UNIVERSITY OF COPENHAGEN DEPARTMENT OF BIOLOGY

PhD thesis Klara Junker

Prospecting a Predator: Exploration of the biocontrol potential of the mycoparasitic yeast Saccharomycopsis schoenii and an integrated microscopic and multiomic characterization of its predatory behaviour

Supervisor: Birgitte Regenberg, University of Copenhagen Co-supervisor: Jürgen Wendland, Vrije Universiteit, Brussel

This thesis has been submitted to the PhD School of The Faculty of Science, University of Copenhagen on the 6th of July, 2018

Name of department: Department of Biology

Author: Klara Junker

Title and subtitle: Prospecting a predator: Exploration of the biocontrol potential of the mycoparasitic yeast Saccharomycopsis schoenii and an integrated microscopic and multiomic characterization of its predatory behaviour

Supervisor: Assoc. Prof. Birgitte Regenberg Department of Biology, University of Copenhagen, Copenhagen, Denmark

Co-supervisor: Prof. Jürgen Wendland Research Group of Microbiology, Vrije Universiteit Brussel, Brussels, Belgium

Submitted on: The 6th of July, 2018

Assessment committee: Assoc. Prof. Rasmus Kjøller Department of Biology, University of Copenhagen, Copenhagen, Denmark

Dr. Alexandra Brand MRC Centre for Medical Mycology, University of Aberdeen, Aberdeen, Scotland

Dr. Gianni Liti Institute of Research on Cancer and Ageing of Nice, CNRS, INSERM, Université de Nice, Nice, France

”Klara klarar allt” ~ Siri Johansson

~

In dedication to my grandmother Siri Johansson, 26/4 1916 ~ 14/11 2017

Det här är resultatet av möjligheter jag önskar du också fått

Prospecting a Predator | Klara Junker

Table of contents

Preface ...... 7

Summary ...... 8

Resumé ...... 10

List of papers ...... 12

Aim & objectives ...... 13

Introduction ...... 14

1. Introduction to fungi ...... 14 1.1. Fungal nutrition ...... 17 1.2. Nutrient acquisition tactics ...... 18 1.3. Mycoparasitism ...... 19 1.4. Biocontrol fungi ...... 20 1.5. The fungal cell wall ...... 22 1.6. Fungal parasite genomics ...... 23 1.7. Human fungal pathogens ...... 24

2. Yeasts ...... 25 2.1. Saccharomyces cerevisiae: the model yeast organism ...... 25 2.2. Human yeast pathogens in the Candida genus ...... 26 2.3. Candida albicans: a commensal yeast in CTG clade ...... 26 2.4. Candida auris: an emerging multidrug resistant pathogen ...... 28 2.5. Saccharomycopsis yeasts: necrotrophic mycoparasites ...... 29

5 Klara Junker | Prospecting a Predator

Materials & methods ...... 32

1. Strains & culture conditions ...... 32

2. Live cell microscopy ...... 34

3. Genome sequencing & assembly ...... 36

4. Genome annotation ...... 37

5. Proteomic analyses ...... 39

6. Transcriptomic analyses ...... 40

7. GO term analysis ...... 41

8. Genetic manipulations of S. schoenii ...... 42

Summary of papers ...... 43

Results & Discussion ...... 46

Perspectives for further research ...... 55

Conclusions ...... 57

Acknowledgements ...... 58

References ...... 62

Papers ...... 68

Paper I ...... 69

Paper II ...... 73

Paper III ...... 77

Paper IV ...... 103

6 Prospecting a Predator | Klara Junker

Preface

This work was funded by the Marie Curie Initial Training Network “FUNGIBRAIN”, Grant agreement no. 607963, from the European Union. The majority of research was carried out in the Yeast and Fermentation group at the Carlsberg Research Laboratory, Copenhagen, Denmark. Work with clinical isolates of Candida species, S. cerevisiae and S. pombe was performed in collaboration with Neil Gow, Alexander Lorenz and Gustavo Bravo of the Aberdeen Fungal Group, at the University of Aberdeen, Aberdeen, UK.

7 Klara Junker | Prospecting a Predator

Summary

Fungal infections have in the recent years gone from being associated to minor annoyances, to being recognized as a diverse group of emerging killers, now even outnumbering the death toll of malaria. The human body is adept at keeping fungal pathogens at bay, but when our immune system is down, pathogenic yeast can infect and even kill us. One of the most recently emerging pathogens, the yeast Candida auris, was first described in Japan in 2009, but has already spread across the globe. In addition, several strains of C. auris are multi drug resistant. Development of antifungal drugs has been even less prioritized than the development of new bacterial antibiotics, but the need for novel strategies to combat fungal infections is real. This thesis explores the antifungal mechanisms and potentials of Saccharomycopsis schoenii, a little-studied mycoparasitic yeast with the rare ability to attack and kill other fungi including, as I demonstrate, multi drug resistant C. auris.

Fungi are diverse in the way they feed themselves. Some filamentous fungi parasitize on plants, whereas yet others are mycoparasites, meaning that they can parasitize on other fungi. A common feature of these parasites is that they have often lost the genes that allow them to synthesise essential nutrients that they instead are able to acquire from their hosts. In parallel, they have increased the copy numbers of genes that allow them to parasitize. For instance, filamentous fungi in the mycoparasitic genus Trichoderma, used as biocontrol agents against fungal plant pathogens, have expanded gene families encoding for cell wall degrading enzymes, that they use when they antagonize or kill other fungi. Similarly, the yeast pathogen C. albicans also use an expanded set of proteases when it infects humans.

In this thesis I made use of state-of-the-art next generation sequencing and bioinformatics as well as live cell imaging techniques with the aim of genetically and functionally characterizing Saccharomycopsis yeasts. We first generated draft genomes of three Saccharomycopsis yeasts, S. fodiens, S. fermentans and S. schoenii and found that the genomic reason that these yeasts, unlike all other yeasts, are unable to assimilate inorganic sulfate is because they lack all genes in the sulfate assimilation pathway. In parallel, we also found that they, analogous to filamentous mycoparasitic fungi, have expanded their sets of genes encoding proteases, chitinases and glucanases that can break down fungal cell walls.

8 Prospecting a Predator | Klara Junker

Through genomic analyses, we also found special tRNA-genes in the Saccharomycopsis yeasts, that suggested they belong to the CTG clade, a group of yeasts that translate the CTG codon to serine instead of leucine. We were able to confirm the translation of CTG to serine, through a proteomic assay.

By means of live microscopy, I provided spatial and temporal visualization of how the S. schoenii attacks and kills S. cerevisiae, a model prey cell. Through microscopy-based assays I set up, I was also able to determine that it is the lack of complex nitrogenous compounds, and not solely the sulfur amino acid methionine, as previously hypothesised, that is the main trigger of predatory activity.

To determine the genetic tools S. schoenii employ during its predatory activity, I performed extensive transcriptomic analyses. These revealed predation-specific upregulation of cell wall specific proteases, transposable elements and sulfur scavenging transporters. In total, my transcriptomic analysis suggested that S. schoenii breaks down the cell wall of its fungal prey through protein hydrolysis, and that methionine and other sulfur compounds were acquired from the prey cells.

To explore if the fungicidal properties of S. schoenii could be applied to medically important yeast pathogens, I set up an assay where I could analyse the susceptibility of yeast pathogens in the Candida genus. Indeed, I found that S. schoenii was able to kill clinical isolates of C. albicans, C. glabrata, C. parapsilopsis, C. tropicalis, and C. lutsitaniae as well as both sensitive and multi drug resistant isolates of C. auris.

9 Klara Junker | Prospecting a Predator

Resumé

Svampeinfektioner er i de seneste år gået fra at være forbundet med mindre irritationer, at blive anerkendt som en forskelligartet gruppe af nye dræbende, nu endda mere end døden af malaria. Den menneskelige krop er dygtig til at holde svampepatogener i skak, men når vores immunforsvar er nede, kan patogen gær inficere og endda dræbe os. En af de senest dukkede op patogener, gæren Candida auris, blev først beskrevet i Japan i 2009, men har allerede spredt sig over hele kloden. Derudover er flere stammer af C. auris multi-resistente. Udvikling af svampedræbende stoffer har været endnu mindre prioriteret end udviklingen af nye bakterielle antibiotika, men behovet for nye strategier til bekæmpelse af svampeinfektioner er reel. Denne afhandling udforsker antifungale mekanismer og potentialer hos Saccharomycopsis schoenii, en lille undersøgt mycoparasitisk gær med den sjældne evne til at angribe og dræbe andre svampe, inklusive, som jeg demonstrerer her, multiresistente C. auris.

Svampe er forskelligt i den måde de spiser på. Nogle filamentøse svampe parasiterer på planter, mens endnu andre er mycoparasitter, hvilket betyder at de kan parasitere på andre svampe. Et fælles træk ved disse parasitter er at de ofte har mistet de gener der giver dem mulighed for at syntetisere essentielle næringsstoffer, som de i stedet kan erhverve fra deres værter. Parallelt har de øget kopienumrene af gener, der tillader dem at parasitere. For eksempel har filamentøse svampe i den mycoparasitiske slægt Trichoderma, der anvendes som biokontrolmidler mod svampeplantepatogener, udvidet genfamilier der koder for cellevægnedbrydende proteiner, som de bruger når de modvirker eller dræber andre svampe. Tilsvarende bruger gærpatogenet C. albicans også et ekspanderet sæt protein nedbrydende enzymer, når de inficerer mennesker.

I denne afhandling anvendte jeg den nyeste generation af sekvensering og bioinformatik samt levende celledannelsesteknikker med det formål at genetisk og funktionelt karakterisere Saccharomycopsis-gær. Vi genererede de første udkast til genomer af tre Saccharomycopsis-gærer, S. fodiens, S. fermentans og S. schoenii og fandt, at den genomiske grund til at disse gær, i modsætning til alle andre gær, ikke er i stand til at optage uorganisk sulfat, er fordi de mangler alle de gener der styrer sulfat optagelse. Parallelt fandt vi også at

10 Prospecting a Predator | Klara Junker de, analogt med filamentøse mycoparasitiske svampe, har udvidet deres sæt af gener kodende for proteaser, chitinaser og glucanaser der kan nedbryde svampecellevægge.

Gennem genomiske analyser fandt vi også specielle tRNA-gener i Saccharomycopsis- gærene, der foreslog at de tilhører CTG-gruppen, en gruppe af gær der oversætter CTG- codonet til serin i stedet for leucin. Vi kunne bekræfte oversættelsen af CTG til serin gennem en proteomisk analyse.

Ved hjælp af levende mikroskopi tilvejebragte jeg rumlig og tidsmæssig visualisering af hvordan S. schoenii angriber og dræber S. cerevisiae, en modelbyttecelle. Gennem mikroskopibaserede analyser der jeg oprettede, var jeg også i stand til at fastslå, at det er manglen på komplekse nitrogenholdige stoffer, og ikke kun svovlaminosyren methionin, som tidligere antydet, det er den primære udløser af rovdyrsaktivitet.

For at bestemme de genetiske værktøjer, der S. schoenii anvender under sin rovaktivitet, udførte jeg omfattende transkriptomiske analyser. Disse afslørede rovdyrsspecifik opregulering af cellevægspecifikke proteaser, transponeringselementer og svovlopfangende transportører. I alt foreslog min transkriptomisks analyse, at S. schoenii nedbryder den svampelagtige cellevæg gennem proteinhydrolyse, og at methionin og andre svovlforbindelser blev erhvervet fra rovcellerne.

For at undersøge om S. schoenii's fungicide egenskaber kunne anvendes til medicinske vigtige gærpatogener satte jeg op en analyse hvor jeg kunne analysere modtageligheden af gærpatogener i Candida-slægten. Faktisk fandt jeg, at S. schoenii var i stand til at dræbe kliniske isolater af C. albicans, C. glabrata, C. parapsilopsis, C. tropicalis og C. lutsitaniae såvel som både følsomme og multiresistente isolater af C. auris.

11 Klara Junker | Prospecting a Predator

List of papers

Paper I Junker K., Hesselbart A. Wendland J. (2017) Draft genome sequence of Saccharomycopsis fodiens CBS 8332, a necrotrophic mycoparasite with biocontrol potential. Genome Announc 5:e01278-17. https://doi.org/10.1128/genomeA.01278-17.

Paper II Hesselbart A., Junker K. and Wendland J. (2018) Draft genome sequence of Saccharomycopsis fermentans CBS 7830, a predacious yeast belonging to the Saccharomycetales. Genome Announc 6:e01445-17. https://doi.org/10.1128/genomeA.01445-17.

Paper III Junker K. Chailyan A., Hesselbart, A. and Wendland, J. (2018) Multi-omic identification of genes involved in the necrotrophic mycoparasitism exerted by the yeast Saccharomycopsis schoenii. Manuscript

Paper IV Junker K, Bravo Ruiz G., Lorenz A., Walker L., Gow N. A. R. and Wendland J. (2018) The mycoparasitic yeast Saccharomycopsis schoenii predates and kills multi-drug resistant Candida auris. Under review, Scientific Reports [manuscript number SREP-18-19093-T]

12 Prospecting a Predator | Klara Junker

Aim & objectives

The overall aim of this thesis was to identify the genetic pathways and molecular mechanisms involved in the predatory behaviour of Saccharomycopsis yeasts, in particular Saccharomycopsis schoenii, from the perspective of its potential use as a fungicidal biocontrol yeast.

Specific objectives were:

1. To de novo sequence, assemble and annotate draft genomes of predatory Saccharomycopsis yeasts, including S. fodiens, S. fermentans and S. schoenii. 2. To mine the draft genomes of S. schoenii, S. fodiens and S. fermentans for any genetic basis of their genus-specific nutritional requirements and to use the draft genomes to validate genetic relationships between Saccharomycopsis species and other fungal species. 3. To establish S. cerevisiae as a model prey cell for studying the predatory behaviour of S. schoenii and to set up a live cell imaging protocol to determine the timing and the physical and nutritional requirements for S. schoenii to attack S. cerevisiae. 4. To discover and verify genetic pathways involved in nutritional stress responses, predatory behaviour and life cycle of S. schoenii, by generating and analysing genome- wide transcriptomic and proteomic data. 5. To explore the range of prey cells susceptible to attack by S. schoenii to uncover its potential as a prospective biocontrol yeast. 6. To establish a protocol for targeted genetic manipulations in S. schoenii to experimentally validate key genetic pathways involved in its predatory behaviour.

13 Klara Junker | Prospecting a Predator

Introduction

Modern medicine has prevented millions of people from dying of common infections, cancers and immunodeficiency viruses. However, when those lifesaving treatments are based on supressing the human immune system, patients are left vulnerable to secondary infections such as fungal infections (Pappas et al., 2018). A healthy human body and immune system is fine-tuned to keep yeast and fungi at bay, but in an immunocompromised state, several species in the Candida genus are quick to take advantage of the situation and spread throughout the human body and blood. Blood stage infections of Candida species can have mortality rate of almost 70% and many isolates of multidrug resistant fungal strains are currently emerging (Fisher et al., 2018). Some isolates of the most recently emerging species, Candida auris, first isolated in 2009, are already drug resistant to several of the few antifungal agents available (Jeffery-Smith et al., 2018).

Because fungal infections historically have not been as devastating or frequent as bacterial or protozoan infections, the development of antifungal drugs has not been prioritized (Fisher et al., 2018). Fungi share many fundamental cellular processes with human cells, making it difficult to develop antifungal drugs without serious side effects. Most antifungal drugs therefore target the fungal cell wall, which is a defining feature of fungal cells (Gow, Latge and Munro, 2017). In the natural world, fungi who attack other fungi also target fungal cell wall components of prey cells (Inglis and Kawchuk, 2002). By better understanding the tools these fungi have developed and fine-tuned for millions of years, researchers might yet again be able to utilize fungal genetics for human health. In this thesis I explore the genetic tools used by a unicellular predatory yeast, as it attacks and kills other yeasts, including the aformentioned C. auris.

1. Introduction to fungi

At first glance, humans and yeast seem quite different, but if you view our relationship us from the global tree of life, we are actually rather close (Hug et al., 2016). Even though animal and fungal lineages parted ways some 1 billion years ago, we still share many central cellular mechanisms (Torruella et al., 2015). A fundamental example; unlike autotrophic

14 Prospecting a Predator | Klara Junker algae and plant cells that themselves can fix inorganic carbon (such as carbon dioxide) by photosynthesis, both animal and fungi are heterotrophs, meaning we need to acquire organic carbon (such as sugar), from sources outside of ourselves.

Figure 1: Different structures of fungi. Left panel, the fruiting body of a fungus. Middle panel, the hyphal structure of filamentous fungi. Right panel, unicellular yeast cells.

The fungal lineage contains at least 1.5 but probably as many as 3 million species, ranging from mushrooms and lichens to filamentous fungi and single celled yeasts, see Figure 1 (Heitman, 2011; Hawksworth, 2012). Fungi have traditionally been classified based on their morphology, phenotype, modes of reproduction and, for single gene comparisons. However, the advent of substantial technical advances in DNA sequencing, that enable comparisons of whole genomes, not just single genes, have revolutionized the fungal family tree. Fungi used to be divided into four phyla, but now eight phyla are recognised, based on molecular phylogenetics and evolutionary genomics, see Figure 2 (Spatafora et al., 2017).

Perhaps contrary to intuition, it was not unicellular yeasts as we know them today that evolved into multicellular fungi, instead, it was multicellular fungi that evolved the ability to live primarily as single-celled organisms (Nagy et al., 2014). Genomic data also suggest that the switch from a multicellular to a primarily unicellular life style, did not just happen once, but is the result of convergent evolution that has generated five distinct yeast clades, see Figure 2. Several yeasts have retained the ability to shapeshift between yeast and hyphal growth forms, depending on environmental conditions, and are called dimorphic yeasts. Of the five yeast clades, the Saccharomycetes class contain the by far most well studied species, including Saccharomyces cerevisiae, also known as baker’s or brewer’s yeast, and the dimorphic human commensal yeast Candida albicans.

15 Klara Junker | Prospecting a Predator

Figure 2: The fungal tree of life, sorted by phyla and classes. Grey boxes; classes containing clades of unicellular yeasts and yeast-like species. Solid black line; Saccharomycetes, containing the Saccharomyces, Candida and Saccharomycopsis genus, that are discussed in this thesis. Figure reprinted with permission from (Spatafora et al., 2017) and adapted according to (Nagy et al., 2014).

16 Prospecting a Predator | Klara Junker

1.1. Fungal nutrition

The basic nutritional needs of fungi are relatively simple under aerobic conditions. The bare minimum typically include fixed carbon, fixed nitrogen, a few growth factors and inorganic ions, including sulfur (Walker and White, 2018). However, there is variation even within primary metabolism. While filamentous fungi are typically able to assimilate nitrate and fix nitrogen, the ability of yeasts to do so is rare, but not unheard of (Siverio, 2002). Sulfur is a central and essential component of numerous cellular functions and components, such as proteins, nucleic acids, vitamins and a number of co-enzymes. The majority of microorganisms, including virtually all fungi, are able to assimilate sulfur in its inorganic form, sulfate (Moat, Foster and Spector, 2002). Nevertheless, a small number of filamentous fungi, the parasitic fungal pathogens Blumeria graminis and Puccinia graminis, have completely lost genes in the sulfate assimilation pathway, which renders them unable to take up sulfate (Spanu et al., 2010; Duplessis, 2011). The only yeast clade known to be unable to assimilate sulfate are yeasts in the Saccharomycopsis clade (Lachance et al., 2000).

In addition to inorganic sulfate, most fungi can assimilate several different sulfur-containing compounds, such as organosulfur compounds or the two sulfur-containing amino acids methionine and cysteine, as sources of sulfur (Marzluf, 1997; Linder, 2018), see Figure 3. For instance, when the yeast Neurospora crassa is deprived of inorganic sulfate, it activates its methionine uptake transport system and starts scavenging organic sulfur (Jennings, 1995). The sulfur-containing amino acid methionine is an essential amino acid and, universally, every protein synthesized starts with methionine, making intracellular methionine fundamental for sustained life. At the same time, methionine is the energetically most expensive amino acid for yeasts to synthesize, and several intermediate molecules in the sulfate assimilation pathway, in which sulfate is reduced and finally incorporated into methionine, are toxic (Thomas and Surdin-Kerjan, 1997; Stephanopoulos, Aristidou and Nielsen, 1998). While the uptake of most other amino acids usually only occurs when the amino acids for some reason are abundant in the environment, the situation is reversed when it comes to methionine. When intracellular and extracellular levels of methionine are low, attempts to increase scavenging increase (Beckerich et al., 2015).

17 Klara Junker | Prospecting a Predator

Figure 3: Genetic pathways related to sulfur metabolism. The pathway is present in most yeasts such as S. cerevisiae and C. albicans, with the exception that S. cerevisiae lacks the OAS pathway. Figure adapted from Marzluf, 1997, Hébert et al. 2011 and Hébert et al. 2013.

The dependence of external availability of a specific organic compound is termed auxotrophy, and a lack of the auxotrophic compound can trigger altered, and often virulent, behaviour in fungi. For instance, Candida glabrata, a yeast that can cause urinary tract infections in humans, lacks the genes for nicotinic acid uptake (Domergue, 2005). Urine is a source of nicotinic acid, and when that source of nicotinic acid is limited, such as when an indwelling catheter is used in a hospital, C. glabrata is triggered to upregulate specific adhesion genes, which impacts its virulence. Rather than being lost more or less randomly, as previously believed, the availability of big genomic datasets are providing evidence that specific gene loss, for instance of genes linked to an auxotrophy, is probably a form of adaptive evolution (Albalat and Cañestro, 2016).

1.2. Nutrient acquisition tactics

Fungi are very diverse in the range of nutrients they can utilize, as well as in their methods of acquiring these nutrients. Based on how they feed themselves, fungi can be broadly categorized as saprotrophic, symbiotic or parasitic (Young, 1994). Saprotrophic fungi gather organic matter from dead material, such as forest mushrooms decomposing dead trees.

18 Prospecting a Predator | Klara Junker

Symbiotic fungi exchange nutrients with other species in a manner where both species benefit, such as fungi and plant roots interacting in soil mycorrhizal associations. Parasitic fungi, in contrast, derive nutrients from a host species at a cost for the host. Several plant pathogenic fungi enable their parasitism of host by means of special, and potentially conserved, invasion structures, called appresoria and haustoria (Kemen and Jones, 2012), see Figure 4. Fungi further described as biotrophic are dependent on a live host for survival, whereas necrotrophic fungi kill the host, or rather prey, and then feeds off it (Young, 1994).

Figure 4: Haustoria from fungal plant parasites (a-e) and the protozoan parasite P. falciparum (f) invading host tissue. Reprinted with permission from (Kemen and Jones, 2012).

Traditionally, species have been classified according to which type of growth media they can live off (Kurtzman, Fell and Boekhout, 2010). These days, clues to specialized rather than generalist ways of sustenance are often found in the genome of species. The filamentous fungi B. grameria is an obligate biotroph of grasses, including barley, and cause powdery mildew. The genome of Blumeria is relatively large genome, around ten times larger than the average ascomycete genome, but despite their large genome size, B. grameria lacks around 100 genes considered core genes in ascomycetes (Spanu et al., 2010). Apart from the previously mentioned lack of genes in the sulfur assimilation pathway, B. grameria also lacks several genes required for inorganic nitrogen assimilation. Spanu et al. concluded that the loss of so many essential genes was a trade-off enabled by Blumeria’s parasitic lifestyle.

1.3. Mycoparasitism

When the prey of a fungal parasite is another fungus, the fungal parasite is called a mycoparasite (Jeffries, 1995). Many necrotrophic mycoparasites are very aggressive

19 Klara Junker | Prospecting a Predator antagonists, have broad prey ranges and are able to attack other fungi at a distance by secreting toxins, antibiotics and lytic enzymes into the environment. So-called contact necrotrophs, on the other hand, mediate their attacks by direct physical contact between cells, and can use hypha to reach, contact and kill their prey. However contact necrotrophs do not necessarily invade the prey cell (Mims, Hanlin and Richardson, 2007). Invasive necrotrophs, on the other hand, physically penetrate their prey species cell wall with haustoria or penetration pegs. Unless a mycoparasite is an obligate parasite of a host species, it has traditionally been hard to determine if the main purpose of the parasitic fungi is to take up nutrients from their prey cells, to kill competitors, or both (Young, 1994; Jennings, 1995). Species in the Trichoderma genus are probably the most well studied mycoparasitic species, and just like all other well studied mycoparasites, they are filamentous fungi (Schmoll et al., 2016). Several yeast species, including S. cerevisiae, are able to secrete killer toxins, proteins that can kill eukaryotic target cells such as other yeasts, as a consequence of themselves being infected by special cytoplasmic-persisting mycoviruses (Schmitt and Breinig, 2006). The only unicellular fungal species appropriately described as necrotrophic mycoparasites are the yeasts in the Saccharomycopsis clade (Lachance and Pang, 1997).

1.4. Biocontrol fungi

The fact that fungi can fight each other can be of advantage for human purposes, mainly in agriculture, where fungal pests are common (Dean et al., 2012; Fisher et al., 2018). Many chemical pesticides have adverse effects on species and ecosystems well beyond their target pest, but biological control agents, biocontrol agents, usually have much narrower target ranges (Butt, Jackson and Magan, 2001; Nicolopoulou-Stamati et al., 2016). With inherent abilities of extracting nutrients from a wide range of environments, often in a specialized manner, fungi often make excellent biocontrol agents (Burge, 1988; Butt, Jackson and Magan, 2001). In addition, pests are more likely to develop resistance to synthetic chemical agents, than to live biocontrol agents. An important caveat with biocontrol agents is that their use often is associated with high costs and that they are not always consistent in their expected results. One reason for inconsistency lies in the fact that many biocontrol agents act by means of antagonism, i.e. their mode of action is based on passively trying to outcompete the fungal pest (El-Tarabily and Sivasithamparam, 2006).

20 Prospecting a Predator | Klara Junker

Mycoparasitic biocontrol fungi that instead directly kill their target fungi can be more efficient. Fungi in Trichoderma have the ability to be both saprotrophic and mycoparasitic, and its prey species include several plant-pathogenic fungi (Druzhinina et al., 2011). Interestingly, Trichoderma strains can even be directly beneficial to plants, by stimulating the plant’s growth and defence response. Genome sequencing of two Trichoderma species widely used as biocontrol agents, T. atroviridis and T. virens, has been central in understanding molecular mechanisms involved during mycoparasitic actions in the Trichoderma genus (Schmoll et al., 2016). For instance, how these species sense their prey cells has been extrapolated from genome data coupled with transcriptomic data on genes expressed before and during contact with prey cells. In particular, how mycoparasites such as Trichoderma degrade the cell wall of their fungal prey species has been the focus of several decades of research. A recent finding suggest horizontal gene transfer of cell wall degrading genes from plants have enabled their mycoparasitism (Druzhinina et al., 2018).

Figure 5: Graphic representation of how Trichoderma species sense, contact, wrap around and kill their prey species. Reprinted with permission from (Druzhinina et al., 2011).

21 Klara Junker | Prospecting a Predator

1.5. The fungal cell wall

Although there are exceptions, virtually all fungi have a characteristic cell wall, with complex and diverse polysaccharides chitin and β-glucans as core components (Gow, Latge and Munro, 2017; Richards, Leonard and Wideman, 2017), see Figure 6. The fungal cell walls present fungi with both opportunities and limitations. For instance, the rigid fungal cell wall allows growth in complex shapes, but it also prevents fungi from engulfing larger particles by means of phagocytosis. As a consequence of the latter, fungi need to first be able to digest nutritional material extracellularly, and then actively and efficiently transport the processed metabolites into their cell, in order to feed off large and structurally complex compounds.

Figure 6: Examples of the structure of fungal cell walls. Reprinted with permission from (Erwig and Gow, 2016).

To be able to cope with environmental changes, fungal cell walls need to be both physically robust and adaptable. Indeed, yeast cells are estimated to dedicate a whopping one-fifth of their genome to the ability to appropriately assemble and regulate their cell walls (Gow, Latge and Munro, 2017). Since many components of the cell wall are shared between fungal species, cell wall and membrane components are major targets for the four current antifungal drug classes (Fisher et al., 2018). Ergosterol in the fungal membrane, a fungal analogue to cholesterol, is targeted by two classes, whereas β-1,3 glucan of the fungal cell wall is targeted by a third class. The fourth class essentially targets fungal DNA synthesis.

22 Prospecting a Predator | Klara Junker

Proteins in fungal cell walls have functions such as enabling attachment to surfaces and maintaining the integrity of the cell wall (Gow, Latge and Munro, 2017). Proteins in the yapsin (YPS) gene family, a class of aspartic proteases, are involved in cell wall remodelling and cell wall integrity (Gagnon-Arsenault, Tremblay and Bourbonnais, 2006). Yapsins are restrictive in which substrates they cleave, are localized to the cell surface, and are only found in fungi. In S. cerevisiae, yapsins are typically glycosyl-phosphatidylinositol (GPI) anchored membrane proteins, whereas homologs in other species, such as C. albicans and Candida glabrata, are secreted aspartyl proteases (Sap) (Krysan, Ting and Abeijon, 2005). SAP genes in human pathogenic Candida species have been extensively studied and have numerous roles besides cell wall maintenance, including roles in causing host tissue damage, evading the host immune system and enabling dispersion through blood vessels (Rapala- Kozik et al., 2017). In Trichoderma, one aspartic protease gene and several subtilisin-like proteases are upregulated during mycoparasitism (Seidl et al., 2009). However, otherwise Trichoderma species mainly employ chitinases and glucanases when they degrade the cell wall of their prey cells. In addition, numerous copies of the genes encoding these cell wall degrading chitinases, glucanases and protases have been found in the genomes of Trichoderma genus mycoparasites (Schmoll et al., 2016).

1.6. Fungal parasite genomics

Recent advances in comparative genomics have revealed that all kinds of parasites, in orders ranging from fungal mycoparasites (Karlsson et al., 2017) to oomycetes (Kemen et al., 2011) to microsporidia (Pan et al., 2013), and even protozoan Trypanosomatids (Reis-Cunha and Bartholomeu, 2018), exhibit specific expansion of genes that enable parasitic behaviour. A proposed reason for this has been the potential for “sharpening of tools”, which gene duplication could give rise to (Pan et al., 2013). That theory suggests that if a gene is duplicated, the copied version could be diversified and optimized through selective mutagenesis for some purpose, while the original copy remains intact. For instance, one chitinase subtype in Trichoderma has been demonstrated to be able to distinguish between self and non-self (Gruber and Seidl-Seiboth, 2012).

Another typical hallmark in the genomes of parasites is distinct gene loss (Spanu, 2012). Substantial gene loss has been reported in plant pathogenic fungi and oomycetes (Spanu et al., 2010; Duplessis, 2011; Kemen et al., 2011). The evolution and mechanisms of gene loss

23 Klara Junker | Prospecting a Predator are not completely understood, but probably consist of pre-adaptation to a host, expansion and diversification of parasitic genes and finally gene loss, through sexual recombination or retrotransposon activity (Spanu, 2012). Despite some mechanistic uncertainty, gene loss is observable in most organisms and probably infers evolutionary fitness advantages, as it for instance could impart energy savings to not biosynthesise something that is readily available in a new environmental condition (Albalat and Cañestro, 2016).

1.7. Human fungal pathogens

Recent global estimates conclude that fungal diseases affect more than one billion people each year and kill over 1.6 million people (Bongomin et al., 2017). The mortality numbers are on par with deaths causes by tuberculosis and three times the number of deaths caused by malaria. However, whereas both tuberculosis and malaria are caused by a handful closely related species, pathogenic fungal species can be found throughout the entire fungal kingdom (James et al., 2006). The majority of fungal pathogens of both animals and plants, however, belong to the Ascomycota (Heitman, 2011). Seven of the ten of the most important fungal plant pathogens (Dean et al., 2012) and three of the four most important systemic human fungal pathogens, Aspergillus, Histoplasma and Candida, are Ascomycetes (Kim, 2016).

Of the several hundred species of fungi that can cause disease in humans, only very few are able to truly parasitize on humans (Köhler, Casadevall and Perfect, 2015). There are at least four extraordinary properties a fungus need, in order to be able to survive and thrive in or on a human; the ability to grow at or above body temperature, the ability to penetrate human tissue, the ability to live off the nutrients available in the human body and the ability to evade the human immune system. Because the human immune system is very efficient in fending off fungi, typically only immunocompromised patients ultimately die from fungal infections.

Fungal infections caused by filamentous fungi are often acquired as airborne fungal spores into the lungs, whereas transcutaneous fungal infection by yeasts, such as C. albicans, are typically transmitted from person-to-person (Kim, 2016). The majority of fungal infections are superficial skin infections without severe consequences, but invasive fungal infections can be life threatening, particularly to immunocompromised patients. Yet, public health mycology is so neglected that it is not even recognised as a discipline (Denning, 2017; Nat. Micro. Editorial, 2017).

24 Prospecting a Predator | Klara Junker

2. Yeasts

Saccharomycetes, of the Saccharomycotina subphylum, make up only one of the five fungal classes that contain yeasts, and it is by far the most successful and numerically abundant class (Nagy et al., 2014; Dujon and Louis, 2017). In moving from classifying yeasts based on single genes to making more thorough phylogenetic comparisons based on whole genomes, taxonomists have recently reassign new clades within the Saccharomycetes class (Kurtzman and Robnett, 2013; Shen et al., 2016). Ascoidea rubescens for instance has been singled out as a unique clade, with Saccharomycopsis yeasts as its closest relatives.

Compared to filamentous fungi that can grow and spread relatively long distances both through sold and the air, unicellular yeasts are considerably more immobile (Buzzini, Lachance and Yurkov, 2017). Still, since many yeasts species can be isolated from all over the world, they are somehow able to spread from their point of origin. S. cerevisiae, for example, is a ubiquitous yeast that was recently proposed to originate in China (Peter et al., 2018). Despite being employed more or less knowingly for millennia by humans who have enjoyed bread and beer, the complete life cycle of man’s favourite yeast, S. cerevisiae has not been fully understood until recently. The first evidence of how these yeasts are able to survive winter and reproduce sexually in nature came in 2012, when social wasps were demonstrated to be vectors of S. cerevisiae (Stefanini et al., 2012). An important reason for yeasts to produce the fragrant aromas we appreciate actually appears to be to attract insects as vectors, who in turn are not only able eat the sugar that the yeast signal is available, but also consume the yeast itself (Christiaens et al., 2014; Madden et al., 2018). Another example of yeast-insect interactions was the observation that the yeast Wicherhamomoyces amolalus was localized to the midgut and even the gonads of the mosquito Anopheles stephensi, a vector of malaria in Asia (Ricci et al., 2011). One recent study even proposed that angiosperms, i.e. flowering plants, only evolved well after yeast-insect interactions was thoroughly established, perhaps even as a consequence of those very interactions (Becher et al., 2018).

2.1. Saccharomyces cerevisiae: the model yeast organism

The very first eukaryote to have its genome fully sequenced was a laboratory strain of S. cerevisiae, a testament to the importance of this yeast as a model organism (Goffeau et al.,

25 Klara Junker | Prospecting a Predator

1996). Scientists have for decades appreciated how easily S. cerevisiae is cultured in the lab and how readily it is genetically manipulated. In fact, S. cerevisiae is so tractable when it comes to incorporation of foreign genetic material that at least two famous S. cerevisiae hybrids, the lager yeasts hybrids, harbour big genomic fragments from the cold-tolerant yeast S. eubayanus, as a result of spontaneous hybridization occurring at least 400 years ago (Libkind et al., 2011). Countless directed genetic manipulations of S. cerevisiae have subsequently laid solid foundations to our overall understanding of genes, genetics and genomics.

2.2. Human yeast pathogens in the Candida genus

Many yeast species in the polyphyletic genus of Candida yeasts meet the requirements needed for infecting warm-blooded animals, and are commonly found as commensals on mammalian mucus membranes (Köhler, Casadevall and Perfect, 2015). In fact, in contrast to most other fungal disease agents, many of the commensal Candida species are rarely found in the environment (Whittington, Gow and Hube, 2014). The first step of several important pathogenic Candida species, including C. glabrata, C. parapsilopsis, C. tropicalis and C. lusitaniae, is to colonize the skin of a new host (Silva et al., 2012; Papon et al., 2013). Candida species typically only becomes opportunistic or pathogenic when the balance between the yeast and host is disrupted, such as when the immune system of the host is impaired, when the use of antibiotics wipe out bacterial populations, which in turn frees up new niches for Candida, or when chemotherapy disrupts intestinal epithelium (Turner and Butler, 2014; Pappas et al., 2018). This can result in invasive candidiasis, i.e. overgrowth of a Candida species on skin, in mucosa, in internal organs, or in the blood (also known as candidaemia). A handful of Candida species are both colonizers and pathogens of humans, such as Candida glabarata and Candida parapsilosis, but the by far most common yeasts species in an on humans is C. albicans, estimated to colonize 50-70% of healthy humans.

2.3. Candida albicans: a commensal yeast in CTG clade

C. albicans is so well adapted to life in and on humans that it is considered a part of the normal flora (Whittington, Gow and Hube, 2014). C. albicans can endure a wide variety of human pH conditions, thrive off the variable nutrient conditions that make up the human gastrointestinal tract, vagina and skin, and by covering its dynamic cell wall with

26 Prospecting a Predator | Klara Junker glycoproteins it escapes the human immune system. Pathogenicity of C. albicans is characterized by invasive growth, mediated through morphological changes (Pappas et al., 2018). In normal conditions, C. albicans colonise and adhere to mucosal surfaces as a yeast, but during conditions that enable pathogenicity it switches to hyphal growth, and invades the underlying tissue. Similar to S. cerevisiae, research on C. albicans has been greatly aided by the ability to genetically modify this yeast, despite the fact that it was considered an obligate diploid (Hickman et al., 2013). One of the most well-studied key virulence factors in C. albicans are secreted aspartyl proteases (Sap), that for example can break down human proteins (Naglik, Challacombe and Hube, 2003). In parallel with classic methods based on genetic manipulation, transcriptomic analyses have successfully validated and identified novel genes involved in pathogenicity (Wilson et al., 2009; Chin et al., 2016). A summary of genetic and genomic insights into C. albicans has been made available in the public database “Candida Genome Database” (Skrzypek et al., 2017).

All organisms “read” DNA and subsequently translate the genetic message into a protein, by identifying the codon made up of three base pairs in a row. Each such codon is translated by a tRNA harbouring a matching anticodon into the appropriate amino acid. The “language” of each codon is virtually identical for all organisms. For instance, all organisms translate the universal start codon ATG to the amino acid methionine, by means of the tRNA with the anticodon UAC, tRNA(UAC)Met. However, 25 years ago the observation was made that C. albicans, and several other Candida yeasts, did not translate the codon CTG to leucine with a tRNA(CAG)Leu, according to the Standard Codon usage (Ohama et al., 1993). Instead, C. albicans translate the CTG codon to serine, according to the Alternative Yeast Nuclear Code (AYNC) and the way C. albicans does so is through a modified tRNA with a CAG anticodon, a tRNA(CAG)Ser. Interestingly, some 3% of CTG codons still gets “mistranslated” to leucine and one hypothesis for this differential translation is that the proportion of “mistranslation” might be related to virulence and pathogenicity in C. albicans, by slightly altering the hydrophobicity of surface molecules (Miranda et al., 2013). Another recent hypothesis is that the use of tRNA(CAG)Leu was drastically abandoned after an ancient virus infection (Krassowski et al., 2018). The proposed virus would in this theory specifically target tRNA(CAG)Leu genes and confer toxicity to carriers of normal tRNA(CAG)Leu genes which would prompt complete loss, or heavy use of introns, in these tRNA(CAG)Leu genes, to mask them from the virus.

27 Klara Junker | Prospecting a Predator

Figure 7: The subdivision of the CTG clade into a Ser1 and a Ser2 clade. Reprinted with permission from (Krassowski et al., 2018) and the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/)

2.4. Candida auris: an emerging multidrug resistant pathogen

As recent as in 2009, a new Candida species, C. auris, was discovered and it has since rapidly spread rapidly around the world (Satoh et al., 2009; Lockhart, Berkow, et al., 2017). Spread of C. auris is mainly contact-mediated and its transmission frequently occur in health care settings. C. auris can frequently become a blood borne infection in immune- compromised patients with central venous catheters, and one study found that 59% of patients infected with C. auris died after an average on 19 days post-admission (Lockhart, Etienne, et al., 2017). On top of this, C. auris is hard to diagnose and, importantly, highly resistant to antifungals. Several clinical isolates are even multi-drug resistant. Any measures

28 Prospecting a Predator | Klara Junker to control the emerging global threat of Candida auris are therefore highly sought after. Genetic manipulation of C. auris has currently not been achieved.

2.5. Saccharomycopsis yeasts: necrotrophic mycoparasites

Saccharomycopsis yeasts are cosmopolitan yeasts, isolated from diverse habitats across the world, such as on bees, beetles, fruits, fermented foods and oak exudate and flux (Lachance et al., 2000; Kurtzman and Smith, 2011a). The first observation that a yeast in this clade was able to invade and parasitize on other yeast were done in 1973, but at that point it was not possible to determine if the prey cells were alive or already dead (Kreger-van Rij and Veenhuis, 1973). It took until 1997, for the predatory capabilities of Saccharomycopsis yeasts to be rediscovered (Lachance and Pang, 1997). A mix-up of strains during mating studies for taxonomic purposes, led Lachance and Pang to observe that Saccharomycopsis fodiens attacked and killed other yeasts in co-culture. Soon after, Lachance et al. made the observations that all Saccharomycopsis yeasts shared a rare auxotrophy for organic sulfur compunds, that virtually all Saccharomycopsis species studied were able to predate on a wide range of prey species, and that methionine was a potential trigger for predatory behaviour (Lachance et al., 2000). In addition, Lachance et al. were able to visualize how Saccharomycopsis species use narrow haustoria-like penetration structures, during host invasion, see Figure 8.

Figure 8: Scanning electron micrograph of the dimorphic yeast Saccharomycopsis javanensis preying upon S. cerevisiae. a) arrows indicate penetration pegs invading prey cells. b) arrows indicate collapsed prey cells. Scale bar 5μm. Reprinted with permission from (Lachance et al., 2000).

29 Klara Junker | Prospecting a Predator

Saccharomycopsis yeasts were first described and defined based on morhoplogy, and was initially assigned to the genus Endomyces Reess, on the basis of presence of true hypha, and then subassigned to Endomycopsis Dekker, on the basis of being a budding yeast (Kurtzman and Smith, 2011b). From the 1930s, many species were further subdivided on basis of physiological features, such as abilities to ferment sugars (von Arx and Yarrow, 1984). After a few decades of improved, but still rather confusing, taxonomic studies, genetic tools paved the way to more accurate phylogenetic family trees, placing Saccharomycopsis yeasts close to Ascoidea rubescens, another little-studied yeast (Kurtzman and Robnett, 2013).

S. fibuligera, isolated from rye bread in Germany and wheat-based alcoholic cultures in Korea, was the first yeast in the Saccharomycopsis clade to have its genome sequence published (Choo et al., 2016). S. fibuligera is predominantly diploid species and attracts biotechnological interest, because of its production of amylases, acid proteases, trehalose and β-glucosidase (Chi et al., 2009). The only genetic manipulation of a yeast in the Saccharomycopsis clade is a reported disruption of a single acid protease gene in a mutant strain of S. fibuligera, that was pre-mutagenised with ethylmethanesulfonate (Chi et al., 2003; Wang et al., 2011) The gene disruption was performed in the context of increased amylase production and was not tested in relation to any predatory behaviour. However, only the promoter and terminator of one reported acid protease gene was targeted and no evidence was provided on whether the promoter and terminator sequences represented unique sequences, or if the coding sequence was intact, i.e. if the resistance marker was integrated ectopically or had indeed disrupted the intended gene. In addition, while Wang et al. were under the impression in 2011 that S. fibuligera only harboured one acid protease gene, subsequent genomic analysis revealed the presence of several acid protease genes, raising additional questions regarding the identity of the reported disrupted gene (Wang et al., 2011; Choo et al., 2016).

Because of the ability of Saccharomycopsis yeasts to predate on other fungi, S. schoenii was selected for the first trial in using Saccharomycopsis yeasts as biocontrol agents (Pimenta et al., 2008). In the trial, S. schoenii was inoculated on oranges infected with spores of the filamentous fungi Penicillium digitatum, P. expansum and P. italicum. All Penicillum species were susceptible to attack and co-inoculation with S. schoenii reduced disease severity on the oranges. In addition, S. schoenii itself did not produce any decay or necrosis symptoms on the oranges. Similarly, S. fibuligera was recently used as a biocontrol of the

30 Prospecting a Predator | Klara Junker toxigenic filamentous fungi Penicillium nordicum and Aspergillus ochraceus on speck. (Iacumin et al., 2017) Co-culture with S. fibuligera resulted in virtual elimination of both the mould and Ochratoxin A. No Saccharomycopsis yeasts, or any other fungi, have, to our knowledge, been prospected for use against clinically relevant fungi, such as Candida yeasts.

31 Klara Junker | Prospecting a Predator

Materials & methods

Fungi has traditionally been described and classified based on morphology and phenotypic characteristics as well as on physiological and nutritional limitations (Zhang, Luo and Bhattacharya, 2017). These fundamental observational studies were greatly aided by developments in light and electron microscopy technologies, but when it comes to assigning interfungal relationships, they remain limited and could introduce bias (Stajich, 2017). In the last decades, DNA based methods such as PCR and gene sequencing have enabled genetic insights that have also been fundamental for modern fungal biology research. Recent developments in high throughput genomic sequencing has stepped up the game of describing yeast and fungi and their phylogenetic family tree even further. Because the currently rapidly developing sequencing techniques are becoming more accurate and require less input material, many fungal researchers today believe that the we could soon go from currently 135 000 described species, to have the genetic blueprints of close the globally estimated 1.5- 3 million fungal species (Hawksworth, 2012; Grube et al., 2017). In parallel, recent developments in microscopic methods such as transmission electron microscopy, fluorescence and live microscopy have enabled important biological studies such as characterizing ultrastructures of fungal cells and dynamic interactions between fungal cells and other cells (Douglas B. Murphy and Michael W. Davidson, 2013; Lewis et al., 2013). Current challenges in fungal research are recognised as the ability of integrating big data from several contemporary methods, such as interpreting biologically relevant results based om both genomic, transcriptomic, proteomic and imaging data (Grube et al., 2017; Haas et al., 2017). The work presented in this thesis is an example of such a challenge.

1. Strains & culture conditions

Strains used in this thesis are listed in Table 1. Media used is listed in Table 2. Prior to all studies, all cells were cultured overnight in YPD, rotating, at 30°C, then resuspended to an optical density (OD) of 0.3 at 600 nm and allowed to grow to log phase. For starvation and predation studies, cells were cultured on YPD, CSM, CSM-Met or SD (see Table 2), solidified with 20 g/L of bacto-agar prior to extraction of DNA, RNA or proteins or solidified with 10 g/L of agarose prior imaging.

32 Prospecting a Predator | Klara Junker Table 1: List of strains used in this thesis

33 Klara Junker | Prospecting a Predator

Table 2: Media used in this thesis

Media Ingredients

“Standard” 10 g/L yeast extract YPD: 20 g/L bacto peptone 20 g/L glucose

“Nutrient 0.79 g/L Complete Supplement Mixture (CSM) Limited” 6.7 g/L YNB w/o amino acids with ammonium sulfate CSM: 20 g/L glucose

“Methionine 0.75 g/L Complete Supplement Mixture -Methionine (CSM-Met) Deprived” 6.7 g/L YNB w/o amino acids with ammonium sulfate CSM-Met: 20 g/L glucose

“Starvation” 6.7 g/L YNB w/o amino acids with ammonium sulfate SD: 20 g/L glucose

YPD is a rich standard media, providing nutrients in the form of various complex molecules. CSM is made up of only the amino acids, vitamins, trace elements and salts that are essential for prototrophic S. cerevisiae. CSM-Met is CSM without methionine, through which any specific responses due to a methionine deprivation can be studied. SD media is a yeast minimal media, that only contain ammonium sulfate as a nitrogen source, and vitamins, trace elements and salts, which necessitates de novo synthesis of all amino acids.

2. Live cell microscopy

Differential interference contrast (DIC) and fluorescent microscopy was performed with a Zeiss Axio Imager M2 Microscope, using a halogen lamp for transmitted-light and UV for fluorescent imaging, and the software Metamorph for image acquisition. Microscopy was also performed using the PerkinElmer UltraVIEW VoX Spinning Disk Confocal Microscope at the University of Aberdeen, UK, and the software Volocity® for image acquisition. To stain cell walls and septa between cells, Calcofluor White (CW, final concentration 10 µg/mL) was used, and to stain nucleic acids of cells with a compromised cell membrane, i.e. dead or dying cells (Sasaki, Dumas and Engleman, 1987; Przybylo et al., 2008), Propidium Iodine (PI, final concentration 1 µg/mL) was used. Prior to live cell imaging and kill curve image collection, cells were seeded on pads of media solidified with 1% agarose. Movies were generated by capturing images at regular intervals for up to 3h. Images for movies were

34 Prospecting a Predator | Klara Junker autofocused on the Spinning Disk Confocal Microscope using the Nikon Perfect Focus System. Kill curve analyses were performed by an initial seeding of cells on several pads with appropriate media, taking out new slides from humid storage conditions at every time point to capture at least three frames, representative of the whole slide. Locations where cells were growing on top of each other were selected against, in favour of locations where individual cells could be distinguished.

Figure 9: Morphology based scoring of attacked and killed prey cells. Top panel illustrate the vacuolarization of attacked C. auris cells (arrows). Bottom panel illustrate killed cells, with a shrunken or flattened morphology, here also stained with PI.

The software FIJI/ImageJ(Eliceiri et al., 2012; Schindelin et al., 2012, 2015) was used for image processing and analysis. For movies, any drift in frames was corrected using either the plugin Image Stabiliser (Li, 2008) and/or the macro NMS fixTranslation v1 (Schneider, 2014). Prey cells were scored as predated if they were in contact with S. schoenii cells and either vacuolarized, shrunken or flattened, see Figure 9. For kill curve analyses, individual cells were counted using the plugin Cell counter and, the plugin Multi Stack Montage enabled simultaneous visualization of DIC, fluorescent channels and overlays of channels as montages. When fluorescent images were used, background noise was reduced and signal enhanced by batch adjustment of brightness/contrast

35 Klara Junker | Prospecting a Predator

3. Genome sequencing & assembly

Genome sequencing, assembly and annotation of three Saccharomycopsis species, S. fodiens, S. fermentans and S. schoenii was carried out using Illumina MiSeq paired-end and mate-pair read sequencing as described in Paper I and II. In short, DNA extraction and sequencing was performed by LGC Genomics (Berlin, Germany). Raw reads from both a 250-bp paired-end library and an 8-kb mate-pair library were quality processed and trimmed to high-quality reads. Assemblies of quality-controlled reads were performed using Bowtie version 2.1.0, and the resulting contigs were refined by further scaffolding. The draft genome assemblies based on Illumina sequencing is summarized in Table 2.

To improve the genome assembly of S. schoenii, additional PacBio sequencing was performed, as outlined in Figure 10. Genomic DNA was prepared using the QIAGEN Blood & Cell Culture DNA Maxi Kit with a QIAGEN Genomic-tip 500/G (QIAGEN Gmbh, Hilden, Germany) according to the manufacturer’s protocol. PacBio sequencing, based on Single Molecule Real-Time (SMRT) technology, was performed by DNA Link Inc. (Seoul, Republic of Korea), using kits and reagents from Pacific Biociences. 5 µg of quality controlled genomic DNA was used to prepare SMRTbell library. Fragments <20 kb were removed using the Blue Pippin Size selection system for large-insert library. The constructed library was validated using the Agilent 2100 Bioanalyzer. A sequencing primer was annealed to the SMRTbell template and DNA polymerase was bound to the complex using DNA/Polymerase Binding kit P6. The resulting polymerase-SMRTbell-adaptor complex was then loaded into 4 SMRT cells and sequenced using C4 chemistry (DNA sequencing Reagent 4.0). 240 minute movies were captured for each SMRT cell using the PacBio RS II sequencing platform. Four raw sequencing subreads were generated, one per SMRT cell.

Raw PacBio sequencing subreads in h5 format were filtered on quality and length, using the RS_subreads protocol in PacBio’s SMRT-Portal software, run through Amazon Web Services Inc (Seattle, USA) and exported as fastq files. Several de novo assembly of the filtered reads was performed using the CLC genomics workbench v.9.5 (QIAGEN Aarhus, Aarhus C, Denmark), and the best assembly was chosen based on highest total length of assembly and lowest number of contigs, using QUAST (Gurevich et al., 2013). For the best assembly, filtered subreads were batch error-corrected using the tool "Correct PacBio Reads (beta)" with default parameters, and de novo assembled using the tool “De Novo Assemble

36 Prospecting a Predator | Klara Junker

PacBio Reads (beta)” with default parameters, generating 78 contigs. These 78 contigs were then further polished using Paired-End and Mate-Pair reads from previous Illumina sequencing, generating a final draft genome assembly of 29 contigs, at 14.3 mega base pairs. Draft genome comparisons of the Illumina read based and Illumina + PacBio read based assemblies are summarized in Table 3.

Figure 10: Workflow of Genome sequencing and assembly of S. schoenii. White box: wet lab input generated by author. Purple boxes: high throughput sequencing performed by LGC genomics (Berlin, Germany) and DNA link (Seoul, Republic of Korea) respectively. Blue boxes: bioinformatics analyses performed by author.

4. Genome annotation

Draft S. fodiens, S. fermentans and S. schoenii genomes based on Illumina sequence reads only were annotated as described in Paper I and Paper II. In short, Open reading frames (ORFs) were determined as sequences with an AUG start codon, spanning >300 bp. Non- overlapping ORFs were first compared to S. cerevisiae genes, translated with the standard codon usage, using blastx. ORFs with no hits against S. cerevisiae genes were subjected to blastx against the non-redundant (nr) database at NCBI, with an e-value cut off at <10-10.

37 Klara Junker | Prospecting a Predator

The workflow for annotating the S. schoenii (Illumina + PacBio) draft genome is summarized in Figure 11. ORFs were predicted as sequences with an AUG start codon, spanning >300 bp. For an unbiased overview of related fungal species, ORFs translated with either standard or alternative codon usage were subjected to a cloud based blastx against all proteins in the non-redundant protein database (nr) deposited by 06.06.2017, using the plugin Blast2GO (Götz et al., 2008) in the software CLC genomics 9.5. To functionally annotate the S. schoenii genes, blastx of ORFs translated with either standard or alternative codon usage was performed against either S. cerevisiae alone, against C. albicans alone or against both S. cerevisiae and C. albicans with a bit score cut off at >55. Draft genome annotation comparisons of all draft genomes are summarized in Table 3. tRNA genes were identified in all Saccaromycopsis genomes using the web-based tool tRNAscan-SE (Lowe and Chan, 2016). In short, fasta sequences (<5,000,000 nucleotides in length) of the entire draft genomes were used as input, with predicted tRNA genes and relevant descriptions, such as their anti-codon sequence and an image of their predicted 2D structure, as output.

Figure 11: Workflow of in silico annotation of S. schoenii draft genome (PacBio + Illumina). Blue box: input draft genome assembly, generated by author. Green boxes: in silico annotations performed by author. Red box: draft genome annotation, generated by author.

38 Prospecting a Predator | Klara Junker

5. Proteomic analyses

A proteomic analysis was performed, to confirm whether S. schoenii translate the CTG codon to leucine or serine, and to analyse the relative abundance of proteins expressed during starvation and predation conditions, as outlined in Figure 12 and described in detail in Paper III. In summary, S. schoenii cells were cultured on YPD media alone, on SD media alone or on SD media with equal numbers of S. cerevisiae cells. After three hours, cells were pelleted and flash frozen in liquid nitrogen. Subsequent proteomic analysis was performed by Phylogene (Bernis, France). Proteins were extracted, purified and measured for concentration. Peptides were generated by trypsin digestion, measured for concentration and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Peptides were identified against two databases; all S. schoenii ORFs translated with Standard Code or Alternative Yeast Nuclear Code respectively. When S. schoenii and S. cerevisiae were co- cultured, peptides were also mapped against a database containing reference proteome of S. cerevisiae mined from UNIPROT.

Figure 12: Workflow of proteomic analysis of S. schoenii Blue box: draft genome assembly, by author. Green boxes: in silico annotations performed by author. White box: wet lab input by author. Yellow boxes: proteomic analyses performed by Phylogene (Bernis, France).

39 Klara Junker | Prospecting a Predator

6. Transcriptomic analyses

To analyse the genome wide gene expression during starvation, predation and mating conditions, a transcriptomic analysis was performed, as outlined in Figure 13. S. schoenii cells were cultured for 3 hours on YPD, CSM, CSM-Met or SD media, either alone or in co- culture with equal numbers of S. cerevisiae as prey cells. In total, 8 different samples were prepared, and all experiments were performed in three biological replicates, see Table 1. At the appropriate time point, cells were flushed off the solid media and washed using sterilized deionised water. Cell pellets were flash frozen in liquid nitrogen. RNA was extracted using the Ambion Ribo-pure Yeast kit (Thermo Fisher Scientific) and DNAse treated. BGI Europe (Copenhagen, Denmark) performed subsequent quality control (RIN ≥ 6.5, 28 S/18 S: ≥ 1.0, OD 260/280 ≥ 1.8, OD 260/230 ≥ 1.8), constructed strand-specific Transcriptomic libraries, and sequenced the samples at 10MB clean reads/sample using Illumina Hiseq4000 PE100.

Figure 13: Workflow of genome-wide transcriptomic analysis of S. schoenii. White box: wet lab input generated by author. Red box: draft genome annotation, generated by author. Purple box: BGI Europe (Copengagen, Denmark) performed RNA sequencing. Orange boxes: bioinformatics analyses performed by author.

Transcriptomic analyses were performed using CLC genomics workbench 9.5. The tool RNAseq was used, mapping reads to genes previously predicted and annotated in silico (as

40 Prospecting a Predator | Klara Junker described above) in the S. schoenii Illumina or PacBio generated draft genomes, as well as mapping reads to inter-genic regions. Reads were aligned with default parameters; a mismatch cost 2, insertion cost 3, deletion cost 3, length fraction 0.8, similarity fraction 0.8, no global alignment, with auto-detected paired distances, strand specific to both strands, and a maximum of 10 hits per read. Paired reads were counted as one, the expression value was total counts, and EM estimation was used. Original expression values were normalized by totals reads, using the tool Normalize, as Reads per 1 000 000, with normalization value Mean, Median mean as reference and trimming percentage set to 5. Fold changes were expressed as the differential gene expression in the different samples against gene expression of S. schoenii cells cultured alone on YPD. In this study, genes with a total range of normalized values below 100 were discarded as noise.

Table 1: Samples used in transcriptomic analysis. Three biological replicates per sample were analysed. Cells were cultured for three hours prior to analysis. Sample Media Cells

1 x3 YPD 1x108 S. schoenii

2 x3 YPD 1x108 S. schoenii +1x108 S. cerevisiae

3 x3 CSM 1x108 S. schoenii

4 x3 CSM 1x108 S. schoenii +1x108 S. cerevisiae

5 x3 CSM-Met 1x108 S. schoenii

6 x3 CSM-Met 1x108 S. schoenii +1x108 S. cerevisiae

7 x3 SD 1x108 S. schoenii

8 x3 SD 1x108 S. schoenii +1x108 S. cerevisiae

7. GO term analysis

To be able to couple and extract biologically meaningful data from different sets of genes and proteins, the Web Tool FungiFun2 was used for gene ontology (GO) term analysis (Priebe et al., 2015). In contrast to similar GO tools, such as Gorilla (Eden et al., 2009) and the Candida Genome Database GO Term Finder (Skrzypek et al., 2017), FungiFun2 extracts category associations and annotations from 240 fungal species and outputs relevant GO

41 Klara Junker | Prospecting a Predator terms pertaining to both Molecular Functions, Biological Processes and Cellular Components into one single list of statistically enriched GO term categories. Since C. albicans was the closest related yeast with a functionally well characterized genome, we used C. albicans gene annotation nomenclature of S. schoenii genes and all S. schoenii genes with a C. albicans homolog as a background list. Significance of over-representation (enrichment) of direct GO terms was calculated using a Hypergeometric distribution test and adjusted with a Benjamini-Hochberg procedure (FDR correction).

8. Genetic manipulations of S. schoenii

To be able to prove the role of certain genes in predation-specific pathways in S. schoenii, several attempts at targeted gene disruption, based on homologous recombination, was performed. Constructions of genetic markers conferring resistance to antibiotics, such as Geneticin (G418) to which S. schoenii is sensitive to, were flanked by genes of interest and amplified in plasmids. Subsequent to learning that S. schoenii might belong to the CTG clade, the kan-ORF marker was codon optimized, and the S. schoenii ACT1 promoter was used. Purified and appropriately digested gene constructs were delivered into S. schoenii cells by electroporation or heat shock at various temperatures. Numerous attempts at targeted gene disruption, based on the variables above, were attempted, but only a few cases of ectopic integration of the selection marker was achieved at best, and we did not achieve targeted gene disruption in S. schoenii.

42 Prospecting a Predator | Klara Junker

Summary of papers

Our aim with Paper I and II was to provide draft genomes of the mycoparasitic yeasts Saccharomycopsis fodiens and Saccharomycopsis fermentans, which could enable further studies and genome comparisons. To this end, we sequenced 250-bp paired end and 8-kb mate pair libraries from both genomes using an Illumina MiSeq platform and assembled genome scaffolds using Bowtie 2. We identified and functionally annotated non-overlapping open reading frames (ORFs) and subjected these to a blastx against S. cerevisiae proteins. In contrast to virtually all other microorganisms, Saccharomycopsis yeasts are unable to assimilate sulfate, making Saccharomycopsis yeasts auxotrophic for organic sulphur. By mining the annotated genomes, we also identified the genomic basis for the observed sulfate auxotrophy in both S. fodiens and S. fermentans as a lack of all genes in the sulfate assimilation pathway.

In Paper III our aim was to identify which genes are used when S. schoenii attack and kill S. cerevisiae, a model prey yeast species, to enable further prospecting of S. schoenii as a biocontrol yeast. To that end, we generated the first draft genome of S. schoenii, set up a quantitative image based predation assay and integrated predatory behaviour and genomic insights with transcriptomic and proteomic analyses.

To generate a draft genome with both high and deep coverage, we de novo sequenced the S. schoenii genome using short, medium, and long read libraries. We assembled scaffolds with long reads and polished these scaffolds with the short and medium reads. To predict genes, we identified all ORFs and subjected all translated ORFs to a number of blastx strategies. To functionally annotate the S. schoenii genome, we subjected all ORFs to blastx against translated genes from S. cerevisiae and C. albicans, and found that S. schoenii had more genes matching C. albicans than S. cerevisiae.

Just as we had found in Paper I and Paper II, we noted a complete lack of genes in the sulfate assimilation pathway. We also noted expansions of genes encoding for instance chitinases and yapsins/secreted aspartic proteases. In addition, we performed a cloud based blastx against all proteins deposited in the non-redundant (nr) protein database. This verified

43 Klara Junker | Prospecting a Predator a close relationship between S. schoenii, Ascoidea rubescens and Wickerhamomyces anomalus and revealed that many other yeast species related to S. schoenii were CTG clade yeasts. Virtually all organisms, including S. cerevisiae, translate the CTG codon to leucine, using a tRNALeu(CAG) gene, but yeasts in the CTG clade, such as C. albicans, translate the CTG to serine, instead of leucine, enabled by a special tRNASer(CAG) gene. In addition, we identified two distinct CAG-tRNA genes in the S. schoenii genome, with one being very similar to the C. albicans tRNASer(CAG), suggesting S. schoenii would translate CTG to serine. To find out if S. schoenii translated CTG to serine, we performed a proteomic analysis and found that >99 % of S. schoenii genes harbouring at least one CTG codon had their CTG codons translated to serine.

To study predatory activity, we set up two microscopy based assays in Paper III. In a qualitative live microscopy study, we observed that S. schoenii rapidly attacked and killed S. cerevisiae upon physical contact on starvation media, by causing prey cells to vacuolarize and shrink. Based on the finding that S. schoenii need to acquire sulphur in its organic form, we hypothesised that the presence or absence of methionine, one of only two sulphur containing amino acids, might impact the predatory propensity of S. schoenii to attack prey cells. To that end, we set up a quantitative microscopy-based predation assay, where S. schoenii was subjected to gradual nutrient-limited conditions, with or without S. cerevisiae available as prey. By enumerating and scoring the viability of S. cerevisiae cells, we found that general nutrient limitations, not primarily methionine deprivation, triggered S. schoenii to attack and kill prey cells.

To mine what genes S. schoenii exclusively employ in its predatory behaviour, we performed a genome-wide transcriptomic analysis, with samples mirroring the samples in the quantitative microscopic predation assay in Paper III. We found that even though a specific lack of methionine did not significantly affect the efficiency of S. schoenii to attack prey cells, S. schoenii nevertheless responded dramatically to an absence of methionine in the media. Transcriptional responses to methionine deprivation included heavy upregulation of genes in methionine scavenging and salvaging pathways. Under predation conditions, genes encoding yapsins and secreted proteases were highly upregulated. We were also able to confirm that yapsins and secreted proteases were also translated into proteins during predation conditions.

44 Prospecting a Predator | Klara Junker

The main limitation with the study is that it proves correlation, not causation. To infer causation, further studies where specific genes or pathways are disrupted are warranted. In conclusion, in Paper III, we provide an annotated draft genome of S. schoenii, a microscopy based assay for predatory activity, a proteomic analysis confirming translation of the CTG codon to serine and proteomic and transcriptomic analysis of starvation and predation specific gene expression and translation.

Fungal infections have increased dramatically in the last decades, and is currently causing more global mortality than malaria. However, as opposed to malaria, a much wider range of different species of yeast and fungi are causing the numerous types of fungal infections. In parallel to antibiotic resistance in bacteria, several fungi are becoming drug resistant, but the development of novel antifungal agents is slow. In Paper IV, our aim was to evaluate S. schoenii as a biocontrol agent against clinically relevant Candida species, with focus on the recently emerged multi-drug resistant pathogen Candida auris.

Using the quantitative microscope assay developed in Paper III, we co-cultured S. schoenii with clinical isolates of C. auris, with variable drug resistance properties, as well as with clinical isolates of pathogenic Candida species; C. albicans, C. glabrata, C. tropicalis, C. lusitaniae and C. parapsilosis in Paper IV. We included S. cerevisiae as a control, and demonstrated the broad host range of S. schoenii by including the distantly related yeast Schizosaccharomyces pombe.

In Paper IV we found that C. auris, irrespective of drug resistance profile, was susceptible to predation and could be killed by S. schoenii. The other pathogenic clinical Candida strains, as well as S. pombe, could also be attacked and killed. These results open up the possibility of further exploitation of using S. schoenii as a biocontrol agent against multi- drug resistant fungi in clinical settings.

45 Klara Junker | Prospecting a Predator

Results & Discussion

The aim of this thesis was to identify the yeast-killing genetic tools employed by S. schoenii, as it attacked and killed other yeasts. My initial work aimed at generating a protocol for targeted gene disruption, in order to be able to infer involvement of certain genetic pathways in the predatory behaviour of S. schoenii, and were based on traditional genetic methods. To guide genetic manipulation, we de novo sequenced, assembled draft genomes for S. schoenii, S. fermentans and S. fodiens using Illumina MiSeq paired-end and mate-pair sequencing reads, and annotated genes based on homology to S. cerevisiae proteins (see Table 3). Subsequent attempts at genetic manipulation of Saccharomycopsis yeast, however, was not straight forward or successful.

Initial genomic insights In draft genomes of S. schoenii, as well as in the draft genomes of S. fodiens and S. fermentans, we were able to identify a complete lack of all genes in the sulfate assimilation pathway as the genomic background to the rare sulfate assimilation deficiency of Saccharomycopsis yeasts. The sulfate assimilation deficiency is unique among yeasts, but shared by a few parasitic filamentous fungi (Spanu, 2012), suggesting a convergent gene loss evolution correlated with and perhaps a result of parasitic abilities in Saccharomycopsis. In further genomic analyses, we identified a potential tRNA (CAG)Ser gene in the genomes of S. fodiens, S. fermentans and S. schoenii, indicating these species belonged to the CTG clade. That hypothesis led us to codon optimize our resistance markers, to allow for an alternative codon usage. When those adjustments did not deliver any major developments in our gene manipulation attempts, we found ourselves at a cross road. Either we could continue to explore additional gene disruption protocols, based on for instance CRISPR-Cas9, or we could adopt new approaches to identify the genetic tools involved in predation.

Next-Generation Sequencing approaches We chose to take advantage of the recent advances of increasingly affordable and precise next-generation sequencing and omic methods. Transciptomic and proteomic approaches might be argued to just yield correlations, not the gold-standard causations, typically derived by means of genetic manipulation. However, such approaches have been useful in not just

46 Prospecting a Predator | Klara Junker validating previous findings, but also in providing profound non-biased and non-targeted insights in understanding host pathogens interactions and C. albicans virulence (Chin et al., 2016). To get a better foundation for subsequent transcriptomic and proteomic analyses, I began by improving our first draft genome of S. schoenii. Short reads, such as those

a) Identified

ion (table 12). d) Manual 6.17. g) ORFs, any blastx non-overlapping

. c) Using alternative translat draft genome assemblies and annotations.genome assemblies and draft

<55

score

bit S. schoenii

hit

and

blastx

astx hit score >55. f) deposited by 06.0 ORFs

S. fermentans

, overlapping S. fodiens -

non

h)

>55

score 2: Comparison of 2: Comparison

bit

Table with tRNA Scan-SE. b) ORF = min 300bp/100aa, ATG start site genes. e) curation of non-overlapping Bl hit

47 Klara Junker | Prospecting a Predator generated by Illumina sequencing, are rarely able to appropriately assemble repetitive regions, such as telomeres. To get a better coverage of such repetitive regions, we de novo sequenced the S. schoenii genome with long PacBio sequencing reads (English et al., 2012). I then assembled a new S. schoenii draft genome by generating scaffolds from the long PacBio reads, and polished the scaffolds with the short, but more accurate, Illumina reads, see Table 3.

Functional genome annotations To functionally annotate the S. schoenii, S. fodiens and S. fermentans ORFs, we subjected all translated ORFs to three blastx strategies; against all proteins from S. cerevisiae, against all proteins from C. albicans, or against proteins from both S. cerevisiae and C. albicans, see Table 3. With this approach, we could identify functional homologs to each putative S. schoenii gene and manually remove overlapping ORFs. In addition, we could observe that some putative genes had only a S. cerevisiae or only a C. albicans homolog. We found that Saccharomycopsis harbours several genes homologous to aspartic protease genes, and that some were closest to the S. cerevisiae yapsin gene YPS3 whereas others were closest to the C. albicans secreted aspartic protease gene SAP6. A blastx of all non-overlapping ORFs against S. cerevisiae proteins only yielded 19 YPS3 genes in both S. schoenii genomes, 23 YPS3 genes in the S. fermentans genome and 25 YPS3 genes in the S. fodiens genome. However, when I blasted all non-overlapping translated S. schoenii ORFs against both S. cerevisiae and C. albicans proteins, 11 of these aspartic protease genes were in closest homology to S. cerevisiae YPS3 while five were closest to C. albicans SAP1. This discrepancy suggest that the majority S. schoenii aspartic protease genes are more similar to those of S. cerevisiae that remodel fungal cell walls (Gagnon-Arsenault, Tremblay and Bourbonnais, 2006) than the secreted aspartic proteases of C. albicans, that target human cells (Naglik, Challacombe and Hube, 2003).

Cloud-based blastx phylogeny analysis To get an unbiased overview of closely related fungi, I blasted all identified ORFs in the draft genome of S. schoenii against all fungal proteins in the nr database, using a cloud-based blast approach, to facilitate an otherwise computationally too demanding task (Angiuoli et al., 2011), see Table 3. With this approach, I discovered a close relationship between S. schoenii and A. rubescens, W. anomalus and P. tannophilus, as well as several CTG clade yeasts, see Paper III. In 2017, P. tannophilus, but not A. rubescens, had been demonstrated

48 Prospecting a Predator | Klara Junker to translate CTG alternatively (Riley et al., 2016). Interestingly, apart from the potential tRNA(CAG)Ser we had also identified a tRNA(CAG)Leu gene with an intron in the S. schoenii and S. fermentans genomes. To resolve the identities of these tRNA genes, I made alignments that indicated that while the tRNA(CAG)Ser genes indeed shared several features with other tRNA(CAG)Ser and tRNASer genes, the tRNA(CAG)Leu gene was barely homologous to any tRNALeu gene. While the tRNA(CAG)Leu genes aligned fairly close to the P. tannophilus tRNA(CAG)Ala, significant differences indicated that the tRNA(CAG)Leu genes from S. schoenii and S. fermentans would not translate to alanine, see Paper III.

The S. schoenii CTG codon is predominantly translated to serine To determine if S. schoenii indeed is able to translate the CTG codon to serine, as predicted by the presence of a tRNA(CAG)Ser gene, or if it translates the CTG codon to leucine, as predicted by the presence of a tRNA(CAG)Leu gene, or even both, as very recently demonstrated in Ascoidea asiatica (Mühlhausen et al., 2018), I set up a proteomic analysis. I snap froze S. schoenii cells that I had cultured for three hours on standard media (YPD) and sent them to Phylogene for subsequent analysis, which included proteins purification, digestion and quantification, peptide identification, quantification and mapping of peptides against either one of two databases I had prepared. In the first database, all S. schoenii ORFs were translated according to the Standard Code, i.e. CTG codon to leucine, and in the second database all S. schoenii ORFs were translated according to the Alternative Yeast Nuclear Code (AYNC), i.e. all CTG codons were translated to serine. In my analysis of the results, the vast majority of CTG codons were translated to serine, not leucine, see Paper III. Specifically, of the translated genes that had peptidic coverage of their CTG codon, 450 (97.2%) were translated to serine and 13 (2.8 %) were translated to leucine, which warrants S. schoenii a reassignment into the CTG clade. Indeed, Krassowski et al. recently reassigned three Saccharomycopsis species into a special Ser1 clade of the CTG clade yeasts, a reassignment which our findings validate (Krassowski et al., 2018). At the same time, we were unable to identify the CTG translation of almost 70 % of the S. schoenii proteins with a CTG that were identified. The main technical reason for this was that there were no peptides covering the CTG codons in these proteins. However, in a small number of cases, it had not been possible to determine the amino acid from a CTG codon. In light of the recent finding by Mülhausen et al. (Mühlhausen et al., 2018), a more stringent reanalysis of the raw data could be warranted.

49 Klara Junker | Prospecting a Predator

It has been hypothesized that C. albicans more or less by purpose ”mistranslate” some 3 % of the CTG codon to leucine in order to generate surface variability, as a stress response or as a part of a virulent behaviour (Miranda et al., 2013). To determine if S. schoenii would “mistranslate” the CTG codon in the event on nutritional stress or as a part of its predatory activity, I snap froze S. schoenii cells that had been cultured on starvation media (SD) for three hours and S. schoenii cells that were predating on S. cerevisiae after three hours of on starvation media (SD + S. cerevisiae). Phylogene performed the proteomic analysis as described above, with the addition that abundance of proteins and peptides from these two samples were also expressed as an abundance ratio, compared to the abundance of proteins and peptides in standard media. In this analysis we were able to detect minor differences in the proportions of the translation of the identified proteins with a serine vs leucine translation. During starvation conditions, 430 (96.8 %) of the proteins with peptide coverage of their CTG position were translated to serine, and 14 (3.2 %) were translated to leucine. During predation conditions 399 (98.8 %) of the proteins with peptide coverage of their CTG position were translated to serine and 5 (1,2 %) were translated to leucine. While these numbers do not exclude the possibility of a targeted differential translation, the individual proteins that were identified with leucine translation differed in between the samples, suggesting to us that leucine is more likely to be incorporated by chance.

Physical necrotrophic mycoparasitism To study the observable predatory behaviour, I set up a live microscopy study, where I could monitor the interaction between individual S. schoenii cells and S. cerevisiae (H4-GFP) cells as prey cells. Histone H4 is a core component of the nucleosome that wraps DNA, and histone modification is associated with cell death and apoptosis in S. cerevisiae (Fahrenkrog, 2016). I captured of co-cultured S. schoenii cells and S. cerevisiae cells on starvation media (SD) images every 5 minutes for up to 3 hours. Prey cells that came into physical contact with S. cerevisiae vacuolarized, shrank significantly and lost their H4-GFP signal, whereas non-predated cells kept their shape and retained their H4-GFP signal, see Paper III. These results verified that that S. schoenii kills S. cerevisiae cells in a contact-mediated manner and demonstrates that it took between 20-135 minutes for a prey S. cerevisiae H4-GFP cell lose its H4-signal after it came into physical contact with a S. schoenii, with a mean of 54 minutes (data not shown).

50 Prospecting a Predator | Klara Junker

Nitrogen source limiting media Several pathogenic fungi, such as C. albicans, modify their morphology and associated virulent behaviour in response to altered nutritional conditions (Brand, 2012). Having observed how S. schoenii would readily attack prey cells under starvation conditions, we hypothesized that, given its lack of genes in the sulfate assimilation pathway, the availability of organic sulfur compounds and methionine would also influence the propensity of S. schoenii to attack and kill prey cells. To tease out how S. schoenii respond to nutritional conditions, we employed four different types of media, YPD, CSM, CSM-Met and SD, representing gradually nutrient-limited media, especially in regards to nitrogen sources and with a specific focus on methionine (Victoria and Kevin, 2008). YPD is a rich and complex media, made up of essentially ground up S. cerevisiae cells, sugar, polypeptides and amino acids, which allows most yeast strains to eat, live and multiply. CSM (Complete Supplement Mixture) is a synthetic mix made up of only the amino acids, vitamins, trace elements and salts that are essential for growth of prototrophic, i.e. wild type S. cerevisiae cells, and sugar. CSM prompts yeasts to focus on biosynthesis of both the absent amino acids as well as on protein synthesis. CSM-Met is CSM without methionine, and allows for analysis of specific responses due to a specific lack of methionine. SD is a minimal media that only contain ammonium sulfate as a nitrogen source, together with vitamins, trace elements, salts, and sugar, which necessitates de novo synthesis of all amino acids, On SD media, cells need to recycle proteins and other cellular components to make any type of new material, and peripheral processes and cell division are typically downregulated.

Predatory activity as a function of nutrient availability To quantify if and how the nutritional conditions would impact the propensity of S. schoenii to attack and kill prey cells I set up quantitative kill curve assay, based on the distinct morphology of killed prey cells. By capturing frames of S. schoenii co-cultured with S. cerevisiae, I could score prey cells based on morphology and enumerate predatory activity every hour over the course of six hours, as a factor of nutrient availability, see Paper III. We had anticipated that the specific removal of methionine, i.e. the difference between predatory activity between CSM and CSM-Met would be detectable, but in this set-up it was not. Instead, the greatest difference could be observed between YPD and CSM, indicating that the removal of complex nutrients was the major trigger of predatory activity in this set up. During conditions of starvation on SD media, access to prey meant that S. schoenii would not go into “hiatus”, but instead redistribute its resources to allow for attack and killing of

51 Klara Junker | Prospecting a Predator prey cells, which suggested that there was most likely a nutritional reward associated with killing other yeasts which would offset the energy spent on fatal attacks. In addition, the observation that S. schoenii, to a small degree, even attacks and kill prey cells when complex nutrients are abundant on YPD, suggest that S. schoenii probably does not kill other yeasts merely as a survival tactic, but that predatory activity could also be a way to eliminate competitors.

Set-up of genome-wide transcriptomic analyses To identify gene overexpressed during starvation or predatory conditions, I seeded S. schoenii alone or together with S. cerevisiae (BY4741) on YPD, CSM, CSM-Met and SD, see Table 1, in triplicates. After three hours of co-culture, when predatory activity had been demonstrated as well under way, but far from over, I snap froze cells, extracted the RNA. BGI Europe constructed single-stranded libraries and sequenced the reads. I mapped reads to S. schoenii ORFs, calculated means from each triplicate, normalized expression values and set the baseline reference as the sample with S. schoenii cultured alone on standard media (YPD). I reduced noise by rejecting genes with expression levels lower than 100, and selected genes with at least 10-fold upregulation during any of the conditions. I manually curated upregulated genes according to shared expression patterns, and used FungiFun2, a web based GO term category finder (Priebe et al., 2015), to identify significantly enriched and shared biological features within the gene sets.

Burn fat or kill to eat? Three distinct expression patterns were discernible in our transcriptomic analysis of S. schoenii; the response to nutrient limitation when S. schoenii was cultured alone (YPD vs CSM), the specific removal of methionine when S. schoenii was cultured alone (CSM vs CSM-Met), and the predation specific responses, when prey was available compared to not on the same media (e.g. CSM vs CSM + S.c.), see Paper III. In response to nutrient limitation, genes enabling energy generation from carnitine were highly upregulated. Carnitines prolong the life span of yeasts (Palermo et al., 2010), suggesting that in response to limited nutrients, S. schoenii was maintaining homeostasis by living of stored energy reserves. During methionine deprivation, gene expression in S. schoenii was more drastically refocused, to enable the uptake of organic sulfur compounds and shuttle any sulfur- containing compounds towards the methionine biosynthesis pathway. For instance, Met4 is the major transcriptional regulator of the sulfur metabolic network in S. cerevisiae

52 Prospecting a Predator | Klara Junker

(Ljungdahl and Daignan-Fornier, 2012), and MET4 was upregulated nearly 10-fold when methionine was removed. However, when prey was present during methionine deprived conditions, MET4 was only 3-fold upregulated, suggesting an increase in available sulfur compounds during predatory conditions. Similar patterns were seen for the other genes that scavenge, salvage and shuttle organosulfur compounds to the methionine biosynthesis pathway (Thomas and Surdin-Kerjan, 1997); highly upregulated during methionine- deprived conditions, but a relaxed upregulation if prey was present. In contrast, the two genes “spinning” the methyl cycle, SAH1 and MET6, were upregulated twice as much if prey was present compared to absent, during methionine-deprived conditions, suggesting that it was only during the presence of prey that enough sulfur compounds made it far enough to allow the methyl cycle to spin. In summary, S. schoenii responds to the absence of complex nutrients by either burning carnitine and fatty acids, or attack and kill other yeasts, if they are available. S. schoenii is highly sensitive to the absence of methionine, and appears to be able to scavenge methionine and/or other sulfur compounds from S. cerevisiae. However, because the methionine deprivation was not discernible when it came to predatory behaviour, I hypothesize that S. schoenii is proactive in acquiring the nutrients it needs, rather than attempting to steal those nutrients as a last resort, in particular when it comes to organic sulfur compounds.

Predatory genes With the transcriptomic assay, I was also able to identify predation specific gene expression. Most notably, aspartic proteases, homologous to YPS3 and SAP1 genes were highly upregulated, together with glucanases and chitinases. This predatory expression profile of S. schoenii resembles that of other invasive parasites. C. albicans upregulate SAP genes during invasive growth in humans (Naglik, Challacombe and Hube, 2003), and several mycoparasites highly upregulate both proteases, chitinases and glucanases during their attack of prey cells (Karlsson et al., 2017). While the mycoparasite Trichoderma predominantly employs glucanases and chitinases, S. schoenii predominantly overexpressed proteases. Interestingly, Trichoderma was recently reported to have acquired genes encoding plant cell-wall degrading enzymes from its plant-associated prey species (Druzhinina et al., 2018).

53 Klara Junker | Prospecting a Predator

Medically relevant prey Initial studies by Lachance et al. on which species were susceptible to attack by predatory Saccharomycopsis yeasts indicated that their prey range is wide, and include both ascomycetous and basidiomycetous yeasts (Lachance et al., 2000). Later studies confirmed that filamentous fungi are also susceptible (Pimenta et al., 2008; Iacumin et al., 2017). However, except for C. albicans, most yeast prey cells tested by Lachance et al. had little biological relevance in terms of the meaningfulness in targeting their elimination. Guided by the facts that several pathogenic Candida species colonize the skin prior to infection, and the observation that S. schoenii is able to grow at skin temperatures (32-35°C) but not body temperature (37°C) (Kurtzman and Smith, 2011a), we hypothesized that S. schoenii might be able to kill pathogenic Candida species on skin, while not presenting an immediate risk factor for systemic infection in itself.

Figure 14: The dimorphic yeast S. schoenii attacks and kills multi-drug resistant Candida auris cells (NCPF8985#20). Arrows indicate penetration sites, and the outline of the penetration pegs made visible by calcofluor white stain (cyan, right-hand panel). Attacked and killed cells are stained by propidium iodide (red, right-hand panel). Images were captured after 60 minutes after of co-culture on SD media. A movie version of this interaction is available as Supplementary Movie 1 for this thesis at https://tinyurl.com/KlaraPhD.

S. schoenii can kill clinical isolates of Candida species To test if S. schoenii could kill clinically isolated Candida species in vitro, I quantified predatory activity on C. albicans, C. glabrata, C. topicalis, C. lusitaniae, C. parapsilopsis as well as four isolates of C. auris with variable drug resistance profiles. I found that S. schoenii was able to attack and kill all species tested, including the multidrug resistant C. auris isolates. Contrary to S. cerevisiae, that could be nearly eliminated by S. schoenii after six hours, no other species were eliminated after six hours, at which point overgrowth made

54 Prospecting a Predator | Klara Junker subsequent time points with this microscopy-based assay impossible. Our results clearly indicate a potential of S. schoenii as a novel antifungal agent against clinically important, including multidrug resistant, yeast pathogens.

Perspectives for further research

In this thesis I demonstrated that S. schoenii is able to attack and kill clinically relevant Candida species in vitro. However, should these finding be explored further, a wide range of additional studies are needed. Under what settings might S. schoenii eliminate C. auris completely? Do varying temperature effect predatory activity? Can S. schoenii killing of C. auris be studied on a skin model system, either in vivo using animal models, or in vitro using human skin?

C. albicans expresses SAP genes differently, depending on infection type (Staib et al., 2000). To better understand whether the predation specific tools S. schoenii employ are generic or prey species specific, one could compare transcriptomic responses when S. schoenii attacks other yeasts, such as C. auris and the phylogenetically distant yeast S. pombe. In addition, studies on target specificity of each aspartic protease from S. schoenii or Saccharomycopsis yeast might also reveal if these enzymes specifically target only fungal proteins or if human proteins are susceptible too. Such studies could be useful in any prospecting of novel antifungal molecules.

When I performed live cell microscopy time-lapse studies on the interactions between S. schoenii cells and S. cerevisiae prey cells, I could observe that S. schoenii appears to direct its growth towards its prey cells. I could also observe similar growth patterns on a macroscopic scale, between colonies on agar plates. If and how S. schoenii senses prey, and what components from both the predator and prey that are involved in such long distance interactions merits further investigations.

To be able to control and/or further understand the predatory behaviour of S. schoenii, genetic manipulation is key. One alleged disruption of an acid protease gene has been reported in S. fibuligera, and testing for any changes in any predatory behaviour of that mutant might be useful, coupled with complete genome sequencing, to confirm which of the several acid protease genes was disrupted.

55 Klara Junker | Prospecting a Predator

The sexual cycle of Saccharomycopsis yeasts has not been described, but an understanding of the complete life cycle of S. schoenii is warranted from a biological perspective. Given that S. cerevisiae has a sexual cycle in wasps (Stefanini et al., 2016), and that a close relative of S. schoenii, W. anomalus, is associated with mosquitoes (Ricci et al., 2011), it could be hypothesized that sexual reproduction of Saccharomycopsis yeasts might be linked to the insects it can be isolated from, e.g. beetles. Preliminary data I have generated include the detection of potential mating type loci in the draft genome of S. schoenii, a cAMP-mediated mating-type switching and a G1-phase, haploid nature of S. schoenii. In addition, I generated microscopy-based and transcriptomic data on mating and sporulation, that needs further analysis, but I could observe that key mating related genes were upregulated, such as mating factor α and several pheromone responsive genes.

In this thesis I have focused on the potential application for S. schoenii against medically relevant Candida species, whereas previous studies have identified biocontrol potential in both agriculture and food protection. Based on its broad prey range, we also hypothesized that it might be useful as a biocontrol agent against important plant pathogens. To that end, I also generated preliminary data showing that S. schoenii is able to attack and kill Fusarium graminearum, an important plant pathogen of cereals such as barley, which should be investigated further.

56 Prospecting a Predator | Klara Junker

Conclusions

 S. schoenii, S. fermentans and S. fodiens lack all genes in the sulfate assimilation pathway in their genomes, probably as a consequence of their mycoparasitic abilities.  The S. schoenii, S. fermentans and S. fodiens genomes harbours several expanded gene families associated with mycoparasitism, such as aspartic proteases and chitinases and, in the S. schoenii genome, transposable elements.  The S. schoenii and S. fermentans genome harbours two distinct tRNA(CAG) genes in its genome, one tRNA(CAG)Ser and one tRNA(CAG)Leu.  S. schoenii translates CTG to serine and is a member of the CTG clade of yeasts. The CTG clade was recently subdivided in to subclades, and our analysis suggest that S. schoenii, S. fermentans and S. fodiens also belong to the proposed Ser2 clade.  S. schoenii is able to kill S. cerevisiae within 20 minutes of physical contact.  S. schoenii is able to kill S. cerevisiae under a range of different nutritional conditions.  Methionine deprivation triggers major transcriptional responses in S. schoenii related to methionine scavenging, salvage and biosynthesis, but is not the sole trigger for predatory behaviour of S. schoenii.  S. schoenii probably acquires methionine and/or other organic sulfur compounds from prey cells during its predatory activity.  S. schoenii upregulates aspartic proteases during its predatory activity.  S. schoenii is able to attack and kill clinical isolates of C. albicans, C. glabrata, C. topicalis, C. lusitaniae and C. parapsilopsis in vitro.  S. schoenii is able to attack and kill clinical isolates of C. auris, including both sensitive and multi-drug resistant isolates, in vitro.

57 Klara Junker | Prospecting a Predator

Acknowledgements

First, I thank Jürgen Wendland for bringing me in to work on and with the fascinating predatory Saccharomycopsis yeasts and for your scientific advice. What would this thesis be without you?

Thank you Birgitte Regenberg for keeping me on track with my formal PhD work and for providing me with relevant new points of view and ideas for my research.

Thank you Jochen Förster for doing what great boss should do, having my back, supporting me and… signing off the payments for all expensive experiments I wanted to do  It paid off, right?

Thank you all past and present members of the Yeast & Fermentation group for advising me on high and low, inspiring me to do a better job every day and for being the best of colleagues. Thank you; Ana Hesselbart for establishing the bioinformatic foundations I could build my work on. Lisa Wasserström, for taking the time to introduce me to all aspects of the lab, I value our time greatly. Klaus Lengeler, for being the yeast expert who always took the time to answer any pressing question I had. Andrea Walther for introducing me how to work the moody big microscope. Claudia Kempf and Davide Ravasio for inspiring me to push on through with my work. Anna Chailyan for introducing me on how to work with bioinformagic..., eh, bioinformatics, for guiding me through my PacBio data set and for supporting me in times of need. Ross Fennessey for guiding me in designing sequencing experiments and for sharing a passion for great beer. Michael Katz for all our conversations throughout the years, you have kept me looking at the horizon ahead. Judita Gartner for ALWAYS being able to put me in a good mood. Heidi Hansen for your radiant kindness and the calm way you take time to small talk about big and small. Rosa Garcia Sanchez and Kathrine Meier Axelsen for being excellent people to talk about the highs and lows of life as a parent in science. Claes, Marc, Natalia, Marta M-K, Gemma, Igor, Marta Z, Anne-Mette and Sara for being Triple A colleagues; Ambitious, Amusing and Awesome 

58 Prospecting a Predator | Klara Junker

My time at the Carlsberg Research Lab has been a delight thanks to all its great employees. First and foremost, thank you Merete Yding, for being excellent in everything you do, you inspire me with your many talents. You have made my life run as smooth as possible in countless ways, such as helping me arrange day care for my children and for hooking me up with apartments and living arrangement countless times! Thank you Anastassia for being the coolest co-worker as well as a dear friend. Thank you Mogens and Mads for providing excellent IT support and being able to solve any kind of problem I bothered you with. Thank you Hanne and Rikke for making sure everything I did got paid for. Thank you Tove, Mette, Berit and Suzanne for feeding both my gut and soul with your care in the canteen. Thank you Mette, Alexander, Christoph and Jesper for conversations about life, science and beer! Everyone, past and present, in the cake club for marking the half time of the working week with splendid company, a good dose of sugar and a little bit of gossip…

Thank you everyone in the Fungibrain network. Thank you all fellows, Valeria, Patrícia S, Tania, Antonio, Mariana, Saskia, Cassandre, Hugo, Luigi, Paola, Pavlos, Patrícia H and Stefania. We have trained, supported and cheered each other on, and the time I have spent with all of you has been priceless. While this type of Pan-European training network has already paid off (hats off to everyone who got or are about to get their PhD title!), I believe its true value will shine even further in the future, thanks to the friendships we have created now. Thank you Colette Inkson for being the spider in the network. Many, many thanks also to the PIs for making Fungibrain both exist and run smoothly; Nick Read, Jason Oliver, Antonio di Pietro, Jürgen Wendland, Gerhard Braus, Nicolas Minc, Rob Arkowitz, Martine Bassilana, Sophie Martin, Pepe Perez-Martin, André Fleissner, Neil Gow and Alex Brand. You have all put such great efforts into this network, and I would not be where I am today without you, thank you again. Thanks also to Darren Thompson, for almost being a part of Fungibrain and introducing me and the other fellows to cool microscopy and FIJI/ImageJ…

Thank you Neil Gow, Alexander Lorenz and Gustavo Bravo for inviting me to your labs, trusting me to play with expensive microscopes in Aberdeen and helping and allowing me to do some truly exciting work with C. auris. Those two weeks turned out to be game changing for my research! Thank you Louise Walker for making excellent TEM pictures of the predators in action.

59 Klara Junker | Prospecting a Predator

Thank you everyone else I have met in science which includes but is not restricted to; My friends and colleagues from my time working on malaria parasites in the Duraisingh lab in Boston, especially Kim & Ulf, Sandra & Per, Phiia-Lotta & Per, Johan A, Elisabeth, Elizabeth, Caeul, Yovany & Zaava and Maura, thank you for showing me that a science career worth pursuing is the result of both hard work and fun times (especially when you get to do it your own way). Akria Kaneko and my friends and colleagues in the Drakely lab in London, thank you for making my first paid business trip a malaria parasite field trip to Vanuatu in the South Pacific! My friends and colleagues in the Engwerda lab in Brisbane, especially Ashraful Haque for being probably the best scientific supervisor I ever had, which resulted in my first co-authored scientific publication on the immune system against malaria parasites! My friends and colleagues in the Huete-Pérez lab who first introduced me to parasite research, especially Carlos Talavera-López, whenever I think I struggle, I think of you and how you successfully pulled through!

Thank you everyone from my times at MedBi in Linköping, for making my life during and after those four undergraduate years some of the best I ever had. You are too many to mention, but the bare minimum of you include Susanne, Lina, Carl-Oscar & Cecilia, Ulrika & Henrik, Anders, Jonas, Patiyan & Lauren, Helena & Petrus, Jonathan & Sofie, Carmen, Tobias, Robert, Joakim, Torsten, Ida, Lotta etc etc etc!!! Patrik Olauson, rest in peace.

My extra special friends from MedBi, “Guns Girls”; Elin Falk, Emma Sund, Anna Forsberg, Hanna Skärstrand, Angelika Holm and Pamela Wintmo. I want to thank you from the bottom of my heart for all the fun we have had and will have, and for all the strong support we have given each other in this group. We all work hard so that we can play hard, but we also know the importance of giving ourselves a break every now and then, preferably in each other’s company.

Thank you to all my “normal” friends who have cheered me on along the way, Josefina, Emelie, Rebecka, Thommie & Lina, Stina, Jill & Erik, Mona & Richard, Kristina and Emma. And everyone who cheered med on through social media – you make a difference!

60 Prospecting a Predator | Klara Junker

Thank you everyone else who inspired me to be a better person, better scientist, better appreciator of the good fermented things in life and those who have spurred my interest in entrepreneurship, in no particular order; everyone on “Science-Twitter” (especially Sophien Kamoun for pushing me to join!), all cool entrepreneurs I’ve met through Instagram (Surtantens, SimranSethi, TheBlackSheepSchool, TeamSilent, Ourcookquest, Adalagard, Butterviking etc), the beer yeast experts Troels Prahl and Kevin Verstrepen, the ambitious people of REBBLS (especially my “colleagues” in the core group of 2016-17), Martin at SUND Hub for believing in my ideas, my brewing technology classmates (especially Jens B), my colleagues at Mystic Brewery, the BABES in Boston, my colleagues at Carlsberg in Falkenberg (especially Elisabeth Janver for allowing me a day off for a visit the Yeast Biology group at the Carlsberg in Copenhagen, that turned out fruitful!), Mariam and Ölgäris. Thank you Hanna A for being an amazing babysitter when I was living as a single mom with two kids in Copenhagen and needed a liiittle bit more time in the lab…

Thank you my brothers, Dr Erik and Dr Jon, you both paved the road for me and showed me paths off the beaten tracks. I love you, but never forget who got published first…! Thank you my sister-in-laws Dr Johanna U and (soon to be) Dr Johanna N, you are brilliant and I am so happy to have you in my family. Tack Stina, Karl, Henning och Felix för att ni är världens finaste brorsbarn! Tack Petter för att du introducerade mig till Köpenhamn och funnits där. Tack svärmor Janicke, för allt stöd genom åren, speciellt då du tog hand om Siri den första tiden i Köpenhamn. Stort tack mamma och pappa för att ni har gjort mig till den jag är, låtit mig vara den jag är och stöttat mig i alla de upptåg jag har ägnat mig åt runt om i världen. Extra tack till mamma för all den ovärderliga hjälp du gett mig med barnen under min sista tid i Köpenhamn. Utan dig hade de bästa resultaten i denna bok inte kommit till!

Tack Johan för att du stöttar, pushar och står ut med mig. Du är bäst (nu är det på pränt!) och jag älskar dig! Tack Siri för att du, sen dagen du var ett blått streck, gett mig världens bästa perspektiv på livet och styrt in min livsväg på nya outforskade och roliga vägar. Du är den smartaste, mest kreativa och underbara dotter man kan ha, dina heja-rop betyder allt för mig. Tack Emil för att du gett mig andrum, tålamod och en blick för nya möjligheter i livet, du är den snällaste, roligaste och mest omtänksamma son som finns. Tack för alla våra ”millekramar” som gör allt annat i livet värdsligt. Jag älskar er mer än ni nånsin kan förstå!

61 Klara Junker | Prospecting a Predator

References

Albalat, R. and Cañestro, C. (2016) ‘Evolution by gene loss’, Nature Reviews Genetics. Nature Publishing Group, 17(7), pp. 379–391. Angiuoli, S. V et al. (2011) ‘Resources and Costs for Microbial Sequence Analysis Evaluated Using Virtual Machines and Cloud Computing’, PLOS ONE. Public Library of Science, 6(10), p. e26624. Available at: https://doi.org/10.1371/journal.pone.0026624. von Arx, J. A. and Yarrow, D. (1984) ‘The adventures of the yeast genus Endomycopsis Dekker’, Antonie van Leeuwenhoek, 50(5), pp. 799–805. Becher, P. G. et al. (2018) ‘Chemical signaling and insect attraction is a conserved trait in yeasts’, Ecology and Evolution, 8(5), pp. 2962–2974. Beckerich, J. et al. (2015) ‘Sulfur Metabolism in Hemiascomycetes Yeast’, in Molecular Physiology and Ecophysiology of Sulfur, pp. 49–63. Bongomin, F. et al. (2017) ‘Global and Multi-National Prevalence of Fungal Diseases—Estimate Precision’, Journal of Fungi, 3(4), p. 57. Brand, A. (2012) ‘Hyphal growth in human fungal pathogens and its role in virulence’, International Journal of Microbiology, 2012. Burge, M. N. (1988) Fungi in biological control systems . Edited by M. N. Burge. Manchester: Manchester University Press. Butt, T. M., Jackson, C. and Magan, N. (2001) Fungi as biocontrol agents: progress, problems and potential. Edited by T. M. Butt, C. Jackson, and N. Magan. Wallingford: CABI. Buzzini, P., Lachance, M. A. and Yurkov, A. (2017) Yeasts in natural ecosystems: Diversity, Yeasts in Natural Ecosystems: Diversity. Chi, Z. et al. (2003) ‘Enhanced conversion of soluble starch to trehalose by a mutant of Saccharomycopsis fibuligera sdu’, Journal of Biotechnology, 102(2), pp. 135–141. Chi, Z. et al. (2009) ‘Saccharomycopsis fibuligera and its applications in biotechnology’, Biotechnology Advances, 27(4), pp. 423–431. Chin, V. K. et al. (2016) ‘Dissecting candida albicans infection from the perspective of c. Albicans virulence and omics approaches on host–pathogen interaction: A review’, International Journal of Molecular Sciences, 17(10). Choo, J. H. et al. (2016) ‘Whole-genome de novo sequencing, combined with RNA-Seq analysis, reveals unique genome and physiological features of the amylolytic yeast Saccharomycopsis fibuligera and its interspecies hybrid’, Biotechnology for Biofuels, 9(246), pp. 1–22. Christiaens, J. F. et al. (2014) ‘The fungal aroma gene ATF1 promotes dispersal of yeast cells through insect vectors’, Cell Reports. The Authors, 9(2), pp. 425–432. Dean, R. et al. (2012) ‘The Top 10 fungal pathogens in molecular plant pathology’, Molecular Plant Pathology, 13(4), pp. 414–430. Denning, D. W. (2017) ‘Calling upon all public health mycologists’, European Journal of Clinical Microbiology & Infectious Diseases. European Journal of Clinical Microbiology & Infectious Diseases, 36(6), pp. 923–924. Domergue, R. (2005) ‘Nicotinic Acid Limitation Regulates Silencing of Candida Adhesins During UTI’, Science, 308(5723), pp. 866–870. Douglas B. Murphy and Michael W. Davidson (2013) ‘Chapter 16: IMAGING LIVING CELLS WITH THE MICROSCOPE’, Fundamentals of Light Microscopy and Electronic Imaging.

62 Prospecting a Predator | Klara Junker

Druzhinina, I. S. et al. (2011) ‘Trichoderma: the genomics of opportunistic success’, Nature Reviews Microbiology, 9(10), pp. 749–759. Druzhinina, I. S. et al. (2018) Massive lateral transfer of genes encoding plant cell wall-degrading enzymes to the mycoparasitic fungus Trichoderma from its plant-associated hosts, PLoS Genetics. Dujon, B. A. and Louis, E. J. (2017) ‘Genome Diversity and Evolution in the Budding Yeasts (Saccharomycotina)’, 206(June), pp. 717–750. Duplessis, S. (2011) ‘Obligate biotrophy features unraveled by the genomic analysis of rust fungi’, Proceedings of the National Academy of Sciences of the United States of America, pp. 1–23. Eden, E. et al. (2009) ‘GOrilla: A tool for discovery and visualization of enriched GO terms in ranked gene lists’, BMC Bioinformatics, 10, pp. 1–7. El-Tarabily, K. A. and Sivasithamparam, K. (2006) ‘Potential of yeasts as biocontrol agents of soil- borne fungal plant pathogens and as plant growth promoters’, Mycoscience, 47(1), pp. 25–35. Eliceiri, K. et al. (2012) ‘NIH Image to ImageJ : 25 years of image analysis HISTORICAL commentary NIH Image to ImageJ : 25 years of image analysis’, Nature Methods, 9(7), pp. 671–675. English, A. C. et al. (2012) ‘Mind the Gap: Upgrading Genomes with Pacific Biosciences RS Long- Read Sequencing Technology’, PLOS ONE. Public Library of Science, 7(11), p. e47768. Available at: https://doi.org/10.1371/journal.pone.0047768. Erwig, L. P. and Gow, N. a. R. (2016) ‘Interactions of fungal pathogens with phagocytes’, Nature Reviews Microbiology. Nature Publishing Group, 14(3), pp. 163–176. Fahrenkrog, B. (2016) ‘Histone modifications as regulators of life and death in Saccharomyces cerevisiae’, Microbial Cell, 3(1), pp. 1–13. Fisher, M. C. et al. (2018) ‘Worldwide emergence of resistance to antifungal drugs challenges human health and food security’, Science, 742(May), pp. 739–742. Gagnon-Arsenault, I., Tremblay, J. and Bourbonnais, Y. (2006) ‘Fungal yapsins and cell wall: A unique family of aspartic peptidases for a distinctive cellular function’, FEMS Yeast Research, 6(7), pp. 966–978. Goffeau, A. et al. (1996) ‘Life with 6000 genes’, Science, 274(5287), pp. 546–567. Götz, S. et al. (2008) ‘High-throughput functional annotation and data mining with the Blast2GO suite’, Nucleic Acids Research, 36(10), pp. 3420–3435. Gow, N. A. R., Latge, J. and Munro, C. A. (2017) ‘The Fungal Cell Wall : Structure , Biosynthesis , and Function’, Microbiol Spectrum, 5(3), pp. 1–25. Grube, M. et al. (2017) ‘The next generation fungal diversity researcher’, 1. Gruber, S. G. and Seidl-Seiboth, V. (2012) ‘Self versus non-self: Fungal cell wall degradation in Trichoderma’, Microbiology, 158(1), pp. 26–34. Gurevich, A. et al. (2013) ‘QUAST: Quality assessment tool for genome assemblies’, Bioinformatics, 29(8), pp. 1072–1075. Haas, R. et al. (2017) ‘Designing and interpreting “multi-omic” experiments that may change our understanding of biology’, Current Opinion in Systems Biology. The Francis Crick Institute, 6(September), pp. 37–45. Hawksworth, D. L. (2012) ‘Global species numbers of fungi: Are tropical studies and molecular approaches contributing to a more robust estimate?’, Biodiversity and Conservation, 21(9), pp. 2425–2433. Heitman, J. (2011) ‘Microbial pathogens in the fungal kingdom’, Fungal Biology Reviews. Elsevier Ltd, 25(1), pp. 48–60. Hickman, M. a et al. (2013) ‘The “obligate diploid” Candida albicans forms mating-competent haploids’, Nature, 494(7435), pp. 55–59. Hug, L. A. et al. (2016) ‘A new view of the tree of life’, Nature Microbiology, 1(5), p. 16048. Iacumin, L. et al. (2017) ‘Biocontrol of ochratoxigenic moulds (Aspergillus ochraceus and Penicillium

63 Klara Junker | Prospecting a Predator

nordicum) by Debaryomyces hansenii and Saccharomycopsis fibuligera during speck production’, Food Microbiology. Elsevier Ltd, 62, pp. 188–195. Inglis, G. D. and Kawchuk, L. M. (2002) ‘Comparative degradation of oomycete, ascomycete, and basidiomycete cell walls by mycoparasitic and biocontrol fungi’, Canadian Journal of Microbiology, 48(1), pp. 60–70. James, T. Y. et al. (2006) ‘Reconstructing the early evolution of Fungi using a six-gene phylogeny’, Nature, 443(7113), pp. 818–822. Jeffery-Smith, A. et al. (2018) ‘Candida auris: a Review of the Literature’, Clinical Microbiology Reviews, 31(1), pp. e00029-17. Jeffries, P. (1995) ‘Biology and ecology of mycoparasitism’, Canadian Journal of Botany. NRC Research Press, 73(S1), pp. 1284–1290. Jennings, D. H. (1995) The physiology of fungal nutrition. Digitally. Edited by D. H. Jennings. Cambridge: Cambridge University Press. Karlsson, M. et al. (2017) ‘Necrotrophic Mycoparasites and Their Genomes’, pp. 1–21. Kemen, E. et al. (2011) ‘Gene gain and loss during evolution of obligate parasitism in the white rust pathogen of Arabidopsis thaliana’, PLoS Biology, 9(7). Kemen, E. and Jones, J. D. G. (2012) ‘Obligate biotroph parasitism: Can we link genomes to lifestyles?’, Trends in Plant Science. Elsevier Ltd, 17(8), pp. 448–457. Kim, J. Y. (2016) ‘Human fungal pathogens: Why should we learn?’, Journal of Microbiology, 54(3), pp. 145–148. Köhler, J. R., Casadevall, A. and Perfect, J. (2015) ‘The spectrum of fungi that infects humans’, Cold Spring Harbor Perspectives in Medicine, 5(1), pp. 1–22. Krassowski, T. et al. (2018) ‘Evolutionary Instability of CUG-Leu in the Genetic Code of Budding Yeasts’, Nature Communications. Springer US, (2018), p. In Press. Kreger-van Rij, N. J. and Veenhuis, M. (1973) ‘Electron microscopy of some special cell contracts in yeasts.’, Journal of Bacteriology, 113(1), pp. 350–356. Krysan, D., Ting, E. and Abeijon, C. (2005) ‘Yapsins are a family of aspartyl proteases required for cell wall integrity in Saccharomyces cerevisiae.’, Eukaryot Cell, 4(8), pp. 1364–74. Kurtzman, C. P., Fell, J. W. and Boekhout, T. (2010) The yeasts : a taxonomic study. Elsevier. Available at: https://www.sciencedirect.com/science/book/9780444521491 (Accessed: 14 May 2018). Kurtzman, C. P. and Robnett, C. J. (2013) ‘Relationships among genera of the Saccharomycotina (Ascomycota) from multigene phylogenetic analysis of type species’, FEMS Yeast Research, 13(1), pp. 23–33. Kurtzman, C. P. and Smith, M. T. (2011a) ‘Chapter 63 – Saccharomycopsis Schiönning (1903)’, in The Yeasts, pp. 751–763. Kurtzman, C. P. and Smith, M. T. (2011b) ‘Saccharomycopsis Schiönning (1903)’, in The Yeasts. Elsevier, pp. 751–763. Lachance, M. A. and Pang, W. M. (1997) ‘Predacious yeasts’, Yeast, 13(3), pp. 225–232. Lachance, M. a et al. (2000) ‘Nutrition and phylogeny of predacious yeasts.’, Canadian journal of microbiology, 46(6), pp. 495–505. Lewis, L. E. et al. (2013) ‘Live-cell video microscopy of fungal pathogen phagocytosis.’, Journal of visualized experiments : JoVE, (71), p. e50196. Li, K. (2008) The image stabilizer plugin for ImageJ. Available at: http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html,. Libkind, D. et al. (2011) ‘Microbe domestication and the identification of the wild genetic stock of lager- brewing yeast’, Proceedings of the National Academy of Sciences, 108(35), pp. 14539–14544. Linder, T. (2018) ‘Assimilation of alternative sulfur sources in fungi’, World Journal of Microbiology and Biotechnology. Springer Netherlands, 34(4), p. 51.

64 Prospecting a Predator | Klara Junker

Ljungdahl, P. O. and Daignan-Fornier, B. (2012) ‘Regulation of amino acid, nucleotide, and phosphate metabolism in Saccharomyces cerevisiae’, Genetics, 190(3), pp. 885–929. Lockhart, S. R., Berkow, E. L., et al. (2017) ‘Candida auris for the Clinical Microbiology Laboratory: Not Your Grandfather’s Candida Species’, Clinical Microbiology Newsletter. Elsevier Inc., 39(13), pp. 99–103. Lockhart, S. R., Etienne, K. A., et al. (2017) ‘Simultaneous emergence of multidrug-resistant candida auris on 3 continents confirmed by whole-genome sequencing and epidemiological analyses’, Clinical Infectious Diseases, 64(2), pp. 134–140. Lowe, T. M. and Chan, P. P. (2016) ‘tRNAscan-SE On-line: integrating search and context for analysis of transfer RNA genes’, Nucleic acids research, 44(W1), pp. W54–W57. Madden, A. A. et al. (2018) ‘The ecology of insect–yeast relationships and its relevance to human industry’, Proceedings of the Royal Society B: Biological Sciences, 285(1875), p. 20172733. Marzluf, G. A. (1997) ‘Molecular genetics of sulfur assimilation in filamentous fungi and yeast’, Annu Rev Microbiol, 51, pp. 73–96. Mims, C. W., Hanlin, R. T. and Richardson, E. A. (2007) ‘Light- and electron-microscopic observations of Cladosporium sp. growing on basidia of Exobasidum camelliae var. gracilis’, Canadian Journal of Botany. NRC Research Press, 85(1), pp. 76–82. Miranda, I. et al. (2013) ‘Candida albicans CUG mistranslation is a mechanism to create cell surface variation.’, mBio. American Society for Microbiology, 4(4), pp. e00285-13. Moat, A., Foster, J. and Spector, M. (2002) ‘Biosynthesis and metabolism of amino acids’, in Microbial Physiology. Fourth edi. Wiley-Liss, Inc., pp. 503–544. Available at: http://onlinelibrary.wiley.com/doi/10.1002/0471223867.ch15/summary. Mühlhausen, S. et al. (2018) ‘Endogenous Stochastic Decoding of the CUG Codon by Competing Ser- and Leu-tRNAs in Ascoidea asiatica’, Current Biology. Elsevier. Naglik, J. R., Challacombe, S. J. and Hube, B. (2003) ‘Candida albicans Secreted Aspartyl Proteinases in Virulence and Pathogenesis’, Microbiology and Molecular Biology Reviews, 67(3), pp. 400– 428. Nagy, L. G. et al. (2014) ‘Latent homology and convergent regulatory evolution underlies the repeated emergence of yeasts’, Nature Communications, 5. Nat. Micro. Editorial (2017) ‘Stop neglecting fungi’, Nature Microbiology, 2(8), p. 17120. Nicolopoulou-Stamati, P. et al. (2016) ‘Chemical Pesticides and Human Health: The Urgent Need for a New Concept in Agriculture’, Frontiers in Public Health, 4(July), pp. 1–8. Ohama, T. et al. (1993) ‘Non-universal decoding of the leucine codon CUG in several Candida species’, Nucleic Acids Research, 21(17), pp. 4039–4045. Palermo, V. et al. (2010) ‘Acetyl-L-carnitine protects yeast cells from apoptosis and aging and inhibits mitochondrial fission’, Aging Cell, 9(4), pp. 570–579. Pan, G. et al. (2013) ‘Comparative genomics of parasitic silkworm microsporidia reveal an association between genome expansion and host adaptation’, BMC Genomics. BMC Genomics, 14(1), p. 1. Papon, N. et al. (2013) ‘Emerging and Emerged Pathogenic Candida Species: Beyond the Candida albicans Paradigm’, PLoS Pathogens, 9(9). Pappas, P. G. et al. (2018) ‘Invasive candidiasis’, Nature Reviews Disease Primers. Macmillan Publishers Limited, 4(May), p. 18026. Peter, J. et al. (2018) ‘Genome evolution across 1,011 Saccharomyces cerevisiae isolates’, Nature, 556(7701), pp. 339–344. Pimenta, R. S. et al. (2008) ‘Biological control of Penicillium italicum, P. digitatum and P. expansum by the predacious yeast Saccharomycopsis schoenii on oranges’, Brazilian Journal of Microbiology, 39(1), pp. 85–90. Priebe, S. et al. (2015) ‘FungiFun2: A comprehensive online resource for systematic analysis of gene

65 Klara Junker | Prospecting a Predator

lists from fungal species’, Bioinformatics, 31(3), pp. 445–446. Przybylo, M. et al. (2008) ‘A method to evaluate a propidium iodide association with oligonucleotides and oligonucleotide-cationic lipid interactions.’, Current drug discovery technologies, 5(2), pp. 162–7. Rapala-Kozik, M. et al. (2017) ‘Extracellular proteinases of Candida species pathogenic yeasts’, Molecular Oral Microbiology, (November 2017), pp. 1–12. Reis-Cunha, J. L. and Bartholomeu, H. O. V. and D. C. (2018) ‘Gene and Chromosomal Copy Number Variations as an Adaptive Mechanism Towards a Parasitic Lifestyle in Trypanosomatids’, Current Genomics, pp. 87–97. Ricci, I. et al. (2011) ‘The yeast Wickerhamomyces anomalus (Pichia anomala) inhabits the midgut and reproductive system of the Asian malaria vector Anopheles stephensi’, Environmental Microbiology, 13(4), pp. 911–921. Richards, T. A., Leonard, G. U. Y. and Wideman, J. G. (2017) ‘What Defines the “ Kingdom ” Fungi ?’, Microbiology Spectrum, 5(3), pp. 1–21. Riley, R. et al. (2016) ‘Comparative genomics of biotechnologically important yeasts’, Proceedings of the National Academy of Sciences, 113(35), pp. 9882–9887. Sasaki, D. T., Dumas, S. E. and Engleman, E. G. (1987) ‘Discrimination of viable and non-viable cells using propidium iodide in two color immunofluorescence’, Cytometry, 8(4), pp. 413–420. Satoh, K. et al. (2009) ‘Candida auris sp. nov., a novel ascomycetous yeast isolated from the external ear canal of an inpatient in a Japanese hospital’, Microbiology and Immunology, 53(1), pp. 41– 44. Schindelin, J. et al. (2012) ‘Fiji: An open-source platform for biological-image analysis’, Nature Methods, 9(7), pp. 676–682. Schindelin, J. et al. (2015) ‘The ImageJ ecosystem: An open platform for biomedical image analysis’, Molecular Reproduction and Development, 82(7–8), pp. 518–529. Schmitt, M. J. and Breinig, F. (2006) ‘Yeast viral killer toxins: lethality and self-protection.’, Nature reviews. Microbiology, 4(3), pp. 212–21. Schmoll, M. et al. (2016) ‘The Genomes of Three Uneven Siblings: Footprints of the Lifestyles of Three Trichoderma Species.’, Microbiology and molecular biology reviews : MMBR, 80(1), pp. 205– 327. Schneider, N. (2014) ‘fixTranslation-Macro-for-ImageJ’. Available at: https://github.com/NMSchneider/fixTranslation-Macro-for-ImageJ. Seidl, V. et al. (2009) ‘Transcriptomic response of the mycoparasitic fungus Trichoderma atroviride to the presence of a fungal prey.’, BMC genomics, 10, p. 567. Shen, X.-X. et al. (2016) ‘Reconstructing the Backbone of the Saccharomycotina Yeast Phylogeny Using Genome-Scale Data’, G3: Genes|Genomes|Genetics, 6(12), p. 3927 LP-3939. Available at: http://www.g3journal.org/content/6/12/3927.abstract. Silva, S. et al. (2012) ‘Candida glabrata, Candida parapsilosis and Candida tropicalis: Biology, epidemiology, pathogenicity and antifungal resistance’, FEMS Microbiology Reviews, 36(2), pp. 288–305. Siverio, J. M. (2002) ‘Assimilation of nitrate by yeasts’, FEMS Microbiology Reviews, 26(3), pp. 277– 284. Skrzypek, M. S. et al. (2017) ‘The Candida Genome Database (CGD): Incorporation of Assembly 22, systematic identifiers and visualization of high throughput sequencing data’, Nucleic Acids Research, 45(D1), pp. D592–D596. Spanu, P. D. et al. (2010) ‘Genome Expansion and Gene Loss in Powdery Mildew Fungi Reveal Tradeoffs in Extreme Parasitism’, Science, 330(6010), pp. 1543–1546. Spanu, P. D. (2012) ‘The Genomics of Obligate (and Nonobligate) Biotrophs’, Annual Review of

66 Prospecting a Predator | Klara Junker

Phytopathology, 50(1), pp. 91–109. Spatafora, J. W. et al. (2017) ‘The Fungal Tree of Life: from Molecular Systematics to Genome-Scale Phylogenies’, Microbiology Spectrum, 5(5), pp. 1–32. Staib, P. et al. (2000) ‘Differential activation of a Candida albicans virulence gene family during infection.’, Proceedings of the National Academy of Sciences of the United States of America, 97(11), pp. 6102–6107. Stajich, J. E. (2017) ‘Fungal Genomes and Insights into the Evolution of the Kingdom’, Microbiology Spectrum, 5(4), pp. 1–15. Stefanini, I. et al. (2012) ‘Role of social wasps in Saccharomyces cerevisiae ecology and evolution’, Proceedings of the National Academy of Sciences, 109(33), pp. 13398–13403. Stefanini, I. et al. (2016) ‘Social wasps are a Saccharomyces mating nest’, Proceedings of the National Academy of Sciences, 113(8), pp. 2247–2251. Stephanopoulos, G., Aristidou, A. and Nielsen, J. (1998) ‘Review of Cellular Metabolism’, Metabolic Engineering, 16(7), pp. 21–79. Thomas, D. and Surdin-Kerjan, Y. (1997) ‘Metabolism of sulfur amino acids in Saccharomyces cerevisiae’, Microbiology and molecular biology reviews : {MMBR}, 61(4), pp. 503–532. Torruella, G. et al. (2015) ‘Phylogenomics Reveals Convergent Evolution of Lifestyles in Close Relatives of Animals and Fungi’, Current Biology, 25(18), pp. 2404–2410. Turner, S. A. and Butler, G. (2014) ‘The Candida pathogenic species complex’, Cold Spring Harbor Perspectives in Medicine, 4(9), pp. 1–18. Victoria, L. and Kevin, S. (2008) ‘Yeast’, Current Protocols in Molecular Biology. Wiley-Blackwell, 82(1), p. 13.0.1-13.0.4. Walker, G. M. and White, N. A. (2018) ‘Introduction to Fungal Physiology’, in Kavanagh, K. (ed.) Fungi: Biology and Applications. Third Edit. John Wiley & Sons, Inc, pp. 1–35. Wang, D. et al. (2011) ‘Disruption of the acid protease gene in Saccharomycopsis fibuligera A11 enhances amylolytic activity and stability as well as trehalose accumulation’, Enzyme and Microbial Technology. Elsevier Inc., 49(1), pp. 88–93. Whittington, A., Gow, N. A. R. and Hube, B. (2014) ‘From Commensal to Pathogen: Candida albicans’, Human Fungal Pathogens, pp. 3–18. Wilson, D. et al. (2009) ‘Identifying infection-associated genes of Candida albicans in the postgenomic era’, FEMS Yeast Research, 9(5), pp. 688–700. Young, T. W. K. (1994) Interfungal parasitic relationships . Edited by T. W. K. Young and P. Jeffries. Wallingford: C.A.B. International. Zhang, N., Luo, J. and Bhattacharya, D. (2017) Advances in Fungal Phylogenomics and Its Impact on Fungal Systematics. 1st edn, Fungal Phylogenetics and Phylogenomics. 1st edn. Elsevier Inc.

67 Klara Junker | Prospecting a Predator

Papers

All material is available at https://tinyurl.com/KlaraPhD

68 Prospecting a Predator | Klara Junker

Paper I

69 EUKARYOTES crossm

Draft Genome Sequence of Saccharomycopsis fodiens CBS 8332, a Necrotrophic Mycoparasite with Biocontrol Potential

Klara Junker,a Ana Hesselbart,a Jürgen Wendlanda,b Carlsberg Research Laboratory, Yeast & Fermentation, Copenhagen V, Denmarka; Department of Bioengineering Sciences, Research Group of Microbiology, Functional Yeast Genomics, Vrije Universiteit Brussel, Brussels, Belgiumb

ABSTRACT Saccharomycopsis fodiens is an ascomycetous necrotrophic mycopara- site. Predator-prey interaction leads to killing of the host cell by a penetration peg Received 12 October 2017 Accepted 20 October 2017 Published 16 November 2017 and utilization of cell content by the predator. Here, we report the 14.9-Mb S. fodi- Citation Junker K, Hesselbart A, Wendland J. ens draft genome sequence assembled into 9 large scaffolds and 13 minor scaffolds 2017. Draft genome sequence of (Ͻ20 kb). Saccharomycopsis fodiens CBS 8332, a necrotrophic mycoparasite with biocontrol potential. Genome Announc 5:e01278-17. https://doi.org/10.1128/genomeA.01278-17. accharomycopsis fodiens is a member of the sole genus in the family Saccharomy- Copyright © 2017 Junker et al. This is an open- Scopsidaceae (1). Previously, this genus gained attention due to the ability of access article distributed under the terms of Saccharomycopsis fibuligera to degrade starch by producing a suite of enzymes, includ- the Creative Commons Attribution 4.0 ing ␣-amylase, glucoamylase, ␤-glucosidase, and acid protease (2). Subsequently, International license. Address correspondence to Jürgen Wendland, S. fibuligera genes encoding these enzymes were used in heterologous hosts, such as [email protected]. Saccharomyces cerevisiae and Yarrowia lipolytica (3–7). This interest in using S. fibuligera K.J. and A.H. contributed equally to this article. also spurred genome sequencing of this yeast (8). A totally different aspect of Saccharomycopsis yeast biology is the ability to act as necrotrophic mycoparasites, killing other fungi via penetration pegs (9). The broad host range includes both ascomycetes and basidiomycetes, yeasts, and filamentous fungi (10). This broad host range apparently enables Saccharomycopsis predators to use their penetration pegs like Swiss army knives. Due to their potential use as biocontrol agents, we are interested in elucidating the molecular biology of their predation. Here, we report the draft genome sequence of S. fodiens. This strain was isolated in 1995 in Queensland, Australia, and was the first saccharomycete for which predacious behavior was described (11). With this draft genome sequence, we can now enter comparative genome biology of Saccharomycopsis species and provide genomic insight for strain improvements of fermentation traits. This knowledge will also fuel our understanding of the predatory behaviors of different Saccharomycopsis species. The draft genome sequence of S. fodiens (CBS 8332 ϭ NRRL Y-48786 ϭ UWOPS 95-697.4) we report here was determined using Illumina MiSeq paired-end read se- quencing. The S. fodiens strain was grown overnight at 30°C in rich medium (YPD, 1% yeast extract, 2% casein peptone, and 2% dextrose). DNA extraction and sequencing were carried out by LGC Genomics (Berlin, Germany). Two paired-end libraries were sequenced, producing 7,771,134 raw reads. These were quality processed and trimmed, resulting in 7,687,844 high-quality reads. The 250-bp paired-end library with short fragments produced 4,605,178 quality reads, and a further 3,082,666 high-quality reads were obtained from an 8-kb mate pair library. All quality-controlled reads were assem- bled using Bowtie 2 version 2.1.0. Initially, 90 contigs were assembled, with a total of

14,879,925 bp and an N50 of 508,788 bp. This assembly was refined by scaffolding these

Volume 5 Issue 46 e01278-17 genomea.asm.org 1 Junker et al.

Ͼ contigs into 9 scaffolds ( 20 kb) comprising 14,908,178 bp and an N50 of 2,606,857 bp. The average GC content in these scaffolds is 51.5%, which is remarkably high compared with that of other ascomycetous yeasts. Additionally, 13 scaffolds with fewer than 12 kb were generated. The longest scaffold is 2,724,259 in size. The large scaffolds were compared with those of the S. cerevisiae genome using BLASTX. This identified 4,725 hits (E value Ͻ 1e-10). BLASTX searches against the nonredundant database at NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi) generated a further 373 hits. As was observed for S. fibuligera, also S. fodiens lacks genes required for sulfate uptake and assimilation (8). Running tRNAscan on the scaffolds identified 162 tRNA genes (12). Accession number(s). This whole-genome shotgun project has been deposited in DDBJ/ENA/GenBank under the accession no. JNFV00000000. The version described in this paper is the first version, JNFV01000000.

ACKNOWLEDGMENTS This research was supported in part by the European Union, Marie Curie Initial Training Network Fungibrain (grant 607963). The S. fodiens strain was kindly provided by Marc-André Lachance.

REFERENCES 1. Suh SO, Blackwell M, Kurtzman CP, Lachance MA. 2006. Phylogenetics of 7. Lee CR, Sung BH, Lim KM, Kim MJ, Sohn MJ, Bae JH, Sohn JH. 2017. Saccharomycetales, the ascomycete yeasts. Mycologia 98:1006–1017. Co-fermentation using recombinant Saccharomyces cerevisiae yeast 2. Chi Z, Chi Z, Liu G, Wang F, Ju L, Zhang T. 2009. Saccharomycopsis strains hyper-secreting different cellulases for the production of cellu- fibuligera and its applications in biotechnology. Biotechnol Adv 27: losic bioethanol. Sci Rep 7:4428. https://doi.org/10.1038/s41598-017 423–431. https://doi.org/10.1016/j.biotechadv.2009.03.003. -04815-1. 3. Gurgu L, Lafraya Á, Polaina J, Marín-Navarro J. 2011. Fermentation of 8. Choo JH, Hong CP, Lim JY, Seo JA, Kim YS, Lee DW, Park SG, Lee GW, cellobiose to ethanol by industrial Saccharomyces strains carrying the Carroll E, Lee YW, Kang HA. 2016. Whole-genome de novo sequencing, beta-glucosidase gene (BGL1) from Saccharomycopsis fibuligera. Biore- combined with RNA-Seq analysis, reveals unique genome and physio- sour Technol 102:5229–5236. https://doi.org/10.1016/j.biortech.2011.01 logical features of the amylolytic yeast Saccharomycopsis fibuligera and .062. its interspecies hybrid. Biotechnol Biofuels 9:246. https://doi.org/10 4. Tang H, Hou J, Shen Y, Xu L, Yang H, Fang X, Bao X. 2013. High beta- .1186/s13068-016-0653-4. glucosidase secretion in Saccharomyces cerevisiae improves the efficiency 9. Lachance MA, Pang WM. 1997. Predacious yeasts. Yeast 13:225–232. of cellulase hydrolysis and ethanol production in simultaneous sacchar- https://doi.org/10.1002/(SICI)1097-0061(19970315)13:3Ͻ225::AID-YEA87 ification and fermentation. J Microbiol Biotechnol 23:1577–1585. https:// Ͼ3.0.CO;2-I. doi.org/10.4014/jmb.1305.05011. 5. Yu XJ, Chi Z, Wang F, Li J, Chi ZM, Madzak C. 2013. Expression of the acid 10. Lachance MA, Pupovac-Velikonja A, Natarajan S, Schlag-Edler B. 2000. protease gene from Saccharomycopsis fibuligera in the marine-derived Nutrition and phylogeny of predacious yeasts. Can J Microbiol 46: Yarrowia lipolytica for both milk clotting and single cell protein produc- 495–505. https://doi.org/10.1139/w00-021. tion. Appl Biochem Biotechnol 169:1993–2003. https://doi.org/10.1007/ 11. Lachance MA, Rosa CA, Carvajal EJ, Freitas LF, Bowles JM. 2012. Saccha- s12010-013-0118-1. romycopsis fodiens sp. nov., a rare predacious yeast from three distant 6. Casa-Villegas M, Marín-Navarro J, Polaina J. 2017. Synergies in coupled localities. Int J Syst Evol Microbiol 62:2793–2798. https://doi.org/10 hydrolysis and fermentation of cellulose using a Trichoderma reesei .1099/ijs.0.043109-0. enzyme preparation and a recombinant Saccharomyces cerevisiae strain. 12. Lowe TM, Eddy SR. 1997. tRNAscan-SE: a program for improved detec- World J Microbiol Biotechnol 33:140. https://doi.org/10.1007/s11274-017 tion of transfer RNA genes in genomic sequence. Nucleic Acids Res -2308-4. 25:955–964.

Volume 5 Issue 46 e01278-17 genomea.asm.org 2 Klara Junker | Prospecting a Predator

72 Prospecting a Predator | Klara Junker

Paper II

73 EUKARYOTES crossm

Draft Genome Sequence of Saccharomycopsis fermentans CBS 7830, a Predacious Yeast Belonging to the Saccharomycetales

Ana Hesselbart,a Klara Junker,a Jürgen Wendlanda,b aCarlsberg Research Laboratory, Yeast & Fermentation, Copenhagen, Denmark bVrije Universiteit Brussel, Functional Yeast Genomics, Brussels, Belgium

ABSTRACT Saccharomycopsis fermentans is an ascomycetous necrotrophic fungal pathogen that penetrates and kills fungal prey cells via targeted penetration pegs. Here, we report the draft genome sequence and scaffold assembly of this mycopara- site.

accharomycopsis is the sole genus of the family Saccharomycopsidaceae (1). One Smember of this genus, the starch-degrading yeast S. fibuligera, is known for its contributions to rice wine fermentation (2). The role of S. fibuligera as an enzyme producer also spurred interest in using this yeast for bioethanol production (3). Recently, whole-genome sequencing studies revealed that the genome of S. fibuligera encompasses seven chromosomes with about 18 Mb (4). Other members of the genus Saccharomycopsis, including S. fermentans, have been described as predacious yeasts that generate penetration pegs with which they kill other fungal cells utilizing prey cell content (5). S. fermentans (formerly classified as Arthroascus fermentans) was isolated in 1994 from the soil of a Taiwanese orchard and was shown to ferment glucose (references 6 and 7 and references therein). Saccharomycopsis species are natural auxotrophs for organic sulfur (8). In S. fibuligera, the absence of genes involved in sulfate assimilation has been observed (4). The usefulness of non-S. cerevisiae yeasts in fermentation and biotechnology de- pends on detailed characterization of the microbial genomes. Thus, additional draft genome sequences are required to not only provide species-specific markers for identification but also initiate pathway analyses and enable targeted strain improve- ments. Furthermore, a deeper understanding of the molecular biology of Saccharomy- copsis species—specifically of their predacious behavior—is warranted. To this end, we recently established the draft genome of another predacious Saccharomycopsis yeast, S. fodiens strain CBS 8332 (9). We confirmed that, just like S. fibuligera, S. fodiens also lacks the genes in the sulfate assimilation pathway. Here, we report the draft genome sequence of S. fermentans, obtained using Illumina MiSeq paired-end read sequencing. S. fermentans was grown in complex Received 20 November 2017 Accepted 27 medium (1% [wt/vol] yeast extract-peptone-dextrose [YPD], 2% [wt/vol] peptone, and November 2017 Published 11 January 2018 2% [wt/vol] dextrose) at 30°C with constant shaking. DNA extraction and sequencing Citation Hesselbart A, Junker K, Wendland J. were performed at LGC Genomics (Berlin, Germany). Two paired-end libraries were 2018. Draft genome sequence of Saccharomycopsis fermentans CBS 7830, a sequenced, producing 10,363,062 raw reads. These were quality trimmed, resulting in predacious yeast belonging to the 4,807,796 high-quality reads with the short fragment library and 4,521,208 high-quality Saccharomycetales. Genome Announc 6: reads with an 8-kb mate pair library. These were assembled using Bowtie2 version 2.1.0. e01445-17. https://doi.org/10.1128/genomeA .01445-17. The initial assembly generated 160 contigs with a total content of 14,266,439 bp and Copyright © 2018 Hesselbart et al. This is an a contig N50 of 265,725 bp. The contigs were then further assembled into 33 scaffolds open-access article distributed under the terms of the Creative Commons Attribution 4.0 harboring 14,461,413 bp, with an N50 of 2,146,288 bp. The longest scaffold contains International license. 3,513,907 bp, with 13 scaffolds larger than 20 kb. The overall GC content is 35.1%. Address correspondence to Jürgen Wendland, For a draft annotation of the nuclear genome, open reading frames (ORFs) with a [email protected]. size of Ͼ300 nt were predicted and compared using blastx to translated proteins in the

Volume 6 Issue 2 e01445-17 genomea.asm.org 1 Hesselbart et al.

Saccharomyces cerevisiae genome. In all, 5,917 nonoverlapping ORFs were detected in S. fermentans, and of those, 3,882 genes produced hits with S. cerevisiae. An additional blast search against other organisms in the nonredundant database at NCBI (https:// blast.ncbi.nlm.nih.gov/Blast.cgi) generated a further 263 hits (E values, Ͻ10Ϫ10). We could also verify the absence of genes required for sulfate assimilation in S. fermentans. Additionally, we identified 149 tRNA genes in the S. fermentans genome using tRNAscan-SE (10). Accession number(s). This whole-genome shotgun project has been deposited in DDBJ/ENA/GenBank under the accession no. JNFW00000000. The version described in this paper is the first version, JNFW01000000.

ACKNOWLEDGMENTS This research was supported in part by the European Union Marie Curie Initial Training Network FungiBrain (grant 607963). The S. fermentans strain was obtained from the Westerdijk Fungal Biodiversity Institute.

REFERENCES 1. Suh SO, Blackwell M, Kurtzman CP, Lachance MA. 2006. Phylogenetics of 6. Lee C-F, Lee F-L, Hsu W-H, Phaff HJ. 1994. Arthroascus fermentans, a new Saccharomycetales, the ascomycete yeasts. Mycologia 98:1006–1017. yeast species isolated from soil in Taiwan. Int J Syst Bacteriol 44:303–307. 2. Chi Z, Chi Z, Liu G, Wang F, Ju L, Zhang T. 2009. Saccharomycopsis https://doi.org/10.1099/00207713-44-2-303. fibuligera and its applications in biotechnology. Biotechnol Adv 27: 7. Naumov GI, Naumova ES, Smith MT, de Hoog GS. 2006. Molecular- 423–431. https://doi.org/10.1016/j.biotechadv.2009.03.003. genetic diversity of the ascomycetous yeast genus Arthroascus: Arthroas- 3. Farh ME, Cho Y, Lim JY, Seo JA. 2017. A diversity study of Saccharomy- cus babjevae sp. nov., Arthroascus fermentans var. arxii var. nov. and copsis fibuligera in rice wine starter nuruk, reveals the evolutionary geographical populations of Arthroascus schoenii. Int J Syst Evol Micro- process associated with its interspecies hybrid. J Microbiol 55:337–343. biol 56:1997–2007. https://doi.org/10.1099/ijs.0.64301-0. https://doi.org/10.1007/s12275-017-7115-y. 8. Lachance MA, Pupovac-Velikonja A, Natarajan S, Schlag-Edler B. 2000. 4. Choo JH, Hong CP, Lim JY, Seo JA, Kim YS, Lee DW, Park SG, Lee GW, Nutrition and phylogeny of predacious yeasts. Can J Microbiol 46: Carroll E, Lee YW, Kang HA. 2016. Whole-genome de novo sequencing, 495–505. https://doi.org/10.1139/w00-021. combined with RNA-Seq analysis, reveals unique genome and physio- 9. Junker K, Hesselbart A, Wendland J. 2017. Draft genome sequence of logical features of the amylolytic yeast Saccharomycopsis fibuligera and Saccharomycopsis fodiens CBS 8332, a necrotrophic mycoparasite with its interspecies hybrid. Biotechnol Biofuels 9:246. https://doi.org/10 biocontrol potential. Genome Announc 5:e01278-17. https://doi.org/10 .1186/s13068-016-0653-4. .1128/genomeA.01278-17. 5. Lachance MA, Pang WM. 1997. Predacious yeasts. Yeast 13:225–232. 10. Lowe TM, Eddy SR. 1997. tRNAscan-SE: a program for improved detec- https://doi.org/10.1002/(SICI)1097-0061(19970315)13:3Ͻ225::AID tion of transfer RNA genes in genomic sequence. Nucleic Acids Res -YEA87Ͼ3.0.CO;2-I. 25:955–964.

Volume 6 Issue 2 e01445-17 genomea.asm.org 2 Klara Junker | Prospecting a Predator

76 Prospecting a Predator | Klara Junker

Paper III

77 Multi-omic identification of genes involved in the necrotrophic mycoparasitism exerted by the yeast Saccharomycopsis schoenii

Klara Junker1, Anna Chailyan1, Ana Hesselbart1 and Jürgen Wendland2 1. Yeast & Fermentation, Carlsberg Research Laboratory, DK-1799 Copenhagen, Denmark 2. Functional Yeast Genomics, Vrije Universiteit Brussel, BE-1050 Brussels, Belgium

Abstract Pathogenic yeast and fungi are an increasing global healthcare burden, but discovery of novel antifungal agents is slow. The mycoparasitic yeast Saccharomycopsis schoenii was recently demonstrated as able to kill the emerging multi-drug resistant yeast pathogen Candida auris. However, the molecular mechanisms involved in the predatory activity of S. schoenii have not been explored. To this end, we de novo sequenced, assembled and annotated a draft genome of S. schoenii. We confirmed that Saccharomycopsis yeasts belong to the CTG clade and that S. schoenii translate the CTG codon to serine instead of leucine. Further, we confirmed an absence of all genes from the sulfate assimilation pathway in the genome of S. schoenii, and detected the expansion of several gene families, including aspartic proteases. Using Saccharomyces cerevisiae as a model prey cell, we honed in on the timing and nutritional conditions under which S. schoenii kills prey cells. We found that a general nutrition limitation, not a specific methionine deficiency, triggered predatory activity. Nevertheless, we observed dramatic responses to methionine deprivation, that were alleviated when S. cerevisiae was available as prey, and therefore postulated that S. schoenii acquired methionine from its prey cells. During predation, S. schoenii highly upregulated and translated aspartic protease genes, probably used to break down prey cell walls. With these fundamental insights into the predatory behavior of S. schoenii, we open up for further exploitation of this yeast as a biocontrol yeast and/or source for novel antifungal agents.

Introduction The burden of fungal pathogens of plants, animals and humans is increasing at an unprecedented rate1. Fungal diseases currently affect more than one billion people and cause over 1.6 million deaths per year2. The increase of immunocompromised patients in hospital has created a breeding ground for multi-drug resistant fungal infections, most severely exemplified by the yeast Candida auris3. In parallel, the usefulness of fungi range from being sources of fundamental antimicrobial agents to being used as live agricultural biocontrol agents4,5. Recent advances in next generation sequencing (NGS) and multi-omic methods have revolutionized the overall understanding of both pathogenic and beneficial fungi, by have enabling more in-depth studies of fungal parasitic systems and opening up the exploration of novel antifungal mechanisms from lesser-known fungi6–10.

Several mycoparasitic filamentous fungi are used as biocontrol agents against plant pathogenic fungi, and are well studied in their abilities to physically attack and even kill other fungi11–13. Necrotrophic mycoparasite yeasts in the Saccharomycopsis clade are the only yeasts known to physically attack and kill other fungi, but few studies have explored their potential as biocontrol agents14–16. Saccharomycopsis yeasts invade fungal prey cells with small haustoria-like penetration pegs, which ultimately kills the prey cells14. S. schoenii is one of the most efficient predators in this clade and was recently demonstrated to attack and kill several clinical isolates of pathogenic Candida species in vitro, including multi-drug resistant isolates of C. auris17. Molecular characterization of the predatory behavior of S. schoenii is essential for further exploration of this yeast a unique biocontrol agent.

78 Saccharomycopsis yeasts are closely related to Ascoidea rubescens18 and Wickerhamomyces anomalus19 and were recently proposed to belong to a subclade of the CTG clade yeasts20. Virtually all organisms translate the CTG codon to leucine, typically using a tRNALeu(CAG), but yeasts in the CTG clade, such as Candida albicans, use a modified tRNASer(CAG) to translate CTG to serine21,22. Three Saccharomycopsis yeasts were found to harbor both a tRNASer(CAG) and a tRNALeu(CAG) in their genome, still, Saccharomycopsis capsularis only translated the CTG codon to serine.

Organic sulfur is essential element for all types of cells, and one of the two sulfur containing amino acids, methionine, is also the most energetically expensive amino acid for yeasts to produce. The vast majority of microorganisms, including yeast, are able to take up sulfur in the form of inorganic sulfate and reduce it in the sulfate assimilation pathway23, but yeasts in the Saccharomycopsis clade share a rare inability to assimilate inorganic sulfur24,25. We recently reported that in Saccharomycopsis fodiens26 and Saccharomycopsis fermentans27, the genomic basis for this rare organic sulfur auxotrophy is a complete absence of all genes in the sulfate assimilation pathway.

Specific auxotrophies are often involved in parasitic behavior in both yeast and fungi28,29. The human pathogenic yeast Candida glabrata is a nicotinic acid auxotroph, and when nicotinic acid is absent in the urinary tract of its host, such as when a catheter is used, C. glabrata becomes virulent and colonizes host tissue30. Similarly, the rust fungi Puccinia graminis have deficiencies in both the nitrate and sulfate assimilation pathways, which might have enabled or adapted it to life as an obligate biotroph31. In preliminary studies, Lachance et al. demonstrated that some Saccharomycopsis species appeared to change their predatory behavior depending on methionine availability, but the results were inconclusive24.

In fungal parasites, gene loss often come hand in hand with gene and genome expansion32,33. Transposable elements (TE) for instance, that shape eukaryotic genomes, are often expanded in biotrophic fungi28,34,35. TEs such as retrotransposons have ancient RNAi-mediated mechanisms, can be involved in genome defense and some appear activated during stressful conditions36. Gene expansion is also often coupled to genes that enable parasitic behavior33. For instance, the genomes of Trichoderma species harbor expanded gene families also include encoding cell wall degrading enzymes, such as chitinases, glucanases and protases10,37. Several Trichoderma species are used as biocontrol agents, and upregulate and release proteases prior to their antagonism of plant- pathogenic fungi, whereas chitinases and glucanases are upregulated during active mycoparasitism5.

Saccharomycopsis species have been successfully trialed as agricultural or food biocontrol agents; Saccharomycopsis schoenii against plant pathogens on oranges15, and Saccharomycopsis fibuligera against toxic molds on speck16. The genome of S. fibuligera was recently sequenced, and its transcriptomic responses to organic sulfur starvation was studied38. To our knowledge, only one successful genetic manipulation of a Saccharomycopsis yeast has been achieved, but the disruption of a single protease gene in a mutagenized S. fibuligera strain was not tested in the context of predatory behavior39.

Here we hypothesized that by taking a genomic, transcriptomic and proteomic approach, and coupling it with quantifiable predatory behavior, we could identify and separate major genetic pathways involved in starvation and predation responses in S. schoenii, with little bias. We further hypothesized that the need for organic sulfur compounds, especially methionine, play a central role in the predatory behavior of S. schoenii. Our aim was to expose the genetic toolkit of S. schoenii, particularly in regards to its predatory behavior, with the purpose of gaining fundamental insights into its potential usefulness as a biocontrol yeast.

79 Materials and methods Strains and culture conditions: Wild-type Saccharomycopsis schoenii (CBS 7425, CBS-KNAW collection, Utrecht, Netherlands) was provided by Marc-André Lachance. Saccharomyces cerevisiae strain BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0, EUROSCARF, Scientific Research and Development GmbH, Oberursel, Germany), S. cerevisiae with hygromycin resistance (BY4741, HSP104::gfp::hyg, Carlsberg Research Laboratory, Denmark) and S. cerevisiae H4-GFP (BY4741, H4- GPF, Carlsberg Research Laboratory, Denmark) were used as model prey cells as indicated. Yeast cells were cultured to log phase in “Standard media” (YPD; 10 g/L yeast extract, 20 g/L Bacto peptone, 20 g/L glucose), at 30°C, rotating. S. schoenii and S. cerevisiae were subsequently cultured alone or co- cultured as indicated on solid YPD, “Nutrient limited media” (Complete Synthetic Media [CSM]; 0.79 g/L Complete Supplement Mixture, 6.7 g/L Yeast Nitrogen Base (YNB) w/o amino acids with ammonium sulfate, 20 g/L glucose), “Methionine deprived media” (CSM-Met; 0.75 g/L Complete Supplement Mixture -Methionine, 6.7 g/L YNB w/o amino acids with ammonium sulfate 20 g/L glucose) or “Starvation media” (Synthetic Defined [SD]; 20g/L glucose 6.7 g/L YNB w/o amino acids with ammonium sulfate). YPD is a rich standard media, providing nutrients as various complex molecules. CSM is made up of only the amino acids, vitamins, trace elements and salts that are essential for prototrophic S. cerevisiae. CSM-Met is CSM without methionine, through which any specific responses due to a methionine deprivation can be studied. SD media is a yeast minimal media, that only contain ammonium sulfate as a nitrogen source, and vitamins, trace elements and salts, which necessitates de novo synthesis of all amino acids. All media were solidified with 10 g/L agarose for microscopy and with 20 g/L bacto-agar for cultures for genome preparation, protein preparation and RNA preparation.

Microscopy: Differential interference contrast (DIC) and fluorescent microscopy was performed with a Zeiss Axio Imager M2 Microscope, using a halogen lamp for transmitted-light and UV for fluorescent imaging, and the software Metamorph for image acquisition. Predation quantification analyses, performed in duplicates, were performed by initially seeding co-cultured cells on several agarose pads with appropriate media, imaging new slides at each hour for six hours. Three frames, representative of the whole slide were captured. Locations where cells were growing on top of each other were selected against, in favor of locations where individual cells could be distinguished. S. cerevisiae prey cells were scored as live/non-predated (regular, round morphology), dead/non- predated (flattened, shrunken, no physical contact with S. schoenii cells) or dead/predated (vacuolarized or flattened, shrunken and in physical contact with S. schoenii cells). S. schoenii cells were scored as live (regular, full morphology) or dead (shrunken and/or flattened). The software FIJI/ImageJ40–42 was used for image processing and analysis. For movies, drift in frames was corrected with the macro NMS fixTranslation v143 and the plugin Image Stabiliser44. Cells were counted with the plugin Cell Counter and area measurements were performed with the elliptical selection tool.

Genome sequencing: For genomic sequencing with Illumina MiSeq, DNA extraction and sequencing was performed by LGC Genomics (Berlin, Germany), generating a 250-bp paired-end library and a 8- kb mate-pair library. For genomic PacBio sequencing, DNA was prepared using the QIAGEN Blood & Cell Culture DNA Maxi Kit with a QIAGEN Genomic-tip 500/G (QIAGEN Gmbh, Hilden, Germany) according to the manufacturer’s protocol. Subsequent PacBio sequencing, based on Single Molecule Real-Time (SMRT) technology, was performed by DNA Link Inc. (Seoul, Republic of Korea), using kits and reagents from Pacific Biociences. Quality controlled genomic DNA was used to prepare the SMRTbell library and fragments smaller than 20kb were removed using the Blue Pippin Size selection system. Polymerase-SMRTbell-adaptor complexes were loaded into four SMRT cells and sequenced using C4 chemistry (DNA sequencing Reagent 4.0). 240-minute movies were captured for each SMRT

80 cell using the PacBio RS II sequencing platform, generating one set of raw sequencing subreads per SMRT cell. Raw genome sequencing reads deposited at the European Nucleotide Archive (ENA), Primary Accessoion # PRJEB23925.

Genome assembly: Raw PacBio sequencing subreads were filtered on quality and length, using the RS_subreads protocol in PacBio’s SMRT-Portal software, run through Amazon Web Services Inc (Seattle, USA) and exported as fastq files. Filtered subreads were batch error-corrected using the tool “Correct PacBio Reads (beta)” and the de novo assembled using the tool “De Novo Assemble PacBio Reads (beta)” in CLC genomics workbench v.9.5 (QIAGEN Aarhus, Aarhus C, Denmark), generating 78 contigs. At this stage the assembly was quality controlled with QUAST analysis45. The 78 contigs were then further polished using Illumina MiSeq generated 250-bp paired-end library and 8-kb mate-pair library, resulting in a final draft genome assembly of 29 contigs with 14.3 mega base pairs. The draft genome assembly is available as Supplementary Material 1.

Genome Annotation: Open Reading Frames (ORFs) were predicted as sequences with an AUG start codon, spanning >300 bp. ORFs translated with the Alternative Yeast Nuclear Code (AYNC) were subjected to a cloud-based blastx against the non-redundant protein database (nr), deposited by 06.06.2017, using the plugin Blast2GO46 inside the software CLC genomics v9.5. To functionally annotate the S. schoenii ORFs, all ORFs were subjected to blastx strategies against proteins from S. cerevisiae, C. albicans or both, and homology was inferred as a hit with a bit score >5547. Overlapping ORFs were removed manually, resulting in 4,660 annotated genes. tRNA genes were identified using tRNAScan-SE48. The ORF identification and our annotation of our S. schoenii genome is available as Supplementary Material 2 and Supplementary Data 1.

Proteomic analysis: For the proteomic analysis, S. schoenii cells were cultured for three hours under three different conditions; on YPD media alone (“standard”), on SD media alone (“starvation”) or on SD media together with equal numbers of S. cerevisiae (“predation”). Cells were pelleted and flash frozen in liquid nitrogen. Subsequent proteomic analysis was performed by Phylogene (Bernis, France). Proteins were extracted, purified and concentration of lysates was determined by Pierce 660 nm assay. Peptides were prepared according to the FASP method, ultrafiltrated, reduced, alkylated and digested by trypsin. Peptides were purified by SPE chromatography and peptide concentration was determined using the BCA method. Liquid chromatography-tandem mass spectrometry (LC- MS/MS) measurements were done in triplicates. Chromatography was performed using Ultimate 3000 (Dionex) and data acquired using the Q-Exactive Plus (Thermo) mass spectrometer. Proteins were identified using SEQUEST-HT algorithm against two custom databases with S. schoenii ORFs translated with Standard Code or AYNC respectively, and when S. schoenii and S. cerevisiae were co- cultured, also against a database containing reference proteome of S. cerevisiae mined from UNIPROT. Data were processed using Minora and feature mapper for Proteome Discoverer 2.2 software. Statistical analyses were performed by using Precursors Ions quantifier node for Proteome Discoverer 2.2 software. Abundances of S. schoenii peptides and proteins were measured against the abundances of S. schoenii peptides and proteins when cultured alone on YPD. Only hits identified to non-overlapping ORFs with a homolog in other yeasts were analyzed further. The raw mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium49 via the PRIDE partner repository50 with the dataset identifier PXD008453. Protein identification, CTG translation and quantification linked to predicted ORFs is available in Supplementary Data 1.

Transcriptomic analysis: For the transcriptomic analysis, 1x108 S. schoenii cells were cultured alone or with 1x108 S. cerevisiae cells for three hours on YPD, CSM, CSM-Met or SD media. Each of the eight conditions were performed in three biological replicates. Cells were pelleted, washed and flash frozen in liquid nitrogen. RNA was extracted using the RiboPure™-Yeast Kit (Ambion, Thermo Fisher

81 Scientific) and DNAse treated as per instructions. BGI Europe A/S, (Copenhagen, Denmark) performed quality control, constructed polyA-selected strand-specific transcriptomic libraries, and sequenced the samples at 10MB clean reads/sample using Illumina Hiseq4000 PE100. Raw reads from each triplicate were pooled, mapped to S. schoenii ORFs and normalized by total reads per sample using CLC genomics workbench v.9.5. Transcription was expressed as mean normalized expression values. To exclude noise, genes with transcription values of >100 and a 10-fold upregulation during any of the conditions, compared to S. schoenii on YPD alone, were selected. Only hits mapped to non-overlapping ORFs with a homolog in other yeasts were analyzed further. Raw transcriptomic data is deposited at ENA under Primary accession # PRJEB23926. Normalized mean expression values of predicted ORFs are available in Supplementary Data 1.

GO term analysis: Subsets of genes and proteins from the proteomic and transcriptomic analyses were subjected to Gene Ontology (GO) term category analysis using FungiFun251. Only genes and proteins with a homologous S. cerevisiae or C. albicans gene were used. Proteins with a 10-fold abundance during starvation (SD) or predation (SD + S.c.) conditions, compared to standard conditions (YPD) were selected. Proteins were further divided as enriched during any of three conditions; “Starvation” (>10-fold higher abundance during starvation compared to predation conditions), “Predation” (>10-fold higher abundance during predation compared to starvation conditions) or “Predation + Starvation” (the rest). For the subsets of transcribed genes, genes with a total experimental range >100 and with a 10-fold upregulation during any of the seven experimental conditions, compared to S. schoenii on YPD alone, were selected. Upregulated genes were curated as primarily upregulated during three conditions; “Nutrient limitation” (>10-fold upregulated when on CSM alone, compared to when on YPD alone), “Methionine deprivation” (>2-fold upregulated when on CSM-Met alone, compared to when on CSM alone) and “Predation” (>2-fold upregulated when co-cultured with S. cerevisiae on any media, compared to when cultured alone on the same media). For the FungiFun2 analysis, the corresponding C. albicans homolog gene name for each gene was entered, using a background list of all S. schoenii genes with their C. albicans homolog. Significance of over-representation (enrichment) of direct GO terms was calculated using a Hypergeometric distribution test and adjusted with a Benjamini-Hochberg procedure (FDR correction).

Results Contact-dependent mycoparasitism enables S. schoenii to eliminate S. cerevisiae S. schoenii is an efficient predator yeasts in the Saccharomycopsis clade and we first validated that S. cerevisiae was susceptible to predation by S. schoenii24. Just like S. fibuligera, S. schoenii is hygromycin sensitive and we therefore chose a hygromycin (hyg) resistant S. cerevisiae strain (HSP104::hyg) as prey. We cultured S. schoenii and S. cerevisiae HSP104::hyg separately or in co- culture at equal ratios on starvation media (SD) for three days. Both S. schoenii and S. cerevisiae remained viable and able form new colonies on standard media (YPD), whereas only S. cerevisiae HSP104::hyg could form colonies on hygromycin media (YPD + hyg) (Suppl. Figure 1). After being co- cultured with S. schoenii, an complete absence of S. cerevisiae colonies on YPD + hyg indicated that S. cerevisiae had been eliminated by S. schoenii.

In order to detect when and how S. schoenii eliminates S. cerevisiae, we co-cultured S. schoenii with a histone 4-GFP labelled S. cerevisiae strain (H4-GFP) on SD media and monitored cellular interactions with live cell microscopy (Figure 1A and Suppl. Movie 1). Upon physical contact with prey cells, predation by S. schoenii triggered the S. cerevisiae cells to first vacuolarize, then to shrink and lose their H4-GFP label (Figure 1A). Non-predated cells out of physical contact with S. schoenii, did not shrink or lose their H4-GFP label indicating that the physical contact between S. schoenii and prey

82 cells was required in killing the prey cells. We also observed a significant change in size between predated and non-predated cells (Figure 1B). In addition, we noticed that predated cells were unable to bud, whereas non-predated cells would bud continuously which supports previous observations that S. schoenii that efficiently kills the model prey cell S. cerevisiae through contact-dependent mycoparasitism.

The draft genome of S. schoenii In order to provide a S. schoenii genome with both good coverage over long repetitive stretches and with high fidelity, we de novo sequenced the genome and assembled scaffolds with long >20 kb PacBio reads, and subsequently polished the scaffolds with short 250-bp paired-end library and 8-kb mate-pair Illumina Miseq reads (Table 1A). We identified 7,999 open reading frames in the draft genome of S. schoenii. The non-redundant (nr) protein database contains protein sequences from all species deposited in GenBank, and we ran all ORFs through a cloud-based blastx strategy against the non-redundant (nr) protein database (Table 1B). This provided us with a match for each ORF to the closest species, of those deposited in the nr database to date, permitting us to validate the close relationships between Saccharomycopsis, A. rubescens and W. anomalus (Figure 2A). Several, but not all, of the most similar yeast species were other CTG clade members. The number of S. schoenii ORFs with functional homologs to S. cerevisiae or C. albicans are listed in Table 1B. After we manually removed overlapping ORFs, the total number of genes with homologs came to 4,660. More ORFs were homologous to genes from C. albicans than to genes from S. cerevisiae, suggesting a closer relationship with C. albicans than S. cerevisiae. 263 non-overlapping ORFs were at least 100 amino acids long but lacked homologs from any of the blastx strategies.

Just as we were unable to detect any genes in the sulfate assimilation pathway in the genomes of S. fodiens26 and S. fermentans27, the genes for sulfate uptake and reduction were all absent in the draft genome of S. schoenii (Table 1C and Figure 6B). Like the majority of yeasts52, S. schoenii, also lacked genes in the nitrate assimilation pathway (YNT1, YNR1 and YNI1). There were several highly expanded gene families in the S. schoenii draft genome, including aspartic proteases (YPS3), transposable elements, permeases and other cell wall related genes (Table 1D). The expanded genes SLA2 and YJR098C typically flank mating type loci in Saccharomycetaceae53 and appear to flank similar loci in S. schoenii (Figure S2).

S. schoenii harbors two tRNA(CAG) genes and translate CTG to serine We were able to identify two tRNA(CAG) genes in the draft genome of S. schoenii, by running the draft genome sequence through tRNAscan-SE 2.048 (Figure 2B). We were also able to identify two tRNA(CAG) genes in the genome of S. fermentans, but only one tRNA-CAG gene in the S. fodiens genome26,27. We aligned these Saccharomycopsis tRNA(CAG) genes with serine and leucine tRNA genes from S. cerevisiae and C. albicans, as well as with tRNA(CAG) genes from other CTG clade yeasts, including yeasts from the recently proposed CTG subclades (Figure 2C)54,55. The tRNA(CAG.1) genes S. schoenii and S. fermentans were identical and aligned close to other Serine-tRNAs, whereas the also identical tRNA(CAG.2) genes aligned somewhat closer to leucine-tRNAs. The S. schoenii and S. fermentans tRNA(CAG.1) bore several hallmarks of the well described tRNASer(CAG), from C. 1 56,57 albicans, including a m G37 Leu-identity element and a (G73) discriminator base (Figure 2B) . Similar to C. tropicalis, neither S. schoenii or S. fermentans tRNA-CAG.1 featured the leucylation-lowering

G33. The tRNA(CAG.2) from S. schoenii and S. fermentans bore no serine-identity elements and were also less similar to other leucine-tRNA.

To find out how S. schoenii translated the CTG codon, we analyzed the S. schoenii proteome. Of the identified proteins that harbored a CTG codon, 450 had peptidic evidence of >1 CTG codon being translated to serine, whereas 13 were mapped to leucine (Suppl. Table 1). In C. albicans, around 3%

83 of CTG codons are “mistranslated” to leucine, which incidence might be correlated with virulence58. To test whether or not S. schoenii varied its translation of the CTG codon during starvation and/or predation conditions, we cultured S. schoenii either alone under starvation (SD) or predation (SD + S. cerevisiae) conditions for three hours. During starvation, 14 proteins had CTG positions translated to leucine, whereas during predation, only 5 proteins had CTG positions translated to leucine (Table 2B). Three proteins (CET1, ZIM17 and ARO8) had peptides supporting translation of CTG to both serine and leucine under any one condition. Only one protein, CAB3, was translated only to serine during one condition (YPD) and leucine under another (SD).

S. schoenii proteins involved in Starvation and Predation Our next aim was to identify, quantify and categorize S. schoenii proteins present during starvation (SD) and/or predation (SD + S. cerevisiae) conditions, relative to standard (YPD) conditions. Protein and peptide concentrations were measured and analyzed and a protein abundance was expressed as the ratio of protein concentration compared to standard (YPD) conditions. We manually curated all proteins with at least 10-fold increase into subsets as relevant to either Starvation, Predation or both conditions (Figure 3A). To identify functionally meaningful enrichment patterns, we subjected the proteins in each subset to a GO term category analysis, using the online resource FungiFun2, against C. albicans proteins51. During Starvation, proteins involved in catabolic processes (CIS2, DUG3 and DUR1,2) as well a protein regulating of sulfur metabolic processes (MET30) were enriched. During Predation, several cell wall related proteins (CHT3, CRH11, RNY11, SAP1, SAP2, SAP6, SIM1, PGA4 and PHO112) were enriched (Figure 3B). During both starvation and predation, proteins in the biosynthesis of methionine (MET2, MET15 and STR3) were enriched, as well as transporters (DAL7, FEN2, SEO1 and TNA1), carnitine (CNT3 and CAT2) and fatty acid catabolic (ICL1 and CAT2) related proteins.

Nutrient limitation is the main trigger for predation To quantify predatory activity, as a consequence of time and nutritional conditions, with specific regards to the absence or presence of methionine, we set up a microscopy-based predation assay. S. schoenii was able to attack and kill S. cerevisiae when co-cultured on all different nutritional conditions, including on the nutrient rich Standard media (Figure 1C). On Starvation media, S. schoenii nearly eliminated S. cerevisiae after 6 hours of co-culture. Non-predated/dead prey cells made up < 0.01% of all prey cells at any time or condition. When cultured alone, neither S. schoenii or S. cerevisiae cells died at any significantly different rate than when in co-culture (data not shown).

To tease out if the absence of methionine would have any effect on the predatory propensity of S. schoenii, we co-cultured S. schoenii and S. cerevisiae on CSM and CSM-Met. CSM and CSM-Met are both nutrient limited media made up primarily of sugars and essential amino acids, which excludes cysteine, and differ only in the presence or absence of methionine. We were unable to detect any difference in the propensity or speed at which S. schoenii predated on S. cerevisiae in regards to the specific absence or presence of methionine, using this setup (Figure 1C). Instead, the greatest difference in predation activity was seen between YPD and CSM. Proliferation of S. cerevisiae when cultured alone, was not significantly affected between YPD and CSM during the 6 hours assayed (data not shown), suggesting that the increased predatory activity was unlikely to primarily be a consequence of a less fit prey cell, but instead a response from S. schoenii to nutrient limited conditions.

S. schoenii respond drastically to methionine deprivation To identify which S. schoenii genes involved in responses to nutrient limitation, methionine deprivation and starvation and to subsequently isolate the genes specifically involved in predation responses, we performed genome-wide transcriptomic analyses. We selected all genes with a 10-fold

84 upregulation during any of the eight conditions, compared to S. schoenii on YPD alone and manually curated most of the upregulated genes into three subsets; genes upregulated mainly under conditions of “Nutrient limitation”, “Methionine deprivation” or “Predation” (Figure 4A)

To extract functionally meaningful enrichment patterns in the subsets of highly upregulated genes, we used the online resource FungiFun2, with a C. albicans gene nomenclature51 (Figure 4B). During Nutrient limitation conditions, GO categories such as carnitine and acetyl-CoA metabolism (CTN3, CAT2 and CRC1), anion and cation transport (DAL8, VHT1 and FGR2) and the Golgi apparatus (SGA1, GAP2, VHT1) were highly enriched, suggesting a focus on peroxisome energy generation and membrane transport facilitation59,60. During Methionine deprivation, membrane transport (DAL7, YCT1, TNA1, FEN2, SEO1, GPT1, C1_10710C, MUP1, HGT19, TPO3, DUR4 and JEN2) GO terms were highly enriched, together with methionine biosynthesis (CYS1, MET17, MET2 and CYS3) and oxidation-reduction processes (C1_01190C, ADI1, MXR1, C2_01540W, ADH4, C2_09850C, GRE3, ADH6, PRX1 and C7_03350C) suggesting, in concert, extensive scavenging and salvaging efforts of methionine and other sulfur compounds (Figure 4B).

S. schoenii probably acquire methionine from prey cells Interestingly, we noticed that the majority of S. schoenii genes that were upregulated in the Methionine deprivation subset were downregulated at least twofold when S. cerevisiae was available as prey (Figure 4A and 6A). When we looked at the genes involved in methionine biosynthesis and uptake, all genes leading up to the methyl cycle and methionine, except SAH1, MET6 and SAM1, were at least twofold upregulated when methionine was absent, suggesting increased need to scavenge and salvage of methionine and other sulfur compounds (Figure 6B). SAH1 was the sole gene downregulated gene when methionine was absent. MET6 and SAH1 stood out as being upregulated at least twofold when prey was present, suggesting that precursors needed to operate the for the methyl cycle might have been acquired during predation.

S. schoenii upregulate aspartic proteases and glucanases during predation During Predation, GO terms categories relating to the cell wall surface (XOG1, ENG1, SIM1, PRY2, MP65, CRH11, GAS1, SAP6, SAP1, SAP2, CHT3 and ALS9), were significantly enriched (Figure 5B). The most outstanding gene family was the secreted aspartic proteases (SAP) genes, homologous to yapsin (YPS) genes in S. cerevisiae. The aspartic protease genes were associated with several GO term categories, such as protein metabolic and catabolic process, signal peptide processing, aspartic-type endopeptidase activity and pathogenesis. Indeed, S. schoenii YPS3 genes were only expressed when S. cerevisiae was present as prey, and correlated with predation efficacy (Figure 5A). Similarly, overexpression of the glucanases MP65, XOG1 and ENG1, the glycosidases GAS1 and CRH11 the chitinase CHT3 also correlated with predation efficacy, but these were also upregulated, albeit to a lesser degree, when prey was absent (Figure 5B). Interestingly, two transposable element genes, associated with DNA integration, C4_03230C and POL99, were significantly enriched during predation (Figure 4B).

Discussion Yeasts in the Saccharomycopsis clade are potent necrotrophic mycoparasites of other yeasts, and S. schoenii has a specific potential as antifungal biocontrol agents against human yeast pathogens, such as multidrug-resistant C. auris. However, the molecular mechanisms involved in the predatory behavior has not been described for any Saccharomycopsis species. Identifying the genetic basis of their mode of action will facilitate both our basic understanding of their unique biology as well as enable prospecting of novel antifungal molecules.

85 In this study, we integrated quantitative live cell microscopy assays with genomic, transcriptomic and proteomic approaches to identify genes and proteins that are overexpressed by S. schoenii during its predation of the model prey cell S. cerevisiae. We found high copy numbers of aspartic proteases in the S. schoenii genome, consistent with conclusions that mycoparasitic fungi typically harbor major gene expansion related to their parasitism12. We found that substantial overexpression of four aspartic protease genes correlated with predatory activity in S. schoenii, analogous to overexpression of SAP genes during C. albicans pathogenesis61. Surprisingly, general nutrient limitation, and not a specific presence or absence of methionine, appeared to be the main trigger predatory activity in S. schoenii. However, the specific removal of methionine did not go unnoticed, as it triggered major transcriptional responses in S. schoenii, including upregulation and translation of MET30. Met30p is part of a ubiquitin ligase that senses methionine and S-Adenosyl methionine (SAM) availability to cell cycle control and transcriptional response62–64. In S. schoenii the dual role of Met30p is might therefore be to both increase protein degradation by the proteasome and to stop cell cycle progression until cellular levels of methionine are sufficiently high again. Interestingly, methionine specific responses in S. schoenii were alleviated when prey cells were present, suggesting methionine might be acquired in some form from prey cells. In comparison, the fungi Puccinia striiformis, also deficient in sulfate uptake, specifically upregulates S-methylmethionine permease in their haustoria during plant infection8.

Our finding that S. schoenii has lost all genes in the sulfate assimilation fit with the rare inability of Saccharomycopsis yeasts to assimilate sulfate24. While no other yeasts are unable to take up sulfate, three of the top ten most important parasitic filamentous fungi, Puccinia spp., B. graminis and Melampsora lini share this sulfate assimilation deficiency65. In addition, even parasites with functional sulfate assimilation, such as Trichoderma species ,increase sulfur metabolism during its mycoparasitic activity66. This points to a convergent evolution of the ability of some yeast and fungi to obtain sulfur from other sources than sulfate, probably associated with their parasitic abilities. Since methionine is the most energetically costly amino acid to biosynthesize67 and several metabolites in the sulfate reduction pathway are toxic23, it can be speculated that organic sulfur compounds is a top bounty for parasites. In this study, S. schoenii was able to nearly eliminate S. cerevisiae after just six hours of co-culture under nutrient limited conditions, demonstrating its potency as a prospective biocontrol agent. However, we had not anticipated that S. schoenii would predate on S. cerevisiae on nutrient rich media, pointing at S. schoenii being facultative parasites not only for survival during stressful conditions, but also to actively eliminate competitors.

Saccharomycopsis yeasts were recently reassigned as a subclade in the CTG clade of yeasts and we confirmed this with our genomic and proteomic analysis of S. schoenii20. Just like Krassowski et al.20 found in other Saccharomycopsis species, we identified two tRNA(CAG) genes in the S. schoenii genome. We conclude that one must be a tRNASer(CAG), since our proteomic analysis supported translation of the CTG codon to serine. We cannot determine if the other tRNA(CAG) is functional, but even though we found that CTG codons were “mistranslated” to leucine 1% of the time, just like in S. capsularis, we are at this point unable to call any real significance to the “mistranslation” of these proteins.

We detected upregulation of two TEs during the predatory, but not nutrient limited conditions of S. schoenii, which could imply that these TEs might have roles in either protecting the genome of S. schoenii during predation, or have roles in silencing any defense mechanisms in the prey cells. Studying these TEs further might therefore either reveal mechanistic insights to why no Saccharomycopsis species have been successfully genetically manipulated so far, or provide further insights into predatory mechanisms.

86 In this study, we were only able to demonstrate correlation between gene expression and abundance with responses to nutritional stresses and predatory activity, not causation. With our experimental this setup we cannot prove if the overexpressed aspartic proteases and glucanases are destroying prey cell walls, or are used in remodeling the S. schoenii cell wall during predation. However, we only observe overexpression of aspartic proteases occurs exclusively when S. schoenii predates on S. cerevisiae. In addition, similar transcriptome based studies on pathogen associated genes have been validated in C. albicans 68. Targeted gene disruption could be performed to validate our findings. Similarly, while we were able to identify a tRNASer(CAG) gene, with high similarity to the well characterized C. albicans tRNASer(CAG), and could detect translation of the CTG codon to serine, our setup could not determine if it was indeed caused by this tRNASer(CAG). To infer likelihood of the use of the S. schoenii tRNASer(CAG) gene, RT-PCR analysis could be performed.

Our findings suggest that S. schoenii acquires methionine and other sulfur compounds from its prey cells. Further studies, enabled by for instance the knock-in of fluorescent tags, could determine whether for instance the upregulated organic sulfur permeases MUP1, SEO1 and YCT1 are specifically distributed to the site of prey penetration. In addition, any role of cysteine as a trigger for predation should be considered. To determine if S. schoenii attacks all prey species with the same set of tools, of if its expanded set of aspartic proteases and glucanases is more like a Swiss army knife, adapted to specific prey cells, future studies using different prey species are needed.

In summary, our study sets the framework for further studies on the use of Saccharomycopsis yeast as potential biocontrol agents. We honed in on the timing and nutritional conditions under which S. schoenii kills the model prey species S. cerevisiae, identified and functionally characterized the genes and proteins involved during the predatory behavior of S. schoenii and provided a multi-omic foundation for further exploration of the ecology and evolution of Saccharomycopsis yeasts.

Acknowledgements This research was supported by the European Union Marie Curie Initial Training Network Fungibrain (Project ID: 607963). Saccharomycopsis schoenii was kindly provided by Marc-André Lachance.

References 1. Fisher, M. C., Hawkins, N. J., Sanglard, D. & Gurr, S. J. Worldwide emergence of resistance to antifungal drugs challenges human health and food security. Science (80-. ). 742, 739–742 (2018). 2. Bongomin, F., Gago, S., Oladele, R. & Denning, D. Global and Multi-National Prevalence of Fungal Diseases—Estimate Precision. J. Fungi 3, 57 (2017). 3. Jeffery-Smith, A. et al. Candida auris: a Review of the Literature. Clin. Microbiol. Rev. 31, e00029-17 (2018). 4. Fleming, A. On the Antibacterial Action of Cultures of a Penicillium, with Special Reference to their Use in the Isolation of B. influenzæ. Br. J. Exp. Pathol. 10, 226–236 (1929). 5. Druzhinina, I. S. et al. Trichoderma: the genomics of opportunistic success. Nat. Rev. Microbiol. 9, 749– 759 (2011). 6. Nagy, L. G. & Szöllősi, G. Fungal Phylogeny in the Age of Genomics: Insights Into Phylogenetic Inference From Genome-Scale Datasets. 1–24 (2017). doi:10.1016/bs.adgen.2017.09.008 7. Riley, R. et al. Comparative genomics of biotechnologically important yeasts. Proc. Natl. Acad. Sci. 113, 9882–9887 (2016). 8. Garnica, D. P., Upadhyaya, N. M., Dodds, P. N. & Rathjen, J. P. Strategies for Wheat Stripe Rust

87 Pathogenicity Identified by Transcriptome Sequencing. PLoS One 8, (2013). 9. Haas, R. et al. Designing and interpreting ‘multi-omic’ experiments that may change our understanding of biology. Curr. Opin. Syst. Biol. 6, 37–45 (2017). 10. Schmoll, M. et al. The Genomes of Three Uneven Siblings: Footprints of the Lifestyles of Three Trichoderma Species. Microbiol. Mol. Biol. Rev. 80, 205–327 (2016). 11. Jeffries, P. Biology and ecology of mycoparasitism. Can. J. Bot. 73, 1284–1290 (1995). 12. Karlsson, M., Atanasova, L. E. A., Jensen, D. A. N. F. & Zeilinger, S. Necrotrophic Mycoparasites and Their Genomes. 1–21 (2017). doi:10.1128/microbiolspec.FUNK 13. Kim, S. H. & Vujanovic, V. Relationship between mycoparasites lifestyles and biocontrol behaviors against Fusarium spp. and mycotoxins production. Appl. Microbiol. Biotechnol. 100, 5257–5272 (2016). 14. Lachance, M. A. & Pang, W. M. Predacious yeasts. Yeast 13, 225–232 (1997). 15. Pimenta, R. S. et al. Biological control of Penicillium italicum, P. digitatum and P. expansum by the predacious yeast Saccharomycopsis schoenii on oranges. Brazilian J. Microbiol. 39, 85–90 (2008). 16. Iacumin, L., Manzano, M., Andyanto, D. & Comi, G. Biocontrol of ochratoxigenic moulds (Aspergillus ochraceus and Penicillium nordicum) by Debaryomyces hansenii and Saccharomycopsis fibuligera during speck production. Food Microbiol. 62, 188–195 (2017). 17. Junker, K. et al. The mycoparasitic yeast Saccharomycopsis schoenii predates and kills multi-drug resistant Candida auris. Sci. Rep. Manuscript ID # SREP-18-19093-T (2018). 18. Kurtzman, C. P. & Robnett, C. J. Relationships among genera of the Saccharomycotina (Ascomycota) from multigene phylogenetic analysis of type species. FEMS Yeast Res. 13, 23–33 (2013). 19. Kurtzman, C. P. & Smith, M. T. Saccharomycopsis Schiönning (1903). in The Yeasts 751–763 (Elsevier, 2011). doi:10.1016/B978-0-444-52149-1.00063-X 20. Krassowski, T. et al. Evolutionary Instability of CUG-Leu in the Genetic Code of Budding Yeasts. Nat. Commun. In Press (2018). doi:10.1038/s41467-018-04374-7 21. Ohama, T. et al. Non-universal decoding of the leucine codon CUG in several Candida species. Nucleic Acids Res. 21, 4039–4045 (1993). 22. Butler, G. et al. Evolution of pathogenicity and sexual reproduction in eight Candida genomes. Nature 459, 657–662 (2009). 23. Thomas, D. & Surdin-Kerjan, Y. Metabolism of sulfur amino acids in Saccharomyces cerevisiae. Microbiol. Mol. Biol. Rev. {MMBR} 61, 503–532 (1997). 24. Lachance, M. a, Pupovac-Velikonja, a, Natarajan, S. & Schlag-Edler, B. Nutrition and phylogeny of predacious yeasts. Can. J. Microbiol. 46, 495–505 (2000). 25. Kurtzman, C. P., Fell, J. W. & Boekhout, T. The yeasts : a taxonomic study. (Elsevier, 2010). 26. Junker, K., Hesselbart, A. & Wendland, J. Draft Genome Sequence of Saccharomycopsis fodiens CBS 8332, a Necrotrophic Mycoparasite with Biocontrol Potential. Genome Announc. 5, 605–606 (2017). 27. Hesselbart, A., Junker, K. & Wendland, J. Draft Genome Sequence of Saccharomycopsis fermentans CBS 7830 , a Predacious Yeast Belonging to the Saccharomycetales. Genome Announc. 6, 4–5 (2018). 28. Spanu, P. D. The Genomics of Obligate (and Nonobligate) Biotrophs. Annu. Rev. Phytopathol. 50, 91– 109 (2012). 29. Albalat, R. & Cañestro, C. Evolution by gene loss. Nat. Rev. Genet. 17, 379–391 (2016). 30. Domergue, R. Nicotinic Acid Limitation Regulates Silencing of Candida Adhesins During UTI. Science (80- . ). 308, 866–870 (2005). 31. Duplessis, S. Obligate biotrophy features unraveled by the genomic analysis of rust fungi. Proc. Natl. Acad. Sci. U. S. A. 1–23 (2011). doi:10.1073/pnas.1019315108/- /DCSupplemental.www.pnas.org/cgi/doi/10.1073/pnas.1019315108 32. Kemen, E. & Jones, J. D. G. Obligate biotroph parasitism: Can we link genomes to lifestyles? Trends

88 Plant Sci. 17, 448–457 (2012). 33. Schmidt, S. M. & Panstruga, R. Pathogenomics of fungal plant parasites: What have we learnt about pathogenesis? Curr. Opin. Plant Biol. 14, 392–399 (2011). 34. Raffaele, S. & Kamoun, S. Genome evolution in filamentous plant pathogens: Why bigger can be better. Nat. Rev. Microbiol. 10, 417–430 (2012). 35. Möller, M. & Stukenbrock, E. H. Evolution and genome architecture in fungal plant pathogens. Nat. Rev. Microbiol. 15, 756–771 (2017). 36. Muszewska, A., Steczkiewicz, K., Stepniewska-Dziubinska, M. & Ginalski, K. Cut-and-Paste Transposons in Fungi with Diverse Lifestyles. Genome Biol. Evol. 9, 3463–3477 (2017). 37. Gruber, S. G. & Seidl-Seiboth, V. Self versus non-self: Fungal cell wall degradation in Trichoderma. Microbiology 158, 26–34 (2012). 38. Choo, J. H. et al. Whole-genome de novo sequencing, combined with RNA-Seq analysis, reveals unique genome and physiological features of the amylolytic yeast Saccharomycopsis fibuligera and its interspecies hybrid. Biotechnol. Biofuels 9, 1–22 (2016). 39. Wang, D., Chi, Z., Zhao, S. & Chi, Z. M. Disruption of the acid protease gene in Saccharomycopsis fibuligera A11 enhances amylolytic activity and stability as well as trehalose accumulation. Enzyme Microb. Technol. 49, 88–93 (2011). 40. Eliceiri, K., Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ : 25 years of image analysis HISTORICAL commentary NIH Image to ImageJ : 25 years of image analysis. Nat. Methods 9, 671–675 (2012). 41. Schindelin, J., Rueden, C. T., Hiner, M. C. & Eliceiri, K. W. The ImageJ ecosystem: An open platform for biomedical image analysis. Mol. Reprod. Dev. 82, 518–529 (2015). 42. Schindelin, J. et al. Fiji: An open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012). 43. Schneider, N. fixTranslation-Macro-for-ImageJ. (2014). 44. Li, K. The image stabilizer plugin for ImageJ. (2008). Available at: http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html,. 45. Gurevich, A., Saveliev, V., Vyahhi, N. & Tesler, G. QUAST: Quality assessment tool for genome assemblies. Bioinformatics 29, 1072–1075 (2013). 46. Götz, S. et al. High-throughput functional annotation and data mining with the Blast2GO suite. Nucleic Acids Res. 36, 3420–3435 (2008). 47. Pearson, W. R. An Introduction to Sequence Similarity (“Homology”) Searching. Curr. Protoc. Bioinformatics 0 3, 10.1002/0471250953.bi0301s42 (2013). 48. Lowe, T. M. & Chan, P. P. tRNAscan-SE On-line: integrating search and context for analysis of transfer RNA genes. Nucleic Acids Res. 44, W54–W57 (2016). 49. Deutsch, E. W. et al. The ProteomeXchange consortium in 2017: Supporting the cultural change in proteomics public data deposition. Nucleic Acids Res. 45, D1100–D1106 (2017). 50. Vizcaíno, J. A. et al. 2016 update of the PRIDE database and its related tools. Nucleic Acids Res. 44, D447–D456 (2016). 51. Priebe, S., Kreisel, C., Horn, F., Guthke, R. & Linde, J. FungiFun2: A comprehensive online resource for systematic analysis of gene lists from fungal species. Bioinformatics 31, 445–446 (2015). 52. Siverio, J. M. Assimilation of nitrate by yeasts. FEMS Microbiol. Rev. 26, 277–284 (2002). 53. Gordon, J. L. et al. Evolutionary erosion of yeast sex chromosomes by mating-type switching accidents. Proc. Natl. Acad. Sci. U. S. A. 108, 20024–9 (2011). 54. Riley, R. et al. Comparative genomics of biotechnologically important yeasts. Proc. Natl. Acad. Sci. U. S. A. 113, 9882–9887 (2016).

89 55. Chan, P. P. & Lowe, T. M. GtRNAdb 2.0: An expanded database of transfer RNA genes identified in complete and draft genomes. Nucleic Acids Res. 44, D184–D189 (2016). 56. Santos, M. A. S., Gomes, A. C., Santos, M. C., Carreto, L. C. & Moura, G. R. The genetic code of the fungal CTG clade. Comptes Rendus - Biol. 334, 607–611 (2011). 57. Mühlhausen, S., Findeisen, P., Plessmann, U., Urlaub, H. & Kollmar, M. A novel nuclear genetic code alteration in yeasts and the evolution of codon reassignment in eukaryotes. Genome Res. 26, 945–955 (2016). 58. Miranda, I. et al. Candida albicans CUG mistranslation is a mechanism to create cell surface variation. MBio 4, e00285-13 (2013). 59. Strijbis, K. & Distel, B. Intracellular acetyl unit transport in fungal carbon metabolism. Eukaryot. Cell 9, 1809–1815 (2010). 60. Canadell, D. & Ariño, J. Interactions Between Monovalent Cations and Nutrient Homeostasis. in Yeast Membrane Transport 892, 271–289 (2016). 61. Naglik, J. R., Challacombe, S. J. & Hube, B. Candida albicans Secreted Aspartyl Proteinases in Virulence and Pathogenesis. Microbiol. Mol. Biol. Rev. 67, 400–428 (2003). 62. Smothers, D. B., Kozubowski, L., Dixon, C., Goebl, M. G. & Mathias, N. The abundance of Met30p limits SCF(Met30p) complex activity and is regulated by methionine availability. Mol. Cell. Biol. 20, 7845–52 (2000). 63. Rouillon, A., Barbey, R., Patton, E. E., Tyers, M. & Thomas, D. Feedback-regulated degradation of the transcriptional activator Met4 is triggered by the SCF(Met30) complex. EMBO J. 19, 282–294 (2000). 64. Ouni, I., Flick, K. & Kaiser, P. Ubiquitin and transcription: The SCFMet30/Met4 pathway, a (protein-) complex issue. Transcription 2, 135–139 (2011). 65. Dean, R. et al. The Top 10 fungal pathogens in molecular plant pathology. Mol. Plant Pathol. 13, 414– 430 (2012). 66. Seidl, V. et al. Transcriptomic response of the mycoparasitic fungus Trichoderma atroviride to the presence of a fungal prey. BMC Genomics 10, 1–13 (2009). 67. Stephanopoulos, G. N., Aristidou, A. A. & Nielsen, J. 2 - Review of Cellular Metabolism. in Metabolic Engineering: Principles and Methadologies (eds. Stephanopoulos, G. N., Aristidou, A. A. & Nielsen, J. B. T.-M. E.) 21–79 (Academic Press, 1998). doi:https://doi.org/10.1016/B978-012666260-3/50003-0 68. Wilson, D. et al. Identifying infection-associated genes of Candida albicans in the postgenomic era. FEMS Yeast Res. 9, 688–700 (2009).

90 Figure and Table legends

Figure 1: S. schoenii predates on and eliminates S. cerevisiae. A) S. cerevisiae (H4-GFP) cells collapse after contact- mediated mycoparasitism. Upon physical contact, S. cerevisiae cells vacuolarize (arrowhead), collapse in size and loses its H4-GFP signal (asterisk). B) The area of non-predated and predated S. cerevisiae. While non-predated cells stayed the same size, predated cells shrank significantly. Error bars = 95% CI. Dotted lines; linear regression of slopes, p-value of slopes <0.0001. C) Quantification and timing of predation during co-culture of S. schoenii and S. cerevisiae under different nutritional conditions. Cells were scored hourly on morphology-based viability. Results from two independent replicates, plotted as means with SD error bars.

Figure 2: In silico analyses suggest S. schoenii is a CTG clade member. A) A cloudblast of translated S. schoenii ORFs against proteins suggested by nr database. S. schoenii was closely related to several CTG clade members. B) Two CAG-tRNAs were found in the draft S. schoenii genome. C) Alignment of the two S. schoenii CAG- tRNAs with CAG-tRNAs from Saccharomycopsis fermentans 27, C. albicans, C. dubliniensis, C. tropicalis as well as other Leu-tRNAs and Ser-tRNAs from S. cerevisiae and C. albicans. The S. schoenii CAG.2-tRNA aligns the cloesest with a S. cerevisiae Leu-GAG-tRNA, whereas S. schoenii CAG.1-tRNA aligns closer to the Ser-CAG-tRNAs of other CTG clade yeasts. CAG-tRNA features are color-coded54,56

Figure 3: Proteins and GO term categories increased during starvation and/or predation conditions. A) Presence and relative abundance of S. schoenii proteins during starvation and predation conditions. Translated genes with homologs to either S. cerevisiae or C. albicans, with a protein abundance at least 10-fold higher during starvation conditions (SD) or predation conditions (SD + S.c.) were selected and listed with its corresponding gene ID. The proteins were curated into three subsets; proteins highly abundant under starvation (green), predation (red) or in both conditions (yellow). B) The top 15 GO term categories of S. schoenii proteins involved in starvation (green), predation (red) or abundant in both conditions, as output from FungiFun251.

Figure 4: S. schoenii gene and GO term categories upregulated during gradual nutrient availability and/or predation conditions. A) Fold upregulation of S. schoenii genes during gradual nutrient availability, compared to S. schoenii on standard media (YPD), without (right) or with (left) S. cerevisiae present as prey. Upregulated genes were manually curated into three categories; mainly upregulated during conditions of Nutrient limitation (CSM, grey), of Methionine deprivation (CSM-Met, blue) or Predation (CSM + S.c., CSM-Met + S.c. or SD + S.c.). Down pointing arrows symbolize genes with two (one arrow) or five (two arrows) fold lower fold increase values when prey was present, compared to when no prey was present, on CSM-Met. B) Top 15 GO term categories of genes upregulated during Nutrient limitation, Methionine deprivation of Predation conditions.

Figure 5: Correlation between predatory activity and overexpression of S. schoenii genes after3 hours of co- culture. Black line, percentage of predated S. cerevisiae cells. A) Transcription values of yapsin/aspartic protease genes during co-culture of S. schoenii and S. cerevisiae in red, and during sole culture of S. schoenii in dotted black. B) Transcription values of glucanases, glycosidases and chitinase genes during co- culture of S. schoenii and S. cerevisiae in green, and during sole culture of S. schoenii in dotted black.

Figure 6: Methionine biosynthesis pathway and sulfur compound uptake genes in S. schoenii. Genes highlighted in blue were upregulated at least two-fold when methionine was missing compared to present (CSM/CSM-Met). Genes highlighted in red were upregulated at least two-fold when prey was present

91 compared to absent, during methionine deprivation (CSM-Met S.c./CSM-Met). A) Fold change of S. schoenii gene expression during methionine deprived conditions, when prey was present compared to absent (CSM-Met + S.c./CSM-Met). Dotted lines indicate two-fold changes up or down. B) S. schoenii lacks all genes in the sulfate assimilation pathway, but have two methionine permeases (MUP1, MUP3), two copies of the cysteine transporter YCT1 and several copies of SEO1, a putative sulfur compound permease.

Table 1: S. schoenii de novo draft genome assembly and annotation. A) S. schoenii de novo draft genome assembly. B) S. schoenii draft genome annotation. For subsequent functional analyses, the 4,660 genes with homologs were used. a) Identified with tRNA Scan-SE. b) ORF = min 300bp/100aa, ATG start site. c) Using alternative translation (table 12). d) Manual curation of non-overlapping genes. e) Blastx hit score >55. f) deposited by 06.06.17. g) non-overlapping ORFs, any blastx hit bit score >55. h) non-overlapping ORFs, blastx hit bit score <55. C) Gene loss in the sulfate assimilation pathway. D) List of gene expansion. Names of homologous S. cerevisiae or C. albicans genes.

Supplemental Figure 1: S. schoenii can eliminate S. cerevisiae. Hygromycin sensitive S. schoenii cells and hygromycin resistant S. cerevisiae cells were cultured alone or co-cultured on SD media, and subsequently stamped onto YPD and YPD with hygromycin. After co-culture, no live S. cerevisiae is left, as indicated by no growth of S. cerevisiae on YPD + hygromycin (red box).

Supplementary Figure 2: Putative mating genes in S. schoenii. The expanded genes YJR098C (yellow) and SLA2 (blue) flank MTLALPHA1 and other potential mating type loci (red) in S. schoenii. Asterisks (*) denotes genes with homology bit score <55. In this study, we were only able to detect MF(ALPHA) (pink), not MF(A). Black half circles represent end of contigs.

Supplementary Movie 1: S. cerevisiae (H4-GFP) cells collapse minues after S. schoenii exerts contact-mediated mycoparasitism. Movie version of Figure 1 (white insert).

Supplementary Table 1: CTG translation in S. schoenii. Predicted and actual translation of CTG positions in S. schoenii genes.

92 Figures

Figure 1:

93 Figure 2:

94 Figure 3:

95 Figure 4:

96 Figure 5:

97 Figure 6:

98 Tables

Table 1:

99 Supplementary Material

Supplementary Figure 1:

Supplementary Figure 2:

Supplementary Material 1: The S. schoenii genome in a fasta-format.

Available at: https://tinyurl.com/KlaraPhD

Supplementary Material 2: The genome annotation of S. schoenii as a gff-file.

Available at: https://tinyurl.com/KlaraPhD

100 Supplementary Data 1: Excel sheet of annotated S. schoenii ORFs linked with transcriptomic and proteomic data.

Available at: https://tinyurl.com/KlaraPhD

Supplementary Movie 1:

Available at: https://tinyurl.com/KlaraPhD

Supplementary Table 1:

101 Klara Junker | Prospecting a Predator

102 Prospecting a Predator | Klara Junker

Paper IV

103 1 Junker et al., SREP-18-19093-T 2 3 The mycoparasitic yeast Saccharomycopsis schoenii predates on

4 and kills multi-drug resistant Candida auris

5

6 Klara Junker1, Gustavo Bravo Ruiz2, Alexander Lorenz2, Louise

7 Walker2, Neil A.R. Gow2 and Jürgen Wendland1,3#

8 9 1Carlsberg Research Laboratory, Yeast & Fermentation, DK-1799 Copenhagen V, Denmark

10 2The Institute of Medical Sciences (IMS), MRC Centre for Medical Mycology at the University 11 of Aberdeen, School of Medicine, Medical Sciences & Nutrition, University of Aberdeen, 12 Foresterhill, Aberdeen, AB25 2ZD, United Kingdom 13 3 Vrije Universiteit Brussel, Functional Yeast Genomics, BE-1050 Brussels, Belgium 14 15 Running title: Mycoparasitism of Saccharomycopsis schoenii on Candida auris 16 17 18 19 # Corresponding authors 20 Prof. Dr. Jürgen Wendland Prof. Dr. Neil Gow 21 Vrije Universiteit Brussels The Institute of Medical Sciences (IMS) 22 Research Group Microbiology MRC Centre for Medical Mycology 23 Functional Yeast Genomics University of Aberdeen 24 Pleinlaan 2 Foresterhill, Aberdeen, AB25 2ZD 25 BE-1050 Brussels, Belgium United Kingdom 26 Tel.: +45-3327-5230 +44-1224-437483 27 Fax: +45-3327-4708 28 email: [email protected] [email protected] 29 30 31 32

104 33 Abstract 34 Candida auris has recently emerged as a multi-drug resistant fungal pathogen that 35 poses a serious global health threat, especially for patients in hospital intensive care 36 units (ICUs). C. auris can colonize human skin and can spread by physical contact or 37 contaminated surfaces and equipment. Here, we show that the mycoparasitic yeast 38 Saccharomycopsis schoenii efficiently kills multi-drug resistant isolates of C. auris, as 39 well as clinical isolates of other pathogenic species of the Candida genus suggesting 40 novel approaches for biocontrol. 41

42 Introduction 43 Candida auris was first seen as an agent of human disease in 2009, when it had 44 been isolated from the ear canal of a patient in Japan1. It has subsequently spread 45 rapidly around the world and is now a major health threat on a global scale having 46 been associated with potentially lethal infections in patients in ICUs in Eastern and 47 South Asia, South Africa, Europe, the USA, and South America. Sequence-based 48 analyses have grouped C. auris isolates from around the world into at least four 49 different clades, represented by clonal populations2. More than 90% of isolates are 50 fluconazole resistant and many isolates are cross-resistant to more than one of the 51 three major classes of antifungals – azoles, echinocandins and polyenes2. A few 52 strains of this fungus are resistant to all of the major classes of antifungals used by 53 doctors to treat fungal infections. The species has a propensity to colonize skin, and 54 it has proven to be difficult to eradicate from ICUs. Worldwide mortality rates for 55 disease cases with C. auris infections of the bloodstream approach 50%. Concern 56 about the emergence and spread of C. auris has resulted in alerts being posted by 57 the CDC (Centers for Disease Control and Prevention, USA), the ECDC (European 58 Centre for Disease Prevention and Control, Sweden), and PHE (Public Health 59 England, UK)3-5. 60 61 C. auris colonizes the skin of patients and can be transmitted via contact with 62 patients or contaminated hospital fixtures, which has already resulted in several 63 health care associated outbreaks6. The horizontal transmission potential of C. auris 64 demands strict decontamination methods and infection prevention protocols, since 65 mortality rates in patients with systemic infections can be up to 60%2,7-8. 66

105 67 Necrotrophic mycoparasitism describes the ability of a fungal species to kill other 68 fungi9. For example, filamentous fungi in the Trichoderma genus have been well 69 characterized over the last decades and are used as biocontrol agents against fungal 70 plant pathogens10. Mycoparasites typically exhibit wide host ranges and due to their 71 general mode of action in killing their prey may use broad-acting lytic enzymes, such 72 as proteases and chitinases. However, some species, e.g. Trichoderma, also 73 generate specialized penetration structures or haustoria that allow themto efficiently 74 target and kill prey cells11-12. Active mycoparasitism in yeast was discovered only in 75 1997, when species of the genus Saccharomycopsis were first described as 76 necrotrophic predacious yeasts13. It has not been studied whether Saccharomycopsis 77 species attack Candida species other than Candida albicans13-14. Predacious 78 behaviour depends on solid structural support, presumably to allow for stable cell-cell 79 contact and has been suggested to be starvation induced14. Specifically, a lack of 80 organic sulfur-containing organic compounds such as methionine has been 81 suggested as a trigger for predation, as all Saccharomycopsis yeasts share the, for 82 microorganisms, rare feature of being unable to assimilate sulfate as their sole 83 source of sulfur. Recently, we reported the lack of eight genes in the sulfate 84 assimilation pathway in draft genomes of Saccharomycopsis fodiens and 85 Saccharomycopsis fermentans15-16. 86 Here we show that S. schoenii efficiently attacks and kills a range of pathogenic 87 Candida species, including the newly emerged human pathogenic fungus C. auris. 88 We follow the predation process using time lapse microscopy in combination with 89 fluorescent dyes. Efficient predation as shown here could be useful for biocontrol 90 purposes in either clinical settings for skin clearance or in agricultural settings for 91 combatting plant pathogens. 92 93 Results and Discussion 94 In this study, we prospected the use of a predatory yeast, Saccharomycopsis 95 schoenii, as a potential biocontrol agent against human fungal pathogens of the 96 Candida clade with a focus on C. auris. To this end we confronted multiple drug 97 resistant strains including C. auris NCPF8985#20, a multi-drug resistant isolate from 98 the South Asian clade (India), with S. schoenii (Supplementary Table 1). Equal 99 numbers of dimorphic S. schoenii and ovoid C. auris NCPF8985#20 cells were 100 seeded on minimal media agar on microscopy slides to offer solid support for a

106 101 potential S. schoenii interaction. This minimal media lacked methionine and thus did 102 not support proliferation of S. schoenii in pure culture. We found that S. schoenii 103 attacked C. auris cells upon contact and killed prey cells using specialized 104 penetration pegs (Figure 1; Supplementary Movies 1-3). The chitin-staining 105 fluorescent dye Calcofluor White, revealed septa at the bases of penetration pegs 106 indicating the sites of entry (Figure 1a; Supplementary Movies 1-2). Within minutes 107 after S. schoenii cells generated penetration pegs, C. auris cells started to 108 vacuolarize, take up dyes such as propidium iodide that are not permeable to living 109 cells and then collapse, presumably as a result of feeding and material transfer to the 110 predating S. schoenii cell (Figure 1a; Supplementary Movies 1-3). We prepared 111 Transmission Electron Microscopy (TEM) images of interactions between S. schoenii 112 and C. auris after 1 h of co-culture on minimal media (Figure 1b-e) and found that C. 113 auris cells attacked by S. schoenii cells were necrotic (Figure 1b). Penetration pegs 114 were directed at prey cells (Figure 1c) and cell wall interactions led to the formation 115 of penetration peg start sites (# Figure 1d). Ultimately this led to to degradation of the 116 prey cell wall (Figure 1e). After killing of prey cells, penetration pegs did not grow 117 further or develop into buds or daughter cells (Supplementary Movie 3). During their 118 interaction with prey cells, S. schoenii cells did not show apical growth, however, 119 S. schoenii cells resumed polar growth after successful attacks (Supplementary 120 Movie 3). 121 122 We determined rates of killing over a 6 h time-course using morphology and/or 123 propidium iodide staining (PI) stain (Supplementary Figure 1, 2). Several prey cells 124 were found to accumulate PI staining upon predation, however, many prey cells were 125 apparently killed without being stained by PI. In these cases, killed prey cells were 126 “flattened” and or shrunken in size. This resulted in the death of around 34 % of 127 C. auris cells within a period of 6 h of co-culture with S. schoenii (Figure 2, middle 128 panel; Supplementary Table 2). As a control, almost none of the C. auris cells (0.6%) 129 had died after 6 h when cultured alone under identical experimental conditions. To 130 examine if predator-prey interactions differ with different C. auris isolates that exhibit 131 variable drug resistance phenotypes, we analysed predator-prey interactions in three 132 additional C. auris isolates (Supplementary Table 1). Furthermore, to elucidate host 133 range of predator-prey interactions within Candida species we included clinical 134 isolates of Candida albicans, Candida glabrata, Candida lusitaniae, Candida

107 135 parapsilosis and Candida tropicalis in this analysis. For reference, we used 136 Saccharomyces cerevisiae and Schizosaccharomyces pombe, two previously known 137 prey species of S. schoenii14. All isolates of Candida species tested, including several 138 drug resistant C. auris strains, were susceptible to predation by S. schoenii (Figure 2 139 and Supplementary Table 2). 140 141 Collectively, these results demonstrate that the predator yeast S. schoenii provides a 142 novel opportunity to develop biocontrol methods for skin disinfection. 143 Saccharomycopsis predator yeasts may therefore have potential as biocontrol agents 144 of other fungi including human and plant pathogens. Based on genome survey 145 sequencing, Saccharomycopsis yeasts, like the distantly related filamentous 146 ascomycete Trichoderma, harbor multi-gene families of proteases and chitinases15-16 147 (our unpublished data). These multi-gene families probably represent a resource for 148 the identification of lytic enzymes that have the potential to generate novel antifungal 149 compounds. 150 151 Methods 152 Strains. Wild-type Saccharomycopsis schoenii (CBS 7425, CBS-KNAW collection, 153 Utrecht, Netherlands) was provided by Marc-André Lachance. Saccharomyces 154 cerevisiae (BY4741) was provided by EUROSCARF (Scientific Research and 155 Development GmbH, Oberursel, Germany). Schizosaccharomyces pombe UoA324, 156 a derivative of FY15112 from NBRP yeast and SO2427 from Snezhana Oliferenko 157 (King’s College London). Clinical isolates of the following C. auris and other Candida 158 strains were used; C.auris (NCPF8980#9), C. auris (NCPF8985#20) and C. auris 159 (NCPF13005#95) provided by Liz Johnson (PHE Bristol) and C. auris (B8441) 160 provided by Shawn Lockhart (CDC Atlanta), Candida albicans (UC820) provided by 161 Mihai Netea (Radboudumc, Nijmegen), Candida glabrata (BG2) provided by Brendan 162 P. Cormack (Johns Hopkins University, Baltimore, MD), and Candida parapsilosis 163 (AM2017/001), Candida lusitaniae (AM2017/006) and Candida tropicalis 164 (AM2017/004) provided by Donna MacCallum (University of Aberdeen). 165 166 Media and Growth conditions. Cells were cultured to log phase in YPD media (10 167 g/L yeast extract, 20 g/L Bacto peptone, 20 g/L glucose), 30°C, rotating. Cells were 168 washed and stained with Calcofluor White (10 µg/mL) and propidium iodide (1

108 169 µg/mL). S. schoenii was mixed with prey cells at roughly 5*107 cells/mL each, in 170 Synthetic Defined (SD) media (6.7 g/L YNB w/o amino acids with ammonium sulfate 171 20 g/L glucose). Prior to imaging, cells were seeded on pads with SD media solidified 172 with 1 % agarose. 173 174 Microscopy. Imaging was performed using the PerkinElmer UltraVIEW VoX 175 Spinning Disk Confocal Microscope controlled by Volocity software. Images for 176 movies were captured 2-4 times/min for up to 2 h, using the Nikon Perfect Focus 177 System to autofocus. For kill curve analyses, three frames were captured every hour 178 per species and time point. FIJI/ImageJ17 was used for image processing and 179 analysis. Drift in movies frames was corrected using the macro NMS fixTranslation v1 180 and the plugin Image Stabiliser. For kill curve analyses, individual cells were counted 181 using the Cell counter plugin. 182 183 For TEM images, S. schoenii and C. auris cells were separately pre-cultured to log 184 phase in YPD media, then washed and mixed together at equal ratios. A total 1*108 185 cells were seeded on SD media solidified with 2% agarose. After 1 h of co-culture, 186 cells were scraped off, washed and pelleted. High Pressure Freezing was carried out 187 using a Leica EM PACT 2 (Leica Microsystems, Milton Keynes, UK) and samples 188 were freeze substituted in a Leica AFS 2. Freeze substitution was carried out using

189 the following program: -95°C to -90°C for 30 h with 2% OsO4 in acetone, -90°C for 10

190 h with 2% OsO4 in acetone, -90°C to -30°C for 8 h with 2% OsO4 in acetone, -30°C to 191 -10°C for 1 h with acetone, -10°C to 4°C for 1 h in acetone, 4°C to 20°C for 1 h in 192 acetone. Samples were then removed and placed in 10% Spurr’s (TAAB, UK): 193 acetone for 72 h, followed by 30% Spurr’s overnight, 50% Spurr’s for 8 h, 70% 194 Spurr’s overnight, 90% Spurr’s for 8 h and embedded in Spurr’s resin at 60°C for at 195 least 24 h. Ultrathin sections were cut to 90 µm using a diamond knife (Diatome Ltd, 196 Switzerland) onto copper grids (TAAB, UK) using a Leica UC6 and were contrast 197 stained with uranyl acetate and lead citrate in a Leica AC20. Samples were imaged 198 on a JEM 1400 plus (JEOL UK) Transmission Electron Microscope and captured 199 using an AMT UltraVUE camera (AMT, USA). 200 All relevant data are available from the authors. 201

109 202 References 203 1. Satoh, K. et al. Candida auris sp. nov., a novel ascomycetous yeast isolated from the external 204 ear canal of an inpatient in a Japanese hospital. Microbiol Immunol 53, 41-4 (2009). 205 2. Lockhart, S.R. et al. Simultaneous Emergence of Multidrug-Resistant Candida auris on 3 206 Continents Confirmed by Whole-Genome Sequencing and Epidemiological Analyses. Clin 207 Infect Dis 64, 134-140 (2017). 208 3. Jeffery-Smith, A. et al. Candida auris: a Review of the Literature. Clin Microbiol Rev 31(2018). 209 4. Chowdhary, A., Sharma, C. & Meis, J.F. Candida auris: A rapidly emerging cause of hospital- 210 acquired multidrug-resistant fungal infections globally. PLoS Pathog 13, e1006290 (2017). 211 5. Lockhart, S.R., Berkow, E.L., Chow, N. & R.M. Welsh. Candida auris for the Clinical 212 Microbiology Laboratory: Not Your Grandfather's Candida Species. Clin Microbiol Newsl 39, 213 99-103 (2017). 214 6. Welsh, R.M. et al. Survival, Persistence, and Isolation of the Emerging Multidrug-Resistant 215 Pathogenic Yeast Candida auris on a Plastic Health Care Surface. J Clin Microbiol 55, 2996- 216 3005 (2017). 217 7. Schelenz, S. et al. First hospital outbreak of the globally emerging Candida auris in a 218 European hospital. Antimicrob Resist Infect Control 5, 35 (2016). 219 8. Kean, R. et al. Surface disinfection challenges for Candida auris: an in-vitro study. J Hosp 220 Infect (2017). 221 9. Karlsson, M., Atanasova, L., Jensen, D.F. & Zeilinger, S. Necrotrophic Mycoparasites and 222 Their Genomes. Microbiol Spectr 5(2017). 223 10. Hermosa, R., Viterbo, A., Chet, I. & Monte, E. Plant-beneficial effects of Trichoderma and of 224 its genes. Microbiology 158, 17-25 (2012). 225 11. Druzhinina, I.S. et al. Trichoderma: the genomics of opportunistic success. Nat Rev Microbiol 226 9, 749-59 (2011). 227 12. Schmoll, M. et al. The Genomes of Three Uneven Siblings: Footprints of the Lifestyles of 228 Three Trichoderma Species. Microbiol Mol Biol Rev 80, 205-327 (2016). 229 13. Lachance, M.A. & Pang, W.M. Predacious yeasts. Yeast 13, 225-32 (1997). 230 14. Lachance, M.A., Pupovac-Velikonja, A., Natarajan, S. & Schlag-Edler, B. Nutrition and 231 phylogeny of predacious yeasts. Can J Microbiol 46, 495-505 (2000). 232 15. Junker, K., Hesselbart, A. & Wendland, J. Draft Genome Sequence of Saccharomycopsis 233 fodiens CBS 8332, a Necrotrophic Mycoparasite with Biocontrol Potential. Genome Announc 234 5(2017). 235 16. Hesselbart, A., Junker, K. & Wendland, J. Draft Genome Sequence of Saccharomycopsis 236 fermentans CBS 7830, a Predacious Yeast Belonging to the Saccharomycetales. Genome 237 Announc 6(2018). 238 17. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat Methods 9, 239 676-82 (2012). 240

110 241 Acknowledgements 242 This research was supported by the European Union Marie Curie Initial Training Network 243 Fungibrain (607963), the Wellcome Trust (101873, 204815, 086827, 075470, 099215, and 244 097377), and the MRC Centre for Medical Mycology at the University of Aberdeen 245 (MR/N006364/1). We thank Gillian Milne of the Microscopy and Histology facility at the 246 University of Aberdeen for expert assistance with TEM. Fungal strains were kindly provided 247 by Brendan P. Cormack, Elizabeth Johnson, Marc-André Lachance Shawn Lockhart, Donna 248 MacCallum, Mihai Netea, Snezhana Oliferenko, and the National BioResource Project 249 (NBRP) Japan. 250 251 Author contributions 252 K.J, J.W., A.L. and N.A.R.G. conceived this study. K.J., G.B.R., A.L., and L.W. conducted the 253 experiments. K.J. and A.L. designed and conducted the predation assays and its analyses. 254 J.W., K.J. and N.A.R.G. drafted the manuscript. All authors read, revised and approved the 255 manuscript. 256 257 Competing interests 258 The authors declare no competing interests

111 259 Figure legends 260 Figure 1: 261 S. schoenii attacks and kills C. auris. a) S. schoenii and C. auris NCPF8985#20 262 stained with Calcoflour White (CW, cyan, bottom panel), a fluorescent dye that stains 263 chitin rich cell walls and septa, and propidium iodide (PI, red, bottom panel), a 264 fluorescent dye that stains nucleic acids of cells with a compromised cell membrane, 265 i.e. dead or dying cells. We captured images twice per minute for two hours and 266 found that at 15 min, a penetration peg [Δ] from S. schoenii is visualized by CW. The 267 C. auris prey cell subsequently collapses in size between 15 and 30 min (Λ). 268 Whereas the attacked C. auris cell was not stained by PI, its daughter cell 269 accumulated PI between 75 min to 120 min (*). b-e) TEM images of S. schoenii and 270 C. auris that had been co-cultured for 1h. Scale bar 500 nm in b, c and e. Scale bar 271 100nm in d. b) A dimorphic S. schoenii cell has formed a penetration peg to contact, 272 attack and kill an ovoid C. auris cell. c) A S. schoenii cell with a penetration peg 273 protruding towards a prey cell. d) Early interactions between A. schoenii and C. auris 274 visualize potential penetration peg start sites (#). e) Partial disintegration of the 275 C. auris cell wall. 276 277 Figure 2: 278 Kill curves of prey species attacked and killed by S. schoenii. Cells were co- 279 cultured on several slides with SD media with 1% agarose, for up to 6 h. Every hour, 280 we took one slide and captured 3 representative frames. Cells were scored on 281 viability based on morphology and PI stain as per Fig S2. 282 283 Supplementary Figure 1: 284 Morphology and Propidium Iodide (PI) guided viability score. We counted and 285 scored prey cells and S. schoenii cells based on morphology and/or PI stain (red). 286 Top row: Live cells were large with smooth morphology and no PI stain, C. auris to 287 the left, S. schoenii to the right. Middle row: Attacked prey cells (Λ) were vacuolarized 288 and in physical contact with S. schoenii cells. Bottom row: Dead prey cells (Δ) were 289 stained by PI (left) and/or were shrunken in size (middle) or flattened (right). 290

112 291 Supplementary Figure 2: 292 Example of overgrowth of S. schoenii cells and C. auris (NCPF8980#9) cells 293 after 6 h of co-culture. Quantitative observations were done hourly for up to 6 h, 294 after which point individual cells could no longer be distinguished. 295 296 Supplementary Table 1: 297 Details of yeast strains used in this article. 298 299 Supplementary Table 2: 300 Number of S. schoenii cell and prey cells counted and viability scored at each time 301 point. 302 303 Supplementary Movie 1: 304 Movie corresponding to Figure 1; S. schoenii attacks and kills C. auris. 305 S. schoenii and C. auris NCPF8985#20 were stained with Calcoflour White (CW, 306 cyan, bottom panel), a fluorescent dye that stains chitin rich cell walls and septa, and 307 propidium iodide (PI, red, bottom panel), a fluorescent dye that stains nucleic acids of 308 cells with a compromised cell membrane, i.e. dead or dying cells. 309 310 Supplementary Movie 2: 311 S. schoenii cells attacking three C. auris cells. S. schoenii and C. auris 312 NCPF8985#20 stained with Calcoflour White (CW, cyan in bottom panel) and 313 propidium iodide (PI, red, bottom panel). Penetration pegs are visualized by 314 appearance of CW staining at the site of penetration, at 20-30 min (Δ, lower panel). 315 Prey cells collapse within minutes of the establishment of the penetration peg, at 25- 316 40 min (Λ, upper panel). Weak PI staining is detected from 80-100 min. 317 318 Supplementary Movie 3: 319 Several S. schoenii cells are sequentially attacking C. auris cells. Arrows 320 indicate attacked and killed C. auris NCPF8985#20 cells. Image acquisition (4/min) 321 over 3 322

113 323 Figures 324 Figure 1

325

114 326 Figure 2

327

115 328 Supplementary Figure 1

329

116 330 Supplementary Figure 2

331 332 333

117 334 Supplementary Table 1:

335 118 336 Supplementary Table 2:

337 119 338 Supplementary Movie 1:

339 340 Available at: https://tinyurl.com/KlaraPhD 341 342 Supplementary Movie 2:

343 344 Available at: https://tinyurl.com/KlaraPhD 345 346 Supplementary Movie 3:

347 348 Available at: https://tinyurl.com/KlaraPhD

120