Candida glabrata and the stress responses

Get Complete Project Material File(s) Now! »

A hectic phylogenetic classification

The classification of the haploid hemiascomycete Candida glabrata has been quite vibrant for a long time. Due to the resemblance with some previously discovered fungi, it was first called Crypto-coccus glabratus after its discovery on grapes (Berlese, 1894) and in human stools (Anderson, 1917), a century ago. Its name switched to Torulopsis glabrata in 1938 (Lodder et al., 1938), after the authors noticed that the yeast wasn’t able to form pseudomycelium and had morphological and physiological characteristics resembling the ones of Torulopsis yeasts. Then, it changed to Candida glabrata in 1978 (Yarrow et al., 1978), following an important change in yeast classification and the disappearance of the Torulopsis genus. Finally, it was attributed to the Nakaseomyces clade by Kurtzman et al., 2003, after they performed a huge phylogenetic work thanks to the study of several genes, including rDNA genes, translation elongation factor, actin, RNA polymerase and mitochondrial genes. This clade also comprises three environmental species (N. bacilisporus, N. delphensis, C. castelli) added by Kurtzman et al., 2003, and two other pathogenic species (C. nivariensis and C. bracarensis) were added later to that clade (Alcoba-Flórez et al., 2005; Correia et al., 2006).

Close to Saccharomyces spp, far from Candida spp ?

The phylogenetic study of Kurtzman et al., 2003 was soon followed by the publication of the whole genomic sequence of C. glabrata ATCC2001 strain by the Genolevures consortium (Dujon et al., 2004). They revealed that C. glabrata genome is composed of 13 chromosomes, totaling around 12.3 Mb. The genome has an average GC content of 38.8% and encodes for 5283 CDS. It resembles the overall structure of S. cerevisiae which is 12.1 Mb long, has an average GC content of 38.3% and encodes for 5807 CDS. Some other genomic similarities between the two species can be found in Table 1.1. This article highlighted the closeness of C. glabrata and S. cerevisiae. They both underwent a Whole Genome Duplication (WGD) event (Wolfe et al., 1997; Dietrich et al., 2004; Kellis et al., 2004) before their divergence as distinct species, as shown by the shared duplicated blocks of sister chromosomal regions (Lalo et al., 1993; Dujon et al., 2004). C. glabrata also displays exactly the same set of 42 tRNA encoding genes than S. cerevisiae. The two species even share an average of 65% of sequence identity between orthologous proteins ; this is translated by the fact that approximately 4800/5300 genes in C. glabrata have an homologue in S. cerevisiae (Gabaldón et al., 2016).
However, as close as they may be, this 35% discrepancy between these two yeasts is the same than the difference between human sequences and zebrafish (Gabaldón et al., 2013). In a general way, Nakaseomyces spp genomes are smaller and contain fewer genes than S. cerevisiae (Gabaldón et al., 2013). One explanation could be that C. glabrata has a much higher rate of loss of duplicated genes (Du-jon et al., 2004). According to this team, C. glabrata lost so much paralogues that it resulted in reductive evolution, associated with loss of functions and decrease genome redundancy. Even more striking, C. glabrata genome redundancy is equivalent to that of Kluyveromyces lactis, a pre-WGD species. Mainly, the genes were lost in galactose metabolism, phosphate metabolism, cell rescue, defence and virulence and nitrogen and sulphur metabolism compared to S. cerevisiae and three other yeast species. Among other genomic differences, S. cerevisiae genome contains active transposon (Krastanova et al., 2005), while all Nakaseomyces species, including C. glabrata have transposon-free genomes (Gabaldón et al., 2013). Last but not least, C. glabrata is pathogenic while S. cerevisiae isn’t, which indicates some pro-found genomic differences in these yeast. As mentioned previously, C. glabrata has some specific genes with no homology to S. cerevisiae. Though one could have thought that these specific genes might be an explanation as to why C. glabrata is pathogenic, while S. cerevisiae is not, Gabaldón et al., 2016 indicated that most differences in gene content was not related to virulence.
Despite these differences and the fact that Candida albicans is also pathogenic, C. glabrata is even more distant from this yeast than it is from S. cerevisiae (Dujon et al., 2004; Fitzpatrick et al., 2006), as represented in Figure 1.1. Another major difference between C. glabrata and C. albicans is the change in C. albicans genetic code : in this species, the codon CTG is translated as a serine, while in other yeasts, including S. cerevisiae and C. glabrata, it is translated as a leucine.
C. albicans and C. glabrata have a common ecological niche, they can both cause infections in human ranging from benign to fatal and they even share a common species name. However, despite these common points, they still are very distant species according to their genomic sequences as well as some of their phenotypic features. For example, C. glabrata cells are usually described as budding yeast or pseudohyphae (under certain specific conditions such as limiting nitrogen conditions (Csank et al., 2000)), while C. albicans cells can be shaped into yeast, pseudohyphae, hyphae or chlamydospores. A more complete comparison of several features in these two yeasts and S. cerevisiae can be found in Table 1.2. Nevertheless, these two Candida species are both pathogen, and this pathogenicity had a big impact in bringing them under the focus of research.

An « emerging » opportunistic pathogen

Candida glabrata began to draw attention only long after its discovery, even though it was identified as early as the beginning of the 20th century. It only received significant consideration in the late 80’s, when it was noticed to be one of the main cause of fungal infections in immunocompromised individuals (Just et al., 1989). Following this statement, the increasing incidence of this yeast was acknowledged in the late 20th century when it was declared as an emerging pathogen (Hazen, 1995). It was also noticed the phyloge-netic tree inferred by X.-X. Shen et al., 2016. The tree was inferred from the concatenation-based analysis of 1233 single-copy orthologues. The Whole Genome Duplication is indicated by a black star. The CUG-clade is depicted by red branches on the tree. C. glabrata, C. albicans and S. cerevisiae are highlighted in bold.
that its prevalence tends to increase with age, antibiotic treatment, length of stays in hospitals, diabetes mellitus. . . (Angoulvant et al., 2016).
Usually, Candida glabrata is a harmless fungi of the human gut microbiota. However, given some specific conditions, it can proliferate and trigger superficial benign infections such as oral or vaginal thrush, gastrointestinal tract infection or urinary bladder infection (Pfaller et al., 1998; Fidel et al., 1999), hence the adjective “opportunistic”. In the worst cases, it can breach the mucosal barriers, enter the bloodstream and disseminate throughout the body, causing systemic candidiasis infections with high mortality rates (40-60%), especially in immunocompromised individuals (cancer patients undergoing chemotherapy, transplanted patients, for example) and elder people (Angoulvant et al., 2016).
et al., 2007; Azie et al., 2012) and Candida spp are in the top five causes of nosocomial bloodstream in-fections (Gudlaugsson et al., 2003; Pfaller et al., 2007). Most of the time, Candida albicans is the main cause of candidiasis (50-70%), followed by C. glabrata (20-30%) (Perlroth et al., 2007). However, the role of C. glabrata as cause of candidiasis is rapidly increasing, perhaps because of its natural resistance to the compounds usually administered as treatment of this disease, namely azoles (Jandric et al., 2011; Pfaller, 2012). It can also easily acquire resistance to another class of antifungal, the echinocandins. It was shown that resistance increases in case of a pre-treatment. Additionally, a phenomenon of cross-resistance to other drugs can appear in strains already resistant to fluconazole (Komshian et al., 1989; Malani et al., 2005). One of the principal cause of the innate resistance of C. glabrata is the increased efflux of drugs driven out of the cells by pumps, usually regulated by PDR1. When this gene is over-expressed, the cells tend to bear an increased resistance and a increased virulence. Also, some changes in the cell wall have been reported to prevent drug diffusion (Parkinson et al., 1995; Clark et al., 1996; Vermitsky et al., 2004; Ferrari et al., 2011a). All these mechanisms are reviewed in Sanglard, 2002.
C. glabrata and C. albicans can both easily infect humans, however their strategies are different on several points (reviewed in Brunke et al., 2013) :
Adhesion : C. albicans have Als and Hwp adhesins, while C. glabrata has Epa adhesins. Also, both species can form biofilms on host cells or medical devices (Iraqui et al., 2005; Nobile et al., 2006).
Invasion : C. albicans mainly use its hyphae form to gain access to epithelial cells. The invasion mechanism in C. glabrata is not known yet, as this yeast doesn’t form true hyphae. However, it was noticed that C. glabrata infections cause reduced inflammatory reaction from the host, contrary to C. albicans. It might suggest a smoother way for C. glabrata to enter its host.
Interaction with immune cells : The two species have very different approach : shortly after internaliza-tion in macrophages, C. albicans uses hyphae to burst them, while C. glabrata uses the phagosome as a safe haven to multiply before bursting it. Once again, this has the advantage to be more quiet against the immune system of the host.
Nutrient acquisition : C. albicans is much better equipped to face the host in nutrients acquisition. It has no known auxotrophy and owns several integration pathways for a lot of nutrients, including metallic ions. On the other hand, C. glabrata loss of genes after WGD caused the disappearance of several important pathways (mentioned in Section 1.2). This lack has to be compensated for the cells to survive and thrive.
All these mechanisms are summarised in Figure 1.2. Finally, given these dissimilarities and the phylogenetic disparity between C. glabrata and C. albicans, it is likely that their pathogenicity and their infection strategies came from different and independent evolutions.

glabrata genome plasticity : a huge impact

Over the years, it became clear that Candida glabrata possesses an impressive genome plasticity and can be rearranged easily. Several teams reported some interesting features and consequences to this very dynamic nature of the C. glabrata genome, which could go as far as considering the rearrangements as an adaptive mechanism. The first clue might be the high rate of loss of paralogues after the WGD, leading to a reductive evolution (Dujon et al., 2004) (Table 1.1). Shin et al., 2007 reported that changes in karyotype can appear quickly in strains isolated at different time intervals in the same patient without antifungal therapy. They also showed that these strains were able to acquire azole resistance under treatment with azole compounds. Ahmad et al., 2013 reported that the number of chromosomes of C. glabrata can vary among different isolates. Especially, they found extra chromosomes created by segmental duplication and translocation. They supposed that it was a way for C. glabrata to improve its fitness around new environments. Poláková et al., 2009 also noticed some size variation in the chromosomes, but more im-portantly, they remarked that new supernumerary chromosomes carried duplicated genes among which  glabrata pathogenicity. Adapted from Brunke et al., 2013.
the orthologues of S. cerevisiae family of ABC transporters, which plays a role in pleiotropic drug re-sistance and yeast-host interaction genes. Overall, these karyotype variations correlated with antifungal drug resistance.
C. glabrata genome contains the family of EPA genes (homologues of the FLO genes in S. cere-visiae), coding for cell-wall proteins, which are known to be responsible for the adherence to human epithelium (Cormack et al., 1999; Gabaldón et al., 2013). Coincidentally, Muller et al., 2009 reported size variation in tandem arrays of repeated genes, which often encode the EPA genes, suggesting a role in adaptation to the host (Figure 1.3). It also suggests a role in virulence because these tandem repeats were also found to encode aspartyl proteases responsible for adherence to mammalian cells and survival in macrophages (Kaur et al., 2007).
However, this genomic changes also happen very quickly in lab strains : Bader et al., 2012 noticed some karyotypic modifications that could be associated to phenotypic variations. It shows that genome plasticity occurs not only during the harsh conditions of host infection, but also in laboratory conditions during a very short time and without strong selective pressures.
Hence, it seems that chromosome remodeling and gene duplications play an important role in the specialization to specific environments, virulence, and interaction with the host (Butler et al., 2009; Moran et al., 2011; Gabaldón et al., 2018).

Breaking C. glabrata clichés

A commensal yeast ?

andida glabrata is often considered as a human commensal, even if its prevalence can vary a lot between studies and is influenced by a range of factors, including the age, the previous use of antifungal treatments and the medical condition of a patient. C. glabrata has been detected in human flora (oropha-ryngeal, digestive, vaginal), on medical devices (Iraqui et al., 2005), phones (Kordecka et al., 2016) but it can also be found in non-human related niches such as fermenting coffee beans (de Melo Pereira et al., 2014) or cloaca of several bird species (Cafarchia et al., 2008; Francesca et al., 2014; Al-Yasiri et al., 2016). These are all diverse isolation sites, sometimes without any link between them (for example, mobile phone and bird cloaca or with human (for example, bird cloaca and human). It suggests that either C. glabrata is a commensal of several sources, or that it contaminated all these secondary sources from a primary source that is still unknown. This supposition is coherent with the fact that C. glabrata is not always found in these niches. This type of behaviour has been shown for other yeasts, such as C.
albicans (Bensasson et al., 2018) or S. cerevisiae (Goddard et al., 2015).
FIGURE 1.4. Phylogenetic structure of 33 Strains of C. glabrata. This tree indicates the phylogenetic structure of the strains based on SNPs data analysis. Total number of SNPs isn’t displayed, but a range of 4.66 to 6.56 SNPs per kb per strain when compared to the reference strain (as sequenced by Dujon et al., 2004) is observed. Each colour designates a clade. Super-indices indicate pairs of strains isolated in the same patient, but on different body site or at a different date. Body site and country of the isolation are displayed, along with the mating type of the strains. Adapted from Carreté et al., 2018.
One strong argument in favour of the human commensality of C. glabrata is the existence of phylo-genetically distinct clades among the yeast population related to specific geographical origins. In other words, some distinct clades were defined according to genetic and phenotypic markers, and these clades were specific to a geographical zone (Dodgson et al., 2003; Dodgson et al., 2005; Brisse et al., 2009; Rolland et al., 2010; Schwarzmüller et al., 2014). However, these studies were only performed on a few markers, and consequently, the number of detected clades varied a lot. This problem was solved by the study of Carreté et al., 2018, who assessed genomic and phenotypic variations in 33 C. glabrata isolates coming from all over the world. They were able to sequence these strains and class them into tree with seven clades (Figure 1.4). On this tree, it appears quite clear that strains of the same clade clustered together despite their geographical origin (for example, cluster I with USA and Belgium), which would go against a human commensality of C. glabrata : this undermines the idea of a genomic co-evolution between C. glabrata and human, as we could have expected if there was a commensal relationship be-tween them. It lets us think that human activities and transportations caused the crossing of strains that have been isolated for a long time, most likely in specific geographical areas.
However, C. glabrata displays several traits that let us think of it as a commensal of human. The optimal growth temperature of C. glabrata is close to 37°C, the normal internal temperature of the human body. This is a huge advantage as both a commensal and an opportunistic pathogen. C. glabrata has a higher stress resistance and an enhanced ability to sustain prolonged starvation. It can withstand very high concentrations of H2O2 (also considered as oxidative stress), varying concentration of bio-available iron, lack of oxygen, among others, and can also resist to shortage of nutrients caused by protective mechanisms of the human host. Especially, the adaptation to use alternative nutrients and survive long periods of starvation is essential for C. glabrata to survive macrophage engulfment (Roetzer et al., 2010) and use the macrophage as a “Trojan horse” to invade the whole body of the host. This type of adaptation reminds the bacteria surviving internalization in amoebae and the subsequent oxidative and osmotic stresses (Greub et al., 2004), as well as the likely selection for bacterial virulence and resistance traits (Tosetti et al., 2014; Hao et al., 2015). Additionally, earlier, C. glabrata can also resist to antifungal drugs and acquire new resistances, which is an useful mechanisms for a human pathogen. Also, C. glabrata genome has remodeled its cell-wall components resulting in a higher adherence. The ability to adhere to the host tissues is mediated by cell-wall-associated proteins called adhesins (reviewed in Groot et al., 2013). This translates into the presence of a high number of particular adhesins encoded by the EPA family of genes . This adhesins proved to be crucial in virulence and to form biofilms (Cormack et al., 1999; Roetzer et al., 2011a), an ability which increases resistance, virulence and persistence in the host. Finally, C. glabrata has lost several hundreds of genes, including genes in the pyridoxine, thiamine and nicotinic acid biosynthetic pathways (Dujon et al., 2004; Kaur et al., 2005). The fact that the human host is a really stable growth environment, because all its parameters (temperature, oxygen, nutrients. . . ) are regulated through homeostasis, could be an explanation to the loss of specific pathways in C. glabrata. It also means that this yeast has a higher dependence of the host, and rely on it to acquire nutriments. All these features tend to show an adaptation of C. glabrata to its human host.
Nonetheless, several of these features (growth at 37°C, loss of the nicotinic acid pathway and pres-ence of auxotrophies) are shared by all Nakaseomyces species, pathogenic and non-pathogenic, and the expansion of the EPA family was linked to virulence in the pathogenic Nakaseomyces (C. glabrata, C. bracarensis, C. nivariensis). This means that not all these changes are specific to C. glabrata and its com-mensalism (or its pathogenicity). Hence, C. glabrata clearly displays commensal traits but the repercus-sions of this commensalism on C. glabrata genome are not clearly identified yet. This could be because C. glabrata might be a recent human commensal (which hasn’t left genomic traces yet) or/and because we still lack experimental data on the comparisons between C. glabrata and other Nakaseomyces.
C. glabrata is believed to be an asexual haploid fungus, which reproduces exclusively by budding. Though it is haploid and can bud, it might not be as asexual as we thought, even if its mating has never been actually observed. Yeast in the same clade than C. glabrata and S. cerevisiae often have two mating types in haploid cells. These haploid cells can merge, forming diploids, which can then undergo meiosis and sporulation (Muller et al., 2008). To increase their adaptive potential, haploid cells can also switch mating types through recombination with silent loci after an HO endonuclease cut (Haber, 2012).
One of the reasons supporting a sexual reproduction of C. glabrata is the existence and the obser-vation of mating types a and a (Muller et al., 2008; Carreté et al., 2018). Even more striking, Brockert et al., 2003 reported mating type switching during infection of a host and Butler et al., 2004 reported mating type switching during laboratory growth. Interestingly, C. glabrata possesses a seemingly intact mating machinery in its genome (S. Wong et al., 2003), composed of orthologues of genes involved in mating in S. cerevisiae : MTL1, MTL2 and MTL3 genes were found in C. glabrata (Srikantha et al., 2003; Butler et al., 2004). MTL1 is the expression locus, while MTL2 and MTL3 encodes a and a mating informations. The genome also contains an homothallic (HO) endonuclease, which is known to be re-sponsible for gene conversion events that underlies mating type switching in S. cerevisiae. Additionally, Butler et al., 2004 found a putative HO endonuclease recognition site in MTL1.
All these findings were further confirmed by Carreté et al., 2018, who furthermore demonstrated the existence of illegal repair events during switching (Figure 1.5). They also found traces of selective con-straints in mating genes. Even more interesting, they found the same levels of constraint in C. glabrata than in S. cerevisiae or C. albicans, which have functional sexual and parasexual cycles, respectively. This tends to show that sexual genes in C. glabrata are still under selective pressure, most likely because they are still functional. Also, switching is lethal to many C. glabrata cells in laboratories Boisnard et al., 2015, which shows that the switch is tightly regulated. Besides, they discovered evidences of chimeric patterns in chromosomes, which is the result of sexual mating between yeasts of different clades.
FIGURE 1.5. Example of illegal repair events during switching in C. glabrata. BS means Before Switching. AS means After Switching. Adapted from Carreté et al., 2018. Case 1 represents a normal conversion event at MTL1, which switched mating type from a to a. Case 2 shows the classic a-to-a switch at MTL1 accompanied by illegitimate conversion at MTL2, resulting in a triple-a strain. Case 3 displays an illegitimate MTL2 conversion without the MTL1 switch. Case 4 represents an illegitimate MTL3 conversion without the MTL1 switch, resulting in a triple-a strain.
To summarize, Candida glabrata keeps distinct a and a haploid mating types and distinct associated cellular identities (Muller et al., 2008) ; it shows evidences of mating type switches in populations issued of isolates and laboratory populations (Brockert et al., 2003; Butler et al., 2004) and sexual recombination (Dodgson et al., 2005; Carreté et al., 2018). All this lets us think that C. glabrata might still be mating but rarely and we just haven’t found yet the adequate conditions to observe it.

Table of contents :

1 Candida glabrata, a yeast with many faces
1.1 A hectic phylogenetic classification
1.2 Close to Saccharomyces spp, far from Candida spp ?
1.3 An « emerging » opportunistic pathogen
1.4 C. glabrata genome plasticity : a huge impact
1.5 Breaking C. glabrata clichés
1.5.1 A commensal yeast ?
1.5.2 An asexual yeast
2 Candida glabrata and the stress responses 
2.1 A general response : the Environmental Stress Response (ESR)
2.2 Condition-specific stress responses
2.2.1 Rox1, a mediator of response to hypoxia
2.2.2 The oxidative stress response (OSR)
2.2.3 The CCAAT-Binding Complex as a link between oxidative stress and respiration
2.2.4 The iron homeostasis is linked with respiration and oxidation
2.2.4.1 Response to iron-depleted media
2.2.4.2 Response to iron-excess
3 Candihub, the study of transcriptional networks of Candida glabrata stress responses
3.1 Networks : a model to represent connections between various elements
3.2 The transcriptional regulatory networks
3.3 The Candihub project : deciphering stress responses in Candida species using transcriptional
regulatory networks
3.3.1 Description and goals of the Candihub project
3.3.2 Building the Candihub networks
3.3.2.1 Types of C. glabrata strains used in this work
3.3.2.1.1 DHTL strain
3.3.2.1.2 TF deleted strains
3.3.2.1.3 myc-tagged TF strains
3.3.2.2 Approaches to build networks
3.3.2.2.1 Mathematical inference of regulatory networks from experimental data
3.3.2.2.2 Direct determination of regulatory networks
3.3.3 Proof of concept of Candihub : the Yap network
3.3.4 Goals of my PhD
4 Stress responses in Candida glabrata : a highly interconnected network 
4.1 Introduction
4.2 Selection of the ChIP conditions
4.3 Sequencing of the immuno-precipitated DNA
4.4 Peak-calling and identification of the targets
4.5 Representing the Candihub network
5 The CBC Impacts Respiratory Genes and Iron Homeostasis in Candida Glabrata 
5.1 Introduction
5.2 Publication
5.3 Supplementary results
5.3.1 Introduction
5.3.2 The YRE and the CCAAT motifs are required to activate the iron excess response
5.3.3 The YRE and the CCAAT motifs are differentially conserved in the Saccharomycetales
5.3.4 Hap4 might still interact with the CBC during iron excess response
5.3.5 Conclusion
6 Comparative Transcriptomics Reveals New Features of Iron Starvation in Candida glabrata 
6.1 Introduction
6.2 Publication
6.3 Supplementary results
6.3.1 Introduction
6.3.2 Aft1 network in Candida glabrata
6.3.3 C. glabrata Aft1 has several functions shared with S. cerevisiae Aft1
6.3.4 C. glabrata Aft1 and Aft2 roles are only partially redundant
6.3.5 Relationship between Aft factors in S. cerevisiae share some features with C. glabrata
6.3.6 Relationship between Aft factors in more distant species
6.3.7 GRX3 and GRX4 are involved in iron-deprivation response under different regulations
6.3.8 Conclusion
Conclusion and perspectives

GET THE COMPLETE PROJECT