direct target of 41F5 (Lee et al. 2016). This highlights a problem with the genetic approach to target identification. Sometimes the mutations that generate resistance are not in the target of the drug itself. For example, in

Candida albicans mutations in the transcription factor that regulates Erg11, UPC2, or mutation in the stress response pathway can also lead to azole resistance (Cowen et al. 2015). Therefore, analysis of the results from whole genome sequencing is often difficult to interpret. In addition,

72 chromosomal abnormalities such as copy number can lead to antifungal resistance (Cowen et al.

2015). These are not detected in SNP calling and add another layer of analysis that can be preformed. Currently we are still working on testing the current potential targets from the SNP analysis; however a chromosomal analysis of the mutants may still be preformed on the data from the whole genome sequencing in the future.

This study also developed a co-fractionation method for identifying drug targets through their potential biochemical interaction with the 41F5. The method showed that 41F5 interacts with a protein target in the 100-300kDa size range. Further fractionation and prioritization identified 34 likely protein target candidates. The candidate list contains several interesting proteins. Acetyl coenzyme A (acetyl-CoA) synthetase (ACS2) is the proposed target of the antifungal AR-12

(Koselny et al. 2016) and was therefore a high priority candidate to test in our overexpression system. However, overexpression of ACS2 did not confer any resistance to 41F5. Preliminary overexpression tests of Erg10 however, did not confer any 41F5 resistance. The other clear pattern seen in this proteomic dataset is the high numbers of tRNA synthetases. Although not currently seen in fungi as drug targets, tRNA synthetases inhibitors are promising antiparasitic drugs (Pham et al. 2013). This holds promise that a similar inhibitor could work in fungi as they are both eukaryotic pathogens. While overexpression of the first two tested tRNA synthetases genes WRS1 and DED81 (tryptophan and aspartate respectively) did not show any resistance, there are 10 more that are still being tested. No other large-scale patterns were seen in the proteomics data.

Very little correlation was seen between the genetics and co-fractionation approaches. No target gene was identified in both sets and the only pathway overlap was in the ergosterol biosynthesis pathway with Erg20 identified in the Cryptococcus SNP analysis of 41F5-resistant mutants and

Erg10 co-fractionating with 41F5 in Histoplasma lysates. However, it is difficult to predict that

41F5 inhibits two separate enzymes within the same pathway. As the Erg10 gene in

Histoplasma does not confer resistance, this is not likely the target in both organisms. Therefore, there are two possible reasons why we do not see overlap in the data sets: the two organisms do not have the same target or the genetics approach only identified indirect resistant mutations in

Cryptococcus. As to the first reason Histoplasma and Cryptococcus are members of two different phyla of fungi (Ascomycota and Basidiomycota respectively) and therefore have a huge amount of evolutionary divergence between them. To address this concern, the co-fractionation approach could be used to generate a proteomics list for Cryptococcus 41F5 co-fractionating proteins.

Initial attempts at this, however, showed that 41F5 in Cryptococcus lysates does not fractionate in the same size range as it does in Histoplamsa lysates. Alternatively, the genetics approach may have identified indirect factors in Cryptococcus that confer resistance not through changes in the target itself but through other interacting proteins or pathways. Therefore overexpressing the Cryptococcus homolog of the Histoplasma target could lead to confirmation in

Cryptococcus. This highlights why the development of this new method is such a benefit to the field of drug discovery. This method is based on direct interaction of target and drug avoiding off-target resistance mechanisms. It also can be adapted for any organism, allowing for use in target identification in any kind of drug development.

Chapter 4. Investigation into the O-linked mannosylated proteome of Histoplasma capsulatum

4.1 Introduction

O-linked glycosylation is one of the two major types of glycosylation found in fungi. It commonly consists of a linear series of mannose residues attached at serines and threonines and is performed by a relatively simple pathway (Loibl and Strahl 2013). The initial mannose attachment is performed by protein mannosyltransferases (PMTs) with subsequent mannose additions by the MNT family of proteins, both of which have been characterized in C. albicans

(Loibl and Strahl 2013; Munro et al. 2005). Mutants of Histoplasma in three PMT proteins

(Pmt1, Pmt2, and Pmt4) and the only identified Histoplasma MNT (Mnt1) were previously generated and examined for their role in virulence (Garfoot et al. 2018). Confirmation of the lack of O-linked mannosylation in the mutants was performed by observing molecular weight shifts in Cfp4, a protein with a mucin-like domain that is predicted to be heavily o-linked mannosylated. The Cfp4 protein of the pmt2 mutant has a reduction in mannose content

(Garfoot et al. 2018), but Sod3, a protein without predicted O-linked mannosylation, is the same molecular weight in mutants and wild type strains (Garfoot et al. 2018). In addition, mass spectrometry was used to directly show that pmt2 mutants had reduced mannosylation on purified Cfp4 proteins (Garfoot et al. 2018). In vivo mouse assays show virulence defects in the pmt1, pmt2, and mnt1 mutants (Garfoot et al. 2018). These results are consistent with reports that in C. albicans the two MNT proteins are necessary for adhesion and virulence (Munro et al.

2005). To explain the virulence defect seen in mice, our lab has worked on characterizing the 75 pmt2 mutant extensively, and have discovered that the mutant is temperature sensitive, which leads to the loss of virulence (Garfoot et al. 2018). This work attempts to expand upon our understanding of O-linked mannosylation in Histoplasma in order to identify mannosylated proteins and determine which proteins play a role in temperature sensitivity and therefore virulence. To do this we performed a preliminary bioinformatics search for O-linked mannosylated proteins, and identified some proteins with mucin-like domains (for example,

Cfp4); however, there is no true consensus sequence for O-linked mannosylation, which significantly hindered this approach. In addition, we are working on developing a novel approach to purification of O-linked mannosylated proteins via biotinylation and then identification by mass spectrometry. Initial attempts and differentiation of O-linked mannosylated proteins by mass spectrometry alone were unsuccessful.

4.2 Materials and Methods

4.2.1 Histoplasma growth

Histoplasma strains used in this study were derived from wild-type strain G217B (ATCC

26032) (Table 4.1) and were grown on HMM (Worsham and Goldman 1988). The uracil auxotrophs of Histoplasma were grown on/in HMM supplemented with 100 ug/mL uracil. For maintenance, cultures were grown at 37 °C under 5% CO2/95% air with shaking at 200 rpm for liquid cultures. Solid media was prepared by addition of 0.6% agarose and 25 µM FeSO4.

Table 4.1 Histoplasma strains used in this study Strain Genotype Other Designation1 G217B Wild type NAm1 isolate WT (ATCC26032) WU15 G217B ura5-42Δ OSU194 G217B ura5-42Δ zzz::pAG21 (G418R,GFP) RNAi sentinel background G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU379 zzz:pED02 (URA5, gfp-RNAi) GFP-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU380 zzz:pKG22 (URA5, gfp:CRH1-RNAi) CRH1-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU381 zzz:pKG16 (URA5, gfp:ENG3-RNAi) ENG3-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU382 zzz:pKG17 (URA5, gfp:PDI-RNAi) PDI-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU383 zzz:pED02 (URA5, gfp-RNAi) GFP-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU384 zzz:pKG23 (URA5, gfp:BGL5-RNAi) BGL5-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU385 zzz:pKG24 (URA5, gfp:CRR1-RNAi) CRR1-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU386 zzz:pED02 (URA5, gfp-RNAi) GFP-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU387 zzz:pKG18 (URA5, gfp:GH-RNAi) GH-RNAi G217B ura5-42Δ zzz::pAG21 (G418R,GFP) OSU388 zzz:pKG19 (URA5, gfp:CTS4-RNAi) CTS4-RNA1 1gene designations: zzz::T-DNA: T-DNA integration at an undetermined chromosomal location URA5: orotate phosphoribosyltransferase apt3: aminoglycoside phosphotransferase (G418 resistance) GFP: green-fluorescence protein CRH1: GH16 fungal CRH1 transglycosylase; predicted role in cell wall architecture, required for the transfer of chitin to 1,6-beta-glucan in the cell wall in S. cerevisiae ENG3: GH16 fungal Lam16A glucanase PDI: Protein Disulfide Isomerase (PDIa) family, YbbN superfamily neg reg of GroEL BGL5: Beta-glucosidase 5, Glycosyl hydrolase family 1 domain CRR1: GH16 fungal CRR1 transglycosylase; Spore specific glycosidase involved in spore wall assembly during sporulation and potentially involved in copper import in S. cerevisiae. GH: Glycosyl hydrolases family 15 (Glucoamylase (glucan1,4-alpha-glucosidase)), C-terminal CBM20 (carbohydrate-binding module, family 20) domain CTS4: glycosyle hydrolase 18 domain; chitinase

4.2.2. Proteomics analysis

Supernatants of saturated Histoplasma pmt2 (RNAi) knockdown, mnt1 (RNAi) knockdown and wild type Histoplasma strains in biological triplicate were collected and concentrated 100X.

Samples of the concentrated supernatants were treated overnight with PNGase F to remove N- linked glycosylation (Holbrook et al. 2014). Proteins were separated by electrophoresis 1(0%

SDS-PAGE gel) and gel slices were submitted for trypsin digestion and proteomic analysis.

In gel digestion took place with sequencing grade-modified trypsin (Promega, Madison WI) added to SDS-PAGE gel slices for 6 hours at 37ºC. The peptides were then extracted from the polyacrylamide with 50% acetonitrile and 5% formic acid. Non-gel samples were also trypsin digested overnight at 37 °C. Peptide concentration was determined by nanodrop (A280nm).