Molecular Tools for the Study of Fungal Aerosols

Molecular tools for the study of fungal aerosols

Thèse

Hamza Mbareche

Doctorat en microbiologie Philosophiæ doctor (Ph. D.)

Québec, Canada

Molecular tools for the study of fungal aerosols

Thèse

Hamza Mbareche

Sous la direction de :

Caroline Duchaine, directrice de recherche Guillaume Bilodeau, codirecteur de recherche Université Laval

Résumé

Depuis le développement rapide des méthodes de séquençage à haut débit (SHD) en écologie moléculaire, les moisissures ont eu moins d’attention que les bactéries et les virus, en particulier dans les études de bioaérosols. Les études d'exposition aux moisissures dans différents environnements sont limitées par les méthodes de culture traditionnelles qui sous- estiment le large spectre de moisissures pouvant être présentes dans l'air. Bien que certains problèmes de santé soient déjà associés à une exposition fongique, le risque peut être sous- estimé en raison des méthodes utilisées. L’application du séquençage à haut débit dans des échantillons de sol par exemple a permis de mieux comprendre le rôle des moisissures dans les écosystèmes. Cependant, la littérature n'est pas clair quant à la région génomique à utiliser comme cible pour l'enrichissement et le séquençage des moisissures. Cette thèse vise à déterminer laquelle des deux régions universellement utilisées, ITS1 et ITS2, convient le mieux pour étudier les moisissures dans l’air. Durant le développement de la méthode moléculaire, un autre défi, touchant la perte de cellules fongiques lors de la centrifugation d'échantillons d'air liquide à des fins de concentration, s’est rajouté. Ainsi, cette thèse décrit une nouvelle méthode de filtration pour remédier à la perte due à la centrifugation. Ces deux objectifs représentent la première partie de la thèse qui se concentre sur le développement de méthodes: le traitement des échantillons d’air avant extraction de l’ADN et la meilleure région à cibler avec la méthode SHD. La deuxième partie consiste à appliquer la méthodologie développée pour caractériser l'exposition aux moisissures dans trois environnements de travail différents: le compost, la biométhanisation et les fermes laitières. Les résultats montrent que la région d’ITS1 a surpasser ITS2 en couvrant davantage de diversité dans les bioaérosols. En raison de profils taxonomiques complémentaires, l'auteur de la thèse suggère d'utiliser les deux régions pour couvrir la plupart des taxons lorsque la taxonomie constitue le principal intérêt de l'étude. Cependant, ITS1 devrait être le premier choix dans les autres études, principalement en raison de la grande diversité et de la similarité des profils taxonomiques obtenus par l’approche métagénomique et l’approche ciblant ITS1. De plus, la nouvelle approche de filtration proposée constitue une meilleure alternative pour compenser la perte fongique due à la centrifugation. Ensemble, ces méthodes ont permis une meilleure description de l’exposition aux moisissures en milieu professionnel.

iii

Abstract

Since the rapid development of high-throughput sequencing methods in molecular ecology, fungi have been the underdogs of the microbial world, especially in bioaerosol studies. Particularly, studies describing fungal exposure in different occupational environments have been limited by traditional culture methods that underestimate the broad spectrum of fungi present in the air. There are potential risks in the human inhalation of fungal spores in an occupational scenario where the quantity and diversity of fungi is high. Although some health problems are already known to be associated with fungal exposure in certain work environments, the risk may be underestimated due to the methods used. Applying high-throughput sequencing in soil samples has helped the explanation of the fungal role in ecosystems. However, the literature is not decisive in terms of the genomic region to use as target for the enrichment and sequencing of fungi. The present thesis deals with the challenge of determining which region from the two universally used regions, ITS1 and ITS2, is best suited for study of fungal aerosols. In tandem with this challenge came another of addressing the loss of fungal cells during the centrifugation of liquid impaction air samples for purposes of concentration. This thesis describes a new filtration-based method to circumvent such losses during centrifugation. These two challenges represent the first part of the thesis, which focuses on methodology development. In synopsis, the treatment of air samples prior to DNA extraction is considered, along with the identification of the best region to target in amplicon-based high throughput sequencing. In the second part of the thesis, the focus turns to the application of the developed methodology to characterize fungal exposure in three different work environments: compost, biomethanization, and dairy farms. All three are of special interest due to potentially high fungal exposure. Results show that ITS1 outperformed ITS2 in disclosing higher levels of fungal diversity in aerosol samples. Due to complementarity in the taxonomic profiles disclosed by the two regions, the author suggests the use of both regions to cover the greatest possible number of taxa when taxonomy is the main interest of the study. However, ITS1 should be the first choice in other studies, mainly because of the high diversity it reveals and its concordance with results obtained via shotgun metagenomic profiling. In addition, the new filtration-based approach proposed in this work might be the best alternative available for compensating the loss of propagules in

centrifugation done prior to DNA extraction. Taken together, these methods allowed a profound characterization of fungal exposure in occupational environments.

Table of Contents Résumé ...... iii Abstract ...... iv List of Figures ...... ix List of Tables ...... xii List of Abbreviations ...... xiv Acknowledgements ...... xvii Preface ...... xix Introduction ...... 1 Chapter 1: Literature Review ...... 5 1.1 Bioaerosols ...... 5 1.2 Air Sampling ...... 10 1.3 Fungal Biology ...... 12 1.4 Health Outcomes after Fungal Exposure ...... 16 1.5 Occupational Exposure to Fungi ...... 22 1.6 Cultivating Fungi ...... 26 1.7 Microbiota Analyses ...... 30 1.8 Bioinformatics ...... 37 1.9 Future of Bioaerosols ...... 41 1.10 Specific Aims of the Thesis ...... 42 Part one: Methods for the Characterization of Fungal Aerosols ...... 44 Chapter 2: Comparison of the performance of ITS1 and ITS2 as barcodes in amplicon- based sequencing of bioaerosols ...... 45 2.1 Résumé ...... 45 2.2 Summary of the Paper ...... 46 2.3 Abstract ...... 48 2.4 Introduction...... 48 2.5 Materials and Methods ...... 52 2.6 Results ...... 60 2.7 Discussion ...... 78 Conclusion ...... 83 2.9 Bibliography ...... 85 Chapter 3: Fungal cell recovery from air samples: a tale of loss and gain ...... 95

3.1 Résumé ...... 95 3.2 Summary of the Paper ...... 96 3.3 Abstract ...... 98 3.4 Introduction...... 99 3.5 Materials and Methods ...... 102 3.6 Results ...... 110 3.7 Discussion ...... 121 3.8 Conclusion ...... 124 3.9 Bibliography ...... 126 Part Two: Evaluation of Workers Exposure to Fungi in Three Environments Affected by Fungal Aerosol problems ...... 134 Chapter 4: A Next Generation Sequencing Approach with a Suitable Bioinformatics Workflow to Study Fungal Diversity in Bioaerosols Released from Two Different Types of Composting plants ...... 135 4.1 Résumé ...... 135 4.2 Summary of the Paper ...... 136 4.3 Abstract ...... 138 4.4 Introduction...... 139 4.5 Materials and Methods ...... 141 4.6 Results ...... 145 4.7 Discussion ...... 154 4.8 Conclusion ...... 157 4.9 Bibliography ...... 158 Chapter 5: Fungal bioaerosols in biomethanization facilities...... 164 5.1 Résumé ...... 164 5.2 Summary of the Paper ...... 165 5.3 Abstract ...... 167 5.4 Introduction...... 168 5.5 Materials and Methods ...... 171 5.6 Results ...... 176 5.7 Discussion ...... 184 5.8 Conclusion ...... 188 5.9 Bibliography ...... 190 Chapter 6: Fungal aerosols at dairy farms using molecular and culture techniques ... 196

vii

6.1 Résumé ...... 196 6.2 Summary of the Paper ...... 197 6.3 Abstract ...... 199 6.4 Introduction...... 199 6.5 Materials and Methods ...... 202 6.6 Results ...... 209 6.7 Discussion ...... 219 6.8 Conclusion ...... 224 6.9 Bibliography ...... 225 Chapter 7: Additional Discussions ...... 233 Conclusion ...... 241 Concluding Remarks ...... 241 Future work ...... 242 Appendix A: Bioinformatics Tools to Study the Microbial Ecology of Bioaerosols ...... 244 A.1 Résumé...... 244 A.2 Summary of the paper ...... 245 A.3 Abstract ...... 247 A.4 Introduction ...... 247 A.5 Methods and Software...... 250 A.6 Sequencing Depth ...... 250 A.7 Alpha and Beta Diversity ...... 252 A.8 Parametric VS. Nonparametric Statistics ...... 254 A.9 Microbial Community Comparison: Statistics and Visualization Tools...... 256 A.9.1 Comparisons using Alpha and Beta Diversity Measures ...... 256 A.9.2 Statistical Significance of Sample Groupings ...... 256 A.9.3 Differential Abundance...... 259 A.9.4 Taxonomic Analyses ...... 260 A.9.5 Additional Visualization Tools ...... 262 A.9.6 Correlations ...... 264 A.10 Conclusion ...... 265 A.8 Bibliography ...... 266 Bibliography...... 274

viii

List of Figures

Figure 2.1: Rarefaction curves obtained from the number of observed OTUs and the sequences per sample for air samples from A) composting sites (the plateaus of the curves started at around 5000 sequences); B) biomethanization (the plateaus of the curves started at around 1500 sequences); C) dairy farms (the plateaus of the curves started at around 5000 sequences)………………………………………………………………………………….62 Figure 2.2: Principal Coordinates Analysis of air samples collected from composting sites (a and b), biomethanization facilities (c and d), and dairy farms (e and f)…………………….66

Figure 2.3: Fungal species with statistically significant differential abundances across compost samples targeting either ITS1 or ITS2 barcodes…………………………………..69

Figure 2.4: Fungal species with statistically significant differential abundances across biomethanization samples targeting either ITS1 or ITS2 barcodes…………………………70

Figure 2.5: Fungal species with statistically significant differential abundances across dairy farm samples targeting either ITS1 or ITS2 barcodes………………………………………71

Figure 2.6: Relative abundances of fungal genera detected in dairy farms by shotgun and amplicon-based (ITS1 and ITS2) HTS…………………………………………………..…77

Figure 3.1: Diagram of the fungal cells concentration protocols prior to DNA extraction..105

Figure 3.2: Concentration of Aspergillus fumigatus using culture and molecular (qPCR) methods in aerosol samples collected from three different composting facilities………….111

Figure 3.4: Boxplot representing the number of observed fungal OTUs in bioaerosol samples collected with the Coriolis in biomethanization facilities…………………………………115

Figure 3.5: Distribution of fungi detected in the centrifuged samples. All 50 fungi listed were present in 100% of the filtered samples……………………………………………………118

Figure 3.6: Triboelectric effect on spore pelleting of A.niger N402 (A), P.roquefortii FM164 (B), and A.niger ΔpptA (C) after centrifugation for 10 min at 1700 g in an Eppendorf fixed angle rotor…………………………………………………………………………………120

Figure 4.1: Overview of the bioinformatics pipeline for the processing of sequencing data before the visualization analyses………………………………………………………….144

Figure 4.2: Rarefaction analysis of air and compost samples from the two types of composting plants…………………………………………………………………………146

Figure 4.3: Principal coordinate analysis of air and compost samples combined from the two types of composting facilities (domestic compost in blue and animal compost in red)……147

Figure 4.4: Relative abundance of the classification of fungi (class and phylum) identified in compost and air samples from the two types of composting plants………………………..149

Figure 4.5: the most abundant fungal genera identified in domestic and animal composting facilities (air and compost)………………………………………………………………...152

Figure 4.6: Aerosolization behavior of some of the most abundant fungi genera identified in domestic composting facilities……………………………………………………………153

Figure 4.7: Aerosolization behavior of some of the most abundant fungi genera identified in animal composting facilities………………………………………………………………153

Figure 5.1: Concentrations of Penicillium/Aspergillus spp. (Pen-Asp) in the air at sampling sites from both biomethanization facilities during summer and winter……………………177

Figure 5.2: Concentrations of Aspergillus fumigatus spores in the air at sampling sites from both biomethanization facilities during summer and winter………………………………177

Figure 5.3: Rarefaction curves obtained from the number of observed OTUs and the sequences per sample for air samples from the two biomethanization facilities visited….178

Figure 5.4: Comparison of species richness estimator Chao1 index values from air samples from two biomethanization facilities collected during summer and winter……………….179

Figure 5.5: Principal coordinates analysis of air samples taken from two different biomethanization facilities visited during summer and winter…………………………….181

Figure 5.6: Average relative abundances of fungal classes in air samples collected from two different biomethanization facilities visited during summer and winter……………..……182

Figure 5.7: Venn diagram showing the 20 most abundant genera of fungi identified in air samples from BF1 during summer and during winter and those present during both seasons…………………………………………….………………………………………183

Figure 5.8: Venn diagram showing the 20 most abundant genera of fungi identified in air samples from BF2 during summer and during winter and those present during both seasons…………………………………………………………………………………….184

Figure 6.2: Rarefaction curves obtained from the number of observed OTUs and the sequences per sample for air samples from the five dairy farms visited………………….212

Figure 6.3: Relative abundance of fungi classes detected in air samples from five dairy farms using high-throughput sequencing…………………………………………………...……214

Figure 6.4: Relative abundance of fungal genera detected in air samples from five dairy farms………………………………………………………………………………………216

Figure 6.5: Relative abundance of fungal genera identified by high-throughput sequencing and culture in air samples collected from five dairy farms…………………………...……218

List of Tables

Table 2.1: Primers used for amplification of ITS1 and ITS2 barcodes and for Illumina Miseq sequencing………………………………………………………………………………….57 Table 2.2: Summary of the HTS data during the bioinformatics treatment process. ITS1 and ITS2 amplicons are compared in the bioaerosols from the three environments studies……..61 Table 2.3: Alpha diversity analysis comparing data obtained from targeting ITS1 and ITS2 barcodes in aerosol samples from three environments…………………………………….64

Table 2.5: Comparison of the 20 most abundant fungal genera identified by HTS targeting ITS1 and ITS2 barcodes, and the fungal genera identified by culture in aerosol samples collected at five dairy farms………………………………………………………………...75

Table 3.1: Primers and probes used for qPCR quantification of selected microorganisms...106

Table 3.2: Primers used for Illumina amplification……………………………………….107

Table 3.3: Eight fungal families with the highest number of representative OTUs and the range of their adjusted p-values representing the statistical significance of their differential abundance between the groups of samples………………………………………………...117

Table 3.4: The reported presence of hydrophobins or oil droplets in the seven fungal genera that were absent in the case of centrifugation……………………………………………...119

Table 4.1: Number of OTUs in each type of composting (air and compost)……………….145

Table 4.2: The Eurotiomycetes found in animal composting (air and compost samples)…………………………………………………………………………………...150

Table 5.1: Primers, probes, and protocols used for qPCR quantification of selected microorganisms………………………………………………………...…………………173

Table 5.2: Primers used for Illumina amplification………………………………………174

Table 6.1: Description of the sampling sites and the parameters affecting the sampling environments…………………………………………………………………………...…203

Table 6.2: Comparison of p-value of the concentrations obtained by qPCR between groups of samples (n=15) within four environmental factors using Kruskal-Wallis one-way analysis of variance………………………………………………………………………………...211

xii

xiii

List of Abbreviations

ABPA: Allergic bronchopulmonary aspergillosis bp: Base pair COPD : Chronic obstructive pulmonary disease DNA: Deoxyribonucleic acid HTS : High-throughput sequencing IARC: International agency for research on cancer ITS: Internal Transcribed Spacer LSU: Large subunit MALDI-TOF : Matrix-assisted laser desorption/ionization-time of flight MVOCs: Microbial volatile organic compounds RNA: Ribonucleic acid PCR: Polymerase chain reaction qPCR: Quantitative polymerase chain reaction rDNA: Ribosomal DNA rRNA: Ribosomal RNA SSU : Small subunit

xiv

Dedicated to Caroline Duchaine

“If I have seen further it is by standing on the shoulders of

Giants”

- Isaac Newton

xvi

Acknowledgements

The progress made throughout the path to the end of doctoral studies is the product of a wide range of influences. I am grateful for many people for their contributions to scientific knowledge and my personal development. Thank you to Caroline Duchaine for taking me in her team in the summer of 2013, where my journey began. Caroline is the reason I want to follow an academic career. The positive influence, and the life-changing experience she helped me navigate through enlightened me on my purpose in life, and how I can serve this purpose with my full potential. I hope that our work together has just begun and that we will continue to accomplish scientific challenges in bioaerosols. Thank you to Marc Veillette for giving me my first tools in molecular biology and for teaching me the steps of problem- solving. Your assistance professionally and your friendship are precious to me. I sincerely hope that we will continue to learn from each other, challenge each other, write papers, attend shows, and enjoy tasty beers. I would like to also thank Valérie Létourneau and Nathalie Turgeon for their care, love, and kindness. Evolving with you was such a pleasant experience. The mentorship and guidance of Caroline, Marc, Valérie, and Nathalie has been primary to my graduate education.

I thank my co-director Guillaume Bilodeau for all thoughtful comments and suggestions on the project, and for reviewing all the papers. Your additions made the publications more valuable. Thank you for welcoming me in your lab in Ottawa for the internship, which opened the door for my passion of bioinformatics. I am grateful to my committee members, Steve Labrie and Michel Frenette for their interest in and feedback on my research work. Their suggestions have made this work better. Thank you to Richard Summerbell for accepting to evaluate this thesis and for being the external jury during the thesis defense.

I would like to acknowledge all the lab members, including undergraduates, graduates, and postdocs, of Caroline’s research group in the last six years for the beautiful moments shared; you are my family. The caring love and appreciation that I felt from you towards me are my power boost during my down moments. I thank you all from the deepest

xvii

of my heart. A special mention to Laetitia Bonifait, Evelyne Brisebois, Vanessa Dion- Dupont, Marie-Eve Dubuis, and Joanie Lemieux for our work and the papers that we wrote.

I thank my mother Fatiha for the unconditional love and support. You always pushed me to pursue higher academic degrees, so I hope that you are proud of me getting to the end of the Ph.D. I appreciate all that you have done for me since my oldest childhood memories until now. I also thank my brother Omar for believing in me. Your love and support are valuable to me. I continue to learn from you, and your motivation is what pushed me always to surpass myself and achieve what everybody thought it was impossible. I thank my sister Fatima for teaching me how to free my spirit, and I thank my father, Mohamed for believing in me. I would like to give a special thanks to my uncles Hafid and Thami for their support throughout the years and for being my role models for success.

Finally, I thank Emilie for being an important person in my life from the beginning of my studies up until now. You helped me develop into the person I am today, and without you, I wouldn’t be where I am. Thank you for pushing me to go meet Caroline to give me an opportunity in her lab. Thank you for not letting me give up on myself during my darkest time. Thank you for your guidance throughout the Ph.D. path.

xviii

Preface

This thesis contains four published papers presented in Chapter 3, Chapter 4, Chapter 5, Chapter 6, and two papers in revision presented in Chapter 2 and Annexe A. The author of the thesis is the principal author for the six papers presented in the dissertation. The author, helped in designing all the studies, participated in all the sampling campaigns, did or participated in all laboratory experiments, analyzed all the data, and wrote the original drafts on the papers. All the supplementary materials referenced in the inserted papers are available with the published version of the papers (a section in each paper is dedicated to the availability of the supplementary materials).

Chapter 3 titled Fungal Cells Recovery From Air Samples: Tale of Loss and Gain was published in Applied and Environmental Microbiology on March 1st, 2019. Modifications between the inserted version and the published version are the re-organization of the sections (introduction followed by materials instead of results), the reference style in the text (name of the first author and the year of publication instead of numbers), and the reference style in the bibliography section. Also, the number of the figures and tables was changed by adding the number of the chapter before the figure or the table (e.g., Figure 3.1; Table 3.1). These modifications were made so that the format of the article is adapted to the general invoice of the thesis. The coauthors are Marc Veillette, Wieke Teertstra, Willem Kegel, Guillaume J. Bilodeau, Han A.B. Wösten, and Caroline Duchaine. Applied and Environmental Microbiology 2019 Mar 1. pii: AEM.02941-18. doi: 10.1128/AEM.02941-18

Chapter 4 titled A next generation sequencing approach with a suitable bioinformatics workflow to study fungal diversity in bioaerosols released from two different types of composting plants was published in Science of the Total Environment on December 1st, 2017. Modifications between the inserted version and the published version are the change of figures and tables numbers by adding the number of the chapter before the figure or the table (e.g., Figure 4.1; Table 4.1). These modifications were made so that the format of the article is adapted to the general invoice of the thesis. The coauthors are Marc Veillette,

xix

Laetitia Bonifait, Marie-Eve Dubuis, Yves Bernard, Geneviève Marchand, Guillaume J. Bilodeau, and Caroline Duchaine. Sci Total Environ. 2017 Dec 1;601-602:1306-1314

Chapter 5 titled Fungal bioaerosols in biomethanization facilities was published in Journal of the Air & Waste Management Association on August 27th, 2018. Modifications between the inserted version and the published version are the reference style in the bibliography section, and the number of the figures and tables that changed by adding the number of the chapter before the figure or the table (e.g., Figure 5.1; Table 5.1). These modifications were made so that the format of the article is adapted to the general invoice of the thesis. The coauthors are Marc Veillette, Marie-Ève Dubuis, Bouchra Bakhiyi, Geneviève Marchand, Joseph Zayed, Jacques Lavoie, Guillaume J. Bilodeau, and Caroline Duchaine. J Air Waste Manag Assoc. 2018 Nov;68(11):1198-1210. doi: 10.1080/10962247.2018.1492472.

Chapter 6 titled Fungal aerosols at dairy farms using molecular and culture techniques was published in Science of the Total Environment on October 26th, 2018. Modifications between the inserted version and the published version are the change of figures and tables numbers by adding the number of the chapter before the figure or the table (e.g., Figure 6.1; Table 6.1). These modifications were made so that the format of the article is adapted to the general invoice of the thesis. The coauthors are Marc Veillette, Guillaume J. Bilodeau, and Caroline Duchaine. Sci Total Environ. 2019 Feb 25;653:253-263. doi:10.1016/j.scitotenv.2018.10.345

Chapter 2 was submitted for peer-review publication on June 27th, 2019 and Annexe A will be submitted in the upcoming weeks.

I hope you will enjoy reading this thesis, as much as I enjoyed writing it.

Introduction

The work presented in this dissertation was motivated by the desire to introduce new approaches to study fungal aerosols. The reason behind this motivation was to compensate the gaps caused by the traditional approaches used in the study of fungal exposure in occupational environments. The traditional methods consist of cultivating fungal species collected from air samples to study their concentration and diversity. The gap in the literature caused by the application of cultivation methods solely is linked to the specific representation of the cultivable and viable portion of fungal aerosols. The underrepresented portion of the fungal community might have a significant role in the adverse health effects related to exposure. Thus, it is necessary to consider not only the tip of the iceberg, but look at the bigger picture of fungal aerosols. The willingness to fill the gap of the occupational studies with the non-viable or the uncultivable portion of fungal aerosols took us to consider the application of molecular tools. These tools consist of the high-throughput sequencing (HTS) of the genomic DNA extracted from aerosol samples taken from environmental settings, as streamlined by targeting various fungal universal barcodes. In addition, culturomics, which consists of the application of multiple culture conditions, combined with amplification and sequencing for the identification of previously unidentified colonies, is used to offer complementary images of fungal diversity identified by molecular tools. In the quest of developing a standardized molecular method to study fungi collected using air samplers, a physical phenomenon was noticed through which fungal cells recovered in liquid samplers were depleted, leading to the artefactual loss of biomass and diversity. This observation led to the development of a more efficient method to recover fungal cells from liquid air samplers. Also, with HTS becoming more popular, more bioinformatic tools are available, and choosing the right tool is a prerequisite to a valid diversity analysis. The author was motivated by the desire to develop a bioinformatic workflow with the freedom to control every step and parameter in the process of sequence treatment and diversity analyses.

The ideas defended by the thesis can be summarized as the use of molecular tools giving insightful information for an improved assessment of fungal exposure in occupational settings. In order that this goal be achieved, air samplers must be optimized in order to not miss crucial parts of the information provided by subsequent specific PCR amplification and HTS. Furthermore, the methods developed must be applied to various occupational environments to confirm their usefulness and robustness in achieving the goal of assessing exposure to fungal aerosols.

The presence of previously undocumented fungi in aerosol samples represent one of the limitations in the understanding of the impacts of fungal exposure on respiratory health. The harmfulness of fungal aerosol mixes could be dependent upon the composition of elements, synergy among the elements, and factors attributable to source and growth conditions. The absence of a universal method to explore fungal exposure in occupational settings is not surprising due to the limitations of the actual methods. To date, many questions are still unanswered in the use of molecular methods to characterize the fungal diversity of aerosols.

The hypothesis is that, although essential in the study of biodiversity, traditional approaches based on the cultivation of fungi induce a bias in the representation of fungi in aerosols. Thus, a method based on HTS with the appropriate barcode could be optimal to reveal the diversity of fungal aerosols and to provide exposure data with a maximally complete image of the presence of fungi. In the accomplishment of this purpose, the design of a customized bioinformatic workflow, enhanced by a revision of the bioinformatic microbial ecology tools adapted for analysis of fungal diversity could be a turning point in effecting the outcomes pursued by HTS. Regarding fungal cell properties influencing the accuracy of liquid air samplers, we hypothesized that hydrophobicity caused significant loss of fungal particles during centrifugation, which is usually the first step for cells recovery from air samples. The proposed filtration-based approach might be the best alternative to countervail this fungal loss prior to DNA extraction. Finally, the last hypothesis states that the application of these methodologies can provide results relevant for assessing workplace exposure in specific studies.

This thesis is organized as follows: Chapter 1 presents a literature review introducing the subjects addressed in the thesis, and presents new developments in bioaerosol studies backed by recent publications. Then, the thesis is divided in two parts: Part one presents a review of the methods developed to study fungal aerosols. Chapter 2 moves on to experimental work. It describes a large-scale study using air samples from waste treatment and dairy barns in order to do a methodical comparison of the performance of two universal fungal barcodes, ITS1 and ITS2, in a metabarcoding analyses of fungal diversity in aerosols. The analysis includes sequence length distribution, richness and diversity indexes, multivariate analyses, differential abundance, species discrimination efficiency and taxonomy analyses. In addition, a shotgun metagenomic approach applied on air samples from dairy farms was compared in outcome to the ITS1 and ITS2-based HTS approach. Diversity as elucidated by traditional culture method was added to the analysis to evaluate its potential contribution to modern fungal bioaerosol studies. Chapter 3 describes methodology used to develop a protocol, compatible with the molecular test methods used, for concentrating fungal spores recovered from air samples in a way that minimized losses occurring during centrifugation in the liquid impaction methods widely used for processing air samples. A real-time PCR and an HTS approach were used on samples from two different environments. Results obtained using the new filtration-based technique were contrasted with results obtained of standard centrifugation protocols in microbial ecology metrics. Part two of the thesis contains three chapters, each describing the application of the proposed methods from part one in evaluation of workers' exposure to fungi in three environments affected by fungal aerosol problems. Chapter 4 describes the analysis of fungal diversity in composting organic matter, and of the aerosols produced in two different composting formats: domestic composting bins and pig carcass composting. Chapter 5 provides a thorough, occupationally oriented characterization of fungal exposure in two biomethanization facilities, one treating primary and secondary sludge from wastewater treatment plants and organic industrial food waste, and the other treating domestic wastes. Chapter 6 presents a characterization, based on qPCR and HTS, of fungal exposure at Eastern Canadian dairy farms, as well as a comparison to traditional culture methods.

This work also includes one appendix. Appendix A is a bioinformatic review article describing the microbial ecology tools used in the diversity analyses conducted throughout this dissertation.

In conclusion, this thesis makes the following novel contributions: • It extends our understanding of the relative effectiveness of the two universal fungal barcodes, ITS1 and ITS2, in describing fungal diversity of aerosols. Thus, it offers aerosol scientists a guided strategy to use in designing a molecularly based study to address fungal aerosol exposure, including considerations of the possible contributions of shotgun metagenomics. • It addresses the issue of fungal cell loss when recovered from air samples and proposes an alternative method to improve evaluation of fungal exposure and diversity. Based on the results of this study, the newly developed filtration protocol should be used to achieve the highest possible fungal cell recovery from air samples. • It provides information on fungal exposure in three different work environments, composting and biomethanization facilities, as well as dairy farms. Substantial insights are given about fungal biomass and diversity in these environments. • It illustrates and promotes the use of selected bioinformatic tools in the study of bioaerosols and serves as a good source for learning the “dos and don’ts” involved in conducting a precise microbial ecology study.

Chapter 1: Literature Review

1.1 Bioaerosols

Bioaerosols are the biological fraction of airborne particles originating from plants, animals and other living organisms. The totality of bioaerosols contains a mixture of viable and nonviable microorganisms, including bacteria, fungi and viruses, and of other biological components such as animal and plant debris, endotoxins, exotoxins, and miscellaneous microbial metabolites; Desprès et al., 2012; Macher et al., 1999; Tuck, 2002). Bioaerosols are ubiquitous in indoor and outdoor environments, and may be generated from various natural sources as well as anthropogenic sources. With their small sizes and low weights, bioaerosols are easily transported from one place to another, and persist in the air for long periods of time. In fact, they may travel kilometres depending on the size of the particle and the incident air movements (Brown and Hovmøller, 2002; Burrows et al., 2009a; 2009b; Womack et al., 2010; Nygard et al., 2008; Desprès et al., 2012).

Bioaerosols are composed of particles ranging from a few nanometres to 200 µm in diameter. For reference, the diameters of viruses are generally between 20 and 300 nm, while bacteria range from 0.3 to 10 µm, and fungal spores from 1 to 100 µm (Heikkinenm et al., 2005). A parameter called the aerodynamic diameter parameter, defined as the diameter of a sphere with a density of 1g/cm3 that has the same settling velocity as the particle of interest, is considered to be the most important parameter for describing the behaviour of particles suspended in the air (Nazaroff, 2014). Although the density of bioaerosol particles may vary, the aerodynamic parameter depends primarily on particle size and shape. Thus, the aerodynamic diameter influences the interactions of the particles with physical processes. Examples include ice nucleation, gravitational settling, building penetration, resuspension from surface into air, and respiratory deposition (Yamamoto et al., 2014). The size distribution of bioaerosols varies considerably with atmospheric (outdoor) and indoor conditions. This variation is generally occurs both between and within sites (Saari et al., 2015; Dong et al., 2016).

The behavior of bioaerosols is dictated by physical laws. Airborne particles move and deposit on surfaces at different rates depending on their aerodynamic diameter; this parameter tends to determine rates of sedimentation, inertial impaction, interception, diffusion, and electrostatic attraction. Sedimentation occurs when gravitational attraction overcomes air resistance to diminish the floatability of an airborne particle. Inertial impaction of a particle in the air involves the appearance of forces causing the mass of the particle to overcome the drag force of the airflow. In cases of insufficient inertial momentum, particles follow the streamlines of airflow. In contrast, interception happens when airborne particles impact on a surface without deviation from the path of the air current. Diffusion occurs when a particle of small size (< 0.5 µm) entrained in Brownian motion in the air due to movement of the gas molecules around it, is brought into surface contact., Electrostatic attraction, as the name suggests, takes place when a charged particle meet an oppositely charged surface (Hinds, 1999).

In terms of biological hazards, particle size of bioaerosols influences the site of deposition in the respiratory tract and the survival of microbes in the airborne particles. Small particles (1 to 5 µm) may reach the alveoli; particles with a size of 4 to 10 µm tend to deposit in the upper airways, while larger particles (> 10 µm) may remain in the nasal cavity (Sturm, 2016; Rawlings et al., 2013; Tong et al., 1998; Lighthart et al., 1997). It is important to mention that the deposition rate is always presented as a ratio. For example, 60% of the particles of 0.01 µm in diameter may deposit in the alveolar region (Carroll, 2009). Some particles with an aerodynamic diameter in the middle of the range can be deposited in both the upper and lower respiratory tract.

Roy and Milton proposed that different sources each generate particles with a characteristic range of sizes. Even particles released from a single infected subject will differ in size according to the origin of the aerosol. (Roy and Milton, 2004). As mentioned before, the size range affects the length of time an airborne particle will spend in the air. Other factors like aerosol composition and environmental factors also influence the length of time a particle spend in the air. The environmental factors include the strength of air currents and the presence of precipitation, as well as the relative humidity and temperature. The dispersal of

bioaerosols may impact the air quality of extensive areas from diverse sources and be a complex public health issue due to the presence of diverse microbial communities changing in time and space.

Because of the pathogenic potential of bioaerosols, it is important to understand their behaviour and their role in disease transmission. Pathogenic transmission can take several routes: person-to-person, waterborne, foodborne, vector-borne (rodent- or insect-to-human), and airborne. As expected, air quality specialists focus on the airborne transmission route. Some pathogens are transmitted primarily through the airborne route, such as the viruses causing the Severe Acute Respiratory Syndrome (SARS) and those causing avian flu (Yu et al., 2004; Sergeev et al., 2012). Probably the best known example of an ‘obligately’ aerially dispersed pathogen is Mycobacterium tuberculosis, infection with which is initiated only through aerosols under natural conditions. Pathogens that considered to be ‘preferentially’ aerially transmitted can cause infection through multiple routes but nonetheless predominantly infect via aerosol deposition in the respiratory tract tissues. Examples are measles and variola viruses. ‘Opportunistic’ airborne pathogens also use diverse routes for disease transmission, but have adaptations exploiting aerosols as an efficient way to propagate in an ideal environment (Roy and Milton, 2004). For purposes of controlling risks to health, scientists in the bioaerosol field have undertaken to understand the component parts of bioaerosols, and the actions of these particles in various environments. Demonstration of the transmission route of an infectious agent is very challenging and, often, mathematical models must be configured to assess the contributions of different transmission pathways.

Bioaerosols carry not just infectious agents, but also, other biomaterials. They may include various combinations of viruses, bacteria, fungi, endotoxins, mycotoxins, b-glucans, and allergens. The success of bioaerosol studies benefits from multidisciplinary approaches brought together by microbiologists, public health specialists, physicists, and engineers working to increase the impact of studies.

Viruses are ubiquitous in mostly all ecosystems; they represent the most exuberantly reproduced biological entities on the planet, and are tenfold more numerous than bacteria (Suttle, 2007). Although, viruses are presumably widely distributed in indoor and outdoor environments, their community structure in bioaerosols remains mostly unexplored. Normally, when studied in water or other substrates, viruses are sampled and concentrated by filtration based on their small size . In aerosols, it is challenging to enrich them based on this basis, since they may be trapped within particles that are of different sizes (Verreault et al., 2008). As with viruses, other organisms may also need to be treated differently when airborne. Evidently, the stress of the aerosolization process may cause them to react differently than they do in other environmental samples. This can be seen in a metagenomics analysis of aerosols from urban spaces described by Be et al., (2015). Viruses were expected to outnumber bacteria but the study showed their sequences were underrepresented, probably because the approach used was more suitable for detecting bacteria than it was for detecting the small genome of the viruses. The types of viruses present in bioaerosols will depend strongly on the viral sources in the local surroundings, or in a combination of more distant sources if the viruses are transported over long distances (Prussin II et al., 2014). Considering that viruses have the potential to cause widespread diseases, from the seasonal flu transmission to more severe cases of atypical pneumonia caused by corona virus, research on virus characterization in aerosols is still considered a priority.

Bacteria are prokaryotic microorganisms with an estimated diversity of millions of species (Amann and Rosselò-Mòra, 2016). Historically, bacteria have been classified in Gram-positive and Gram-negative groups, distinguished by outer membrane chemistry. Because bacterial growth depends on various factors like water availability, pH, temperature, oxygen, and nutrients, environmental conditions that favour optimal growth are used in the identification of bacterial species. Bacteria can become airborne in aqueous or dry particles, and can carry volatile organic compounds with them that were produced during growth on food or other organic materials like wood, paper, and textiles.

The development of HTS approaches in recent years has led to a new era for microbial ecology studies. Bioaerosol studies have used these methods to make new developments

(Yoo et al., 2017). Characterizing the aerosol microbiota is necessary to discern the microbial processes occurring in bioaerosols. To extend the analyses, metatranscriptomic approaches have provided an means of explaining microbial community changes in different environments; this is seen, for example, in the analyses of Coulon and Colbeck (2018) on complex bioaerosol communities and their interactions with environmental factors. In addition, the new field of ecological metabolomics is about to further enrich our understanding. In the same way as metatranscriptomic comprehends the analyses of all transcripts from a given sample, metabolomic approaches look at the naturally occurring metabolites from the entire bacterial population of an environmental sample (Jones et al., 2016). Designing bioaerosol studies that include these approaches will be beneficial for the future of the field. At present, there is a paucity of insight into how the genetics, physiology and biochemistry of bioaerosols differ in different conditions.

Fungi are the microorganisms of interest in this work, and will be introduced in detail in the Fungal Biology section of this chapter (Section 1.3). Also, fungi will be the main focus in all the remaining chapters of this dissertation.

Sources of bioaerosol exposure in occupational settings are diverse, and include waste treatment, agricultural activities and food processing plants. The sources of bioaerosols indoors include outdoor air, building materials, furnishings, occupants (coughing, sneezing, talking, walking) animals, plants, and organic wastes (Prussin II and Marr, 2015). Assessments of exposure to bioaerosols in occupational, indoor and outdoor environments play a central role in studies seeking to characterize risk. However, the assessment of exposure to bioaerosols remains difficult, and there is little agreement on standard methods for analysis of bioaerosol components. Many technical details need to be fine-tuned in order to obtain results that are representative of the sampled environment. For example, air sampling devices are crucial as they represent the first link between air and analytical methods to analyze it.

1.2 Air Sampling

Air sampling methods for bioaerosols must be functional in different aspects. First, the samplers must have a high inlet efficiency at collecting aerosol particles. Second, the samplers must have the efficiency to physically collect airborne particles. The physical collection efficiency is defined as the sampler's capability to collect the bioaerosols into the sampler. This condition is affected by the size, shape, and aerodynamic diameter of the airborne particles. In specific cases, the sampler must also be able to maintain the viability of microbes in bioaerosols (Kulkarni et al., 2011; Nevalainen et al, 1992). Obviously, this last condition is required only when subsequent analytical analyses include cultivation methods yielding the appropriate conditions to determine viable counts. In such cases, the microbes should not face any harsh conditions, like impaction through the sampling device. Prompt care of samples plays a crucial role in the analyses. When molecular methods are applied, maintaining viability is not mandatory. Currently, there is only one commercially available, real-time instrument for the live measurement of bioaerosols (UV-APS, time-of- flight spectrometer; Agranovski et al., 2003). All other sampling methods concentrate particles for subsequent analysis in a liquid, on agar, on an adhesive surface, or on a filter.

All the samplers operate at different airflow rates, over different time periods, collecting different volumes of air. In molecular biology studies, high airflow samplers are the most used, as low airflow samplers lead to low biomass concentrations that may reduce the chances of successful molecular analyses. Measuring the concentration of only the viable fraction of airborne particles underestimates the total bioaerosol concentration and diversity because most bacteria and fungi in the air are not in a culturable state and will not develop on growth media in a laboratory environment (Amann et al., 1995; Pace et al., 1997). In addition, sampling without particle size discrimination or information on the cut point of an instrument may also over- or underestimate exposure (Heikkinenm et al., 2005). The selection of air sampling methods for airborne bioaerosols should be more delicate than those commonly employed for the general analysis of airborne particle composition. As anticipated, each type of bioaerosol reflects unique characteristics and is most suited for a particular type of sampling. No single air sampling method can consider the entire range of analytical

procedures that are needed to provide a complete picture of the bioaerosol composition. However, optimal collection methods can be chosen for individual biological agents or groups of agents that are identified in advance as the focus of an environmental assessment. Investigators designing a study must consider the limitations of the air sampling method, the sampler performance, the temporal and spatial variability of bioaerosol composition and the concentration of particles in the studied environment. As the present work focuses on molecular methods, two high airflow rate samplers were used: (1) The liquid cyclonic impactor Coriolis µ® (Bertin Technologies, Montigny-le-Bretonneux, France) and (2) the dry/electrostatic filter SASS® 3100 (Research International, Monroe, WA, USA). In the former, air is aspirated and drawn into the cone forming a vortex where particles are impacted into the liquid in the cone. The Coriolis µ® has a cut-off size of 0.5µm for an operation flow of 300 litres/min, which means that 0.5µm particles are sampled at 50% efficiency. Larger particles are sampled at a higher efficiency. Then, 15ml of phosphate-buffered saline set at a 7.4 pH and a concentration of 50µM are placed in the sampling collection recipient of the Coriolis µ® sampler and used as the collection liquid. The SASS® 3100, by contrast, contains a 44 mm diameter capture electret filter. The collection efficiency for 0.5 to 5µm particles is approximately 92 % at an air flow rate of 120 litres/min. At the maximum airflow rate (310 litres/min), the collection efficiency for the same particle size range is 80%. However, the overall capture rate is much greater at a higher airflow. After particle collection on the electret filter, a SASS® 3010 Particle Extractor (Research International, Monroe, WA, USA) is required to elute the captured particles in a buffer (SASS® 3010 Extraction Fluid). The cultivability analyses in the present study were performed using counts retrieved from liquid sampling.

Selection of the appropriate analytical methods is linked to the air sampling method used. Depending on the information desired, differing analytical methods can be used, including direct counts (e.g., microscopic analyses on spore traps), cultivation-based methods, biological assays (e.g., infectivity and cytotoxicity), immunoassays, chemical and biochemical analyses (e.g., detection of carbohydrates, fatty acids, lipids, proteins, and metabolic products), and molecular assays (usually based on RNA or DNA sequence analysis). These analytical methods vary widely in their sensitivity and specificity.

Microscopy and cultivation assays measure intact and viable microorganisms, respectively, while bioassays, immunoassays, and chemical analyses measure different compounds or biologically active molecules in whole cells, cell fragments, or extracts of extracellular metabolites. Molecular methods that detect and amplify selected genes or that generate millions of sequences to study DNA, RNA, transcriptomes and metabolomes are becoming increasingly available and their superior utility compared to the traditional methods has been demonstrated, especially for bacteria. However, there are still many challenges and unanswered questions for fungi that must be answered in order to take full advantage of this new era of molecular methods to better understand occupational exposure to fungal bioaerosols.

1.3 Fungal Biology

The estimates of fungal diversity were recently revisited and the biosphere was calculated to contain up to 3.8 million species (Hawksworth and Lücking, 2017). This diversity of the fungal kingdom opens the doors to boundless associations with other organisms. The origin of fungi goes back to marine ancestors 1 billion years ago in the Mesoproterozoic era (Berbee et al., 2017). Since then, fungi have developed in almost every habitat on earth (Peay et al., 2016). They exert influence on all ecosystems by producing and using nutrients across all trophic levels of the food chain (Stajich et al., 2009). Even though fungi can support human benefit by sustaining diverse ecosystem services and providing medicines and a myriad of food products, an extremely diverse group of fungi can have devastating impacts on the ecosystem by spoilage of food and materials, and some can represent direct threats to exposed humans (Meyer et al., 2016; Baxi et al., 2016). In addition, the vast bulk of fungal diversity remains undetected and even the detected portion still needs extensive taxonomic characterization (Hibbett et al., 2016). This applies to different taxonomic levels, and includes cryptic diversity within species or even phenotypic heterogeneity (Lücking et al., 2014; Hewitt et al., 2016). Cryptic diversity occurs when two different species are identified as the same species because of similarity in their morphology. This phenomenon could be a significant factor in determining the number of taxa identified

in future identifications (Trontelj et al., 2009). In contrast, phenotypic heterogeneity describes the variation that can exist between individuals within a single strain.

Although all fungi are heterotrophs, which means they get their nutrients by absorbing them, the fungal kingdom contains a wide range of life strategies. Saprotrophic fungi mediate the decomposition of organic matter. Mycorrhizal fungi associate with plants in a symbiotic relationship, where the hyphae of fungi interact with plant root cells, which allows the plant and the fungi to receive more nutrients. Meanwhile, endophytes are found inside of plants, and they have a range of poorly understood nutritional strategies and taxonomic affiliations (Deacon, 2006). HTS methods have the potential to elucidate many other life strategies of fungi by continuing the exploration of fungi in many ecosystems.

In the past, fungi were classified according to their macroscopic characteristics such as. physiology, shape, and colour. With the new sequencing technologies, classification now relies on molecular genetics and on reproductive strategies. Before 2013, teleomorphic fungi had separate names based on sexual structures, while anamorphic names designated the asexual reproductive structures. For instance, Davidiella was considered to be the teleomorphic state of Cladosporium. Likewise, Talaromyces and Eupenicillium were the teleomorphs corresponding to the anamorph Penicillium, while Emericella, Eurotium, Neosartorya, and Petromcyes were among several teleomorph states corresponding to the anamorph genus Aspergillus (Houbraken and Samson, 2011; Samson et al., 2014; Yilmaz et al., 2014). Some fungi had only an asexual state. With molecular methods, new links between sexual and asexual states were identified, which led to mycologists having to cope with the dual nomenclature problem. This awkward situation was resolved in 2013 with the introduction of a one-fungus-one-name system of nomenclature. In addition to the fungi known from morphological studies, some lineages are known entirely only through DNA or RNA sequences, like Cryptomycota, Archaeorhizomycetes, and Cyphobasidiales (Jones et al., 2011; James et al., 2013; Rosling et al., 2011; Spribille et al., 2016).

Despite the fact that fungi classifications are not unanimous agreed upon among mycologists, researchers generally recognize seven phyla of fungi. Dikarya is a sub-kingdom divided into two phyla, Ascomycota and Basidiomycota. In fact, the largest fungal phylum is Ascomycota, which got its name from the ascus that contains the sexual spores, called ascospores. By contrast, Basidiomycota are named consistently with the basidiospores that grow on club-shaped structures named basidia. Taken together, Ascomycota and Basidiomycota contain far more species than the five other phyla. Chytridiomycota are mainly aquatic fungi that are known for degradation of chitin and keratin. Blastocladiomycota were considered as the same phylum as the Chytridiomycota, until molecular methods showed they constituted a distinct phylum. All the members of the phylum Glomeromycota appear to have a purely asexual reproduction cycle. As for the phylum Microsporidia, they are parasites of unicellular animals and protists. Finally, the Neocallimastigomycota constitute the smallest phylum of the fungal kingdom and consist of anaerobic rumen symbionts (Guarro et al., 1999).

Recent phylogenetic studies in the last decade driven by molecular methods have led to the reshaping of the fungal kingdom. For example, Zygomycota, a phylum in the old morphologically based scheme of the kingdom Fungi, were reclassified into two subphyla, Mucoromycota and Zoopagomycota (Spatafora et al., 2016). Based on the enormous amount of taxonomic data provided by HTS, Tedersoo and collaborators proposed “an updated phylum- and class-level fungal classification accounting for monophyly and divergence time so taxonomic ranks are more informative. The phyla proposed are Aphelidiomycota, Basidiobolomycota, Calcarisporiellomycota, Glomeromycota, Entomophthoromycota, Entorrhizomycota, Kickxellomycota, Monoblepharomycota, Mortierellomycota, and Olpidiomycota; under nine sub-kingdoms” (Tedersoo et al., 2018).

There are other mould-like organisms like slime moulds and Oomycetes (water moulds) that act like fungi but do not constitute a phylum in the fungal kingdom (Rossman and Palm, 2006). Considering the astonishing diversity and the rapid pace of taxonomic changes in mycology, exploring fungal diversity in bioaerosols is far from an esoteric objective, but

represent a crucial achievement to management and human health concerns over the next centuries.

Fungi have distinct features that may affect their behaviour when in contact with water, including charge (Gregory, 1957) and hydrophobicity (Wösten et al., 1999; Linder et al., 2005). Hydrophobicity is caused by hydrophobins (Wösten et al., 1994) or by hydrophobin- like proteins such as repellents (Teertstra et al., 2006). These surface proteins make fungal spores water-repellent, which may cause the spores to be relatively easy to aerosolize at a water/air interface when affected by natural bubbling or by a actions involving water splash. Other factors that affect the tendency of fungal spores to become concentrated in bioaerosols include the quantity of sporulation, which is dependent on growth temperature, the rate of spore release from conidiophores as influenced by relative humidity and air streams, the weights and dimensions of spores, surface adherence and germination; and the surface chemistry of spores (dry or mucoid; Fröhlich-Nowoisky et al., 2016). If we survey fungi commonly found in bioaerosols in general, Aspergillus, Penicillium and Cladosporium produces many small and light spores, while Alternaria and Stachybotrys produce fewer, but heavier spores. In terms of the distance travel, fungal spores are known to travel great distances, and at high altitudes. Indeed, early aerobiology studies detected fungal spores at 10,000 feet from the ground, even above the oceans (Damialis et al., 2017).

Ideal growth conditions for fungi are influenced essentially by humidity, water activity, and temperature. Therefore, these organisms can grow on a multitude of surfaces and environments. For example, when a fungal growth amplifier is present indoors, the mycoflora associated with it differs from the outdoor mycota qualitatively and quantitatively. Whether the source of bioaerosols is carpets, furniture, walls, or dusts, any indoor fungal growth contributes to the statistical changes observed in comparisons with outdoor bioaerosols (Haleem Khan and Mohan Karuppayil, 2012). Sources in outdoor environments are many and include herbaceous plants, woodlands, and bodies of water, including oceans. In fact, the assemblages of bioaerosols seen in outdoor air is as diverse as collective sources in all the ecosystems of the planet. Occupational environments with a specific presence of materials that encourage fungal presence or growth, like wastes undergoing treatment, or food being

processed, can emit spores via ventilation and become important sources of outdoor fungal spores. A specific Occupational Exposure section (section 1.5) will focus on bioaerosol fungal exposure in various work environments. The purely outdoor sources of airborne fungi are not without risk, as most outbreaks of nosocomial fungal diseases have been associated with fungal spores from outside the hospitals (Shams-Ghahfarokhi et al., 2014; Ekhaise et al., 2008; Sautour et al., 2009; Kim et al., 2010). Therefore, describing the wide range of airborne fungi, using methods that assure accuracy of the results, is of great importance for understanding epidemiology and trying to diminish exposure.

Apart from fungal biology, scientists in the field have been including other disciplines like bioinformatics, biostatistics, big data visualization analyses to explore ecological dimensions of fungi. Combining different approaches and expertise can help fill the gaps in our knowledge of fungi. However, integrating data from various approaches is challenging and requires knowledge of the different disciplines involved. Fungal diversity researchers need to adopt new solutions for compiling and interpreting the data obtained by using advanced visualization tools to obtain multidimensional information. Networking and collaborations must not be underestimated as means to increase the sophistication of fungal diversity research. Indeed, to profit from the large-scale outputs generated by HTS approaches, hypotheses should be validated by comparison with more traditional methods. This combination of new and traditional could take fungal studies from the purely observational to the causal level of understanding by revealing better ways to implicate fungal exposure in defined health outcomes.

1.4 Health Outcomes after Fungal Exposure

Fungal infections are probably the best understood health problems related to fungal exposure. Between 300 and 600 fungal species can be pathogenic to humans. Thirty of these are listed as the causative agents of the most important fungal infections (Bongomin et al., 2017). Immunocompromised patients are the most severely affected by fungal infections. Persons with stable immune systems who live outside of the specific regions where endemic

mycoses are prevalent may have only superficial fungal infections, and sometimes, even if infected, don’t experience any symptoms. In exceptional cases, fungi can cause serious, even deadly, infections in healthy individuals. Blastomycosis caused by Blastomyces dermatitidis is an example of an endemic fungal infection that lead to pneumonia-like symptoms, as well as haemoptysis, fever, dyspnea, muscle ache, and fatigue (Bariola et al., 2010). Coccidiomycosis is caused by inhaling spores of Coccidioides immitis or C. posadasii inducing flu-like or pneumonia-like symptoms, sometimes accompanied by skin rashes. The infection can sometimes worsen and disseminate throughout the body, causing life- threatening symptoms. In some situations, e.g., coccidioidal meningitis, antifungal drugs may have to be taken for life (Hector and Laniado-Laborin, 2005). Cryptococcosis, caused by Cryptococcus gattii, C. neoformans and several related species, is another example of fungal infection acquired through inhalation of spores ( Some of its etiologic agents are cosmopolitan, while others are found primarily in tropical and subtropical climates. Cryptococcal infection frequently affects the brain, causing meningitis (Abassi et al., 2015). In addition, inhaling spores of Histoplasma capsulatum causes histoplasmosis, a fungal infection that particularly affects immunocompromised individuals. Symptoms can become severe and the infection can disseminate outside the lung to various parts of the body (Kauffman, 2007). It is important to keep in mind that case presentations in the four examples of fungal infections mentioned here are highly dependent on the patient’s susceptibility, and that each individual represents a unique case.

Another example of fungal infection is related to ubiquitous environmental fungi that may cause disorder in healthy humans who have immune system disturbances. A great example of such case is yeast infections caused by Candida albicans, a natural colonizer of mucosal surfaces of humans (Jabra-Rizk et al., 2016). Cryptococcus neoformans is another yeast implicated in opportunistic pathogenesis, overwhelmingly in HIV and AIDS patients (Kwon-Chung et al., 2014). Other than yeasts, filamentous fungi are also involved in opportunistic infections. Invasive aspergillosis is caused mainly by Aspergillus fumigatus, but other Aspergillus species have been implicated in the infection. This type of infection has been noted in individuals undergoing chemotherapy, immunosuppressive therapy, and organ transplants (Patterson et al., 2016). In addition, Fusarium species also cause invasive

infections, particularly in leukemia patients, and in patients with hematopoietic stem cell transplants (Liu et al., 2014). In general, factors like age, smoking history or habits, and the general immune system functionality affect a person's susceptibility to fungal infections. Many of the fungi mentioned above are widely distributed in the environment, thus adequate monitoring of fungal diversity in various environments is recommended to continually assess the type of exposure.

Many systematic reviews and epidemiological studies have associated fungal exposure and the risk of developing specific asthma symptoms (Sharpe et al., 2015). The indicators of probable exposure to allergenic fungal material measured in these studies include visible water damage or condensation caused by flooding or leaks, mouldy odours, visible fungal growth, and quantitative measure of fungi in air samples by culturing techniques or by microscopic counts. Some studies have associated outdoor fungal exposure in the development of asthma, as well as in its persistence from childhood to adulthood, and in the exacerbation of symptoms (Tham et al., 2014). Sensitization to fungal allergens is also seen in allergic rhinitis (Lee et al., 2014). However, insufficient evidence is available to indicate whether indoor or outdoor fungal exposure is responsible for the initial appearance of symptoms.

Linking fungal exposure to the induction of allergic asthma (as opposed to individual asthma attacks in a person who already suffers from the condition) is complicated because many different fungal species contribute to the exposure, while the patient’s physiological response is dependent on the individual exposed. In addition, all individuals are exposed to numerous fungal species throughout their life spans, a factor that complicates the implication of particular fungi in the induction of sensitization. The only way to prove a direct link between exposure to fungi and development of allergic diseases like asthma and rhinitis is to artificially carry out exposures of individuals to selected fungi. In managing attacks in previously sensitized patients, researchers may find ways to selectively reduce fungal exposure in symptomatic patients so that exacerbations or abatements of symptoms can be documented. Considering the difficulty of conducting such research, observational studies

will be need to continue in order to provide us with fortuitously acquired, insightful information.

One of the most severe allergic diseases caused by fungi is allergic bronchopulmonary aspergillosis (ABPA). It can be induced in asthmatics, chronic obstructive pulmonary disease (COPD) patients and cystic fibrosis patients after exposure to Aspergillus fumigatus (Greenberger, 2002). The difference between the infectious and allergic aspergillosis is that, in the latter, colonization of the lung by Aspergillus fumigatus is required for the generation of IgE and IgG antibodies specific to Aspergillus, a process connected with sensitization. In other words, first the germination of fungal spores occurs in the lung, then colonization, and then chronic exposure to locally produced fungal allergens (Kosmidis and Denning, 2015). Other than Aspergillus, a few fungal species can be responsible for similar allergic reactions called allergic bronchopulmonary mycosis (ABPM). These fungi include, Candida, Bipolaris, Schizophyllum commune, and Curvularia (Sarkar et al., 2010).

Allergic fungal rhinosinusitis (AFRS) causes a hypersensitivity reaction similar to that seen in ABPA. The main difference is that AFRS occur after fungal inoculum colonizes the sinuses, instead of the lung. The individuals affected are usually allergic patients with altered tissues. In contrast to ABPA, where Aspergillus fumigatus is the main etiologic agent, other fungi are commonly linked to AFRS, like Bipolaris, Exserohilum, Curvularia, Alternaria, Epicoccum, Ulocladium, and Botrytis (Schubert, 2009). Unlike the lung, the sinuses may often be colonized by fungal inoculum in healthy individuals. Many other conditions may be involved in the development of AFRS. Taking into account the astonishing fungal diversity of bioaerosols in different environments, many other genera and species of fungi could be suspected of playing a role in ABPM and AFRS.

Another fungal-related disease that falls into the category of allergic disorders is hypersensitivity pneumonitis (HP). The disease is defined as an interstitial granulomatous lung disease that develops in exposed subjects who become sensitized to a fungal antigen after multiple exposures to organic particles. The interstitial granulomatous condition is

characterized by the formation of granulomas and interstitial infiltrates composed of lymphocytes, plasma and cells, and macrophages. The symptoms of the disease are variable, and the diagnosis is difficult because no specific test or biomarker gives a consistent diagnosis. Even though patients with HP have generally considerable levels of antigen specific IgG antibodies, it is uncertain if these antibodies represent a marker of exposure or are part of the pathogenesis (Riario Sforza and Marinou, 2017). In general, HP cases are either acute, subacute or chronic. Many individuals may be exposed to the same levels of antigens, and even become sensitized, while only few develop HP. This observation strongly suggests that genetics and gene-environment interactions play a major role in the development of the disease (Fitzpatrick et al., 2017). Species of Aspergillus, Penicillium, Cladosporium, Alternaria, Bjerkandera and Fusarium have been suspected in many HP diagnoses (Chowdhary, et al 2016). The genus Trichosporon was particularly frequently linked to HP in Japan, where some individuals from the northern islands were exposed to the fungi during summer, after the rainy season, in indoor growth (Ando et al., 1989). As with ABPM and AFRS, broad-ranging investigations of a airborne fungal diversity in different affected environments may help to associate additional species with HP.

Mycotoxins are secondary metabolites of low molecular weight produced by filamentous fungi. Mycotoxins can be named in ways that reflect their structure, the taxonomy of the fungi producing them, or disease condition they cause (e.g., the tissue affected; Bennett and Klich, 2003). The microbiological context of the present study means that it is sensible to discuss mycotoxins as we come to the taxa involved in their production. Hundreds of mycotoxins have been identified, a limited number of which have known environmental importance. In agriculture, which is the best characterized environment in terms of mycotoxin exposure, the major players are aflatoxins (AF B1, AFB2, AFG1, and AFG2); ochratoxin A (OTA); trichothecenes (deoxynivalenol [DON], nivalenol, T-2 toxin, HT-2 toxin); the fumonisins B (FB1 and FB2); and zearalenone. These mycotoxins are produced by species of Aspergillus, Penicillium, and Fusarium (Alshannaq and Yu, 2017). The complexity of mycotoxin ecology is that different fungal species can produce the same mycotoxin and one fungal species can produce multiple mycotoxins (Freire and Sant’Ana, 2018). Exposure in agriculture gets the bulk of scientific attention, and the World Health

Organization and Food and Agricultural Organization of the United Nations have produced regulations outlining acceptable dietary intake levels. It is not surprising that human susceptibility to mycotoxin exposure depends on differing factors like, age, sex, weight, dietand diverse lifestyle habits. Any of these factors may alter the efficiency of the enzymatic pathways involved in the metabolic breakdown of mycotoxins (Milićević et al., 2010). The International Agency for Research on Cancer (IARC) has classified many aflatoxins as carcinogenic (IARC, 1972).

Exposure to mycotoxins causes mycotoxicosis in humans and animals primarily after the ingestion of contaminated food and feed. However, airborne transmission is not excluded as a source of exposure: there are some reported cases, even if they are poorly documented (Hintika and Nikulin, 2010; Engels, 2012; Aleksic, et al., 2017). Mycotoxins are not volatile per se, but can be aerosolized on conidia. For example, the conidia of Stachybotrys chartarum produce mycotoxin-containing fluid droplets through a process called guttation (Gareis and Gotschalk, 2014). Hyphae and other fragments are also suspected to carry mycotoxins, but not enough evidence has been provided, as yet, to fully demonstrate this (Aleksic, et al., 2017). Few studies have tackled mycotoxins in air samples, where concentrations varied from 0.0003 to 0.43 ng/m3 (Borchers et al., 2017). This concentration is low compared to the concentrations detected in food products.

Several human disorders are known to be caused or triggered by forms of fungal exposure other than those mentioned above. Fungal spores contain beta 1,3-D glucan, an antigenic compound, in their cell wall. Inhalation of fungal glucan is involved in changes of neutrophils, macrophages, and eosinophils, as well as in altering concentrations of inflammatory markers in blood samples (Palić et al., 2006). In agricultural and related industries, fungal glucan exposure has been tentatively linked to non-allergic asthma, non- allergic rhinitis, mucous irritation and chronic bronchitis, as well as to changes in airway sensitivity after exposure (Douwes et al., 2003; Douwes, 2005). Organic dust toxic syndrome (ODTS) is another example of a direct toxicosis in which fungal materials are involved. In these cases, persons are exposed to a highly diverse mixture of bacteriaand fungi, plus their toxins and other components (Seifert et al., 2003). In addition, exposure to microbial volatile

organic compounds (MVOCs), which are metabolites of fungi responsible for characteristic mouldy odours, has been suggested to cause irritations of the eyes and upper airways (Korpi et al., 1999).

Fungi are omnipresent and generally are harmless to healthy humans. However, the health conditions discussed in this section, including infections, allergies and toxicoses, still represent a threat to public health. This threat is most important in occupational environments where there may be multiple sources of fungi, and where the types of activity undertakenmay open the door to significant exposures. Thus, assessing occupational exposure to fungi is of great importance.

1.5 Occupational Exposure to Fungi

From an occupational health perspective, enhancing our understanding of the risks of exposure requires attention. The relevance of bioaerosol data is that these data are needed for risk assessment and risk management. Industrial and agricultural environments are major concerns in occupational health due to the potential presence of fungal amplifiers (e.g., in raw organic materials), and the occurrence of operations releasing harmful bioaerosols (e.g., mechanical operations such as wood planing, straw chopping, animal bedding, hay handling, and compost turning). The generation of large amounts of bioaerosols in confined spaces is what makes this type of exposure important in occupational health. Bio-waste facilities are characterized by notable concentrations of fungal aerosols due to the fungal activity responsible of the waste degradation and the daily activities related to composting (Wéry, 2014; Bonifait et al., 2017; Dubuis et al., 2017; Mbareche et al., 2017). Intensive animal farming practices in confined buildings containing many animals (e.g., cows, pigs, poultry, cattle) are also associated with extreme exposure to airborne microbes. The variety of sources (e.g., animals, feces, feed, litter) present on farms leads to the emission of complex mixtures of biological particles (Gilbert and Duchaine, 2009; Lanier et al., 2010; Létourneau et al., 2010; Tsapko et al., 2011; Douglas et al., 2018). Moreover, the constantly changing nature of the microbial composition over time and space makes health risk evaluation complicated

for farm workers and nearby residents (Iversen et al., 2000; Schiffman et al., 2005). Environmental scientists, public health practitioners, and epidemiologists continue to insist that insufficient exposure assessments are one of the main reasons for the absence of bioaerosol exposure limits and strategies to mitigate risk (Douglas et al., 2018; Walser et al., 2015; Mubareka et al., 2019).

For the remaining paragraphs of this section, a description of fungal exposure in biowaste, agricultural, and food processing work environments is presented.

For the last several years, composting has been commonly used for managing organic residue (Sykes et al, 2007). The advantage lies in using the final product as a soil fertilizer in agricultural practices when quality control allows it. The quality of the finished product is linked to the microbiological and chemical characteristics of the compost after the maturation stages (Fuchs et al, 2000). For the microbiological part, the degradation of organic matter is associated with complex microbial communities in aerobic conditions (Ryckeboer et al, 2003). Moreover, filamentous fungi serve as the main actors in composting activities by breaking down complex molecules present in organic matter that are not easily degraded by other microorganisms (Anastasi et al., 2005; Hoorman, 2011; Floudas et al., 2012). As mentioned above, the necessary composting activities of shredding, pile turning and screening release high concentrations of bioaerosols, a risk for workers working eight-hour days (Sanchez-Monedero et al., 2003; Persoons et al., 2012; Taha et al., 2005). Previous studies have linked exposure to bioaerosols from composting facilities to a multitude of diseases and symptoms like tracheobronchitis, sinusitis, mucosal irritations, dermatomycosis and gastro-intestinal problems (Bünger et al., 2000). Considering the importance of fungal diversity in compost piles, fungal bioaerosols are more likely than other types of bioaerosols to cause health issues among compost workers. As most studies describing the diversity of fungal aerosols generated by compost have used traditional culture-based methods, analyzing the this diversity using molecular approaches clearly has potential to yield new epidemiological information.

Biomethanization is another contemporary technology for the management of organic residue, where the waste degraded by the microbial communities under anaerobic condition used to produce biogas made up of approximately 65% methane and 35% carbon dioxide (Mata-Alvarez, 2002). The biogas produced serves as an energy source in the same way as natural gases do (Amon et al., 2007; Lee and Holder, 2001; Karakurt et al., 2011; Dai et al., 2017). In some cases, the gas produced from biomethanization has been used to provide neighbour industries with heat during winter. Among the possible ways of treating various types of organic waste, ranging from municipal green waste to industrial food waste to sewage sludge to animal manure, biomethanization will have the advantage over composting since it is superior in eco-friendliness and cost-efficiency. As with composting environments, biomethanization environments are not exempt from the presence of fungi. Although green- waste degradation happens under anaerobic conditions in immense closed barrels, workers in biomethanization facilities still face a risk of airborne fungal exposure during the reception, and the handling of waste. Considering the speed with which biomethanization is growing as a business activity in Eastern Canada and around the world, a priority needs to be set on describing the fungal exposures encountered in facilities. Dubuis et al. (2017) described the airborne bacterial concentrations and bacterial composition of air samplers from two biomethanization facilities. We believe that our understanding of biomethanization environments could profit from the broad scope of characterizations provided by HTS approaches to fungal diversity.

With the world population rapidly on the increase, food production through farming is set to increase as well. The agricultural practices holding many animals like cows, pigs, poultry, and cattle in densely packed, indoor environments represent an ideal scheme for developing occupational health problems. Animal farming has been linked to a broad range of adverse health problems, consisting mainly of respiratory diseases and gastrointestinal disorders. In addition, residents dwelling near sites offarming activity have been reported to experience unfavourable respiratory conditions. Dairy farms are a particular concern for y fungal exposure owing to the presence of hay and straw, which are naturally colonized by fungi. Under ideally humid conditions, in connection with activities like animal bedding, fungal aerosols can reach high concentrations. Some dairy farmers develop a specific disease

called farmer’s lung, an allergic alveolitis caused by inhalation of a mixture of microbial components, mainly fungi (Madsen et al., 1976; Cormier et al., 1985). Similarly to bio-waste environments, studies describing airborne fungi in dairy farms, qualitatively and quantitatively, have used mainly traditional culture methods. Particularly in studies characterizing diversity, culture methods are constrained in value by their well-known bias of showcasing only the viable and culturable portion of fungal aerosols. Using culture- independent molecular methods offers a better solution in the form of a more exhaustive portrait of fungal diversity. For example, the application of HTS methods can not only be observational, yielding a portrait of fungi present, but can also shed light on the variables affecting the composition of bioaerosol assemblages, and the factors favouring growth of potentially harmful fungi in specific locations. Ultimately, discussions with engineers and farm decision makers could lead to the development of controls on sources of exposure.

Fungi can also become airborne in other environments where they are deliberately introduced, such as in the food industry where fungi are used for production (Morell et al., 2011; Simon & Duquenne, 2014). As an example, cheese factory workers, especially those who brush, clean and pack the final product, are exposed to a large concentration of fungi, and this can alter their respiratory health (Simon & Duquenne 2014). In addition, handling salami is also an occupational situation with high potential exposure to fungi (Morell et al., 2011). Warehouses that stock fruits and vegetables are can be affected by the presence of fungi in the air. This is not surprising, as most of the spoilage organisms affecting fruits and vegetables are fungi (Sherf, 1987). Handling activities in such facilities could lead to the generation of fungal aerosols and workers’ exposure could be amplified. Although the last statement is a plausible hypothesis, no confirmatory evidence has been provided so far.

Recently, a chocolate factory encountered a fungal problem in some of their chocolate ganache recipes. Investigation showed that the problem came from the environment, as multiple air and dust samples showed a considerable quantity and diversity of fungi. The source was probably traceable to straw that the factory had introduced into the warehouse as a part of a special packaging design. As mentioned above, straw is naturally colonized by fungi. With the right temperature and humidity conditions, fungi could have proliferated to

colonize various sites in the factory causing the spoilage of chocolate ganache and eventually putting susceptible workers with impaired immune systems at risk (Mbareche and Duchaine, unpublished data).

Studies in occupational health point towards adverse health effects, especially respiratory symptoms, among workers exposed to high levels of bioaerosols. Establishing health-related exposure limits is crucial in the process of implementing procedures to decrease workers’ exposure to an acceptable level. This field of bioaerosol research could make great progress through use of suitable measurable parameters, development of standardized detection methods for exposure markers, comparison of different detection methods to compensate for the limitations of individual approaches, and application of new analysis methods to identify the potential hazards incurred by workers. As stated by Mubareka et al. (2019), the field of applied bioaerosol research can benefit from having an open research network based on collaboration of different specialists, facilitating the sharing of technical protocol and training programs; the engagement of interested knowledge user groups; and the enhancement of capacity for response measures, including development of best-practices guidelines. Particularly for fungi assessed in workplaces, culture methods have, so far, been the gold standard for training programs, as well as the foundation of knowledge on identification and abundances of relevant fungi.

1.6 Cultivating Fungi

Although HTS methods are increasing the sophistication of fungal studies, the usefulness of data derived from cultivating fungi cannot be underestimated. Isolation of fungal species is essential for understanding phenotypic and genotypic characteristics of individual isolates as well as for a robust assessment of biodiversity. Another great example of the importance of culture methiods is antifungal resistance. Antifungal resistance is a current trend that represents a major clinical challenge in treating invasive fungal infections. As is seen in diseases caused by bacteria, clinicians are faced with a limited stock of available antifungal agents. Antifungal resistance has been found in fungi such as Candida albicans, Aspergillus

fumigatus, Scedosporium, and Fusarium worldwide (Wiederhold, 2017). Even if molecular methods allow the identification of resistance genes, cultivatation of fungi carrying the genes of interest is key for the determination of antifungal resistance phenotypes. Also, culture methods provide a link between antifungal resistance measured in the environment, and antifungal resistance detected in human clinical specimens. In addition, cultivation of fungi in vitro yields information on the range of antifungal minimum inhibitory concentration (MIC) values s, an important factor in invasive fungal infections treatments. In summary, using both molecular and cultivation methods is needed to obtain the full spectrum of resistomes. Another example of the advantages of cultivating fungi is seen with the intracellular parasites, microbes that are capable of growing and reproducing inside a host cell. Some fungal intracellular parasites include Histoplasma capsulatum and Cryptococcus neoformans (sebghati, et al., 2000; Feldmesser et al., 2000) . The combination of optimized culture conditions with molecular detection methods allows a better understanding of the mechanisms used by intracellular pathogens (Orfila, 1996). In general, subjects with a T cell deficiency are particularly affected by intracellular pathogens (Carneiro-Sampaio and Coutinho, 2007). One common example – more related to occupational exposure – of the importance of cultivating fungi is seen in the production and characterization of antibodies to environmentally important fungal species. Such a technique has been used to describe exposure and to gauge workers' susceptibility to fungi in various workplaces (Eduard, 2009). Finally, even some molecular techniques require the cultivation of fungi. Reverse Sample Genome Probing (RSGP), is a method used to analyze microbial composition of the most dominant culturable fungal species by using genome microarrays. One of the steps of the RSGP is the isolation of genomic DNA from pure cultures (Voordouw et al., 1991).

Now that the conclusion is undeniable that fungal culture remains essential in many studies, it must be conceded that effecting this cultivation in diversity studies is problematic. Indeed, most fungal species are difficult to isolate using common culture methods. Growth media and conditions are not known for most fungi. The media that are used commonly (Sabouraud and Rose-Dextrose Agar; Malt Extract Agar) may advantage some species and be less fit for others. The bias could create an imbalance in the census of fungal species appearing to be represented in bioaerosols. A factor making culturing complicated is that the

fungal kingdom is extremely diverse (Blackwell, 2011). As described in the section discussing health outcomes after fungal exposure, viability is not the only condition in which fungi can cause harm once inhaled. Cultivating airborne fungi for concentration measures and diversity analyses is still current amongst bioaerosol researchers. Distinctly, exposure to bioaerosols in waste work environments has has predominantly been studied via culture methods. In a 2016 study, Madsen and collaborators identified 23 fungal species, and a total concentration of fungi up to 4.6 x 104 fungi/m3 inside the cab of a truck during waste collection (Madsen et al., 2016). A more recent study in 2018 has evaluated the risk of occupational exposure to viable fungi in five waste-sorting industries, identifying a total of 13 species with a mean count of 1.6 x 104 total fungi/m3 (Santos et al., 2018). In beginning of 2019, Madsen and collaborators identified 21 fungal species within a personal exposure of 6.5 x 104 CFU/m3 in cardboard waste sorting facilities.

Outdoor environments are also often still characterized using culture methods. In 2016, a fungal exposure analysis in various outdoor settings identified 18 fungal genera, with Cladosporium, Penicillium, Aspergillus, Alternaria, and Rhizopus being the most abundant (Ghiasian et al., 2016). Also in 2018, a study monitoring airborne fungi in different outdoor spaces identified Aspergillus, Penicillium, and Cladosporium as the dominant genera (Ziaee et al., 2018). In addition, a thesis from the University of Iowa, published in 2018, described concentrations and diversity of cultivable airborne fungi in 100 indoor and outdoor locations spanning different climate zones in the United States. The study identified close to 40 fungal species (Messer, 2018). Researchers following the progress of indoor studies are also familiar with the use of culture methods for the assessment of cultivable airborne fungi (Liu et al., 2018). A recent review of common indoor fungi has identified Aspergillus, Penicillium, Cladosporium, Alternaria, and Fusarium as the major fungi in hospital interior environments (Marques do Nascimento et al., 2019). A survey of 71 classrooms has also used cultivation methods to address schoolchildren indoor exposure to fungi (Madureira et al., 2018).

In consideration of the complexity and the diversity of fungi in different ecosystems, one would expect such diversity to be mirrored in bioaerosols too. Yet, occupational, indoor and outdoor bioaerosol studies characterizing fungal exposure around the world refer to the same

major key players in the fungal kingdom, as demonstrated above. Could it be that the underdogs of the fungal kingdom are waiting to be put in the spotlight through application of the right methods? Shade and his collaborators (2012) have demonstrated brilliantly the complementarity of culture-dependent and culture-independent approaches to studying bacterial diversity in soils. The premise of their study is that culture-dependent methods reveal bacteria from among the rare elements of the biosphere and provide information supplemental to that obtained using an HTS approach. The latter method would be beneficial for elucidating the most abundant taxa. This information generates the hypothesis that all these years microbiologist have been studying relatively rare biosphere components (Shade et al., 2012). Thus, combining culture-based methods with HTS can provide a richer portrait of microbial diversity than does either approach alone. Airborne fungi in different environments and conditions deserve the same attention.

A rookie in the -omics team (metagenomics, metatranscriptomics, metabolomics, etc.), which is a part of the culture methods family, appeared in recent years as a promising new technique in microbiology. As indicated in the name, culturomics consists of the application of high-throughput culture conditions combined with 16S rRNA amplification and sequencing or matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) for the identification of previously unidentified colonies. Although the first mention of the term goes back to 2012, culturomics has gained popularity in recent years (Lagier et al., 2012; Lagier et al., 2015; Lagier et al., 2016; Amrane et al., 2018; Kambouris et al., 2018; Hosny et al., 2019). Most of the studies using culturomics have been done to explore the gut microbiota, with a focus on bacterial communities. Future efforts in culturomics could include fungi, as actual knowledge on high-throughput culture conditions optimal for fungi is very limited compared to those applicable to bacteria. For fungi, studies combining culturomics and HTS in the human gut have arrived at the expected conclusions about the complementarity of using both approaches in describing fungal diversity (Hamad et al., 2017).

In conclusion, cultivating fungi alone underrepresents the breadth of airborne fungal diversity and of the complexity of exposure in occupational settings. The application of

microbial molecular ecology methods will provide us with necessary information in order to complete the portrait of fungal diversity.

1.7 Microbiota Analyses

The use of molecular methods in microbial studies is based on the detection of genetic material of organisms present in a given sample. Applying these methods to different environments permits the identification of the microorganisms present and yields a nuanced comprehension of the environmental impacts and ecological roles ascribable to microbial communities. The contrast between molecular biology and culture approaches can be seen in the higher representation of the non-viable or the uncultivable (unknown culture conditions) component of the microbiota in the former approach. When used quantitatively, the molecular approach extends to allowing a calculation of total biomass, including uncultured and cultured components. In bacteriology, it became common knowledge that only a small percentage, less than 1%, of microbes could be recovered by enrichment culture. Although this unexpected selectiveness was first demonstrated for bacteria from soil and aquatic environments (Amann et al., 1995; Pace et al., 1997), similar results for both total and culturable microbes were later obtained for bacteria and fungi in bioaerosols (Peccia and Hernandez, 2006). For example, in 2005, Fabian and collaborators showed that the percentage of culturability for biological structures in outdoor air was 0.1% for bacteria (trypticase soy agar), and 0.001% for fungi (malt extract agar; Fabian et al., 2005). Without a doubt, the reliance on culturing alone is a limitation in bioaerosol studies.

Molecular methods do not assess organisms’ viability or integrity. However, some negative health impacts, like allergies, do not rely exclusively on the viability of microbes. The enormous potential of molecular biology methods for characterizing bioaerosols, via expanding the diversity of microbes that can be detected, identified, and quantified, counterbalance any drawbacks cited by critics. There are major challenges associated with the detection of the nucleic acids used to identify and quantify microbes, and they can be summarized as follows: 1) concentrating and retrieving the biological material from aerosol

samples prior to analyses; 2) choosing the right genomic regions to amplify or sequence in order to obtain the most accurate information about the microbial content of the environment sampled. Specific genes from different organisms can be detected using quantitative real- time amplification (qPCR), which yields a measurement of the concentrations of specific microbes in air. Virulence factor genes, metabolic operons or antibiotic resistance genes can also be quantified from environmental samples (Chizhikov et al., 2001; Martìnez, 2008; Richardson et al., 2004).

David Stahl developed the concept of molecular microbial ecology based on the sequencing of the 16S rRNA gene regions of whole communities and performing subsequent comparisons with databases (Amann et al., 1990). This concept has been used for the last decades to study microbial diversity in environmental samples. Detection and identification of bacteria can be achieved with bioaerosols by using the conserved 16S rRNA-encoding prokaryotic gene, after centrifugation of liquid or filtered-and-resuspended air samples. For fungi, two major challenges are encountered: firstly, their particulates are not efficiently recovered from liquid air samples using the centrifugation protocols used for bacteria and archaea (Mbareche et al., 2019), and, secondly, the choice of a genomic region that will yield a representative image of bioaerosol content is difficult. Two eukaryotic nuclear ribosomal conserved regions, 18S rRNA-encoding gene and ITS regions, can be used depending on the scope of organisms considered in the study. Even though those regions have highly conserved functions in their both eukaryotes and prokaryotes, there are some distinct differences among taxonomic groups in how variable they are. For example, while the V6 region in 16S rRNA gene has been considered to be variable and thus well-suited for assessing bacterial diversity (Huse et al., 2008), its equivalent in the 18S rRNA in eukaryotes is more conserved and is often avoided for that reason (Sogin, 1991; Hadziavdic et al., 2014). Also, one of the difficulties in using molecular methods is that the majority of protocols are specific to individual projects and differ from one study to another.

Although they might not give a direct information on dissemination and transmission of diseases, molecular approaches used to detect DNA sequences from an environment give crucial information.They enable us to link aerosol content directly to the source and thus

yield an improvedevaluation of levels and types of air contamination. For example, using qPCR and detecting high concentrations of Aspergillus fumigatus in a composting environment is a good indicator that workers might be at great risk of developing health problems, regardless of the viability of A. fumigatus material. Thus, using an HTS approach can give an indication about the most abundant microorganisms in an environment. Then, qPCR can be applied to have a concentration per cubic metre of a specific genus. This will add a biomass indicator to the diversity portrait. Finally, the type of exposure can be investigated in regards to the transmission and the dissemination of microbial agents.

The manufacturers of many commercial HTS platforms have become involved in the sequencing business since the completion of the first draft of the human genome by the traditional Sanger sequencing. As a consequence, the cost of genome sequencing has dropped drastically (Schloss et al., 2008). The specific paradigm applied by each platform includes sample preparation andamplification (when used), followed by a massive cycle of parallel sequencing. These steps influence the quality, quantity, and biases of the output sequences, and these factors in turn determine the platforms’ usefulness for specific applications. Since the HTS breakout, some sequencing providers have stopped offering services; therefore, they will not be covered in this section. Some other platforms, although important, are not mentioned because of their infrequent use in microbial ecology studies. Examples include 454 pyrosequencing by 454 Life Sciences, part of Roche Diagnostics; Qiagen-intelligent bio- systems sequencing-by-synthesis by IBS, acquired by Qiagen; Polony sequencing by Polonator; Sequencing-by-ligation by ABI SOLiD; and DNA nanoball sequencing by Complete Genomics, acquired by the Beijing Genomics Institute. Detailed information about the chemistry and sequencing technologies used by these platforms can be found in reviews like the one published by Morey et al., 2013.

Currently, Illumina platforms dominate the HTS market. The sequencing by synthesis is performed on bridge amplicons obtained by PCR amplification. This type of sequencing consists of the polymerase-catalyzed addition of reverse terminator fluorescent-labelled bases. The bases are added simultaneously to the reaction and compete to link with oligo- primed cluster fragments. Only one base can be attached in a cycle, preventing other bases

from subsequent attaching. A coupled-charge device (CCD) camera scans the flow cell capture the image. After each imaging, 3’ blocking is chemically removed, and the process is repeated. Illumina have a set of sequencers, MiSeq, NextSeq, and HiSeq, optimized for various applications. All Illumina models have overall error rates below 0.1%, with the most common one being substitutions (Dohm et al., 2008). The MiSeq is set as a fast, personal benchtop sequencer, with a 4-hour run time and designed for amplicon-based sequencing, and also suited for sequencing of small genomes. The HiSeq series were created for application where a deeper sequencing efficiency is needed. For example, the HiSeq 2500 gives an output of 1Tb in 6 days, while the HiSeq X Ten is capable of giving 1.8Tb in 3 days. Recently, Illumina released the HiSeq 3000/4000 platforms with a run time and an output between the HiSeq 2500 and the HiSeq X Ten (Reuter et al., 2015).

Ion torrent sequencing is another popular platform amongst microbial ecology researchers. Amplification is obtained by emulsion-PCR, similar to the 454 pyrosequencing platform. The sequencing process is based on a semiconductor where hydrogen ions are generated and detected after dNTP insertion, rather than by emission of light like Illumina. The detection is performed by a semiconductor chip that measures pH changes induced by the release of hydrogen ions during DNA extension. Ion Torrent Systems wanted to improve sequencing times by omitting time-consuming imaging steps, which also reduces the overall cost. The original benchtop sequencer is Ion PGM with an output capability of up to 1 Gb and a run speed of 2 hours. A second sequencer was released in 2012, the Ion Proton, yielding tenfold more outputs (10 Gb). However, the read length is 200bp compared to 400 bp for Ion PGM. Ion Proton is more suitable for whole-transcriptome analyses, whereas Ion PGM is suited for targeted sequencing. Recently, a third platform was released under the name of Ion GeneStudio S5; it increased the maximum read length to 600 bp and the output to 1.5 to 4.5 Gb with a run time of 22 hours. Insertions and deletions are the most common error type affecting Ion Torrent chips, occurring at rate of 1% (Quail et al., 2012).

There are two other sequencing technologies, both less popular than Illumina and Ion Torrent, that are worth a mention because of their potential for eclipsing current sequencing technologies in the near future, at least for certain applications. Pacific Biosciences has

commercialized the single-molecule real time (SMRT) system designed by Nanofluidics. Briefly, a capped DNA template is produced by ligation of a single-stranded hairpin adapter onto the ends of the digested DNA. By using a strand displacing polymerase, the DNA molecule is sequenced. While polymerization is occurring continuously, the DNA sequence is read in real time with a video recording the fluorescent signals emitted. Each SMRT cell produces 50k reads (up to 1 Gb) in 4 hours, and an average read length of more than 14kb. But, as with most SMRT technoloies, error rates are high (11%), and consist primarily of miscalls of insertions and deletions. Oxford Nanopore Technologies have developed a nanopore-based single molecule sequencing platform. In brief, DNA is fragmented and adapters are ligated: the first adapter is linked to a motor enzyme and the second is a hairpin oligonucleotide that is linked to a motor protein. The nanopore flow cell contains hundreds of microwells, and each comprises a biological nanopore. A molecular motor protein measures the changes in electrical currents induced when the bases pass through the pores. The output is >90 Mbp in 18 hours’ run time, with a maximum length >60 kb. SMRT sequencing is usually suited for small bacterial and viral whole genome sequencing projects.

HTS methods, after being introduced for the study of single organism genomes, were soon adapted to analyse microbial diversity in bulk environmental samples. One of the most used techniques is the targeted sequencing approach, where a specific gene marker is amplified and sequenced, consistent with the defined goal of species identification. This method is also referred to as metabarcoding, environmental barcoding, or amplicon sequencing. Basically, this approach changed the field of microbial molecular ecology by elucidating whole communities, instead of detecting individual members. In the particular case of the barcode genes, the amplicons produced must be suitable for phylogenetic analyses, a factor that is important for revealing phylogenetic diversity, in the large number of unfamiliar sequences generated in various environments. The accuracy of the results obtained from the amplicon-based HTS approach is dependent on the gene marker selection. These markers are selected based on their presence across the spectrum of organisms of interest and on their possession of sufficient sequence variation to distinguish species. The 16S rRNA gene is the most used gene marker for prokaryotes. However, eukaryotes have many markers options depending on the organisms studied. For example, animals are mostly

studied by targeting the mitochondrial gene COI (Hebert et al., 2003). For plants, the rbcl and matK loci have been used as markers in diversity studies (Hollingsworth et al., 2009). For fungi, the ITS region within the rDNA is considered to be the closest available facsimile of a universal fungal gene marker (Roe et al., 2009; Dentinger et al., 2011; Schoch et al., 2012). The ITS region is composed of three sub-regions, ITS1, 5.8S and ITS2; it is a spacer region separating the small ribosomal subunit 18S from the large ribosomal subunit 28S. One peculiarity of the ITS region is that the sequence length varies between fungal taxa, in contrast to the uniformity in length of the 16S rRNA gene used for bacteria. In Ascomycota and Basidiomycota, ITS lengths range between 600 and 900 bp (Toju et al., 2012). Amplicon- based HTS approaches involve the amplification of the target gene marker before its sequencing. The most commonly used high-throughput sequencers have a maximum read length limitation. For Illumina MiSeq and HiSeq, maximum sequence read lengths are 2 x 300 bp, and 2 x 125 bp, respectively. For Ion Torrent Ion PGM, the maximum sequence length is 400 bp. This restriction forces the choice between ITS1 or ITS2 when applying the amplicon-based HTS approach, as the 5.8S gene does not contain enough informative sites to allow design of primers generating amplicons that can be used in species distinction.

Another molecular ecology method used to profile microbes is metagenomics. This HTS method involves sequencing of all the genomic material in a sample without a specific target enrichment. This method allows the analyses of mixed communities in environmental samples. Other names for metagenomics are environmental genomics or shotgun sequencing; these labels distinguish this method from amplicon sequencing. An advantage of the metagenomics approach is the absence of an amplification step prior to DNA sequencing, avoiding PCR amplification biases and selective primer biases. Because no particular gene is targeted, the functional potential of a sample can be explored, in addition to its taxonomicproperties. Also, the metabolic potentials of organisms from different environmental samples under different conditions can be compared. Depending on the sequencing depth and the genome size, reconstruction of individual microbial genomes can be achieved by bioinformatics’ de novo assembly programs. A recent study published in Nature Microbiology assembled 8000 prokaryote genomes from 1500 metagenome sequences available in public databases (Parks et al., 2017). However, mycologists have not

yet benefited from this advance as the size of eukaryotic genomes and their complexity makes genome assembly from metagenomic samples difficult. Improvements in sequencing platforms and in machine-learning-assisted bioinformatics may make such achievements possible in the future. In addition, taxonomic assignment may be biased towards microbes with the whole genomes available in the database used.

When many genes other than barcode markers are sequenced, the miscellaneous sequenced genes within the samples may be hard to place confidently at the species level. Also, the depth of sequencing needed to give adequate coverage of the diversity contained within an environmental sample is technically higher if shotgun sequencing is used rather than amplicon sequencing. As mentioned above, as HTS and SMRT sequencing technologies advance, output increases, and costs decrease. Given the attractive advantage of PCR-free processing, and the availability of continually growing databases of annotated genomes, the popularity of metagenomics is expected to continue to grow among bioaerosol researchers.

Although the sequencing-based approaches described in this section are popular and their use is widespread amongst microbial ecology investigators in general, though only recently for bioaerosol specialists, they have important limitations. Certainly, bioaerosol studies applying modern microbial molecular ecology methods should move from purely descriptive studies to bringing causally connected answers and solutions to decision makers, ultimately meeting the objective of minimizing exposures and improving occupational health. This goal can be achieved through continuing the development of standardized laboratory methods, improving bioinformatics algorithms and the skills deployed in their use, curating more complete and better annotated databases, and combining data from different analyses. A perfect format for data integration would be combining HTS outputs and culturomics to improve the understanding of microbial communities. Reference databases could be improved via the help of professional taxonomist, whose expertise can facilitate the identification of new species. If such steps are taken, the relative abundance of unidentified taxa will decrease. As amplicon-based HTS approaches depend on databases for taxonomic identification, bioaerosol exposure studies applying these approaches while linked to

taxonomically curated databases could identify an ever larger number of airborne microbes, thus yielding an increasingly accurate picture of the real diversity of bioaerosols.

In addition to problems caused by the characteristic error types and rates associated with each sequencing platform, the high accuracy of homopolymers, and the difficulty of amplifying GC-rich regions, most current platforms suffer from yielding short read lengths that limit our ability to characterize large regions, particularly where there are insertions and deletions, as well as structural variation (Ross et al., 2013). Improvements in the long read sequencing technologies can help overcome these limitations.

The challenges arising during use of sequencing technologies can be managed from a bioinformatics point of view. Indeed, bioinformatics analyses can have an impact on diversity analyses such as clustering algorithms, the percent identity threshold and taxonomy assignment tools (BLASTn vs. Naïve Bayesian Classifier). The bioinformatics part of HTS studies has become an unavoidable component of this field of science devoted to integrating microbial molecular ecology studies, and studies focusing on bioaerosols are no exception.

1.8 Bioinformatics

The avalanche of big data generated at an extraordinary rate by HTS platforms has made bioinformatics indispensable to modern molecular ecology. Actually, there is a certain reluctance about bioinformatics among some scientists due to the command line tools and programming skills needed for sequence processing and diversity analyses. Therefore, bioinformatics should be rendered accessible to the majority of potentially sufficiently daring users. Knowing which analyses to conduct and which tools to apply remains confusing for bioaerosol scientists, as a litany of tools and data resources are now available for characterizing microbial communities. Currently, microbial molecular data from bioaerosol studies are obtained predominantly using 16S rRNA gene sequencing surveys that provide a characterization of bacterial and archaeal diversity. In studies using 16S rRNA-encoding gene sequencing, the choice of primer set depends on a number of factors, including

maintaining compatibility with previous studies and the managing the specificities of the primers (Soergel et al., 2012). 16S rRNA gene sequence data from different microbial environments present bioinformatical, statistical, and computational challenges. The most widely used bioinformatics tools are QIIME and mothur (Schloss et al., 2009; Caporaso et al., 2010b). Both packages are open source and have online tutorials and forums that take the users step-by-step through an analysis of the 16S rRNA gene sequences.

When it comes to HTS, PCR errors are common, and are sometimes difficult to detect. Chimeras, which are caused by incomplete extension of DNA strand during amplification that make up a recombination between two sequences, can cause biases in diversity results, particularly the alpha metrics (Ley et al.., 2008; Haas et al., 2011). Multiple chimera checking softwares are available and they often have different filtering methods. One should look for the one that fits adequately with the project (Ley et al.., 2008; Edgar et al., 2011; Quince et al., 2011; Wright et al., 2012).

Data filtering is often made at the same stage as demultiplexing where each sequence is assigned to a sample based on barcode information. Quality threshold parameters include quality scores, read length and the presence of ambiguous base calls. After quality filtering, one way to identify the microbial groups is through OTU analyses where sequences are clustered together based on sequence identity. As a matter of fact, there are several OTU clustering algorithms and the use of any one greatly affects the data interpretation. In de novo OTU clustering, sequences are clustered into OTUs by comparing them to the whole dataset, without the use of a reference (Schloss et al., 2005). In contrast, closed OTU clustering uses a reference sequence database, where the sequences that don’t match members of the reference database are discarded. Open-reference OTU clustering is a two-step process combining both the algorithms just described. First, closed-reference OTU clustering is done, followed by de novo clustering of sequences that fail to match to the reference database. Each of the clustering methods have pros and cons and a choice should be made with particular care. For example, if one must combine datasets that have sequences from different regions of the 16S rRNA gene, the closed reference algorithm should be used, as sequences from different regions of the same 16S rRNA gene would cluster into different de novo OTUs. In

contrast, using the closed reference algorithm in samples coming from an undiscovered environment may lead to a large percentage of the sequences being discarded due to failure of the sequences to match anything in the reference database. A new method has emerged that produces exact sequence variants (ESVs) instead of OTUs for a greater resolution than OTU-based methods. DADA2 processes data from fastq files, removes errors and chimeras, and produces sample abundances and taxonomic assignments (Callahan et al., 2016). Other synonyms of ESVs are amplicon sequence variant (ASV), and zero radius OTU (ZOTU). Another way of expressing this concept is simply as an OTU defined by 100% sequence similarity.

Labelling the OTUs is the next step, accomplished using a taxonomy assignment algorithm combined with a reference database. For the 16S rRNA gene, the three main databases are Greengenes, Ribosomal Database Project (RDP) and SILVA (Cole et al., 2009; McDonald et al., 2012; Quast et al., 2013). Microbial diversity metrics are typically calculated within samples (alpha diversity) and between samples (beta diversity). Multiple scripts for data visualization are available through QIIME, mothur and R. Particular attention should be given when choosing the classification and clustering methods depending on the metadata information available about the samples. For example, classification methods can be used to determine which taxa differ between predefined groups of samples. On the other hand, clustering methods does not make use of any prior knowledge about the samples. Both methods use in-between sample distance metrics. The specificities of the methods used can affect the interpretation of the analyses. For this reason, it is mandatory to use several different ways of classification and clustering to ensure that the existence of clusters is not dependent on just one set of parameters.

Diversity analyses include alpha and beta measures, statistical significance of sample groupings, estimations of differential abundance of taxa in different groups of samples, taxonomic analyses, and the search for correlations between abundances (relative or absolute) and numerical metadata (e.g., days, temperature, relative humidity, etc.). Different taxonomic assignment methods exist , and are not specific to amplicon-based HTS: similarity-based (BLAST, BOLD, METAPHYLER, MG-RAST); phylogeny-based (NJ,

SAP, PPLACER, EPA); composition-based (KNN Classy method by mothur, OTU-picking by QIIME, RDP Classifier, UCLUST); hybrid methods (FUZZYD2, MEGAN); Bayesian treeless methods (Coalescent Assigner, Segregating Sites Assigner); and, Machine learning (BPSI, SVM classifier). Finally, statistical modelling can be used to explain the variation among different samples. Since many covariates can be included, and just as many parameters can be statistically controlled, we strongly recommend the consultation of a statistician or a biostatistics expert during experimental design and analysis.

Characterization of airborne fungal communities is a research area that needs more activity. Although the bioinformatics pipeline may seem to be the same for eukaryotic marker genes as for bacterial marker genes, there are no standardized bioinformatics protocols as there are for 16S rRNA gene analyses. The lack of a standard marker gene and a reference database may be the reason behind that. As mentioned earlier in this chapter, several marker gene options exist. However, ITS region is generally preferred for obtaining high taxonomic resolution. Due to amplicon length limitations imposed by the sequencing platforms, only one of the two ITS subregions (ITS1 or ITS2) can be used in a sequencing run. To the best of our knowledge, there are no relevant studies that explored the potential of both sub-regions in the characterization of fungal communities in bioaerosols. A strategy explored in this thesis is using both subregions, coupled with shotgun metagenomics, to demonstrate their potential in a global fungal characterization leading to a standardized methodology for the assessment of fungal exposure in occupational settings.

The bioinformatics workflow used for the ITS sequences differs greatly from the one for the 16S rRNA gene in the alignment algorithm, since the ITS region differs in length from the 16S rRNA gene. For this reason, amongst others, when ITS sequences analyses are used for the characterization of fungal communities, use should be made of specific OTU clustering, reference databases, taxonomy assigning methods, diversity metrics, data visualization and statistical modelling The UNITE database is often used for ITS sequence- based analyses of fungal sequences (Abarenkov et al., 2010). On the other hand, the 18S rRNA gene can, generally, be used to analyze eukaryotic communities in the same manner as 16S rRNA genes are used. The SILVA database contains many eukaryotic regions and

can be used for such analyses. However, one should confirm that the region of the 18S gene amplified distinguishes the taxa studied and that the 18S rRNA gene is generally not adequate to characterize the fungal diversity and will likely end up to be of questionable utility.

Comparing taxonomy information of bioaerosol samples to a mock community sample can help in determining technical biases linked to sequencing approaches. For example, the comparison of the expected mock community to the actual sequencing results of the mock community can help inform on the taxa that were higher or lower in abundance. These differences between abundances of taxa may be due to the advantage giving to some taxa by the primers used for the amplification step prior to sequencing. Then, the relative abundance of the different taxa identified in the actual samples can be adjusted to take into account the primers choice bias.

Although the bioinformatics workflows described above were developed for certain target environments, many of the same issues apply to microbial communities of different habitats. Thus, they can be applied to bioaerosol samples as well. In general, the different projects presented in this thesis encourage researchers entering the field to use bioinformatics tools to analyze the microbial molecular ecology results obtained in different bioaerosol environments. This active research area needs more standardized protocols that will help build a bioaerosol public database that will allow cross-comparative studies around the world and give this field of study great dash.

1.9 Future of Bioaerosols

Although the importance of bioaerosols in various human diseases including infections, respiratory symptoms, and cancers has been long-acknowledged, many uncertainties still remain in exposure assessment. Bioaerosol study is made difficult by the multiple factors affecting bioaerosol content, the variability in the focus of different exposure studies, the variations seen in experimental design, and the lack of standardization in methods. Therefore, the health impacts of bioaerosol exposure are still poorly understood.

The progress made with molecular methods in recent years, including real-time quantitative PCR, HTS (amplicon-based, shotgun, whole genome), and biosensors is considerable. Undoubtedly, these methods have helped to paint a more extensive portrait of occupational exposure to bioaerosols. For example, understanding the effect of air samplers on the perceived microbial content of bioaerosols allowed for wiser choices among field sampling devices, depending on the application. This statement is crucial, as air sampling devices constitute our first epistemologic contact with air; thus, a biased air sampling device equals a biased analysis. That said, having a broader view of the microbial content of bioaerosols alone does not provide concrete solutions for evaluation of risk. The future of bioaerosol science relies on designing new studies that encompass all of the key elements, going beyond the descriptive alone, and taking the steps necessary to understand the real impacts of exposure in the short-term and long-term, as well as preventing the replication of previous errors.

Recommendations for future work include: the identification of a core microbiome associated with each type of environment; identifying environmental exposure markers; enhancing and encouraging collaborations between specialists among all fields related to bioaerosol exposure (scientists, physicians, engineers, managers, etc.); using peer-reviewed, reproducible, and reliable standard methods; and finding new ways to make the results visible and accessible to all concerned parties. With collaborative efforts, clear health associations with bioaerosol exposure will begin to emerge. Therefore, more research is desirable to properly establish better assessment tools for measuring the exposure to bioaerosols and validation of results. As indicated above, bacteria have been the stars of microbial molecular ecology studies, especially in studies about bioaerosols. Fungi clearly need more research in this area to improve our understanding of their significance in bioaerosol exposures. As published by the editorial board of Nature Microbiology and several other high-impact factor journals lately, biomedicine must stop forgetting about and underestimating fungal diseases (Almeida et al., 2019; Casadevall, 2017; Nature Microbiology, 2017).

1.10 Specific Aims of the Thesis

The present thesis was conceived with three primary objectives. First, this research shall extend the comprehension of the potentials of the two universal fungal barcodes, ITS1 and ITS2, in capturing the fungal diversity of aerosols, including the role of shotgun metagenomics in comparing the performance of the two markers. A second objective of this investigation is to describe the issue of fungal cell loss when propagules are recovered using air liquid samplers, and to propose an alternative method that will better evaluate fungal exposure and diversity. Finally, the thesis aims to use the methodology developed in fulfilling the first two objectives to describe fungal occupational exposure in three environments of concern. Along the way, a final project will be presented: a review on the use of bioinformatics tools to carry out successful bioaerosol studies that use microbial molecular methods. The methodology described was applied in order to treat sequences and analyze their diversity in all the chapters of this thesis.

Part one: Methods for the Characterization of Fungal Aerosols

Chapter 2: Comparison of the performance of ITS1 and ITS2 as barcodes in amplicon-based sequencing of bioaerosols

2.1 Résumé

De par les activités qui y sont menées, certains milieux de travail génèrent de grandes quantités d’aérosols, incluant des moisissures, qui proviennent de la matière présente dans l’environnement. La présence de moisissures dans l’air a d’abord été confirmée par l’emploi de méthodes d’analyse par culture. Ces méthodes classiques induisent un biais dans la diversité fongique tant au niveau qualitatif que quantitative. L’objectif de cette étude est d’appliquer une méthode moléculaire basée sur le séquençage d’ADN à haut débit pour combler les lacunes des méthodes de culture. Deux marqueurs génomiques fongiques ont été ciblés par l’approche de séquençage, soit l’ITS1 et l’ITS2, afin de comparer leur efficacité à décrire la diversité fongique d’un environnement. Les résultats obtenus suggèrent l’utilisation de la région ITS1 comme marqueur fongique universel, en raison de sa capacité à couvrir plus de diversité qu’ITS2 et à capturer des profils taxonomiques semblables à ceux obtenus par l’approche métagénomique.

2.2 Summary of the Paper

This paper present the performance of two eukaryote genomic ribosomal regions, ITS1 and ITS2, in describing fungal diversity in aerosol samples using amplicon-based HTS. Composting sites, biomethanization facilities, and dairy farms, all affected by the presence of fungi, were visited to collect air samples. The amplicon-based HTS approach is target enrichment method that relies on the amplification of a specific target using particular primers before sequencing. Thus, the results are highly dependent on the quality of amplification. For this reason, the authors of this paper used a shotgun metagenomic approach to compare its outcome with the amplicon-based method. Indeed, shotgun metagenomic does not rely on any amplification prior to sequencing, because all genes are sequenced without a specific target. In addition, culture method were added to the analyses in biomethanization and dairy farms samples to validate their contribution to fungal diversity of aerosols. The bioinformatics protocol applied for diversity analyses include richness and diversity measures, multivariate analyses, differential abundance, taxonomic profiles, and statistical validation.

The results obtained are unequivocal towards ITS1 outperformance to ITS2 in terms of richness, and taxonomic coverage. The differential abundance analysis did demonstrate that some taxa were exclusively detected only by ITS2, and vice-versa for ITS1. However, the shotgun metagenomic approach showed a taxonomic profile more resembling to ITS1 than ITS2. Based on these results, neither of the barcodes evaluated is perfect in terms of distinguishing all species. Using both barcodes offers a broader view of the fungal aerosol population. However, with the actual knowledge, we strongly recommend using ITS1 as a universal fungal barcode for quick general analyses of diversity and when limited financial resources are available, primarily due its ability to capture taxonomic profiles similar to those obtained using the shotgun metagenomic approach.

Comparison of the performance of ITS1 and ITS2 as barcodes in amplicon-based sequencing of bioaerosols

RUNNING TITLE Metabarcoding analysis of fungal aerosol AUTHORS Hamza Mbareche1,2, Marc Veillette1, Guillaume J. Bilodeau3, and Caroline Duchaine1,2 AUTHORS’ AFFILIATION 1. Centre de recherche de l’institut universitaire de cardiologie et de pneumologie de Québec, Quebec City (QC), Canada 2. Département de biochimie, de microbiologie et de bio-informatique, Faculté des sciences et de génie, Université Laval, Quebec City (QC), Canada 3. Pathogen Identification Research Lab, Canadian Food Inspection Agency (CFIA), Ottawa, Canada

KEYWORDS Bioaerosols; Fungi; ITS1; ITS2; Compost; Biomethanization; Dairy farms; Illumina MiSeq sequencing

CORRESPONDING AUTHOR Mailing address: Caroline Duchaine, Ph.D., Centre de Recherche de l’Institut Universitaire de Cardiologie et de Pneumologie de Québec, 2725 Chemin Ste-Foy, Québec, Canada, G1V 4G5. Phone: (418) 656-8711 ext. 5837. Fax: 418 656-4509. E-mail: [email protected]

Submitted for peer review

2.3 Abstract

Fungal spores are ubiquitous in the air and their diversity and quantity vary depending on growth, the geographical location, environmental conditions, and the presence of sources. The health effects linked to this exposure is likely underestimated due to the presence of undocumented fungi. Culture methods are still widely used to describe fungal aerosols. However, there are inherent biases associated with these methods when applied to fungal diversity analyses because most fungal species are difficult to isolate from culture. Amplicon- based sequencing approaches depend on the critical choice of which DNA region is used as the barcode. These universal markers are selected based on different criteria such as their presence across taxa and sufficient sequence variation between taxa. The limitations imposed by the sequencers in term of amplicon size forces the use of ITS1 or ITS2 to study fungal diversity of bioaerosols. The objective of this work is to use air samples from composting sites, biomethanization facilities, and dairy farms to make a systematic comparison of the performance of ITS1 and ITS2 in determining the fungal diversity of bioaerosols. In addition, shotgun metagenomics was applied to air samples from dairy farms in order to compare its results with the HTS approach based on ITS1 and ITS2. Furthermore, diversity was measured using the culture method to evaluate its contribution to fungal bioaerosol studies. The results obtained suggest that neither of the barcodes evaluated is perfect for distinguishing all species. However, we strongly recommend the use of ITS1 as a universal fungal barcode for quick general analyses of diversity, and when limited financial resources are available (because of its capacity to cover more richness and diversity than ITS2). We recommend using ITS1 mainly because of its ability to capture taxonomic profiles similar to those obtained using the shotgun metagenomic approach. The culture comparison with amplicon- based sequencing showed the complementarity of both approaches in describing the most abundant taxa.

2.4 Introduction

Natural air contains a class of particulate matter of biological origin referred to as bioaerosols. Bioaerosols include: living and dead fungi and bacteria, viruses, bacterial

endotoxins, mycotoxins, β (1, 3)-glucans, pollens and other airborne allergens, etc. (Macher et al., 1999; Douwes et al., 2003; Després et al., 2012). Both natural and artificial (manmade), bioaerosols are ubiquitous, highly variable and complex to identify, mainly because they can originate from any source (e.g. plants, soil, water, humans; Paez-Rubio et al., 2005; Taha et al., 2006; Hospodsky et al., 2012). Their composition depends on the source, aerosolization mechanisms and environmental conditions (Foarde et al., 1993; Pasanen et al., 2000; Pillai and Ricke, 2002; Jones and Harrison, 2004; Bonifait et al., 2017; Mbareche et al., 2017). The dispersal of bioaerosols can have major impacts on public health through their effects from inhalation and potential ingestion. The inhalable particles can reach deep parts of the respiratory system, causing a wide range of acute and chronic diseases such as allergies, asthma, rhinitis, sinusitis and bronchitis. The particles can also lead to the dispersal of pathogens and adverse health effects from occupational exposure (Brown et al., 2002; Douwes, 2003; Brodie et al., 2007; Eduard et al., 2012; Heederik et al., 2012). Bioaerosols are also recognized as transmission routes for infectious diseases (Roy and Milton, 2004; Yu et al., 2004; Eames et al., 2009; Li et al., 2007).

Fungal spores are ubiquitous in the air and their diversity and/or concentration vary depending on the climate, geographical conditions and the presence of fungal growth sources in the environment (Ruzer and Harley, 2005; Kakde, 2012). Industrial activities that are mainly linked to agriculture have some of the highest rates of fungal exposure in their environments. These conditions are created by the presence of decaying materials such as hay, peat, wood dust, manure, biosolids, and organic wastes, like compost (Gilbert and Duchaine, 2009). Dairy farms and waste treatment sites are examples of such environments where humans are exposed to a wide variety of fungi (Mbareche et al., 2017; Mbareche et al., 2018a; Mbareche et al., 2019a). Fungi can also become airborne in environments where they are deliberately introduced, such as in the food industry where fungi are used for production (Morell et al., 2011; Simon & Duquenne, 2014). The health effects of fungal exposure range from relatively serious effects such as allergy related diseases, pulmonary inflammation, increased sensitivity to endotoxins, and pulmonary embolisms to milder effects such as bronchial irritation, mucous membrane irritation syndrome, nasal congestion, sore throat, and irritation of the nose and eyes (Wyngaarden et al., 1992; Fogelmark et al.,

1994; Rylander, 1996; Burge and Rogers, 2000; Arshad et al., 2001; Hardin et al., 2003; Daisey et al., 2003; Pieckova and Wilkins, 2004; Stark et al., 2005; Zekovic et al., 2005; Beezehold et al., 2008; Bush et al., 2008; Porter et al., 2009; Sarkar et al., 2010; Selman et al., 2009; Glass and Amedee, 2011; Chowdhary et al., 2014). Exposure to a variety of fungi can also result in infections, especially in people with impaired immune systems (Latgé, 1999; Nucci and Anaissie, 2007; Rodriguez and Ramos, 2014; Vlegraki et al., 2015). However, the impact of fungi on occupational health is still largely underestimated. The numerous fungi that are still undocumented present a barrier to establishing a clear link between respiratory problems and fungal exposure (Bush et al., 2006; Tischer et al., 2013).

Historically, fungi have been identified based on the morphological characteristics of pure cultures in agar media. This process has also been used in more recent exposure studies (Sanchez-Monedero et al., 2006; Taha et al., 2006; Schlosser et al., 2012; Park et al., 2013; Ghiasian et al., 2017; Liu et al., 2018). However, most fungal species are difficult to isolate using common culture methods. Therefore, using these techniques may lead to an underestimation of the fungal diversity in bioaerosols, especially considering that the kingdom of Fungi is one of the most diverse (Blackwell, 2011). Molecular methods are offering new perspectives on the occurrence and the ecological impact of microbes. One of these methods includes high-throughput sequencing (HTS), which offers a more thorough analysis of the microbial content of a sample because of the millions of sequences that are generated. Using the appropriate bioinformatics tools, this technology can characterize thousands of species, referred to as OTUs (operational taxonomic units), from environmental samples. The success of the amplicon-based HTS approach resides in the critical decision of which DNA region to use as the barcode. These universal phylogenetic markers are selected based on a number of criteria, including their ubiquitous presence across taxa and having sufficient sequence variation between taxa. While using the small ribosomal DNA (rDNA) subunit 16S is the obvious choice for prokaryotes, eukaryotic species present more challenges for the metabarcoding community. For example, the mitochondrial gene COI has been used as a universal barcode for animals (Hebert et al., 2003). The combination of the rbcl and matK genes has been proposed as the universal plant barcode (Hollingsworth et al., 2009). For fungi, the internal transcribed spacer (ITS) region of rDNA is considered to be the best

barcode for most fungal groups (Roe et al., 2009; Dentinger et al., 2011; Schoch et al., 2012). The ITS region contains three partitions: ITS1, 5.8S and ITS2. The length of the ITS sequence is highly variable from one fungal species to another and it is strongly dependent on the primers used to target the DNA sequence. For example, in Ascomycota and Basidiomycota the sequence lengths range between 600 and 900 bp (Toju et al., 2012). Amplicon-based HTS approaches involve an enrichment step prior to sequencing, which involves using PCR amplification of the targeted barcode. The most commonly used high- throughput sequencers have a maximum read length limitation (MiSeq Illumina – 2 x 300 bp; HiSeq 2500 Illumina – up to 2 x 125 bp; Ion PGM – 400 pb). This limitation forces the use of only one of the two sub regions (ITS1 or ITS2) when applying the amplicon-based HTS approach for determining fungal diversity. The 5.8S region does not contain a sufficient number of informative sites that can be used for phylogenetic studies and DNA barcoding. Studies using environmental samples from soil, mangroves, plants and aquatic ecosystems or retrieved ITS sequences from GenBank (using ‘internal transcribed spacer’ as the keyword) gave mixed reviews on the performances of ITS1 and ITS2 in documenting and characterizing fungal biodiversity (Kelly et al., 2011; Arfi et al., 2012; Ihrmark et al., 2012; Heinrichs et al., 2012; Osmundson et al., 2013; Blaalid et al., 2013; Bazzicalupo et al., 2013; Monard et al., 2013; Kohout et al., 2014; Op De Beeck et al., 2014; Wang et al., 2014; Ishii et al., 2015; Tedersoo et al., 2015). Comparing studies on fungal diversity from different scientific fields is challenging/difficult/nearly impossible, due to their use of different barcodes and methods of analyses.

Another HTS approach, referred to as shotgun metagenomics, consists of the untargeted sequencing of all microbial genomes in a sample (Quince et al., 2017). Shotgun sequencing is not subject to amplicon length limitations or the PCR biases imposed by HTS approaches based on ITS1 or ITS2. However, shotgun sequencing presents many other known limitations. The most important one is that there can be a low relative proportion of ribosomal DNA from the metagenomes of the microbes of interest (fungi in this case) compared to the pool of genes that are present in samples. This may complicate the detection of fungal species when there is no enrichment.

This large-scale study uses air samples from waste treatment sites and dairy farms to systematically compare the performance of ITS1 and ITS2 in metabarcoding analyses of fungal diversity in aerosols. The study was designed as a result of discussions about which amplicon-based HTS approaches would best describe aerosol fungal exposure. The analyses include sequence length distribution, richness and diversity indexes, multivariate analyses, differential abundance, species discrimination efficiency and taxonomy analyses. In addition, shotgun metagenomics was applied to air samples from dairy farms in order to compare its results with the HTS approach based on ITS1 and ITS2. Furthermore, diversity was measured using the culture method to evaluate its contribution to fungal bioaerosol studies. This work provides new insights into the use of both ITS sub-regions in order to assess fungal aerosol populations and also provides a guide for which strategies to use for analyzing particular taxonomic groups. The results obtained suggest that neither of the barcodes evaluated is perfect for distinguishing all species. However, we strongly recommend the use of ITS1 as a universal fungal barcode for quick general analyses of diversity, and when limited financial resources are available (because of its capacity to cover more richness and diversity than ITS2). We recommend using ITS1 mainly because of its ability to capture taxonomic profiles similar to those obtained using the shotgun metagenomic approach. The culture comparison with amplicon-based sequencing showed the complementarity of both approaches in describing the most abundant taxa.

2.5 Materials and Methods

Description of the Environmental Conditions of Field Sampling

Compost

In 2014-2015, two different composting plants located in the province of Quebec, Canada were visited during a year-long sampling schedule to monitor the composting processes. The composting plants treat two different types of raw materials: household green waste (domestic) and pig carcasses and placenta (animal). Detailed information about the

sampling schedule and conditions are presented in the original composting study report (Bonifait et al., 2017).

Biomethanization

Samples were collected from two different biomethanization facilities during the summer of 2015 and the winter of 2016. One facility (BF1) processes primary and secondary sludge from wastewater treatment plants, as well as industrial waste. The second one (BF2) handles municipal waste from domestic sources. Detailed information about the sampling sites and conditions can be found in the original study report (Dubuis et al., 2017).

Dairy farms

Air samples were collected from five dairy farms in Eastern Canada during the summer of 2016. At each farm, a sampling site was designated based on where activities that generate the most bioaerosols took place. The buildings at each farm exhibited differences in: building type and characteristics (age, volume, ventilation), number of animals present (cows), method of milking (automatic or manual), and type of animal feed given. Detailed information about the sampling sites and conditions can be found in the original study report (Mbareche et al., 2019a).

Air Sampling

A liquid cyclonic impactor Coriolis µ® (Bertin Technologies, Montigny-le- Bretonneux, France) was used for collecting air samples. The sampler was set at 200 L/min for 10 minutes (2m3 of air per sample) and placed within 1-2 meters of the bioaerosol source. The sampling sites were chosen according to workers’ activities. The airflow in the sampler creates a vortex through which air particles enter the Coriolis cone and are impacted in the liquid. Fifteen milliliters of a phosphate buffer saline (PBS) solution with a concentration of 50 mM and a pH of 7.4 were used to fill the sampling cone.

Culture-based Approach to Study Fungal Diversity

One millilitre of the 15ml Coriolis sampling liquid was used to perform a serial dilution from 100 to 10-4 concentration/ml. The dilutions were made using 0.9% saline and 0.1% Tween20 solution and were performed in triplicate. Tween20 is a detergent that makes spores less hydrophobic and easier to collect. One hundred microliters of each triplicate were plated on Rose Bengal Agar with chloramphenicol at a concentration of 50 µg/ml. Half of the petri dishes were incubated at 25°C for mesophilic mould growth and the other half at 50°C for thermophilic mould growth, specifically the fungus/mould Aspergillus fumigatus. After 5 days of incubation, the moulds were identified and the counts were translated into CFU/m3.

Identification of Isolates

Spores from cultured fungi were recovered in one milliliter of a 0.9 % saline and 0.1% Tween20 solution and stored in an Eppendorf tube. Two hundred microliters of the collection liquid were placed in an FTA card (sample collection card; Qiagen, Mississauga, Ontario, Canada). Five punches from the spiked zone of the FTA card were placed in a microtube and washed three times with the FTA purification agent. The washing step is mandatory as it removes the chemical substrates in the FTA card that may alter the subsequent amplification step. Forty-eight microliters of the master mix solution (described in Suppl. Material 1) were placed in each microtube followed by the amplification and sequencing of the ITS genomic region. The protocol described by White and his collaborators (White et al., 1990) was performed at CHU (Centre hospitalier de l’Université Laval). The following oligonucleotides were used for the ITS region amplification: ITS1: 5’-TCCGTAGGTGAACCTGCGG-3’ ITS4: 5’-TCCTCCGCTTATTGATATGC-3’ The isolates were identified by comparing the sequences obtained with sequences in the UNITE 7.2 database with the BLASTn option.

Concentration of Fungal Spores in Aerosols

The following method was used because it is optimal for recovering fungal spores from air samples, as described in detail by Mbareche and his coauthors in 2019 (Mbareche et al., 2019b). Briefly, the liquid suspension from the Coriolis cone was filtered through a 2.5cm polycarbonate membrane (0.2-mm pore size; Millipore) using a vacuum filtration unit. The filters were flash-frozen and pulverized using a tungsten steel bead in an Eppendorf tube in a bead-beating machine (a Mixer Mill MM301, Retsch, Düsseldorf, Germany). Aliquots of the liquid containing the pulverized filter particles were used for the first step of the DNA extraction procedure.

DNA Extraction

Using the same apparatus, bead-beating was performed a second time using glass beads at a frequency of 20 movements per second for 10 minutes to ensure that all of the cells were ruptured. Next, a MoBio PowerLyserÒ PowersoilÒ Isolation DNA kit (Carlsbad, CA, U.S.A) was used to extract the total genomic DNA from the samples following the manufacturer’s instructions. Then, DNA was eluted in a 100µl MoBio buffer and stored at - 20°C until subsequent analyses.

MiSeq Illumina Sequencing

The amplification primers chosen for amplicon-based HTS were based on Tedersoo and his colleagues in their analyses of primer biases in fungal metabarcoding (Tedersoo et al., 2015). Amplification of the amplicons, equimolar pooling and sequencing were performed at the Plateforme d’analyses génomiques (IBIS, Université Laval, Quebec City, Canada). Briefly, amplification of the ITS regions was performed using the sequence-specific regions (ITS1 and ITS2) described by Tedersoo et al. (2015) and references therein, using a two-step dual-indexed PCR approach specifically designed for Illumina instruments. First, the gene-specific sequence was fused to the Illumina TruSeq sequencing primers. Next, PCR

was carried out on a total volume of 25µL of liquid made up of: 1X Q5 buffer (NEB), 0.25µM of each primer, 200µM of each of the dNTPs, 1U of Q5 High-Fidelity DNA polymerase (NEB) and 1µL of template cDNA. The PCR started with an initial denaturation at 98°C for 30s followed by 35 cycles of denaturation at 98°C for 10s, annealing at 55°C for 10s, extension at 72°C for 30s and a final extension step at 72°C for 2min. The PCR reaction was purified using an Axygen PCR cleanup kit (Axygen). The quality of the purified PCR products was verified with electrophoresis (1% agarose gel). A dilution of 50 to 100-fold of this purified product was used as a template for a second round of PCR with the goal of adding barcodes (dual-indexed) and the missing sequences required for Illumina sequencing. The conditions for the second round of PCR cycling were identical to the first PCR, but with 12 cycles. The PCR reactions were purified as above, checked for quality on a DNA7500 Bioanlayzer chip (Agilent) and then quantified spectrophotometrically with a Nanodrop 1000 (Thermo Fisher Scientific). Barcoded Amplicons were pooled in equimolar concentration for sequencing on the illumina Miseq. The primer sequences used for amplification are presented in Table 2.1.

For the shotgun metagenomics, library preparation and sequencing was also performed at the Plateforme d’analyses génomiques (IBIS, Université Laval, Quebec, Canada). In brief, Genomic DNA (500 ng in 55ul) was mechanically fragmented for 40 sec using a covaris M220 (Covaris, Woburn MA, USA) with default settings. Fragmented DNA was transferred to PCR tubes and library synthesis was performed with the NEB Next Ultra II (New England Biolabs) according to manufacturer’s instructions. TruSeq HT adapters (Illumina, SanDiego, CA, USA) were used to barcode the samples. The libraries were quantified and pooled using an equimolar ratio and sequenced on an Illumina MiSeq 300 base pair paired-end run (600 cycle, v3 kit).

Table 2.1 Primers used for amplification of ITS1 and ITS2 barcodes and for Illumina Miseq sequencing

Primers name Features Sequence Barcode PCR ITS1Fngs Fwd, tagged ACACTCTTTCCCTACACGACGCTCTTCCGATCTGGTCATTTAGAGGAAGTAA ITS1 First ITS2 Rev GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGCTGCGTTCTTCATCGATG ITS1 First C

ITS3tagmix1 Fwd ACACTCTTTCCCTACACGACGCTCTTCCGATCTTAGACTCGTCATCGATGAAG ITS2 First AACGCAG

ITS3tagmix2 Fwd ACACTCTTTCCCTACACGACGCTCTTCCGATCTTAGACTCGTCAACGATGAAG ITS2 First AACGCAG

ITS3tagmix3 Fwd ACACTCTTTCCCTACACGACGCTCTTCCGATCTTAGACTCGTCACCGATGAAG ITS2 First AACGCAG

ITS3tagmix4 Fwd ACACTCTTTCCCTACACGACGCTCTTCCGATCTTAGACTCGTCATCGATGAAG ITS2 First AACGTAG

ITS3tagmix5 Fwd ACACTCTTTCCCTACACGACGCTCTTCCGATCTTAGACTCGTCATCGATGAAG ITS2 First AACGTGG

ITS4ngs Rev GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTTTCCTSCGCTTATTGATAT ITS2 First GC Generic Fwd AATGATACGGCGACCACCGAGATCTACAC[index1]ACACTCTTTCCCTCACGA ITS1&2 Second C forward Generic Rev CAAGCAGAAGACGGCATACGAGAT[index2]GTGACTGGAGTTCAGACGTGT. ITS1&2 Second

reverse

Please note that the primers used in this work contain Illumina specific sequences protected by intellectual property (Oligonucleotide sequences © 2007-2013 Illumina, Inc. All rights reserved. Derivative works created by Illumina customers are authorized for use with Illumina instruments and products only. All other uses are strictly prohibited).

Bioinformatics

After demultiplexing the raw FASTQ files, the reads generated from the paired-end sequencing were paired and quality-filtered using MOTHUR 1.35.1 (Schloss et al., 2009).

The quality-filtering consisted of discarding reads with ambiguous sequences, sequence length ranges from 100 bp to 600 bp, and maximum homopolymer lengths of 8. Similar sequences were combined to reduce the computational burden, and the number of copies of the same sequence was displayed. Next, sequences that occurred only one time (singletons) were discarded. This dereplication step was performed using USEARCH 7.0.1090 (Edgar, 2010). The ITS1 and ITS2 sequences of fungal origin were then extracted from the dataset with ITSx. ITSx uses HMMER3 (Mistry et al., 2013) to compare input sequences against a set of models built from a number of different ITS region sequences found in various organisms (Bengtsson-Palme et al., 2013). Only the sequences belonging to the kingdom Fungi were kept for further analyses. Then, the sequences were truncated (200 bp for ITS1 and 300 bp for ITS2) in order to equalize their lengths to enable further clustering analyses. Operational taxonomic units (OTUs) with a 97% similarity cut-off were clustered using UPARSE 7.1 (Edgar, 2013). The similarity threshold (97%) is commonly used in OTU-based analyses and has been shown to be an optimal threshold when using ITS to identify fungi (Koljalg et al., 2013). UNITE-UCHIME was used to identify and remove chimeric sequences (Nilsson et al., 2015). QIIME 1.9.1 (Caporaso et al., 2010) was used to assign taxonomy to OTUs based on the UNITE 7.2 fungal ITS reference training data set. QIIME 1.9.1 was also used to generate an OTU table. Fungal diversity was analyzed by using several different QIIME scripts. The scripts used for alpha/beta diversity, multivariate analyses, differential abundance and taxonomy analyses are listed on the following website: http://qiime.org/scripts/.

For the shotgun metagenome sequences, samples were demultiplexed, quality- controlled and assembled for taxonomic profiling. This was done using the standard MEGAN6 pipeline (Huson et al., 2007) and the default MetaPhlan 2.0 analyses pipeline, which was run on the PyCharm CE platform for python 2.7 (Truong et al., 2015). After comparing the outcomes from both programs, the results obtained by MetaPhlan 2.0 were used in this work because of the flexibility of the command line tool. In addition, all the organisms’ ITS regions were extracted from the quality-controlled metagenome sequences using SORTMERNA v2.1 (Kopylova et al. 2012). The extracted ITS sequences were classified against the UNITE 7.2 reference database for taxonomic identification.

The following are links to the bioinformatics protocols that were applied for shotgun metagenomic analyses: MetaPhlAn2 pipeline: https://bitbucket.org/biobakery/metaphlan2 MEGAN6 tutorial: http://ab.inf.uni- tuebingen.de/data/software/megan6/download/MeganTutorialApril2018.pdf Source code for SORTMERNA: https://github.com/biocore/sortmerna/releases/tag/2.1

Statistical Analyses

For alpha diversity measures, normality was verified by the D′Agostino and Pearson omnibus normality test. As normality was not demonstrated, the non-parametric Mann- Whitney U test was used to assess the significance of the differences between ITS1 and ITS2 in air samples from compost, biomethanization and dairy farms. A p-value ≤ 0.05 was considered statistically significant. The results were analyzed using the software GraphPad Prism 5.03 (GraphPad Software, Inc.). To determine the statistical significance of the variation observed with the multivariate analyses (PCoA figures), a PERMANOVA test was performed on the Bray-Curtis dissimilarity matrix. The compare categories QIIME script was used to generate the statistical results. Because PERMANOVA is a non-parametric test, significance is determined through permutations. The number of permutations used was 999. A p-value ≤ 0.05 was considered statistically significant. Detailed information about the performance of the test is presented in the multivariate section of the results. The nonparametric Mann-Whitney U test was used to ascertain whether or not the differences in OTU abundances are statistically significant between ITS1 and ITS2. To test OTU differential abundance, the null hypothesis was that the populations that the two groups of samples were collected from have equal means The range of p-values obtained for the 50 most differentially abundant OTUs between ITS1 and ITS2 are presented in the differential abundance section of the results.

Data Availability

The raw sequence reads of all of the samples used and that support this study’s findings are accessible in the National Center for Biotechnology Information (NCBI) under the BioProject ID: coming soon.

2.6 Results

Summary of Sequencing Data Processing

Table 2.2 presents a summary of the sequencing information from the number of raw reads to the number of OTU clusters recorded at different steps of the bioinformatics data processing. Samples from composting sites (27), biomethanization facilities (16) and dairy farms (5) were compared based on the barcode used (ITS1 & ITS2). The number of raw reads from MiSeq sequencer was comparable when either one of the barcodes were used (same order of magnitude). The highest numbers of reads corresponded with the highest numbers of samples in each environment: » 3 million for compost, » 600,000 for biomethanization and » 30,000 for dairy farms. The sequence lengths were different for ITS1 and ITS2. ITS2 sequences were systematically longer than ITS1 sequences in the three environments. The mean length of ITS1 sequences ranged from 278 bp to 294 bp and they ranged from 364 bp to 398 bp for ITS2 sequences. Unexpectedly, the mean length of ITS2 sequences was exactly the same (364 bp) across compost and biomethanization samples. After quality filtering, all singletons were excluded from the data set (Brown et al., 2015). In general, the percentage of singletons was the same for ITS1 and ITS2, except in compost samples. Singletons represented only 5.6% of the sequences when ITS1 was used compared to 14% for ITS2. For biomethanization plants and dairy farms, the percentage of singletons was comparable between ITS1 and ITS2 and represented 16% to 19% of the sequences. At the end of data processing, clusters of OTUs were formed. The number of OTUs was two to five times higher when ITS1 was targeted compared to ITS2.

Table 2.2 Summary of the HTS data during the bioinformatics treatment process. ITS1 and ITS2 amplicons are compared in the bioaerosols from the three environments studies

Number of raw seq. Mean length of Number of seq. after Number of from MiSeq seq. after paired- quality filtering (% of OTUs platform end assembly singletons) ITS1 ITS2 ITS1 ITS2 ITS1 ITS2 ITS1 ITS2 Compost 3871313 3680926 294 364 44438 53001 1208 772 (5.6%) (14%) Biomethanization 675642 730688 281 398 18080 25658 1149 330 (18%) (19%) Dairy farms 354262 310362 278 364 10502 11427 1015 218 (16%) (17%)

Rarefaction

A rarefaction analysis using the observed OTU alpha diversity metric was conducted to validate the sequencing depth and confirm the effective sampling of the biological content of the aerosol samples that were collected in the three environments studied,. The lowest depth was used to as the sequencing depth of the rarefaction analyses. This procedure allows the rarefaction of all the samples to the same number of sequences. In other words, samples with a lower sequencing depth than the one chosen were excluded from the analyses. The higher the sequencing depth, the more likely diversity coverage is attained. In this case, the sequencing depth was 40,000 sequences per sample for compost and dairy farms, and 9000 sequences for biomethanization. All the samples were included in the analyses, except the outdoor controls due to low sequence numbers. The points shown in Fig. 2.1a (compost) and Fig. 2.1b (biomethanization) were calculated using ten randomly selected values from 10 to 40,000 sequences. Points shown in Fig. 2.1c (dairy farms) were calculated similarly but with values selected from 10 to 9000 sequences. The corresponding number of OTUs observed for each of these values was noted for all of the samples. Then, the average number of OTUs

observed with the standard deviation was calculated for each one of the ten values. To analyze the rarefaction data, samples were grouped according to the barcode used. The plateaus of the curves in Fig. 2.1 indicate efficient coverage of the fungal diversity with ITS1 and ITS2, as no more OTUs were observed, even with greater numbers of sequences per sample

Alpha Diversity

The species diversity measurement was introduced by Whittaker and defined as the number of species and their proportions within one sample (Whittaker, 1972). Different alpha diversity measures have been proposed and the choice of measure depends on the context of

the study. To help make informed choices, a list of indexes and explanations on how these measures are used is presented in Magurran and McGill’s book (Magurran and McGill, 2011).

For this work, three indexes were used to measure alpha diversity: Chao1, Shannon and Simpson. More specifically, Chao1 is a richness estimator. The higher the number of different OTUs in a sample, the higher the value of the Chao1 index; there is no value limit. For Shannon and Simpson, the species richness is combined with the abundance to give one diversity measure. The Simpson index represents the probability that two randomly selected OTUs in a sample are from the same species. The values are bounded between 0 and 1, where 1 represents the most diverse case. Shannon values are bounded between 0 and 10, where 10 represents the highest diversity. The diversity measures were obtained using the alpha diversity QIIME script.

Table 2.3 presents a summary of the alpha diversity measures with the systematic comparison of ITS1 and ITS2 barcodes in the three environments studied. The ITS1 barcode produced significantly more richness per sample compared to the ITS2 barcode for composting (domestic P = 0.0009; animal P = 0.0006) and biomethanization (BF1 P = 0.001; BF2 P = 0.0007). The estimated richness was also higher when targeting ITS1 compared to ITS2 in the five dairy farms. However, no statistics could be calculated because there was only a single value for each dairy farm. Similarly, the Shannon diversity index was significantly higher for the ITS1 barcode for compost (domestic P = 0.0008; animal P = 0.001) and for biomethanization (BF1 P = 0.001). For BF2, no significant difference was noted between ITS1 and ITS2. The Shannon diversity indexes obtained for the five dairy farms were mixed: three out of five dairy farms had higher values for the ITS2 barcode compared to ITS1. The ITS1 barcode in domestic compost resulted in a significantly higher value compared to ITS2 for the Simpson diversity index (P = 0.00001). Furthermore, two out of five dairy farms had higher Simpson values for the ITS1 barcode compared to ITS2. The rest of the Simpson values were similar for both the ITS1 and ITS2 barcodes. Overall, the ITS1 barcode performed better than ITS2 in estimating richness and measuring diversity when the three environments were considered.

Table 2.3: Alpha diversity analysis comparing data obtained from targeting ITS1 and ITS2 barcodes in aerosol samples from three environments. The numbers represent the mean values with the standard deviation for each group of samples. When the standard deviation is not shown, its value is zero. The highest values obtained from the comparisons between ITS1 and ITS2 are highlighted in bold type. The asterisk (*) indicates the statistical significance of the Mann-Whitney U test (ns P > 0.05; * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001; **** P≤ 0.0001)

Chao1 Shannon Simpson ITS1 ITS2 ITS1 ITS2 ITS1 ITS2 Compost Domestic 292±83*** 214±101 5±0.5*** 4±1 0.9**** 0.8 Animal 289±64*** 214±64 5±0.1*** 4.5±0.5 0.9±0.1 0.9 Biomethanization BF1 161±13** 119±54 4±0.7*** 3.8±0.4 0.8 0.8 BF2 273±73*** 117±48 4±1 4±0.7 0.9 0.9 Dairy farms DF1 480 201 4.4 4.6 0.9 0.9 DF2 626 168 4 4.3 0.8 0.7 DF3 646 193 3.5 2.7 0.7 0.7 DF4 571 186 5.6 4.2 0.9 0.8 DF5 420 193 3.9 4.7 0.9 0.9

Multivariate Analysis

An ecological analysis was conducted to determine which variable is the strongest predictor of the variation in a fungal community/fungal communities: environmental factors or the choice of barcode. One common technique used to determine the more influential variable relies on the creation of a (dis)similarity matrix to calculate the distances between samples. In this case, the Bray-Curtis dissimilarity measure was used to try and explain community variation. The Bray-Curtis index is bounded between 0 and 1, where 0 means the two samples have the same composition and 1 means that they do not share any species. The QIIME script for beta diversity analyses was used to produce the Bray-Curtis matrix, which includes information about OTU abundance. Because the Bray-Curtis dissimilarity uses the absolute abundances of the OTUs, it is necessary to use a rarefied OTU table as the input for

the dissimilarity calculation. The same rarefaction depth from the rarefaction curves described previously was used for the multivariate analysis (40,000 sequences for compost and dairy farm samples and 9000 sequences for biomethanization samples). Inter-sample distances were represented in a dimensional space using ordination. One of the most commonly used methods to evaluate ordination patterns is the principal coordinate analyses (PCoA). The dissimilarity matrix is used as an input for ordination calculation. The matrices were transformed to coordinates and then plotted using the principal coordinates script from QIIME. For each environment, samples were separated according to the barcode used (ITS1 and ITS2) and the environmental factors that could explain community variation (composting sites: domestic and animal; biomethanization facilities: BF1 and BF2; dairy farms: DF1 to DF5).

Figure 2.2 shows the three principal coordinate axes capturing more than 70% of the variation for compost (Fig. 2.2a and Fig. 2.2b), more than 63% for biomethanization (Fig. 2.2c and Fig. 2.2d), and more than 76% for dairy farms (Fig. 2e and Fig. 2f). Samples were colored according to two variables (choice of barcode and environmental factor) to better visualize sample clustering. Samples plotted closer to one another are more similar than those ordinated further away. In each of the three environments, the choice of barcode consistently led to the best sample clustering (compost P = 0.001; biomethanization P = 0.001; dairy farms P = 0.007; Fig. 2b, Fig. 2d, Fig. 2f, respectively). When the environmental factor variable was used, samples were randomly dispersed with no particular color grouping (compost P = 0.08; biomethanization P = 0.22; dairy farms P = 0.98; Fig. 2.2a, Fig. 2.2c, Fig. 2.2e, respectively). Across all three environments, the strongest predictor of fungal composition in samples was the choice of barcode used. This was a stronger predictor than the potential fungal sources present during air sampling.

Figure 2.2: Principal Coordinates Analysis of air samples collected from composting sites (a and b), biomethanization facilities (c and d), and dairy farms (e and f). The PCoA plots were generated using the Bray-Curtis dissimilarity measure to calculate the distances between samples. Fig. 2.2a, Fig. 2.2c and Fig. 2.2e show samples colored according to the type of bioaerosol source (a: domestic in blue and animal in red; c: BF1 in blue and BF2 in red; e: DF1 in blue, DF2 in orange, DF3 in red, DF4 in green, DF5 in purple). Fig. 2.2b, Fig. 2.2d and Fig. 2.2f show samples colored according to the barcode used (ITS1 in blue and ITS2 in red).

Differential abundances of species

After measuring the fungal community variation across samples, the next step was to try to identify species that had significantly different abundances depending on the barcode used. To accomplish this goal, a statistical test designed specifically for the differential analyses of count data was used. Using this Mann-Whitney U test, OTU frequencies can be compared in groups of samples and whether or not the two groups of samples have statistically different OTU abundances can be ascertained. This test is non-parametric and uses absolute data counts rather than relative abundances. More specifically, the output of the test contains the test statistic, the p-value corrected for multiple comparisons and a mean count for each OTU in the given sample group. The Mann-Whitney U test was used following instructions from the group significance QIIME script. The 50 OTUs with the most significant differences in abundance when ITS1 (25 OTUs) or ITS2 (25 OTUs) was targeted in samples from compost, biomethanization facilities and dairy farms, are shown in figures 2.3 through 2.5, respectively. All the OTUs presented in Fig. 2.3 to Fig. 2.5 have large differences in mean counts between ITS1 and ITS2. However, these are not exhaustive lists of the OTUs with differential abundances. The complete outputs of the differential abundance analyses are presented in supplementary datasets 1 to 3 for compost, biomethanization and dairy farms, respectively. For compost, the first 25 OTUs that were more abundant in the ITS1 group compared to ITS2 had p-values ranging from 0.0001 to 0.0000004. The ones that were more abundant in the ITS2 group had p-values ranging from 0.0008 to 0.000004. For biomethanization facilities, p-values of the first 25 OTUs with differences in abundance ranged from 0.004 to 0.0003 for ITS1 and from 0.0002 to 0.00003 for ITS2. For dairy farms, these numbers ranged from 0.04 to 0.008 for ITS1 and 0.05 to 0.003 for ITS2. The most striking example in the list compiled from the compost samples is Penicillium cinnamopurpureum, with a mean count of 5000 sequences across the ITS1 group and fewer than 10 sequences in the ITS2 group. Similarly, Cladosporium arthropodii was present with a mean count of 5000 sequences in ITS2 and less than 5 sequences in ITS1. Surprisingly, Cladosporium arthropodii was also the most differentially abundant species in ITS2 samples from biomethanization facilities and dairy farms with mean counts of 5800 sequences and more than 7000 sequences, respectively. For ITS1, the most differentially abundant species

in biomethanization samples was Penicillium polonicum with a mean count of 5900 sequences and Aspergillus intermedius in dairy farms with a mean count of more than 8000 sequences. These species were not detected by ITS2. In the three environments studied, the differential abundances of the 50 species were obvious as there were considerable margins in the sequence counts between ITS1 and ITS2. Penicillium polonicum, Mycosphaerella tassina, Penicillium vanderhammenii, and Apergillus intermedius were consistently more abundant in the ITS1 group and either under or not represented in the ITS2 group in all three environments. The same observation was made for Aspergillus terreus, Penicillium cinnamopurpureum, Cladosporium delicatum, Aspergillus piperis, Aureobasidium microsticum, Malassezia restricta, Ganoderma sichuanense, Irpex hypnoides, Microascus brevicaulis, Mrakia frigida, Trametes versicolor and Bullera unica in two of the three environments studied. Similarly, Hydnellum suaveolens, Cryptococcus penaeus and Penicillium herquei were consistently more represented by ITS2 compared to ITS1 (under or not represented) in all three environments. This was also the case for Paraphoma dioscoreae, Candida sake, Alternaria eureka, Pichia fermentans, Cytospora abyssinica, Neopestalotiopsis foedans and Yarro wialipolytica in two of the three environments.

Figure 2.3: Fungal species with statistically significant differential abundances across compost samples targeting either ITS1 or ITS2 barcodes. From the bottom to the top: the first 25 species were the most abundant with ITS1 and the last 25 were more abundant with ITS2.

Figure 2.4: Fungal species with statistically significant differential abundances across biomethanization samples targeting either ITS1 or ITS2 barcodes. From the bottom to the top: the first 25 species were the most abundant with ITS1 and the last 25 were more abundant with ITS2.

Figure 2.5: Fungal species with statistically significant differential abundances across dairy farm samples targeting either ITS1 or ITS2 barcodes. From the bottom to the top: the first 25 species were the most abundant with ITS1 and the last 25 were more abundant with ITS2.

Taxonomic Analyses

In order to compare classes of fungi across samples, the taxonomic profiles were analyzed more carefully according to which barcode was used and the environmental factors of each one of the three environments that were studied. From the 20 classes of fungi that were represented in compost, Dothideomycetes, Eurotiomycetes, Saccharomycetes, Sordariomycetes and Agaricomycetes are the most abundant, accounting for 90% of the total relative abundance (Fig. S1). Among these classes, the only strong class-level difference between ITS1 and ITS2 was observed for Saccharomycetes. It was 2.5 times more abundant for ITS2 (20%) compared to ITS1 (8%). However, the less abundant classes, Wallemiomycetes, Exobasidiomycetes and Taphrinomycetes were detected only by ITS1 and Glomeromycetes, Tritirachiomycetes, Mucoromycotina, Rozellomycota and Lecanoromycetes were detected only by ITS2. In biomethanization facilities/samples, four out of 14 classes represented 90% of the total relative abundance (Eurotiomycetes, Dothideomycetes, Sordariomycetes and Agaricomycetes). Saccharomycetes were five times more abundant for ITS2 compared to ITS1. Wallemiomycetes, Exobasidiomycetes, Ustilaginomycotina and Cystobasidiomycetes were specific to ITS1 and Mucoromycotina was specific to ITS2 (Fig. S2). Similarly, four of the 14 classes that were present in dairy farm samples accounted for more than 90% of the relative abundance (Eurotiomycetes, Dothideomycetes, Sordariomycetes and Agaricomycetes). Wallemiomycetes, Exobasidiomycetes, Ustilaginomycotina and Microbotryomycetes were only present in ITS1. Lecanoromycetes and Ciliophora were specific to ITS2 (Fig. S3). Similar to compost and biomethanization facilities, the only major difference between ITS1 and ITS2 in dairy farms was the abundance of Saccharomycetes (four times more abundant in ITS2). The conclusions were the same when samples were compared according to environmental factors rather than which barcode was used (animal vs domestic for compost; BF1 vs BF2 for biomethanization; DF1 to DF5 for dairy farms; Fig. S1 to Fig. S3, respectively). Other notable differences were that Wallemiomycetes and Exobasidiomycetes were consistently only detected by ITS1 in the three environments. Lecanoromycetes was consistently present only in ITS2 across samples from the three environments. Furthermore, unidentified sequences of plants were found to be exclusive to ITS2.

Culture vs HTS

The diversity of fungi identified using the culture method was compared with the fungal diversity identified using HTS at the genus level. Table 2.4 shows a summary of this comparison in biomethanization samples. Culture methods were able to detect six fungal genera. For HTS, the first 20 most abundant genera were considered. They represented 81% and 84% of the total relative abundance of ITS1 and ITS2, respectively. The most abundant genera were the same using both methods (Penicillium, Aspergillus, Cladosporium and Talaromyces). Although Phialocephala is shown to be present only in the profile obtained from the culture method, it was also detected by ITS1 and ITS2 (but not present in the top 20). The diversity profile obtained by the culture method in dairy farms was more exhaustive compared to biomethanization facilities (Table 2.5). The fungal genera that were identified at more than one dairy farm were grouped together. The fungi that were detected only once by culture were Trichoderma, Microdochium, Phoma, Apiospora, Botrytis, Coniothyrium, Millerozyma, Neosetophoma, Irpex, and Debaryomyces. The relative abundance of the fungi identified by culture was calculated as follows: For each genus, the number of times that it was isolated from the five dairy farms was calculated. Based on this sum, a percentage of overall relative abundance was calculated for each one, as appears on the list in table 2.5. For HTS, the top 20 most abundant genera accounted for 67% and 90% of the total relative abundance for ITS1 and ITS2, respectively. As in biomethanization samples, only four genera were detected by both approaches in dairy farms: Penicillium, Cladosporium, Bipolaris and Fusarium. However, some genera were shared only between ITS1 and culture (Sarocladium and Aspergillus) and between ITS2 and culture (Wickerhamomyces and Alternaria). The HTS profile of fungal genera is much more exhaustive than what is shown on the lists in table 2.4 and table 2.5. Of the fungal genera that were isolated using culture techniques, three (Hyphopichia, Gibellulopsis and Myceliophthora) were not detected by HTS.

Table 2.4: Comparison of the 20 most abundant fungal genera identified by HTS targeting ITS1 and ITS2 barcodes, and the fungal genera identified by culture in aerosol samples collected in two biomethanization facilities. The taxa in bold and underlined are shared between the three columns (ITS1, ITS2 and culture). Taxa in bold are shared between one of the two HTS columns (ITS1 or ITS2) and culture.

HTS ITS1 (% relative abundance) ITS2 (% relative abundance) Culture (% relative abundance) Penicillium (41.4) Penicillium (22) Penicillium (65.5) Aspergillus (13.8) Cladosporium (19.8) Aspergillus (18.9) Mycosphaerella (9.9) Aspergillus (18.1) Cladosporium (11.1) Cladosporium (3) Talaromyces (4) Phialocephala (1.1) Aureobasidium (1.7) Hydnellum (4) Talaromyces (2.2) Cryptococcus (1.7) Paraphoma (3.8) Fusarium (1.1) Malassezia (1.5) Candida (1.6) Botrytis (1.4) Peniophora (1.6) Ganoderma (1) Cryptococcus (1.2) Talaromyces (0.9) Fusarium (1.1) Capnobotryella (0.7) Alternaria (1.1) Neuropsora (0.7) Neokalmusia (0.9) Guehomyces (0.7) Geotrichum (0.9) Microascus (0.5) Sporobolomyces (0.7) Irpex (0.5) Mucor (0.7) Naganishia (0.5) Neurospora (0.5) Sagenomella (0.4) Bullera (0.5) Sporobolomyces (0.4) Kazachstania (0.5) Mrakia (0.4) Microascus (0.4) Candida (0.3) Udeniomyces (0.3)

*The relative abundance of the HTS data represents the relative abundance of the total taxa detected (not only the top 20). For ITS1, the relative abundance of the top 20 fungal genera represents 81.4% of the total genera identified. For ITS2, the top 20 fungal genera represent 83.7% of the total genera identified.

Table 2.5: Comparison of the 20 most abundant fungal genera identified by HTS targeting ITS1 and ITS2 barcodes, and the fungal genera identified by culture in aerosol samples collected at five dairy farms. The taxa in bold and underlined are shared between the three columns (ITS1, ITS2 and culture). Taxa in bold are shared between one of the two HTS columns (ITS1 or ITS2) and culture.

HTS* Culture (% relative abundance) ITS1 (% relative abundance) ITS2 (% relative abundance) Aspergillus (19.5) Penicillium (39) Penicillium (39.5) Penicillium (13.5) Cladosporium (13.7) Aspergillus (22.4) Fusarium (5.6) Paraphoma (9) Cladosporium (10.7) Mycosphaerella (4.7) Bipolaris (6.7) Rhodosporidium (4.6) Bipolaris (4.6) Fusarium (6.3) Sarocladium (3.9) Aureobasidium (3.4) Hyphoderma (2.4) Hormographiella (2.5) Cladosporium (2.2) Alternaria (2.2) Phaeosphaeria (2.5) Parastagonospora (2.2) Peniophora (1.7) Wickerhamomyces (1.8) Capnobotryella (2.1) Hydnellum (1.6) Alternaria (1.8) Sarocladium (1.6) Neopestalotiopsis (1.3) Epicoccum (1.4) Wallemia (1.3) Cryptococcus (1) Meyerozyma (1.4) Ganoderma (1) Phaeoacremonium (1) Bipolaris (1.4) Tubulicrinis (0.8) Candida (0.8) Lichtheimia (1.4) Neoascochyta (0.7) Pichia (0.8) Myceliophthora (0.7) Bjerkandera (0.7) Bjerkandera (0.6) Gibellulopsis (0.7) Irpex (0.6) Wickerhamomyces (0.5) Fusarium (0.7) Trametes (0.6) Chrysosporium (0.5) Hyphopichia (0.7) Cryptococcus (0.6) Phellinus (0.4) Trichosporon (0.7) Candida (0.4) Phyllosticta (0.4) Rhizomucor (0.7) Monographella (0.4) Verrucladosporium (0.4) Thermomyces (0.36)

*The relative abundance of the next-generation sequencing data represents the relative abundance of the total taxa detected (not only the top 20). For ITS1, the relative abundance of the top 20 fungal genera represents 66.6% of the total genera identified. For ITS2, the top 20 fungal genera represent 90.3% of the total genera identified.

Shotgun versus Amplicon Sequencing

Five air samples collected from five dairy farms yielded 101,652,459 high-quality metagenome sequences. Only 2,338,007 of these sequences were of fungal origin,

representing 2.3% of the total sequences. Interestingly, the four most abundant classes of fungi detected using shotgun metagenomics (Dothideomycetes 39%; Eurotiomycetes 18%; Agaricomycetes 15%; Sordariomycetes 11%) correspond to the most abundant classes detected in ITS1 and ITS2 sequences. However, some fungal classes could only be recovered from metagenomes (Mixiomycetes 4%; Geoglossomycetes 2%; Orbiliomycetes 1%) and some were retrieved only from amplicon-based sequencing (Ustilaginomycotina, ITS1; Microbotryomycetes, ITS1; Ciliophora, ITS2; Leotiomycetes, ITS1 and ITS2). One of the most striking differences between shotgun and amplicon-based sequencing lies is the relative abundance of unidentified fungi, which was 10% for metagenomes and less than 2% for amplicon-based sequences. Metagenome sequences were blasted against the ITS sequences in the UNITE database in order to extract the sequences corresponding to the whole ITS region from the shotgun sequencing. Then, the taxonomic identifications of these sequences were compared to those obtained by the amplicon-based (ITS1 and ITS2) HTS approach (Fig.S4). Some differences were observed in the classes identified from ITS shotgun sequences and the ITS1 and ITS2 sequences. For instance, Agaricomycetes and Pezizomycetes were significantly less abundant in the amplicon-based sequences compared to the shotgun metagenomes. In contrast, Dothideomycetes, Sordariomycetes, and Eurotiomycetes were more common in ITS1 and ITS2 sequences compared to ITS shotgun sequences. To more thoroughly examine the relationship between ITS1, ITS2 and ITS retrieved from shotgun sequencing when identifying fungal genera, we generated a taxonomic profile based on genus comparing the three components (Fig. 2.6). For more effective visualization, we considered only the genera that represent more than 1% of the total relative abundance for each component represented in Fig. 2.6. For ITS2, the genera that make up more than 1% of the total relative abundance represented 86% of the total relative abundance when combined. However, for ITS1 and ITS-shotgun, the genera that make up more than 1% of the total relative abundance represented only 62% and 58% of the abundance, respectively. This means that a more diverse genus profile is expected in ITS1 and ITS-shotgun sequences that comprise less than 1% of the total abundance. The genera identified by ITS shotgun sequences had more similarities with the ITS1 profile compared to ITS2. A striking example is the absence of Aspergillus from the ITS2 profile while it was present in more than 25% of ITS1 and ITS-shotgun sequences. Also, Cladosporium was

significantly more abundant in ITS2 sequences compared to those of ITS1 and ITS-shotgun. As expected, some taxa were only detected using ITS-shotgun approach, like Trichaptum and Tubilicrinis. These genera are a part of the class Agaricomycetes, and were statistically more abundant in the metagenomes compared to the amplicons.

Figure 2.6: Relative abundances of fungal genera detected in dairy farms by shotgun and amplicon-based (ITS1 and ITS2) HTS. The whole ITS region (ITS1-5.8S-ITS2) was extracted from the shotgun metagenomes (ITS-shotgun) and relative abundance was recalculated based on the representative number of sequences.

2.7 Discussion

The internal transcribed spacer (ITS) region has been proposed as a standard genetic marker for fungi (Schoch et al., 2012). The amplicon-based HTS approach relies on the use of one of the two sub regions (ITS1 and ITS2). A comprehensive investigation validating their value as DNA barcodes in different environments is necessary for building effective strategies for characterizing fungal diversity. In this large-scale study, the systematic comparison of ITS1 and ITS2 barcodes in three different environments led to consistent results in regards to fungal diversity in bioaerosols.

In terms of sequence length, ITS2 had longer sequences than ITS1 throughout all samples from the three environments. Because longer amplicons represent a challenge for PCR amplification and DNA sequencing, this observation implies that ITS2 is more difficult to amplify than ITS1. This difference in length may be due to the fact that the 5.8S rRNA gene is included in the primers amplifying ITS2 in this study. It has been previously demonstrated that long ITS barcodes may affect amplification and sequencing (Lindhal et al., 2013; Tedersoo et al., 2015b). Furthermore, sequence length may have a strong inverse correlation with abundance recovery (Ihmark et al., 2012). Relatively longer reads are subject to low quality at the end of the sequences, which may cause problems during paired-end ligation. The percentages of singletons were comparable between ITS1 and ITS2 across samples from all three environments. Neither barcode outperformed the other in terms of generating singletons. The importance of eliminating singletons in HTS analyses is best described with Tolstoy’s rule which states that most unique sequences are bad: « If most bases are good, most unique sequences are bad, because good reads are all alike, but every bad read is bad in its own way ».

Alpha diversity metrics indicate that ITS1 was consistently able to recover more OTUs and better estimate richness compared to ITS2 in all three environments. Some exceptions were noted for species diversity measurements, as ITS2 outperformed ITS1 in a few samples. However, only one sample was used for comparison in each dairy farm, which may explain the overall contradictory observations. The diversity measurements are known

to perform better when a higher number of samples is used for making comparisons (Veech and Crist, 2010; Jost, 2006, 2007, 2010; Moreno and Rodriguez, 2010).

The multivariate analyses coupled with the PERMANOVA test provided robust analyses demonstrating the statistical significance of sample clusters using distance matrices. The fact that the two analyses yielded the same conclusions about the clusters formed by the samples confirms their combined usefulness as tools to visualize and measure sample clustering. The overall analyses allow the study of variables that may explain community composition. In this work, the choice of barcode was the principal factor responsible for detecting trends in the fungal composition of samples from compost, biomethanization facilities and dairy farms. These observations argue against the notion that environmental factors (e.g. source of bioaerosols) are the main variables that influence fungal composition, as previously described (Mbareche et al., 2017; Mbareche et al., 2018, Mbareche et al., 2019a, Mbareche et al., 2019b). In light of these findings, researchers should be aware that the fungal diversity detected by amplicon-based HTS is highly dependent on the barcode used, and this should be considered in future discussions.

The differential abundance analyses of OTUs across samples from multiple environments and grouped into ITS1 and ITS2 were essential in determining which species were affected by the choice of barcode. Certainly, differences between ITS1 and ITS2 are species-dependent. The fungal species that were consistently only found in the ITS1 list or ITS2 list in all three environments or at least two of the three environments should be examined more closely when planning a strategy to study the fungal diversity in aerosol samples. Moreover, rare species that were unique to each environment and only detected by ITS1 or ITS2 should also be taken into consideration. The differential abundance results presented herein are not exhaustive. The complete information is presented in the supporting material. Primer mismatches could explain the potential biases linked to the taxa positively influenced by the choice of either ITS1 or ITS2 (Bellemain et al., 2010; Han et al., 2013). The advantage of ITS2 for the class Saccharomycetes in the three environments, as showcased in the taxonomy analysis, could be linked to a 3’ terminal mismatch to ITS1 primers (Tedersoo et al., 2015). For example, the ITS1F forward primer specific to fungi is

known to have mismatches to some classes of fungi including Chytridiomycota, Saccharomycetes and some genera of Dothideomycetes. The occurrence of introns between primer sites in many taxonomic groups of Ascomycota could explain the difference in how ITS1 and ITS2 perform (Perotto et al., 2000; Bhattacharya et al., 2001; Vralstad et al., 2002). The differential abundances of species between ITS1 and ITS2 could also be explained by the targeted rRNA gene copy number. This can lead to a consistent over or under representation of particular taxa (Taylor et al., 2016). Likewise, the presence/absence of fungal classes depending on the type of barcode used should be noted for future environmental studies targeting a specific class of fungi.

Another striking difference in the results from the two barcodes was that when the ITS2 barcode was used in the three environments, consistent unidentified sequences belonging to plants were detected. While this observation suggests that there is a need to design primers that are more specific to fungi, it raises the question of erroneous taxonomic classification. The quality and abundance of data in references may affect the identification of particular taxa, thus causing identification bias. The quality of sequences from public databases is unknown, and it is sometimes hard to distinguish which sequences cover only parts of ITS and which cover the entire region. It has been previously demonstrated that up to 20% of fungal sequences in databases may be erroneously designated (Nilsson et al., 2006; Bidartondo et al., 2008; Koljalg et al., 2013). Two good examples of reference-based biases are related to the identification of Cryptomycota and Microsporidia. These groups of fungi are not well described in ITS databases due to the fact that their descriptions relied on taxonomic studies focused on the small subunit 18S (Lazarus and James, 2014; Bass et al., 2018). Moreover, sequences designated as unidentified fungi (with ITS1 and ITS2) probably belong to early divergent fungal lineages that are underrepresented in ITS databases, which are valued for their intraspecific variability quality (Nilsson et al., 2008; Schoch et al., 2012). Finally, the recording of Talaromyces marneffei, a notable human pathogen that only occurs in Southeast Asia in conjunction with bamboo rat habitats, shows that some of these sequence-based identifications will certainly be erroneous due to closely related species wrongly identified in databases. the In this work, using ITS1 led to an overall identification success rate that was higher than that of ITS2 in the three environments studied.

Comparing the taxonomic profiles obtained by HTS and culture methods confirmed the expected biases in the determination of fungal diversity by culture. Extremely low numbers of species were identified by culture compared to HTS. The differences observed in the diversity profiles for relative abundance that were obtained by the two methods may be explained by the hypothesis that the culture approach may be biased toward fungi from the rare biosphere. These results are consistent with the conclusions made by Shade and his collaborators (2012) regarding the complementarity of culture-dependent and culture- independent approaches for studying bacterial diversity. The premise of their study is that culture-dependent methods reveal bacteria from the rare biosphere and provide supplemental information to data obtained using an NGS approach. In the current case, this complementarity is true only for abundance. As mentioned previously, only three fungi were detected exclusively by using culture methods. In contrast, hundreds of fungi were identified by HTS alone.

Other features may impact the performance of ITS1 and ITS2 as fungal barcodes in amplicon-based HTS. For example, GC content is known to have an effect on PCR and sequencing efficiencies (McDowell et al., 1998). While the GC content was not addressed in this work, a recent study used sequences retrieved from an ITS database to examine the GC content of ITS1 and ITS2 sequences. In the taxa studied, ITS1 had a significantly lower GC content than ITS2 which may give ITS1 an advantage in amplification and sequencing compared to ITS2 (Wang et al., 2014). Also, bioinformatics analyses can have an impact on diversity analyses such as the clustering algorithms, the percent identity threshold and taxonomy assignment tools (BLASTn vs. Naïve Bayesian Classifier). However, the performance of ITS1 and ITS2 was evaluated using the same bioinformatics tools in order to reduce any additional biases in the diversity analyses.

The difference between the shotgun and amplicon-based HTS approaches in exclusively detecting some classes of fungi (mentioned in the Results section) was expected. In fact, PCR biases related to amplicon sequencing are well known, and were previously

addressed. However, shotgun metagenomes also present biases related to the low amounts numbers of overlapping shotgun sequences across the ribosomal DNA, which can make OTU assignment and taxonomic identification inaccurate (Tedersoo et al., 2015; Bengtsson et al., 2012; Gomez-Alvarez et al., 2009). In other words, metagenome sequences may fall into any genomic region and not into any part of the SSU, ITS, LSU, or into the intergenic spacers (IGS) that are typically considered in studies of this kind. This may also explain the high percentage of unidentified fungi in shotgun metagenomes when compared to amplicon-based HTS in this study. Because the SSU, ITS, and LSU all include variable as well as highly conserved regions, it is difficult to accurately assign taxonomy to short fragments and to low taxonomic levels. Thus, there is a higher likelihood of erroneous taxonomic assignments when using the shotgun metagenomic HTS approach compared to the amplicon-based HTS approach. To improve taxonomic identification in metagenomes, databases should be revised to include full-length rDNA sequences originating from genomic studies and that cover all classes of fungi and closely related organisms. One reassuring observation that was made is that the class taxonomic profile remained highly comparable when ITS sequences were extracted from the metagenomes and when all metagenome sequences were considered. It confirms the usefulness of the ITS region in predicting fungal diversity. In this sense, although shotgun metagenomics and amplicon-based HTS approaches resulted in equivalent taxonomic profiles for the most abundant fungal classes, there were substantial similarities between the ITS region retrieved from metagenomes and ITS1-based sequencing when compared to ITS2 with consideration to the fungal genera. This makes ITS1 a more popular choice for/as the fungal barcode when limited resources are available. Based on these results, shotgun metagenomics is a waste of financial and computational resources when its sole intention is to profile fungal taxonomy. In addition, this method presents considerable biases. However, the functional big data produced in metagenomic analyses makes this approach a promising tool for the future of fungal ecology (Sharpton, 2014).

Continuous advances in bioinformatics tools and experimental designs are necessary to further improve the usage of ITS1 and/or ITS2 as universal fungal barcodes. New PCR primers for ITS1 and ITS2 must continue to be developed and tested (Bellemain et al., 2010; Han et al., 2013; Op de Beeck et al., 2014; Tedersoo et al., 2015). Advances in sequencing

technology that improve the quality and length of reads may lead to the use of the full ITS region. This would thus reduce the biases related to choosing ITS1 or ITS2. Also, developing a method that combines the use of multiple reference datasets can help to identify the vast majority of the OTUs at all taxonomic levels. Finally, strategies to correct the gene copy number may help improve the abundance estimates from amplicon-based HTS targeting ITS1 and ITS2. This gene copy number correction had positive outcomes on the prokaryote 16S in HTS studies (Kembel et al., 2012; Angly et al., 2014).

Conclusion

There is no universal solution to cover all fungal taxonomic groups. The goal of this work was to offer a guide for aerosol scientists to use to design studies addressing the fungal population in aerosols using molecular approaches. The large-scale environmental samples used for the systematic comparisons of the performances of ITS1 and ITS2 as barcodes makes this research unequivocal. The results obtained suggest that neither of the barcodes evaluated is perfect in terms of distinguishing all species. Using both barcodes offers a broader view of the fungal aerosol population. However, with the actual knowledge, we strongly recommend using ITS1 as a universal fungal barcode for quick general analyses of diversity and when limited financial resources are available. Finally, the culture comparison with amplicon-based sequencing showed the complementarity of both approaches in describing the most abundant taxa.

Acknowledgements

HM is a recipient of the FRQNT PhD scholarship and received a short internship scholarship from the Quebec Respiratory Health Network. This work was supported by the Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail (2014-0057, 2012- 0029 and 2013-0013) and from NSERC Discovery Grant (CD) RGPIN-2014-05900. We are grateful to all of the workers from the composting sites, biomethanization facilities and dairy farms that participated in this study. We are also grateful to Laetitia Bonifait, Marie-Eve Dubuis, Maude Talbot, Julien Duchaine, and Philippe Bercier, for their technical assistance

in the field. The authors are thankful to Amanda Kate Toperoff and Michi Waygood for English revision of the manuscript. CD is the head of the Bioaerosols and respiratory viruses strategic group of the Quebec Respiratory Health Network.

Funding

HM is a recipient of the FRQNT and IRSST PhD scholarship and received a short internship scholarship from the Quebec Respiratory Health Network. This work was supported by the Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail

(2014-0057, 2012-0029 and 2013-0013) and from NSERC Discovery Grant (CD) RGPIN-

2014-05900.

Authors contribution

H.M. designed the study, performed all experiments including field sampling, analyzed the data and wrote the paper. M.V. performed field sampling and designed the study. G.J.B. and C.D. designed the study and all authors edited the manuscript.

Competing interests

The authors declare no competing financial interests.

Materials & Correspondence

All requests for materials and correspondence should be addressed to Caroline Duchaine.

Supplementary Information

Supplementary information is available at journal website

2.9 Bibliography

Angly, F.E., Dennis, P.G., Skarshewski, A., Vanwonterghem, I., Hugenholtz, P., Tyson, G.W., 2014. CopyRighter: a rapid tool for improving the accuracy of microbial community profiles through lineage-specific gene copy number correction. Microbiome 2:1–13.

Arfi, Y., Buee, M., Marchand, C., Levasseur, A., Record, E., 2012. Multiple markers pyrosequencing reveals highly diverse and host specific fungal communities on the mangrove trees Avicennia marina and Rhizophora stylosa. FEMS Microbiol. Ecol. 79: 433– 444.

Arshad, S.H., Tariq, S.M., Matthews, S., Hakim E., 2001. Sensitization to common allergens and its association with allergic disorders at age 4 years: a whole population birth cohort study. Pediatrics. 08:E33–E37.

Bass, D., Czech, L., Williams, B., Berney, C., Dunthorn, M., Mahé, F., et al., 2018. Clarifying the relationships between Microsporidia and Cryptomycota. J Eukaryot. Microbiol. 65(6):773-782.

Bazzicalupo, A.L., Bàlint, M., Schmitt, I., 2013. Comparison of ITS1 and ITS2 rDNA in 454 sequencing of hyperdiverse fungal communities. Fungal Ecology. 6(1):102-109.

Beezhold, D.H., Green, B.J., Blachere, F.M., Schmechel, D., Weissman, D.N., Velickoff, D., Hogan, M.B., Wilson, N.W., 2008. Prevalence of allergic sensitization to indoor fungi in 70. West Virginia. Allergy Asthma Proc. 29(1):29–34.

Bellemain, E., Carlsen, T., Brochmann, C., Coissac, E., Taberlet, P., Kauserud, H., 2010. ITS as an environmental DNA barcode for fungi: an in silico approach reveals potential PCR biases. BMC Microbiol. 10:189.

Bengtsson-Palme, J., Ryberg, M., Hartmann, M., Branco, S., Wang, Z., Godhe, A.,et al., 2013. Improved software detection and extraction of ITS1 and ITS2 from ribosomal ITS sequences of fungi and other eukaryotes for analysis of environmental sequencing data. Methods Ecol Evol. 4(10): 914–919.

Bengtsson, J., Hartmann, M., Unterseher, M., Vaishampayan, P., Abarenkov, K., Durso, L., et al., 2012. Megraft: a software package to graft ribosomal small subunit (16S/18S) fragments onto full-length sequences for accurate species richness and sequencing depth analysis in pyrosequencing-length metagenomes and similar environmental datasets. Res. Microbiol. 163(6-7):407-412.

Bhattacharya, D., Lutzoni, F., Reeb, V., Simon, D., Nason, J., Fernandez, F., 2000. Widespread occurrence of spliceosomal introns in the rDNA genes of Ascomycetes. Mol. Biol. Evol. 17(12): 1971–1984.

Bidartondo, M.I., 2008. Preserving accuracy in GenBank. Science. 319(5870):1616.

Blaalid, R., Kumar, S., Nilsson, R.H., Abarenkov, K., Kirk, P.M., Klauserud, H., 2013. ITS1 versus ITS2 as DNA metabarcodes for fungi. Mol. Ecol. Res. 13(2): 218–224

Blackwell, M., 2011. The Fungi: 1, 2, 3 … 5.1 million species? Am J. Bot. 98:426–428.

Bonifait, L., Marchand, G., Veillette, M., M’Bareche, H., Dubuis, M.E., Pépin, C., Cloutier, Y., Bernard, Y., Duchaine, C., 2017. Workers’exposure to bioaerosols from three different types of composting facilities. J. Occup. Environ. Hyg.14(10):815-822.

Brodie, E.L, DeSantis, T.Z., Parker, J.P., Zubietta, I.X., Piceno, Y.M., Andersen, G.L., 2007. Urban aerosols harbor diverse and dynamic bacterial populations. Proc. Natl. Acad. Sci. 104(1):299–304.

Brown, J.K.M., Hovmøller, M.S., 2002. Aerial dispersal of pathogens on the global and continental scales and its impact on plant disease. Science. 297(5581):537–541.

Brown, S.P., Veach, A.M., Rigdon-Huss, A.R., Grond, K., Lickteig, S.K., Lothamer, K., et al., 2015. Scraping the bottom of the barrel: are rare high-throughput sequences artifacts? Fungal Ecol. 13:221–225.

Burge, H.A., Rogers, C.A., 2000. Outdoor allergens. Environ. Health Perspect. 108:653–659.

Bush R.K., Portnoy J.M., Saxon A., Terr A.I., Wood R.A. 2006. The medical effects of mold exposure. J. allergy immunol. 117(2): 326-333.

Bush, R.K., 2008. Indoor allergens, environmental avoidance, and allergic respiratory disease. Allergy and Asthma Proc. 29(6):575–579.

Bush, R.K., Portnoy, J.M., Saxon. A., Terr, A.I., Wood, R.A., 2006. The medical effects of mold exposure. J. Allergy Clin. Immunol. 117(2):326–333.

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., et al., 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods. 7(5): 335-336.

CBOL Plant Working Group 1, Hollingsworth, P.M., Forrest, L.L., Spouge, J.L., Hajibabaei, M., Ratnasingham, S., et al., 2009. A DNA barcode for land plants. Proc. Natl. Acad. Sci. U.S.A. 106(31):12794–12797.

Chowdhary, A., Agarwal, K., Kathuria, S., Gaur, S.N., Randhawa, H.S. and Meis, J.F., Allergic bronchopulmonary mycosis due to fungi other than Aspergillus: a global overreview 2014. Crit. Rev. Microbiol. 40(1):30-48.

Daisey, J.M., Angell, W.J. and Apte, M.G., 2003. Indoor air quality, ventilation and health symptoms in schools: an analysis of existing information. Indoor Air. 13:53-64.

Dentinger, B.T., Didukh, M.Y., Moncalvo, J.M., 2011. Comparing COI and ITS as DNA barcode markers for mushrooms and allies (Agaricomycotina). PLoS One. 6(9):e25081.

Després, V.R., Huffman, J.A., Burrows, S.M., Hoose, C., Safatov, A.S., Buryak, G., et al., 2012. Primary biological aerosol particles in the atmosphere: a review. Tellus B. 64(1).

Douwes, J., Thorne, P., Pearce, N., Heederik, D., 2003. Bioaerosol health effects and exposure assessment: progress and prospects. Ann. Occup. Hyg. 47(3):187–200.

Dubuis, M.E., M’Bareche, H., Veillette, M., Bakhiyi, B., Zayed, J., Lavoie, J., Duchaine, C. 2017. Bioaerosols concentrations in working areas in biomethanization facilities. J. Air Manag. Assoc. 67(11):1258-1271.

Eames, I., Tang, J.W., Li, Y. and Wilson, P., 2009. Airborne transmission of disease in hospitals. J. Royal Soc. Interface. 6(suppl 6). Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 26(19):2460-2461.

Edgar, R.C., 2013. UPARSE: Highly accurate OTU sequences from microbial amplicon reads. Nat. Methods. 10:996-996.

Eduard, W., Heederik, D., Duchaine, C., Green, B.J., 2012. Bioaerosol exposure assessment in the workplace: the past, present and recent advances. J. Environ. Monit. 14(2):334–339.

Foarde, K., Dulaney, P., Cole, E., VanOsdell, D., Ensor, D., Chang, J., 1993. Assessment of fungal growth on ceiling tiles under environmentally characterized conditions. Proceedings of indoor air’93. Helsinki, Finland. 4:357–362.

Fogelmark, B., Sjostrand M. Rylander, R., 1994. Pulmonary inflammation induced by repeated inhalations of (1,3)-D-glucan and endotoxin. Int. J. Exp. Path. 75(2):85-90.

Ghiasian, S. D., Maghood, A.H., Aghamirian, M.R. 2017. Aeromycological analysis of allergenic airborne fungi in Qazvin, Iran. Curr. Med. Mycol. 2(3):5-9.

Gilbert, Y., Duchaine, C., 2009. Bioaerosols in industrial environments: a review. Can. J. Civil Eng. 36(12):1873-1886.

Glass, D., Amedee, R.G., 2011. Allergic fungal rhinosinusitis: a review. Ochsner J. 11(3):271–275.

Gomez-Alvarez, V., Teal, T.K., Schmidt, T.M., 2009. Systematic artifacts in metagenomes from complex microbial communities. ISME J.23(11):1314-7.

Han, J., Zhu, Y., Chen, X., Liao, B., Song, J., Chen, S., Meng, F., 2013. The short ITS2 sequence serves as an efficient taxonomic sequence tag in comparison with the full-length ITS. Biomed Res. Int. 2013:741476.

Hardin, B.D., Kelman B.J., Saxon, A., 2003. Adverse human health effects associated with molds in the indoor environment. J. Occup. Environ. Med. 45:470–478.

Hebert, P.D., Ratnasingham, S., deWaard, J.R., 2003. Barcoding animal life: cytochrome c oxidase subunit 1 divergences among closely related species. Proc. Royal Soc. Lond. 270(Suppl 1):S96– S99.

Heederik, D., Von Mutius, E., 2012. Does diversity of environmental microbial exposure matter for the occurrence of allergy and asthma? J. Allergy Clin. Immunol. 130(1):44–50.

Heinrichs, G., de Hoog, G.S., Haase, G., 2012. Barcode identifiers as a practical tool for reliable species assignment of medically important black yeast species. J. Clin. Microbiol. 50(9)3023–3030. Hospodsky, D., Qian, J., Nazaroff, W. W., Yamamoto, N., Bibby, K., Rismani-Yazdi, H. et al., 2012. Human occupancy as a source of indoor airborne bacteria. PLoS ONE. 7(4): e34867.

Huson, D.H., Auch, A.F., Qi, J., Schuster, S.C., 2007. MEGAN analysis of metagenomic data. Genome Res. 17(3):377-86.

Ihrmark, K., Bödeker, I.T.M., Cruz-Martinez, K., Friberg, H., Kubartova, A., Schenk, J., et al., 2012. New primers to amplify the fungal ITS2 region - evaluation by 454-sequencing of artificial and natural communities. FEMS Microbio. Ecol. 82(3): 666–677.

Ishii, N., Ishida, S., Kagami, M., 2015. PCR primers for assessing community structure of aquatic fungi including Chytridiomycota and Cryptomycota. Fungal Ecology 13:33–43.

Jones, A.M., Harrison, R.M., 2004. The effects of meteorological factors on atmospheric bioaerosol concentrations — a review. Sci. Total Environ. 326 (1–3):151–180.

Jost, L. DeVries, P., Walla, T., Greeney, H., Chao, A., Ricotta, C., 2010. Partitioning diversity for conservation analyses. Divers. Distrib. 16(1): 65-76.

Jost, L., 2006. Entropy and diversity. Oikos. 113(2):363-375.

Jost, L., 2007. Partitioning diversity into independent alpha and beta components. Ecology. 88(10): 2427-2439.

Kakde, U.B. 2012., Fungal bioaerosols: global diversity, distribution and its impact on human beings and agricultural crops. Bionano. Frontier. 5(2):323-329.

Kelly, L.J., Hollingsworth, P.M., Coppins, B.J., Ellis, C.J., Harrold, P., Tosh, J., et al., 2011. DNA barcoding of lichenized fungi demonstrates high identification success in a floristic context. New Phytol. 191(1):288–300.

Kembel, S.W., Wu, M., Eisen, J.A., Green, J.L., 2012. Incorporating 16S gene copy number information improves estimates of microbial diversity and abundance. PLoS Comput. Biol. 8910):e1002743.

Kohout, P., Sudova, R., Janouskova, M., Ctvrtlikova, M., Hejda, M., Pankova, H., et al., 2014. Comparison of commonly used primer sets for evaluating arbuscular mycorrhizal fungal communities: Is there a universal solution? Soil Biol. Biochem. 68:482–493.

Koljalg, U., Nilsson, R.H., Abarenkov, K., Tedersoo, L., Taylor, A.F., Bahram, M., et al., 2013. Towards a unified paradigm for sequence-based identification of fungi. Mol. Ecol. 22(21):5271-5277.

Kopylova, E., Noé, L., Touzet, H., 2012. SortMeRNA: Fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics 28:3211–3217.

Latgé, J.P., 1999. Aspergillus fumigatus and Aspergillosis. Clin. Microbio. Rev. 12 (2):310- 350.

Lazarus, K.L., James, T.Y., 2014. Surveying the biodiversity of the Cryptomycota using a targeted PCR approach. Fungal Ecol. 14:62-70.

Li, Y., Leung, G. M., Tang, J. W., Yang, X., Chao, C. Y. H., Lin, J. Z., et al., 2007. Role of ventilation in airborne transmission of infectious agents in the built environment – a multidisciplinary systematic review. Indoor Air. 17(1): 2–18.

Lindahl, B.D., Nilsson, R.H., Tedersoo, L., Abarenkov, K., Carlsen, T., Rasmus, K., et al., 2013. Fungal community analysis by high-throughput sequencing of amplified markers – a user’s guide. New phytol. 199(1):288-299.

Liu. Z., Cheng, K., Li, H., Cao, G., Wu, D. and Shi, Y. 2018. Exploring the potential relationship between indoor air quality and the concentration of airborne culturable fungi: a combined experimental and neural network modeling study. Environ. Sci. Poll. Res.

25(4):3510-3517.

Livermore, J.A., Mattes, T.E. 2013. Phylogenetic detection of novel Cryptomycota in an Iowa (United States) aquifer and from previously collected marine and freshwater targeted high-throughput sequencing sets. Environ. Microbiol. 15(8): 2333–2341.

Macher, J., Ammann, H.A., Milton, D.K., Burge, H.A., and Morey, P.R. (Editors)., 1999. Bioaerosols: assessment and control. American Conference of Governmental Industrial Hygienists (ACGIH). Cincinnati, Ohio, USA.

Magurran, A.E., McGill, B.J., 2011. (Editors) Biological diversity. Oxford University Press, London, UK.

Mbareche, H., Veillette, M., Bilodeau, G.J. and Duchaine, C. 2019b. Fungal spore recovery from air samples: tale of loss and gain. Appl. Environ. Microbiol. AEM.02941-18.

Mbareche, H., Veillette, M., Bilodeau, G.J., Duchaine, C., 2018. Bioaerosol Sampler Choice Should Consider Efficiency and Ability of Samplers to Cover Microbial Diversity. Appl. Environ. Microbiol. 10.1128/AEM.01589-18

Mbareche, H., Veillette, M., Bilodeau, G.J., Duchaine, C., 2019a. Fungal aerosols at dairy farms using molecular and culture techniques. Sci. Total Environ. 653:253-263.

Mbareche, H., Veillette, M., Bonifait, L., Dubuis, M.E., Bernard, Y., Marchand G., Bilodeau, G.J., Duchaine, C., 2017. A next generation sequencing approach with a suitable bioinformatics workflow to study fungal diversity in bioaerosols released from two different types of compost plants. Sci. Total Environ. 601-602:1306-1314.

Mbareche, H., Veillette, M., Dubuis, M.E., Bakhiyi, B., Zayed, J., Lavoie, J., et al., 2018. Fungal bioaerosols in biomethanization facilities. J. Air Waste Mang Assoc. 68(11):1198- 1210.

McDowell, D.G., Burns, N.A., Parkes, H.C., 1998. Localised sequence regions possessing high melting temperatures prevent the amplification of a DNA mimic in competitive PCR. Nucleic Acids Res. 26(14):3340–3347.

Mistry, J., Finn, R.D., Eddy, S.R., Bateman, A., Punta, M., 2013. Challenges in Homology Search: HMMER3 and Convergent Evolution of Coiled-Coil Regions. Nucleic Acids Res. 41: e121, 2013

Monard, C., Gantner, S., Stenild, J., 2013. Utilizing ITS1 and ITS2 to study environmental fungal diversity using pyrosequencing. FEMS Microbiol. Ecol. 84(1):165–175.

Morell, F., Cruz, M.J., Gòmez, F.P., Rodriquez-Jerez, F., Xaubet, A. and Munoz, X., 2011. Chacinero’s lung – hypersensitivity pneumonitis due to dry sausage dust. Scand. J. Work Environ. Health. 37(4):349-56.

Moreno, C.E., Rodrìguez, P., 2010. A consistent terminology for quantifying species diversity? Oecologia, 163:279-282.

Nilsson, R.H., Kristiansson, E., Ryberg, M., Hallenberg, N., Larsson, K.H., 2008. Intraspecific ITS variability in the kingdom Fungi as expressed in the International Sequence Databases and Its implications for molecular species identification. Evol. Bioinform. Online. 4:193-201.

Nilsson, R.H., Ryberg, M., Kristiansson, E., Abarenkov, K., Larsson, K.H, Kojalg, U., 2006. Taxonomic reliability of DNA sequences in public sequence databases: a fungal perspective. PLoS One. 1(1):e59.

Nilsson, R.H., Tedersoo, L., Ryberg, M., Kristiansson, E., Hartmann, M., Unterseher, M., et al., 2015. A Comprehensive, Automatically Updated Fungal ITS Sequence Dataset for Reference-Based Chimera Control in Environmental Sequencing Efforts. Microbes Environ. 30(2):145–150.

Nucci, M., Anaissie, E., 2007. Fusarium Infections in Immunocomprised Patients. Clin. Microbiol. Rev. 20(4):695-704.

Op De Beeck, M., Lievens, B., Busschaert, P., Declerck, S., Vangronsveld, J., Colpaert, J.V., 2014. Comparison and validation of some ITS primer pairs useful for fungal metabarcoding studies. PLoS ONE 9(6):e97629.

Osmundson, T.W., Robert, V.A., Schoch, C.L., Baker, L.J., Smith, A., Robich, G., et al. 2013. Filling gaps in biodiversity knowledge for macrofungi: contributions and assessment of an herbarium collection DNA barcode sequencing project. PLoS One. 8(4):e62419.

Paez-Rubio, T., Viau, E., Romero-Hernandez, S., Peccia, J., 2005. Source bioaerosol concentration and rRNA gene-based identification of microorganisms aerosolized at a flood irrigation wastewater reuse site. Appl. Environ. Microbiol. 71(2):804-810. Park, D., Ryu, S., Kim, S., Byun, H., Yoon, C., Lee, K., 2013. Airborne bacteria and fungi associated with waste-handling work. Int. J. Occup. Environ. Health. 19(4):313-317.

Pasanen, P., Pasanen, A.L., Janunen, M., 2000. Water condensation promotes fungal growth in ventilation ducts. Indoor Air. 3:106–112.

Perotto, S., Nepote-Fus, P., Saletta, L., Bandi, C., Young, J.P.W., 2000. A diverse population of introns in the nuclear ribosomal genes of ericoid mycorrhizal fungi includes elements with sequence similarity to endonuclease-coding genes. Mol. Biol. Evol. 17(1):44–59.

Pieckova E., Wilkins K., 2004. Airway toxicity of house dust and its fungal composition. Ann. Agric. Environ. Med. 11:67–73.

Pillai, S.D., Ricke, S.C., 2002. Bioaerosols from municipal and animal wastes: background and contemporary issues. Can. J. Microbiol. 48(8):681–696.

Porter, P., Susarla, S., Polikepahad, S., Qian, Y., Hampton, J., Kiss, A. and Corry, D., 2009. Link between allergic asthma and airway mucosal infection suggested by proteinase- secreting household fungi. Mucosal Immunol. 2(6):504–517.

Quince, C., Walker, A.W., Simpson, J.T., Loman, N.J. Segata, N., 2017. Shotgun metagenomics, from sampling to analysis. Nat. Biotechnol. 35(9):833-844.

Rodriguez, Z.C., Ramos, M.G., 2014. Acremonium species associated fungemia: a novel pathogen in the immunosuppressed patient. Bol. Asoc. Med. P. R. 106(3):29-31.

Roe, A.D., Rice, A.V., Bromilow, S.E., Cooke, J.E., Sperling, F.A., 2010. Multilocus species identification and fungal DNA barcoding: insights from blue stain fungal symbionts of the mountain pine beetle. Mol. Ecol. Res. 10(6):946–959.

Roy, C.J., Milton, D.K., 2004. Airborne transmission of communicable infection-the elusive pathway. N. Engl. J. Med. 530(17):1710-1712.

Ruzer, L.S., Harley, N.H., 2005. Aerosols handbook: measurement, dosimetry and health effects. CRC Press. Florida, USA.

Rylander, R., 1996. Airway Responsiveness and Chest Symptoms after Inhalation of Endotoxin or (1→3)-β-D-Glucan. Indoor Built Environ. 5:106-111.

Sanchez-Monedero, M.A., Stentiford, E.I. 2003. Generation and dispersion of airborne microorganisms from composting facilities. Trans IChem. 81(3):166-170.

Sarkar, A., Mukherjee, A., Ghoshal, A. G., Kundu, S., Mitra, S., 2010. Occurrence of allergic bronchopulmonary mycosis in patients with asthma: An Eastern India experience. Lung India. 27(4):212–216.

Schloss, P.D, Westcott, S.L., Ryabin, T., Hall, JR, Hartmann M, Hollister EB, et al. 2009. Introducing mother: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ Microbiol. 75(23):7537- 41.

Schoch, C.L., Seifert, K.A., Huhndorf, S., Robert, V., Spouge, J.L., Levesque, C.A., et al. 2012. Nuclear ribosomal internal transcribed spacer (ITS) region as a universal DNA barcode marker for Fungi. Proc. Natl. Acad. Sci. U.S.A 109(16):6241–6246.

Selman, M., Lacasse, Y., Pardo, A., Cormier, Y., 2010. Hypersensitivity pneumonitis caused by fungi. Proc Am Thorac. Soc. 7:229–23.

Shade, A., Hogan, C.S., Klimowiwicz, A.K., Linske, M., McManus, P.S., Handelsman, J., 2012. Culturing captures members of the rare biosphere. Environ. Microbiol. 14(9):2247– 2252.

Sharpton, T.J. 2014. An introduction to the analysis of shotgun metagenomic data. Front Plant Sci. 5:209.

Simon, X., Duquenne, P., 2014. Assessment of workers’ exposure to bioaerosols in a French cheese factory. Ann. Occup. Hyg. 58(6):677-92.

Stark, P. C., Celedón, J. C., Chew, G. L., Ryan, L. M., Burge, H. A., Muilenberg, M. L., & Gold, D. R. 2005. Fungal Levels in the Home and Allergic Rhinitis by 5 Years of Age. Environ. Health Perspect. 113(10):1405–1409.

Taha, M. P. M., Drew, G. H., Longhurst, P. J., Smith, R., Pollard, S.J.T., 2006. Bioaerosol releases from compost facilities: evaluating passive and active source terms at a green waste facility for improved risk assessments. Atmos. Environ. 40(6):1159–1169

Taha, M.P.M, Drew, G.H., Longhurst, P.J., Smith, R., Pollard, S.J.T., 2006. Bioaerosol releases from compost facilities: Evaluating passive and active sources terms at a green waste facility for improved risk assessments. Atmos. Environ. 40(6):1159-1169.

Taylor, D.L., Walters, W.A., Lennon, N.J., Bochicchio, J., Krohn, A., Caporaso, J.G., Pennanen, T., 2016. Accurate estimation of fungal diversity and abundance through improved lineage-specific primers optimized for Illumina amplicon sequencing. Appl. Environ. Microbiol. 82:7217–7226.

Tedersoo, L., Anslan, S., Bahram, M., Polme, S., Riit, T., Liiv, I., Koljalg, U., Kisand, V., Nilsson, R.H. Hildbrand, F., Bork, P., Abarenkov, K., 2015. Shotgun metagenomes and multiple pair-barcode combinations of amplicons reveal biases in metabarcoding analyses of fungi. MycoKeys. 10:1-43.

Tedersoo, L., Bahram, M., Põlme, S., Anslan, S., Riit, T., Kõljalg, U., et al., 2015. Response to comment on "Global diversity and geography of soil fungi": Analytical biases in microbial diversity studies. Science. 359:936.

Tischer C.G., Heinrich J., 2013. Exposure assessment of residential mould, fungi and microbial components in relation to children’s health: achievements and challenges. Int. J. Hyg. Environ. Health. 216(2):109-114.

Toju, H., Tanabe, A.S., Yamamoto, S., Sato, H., 2012. High-coverage ITS primers for the DNA-based identification of Ascomycetes and Basidiomycetes in Environmental Samples. Plos One. 7(7): e40863.

Truong, D.T., Tett, A., Pasolli, E., Huttenhower, C. Segata, N., 2015. MetaPhlAn2 for enhanced metagenomic taxonomic profiling. Nat. Methods . 2:902-903.

Veech, J.A., Crist, T.O. 2010. Diversity partitioning without statistical independence of alpha and beta. Ecology. 91(7):1964-1969.

Velegraki, A., Cafarchia, C., Gaitanis, G., Iatta, R., Boekhou, T., 2015. Malassezia infections in humans and animals: pathophysiology, detection, and treatment. PLoS Pathog. 11(1):e1004523.

Vrålstad, T., Myhre, E., Schumacher, T., 2002. Molecular diversity and phylogenetic affinities of symbiotic root-associated ascomycetes of the Helotiales in burnt and metal polluted habitats. New Phytol. 155(1):131–148.

Wang, X.C., Liu, C., Huang, L., Bengtsson-Palme, J., Chen, H., Zhang, J.H., et al., 2014. ITS1: a DNA barcode better than ITS2 in eukaryotes? Mol. Ecol. Res. 15(3):573-586.

White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Book: PCR protocols: A guide to Methods and Applications. Academic Press. Cambridge, MA, USA.

Whittaker, R.H. 1972. Evolution and Measurement of Species Diversity. Taxon. 21:213- 251.

Wyngaarden, J. B., Smith, L.H., Bennett, J.C., (ed.), 1992. Cecil textbook of medicine, 19th ed., p.399and 407. W.B. Saunders Co.. Philadelphia, Pa, USA.

Yu, I.T.S, Li, Y., Wong, T.W., Tam, W., Chan, A.T., Lee, J.H.W., et al., 2004. Evidence of Airborne Transmission of the Severe Acute Respiratory Syndrome Virus. N. Engl. J. Med. 350:1731-1739.

Zekovic D.B., Kwiatkowski S., Vrvic M.M., Jakovljevic D., Moran C.A., 2005. Natural and modified (1→3)-β-D-Glucans in health promotion and disease alleviation. Crit. Rev. Biotechnol. 25(4):205-230.

Chapter 3: Fungal cell recovery from air samples: a tale of loss and gain

3.1 Résumé

Les méthodes moléculaires permettent de mieux comprendre l’impact des communautés microbiennes dans un contexte d’exposition occupationnelle. Cependant, l'application de ces techniques aux bioaérosols, en particulier aux moisissures, présente des difficultés. Cette étude révèle qu'il existe une perte de cellules fongiques lorsque des échantillons d’air liquides sont concentrés par centrifugation. L'objectif est de développer un nouveau protocole pour la concentration des spores de moisissures par filtration avant l'extraction de l'ADN en utilisant des échantillons environnementaux. Les résultats obtenus avec la nouvelle technique de filtration ont montré une concentration de moisissures plus élevés comparativement à la centrifugation. De plus, le séquençage haut débit a révélé un profil de diversité différent selon la méthodologie utilisée. Ces travaux ont permis d'identifier la perte de cellules fongiques lors du traitement des échantillons d'air par centrifugation, et de proposer une méthode alternative pour une évaluation plus précise de l'exposition et de la diversité des moisissures.

3.2 Summary of the Paper

In this paper, we tell the story of how the comparison of Aspergillus fumigatus concentrations in air samples with qPCR and culture methods led to the discovery of a methodology problem that caused the loss of fungal cells, thus, affecting the overall biomass and diversity outcomes. The fact that the concentration of cultivable A. fumigatus was higher than the biological material detected by a specific qPCR made us reconsider the centrifugation step that we apply to liquid air samples to concentrate the biomass collected prior to DNA extraction. The hypothesis of the loss due to centrifugation came from the hydrophobic characteristic of fungi, the electric charges on the outer membrane of the cells and their know behaviour in air/water surfaces. The objective of the study was to develop a new protocol to concentrate air samples prior to DNA extraction using a filtration method. This new alternative protocol, presented in figure 3.1 of this paper, was compared to the centrifugation protocol usually used, and the comparison was applied to environmental field samples. The authors used specific qPCR targeting Penicillium/Aspergillus and A. fumigatus combined to HTS targeting ITS1 to validate the hypothesis of the fungal biomass and diversity loss caused by centrifugation.

The results obtained confirmed the loss as the new filtration-based method allowed a significant gain of fungal cells. The diversity analyses obtained after the sequencing of ITS1 amplicons demonstrated specific taxa that were affected by the centrifugation loss. In a laboratory experiment, the authors used strains of Aspergillus niger and Penicillium roquefortii grown in laboratory conditions to support the environmental results, and prove, once again, the effect of centrifugation fungal loss. The filtration protocol described in this work offers a better description of the fungal diversity of aerosols and should be used in fungal aerosol studies. This research is relevant for any fungal study that include resuspension of samples into a liquid.

Fungal cell recovery from air samples: a tale of loss and gain

RUNNING TITLE Fungal Cells Recovery From Air AUTHORS Hamza Mbareche1,2, Marc Veillette1, Wieke Teertstra3, Willem Kegel4, Guillaume J. Bilodeau5, Han A.B. Wösten3 and Caroline Duchaine1,2 AUTHORS’ AFFILIATION 1. Centre de recherche de l’institut universitaire de cardiologie et de pneumologie de Québec, Quebec City (Qc), Canada 2. Département de biochimie, de microbiologie et de bio-informatique, Faculté des sciences et de génie, Université Laval, Quebec City (Qc), Canada 3. Microbiology, Department of Biology, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands 4. Department of Physical and Colloid Chemistry, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands 5. Pathogen Identification Research Lab, Canadian Food Inspection Agency (CFIA). Ottawa, Canada KEYWORDS Bioaerosols, fungal cells, air samples, recovery method, centrifugation, filtration, taxon loss CORRESPONDING AUTHOR Mailing address: Caroline Duchaine, Ph.D., Centre de Recherche de l’Institut Universitaire de Cardiologie et de Pneumologie de Québec, 2725 Chemin Ste-Foy, Québec, Canada, G1V 4G5. Phone: (418) 656-8711 ext. 5837. Fax: 418 656-4509. E-mail: [email protected]

PUBLISHED: Applied and Environmental Microbiology 2019 Mar 1. pii: AEM.02941-18. doi: 10.1128/AEM.02941-18

3.3 Abstract

There are limitations in establishing a direct link between fungal exposure and health effects due to the methodology used, amongst other reasons. Culture methods ignore the nonviable/uncultivable fraction of airborne fungi. Molecular methods allow for a better understanding of the environmental health impacts of microbial communities. However, there are challenges when applying these techniques to bioaerosols, particularly to fungal cells. This study reveals that there is a loss of fungal cells when samples are recovered from air using wet samplers and aimed to create and test an improved protocol for concentrating mold spores via filtration prior to DNA extraction. Results obtained using the new technique showed that up to 3 orders of magnitude more fungal DNA was retrieved from the samples using qPCR. A sequencing approach with MiSeq revealed a different diversity profile depending on the methodology used. Specifically, eight fungal families were highlighted to be differentially abundant in centrifuged and filtered samples, out of 19 families tested. An experiment using laboratory settings showed the same spore loss during centrifugation for Aspergillus niger and Penicillium roquefortii strains. We believe that this work helped identify and address fungal cell loss during processing of air samples including centrifugation step and propose an alternative method for a more accurate evaluation of fungal exposure and diversity.

Importance

This work shed light on a significant issue regarding the loss of fungal spores when recovered from air samples using liquid medium and centrifugation to concentrate air particles before DNA extraction. We provide proof that the loss affect the overall fungal diversity of aerosols, and that some taxa are differentially more affected than others. Furthermore, a laboratory experiment confirmed the environmental results obtained during the field sampling. The filtration protocol described in this work offers a better description of the fungal diversity of aerosols and should be used in fungal aerosol studies.

3.4 Introduction

Human exposure to diverse and dynamic airborne microbial communities have major impacts on public health in both urban and rural environments. These impacts range from allergies and asthma to the dispersal of pathogens and health effects from occupational exposure (Brodie et al., 2007; Brown and Hovmøller, 2002; Douwes, 2003; Heederik and Von Mutius, 2012). Industrial environments are at the centre of occupational health issues due to the many types of raw materials used, the prevalence of operations that release harmful bioaerosols and the eventual high bioaerosol concentration in confined space. (Bonifait et al., 2017; Dubuis et al., 2017; Lecours et al., 2012; Nehme et al., 2008; Just et al., 2011; Veillette et al., 2004). To better understand the potential risks associated with complex environments, it is essential to be aware of the nature of the airborne microorganisms in order to better control or prevent the potential health effects.

Fungal bioaerosols consisting of spores and hyphal fragments may be respirable and are potent elicitors of bronchial irritation and allergies, pulmonary inflammation, increased sensitivity to endotoxins, mucous membrane irritation syndrome, nasal congestion, sore throat, and irritation of the nose and eyes (Porter et al., 2009; Hardin et al., 2003; Glass and Amedee, 2011; Chowdary et al., 2014; Pieckova and Wilkins, 2004; Selman et al., 2010; Rylander, 1996; Zekovic et al., 2005; Fogelmark et al., 1994; Beezhold et al., 2008; Bush, 2008). The fungal impact on occupational health is largely underestimated (Oppliger and Duquenne, 2016). The incomplete portrait of fungal community documented using standard exposure assessment approaches still represents a barrier in establishing the link between respiratory problems and exposure to fungi (Bush et al., 2006; Tischer and Heinrich, 2013).

Typically, molds are collected from air using different sampling devices and studied through both culture and/or non-culture based methods. For instance, fungal load has been established by monitoring microscopic spore counts in aerosols (O’connor et al., 2015; Vestlund et al., 2014). Other studies have used culture methods for detecting specific fungal species (Sanchez-Monedero and Stentiford, 2003; Taha et al. 2003; Park et al., 2013; Schlosser et al., 2013). However, it is well documented that only 1% of fungi recovered from

air samples are culturable therefore using these techniques may lead to an underestimation of the real fungal diversity and biomass in bioaerosols (Peccia and Hernandez, 2006). Molecular approaches are a good alternative for getting around the non-viable/non-culturable limits of the commonly used methods mentioned previously. One widely used molecular approach is the specific real-time PCR amplification of targeted genes. This method is used to quantify specific fungi (Bellanger et al., 2010) and/or targets conserved regions of the 18S rRNA gene for the quantification of a greater diversity of fungi (Qian et al., 2013). Another molecular technique includes a high-throughput sequencing approach of taxonomy- meaningful genes, which offers a more thorough analysis of the microbial content, thanks to the millions of sequences that are generated. In this latter case, the eukaryotic ITS1 region, considered to be the universal fungal barcode, was amplified and analyzed through high- throughput sequencing (Schoch et al., 2012). This allowed for an in-depth characterization of fungal diversity in the collected samples (Mbareche et al., 2013).

When working with air samples, microorganisms concentration is commonly needed before applying DNA purification and further molecular techniques to a sample, especially with air samples from liquid impingers or filters eluted in a liquid. One of the common ways to recover fungal cells from the air for subsequent molecular analysis is to use a sampler where the fungal material is collected in a liquid or on a filter membrane and then eluted in a liquid buffer. It is then concentrated by centrifuging the liquid samples and resuspending the pellets in a smaller volume of buffer solution (Vaccari et al., 2006). Using centrifugation to concentrate microorganisms is a common practice in commercial columns-based extraction kits and it is often used with bacterial and archaea specimens (Jacobsen and Rasmussen, 1992; Cullison and Jaykus, 2002; Lucore et al., 2000) for several types of samples including environmental aerosols (Cayer et al., 2007; Nehmé et al., 2008; Nehmé et al., 2009; Lecours et al., 2012).

Recently, as a part of a study describing bioaerosols in composting sites, Aspergillus fumigatus was quantified using culture and molecular approaches (qPCR) (Bonifait et al., 2017). The comparison between culture and real-time PCR using a centrifugation protocol prior to DNA extraction led to inconsistent results, with higher concentrations of Aspergillus

100

fumigatus found in samples using culture. This observation contradicts the actual detection efficiency of both methods, and the results observed from bacteria successfully concentrated using centrifugation of air samples (Nehmé et al., 2008; Lecours et al., 2012; Cayer et al., 2007; Nehmé et al., 2009). One hypothesis that might explain these results is that fungi have many distinct features including charge (polarity) (Gregory, 1957) and hydrophobicity (Wösten et al., 1999; Linder et al., 2005) that may cause a different behaviour during centrifugation. Hydrophobicity is caused by hydrophobins (Wösten et al., 1994) or by hydrophobin-like proteins such as repellents (Teertstra et al., 2006). These surface proteins make fungal spores water repellent, which may cause these structures to be recovered less effectively through centrifugation, as a portion of them may be/are lost during the disposal of the supernatant. Spores with a lower density as water may also be lost when the supernatant is removed.

Previous studies may have underestimated the fungal load and diversity in aerosol samples probably due to the method used for aerosol sample treatment prior to DNA extraction: centrifugation, vortexing/shaking and enzymatic/chemical lysis (Méheust et al., 2013; Bellanger et al., 2012; Madsen et al., 2015; Hospodsky et al., 2014; Rittenour et al., 2012; Méheust et al., 2012). In some cases, fungi were underrepresented compared to bacteria in bioaerosols. This can lead to an underestimation of the fungal exposure in workplaces, schools, hospitals and homes. Hence, the possibility of the health impact of fungi can be still undervalued.

The goal of this research is to develop a new molecular-method targeted protocol for concentrating mold spores recovered from air samples prior to nucleic acid extraction in order to overcome the loss during centrifugation, which is a widely used method for processing air samples. A real-time PCR and a next-generation sequencing approach was applied to samples from two different environments. Results obtained using the new technique were compared with results using standard centrifugation protocols applying microbial ecology models. This research raise the question of fungal cell loss when recovered from air samples, due to centrifugation, and propose an alternative method to better evaluate fungal diversity and human occupational exposure. This research is relevant for occupational and non-

101

occupational environments where liquid analyses is expected no matter what air sampling regime is used. Also, studies involving bulk dust or settled dust sampling (vacuuming or electrostatic dust collection), that include resuspension of dust into a liquid, will face the same problem of sample concentration.

3.5 Materials and Methods

Field Work

Environmental Field Samples

The samples used for this study were collected as part of two different other studies. The common goal was to assess occupational exposure to bioaerosols. Thus, the sampling sites were chosen according to workers’ activities. Below is a summary of the field sampling methods.

Compost. The compost samples come from a year-long sampling schedule of compost piles from three different composting plants in order to monitor their composting processes. Each plant treats different raw materials: Household green waste (domestic); Manure and hay (vegetal);Pig carcasses and placenta (animal). All of the composting plants were located in the province of Quebec, Canada. Outdoor temperature varied from 7oC to 10oC during spring visits, 19oC to 28oC during summer and 1oC to 5oC during the fall visit. Sampling took place in the morning and could last until 1pm, depending on workers’ activity. At each visit, three samples were taken in the beginning, the middle and the end of the working shifts. Detailed information about the sampling schedule is presented in the original composting study report (Bonifait et al., 2017).

Biomethanization. Samples were collected from two different facilities as a part of a biomethanization study. One facility (BF1) processes primary and secondary sludge from wastewater treatment plants, as well as industrial waste. The second one (BF2) handles

102

municipal waste from domestic sources. Both facilities were visited once during summer and once during winter. Five sites were sampled in the 1st BF and two sites in the second one. At each site, three samples were taken in the beginning, middle and the end of the process. Samples were then pooled together to obtain a final volume of 45ml for each site sampled. The time of the sampling campaign was dependent on workers’ activity and was comprised between 8am and 4pm. During the summer sampling time, temperature ranged from 22°C to 26°C. During winter, temperatures ranged between 18°C and 23°C. Detailed information about the sampling sites can be found in to the original study report (Dubuis et al., 2017).

Air Sampling

A liquid cyclonic impactor Coriolis µâ (Bertin Technologies, Montigny-le- Bretonneux, France) set at 300L/min for 10 minutes was used to collect bioaerosols for this study. Fifteen millilitres of sterile 5mM phosphate-buffered saline (pH 7.4) were used to fill the sampling cone of the Coriolis. In both environments and in all the sampling sites, the sampler was placed within 1-2 m from the source.

Culture Approach

One millilitre of the 15ml of Coriolis sampling liquid (5mM phosphate-buffered saline) was used to perform a serial dilution from 100 to 10-4 concentration/ml. The dilutions were made using a saline and tween solution and were performed in triplicate. One hundred microlitres of each triplicate were plated on Rose bengal media with chloramphenicol at a concentration of 50µg/ml. Half of the petri dishes were incubated at 25°C for mesophilic mold growth and the other half at 50°C for thermophilic mold growth, specifically Aspergillus fumigatus. After 5 days of incubation, molds were counted and the counts were translated into CFU/m3. The concentrations were obtained after calculating a mean count of the triplicate plated dilutions. Aspergillus fumigatus was identified according to macroscopic and microscopic specific features. Spores were observed with Leitz Laborlux S microscope and lactophenol blue staining.

103

Fungal Cell Concentration Prior to DNA Extraction

Standard centrifugation protocol. In this paper, the centrifugation protocol is referred to as the standard because of its use in most prior bioaerosol studies including molecular biology approaches. Each sample was divided into three aliquots of 1.5 ml each which were centrifuged for 10 min at 14 000 x g. The supernatant was discarded and the pellets were kept at -20°C until the DNA extraction.

Newly developed filtration-based protocol. Ten ml aliquots of Coriolis suspension was filtered through a 2.5 cm polycarbonate membrane (0.2-mm pore size; Millipore) using a vacuum filtration unit. The filters were placed in a 1.5 ml Eppendorf tube with 750µl of the extraction buffer (bead solution) from a MoBio PowerLyserÒ PowersoilÒ Isolation DNA kit (Carlsbad, CA, U.S.A.) and a tungsten bead with a diameter of 0.3 cm. The filters were flash-frozen by placing the Eppendorf tube in a 99% ethanol solution and dry ice. Next, the frozen filters were pulverized using tungsten steel beads in an Eppendorf tube in a bead- beating machine (Mixer Mill MM301, Retsch, Düsseldorf, Germany) at a frequency of 20 revolutions per second for 20 minutes. An aliquot of 1.5 ml of the liquid containing the pulverized filter particles was used for the first step of the DNA extraction kit (Fig. 3.1).

104

Figure 3.1: Diagram of the fungal cells concentration protocols prior to DNA extraction. The diagram shows each step from the sampling to the DNA extraction. The red circle shows the hypothesis of the fungal cell loss during the disposal of the supernatant in the centrifugation protocol. The SASS 3100 dry sampler was used as a control condition presented in Fig. S1 to Fig.S3. After particle collection on the electret filter, a SASS® 3010 Particle Extractor is required to elute the captured particles in a buffer. The particles are trapped in the filter via electric charges, and the use of the buffer changes the charges of the particles, which are collected in the liquid buffer

DNA Extraction

A second bead beating using glass beads was conducted at a frequency of 20 revolutions per second for 10 minutes. Next, a MoBio PowerLyserÒ PowersoilÒ Isolation DNA kit (Carlsbad, CA, U.S.A.) was used to extract the total genomic DNA from the samples following the manufacturer’s instructions. After the DNA elution, samples were stored at - 20°C until subsequent analyses.

105

Real-Time PCR Quantification

A Bio-Rad CFX 96 thermocycler (Bio-Rad Laboratories, Mississauga, CANADA) was used for DNA amplification. The PCR mixture contained 2 µl of DNA template, 0.150 µmol/liter per primer, a 0.150 µmol/liter probe, and 7.5µl of 2× QuantiTect Probe PCR master mix (QuantiTect Probe PCR kit; Qiagen, Mississauga, Ontario, Canada) in a 15-µl reaction mixture. Standard curves were made with a strain of Aspergillus fumigatus (isolated from an environmental sample and identified with key identification tools). After extraction of the genomic DNA, a spectrophotometer Nanodrop (Thermo Scientific, Massachusetts, USA) was used to quantify the concentration of the extracted DNA. Then, dilutions from 106 to 100 were used in triplicates for the standard curve calculations. The results were analyzed using Bio-Rad CFX Manager software version 3.0.1224.1015 (Bio-Rad Laboratories), including efficiency and R2 of the standard curve slope. Table 3.1 lists the primers, probes and PCR protocol used for this study.

Table 3.1: Primers and probes used for qPCR quantification of selected microorganisms Microorganisms and references Primers and probes PCR protocol Penicillium, Aspergillus and Paecilomyces PenAsp1mgb (Taqman) Activation: 94°C-3min variotii PenAspR1: 5’-GCCCGCCGAAGCAAC-3’ Denaturation: 94°C-15sec PenAspF1: 5’-CGGAAGGATCATTACTGAGTG- Annealing/extension: 60°C-60sec http://www.epa.gov/microbes/moldtech.html 3’ Cycles: 40 PenAspP1mgb: 5’-FAM- CCAACCTCCCACCCGTG-TAMRA-3’ Aspergillus fumigatus and Neosartoya Afumi (Taqman) Activation: 94°C-3min fischeri AfumiR1: 5’- Denaturation: 94°C-15sec CCGTTGTTGAAAGTTTTAACTGATTAC-3’ Annealing/extension: 60°C-60sec http://www.epa.gov/microbes/moldtech.html AfumiF1: 5’-GCCCGCCGTTTCGAC-3’ Cycles: 40 AfumiP1: 5’-CCCGCCGAAGACCCCAACATG- 3’

High-Throughput Sequencing

Amplification of the fungal ITS1 region, equimolar pooling and sequencing was performed at the Plateforme d’analyses génomiques (IBIS, Université Laval, Quebec City, Canada). Briefly, amplification of the ITS1 region was performed using the sequence- specific regions references therein (Tedersoo et al., 2015), using a two-step dual-indexed

106

PCR approach specifically designed for Illumina instruments. In the first step, the gene- specific sequence was fused to the Illumina TruSeq sequencing primers and PCR was performed on a total volume of 25µL containing 1X Q5 buffer (NEB), 0.25µM of each primer, 200µM of each dNTPs, 1U of Q5 High-Fidelity DNA polymerase (NEB) and 1µL of template cDNA. The PCR started with an initial denaturation at 98°C for 30s followed by 35 cycles of denaturation at 98°C for 10s, annealing at 55°C for 10s, extension at 72°C for 30s and a final extension at 72°C for 2 min. The PCR reaction was purified using the Axygen PCR cleanup kit (Axygen, New York, USA). Quality of the purified PCR products was checked on a 1% agarose gel. Fifty to 100-fold dilution of this purified product was used as a template for a second PCR step with the goal of adding barcodes (dual-indexed) and missing sequences required for Illumina sequencing. Cycling for the second PCR was identical to the first PCR but with 12 cycles. PCR reactions were purified as above, checked for quality on a DNA7500 Bioanlayzer chip (Agilent) and then quantified spectrophotometrically with the Nanodrop 1000 (Thermo Fisher Scientific). Barcoded Amplicons were pooled in equimolar concentration for sequencing on the Illumina Miseq. The oligonucleotide sequences that were used for amplification are presented in Table 3.2. Please note that primers used in this work contain Illumina specific sequences protected by intellectual property (Oligonucleotide sequences © 2007-2013 Illumina, Inc. All rights reserved. Derivative works created by Illumina customers are authorized for use with Illumina instruments and products only. All other uses are strictly prohibited.)

Table 3.2: Primers used for Illumina amplification. First-PCR primer ITS1Fngs: 5’-ACACTCTTTCCCTACACGACGCTCTTCCGATCTGGTCATTTAGAGGAAGTAA-3’ ITS2: 5’-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGCTGCGTTCTTATCGATGC-3’ Second-PCR Generic forward: 5’AATGATACGGCGACCACCGAGATCTACAC[index1]ACACTCTTTCCCTACACGAC-3’ primer Generic reverse: 5’CAAGCAGAAGACGGCATACGAGAT[indexe2]GTGACYGGAGTTCAGACGTGT-3’

Bioinformatics Workflow

Briefly, after demultiplexing the raw FASTQ files, the reads generated from the paired-end sequencing using Mothur v 1.35.1 were combined (Schloss et al., 2009). The

107

quality filtering was performed using Mothur discarding reads with ambiguous sequences. Reads shorter than 100 bp and longer than 450 bp were also discarded. Similar sequences were gathered together to reduce the computational burden, and the number of copies of the same sequence was displayed. This dereplication step was performed using USEARCH (version 7.0.1090) (Edgar, 2010). The selected region of fungal origin was then extracted from the sequences with ITSx which uses HMMER3 (Mistry et al., 2013) to compare input sequences against a set of models built from a number of different ITS region sequences found in various organisms. Only the sequences belonging to fungi were kept for further analyses. Operational taxonomic units (OTUs) with 97% similarity cut-off were clustered using UPARSE (VERSION 7.1) (Edgar, 2013). The similarity threshold (97%) is commonly used in all OTU based analyses and shown to be optimal when using ITS for fungal identification (Kõljalg et al., 2013). UCHIME was used to identify and remove chimeric sequences (Edgar, 2011). QIIME (version 1.9.0) (Caporaso et al., 2010) was used to assign taxonomy to OTUs based on a UNITE fungal ITS reference training data set for taxonomic assignment and to generate an OTU table. The microbial diversity analysis conducted in this study was achieved by using QIIME commands described in the QIIME scripts (http://qiime.org/scripts/).

Laboratory Experiments

Strains and Culture Conditions

Wild-type A. niger strain N402 (Bos et al., 1988) and wild-type P. roquefortii strain FM164 (also known as LCP6094) (Ropars et al., 2011) were grown on 20 ml minimal medium (MM) (de Vries et al., 2004) with 2% glucose and 1.5% agar at 30°C and 25°C, respectively. The melanin deficient A. niger strain ΔpptA was grown on 20 ml MM/siderophore spent medium (50/50) with 2% glucose, 20mg/ml L-leucine and 1.5 % agar (Jørgensen et al., 2001). Plates were inoculated by spreading conidia confluently over the agar surface. After 7 days conidia were harvested with 10ml H2O per plate and filtered over Miracloth (Millipore, www.merckmillipore.com). Spores were counted in a 100 fold dilution in Saline/Tween (0.9% NaCl, 0.005% Tween80) (S/T). Accordingly, spore suspensions were

108

brought to a final number of 1.108/ml. Spores were used in experiments within 2 hours after

7 harvesting at a concentration of 1.10 /ml H2O or 150 mM NaCl.

Centrifugation Analysis

The effect of triboelectricity was assessed qualitatively by charging polypropylene tubes by rubbing with wool. Charging was confirmed by attraction of a slip of paper to the tubes. Two aliquots of 1 ml were transferred via a charged 15 ml polypropylene tube (Sarstedt 62.554.502.PP; www.sarstedt.com) to a 1.5 ml polypropylene micro tube (Sarstedt 72.690.001), while another aliquot was directly added to the microtube. Spores were centrifuged for 10 min at 1700 g at 4°C in an fixed angled Eppendorf rotor. The experiments were performed at the 3rd (A.niger N402), 8th (P.roquefortii FM164), and 9th (A.niger ΔpptA) of May 2018. In the next set of experiments, the effect of triboelectricity was assessed quantitatively by performing the same experiments on the 15th (A. niger) and 23rd (P. roquefortii) of May 2018. The supernatant was carefully pipetted off by first taking the upper

900µl followed by the lower 100 µl. The pellet was resuspended in 100 µl H2O by pipetting and transferred to a clean microtube. Spores were counted using a haemocytometer. When necessary 10 or 100 fold dilutions were made using S/T.

Statistical Analysis

Descriptive statistics were used on qPCR data to highlight significant differences. The normality was verified by the D′Agostino and Pearson omnibus normality test. The normality assumption on data was not fulfilled. Non-parametric Mann-Whitney U test (two- tailed) analyses were performed to highlight significant differences showing a p-value ≤ 0.05. All the results were analyzed using the software GraphPad Prism 5.03 (GraphPad Software, Inc.).

For sequencing data, the differences in the observed OTUs showed with boxplots were also analyzed with non-parametric Mann-Whitney U test using the software GraphPad Prism 5.03 (GraphPad Software, Inc.). Deseq2 statistical tests was used to determine the

109

statistical significance of the differential abundance of OTUs across samples. Deseq2 test was used as a part of the QIIME script for differential abundance analyses. Detailed information about the performance of the test are presented in the differential abundance section of the results. Outdoor control samples were excluded from the analyses due to the low number of sequences, and the low number of OTUs exhibited. This way, the rarefaction depth is high, and we eliminate the possible biases related to samples clustering according to the number of sequences rather than the diversity.

Data Availability

Raw sequence reads of every sample used in this study and that support its findings have been deposited in the National Center for Biotechnology Information (NCBI) under the

BioProject ID : PRJNA450069 https://www.ncbi.nlm.nih.gov/bioproject/450069.

3.6 Results

Field Work

Compost

To study compost, culture methods and real-time PCR were used to quantify the concentration of Aspergillus fumigatus in aerosol samples collected with the Coriolis sampler during various visits to three different composting facilities (Fig. 3.2). The standard centrifugation protocol was used for fungal cell concentration prior to DNA extraction and real-time PCR quantification. In Fig. 3.2, the triangles and squares represent the concentration of A. fumigatus obtained from samples collected while workers were performing their normal duties. For samples from the domestic compost facility, culture methods yielded higher concentrations than qPCR during the four visits. Samples from two out of four visits were positive for A. fumigatus in the vegetal composting facility and they

110

were both results obtained using culture method. qPCR analyses of the same samples were negative for A. fumigatus. A similar observation was made for samples from the animal composting facility. Results from three out of four visits exhibited higher concentrations of A. fumigatus in cultured samples compared to those from qPCR.

Figure 3.2: Concentration of Aspergillus fumigatus using culture and molecular (qPCR) methods in aerosol samples collected from three different composting facilities. Each point represents a different visit. The bars represent the mean value for each condition. The biological replicates are represented by the site sampled at each visit. Two visits in vegetal composting and one visit in animal composting were negative for A. fumigatus. Therefore, they do not appear on the figure.

Biomethanization

The bioaerosol samples collected from the two biomethanization facilities using the Coriolis sampler were separated into two groups. The first group was subjected to the new

111

filtration-based protocol and the second group was centrifuged using the standard protocol. After extracting the DNA from the concentrated samples, the efficiency of both concentration protocols was evaluated using qPCR targeting a region in the ITS gene common to the Aspergillus and Penicillium genera (Pen-Asp). Fig. 3.3 presents a comparison of the concentration of Penicillium and Aspergillus in the centrifuged and filtered samples. In samples from the first biomethanization facility, the five sites sampled showed a higher concentration of Penicillium and Aspergillus when filtration was used for concentrating the sample. Differences up to 2 orders of magnitude (102 to 104) were observed between the filtration and the centrifugation methods. One data point is missing because one site in BF1 showed no mesophilic fungal growth. Similarly, the two sites sampled in the second biomethanization facility exhibited concentrations of Penicillium and Aspergillus 100 times higher when fungal cells were concentrated using the filtration protocol. Furthermore, concentrations of Penicillium and Aspergillus obtained by qPCR in the filtered samples were compared to the concentrations of other mesophilic molds obtained by culture counts. Fig. 3.3 shows a difference of approximately one log between the concentrations obtained by the two methods with the qPCR method yielding the higher concentrations. This difference is consistent throughout the seven sites sampled (5 in BF1 and 2 in BF2). These results strongly suggest that the centrifugation protocol applied to the compost aerosols samples led to the loss of Aspergillus fumigatus spores, and this is supported by the results presented in Fig. 3.2.

Because the newly developed filtration-based protocol has multiple steps, it was necessary to set up control conditions to ensure that there is no bias due to the tungsten bead- beating step. In other words, the control combining centrifugation and tungsten bead-beating confirmed that the spore gain observed (with qPCR and high-throughput sequencing) was not due to more cells being disrupted by the bead beating step with the tungsten bead, but it is the result of the complete filtration protocol (Fig.S1, Fig.S2, Fig.S3). More information on the control experiments is provided in the supplement file (text under the section: Tungsten Bead Control Using Two Samplers).

112

Figure 3.3: Concentrations of Penicillium and Aspergillus (PenAsp/m3) using qPCR on filtered and centrifuged samples (left y-axis) compared with concentrations of mesophilic molds (CFU/m3) using culture counts (right y-axis) from bioaerosol samples collected from two different biomethanization facilities. The points on the graph represent the sites sampled in each facility. The bars represent the mean value for each condition. The biological replicates are represented by the site sampled during the different visits. The bar represent the mean concentration.

High-Throughput Sequencing

Bioaerosol samples collected at the biomethanization facilities were sequenced using an Illumina Miseq platform. The purpose of this analysis was to demonstrate the effect of centrifugation on the fungal diversity obtained from environmental samples. For

113

biomethanization facilities, 32 samples were sequenced and 1 812 622 raw sequences were generated. After quality filtering and chimera checking, sequences clustered into 5 132 OTUs.

Alpha Diversity

The numbers of observed OTUs in each set of samples are presented in Fig. 3.4. The filtered samples contained a higher number of OTUs compared to the centrifuged ones. The maximum number of OTUs obtained from filtration was two times higher than the number obtained from centrifugation. Alpha richness (observed OTUs) was calculated at a value of 30,000 sequences. This number was chosen based on the lowest-depth sample parameter, which represents the lowest number of sequences in a sample. Samples with a lower number than what is chosen are excluded from the analyses. The higher the number, the more accurate the results are. In the currently presented case, all the samples were included as they have more than 30,000 sequences per sample, except for outdoor control samples. The negative controls are samples taken outside the facilities at each visit during the summertime. The gap between the number of sequences in samples taken from the working sites in the facilities and the outdoor negative controls was too large (more than 28,000 sequences). For this reason, outdoor controls were excluded from the analyses. This last paragraph was added to the alpha diversity section of the results.

114

Figure 3.4: Boxplot representing the number of observed fungal OTUs in bioaerosol samples collected with the Coriolis in biomethanization facilities. The samples were categorized into two groups, centrifugation and filtration, according to the concentration protocol used.

Differential Abundance

After observing a difference in the number of fungal OTUs between the samples treated with different concentration protocols, it seemed a logical next step to try to identify the OTUs that had significantly different abundances across the sample categories. To accomplish this goal, a statistical test designed specifically for differential analyses of count data was used. DESeq2 is a statistical test package that estimates the variance-mean dependence in count data from high-throughput sequencing assays and tests for differential expression based on a model using the negative binomial distribution (Love et al., 2014). The DESeq2 package is available at http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html. In this context, it was used to test for differences in OTU abundance between groups of samples. The samples

115

were grouped into two categories based on concentration method: filtration and centrifugation. To ensure the test was appropriate for these data, a diagnostic plot was generated to validate the fit line on the dispersion plot. The results showed that the data fit with the dispersion values when plotted against the mean of the normalized counts (Fig. S5). The detailed output was in the form of an OTU text file containing a list of all the OTUs in the raw input matrix and their taxonomic identification along with their associated statistics. It also included p-values representing the statistical significance of the differential abundance in the group of samples (data set S1). After a detailed analysis, eight families of fungi had significantly different abundances of OTUs between the centrifuged and filtered samples. The OTUs that were associated with unidentified taxa were not analyzed. Table 3.3 lists the eight fungal families and the range of adjusted p-values obtained for their representative OTUs. The differential abundance was performed to the species level (whenever it was possible), or at least to the genera level. The detailed analyses refer to the fact that we scrutinized the OTUs that had the highest and significant log2Fold Change in the differential abundance analyses, and we summarized the findings to the family level to make it easier to visualize. In other words, the families that are listed in table 3.3 had the highest number of OTUs (identified to the genera or specie level) that had a significant differential abundance.

116

Table 3.3: Eight fungal families with the highest number of representative OTUs and the range of their adjusted p-values representing the statistical significance of their differential abundance between the groups of samples. The samples were separated according to the concentration protocol (centrifugation and filtration).

Taxonomy (family) P-values Number of OTUs (genera or species level)

Davidiellaceae 7.41x10-6 – 5.18x10-5 17

Pleosporaceae 3.69x10-15 – 2.40x10-5 14

Cystofilobasidiaceae 3.29x10-9 – 2.08x10-5 13

Ganodermataceae 5.43x10-5 - 1.15x10-5 10

Meruliaceae 9.28x10-6 – 2.22x10-3 20

Polyporaceae 3.60x10-3 - 3.85x10-2 15

Psathyrellaceae 1.91x10-2 - 1.28x10-2 7

Ophiocordycipitaceae 1.84x10-2 – 1.15x10-2 8

Taxonomic Distribution

To study differences in taxa between the centrifuged and filtered samples at the genus level, each sample was analyzed individually. Based on the OTUs, a list of genera was identified for each sample. Fifty taxa were identified in the 16 samples from the filtration protocol group. All of which were present in every single sample. Samples treated with the centrifugation protocol showed a different distribution. The most striking difference between samples from the two different protocols was that there were nine fungi not detected in any samples from the centrifuged group that were present in every filtered sample. The taxonomic distribution is described in Fig. 3.5. Briefly, 6 fungi were detected in 26% of the centrifuged samples, 18 fungi were detected in 40% of the centrifuged samples and finally 17 fungi were detected in more than half of the centrifuged samples.

117

A literature review about the presence of hydrophobins or oil droplets in the 7 fungal genera that were constantly absent when using the centrifugation protocol is presented in Table 3.4. For the majority, studies have reported the presence of oil droplets by direct observation on fungal cells using microscopy (Davidiella, Alternaria, Fusarium and Capnobotryella). In some cases, oil droplets and lipid production was assessed by chromatography profiles (Epicoccum and Cryptococcus). For Botrytis, presence of hydrophobin was confirmed by analysing gene expression during the sexual developement of the cells.

Figure 3.5: Distribution of fungi detected in the centrifuged samples. All 50 fungi listed were present in 100% of the filtered samples. Each pie represent the percentage of detection of the taxa listed under in the centrifuged samples. The taxa listed under each pie chart were present in the percentage of centrifuged samples shown above. Numbers in parenthesis represent the number of taxa listed above.

118

Table 3.4: The reported presence of hydrophobins or oil droplets in the seven fungal genera that were absent in the case of centrifugation.

Fungal genera Mention of oil droplets or Reference hydrophobins Epicoccum Production of fungal dye oil in Cretu et al., 2011; Wösten et water emulsions/hydrophobins al., 2001 absent Davidiella Presence of small oil droplets in Braun and Schubert, 2007 fungal cells/hydrophobins absent Alternaria Presence of numerous oil Kwasna and Kosiak, 2003 droplets in the vegetative hyphae/hydrophobins absent Botrytis Expression of hydrophobin Terhem and Van Kan, 2014 genes in sexual development of fungal cells Fusarium Observation of oil droplets Feng et al., 2005; Fuchs et al., surrounding wild type 2004 mycelium/hydrophobins present Cryptococcus Microbial lipid production by Zhang et al., 2011 fungal cells in different culture conditions/hydrophobins absent Capnobotryella Observation of oil droplets in Sterflinger et al., 1992 fungal cells/hydrophobins absent

Laboratory Settings

Strains of Aspergillus niger and Penicillium roquefortii grown in laboratory settings were tested for the loss due to centrifugation as observed for fungal species with environment field samples. Spores of A. niger wild-type did not pellet upon centrifugation at 1700 g when harvested in H2O at March 14, 2018 (calculated indoor RH 30%; Table S1). However, spores did pellet after they had been in contact with aluminum. This indicates that static electricity

119

prevented the spores to pellet (data not shown). To expand on this result, tubes were rubbed with wool creating a charged surface; the so-called triboelectric effect. Spore concentration was severely reduced by transferring the spores of A. niger wild-type, A. niger ΔpptA, and P. roquefortii via a charged 15 ml polypropylene tube (Fig. 3.6). Addition of 150 mM NaCl did not alleviate the spore loss. These data indicate that spores had adsorbed to the charged surface of the 15 ml tubes mainly due to the triboelectric effect. In the next set of experiments, we counted the spore loss (Table S2). This time, no effect was observed of the transfer via a charged 15ml tube, possibly due to a higher RH (i.e 48 and 56%) at the days of the experiments (May 15 and 23; Table S1). In the case of A. niger wt about 50% of the spores were not recovered, explained by adsorption to the wall of the micro tube during centrifugation. Notable, about 90% and 98% of the ΔpptA and P. roquefortii spores, respectively, were not recovered. These data show that spores adsorb to the wall of the tube during centrifugation and that melanin is not the cause of this effect. Spores may also adsorb to the tube wall during transfer but this requires static electricity.

The effect of adsorption to the wall of the tube during centrifugation was further assessed by using a fixed angle and a swing out rotor. Fixed angle rotor caused more spore loss due to the g-force that brings the spores in close contact with the side walls of the tube (Fig.S4).

Figure 3.6: Triboelectric effect on spore pelleting of A.niger N402 (A), P.roquefortii FM164 (B), and A.niger ΔpptA (C) after centrifugation for 10 min at 1700 g in an

Eppendorf fixed angle rotor. Spores in H2O (1,2) or 150 mM NaCl (3) were added to a 1. ml tube either (1,3) or not (2) after the transfer via a charged 15ml tube.

120

3.7 Discussion

This study demonstrates an underestimation of fungal load and diversity in aerosol samples due to centrifugation steps prior to DNA extraction and propose a new method to overcome this bias. The idea that fungal cells may be lost during the centrifugation step is supported by the water-repellent nature of fungi related to fungal hydrophobins and hydrophobin like proteins (Gebbink et al., 2005). Fungal hydrophobins are secreted proteins that fold into amphipathic membranes when fungal cell walls come in contact with water, oil, or hydrophobic solid surfaces (Walther and Farese, 2012). This is the case when spores are produced on aerial reproductive structures such as conidiophores. The hydrophobic nature makes that spores collect on the water surface or attach to the hydrophobic surface of plastic such as that of Eppendorf tubes. After the centrifugation step, the supernatant, which contains centrifugation-resistant fungal cells, is discarded resulting in an overall loss of spores. Another explanation of loss of spores after centrifugation is the presence of lipid droplets (LDs). These dynamic intracellular organelles can contain neutral lipids like triacylglycerols (TAGs) and sterol esters (SEs) (Brasaemle, 2007; Bickel et al., 2009; Welte, 2015). Apart from lipid storage, their role have been extended to lipid metabolism, energy homeostasis and their dysfunction is linked to many human diseases (Murphy, 2001). LDs are ubiquitous in eukaryotes and prokaryotes cells (Radulovic et al., 2013; Kobae et al., 2014). In fungi, they are believed to support growth and propagation of mycelia and are implicated in fungal spore development (Yang et al., 2017). However, when and where the LDs are produced in fungal cells is still unclear. Their presence in the fungal cell could be an explanation of the loss of certain fungi in the supernatant during the centrifugation protocol by lowering cell density. The newly developed filtration-based protocol circumvents this issue, as all the liquid is filtered and its content is fully recovered before the DNA extraction.

Using a next generation sequencing approach, we aimed to characterize the fungal cell loss caused by the centrifugation protocol. Assessments of airborne fungal diversity in various environments have been the focus of several recent studies (Kumari et al., 2016; Green et al., 2017; O’brien et al., 2016). There is an obvious need to document the effect of fungal cell loss due to centrifugation and to highlight the necessary precautions when

121

working with fungi from aerosols sampled in liquid and used with molecular techniques/methods. As expected, the fungal cell loss was not limited to Aspergillus and Penicillium species. The chao1 richness index and the Shannon/Simpson diversity indexes yielded similar results as those obtained by comparing the observed OTUs in Fig.4. Centrifugation clearly affects the diversity/richness of fungal bioaerosols. These results are consistent with the qPCR quantification of Aspergillus and Penicillium from the same samples. Overall, the filtration protocol allowed for a better recovery of specific fungi (Aspergillus and Penicillium) and for a greater diversity (number of OTUs, Chao1, Shannon and Simpson).

Taxonomic analyses demonstrated very compelling results as 100% of taxa detected by centrifugation were also detected by filtration. In contrast, taxa identified by filtration were present in portions of centrifuged samples ranging from 0% to 53 %. Although most of the identified fungal cells are filamentous and may be considered hydrophobic, the degree may vary from mildly to highly hydrophobic (Beever and Dempsey, 1978). This could explain the observed variations and losses among taxa depending on the concentration protocol used. For example, Aspergillus fumigatus conidia are considered to be highly hydrophobic compared to other fungi such as Cladosporium (Kwon-Chung and Sugui, 2013). This can impact the efficiency of spore dispersion and, in this case, their behaviour in a liquid/surface environment. Wösten et al. (1999) described clearly how a fungus escapes the water to grow in the air and how this behaviour is linked to the hydrophobicity of fungal cells. Hydrophobins have been identified in most, if not all, filamentous ascomycetes and basidiomycetes. Yet, they are absent in yeasts and in other phyla of the fungal kingdom. It was shown that in the dimorphic fungus Ustilago maydis surface hydrophobicity is not caused by hydrophobins but by hydrophobin-like proteins called repellents (Teertstra et al., 2006). This may also be the case for other yeasts or non-ascomycete or non-basidiomycete species. Spores with a wettable surface could be lost during the centrifugation procedure due to lipid droplets that lower the density of the cell. Reversely, some fungi may be preferentially centrifuged compared to other due to their wettable nature or the absence of lipid droplets. Hypothesis of the fungal cell loss due to hydrophobicity caused by hydrophobin and/or oil droplets was supported by previous studies reporting the presence of hydrophobins or oil

122

droplets in the seven fungal genera identified in this study as the ones that are the most affected by centrifugation.

The in vitro laboratory experiment confirmed the loss of fungal spores observed during centrifugation of environmental air samples. Another mechanism of spore sorption is via electrostatic interactions. Spores of A. niger have both a hydrophobic nature and a surface charge (Wagernau et al., 2011). Addition of salt reduces the long distance electrostatic interactions enabling the short distance hydrophobic interactions with the micro tube to occur. However, addition of salt did not increase the incidence of sorption to the wall, implying that the long range electrostatic interactions are low compared to the short range of hydrophobic interactions. Presence of melanin also did not increase spore sorption. In fact, more spores adhered to the wall of the tube in a strain not producing melanin, likely explained by the fact that the ΔpptA spores are still hydrophobic (our unpublished results). Melanin contributes to spore surface characteristics in some fungal species like A. niger. Thus, it is believed that melanin could play a role in the initial aggregation/adhesion of spores to pellet.

Applying the standard centrifugation protocol when studying fungal bioaerosols may lead to an underestimation of some species including the taxa identified in this research. In certain studies fungal load has been underestimated by using the centrifugation method (Méheust et al., 2013; Bellanger et al., 2012; Bellanger et al., 2009). Studies using settled dust sampling (via EDC) that also require handling samples in liquid buffers and centrifugation may also be affected by the fungal loss. In addition to being underestimated in diversity and exposure studies, these fungi may represent greater health risks because of the characteristics associated with their hydrophobicity. In this context, a study showed that a hydrophobin layer may affect immune recognition of fungi (Aimanianda et al., 2009). Of the nine taxa not detected by the centrifugation protocol some are quite common including Fusarium, Cryptococcus and Alternaria. Fusarium species are responsible for a broad range of health problems, from local and systemic infections to allergy related diseases, in immunodepressed individuals (Nucci and Anaissie, 2007). Lipid droplets seems to have a

123

particular role in Fusarium species virulence (Zhang et al., 2017; Nguyen et al., 2011). Also, lipid droplets in Cryptococcus affect the host-pathogen interaction (Nolan et al., 2017).

Further research can help us better understand fungal aerosol behaviour. Future in vitro studies could be extended to include all the taxa identified in this research to determine if in vitro samples constantly yield the same results as environmental samples. Additionally, during the centrifugation protocol, the pellet and supernatant could be used for lipid droplets identification. It would confirm the hypothesis of fungal taxa loss during centrifugation due to lipid droplet production.

3.8 Conclusion

Although this paper seems to be a collection of three different studies with many variables, the storyline and the controls used tells the story of a particular fungal cell behaviour when retrieved from air and concentrated using liquid medium. Here, the three environments studied represent biological replicates of the observed tendency of the fungal cell behaviour. Even more striking, the same conclusions were drawn when the experiment was repeated in a laboratory setting using strains cultured in a media instead of air samples collected from the field. The data presented in this research provide a unique framework for understanding the role of the concentration protocol prior to DNA extraction on fungal composition obtained from bioaerosol samples. We believe that this work identified the issue of fungal cell loss when recovered from air samples and propose an alternative method to better evaluate fungal exposure and diversity. Based on the results of this investigation, the newly developed filtration protocol should be used to achieve the highest possible fungal cell recovery from air samples.

Acknowledgements

HM is a recipient of the FRQNT PhD scholarship and received a short internship scholarship from the Quebec Respiratory Health Network. This work was supported by the

124

Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail (2014-0057). Sampling of composting plants and Biomethanization facilities was performed as part of two distinct projects (IRSST 2012-0029 and 2013-0013). Swine building sampling was performed as part of the AgriVita-IRSST project (Agriculture and AgriFood Canada and IRSST 2014-0058). Penicillium roquefortii strain FM164 was kindly provided by Dr. Joëlle Dupont, MHNM, Sorbonne, Paris. The authors are thankful to Amanda Kate Toperoff and Michi Waygood for English revision of the manuscript. CD is the head of the Bioaerosols and respiratory viruses strategic group of the Quebec Respiratory Health Network.

Authors contribution

H.M. helped designed the study, performed all experiments except for the experiments with spores isolated from lab strains, including field sampling, analyzed the data and wrote the paper; M.V. performed field sampling and designed the study; W.T designed and performed the experiments with the spores of the lab strains; H.W., W.R, and W.K contributed to manuscript revision and made suggestions to improve the discussion and the scope of the paper overall; G.J.B. and C.D. designed the study; and all authors edited the manuscript.

Competing Interests

The authors declare no competing financial interests.

Materials & Correspondence

All material requests and correspondence should be addressed to Caroline Duchaine. Supplementary Information

Supplementary information is available at Applied and Environmental Microbiology website

125

3.9 Bibliography

Aimanianda, V., Bayry, J., Bozza, S., Kniemeyer, O., Perruccio, K., Elluru, S.R., Clavaud, C., et al., 2009. Surface hydrophobin prevents immune recognition of airborne fungal spores. Nature. 460(7259):1117–1121.

Beever, R.E., Dempsey, G.P., 1978. Function of rodlets on the surface of fungal spores. Nature. 272(5654):608-10.

Bellanger, A.P., Reboux, G., Murat, J.B., Bex, V., Million, L., 2010. Detection of Aspergillus fumigatus by quantitative polymerase chain reaction in air samples impacted on low-melt agar. Am. J. Infect. Control. 38(3):195–198.

Bellanger, A.P., Reboux, G., Roussel, S., Grenouillet, F., Scherer, E.D., Dalphin, J.C., Million, L., 2009. Indoor fungal contamination of moisture-damaged and allergic patient housing analysed using real-time PCR. Lett. Appl. Microbiol. 49(2):260-260.

Bellanger, A.P., Reboux, G., Scherer, E., Vacheyrou, M., Millon, L., 2012. Contribution of a cyclonic-based liquid air collector for detecting Aspergillus Fumigatus by QPCR in air Samples. J. Occup. Environ. Hyg. 9(1):D7-D11.

Bickel, P.E., Tansey, J.T., Welte, M.A., 2009. PAT proteins, an ancient family of lipid droplet proteins that regulate cellular lipid stores. Biochim. Biophys. Acta. 1791(6):419–440.

Bonifait, L., Marchand, G., Veillette, M., M’Bareche, H., Dubuis, M.E., Pépin, C., et al., 2017. Workers’exposure to bioaerosols from three different types of composting facilities. J Occup. Environ. Hyg. 14(10):815-822.

Bos, C.J., Debets, A.J., Swart, K., Huybers, A., Kobus, G., Slakhorst, S.M., 1988. Genetic analysis and the construction of master strains for assignment of genes to six linkage groups in Aspergillus niger. Curr. Genet. 14(5):437-443.

Braun, U., Schubert, K., 2007. Taxonomic revision of the genus Cladosporium s. lat. 7. Descriptions of new species, a new combination and further new data. Schlechtendalia. 16: 61–76.

Brodie, E.L., DeSantis, T.Z., Parker, J.P.M., Zubietta, I.X., Piceno, Y.M., Andersen, G.L., 2007. Urban aerosols harbor diverse and dynamic bacterial populations. Proc. Natl. Acad.

126

Sci. 104, 299–304.

Brown, J.K.M., Hovmøller, M.S., 2002. Aerial dispersal of pathogens on the global and continental scales and its impact on plant disease. Science. 297(5581):537–541.

Bush, R.K., 2008. Indoor allergens, environmental avoidance, and allergic respiratory disease. Allergy Asthma Proc. 29(6):575–579.

Bush, R.K., Portnoy, J.M., Saxon, A., Terr, A.I., Wood, R.A., 2006. The medical effects of mold exposure. J. Allergy Clin. Immunol. 117(2):326-333.

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., et al., . 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods. 7(5): 335-336.

Cayer, M.P., Veillette, M., Pageau, P., Hamelin, R., Bergeron, M.J., Mériaux, A., Cormier, Y., Duchaine, C., 2007. Identification of mycobacteria in peat moss processing plants: application of molecular biology approaches. Can. J. Microbiol. 53(1):92-99.

Chowdhary, A., Agarwal, K., Kathuria, S., Gaur, S.N., Randhawa, H.S., Meis, J.F., 2014. Allergic bronchopulmonary mycosis due to fungi other than Aspergillus: a global overview. Crit. Rev. Microbiol. 40(1):30-48.

Cretu, R., Dima, S., Bahrim, G., 2011. Antioxidant activity evaluation of the fungal dye oil- in-water emulsions obtained by cyclic voltammetric technique. Rom. Biotechnol. Lett. 16(6)19-33.

Cullison, M.A., Jaykus, L., 2002. Magnetized carbonyl iron and insoluble zirconium hydroxide mixture facilitates bacterial concentration and separation from non-fat dry milk. J. Food Prot. 65(11):1806-1810. de Vries, R.P., Burgers, K., Van De Vondervoort, P.J.I., Frisvad, J.C., Samson, R.A.,Visser, J., 2004. A new black aspergillus species, A. vadensis, is a promising host for homologous and heterologous protein production. Appl. Environ. Microbiol. 70(7)3954-3959.

Douwes, J., Thorne, P., Pearce, N., Heederik, D., 2003. Bioaerosol health effects and exposure assessment: progress and prospects. Ann. Occup. Hyg. 47(3):187–200.

Dubuis, M.E., M’Bareche, H., Veillette, M., Bakhiyi, B., Zayed, J., Lavoie, J., et al., 2017. Bioaerosols concentrations in working areas in biomethanization facilities. J. Air Waste Manag. Assoc. 67(11):1258-1271.

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 26(19):2460-2461.

Edgar, R.C., 2013. UPARSE: Highly accurate OTU sequences from microbial amplicon reads. Nat. Methods. 10:996-998.

127

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 27(16):2194-2200.

Feng, J., Liu, G., Selvaraj, G., Hughes, G.R., Wei. Y., 2005. A secreted lipase encoded by LIP1 is necessary for efficient use of saturated triglyceride lipids in Fusarium graminearum. Microbiology. 151(pt 12):3911–3921.

Fogelmark, B., Sjostrand, M., Rylander, R., 1994. Pulmonary inflammation induced by repeated inhalations of (1,3)-D-glucan and endotoxin. Int. J. Exp. Path. 75(2):85-90.

Fuchs, U., Czymmek, K.J., Sweigard, J.A., 2004. Five hydrophobins genes in Fusarium verticillioides include two required for microconidial chain formation. Fungal Genet Biol. 41(9):852-864.

Gebbink, M.F.B.G., Claessen, D., Bouma, B., Dijkhuizen, L., Wösten, H.A.B., 2005. Amyloids – a functional coat for microorganisms. Nat. Rev. Microbiol. 3(4):333-334.

Glass, D., Amedee, R.G., 2011. Allergic fungal rhinosinusitis: a Review. Ochsner J. 2011. 1(3):271–275.

Green, B.J., Lemons, A.R., Park, Y., Cox-Ganger, M., Park, .JH., 2017. Assessment of fungal diversity in a water-damaged office building. J. Occup. Environ. Hyg. 14(4):285-293.

Gregory, P.H., 1957. Electrostatic charges on spores of fungi in air. Nature. 180:330

Hardin, B.D., Kelman, B.J., Saxon, A. 2003. Adverse human health effects associated with molds in the indoor environment. J. Occup. Environ. Med. 45(5):470–478.

Heederik, D., von Mutius, E., .2012. Does diversity of environmental microbial exposure matter for the occurrence of allergy and asthma. J. Allergy Clin. Immunol. 130(1):44–50.

Jacobsen, C., Rasmussen, O., 1992. Development and application of a new method to extract bacterial DNA from soil based on a separation of bacteria from soil with cation-exchange resin. Appl. Environ. Microbiol. 58(8):2458-2462.

Jørgensen, T.R., Park, J., Arentshorst, M., van Welzen, A.M., Lamers, G., Vankuyk, P.A., et al., 2001. The molecular and genetic basis of conidial pigmentation in Aspergillus niger. Fungal Genet. Biol. 48(5):544–553.

Just, N., Kirychuk, S., Gilbert, Y., Létourneau, V., Veillette, M., Singh, B., et al., 2011. Bacterial diversity characterization of bioaerosols from cage-housed and floor-house poultry operations. Environ. Res. 111(4):492–498

128

Kobae, Y., Gutjahr, C., Paszkowski, U., Kojima, T., Fujiwara, T., Hata, S., 2014. Lipid Droplets of arbuscular mycorrhizal fungi emerge in concert with arbuscule collapse. Plant Cell Physiol. 55(11):1945–1953.

Kõljalg U, Nilsson R H, Abarenkov K, Tedersoo L, Taylor AFS, Bahram M, et al. 2013.Towards a unified paradigm for sequence-based identification of fungi. Mol. Ecol. 22(21): 5271–5277.

Kumari, P., Woo, C., Yamamoto, N., Choi, H.L., 2016. Variations in abundance, diversity and community composition of airborne fungi in swine houses across seasons. Sci. Rep. 6:37929.

Kwasna, H., Kosiak, B., 2003. Lewia avenicola sp. nov. and its Alternaria anamorph from oat grain, with a key to the species of Lewia. Mycol. Res. 107(3):371-376.

Kwon-Chung, K.J., Sugui, J.A., 2013. Aspergillus fumigatus—What Makes the Species a Ubiquitous Human Fungal Pathogen? PLoS Pathog. 9(12):e1003743.

Lecours, P.B., Veillette, M., Marsolais, D., Duchaine, C., 2012. Characterization of Bioaerosols from Dairy Barns: Reconstructing the Puzzle of Occupational Respiratory Diseases by Using Molecular Approaches. Appl. Environ. Microbiol. 78(9):3242-3248.

Linder, M.B., Szilvay, G.R., Nakari-Setälä, T., Penttilä, M.E., 2005. Hydrophobins: the protein-amphiphiles of filamentous fungi. FEMS Microbiol Rev. 29(5): 877–896.

Love, M., Huber, W., Anders, S., Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15(12):550.

Lucore, L.A., Cullison, M.A., Jaykus, L.A., 2000. Immobilization with metal hydroxides as a means to concentrate foodborne bacteria for detection by cultural and molecular methods. Appl. Environ. Microbiol.66(5):1769-1776.

Madsen, A.M., Zervas, A., Tendal, K., Nielsen, J.L., 2015. Microbial diversity in bioaerosol samples causing ODTS compared to reference bioaerosol samples measured using Illumina sequencing and MALDI-TOF. Environ. Res. 140:255–267. Mbareche, H., Veillette, M., Bonifait, L., Dubuis, M.E., Bernard, Y., Marchand, G., et al., 2017. A next generation sequencing approach with a suitable bioinformatics workflow to study fungal diversity in bioaerosols released from two different types of compost plants. Sci. Total Environ. 1(601-602):1306-1314.

Méheust, D., Gangneux, J.P., Reponen, T., Wymer, L., Vesper, S., LeCann, P., 2012. Correlation between Environmental Moldiness Index (ERMI) values in French dwellings and other measures of fungal contamination. Sci. Total Environ. 438:319-24.

129

Méheust, D., LeCann, P., Reboux, G., Million, L., Gangneux, J.P., 2013. Indoor fungal contamination: health risks and measurement methods in hospitals, homes and workplaces. Crit. Rev. Microbiol. 40(3):248-260.

Mistry, J., Finn, R.D., Eddy, S.R., Bateman, A., Punta, M., 2013. Challenges in Homology Search: HMMER3 and Convergent Evolution of Coiled-Coil Regions. Nucleic Acids Res. 41:e121.

Murphy, D.J., 2001. The biogenesis and functions of lipid bodies in animals, plants and microorganisms. Prog. Lipid Res. 40(5):325–438.

Nehmé, B., Gilbert, Y., Létourneau, V., Forster, R.J., Veillette M, Villemur R, Duchaine C. 2009. Culture-Independent Characterization of Archaeal Biodiversity in Swine Confinement Building Bioaerosols. Appl. Environ. Microbiol. 75(17):5445-5450.

Nehme, B., Létourneau, V., Forster, R.J., Veillette, M., Duchaine, C., 2008. Culture- independent approach of the bacterial bioaerosol diversity in the standard swine confinement buildings, and assessment of the seasonal effect. Environ. Microbiol. 10(3):665-75.

Nguyen, L.N., Bormann, J., Le, G.T., Stärkel, C., Olsson, S., Nosanchuk, J.D., et al., 2011. Autophagy-related lipase FgATG15 of Fusarium graminearum is important for lipid turnover and plant infection. Fungal Genet Biol. 48(3):217–224.

Nolan, S.J., Fu, M.S., Coppens, I., Casadevall, A., 2017. Lipids Affect the Cryptococcus neoformans-Macrophage Interaction and Promote Nonlytic Exocytosis. Infect. Immun. 85(12):e00564-17.

Nucci, M., Anaissie E., 2007. Fusarium Infections in Immunocomprised Patients. Clin. Microbiol. Rev. 20(4):695-704.. O’brien, K.M., Chimenti, M.S., Farnell, M., Tabler, T., Biar, T., Bray, J.L., Nonnenmann, M.W., 2016. High throughput genomic sequencing of bioaerosols in broiler chicken production facilities. Microb. Biotechnol. 9(6):789-791.

O’Connor, J.D., Daly, S.M., Sodeau, J.R., 2015. On-line monitoring or airborne bioaerosols released from a composting/green waste site. Waste manag. 42:23-30.

Oppliger, A., Duquenne, P., 2016. Section II : Occupational Settings. Environmental Mycology in Public Health: Fungi and Mycotoxins Risk. Ed. Carla Viegas, Ana Catarina Pinheiro, Raquel Sabino, Susana Viegas, Joao Brandao, Cristina Verissimo. Elsevier Inc. Netherlands.

Park, D., Ryu, S., Kim, S., Byun, H., Yoon, C., Lee, K., 2013. Airborne bacteria and fungi associated with waste-handling work. Int. J. Occup. Environ. Health. 19(4):313-317.

Peccia, J., Hernandez, M., 2006. Incorporating polymerase chain reaction-based identification, population characterization, and quantification of microorganisms into aerosol science: A review. Atmos. Environ. 40(21):3941–396.

130

Pieckova, E., Wilkins, K., 2004. Airway toxicity of house dust and its fungal composition. Ann. Agric. Environ. Med. 11(1):67–73.

Porter, P., Susarla, S., Polikepahad, S., Qian, Y., Hampton, J., Kiss, A., et al., 2009. Link between allergic asthma and airway mucosal infection suggested by proteinase-secreting household fungi. Mucosal Immunol. 2(6):504–517.

Qian, J., Hospodsky, D., Yamamoto, N., Nazaroff, W.W., Peccia, J., 2012. Size-resolved emission rates of airborne bacteria and fungi in an occupied classroom. Indoor Air. 22(4):339–351.

Radulovic, M., Knittelfelder, O., Cristobal-Sarramian, A., Kolb, D., Wolinski, H., Kohlwein, S.D., 2013. The emergence of lipid droplets in yeast: current status and experimental approaches. Curr. Genet. 59(4): 231–242.

Rittenour, R.W., Park, J.H., Cox-Ganser, J.M., Beezhold, D.H., Green, B.J., 2012. Comparison of DNA extraction methodologies used for assessing fungal diversity via ITS sequencing. J. Environ. Monit. 14(3):766-774.

Ropars, J., Dupont, J., Fontanillas, E., Rodríguez de la Vega, R.C., Malagnac, F., et al., 2011. Sex in Cheese: evidence for sexuality in the fungus Penicillium roqueforti. PLoS ONE. 7(11):e49665.

Rylander, R., 1996. Airway Responsiveness and Chest Symptoms after Inhalation of Endotoxin or (1→3)-β-D-Glucan. Indoor Built Environ. 5:106-111.

Sanchez-Monedero, M.A., Stentiford, E.I., 2003. Generation and dispersion of airborne microorganisms from composting facilities. Trans IChem. 81(3):166-170. Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., et al., 2009. Introducing mother: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75(23):7537-41.

Schlosser, O., Huyard, A., Rybacki, D., Do Quang, Z., 2012. Protection of the vehicle cab environement against bacteria, fungi and endotoxins in composting facilities. Waste manag. 32(6):1106-1115.

Schoch, C.L., Seifert, K.A., Huhndorf, S., Robert, V., Spouge, J.L., Levesque, A.C., et al., 2012. Nuclear ribosomal internal transcribed spacer (ITS) region as a universal DNA barcode marker for Fungi. Proc. Nat. Acad. Sci. 109(16):6241-6246.

Selman, M., Lacasse, Y., Pardo, A., Cormier, Y., 2010. Hypersensitivity pneumonitis caused by fungi. Proc. Am. Thorac. Soc. 7(3):229–236

Sterflinger, K., De Baere, R., de Hoog, G., De Wachter, R., Krumbein, W.E., Haase. G., 1992. Coniosporium perforans and C. apollinis, two new rock-inhabiting fungi isolated

131

from marble in the Sanctuary of Delos (Cyclades, Greece). Antonie Van Leeuwenhoek; 72(4):349-363.

Taha, M.P.M., Drew, G.H., Longhurst, P.J., Smith, R., Pollard, S.J.T., 2003. Bioaerosol releases from compost facilities: evaluating passive and active source terms at a green waste facility for improved risk assessments. Atmos. Environ. 40(6):1159–1169.

Tedersoo, L., Anslan, S., Bahram, M., Põlme, S., Riit, T., Liiv, I., et al., 2015. Shotgun metagenomes and multiple primer pair barcode combinations of amplicons reveal biases in metabarcoding analyses of fungi. MycoKeys. 10:1-43.

Teertstra, W., Deelstra, H., Vranes, M., Bohlmann, R., Kahmann, R., Kämper, J., Wösten, H.A.B., 2006. Repelents have functionally replaced hydrophobins in mediating attachement to a hydrophobic surface and in formation of hydrophobic aerial hyphae in Ustilago maydis. Microbiology. 152(12):3607-3612.

Terhem, R.B., Van Kan, J.A.L., 2014. Functional analysis of hydrophobin genes in sexual development of Botrytis cinerea. Fungal Genet. Biol. 71:42-51.

Tischer, C.G., Heinrich, J., 2013. Exposure assessment of residential mould, Fungi and microbial components in relation to children’s health: achievements and challenges. Int. J. Hyg. Environ. Health. 216(2):109-114.

Vaccari, D.A., Strom, P.F., Alleman, J.E., 2006. Environmental biology for engineers and scientists. Chapter 11: Quantifying Microorganisms and their Activity. Section 11.3: Sampling, Storage and Preparation, Pages 299-300. Wiley-Interscience. New York, NY, USA. Veillette, M., Thorne, P.S., Gordon, T., Duchaine, C., 2004. Six Month Tracking of Microbial Growth in a Metalworking Fluid After System Cleaning and Recharging. Ann. Occup. Hyg. 48(6):541-546.

Vestlund, A.T., Al-Ashaab, R., Tyrrel, S.F., Longhurst, P.J., Pollard, S.J.T., Drew, G.H.. 2014. Morphological classification of bioaerosols from composting using scanning electron microscopy. Waste Manag. 34:1101–1108.

Walther, T.C., Farese, R.V. Jr., 2012. Lipid droplets and cellular lipid metabolism. Annu. Rev. Biochem. 81:687–714.

Wargenau, A., Fleißner, A., Bolten, C.J., Rohde, M., Kampen, I., Kwade, A., 2011. On the origin of the electrostatic surface potential of Aspergillus niger spores in acidic environments. Res. Microbiol. 162(10):1011-1017.

Welte, M.A., 2015. Expanding roles for lipid droplets. Curr. Biol. 25(11):R470–R481.

Wösten, H.A.B., 2001. Hydrophobins: Multipurpose Proteins. Annu. Rev. Microbiol. 55:625-646.

132

Wösten, H.A.B., Schuren, F.H., Wessels, J.G., 1994. Interfacial self-assembly of a hydrophobin into an amphipathic protein membrane mediates fungal attachment to hydrophobic surfaces. EMBO J. 13(24):5848–5854.

Wösten, H.A.B., Van Wetter, M.A., Lugones, L.G., Van Der Mei, H.C., Busscher, H.J., et al., 1999. How a fungus escape water to grow into the air. Curr. Biol. 9(2):85-88.

Yang, H.J., Osakada, H., Kojidani, T., Haraguchi, T., Hiraoka, Y., 2017. Lipid droplet dynamics during Schizosaccharomyces pombe sporulation and their role in spore survival. Biol. Open. 6:217-222.

Zekovic, D.B., Kwiatkowski, S., Vrvic, M.M., Jakovljevic, D., Moran, C.A., 2005. Natural and Modified (1→3)-β-D-Glucans in Health Promotion and Disease Alleviation. Crit. Rev. Biotechnol. 25(4):205-230.

Zhang, J., Fang, X., Zhu, X.L., Li, Y., Xu, H.P., Zhao, B.F., et al., 2011. Microbial lipid production by the oleaginous yeast Cryptococcus curvatus O3 grown in fed-batch culture. Biomass and Bioenergy. 35(5):1906-1911.

Zhang, Y., Gao, T., Shao, W., Zheng, Z., Zhou, M., Chen, C., 2017. The septins FaCdc3 and FaCdc12 are required for cytokinesis and affect asexual and sexual development, lipid metabolism and virulence in Fusarium asiaticum. Mol. Plant Pathol. 18(9):1282–1294.

133

Part Two: Evaluation of Workers Exposure to Fungi in Three Environments Affected by Fungal Aerosol problems

134

Chapter 4: A Next Generation Sequencing Approach with a Suitable Bioinformatics Workflow to Study Fungal Diversity in Bioaerosols Released from Two Different Types of Composting plants

4.1 Résumé

Le compostage est utilisé partout dans le monde pour transformer différents types de matière organique par l’action de communautés microbiennes complexes. Des niveaux d'exposition élevés sont associés à des effets néfastes sur la santé des travailleurs de l'industrie du compost. Il existe peu de données sur la diversité fongique détaillée dans les aérosols émis lors des activités de compostage. Le but de cette étude est d'analyser la diversité fongique du compost et des aérosols émis dans les deux types de compostage. Cela a été accompli en utilisant une approche de séquençage haut débit qui cible la région génomique ITS1. Les résultats montrent que le type de compost affecte la diversité fongique des aérosols émis. Un profil de diversité important a été observé dans les bioaérosols des deux types de compost, montrant la présence d'un certain nombre de pathogènes nouvellement identifiés dans les bioaérosols émis par les usines de compostage.

135

4.2 Summary of the Paper

This paper present a description of the fungal diversity released from two different composting environments using HTS methods with a personalized bioinformatic workflow for the diversity analyses. The project came from the desire to have a better understanding of the type of fungal exposure workers are facing during the different steps of composting activities generating bioaerosols. Indeed, fungi are of a particular interest in composting environments because of their crucial role in degrading organic matter. We visited two composting sites, one using domestic waste, and the other was for swine carcasses and placenta. Samples were collected from air and composting piles to study the link between the source and bioaerosols. The authors used an HTS approach targeting ITS1 to describe fungal exposure qualitatively. The bioinformatic workflow was personalized for fungal analyses, and for its friendly-user approach.

The results obtained showed that the fungal composition was dependent on the type of waste treated. The differences between the two types of compost were statistically significant, and were driven by the class of Eurotiomycetes, which were more abundant in carcass and placenta composting, while Sordariomycetes were dominant in domestic composting. Very diverse fungal profiles were obtained for the two types of composting environments, in the compost pile and air samples. Although, the taxonomic profiles showed a clear link between the source and bioaerosols, some taxa of interest were detected in higher proportions in air samples compared to the composting piles. This suggest a higher risk of inhalation by daily exposed workers by these particular fungi.

We are aware that changing the infrastructure of a work environment to reduce bioaerosol emission is not always easy, especially in composting sites where microbes are the main actors of the activity. But, we strongly recommend the use of protective equipment by the workers to prevent adverse health effect due to a continuous exposure to potentially harmful fungi.

136

A Next Generation Sequencing Approach with a Suitable Bioinformatics Workflow to Study Fungal Diversity in Bioaerosols Released from Two Different Types of Composting plants RUNNING TITLE Exposure to Fungal Bioaerosols in Composting Plants Using NGS

AUTHORS Hamza Mbareche1,5, Marc Veillette1, Laetitia Bonifait1, Marie-Eve Dubuis1,5, Yves Benard3, Geneviève Marchand2, Guillaume J. Bilodeau4 and Caroline Duchaine1,5*

AUTHORS’ AFFILIATION 1. Centre de recherche de l’institut universitaire de cardiologie et de pneumologie de Québec, Quebec City (Qc), Canada 2. Institut de Recherche Robert-Sauvé en Santé et en Sécurité du travail (IRSST), Montreal (Qc), Canada 3. Centre de Recherche Industrielle du Québec (CRIQ), Quebec City (Qc), Canada 4. Pathogen Identification Research Lab, Canadian Food Inspection Agency (CFIA). Ottawa, Canada 5. Département de biochimie, de microbiologie et de bio-informatique, Faculté des sciences et de génie, Laval university, Quebec City (Qc), Canada

KEYWORDS Fungal exposure, bioaerosols, composting plants, next generation sequencing

PUBLISHED : Sci Total Environ. 2017 Dec 1;601-602:1306-1314

137

4.3 Abstract

Composting is used all over the world to transform different types of organic matter through the actions of complex microbial communities. Moving and handling composting material may lead to the emission of high concentrations of bioaerosols. High exposure levels are associated with adverse health effects among compost industry workers. Fungal spores are suspected to play a role in many respiratory illnesses. There is a paucity of information related to the detailed fungal diversity in compost as well as in the aerosols emitted through composting activities. The aim of this study was to analyze the fungal diversity of both organic matter and aerosols present in facilities that process domestic compost and facilities that process pig carcasses. This was accomplished using a next generation sequencing approach that targets the ITS1 genomic region. Multivariate analyses revealed differences in the fungal community present in samples coming from compost treating both raw materials. Furthermore, results show that the compost type affects the fungal diversity of aerosols emitted. Although 8 classes were evenly distributed in all samples, Eurotiomycetes were more dominant in carcass compost while Sordariomycetes were dominant in domestic compost. A large diversity profile was observed in bioaerosols from both compost types showing the presence of a number of pathogenic fungi newly identified in bioaerosols emitted from composting plants. Members of the family Herpotrichiellaceae and Gymnoascaceae which have been shown to cause human diseases were detected in compost and air samples. Moreover, some fungi were identified in higher proportion in air compared to compost. This is the first study to identify a high level of fungal diversity in bioaerosols present in composting plants suggesting a potential exposure risk for workers. This study suggests the need for creating guidelines that address human exposure to bioaerosols. The implementation of technical and organizational measure should be a top priority. However, skin and respiratory protection for compost workers could be used to reduce the exposure as a second resort.

138

4.4 Introduction

Composting is one of the most promising techniques for managing organic residues in future years (Sykes et al., 2007). The final product can be used as soil amendment in agricultural practices and the quality of that product is linked to the quality of the compost after the maturation stages (Fuchs and Bieri, 2000). The degradation of organic matter requires the presence of a complex microbial community under aerobic conditions (Ryckeboer et al., 2003). More specifically, filamentous fungi play a major role in composting by releasing enzymes that break down complex molecules present in organic matter that are not easily degraded by other microorganisms (Anastasi et al., 2005; Hoorman, 2011; Floudas et al., 2012). Previous studies have revealed that the fungal diversity present during composting is influenced by changes in physico-chemical conditions like temperature and moisture content during different stages of compost maturation (Hansgate et al., 2005; Langarica-Fuentes et al., 2014; De Gannes et al., 2013).

Bioaerosols are released during composting activities. The type of organic matter being composted influences the microbial content of the compost and, consequently, of the emitted bioaerosols. No matter the composting procedure used, reactive or non-reactive, the compost is subject to various actions including delivery, shredding, pile turning and compost screening. High concentrations of bioaerosols are released during those activities suggesting potential health impacts related to exposure (Epstein et al., 2001; Sánchez-Monedero et al., 2005; Persoons et al., 2010; Taha et al., 2005).

Occupational exposure to bioaerosols is associated with a wide range of health effects including infectious diseases, toxic effects, allergies and cancer (Douwes et al., 2003). Respiratory symptoms are among the most prevalent bioaerosol-associated health effects (Domingo and Nadal, 2009).

Heavy exposure to bioaerosols has led to a series of diseases, symptoms and complaints among compost workers. Health problems include tracheobronchitis, mucosal

139

irritations, sinusitis, eczema, dermatomycosis and gastro-intestinal problems (Bünger et al., 2000). Compared to other industrial activities, composting sites represent a greater risk to workers due to the presence of higher concentrations of actinomycetes and thermophilic and/or thermo-tolerant fungi (Bünger et al., 2007). Furthermore, fungi, their secondary metabolites, and other components are believed to be associated with asthma, allergic alveolitis and chronic bronchitis (Swan et al., 2003; Fung and Clark, 2004). Van Kampen et al. affirmed the need for adequate monitoring of exposure in order to reduce health risks (Van Kampen et al., 2014).

The fungal diversity of bioaerosols emitted from composting plants has not been thoroughly investigated, and no previous studies have compared the bioaerosol emissions from different compost types. Most studies have used culture methods or the detection of specific fungal species (Sanchez-Monedero and Stentiford, 2003; Taha et al., 2006; Park et al., 2013; Schlosser et al., 2012). Other studies have focused mainly on monitoring particles or microscopic spore counts in aerosols (O'Connor et al., 2015; Vestlund et al., 2014). It is well known that only 1% of fungi recovered from air samples are cultivable therefore using this technique may lead to an underestimation of the real fungal diversity in bioaerosols (Peccia and Hernandez, 2006). More recent studies using molecular tools to study bioaerosols in composting plants focused on the 16S RNA region for bacteria and the 18S RNA for fungi. The results showed limited fungal diversity compared to bacterial diversity. However, the results from sequencing methods showed a higher number of fungal species compared to the results from culture methods (Le Goff et al., 2010; Bru-Adan et al., 2009). Moreover, molecular methods have been used to target specific pathogens or to identify specific biomarkers for the study of bioaerosol dispersion from composting sites. (Le Goff et al., 2011). To the best of our knowledge, the fungal diversity of bioaerosols released from composting plants has never been fully examined.

The aim of this study was to analyze the fungal diversity in composted organic matter and of the aerosols produced from two different compost activities (domestic composting bins and pig carcasses) using a next generation sequencing approach targeting the ITS1 genomic region.

140

4.5 Materials and Methods

Field study

Site sampling

The two composting plants were located in the province of Quebec (eastern Canada). Both were located in industrial areas and on agricultural/ farm lands away from urban areas. The type of organic matter treated at the two plants was different. One plant treated domestic compost which we considered to be green waste and the other treated pig carcasses with placenta, called animal compost. Composting piles were followed over one year. The domestic composting process lasted 6 weeks and the animal composting took 24 weeks. The two main steps of the domestic composting process were sorting and screening, whereas filling and brewing were used to process animal compost. The domestic composting plant was visited two times in the spring and two times in the summer while the animal composting plant was visited once in the summer and once during the fall. Outdoor temperature varied from 7 °C to 10 °C during spring visits, 19 °C to 28 °C during summer and 1 °C to 5 °C during the fall visit.

Air sampling

Air samples were taken during composting activities where workers were performing the tasks consisting of sorting, screening, filling and brewing the compost piles. Air samples were collected using the liquid cyclonic impactor Coriolis µ® (Bertin Technologies, Montigny-le- Bretonneux, France) set at 300 l/min for 10 min and placed in the center of the handling operations. Fifteen milliliters of sterile 50 mM phosphate-buffered saline (pH 7.4) were used to fill the sampling cone of the Coriolis. At each visit, 3 sampleswere taken in the beginning, the middle and the end of the working shifts. Additionally, 2 control samples were taken: one prior to worker's activity to set the background and one off-site upwind. Plus, field blank samples are always a part of the protocols.

141

Compost sampling

During the composting process, both samples were collected concurrently with air samples. The same sampling protocol was applied to both types of composting plants. Three samples of both type of samples were collected from the compost piles or the excavator during brewing activities. Both samples collection time was distributed such that it coincides with the air sampling period. Each sample corresponds to 1 l of organic matter. After samples were homogenized through manual mixing, a 25 g subsample was taken and homogenized again with 200 ml of phosphate buffered saline containing 0,05% tween 20 using a sterile stomacher Filtra-Bag (Labplas, Quebec, Canada) in a stomacher Mix 1 (Aes Laboratoire, Bruz, France).

DNA extraction

A 1.5 ml aliquots from the Coriolis air samples were centrifuged for 10min at 14,000 ×g. The pellets were kept at−20 °C until the DNA extraction. A Mixer Mill MM301 (Retsch, Düsseldorf, Germany) at a frequency of 20 movements per seconds for 10 min was used for cell mechanical damage with glass beads. Afterwards, a MoBio PowerLyser® UltraClean® Microbial DNA kit (Carlsbad, CA, U.S.A) was used to extract the total genomic DNA from the samples following the manufacturer's instructions. After the DNA elution, it was stored at −20 °C.

Miseq sequencing

Amplification of the fungal ITS1 gene, equimolar pooling and sequencing was performed at the Plateforme d'analyses génomiques (IBIS, Université Laval, Quebec City, Canada). Briefly, amplification of the ITS1 regions was performed using the sequence specific regions described in Tedersoo et al. (2015) and references therein using a two-step dual-indexed PCR approach specifically designed for Illumina instruments. In a first step, the gene specific sequence is fused to the Illumina TruSeq sequencing primers and PCR was carried out in a total volume of 25 µL that contains 1× Q5 buffer (NEB), 0.25 µMof each

142

primer, 200 µM of each dNTPs, 1 U of Q5 High-Fidelity DNA polymerase (NEB) and 1 µL of template cDNA. The PCR started with an initial denaturation at 98 °C for 30 s followed by 35 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 10 s, extension at 72 °C for 30s and a final extension at 72 °C for 2 min. The PCR reaction was purified using the Axygen PCR cleanup kit (Axygen). Quality of the purified PCR product were checked on a 1% agarose gel. Fifty to 100-fold dilution of this purified product was used as a template for a second PCR step with the goal of adding barcodes (dual-indexed) and missing sequence required for Illumina sequencing. Cycling for the second PCR were identical to the first PCR but with 12 cycles. PCR reaction were purified as above, checked for quality on a DNA7500 Bioanlayzer chip (Agilent) and then quantified spectrophotometrically with the Nanodrop 1000 (Thermo Fisher Scientific). Barcoded Amplicons were pooled in equimolar concentration for sequencing on the illumina Miseq.

The following oligonucleotide sequences were used for amplification:

ITS1 region ITS1Fngs: ACACTCTTTCCCTACACGACGCTCTTCCGATCTGGTCATTTAGAGGAAGTAA ITS2: GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGCTGCGTTCTTCATCGATGC

generic forward second-PCR primer AATGATACGGCGACCACCGAGATCTACAC[index1]ACACTCTTTCCCTACACGAC and generic reverse second-PCR primer CAAGCAGAAGACGGCATAC GAGAT[index2]GTGACTGGAGTTCAGACGTGT.

Please note that primers used in this work contain Illumina specific sequences protected by intellectual property (Oligonucleotide sequences© 2007–2013 Illumina, Inc. All rights reserved. Derivative works created by Illumina customers are authorized for use with Illumina instruments and products only. All other uses are strictly prohibitEd.)

143

Processing and analyzing of sequencing data

Fig. 4.1 describes the bioinformatics workflow and shows which programs were used for each step of the data treatment. After demultiplexing the raw FASTQ files, the reads generated from the paired-end sequencing using Mothur v 1.35.1 were combined (Schloss et al., 2009). The quality filtering was also performed using Mothur discarding reads with ambiguous sequences. Reads shorter than 100 bp and longer than 450 bp were also discarded. Similar sequences were gathered together to reduce the computational burden and the number of copies of the same sequence was displayed. This dereplication step was performed using USEARCH (version 7.0.1090) (Edgar, 2010). The selected region of fungal origin was then extracted from the sequences with ITSx which uses HMMER3 (Mistry et al., 2013) to compare input sequences against a set of models built from a number of different ITS region sequences found in various organisms. Only the sequences belonging to fungi were kept for further analyses. Operational taxonomic units (OTUs) with 97% similarity cut-off were clustered using UPARSE (VERSION 7.1) (Edgar, 2013). UCHIME was used to identify and remove chimeric sequences (Edgar et al., 2011). QIIME (version 1.9.0) (Caporaso et al., 2010) was used to assign taxonomy to OTUs based on UNITE fungal ITS reference training data set for taxonomic assignment and to generate an OTU table. All of the visualization analyses were performed using QIIME scripts (version 1.9.0, Rarefaction curves, Principal coordinate analyses, Statistical analyses and Taxonomic analyses).

Figure 4.1: Overview of the bioinformatics pipeline for the processing of sequencing data before the visualization analyses.

144

4.6 Results

Fifty-four fastq files were generated and corresponded to the pair-end (forward and reverse) sequencing of 27 compost and air samples. The MiSeq sequencing led to the generation of 3,833,981 sequences. After the quality filtering, dereplication and chimera checking, 855,172 sequences were clustered into 5255 OTUs.

Table 4.1 is a summary of the number of OTUs in each type of composting (air and compost). Rarefaction curves were constructed using the number of observed OTUs in air and compost samples. As shown in Fig. 4.2, the maximum sequencing depth was approximately 40,000 sequences per sample. This number was chosen based on the lowest- depth sample parameter, which represents the lowest number of sequences in a sample. Samples with a lower number than this would've been excluded from analyses. The higher the number, the more accurate the results are. In the currently presented case, all the samples were included as they have > 40,000 counts per sample. The values shown in Fig. 4.2 were calculated as follows: ten values from 10 to 40,000 sequences per sample were randomly selected. For each of these values the corresponding number of OTUs observed was noted for all of the samples. Then, the average number of OTUs observed and the standard deviation were calculated for each one of the ten values. The plateau shown on the graph indicates effective sampling of the fungal diversity, as no more OTUs were observed even with greater numbers of sequences per sample.

Table 4.1: Number of OTUs in each type of composting (air and compost)

Compost type Sample type Average OTU number Standard deviation

Air 218 80 Domestic Compost 167 94 Air 195 44 Animal Compost 141 37

145

Figure 4.2: Rarefaction analysis of air and compost samples from the two types of composting plants. An average of the OTUs observed in each sample was calculated for all air and compost samples respectively with a standard deviation (1SD).

A multivariate analysis was conducted to determine the variation between samples according to the raw material used in composting. Using a Principal Coordinate Analysis (PCoA) samples from the two types of composting plants were compared based on a Bray- Curtis dissimilarity metric. A distance matrix was created containing the Bray- Curtis index to evaluate the distance, taken pairwise, between the samples. This calculation incorporates information about the abundance of the observed OTUs in each pair of samples. It is necessary to use a rarefied OTU table, in which samples have the same size, to calculate the Bray-Curtis index. Indeed, the Bray-Curtis dissimilarity is based on the absolute abundance of the different taxa present in a sample. This method is widely used in microbial ecology to study the abundance of the observed organisms and their variation fromone sample to another. As shown in Fig. 4.3, fungal communities varied according to the type of raw material composted. The PC1 axis of the PCoA plot explained 38% of the total variation and the PC2 axis explained 15% of the total variation between samples. These results suggest the possibility of a fungal signature specific to the type of waste treated. To determine the

146

statistical significance of the variation observed, a permanova analyses was performed on the Bray-Curtis distance matrix. This non-parametric multivariate analysis of variance method separates the distance matrix among sources of variation to describe the robustness and significance that a variable has in explaining the variations between samples. It is based on the ANOVA experimental design but analyses the variance and determine the significance by permutations as it is a nonparametric test (Anderson, 2005). The number of permutations used to calculate the statistical significance is 999. The compost type variable tested separates the samples in 2 groups (compost and animal) and the p-value obtained corresponds to 0.001.

Figure 4.3: Principal coordinate analysis of air and compost samples combined from the two types of composting facilities (domestic compost in blue and animal compost in red). The plot was made using the Bray-Curtis measures from the Bray-Curtis distance matrix that calculates the dissimilarity between each sample. The statistical analyses of the variance using permanova indicate that communities in each composting type are significantly different (p-value 0.001).

147

The taxa were examined in order to determine their relative abundance in compost and air samples from the two types of composting plants. Ascomycota and Basidiomycota were the two most abundant phyla while Chytridiomycota, Rozellomycota and Zygomycota were present at lower percentages. The distribution according to the common features forming the classes of fungi from air and compost samples show a dominance of 8 of the 11 classes that were most represented. The overall similarity of the diversity profiles between air and compost samples for both domestic and animal compost is an indication of the influence of the source on detected bioaerosols (Fig. 4.4). Although Eurotiomycetes, Saccharomycetes, Agaricomycetes, Sordariomycetes, Pezizomycetes, Tremellomycetes, Dothideomycetes and Microbotryomycetes were present throughout all samples, Eurotiomycetes was the most common fungal class in animal compost air samples while Sordariomycetes prevailed in domestic compost air samples. Sordariomycetes and Eurotiomycetes are both classes of fungi involved in the decomposition of organic matter. Composting animal carcasses is a recent practice used as an alternative to the expensive and nonenvironmental friendly incineration approach. Table 4.2 presents the taxonomic rankings of the phyla belonging to the class Eurotiomycetes found in animal compost.

148

Figure 4.4: Relative abundance of the classification of fungi (class and phylum) identified in compost and air samples from the two types of composting plants.

149

Table 4.2: The Eurotiomycetes found in animal composting (air and compost samples)

Phyla Taxonomic rank Penicillium genus Thermomyces genus Eurotiales order Talaromyces genus Trichocomaceae family Arachnomyces genus Emericella genus Cyphellophora genus Capronia genus Cladophialophora genus Exophiala genus Phaecoccomyces genus Rhinocladiella genus Herpotrichiellaceae family Coniosporium genus Chaetothyriales order Byssochlamys genus Paecilomyces genus Sagenomella genus Monascus genus Gymnoascaceae family

When carefully analyzing the relative abundance, a large diversity profile was observed below 1% of total relative abundance. The most abundant taxa comprising the fungal communities were examined more closely. Lists were created of the dominant genera identified in the samples (Fig. 4.5) and were compared. Samples from domestic and animal composting showed several similarities. However, differences were observed in the less abundant phyla from both types of composting, causing the observed variation between their respective fungal communities. Some taxa were specific to the type of waste that was treated. Those taxa included Trametes, Phelinus and Postia for domestic composting and Rasamsonia, Piptoporus, Neurospora, Talaromyces and Fusarium for animal composting (indicated by colored text in Fig. 4.5). These signature taxa are found among the top 20 most abundant fungi according to the type of waste treated as shown by the multivariate analyses (Fig. 4.3). Penicillium was the most abundant fungi in samples from both types of compost (20% to 30%). The rest of the list showed some variation as Candida was the second most abundant genera in domestic composting (>20% for air and <20% for compost) while

150

Blastobotrys was the second most abundant genera (20% for air and 10% for compost) present in animal composting. A globally identical diversity profile was observed in compost and air samples, for animal composting. The same is true for air and compost profiles for domestic composting. These results are supported by the permanova statistical analyses as described earlier in this manuscript. The variation between air and compost samples is not significant as the p-values are higher than 0.05 (p-value 0.06 for domestic compost and p- value 0.2 for animal compost). However, diversity profiles compared between the two composting sources were different. This supports the theory that the source influences the fungal diversity in bioaerosols. However, three exceptions were noted in the animal composting samples. Three genera from the list of abundant fungi were present in the air samples but were not present in the compost. These were Neurospora, Fusarium and Emericella. Two of these genera, Fusarium and Emericella, are associated with human diseases. Additionally, several of the fungal genera were present in higher percentages in air compared to compost and vice-versa. Fig. 4.6 and Fig. 4.7 illustrate the aerosolization behavior showing that the relative abundance of each genera is greater in air samples compared to compost samples in all cases, and in both domestic and animal composting. The more obvious examples from domestic composting included Acremonium, Sistotremastrum, Malassezia, Leucosporidium, Capnobotryella and Davidiella. The more obvious examples from animal composting included Blastobotrys, Malassezia, Leucosporidium, and Talaromyces.

151

Figure 4.5: the most abundant fungal genera identified in domestic and animal composting facilities (air and compost). Fungi that were identified in only one type of composting facility are written in colored text. For animal composting, genera found only in air are written in larger font and those only in compost are boxed.

152

Figure 4.6: Aerosolization behavior of some of the most abundant fungi genera identified in domestic composting facilities. The y-axis represents the percentage of relative abundance compared to the total most abundant fungi genera.

Figure 4.7: Aerosolization behavior of some of the most abundant fungi genera identified in animal composting facilities. The y-axis represents the percentage of relative abundance compared to the total most abundant fungi genera.

153

4.7 Discussion

Using a next generation sequencing method, we aimed to do an in depth characterization of the fungal diversity found in compost and in the bioaerosols emitted during composting in two different environments that treat different types of organic waste. There is a concern that people working in composting plants are more likely to develop respiratory illnesses than other working environments (Bünger et al., 2007; Domingo and Nadal, 2009). Even though microbial diversity in bioaerosols is getting more attention, there is a lack of information about the total fungal diversity (Le Goff et al., 2010; Bru-Adan et al., 2009). The methodology most often used to assess fungal diversity has some weaknesses and there is a call for the reassessment of fungal exposure in bioaerosols emitted from composting environments (Bonito et al., 2010). Our results show an unexpectedly high fungal diversity in the bioaerosols released from both composting environments providing new insights into the fungal community present in organic matter and those aerosolized from it, causing us to consider the potential health risks to workers. The fact that the number of OTUs observed in air is higher than in compost suggests that a certain heterogeneity of compost makes difficult the representativeness of the complete compost pile. Thus, it is not quiet representative of the whole source, as it represents a specific zone in the compost. Air is more homogenous as it captures different sources at the same time making them into one. This hypothesis points out the importance of air sampling in order to get a better view of the microbial content of a contaminant source. The variation in fungal communities corresponding to different stages of composting has been asserted in previous studies (Hansgate et al., 2005; Langarica- Fuentes et al., 2014; De Gannes et al., 2013). Those variations are explained by changes in the physico-chemical characteristics of the compost, including pH, temperature and moisture content (Ishii et al., 2000; Zhang et al., 2010). Consequently, the type of raw material composted might affect those characteristics, which could explain the variation of fungal communities observed through the multivariate analyses used in this study. Notably, decomposing an animal carcass requires more microbial activity than conventional composting (Berge et al., 2009), which can lead to higher temperatures and affect other physico-chemical characteristics. The considerable presence of Eurotiomycetes in animal compost might be the consequence of those variations. However, since composting animal carcasses is a very recent practice used as an alternative to current incineration methods, less

154

is known about the microbial community involved in this type of composting (Gwyther et al., 2011). Further studies will help confirm the role of the Eurotiomycetes in animal composting. Because of the emergent nature of composting, this study represents a valuable contribution of information about the roles of fungi involved in this process. The variation observed between the fungal communities according to the type of composting environment (air or compost) can be explained by the big differences in the diversity profile in the less abundant phyla. A large number of fungal species was detected below 1% of all sequencing hits. It is evident that the real fungal diversity was underestimated in the air examined from waste related working environments. On a taxonomic level, previous studies have reported that Ascomycota and Basidiomycota are the most abundant phyla of fungi present in composting facilities (Bru-Adan et al., 2009; De Gannes et al., 2013; Langarica-Fuentes et al., 2014). This was confirmed in the current study. However, we believe that this is the first study to illustrate such high fungal diversity in compost and in bioaerosols released from composting activities. Members of Sordariomycetes and Eurotiomycetes are known decomposers (Maharachchikumbura et al., 2015; Geiser et al., 2006). However, the higher proportion of Sordariomycetes in domestic compost may be explained by the fact that they are also known particularly as insect and plant pathogens (Zhang et al., 2006). As expected, domestic waste is linked with vegetal products and may contain several plant pathogens. The fungal link between the source (type of waste) and its decomposition (composting) is hard to make for animal composting. As previously stated, not much is known about fungal activity in carcass decomposition. Nonetheless, this study highlights the bigger proportion of Eurotiomycetes in animal composting. This finding is supported by results from a previous study in which the authors found the Eurotiomycetes to be the most abundant eukaryote in the advanced stages of vertebrate decay (Lauber et al., 2014). To the best of our knowledge, many of the fungi genera in Table 4.2 have never before been identified in a composting environment. Thus, opening the door for new hypotheses about their specific role in composting animal carcasses. Another example illustrating the relationship between the fungal community and the source is the detection of Sagenomella in animal composting samples. This fungus is associated with systemic animal illnesses (Gené et al., 2003) supporting the postulation of the presence of fungal animal parasites in animal composting environments and the bioaerosols emitted from them. It should also be noted that, the allergen

155

activity and/or the infectiousness of some of the fungi detected in this study have not been shown to pose any human health risks.

Because domestic compost likely includes decaying fruits and vegetables, it is logical that Penicillium is the most abundant genera, all samples from both sources. Penicillium is ubiquitous in the environment and plays an important role in natural food decomposition (Perrone and Susca, 2016). Its decaying capacity might include vegetal and animal products since Penicillium was the most abundant genera in animal composting as well (air and compost samples).

As indicated in the results (Fig. 4.4 and Fig. 4.5), airborne particles in each environment are often linked to the source. In some cases, the proportion of organisms present at that source may not be the same in the air. This is a phenomenon described as preferential aerosolization. The hypothesis suggests that some bacteria may be preferentially aerosolized compared to others. Parker et al. introduced the idea that respiratory pathogens such as Mycobacterium spp. and Legionella spp. could be enriched in aerosols from the environment (Parker et al., 1983). In addition, a study comparing the microbial composition of biogas and anaerobic digestors showed that some microbial phyla had higher proportions in the air compared to the source (Moletta et al., 2007). This study showed that the proportions of Proteobacteria in compost piles differed compared to air samples in compost facilities. Some species like Methylobacterium were higher in proportion in air samples than in compost (Veillette et al., 2016; Mbareche et al., 2015). To the best of our knowledge, no other study has demonstrated this aerosolization behavior in fungi. This study underscores the higher percentage of some fungi species found in air compared to compost. For Aspergillus, Malassezia, Leucosporidium and Resinicium, the air enrichment was noted in both animal and domestic composting. It is important to consider the potential health risks for people with frequent exposure to composting as some of these identified fungi are opportunistic pathogens or allergenic and are undoubtedly inhaled by workers. The presence of opportunistic pathogens in aerosols released from composting environments has been previously reported (Wéry, 2014; Le Goff et al., 2010; Bru-Adan et al., 2009) but, this is the first study to reveal the presence of fungi in bioaerosols emitted from a composting

156

environment known to cause health problems. This study identified several members of the family Herpotrichiellaceae and Gymnoascaceae which have been reported to cause diseases in humans (de Hoog et al., 2000; Crous et al., 2007; Ghosh, 1985). Other opportunistic, potential pathogens were detected for the first time in composting air samples. Coniosporium is an environmental fungal genus that forms black patches on plants and rotten wood raising new dermatological concerns (Li et al., 2008). Malassezia causes skin disorders and can lead to invasive infections in immunocompetent individuals (Velegraki et al., 2015). Emericella is a taxon of teleomorphs related to Aspergillus. Species of this group are known agents of chronic granulomatous disease (CGD) (Matsuzawa et al., 2010). Acremonium causes fungemia in immunosuppressed patients (Rodriguez and Ramos, 2014). Fusarium species are responsible for a broad range of health problems, from local and systemic infections to allergy related diseases such as sinusitis, in immunodepressed individuals (Nucci and Anaissie, 2007). Some of the fungi were found at higher percentages in air compared to compost which may be harmful to people constantly in contact with those fungi. Indeed, the air enrichment of particular fungal spores makes them more likely to be inhaled and therefor responsible of respiratory health problems. These results indicate that protective measures should be taken in order to minimize human exposure to these aerosols. As composting becomes more important as an ecofriendly alternative for waste management, it is crucial to consider the harmful fungi released during composting processes. This study is the first to provide a comprehensive analysis of fungal diversity from two different composting environments and to identify the fungal diversity present in bioaerosols including many known pathogenic fungi. The information from this study will allow for better risk assessments for composting workers.

4.8 Conclusion

This study highlights the importance of using the ITS1 region in the high-throughput sequencing method for an in-depth characterization of fungal diversity in bioaerosols with a bioinformatics workflow to facilitated the data analyses. Based on the results of this investigation, the authors strongly recommend taking action to reduce the worker exposure. Technical and organizational measure should be implemented in composting facilities as a

157

first resort. Such measures include building efficient ventilation systems, better confinement and capturing air contaminants from the source. If the first measure can't be taken, we recommend skin and respiratory protection for compost workers as a means to reduce continuous exposure to harmful fungi present in bioaerosols. The broad spectrum of fungi detected in this study includes many know pathogenic agents and adequate monitoring of exposure is necessary to diminish risks. Additional studies should be conducted to further support the conclusion that there are fungal signatures associated with the type of waste treated.

Acknowledgements

We are grateful to all of the employees of the composting sites that participated in this study. We are also grateful to Maude Talbot, Eric Légaré and the members of the IRSST that were in the field for their technical assistance. This work was supported by the Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail (grant number: 2014-0057). The authors are thankful to Amanda Kate Toperoff for English revision of the manuscript.

4.9 Bibliography

Anastasi, A., Varese, G. C., Marchisio, V.F. 2005. Isolation and identification of fungal communities in compost and vermicompost. Mycologia 97(1)33-44.

Anderson, M.J. 2005. PERMANOVA: a FORTRAN computer program for permutational multivariate analysis of variance. Department of Statistics, University of Auckland, New Zealand.

Berge, A.C.B., Glanville, T.D., Millner, P.D., Klingborg, D.J., 2009. Methods and microbial risks associated with composting of animal carcasses in the United States. J. Ame. Vet. Med. Assoc. 234(1):47-56.

Bonito,G., Isikhuenmhen,O. S., Vilgalys, R., 2010. Identification of fungi associated with municipal compost using DNA based techniques. Bioresour. Technol. 101(3):1021-1027. Bru-Adan, V., Wéry, N., Moletta-Denat, M., Boiron, P., Delgènes, J.P., Godon J.J., 2009. Diversity of bacteria and fungi in aerosols during screening in a green waste composting plant. Curr. Microbiol. 59(3):326-335.

158

Bünger, J., Antlauf-Lammers, M., Schulz, T.G., Westphal, A.G. Müller, M. M., Ruhnau, P., Hallier, E., 2000. Health complaints and immunological markers of exposure to bioaerosols among biowaste collectors and compost workers. Occup. Environ. Med. 57(7):458–464.

Bünger, J., Schappler-Scheele, B., Hilgers, R., Hallier, E. 2007. A 5-year follow-up study on respiratory disorders and lung function in workers exposed to organic dust from composting Plants. Int. Arch. Occup. Environ. Health. 80(4):306–312.

Crous, P.W., Schubert, K., Braun, U., de Hoog, G.S., Hocking, A.D., Shin, H.D., et al., 2007. Opportunistic, human-pathogenic species in the Herpotrichiellaceae are phenotypically similar to saprobic or phytopathogenic species in the Venturiaceae. Stud. Mycol. 58:185– 217.

De Gannes, V., Eudoxie, G., Hickey William, J., 2013. Insights into fungal communities in composts revealed by 454-pyrosequencing: implications for human health and safety. Front Microbiol. 4:164. de Hoog, G.S., Guarro, J., Gené, J., Figueras, M.J., 2000. Atlas of Clinical Fungi. Utrecht/Reus:Centraalbureau voor Schimmelcultures/Universitat Rovira i Virgili. p. 1126. 2nd ed.

Domingo, J.L., Nadal, M., 2009. Domestic waste composting facilities: a review of human health risks. Environ. Int. 35(2):382-389.

Douwes, J., Thorne, P., Pearce, N., Heederik, D. 2002. Bioaerosol Health Effects and Exposure Assessment: Progress and Prospects. Ann. occup. Hyg. 47(3):187–200.

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26(19): 2460-2461.

Edgar, R.C., 2013. UPARSE: Highly accurate OTU sequences from microbial amplicon reads. Nat. Methods. 10:996-998.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 27(16):2194-2200.

Epstein, E., Wu, N., Youngberg, C., Croteau, G., 2001. Dust and bioaerosols at a biosolids composting facility. Compost Sci. Util. 9(3): 250-255,

Floudas, D., Binder, M., Riley, R., Barry, K., Blanchette, R.A., Henrissat, B., et al., 2012. The paleozoic origin of enzymatic lignin decomposition reconstructed from 31 Fungal Genomes. Science. 336(6089):1715-1719,

159

Fuchs, J.G., Bieri, B., 2000. New biotests to measure the biological qualities of composts. Agrarforschung Schweiz. 7(7):314-319.

Fung, F., Clark, R.F., 2004. Health Effects of Mycotoxins: A Toxicological Overview. J. Toxicol. Clin. Toxicol. 42(2):217–234.

Geiser, D.M., Gueidan, C., Miadlikowska, J., Lutzoni, F., Kauff, F., Hofstetter, V. et al., 2006. Eurotiomycetes: Eurotiomycetidae and Chaetothyriomycetidae. Mycologia. 98(6):1053-1064.

Gené, J., Blanco, J.L., Cano, J., García, M.E., Guarro, J., 2003. New Filamentous fungus Sagenomella chlamydospora responsible for a disseminated infection in a dog. J. Clin. Microbiol. 41(4):1722–1725.

Ghosh, R.R., 1985. Physiological studies of the Gymnoascaceae with an overview on their ecology. Indian Acad. Sci. 94:197.

Gwyther, C.L., Williams, P., Golyshin, P.N., Edwards-Jones, G., Jones, D.L., 2011. The environmental and biosecurity characteristics of livestock carcass disposal methods: A review. Waste manag. 31(4):767–778.

Hansgate, A.M., Schloss, P.D., Hay, A.G., Walker, L.P., 2005. Molecular characterization of fungal community dynamics in the initial stages of composting. FEMS Microbiol. Ecol. 51(2):209-214.

Hoorman, J.J., 2011. The role of soil fungus. Fact sheet. Agricultural and Natural Resources. Ohio State University. SAG-14-11.

Ishii, K. , Fukui, M., Takii, S., 2000. Microbial succession during a composting process as evaluated by denaturing gradient gel electrophoresis analysis. J. Appl. Microbiol. 89(5):768- 77.

Langarica-Fuentes, A., Zafar, U., Heyworth, A., Brown, T., Fox, G., and Robson, D.G., 2014. Fungal succession in an in-vessel composting system characterized using 454 pyrosequencing. FEMS Microbiol. Ecol. 88(2):296-308.

Lauber, C.L., Metcalf, J.L., Keerpers, K., Ackermann, G., Carter, D.O., Knight, R., 2014. Vertebrate decomposition is accelerated by soil microbes. Appl. Environ. Microbiol 80(16):4920-4929.

Le Goff, O., Bru-Adan, V., Bacheley, H., Godon, J.J., Wéry, N. 2010. The microbial signature of aerosols produced during the thermophilic phase of composting. J Appl Microbiol. 108(1):325-40.

Le Goff, O., Godon, J.J., Steyer, J.P., Wéry, N., 2011. New specific indicators for qPCR monitoring of airborne microorganisms emitted by composting plants. Atmos. Environ 45(30):5342-5350.

160

Li, D.M., de Hoog, G.S., Saunte, D.M.L., van den Ende, A.H.G.G., Chen, X.R., 2008. Coniosporium epidermidis sp. nov., a new species from human skin. Stud. Mycol. 61:131– 136.

Maharachchikumbura, S.S., Hyde, K.D., Jones, E.G., McKenzie, E.H., Huang, S.K., Abdel- Wahab, M.A., et al., 2015. Towards a natural classification and backbone tree for Sordariomycetes. Fungal Divers. 72(1):199-301.

Matsuzawa, T., Tanka, R., Horie, Y., Gonoi, T., Yaguchi, T., 2010. Development of rapid and specific molecular discrimination methods for pathogenic Emericella species. Jpn. J. Med. Mycol. 51:109-116.

Mbareche, H., Veillette, M., Bonifait, L., Bernard, Y., Marchand, G., Duchaine, C., 2015. Preferential aerosolization of bacteria in the air of different composting plants. 4th International Conference and Exhibition on Occupational Health & Safety August 24-25 201. Toronto, Canada.

Mistry, J., Finn, R.D., Eddy, S.R., Bateman, A., Punta, M., 2013. Challenges in Homology Search: HMMER3 and Convergent Evolution of Coiled-Coil Regions. Nucleic Acids Res. 41:e121.

Moletta, M., Delgenes, J., Godon, J., 2007. Differences in the aerosolization behavior of microorganisms as revealed through their transport by biogas. Sci. Total Environ. 379(1): 75–88.

Nucci, M., Anaissie, E., 2007. Fusarium Infections in Immunocomprised Patients. Clin. Microbiol. Rev. 20(4):695-704.

O’Connor, J.D., Daly, S.M., Sodeau, J.R., 2015. On-line monitoring or airborne bioaerosols released from a composting/green waste site. Waste manag. 42:23-30

Park, D., Ryu, S., Kim, S., Byun, H., Yoon, C., Lee, K., 2013. Airborne bacteria and fungi associated with waste-handling work. Int. J. Occup. Environ. Health. 19(4):313-317.

Perrone, G., Susca, A., 2016. Penicillium Species and Their Associated Mycotoxins. Methods Mol. Biol. 1542:107-119.

Persoons, R., Parat, S., Stoklov, M., Perdrix, A., Maitre, A., 2010. Critical working tasks and determinants of exposure to bioaerosols and MVOC at composting facilities. Int. J. Hyg.

161

Environ. Health. 213(5):338-247.

Rodriguez, Z.C., Ramos, M.G., 2014. Acremonium species associated fungemia: a novel pathogen in the immunosuppressed patient. Bol. Asoc, Med P. R. 106(3):29-31.

Ryckeboer, J., Mergaert, J., Vaes, K., Klammer, S., De Clercq, D., Coosemans, J., et al. 2003. A survey of bacteria and fungi occurring during composting and self-heating processes. Ann. Microbial. 53(4):349-410.

Sanchez-Monedero M.A., Stentiford, E.I. 2003. Generation and dispersion of airborne microorganisms from composting facilities. Trans IChem 81(3):166-170.

Sánchez-Monedero, M.A., Stentiford, E.I., Urpilainen, S.T., 2005. Bioaerosol Generation at Large-Scale Green Waste Composting Plants. J. Air Waste Manag. Assoc. 55:5:612-618.

Schlosser, O., Huyard, A., Rybacki, D., Do Quang, Z., 2012. Protection of the vehicle cab environement against bacteria, fungi and endotoxins in composting facilities. Waste management. 32:1106-1115.

Swan, J.R.M., Kelsey, A., Crook, B., 2003. Occupational and environmental exposure to bioaerosols from composts and potential health effects -A critical review of published data. Research report 130. Health & Safety Executive. The composting association.

Sykes, P., Jones, K., Wildsmith, J.D., 2007. Managing the potential public health risks from bioaerosol liberation at commercial composting sites in the UK: an analysis of the evidence base. Resour. Conserv. Recycl. 52(2):410.

Taha, M.P.M., Drew, G.H., Longhurst, P. J., Smith, R., Pollard, S.J.T., 2006. Bioaerosol releases from compost facilities: evaluating passive and active source terms at a green waste facility for improved risk assessments. Atmos. Environ. 40(6):1159–1169.

Taha, P.M.P., Pollard, S.J.T., Sarkar, U., Longhurst, P. 2005. Estimating fugitive bioaerosol releases from static compost windrows: Feasibility of a portable wind tunnel approach. Waste manag. 25(4). 445-450.

Van Kampen, V., I. Sander, V. Liebers, A. Deckert, H.D. Neumann. M. Buxtrup, et al., 2014. Concentration of bioaerosols in composting plants using different quantification methods. Ann. Occup. Hyg. 58(6):693-706.

162

Veillette, M., Bonifait, M., Marchand, G., Duchaine, C., 2016. Bioaerosols exposure in indoor composting installations: classical and molecular approaches and preferential aerosolization. 4th Workplace and Indoor Aerosols Conference 20-22 April 2016. Barcelona, Spain.

Velegraki, A., Cafarchia, C., Gaitanis, G., Iatta, R., Boekhou, T., 2015. Malassezia Infections in Humans and Animals: Pathophysiology, Detection, and Treatment. PLoS Pathog. 11(1): e1004523.

Vestlund, A. T., Al-Ashaab, R., Tyrrel, S. F., Longhurst, P. J., Pollard S.J.T., Drew, G.H., 2014. Morphological classification of bioaerosols from composting using scanning electron microscopy. Waste Manag. 32(6):1101–1108.

Wéry, N. 2014., Bioaerosols from composting facilities – a review. Front. Cell Infection Microbiol. 4:42.

Zhang, J., Zeng, G., Chen, Y., Yu, M., Yu, Z., Li, H., Yu, Y., Huang, H., 2010. Effects of physico-chemical parameters on the bacterial and fungal communities during agricultural waste composting. Bioresour. Technol. 102(3)2950–2956.

Zhang, N., Castlebury, L.A., Miller, A.N., Huhndorf, S.M., Schoch, C.L., Seifert, K.A., et al., 2006. An overview of the systematics of the Sordariomycetes based on a four-gene phylogeny. Mycologia. 98(6)1076-1087.

163

Chapter 5: Fungal bioaerosols in biomethanization facilities

5.1 Résumé

La biométhanisation est une nouvelle technologie utilisée pour la valorisation des déchets verts, dans laquelle les déchets sont biodégradés par des microorganismes dans des conditions anaérobies. Il existe peu de données sur la diversité fongique dans l’air des usines biométhanisation. Le but de cette étude est de fournir une description des moisissures l’Air dans des usines de biométhanisation en utilisant une approche de séquençage haut débit combinée à une PCR en temps réel. Deux usines de biométhanisation traitant différents déchets ont été visitées lors de la campagne d'échantillonnage. La quantification de Penicillium/Aspergillus et d’Aspergillus fumigatus a révélé un risque d'exposition plus important en été. Les analyses taxonomiques ont montré que le type de déchet traité affecte la diversité fongique des aérosols émis et que certains pathogènes ont été identifiés. Les résultats de cette étude peuvent servir de référence pour minimiser l'exposition professionnelle dans de futures installations de biométhanisation.

164

5.2 Summary of the Paper

This paper tackles the question of the type of fungal exposure in biomethanization facilities. It is a waste treatment environment where workers could be exposed to bioaerosols, especially, during the reception/handling of the waste, and during the final maturation step of the compost piles. Two biomethanization facilities treating different types of waste were visited during the air sampling campaign. To have a complete portrait of fungal exposure quantitatively and qualitatively, the authors used a specific qPCR targeting Penicillium/Aspergillus and A. fumigatus combined to HTS targeting ITS1. The same bioinformatic protocol as the compost study was used to confirm its reproducibility and robustness.

Concentrations of the targeted fungi were the same for all the work stations. However, concentrations were higher during summer compared to winter. The results obtained from the diversity analyses confirmed the relation between the fungal composition and the type of waste treated, previously observed in the compost study, as the taxonomic profiles were different for both facilities. Like the compost environment, a large diversity profile was observed in bioaerosols from both facilities showing the presence of pathogenic fungi. This observation served as a proof of the validity of the approach applied because, it was able to catch the difference in fungal composition of bioaerosols released from both facilities. Plus, quantifying the fugal biomass with qPCR allowed the addition of an absolute quantification, rather than just a relative abundance.

The combination of both approaches offers a more complete idea on the type of fungal exposure. As biomethanization facilities are growing rapidly in the waste treatment market, the results presented in this paper can serve for future decision make to try and reduce bioaerosol emission from the source.

165

Fungal bioaerosols in biomethanization facilities

RUNNING TITLE Fungi in biomethanization facilities

AUTHORS Hamza Mbareche1,3, Marc Veillette1, Marie-Eve Dubuis1,3, Bouchra Bakhiyi4, Geneviève Marchand2, Joseph Zayed2,4, Jacques Lavoie2, Guillaume J. Bilodeau5 and Caroline Duchaine1,3*

AUTHORS’ AFFILIATION 1. Centre de Recherche de l’Institut Universitaire de Cardiologie et de Pneumologie de Québec (CRIUCPQ), Quebec City (Qc), Canada 2. Institut de Recherche Robert-Sauvé en Santé et en Sécurité du travail (IRSST), Montreal (Qc), Canada 3. Département de biochimie, de microbiologie et de bio-informatique, Faculté des sciences et de génie, Laval university, Quebec City (Qc), Canada 4. Department of Environmental and Occupational Health, School of Public Health, University of Montreal, Montreal, (Qc), Canada 5. Pathogen Identification Research Lab, Canadian Food Inspection Agency (CFIA). Ottawa, Canada

KEYWORDS Fungal spores, bioaerosols, biomethanization facilities, worker exposure

CORRESPONDING AUTHOR *Mailing address : Caroline Duchaine, Ph.D., Centre de Recherche de l’Institut Universitaire de Cardiologie et de Pneumologie de Québec, 2725 Chemin Ste-Foy, Québec, Canada, G1V 4G5. Phone: (418) 656-8711 ext. 5837. Fax: (418) 656-4509. E-mail: [email protected] PUBLISHED : J Air Waste Manag Assoc. 2018 Nov;68(11):1198-1210. doi: 10.1080/10962247.2018.1492472.

166

5.3 Abstract

Biomethanization is a new technology used for green-waste valorization where organic waste is biodegraded by microbial communities under anaerobic conditions. The main product of this type of anaerobic digestion is a biogas used as an energy source. Moving and handling organic waste may lead to the emission of high concentrations of bioaerosols. High exposure levels are associated with adverse health effects amongst green environment workers. Fungal spores are suspected to play a role in many respiratory illnesses. There is a paucity of information related to the detailed fungal diversity in biomethanization facilities. The aim of this study was to provide an in-depth description of fungal bioaerosols in biomethanization work environments using a next-generation sequencing approach combined with real-time polymerase chain reaction (PCR). Two biomethanization facilities treating different wastes were visited during the sampling campaign (n = 16). Quantification of Penicillium/Aspergillus and Aspergillus fumigatus revealed a greater exposure risk during summer for both facilities visited. Concentrations of Penicillium and Aspergillus were similar in all work areas in both biomethanization facilities. Taxonomy analyses showed that the type of waste treated affects the fungal diversity of aerosols emitted. Although eight classes were evenly distributed in all samples, Eurotiomycetes were more dominant in the first facility and Agaricomycetes were dominant in the second one. A large diversity profile was observed in bioaerosols from both facilities showing the presence of pathogenic fungi. The following fungi detected are known allergens and/or are opportunistic pathogens: Aspergillus, Malassezia, Emericella, Fusarium, Acremonium, and Candida. Daily exposure to these fungi may put workers at risk. The information from this study can be used as a reference for minimizing occupational exposure in future biomethanization facilities.

Implications

Biomethanization is a new technology used for green-waste valorization where organic waste is biodegraded by microbial communities. Effective waste management is increasingly recognized as a strategic approach for achieving newly created regulations

167

concerning the disposal of organic residues; therefore, an expansion of facilities is expected. Workers’ exposure to diverse fungal communities is certain, as fungi are ubiquitous and necessary in organic matter decomposition. Monitoring this occupational exposure is important in order to prevent workers’ health problems.

5.4 Introduction

Approximately 1.3 billion tons of solid waste is collected worldwide every year. This number is expected to increase to 2.2 billion tons by 2025, with almost all of the increase coming from developing countries (Hoornweg and Bhada-Tata, 2012). Effective waste management is increasingly recognized as a strategic approach for achieving newly created regulations concerning the disposal of organic residues (Collagher et al., 2017; Mahajan and Vakhariya, 2016; United Nations Department of Social and Economic Development n.d.).

Composting has become a promising technology for managing different types of organic waste (Sykes et al., 2007). Composting involves the degradation of organic matter by a complex microbial community under aerobic conditions where gram-positive and gram- negative bacteria and fungi are present (Hansgate et al. 2005; Ishii, Fukui, and Takii 2000; Ryckeboer et al. 2003). Unfortunately, composting is also linked to health hazards associated with bioaerosols emitted during several of the necessary composting steps (Epstein et al., 2001; Sanchez-Monedero et al., 2005; Persoons et al., 2010; Taha et al., 2005). Greenhouse gases are also emitted, and depending on the quality, the final product of composting cannot always be used as fertilizer (Fuchs and Bieri, 2000; Lim et al., 2016).

Biomethanization is another technology used for green-waste valorization where organic waste is bio-degraded by microbial communities under anaerobic conditions. The main product of this type of anaerobic digestion is a biogas made up of approximately 65% methane and 35% carbon dioxide (Mata-Alvarez, 2003). Because the biogas contains methane, it can be used as an energy source in the same way as natural gases (Amon et al., 2007; Dai et al., 2017; Karakurt et al., 2011; Lee and Holder, 2001). For example, the biogas

168

produced from biomethanization can be used to heat facilities, operate generators, or fuel vehicles.

Another advantage of biomethanization is the ability to process all types of organic waste ranging from municipal green waste, industrial food waste, to sewage sludge and animal (cattle, cow, and pig) manure (Bouallagui et al., 2005; El-Mashad et al., 2003; Macias-Corral et al., 2008; Murto et al., 2004; Sosnowski et al., 2003). Waste management using this type of anaerobic digestion is more cost-effective and less harmful to the environment compared with composting.

Waste management environments are often associated with increased bioaerosols. Composting activities such as load reception, handling, shredding, pile-turning, and screening of organic matter all contribute to bioaerosol emissions (Viegas et al., 2015). Exposure to high concentrations of these bioaerosols, composed predominantly of bacteria and fungi, suggests a potential health risk (Epstein et al., 2001; Sanchez-Monedero et al., 2005; Persoons et al., 2010; Taha et al., 2005). Respiratory symptoms are among the most prevalent bioaerosol-associated health effects (Domingo and Nadal, 2009) reported. Tracheobronchitis, mucosal irri-ations, sinusitis, eczema, dermatomycosis, and gastro- intestinal problems are some examples of health-related complaints from compost workers and biowaste collectors (Bünger et al., 2000). Compared with other industrial activities, composting represents a greater risk to workers due to the presence of higher concentrations of Actinomycetes and thermophilic and/or thermotolerant fungi (Bünger et al., 2007).

In biomethanization facilities, exposure to high concentrations of bioaerosols is comparable to that at composting sites, as the delivery and the handling of the organic matter are similar in both settings. Because of the rapid development of biogas production using anaerobic digestion in eastern Canada, assessing workers’ exposure in biomethanization facilities is imperative. Even with this rapid development, few studies have addressed the issue of exposure. Dubuis et al. (2017) described the bioaerosol concentrations and microbial composition of air samplers from a biomethanization environment, with a particular focus on

169

bacteria. To the best of our knowledge, no previous studies have described the fungal composition of aero-sols in this type of environment.

Most fungi are aerobic, but anaerobic fungi have also been shown to be a part of the microbial community in biogas reactors (Haitjema et al. 2014). These fungi possess highly active enzymes (plant cell wall–degrading enzymes) capable of degrading polysaccharides, which makes them likely participants in biomass degradation and biogas production (Kazda et al., 2014). Therefore, their presence during the biomethanization process is expected. Workers’ exposure to diverse fungal communities is certain, as fungi are ubiquitous and necessary in organic matter decomposition (Anastasi et al., 2005; Hoorman, 2011; Floudas et al., 2012; Hansgate et al., 2005; Langarica-Fuentes et al. ,2014; De Gannes et al., 2013). Exposure occurs during the reception and hand-ling of the organic matter and can lead to a variety of health issues. Particularly, exposures to fungi, their secondary metabolites, and other components are believed to be associated with asthma, allergic alveolitis, chronic bronchitis, hypersensitivity pneumonitis, and infections, especially in people with impaired immune systems (Swan et al., 2003; Fung and Clark, 2004; Pieckova and Wilkins, 2004; Selman et al., 2010; Wyngaarden et al., 1992; Latgé, 1999; Velegraki et al., 2015; Rodriguez and Ramos, 2014; Nucci and Anaissie, 2007).

Fungi represent certain challenges when recovered from air, possibly due to their biochemical characteristics. A recent study identified the problem of fungal spore loss when recovered from air samples and pro-posed a new protocol that helped recover higher fungal concentrations than the traditional approaches. The newly developed protocol led to a more accurate assessment of fungal bioaerosol exposure (Mbareche et al., manuscript in preparation).

The aim of this study is to use the optimized protocol for fungal recovery from air samples using a next-generation sequencing approach combined with real-time polymerase chain reaction (PCR) to provide an in-depth description of fungal exposure in biomethanization work environments.

170

5.5 Materials and Methods

Biomethanization facilities and site sampling

The two biomethanization facilities (BFs) visited were located in the province of Quebec (eastern Canada). Both plants treat different types of waste under different conditions. The first BF treats primary and secondary sludge from wastewater treatment plants and organic industrial food waste. At this facility, waste is treated under mesophilic conditions, and it has an annual capacity of 40,000 tons of waste. The second plant treats domestic waste under thermophilic conditions, with an annual capacity of 27,000 tons. In order to evaluate the effect of seasons on fungal exposure, the two facilities were visited once in the summer and again in the winter.

Five sites were sampled in the first BF: Reception, Treatment (shredding and mixing), Storage (of dry organic matter) and Maturation (of dry organic matter), Press filters (water removal of the sorting product), and Output (of finished products where the dried digestate is stored and loaded into trucks for further maturation).

In the second BF, two sites were sampled: the Reception site that included shredding and the Mixing site where the organic matter is mixed with a buffer before it is put into the digesters.

Air sampling

Air samples were collected during activities where workers were likely exposed to the aerosols associated with composting. To collect air samples a liquid cyclonic impactor, Coriolis µ (Bertin Technologies, Montigny-le-Bretonneux, France) set at 300 L/min for 10 min was placed in the center of the handling operations. Fifteen milliliters of sterile 50 mM phosphate-buffered saline (pH 7.4) were used to fill the sampling cone of the Coriolis.

171

At each site, three samples were collected. One was collected at the beginning, one in the middle, and one at the end of the process. Samples were then pooled to obtain a final volume of 45 mL for each sampled site, representing 9 m3 of air. An outdoor control air sample was also collected outside of each facility during the summer. The cold eastern Canadian weather did not allow for outdoor sampling during the winter because the Coriolis sampler is not designed to function in temperatures below zero degrees Celsius. Because of these same constraints, no samples were collected at the Storage and Maturation site in the first BF during winter.

Fungal spore concentration using the filtration protocol

The 45-mL samples from the Coriolis suspension were filtered through a 2.5-cm polycarbonate mem-brane (0.2 mm pore size; Millipore, Etobicoke, Ontario, Canada) using a vacuum filtration unit. The filters were placed in a 1.5-mL Eppendorf tube with 750 µL of extraction buffer (bead solution) from a MO BIO PowerLyzer Powersoil DNA isolation kit (Carlsbad, CA, USA) and a 0.3-cm tungsten bead. The filters were flash-frozen by placing the Eppendorf tube in a 99% ethanol solution and dry ice. The frozen filters were then pulverized using the tungsten steel bead in the Eppendorf tube in a bead-beating machine (Mixer Mill MM301; Retsch, Düsseldorf, Germany) set at a frequency of 20 movements per second for 20 min. The liquid containing the pulverized filter particles was used as aliquot for the first step of the DNA extraction procedure.

DNA extraction

Using the same bead-beating apparatus, a second bead-beating step using glass beads at a frequency of 20 movements per seconds for 10 min was performed to ensure that all of the cells were ruptured. Next, a MO BIO PowerLyzer Powersoil DNA isolation kit was used to extract the total genomic DNA from the samples following the manufacturer’s instructions. Next, the DNA was eluted in a 100 µL buffer, and stored at −20 °C until subsequent analyses. Real-time PCR quantification

172

PCR was performed with a Bio-Rad CFX 96 thermo-cycler (Bio-Rad Laboratories, Mississauga, Ontario, Canada). The PCR mixture contained 2 µL of DNA template, 0.150 µmol/L per primer, 0.150 µmol/L probe, and 7.5 µL of 2× QuantiTect Probe PCR master mix (QuantiTect Probe PCR kit; Qiagen, Mississauga, Ontario, Canada) in a 15-µL reaction mixture. The results were analyzed using Bio-Rad CFX Manager soft-ware version 3.0.1224.1015 (Bio-Rad Laboratories). Aspergillus fumigatus was used for the standard curves of both quantitative PCR (qPCR) analyses (Penicillium/ Aspergillus and Aspergillus fumigatus). Table 5.1 presents the primers, probes, and PCR protocols used in this study.

Table 5.1: Primers, probes, and protocols used for qPCR quantification of selected microorganisms. Microorganisms and references Primers and probes PCR protocol Penicillium, Aspergillus and Paecilomyces PenAsp1mgb (Taqman) Activation: 94°C-3min variotii PenAspR1: 5’-GCCCGCCGAAGCAAC-3’ Denaturation: 94°C-15sec PenAspF1: 5’-CGGAAGGATCATTACTGAGTG- Annealing/extension: 60°C-60sec http://www.epa.gov/microbes/moldtech.html 3’ Cycles: 40 PenAspP1mgb: 5’-FAM- CCAACCTCCCACCCGTG-TAMRA-3’ Aspergillus fumigatus and Neosartoya Afumi (Taqman) Activation: 94°C-3min fischeri AfumiR1: 5’- Denaturation: 94°C-15sec CCGTTGTTGAAAGTTTTAACTGATTAC-3’ Annealing/extension: 60°C-60sec http://www.epa.gov/microbes/moldtech.html AfumiF1: 5’-GCCCGCCGTTTCGAC-3’ Cycles: 40 AfumiP1: 5’-CCCGCCGAAGACCCCAACATG- 3’

Next-generation sequencing

The rRNA fungal gene ITS1 was used for the next-generation sequencing analyses. Amplification of the amplicons, equimolar pooling, and sequencing were performed at the Plateforme d’analyses génomiques (IBIS, Université Laval, Quebec City, Quebec, Canada). Briefly, amplification of the ITS1 regions was performed using the sequence-specific regions described by Tedersoo et al. (2015) and the references therein, using a two-step dual-indexed PCR approach specifically designed for Illumina instruments. First, the gene-specific sequence was fused to the Illumina TruSeq sequencing primers and PCR was carried out on a total volume of 25 µL of liquid made up of 1× Q5 buffer (NEB, xx, xx), 0.25 µM of each primer, 200 µM of each of the dNTPs, 1 U of Q5 High-Fidelity DNA polymerase (NEB),

173

and 1 µL of template cDNA. The PCR started with an initial denaturation at 98 °C for 30 sec followed by 35 cycles of denaturation at 98 °C for 10 sec, annealing at 55 °C for 10 sec, extension at 72 °C for 30 sec and a final extension step at 72 °C for 2 min. The PCR reaction was purified using an Axygen PCR cleanup kit (Axygen, xx, xx). Quality of the purified PCR products was verified with electrophoresis (1% agarose gel). Fifty- to 100-fold dilution of this purified product was used as a template for a second round of PCR with the goal of adding barcodes dual-indexed) and missing sequences required for Illumina sequencing. Cycling conditions for the second PCR were identical to the first PCR but with 12 cycles. The PCR products were purified as above, checked for quality on a DNA7500 Bioanlayzer chip (Agilent, Santa Clara, CA, USA), and then quantified spectrophotometrically with a Nanodrop 1000 (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Barcoded amplicons were pooled in equimolar concentrations for sequencing on the Illumina Miseq. The oligonucleotide sequences used for amplification are presented in Table 5.2.

Please note that primers used in this work contain Illumina specific sequences protected by intellectual property (Oligonucleotide Sequences © 2007–2013 Illumina, Inc. All rights reserved). Derivative works created by Illumina customers are authorized for use with Illumina instruments and products only. All other uses are strictly prohibited.

Table 5.2: Primers used for Illumina amplification.

First-PCR primer ITS1Fngs: 5’-ACACTCTTTCCCTACACGACGCTCTTCCGATCTGGTCATTTAGAGGAAGTAA-3’ ITS2: 5’-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGCTGCGTTCTTATCGATGC-3’ Second-PCR Generic forward: 5’AATGATACGGCGACCACCGAGATCTACAC[index1]ACACTCTTTCCCTACACGAC-3’ primer Generic reverse: 5’CAAGCAGAAGACGGCATACGAGAT[indexe2]GTGACYGGAGTTCAGACGTGT-3’

Sequencing data processing

The bioinformatics workflow used in this study was developed during a compost study by Mbareche et al. (2017). Briefly, after demultiplexing the raw FASTQ files, the reads generated from the paired-end sequencing using MOTHUR 1.35.1 were combined (Schloss

174

et al. 2009). Quality filtering was performed using MOTHUR by discarding reads with ambiguous sequences. Reads shorter than 100 bp and longer than 450 bp were also discarded. Similar sequences were combined to reduce the computational burden, and the number of copies of the same sequence was displayed. This dereplication step was performed using USEARCH (version 7.0.1090; Edgar 2010). The selected region of fungal origin was then extracted from the sequences with ITSx, which uses HMMER3 (Mistry et al. 2013) to compare input sequences against a set of models built from a number of dif-ferent internal transcribed spacer (ITS) region sequences found in various organisms. Only the sequences belonging to fungi were kept for further analyses. Operational taxonomic units (OTUs) with a 97% similarity cutoff were clustered using UPARSE (version 7.1; Edgar 2013). UCHIME was used to identify and remove chimeric sequences (Edgar 2011). QIIME (version 1.9.1; Caporaso et al. 2010) was used to assign taxonomy to OTUs based on a UNITE fungal ITS reference training data set for taxonomic assignment and to generate an OTU table. The fungal diversity analysis was achieved by using different QIIME scripts (http://qiime.org/scripts/).

Statistical analysis

Descriptive statistics were used on sequencing data to high-light significant differences in the observed OTUs showed with box plots. The normality was verified by the D’Agostino-Pearson omnibus normality test. The normal-ity assumption on data was not fulfilled. Nonparametric Mann-Whitney U test analyses were performed to highlight significant differences showing a P value of less than 0.05. The results were analyzed using the software GraphPad Prism 5.03 (GraphPad Software, La Jolla, CA, USA).

To determine the statistical significance of the variation observed with the principal coordinates analysis (PCoA), a permutational multivariate analysis of variance (PERMANOVA) test was performed on the Bray-Curtis dissimilarity matrix. The QIIME script of the compared categories was used to generate the statistical results.

175

5.6 Results

Concentrations of fungi in biomethanization facilities were quantified per cubic meter of air. Penicillium/ Aspergillus spp. and Aspergillus fumigatus were the microorganisms selected for qPCR-based analyses because of their potentially hazardous effects. Concentrations of Penicillium/Aspergillus spp. ranged from 6.4 × 102 to 1.2 × 104 ITS genes/m3 (Figure 5.1). No significant difference was observed between the two biomethanization facilities. However, all of the measured concentrations were higher during summer com-pared with winter for both facilities. For BF1, Reception, Storage, and Output sites represented the highest concentrations of Penicillium/Aspergillus spp. (104 ITS genes/m3). For BF2, similar concentrations were noted at the Reception/Shredding site and were higher than concentrations identified at the Mixing site. Aspergillus fumigatus was detected in samples from all of the sampling sites in both facilities (Figure 5.2). Concentrations ranged from 9.6 × 101 to 1.2 × 104 ITS genes/m3. Comparisons between the facilities during summer and winter showed trends similar to the Penicillium/Aspergillus spp. results. The highest concentrations (103 ITS genes/m3) of Aspergillus fumigatus were found in the Storage and Output sites in BF1 during summer. In BF2, the Reception/Shredding site had the highest concentration (1.2 × 104 ITS genes/m3) of Aspergillus fumigatus during summer. However, this same site exhibited a lower concentration (2.7 × 102 ITS genes/m3) during winter.

Differences between concentrations of Penicillium/Aspergillus spp. and Aspergillus fumigatus from each sampling site were approximately 1 log, which indicates a dominance of Aspergillus fumigatus in bioaerosol samples. Outdoor control samples collected during each visit were below the qPCR detection limit.

176

Figure 5.1: Concentrations of Penicillium/Aspergillus spp. (Pen-Asp) in the air at sampling sites from both biomethanization facilities during summer and winter. The Storage site was not sampled during winter due to cold temperatures. The detection limit was 3 × 101 Pen-Asp/m3

Figure 5.2: Concentrations of Aspergillus fumigatus spores in the air at sampling sites from both biomethanization facilities during summer and winter. The Storage site was not sampled during winter due to cold temperatures. The detection limit was 5 × 101 A.fumigatus/m3.

177

To further assess fungal bioaerosols, fungal communities were analyzed by Illumina Miseq sequencing of the ITS1 region of the fungal ribosomal RNA encoding genes. After quality filtering, dereplication, and chimera checking, 507,842 sequences were clustered into 5132 OTUs. In order to confirm that the sequencing depth was adequate to describe the fungal diversity at each of the sites, rarefaction analyses were performed using the observed OTUs alpha diversity metric. The lowest-depth sample parameter was used to determine the sequencing depth threshold for rarefaction analyses. Samples with a lower sequencing depth than the one determined were excluded from analyses. The higher the sequencing depth, the more likely diversity coverage will be attained. In this case, the sequencing depth was approximately 8000 sequences per sample. All samples met this criterion and were included in the analyses except the outdoor all control samples exhibited sequence numbers that were too low. The values shown in Figure 5.3 were calculated as follows: 10 values from 10–8000 sequences per sample were randomly selected. For each of these values, the corresponding number of OTUs observed was noted for all of the samples. Then, the average number of OTUs observed (±1 standard deviation) was calculated for each of the 10 values. The plateau of the curves in Figure 5.3 indicates efficient coverage of the fungal diversity, as no more OTUs were observed even with greater numbers of sequences per sample.

Figure 5.3: Rarefaction curves obtained from the number of observed OTUs and the sequences per sample for air samples from the two biomethanization facilities visited. An average of the OTUs observed in each sample was calculated for all the samples (±1 standard deviation).

178

A Chao1 index was used to estimate species richness and to make comparisons between the two biomethanization facilities in the summer and winter. Figure 5.4 shows greater species richness in air samples collected during summer compared with winter. However, the medians and standard deviations obtained (780 ± 425 for summer and 410 ± 364 for winter) show an important variation of species richness between the samples taken in both seasons.

Figure 5.4: Comparison of species richness estimator Chao1 index values from air samples from two biomethanization facilities collected during summer and winter. Statistical significance shown was by Mann-Whitney U test analyses.

After estimating the overall within-community diversity (alpha), many next- generation sequencing surveys of microbial communities aim to compare the composition of different groups of samples (beta). In the current study, the two variables that may be responsible for the variation in fungal communities between samples are season and type of biomethanization facility. Each of the plants sampled treats a different type of waste. Therefore, samples could be grouped by season (summer/winter) and by facility (BF1/BF2). Ordination and the hierarchical clustering of the samples were used to visually compare community composition. One of the techniques commonly used by microbial ecologists

179

relies on the creation of a dissimilarity matrix such as the Bray-Curtis index. This index was used to evaluate the distance, taken pairwise between samples. The index uses numbers between 0 and 1, where 0 means the two samples have the same composition and 1 means that they do not share any species. Because the Bray-Curtis dissimilarity matrix uses the absolute abundance of OTUs, it is necessary to use a rarefied OTU table as the input for the dissimilarity calculation. To evaluate ordination pat-terns, one of the most common methods used is principal coordinates analysis (PCoA). The dissimilarity matrix is used as an input for ordination calculation and clustering. Figure 5.5 shows the PCoA results obtained, with the samples colored according to season (Figure 5.5a) and the type of facility from which the samples were collected (Figure 5.5b). The season appears to have a more pronounced effect on the fungal composition of the samples than the type of facility visited. In Figure 5.5a, samples collected during summer (colored red) are clustered together compared with the more scattered samples taken during winter (colored in blue). However, no particular pattern was observed when the samples were colored according to the type of facility (BF1 in blue and BF2 in red). All the dots on the figure are randomly dispersed (Figure 5.5b). The same variables used for color clustering in the PCoA analyses were used for the PERMANOVA test for the statistical significance of the sample clustering. This nonparametric multivariate analysis of variance method separates the distance matrix among sources of variation to describe the robustness and significance that a variable has in explaining the variations observed between samples. It is based on the ANOVA experimental design but analyses the variance and deter-mine the significance by permutations, as it is a nonparametric test (Anderson et al. 2005). Whereas ANOVA/ MANOVA assumes normal distributions and a Euclidean distance, PERMANOVA can be used with any distance measure as long as it is appropriate to the data set. The PERMANOVA testing results concur with the color grouping observation made with the PCoA analyses. The season clustering had a significant P value of 0.001, and the type of facility variable had a P value of 0.108.

180

Figure 5.5: Principal coordinates analysis of air samples taken from two different biomethanization facilities visited during summer and winter. (a) Samples colored according to the season (winter in blue and summer in red). (b) Samples colored according to the facility visited (BF1 in blue and BF2 in red)

Taxonomic analysis allows for the identification of fungal communities present in the samples collected. Figure 5.6 shows the relative abundance of different taxa present in BF1 and BF2 during summer and winter. The taxonomic distributions represent the classes of fungi present at each facility, grouping all samples together and also separating them by season. Different fungal diversity profiles were observed in the air from both of the facilities visited. Agaricomycetes were the most abundant class in BF1, whereas Eurotiomycetes were dominant in BF2 air samples. This dominance in BF2 air samples was present during summer and winter, although summer samples were slightly more diverse than winter samples, overall. In samples from BF1, a more equal fungal distribution was observed compared with BF2 and more obvious differences were noted between summer and winter samples. Air from BF1 was composed of higher proportions of Agaricomycetes compared with other classes of fungi during summer and higher proportions of Eurotiomycetes compared with other classes

181

of fungi during winter. Additionally, the occurrence of unidentified fungi was much greater during winter compared with summer.

Figure 5.6: Average relative abundances of fungal classes in air samples collected from two different biomethanization facilities visited during summer and winter.

The relative abundance of taxa was analyzed more thoroughly by identifying the 20 most abundant genera present in the biomethanization facilities during summer and winter. In BF1, only eight fungi were present in air samples collected during both seasons (Figure 5.7). During summer, Hyphodontia, part of the Agaricomycetes class, represented 40% of all fungal genera present, followed by Davidiella representing 15% of genera. During winter, Penicillium and Talaromyces were most abundant, with relative abundances of 62% and 13%, respectively. In BF2, Penicillium was the dominant genera during summer and winter, with relative abundances of 82% and 90%, respectively (Figure 5.8). Air samples collected during summer were more diverse, as only 10 fungi were identified during winter compared

182

with the 20 identified during summer. The diversity profiles were larger than what are shown in the figures, but due to graphical limitations only the most abundant fungi are represented.

Figure 5.7: Venn diagram showing the 20 most abundant genera of fungi identified in air samples from BF1 during summer and during winter and those present during both seasons. The numbers are percentages of relative abundance. In the middle section, percentages of relative abundance of each fungus in summer and winter are separated by a vertical bar

183

Figure 5.8: Venn diagram showing the 20 most abundant genera of fungi identified in air samples from BF2 during summer and during winter and those present during both seasons. The numbers are percentages of relative abundance. In the middle section, percentages of relative abundance of each fungus in summer and winter are separated by a vertical bar.

5.7 Discussion

Managing organic putrescible residual material is one of the major challenges of this century, as human population is still growing and so is consumption. Biomethanization facilities use the composting process to make renewable energy from waste materials. Workers’ exposure to fungal bioaerosols in composting environments has been previously

184

investigated (Wéry, 2014; Goff et al., 2010; Mbareche et al., 2017). However, exposure in biomethanization plants is less known. This study evaluates fungal exposure in two different biomethanization facilities. Air samples were collected from areas where workers are at risk and were analyzed by combining qPCR and a next-generation sequencing approach to thoroughly describe bioaerosol fungal composition. The combination of qPCR and next- generation sequencing adds a quantitative aspect to the quality feature brought by sequencing.

Penicillium/Aspergillus qPCR assay is an effective tool for measuring total quantities of Aspergillus/ Penicillium conidia in aerosol samples (Haugland et al., 2004). Concentrations of these fungi revealed that there is a greater exposure risk during summer for both facilities due to input from the outside environment. The air surrounding a plant can influence bioaerosol compositions inside the plant depending on the type of sources that are present and the amount of air exchanged between the two environments. For example, BF2 had compost piles near the biomethanization plant, which may have influenced the fungal concentrations inside the facility. During winter, when there is less air exchange, quantities of fungi in aerosols were linked only to the organic waste treated. Concentrations of Penicillium and Aspergillus were similar in all work areas and in both biomethanization facilities. However, the zones of greatest concern are the Storage and the Output sites in BF1 and the Reception/Shredding site in BF2. The storage and out-put steps are both associated with the maturation pro-cess (the breakdown of material by microorganisms) of organic matter. And, during the reception and the shredding steps, workers come in direct contact with waste. Therefore, these steps are associated with the highest risks of fungal exposure in the two plants. Aspergillus fumigatus was specifically identified and quantified in areas of the biomethanization facilities where humans regularly work. This fungus is a known pathogenic agent that causes aspergillosis, allergic bronchopulmonary aspergillosis, and is associated with other pulmonary diseases (Dogra et al., 2016; Greenberger, 2002). Similar conclusions can be drawn from the Aspergillus fumigatus and Penicillium and Aspergillus assays, as summer samples showed higher concentrations of these fungi compared with winter samples. Also, the Output/Storage and Reception/ Shredding sites were the most exposed areas to Aspergillus fumigatus. qPCR analyses allowed the quantification of

185

potentially hazardous fungal spores in bioaerosols and helped identify the work areas where human exposure, and consequently health effects, should be considered.

A previous study (Mbareche et al., 2017) used a next-generation sequencing approach targeting the ITS1 genomic region to produce an in-depth assessment of the fungal composition of bioaerosols released from composting sites. The analysis identified variations in fungal communities associated with the type of waste treated. The same approach was applied in this study to describe the fungal communities present in the air from biomethanization facilities. As expected, variations were observed between the two plants. BF1 treats sludge from wastewater plants, and BF2 treats domestic waste. Although the season seemed to influence variations in fungal communities more than the type of facility visited, as seen in results from the PCoA analysis, these results are likely influenced by the different input sources during each season due to changes in airflow from the outdoor environment. These results were consistent with those obtained from composting sites (Mbareche et al., 2017), supporting the idea that the type of waste treated influences the fungal composition of the bioaerosols released.

The type of waste product being processed can influence the genera present in aerosol samples. Agaricomycetes are a group of fungi known for their role in wood-decaying activities and in ectomycorrhizal symbiosis (Hibbett and Matheny, 2009; Morgenstern et al., 2008). A large portion of agricultural planting material in the waste processed at BF1 might explain the presence of this class of fungi. Eurotiomycetes are a class of fungi linked to processes such as fermentation used in food processing. Many genera of this class are natural decomposers and are involved in food spoilage (Geiser et al., 2006; Pitt and Hocking, 2009). The presence of natural or processed foods (e.g., fruits, vegetables, dairy, etc.) at BF2 might explain the large abundance of Eurotiomycetes detected in the air there. Some fungal genera identified in this study are commonly found in waste treatment environments: Aspergillus, Penicillium, Davidiella, Alternaria, Talaromyces, Neurospora, Capnobotryella, and Fusarium (Mbareche et al., 2017; Wéry, 2014). Fungi detected in samples from summer and winter visits are ubiquitous in wood and plant debris, and several are crop pathogens. The differences in the relative abundance between the summer and winter visits can be explained

186

either by the direct influence of the source (organic waste present at the time of sampling) or by the outdoor environment.

Overall, the application of the next-generation sequencing approach revealed a large fungal diversity profile in bioaerosols released from biomethanization activities. The presence of a diverse portrait of fungi in air may represent a risk for workers who are exposed on a daily basis. In some cases, the allergen/infective activity of the fungi may not be known and can increase the risks to workers. More specifically, the following fungi detected are known allergens and/or are opportunistic pathogens: Aspergillus, Malassezia, Emericella, Fusarium, Acremonium, and Candida. Malassezia causes skin disorders and can lead to invasive infections in immunocompetent individuals (Vlegraki et al., 2015). Emericella is a taxon of teleomorphs related to Aspergillus. Species of this group are known agents of chronic granulomatous disease (CGD; Matsuzawa et al., 2010). Acremonium causes fungemia in immunosuppressed patients (Rodriguez and Ramos, 2014). Fusarium species are responsible for a broad range of health problems, from local and systemic infections to allergy-related diseases such as sinusitis, in immunodepressed individuals (Nucci and Anaissie, 2007).

This study has some limitations: it included only molecular techniques and did not include traditional culture methods used in earlier studies, which would have provided a possibility for comparison with previous studies in waste management environments. In addition, there is a limitation linked to the database used and its taxonomic nomenclature. Cladosporium is the anamorph form of Davidiella. In this study, Davidiella was commonly found during summer and winter in the two facilities visited. The nonpresence of Cladosporium in our samples may be due to the well-described problem of double nomenclature in mycology. For example, in UniProt, Cladosporium herbarum have the scientific name of Davidiella tassiana.

As biomethanization activities represent an eco-friendly and energy-effective alternative to waste management, it is crucial to evaluate the exposure risk for biomethanization workers. This study addresses the concerns and risks associated with fungal

187

exposure in bioaerosols released from biomethanization activities. The exposure risk to workers may increase, as the facilities included in this research will increase the amount of waste treated yearly. The information from this study can be used as a reference for minimizing occupational exposure in future biomethanization facilities.

5.8 Conclusion

This study highlights the importance of using a high-throughput sequencing method combined with a real-time qPCR assay for quantification and an in-depth characterization of fungal diversity in bioaerosols in order to assess occupational exposure. Based on the results of this investigation, the authors strongly recommend taking action to reduce workers’ exposure to these aerosols. Technical and organizational measures should be implemented in biomethanization facilities as a first resort. Such measures include better air exchange rates, better confinement, and source ventilation. If these initial measures cannot be taken, we recommend skin and respiratory protection for workers exposed on a daily basis as a means to reduce continuous expo-sure to harmful fungi present in bioaerosols. The broad spectrum of fungi detected in this study includes many know pathogenic agents, and adequate monitoring of exposure is necessary to diminish risks. This study supports the conclusion that there are fungal signatures associated with the type of waste treated. Additional studies should be conducted in any environment involving fungi susceptible to aerosolization in order to continue documenting fungal bioaerosols.

Acknowledgment

We are grateful to all of the employees of the biomethanization facilities who participated in this study. We are also grateful to the members of the IRSST who were in the field for their technical assistance. The authors are thankful to Amanda Kate Toperoff and Michi Waygood for English revision of the manuscript.

188

Funding

H.M. is a recipient of the FRQNT Ph.D. scholarship and received a short internship scholarship from the Quebec Respiratory Health Network. This work was supported by the Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail (2013-0013).

About the authors

Hamza Mbareche is a Ph.D. student in the Department of Biochemistry, Microbiology and Bioinformatics, Laval University, Quebec City, Quebec, Canada.

Marc Veillette is a researcher in the laboratory of Caroline Duchaine at the Centre de Recherche de l’Institut Universitaire de Cardiologie et de Pneumologie de Québec (CRIUCPQ), Quebec City, Quebec, Canada.

Marie-Ève Dubuis is a Ph.D. student in the Department of Biochemistry, Microbiology and Bioinformatics, Laval University, Quebec City, Quebec, Canada.

Bouchra Bakhiyi is a research scientist in the Department of Environmental and Occupational Health, School of Public Health, University of Montreal, Montreal, Quebec, Canada.

Geneviève Marchand is an associate professor in the Department of Environmental and Occupational Health, School of Public Health, University of Montreal, Montreal, Quebec, Canada, and a research scientist at the Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail (IRSST), Montreal, Quebec, Canada.

Joseph Zayed is an associate professor in the Department of Environmental and Occupational Health, School of Public Health, University of Montreal, Montreal, Quebec, Canada, and the leader of the IRSST Chemical and Biological Hazard Prevention research field.

189

Jacques Lavoie is an associate professor in the Department of Environmental and Occupational Health, School of Public Health, University of Montreal, Montreal, Quebec, Canada, and research scientist at the IRSST. Guillaume J. Bilodeau is a research scientist at the Canadian Food Inspection Agency (CFIA), Ottawa Plant Laboratory, Fallowfield, Ontario, Canada, and the head of the Pathogen Identification Research Laboratory.

Caroline Duchaine is a professor in the Department of Biochemistry, Microbiology and Bioinformatics, Laval University, Quebec City, Quebec, Canada, and senior researcher at CRIUCPQ where she leads the bioaerosols research group. She is also head of the Quebec Bioaerosols and Respiratory Viruses Strategic Group of the Quebec Respiratory Health Network.

5.9 Bibliography

Amon, T., Amon, B., Kryvoruchko, V., Zollitsch, W., Mayer, K., Bruber, L., 2007. Biogas production from maize and dairy cattle manure-Influence of biomass composition on the methane yield. Agric. Ecosyst. Environ. 118 (1-4):174-182.

Anastasi, A., Varese G.C., Marchisio, V.F., 2005. Isolation and identification of fungal communities in compost and vermicompost. Mycologia. 97(1):33-44.

Anderson, M.J., 2005. PERMANOVA: a FORTRAN computer program for permutational multivariate analysis of variance. Department of Statistics, University of Auckland, New Zealand.

Bouallagui, H., Y. Touhami, R. B. Cheikh and M. Hamdi. 2005. “Bioreactor performance in anaerobic digestion of fruit and vegetable wastes.” Process Biochemistry. Volume 40, Issue 3-4, Pages 989-995

Bünger, J., Schappler-Scheele, B., Hilgers, R., Hallier, E., 2007. A 5-year follow-up study on respiratory disorders and lung function in workers exposed to organic dust from composting Plants. Int. Arch. Occup. Environ. Health. 80(4):306–312.

Bünger, J., Antlauf-Lammers, M., Schulz, T.G., Westphal, A.G., Müller, M.M., Ruhnau, P., et al., 2000. Health complaints and immunological markers of exposure to bioaerosols among biowaste collectors and compost workers. Occup. Environ. Med. 57(7):458–464.

190

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello E. K., et al., 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods. 7(5): 335-336.

Collagher, M., Campbell, J., Brenner, A.M., 2017. Collaboration achieves effective management design at Brookfield Place Berth, Western Australia. Procedia Eng. 180:1763– 1772.

Dai, X., Hu, C., Zhang, D., Chen, Y., 2017. A new method for the simultaneous enhancement of methane yield and reduction of hydrogen sulfide production in the anaerobic digestion of waste activated sludge. Bioresour .Technol. 243:914-921.

De Gannes, V., Eudoxie, G., William, H.J., 2013. Insights into fungal communities in composts revealed by 454-pyrosequencing: implications for human health and safety. Front. Microbiol. 4:164.

Dogra, V., Sinha, A.K., Saxena, R., Talwar, D., 2016. Aspergillus march: from ABPA to aspergilloma to subacute invasive aspergillosis. Allergy Asthma Clin. Immunol. 12:64.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 27(16):2194-200.

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 26(19):2460-2462

Edgar, R.C., 2013. UPARSE: Highly accurate OTU sequences from microbial amplicon reads. Nat. Methods. 10(10):996-8.

El-Mashad, H.M., Zeeman, G., Van Loon, W.K.P., Bot, G.P.A, Lettinga, G., 2003. Effect of temperature and temperature fluctuation on thermophilic anaerobic digestion of cattle manure. 95(2):191-201.

Epstein, E., Wu, N.C., Youngberg, C., Croteau, G., 2001. Dust and bioaerosols at a biosolids composting facility. Compost Sci. Util. 9(3): 250-255.

Floudas, D., Binder, M., Riley, R., Barry, K., Blanchette, R.A., Henrissat, B., et al. 2012. The Paleozoic Origin of Enzymatic Lignin Decomposition Reconstructed from 31 Fungal Genomes. Science. 336(6089):1715-1719.

Fuchs, J.G., Bieri, B., 2000. New biotests to measure the biological qualities of composts. Agrarforschung Schweiz 7(7):314-319.

Fung, F., Clark, R.F., 2004. Health Effects of Mycotoxins: A Toxicological Overview.

191

Journal of Toxicology. Clin. Toxicol. 42(2):217–234.

Geiser, D.M., Gueidan, C., Miadlikowska, J., Lutzoni, F., Kauff, F., Hofstetter, V., et al., 2006. Eurotiomycetes: Eurotiomycetidae and Chaetothyriomycetidae. Mycologia. 98(6):1053-1064.

Greenberger, P.A. 2002. Allergic bronchopulmonary aspergillosis. Journal of Allergy and Clinical Immunology. 110(5): 685-692.

Haitjema, C.H., Solomon, K.V., Henske, J.K., Theodorou, M.K., O’Malley, M.A., 2014. Anaerobic gut fungi: advances in isolation, culture, and cellulolytic enzyme discovery for biofuel production. Biotechnol. Bioeng.111(8):1471–1482.

Hansgate, A.M., Schloss, P.D., Hay, A.G., Walker, L. P., 2005. Molecular characterization of fungal community dynamics in the initial stages of composting. FEMS Microbiol. Ecol. 51(2):209–214.

Haugland, R.A., Varma, M., Wymer, L.J., Vesper, S.J., 2004. Quantitative PCR analyses of selected Aspergillus, Penicillium and Paecilomyces species. Syst. Appl. Microbiol. 27(2):198-210.

Hibbett, D.S., Matheny, P.B., 2009. The relative ages of ectomycorrhizal mushrooms and their plant hosts estimated using Bayesian relaxed molecular clock analyses. BMC Biol. 7:13.

Hoorman, J.J., 2011. The role of soil fungus. Fact sheet. Agricultural and Natural Resources. Ohio State University. SAG-14-11.

Hoornweg, D., Bhada-Tata, P., 2012. What a Waste: A Global Review of Solid Waste Management. Urban development series; knowledge papers no. 15. World Bank, Washington, DC, USA.

Ishii, K., Fukui, M., Takii, S., 2000. Microbial succession during a composting process as evaluated by denaturing gradient gel electrophoresis analysis. J. Appl. Microbiol. 89(5):768– 777.

Karakurt, I., Aydin, G., Aydiner, K., 2011. Mine ventilation air as a sustainable energy source. Renewable and Sustainable Energy Reviews. 15(2):1042-1049.

Kazda M., Langer, S., Bengelsdorf, F.R., 2014. Fungi open new possibilities for anaerobic fermentation of organic residues. Energy Substain. Soc. 4:6.

Langarica-Fuentes, A., Zafar, U., Heyworth, A., Brown, T., Fox, G., Robson, D.G., 2014. Fungal succession in an in-vessel composting system characterized using 454 pyrosequencing. FEMS Microbial. Ecol. 88(2):296-308.

Latgé, J.P. 1999. Aspergillus fumigatus and Aspergillosis. Clin. Microbio. Rev. 12(2):310-

192

350.

Le Goff, O., Bru-Adan, V., Bacheley, H., Godon, J.J., Wéry. N., 2010. The microbial signature of aerosols produced during the thermophilic phase of composting. J. Appl. Microbio. 108: 325–340.

Lee, S.Y., Holder, G.D., 2001. Methane hydrates potential as a future energy source. Fuel Process Technol. 71(1-3):181–186

Lim, S.L., Lee, L.H., Wu, T.Y., 2016. Sustainability of Using Composting and Vermicomposting Technologies for Organic Solid Waste Biotransformation: Recent Overview, Greenhouse Gases, Emissions and Economic Analysis. J. Clean. Prod. 111(Part A):262-278.

Macias-Corral, M., Samani, Z., Hanson, A., Smith, G., Funk, P., Yu, H., et al.,2008. Anaerobic digestion of municipal solid waste and agricultural waste and the effect of co- digestion with dairy cow manure. Bioresour. Technol. 99(17):8288-8293.

Mahajan, J., Vakhariya, J.A., 2016. Waste management: A reverse supply chain perspective. vikalpa. 41(3):197–208.

Mata-Alvarez, J., 2003. Biomethanization of the organic fraction of municipal solid waste. IWA publishing. London, UK.

Mbareche, H., Veillette, M., Bonifait, L., Dubuis, M.E., Bernard, Y., Marchand, G., et al., 2017. A next generation sequencing approach with a suitable bioinformatics workflow to study fungal diversity in bioaerosols released from two different types of compost plants. Sci. Total Environ. 601-602:1306-1314.

Mistry, J.,. Finn, R.D ., Eddy, S.R., Bateman, A., Punta, M., 2013. Challenges in Homology Search: HMMER3 and Convergent Evolution of Coiled-Coil Regions. Nucleic Acids Res. 41(12): e121, 2013.

Morgenstern, I., Klopman, S., Hibbett, D.S., 2008. Molecular evolution and diversity of lignin degrading heme peroxidases in the Agaricomycetes. J. Mol. Evol. 66(3): 243.

Murto, M., Björnsson, L., Mattiasson, B. 2004. Impact of food industrial waste on anaerobic co-digestion of sewage sludge and pig manure. Journal of Environmental Management. 70(2):101-107.

Nucci, M., Anaissie, E., 2007. Fusarium infections in immunocomprised Patients. Clin. Microbiol. Rev. 20(4):695-704.

193

Persoons, R., Parat, S., Stoklov, M., Perdrix, A., Maitre, A., 2010. Critical working tasks and determinants of exposure to bioaerosols and MVOC at composting facilities. Int. J. Hyg. Environ. Health. 213(5):338-247.

Pieckova, E., Wilkins, K., 2004. Airway toxicity of house dust and its fungal composition. Ann. Agric. Environ. Med. 11(1):67–73

Pitt, J. I., Hocking, A.D., 2009. The ecology of fungal food spoilage. Book: Fungi and Food Spoilage. Springer, Boston, MA, USA.

Rodriguez, Z.C., Ramos, M.G., 2014. Acremonium species associated fungemia: a novel pathogen in the immunosuppressed patient. Bol. Asoc. Med. P. R. 106(3):29-31.

Sánchez-Monedero, M.A., Stentiford, E.I.,. Urpilainen, S.T. 2005. Bioaerosol generation at large-scale green waste composting plants. J. Air Waste Manag. Assoc. 55(5):612-618.

Selman, M., Lacasse, Y., Pardo, A., Cormier, Y., 2010. Hypersensitivity pneumonitis caused by fungi. Proc. Am. Thorac. Soc. 7(3):229–236.

Sosnowski, P., Wieczorek, A., Ledakowicz, S., 2003. Anaerobic co-digestion of sewage sludge and organic fraction of municipal solid wastes. Adv. Environ. Res. 7(3):609-616.

Taha, P.M.P.,. Pollard, S.J.T., Sarkar, V., Longhurst, P., 2005. Estimating fugitive bioaerosol releases from static compost windrows: Feasibility of a portable wind tunnel approach. Waste manag. 25(4):445-450

194

United Nations Department of Social and Economic Development, Sustainable Development Knowledge Platform Goal 11 Make Cities and Human Settlements Inclusive, Safe, Resilient and Sustainable, United Nations, n.d. [Online]. Available: https://sustainabledevelopment.un.org/sdg11. [Accessed 1-08-2017]

Velegraki, A., Cafarchia, C., Gaitanis, G., Iatta, R., Boekhou, T., 2015. Malassezia infections in humans and animals: pathophysiology, detection, and treatment. PLoS Pathog. 11(1): e1004523.

Wéry, N. 2014. Bioaerosols from composting facilities—a review. Front. Cell. Infect. Microbiol. 4:42.

Wyngaarden, J.B., Smith, L.H., Bennett, J.C. (ed.). 1992. Cecil textbook of medicine, 19th ed., p. 399 and 407. W.B. Saunders Co., Philadelphia, Pa, USA.

195

Chapter 6: Fungal aerosols at dairy farms using molecular and culture techniques

6.1 Résumé

Les travailleurs des fermes laitières sont exposés à des niveaux élevés de moisissures dans l’air. L’application des méthodes de culture uniquement pourrait induire un biais dans l’étude de la diversité des moisissures. Le but de cette étude est de caractériser l'exposition aux moisissures dans les fermes laitières à l'aide de méthodes de qPCR et de séquençage à haut débit. Les concentrations minimales et maximales de Penicillium/Aspergillus allaient de 4,6 × 106 à 9,4 × 106 copies de gènes/m3 et de 1 × 104 copies de gènes/m3 à 4,8 × 105 copies de gènes/m3 pour Aspergillus fumigatus. Le large spectre de moisissures détectées dans cette étude inclut de nombreux agents pathogènes connus tels que Acremonium, Alternaria et Fusarium. Une évaluation de l'exposition aux bioaérosols est nécessaire pour minimiser les risques. La présence d'un portrait aussi diversifié de moisissures dans l'air peut représenter un risque pour la santé des travailleurs exposés quotidiennement.

196

6.2 Summary of the Paper

With this paper, the authors wished to apply the methodology developed so far on the characterization of fungi in aerosols on another environment affected by fungal exposure. In addition, the authors wished to evaluate the intake of culture methods on the study of fungal diversity in aerosols. Dairy farms are important sources of fungal aerosols because of the presence of hay and straw, which are naturally colonized by fungi. The sampling campaign consisted of collecting air samples from five different farms. Similarly to the biomethanization project, the authors applied a specific qPCR targeting Penicillium/Aspergillus and A. fumigatus combined to HTS targeting ITS for biomass quantification and diversity analyses. The same bioinformatic protocol as the compost, and biomethanization studies was again used for validation. To add more value to the study, mycobiota was compared according to variables like the type of animal feed, the type of ventilation, and the animal space.

Concentrations of Penicillium and Aspergillus reached a maximum value of 106 gene/m3 of air. The taxonomic profiles obtained from the five dairy farms were different, proving that the method is able to capture important players in the fungal community, and that fungal composition may vary depending on different conditions of each farm. Many opportunistic fungi were identified in air samples of dairy farms, suggesting a potential harm on daily exposed individuals. The intake of culture methods on fungal diversity showed a complementarity of both approaches in the most abundant taxa. In other words, when the most abundant fungi were considered HTS and culture gave complementary taxonomic profiles. However, when all the taxa were considered, HTS undoubtedly outperformed the culture approach in identifying fungi in aerosols from dairy farms.

This paper serves as an endorsement of the methodology applied in terms of fungal analyses in aerosol samples, from the sample treatment to the diversity analyses.

197

Fungal aerosols at dairy farms using molecular and culture techniques

RUNNING TITLE Fungal Aerosols at Dairy Farms AUTHORS Hamza Mbareche1,2, Marc Veillette1, Guillaume J Bilodeau3 and Caroline Duchaine1,2 AUTHORS’ AFFILIATION 1. Centre de recherche de l’institut universitaire de cardiologie et de pneumologie de Québec, Quebec City (Qc), Canada 2. Département de biochimie, de microbiologie et de bio-informatique, Faculté des sciences et de génie, Université Laval, Quebec City (Qc), Canada 3.Pathogen Identification Research Lab, Canadian Food Inspection Agency (CFIA). Ottawa, Canada KEYWORDS Bioaerosols, fungi, dairy farms, high-throughput sequencing, culture dependent, Occupational exposure

PUBLISHED: Sci Total Environ. 2019 Feb 25;653:253-263. doi:10.1016/j.scitotenv.2018.10.345

198

6.3 Abstract

Occupational exposure to harmful bioaerosols in industrial environments is a real threat to the workers. In particular, dairy-farm workers are exposed to high levels of fungal bioaerosols on a daily basis. Associating bioaerosol exposure and health problems is challenging and adequate exposure monitoring is a top priority for aerosol scientists. Using only culture-based tools does not express the overall microbial diversity and underestimate the large spectrum of microbes in bioaerosols and therefore the extended fungal profile that farmers are exposed to. The aim of this study was to provide an in-depth characterization of fungal exposure at Eastern Canadian dairy farms using qPCR and high-throughput sequencing methods. Specific primers were used for the quantification of Penicillium/Aspergillus and Aspergillus fumigatus in dairy farms air samples. Illumina Miseq sequencing of the ITS1 region provided sequences for the diversity analyses. The minimum and maximum concentration of Penicillium/Aspergillus ranged from 4.6 × 106 to 9.4 × 106 gene copies/m3 and from 1 × 104 gene copies/m3 to 4.8 × 105 gene copies/m3 for Aspergillus fumigatus, respectively. Differences in the diversity profiles of the five dairy farms support the idea that the novel approach identifies a large number of fungal taxa. The most striking differences include Microascus, Piptoporus, Parastagonospora, Dissoconium, Microdochium, Tubilicrinis, Ganoderma, Ustilago, Phlebia and Wickerhamomyces. The presence of a diverse portrait of fungi in air may represent a health risk for workers who are exposed on a daily basis. The broad spectrum of fungi detected in this study includes many known pathogens like Aspergillus, Acremonium, Alternaria and Fusarium. Adequate monitoring of bioaerosol exposure is necessary to evaluate and minimize risks.

6.4 Introduction

Exposure to airborne microbial flora or bioaerosols in the environment, whether from indoor or outdoor sources, is an everyday phenomenon that may lead to a wide range of human diseases. Compared to other well-described microbial habitats, such as water and soil, little is known about the diversity of airborne microbes (Venter et al., 2004; Delmont et al., 2011; Be et al., 2015). Whether aerosolized from natural sources (e.g., wind) or human

199

activities (e.g., industrial processes), the dispersal of bioaerosols can impact public health due to the presence of highly diverse and dynamic microbial communities in urban and rural environments. These impacts range from allergies to non-allergic asthma and can lead to exposure to pathogens (Brown and Hovmøller, 2002; Douwes, 2003; Brodie et al., 2007; Heederik and Von Mutius, 2012). Occupational exposure to harmful bioaerosols in industrial environments can be worrisome depending on the types of raw materials present, and the disturbance and the intensity of air movement and ventilation. For example, animal feeding operations involve various sources of biological material potentially associated with respiratory problems (Nehme et al., 2008; Gilbert and Duchaine, 2009; Milner, 2009; Létourneau et al., 2010; Lanier et al., 2010; Tsapko et al., 2011).

Fungal bioaerosols consist of spores, mycelium fragments and debris which are easily inhaled by workers and cause numerous symptoms including allergies, irritation and opportunistic infections. Long-term lung exposure to fungal bioaerosols can be associated with chronic diseases while the effects of short-term exposure range from irritation of the eyes and nose to coughing and a sore throat (Fung and Hughson, 2003; Eduard, 2008). Dairy- farm workers are exposed to high levels of fungal bioaerosols on a daily basis. In fact, concentration of culturable fungi in the air at dairy farms were reported to be higher than bacterial concentrations and may reach up to 1011 colony-forming units/m3 (Duchaine et al., 1999). At dairy farms, hay and straw are important sources of fungal bioaerosols, as fungi naturally colonize those substrates, especially if there are high moisture levels (Gregory et al., 1963; Festenstein et al., 1968; Kotimaa et al., 1991). Building type and management practices (e.g. free stall, use of various bedding materials, ventilation type) also influence the fungal load in bioaerosols.

The inhalation of large concentrations of fungal bioaerosols can lead to a variety of respiratory problems. The major allergy-related diseases caused by fungi are allergic asthma (Hardin et al., 2003; Bush et al., 2006; Porter et al., 2009), allergic rhinitis (Arshad et al., 2001; Stark et al., 2005), allergic sinusitis (Glass and Amedee, 2011), bronchopulmonary mycoses (Sarkar et al., 2010; Chowdhary et al., 2014), and hypersensitivity pneumonitis (Wyngaarden et al., 1992; Pieckova and Wilkins, 2004; Selman et al., 2012). The latter

200

includes farmer's lung disease (allergic alveolitis), a disease specific to dairy farm workers (Madsen et al., 1976; Cormier et al., 1985). Furthermore, a component of the fungal cell wall ((1–3)-β-D glucan), is believed to play a role in pulmonary inflammation, increased sensitivity to endo-toxins and pulmonary embolisms (Fogelmark et al., 1994; Rylander, 1996; Zekovic et al., 2005). Some respiratory symptoms are also associated with fungal exposure including mucous membrane irritation syndrome, nasal congestion, sore throat, and irritation of the nose and eyes (Burge and Rogers, 2000; Daisey et al., 2003; Bush, 2008; Haleem Khan and Karuppayil, 2012).

The link between exposure to fungi and occupational diseases is often difficult to prove due to undocumented fungi in bioaerosols. This lack of information is primarily due to the methods used to describe fungi present in the workplace. In diversity studies, culture methods are associated with well-known biases as only the viable/culturable portion of the samples is represented. Using culture-independent molecular methods is a good solution for getting around the non-viable/ non-culturable limits of the commonly used culture-based methods. Molecular methods are based on the detection of the genetic material of organisms present in a given sample. Applying these methods to samples from composting and biomethanization environments allowed the identification and quantification of fungal bioaerosols present and a better understanding of human exposure (Mbareche et al., 2017; Mbareche et al., 2018a). In dairy farms, only culture-dependent methods have been used to assess occupational exposure or ambient fungal aerosols (Adhikari et al., 2004; Lee et al., 2007; Poperscu et al., 2011; Lanier et al., 2012).

While bioaerosols are a major concern for public health, accurately assessing human exposure is challenging. Highly contaminated environments, such as agricultural facilities, contain a broad diversity of aerosolized fungi that may impact human health. Effective bioaerosol monitoring is increasingly recognized as a strategic approach for achieving occupational exposure description. Workers exposure to diverse fungal communities is certain, as fungi are ubiquitous in the environments and the presence of potential sources increases their presence in the air. Applying new molecular approaches to describe occupational exposure is a necessary work around the traditional culture approaches and the

201

biases they introduce to such studies. The importance of the newly developed approach can help to prevent worker's health problems.

Because of the dearth of information about fungal diversity and concentrations in bioaerosols at dairy farms, the aim of this study was to provide an in-depth characterization of fungal exposure at Eastern Canadian dairy farms using qPCR and High-Throughput Sequencing (HTS) methods. Penicillium/Aspergillus spp. and Aspergillus fumigatus were the microorganisms selected for qPCR-based analyses because of their potentially hazardous effects (Fung and Hughson, 2003).

6.5 Materials and Methods

Environmental field samples

Indoor air samples were collected from five dairy farms in Eastern Canada during summer 2016. All the dairy farms were located within a distance of 100 km. At each farm, a sampling site was designated based on where activities that generate the most bioaerosols took place. The buildings at each farm exhibited differences in building type and characteristics (age, volume, ventilation), number of animals present (cows), methods of milking (automatic or manual) and types of animal feed animal were given. Table 6.1 presents a description of the sampling sites at each dairy farm. At each sampling site, three air samples representing biological replicates were collected during the morning milking activity, when workers are exposed to the most bioaerosols. However, only one replicate in each dairy farm was used for library preparation and sequencing. Thus, the five sequencing datasets correspond to the five dairy farms visited. A total of 15 samples were collected for the five dairy farms visited between 31st May and 5th July 2016. Although the season would definitely affect the fungal diversity, this study does not intend to show a timescale variation of fungi in dairy farms during different seasons. The aim is to provide information about the type of fungal exposure the dairy farmers face during the busiest time of their day. The manager of each farm was contacted by phone prior to the sampling to inform the authors

202

about the farm activities. All the farmers were most active during the morning milking activity. An outdoor control air sample was also collected outside the dairy farms.

Table 6.1: Description of the sampling sites and the parameters affecting the sampling environments.

Type of Animal Cattle feed Ventilation Temperature Date of sampling Time of sampling

milking space

DF1 manual confined forage natural 22°C May 31st 6 am – 9 am

DF2 automatic confined forage mechanical 21°C June 7th 7 am – 10 am

DF3 manual confined concentrates natural 19°C June 14th 7 am – 10 am

DF4 automatic confined concentrates mechanical 20°C June 21th 6 am – 9 am

& forage

DF5 manual semi- forage mechanical 23°C July 5th 11 am – 2 pm

confined

Air sampling

A liquid cyclonic impactor Coriolis µ® (Bertin Technologies, Montigny-le- Bretonneux, France) was used for collecting air samples. The sampler was set at 200 l/min for 10 min (2 m3 of air per sample) and placed within 1–2 m of the source. The air flow in the sampler creates a vortex through which air particles enter the Coriolis cone and are impacted in the liquid. Fifteen millilitres of a phosphate buffer saline (PBS) solution with a concentration of 50 mM and a pH of 7.4 were used to fill the sampling cone.

203

Culture-based approach to study fungal diversity

One millilitre of the 15 ml Coriolis sampling liquid was used to per-form a serial dilution from 100 to 10−4 concentration/ml. The dilutions were made using 0.9% saline and 0.1% Tween 20 solution and were per-formed in triplicate. Tween 20 is a detergent that makes spores less hydrophobic and easier to collect. One hundred microlitres of each triplicate were plated on Rose Bengal Agar with chloramphenicol at a concentration of 50 µg/ml. Half of the petri dishes were incubated at 25 °C for mesophilic mould growth and the other half at 50 °C for thermophilic mould growth, specifically Aspergillus fumigatus. After 5 days of incubation, moulds were identified and counts were translated into CFU/m3.

Identification of isolates

Spores from cultured fungi were recovered in one millilitre of a 0.9% saline and 0.1% Tween20 solution and stored in an Eppendorf tube. Two hundred microlitres of the collection liquid were placed in an FTA card (sample collection card; Qiagen, Mississauga, Ontario, Canada). Five punches from the inoculated zone of the FTA card were placed in a microtube and washed three times with the FTA purification agent. The washing step is mandatory as it allows the removal of the chemical substrates in the FTA card that may alter the subsequent amplification step. Forty-eight microlitres of the master mix solution described in Supplementary File 1 were placed in each microtube followed by amplification and sequencing of the ITS genomic region. The protocol de-scribed by White et al. (1990) was performed at the CHU (Centre hospitalier de l'Université Laval). The following oligonucleotides were used for the ITS region amplification:

ITS1: 5′-TCCGTAGGTGAACCTGCGG-3′

ITS4: 5′-TCCTCCGCTTATTGATATGC-3′

The identification of the isolates was made by comparing the sequences obtained with sequences in the UNITE database.

204

Fungal spore concentration by filtration

The following methods are described in detail by Mbareche et al. (2018b). Briefly, a 2.5 cm polycarbonate membrane (0.2-mm pore size; Millipore) was used to filter the 45 ml Coriolis suspension through using a vacuum filtration unit. The filters were placed in a 1.5 ml Eppendorf tube with 750 µl of extraction buffer (bead solution) from a MoBio PowerLyser® Powersoil® Isolation DNA kit (Carlsbad, CA, U.S.A) and a 0.3 cm tungsten bead. Dry ice and 99% ethanol solution were used to freeze instantly the Eppendorf tube. The frozen filters were then pulverized in the Eppendorf tube in a bead-beating machine (a Mixer Mill MM301, Retsch, Düsseldorf, Germany) set at a frequency of 20 movements per second for 20 min. The pulverization was achieved by the tungsten bead hitting the frozen filter. The pulverized filter particles contained in the Eppendorf tube were used as aliquot for the first step of the DNA extraction procedure of MoBio PowerLyser® Powersoil® Isolation DNA kit.

DNA extraction

A second bead-beating step at a frequency of 20 movements per sec-ond for 10 min using glass beads was performed to ensure that all of the cells were ruptured. Then, a MoBio PowerLyser® Powersoil® Isolation DNA kit (Carlsbad, CA, U.S.A) was used to extract the total genomic DNA from the samples following the manufacturer's instructions. After the DNA elution in a 100 µl buffer, DNA was stored at −20 °C until sub-sequent analyses.

Real-time PCR quantification

PCR was performed with a Bio-Rad CFX 96 thermocycler (Bio-Rad Laboratories, Mississauga, CANADA). The PCR mixture contained 2 µl of DNA template, 0.150 µM per primer, 0.150 µM probe, and 7.5 µl of 2× QuantiTect Probe PCR master mix (QuantiTect Probe PCR kit; Qiagen, Mississauga, Ontario, Canada) in a 15-µl reaction mixture. The results were analyzed using Bio-Rad CFX Manager software version 3.0.1224.1015 (Bio-

205

Rad Laboratories). Positive control and standard curves of both qPCR analyses (Penicillium/Aspergillus and Aspergillus fumigatus) was tuned using Aspergillus fumigatus. Supplementary File 2 presents the primers, probes and PCR protocol used in this study.

High-throughput sequencing

Amplification of the DNA marker ITS1, equimolar pooling and sequencing were performed at the Plateforme d'analyses génomiques (IBIS, Université Laval, Quebec City, Canada). Briefly, amplification of the targeted gene was performed using the sequence specific regions described by Tedersoo et al. (2015) using a two-step dual-indexed PCR approach specifically designed for Illumina instruments. First, the gene-specific sequence was fused to the Illumina TruSeq sequencing primers and PCR was carried out on a total volume of 25 µl of liquid made up of 1× Q5 buffer (NEB), 0.25 µM of each primer, 200 µM of each of the dNTPs, 1 U of Q5 High-Fidelity DNA polymerase (NEB) and 1 µl of template cDNA. The PCR started with an initial denaturation at 98 °C for 30 s followed by 35 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 10 s, extension at 72 °C for 30 s and a final extension step at 72 °C for 2 min. The PCR reaction was purified using an Axygen PCR cleanup kit (Axygen). Quality of the purified PCR products was verified with electrophoresis (1% agarose gel). Fifty to 100-fold dilution of this purified product was used as a template for a second round of PCR with the goal of adding barcodes (dual-indexed) and missing sequence required for Illumina sequencing. Identical cycling conditions were used for the second PCR with the exception of 12 cycles. The PCR reactions were purified as above, checked for quality on a DNA7500 Bioanlayzer chip (Agilent) and then quantified spectrophotometrically with a Nanodrop 1000 (Thermo Fisher Scientific). Barcoded Amplicons were pooled in equimolar concentration (85 ng/µl) for sequencing on the illumina Miseq. The oligonucleotide sequences used for amplification are presented in Supplementary File 3.

206

Bioinformatics workflow

The bioinformatics workflow applied to analyze sequences of this study was first used during a compost study by Mbareche et al. (2017). In brief, after demultiplexing the raw FASTQ files, the reads generated from the paired-end sequencing were combined using Mothur v 1.35.1 (Schloss et al., 2009). Mothur was used for quality filtering and discarding reads with ambiguous sequences. Reads shorter than 100 bp and longer than 450 bp were also discarded. Similar sequences were put together to reduce the computational burden, and the number of copies of similar sequences was displayed. After quality filtering, all singletons were excluded from the data set (Brown et al., 2015). This dereplication step was performed using USEARCH version 7.0.1090 (Edgar, 2010). The ITS1 fungal region was then extracted from the dataset with ITSx which uses HMMER3 (Mistry et al., 2013) to compare input sequences against a set of models built from a number of different ITS region sequences found in various organisms. Only the sequences belonging to the fungal domain were kept for further analyses. A 97% threshold was used to cluster OTUs (Operational Taxonomic Units) using UPARSE 7.1 (Edgar, 2013). UCHIME was used to identify and re-move chimeric sequences (Edgar et al., 2011). QIIME version 1.9.1 (Caporaso et al., 2010) was used to assign taxonomy to OTUs based on the UNITE fungal ITS reference database for taxonomic assignment and to generate an OTU table. The fungal diversity analysis was achieved by using different QIIME scripts. The alpha and beta diversity scripts used are listed in the following link: http://qiime.org/scripts/.

Diversity analyses

Fungal communities were described by Illumina Miseq sequencing of the ITS1 region. After quality filtering, dereplication and chimera checking, 307,304 sequences were clustered into 188 OTUs. In order to confirm that the sequencing depth was adequate to describe the fun-gal diversity at each of the sampling sites, rarefaction analyses were performed using the observed OTU alpha diversity metric. The lowest-depth sample parameter was used to determine the sequencing depth of the rarefaction analyses which was

207

approximately 40,000 sequences per sample. Samples with a sequencing depth lower than 40,000 were excluded from analyses. The higher the sequencing depth, the more likely it is that the true diversity of the fungi in aerosols is captured. All of the samples from the five dairy farms met this criterion and were included in the analyses, except the outdoor control samples that exhibited sequence numbers that were too low.

Numbers in the dissimilarity matrix are bound between 0 and 1, where 0 means the two samples have the same composition and 1 means that they do not share any species. The Bray-Curtis dissimilarity matrix uses absolute abundance of OTUs, therefore it is necessary to use a rarefied OTU table as the input for the dissimilarity calculation. One function of multivariate analyses is to represent inter-sample distances in a 2-dimensional space using ordination (Ramette, 2007). To evaluate ordination patterns, one of the most common methods used is the Principal coordinate analyses (PCoA). In this case, the input used for ordination calculation and clustering was the dissimilarity matrix calculated above. The matrix was transformed to coordinates and then plotted using the principal coordinates script in QIIME.

Statistical analyses

Concentrations of PenAsp were compared with the Kruskal-Wallis one-way analysis of variance. The test was performed using the soft-ware R version 3.3.2 with RStudio Version 0.99.486. The same analysis was performed comparing concentrations of Aspergillus fumigatus.

To determine the statistical significance of the variance observed in the PCoA analyses, a PERMANOVA test was performed on the Bray-Curtis dissimilarity matrix. This nonparametric test allows for the analysis of the strength that each variable have in explaining the variations observed between samples (sample clustering). It is based on the ANOVA experimental design but analyzes the variance and determines the significance using permutations, as it is a non-parametric test (Anderson, 2005). Whereas ANOVA/MANOVA

208

assumes normal distributions and a Euclidean distance, PERMANOVA can be used with any distance measure as long as it is appropriate to the dataset. The same variables used for color clustering in the PCoA analyses were used with the PERMANOVA test for statistical significance of sample clustering. The QIIME compare categories script was used to generate the statistical results. Results from the PERMANOVA are consistent with the color clustering observations made based on the PCoA analyses. Because PERMANOVA is a non-parametric test, significance is determined through permutations. The number of permutations used is 999. p-Value ≤0.05 was considered statistically significant. Four variables were chosen to examine the differences more closely. The use of multi-variate analyses, PCoA, coupled with a PERMANOVA test, offers a robust statistical significance of sample clustering using distance matrices. Both analyses (PCoA and PERMANOVA) resulted in the same conclusions in regards of sample clustering confirming their usefulness as tools to visualize and measure sample clustering. Detailed information about the performance of the test are presented in the multivariate section of the results.

Data availability

Raw sequence reads of every sample used in this study and that sup-port its findings have been deposited in the National Center for Biotech-nology Information (NCBI) under the BioProject ID: PRJNA473493 https://www.ncbi.nlm.nih.gov/bioproject/473493

6.6 Results

Concentrations of fungal aerosols

Blank filters were taken to the farms, handled out on the field but did not undergo air sampling. They underwent the same extraction and PCR conditions as field samples. No detection was observed using the PenAsp qPCR protocol in three out of five of our blanks. The two positive blanks were under the detection limit. Concentrations of fungal bioaerosols using culture methods to capture the viable fungi and qPCR for DNA quantification of

209

Penicillium/Aspergillus genera and Aspergillus fumigatus species are shown in Fig. 6.1. Using culture methods, the concentrations of culturable fungi ranged from 3.2 × 106 to 8.2 × 106 CFU/m3 in samples from the five dairy farms (DF1 to DF5). The same tendency was observed between concentrations obtained by culture methods and those obtained by qPCR targeting Penicillium and Aspergillus (PenAsp). Concentrations of PenAsp ranged from 4.6 × 106 to 9.4 × 106 gene copies/m3 at the five dairy farms. Greater variance was observed in concentrations of Aspergillus fumigatus which, ranged from 1 × 104 gene copies/m3 at DF3 to 4.8 × 105 gene copies/m3 at DF2. Concentrations of Aspergillus fumigatus at DF1, DF4 and DF5 were 3 × 104, 2.9 × 104 and 3.9 × 105 gene copies/m3, respectively. The highest concentrations of PenAsp coincided with the highest concentrations of Aspergillus fumigatus as observed at DF2 and DF5 (Fig. 6.1). The gap be-tween the two concentrations was more notable in results from DF1, DF3 and DF4 where concentrations of Aspergillus fumigatus were lower.

Figure 6.1: Concentrations of viable spores of mesophilic fungi (from culture), Penicillium/Aspergillus (PenAsp from qPCR) and Aspergillus fumigatus (from qPCR) in air samples collected from five different dairy farms. Error bars represent SD calculated from the three biological replicates represented by the three air samples collected. The detection limit of the qPCR run was 2 × 103 Gene copies/m3 for PenAsp and 5 × 102 Gene copies/m3 for Aspergillus fumigatus.

210

Samples were separate by four categorical variables: Type of milking, animal space, cattle feed, and type of ventilation. Concentrations of PenAsp between groups of samples within each of those categories were compared. The same comparison was made for Aspergillus fumigatus concentrations. No significant differences (p < 0.05) in concentrations were found between the groups of samples for any of the four variables for either PenAsp or Aspergillus fumigatus (Table 6.2).

Table 6.2: Comparison of p-value of the concentrations obtained by qPCR between groups of samples (n=15) within four environmental factors using Kruskal-Wallis one- way analysis of variance.

Environmental factors PenAsp (p-value) Aspergillus fumigatus Animal Space 0.09 0.06 (confined vs semi confined Cattle Feed (forage vs 1 0.07 concentrates vs concentrates & forage) Type of Milking 0.74 0.07 (automatic vs manual) Type of Ventilation 0.8 0.08 (mechanical vs natural)

Diversity of fungal aerosols

The values shown in Fig. 6.2 were calculated following these steps: ten values from 10 to 40,000 sequences per sample were randomly selected. For each of these values the corresponding number of OTUs observed was noted for all of the samples. The plateaus observed in the five curves shown in Fig. 6.2 indicate an efficient coverage of the fungal diversity, as no more OTUs were observed even with much greater numbers of sequences per sample.

211

Figure 6.2: Rarefaction curves obtained from the number of observed OTUs and the sequences per sample for air samples from the five dairy farms visited. The plateau of the curves started at around 5000 sequences.

Many HTS surveys of microbial communities aim to compare the composition of different groups of samples (beta diversity). This multi-variate approach can be used to assess the effects of several environ-mental factors on the microbial content of the samples. The environmental factors, or “variables”, are used to separate the samples into different groups. In this case, the same four variables used to categorize qPCR concentrations were also used for the HTS multivariate analysis and included: Animal space, Cattle feed, Type of milking and Type of ventilation. A dissimilarity matrix was created based on Bray-Curtis index. This index was used to calculate distances between pairs of samples (representing how closely related samples are). Table 6.3 shows a summary of the results from the PCoA analyses (the PCoA figure is presented as a Supplementary File 4). The three principal coordinate axes captured > 90% of the variation in the DF samples. Samples were colored according to the four variables to visualize and identify sample clustering. Samples closer to one another are more similar than those that are further away from each other. No obvious sample clustering was observed for any of the four variables. Though they were not clearly clustered, calculations based on Animal space and Cattle feed were close together than the others. The

212

samples from confined spaces were grouped far from those from the semi-confined space. The forage samples were more closely grouped compared to the samples with concentrates and forage & concentrates combinations. No patterns were observed when samples were colored according to the Type of milking or Type of ventilation.

Using a significance of 0.05, the only variables that exhibited significant differences among sample groupings were Animal space (p-value = 0.04) and Cattle feed (p-value = 0.05). The two other variables tested did not exhibit significant differences (Type of milking p-value = 0.61 and Ventilation p-value = 0.90). It is important to note that only one sample correspond to the semi-confined group of the Animal space variable. Thus, the intra-group variability could not be taken in consideration in the statistical test.

Table 6.3: Summary of the parameters and results of the principal coordinates analysis of air samples collected from five dairy farms including the statistical significance of the sample clustering. The PCoA was calculated using the Bray-Curtis dissimilarity based on ITS1 sequences. The three principal coordinate axes captured over 90% of the variation in the input DF samples. The statistical significance of the variation observed with the PCoA analyses was determined using a PERMANOVA statistical test. Four environmental variables were used for sample clustering, only two were statistically significant (Animal space and Cattle feed).

Environmental factors Sample clustering PERMANOVA (p-value) Animal Space Ö 0.04 (confined vs semi confined Cattle Feed (forage vs Ö 0.05 concentrates vs concentrates & forage) Type of Milking X 0.61 (automatic vs manual) Type of Ventilation X 0.90 (mechanical vs natural)

213

Taxonomy of fungal aerosols

The taxonomy of the microbes in the air samples collected from the dairy farms was determined by comparing Illumina sequences to the UNITE database. Of the 12 fungal classes detected in samples from the dairy farms (Fig. 6.3) six classes seem to be dominant: Eurotiomycetes, Dothideomycetes, Wallemiomycetes, Agaricomycetes, Sordariomycetes and Tremellomycetes. However, there is variability in this dominance between the diversity profiles from the five dairy farms. At DF3 and DF5 the class Eurotiomycetes have much greater relative abundance than the other classes. In DF2 samples, Dothideomycetes and Wallemiomycetes are more abundant than the other classes. The Sordariomycetes class is particularly more abundant at DF1 compared to the other farms. Fungi from the class Tremellomycetes have greater relative abundance at DF4 than any of the other farms. In fact, DF4 has the most diverse profile, in contrast to samples from DF3 where the class Eurotiomycetes represents 70% of the relative abundance. Ustilaginomycotina were detected only in samples from DF4.

Figure 6.3: Relative abundance of fungi classes detected in air samples from five dairy farms using high-throughput sequencing. While 12 classes were detected, six dominate the diversity profiles. Those six classes have different distributions among the five dairy farms visited

214

Relative abundance of taxa was analyzed more thoroughly by examining the 20 most abundant genera at each dairy farm (Fig. 6.4). From this list, only six fungi were present at all five of the dairy farms: Aspergillus, Penicillium, Wallemia, Aureobasidium, Pleosporales and Tremellomycetes OTUs that were not identifiable to the genus level were identified to the highest taxonomic level (e.g. class Tremellomycetes and order Pleosporales). Similar to observations made based on fungal class, diversity profiles of the genera present were quite variable between the five farms. The least diverse profile was observed in samples from DF3 where Penicillium occupied 67% of the abundance. The most diverse pro-files were from DF1, DF4 and DF5 as they exhibited the greatest variety of fungal genera. In DF2 samples, 52% of the abundance was made up of Wallemia (31%) and Bipolaris (21%). The diversity profiles from the five dairy farms are larger than what is shown in Fig. 6.4. Due to graphical limitations, only the most abundant fungi are represented. Piptoporus and Microascus were identified only at DF1. Exobasidiomycetes, Microdochium, Dissoconium and Parastagonospora were present at DF2 exclusively. Tubilicrinis was detected only at DF3. Mycoacia, Phlebia, Ustilago and Ganoderma were identified solely at DF4. Finally, Whickerhamomyces was specific only to samples from DF5.

215

Figure 6.4: Relative abundance of fungal genera detected in air samples from five dairy farms. The 20 most abundant genera from each dairy farm were included in the analyses. The underlined bold fungi were common in all the dairy farms. The fungi that were detected only in one dairy farm have the DF identification after their name. Comparison of fungal aerosols detected by culture and HTS

216

The diversity of fungi identified using the culture method was com-pared with the fungal diversity obtained using HTS. Using HTS, fungal genera representing N1% of the total abundance of the five dairy farms combined are presented in Fig. 6.5. For species identified using the culture approach, the fungi identified at more than one dairy farm were grouped together. The fungi that were detected only once by culture were Trichoderma, Microdochium, Phoma, Apiospora, Botrytis, Conyothirium, Millerozyma, Neosetophoma, Irpex, and Debaryomyces. Those genera detected at more than one farm and their relative abundances are presented in Fig. 6.5. The relative abundance of fungi identified by culture was calculated as follows: for each fungus, the number of times that it was isolated from the five dairy farms was calculated. Based on this sum, a percentage of relative abundance was calculated for each fungus and appears in the list in Fig. 6.5. Only four fungi were detected by both approaches: Penicillium, Aspergillus, Bipolaris and Sarocladium. Of the 16 fungi isolated using culture techniques, three (Hyphopichia, Gibellulopsis and Myceliophtora) were not detected by HTS. The remaining 13, though they do not appear on the list, were detected with a total abundance of < 1%. Many fungi genera were present but with a total relative abundance of < 1% making the diversity profile more exhaustive than what is shown in the figures.

217

Figure 6.5: Relative abundance of fungal genera identified by high-throughput sequencing and culture in air samples collected from five dairy farms. Fungi in bold character are common to both approaches

218

6.7 Discussion

Concentrations of fungal aerosols

The PenAsp qPCR assay is a good indicator of the total quantities of Aspergillus, Penicillium and Paecilomyces conidia in air samples (Haugland et al., 2004). Results of this study showed a strong connection between concentrations of culturable fungi and PenAsp. This connection supports the idea of using the qPCR PenAsp assay as an indicator of total fungal concentration in exposure studies. No significant differences were observed in fungal concentrations obtained from the five dairy farms using qPCR and culture. These concentrations are comparable to concentrations obtained using culture methods al-most two decades ago from Eastern Canadian dairy farms (Duchaine et al., 1999; Blais Lecours et al., 2012). This suggests that dairy-farm workers are still at risk for developing diseases linked to fungal expo-sure. Furthermore, Aspergillus fumigatus was specifically quantified in aerosols from areas at dairy farms where humans work because it is a known pathogen that causes aspergillosis, allergic bronchopulmonary aspergillosis and is involved in other pulmonary diseases (Greenberger, 2002; Dogra et al., 2016). In some cases, the gap between the concentrations of PenAsp and Aspergillus fumigatus can be used as an indicator of the diversity of Aspergillus and Penicillium genera in air samples. The difference in concentration between Penicillium/Aspergillus (higher) and Aspergillus fumigatus (lower) can be a possible indication of the presence of other species than Aspergillus fumigatus. It is not a verified statement, but an interesting observation that adds up to the comparison of the two qPCR assays.

The qPCR analysis allowed the quantification of potentially hazardous fungal spores in bioaerosols. No particular correlation was found be-tween the types of ventilation, animal confinement, cattle feed and milking methods, and concentrations of PenAsp and Aspergillus fumigatus in aerosols from dairy farms. These results prove that no matter how different the building attributes, animal confinement and types of milking activities are, exposure to fungal bioaerosols should be considered regardless of the modernity of the method used.

219

Variations in the diversity of fungal aerosols

The Illumina MiSeq sequencing depth used in this study was adequate for covering the true diversity of fungi in the samples. Targeting the ITS1 genomic region provided an in- depth analysis of the fungal composition of bioaerosols at the five dairy farms. The methodology applied also revealed the variations in fungal communities present in the air (Mbareche et al., 2017; Mbareche et al., 2018a). Differences in the diversity profiles support the idea that this approach identifies a large number, if not all of the taxa that are responsible for the fungal community changes. These variations in diversity profiles may be explained by the presence of different and multiple sources of fungal bioaerosols at dairy farms. The main source of the variation in diversity is associated with cattle feed type. Dairy cattle are fed a wide range of feedstuffs, from forage (grasses, legumes, hay, straw, grass silage and corn silage) to concentrates (barley and maize). The presence of Ustilaginomycotina and Exobasidiomycetes could be explained by the presence of wheat and other grasses. These classes of fungi include the plant pathogen Tilletia known to affect various grasses. Biochemical changes in these products, like pH and water content, can affect their fungal composition (Seglar, 2013; Adams, 2017). Animal confinement also affected the fungal composition of bioaerosols. The semi-confined environment consists of an enclosure where dairy cattle have freedom to move around inside the enclosed space. The confined spaces allow no freedom of movement and each cow has its own space. These differences in the density of cows seem to have an impact on the fungal bioaerosols. The type of milking whether automated or manual, and the type of ventilation, either automatic or manual does not seem to have an effect on the fungal content of the bioaerosols collected. However, a limited number of dairy farms were visited during this study and multivariate analyses and sample clustering methods are known to perform better with a large number of samples. A larger number of air samples collected from different dairy farms would be useful to support the findings that milking method and or types of ventilation influences fungal bioaerosol variability. Also, sampling was done only once in each farm; multiple sampling during the year would capture the full extent of the exposure of farmers.

220

Other factors like building attributes, handling of feed, seed and silage, and method of spreading the bedding can affect the fungal content of the bioaerosols at dairy farms (Poperscu et al., 2011; Samadi et al., 2012; Garcia et al., 2013; Basinas et al., 2014). While PCoA gives a cursory assessment of the variables that affect sample clustering, these variables can often be harder to define. A set of chosen explanatory environmental factors does not guarantee that they have true explanatory power. There is always the possibility that an unexplored covariate is the real causal influence on the microbial ecology of the samples (Buttigieg and Ramette, 2014). Further research including larger sample sizes and additional variables should be conducted.

Agaricomycetes are a group of fungi known for their role in wood-decaying activities and in ectomycorrhizal symbiosis (Hibbett and Matheny, 2009; Morgenstern et al., 2008). The presence of agricultural planting material/products may explain the larger proportions of Agaricomycetes identified at DF1 and DF2 compared to the three other farms. Conversely, Eurotiomycetes are a class of fungi linked to processes like fermentation used in food processing. Many genera of this class are natural decomposers and are involved in food spoilage (Geiser et al., 2006; Pitt and Hocking, 2009). The presence of natural or processed foods at DF3, DF4 and DF5 might explain the greater abundance of Eurotiomycetes detected in the air at those farms. Additionally, the prevalence of Eurotiomycetes might also be explained by the presence of silage which is a fermented, high-moisture stored fodder used to feed cattle (Woolford and Pahlow, 1998). Members of Dothideomycetes and Tremellomycetes include several important plant pathogens that grow on wood debris or decaying leaves (Schoch et al., 2009; Hibbett et al., 2007). Wallemiomycetes were detected at all five dairy farms. They were most prevalent at DF1 and DF2, representing 20% and 32% of genera detected, respectively. This class includes one order (Wallemiales), containing one family (Wallemiaceae), which in turn contains one genus (Wallemia; Zalar et al., 2005).

These fungi can grow over a wide range of water activity from 0.69 aw to 0.997 aw (Pitt and Hocking, 1977). Water activity is the vapour pressure of water in the product divided by vapour pressure of pure water at the same temperature. High aw support more microbial growth. Wallemia have been isolated in air samples from dairy farms in previous studies (Hanhela et al., 1995). Airborne Wallemia are suspected of playing a role in human allergies

221

like bronchial asthma (Sakamoto et al., 1989). A study conducted in France identified Wallemia as a causative agent of farmer's lung disease (Reboux et al., 2001). Other prevalent fungal genera commonly found at dairy farms were identified in this study: Aspergillus, Penicillium, Cladosporium, Alternaria, Nigrospora and Periconia.

Comparison of fungal aerosols detected by culture and HTS

For relative abundance, differences observed in the diversity profiles obtained by HTS and culture methods may be explained by the hypothesis that the culture approach may be biased toward fungi from the rare biosphere. These results are consistent with the conclusions made by Shade and his collaborators (Shade et al., 2012) regarding the complementarity of culture-dependent and culture-independent approaches to studying bacterial diversity. The premise of their study is that culture-dependent methods reveal bacteria from the rare biosphere and provide supplemental information to that obtained using a HTS approach. In the current case, this complementarity is true only for abundance. As mentioned previously, only three fungi were detected exclusively by culture, while more than a hundred fungi were identified by HTS and not by culture. This is consistent with the concept that culture methods may reveal less abundant taxa in an environment while HTS provides a more exhaustive diversity profile. To the best of our knowledge this is the first research to compare both approaches for examining aerosols at dairy farms.

Impact of the study

The application of the HTS approach revealed a large fungal diversity profile in bioaerosols released from five dairy farms. All of the previous studies identified the same frequently encountered genera including Aspergillus, Penicillium, Cladosporium and Alternaria. In Canada, the last study that described the airborne fungal microflora in dairy farms is from 1999 (Duchaine et al., 1999). Although most isolates are commonly encountered and are not new, this is the first time the presence of a diverse portrait of fungi in air from dairy farms is demonstrated and it may represent a health risk for workers who

222

are exposed on a daily basis. In some cases, the allergen/infective activity of the fungi may not be known and can increase the risks to workers. More specifically, the following fungi detected are known allergens and/or are opportunistic pathogens: Aspergillus, Malassezia, Wallemia, Emericella, Fusarium, Alternaria and Candida. Malassezia causes skin disorders and can lead to invasive infections in immunocompetent individuals (Velegraki et al., 2015). Emericella is a taxon of teleomorphs related to Aspergillus. Species of this group are known agents of chronic granulomatous disease (CGD; Matsuzawa et al., 2010). Acremonium causes fungemia in immunosuppressed patients (Rodriguez and Ramos, 2014). Fusarium species are responsible for a broad range of health problems, from local and systemic infections to allergy-related diseases such as sinusitis, in immunodepressed individuals (Nucci and Anaissie, 2007). Alternaria is an important allergen related to asthma (Peat et al., 1993). Furthermore, mycotoxins, a secondary metabolite produced by fungi, can become airborne on conidia or smaller fragments suggesting a potential inhalation or ingestion by farm workers (Brochers et al., 2017). Mycotoxins are known to have adverse health effects like mycotoxycosis or being carcinogenic to humans. However, significant exposure to mycotoxins is more likely to occur in places where there is no regulation to protect exposed humans and where methods of handling food is a problem (Bennett and Klich, 2013). Respiratory health of farmers has been the subject of controversy with the hygiene hypothesis that stipulates that exposure to microbes resulting from intensive farming during early life could be beneficial to later health (Strachan, 1989; Stiemsma et al., 2015). However, this hypothesis is not unanimous in the scientific community (Bloomfield et al., 2016). In this study, taxa identified among farms could hypothetically play a role in the protective aspect of allergy and atopic disease development.

Comparison of fungal airborne microbiota using culture-independent techniques in other farms present different taxonomy results than dairy farms. For example, Sagenomella was the most abundant taxa in a poultry farm (Nonnemann et al., 2010) and Clavaria was the most abundant taxa in swine houses (Priyanka et al., 2016). These results confirm the hypothesis that airborne fungal diversity is influenced by numerous factors including the seasons, geographical locations and surrounding sources.

223

6.8 Conclusion

Bioaerosols from Eastern Canadian dairy farms contain high concentrations of highly diverse fungi. This study demonstrated that fungal bioaerosols have large diversity profiles. It also adds another piece to the puzzle regarding the complexity of bioaerosols relative to the sources present. This study also highlights the importance of using a high-throughput sequencing method combined with real-time PCR assay for quantification and an in-depth characterization of fungal diversity in bioaerosols to better assess occupational exposure. As air samples were collected during activities where workers are present, this study shows that human exposure to harmful fungi may be higher during milking activities (automatic or manual), as well as during handling of feed and silage and when spreading bedding. Based on the results of this investigation, the authors strongly recommend taking action to reduce workers' exposure to bioaerosols. Such measures include in-creased air-exchange rates, better confinement and source ventilation. The feasibility of the recommendations proposed depends on the infra-structure and the farming methods used by the farms and should be considered by the management team of each farm. If these measures cannot be applied, we recommend skin and respiratory protection for workers who are exposed on a daily basis as a means to reduce continuous exposure to harmful fungi present in bioaerosols. The broad spectrum of fungi detected in this study includes many know pathogens and proves that adequate monitoring of bioaerosol exposure is necessary to evaluate and minimize risks.

Acknowledgements

HM is a recipient of the FRQNT PhD scholarship and received a short internship scholarship from the Quebec Respiratory Health Network. This work was supported by the Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail (2014-0057). We are grateful to all of the employees of the dairy farms that participated in this study. We are also grateful to Julien Duchaine and Philippe Bercier that were in the field for their technical assistance. The authors are thankful to Amanda Kate Toperoff and Michi Waygood for English revision of the manuscript. CD is the head of the Bioaerosols and respiratory viruses strategic group of the Quebec Respiratory Health Network.

224

Supplementary Data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.scitotenv.2018.10.345.

Competing interests

The authors declare no competing financial interests.

Materials & correspondence

All material requests and correspondence should be addressed to Caroline Duchaine.

6.9 Bibliography

Adams, R.S., 2017. Mold and mycotoxin problems in livestock feeding. PennState Extension. Department of Veterinary Science. https://extension.psu.edu/mold-and- mycotoxin-problems-in-livestock-feeding Visited: 15 November 2017.

Adhikari, A., Reponen, T., Lee, S.A., Grinshpun, S.A., 2004. Assessment of human exposure to airborne fungi in agricultural confinements: personal inhalable sampling versus stationary sampling. Ann. Agric. Environ. Med.11(2):269-77.

Anderson, M.J., 2005. PERMANOVA: a FORTRAN computer program for permutational multivariate analysis of variance. Department of Statistics, University of Auckland, New Zealand.

Arshad, S.H., Tariq, S.M., Matthews, S., Hakim, E., 2001. Sensitization to common allergens and its association with allergic disorders at age 4 years: a whole population birth cohort study. Pediatrics. 108:E33–E37.

Basinas, I., Sigsgaard, T., Erlandsen, M., Andersen, N.T., Takai, H., Heederik, D., Omland, O., Kromhout, H., Schlünssen., 2014. Exposure-affecting factors of dairy farmers’ exposure to inhalable dust and endotoxin. Ann. Occup. Hyg. 58(6):707-23.

Be, N.A., Thissen, J.B., Fofanov, V.Y., Allen, J.E., Rojas, M., Golovko., G., et al., 2015. Metagenomic analysis of the airborne environment in urban spaces. Microb Ecol. 69(2):346– 355.

225

Bennett, J.W., Klich, M., 2013. Mycotoxins. Clin. Microbiol. Rev. 16(3):497-516.

Blais Lecours, P., Veillette, M., Marsolais, D., Duchaine, C., 2012. Characterization of bioaerosols from dairy barns: reconstructing the puzzle of occupational respiratory diseases by using molecular approaches. Appl. Environ. Microbiol. 78(9):3242-3248.

Bloomfield, S.F., Rook, G.A., Scott, E.A., Shanahan, F., Stanwell-Smith, R., Turner, P., 2016. Time to abandon the hygiene hypothesis: new perspectives on allergic disease, the human microbiome, infectious disease prevention and the role of targeted hygiene. Perspect. Public Health. 136(4):213–224.

Brochers, A.T., Chang, C., Gershwin, E., 2017. Mold and human health: a reality check. Clin. Rev. Allergy Immunol. 52(3):305-322.

Brodie, E.L., DeSantis, T.Z., Parker, J.P.M., Zubietta, I.X., Piceno, Y.M., Andersen, G.L., 2007. Urban aerosols harbor diverse and dynamic bacterial populations. Proc. Nat. Acad. Sci. U.S.A. 104(1):299-304.

Brown, J.K.M., Hovmøller, M.S., 2002. Aerial dispersal of pathogens on the global and continental scales and its impact on plant disease. Science. 297(5581):537–541.

Brown, S.P., Veach, A.M., Rigdon-Huss, A.R., Grond, K., Lickteig, S.K., Lothamer, K., Oliver, A.K., et al., 2015. Scraping the bottom of the barrel: are rare high-throughput sequences artifacts? Fungal Ecol. 13:221–225.

Burge, H.A., Rogers, C.A., 2000. Outdoor allergens. Environ. Health Perspect. 108(Suppl 4):653–659.

Bush, R.K., 2008. Indoor allergens, environmental avoidance, and allergic respiratory disease. Allergy Asthma Proc. 29(6):575–579.

Bush, R.K., Portnoy, J.M., Saxon, A., Terr, A.L., Wood, R.A., 2006. The medical effects of mould exposure. J. Allergy Clin. Immunol. 117(2):326–333.

Buttigieg, P.L., Ramette, A., 2014. A Guide to Statistical Analysis in Microbial Ecology: a community-focused, living review of multivariate data analyses. FEMS Microbiol. Ecol. 90(3):543–550.

226

Cormier, Y., Bélanger, J., Durand, P., 1985. Factors influencing the development of serum precipitins to farmer’s lung antigen in Quebec dairy farmers. Thorax. 40(2):138-142.

Daisey, J.M., Angell, W.J., Apte, M.G., 2003. Indoor air quality, ventilation and health symptoms in schools: an analysis of existing information. Indoor Air. 13(1):53-64.

Delmont, T.O., Robe, P., Cecillon, S., Clark, I.M., Constancias, F., Simonet, P., et al., 2011. Accessing the soil metagenome for studies of microbial diversity. Appl. Environ. Microbiol. 77(4):1315–1324.

Dogra, V., Sinha, A.K., Saxena, R., Talwar, D., 2016. Aspergillus march: from ABPA to aspergilloma to subacute invasive aspergillosis. Allergy Asthma Clin. Immunol. 12:64.

Douwes, J., 2003. Bioaerosol health effects and exposure assessment: progress and prospects. Ann. Occup. Hyg. 47(3):187–200.

Duchaine, C., Mériaux, A., Brochu, G., Cormier, Y., 1999. Airborne microflora in Quebec dairy farms: Lack of effect of bacterial hay preservatives. Am. Ind. Hyg. Assoc. J. 60(1):89- 95.

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 26(19):2460-2461

Edgar, R.C., 2013. UPARSE: Highly accurate OTU sequences from microbial amplicon reads. Nat. Methods. 10:996-998.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 27(16):2194-200.

Eduard, W., 2008. A health based criteria document on fungal spore exposure in the working population. Is it relevant for the general population? Indoor Air. 18(3):257-2588.

Festenstein, G., Lacey, J., Skinner, F.A., 1968. Self-heating of hay and grain in Dewar flask and the development of farmer’s lung antigens. J. Gen. Microbiol. 41(3):389–407.

Fogelmark, B., Sjöstrand, M., Rylander, R., 1994. Pulmonary inflammation induced by repeated inhalations of (1,3)-D-glucan and endotoxin. Int. J. Exp. Path. 75(2):85-90.

Fung, F., Hughson, W.G., 2003. Health effects of indoor fungal bioaerosol exposure. Appl. Occup. Environ. Hyg. 18(7):535-544.

Garcia, J., Benett, D.H., Tancredi, D., Schenker, M.D., Mitchell, D., Reynolds, S.J., Mitloehner, F.M., 2013. Occupational exposure to particulate matter and endotoxin for California Dairy workers. Int. J. Hyg. Environ. Health. 216(1):56-63.

227

Geiser, D.M., Gueidan, C., Miadlikowska, J., Lutzoni, F., Kauff, F., Hofstetter, et al., 2006. Eurotiomycetes: Eurotiomycetidae and Chaetothyriomycetidae. Mycologia. 98(6):1053- 1064.

Gilbert, Y., Duchaine, C., 2009. Bioaerosols in industrial environments: a review. Can. J. Civ. Eng. 36: 1873–1886.

Glass, D., Amedee, R.G., 2011. Allergic Fungal Rhinosinusitis: A Review. Ochsner J. 11(3):271–275.

Greenberger, P.A., 2002. Allergic bronchopulmonary aspergillosis. J. Allergy Clin. Immunol. 110(5):685-692.

Gregory, P.H., Lacey, M.E., Festenstein, G.N., Skinner, F.A., 1963. Microbial and biochemical changes during the molding of hay. J. Gen. Microbiol. 33:147–174.

Haleem Khan, A.A., Karuppayil, M.S., 2012. Fungal pollution of indoor environments and its management. Saudi J. Biol. Sci. 19(4):405-426.

Hanhela, R., Louhelainen, K., Pasanen, A.L., 1995. Prevalence of microfungi in Finnish cow barns and some aspects of the occurrence of Wallemia sebi and Fusaria. Scand. J Work Environ. Health. 21(3):223-228. Hardin, B.D., Kelman, B.J., Saxon, A., 2003. Adverse human health effects associated with moulds in the indoor environment. J. Occup. Environ. Med. 45:470–478.

Haugland, R.A., Varma, M., Wymer, L.J., Vesper, S.J., 2004. Quantitative PCR analyses of selected Aspergillus, Penicillium and Paecilomyces species. Syst. Appl. Microbiol. 27(2):198-210.

Heederik, D., Von Mutius, E., 2012. Does diversity of environmental microbial exposure matter for the occurrence of allergy and asthma? J. Allergy Clin. Immunol. 130(1):44–50.

Hibbett, D.S., Binder, M., Bischoff, J.F., Blackwell, M., Cannon, P.F., Eriksson, O.E et al., 2007. A higher level phylogenetic classification of the Fungi. Mycological Research. 111(5): 509–547.

Hibbett, D.S., Matheny, P.B., 2009. The relative ages of ectomycorrhizal mushrooms and their plant hosts estimated using Bayesian relaxed molecular clock analyses. BMC Biol. 7:13.

Kotimaa, M.H., Oksanen, L., Koskela, P., 1991. Feeding and bedding materials as sources of microbial exposure on dairy farms. Scand. J. Work Environ. Health. 17(2):117-122.

Lanier, C., André, C., Séguin, V., Heutte, N., El kaddoumi, A., Bouchart, V., Picquet, R., Garon, D., 2012. Recurrence of Stachybotrys chartrum during mycological and toxicological study of bioaerosols collected in dairy cattle shed. Ann Agric Environ Med 2012;19(1):61- 7.

228

Lanier, C., Richard, E., Heutte, N., Picquet, R., Bouchart, V., Caron, D., 2010. Airborne moulds and mycotoxins associated with handling of corn silage and oilseed cakes in agricultural environment. Atmos. Environ. 44(16):1980-1986.

Lee, J.A., Thorne, P.S., Reynolds, S.J., O’Shaughnessy, P.T., 2007. Monitoring risks in association with exposure levels among wastewater treatment plant workers. Occup. Environ. Hyg. 49(11):1235-48.

Létourneau, V., Nehmé, B., Mériaux, A., Daniel, M., Cormier, Y., Duchaine, C., 2010. Human pathogens and tetracycline-resistant bacteria in bioaerosols of swine confinement buildings and in nasal flora of hog producers. Int. J. Hyg. Environ Health. 213(6):444-449.

Madsen, D., Klock, L.E., Wenzel, F.J., Robbins, J.L., Schmidt, C.D., 1976. The prevalence of farmer’s lung in an agricultural population. Am. Rev. Respir. Dis. 113:171–174 (1976).

Matsuzawa, T., Tanaka, R., Horie, Y., Gonoi, T., Yaguchi, T., 2010. Development of rapid and specific molecular discrimination methods for pathogenic Emericella species. Jpn. J. Med. Mycol. 51(2):109-116.

Mbareche, H., Veillette, M., Bilodeau, G.J., Wösten, H., Duchaine, C., 2018b. Fungal spore recovery from air samples: tale of loss and gain. Appl. Environ. Microbiol. 10.1128/AEM.02941-18

Mbareche, H., Veillette, M., Dubuis, M.E., Bakhiyi, B., Marchand, G., Zayed, J et al., 2018a. Fungal bioaerosols in biomethanization facilities. J. Air Waste Manag. Assoc. 68(11):1198- 1210.

Milner, P.D., 2009. Bioaerosols associated with animal production. Bioresour. Technol. 100(22):5379-5385.

Mistry, J., Finn, R.D., Eddy, S.R., Bateman, A., Punta, M., 2013. Challenges in Homology Search: HMMER3 and Convergent Evolution of Coiled-Coil Regions. Nucleic Acids Res. 41(12): e121.

Morgenstern, I., Klopman, S., Hibbett, D.S., 2008. Molecular evolution and diversity of lignin degrading heme peroxidases in the Agaricomycetes. J. Mol. Evol. 66(3): 243.

Nehme, B., Létourneau, V., Foster, R.J., Veillette, M., Duchaine, C., 2008. Culture- Independent approach of the bacterial bioaerosol diversity in the standard swine confinement buildings, and assessment of the seasonal effect. Environ. Microbiol. 10(3):665-675.

229

Nonnemann, M.W., Brextine, B., Dowd, S.E., Gilmore, K., Levin, J.L., 2010. Culture- independent characterization of bacteria and fungi in a poultry bioaerosol using pyrosequencing: a new approach. J. Occup. Environ. Hyg. 7(12): 693-699.

Nucci, M., Anaissie, E., 2007. Fusarium Infections in Immuno comprised Patients. Clin. Microbiol. Rev. 20(4):695-704.

Peat, J.K., Tovey, E., Mellis, C.M., Leeder, S.R., Woolcock, A.J., 1993. Important of house dust mite and Alternaria allergens in childhood asthma: an epidemiological study in two climatic regions of Australia. Clin. Exp. Allergy. 23(10):812-820.

Pieckova, E., Wilkins, K., 2004. Airway toxicity of house dust and its fungal composition. Ann. Agric. Environ Med. 11(1):67–73.

Pitt, J.I., Hocking, A.D., 1977. Influence of solute and hydrogen ion concentration on the water relations of some xerophilic fungi. J. Gen. Microbiol. 101(1):35-40

Pitt, J.I., Hocking, A.D., 2009. The Ecology of Fungal Food Spoilage. In: Fungi and Food Spoilage. Springer, Boston, MA, USA.

Poperscu, S., Borda, C., Diugan, E., 2011. Microbiological air quality in tie-stall dairy barns and some factors that influence it. Afr. J. Agric. Res. 6(32):6726-34.

Porter, P., Susaria, S.C., Polikepahad, S., Qian, Y., Hampton, J., Kiss, A., et al., 2009. Link between allergic asthma and airway mucosal infection suggested by proteinase-secreting household fungi. Mucosal Immunol. 2(6):504–517.

Priyanka, K., Woo, C., Yamamoto, N., Choi, H.L., 2016. Variations in abundance, diversity and community composition of airborne fungi in swine houses across seasons. Sci. Rep. 6:37929

Ramette, A., 2007. Multivariate analyses in microbial ecology. FEMS Microbiol. Ecol. 62(2):142–160

Reboux, G., Piarroux, R., Mauny, F., Madroszyk, A., Millon, L., Bardonnet, K., Dalphin, J.C., 2001. Role of moulds in farmer's lung disease in eastern France. Am. J. Respir. Crit. Care Med. 163(7):1534-153

Rodriguez, Z.C., Ramos, M.G., 2014. Acremonium species associated fungemia: a novel pathogen in the immunosuppressed patient. Bol. Asoc. Med. P. R. 106(3):29-31.

Rylander, R., 1996. Airway Responsiveness and Chest Symptoms after Inhalation of Endotoxin or (1→3)-β-D-Glucan. Indoor Built Environ. 5:106-111.

Sakamoto, T., Urisu, A., Yamada, M., Matsuda, Y., Tanaka, K., Torii, S., 1989. Studies on the osmophilic fungus Wallemia sebi as an allergen evaluated by skin prick test and radioallergosorbent test. Int. Arch. Allergy Appl. Immunol. 90(4):368-372.

230

Samadi, S., Van Eerdenburg, F.J., Jamshidifard, A.R., Otten, G.P., Droppert, M., Heederick, D.J., Wouters, I.M., 2012. The influence of bedding materials in bio-aerosol exposure in dairy barns. J. Expo. Sci. Environ. Epidemiol. 22(4):361-8.

Sarkar, A., Mukherjee, A., Ghoshal, A.G., Kundu, S., Mitra, S., 2010. Occurrence of allergic bronchopulmonary mycosis in patients with asthma: An Eastern India experience. Lung India : Official Organ of Indian Chest Society; 27(4):212-216.

Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., et al., 2009. Introducing mother: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75(23):7537-7541

Schoch, C.L., Spatafora, J.W., Lumbsch, T., Hundorf, S.M., Kevin, D.H., Johannes, Z.G., Crous, W.D., 2009. A phylogenetic re-evaluation of Dothideomycetes Studies in Mycology. 64. CBS-KNAW Fungal Biodiversity Centre.

Seglar, B., 2013. Diversity of silage moulds produces harmful mycotoxins. https://www.progressiveforage.com/forage-types/silage/diversity-of-silage-molds- produces-harmful-mycotoxins Visited: 15 November 2017.

Selman, M., Padro, A., King, T.E. jr., 2012. Hypersensitivity pneumonitis: insights in diagnosis and pathobiology. Am. J. Respir. Crit. Care Med. 186(4):314-324.

Shade, A., Hogan, C.S., Klimowicz, A.K., Linske, M., McManus, P.S., Handelsman, J., 2012. Culturing captures members of the rare biosphere. Environ. Microbiol. 14(9):2247– 2252.

Stark, P.C., Celedón, J.C., Chew, G.L., Ryan, L.M., Burge, H.A., Muilenberg, M.L., Gold, D.R., 2005. Fungal Levels in the Home and Allergic Rhinitis by 5 Years of Age. Environ. Health Perspect. 113(10):1405–1409.

Stiemsma, L.T., Reynolds, L.A., Turvey, S.E., Finlay, B.B., 2015. The hygiene hypothesis: current perspectives and future therapies. Immunotargets Ther. 4:143–157.

Strachan, D.P., 1989. Hay fever, hygiene, and household size. BMJ. 299(6710):1259–1260.

Tedersoo, L., Anslan, S., Bahram, M., Polme, S., Riit, T., Liiv, I., et al., 2015. Shotgun metagenomes and multiple primer pair barcode combinations of amplicons reveal biases in metabarcoding analyses of fungi. MycoKeys 10: 1-43.

Tsapko, V.G., Chudnovets, A.J., Sterenbogen, M.J., Papach, V.V., Dutckiewicz, J., Skorska, C., et al., 2011. Exposure to bioaerosols in the selected agricultural facilities of the Ukraine and Poland - A review. Ann. Agric. Environ. Med. 18(1):19–27.

231

Velegraki, A., Cafarchia, C., Gaitanis, G., Iatta, R., Boekhout, T., 2015. Malassezia Infections in Humans and Animals: Pathophysiology, Detection, and Treatment. PLoS Pathog. 11(1): e1004523.

Venter, J.C., Remington, K., Heidleberg, J.F., Halpern, A.L., Rusch, D., Eisen, J.A., et al., 2004. Environmental genome shotgun sequencing of the Sargasso Sea. Science. 304(5667):66–74

White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. PCR protocols: A guide to Methods and Applications. Wiley. New Yor, NY, USA.

Woolford, M.K., Pahlow, G., 1998. The silage fermentation. Microbiology of Fermented Foods, pp. 73–102, Springer US

Wyngaarden, J.B., Smith, L.H., Bennett, J.C., (ed.)., 1992. Cecil textbook of medicine, 19th ed., p. 399 and 407. W.B. Saunders Co.. Philadelphia, Pa, USA.

Zalar, P., de Hoog, S.G., Schroers, H.J., Frank, J.M., Gunde-Cimerman, N., 2005. Taxonomy and phylogeny of the xerophilic genus Wallemia (Wallemiomycetes and Wallemiales, cl. et ord. nov.). Antonie Van Leeuwenhoek. 87(4): 311–328

232

Chapter 7: Additional Discussions

This dissertation investigated the use of molecular tools to study fungal aerosols. The main question regards the choice of the eukaryotic genomic regions (barcodes) when applying HTS to best portray fungi in aerosols. In fact, the first part consisted on developing the methodology of HTS (choice of barcode), and the optimal way to recover fungi from aerosol samples, which were described in chapter 2 and 3, respectively. Afterwards, the next three chapters (4, 5 and 6) subsisted on the application of the methodology to describe fungal exposure in three different environments particularly affected by fungi. Furthermore, the chapter 6 added culture methods for comparison with HTS approaches. Lastly, the bioinformatic analysis applied in the whole thesis is discussed on the appendix A. This actual chapter is a gathering of undiscussed points on the previous chapters consisting of published articles.

The validity of HTS data obtained from the PCR-amplified marker genes depend greatly on bioinformatic tools and parameters used for data analyses. Commonly, the analyses are conducted by creating OTUs according to different algorithms. In de novo OTU clustering, sequences are clustered into OTUs by comparing them to the whole dataset with a fixed sequence dissimilarity threshold (e.g., 3%); without the use of a reference. In closed- reference methods, OTU clustering uses a reference sequence database, whereby the sequences that are sufficiently similar to sequences in the reference database form an OTU. Recently, a new trendy method don’t impose a dissimilarity threshold to analyse HTS gene markers nor a reference database. ASVs prone a better amplicon resolution by distinguishing sequence variants differing by one nucleotide (Eren et al., 2015; Tikhonov et al., 2015; Callahan et al., 2016; Edgar et al., 2016; Amir et al., 2017; Callahan et al., 2017). ASVs most prominent advantage is the combination of the benefits from overcoming limitations inherent to closed-reference and de novo methods. For instance, closed-reference OTUs cannot document biological variations outside of the reference database used for their construction. On the other hand, the validity of de novo OTUs outside of the dataset in which they were defined is also questionable, which make cross-studies comparison invalid. While ASVs capture all biological variations present in a dataset, and ASVs

233

inferred from a given dataset can be reproduced in future datasets and validly compared (Callahan et al., 2017). However, ASVs method also comes with its share of limitations. Allowing 100% sequence similarity may lead to a wrong differentiation between the SNPs of the same species. In addition, the zero percent difference may give an extremely high number of ASVs in a sample, which, in return, causes the missing of the core microbiome information’s (unpublished data). Above all, the same genome can contain multiple ASVs if there is multiple copies of the targeted gene. For this matter, ASVs can be validly compared between studies, only when the same primers were used on the targeted gene. Furthermore, the high variability of the ITS region makes us reconsider the automatic replacement of the traditional OTUs by ASVs. To sum up, ASVs and de novo OTUs are more precise in describing diverse biological sequences in a less represented environment in reference databases like bioaerosols, compared to closed-reference OTUs. Most importantly, no matter the methodology used, downstream analyses should consider the methodological differences, accordingly.

The diverse biological sequences are identified by comparison to a database. As mentioned throughout this dissertation, UNITE is the database used for the molecular identification of fungi. Without a doubt, it is a better choice when working with the fungal ITS region, as other databases, like NCBI or SILVA are not specifically curated for ITS identification. The UNITE database is regularly updated and annotated, thus, it continues to show tremendous improvements in the sequence management of fungal ITS (Nilsson et al., 2019). Although some databases contain an enormous number of sequences due to the fact that sequence deposition is easy an open to anybody, the risk of erroneous identification is also multiplied. In the contrary, UNITE is subject to quality control by ITSx and UCHIME in an attempt to reject non-ITS and chimeric sequences. When working with a less characterized environment like air (in comparison to soil and water), the quality control of sequences in a database is greatly appreciated. However, the representative sequences of ITS1 and ITS2 in UNITE is not known. According to the results obtained in chapter 2, more unidentified fungi and plant identification were associated to ITS2 in datasets obtained from air samples in compost, biomethanization, and dairy farms. Earlier this year, UNITE curators stated that they are working on adding more ITS2-derived HTS studies to UNITE (Nilsson

234

et al., 2019). It would be valuable to address the current state of ITS2 sequences in the UNITE databases. Actually, the author is conducting an analyses on the actual state of UNITE database. The first results indicate that ITS1 and ITS2 have equal representatives in all of the nine phylum present in UNITE. Few exceptions were noted like in Rozellomycota and Zoopagomycota where ITS1 had more representative taxa. This lead to the belief that the difference observed between ITS1 and ITS2 barcode is certainly dependent on the primers choice. The next step of the analyses will include an in silico analyses where the author tested multiple pairs of primers targeting ITS1 and ITS2 to evaluate the effect of the primers choice on the amplified taxa from UNITE. The results will be a part of a short communication (work in preparation). In addition, annotated UNITE sequences covers not only taxonomic names, but also country of collection, and substrate of collection. Currently, UNITE uses the default NCBI Taxonomy classification complemented by Index Fungorum and MycoBank. Sequences with a confusing taxonomic annotation are flagged by experienced users for manual correction by UNITE curators. In the long term, UNITE is re-annotating, and re- identifying conflicting taxonomic information. The original names are kept for reference tracking accuracy.

In molecular microbial ecology studies, phylogeny allows the study of the evolutionary relation between species based on mathematical algorithms that compares DNA sequences. Particularly, phylogeny analysis is useful when comparing microbial community members between different samples, and by incorporating information on the relative relatedness between the microbes. Thus, the use of a reliable phylogenetic tree is necessary when analyzing diversity with a phylogenetic measure. A phylogenetic tree is a diagram with branches showing the evolutionary relationships between species. Because the 16S rRNA gene sequence contains both highly conserved regions for primer design (PCR-amplification) and hypervariable regions to identify phylogenetic characteristics of microbes, it became the most widely used marker gene for profiling bacterial communities (Lane et al., 1985; Tringe et a., 2008; Yang et al., 2016). As ITS sequences are subject to intraspecific variability the construction of a phylogenetic tree is not recommended (Nilsson et al., 2008). Actually, different tree construction methods (clustalw, and different programs within the PHYLIP package) led to conflicting phylogenetic results using the same dataset (unpublished data).

235

This issue is known amongst the scientific community studying fungal diversity. Therefore, when analyzing fungal diversity with ITS datasets, measures including phylogenetic information are not recommended. Hopefully, many alternatives are available to compare microbial composition between samples without the use of phylogeny. These methods are discussed in greater detail in Appendix A. Another popular alternative amongst phylogeny aficionados is the use of the 18S rRNA gene sequences as a phylogenetic marker for the fungal kingdom. Although the 18S rRNA have promising characteristics to build a phylogenetic framework for fungi (Yarza et al., 2017), it shows a poor potential in describing the full extent of fungal diversity (Liu et al., 2015).

In recent years, constructing phylogenetic trees with whole genomes instead of targeting the 16S rRNA-encoding gene gained more attention. A compelling study by Hug et al., 2016 showed the limits of targeting the ribosomal gene as a phylogenetic marker because it does not carry all the phylogenetic information of the studied organisms. Compared to genomic trees, the gene catalog sequenced can give more accurate information on the evolutionary link between microorganisms. Moreover, the results of the latter study showed a phylogenetic tree completely different from the common one used with 16S rRNA- encoding gene (Hug et al., 2016). Interestingly, the same conclusion was demonstrated for fungi where the gene trees showed differences compared to the proteome trees (gene catalog proteins; Choi and Kim, 2017).

As discussed in the Literature Review chapter (section 1.7), long HTS reads from technologies such as PacBio (https://www.pacb.com/) and Oxford Nanopore (https://nanoporetech.com/) are gradually becoming available. To put it in context, one downfall of the actually most used sequencing platforms is the short sequencing reads. For example, we often choose a partial sub-region among the nine variable sub-regions (V1 to V9) of the full 16S rRNA gene so that the chosen region is fully sequenced by Illumina MiSeq (2 x 300 bp), HiSeq 2500 Illumina (2 x 125 bp) or Ion PGM (400 pb). With the increasing demand of more accurate methods to analyze microbial communities from different environments, sequencing longer amplicons (e.g., the full-length 16S rRNA gene) is appealing. Primarily, because of its potential to better identify bacteria at the species level.

236

Indeed, the high similarity of the 16S rRNA amplicon sequences usually used with Illumina or Ion Torrent make the accurate species level identification difficult. To overcome this situation, recent microbiome studies have used the nanopore sequencing platform to sequence the full-length 16S rRNA amplicon from the mouse gut microbiota. A comparison of the nanopore (full-length 16 rRNA gene) and short-read (V3-V4) Illumina sequencing data showed that there were no significant differences in major taxonomic units, with 89% of similarity between the two approaches. Moreover, both sequencing data were highly similar at all taxonomic resolutions except the species level. At the species level, nanopore sequencing allowed identification of more species than short-read sequencing, facilitating the accurate classification of the bacterial community composition (Shin et al., 2016). With these results in mind, applying the same approach to sequence the full-length ITS region represent a compelling avenue for fungal amplicon-based HTS methods, without having to choose between ITS1 or ITS2. However, a proof of concept should be conducted to fine-tune library preparation, including the best set of primers to amplify the ITS region. In addition, UNITE has already started to include long PacBio reads comprising the full ITS region, including few hundreds bases of the LSU gene, to the database (Nilsson et al., 2019).

Like for other environments, fungi in aerosols could benefit from the myriad of methods available to answer questions related to their occupational impact. Amplicon-HTS, shotgun metagenomics, metatranscriptomics, culturomics, and qPCR can all bring insightful information depending on the questions asked. What is the source of bioaerosols? What are the environmental factors favorizing aerosolization? Who is there? How are they interacting? Is this interaction susceptible to amplify the harm? What is the metabolic potential? Are there any differential expression of genes involved directly or indirectly to fungal pathogenesis? The answers to these questions could take us to answer other types of occupational questions. Is it possible to eliminate the source? How can we reduce the exposure? Can we control the factors favorizing aerosolization? Could the measures proposed change the quality and quantity of fungal exposure?

237

The success of the combination of multiple methods to study fungi in aerosols reside in the creativity applied to analyze the data. The importance of bioinformatics was already mentioned in chapter 1 (section 1.8), and the appendix A is dedicated to a bioinformatic review paper. However, one significant subject matter related to bioinformatics was not yet discussed. Machine learning is a subset of artificial intelligence based of algorithms and statistical models that computers use to perform specific tasks without a definite instruction for each task (Bishop, 2006). Concretely, the algorithms build mathematical models from a set of training data to predict the next decision without being programmed to do so. In the context of molecular ecology, data mining is the machine learning field of study closely related to the analyses needed for HTS data. Data mining focuses on discovering patterns in large datasets with the use of machine learning, statistics and databases. It allows the exploration of data analyses through unsupervised learning. Supervised machine learning include support-vector machines and it is possible when more data information are available.

Recent work on HTS environmental DNA (eDNA) have showed that machine learning could overcome many limitations of taxonomy-based eDNA bioassessment (Cordier et al., 2018). The algorithms used could predict accurately biotic indices values from HTS data, regardless of the taxonomic identification of the sequences. In the Cordier study, prokaryotic and eukaryotic ribosomal markers yielded the same accuracy of the predictive models to assess the environmental impact of marine aquaculture. This suggest that the use of predictive models with training datasets from HTS on relevant molecular markers that comprises different potential bioindicator taxa have the potential to overcome some limitations of the traditional methods. Thus, the combination of microbial HTS with machine learning algorithms can be highly effective in filling important gaps of microbial ecology studies. For example, machine learning models were previously used to predict effluent concentration in a wastewater treatment plant (Guo et al., 2015). In addition, microbial community compositions helped the establishment of predictable effluent parameters, like temperature, suspended solids, biochemical oxygen demand, using machine learning’s vector regression models (Liu et al., 2016). The application of machine learning to bioaerosols could include scaling-up both spatial and temporal resolution for larger and more ambitious

238

monitoring of exposure. It could be used to predict microbial source tracking, and provide a clearer vision of the environmental impact of bioaerosols.

Bioaerosol worker exposure studies suffer from the lack of exposure-response relationship. As an analogy to particulate matter of inhalable size PM2.5 and PM10, important information is collected from standard measurement methods, thus more rigorous exposure– response curves are generated, which are used by WHO (World Health Organization) to set guidelines on exposure limits. The absence of accurate dose-response data for bioaerosol exposure is due to the diversity of health effects where each person can react differently to the same exposure, and to inadequate exposure assessment. Occupational exposure studies can take a step forward and gain much more information by applying HTS methods to better understand who is present, what are they doing and how are they interacting with each other and with the different parameters of the environment. Ultimately, the information cumulated using these approaches in different types of environments can lead to find better environmental markers to assess exposure. Thus, these environmental markers, which can be specific to different environments, could be used to obtain health-related exposure limits. Furthermore, the environmental markers and the exposure-limits can be specific to different health outcomes. Another key element for the elaboration of exposure-response relationships and exposure limits is the epidemiological studies. Indeed, one of the main goal of the exposure characterization is to provide information for epidemiological studies, which are essential tools for health risk assessment and elaboration of guidelines. The data collected using the proposed multi-omics approaches herein can be used to carry out epidemiological investigations by describing the distribution of bioaerosols.

Studies applying HTS methods to assess occupational exposure should move from being solely descriptive (stating who is there) to a more thorough analysis (explaining the implication of such microbial presence) by making hypothesis and experiments to confirm them. For example, the chapter 3 described the loss of fungal taxa from environmental air samples by hypothesizing on the hydrophobicity of fungi, but also proved the hypothesis with a laboratory experiment confirming the environmental field results. Also, chapter 4 did not only describe fungi in a composting environment, but linked fungal composition to the

239

type of composting product, and demonstrated a higher occurrence of some fungi of interest in aerosols compared to the source (preferential aerosolization). Likewise, chapter 6 used different environmental variables to explain the difference in the fungal composition of five dairy farms. Although the number of farms visited was low to make statistically significant associations, this work could inform future studies on the type of variables to evaluate and the associations that are worth the validation. Plus, the comparison between HTS and culture allowed the answer of the complementarity of both method hypothesis.

Bioaerosol researchers could conduct an exposure study with the goal of determining whether there is an association between specific symptoms and the type of bioaerosols workers are exposed to. Information collected and stocked from long-term association studies could help to determine specific environmental markers of exposure. Associations between a type of environment, its multiple markers and the health outcomes observed could be used to accumulate knowledge for these association studies. In the long-term, the compilation of this massive amount of information and the use of advanced bioinformatics algorithms to treat that information, can lead to wide-scale association studies bringing bioaerosol exposure assessment into a new era.

240

Conclusion

Concluding Remarks

This thesis presented a comprehensive framework for the study of fungi in aerosols. The concern came mainly from the significant gap in fungal diversity of aerosols in occupational environments. Certainly, having a wider portrait of fungi that could represent a risk for exposed individuals in concerned environments represent an important piece of the puzzle for occupational exposure studies. Plus, combining the description of the fungal profiles of bioaerosols to other components like quantifying the fungal biomass, the association with traditional culture methods, the association of aerosols with the source, and the environmental variables explaining the fungal community composition served as a methodology approval and has brought new answers to fill the gap of fungi exposure in occupational environments. The primary objective of comparing the performance of two universal fungal barcodes to study fungi in aerosols using HTS opened the door to an unknown problem of recovering fungal cells from liquid air samples. A second objective was then to understand the behaviour causing the loss of fungi in liquid samples and to come up with a new method that allows a better qualitative and quantitative recovery. This latter work give the opportunity for future studies to acknowledge the problem of fungal cell loss, and offers an alternative method no matter the types of sampler used, as far as it involves liquid suspension. The third objective was to evaluate the potential of the developed methodologies to describe exposure in composting sites, biomethanization plants, and dairy farms, three environments affected by the fungal presence.

In general, the work presented herein tells a story that covers the study of fungal aerosols starting from optimal method of treating air samples, going through the appropriate microbial molecular tools to describe fungi, and finishing by their concrete application on occupational studies. Looking behind, this thesis has addressed some of the major challenges in fungal aerosol studies, and the results obtained could be useful not only for occupational environments, but indoor/outdoor as well. In addition, chapter 1 (Literature Review) and chapter 7 (Additional Discussions) put the thesis in the actual scientific context and open

241

discussions about the forces, limitations, and important considerations that the readers and users of this work could use to design and conduct future work.

Future work

Looking ahead, key research directions which could follow up on the methodological contributions of this thesis could include the HTS of long reads using the PacBio or Nanopore technology covering the whole ITS region. The outcome could be compared to the approach targeting ITS1 or ITS2 to characterize fungal diversity in aerosols. In terms of occupational exposure, future works could include the measuring of health problems amongst the workers of the sampled environment. This way, fungal composition and concentration can be linked to specific health conditions, especially if the number of visits in the sampling campaign is very high. For this matter, it is important to shift attention to new ways of designing bioaerosol exposure studies in order to prevent the replication of previous flaws (e.g. different study designs, methods and analyses which makes it impossible to compare the results). Eventually, the goal is to set exposure guidelines referring to health-related exposure limits.

Microbes are now considered a community instead of single, independent microorganisms. For example, pathogenesis in the gut is now examined by looking at the microbial community as a whole instead of at individual pathogens. Similarly, a core mycobiome of aerosols may be specific to particular occupational environments that are influenced by specific types of sources. A collaborative effort could allow for the identification of core mycobiomes for bioaerosols through a public database specifically for bioaerosols. Verified high-quality sequences obtained from HTS platforms could be deposited in a particular section of the database according to the environment sampled. Additionally, metagenomics data could serve as potential environmental exposure markers in the future, as environmental genes from thousands of samples would be available to all researchers interested in evaluating the new markers. The same idea can be applied to any indoor/outdoor occupational and general community exposure environment of interest. Furthermore, information obtained by HTS methods can help make better cultivation

242

conditions and improve culturomics methods, a field of study not exploited for fungal analyses.

Recent studies have established a connection between the bacterial community of the upper respiratory tract (nasal cavity and nasopharynx) of exposed individuals and the bioaerosols collected from the environment they were exposed to, using an HTS approach. It would be of great interest to verify this connection for fungi. The mutual fungal community of the bioaerosol core mycobiome and the upper respiratory tract of exposed individuals could help in the identification of environmental markers specific to each working environment. Future work should include nasal or nasopharynx sampling regime along with bioaerosol sampling in the design of all occupational exposure studies, which can help with the creation of a list of potential biological markers.

The idea of creating a bioaerosol public database was discussed in a recently published opinion paper (Mbareche et al., 2019). All the information generated from the methods discussed above could be incorporated in the Bioaerosol Public Database to keep track of what is measured and how, and accumulate the large dataset on bioaerosol exposure studies. An international scientific organisation could be responsible for the creation of the database like the NIH (National Institute of Health ) for the Human Microbiome Project. Else, the creation of the database could be an effort from courageous and ambitious bioaerosol scientists, working with third-parties specializing in database creation and maintenance. Ultimately, the implementation of such work will lead to a deeper understanding and more efficient utilization of bioaerosol studies.

The author hopes that the foundations established throughout this thesis will be valuable for future fungal aerosol studies. The story told encourages the application of the filtration-based method on air samples prior to DNA extraction, targeting ITS1 as a universal fungal barcode, especially with limited financial resources, and the application of an accurate bioinformatic analyses that allows hypothesis testing and statistical associations to explore fungal exposure in various environments.

243

Appendix A: Bioinformatics Tools to Study the Microbial Ecology of Bioaerosols

A.1 Résumé

Le séquençage à haut débit a changé notre façon d’étudier les microbes. Les analyses bio-informatiques permettent de tirer pleinement profit de la quantité gigantesque de données de séquençage. Certains scientifiques ont une réticence à l'égard de la bio- informatique, en particulier dans le domaine des bioaérosols, en raison de l’absence d’outils d’analyse avec des interfaces conviviales et des compétences de programmation nécessaires au traitement des séquences et à l’analyse de la diversité. Savoir quelles analyses effectuer et quels outils appliquer reste une source de confusion pour les scientifiques, car une panoplie d'outils et de ressources sont disponibles pour caractériser les communautés microbiennes. Le présent article a pour objectif de proposer une visite guidée des outils bio-informatiques utiles à l’étude de l’écologie microbienne des bioaérosols. Les auteurs espèrent que le travail présenté constituera une source d’information pertinente pour mener à bien une étude critique en écologie microbienne impliquant la bio-informatique.

244

A.2 Summary of the paper

This paper is a review on the use of bioinformatics in bioaerosol studies. The concepts discussed can be applied to any type of environment, but examples cited herein are related to bioaerosol studies. The idea of this project came from the trial and error approach that the author of the thesis had to go through to learn the basics of bioinformatics in microbial ecology, and to adapt the knowledge to study of fungi. Facing a panoply of tools and a wide range of parameters that affects the outcome of the diversity analyses, the author wished to synthesize all the knowledge learned on bioinformatics throughout the PhD studies to make bioinformatics more accessible, and to offer new users a platform to start from.

This review contains five sections covering most important subject of microbial ecology like sequencing depth, data normalization, alpha/beta diversity, multivariate analyses, parametric/non-parametric tests, differential abundance, microbial dysbiosis, taxonomic analyses, and correlations.

245

Bioinformatics Tools to Study the Microbial Ecology of Bioaerosols

AUTHORS Hamza Mbareche1,2, Marc Veillette1, Guillaume J Bilodeau3 and Caroline Duchaine1,2

AUTHORS’ AFFILIATION 1. Centre de recherche de l’institut universitaire de cardiologie et de pneumologie de Québec, Quebec City (QC), Canada 2. Département de biochimie, de microbiologie et de bio-informatique, Faculté des sciences et de génie, Université Laval, Quebec City (QC), Canada 3. Pathogen Identification Research Lab, Canadian Food Inspection Agency (CFIA), Ottawa, Canada

KEYWORDS Bioaerosols, bioinformatics, microbial ecology

IN PREPARATION : The paper is in revision by coauthors

246

A.3 Abstract

High-throughput sequencing has changed our understanding of the microbial composition present in a wide range of environments. Bioinformatic analyses make it possible to take full advantage of the avalanche of information from sequencing and help discover general patterns that govern microbial ecosystems. Actually, there is a certain reluctance towards bioinformatics by some scientists, especially in the bioaerosol field, due to the command line tools and programming skills needed for sequence processing and diversity analyses. Knowing which analyses to conduct and which tools to apply remains confusing for bioaerosol scientists, as a litany of tools and data resources are now available for characterizing microbial communities. The goal of this review paper is to offer a guided tour through the bioinformatics tools that are useful in studying the microbial ecology of bioaerosols. The authors hope that this work represents a conduit for the popularization of bioinformatics in the study of bioaerosols and will provide a good source for the «dos and don’ts» when conducting a critical microbial community study.

A.4 Introduction

The development of next-generation sequencing (NGS) platforms has been grown exponentially in recent years (Mardis, 2008; Novais & Thorstenson, 2011; Caporaso et al., 2012). This burst in high-throughput sequencing (HTS) has revolutionized our understanding of the microbial composition present in a wide range of environments (Caporaso et al., 2011; Bolhuis et al., 2014; Huttenhower et al., 2014; Goodrich et al., 2014; Yoon et al., 2015; Gilbert et al., 2018). More specifically, amplicon-based sequencing is the most commonly used method for characterizing microbial diversity (Benìtez-Pàez and Sanz, 2017; De Filippis et al., 2017; Pollock et al., 2018; Naqib et al., 2018). This method includes the use of a taxonomically informative genomic marker that is common to all microorganisms of interest and that is targeted by an amplification step prior to sequencing. For bacteria and archaea, amplicon-based sequencing studies target the gene that codes for the small ribosomal subunit 16S (Hugenholtz and Pace, 1996). For fungi, the gene that codes for the Internal Transcribed Spacer (ITS) is considered the universal maker for molecular approaches to studying fungal

247

diversity (Schoch et al., 2012). The sequenced amplicons are characterized using bioinformatics tools to determine which microbes are present in a sample and at what relative abundance. Comparing the targeted sequences across samples gives insight into how microbial diversity associates with and scales across environmental conditions.

HTS approaches have been used to characterize the microbial composition of various environments, from soil to water to the rhizosphere to the human gut (Venter et al., 2004; Delmont et al., 2011; Walter and Ley, 2011; Philippot et al., 2013). In 2010, Peccia and his collaborators (2010) highlighted the importance of incorporating DNA sequencing methods into the study of aerosol science. Applying HTS methods to air samples from different environments allows for the identification and quantification (relative abundance) of the microorganisms present and for a better understanding of human exposure to indoor and outdoor bioaerosols. Using HTS approaches offers a thorough picture of the microbial content of aerosols, and leads to millions of sequences generated from that single sample (Gandolfi et al., 2013; Yamamoto et al., 2014; Madsen et al., 2015; Mbareche et al., 2017a; Mbareche et al., 2017b; Dubuis et al., 2017). In order to make full use of the information made available by these sequences, repeated measurements must be taken, community composition described, error estimates made, correlations of microbiota with covariates (variables) must be examined, and increasingly sophisticated statistical tests must be conducted, all by using bioinformatics tools (Knight et al., 2012).

Bioinformatics is not new to science, as it was first mentioned back in 1970 in a conversation between Dutch scientist Paulien Hogeweg and her colleague Ben Hesper to describe their work on the study of informatic processes in biotic systems (Hogeweg, 2011). Consistent with the rise in NGS, the past few years represent a surge in bioinformatics tool development for analyzing the large amounts of data generated by amplicon-based sequencing approaches (Schloss et al., 2009; Edgar, 2010; Caporaso et al., 2010; Edgar, 2011; Edgar, 2013; Rognes et al., 2016). Bioinformatics can be divided into computational biology, which uses algorithms to build mathematical models to solve biological problems using a computational method, and analytical bioinformatics, which uses bioinformatics tools to analyze biological data (Jamison, 2003). This definition of bioinformatics inspired

248

conversations about the status of bioinformaticians. Vincent and Charette tried to answer the question “Who qualifies as a bioinformatician?” by suggesting that the status should be reserved for experts who develop bioinformatics algorithms and tools (software) and for those who design architectural models to maintain databases (Vincent and Charette, 2015). This definition did not elicit unanimity amongst the scientific community who/that does not develop algorithms, but that uses bioinformatics tools on a daily basis to analyze data, generate results and solve problems (Smith, 2015). While this distinction is important as it allows universities, human resources and governments to accurately recognize and certify students, employees and others as bioinformatics experts, it is important to remember that using computers to understand biological concepts is as important and necessary as using any other laboratory tool/equipment. Therefore, the bioinformatics field should lean toward inclusion rather than exclusion. Because microbiology is entering a new era, bioaerosol scientists, among others, should not fear using bioinformatics tools to conduct microbial community studies.

Knowing which analyses to conduct and which tools to apply remains confusing for bioaerosol scientists, as a litany of tools and data resources are now available for characterizing microbial communities. The goal of this review paper is to offer a guided tour through the bioinformatics tools that are useful in studying the microbial ecology of bioaerosols. This paper does not focus on sequence data processing (quality filtering, Operational Taxonomic Unit clustering, etc.) as this information is described in previously published work (Mbareche et al., 2017a; Mbareche et al., 2017b) and there is ample literature available on bioinformatics pipelines for processing sequences (Schloss et al., 2009; Caporaso et al., 2010; Human Microbiome Project Consortium, 2012; Davenport and Tümmler, 2013; Sinha et al., 2017). This work explains microbial ecology features like alpha and beta diversity, multivariate analyses, differential abundances, taxonomic analyses, visualization tools and statistical tests using bioinformatics tools for bioaerosol scientists new to the field.

249

A.5 Methods and Software

The methodological bioinformatics approaches proposed in this manuscript for studying the microbial ecology of bioaerosols rely on the use of widely adopted QIIME pipelines, Mothur software (Schloss et al., 2009; Caporaso et al., 2010) and R packages; particularly, the vegan (Oksanen et al., 2007) and phyloseq packages (McMurdie and Holmes, 2013). All of the analyses proposed in this manuscript can be done using these software programs and R packages. Detailed documentation about their usage is available online. Additionally, Bioconductor is an open-source software package for bioinformatics that offers different features, courses and training on the usage of R for sequencing data associated with microbial ecology (https://www.bioconductor.org/).

A.6 Sequencing Depth

Sequencing depth can be defined as the number of reads obtained in a sample. The sequencing effort depends on the NGS platform used and the higher the sequencing depth, the more likely it is that diversity coverage will be attained (Sims et al., 2014). It is essential that the data be normalized so that all samples have the same sequencing depth. Sequencing depth (reads per sample) can affect diversity measures, as samples with more reads may appear more rich and will cluster together in multivariate analyses. It is always recommended to try different methods for normalizing the data because the mentioned biases are always present and sometimes considerable. One way to verify this tendency is to add information about the number of reads per sample into the metadata before normalizing and see if samples with higher numbers of reads cluster together.

Methods of data normalization can include rarefying or normalizing. Rarefaction creates a subsampled data set by randomly sampling the input sequences. Samples with fewer sequences than the requested rarefaction depth are not included in the analyses. The output can be used to provide diversity curves based on the number of sequences in a sample. These types of curves provide insightful information about how much microbial diversity is covered. If plateaus of richness and diversity are attained after a certain number of sequences

250

per sample, they signify that sequencing efforts were sufficient enough to cover all of the diversity in the sample. Different sequencing depth values should be tested. Two important considerations are: 1) finding the highest value for which the majority of samples would be included, and 2) finding the highest value that provides the best coverage plateau. As an alternative to rarefaction, normalization accounts for uneven sample sums and attempts to correct compositionality. In other words, samples represent a fraction of the ecosystem and the observed sequences are relative abundances; therefore, the data are compositional. In general, normalization procedures attempt to minimize the technical variability between samples and sample-specific dispersion (Dillies et al., 2013). A novel normalization technique, CSS (cumulative sum scaling) by metagenomeSeq, corrects the bias associated with the assessment of differential abundance to a pre-determined percentile by dividing raw counts by the cumulative sum of counts (Paulson et al., 2013). It is not recommended to use normalized data with presence/absence metrics like binary metrics or unweighted UniFrac, because CSS methods are abundance-based. Although used mainly for differential abundance analysis (statistically significant differences in microbe abundance across samples), DESeq can also be used as another data normalization alternative to rarefaction (Dillies et al., 2013; Weiss et al., 2015). The microbial communities section of this paper addresses the DeSeq method in the context of differential abundance analysis.

Normalizing and rarefying present both advantages and disadvantages. When a subsample is generated to an even depth, some observations are discarded which reduces the ability to detect differences in diversity measures (McMurdie and Holmes, 2014). And, although there is a definite reduction in resolution, the simplicity and clarity of the method can be worth the loss of a few reads. Furthermore, microbial communities are often different enough that the loss of a few reads won’t affect the overall measure of diversity (Weiss et al., 2015). Despite normalizing data using CSS being a promising technique, it should be used with caution as it can dramatically exaggerate the low-abundance taxa which can lead to their over-representation in a CSS normalized data set (Costea et al., 2014). Also, DESeq produces negative values for OTUs (Operational Taxonomic Units) with low abundances as a result of its log transformation. Some diversity metrics, like Bray-Curtis, cannot be used with negative values and therefore can’t be used to analyze a data set normalized by DESeq.

251

The key is to verify the results using multiple normalizing approaches, as different methods can complement each other depending on the goal of the research. It is important to consider, that normalization is a highly debated topic and there is currently no consensus from experts on which normalization method is better (Paulson et al., 2013).

A.7 Alpha and Beta Diversity

The measurement of species diversity was first introduced by Whittaker and defined as the number of species and their proportion within one sampling site (Whittaker, 1972). There are different ways to measure alpha diversity depending on the context of the study. A list of indexes is presented by Magurran and McGill (2011). The Number of observed OTUs, Chao1, Shannon and Simpson are commonly used alpha diversity measures and have been shown to perform well in the context of bioaerosol exposure studies (Mbareche et al., 2017b, Dubuis et al., 2017; Mbareche et al., 2018). More specifically, Chao1 is a richness estimator. The higher the number of unique OTUs in a sample, the higher the value of the Chao1 index. For Shannon and Simpson, the species richness is combined with the abundance to give one diversity measure. The Simpson index represents the probability of two randomly selected OTUs from the same sample, being of/from the same species. The output values are bounded between 0 and 1, where 1 represents the highest diversity. Shannon output values start at 0, and higher values are associated with higher diversity. An important factor to consider when choosing an alpha diversity measure for comparing sets of samples is the gene marker used for HTS. For example, PD Whole Tree is a phylogenetic alpha diversity measure and is defined as the minimum length of all phylogenetic branches reacquired to span a given set of taxa on the phylogenetic tree (Faith and Baker, 2006). Thus, the use of a reliable phylogenetic tree is necessary when applying the PD Whole Tree analysis. Compared to the markers for 16S bacterial and archaea genes, the fungal ITS gene marker is subject to intraspecific variability (Nilsson et al., 2008). The construction of a phylogenetic tree is not recommended due to the possibility of obtaining different results using the same dataset but with different tree construction methods (data not shown). Every metric has different strengths and limitations. Technical information on each metric is available in ecology textbooks and is beyond the scope of this paper.

252

Beta diversity compares the microbial composition between samples for different environments (Tuomisto, 2010). The output of beta diversity measures is a distance matrix containing a dissimilarity value for each pairwise comparison (each sample compared to another). Before any comparison can be accurately made, samples must be normalized or rarefied so that they are all the same size (Dillies et al., 2013; Paulson et al., 2013). There are a number of metrics for beta diversity measurements that can be classified into two categories: those that use phylogenetic information (rely on the quality of the constructed phylogenetic tree) and those that do not, which are formally known as non-phylogenetic methods (Lozupone and Knight, 2008; Kuczynski et al., 2010; Leprieur et al., 2012; Wang et al., 2013). One of the most used phylogenetic beta diversity measures is UniFrac (Unique Fraction), which measures the degree of unique evolution of one microbial community compared to others (Navas-Molina et al., 2013). With the assumption that closely related species have similar genetic functions, the abundances of phylogenetically similar taxa have less importance when using UniFrac for beta diversity measurements (Lozupone et al., 2006). Quantitative measures (e.g., weighted UniFrac) are suited for revealing community differences that are due to changes in relative taxon abundance (e.g., when a particular set of taxa is more abundant in bioaerosol samples compared to the source of aerosolization). Qualitative measures (e.g., unweighted UniFrac) consider the presence/absence of OTUs and are most informative when bioaerosol microbial communities differ according to various factors such as temperature, relative humidity, season, and time. This is because information on relative abundance can sometimes mask significant patterns of variation in which taxa are present (Lozupone and knight, 2007). The Bray-Curtis Dissimilarity Index is one of the most popular non-phylogenetic measures (Bray and Curtis, 1957). It quantifies the compositional dissimilarity between two different samples, based on the counts from each sample. The Bray–Curtis dissimilarity is bounded between 0 and 1, where 0 means the two samples have the same composition and 1 means the two samples do not share any taxa. It is not considered a distance because it does not satisfy the triangle inequality rule, and should be called a dissimilarity to avoid confusion. Bray–Curtis and Jaccard indices both use rank-order but the Jaccard index is metric while Bray-Curtis is semi-metric.

253

Once distances/dissimilarities between samples are computed, hierarchical clustering can be used to detect patterns of sample grouping. Samples with similar microbial compositions are grouped together in branches of a dendrogram (Rokach et Maimon, 2005). Hierarchical clustering is a useful tool for sample grouping visualization but should be coupled with additional statistical tests (Caporaso et al., 2010). Moreover, the information in the distance matrices generated can be displayed in a dimensional space (two or three orthogonal axes) for better visualization of the sample closeness. Two popular ordination techniques in microbial ecology are non-metric multidimensional scaling (NMDS) and metric multidimensional scaling (MDS). The classic example of multidimensional scaling is the Principal Coordinates Analyses (PCoA; Quinn and Keough, 2002, Caporaso et al., 2010; Navas-Molina et al., 2013). MDS algorithms aim to place each sample in N- dimensional space such that the inter-sample distances are preserved as much as possible. Each sample is assigned coordinates in each of the N dimensions. The number of dimensions on an MDS plot can exceed 2 and is specified a priori. Choosing N=2 optimizes the object locations for a two-dimensional scatterplot. The accuracy of the PCoA plot can be evaluated using jackknifing which is an iterative resampling procedure where one OTU from the data set is omitted in each iteration. Then, the average is represented on a PCoA plot with variance represented as confidence ellipsoids (Navas-Molina et al., 2013). On the contrary, the position of samples in NMDS represents the rank order of inter-sample distances. In general, both ordination techniques should lead to similar conclusions and it is recommended to test both methods on each data set. To choose the method that is most appropriate for the dataset, there are several papers that are dedicated to the subject and that go into greater detail (Ramette, 2007; Zur et al., 2007; Buttigieg and Ramette, 2014).

A.8 Parametric VS. Nonparametric Statistics

Nonparametric statistics are not based on parameterized families of probability distributions (Saltelli and Marivoet). Some examples of the typically used parameters are mean, median, mode, variance, range, and standard deviation. Unlike parametric statistics, nonparametric statistics make no assumptions about the probability distributions of the variables being assessed. The difference between parametric and nonparametric models is

254

that the former has a pre-established number of parameters, while the latter determines the number of parameters depending on the dataset. In other words, the parameters are determined by the dataset in nonparametric statistics, and by the model in parametric statistics. Counterintuitively, parametric tests are able to perform well with skewed and nonnormal distributions (Vickers, 2005). Sample size and dispersion (data spread in all groups) should also be checked before using a parametric test with data that do not have a normal distribution in order to choose the right test. For example, the 2-sample t-test and One-Way ANOVA assume equal variances and these options should not be selected when the dispersion of data in each group of samples is different. Usually, parametric tests have equivalent nonparametric tests that can be used as alternatives. Here are a few examples of related pairs of tests: 1-sample t-test and Wilcoxon; 2-sample t-test and Mann-Whitney test; One-Way ANOVA and Kruskal-Wallis. Even though parametric tests have more statistical power for detecting significance, nonparametric tests can be more suitable when a dataset is better represented by the median rather than the mean (Garcìa et al., 2010). Also, nonparametric tests perform better with ordinal and ranked data compared to parametric tests that can only assess continuous data. Thus, nonparametric tests can better handle exceptions that cannot be removed (Zimmerman, 1994).

To conclude, if the mean accurately represents the center of the distribution of the dataset and the sample size is large enough, one might consider a parametric test even with a nonnormal distribution. However, if the median is a better representative of the center of the distribution of the dataset, nonparametric tests can give more accurate results even with a large number of samples. It should be noted that when the sample size is very small, nonparametric tests are the only option. Overall, checking the assumptions associated with the statistical test is crucial for making the best choice as each one has its own data requirements (Erceg-Hurn and Mirosevich, 2008).

255

A.9 Microbial Community Comparison: Statistics and Visualization Tools

A.9.1 Comparisons using Alpha and Beta Diversity Measures

Alpha diversity index values obtained for each sample can be compared based on parametric or nonparametric tests that use multiple groupings of sample data. For example, air samples may be labeled as one of three types: outdoor control, sampling site 1 or sampling site 2. Statistics comparing each combination of two sample groups (outdoor control and sampling site 1; outdoor control and sampling site 2; sampling site 1 and sampling site 2). The results include the means and standard deviations of the alpha diversities of the two groups, along with the p-value of the statistical test. Based on these results, one can determine which groups of samples are significantly more rich and diverse than the others.

Similarly, a common set of simple statistics can be used to determine if a sample is statistically significantly different from another sample in terms of its microbial content (beta diversity). This is achieved by performing a determined number of randomizations of sample/sequence assignments, and recording the probability that sample 1 is phylogenetically different from sample 2. The p-value would represent the probability that a random sample/sequence assignment will result in more dissimilar samples than a pair of the actual samples being compared. In other words, after n random sample/sequence assignments, a probability of the dissimilarity of the samples is recorded. If this probability (of samples being more dissimilar) is higher than the probability of the actual pair of samples tested, we don’t reject the null hypothesis that there is no difference between the samples being compared.

A.9.2 Statistical Significance of Sample Groupings

The analysis of the strength and statistical significance of sample groupings using a distance matrix as the primary input can be used in combination with the previously discussed NMDS or MDS (PCoA) to further validate that the detected patterns of sample groupings are statistically robust. As to NMDS and MDS, the input is a distance matrix with values

256

measuring dissimilarities between pairs of samples. Dissimilarity measures may be defined as distances in a Euclidean space or in a space that may be interpreted as Euclidean distance. Euclidean distances satisfy the triangle inequality: the direct distance between two points is less than any detour. Thereby, the distances are considered to be metric. A semi-metric distance is similar to a metric distance but it accepts the discernibility between two points, even if the distance is equal to zero (Gower and Legendre, 1986). Also, metric and semi- metric dissimilarities satisfy the non-negative rule. In a non-Euclidean distance, it is not possible to find a representation in a two- or higher-dimensional Euclidean space in which the distances between points equal the given pairwise dissimilarities. However, non- Euclidian distances can still be metric (Lu et al., 2011). There are several methods available for analyzing the statistical significance of sample groupings using distance matrices. The suitability of these methods should be evaluated based on parametric or nonparametric features and on distance matrices that are constructed with metric, semi-metric or non-metric dissimilarities. The following tests are among the most used in microbial ecology studies, and are well suited for bioaerosol studies more specifically: Adonis, ANOSIM, BIO-ENV, Moran’s I, MRPP, PERMANOVA, PERMDISP, and db-RDA.

Adonis partitions distance matrices among sources of variation in order to describe the strength and significance that a categorical or continuous variable has in determining variation of distances. This is a nonparametric method and is almost equivalent to db-RDA, except when distance matrices are constructed with semi-metric or non-metric dissimilarities, which may result in negative eigenvalues. Adonis is very similar to PERMANOVA, though it is more robust because it accepts both categorical and continuous variables in the metadata mapping file, while PERMANOVA only accepts categorical variables (Legendre and Anderson, 1999). Moreover, PERMANOVA is based on the ANOVA experimental design, but because it is a non-parametric test it analyzes the variance and determines the significance using permutations (Anderson et al., 2005). While ANOVA/MANOVA assumes normal distributions and Euclidean distance, PERMANOVA can be used with any distance measure as long as it is appropriate to the dataset. PERMDISP is a method that analyzes the multivariate homogeneity of group dispersions (variances). It determines whether the variances of groups of samples are significantly different. The results of both parametric and

257

nonparametric significance tests are provided in the output. This method is generally used in combination with PERMANOVA (Anderson and Walsh, 2013). MRPP is another method that tests whether two or more groups of samples are significantly different based on a categorical variable found in the metadata mapping file. Since MRPP is nonparametric, significance is determined through permutations (Berry and Wong, 1986). ANOSIM tests whether two or more groups of samples are significantly different based on a categorical variable found in the metadata mapping file. Since ANOSIM is nonparametric, significance is also determined through permutations (LeRoy Poff et al., 2007). Similar to Adonis, db- RDA differs if certain non-Euclidean semi or non-metrics are used to produce the distance matrix, and negative eigenvalues are encountered. This difference will be apparent in the p- values, not the R^2 values. As part of the output, an ordination plot (like PCoA) can be generated to visualize sample grouping based on a category in the metadata mapping file. The ordination plot of db-RDA is similar to PCoA except that it is constrained, while PCoA is unconstrained. In other words, with db-RDA a category must be specified to explain the variability in the dataset (McArdle and Anderson, 2001). BIO-ENV (BEST) finds subsets of variables whose Euclidean distances are maximally rank-correlated with the distance matrix. For example, the distance matrix might contain UniFrac distances between communities, and the variables might be numeric environmental variables (e.g., pH and latitude). Correlations between the community distance matrix and Euclidean environmental distance matrix is computed using Spearman’s rank correlation coefficient (rho). This method will only accept continuous or discrete numerical categories (Clarke et al., 2008). Interestingly, this method accepts more than one category to explain variation between groups of samples. Moran’s I is another method that uses numerical data to identify which type of numerical variables explain sample grouping (Braz et al., 2009).

Computing sample grouping quality can also be assessed by measuring the ratio of the mean distances between samples from different groups by the mean distances between samples from the same group. The closer the ratio is to one, the less difference exists between the groups. This can be useful in comparing the ratios of different groupings represented by different categories in the mapping file. Again, distance matrices can be used as an input for

258

the ratio calculation. Normally, the results should match the groupings observed by transforming the distance matrix to principal coordinates and plotting the result.

A.9.3 Differential Abundance

Differential abundance analyses allow for the identification of OTUs that are differentially abundant across two sample categories in the mapping file (e.g., outdoor and indoor air samples). Two parametric tests are available for such analyses: MetagenomeSeq zero-inflated Gaussian (ZIG) and DESeq2 negative binomial Wald test. It is recommended to have at least five samples in each category to apply these methods. However, caution is required as parametric tests assume a normal distribution and perform poorly when assumptions about the data are not met. The input is a raw (not normalized, not rarefied) matrix with uneven column sums. With these techniques, it is still recommend to remove low depth samples (e.g., below 1000 sequences per sample), and low abundance/rare OTUs from the data sets. QIIME offers a diagnostic plot along with the differential abundance analyses. The DESeq2 method should not be used if the fit line on the dispersion plot is not smooth, if there are big gaps in the point spacing, or if the fitted line does not look appropriate to the data (Caporaso et al., 2010). DESeq2 is stronger when used with very small data sets, while MetagenomeSeq’s fitZIG uses an algorithm better suited for larger sized libraries with over 50 samples per category (the more the better). The results are presented in the form of a list of all of the OTUs in the input matrix, along with their associated statistics and the p-values that determine the statistical power of the differential abundance in the compared categories. These methods can be used in combination with the rarefied approaches to compare their outcomes. This manuscript is meant as a guide presenting recommended analyses for use in bioaerosol microbial ecology studies and the tools to achieve them. However, more technical detailed information can be found in the original papers describing the methods (Anders and Huber, 2010; Paulson et al., 2013; McMurdie and Holmes, 2014; Love et al., 2014).

259

In the context of differential abundance analyses, rarefied approaches are statistical tests that compare OTU frequencies in sample groups and ascertain whether or not there are statistically significant differences between the OTU abundances of different sample groups. Rarefying the samples prevents zero-variance errors and spurious significance for low abundance OTUs and focuses on the abundant OTUs, which likely play the most important role in the differential abundance. Examples of these approaches are the G-test, Kruskal- Wallis, ANOVA, Mann-Whitney U and T-test. Each test has its own null and alternate hypotheses and its own assumptions. It is important to check the sample size requirements, assumptions, and the null and alternate hypotheses of each test in order determine which is most appropriate for the dataset. Documentation on QIIME and R packages provides useful information on the subject, as does key literature on the subject of statistics in ecology (McDonald, 2014; Hollander et al., 2015). The three non-parametric tests (Kurskal-Wallis, nonparametric t-test and Mann-Whitney U) are most suited for bioaerosol sequencing data when the statistical distribution is not known. The t-test and Mann-Whitney U test may only be used when there are two sample groups, while Kruskal-Wallis can also be used when three or more groups of samples are compared (e.g., outdoor, indoor, source and samples).

A.9.4 Taxonomic Analyses

In general, after calculating diversity and richness indexes, visualizing sample grouping, and determining differential abundance, providing summary information of taxonomic groups represented within each sample is the next logical step. The taxonomic analysis uses an OTU table containing taxonomic information as input data. This information is based on public databases. The databases used should be chosen based on the gene marker used for the study. Greengenes is a 16S rRNA gene database suited for bacterial diversity (DeSantis et al., 2006). UNITE is more appropriate for the fungal ITS gene (Koljalg et al., 2005). SILVA is a wider database of small (16S/18S, SSU) and large subunit (23S/28S, LSU) rRNA sequences for all three domains of life (Bacteria, Archaea and Eukarya; Quast et al., 2013). Next, the taxonomic level for which the summary information is provided is designated. This level will depend on the format of the taxon strings that are returned from the taxonomy assignment step. The taxonomy strings that are most useful are those that

260

standardize the taxonomic level with the depth in the taxonomic strings. For instance, for the RDP classifier taxonomy: level 2 = Domain (e.g., Bacteria), 3 = Phylum (e.g., Firmicutes), 4 = Class (e.g., Clostridia), 5 = Order (e.g., Clostridiales), 6 = Family (e.g., Clostridiaceae), and 7 = Genus (e.g., Clostridium). Although, the relative abundance of each taxonomic group is the most used technique to compare taxa, raw counts can also be used for an absolute abundance. Results can be displayed with bar or area charts comparing taxonomy between groups of samples or between all individual samples. In addition, each pair of samples can be compared and the number of their shared OTUs is displayed in order to focus only on common OTUs between groups of samples. Comparing taxonomy information of bioaerosol samples to a mock community sample can help in determining technical biases linked to sequencing approaches.

Furthermore, the inclusion of taxonomic information in the mapping file allows NMDS or MDS plots to be colored based on taxonomy. More specifically, results displayed on principal coordinate plots can be colored based on any of the metadata fields in the mapping file. Coloration of the plots based on the relative abundances of each taxa can help in distinguishing which taxonomic groups are responsible for the sample grouping patterns.

Taxonomic analyses can also include the calculation of the ratio of abundance of specified taxonomic groups. This method is based on the microbial dysbiosis index described by Gevers and his coauthors (2014). Microbial Dysbiosis index (MD-index) is used as an indicator of the microbial imbalance within samples. One should specify the taxonomic groups to be used for the analyses according to their susceptibility to being affected by the different environmental conditions that define the samples. This index provides the option to choose the numerator and the denominator of the log ratio. The index must include the taxonomic groups that will be tested for increase (numerator) and decrease (denominator). For example, the ratio comparing firmicutes and proteobacteria would have firmicutes as the numerator and the proteobacteria as the denominator. To determine the taxonomic biomarkers, one can use a distance matrix plotted on ordination and validate which variable in the metadata mapping file best/most explains the variation observed, and then use taxonomic analyses to visualize the taxonomic composition of the samples based on the

261

variable chosen. That way, it is possible to determine which taxonomic groups exhibit differential abundance and can be used for the specified MD-index. The comparisons between samples based on microbial dysbiosis and the categories they belong to in the metadata mapping file can help determine which environmental condition creates a microbial dysbiosis. In bioaerosol studies, the analyses of dysbiosis can be very useful in determining if there is a microbial imbalance between a given source and the aerosols released.

Finally, identification of the core microbiome is another example of taxonomic analyses that provide useful information on the ecology of bioaerosols. The core of a microbiome is defined as the minimum community of microbes that is essential for a well- functioning ecosystem. This concept that has mostly been applied to the gut ecosystem may also be applicable to bioaerosols (Turnbaugh et al., 2009; Shade and Handelsman, 2011). The identification of the species that are found in a certain percentage (e.g., 50% to 95%) of all aerosol samples from a specific environment can determine the core microbial composition (core microbiome) of the environment being investigated. The importance of characterizing a core microbiome for each environment is extremely evident when searching for biomarkers of bioaerosol exposure in hazardous environments. The characterization of these biomarkers plays a key role for better evaluating the risk of bioaerosol exposure and will help in the standardization of bioaerosol studies.

A.9.5 Additional Visualization Tools

Creating a scatterplot representing average distances between samples, broken down by specified parameters (categories) is an alternative way to compare the microbial compositions of samples. The inputs are a distance matrix and a mapping file. The x-axis represents a category and must be numerical. In the primary state, each sample within the category will be compared to the other samples (or the one representing the secondary state) and an average of their distances will be calculated. The average distances will be plotted against a numerical category and are represented in the y-axis. The numerical category in the x-axis should preferably be linear and correlated somehow to the primary state. The points on the plot can then be colored according to another defined category. Thus, we have average

262

distances between the groups we are comparing according to a linear parameter (e.g., variation of the microbial composition of bioaerosols according to days, temperature etc.).

Similar to scatterplots, boxplots can be used to compare distances between categories of samples. The boxplots can compare distances within all samples of a category, as well as between different categories. Thus, individual-, within- and between-distances can be plotted. The input for a scatterplot is a distance matrix with the mapping file explaining the categories of samples. Statistical test comparing all combinations of paired boxplots can help determine which microbial distributions are significantly different from the others

In addition to using NMDS and MDS plots, building a neighbor joining tree or a UPGMA (Unweighted Pair Group Method with Arithmetic mean) tree that compares samples, using a distance matrix as input, is another way to examine sample grouping. Neighbor joining is an agglomerative clustering method for creating phylogenetic trees. Typically used for trees based on DNA data, the algorithm requires knowledge of the distance between each pair of taxa. In this case, it is used to cluster samples. Compared to UPGMA, the advantage of neighbor joining is that it does not assume that all lineages evolve at the same rate (Saitou and Nei, 1987).

Information in an OTU table can be visualized as a heatmap where each row corresponds to an OTU and each column corresponds to a sample. The higher the relative abundance of an OTU in a sample, the more intense the color at the corresponding position on the heatmap. The OTUs can be clustered by UPGMA hierarchical clustering, and the samples are presented in the order in which they appear in the OTU table. This is useful for establishing a general overview of the samples that have equal abundance of OTUs and are clustered together. However, identification of specific OTUs is difficult to visualize when the number of OTUs from the OTU table is very high. Therefore, taxonomic analyses are preferred for OTU identification.

263

A.9.6 Correlations

One common application of distance matrix comparison techniques is to determine if a correlation exists between an ecological distance matrix (e.g., UniFrac distance matrix) and a second matrix derived from an environmental parameter that is numeric/continuous (e.g., differences in pH, temperature, or geographical location). For example, one might be interested in knowing if aerosol samples with different pH levels are more different from one another than from aerosol samples with similar pH levels. If so, this would indicate a positive correlation between the two distance matrices. Mantel correlation tests allow for the comparison of two or more distance/dissimilarity matrices to determine if there is a correlation. It tests the hypothesis that distances between samples within a given matrix are linearly independent of the distances within those same samples in a separate matrix. A Mantel correlogram produces a plot of distance classes versus Mantel statistics. Briefly, an ecological distance matrix and a second distance matrix (e.g., spatial distances, pH distances, etc.) are provided. In the second distance matrix distances are split into a number of distance classes (this number is determined by Sturge’s rule). A Mantel test is applied to these distance classes versus the ecological distance matrix. The Mantel statistics obtained from each of these tests can then be plotted in a correlogram. A filled symbol on the plot indicates that the Mantel statistic was statistically significant (Legendre and Fortin, 2010).

Moreover, correlations between abundances (relative or absolute) and numerical metadata can also be used to correlate features to sample metadata values. Several methods are available to accomplish this. Pearson is a parametric and linear measure of correlation. It is a scaled measure of the degree to which two sequences of numbers co-vary. For correlated sequences, Pearson > 0, and for anticorrelated sequences, Pearson < 0 (uncorrelated implies Pearson = 0). The Spearman correlation is a nonparametric measure of the correlation between two sequences of numbers. Kendall’s Tau is an alternative method of calculating correlations between two sequences of numbers. However, it is slower and utilized less often than Spearman or Pearson scores (Rohlf and Sokal, 1995). Statistics can be added to these correlation approaches in order to generate p-values to confirm the correlation scores obtained. Bootstrapping is the most robust procedure for calculating the p-value of a given

264

correlation score. Bootstrapping takes the input sequences, randomly changes the order of one, and then recomputes the correlation score. The p-value represents the number of times (out of the given number of permutations) that the score of the permuted sequence pair was more extreme than the observed pair. Bootstrapping is preferred when information about statistical distributions is unknown.

Finally, the correlation between samples in terms of their taxonomic composition can also be computed. This is useful for determining if the taxonomic compositions of mock communities that were assigned using different taxonomy assigners are correlated. Another usage is to compare the taxonomic compositions of several mock community samples to a single known sample community. In general, correlations in the taxonomic composition between different groups of samples can be useful (e.g., aerosol samples collected from different sites). The correlation coefficient, an associated confidence interval, and p-values (nonparametric or parametric) should also be included using the method discussed previously.

A.10 Conclusion

The analysis of microbial diversity is becoming a crucial component in several fields of scientific research, and bioaerosols is no exception. Many of the bioinformatics tools used to study microbial diversity were developed for researchers comfortable with a command line environment. This manuscript is intended as a guide to the types of useful bioinformatics tools that provide a thorough investigation of the microbial communities of bioaerosols. Many questions can be answered, hypotheses confirmed and critical thinking can be triggered by such analyses. Thus, the main goal is not to provide command lines about how to perform the analyses, but to offer important information and insight on tests typically used in microbial ecology. We do this by providing examples of their application in bioaerosols studies. Bioinformatics tools are still underutilized by bioaerosol scientists and they can, in some cases, lead to spurious analyses and interpretations. The authors hope that this work represents a conduit for the popularization of bioinformatics in the study of bioaerosols and

265

will provide a good source for the «dos and don’ts» when conducting a critical microbial community study.

Acknowledgements

HM is a recipient of the FRQNT PhD scholarship as well as a scholarship for a short internship from the Quebec Respiratory Health Network. The authors are thankful to Amanda Kate Toperoff and Michi Waygood for English revision of the manuscript. CD is the head of the Bioaerosols and respiratory viruses strategic group of the Quebec Respiratory Health Network.

Competing interests

The authors declare no competing financial interests.

A.8 Bibliography

Anders, S., Huber, W., 2010. Differential expression analysis sequence count data. Genome Biol. 11(10):R106.

Anderson, M.J., 2005. PERMANOVA: a FORTRAN computer program for permutational multivariate analysis of variance. Department of Statistics, University of Auckland, New Zealand.

Anderson, M.J., Walsh, D.C.I., 2013. PERMANOVA, ANOSIM, and the Mantel test in the face of heterogeneous dispersions: What null hypothesis are you testing? Ecol. Monogr. 83(4):557-574.

Benìtez-Pàez, A., Sanz, Y., 2017. Multi-locus and long amplicon sequencing approach to study microbial diversity at species level using the MinIon™ portable nanopore sequencer. Gigascience. 6(7):1-12.

Berry, K.J., Wong, R.K.W., 1986. Approximate mrpp p-values obtained from four exact moments. Commun. Stat. Simul. Comput. 15(2):581-589.

Bolhuis, H., Cretoiu, M.S. and Stal, L.J., 2014. Molecular ecology of microbial mats. FEMS Microbiol. Ecol. 90(2):335-350.

266

Bray, J.R., Curtis, J.T., 1957. An ordination of upland forest communities of southern Wisconsin. Ecol. Monogr. 27(4):325-349.

Braz, G.J., de Paiva, A.C., Silva, A.C., de Oliveira, A.C.M., 2009. Classification of breast tissues using Moran’s Index and Geary’s coefficient as texture signatures and SVM. Comput. Biol. Med. 39(12):1063-1072.

Buttigieg, P.L., Ramette, A., 2014. A Guide to Statistical Analysis in Microbial Ecology: a community-focused, living review of multivariate data analyses. FEMS Microbiol Ecol. 90(3):543–550.

Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Huntley, J., Fierer, N., et al., 2012. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 6(8):1621-1624.

Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Lozupone, C.A., Turnbaugh, P.J., et al., 2011. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc. Natl. Acad. Sci. U.S.A. 108(Suppl 1):4516-4522.

Clarke, K.R. Somerfield, P.J., Gorley, R.N., 2008. Testing null hypotheses in exploratory community analyses: similarity profiles and biota-environment linkage. J. Exp. Mar. Biol. Ecol. 366(1-2):56-69.

Costea, P.I., Zeller, G., Sunagawa, S., Bork, P., 2014. A fair comparison. Nat. Methods 11:359

Davenport, C.F., Tümmler, B., 2013. Advances in computational analysis of metagenomes sequences. Environ. Microbiol. 15(1):1-5.

De Filippis, F., Laiola, M., Blaiotta, G., Ercolini, D., 2017. Different Amplicon Targets for Sequencing-Based Studies of Fungal Diversity. Appl. Enviro. Microbiol. 83(17):e00905-17.

DeSantis, T.Z., Hugenholtz, P., Larsen, N., Rojas, M., Brodies, E.L., Keller, K., et al., 2006. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microbiol. 72(7):5069-5072.

Dillies, M.A., Rau, A., Aubert, J., Hennequet-Antier, C., Jeanmougin, M., Servant, N., et al., 2013. A comprehensive evaluation of normalization methods for Illumina high-throughput RNA sequencing data analysis. Brief. Bioinform. 14(6):671-683.

267

Dubuis, M.E., M’Bareche, H., Veillette, M., Bakhiyi, B., Zayed, J., Lavoie, et al., 2017. Bioaerosols concentrations in working areas in biomethanization facilities. J. Air Waste Manag. Assoc. 67(11):1258-1271.

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 26(19):2460-2461.

Edgar, R.C., 2013 UPARSE: Highly accurate OTU sequences from microbial amplicon reads. Nat. Methods. 10:996-998.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 27(16):2194-2200.

Erceg-Hurn, D.M., Mirosevich, V.M., 2008. Modern robust statistical methods: An easy way to maximize the accuracy and power of your research. Am. Psychol. 63(7):591-601.

Faith, D. P., Baker, A. M. 2006. Phylogenetic diversity (PD) and biodiversity conservation: some bioinformatics challenges. Evol. Bioinform. Online. 2:121–128.

Gandolfi, I., Bertolini, V., Ambrosini, R., Besttetti, G., Franzetti, A., 2013. Unravelling the bacterial diversity in the atmosphere. Appl. Microbiol. Biotechnol. 97(11):4727-4736.

Garcìa, S., Fernàndez, A., Luengo, J., Herrera, F., 2010. Advanced nonparametric tests for multiple comparisons in the design of experiments in computational intelligence and data mining: Experimental analysis of power. Inf. Sci. 180(10):2044-2064.

Gevers, D., Kugathasan, S., Denson, L.A., Vasquez-Baeza, Y., Van Treuren, W., Ren, B. al., 2014. The treatment-naïve microbiome in new-onset Crohn’s disease. Cell Host Microbe. 15(3):382-392.

Gilbert, J.A., Blaser, M.J., Caporaso, J.G., Jansson, J.K., Lynch, S.V., Knight, R., 2018. Current understanding of the human microbiome. Nat. Med. 24(4):392-400.

Goodrich, J.K., Di Rienzi, S.D., Poole, A.C., Koren, O., Walters, W.A., Caporaso, J.G., et al., 2014. Conducting a Microbiome Study. Cell. 158(2):250-262.

Gower, J.C., Legendre, P., 1986. Metric and Euclidean properties of dissimilarity coefficients. J. Class. 3:5.

Hogeweg, P., 2011. The roots of bioinformatics in theoretical biology. PLoS Comput. Biol. 7(3):e1002021.

Hollander, M., Wolf, D.A., Chicken, E. 2015. Nonparametric statistical methods (2nd ed.). John Wiley and Sons Inc. New York, NY, USA.

268

Hugenholtz, P., Pace, N.R., 1996. Identifying microbial diversity in the natural environment: a molecular phylogenetic approach. Trends Biotechnol. 14(6):190-197.

Human Microbiome Project Consortium, 2012. A framework for human microbiome research. Nature. 486:215-221.

Huttenhower, C., Knight, R., Brown, C.T., Caporaso, J.G., Clemente, J.C., Gevers, D. et al., 2014. Advancing the microbiome research community. Cell. 159(2):277-230.

Jamison, D.C., 2003. Introduction, in Perl Programming for Biologists. Hoboken, NJ: John Wiley & Sons, Inc. New York, NY, USA. Knight, R., Jansson, J., Field, D., Fierer, N., Desai, N., Fuhrman, J.A., al., 2012. Unlocking the potential of metagenomics through replicated experimental design. Nat. Biotechnol. 30(6):513–520.

Koljalg, U., Larsson, K.H., Abarenkov, K., Nilsson, R.H., Alexander, I.J., Eberhardt, U., et al., 2005. UNITE: a database providing web-based methods for the molecular identification of ectomycorrhizal fungi. New phytol. 166(3):1063-8.

Kuczynski, J., Liu, Z., Lozupone, C., McDonald, D., Fierer, N., knight, R., 2010. Microbial community resemblance methods differ in their ability to detect biologically relevant patterns. Nat. Methods. 7:813-819.

Legendre, P., Anderson M.J. 1999. Distance-based redundancy analysis: Testing multispecies responses in multifactorial ecological experiments. Ecol. Monograr. 69:1–24.

Legendre, P., Fortin, M.J., 2010. Comparison of the Mantel test and alternative approaches for detecting complex multivariate relationships in the spatial analyses of genetic analyses. Mol. Ecol. Resour. 10(5):831-844.

Leprieur, F., Albouy, C., De Bortoli, J., Cowman, P.F., Bellwood, D.R., Mouillot, D., 2012. Quantifying phylogenetic beta diversity: distinguishing between ‘true’ turnover of lineages and phylogenetic diversity gradients. PLos ONE. 7:e42760.

LeRoy Poff, N., Olden, J.D., Merritt, D.M., Pepin, D.M., 2007. Homogenization of regional river dynamics by dams and global biodiversity implications Proc. Nat. Acad. Sci. 104(14):5732-5737.

Love, M., Huber, W., Anders, S., 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15(12):550.

Lozupone, C. A., Knight, R., 2007. Global patterns in bacterial diversity. Proc. Nat. Acad. Sci. U. S. A. 104(27):11436–11440.

Lozupone, C., Hamady, M., Knight, R., 2006. UniFrac- an online tool for comparing microbial community diversity in a phylogenetic context. BMC Bioinformatics. 7:371.

269

Lozupone, C.A., Knight, R., 2008. Species divergence and the measurement of microbial diversity. FEMS Microbiol. Rev. 32(4):557-578.

Lu, B., Charlton, M., Fotheringhama, A.A., 2011. Geographically Weighted Regression Using a Non-Euclidean Distance Metric with a Study on London House Price Data. Procedia Environ. Sci. 7:92-97.

Madsen, A.M., Zervas, A., Tendal, K., Nielsen, J.L., 2015. Microbial diversity in bioaerosol samples causing ODTS compared to reference bioaerosol samples as measured using Illumina sequencing and MALDI-TOF. Environ. Res. 140:255-267.

Magurran & McGill. 2011. (Editors) Biological diversity. Oxford University Press. London, UK.

Mardis, E.R., 2008. Next-Generation DNA sequencing methods. Annu. Rev. Genomics Hum. Genet. 9:387-402.

Mbareche, H., Brisebois, E., Veillette, M., Duchaine, C., 2017a. Bioaerosol sampling and detection methods based on molecular approaches: No pain no gain. Sci. Total Environ. 599- 600:2095-2104.

Mbareche, H., Veillette, M., Bonifait, L., Dubuis, M.E., Bernard, Y., Marchand, G., et al., 2017b. A next generation sequencing approach with a suitable bioinformatics workflow to study fungal diversity in bioaerosols released from two different types of composting plants. Sci. Total Environ. 601-602:1306-1314.

Mbareche, H., Veillette, M., Pilote, J., Létourneau, V., Duchaine, C., 2018. Corrélation entre la diversité bactérienne des bioaérosols de bâtiments de porcheries et des nasopharynx des travailleurs exposés. presented at Journées de la recherché 2018 de l’Institut Universitaire de Cardiologie et de Pneumologie de Québec, 30th may 2018, Québec, Canada.

McArdle, B.H., Anderson, M.J., 2001. Fitting multivariate models to community data: a comment on distance-based redundancy analysis. Ecology. 82(1):290-297.

McDonald, J.H., 2014. Handbook of Biological Statistics (3rd ed.). Sparky House Publishing. Baltimore, Maryland, USA.

McMurdie, P.J., Holmes, S., 2014. Waste not, want not: why rarefying microbiome data is inadmissible. PLoS Comput. Biol. 10(4): e1003531.

McMurdie, P.S., Holmes, S. 2013. phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS ONE. 8(4):e61217.

Naqib, A., Silvana, P., Wang, W., Hyde, M., Kuntsman, K., Green, S.J., 2018. Making and sequencing heavily multiplexed, high-throughput 16S ribosomal RNA gene Amplicon libraries using a flexible, two-stage PCR protocol. Methods Mol. Biol. 1783:149-169.

270

Navas-Molina, J.A., Peralta-Sànchez, J.M., Gonzàlez, A., McMurdie, P. J., Vàsquez-Baeza, Y., Xu, Z. et al., 2013. Advancing our understanding of the human microbiome using QIIME. Meth. Enzymol. 531:371-444.

Nilsson, R.H., Kristiansson, E., Ryberg, M., Hallenberg, N., Larsson, K.H., 2008. Intraspecific ITS Variability in the Kingdom Fungi as Expressed in the International Sequence Databases and Its Implications for Molecular Species Identification. Evol. Bioinform. Online. 4:193–201.

Novais, R.C., Thorstenson, Y.R., 2011. The evolution of pyrosequencing® for microbiology: from genes to genomes. J. Microbiol. Methods. 86(1):1-7.

Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Stevens, M.H.H., 2007. The vegan package. Community ecology package. 10:631-637 http://r-forge.r-project.org/projects/vegan/.

Paulson, J.N., Stine, O.O., Bravo, H.O., Pop, M., 2013. Robust methods for differential analysis in marker gene surveys. Nat. Methods. 10(12): 1200-1202.

Peccia, J., Hospodsky, D., Kyle, B., 2010. New Directions: A revolution in DNA sequencing now allows for the meaningful integration of biology with aerosol science. Atmos. Environ. 45(10):1896-1897.

Philippot, L., Raaijmakers, J.M., Lemanceau, P., and van der putten, W.H., 2013. Going back to the roots: the microbial ecology of the rhizosphere. Nat. Rev. Microbio. 11(11):789-799.

Pollock, J., Glendinning L., Wisedschanwet, T., Watson, M., 2018. The madness of microbiome: Attempting to find consensus “best practice” for 16S microbiome studies. Appl. Environ. Microbiol. 84(7):e02627-17.

Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., et al., 2013. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 41(Database issue):D590–D596.

Quinn, G.P., Keough, M.J., 2002. Experimental design and data analyses for biologists. Cambridge: Cambridge University Press. Cambridge, UK.

Ramette, A., 2007. Multivariate analyses in microbial ecology. FEMS Microbiol. Ecol. 62(2):142-160.

Rognes, T., Flouri, T., Nichols, B., Quince, C., Mahé, F., 2016. VSEARCH: a versatile open source tool for metagenomics. PeerJ. 4:e2584.

Rokach, L., Maimon, O., 2005. Clustering methods. Data mining and knowledge discovery handbook. Springer. New York, NY, USA.

271

Saitou, N., Nei, M., 1987. The neighbor-joining method: an new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4(4):406-25.

Saltelli, A., Marivoet, J., 1990. Non-parametric statistics in sensitivity analysis for model output: A comparison of selected techniques. Reliability Engineering & System Safety. 28(2):229-253

Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., al., 2009. Introducing mothur: open-source, platform-independent, community supported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75:7537–7541. Schoch, C.L., Seifert, K.A., Huhndorf, S., Robert, V., Spouge, J.L., Levesque, A.C., et al., 2012. Nuclear ribosomal internal transcribed spacer (ITS) region as a universal DNA barcode marker for Fungi. Proc. Nat. Acad. Sci. 109(16):6241-6246.

Shade, A., Handelsman, J., 2011. Beyond the Venn diagram: the hunt for a core microbiome. Environ Microbiol. 14(1):4-12.

Sims, D., Sudbury, I., Ilott, N.E., Heger, A., Ponting, C.P., 2014. Sequencing depth and coverage: key considerations in genomic analyses. Nat. Rev. Genet. 15(2);121-132.

Sinha, R., Abu-Ali, G., Vogtmann, E., Fodor, A.A., Ren, B., Amir, A., et al., 2017. Assessment of variation in microbial community amplicon sequencing by the Microbiome Quality Control (MBQC) project consortium. Nat. Biotechnol. 35:1077-1086.

Smith, D.R., 2015. Broadening the definition of a bioinformatician. Front. Genet. 6:258.

Sokal, R.R., Rohlf, F.J., 1995. The principles and practices of statistics in biological research. 4th ed. Stony Brook University. WH Freeman and Company. New York, NY, USA.

Tuomisto, H., 2010. A diversity of beta diversities: straightening up a concept gone awry. Part 1. Defining beta diversity as a function of alpha and gamma diversity. Ecography. 33(1):2-22.

Turnbaugh, P.J., Hamady, M., Yatsunenko, T., Cantarel, B.L., Duncan, A., Ley, R.E., et al., 2009. A core gut microbiome in obese and lean twins. Nature. 457(7228):480-484.

Venter, J.C., Remington, K., Heidleberg, J.F., Halpern, A.L., Rusch, D., Eisen, J.A., et al., 2004. Environmental genome shotgun sequencing of the Sargasso Sea. Science. 304(5667):66–74.

Vickers, A.J., 2005. Parametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data. BMC Med. Res. Methodol. 5:35.

Vincent, A.T., Charette, S.J., 2015. Who qualifies to be a bioinformatician? Front. Genet. 6:164.

272

Walter, J., Ley, R., 2011. The human gut microbiome: ecology and recent evolutionary changes. Annu. Rev. Microbiol. 65:411-429.

Wang, J., Shen, J., Wu, Y., Tu, C., Soinine, J., Stegen, J.C. et al., 2013. Phylogenetic beta diversity in bacterial assemblages across ecosystems: deterministic versus stochastic processes. ISME J. 7(7):1310-1321.

Weiss, S.J., Xu, Z.Z., Amir, A., Peddada, S., Bittinger, K., Gonzalez, A., et al., 2015. Effect of library size variance, sparity, and compositionality on the analysis of microbiome data. PeerJ PrePrints. 3e1157v1.

Whittaker, R.H., 1972. Evolution and Measurement of Species Diversity. Taxon. 21(2/3):213-251.

Yamamoto, N., Nazaroff, W., Peccia, J., 2014. Assessing Aerodynamic diameters of taxon- specific fungal bioaerosols by quantitative PCR and next-generation DNA sequencing. J. Aerosol Sci. 78:1-10.

Yoon, S.S., Kim, E.K., Lee, W.J., 2015. Functional genomic and metagenomic approaches to understanding gut microbiota-animal mutualism. Curr. Opin. Microbiol. 24:38-46.

Zimmerman, D.W., 1994. A note one the influence of outliers on parametric and nonparametric tests. J. Gen. Psychol. 121(4):391-401.

Zur, E.F., Ieno, E.N., Smith, G.M., 2007. Analyzing ecological data. Springer. New York, NY, USA.

273

Bibliography

Abarenkov, K., Henrik Nilsson, R., Larsson, K.H., Alexander, I.J., Eberhardt, U., Erland, S., et al., 2010. The UNITE database for molecular identification of fungi—recent updates and future perspectives. New Phytol. 186:281–28.

Abassi, M., Boulware, D.R., Rhein, J., 2015. Cryptococcal Meningitis: Diagnosis and Management Update. Curr. Trop. Med. Rep. 2(2):90-99.

Agranovski, V., Ristovski, Z., Hargreaves, M., Blackall, P.J., Morawska, L., 2003. Real-time measurement of bacterial aerosols with UVAPS: performance evaluation. J. Aerosol Sci. 34(3):301-317.

Aleksic, B., Draghi, M., Ritoux, S., Bailly, S., Lacroix, M., Oswald, I.P., 2017. Aerosolization of mycotoxins after growth of toxicogenic fungi on wallpaper. App. Environ. Microbio. 83(16): e01001-17.

Almeida, F., Rodrigues, M.L., Coelho, C., 2019. The Still Underestimated Problem of Fungal Diseases Worldwide. Front. Microbiol. 10:214.

Alshannaq, A., Yu, J.H., 2017. Occurrence, toxicity, and analysis of major mycotoxins in food. Int. J. Environ. Res. Public Health. 14(6):632.

Amann, R., Ludwig, W., Schleifer, K.H., 1995. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiol. Rev. 59:143–169.

Amann, R., Rosselló-Móra, R., 2016. After All, Only Millions? mBio. 7(4):e00999-16.

Amann, R.I., Binder, B.J., Olson, R.J., Chisholm, S.W., Devereux, R., Stahl, D.A., 1990. Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Appl. Environ. Microbiol. 56(6):1919-1925.

274

Amir, A., McDonald, D., Navas-Molina, J.A., Kopylova, E., Morton, J.T., Xu, Z.Z. et al., 2017. Deblur rapidly resolves single-nucleotide community sequence patterns. mSystems 2: e00191–16.

Amrane, S., Raoult, D., Lagier, J.C., 2018. Metagenomics, culturomics, and the human gut microbiota. Expert Rev. Anti-infect. Ther. 16(5): 373-375.

Anastasi, A., Varese G.C., Marchisio, V.F., 2005. Isolation and identification of fungal communities in compost and vermicompost. Mycologia. 97(1):33-44.

Anders, S., Huber, W., 2010. Differential expression analysis sequence count data. Genome Biol. 11(10):R106.

Anderson, M.J. 2005. PERMANOVA: a FORTRAN computer program for permutational multivariate analysis of variance. Department of Statistics, University of Auckland, New Zealand.

Anderson, M.J., Walsh, D.C.I., 2013. PERMANOVA, ANOSIM, and the Mantel test in the face of heterogeneous dispersions: What null hypothesis are you testing? Ecol. Monogr. 83(4):557-574.

Ando, M., Hirayama, K., Soda, K., Okudo, R., Araki, S., Sasazuki, T., 1989. HLA-DQw3 in Japanese summer-type hypersensitivity pneumonitis induced by Trichosporon cutaneum. Am. Rev. Respir. Dis. 140(4):948-950.

Bariola, J.R., Perry, P., Pappas, P.G., Proia, L., Shealey, W., Wright, P.W., 2010. Blastomycosis of the central nervous system: A multicenter review of diagnosis and treatment in the modern era. Clin. Infect. Dis. 50(6):797-804.

275

Bass, D., Czech, L., Williams, B., Berney, C., Dunthorn, M., Mahé, F., et al., 2018. Clarifying the relationships between Microsporidia and Cryptomycota. J Eukaryot. Microbiol. 65(6):773-782.

Baxi, S.N., Portnoy, J.M., Larenas-Linnemann, D., Phipatanakul, W., 2016. Environmental Allergens Workgroup. Exposure and health effects of fungi on humans. J. Allergy Clin. Immunol. Pract. 4(3):396-404.

Bazzicalupo, A.L., Bàlint, M., Schmitt, I., 2013. Comparison of ITS1 and ITS2 rDNA in 454 sequencing of hyperdiverse fungal communities. Fungal Ecology. 6(1):102-109.

Be, N.A., Thissen, J.B., Fofanov, V.Y., Allen, J.E, Rojas, M., Golovko, G., et al., 2015. Metagenomic analysis of the airborne environment in urban spaces. Microb. Ecol. 69(2):346- 55.

Beever, R.E., Dempsey, G.P., 1978. Function of rodlets on the surface of fungal spores. Nature. 272(5654):608-10.

276

Bennett, J.W., Klich, M., 2013. Mycotoxins. Clin. Microbiol. Rev. 16(3):497-516.

Berbee, M.L., James, T.Y.m Strullu-Derrien, C., 2017. Early diverging fungi: diversity and impact at the dawn of terrestrial life. Ann. Rev. Microbiol. 71:41–60.

Berry, K.J., Wong, R.K.W., 1986. Approximate mrpp p-values obtained from four exact moments. Commun. Stat. Simul. Comput. 15(2):581-589.

Bickel, P.E., Tansey, J.T., Welte, M.A., 2009. PAT proteins, an ancient family of lipid droplet proteins that regulate cellular lipid stores. Biochim. Biophys. Acta. 1791(6):419–440.

Bidartondo, M.I., 2008. Preserving accuracy in GenBank. Science. 319(5870):1616.

Bishop, C.M., 2006. Book: Pattern recognition and machine learning. Ed. Jordan, M., Kleinberg, J., Schölkopf. Springer, New York, NY, USA.

Blaalid, R., Kumar, S., Nilsson, R.H., Abarenkov, K., Kirk, P.M., Klauserud, H., 2013. ITS1 versus ITS2 as DNA metabarcodes for fungi. Mol. Ecol. Res. 13(2): 218–224

Blackwell, M., 2011. The Fungi: 1, 2, 3 … 5.1 million species? Am. J.Bot. 98: 426–428.

Bladacci, S., Maio, S., Cerrai, S., Samo, G., Baïz, N., Simoni, M., Annesi-Maesano, I., Viegi, G., HEALS Study, 2015. Allergy and asthma: Effects of the exposure to particulate matter and biological allergens. Respir. Med. 109(9): 1089-1104.

Blais Lecours, P., Veillette, M., Marsolais, D., Duchaine, C., 2012. Characterization of Bioaerosols from Dairy Barns: Reconstructing the Puzzle of Occupational Respiratory Diseases by Using Molecular Approaches. Appl. Environ. Microbiol. 78(9):3242-3248.

277

Bolhuis, H., Cretoiu, M.S. and Stal, L.J., 2014. Molecular ecology of microbial mats. FEMS Microbiol. Ecol. 90(2):335-350.

Bongomin, F., Gago, S., Oladele, R.O., Denning, D.W., 2017. Global and Multi-National Prevalence of Fungal Diseases-Estimate Precision. J. Fungi (Basel). 3(4):E57.

Bonifait, L., Marchand, G., Veillette, M., M’Bareche, H., Dubuis, M.E., Pépin, C., et al., 2017. Workers’ exposure to bioaerosols from three different types of composting facilities. J. Occup. Environ. Hyg. 14(10):815-822.

Bonito,G., Isikhuenmhen,O. S., Vilgalys, R., 2010. Identification of fungi associated with municipal compost using DNA based techniques. Bioresour. Technol. 101(3):1021-1027.

Borchers, A.T., Chang, C., Gershwin, M.E., 2017. Mold and human health: a reality check. Clinic. Rev. Allerg. Immunol. 52:305-322.

Brasaemle, D.L., 2007. Thematic review series: adipocyte biology. The perilipin family of structural lipid droplet proteins: stabilization of lipid droplets and control of lipolysis. J. Lipid Res. 48(12):2547–2559. Braun, U., Schubert, K., 2007. Taxonomic revision of the genus Cladosporium s. lat. 7. Descriptions of new species, a new combination and further new data. Schlechtendalia. 16: 61–76.

Bray, J.R., Curtis, J.T., 1957. An ordination of upland forest communities of southern Wisconsin. Ecol. Monogr. 27(4):325-349.

Brisebois, E., Veillette, M., Dion-Dupont, V., Lavoie, J., Corbeil, J., Culley, A., et al., 2018. Human viral pathogens are pervasive in wastewater treatment center aerosols. J. Environ. Sci. 67:45-53

278

Brochers, A.T., Chang, C., Gershwin, E., 2017. Mold and human health: a reality check. Clin. Rev. Allergy Immunol. 52(3):305-322.

Brown, J.K.M., Hovmøller, M.S. 2002. Aerial dispersal of pathogens on the global and continental scales and its impact on plant disease. Science. 297:537–541.

Bru-Adan, V., Wéry, N., Moletta-Denat, M., Boiron, P., Delgènes, J.P., Godon J.J., 2009. Diversity of bacteria and fungi in aerosols during screening in a green waste composting plant. Curr. Microbiol. 59(3):326-335.

Bünger, J., Antlauf-Lammers, M., Schultz, T.G., Westphal, G.A., Müller, M.M., Ruhnau, P., et al., 2000. Health complaints and immunological markers of exposure to bioaerosols among biowaste collectors and compost workers. Occup. Environ. Med. 57(7):458-64.

Burge, H.A., Rogers, C.A., 2000. Outdoor allergens. Environ. Health Perspect. 108:653–659. Burge, H.A., Rogers, C.A., 2000. Outdoor allergens. Environ. Health Perspect. 108(Suppl 4):653–659.

Burrows, S.M., Butler, T., Jöckel, P., Tost, H., Kerkweg, A., Pöschl, U., Lawrence, M.G., 2009a. Bacteria in the global atmosphere-part 2: modeling of emissions and transport between different ecosystems. Atmos. Chem. Phys. 9:9281–9297.

279

Burrows, S.M., Elbert, W., Lawrence, M.G., Pöschl, U., 2009b. Bacteria in the global atmosphere–part 1: review and synthesis of literature data for different ecosystems. Atmos. Chem. Phys. 9:9263–9280.

Bush R.K., Portnoy J.M., Saxon A., Terr A.I., Wood R.A. 2006. The medical effects of mold exposure. J. allergy immunol. 117(2): 326-333.

Bush, R.K., 2008. Indoor allergens, environmental avoidance, and allergic respiratory disease. Allergy and Asthma Proc. 29(6):575–579.

Bush, R.K., Portnoy, J.M., Saxon, A., Terr, A.I., Wood, R.A., 2006. The medical effects of mold exposure. J. Allergy Clin. Immunol. 117(2):326-333.

Buttigieg, P.L., Ramette, A., 2014. A Guide to Statistical Analysis in Microbial Ecology: a community-focused, living review of multivariate data analyses. FEMS Microbiol. Ecol. 90(3):543–550.

Buttigieg, P.L., Ramette, A., 2014. A Guide to Statistical Analysis in Microbial Ecology: a community-focused, living review of multivariate data analyses. FEMS Microbiol Ecol. 90(3):543–550.

Callahan, B. J., McMurdie, P. J., Rosen, M. J., Han, A. W., Johnson, A. J. A., Holmes, S. P., 2016. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods. 13(7):581–583. Callahan, B.J., McMurdie, P.J., Holmes, S.P., 2017. Exact sequence variants should replace operational taxonomic units in marker-gene data analyses. ISME J. 11(12): 2639-2643.

Carducci, A., Donzelli, G., Cioni, L., Federigi, I., Lombardi, R., Verani, M., 2018. Quantitative microbial risk assessment for workers exposed to bioaerosol in wastewater treatment plants aimed at the choice and setup of safety measures. Int. J. Environ. Red. Public Health. 15(7):E1490.

Carneiro-Sampaio, M., Coutinho, A., 2007. Immunity to microbes: lessons from primary immunodeficiencies. Infect. Immun. 75(4):1545-1555.

280

Carroll, C. 2009. EPA nanoparticle air monitoring workshop. 2-3 March 2009. Session #2. Technology Needs and Gaps [pdf slides]. Retrieved from https://www3.epa.gov/ttnemc01/meetnw/2009/nano7.pdf

Casadevall, A., 2017. Don’t forget the fungi when considering global catastrophic biorisks. Health Secur. 15:341–342.

Chizhikov, V., Rasooly, A., Chumakov, K., Levy, D.D. 2001. Microarray analysis of microbial virulence factors. Appl. Environ. Microbiol. 67(7):258-3263.

Choi, J.J., Kim, S.H., 2017. A genomic tree of life for the fungi kingdom. PNAS. 114(35):9391-9396.

Chowdhary, A., Agarwal, K., Meis, J.F., 2016. Filamentous fungi in respiratory infections. What lies beyond aspergillosis and mucormycosis? PLoS Pathog. 12(4): e1005491. Clarke, K.R. Somerfield, P.J., Gorley, R.N., 2008. Testing null hypotheses in exploratory community analyses: similarity profiles and biota-environment linkage. J. Exp. Mar. Biol. Ecol. 366(1-2):56-69.

Cole, J.R., Wang, Q., Cardenas, E., Fish, J., Chai, B., Farris, R.J., et al., 2009. The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucleic Acids Res. 37 (Database issue):D141–D145.

Collagher, M., Campbell, J., Brenner, A.M., 2017. Collaboration achieves effective management design at Brookfield Place Berth, Western Australia. Procedia Eng. 180:1763– 1772.

Cordier, T., Forster, D., Dufresne, Y., Martins, C.I.M, Stoeck, T., Pawlowski, J., 2018. Supervised machine learning outperforms taxonomy-based environmental DNA metabarcoding applied to biomonitoring. Mol. Ecol. Res. 18(6):1381-1391.

Cormier, Y., Bélanger, J., Durand, P., 1985. Factors influencing the development of serum precipitins to farmer’s lung antigen in Quebec dairy farmers. Thorax. 40(2):138-142.

281

Costea, P.I., Zeller, G., Sunagawa, S., Bork, P., 2014. A fair comparison. Nat. Methods 11:359

Coulon, F., Colbeck, I., 2018. RAMBIE, Rapid monitoring of bioaerosols in Urban, Agricultural and Industrial Environments, NERC. 2018(6):14-26.

Cretu, R., Dima, S., Bahrim, G., 2011. Antioxidant activity evaluation of the fungal dye oil- in-water emulsions obtained by cyclic voltammetric technique. Rom. Biotechnol. Lett. 16(6)19-33.

Daisey, J.M., Angell, W.J. and Apte, M.G., 2003. Indoor air quality, ventilation and health symptoms in schools: an analysis of existing information. Indoor Air. 13:53-64.

Damialis, A., Kaimakamis, E., Konoglou, M., Akritidis, I., Traidl-Hoffmann, C., Gioulekas, D., 2017. Estimating the abundance of airborne pollen and fungal spores at variable elevations using an aircraft: how high can they fly? Sci. Rep. 7:44535.

Davenport, C.F., Tümmler, B., 2013. Advances in computational analysis of metagenomes sequences. Environ. Microbiol. 15(1):1-5.

De Filippis, F., Laiola, M., Blaiotta, G., Ercolini, D., 2017. Different Amplicon Targets for Sequencing-Based Studies of Fungal Diversity. Appl. Enviro. Microbiol. 83(17):e00905-17.

De Gannes, V., Eudoxie, G., William, H.J., 2013. Insights into fungal communities in composts revealed by 454-pyrosequencing: implications for human health and safety. Front. Microbiol. 4:164. de Hoog, G.S., Guarro, J., Gené, J., Figueras, M.J., 2000. Atlas of Clinical Fungi. Utrecht/Reus:Centraalbureau voor Schimmelcultures/Universitat Rovira i Virgili. p. 1126. 2nd ed. de Vries, R.P., Burgers, K., Van De Vondervoort, P.J.I., Frisvad, J.C., Samson, R.A.,Visser, J., 2004. A new black aspergillus species, A. vadensis, is a promising host for homologous and heterologous protein production. Appl. Environ. Microbiol. 70(7)3954-3959.

282

Deacon, J.W., 2006. Chapter 1: Introduction: the fungi and fungal activities. In book: Fungal biology (4th ed.). Blackwell Publishing, Malden, MA, USA.

Dentinger, B.T., Didukh, M.Y., Moncalvo, J.M., 2011. Comparing COI and ITS as DNA barcode markers for mushrooms and allies (Agaricomycotina). PLoS One. 6:e25081.

Després, V.R., Huffman, J.A., Burrows, S.M., Hoose, C., Safatov, A.S., Buryak, G., et al., 2012. Primary biological aerosol particles in the atmosphere: a review. Tellus B. 64(1). 15598.

Dogra, V., Sinha, A.K., Saxena, R., Talwar, D., 2016. Aspergillus march: from ABPA to aspergilloma to subacute invasive aspergillosis. Allergy Asthma Clin. Immunol. 12:64.

Dohm, J.C., Lottaz, C., Borodina, T., Himmelbauer, H., 2008. Substantial biases in ultra- short read data sets from high-throughput DNA sequencing. Nucleic Acids Res. 36(16):e105

Domingo, J.L., Nadal, M., 2009. Domestic waste composting facilities: a review of human health risks. Environ. Int. 35(2):382-389.

Dong, L., Qi, J., Shao, C., Zhong, X., Gao, D., Cao, W., et al., 2016. Concentration and size distribution of total airborne microbes in hazy and foggy weather. Sci. Total Environ. 541:1011-1018.

Douglas, P., Robertson, S., Gay, R., Hansell, A.L., Gant, T.W., 2018. A systematic review of the public health risks of bioaerosols from intensive farming. Int. J. Hyg. Environ. Health. 221(2):134-173.

Douwes, J., P. Thorne, N. Pearce, D. Heederik., 2003. Bioaerosol health effects and exposure assessment: Progress and prospects. Ann. Occup. Hyg. 47(3):187-200.

283

Douwes, J., Thorne, P., Pearce, N., Heederik, D. 2002. Bioaerosol Health Effects and Exposure Assessment: Progress and Prospects. Ann. occup. Hyg. 47(3):187–200.

Dubuis, M.E., M’Bareche, H., Veillette, M., Bakhiyi, B., Zayed, J., Lavoie, J., et al. Bioaerosols concentrations in working areas in biomethanization facilities. J. Air Waste Manag. Assoc. 67(11):1258-1271.

Duchaine, C., Mériaux, A., Brochu, G., Cormier, Y., 1999. Airborne microflora in Quebec dairy farms: Lack of effect of bacterial hay preservatives. Am. Ind. Hyg. Assoc. J. 60(1):89- 95.

Eames, I., Tang, J.W., Li, Y. and Wilson, P., 2009. Airborne transmission of disease in hospitals. J. Royal Soc. Interface. 6(suppl 6).

Edgar, R.C., 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 26(19):2460-2461.

Edgar, R.C., 2016. UNOISE2: improved error-correction for Illumina 16 S and ITS amplicon sequencing. bioRxiv 081257.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 27:2194–2200.

Eduard, W., 2008. A health based criteria document on fungal spore exposure in the working population. Is it relevant for the general population? Indoor Air. 18(3):257-2588.

Eduard, W., 2009. Fungal spores: a critical review of the toxicological and epidemiological evidence as a basis for occupational exposure limit setting. Crit. Rev. Toxicol. 39(10):799- 864.

Eduard, W., Heederik, D., Duchaine, C., Green, B.J., 2012. Bioaerosol exposure assessment in the workplace: the past, present and recent advances. J. Environ. Monit. 14(2):334–339.

Ekhaise, F.O., Ighosewe, O.U., 2008. Hospital indoor airborne microflora in private and government owned hospitals in Benin city, Nigeria. World J. Med. Sci. 3:19–23.

El-Mashad, H.M., Zeeman, G., Van Loon, W.K.P., Bot, G.P.A, Lettinga, G., 2003. Effect of temperature and temperature fluctuation on thermophilic anaerobic digestion of cattle manure. 95(2):191-201.

Engels, S., 2012. Airborne exposure to mycotoxins. Master thesis, Faculty of Medicine Theses. Utrecht University Repository.

Epstein, E., Wu, N., Youngberg, C., Croteau, G., 2001. Dust and bioaerosols at a biosolids composting facility. Compost Sci. Util. 9(3): 250-255,

284

Erceg-Hurn, D.M., Mirosevich, V.M., 2008. Modern robust statistical methods: An easy way to maximize the accuracy and power of your research. Am. Psychol. 63(7):591-601.

Eren, A.M., Morrison, H.G., Lescault, P.J., Reveillaud, J., Vineis, J.H., Sogin, M.L., 2015. Minimum entropy decomposition: unsupervised oligotyping for sensitive partitioning of high-throughput marker gene sequences. ISME J. 9(4): 968–979.

Fabian, M.P., Miller, S.L., Reponen, T., Hernanadez, M., 2005. Ambient bioaerosol indices for indoor air quality assessments in flood reclamation. J. Aerosol Sci. 36 (5-6):763–783.

Faith, D. P., Baker, A. M. 2006. Phylogenetic diversity (PD) and biodiversity conservation: some bioinformatics challenges. Evol. Bioinform. Online. 2:121–128.

Feldmesser, M., Kress, Y., Novikoff, P., Casadevall, A., 2000. Cryptococcus neoformans is a facultative intracellular pathogen in murine pulmonary infection. Inf. Immun. 68(7):4225- 37.

Festenstein, G., Lacey, J., Skinner, F.A., 1968. Self-heating of hay and grain in Dewar flask and the development of farmer’s lung antigens. J. Gen. Microbiol. 41(3):389–407.

Fitzpatrick, E., Yoon, T.W., Williams, R., Yi, A.K., 2017., Effect of genetics on the development of Hypersensitivity Pneumonitis (HP) using the BXD recombinant inbred mouse reference population. J. Immunol. 198:53-10.

Floudas, D., Binder, M., Riley, R., Barry, K., Blanchette, R.A., Henrissat, B., 2012. The Paleozoic Origin of Enzymatic Lignin Decomposition Reconstructed from 31 Fungal Genomes. Science. 336(6089):1715-1719.

Fogelmark, B., Sjostrand M. Rylander, R., 1994. Pulmonary inflammation induced by repeated inhalations of (1,3)-D-glucan and endotoxin. Int. J. Exp. Path. 75(2):85-90.

Freira, L., Sant’Ana, A.S., 2018. Modified mycotoxins: an updated review on their formation, detection, occurrence, and toxic effects. Food Chem. Toxicol. 111:189-205.

Fröhlich-Nowoisky, J., Kampf, C.J., Weber, B., Huffman, J.A., Pöhlker, C., Andreae, M.O., et al., 2016. Bioaerosols in the earth system: climate, health, and ecosystem interactions. Atmos. Res. 182:346-376.

285

Fuchs, J.G., Bieri, B., 2000. New biotests to measure the biological qualities of composts. Agrarforschung Schweiz 7(7):314-319.

Fuchs, U., Czymmek, K.J., Sweigard, J.A., 2004. Five hydrophobins genes in Fusarium verticillioides include two required for microconidial chain formation. Fungal Genet Biol. 41(9):852-864.

Fung, F., Clark, R.F., 2004. Health effects of mycotoxins: a toxicological overview. J. Toxicol. Clin. Toxicol. 42(2):217–234.

Fung, F., Hughson, W.G., 2003. Health Effects of Indoor Fungal Bioaerosol Exposure. Appl. Occup. Environ. Hyg. 18(7):535-544.

Gandolfi, I., Bertolini, V., Ambrosini, R., Besttetti, G., Franzetti, A., 2013. Unravelling the bacterial diversity in the atmosphere. Appl. Microbiol. Biotechnol. 97(11):4727-4736.

Gareis, M., Gottschalk, C., 2014. Stachybotrys spp. and the guttation phenomenon. Mycotoxin Res. 30(3): 151-159.

Gebbink, M.F.B.G., Claessen, D., Bouma, B., Dijkhuizen, L., Wösten, H.A.B., 2005. Amyloids – a functional coat for microorganisms. Nat. Rev. Microbiol. 3(4):333-334.

Geiser, D.M., Gueidan, C., Miadlikowska, J., Lutzoni, F., Kauff, F., Hofstetter, V., et al., 2006. Eurotiomycetes: Eurotiomycetidae and Chaetothyriomycetidae. Mycologia. 98(6):1053-1064.

Gevers, D., Kugathasan, S., Denson, L.A., Vasquez-Baeza, Y., Van Treuren, W., Ren, B. al., 2014. The treatment-naïve microbiome in new-onset Crohn’s disease. Cell Host Microbe. 15(3):382-392.

Ghiasian, S. D., Maghood, A.H., Aghamirian, M.R. 2017. Aeromycological analysis of allergenic airborne fungi in Qazvin, Iran. Curr. Med. Mycol. 2(3):5-9.

Ghosh, R.R., 1985. Physiological studies of the Gymnoascaceae with an overview on their ecology. Indian Acad. Sci. 94:197.

286

Gilbert, J.A., Blaser, M.J., Caporaso, J.G., Jansson, J.K., Lynch, S.V., Knight, R., 2018. Current understanding of the human microbiome. Nat. Med. 24(4):392-400.

Gilbert, Y., Duchaine, C., 2009. Bioaerosols in industrial environments: a review. Can. J. Civ. Eng. 36(12):1873-1886.

Glass, D., Amedee, R.G., 2011. Allergic fungal rhinosinusitis: a review. Ochsner J. 11(3):271–275. Gomez-Alvarez, V., Teal, T.K., Schmidt, T.M., 2009. Systematic artifacts in metagenomes from complex microbial communities. ISME J.23(11):1314-7.

Goodrich, J.K., Di Rienzi, S.D., Poole, A.C., Koren, O., Walters, W.A., Caporaso, J.G., et al., Conducting a Microbiome Study. Cell. 158(2):250-262.

Gower, J.C., Legendre, P., 1986. Metric and Euclidean properties of dissimilarity coefficients. J. Class. 3:5.

Green, B.J., Lemons, A.R., Park, Y., Cox-Ganger, M., Park, .JH., 2017. Assessment of fungal diversity in a water-damaged office building. J. Occup. Environ. Hyg. 14(4):285-293.

Greenberger, P.A. 2002. Allergic bronchopulmonary aspergillosis. Journal of Allergy and Clinical Immunology. 110(5): 685-692.

Gregory, P.H., 1957. Electrostatic charges on spores of fungi in air. Nature. 180:330.

Gregory, P.H., Lacey, M.E., Festenstein, G.N., Skinner, F.A., 1963. Microbial and biochemical changes during the molding of hay. J. Gen. Microbiol. 33:147–174.

Guarro, J., Gené J., Stchigel, A.M., 1999. Developments in fungal taxonomy. Clin. Microbiol Rev. 12(3):454-500.

Guo, H., Jeong, K., Lim, J., Jo, J., Kim, Y.M., Park, J., et al. Prediction of effluent concentration in a wastewater treatment plant using machine learning models. J. Environ. Sci. 32(1):90–101.

Haas, B.J., Gevers, D., Earl, A.M., Feldgarden, M., Ward, D.V., Giannoukos, G., et al., 2011. Chimeric 16S rRNA sequence formation and detection in Sanger and 454-pyrosequenced PCR amplicons. Genome Res. 21:494–504

287

Hadziavdic, K., Lekang, K., Lanzen, A., Jonassen, I., Thompson, E. M., Troedsson, C., 2014. Characterization of the 18S rRNA Gene for Designing Universal Eukaryote Specific Primers. PLoS ONE. 9(2): e87624.

Haleem Khan, A.A., Karuppayil, M.S., 2012. Fungal pollution of indoor environments and its management. Saudi J. Biol. Sci. 19(4):405-426.

Hamad, I., Ranque, S., Azhar, E.I., Yasir, M., Jiman-Fatani, A.A., Tissot-Dupont, H., et al. 2017. Culturomics and amplicon-based metagenomic approaches for the study of fungal population in human gut microbiota. Sci. Rep. 16788.

Hansgate, A.M., Schloss, P.D., Hay, A.G., Walker, L. P., 2005. Molecular characterization of fungal community dynamics in the initial stages of composting. FEMS Microbiol. Ecol. 51(2):209–214.

Hardin, B.D., Kelman B.J., Saxon, A., 2003. Adverse human health effects associated with molds in the indoor environment. J. Occup. Environ. Med. 45:470–478.

Haugland, R.A., Varma, M., Wymer, L.J., Vesper, S.J., 2004. Quantitative PCR analyses of selected Aspergillus, Penicillium and Paecilomyces species. Syst. Appl. Microbiol. 27(2):198-210.

Hawksworth, D. L., Lücking, R., 2017. Fungal diversity revisited: 2.2 to 3.8 million species. Microbiol. Spectr. 5(4).

Hebert, P.D., Ratnasingham, S., deWaard, J.R., 2003. Barcoding animal life: cytochrome c oxidase subunit 1 divergences among closely related species. Proc Biol. Sci. 270(Suppl 1:S96– S99.

Hector, R.F., Laniado-Laborin, R., 2005. Coccidioidomycosis—A Fungal Disease of the Americas. PLoS Med. 2(1): e2.

Heederik, D., von Mutius, E., .2012. Does diversity of environmental microbial exposure matter for the occurrence of allergy and asthma. J. Allergy Clin. Immunol. 130(1):44–50.

288

Heikkinenm, M.S.A., Hjelmroos-koski, M., Häggblom, M.M., Macher, J.M., 2005. Bioaerosols. Chapter 13. In book: Aerosols Handbook: Measurement, Dosimetry, and health Effects. CRC Press, Boca Raton, FL, USA. Heinrichs, G., de Hoog, G.S., Haase, G., 2012. Barcode identifiers as a practical tool for reliable species assignment of medically important black yeast species. J. Clin. Microbiol. 50(9)3023–3030.

Hewitt, S.K., Foster, D.S., Dyer, P.S., Avery, S.V., 2016. Phenotypic heterogeneity in fungi: importance and methodology. Fungal boil. Rev. 30:176-184.

Hibbett, D., Abarenkov, K., Koljalg, U., Opik, M., Chai, B., Cole, J.R., et al., 2016. Sequence-based classification and identification of fungi. Mycologia. 108(6):1049-1068.

Hibbett, D.S., Binder, M., Bischoff, J.F., Blackwell, M., Cannon, P.F., Eriksson, O.E et al., 2007. A higher level phylogenetic classification of the Fungi. Mycological Research. 111(5): 509–547.

Hibbett, D.S., Matheny, P.B., 2009. The relative ages of ectomycorrhizal mushrooms and their plant hosts estimated using Bayesian relaxed molecular clock analyses. BMC Biol. 7:13.

Hinds, W. 1999. Aerosol Technology: Properties, Behavior, and measurement of airborne particles (2nd edition ed.).John Wiley and Sons, Inc. New York, NY, USA.

Hintikka, E.L., Nikulin, M., 2010. Airborne mycotoxins in agricultural and indoor environments. Indoor Air. 8(64):66-70.

Hogeweg, P., 2011. The roots of bioinformatics in theoretical biology. PLoS Comput. Biol. 7(3):e1002021.

Hollander, M., Wolf, D.A., Chicken, E. 2015. Nonparametric statistical methods (2nd ed.). John Wiley and Sons Inc. New York, NY, USA.

Hoorman, J.J., 2011. The role of soil fungus. Fact sheet. Agricultural and Natural Resources. Ohio State University. SAG-14-11.

Hoornweg, D., Bhada-Tata, P., 2012. What a Waste: A Global Review of Solid Waste Management. Urban development series; knowledge papers no. 15. World Bank, Washington, DC, USA.

Hosny, M., Abou abdallah, R., Khalil, J.B., Fontanini, A., Baptiste, E., Armstrong, N., et al., 2019. Clostridium pacaense: a new species within the genus Clostridium. New Microbes New Infect. 28:6-10.

Hospodsky, D., Qian, J., Nazaroff, W. W., Yamamoto, N., Bibby, K., Rismani-Yazdi, H. et al., 2012. Human occupancy as a source of indoor airborne bacteria. PLoS ONE. 7(4): e34867.

289

Hospodsky, D., Yamamoto, N., Nazaroff, W.W., Miller, D., Gorthala, S., Peccia, J., 2014. Characterizing airborne fungal and bacterial concentrations and emission rates in six occupied children’s classrooms. Indoor Air. 25(6):641-652. Houbraken, J., Samson, R.A., 2011. Phylogeny of Penicillium and the segregation of Trichocomaceae into three families. Stud. Mycol. 70(1):1-51.

Hug, L.A., Baker, B.J., Anantharaman, K., Brown , C.T., Probst, A.J., Castelle, C.J., et al., 2016. A new view of the tree of life. Nat. Microbiol. 1:16048.

Hugenholtz, P., Pace, N.R., 1996. Identifying microbial diversity in the natural environment: a molecular phylogenetic approach. Trends Biotechnol. 14(6):190-197.

Human Microbiome Project Consortium, 2012. A framework for human microbiome research. Nature. 486:215-221.

Huse, S.M., Dethlefsen L., Huber, J.A., Welch D.M., Relman D.A., Sogin L.M., 2008. Exploring microbial diversity and taxonomy using SSU rRNA Hypervariable Tag Sequencing. PLoS Genet. 4(11): e1000255.

Huson, D.H., Auch, A.F., Qi, J., Schuster, S.C., 2007. MEGAN analysis of metagenomic data. Genome Res. 17(3):377-86.

Huttenhower, C., Knight, R., Brown, C.T., Caporaso, J.G., Clemente, J.C., Gevers, D. et al., 2014. Advancing the microbiome research community. Cell. 159(2):277-230.

IARC. 1972. Some inorganic substances, chlorinated hydrocarbons, aromatic amines, N- nitroso compounds and natural products. IARC Monogr. Eval. Carcinog. Risk Chem. Man. 1:1–184.

Ishii, K. , Fukui, M., Takii, S., 2000. Microbial succession during a composting process as evaluated by denaturing gradient gel electrophoresis analysis. J. Appl. Microbiol. 89(5):768- 77.

Ishii, N., Ishida, S., Kagami, M., 2015. PCR primers for assessing community structure of aquatic fungi including Chytridiomycota and Cryptomycota. Fungal Ecology 13:33–43.

Iversen, M., Kirychuk, S., Drost, H., Jacobson, L., 2000. Human health effects of dust exposure in animal confinement buildings. J. Agric. Saf. 6(4):283-288.

Jabra-Rizk, M.A., Kong, E.F., Tsui, C., Nguyen, M.H., Clancy, C.J., Fidel, P.L., et al., 2016. Candida albicans pathogenesis: fitting within the host-microbe damage response framework. Inf. Immun. 84(10): 2724-39.

290

Jamison, D.C., 2003. Introduction, in Perl Programming for Biologists. Hoboken, NJ: John Wiley & Sons, Inc. New York, NY, USA.

Jones, A.M., Harrison, R.M., 2004. The effects of meteorological factors on atmospheric bioaerosol concentrations — a review. Sci. Total Environ. 326 (1–3):151–180.

Jones, O.A.H., Lear G., Welji, A.M., Collins G., Quince C., 2016. Community metabolomics in environmental microbiology. In Book: Microbial Metabolomics. Ed.: Beale D., Kouremenos K., Palombo E. Springer. Switzerland.

Jost, L. DeVries, P., Walla, T., Greeney, H., Chao, A., Ricotta, C., 2010. Partitioning diversity for conservation analyses. Divers. Distrib. 16(1): 65-76.

Jost, L., 2006. Entropy and diversity. Oikos. 113(2):363-375.

Jost, L., 2007. Partitioning diversity into independent alpha and beta components. Ecology. 88(10): 2427-2439.

Kakde, U.B. 2012., Fungal bioaerosols: global diversity, distribution and its impact on human beings and agricultural crops. Bionano. Frontier. 5(2):323-329.

Kambouris, M.E., Pavlidis, C., Skoufas, E., Arabatzis, M., Kantzanou, M., Velegraki, A., et al., 2018. Culturomics: a new kid on the block of omics to enable personalized medicine. OMICS. 22(2).

Karakurt, I., Aydin, G., Aydiner, K., 2011. Mine ventilation air as a sustainable energy source. Renew. Sust. Energ. Rev. 15(2):1042-1049.

Kauffman, C.A., 2007. Histoplasmosis: a clinical and laboratory update. Clin. Microbiol. Rev. 20(1):115-32.

Kazda M., Langer, S., Bengelsdorf, F.R., 2014. Fungi open new possibilities for anaerobic fermentation of organic residues. Energy Substain. Soc. 4:6.

291

Kembel, S.W., Wu, M., Eisen, J.A., Green, J.L., 2012. Incorporating 16S gene copy number information improves estimates of microbial diversity and abundance. PLoS Comput. Biol. 8910):e1002743.

Kim, K.Y., Kim, Y.S., Kim, D., 2010. Distribution characteristics of airborne bacteria and fungi in the general hospitals of Korea. Indust. Health. 48:236–243.

Knight, R., Jansson, J., Field, D., Fierer, N., Desai, N., Fuhrman, J.A., al., 2012. Unlocking the potential of metagenomics through replicated experimental design. Nat. Biotechnol. 30(6):513–520.

Kõljalg U, Nilsson R H, Abarenkov K, Tedersoo L, Taylor AFS, Bahram M, et al. 2013.Towards a unified paradigm for sequence-based identification of fungi. Mol. Ecol. 22(21): 5271–5277.

Koljalg, U., Nilsson, R.H., Abarenkov, K., Tedersoo, L., Taylor, A.F., Bahram, M., et al., 2013. Towards a unified paradigm for sequence-based identification of fungi. Mol. Ecol. 22(21):5271-5277.

Kopylova, E., Noé, L., Touzet, H., 2012. SortMeRNA: Fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics 28:3211–3217.

Korpi, A., Kasanen, J.P., Alarie, Y., Kosma, V.M., Pasanen, A.L., 1999. Sensory irritating potency of some Microbial Volatile Organic Compounds (MVOCs) and a mixture of five MVOCs. Arch. Environ Health. 54(5):347-352. Kosmidis, C., Denning, D.W., 2015. The clinical spectrum of pulmonary aspergillosis. Thorax. 70:270-277.

292

Kotimaa, M.H., Oksanen, L., Koskela, P., 1991. Feeding and bedding materials as sources of microbial exposure on dairy farms. Scand. J. Work Environ. Health. 17(2):117-122.

Kulkarni, P., Baron, P.A., Willeke, K., 2011. Chapter 5: An approach to performing aerosol measurements. Book: Aerosol Measurement: Principles, Techniques, and Applications. Third Edition. Ed. Pramod Kulkarni, Paul A. Baron, Klaus Willeke. Wiley. New Yor, NY, USA.

Kumari, P., Woo, C., Yamamoto, N., Choi, H.L., 2016. Variations in abundance, diversity and community composition of airborne fungi in swine houses across seasons. Sci. Rep. 6:37929.

Kwasna, H., Kosiak, B., 2003. Lewia avenicola sp. nov. and its Alternaria anamorph from oat grain, with a key to the species of Lewia. Mycol. Res. 107(3):371-376.

Kwon-Chung, K.J., Fraser, J.A., Doering, T.L., Wang, Z., Janbon, G., Idnurm, H. et al., 2014. Cryptococcus neoformans and Cryptococcus gattii, the etiologic agents of cryptococcosis. Cold Spring Harb. Perspect. Med. 4(7):a019760.

Kwon-Chung, K.J., Sugui, J.A., 2013. Aspergillus fumigatus—What Makes the Species a Ubiquitous Human Fungal Pathogen? PLoS Pathog. 9(12):e1003743.

Lagier, J.C., Armouggom, F., Million, M., Hugon, P., Pagnier, I., Robert, C. et al., 2012. Microbial culturomics : paradigm shift in the human gut microbiome study. Clin. Microbiol. Infect. 18(12):1185-1193.

Lagier, J.C., Hugon, P., Khelaifia, S., Fournier, P.E., La Scola, B., Raoult, D., 2015. The rebirth of culture in microbiology through the example of culturomics to study human gut microbiota. Clin. Microbiol Rev. 28(1):237-264.

Lagier, J.C., Khelaifia, S., Alou, M.T., Ndongo, S., Dione, N., Hugon, P., et al., 2016. Culture of previously uncultured members of the human gut microbiota by culturomics. Nat. Microbiol. 16203.

Lane, D.J., Pace, B., Olsen, G.J., Stahl, D.A., Sogin, M.L., Pace NR. Rapid determination of 16S ribosomal RNA sequences for phylogenetic analyses. Proc. Natl. Acad. Sci. U S A. 82(20):6955–6959.

293

Lanier, C., Richard, E., Heutte, N., Picquet, R., Bouchart, V., Caron., D., 2010. Airborne moulds and mycotoxins associated with handling of corn silage and oilseed cakes in agricultural environment. Atmospheric Environ. 44(16):1980-1986.

Latgé, J.P. 1999. Aspergillus fumigatus and Aspergillosis. Clin. Microbio. Rev. 12(2):310- 350.

Lauber, C.L., Metcalf, J.L., Keerpers, K., Ackermann, G., Carter, D.O., Knight, R., 2014. Vertebrate decomposition is accelerated by soil microbes. Appl. Environ. Microbiol 80(16):4920-4929.

Lazarus, K.L., James, T.Y., 2014. Surveying the biodiversity of the Cryptomycota using a targeted PCR approach. Fungal Ecol. 14:62-70.

Le Goff, O., Bru-Adan, V., Bacheley, H., Godon, J.J., Wéry, N. 2010. The microbial signature of aerosols produced during the thermophilic phase of composting. J Appl Microbiol. 108(1):325-40.

Le Goff, O., Godon, J.J., Steyer, J.P., Wéry, N., 2011. New specific indicators for qPCR monitoring of airborne microorganisms emitted by composting plants. Atmos. Environ 45(30):5342-5350.

Lee, J.E., Ahn, J.C., Han, D.H., Kim, D.Y., Kim, J.W., Cho, S.H., et al., 2013. Variability of offending allergens of allergic rhinitis according to age: optimization of skin prick test allergens. Allergy Asthma Immunol Res. 6(1): 47-54.

Lee, S. Y., Holder, G.D. 2001. Methane hydrates potential as a future energy source. Fuel Process. Technol. 71(1-3):181–186.

Legendre, P., Anderson M.J. 1999. Distance-based redundancy analysis: Testing multispecies responses in multifactorial ecological experiments. Ecol. Monograr. 69:1–24.

294

LeRoy Poff, N., Olden, J.D., Merritt, D.M., Pepin, D.M., 2007. Homogenization of regional river dynamics by dams and global biodiversity implications Proc. Nat. Acad. Sci. 104(14):5732-5737.

Létourneau, V., Nehmé, B., Mériaux, A., Daniel, M., Cormier, Y., Duchaine, C. 2010. Human pathogens and tetracycline-resistant bacteria in bioaerosols of swine confinement buildings and in nasal flora of hog producers. Int. J. Hyg. Environ. Health. 213(6):444-449.

Ley, R.E., Hamady, M., Lozupone, C., Turnbaugh, P.J., Ramey, R.R., Bircher, J.S., Schlegel, M.L., Tucker, T.A., Schrenzel, M.D., Knight, R., Gordon, J.I., 2008. Evolution of mammals and their gut microbes. Science. 320:1647–1651.

Li, D.M., de Hoog, G.S., Saunte, D.M.L., van den Ende, A.H.G.G., Chen, X.R., 2008. Coniosporium epidermidis sp. nov., a new species from human skin. Stud. Mycol. 61:131– 136.

Lighthart, B., Shaffer, B.T., 1997. Increased airborne bacterial survival as a function of particle content and size. Aerosol Sci. Technol. 27:439-446.

Linder, M.B., Szilvay, G.R., Nakari-Setälä, T., Penttilä, M.E., 2005. Hydrophobins: the protein-amphiphiles of filamentous fungi. FEMS Microbiol. Rev. 29(5): 877–896.

Liu, J., Yu, Y., Cai, Z., Bartlam, M., Wang, Y., 2015. Comparison of ITS and 18S rDNA for estimating fungal diversity using PCR-DGGE. World J. Microbiol. Biotechnol. 31(9):1387- 1395.

295

Liu, T., Liu S., Zheng, M., Chen, Q., Ni, J., 2016. Performance assessment of full-scale wastewater treatment plants based on seasonal variability of microbial communities via high- throughput sequencing. PLOS one. 11(4):e0152998.

Liu, Y.S., Wang, N.C., Ye, R.H., Kao, W.Y. 2013. Disseminated Fusarium infection in a patient with acute lymphoblastic leukemia: A case report and review of the literature. Oncology letters. 7(2):334-336.

Liu, Z., Cheng, K., Li, H., Cao, G., Wu, D., Shi, Y. 2018. Exploring the potential relationship between indoor air quality and the concentration of airborne culturable fungi: a combined experimental and neural network modeling study. Environ Sci Pollut R. 25(4):3510-3517.

Love, M., Huber, W., Anders, S., 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15(12):550.

Lozupone, C. A., Knight, R., 2007. Global patterns in bacterial diversity. Proc. Nat. Acad. Sci. U. S. A. 104(27):11436–11440.

Lozupone, C., Hamady, M., Knight, R., 2006. UniFrac- an online tool for comparing microbial community diversity in a phylogenetic context. BMC Bioinformatics. 7:371.

Lozupone, C.A., Knight, R., 2008. Species divergence and the measurement of microbial diversity. FEMS Microbiol. Rev. 32(4):557-578.

Lu, B., Charlton, M., Fotheringhama, A.A., 2011. Geographically Weighted Regression Using a Non-Euclidean Distance Metric with a Study on London House Price Data. Procedia Environ. Sci. 7:92-97.

Lücking, R., Dal-Forno, M., Sikaroodi, M., Gillevet, P.M., Bungartz, F., Moncada, B., Lawrey, J.D. 2014. A single macrolichen constitutes hundreds of unrecognized species. Proc. Natl. Acad. Sci. 111:11091-11096.

Macias-Corral, M., Samani, Z., Hanson, A., Smith, G., Funk, P., Yu, H., et al.,2008. Anaerobic digestion of municipal solid waste and agricultural waste and the effect of co-

296

digestion with dairy cow manure. Bioresour. Technol. 99(17):8288-8293.

Madsen, A.M., Alwan, T., Orberg, A., Uhrbrand, K., Jorgensen, M.B., 2016. Waste workers’ exposure to airborne fungal and bacterial species in the truck cab during waste collection. Ann. Occup. Hyg. 60(6):651-668.

Madsen, D., Klock, L.E., Wenzel, F.J., Robbins, J.L., Schmidt, C.D., 1976. The prevalence of farmer’s lung in an agricultural population. Am. Rev. Respir. Dis. 113:171–174.

Madureira, J., Aguiar, L., Pereira, C., Mendes, A., Querido, M.M., Neves, P., Teixeira, J.P., 2018. Indoor exposure to bioaerosol particles: levels and implications for inhalation dose rates in school children. Air Qual. Atmos. Health. 11(8): 955-964.

Magurran & McGill. 2011. (Editors) Biological diversity. Oxford University Press. London, UK.

Mahajan, J., Vakhariya, J.A., 2016. Waste Management: A Reverse Supply Chain Perspective. Vikalpa. 41(3):197–208.

Mardis, E.R., 2008. Next-Generation DNA sequencing methods. Annu. Rev. Genomics Hum. Genet. 9:387-402.

Marques do Nascimento, J.P., Lòpez, A.M.Q., Andrade de Araùjo, M., Anhezini de Araujo, L., Alves da Silva Filho, E., 2019. Airborne fungi in indoor hospital environnements. Int J Curr. Microbiol. App. Sci. 8(1):2749-2772.

Martínez, J.L. 2008. Antibiotics and Antibiotic Resistance Genes in Natural Environments. Sci. 32(5887):365-367.

Mata-Alvarez, J. 2003. Biomethanization of The Organic Fraction of Municipal Solid Waste. IWA publishing. London, UK.

Mbareche, H., Brisebois, E., Veillette, M., Duchaine, C., 2017a. Bioaerosol sampling and detection methods based on molecular approaches: No pain no gain. Sci. Total Environ. 599- 600:2095-2104.

297

Mbareche, H., Veillette, M., Bilodeau, G.J. and Duchaine, C. 2019b. Fungal spore recovery from air samples: tale of loss and gain. Appl. Environ. Microbiol. AEM.02941-18.

Mbareche, H., Veillette, M., Bilodeau, G.J., Duchaine, C., 2018. Bioaerosol sampler choice should consider efficiency and ability of samplers to cover microbial diversity. Appl. Environ. Microbiol. 10.1128/AEM.01589-18.

Mbareche, H., Veillette, M., Bilodeau, G.J., Duchaine, C., 2019a. Fungal aerosols at dairy farms using molecular and culture techniques. Sci. Total Environ. 653:253-263.

Mbareche, H., Veillette, M., Dubuis, M.E., Bakhiyi, B., Marchand, G., Zayed, J et al., 2018a. Fungal bioaerosols in biomethanization facilities. J. Air Waste Manag. Assoc. 68(11):1198- 1210.

McArdle, B.H., Anderson, M.J., 2001. Fitting multivariate models to community data: a comment on distance-based redundancy analysis. Ecology. 82(1):290-297.

McDonald, D., Price, M.N., Goodrich, J., Nawrocki, E.P., DeSantis, T.Z., Probst, A., et al. 2012. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J. 6:610–618.

McDonald, J.H., 2014. Handbook of Biological Statistics (3rd ed.). Sparky House Publishing. Baltimore, Maryland, USA.

McMurdie, P.J., Holmes, S., 2014. Waste not, want not: why rarefying microbiome data is inadmissible. PLoS Comput. Biol. 10(4): e1003531.

298

McMurdie, P.S., Holmes, S. 2013. phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS ONE. 8(4):e61217. Méheust, D., Gangneux, J.P., Reponen, T., Wymer, L., Vesper, S., LeCann, P., 2012. Correlation between Environmental Moldiness Index (ERMI) values in French dwellings and other measures of fungal contamination. Sci. Total Environ. 438:319-24.

Messer, S.A., 2018. Assessment of regional fungal concentrations and diversity and their possible association with self-reported health effects among a national sample of office building occupants in the United States. PhD (Doctor of Philosophy) thesis, University of Iowa.

Meyer, V., Andersen, M.R., Brakhage, A.A., Braus, G.H., Caddick, M.X., Caims T.C., et al., 2016. Current challenges of research on filamentous fungi in relation to human welfare and a sustainable bio-economy: a white paper. Fungal Biol. Biotechnol. 3:6.

Milićević, D.R., Skrinjar, M., Baltić, T., 2010. Real and perceived risks for mycotoxin contamination in foods and feeds: challenges for food safety control. Toxins. 2(4):572-92. Milner, P.D., 2009. Bioaerosols associated with animal production. Bioresour. Technol. 100(22):5379-5385.

Mistry, J., Finn, R.D., Eddy, S.R., Bateman, A., Punta, M., 2013. Challenges in Homology Search: HMMER3 and Convergent Evolution of Coiled-Coil Regions. Nucleic Acids Res. 41: e121.

Moletta, M., Delgenes, J., Godon, J., 2007. Differences in the aerosolization behavior of microorganisms as revealed through their transport by biogas. Sci. Total Environ. 379(1): 75–88.

Monard, C., Gantner, S., Stenild, J., 2013. Utilizing ITS1 and ITS2 to study environmental fungal diversity using pyrosequencing. FEMS Microbiol. Ecol. 84(1):165–175.

Morell, F., Cruz, M.J., Gòmez, F.P., Rodriquez-Jerez, F., Xaubet, A., Munoz, X., 2011. Chacinero’s lung – hypersensitivity pneumonitis due to dry sausage dust. Scand. J. Work Environ. Health. 37(4):349-56.

Moreno, C.E., Rodrìguez, P., 2010. A consistent terminology for quantifying species diversity? Oecologia, 163:279-282.

Morey, M., Fernàndez-Marmiesse, A., Catiñeiras, D., Fraga, J.M., Couce, M.L., Cocho, J.A, 2013. A glimpse into past, present, and future DNA sequencing. Mol. Genet. Metab. Rep. 110:3-24.

299

Morgenstern, I., Klopman, S., Hibbett, D.S., 2008. Molecular evolution and diversity of lignin degrading heme peroxidases in the Agaricomycetes. J. Mol. Evol. 66(3): 243.

Mubareka, S., Groulx, N., Savory, E., Cutts, T., Theriault, S., Scott, J.A., et al., 2019. Bioaerosols and transmission, a diverse and growing community of practice. Front. Public Health. 7(23).

Murphy, D.J., 2001. The biogenesis and functions of lipid bodies in animals, plants and microorganisms. Prog. Lipid Res. 40(5):325–438.

Murto, M., Björnsson, L., Mattiasson, B. 2004. Impact of food industrial waste on anaerobic co-digestion of sewage sludge and pig manure. Journal of Environmental Management. 70(2):101-107.

Nature Microbiology Editorial. 2017. Stop neglecting fungi. Nature Microbiology. 17120.

Nazaroff, W.W., 2014. Indoor bioaerosol dynamics. Indoor Air. 26(1):61-78.

Nevalainen, A., Pastuszka, J., Liebhaber, F., Willeke, K., 1992. Performance of bioaerosol samplers: collection characteristics and sampler design considerations. Atmos. Environ. 26(4):531-540.

Nilsson, R.H., Kristiansson, E., Ryberg, M., Hallenberg, N., Larsson, K.H., 2008. Intraspecific ITS variability in the kingdom fungi as Expressed in the International Sequence Databases and its implications for molecular species identification. Evol. Bioinform. Online. 4:193–201.

300

Nilsson, R.H., Larsson, K.H., Taylor, A.F.S., Bengtsson-Palme, J., Jeppesen, T.S., Schigel, D., et al., 2019. The UNITE database for molecular identification of fungi: handling dark taxa and parallel taxonomic classifications. Nucleic Acids Red. 47(D1):D259-D264.

Nolan, S.J., Fu, M.S., Coppens, I., Casadevall, A., 2017. Lipids Affect the Cryptococcus neoformans-Macrophage Interaction and Promote Nonlytic Exocytosis. Infect. Immun. 85(12):e00564-17.

Nonnemann, M.W., Brextine, B., Dowd, S.E., Gilmore, K., Levin, J.L., 2010. Culture- Independent Characterization of Bacteria and Fungi in a Poultry Bioaerosol Using Pyrosequencing: A New Approach. J. Occup. Environ. Hyg. 7(12): 693-699.

Novais, R.C., Thorstenson, Y.R., 2011. The evolution of pyrosequencing® for microbiology: from genes to genomes. J. Microbiol. Methods. 86(1):1-7.

Nucci, M., Anaissie E., 2007. Fusarium Infections in Immunocomprised Patients. Clin. Microbiol. Rev. 20(4):695-704.

Nygard, K., Werner-Johansen, O., Ronsen, S., Caugant, D.A., Simonsen, O., Kanestrom, A., et al., 2008. An outbreak of legionnaires disease caused by long-distance spread from an industrial air scrubber in Sarpsborg, Norway. Clin. Infect. Dis. 46: 61–69. O’brien, K.M., Chimenti, M.S., Farnell, M., Tabler, T., Biar, T., Bray, J.L., Nonnenmann, M.W., 2016. High throughput genomic sequencing of bioaerosols in broiler chicken production facilities. Microb. Biotechnol. 9(6):789-791.

O’Connor, J.D., Daly, S.M., Sodeau, J.R., 2015. On-line monitoring or airborne bioaerosols released from a composting/green waste site. Waste manag. 42:23-30.

O’Connor, J.D., Daly, S.M., Sodeau, J.R., 2015. On-line monitoring or airborne bioaerosols released from a composting/green waste site. Waste manag. 42:23-30

Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Stevens, M.H.H., 2007. The vegan package. Community ecology package. 10:631-637 http://r-forge.r-project.org/projects/vegan/.

Op De Beeck, M., Lievens, B., Busschaert, P., Declerck, S., Vangronsveld, J., Colpaert, J.V.,

301

2014. Comparison and validation of some ITS primer pairs useful for fungal metabarcoding studies. PLoS ONE 9(6):e97629.

Orfila, J., 1996. Definition of intracellular pathogens. Clin. Microbiol. Infect. 1(1):S1-S2.

Pace, N.R., 1997. A molecular view of microbial diversity and the biosphere. Sci. 276:734– 740.

Palić, D., Andreasen C.B., Herolt., D.M., Menzel, B.W., Roth, J.A., 2006. Immunomodulatory effects of beta-glucan on neutrophil function in fathead minnows (Pimephales promelas Rafinesque, 1820).

Park, D., Ryu, S., Kim, S., Byun, H., Yoon, C., Lee, K., 2013. Airborne bacteria and fungi associated with waste-handling work. Int. J. Occup. Environ. Health. 19(4):313-317.

Parker, B.C., Ford, M.A., Gruft, H., Falkinham, J.O., 1983. Epidemiology of infection by nontuberculous mycobacteria. IV. Preferential aerosolization of Mycobacterium intracellulare from natural waters. Am. Rev. Respir. Dis. 128(4): 652–656. Parks, D.H., Rinke, C., Chuvochina, M., Chaumeil, P.A., Woodcroft, B.J., Evans, P., et al. 2017. Recovery of nearly 8,000 metagenome-assembled genomes substantially expands the tree of life. Nat. Microbiol. 2(11):1533–1542.

Pasanen, P., Pasanen, A.L., Janunen, M., 2000. Water condensation promotes fungal growth in ventilation ducts. Indoor Air. 3:106–112.

Patterson, T.F., Thompson, G.R., Denning, D.W., Fishman, J.A., Hadley, S., Herbrecht, R., et al., 2016. Practice Guidelines for the Diagnosis and Management of Aspergillosis: 2016 Update by the Infectious Diseases Society of America. Clin. Infect. Dis. 63(4):e1-e60.

Paulson, J.N., Stine, O.O., Bravo, H.O., Pop, M., 2013. Robust methods for differential analysis in marker gene surveys. Nat. Methods. 10(12): 1200-1202.

302

Peay, K.G., Kennedy, P.G., Talbot, J.M. 2016. Dimensions of biodiversity in the earth mycobiome. Nat. Rev. Microbiol. 14: 434-447.

Peccia, J., Hospodsky, D., Kyle, B., 2010. New Directions: A revolution in DNA sequencing now allows for the meaningful integration of biology with aerosol science. Atmos. Environ. 45(10):1896-1897.

Perrone, G., Susca, A., 2016. Penicillium Species and Their Associated Mycotoxins. Methods Mol. Biol. 1542:107-119.

Persoons, R., Parat, S., Stoklov, M., Perdrix, A. and Maitre, A., 2010. Critical working tasks and determinants of exposure to bioaerosols and MVOC at composting facilities. Int. J. Hyg. Environ. Health. 213:338-247.

Philippot, L., Raaijmakers, J.M., Lemanceau, P., and van der putten, W.H., 2013. Going back to the roots: the microbial ecology of the rhizosphere. Nat. Rev. Microbio. 11(11):789-799.

Pieckova E., Wilkins K., 2004. Airway toxicity of house dust and its fungal composition. Ann. Agric. Environ. Med. 11:67–73.

Pillai, S.D., Ricke, S.C., 2002. Bioaerosols from municipal and animal wastes: background and contemporary issues. Can. J. Microbiol. 48(8):681–696.

Pitt, J.I., Hocking, A.D., 1977. Influence of solute and hydrogen ion concentration on the water relations of some xerophilic fungi. J. Gen. Microbiol. 101(1):35-40.

Pitt, J.I., Hocking, A.D., 2009. The Ecology of Fungal Food Spoilage. In: Fungi and Food Spoilage. Springer, Boston, MA, USA.

303

Poperscu, S., Borda, C., Diugan, E., 2011. Microbiological air quality in tie-stall dairy barns and some factors that influence it. Afr. J. Agric. Res. 6(32):6726-34. Porter, P., Susarla, S., Polikepahad, S., Qian, Y., Hampton, J., Kiss, A., et al., 2009. Link between allergic asthma and airway mucosal infection suggested by proteinase-secreting household fungi. Mucosal Immunol. 2(6):504–517.

Priyanka, K., Woo, C., Yamamoto, N., Choi, H.L., 2016. Variations in abundance, diversity and community composition of airborne fungi in swine houses across seasons. Sci. Rep. 6:37929.

Prussin II, A.J., Marr, L.C., 2015. Sources of airborne microorganisms in the built environment. Microbiome. 3:78.

Prussin II, A.J., Marr, L.C., Bibby, K.J., 2014 Challenges of studying viral aerosol metagenomics and communities in comparison with bacterial and fungal aerosols. FEMS Microbiol. 357(1):1-9.

Qian, J., Hospodsky, D., Yamamoto, N., Nazaroff, W.W., Peccia, J., 2012. Size-resolved emission rates of airborne bacteria and fungi in an occupied classroom. Indoor Air. 22(4):339–351.

Quail, M.A., Smith, M., Coupland, P., Otto, T.D., Harris, S.R., Connor, T.R., 2012. A tale of three next generation sequencing platforms: comparison of Ion Torrent, Pacific Biosciences and Illumina MiSeq sequencers. BMC Genomics. 13(341): 22827831.

Quince, C., Lanzen, A., Davenport, R.J., Turnbaugh, P.J. 2011. Removing noise from pyrosequenced amplicons. BMC Bioinformatics.12:38.

Quince, C., Walker, A.W., Simpson, J.T., Loman, N.J. Segata, N., 2017. Shotgun metagenomics, from sampling to analysis. Nat. Biotechnol. 35(9):833-844.

Quinn, G.P., Keough, M.J., 2002. Experimental design and data analyses for biologists. Cambridge: Cambridge University Press. Cambridge, UK.

Ramette, A., 2007. Multivariate analyses in microbial ecology. FEMS Microbiol. Ecol. 62(2):142–160.

304

Rawlings, B. A., Higgins, T. S., Han, J. K., 2013. Bacterial pathogens in the nasopharynx, nasal cavity, and osteomeatal complex during wellness and viral infection. Am. J. Rhinol. Allergy. 27(1):39–42.

Reuter, J.A., Spacek, D., Snyder, M.P., 2015. High-Throughput Sequencing Technologies. Mol. Cell. 58(4):586-597.

Riario Sforza, G.G., Marinou, A., 2017. Hypersensitivity pneumonitis: a complex lung disease. Clin. Mol. Allergy. 15:6.

Rodriguez, Z.C., Ramos, M.G., 2014. Acremonium species associated fungemia: a novel pathogen in the immunosuppressed patient. Bol. Asoc. Med. P. R. 106(3):29-31.

Roe, A.D., Rice, A.V., Bromilow, S.E., Cooke, J.E., Sperling, F.A. 2010. Multilocus species identification and fungal DNA barcoding: insights from blue stain fungal symbionts of the mountain pine beetle. Molecular Ecology Resources, 10, 946–959.

Rognes, T., Flouri, T., Nichols, B., Quince, C., Mahé, F., 2016. VSEARCH: a versatile open source tool for metagenomics. PeerJ. 4:e2584.

Rokach, L., Maimon, O., 2005. Clustering methods. Data mining and knowledge discovery handbook. Springer. New York, NY, USA.

Ropars, J., Dupont, J., Fontanillas, E., Rodríguez de la Vega, R.C., Malagnac, F., et al., 2011. Sex in Cheese: evidence for sexuality in the fungus Penicillium roqueforti. PLoS ONE. 7(11):e49665.

Ross, M.G., Russ, C., Costello, M., Hollinger, A., Lennon, N.J., Hegarty, R., et al. 2013. Characterizing and measuring bias in sequence data. Genome Biol. 14(5):R51.

Rossman, A.Y., Palm, M.E., 2006. Why are Phytophtora and other Oomycota not true Fungi? USDA Agricultural Research Service, USDA Animal and Plant Health Inspection Service, Systematic Botany & Mycology Laboratory, Beltsville, Maryland 20705.

Roy, C.J., Milton, D.K., 2004. Airborne Transmission of Communicable Infection - The Elusive Pathway. N. Engl. J. Med. 20(350):1710-12.

Ruzer, L.S., Harley, N.H., 2005. Aerosols handbook: measurement, dosimetry and health effects. CRC Press. Florida, USA.

305

Ryckeboer, J., Mergaert, J., Vaes, K., Klammer, S., De Clercq, D., Coosemans, J., et al., 2003. A survey of bacteria and fungi occurring during composting and self-heating processes. Ann. Microbial. 53:349-410.

Rylander, R., 1996. Airway Responsiveness and Chest Symptoms after Inhalation of Endotoxin or (1→3)-β-D-Glucan. Indoor Built Environ. 5:106-111.

Saari, S., Niemi, J.V., Rönkkö, T., Kuuluvainen, H., Järvinen, A., Pirjola, L., Aurela, M., Hillamo, R., Keskinen, J., 2015. Seasonal and diurnal variations of fluorescent bioaerosol concentration and size distribution in the urban environment. Aerosol Air Qual. Res. 15: 572- 581.

Saitou, N., Nei, M., 1987. The neighbor-joining method: an new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4(4):406-25.

Saltelli, A., Marivoet, J., 1990. Non-parametric statistics in sensitivity analysis for model output: A comparison of selected techniques. Reliability Engineering & System Safety. 28(2):229-253.

Samson, R.A., Visagie, C.M., Houbraken, J., Hong, S.B., Hubka, V., Klaassen, C.H., et al., 2014. Phylogeny, identification and nomenclature of the genus Aspergillus. Stud Mycol. 78: 141-173.

Sanchez-Monedero M.A., Stentiford, E.I. 2003. Generation and dispersion of airborne microorganisms from composting facilities. Trans IChem 81(3):166-170.

Sánchez-Monedero, M.A., Stentiford, E.I.,. Urpilainen, S.T. 2005. Bioaerosol Generation at Large-Scale Green Waste Composting Plants. J. Air Waste Manag. Assoc. 55(5):612-618.

Santos, V., Figueiredo, J.P., Vasconcelos, P., 2018. Occupational exposure to bioaerosols in the waste sorting industry. Occupational Safety and Hygiene VI - Arezes et al., (Eds). Taylor & Francis Group, London, UK.

Sarkar, A., Mukherjee, A., Ghoshal, A.G., Kundu, S., Mitra, S., 2010. Occurrence of allergic bronchopulmonary mycosis in patients with asthma: An Eastern India experience. Lung India : Official Organ of Indian Chest Society; 27(4):212-216.

306

Sautour, M., Sixt, N., Dalle, F., L’Ollivier, C., Fourquenet, V., Calinon, C., et al., 2009. Profiles and seasonal distribution of airborne fungi in indoor and outdoor environments at a French hospital. Sci. Total Environ. 407:3766–37671.

Schiffman, S.S., Studwell, C.E., Landerman, L.R., Berman, K., Sundy, J.S. 2005. Symptomatic effects of exposure to diluted air sampled from a swine confinement atmosphere on healthy human subjects. Environ. Health Perspect. 113(5):567-576.

Schloss, J.A., 2008. How to get genomes at one ten-thousandth the cost. Nat. Biotechnol. 26:1113–1115.

Schlosser, O., Huyard, A., Rybacki, D., Do Quang, Z., 2012. Protection of the vehicle cab environement against bacteria, fungi and endotoxins in composting facilities. Waste manag. 32(6):1106-1115.

Schubert, M.S., 2009. Allergic fungal sinusitis: pathophysiology, diagnosis and management. Med. Mycol. 47( Suppl 1):S324-330.

Sebghati, T.S., Engle, J.T., Goldman, W.E., 2000. Intracellular parasitism by Histoplasma capsulatum: fungal virulence and calcium dependence. Science. 290(5495):1368-1372.

Seifert, S.A., Von Essen, S., Jacobitz, K., Crouch, R., Lintner, C.P., 2003. Organic dust toxic syndrome: a review. J. Toxicol. 41(2):285-193.

Selman, M., Lacasse, Y., Pardo, A., Cormier, Y., 2010. Hypersensitivity pneumonitis caused by fungi. Proc Am Thorac. Soc. 7(3):229–23.

Selman, M., Padro, A., King, T.E. jr., 2012. Hypersensitivity pneumonitis: insights in diagnosis and pathobiology. Am. J. Respir. Crit. Care Med. 186(4):314-324.

307

Sergeev, A.R., Demina, O.K., Pyankov, O.V., Pyankova, O.G., Agafonov, A.P., Kiselev, et al., 2013. Infection of chickens caused by avian Influenza virus A/H5N1 delivered by aerosol and other routes. Transbound. Emerg. Dis. 60(2):159-165.

Shade, A., Handelsman, J., 2011. Beyond the Venn diagram: the hunt for a core microbiome. Environ Microbiol. 14(1):4-12.

Shade, A., Hogan, C.S., Klimowicz, A.K., Linske, M., McManus, P.S., Handelsman, J., 2012. Culturing captures members of the rare biosphere. Environ. Microbiol. 14(9):2247– 2252.

Shams-Ghahfarokhi, M., Aghaei-Gharehbolagh, S., Aslani, N., Razzaghi-Abyaneh, M., 2014. Investigation on distribution of airborne fungi in outdoor environment in Tehran, Iran. J. Environ Health Sci Eng. 12(1):54.

Sharpe, R.A., Bearman, N., Thornton, C.R., Husk, K., Osborne, N.J., 2015. Indoor fungal diversity and asthma: A meta-analysis and systematic review of risk factors. J. Allergy Clin. Immuno. 135(1):110-122.

Sharpton, T.J. 2014. An introduction to the analysis of shotgun metagenomic data. Front Plant Sci. 5:209.

Sherf, A.F., 1987. Vegetable Diseases and Their Control. (éd. Second Edition ). Wiley & sons. New York, NY, USA.

Shin, J., Lee, S., Go, M.J., Lee, S.Y., Kim, S.C., Lee, C.H., et al., 2016. Analysis of the mouse gut microbiome using full-length 16S rRNA amplicon sequencing. Sci. Rep. 6:29681.

Simon, X., Duquenne, P., 2014. Assessment of workers’ exposure to bioaerosols in a French cheese factory. Ann. Occup. Hyg. 58(6):677-92.

Sims, D., Sudbury, I., Ilott, N.E., Heger, A., Ponting, C.P., 2014. Sequencing depth and coverage: key considerations in genomic analyses. Nat. Rev. Genet. 15(2);121-132.

Smith, D.R., 2015. Broadening the definition of a bioinformatician. Front. Genet. 6:258.

Soergel, D.A., Dey, N., Knight, R., Brenner, S.E., 2012. Selection of primers for optimal taxonomic classification of environmental 16S rRNA gene sequences. ISME J. 6:1440–1444.

Sogin, M.L., 1991. Early evolution and the origin of eukaryotes. Curr. Opin. Genet. Dev. 1(4):457-463.

308

Sokal, R.R., Rohlf, F.J., 1995. The principles and practices of statistics in biological research. 4th ed. Stony Brook University. WH Freeman and Company. New York, NY, USA.

Sosnowski, P., Wieczorek, A., Ledakowicz, S., 2003. Anaerobic co-digestion of sewage sludge and organic fraction of municipal solid wastes. Adv. Environ. Res. 7(3):609-616.

Spatafora, J.W., Chang, Y., Benny, G.L., Lazarus, K., Smith, M.E., Berbee, M.L., et al. 2016. A phylum-level phylogenetic classification of Zygomycetes fungi based on genome-scale data. Mycologia. 108(5):1028-1046.

Stajich, J.E., Berbee, M.I., Blackwell, M., Hibbett, D.S., James, T.Y., Spatafora, J.W., et al., 2009. The fungi. Curr. Biol. 19(18):R840-R845.

Sterflinger, K., De Baere, R., de Hoog, G., De Wachter, R., Krumbein, W.E., Haase. G., 1992. Coniosporium perforans and C. apollinis, two new rock-inhabiting fungi isolated from marble in the Sanctuary of Delos (Cyclades, Greece). Antonie Van Leeuwenhoek; 72(4):349-363.

Stiemsma, L.T., Reynolds, L.A., Turvey, S.E., Finlay, B.B., 2015. The hygiene hypothesis: current perspectives and future therapies. Immunotargets Ther. 4:143–157.

Strachan, D.P., 1989. Hay fever, hygiene, and household size. BMJ. 299(6710):1259–1260.

Sturm, R., 2016. Bioaerosols in the lungs of subjects with different ages-part 1: deposition modeling. Ann. Transl. Med. 4(11):211.

Suttle, C.A., 2007. Marine viruses–major players in the global ecosystem. Nat. Rev. Microbiol. 5(10):801–812.

Sykes, P., Jones, K., Wildsmith, J. D., 2007. Managing the potential public health risks from bioaerosol liberation at commercial composting sites in the UK: an analysis of the evidence base. Resour. Conserv. Recycl. 52(2):410-424.

309

Taha, M. P. M., Drew, G. H., Longhurst, P. J., Smith, R., Pollard, S. J. T., 2006. Bioaerosol releases from compost facilities: evaluating passive and active source terms at a green waste facility for improved risk assessments. Atmos. Environ. 40(6):1159–1169.

Tedersoo, L., Sánchez-Ramírez, S., Kõljalg, U., Bahram, M., Döring, M., Schigel, D., et al., 2018. High-level classification of the Fungi and a tool for evolutionary ecological analyses. Fungal Diversity. 90(1):135-159.

Teertstra, W., Deelstra, H., Vranes, M., Bohlmann, R., Kahmann, R., Kämper, J., Wösten. H.A., 2006. Repelents have functionally replaced hydrophobins in mediating attachement to a hydrophobic surface and in formation of hydrophobic aerial hyphae in Ustilago maydis. Microbiology. 152(12):3607-3612.

Terhem, R.B., Van Kan, J.A.L., 2014. Functional analysis of hydrophobin genes in sexual development of Botrytis cinerea. Fungal Genet. Biol. 71:42-51.

Tham, R., Dharmage, S.C., Taylor, P.E., Katelaris, C.H., Vicendese, D., Abramson, M.J., 2014. Outdoor fungi and child asthma health service attendances. Pediatr. Allergy Immunol. 25(5):439-449.

Tikhonov, M., Leach, R.W., Wingreen, N.S., 2015. Interpreting 16 S metagenomic data without clustering to achieve sub-OTU resolution. ISME J. 9(1):68–80.

310

Toju, H., Tanabe, A. S., Yamamoto, S., Sato, H., 2012. High-coverage ITS primers for the DNA-based identification of Ascomycetes and Basidiomycetes in Environmental Samples. Plos One. 7(7): e40863.

Tong, Y., Lighthart, B., 1998. Effect of simulated solar radiation on mixed outdoor atmospheric bacterial populations. FEMS Microbiol. Ecol. 26(4):311-316.

Tringe, S.G., Hugenholtz, P., 2008. A renaissance for the pioneering 16S rRNA gene. Curr. Opin. Microbiol. 11(5):442–446.

Trontelj, P., Douady, C.J., Fiser, C., Gibert, J., Spela, G., Lefébure, T., et al., 2009. A molecular test for cryptic diversity in ground water: how large are the ranges of macro- stygobionts? Freshw. Biol. 54(4):727-744.

Truong, D.T., Tett, A., Pasolli, E., Huttenhower, C. Segata, N., 2015. MetaPhlAn2 for enhanced metagenomic taxonomic profiling. Nat. Methods . 2:902-903.

Tsapko, V.G., Chudnovets, A.J., Sterenbogen, M.J., Papach, V.V., Dutckiewicz, J., Skorska, C, et al., 2011. Exposure to bioaerosols in the selected agricultural facilities of the Ukraine and Poland - A review. Ann. Agric. Environ. Med. 18(1):19–27.

Tuck, A., 2002. The role of atmospheric aerosols in the origin of life. Surv. Geophys. 23(5):379-409.

Tuomisto, H., 2010. A diversity of beta diversities: straightening up a concept gone awry. Part 1. Defining beta diversity as a function of alpha and gamma diversity. Ecography. 33(1):2-22.

Turnbaugh, P.J., Hamady, M., Yatsunenko, T., Cantarel, B.L., Duncan, A., Ley, R.E., et al., 2009. A core gut microbiome in obese and lean twins. Nature. 457(7228):480-484.

Tyson, G.W., Chapman, J., Hugenholtz, P., Allen, E.E., Ram, R.J., Richardson, P.M., et al., 2004. Community structure and metabolism through reconstruction of microbial genomes from the environment. Nature. 428:37-43.

311

Veech, J.A., Crist, T.O. 2010. Diversity partitioning without statistical independence of alpha and beta. Ecology. 91(7):1964-1969.

Veillette, M., Thorne, P.S., Gordon, T., Duchaine, C., 2004. Six month tracking of microbial growth in a metalworking fluid after system cleaning and recharging. Ann. Occup. Hyg. 48(6):541-546.

Velegraki, A., Cafarchia, C., Gaitanis, G., Iatta, R., Boekhout, T., 2015. Malassezia Infections in Humans and Animals: Pathophysiology, Detection, and Treatment. PLoS Pathog. 11(1): e1004523.

Venter, J.C., Remington, K., Heidleberg, J.F., Halpern, A.L., Rusch, D., Eisen, J.A., et al., 2004. Environmental genome shotgun sequencing of the Sargasso Sea. Science. 304(5667):66–74.

Verreault, D., Moineau, S., Duchaine C., 2008. Methods for Sampling of Airborne Viruses. Microbiol Mol. Biol. Rev. 72(3):413-444.

Vickers, A.J., 2005. Parametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data. BMC Med. Res. Methodol. 5:35.

Vincent, A.T., Charette, S.J., 2015. Who qualifies to be a bioinformatician? Front. Genet. 6:164.

Voordouw, G., Voordouw, J.K., Karkhoff-Schweizer, R.R., Fedorak, P.M., Westlake, D.W.S., 1991. Reverse Sample Genome Probing, a new technique for identification of bacteria in environmental samples by DNA hybridization, and its application to the identification of sulfate-reducing bacteria in oil field samples. Appl. Environ. Microbiol. 57(11):3070-3078.

312

Walser, S.M., Gerstner, D.G., Brenner, B., Bünger, J., Eikmann, T., Janssen, B., et al., 2015. Evaluation of exposure-response relationships for health effects of microbial bioaerosols – A systematic review. Int. J. Hyg. Environ. Health. 218(7):577-589.

Walter, J., Ley, R., 2011. The human gut microbiome: ecology and recent evolutionary changes. Annu. Rev. Microbiol. 65:411-429.

Walther, T.C., Farese, R.V. Jr., 2012. Lipid droplets and cellular lipid metabolism. Annu. Rev. Biochem. 81:687–714.

Wang, X.C., Liu, C., Huang, L., Bengtsson-Palme, J., Chen, H., Zhang, J.H., et al., 2014. ITS1: a DNA barcode better than ITS2 in eukaryotes? Mol. Ecol. Res. 15(3):573-586.

Welte, M.A., 2015. Expanding roles for lipid droplets. Curr. Biol. 25(11):R470–R481.

Wéry, N. 2014., Bioaerosols from composting facilities – a review. Front. Cell Infection Microbiol. 4:42.

Whittaker, R.H. 1972. Evolution and Measurement of Species Diversity. Taxon. 21:213- 251.

Wiederhold, N.P., 2017. Antifungal resistance: current trends and future strategies to combat. Infection and drug resistance. Infect. Drug. Resist. 10:249-259.

Womack, A.M., Bohannan, B.J.M., Green, J.L., 2010. Biodiversity and biogeography of the atmosphere. Philos. Trans. R. Soc. B. 365(1558):3645–3653. Woolford, M.K., Pahlow, G., 1998. The silage fermentation. Microbiology of Fermented Foods, pp. 73–102. Springer US.

313

Wösten, H.A.B., 2001. Hydrophobins: Multipurpose Proteins. Annu. Rev. Microbiol. 55:625-646.

Wösten, H.A.B., Van Wetter, M.A., Lugones, L.G., Van Der Mei, H.C., Busscher, H.J., Wessels, J.G.H., 1999. How a fungus escape water to grow into the air. Curr. Biol. 9 (2):85- 88.

Wright, E.S., Yilmaz, L.S., Noguera, D.R., 2012. DECIPHER, a search based approach to chimera identification for 16S rRNA sequences. Appl. Environ. Microbiol. 78:717–725.

Wyngaarden, J. B., Smith, L.H., Bennett, J.C., (ed.), 1992. Cecil textbook of medicine, 19th ed., p.399and 407. W.B. Saunders Co.. Philadelphia, Pa, USA.

Yamamoto, N., Nazaroff, W., Peccia, J., 2014. Assessing Aerodynamic diameters of taxon- specific fungal bioaerosols by quantitative PCR and next-generation DNA sequencing. J. Aerosol Sci. 78:1-10.

Yang, B., Wang, Y., Qian, P.Y., 2016. Sensitivity and correlation of hypervariable regions in 16S rRNA genes in phylogenetic analysis. BMC Bioinformatics. 17:135.

Yang, H.J., Osakada, H., Kojidani, T., Haraguchi, T., Hiraoka, Y., 2017. Lipid droplet dynamics during Schizosaccharomyces pombe sporulation and their role in spore survival. Biol. Open. 6:217-222.

Yarza, P., Yilmaz, P., Panzer, K., Glöckner, F.O., Reich, M. 2017. A phylogenetic framework for the kingdom Fungi based on 18S rRNA gene sequences. Mar Genomics. 36:33-39.

Yilmaz, N., Visagie, C.M., Houbraken, J., Samson, R.A., 2014. Polyphasic taxonomy of the genus Talaromyces. Stud. Mycol. 78:175-341.

Yoo, K., Lee, T.K., Choi, E.J., Yang, J., Shukla, S.K., Hwang, S.I., et al., 2017. Molecular approaches for the detection and monitoring of microbial communities in bioaerosols: A review. J. Environ. Sci. 51:234-247.

Yoon, S.S., Kim, E.K., Lee, W.J., 2015. Functional genomic and metagenomic approaches to understanding gut microbiota-animal mutualism. Curr. Opin. Microbiol. 24:38-46. Yu, I.T.S, Li, Y., Wong, T.W., Tam, W., Chan, A.T., Lee, J.H.W., et al., 2004. Evidence of airborne transmission of the Severe Acute Respiratory Syndrome Virus. N. Engl. J. Med. 350:1731-1739.

314

Ziaee, A., Zia, M. Goli, M., 2018. Identification of saprophytic and allergenic fungi in indoor and outdoor environments. Environ. Monit. Assess.190(10): 574.

Zimmerman, D.W., 1994. A note one the influence of outliers on parametric and nonparametric tests. J. Gen. Psychol. 121(4):391-401.

Zur, E.F., Ieno, E.N., Smith, G.M., 2007. Analyzing ecological data. Springer. New York, NY, USA.

315