Contribution of Fdh3 and Glr1 to Glutathione Redox State, Stress Adaptation and Virulence in Candida albicans

The major fungal pathogen of humans, Candida albicans, is exposed to reactive nitrogen and oxygen species following phagocytosis by host immune cells. In response to these toxins, this fungus activates potent anti-stress responses that include scavenging of reactive nitrosative and oxidative species via the glutathione system. Here we examine the differential roles of two glutathione recycling enzymes in redox homeostasis, stress adaptation and virulence in C. albicans: glutathione reductase (Glr1) and the S-nitrosoglutathione reductase (GSNOR), Fdh3. We show that the NADPH-dependent Glr1 recycles GSSG to GSH, is induced in response to oxidative stress and is required for resistance to macrophage killing. GLR1 deletion increases the sensitivity of C. albicans cells to H2O2, but not to formaldehyde or NO. In contrast, Fdh3 detoxifies GSNO to GSSG and NH3, and FDH3 inactivation delays NO adaptation and increases NO sensitivity. C. albicans fdh3⎔ cells are also sensitive to formaldehyde, suggesting that Fdh3 also contributes to formaldehyde detoxification. FDH3 is induced in response to nitrosative, oxidative and formaldehyde stress, and fdh3Δ cells are more sensitive to killing by macrophages. Both Glr1 and Fdh3 contribute to virulence in the Galleria mellonella and mouse models of systemic infection. We conclude that Glr1 and Fdh3 play differential roles during the adaptation of C. albicans cells to oxidative, nitrosative and formaldehyde stress, and hence during the colonisation of the host. Our findings emphasise the importance of the glutathione system and the maintenance of intracellular redox homeostasis in this major pathogen.


Introduction
The major fungal pathogen, Candida albicans, has the capacity to colonise diverse niches in its human host. This fungus is part of the microflora of the skin, mouth, gut and urogenital tracts of humans. C. albicans is a frequent cause of mucosal infections (thrush) in otherwise healthy individuals, with most women suffering at least one episode of vaginitis in their lifetime. C. albicans is also the most common fungal species responsible for life-threatening hospital-acquired bloodstream infections in immunocompromised patients [1][2][3].
The ability of C. albicans to thrive in diverse niches is dependent upon its robust adaptive responses to the local environmental stresses encountered in these niches. For example, immune cells such as macrophages combat microbial infection by exposing invading microbes to a battery of insults that include reactive nitrogen species (RNS), reactive oxygen species (ROS) and cationic fluxes [4][5][6]. Macrophages have been reported to generate up to 57 μM nitric oxide [7] and up to 14 mM hydrogen peroxide (H 2 O 2 ) [8,9], although estimating the levels of specific ROS species during the oxidative burst is challenging [10,11]. Cation concentrations are around 150 mM in human blood, have been reported reach 0.2-0.3 M in phagocytes [5] and can increase to 600 mM in the kidney [12]. Therefore, C. albicans cells are exposed to significant oxidative, nitrosative and cationic stresses during host colonization and invasion [13][14][15].
Glutathione (L-γ-glutamylcysteinylglycine; GSH), the most abundant non-protein thiol in eukaryotic cells, plays a major role in protective responses to oxidative and nitrosative stress. Glutathione reacts with reactive oxygen and nitrogen species to generate glutathione adducts, such as glutathione disulphide (GSSG) and S-nitrosoglutathione (GSNO). In addition to detoxifying xenobiotics and free radicals, glutathione functions as a co-factor in many enzymatic reactions and is involved in amino acid transport and signalling [16,17]. S-glutathionylation protects protein thiols from irreversible over-oxidation [18]. Glutathione maintains intracellular redox homeostasis through the oxidation of its cysteine sulphydryl moiety and disulphide bond formation [19]. The low redox potential of glutathione (E = -240 mV), combined with its high intracellular concentration (1-10 mM), contribute to its large redox buffering capacity [20,21]. Fungal genes involved in both glutathione synthesis and the recycling of GSSG and GSNO are tightly regulated in response to stress exposure [21,22].
Glutathione is synthesised in two ATP-dependent steps. In the first rate-limiting step, γ-glutamylcysteine synthetase (Gsh1 in S. cerevisiae) ligates glutamate and cysteine to form the dipeptide γ-glutamylcysteine [23]. In the second enzymatic step glutathione synthetase (Gsh2 in S. cerevisiae) converts γ-glutamylcysteine and glycine into glutathione [24]. In S. cerevisiae, Gsh1 catalyses the rate-limiting step in GSH synthesis and GSH1 deletion significantly increases the intracellular redox potential to -178 mV, compared with -235 mV in wild type cells [25]. Consequently, S. cerevisiae gsh1 cells are sensitive to oxidative stress. This mutant also displays growth defects on minimal medium in the absence of stress [23], a phenotype that can be suppressed by the addition of exogenous GSH [26]. These results indicate that GSH is an essential metabolite for the growth of S. cerevisiae. Exogenous GSH also confers protection against oxidative stresses mediated by disulfiram [27], hypochlorite, chlorite [28] and heavy metals [29]. In C. albicans, the deletion of GCS1 (the homologue of S. cerevisiae GSH1) causes cells to upregulate typical apoptotic markers, reduces their resistance to killing by human macrophages and decreases their virulence in a murine model of disseminated candidiasis [30], [31]. These observations highlight the importance of glutathione in general and for C. albicans pathogenicity in particular.
Following the formation of glutathione adducts, such as GSSG and GSNO, these molecules are recycled to reform glutathione via glutathione reductase and S-nitrosoglutathione reductase

Results
The C. albicans genome encodes a putative glutathione reductase and an S-nitrosoglutathione reductase A single glutathione reductase gene has been annotated in the C. albicans genome on the basis of its sequence similarity to glutathione reductases (GRs) from other species [39]. C. albicans GLR1 (C5_01520C) encodes a highly conserved NADPH-dependent glutathione reductase (S1 Fig) [40] that carries the classical NADH and FAD binding domains and dimerization domain of GRs ( Fig 1A). Phylogenetically, C. albicans Glr1 is most closely related to GRs from other fungi, displaying 66% amino acid sequence identity to S. cerevisiae Glr1 (Fig 1B). Fungal GRs cluster with human, mouse and Caenorhabditis elegans GRs, clearly separable from GSNORs from the same organisms ( Fig 1B).
C. albicans also contains a single GSNOR gene, FDH3 (CR_10250C_A) [39]. Based on its sequence similarity to GSNORs from other organisms (S2 Fig), FDH3 appears to encode a GSH-dependent formaldehyde dehydrogenase class III that contains alcohol dehydrogenaselike, NADH binding and dimerization domains (Fig 1A). GSNORs are highly conserved from bacteria to man [21]. Indeed, the sequence of C. albicans Fdh3 is 65% identical to human liver ChiChi alcohol dehydrogenase (Hs Adh5) (Fig 1B), which is a GSH-dependent formaldehyde dehydrogenase.
C. albicans Fdh3 and Glr1 play differential roles in conferring formaldehyde, nitrosative and oxidative stress resistance nitrosative and oxidative stress [21]. In C. albicans, Glr1 is presumed to act as a GR [39], and we reasoned that Fdh3 might act as the GSNOR, detoxifying GSNO and catalysing the detoxification of formaldehyde (CH 2 O) to formate (CHOO -) (Fig 2).
To test these predictions we constructed C. albicans glr1Δ and fdh3Δ null mutants, and then reintegrated wild type GLR1 and FDH3 genes to generate isogenic control strains. As predicted, the glr1Δ mutant was sensitive to hydrogen peroxide (7.5 mM), a phenotype that was suppressed by reintegration of GLR1. The glr1Δ mutant was resistant to formaldehyde (5 mM) ( Fig 3A). In contrast, the fdh3Δ mutant was resistant to hydrogen peroxide, but sensitive to formaldehyde (Fig 3A and 3B). Reintegration of FDH3 suppressed this sensitivity, confirming that Fdh3 is critical for formaldehyde resistance. These observations were consistent with the predicted differential roles for Fdh3 and Glr1 in the detoxification of formaldehyde and peroxide-induced GSSG, respectively.
We also investigated the impact of inactivating Fdh3 or Glr1 on the resistance of C. albicans cells to nitrosative stress (2.5 mM DPTA NONOate). As predicted (Fig 2), fdh3Δ cells were relatively sensitive to nitrosative stress, taking significantly longer than wild type cells to adapt to this stress (Fig 3C). Taken together, our data are consistent with the idea that the Fdh3 is a bifunctional enzyme involved in the detoxification of both formaldehyde and GSNO. Phylogenetic tree of GSNOR-and GR-related proteins generated using ClustalW: homologs are presented from Candida albicans (CaFdh3, CaGlr1), Saccharomyces cerevisiae (ScSfa1, ScGlr1), Schizosaccharomyces pombe (SpSPCC13B11.04c, SpPgr1), Mus musculus (MsAdh5; MsGsr1), Homo sapiens (HsAdh5; HsGsr), Drosophila melanogaster (DmFdh) and Caenorhabditis elegans (CeH24K24.3; CeGsr-1). Structures are presented for human liver ChiChi alcohol dehydrogenase (protein data bank (pdb) accession code 1TEH; a GSNOR that has 65% sequence identity to CaFdh3p), and S. cerevisiae Glr1 (pdb accession code 2HQM; a GR with 66% sequence identity to CaGlr1p). Structure representations were made with PyMOL (http://www.pymol.org).  Surprisingly, glr1Δ cells were relatively resistant to nitrosative stress, adapting more quickly to DPTA NONOate than the wild type and GLR1 reintegrant control strains ( Fig 3C). To examine the basis for this we tested the impact of GLR1 or FDH3 deletion upon the expression of each mRNA by qRT-PCR (Fig 4). As expected, the GLR1 and FDH3 mRNAs were not detectable in their respective mutant strains. GLR1 mRNA levels were reduced about 4-fold in fdh3Δ cells, relative to the wild type control. However, FDH3 mRNA levels were elevated about 2.5-fold in glr1Δ cells (Fig 4). This increased FDH3 expression could account for the increased nitrosative stress resistance of the glr1Δ strain ( Fig 3C).

Impact of Fdh3 and Glr1 deletion upon gene expression
The deletion of either FDH3 or GLR1 affects the expression of the other gene under basal conditions in the absence of stress (Fig 4). Therefore, we reasoned that the expression of these genes might also be perturbed in glr1Δ and fdh3Δ cells following exposure to oxidative, and nitrosative or formaldehyde stress. To test this we measured the expression levels of selected genes involved in stress defence by qRT-PCR, relative to the ACT1 mRNA internal control, following a 10 min exposure to 5 mM H 2 O 2 , 2.5 mM CySNO or 5 mM formaldehyde. GLR1 is known to be induced in response to oxidative stress [41], and therefore we examined its expression in response to these stresses in wild type and fdh3Δ cells (Fig 5A). The GLR1 mRNA was up-regulated in response to oxidative and nitrosative stress, but was down-regulated following formaldehyde stress. Deletion of FDH3 had a minimal effect upon this expression profile, except that GLR1 expression was no longer reduced in response to formaldehyde stress. FDH3 displayed a similar expression profile to GLR1 in wild type cells (Fig 5B). FDH3 was up-regulated following exposure to both oxidative and nitrosative stress, and down-regulated following formaldehyde stress. This regulation was perturbed in glr1Δ cells, primarily because basal FDH3 mRNA levels were increased in the absence of stress, which was consistent with our previous observations (Fig 4).
We also examined the mRNA levels for TRX1 (thioredoxin) under equivalent conditions ( Fig 5C). As reported previously [41][42][43], TRX1 was induced in response to oxidative and and GSSG detoxification. A shows the working hypothesis of the major enzymes involved in the detoxification of GSSG (glutathione disulphide) and GSNO (S-nitrosoglutathione). When glutathione (GSH) is oxidised via H 2 O 2 to GSSG, GSSG can be reduced with the help of the NADPH-dependent glutathione reductase (GR). We predict that the glutathione reductase of Candida albicans is GLR1. When GSH is exposed to NO, GSH is S-nitrosylated to GSNO. We predict that the S-nitrosoglutathione reductase (GSNOR) of Candida albicans is FDH3. nitrosative stress. Interestingly, the response to oxidative stress was unaffected in fdh3Δ cells, but the response to nitrosative stress was attenuated. This is consistent with the idea that in the absence of Fdh3, GSSG generation might be reduced following nitrosative stress (Fig 2). In glr1Δ cells, but not in fdh3Δ cells, TRX1 was strongly induced in response to formaldehyde stress ( Fig 5C). This supports the hypothesis that intracellular GSSG levels might accumulate following formaldehyde stress in cells that lack Glr1, but contain Fdh3 (Fig 2).

Effect of Glr1 and Fdh3 inactivation upon GSSG and GSNO detoxification
The data obtained are consistent with the predicted roles for Glr1 and Fdh3 in GSSG and GSNO detoxification (Fig 2). We then tested these predictions directly by examining the impact of deleting GLR1 or FDH3 upon GSSG and GSNO detoxification rates.
GSSG detoxification was assayed by LC-MS/MS. GSSG detoxification rates were minimal in the absence of protein extract (buffer alone), but the majority of GSSG was detoxified within 10 min by wild type protein extracts ( Fig 6A). Deletion of GLR1 completely blocked GSSG detoxification rates, and these rates were restored by reintegration of the wild type GLR1 gene. GSSG detoxification rates were also reduced in fdh3Δ cells, which could be due to the 4-fold reduction in GLR1 expression in these cells (Fig 4). We conclude that GLR1 is essential for rapid GSSG detoxification in C. albicans, which is consistent with the view that this gene does indeed encode a glutathione reductase.
GSNO detoxification was also assayed in wild-type and mutant cells ( Fig 6B). As expected (Fig 2), GLR1 inactivation had no major effect upon GSNO detoxification, whereas rates of GSNO detoxification were reduced in fdh3Δ cells ( Fig 6B). However, GSNO detoxification was not completely blocked following Fdh3 deletion, suggesting that other C. albicans enzymes, in addition to Fdh3, must have denitrosylation or trans-nitrosylation activity. The identity of these enzymes remains to be determined.
Given the impact of Glr1 inactivation upon GSSG detoxification in particular, the glr1Δ mutation is likely to perturb redox homeostasis in C. albicans cells. The GSSG/GSH redox potential (ΔE), which essentially reflects the relative amounts of reduced versus oxidized glutathione, was measured in mutant and wild type cells (Fig 6C). The GSSG/GSH redox potential of C. albicans fdh3Δ cells displayed no significant difference from wild type or FDH3 reintegrant controls, despite the reduced ability of this mutant to detoxify GSSG ( Fig 6A). However, the GSSG/GSH redox potential of glr1Δ cells was significantly elevated, which correlated with their inability to detoxify GSSG (Fig 6A and 6C). Indeed, the redox potential of glr1Δ cells was close to the value of -180 mV, which has been associated with cell death in other systems [20,44]. This was consistent with the relatively slow growth of glr1Δ colonies (Fig 1A). Both FDH3 and GLR1 influence C. albicans virulence The above data support the hypothesis that Glr1 is a glutathione reductase playing a critical role in the detoxification of GSSG and maintenance of redox homeostasis in C. albicans, and that Fdh3 is a bi-functional enzyme playing critical roles in detoxification of GSNO and formaldehyde (Fig 2). Given the importance of oxidative and nitrosative stress in C. albicans-phagocyte interactions during infection [45,46], we reasoned that changes in GLR1 and FDH3 functionality might affect the ability of C. albicans cells to kill phagocytes. To test this, we measured the percentage of cultured murine macrophages (RAW264.7 cells) that were killed by wild type and mutant C. albicans cells. The glr1Δ and fdh3Δ mutants displayed significantly reduced ability to kill macrophages when compared to the wild type and reintegrant controls (Fig 7). Furthermore, doxycycline-dependent overexpression of GLR1 or FDH3 from the tetON promoter [47,48] increased the ability of C. albicans to kill macrophages (Fig 7), when compared to the parental strain. We conclude that the functions of Glr1 and Fdh3, and their contribution to the maintenance of redox homeostasis, are integral to C. albicans viability and potency during fungus-phagocyte interactions.
Based on their effects during fungus-phagocyte interactions, it was conceivable that GLR1 and FDH3 might influence the virulence of C. albicans during systemic infection. This was first tested in the invertebrate wax moth (Galleria mellonella) larval infection model, which has been established as a suitable proxy for systemic infection of the mammalian host [49]. Wild type C. albicans cells killed more than 90% of larvae within two days (Fig 8A). The fdh3Δ cells displayed a small, but reproducible, statistically significant reduction in virulence, whilst deletion of GLR1 resulted in a much greater effect on virulence (Fig 8A). Doxycycline-induced overexpression of GLR1 and FDH3 dramatically increased the virulence of C. albicans (Fig 8B). Experiments utilising the overexpression mutants was performed with a reduced C. albicans inoculum to permit resolution of the increased virulence (1 x 10 4 cells versus 2.5 x 10 5 cells). These infection data are consistent with the macrophage killing assays (Fig 7) and show that both Glr1 and Fdh3 contribute to the virulence of C. albicans.
Having established the impact of GLR1 and FDH3 on C. albicans virulence in the G. mellonella infection model, we then progressed to a mammalian model of systemic candidiasis. The virulence of the mutants was assayed using a three-day murine intravenous challenge model [50], which allows calculation of an outcome score. In this model, fdh3Δ cells displayed no significant difference in outcome score compared to the wild type and reintegrant controls ( Fig  9A). However, the glr1Δ mutant showed a significant reduction in virulence. It is worth noting that Fdh3 inactivation had less of an impact than Glr1 inactivation upon virulence in the G. Fdh3 and Glr1 Contribute to Candida Stress Adaptation and Virulence mellonella model (Fig 8A). This subtle effect might not have been detectable in the mouse experiments ( Fig 9A).
We then examined the virulence of the GLR1 and FDH3 overexpression strains. We noted that the treatment of animals with doxycycline, which was required to induce GLR1 and FDH3 overexpression, resulted in increased outcome score for infection with the wild type C. albicans strain. Therefore, the virulence of the test strains was compared to this control (WT + DOX: Fig 9B). Once again, GLR1 overexpression resulted in increased virulence of C. albicans. However, for reasons that are not clear,

Discussion
Numerous studies support the view that robust oxidative stress responses contribute to the pathogenicity of the major fungal pathogen, C. albicans. For example, inactivation of certain regulators that contribute to oxidative stress adaptation, such as the AP-1-like transcription factor Cap1 and the stress activated protein kinase Hog1, affect C. albicans virulence, albeit Fdh3 and Glr1 Contribute to Candida Stress Adaptation and Virulence only slightly in the case of Cap1 [51][52][53]. Key enzymes involved in ROS detoxification, such as catalase (Cat1) and superoxide dismutase (Sod1, Sod5), are also essential for full virulence [13,[54][55][56]. In addition, C. albicans mutants lacking certain components of the glutaredoxin (Gcs1, Grx2) and thioredoxin systems (Trx1) or mitochondrial-associated stress functions (Goa1) display attenuated virulence [31,[57][58][59]. Therefore, we reasoned that key enzymes involved in the detoxification and recycling of glutathione adducts formed during oxidative stress are likely to be important for the maintenance of redox homeostasis in C. albicans and its  Fdh3 and Glr1 Contribute to Candida Stress Adaptation and Virulence virulence. When this work was initiated, we predicted that two annotated C. albicans genes might play key roles in GSSG and GSNO recycling. The first was Fdh3, a putative glutathionedependent S-nitrosoglutathione reductase (GSNOR), which catalyses the generation of GSSG during the recycling of GSNO adducts. The second was a putative glutathione reductase (GR), Glr1, which was predicted to regenerate GSH from GSSG in an NADPH-dependent manner. Our bioinformatics analyses of C. albicans Glr1 and Fdh3 indicated that they are members of phylogenetically distinct GR and GSNOR families (Fig 1, and S1 and S2 Figs). We also noted that Fdh3 is related to bi-functional enzymes that also have formaldehyde dehydrogenase activity. Therefore we predicted that these highly conserved enzymes play differential but key roles in GSSG and GSNO recycling and formaldehyde detoxification (Fig 2), and in the maintenance of the redox homeostasis. Hence we reasoned that Glr1 and Fdh3 are probably crucial for C. albicans survival during host-pathogen interaction.
To test this, we generated C. albicans glr1Δ and fdh3Δ null mutants and compared their phenotypes to those of wild type and reintegrant control strains. Three key observations have confirmed our predictions. Firstly, glr1Δ cells are sensitive to oxidative stress, but not to nitrosative or formaldehyde stress. Fdh3Δ cells on the other hand display sensitivity to nitrosative or formaldehyde stress, but not to oxidative stress (Fig 3). Secondly, deletion of Glr1 completely blocks GSSG detoxification, and Fdh3 deletion slows GSNO detoxification (Fig 6). Thirdly, the glutathione redox potential is significantly perturbed in glr1Δ cells (Fig 6C).
The functions of Glr1 and Fdh3 are intimately linked, for example through their recycling of glutathione adducts, and in particular by the generation of GSSG by Fdh3-mediated recycling of GSNO (Fig 2). This linkage is reflected in several observations. For example, compensatory changes in GLR1 and FDH3 gene expression occur in fdh3Δ and glr1Δ cells, respectively (Figs 3 and 4). The increased FDH3 expression in glr1Δ cells appears to result in elevated resistance to nitrosative stress (Fig 3C). Also, deletion of Glr1 leads to the induction of oxidative stress genes (TRX1) and redox-dependent protein folding functions (PDI1) in response to formaldehyde stress (Fig 5), presumably due to the excessive accumulation of GSSG under these conditions. Furthermore, deletion of Fdh3 slows GSSG detoxification (Fig 6A), presumably because GLR1 expression is reduced approximately 4-fold in fdh3Δ cells (Fig 4).
Fdh3 does not appear to be the only enzyme in C. albicans that has denitrosylation or transnitrosylation activity because GSNO detoxification is only partially inhibited in fdh3Δ cells ( Fig  6B). Nitric oxide (NO) release and NO signalling are under tight regulatory control in eukaryotic cells [60]. GSNO can act as a cellular NO "reservoir" as it can release NO or act as a substrate for the transfer of NO to target cysteine residues via trans-nitrosylation [61]. Moreover, several types of eukaryotic enzymes in addition to GSNORs [62][63][64] have trans-nitrosylase or denitrosylase activity, including thioredoxin [65], thioredoxin reductase [66][67][68], superoxide dismutase 1 [69,70], haemoglobin [71] and protein-disulphide isomerase [72,73]. However, the potential roles of these systems in GSNO detoxification have not yet been defined in C. albicans. In contrast, the identity of the key NO detoxification enzyme in C. albicans has been determined. This role has been assigned to the haem oxygenase, Yhb1 [43,74,75].
Given their key roles in the maintenance of glutathione redox homeostasis, it was to be expected that Glr1 and Fdh3 influence fungus-phagocyte interactions and contribute to the virulence of C. albicans. The deletion of Glr1 or Fdh3 reduced the ability of C. albicans cells to kill macrophages, and overexpression of GLR1 or FDH3 increased the potency of C. albicans (Fig  7). Also, glr1Δ and fdh3Δ cells displayed reduced virulence in the invertebrate G. mellonella infection model, whilst GLR1 and FDH3 overexpressing cells were more virulent (Fig 8). Furthermore, in a mammalian model of systemic candidiasis, glr1Δ cells displayed attenuated virulence, and GLR1 overexpressing strain showed increased virulence (Fig 9). However, the behaviour of FDH3 mutants differed between the invertebrate and mammalian models, fdh3Δ cells displaying no significant reduction in virulence and FDH3 overexpressing cells having reduced virulence.
Why might fdh3Δ and FDH3 over-expressing cells display differences in their virulence phenotypes between the G. mellonella and murine infection models? A number of studies have shown that there is a good correlation between the outcome of insect and mouse infection models [76][77][78]. However, this might not always be the case. Possible differences in the NO concentrations experienced by C. albicans cells in these infection models might account for the differential impact of FDH3 mutations. However, Bergin et al. (2005) have reported that the G. mellonella haemocytes act in a similar fashion to human neutrophils in that they are able to phagocytose bacteria and fungi and generate superoxide [79][80][81]. Alternatively, the differential virulence phenotypes for FDH3 mutants might relate to the impact of FDH3 upon GLR1 expression (Figs 4 and 5). This might be significant in the context of the strong oxidising potential of mouse immune cells, and the complexity of mammalian niches [46]. Why might fdh3Δ cells, but not glr1Δ cells display contrasting phenotypes in the invertebrate and mammalian infection models? The basis for this probably lies in the observation that glr1Δ cells have a lower glutathione redox buffering capacity than fdh3Δ cells (Fig 6C). Therefore, glr1Δ cells are more vulnerable to ROS or RNS than fdh3Δ cells. Nevertheless, despite the complexities of FDH3 virulence phenotypes, it is clear that FHD3 and GLR1 functionality strongly influences C. albicans pathogenicity.
To conclude, this study has demonstrated that Glr1 and Fdh3 play important roles in oxidative, nitrosative and formaldehyde stress response in C. albicans, and that this contributes to C. albicans pathogenicity. Our findings emphasise the importance of the glutathione system and the key role of enzymes involved in the maintenance of intracellular redox homeostasis during host infection.

Stress conditions
To examine stress responses, C. albicans cells were grown overnight at 30°C and 200 rpm in YPDT. Cells were then re-inoculated into YPDT to an OD 600 of 0.2 and grown to an OD 600 of 0.8 at 30°C and 200 rpm. Cultures were then diluted four-fold in fresh YPDT, mixed with the appropriate stressors, and incubated at 30°C and 200 rpm. Stress was imposed using 1 M NaCl, 5 mM formaldehyde, 5 mM H 2 O 2 , 2.5 mM DPTA NONOate, or 2.5 mM freshly generated CysNO [82].
To examine oxidative and formaldehyde stress resistance, serial two-fold dilutions of C. albicans cells were plated onto YPD plates supplemented with 7.5 mM H 2 O 2 or 5 mM formaldehyde. Plates were incubated for 48 h at 30°C. Results shown are representative of data from at least three independent experiments. Formaldehyde sensitivity was also tested by growth at 30°C in YPD containing a range of formaldehyde concentrations using 96 well microtiter plates. Plates were incubated with shaking at 30°C, and growth assessed relative to untreated controls by measuring the OD 620 after 48 h. Experiments were performed in triplicate and one representative experiment was shown.
Nitrosative stress resistance was assessed by measuring adaptation times. Strains were grown overnight in 10 mL YPDT at 30°C, and diluted to an OD 600 of 0.1 in fresh YPDT. Aliquots (100 μL) were added to 96 well microtiter plates (Costar) containing 100 μL YPDT containing 5 mM DPTA NONOate or CysNO (final concentration 2.5 mM). Plates were incubated with shaking at 30°C and the OD 620 measured every 20 min for 48-72 h in a FluroStar Optima plate reader (BMG Labtech). Stress adaptation times were determined by measuring the inflection times for each growth curve, as described previously [84]. Data represent the means and standard deviations from at least three independent experiments.
To prepare CysNO, 4 mL of 0.34 mM L-cysteine, 0.75 M HCl was added to 5 mL of 0.5 mM NaNO 2 . After the solution turned deep red, 2.5 mL 1 M NaOH was added and the concentration of CysNO measured at 335 nm (ε 335nm 503 M -1 cm -1 ) [85].
The doxycycline-conditional over-expression strains tetON-GLR1 (ATT6) and tetON-FDH3 (ATT7) were generated as described previously [48,89]. The CIp10-based tetON expression plasmids were linearized with StuI, and transformed into C. albicans CEC2175 ( Table 1). The genotypes of these strains were confirmed by diagnostic PCR using the primers described in S1 Table. Transcript analyses C. albicans cells were harvested, flash-frozen in liquid N 2, and RNA extracted as described previously [90,91]. RNA integrity was confirmed using Bioanalyzer RNA 6000 Nano Assay Protocol according to the manufacturer's instructions (Agilent; Stockport, UK) [92]. cDNA was prepared using Superscript II as per the manufacturer's protocols (Invitrogen Ltd.; Paisley, UK). Transcript levels were measured by qRT-PCR relative to the internal ACT1 mRNA control using the primers listed in S1 Table with a LightCycler480 system (Roche Applied Science) using the Roche Universal Probe library [15]. Data represent the means and standard deviations from at least three independent experiments. GSH, GSSG and GSNO assays C. albicans cells were harvested [82] and protein extracts prepared in 0.5 M citrate, pH 5.0) as GSNO was found to be unstable at pH 8.0. For GSNO assays, 5 μL of 1 mM NADH and 10 μL of 250 μM GSNO were added to 10 μL of protein extract, and samples were derivatized at 0, 2, 5, 10, 20 and 30 min. For GSSG assays, 5 μL of 1 mM NADPH and 10 μL of 250 μM GSSG were added to 10 μL of protein extract, and samples were derivatized at 0, 2, 5, 10, 20 and 30 min. The derivatization and analysis were performed as previously described [93] using a Thermo Surveyor LC system coupled to a TSQ Quantum, triple quadrupole mass spectrometer (Thermo Scientific, UK). The following SRM transitions were used for quantification; Glu-Glu (internal standard) m/z 277-241, GSNO m/z 337-307, GSNEM m/z 433-304 and GSSG m/z 613-355. Peak integration and quantification was performed using Xcalibur software (Version 2.0.7.SP2). GSH, GSSG and GSNO concentrations were then calculated relative to authentic standards. Data were normalised against total protein. Data represent the means and standard deviations from at least three independent experiments.
The redox potential was calculated using the GSH and GSSG concentrations calculated using the following equation based on the Nernst equation [20]. Macrophage killing assay RAW264.7 murine macrophages were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. C. albicans cells (1x10 6 ) were co-incubated with RAW264.7 macrophages (2x10 5 ) for 3 h at 37°C. For doxycycline-induced over-expression, C. albicans cells were prepared by growing overnight in YPDT containing 100 μg/mL doxycycline. Wells were washed with PBS and stained with trypan blue. Macrophages were fixed in 3% formaldehyde for 3 min and images were recorded with an Zeiss Axio Observer Z1 inverted microscope and Zeiss Hrm camera. Images were analysed using Axiovision 4.8.2 software, and the percentage of killed macrophages calculated by analysing at least 200 macrophages per well [94]. All data were generated in triplicate for at least three independent experiments and one representative experiment was represented.

Galleria mellonella virulence assay
The virulence of C. albicans strains was assessed using groups of 20 G. mellonella larvae (Livefoods, UK) in the sixth instar as described previously [95]. C. albicans inocula were grown in YPDT at 30°C overnight with agitation. For doxycycline-induced over-expression, 100 μg/mL doxycycline was added to the medium. C. albicans cells were washed three times in PBS, re-suspended in PBS at 2.5 x 10 5 cells/mL and 10 μL of suspension injected via the last proleg [95]. Untouched larvae and larvae injected with PBS +/-doxycycline served as controls. After injection larvae were incubated at 37°C and the number of dead larvae scored every 12 h. Statistical analyses were performed using the Kaplan-Meier log rank test. Differences with a P value < 0.05 were considered significant. Three independent experiments were performed, and data from one representative experiment are presented.

Murine virulence assay
The three-day murine intravenous challenge model of C. albicans infection [50] was used to determine the impact of GLR1 and FDH3 upon virulence. Female BALB/c mice (6-8 weeks old; Harlan, UK) were randomly assigned into groups of six mice, were housed in individually ventilated cages (IVCs) and were provided with food and drinking fluid ad libitum. Each group of mice was inoculated via the lateral tail vein with C. albicans cells (3 x 10 4 CFU g -1 mouse body weight) grown at 30°C in NGY +/-50 μg/mL doxycycline [96]. The fungal inoculum used for each mouse group was also randomised. For doxycycline treated mice, mice were either provided with 5% sucrose or 5% sucrose containing 2 mg/mL doxycycline as their drinking fluid. Mice were monitored and weighed daily. After 72 h, mice were weighed, terminated by cervical dislocation, and renal fungal burdens determined. The outcome score [50] was calculated based on the fungal kidney burdens and percentage weight change at 72 h. Statistical differences between body weight changes, kidney burdens and outcome scores were determined by Kruskall-Wallis and Mann-Whitney U tests using IBM SPSS (version 20).

Ethics statement
All animal experimentation was performed under UK Home Office Project license 60/4135 and was approved by the UK Home Office and by the Animal Welfare and Ethical Review Body of the University of Aberdeen. All work conformed to European Directive 2010/63/EU. Animals were carefully monitored for signs of distress during the infection studies. Distress was minimised by expert handling. Animals were weighed once daily, and monitored for changes in condition at least twice daily. Animals were euthanized humanely by cervical dislocation when they showed signs of severe illness, i.e. they had a ruffled coat, displayed a hunched posture, were unwilling to move around the cage and had lost 20% of their initial body weight. There were no unexpected incidents of mortality during this study. Analgesia and anaesthetics were not required.

Statistical analyses
Unless otherwise stated, data are expressed as means plus standard deviations from at least three independent experiments: Ã p< 0.05; ÃÃ p<0.001; and ÃÃÃ p <0.0001. Statistical significance was determined by one-way ANOVA with post hoc analysis using Dunnett's t-tests with a 95% confidence level. Analyses were carried out using Prism 5.0 (Graphpad). Data are represented in means ± SD.
Supporting Information S1 Fig. Candida albicans GLR1 encodes a NADPH-dependent glutathione reductase. The sequence alignment of homologs of the NADPH-dependent glutathione reductase GLR1 (C5_01520C) was generated using ClustalW. The homologs from Saccharomyces cerevisiae, Schizosaccharomyces pombe, Mus musculus, Homo sapiens and Caenorhabditis elegans used for the multiple sequence alignments were obtained from NCBI/ BLAST are shown. The conserved NADH binding domain within a larger FAD bindingdomain and the C-terminal dimerisation domain are illustrated in red and green, respectively. (TIF) S2 Fig. Candida albicans FDH3 encodes a GSH-dependent formaldehyde dehydrogenase class III. The sequence alignment of homologs of the Candida albicans GSH-dependent formaldehyde dehydrogenase FDH3 (CR_10250C_A) was generated using ClustalW. The homologs from Saccharomyces cerevisiae, Schizosaccharomyces pombe, Mus musculus, Homo sapiens, Drosophila melanogaster and Caenorhabditis elegans used for the multiple sequence alignments were obtained from NCBI/ BLAST are shown. The conserved catalytic domain of the alcohol dehydrogenases class III and the C-terminal cofactor-binding domain that reversibly binds NAD(H) are illustrated in red and green, respectively. (TIF) S1