Unique, Diverged, and Conserved Mitochondrial Functions Influencing Candida albicans Respiration

Candida albicans is an opportunistic fungal pathogen of major clinical concern. The virulence of this pathogen is intimately intertwined with its metabolic behavior, and mitochondria have a central role in that metabolism. Mitochondria have undergone many evolutionary changes, which include lineage-specific adaptations in association with their eukaryotic host. Seven lineage-specific genes required for electron transport chain function were identified in the CTG clade of fungi, of which C. albicans is a member. Additionally, examination of several highly diverged orthologs encoding mitochondrial proteins demonstrated functional reassignment for one of these. Deficits imparted by deletion of these genes revealed the critical role of respiration in virulence attributes of the fungus and highlight important evolutionary adaptations in C. albicans metabolism.

At the center of cellular metabolism are the mitochondria. They house and integrate multiple anabolic and catabolic functions involved in energy production, carbon, nitrogen, lipid and iron metabolism, and biosynthesis of amino acids, nucleic acids, and other cellular constituents. The integrity and function of mitochondria are essential to the virulence of C. albicans. Mutations affecting any one of a number of mitochondrial functions, including mitochondrial ribosome synthesis, mitochondrial transcription or genome maintenance, protein import, or functioning of the electron transport chain (ETC), result in avirulence (6)(7)(8)(9)(10)(11)(12). Avirulence is likely multifactorial, as mitochondrial mutations negatively impact growth rates and carbon source utilization (7,8,13), cell wall structure (7,9,14,15), membrane composition (15), morphogenesis (7,8,13,16,17), and biofilm formation (18), as well as interactions with neutrophils (19) and macrophages (20). These observations emphasize the decisive role of mitochondria in the pathobiology of C. albicans (reviewed in reference 21). The significance of mitochondria extends well beyond a basic understanding of metabolism and virulence, as mitochondrial mutations have a meaningful impact on antifungal susceptibility (7,11,22).
Information about the structure and function of C. albicans mitochondria is limited, and, although mapping of basic information from other organisms is useful, mitochondria have experienced many adaptive evolutionary changes to accommodate the various habitats and the biology of their eukaryotic host (23)(24)(25)(26). Even within the narrow scope of fungal phyla, notable differences occur (27)(28)(29). Comparison of C. albicans and Saccharomyces cerevisiae exemplifies the species-specific adaptations of mitochondria. Unlike S. cerevisiae, a Crabtree-positive yeast, C. albicans has been considered a Crabtree-negative species that maintains respiratory metabolism even in the presence of glucose (30,31), although a recent report indicates that glucose can repress C. albicans respiration (32). This physiological difference is reflected in the dramatically different transcription networks controlling mitochondrial gene expression (33). The posttranscriptional control of mitochondrial protein synthesis has also undergone revision (15,18), and C. albicans contains a complete ETC, while S. cerevisiae is devoid of complex I (NADH: ubiquinone oxidoreductase) (26,28,29).
A unique facet of complex I synthesis in C. albicans is the requirement for GOA1. GOA1 was initially identified in a screen for mutants hypersensitive to oxidative stress (8). GOA1 deletion mutants fail to make complex I, resulting in reduced respiration and multiple attendant deficits, including a reduced ability to form hyphae, cell wall alterations, and hypersensitivity to antifungals (8,14,19,22,34). The mutants also showed an enhanced sensitivity to killing by neutrophils, reduced immune recognition, and reduced virulence in a murine model of disseminated disease (8,14,19,22,34).
Interestingly, orthologs of GOA1 are restricted to members of the "CTG clade" of fungi, whose members decode CTG codons as serine rather than leucine (35), suggesting that it represents a lineage-specific mitochondrial adaptation. Important issues are whether GOA1 is a unique evolutionary adaptation or is representative of broader lineage-specific changes in mitochondrial function and what distinctive attributes might these genes bring to the pathobiology of C. albicans?
As a first step in addressing these issues, an in silico search was conducted for CTG clade-specific genes encoding putative mitochondrial proteins. Twenty-five potential genes of interest were identified, seven of which localized to mitochondria and were required for expression of various ETC complexes. Parallel analysis of three highly diverged mitochondrial genes with apparent orthologs in S. cerevisiae showed conservation of function for only two, PET111 and AEP1, while MNE1 demonstrated evolutionary reassignment. Regardless of the ETC complex affected, all of the respiratory mutants exhibited deficits in fitness, virulence, and antifungal susceptibility. The results revealed multiple lineage-specific adaptations unique to the CTG clade of fungi and suggest that additional adaptations may be found in the form of functional reassignment of mitochondrial proteins.
The data further emphasize the importance of mitochondrial function in the pathobiology of C. albicans.

RESULTS
In silico identification of putative clade-specific mitochondrial proteins. Orthologs of GOA1, which is required for complex I function in C. albicans, are confined to the CTG clade of fungi (8). To assess the scope of clade-specific mitochondrial functions, an in silico screen was conducted. Clade-specific orthologs across eight species were identified and assessed for potential mitochondrial localization using three different algorithms, MitoPred (36), TargetP (37), and Predotar (38). Assessing multiple orthologs of slightly differing sequence was intended to eliminate fortuitous false positives, as all orthologs should embody the characteristics of a mitochondrial protein regardless of sequence variation. The use of multiple prediction algorithms provided a more rigorous assessment to address the low sensitivity of the prediction methods (39). The screen identified 25 genes of interest (Table 1). Importantly, the list included GOA1, validating the screening process. A total of 23 of the 25 genes (with the exceptions being GOA1 and FGR39) were uncharacterized open reading frames (ORFs) encoding proteins of unknown function ( Table 1). As evidenced here, several of these genes influenced expression of electron transport chain complex I, III, or IV and were the focus of this work. These included NUO3 (orf19.1179, NADH:ubiquinone oxidoreductase), NUO4 (orf19.5077), NUE1 (orf19.2819, NADH:ubiquinone oxidoreductase expression), and NUE2 (orf19.4467), which affect complex I; QCE1 (orf19.6918, coenzyme Q:cytochrome c oxidoreductase expression), affecting complex III; and two genes affecting complex IV expression, COE1 (orf19.1371, cytochrome c oxidase expression) and COE2 (orf19.6566).
Identification of genes affecting respiration. As an initial screen for respiratory, defects homozygous deletion mutants were constructed for each gene of interest and tested for growth on nonfermentative carbon sources. Also included in the analysis were orf19.4467 (NUE2), which had only a single nonclade ortholog in Yarrowia lipolytica, and several genes distantly related to mitochondrial genes of S. cerevisiae, including MNE1, PET111, and AEP1 (Table 1). These diverged proteins had Յ15% identity with the C. albicans ortholog, but the orthology data were supported by limited local synteny (40,41). Their inclusion, with the objective of assessing whether these highly diverged orthologs retain analogous function, was motivated by the data showing gross differences between C. albicans and S. cerevisiae in mitochondrial composition and regulation (26,29,30,33).
Homozygous deletion mutants were obtained for 24 of the 25 clade-specific genes. As expected, deletion of GOA1 resulted in the inability to grow on glycerol. Among the other 23 deletion mutants, 7 were either unable to grow or showed severely limited growth on glycerol (Table 1; see also Fig. 1). Other mutants were deficient with respect to lacking NUE2 or any one of the three S. cerevisiae orthologs (Table 1; see also Fig. 1). To rule out glycerol-specific metabolic defects, the mutants were tested for utilization of several nonfermentable carbon sources, including pyruvate, ethanol, and acetate. Consistent with an alteration in respiratory metabolism, the mutants were also unable to grow on these carbon sources. Complementation of the mutants with a wild-type copy of the cognate gene restored the ability to grow on nonfermentable carbon sources, demonstrating the link between genotype and phenotype ( Fig. 1).
All of the putative respiratory mutants grew with glucose as the carbon source, but mutants lacking orf19.5607 showed extremely slow growth and a very low level of plating efficiency (Fig. 1). Because of their poor growth properties, these strains were not considered further and subsequent study focused on the other 10 mutants unable to grow on nonfermentative carbon sources.
Respiration is reduced in the mutants. The growth phenotype of the mutants suggested a deficiency in one or more respiratory pathways. C. albicans is capable of both cyanide-sensitive and cyanide-insensitive respiration, which represent the classic ETC and a short circuit of the ETC from ubiquinone to oxygen via alternative oxidases (42)(43)(44). The contribution of each was assessed using potassium cyanide (KCN) to inhibit cytochrome oxidase, complex IV, of the traditional electron transport chain and using salicylhydroxamic acid (SHAM) to inhibit alternative oxidase activity. As shown in Table 2, addition of KCN inhibited Ͼ95% of respiration in the control strain, in agreement with prior studies (42,43). Subsequent addition of SHAM resulted in only a slight reduction in respiration.
The deletion mutants formed three classes based on O 2 consumption rates and response to inhibitors. Class I mutants were similar to goa1⌬/⌬ mutants. These retained only 35% of the wild-type respiratory capacity. Addition of KCN inhibited 60% to 65% of this activity, and addition of SHAM inhibited the majority of the remaining activity. This phenotype was conferred by deletion of NUO3, NUO4, NUE1, NUE2, or MNE1 ( Table 2). Approximately 70% of respiration in C. albicans is sensitive to rotenone, a specific inhibitor of complex I, and the remaining 30% is contributed by an alternative, rotenone-insensitive NADH-Q oxidoreductase (45). Rotenone treatment of the parental strain inhibited respiration by about 60% but had no measurable effect on class I mutants. Together, these results indicate that class I mutants are deficient in complex I and that residual respiration is due to the presence of the alternative, rotenone-insensitive NADH-Q oxidoreductase in conjunction with the SHAM-sensitive alternative oxidase.
The remaining five mutants formed classes II and III. These showed a more severe reduction in respiration, with O 2 consumptions rates approximately 7% of control rates, and differed in their response to inhibitors. O 2 utilization was inhibited only 15% by KCN but was inhibited 65% by SHAM. This indicates that approximately 99% of ETC-mediated respiration was eliminated in these mutants and that the bulk of oxygen consumption proceeded via the alternative oxidase. Deletion of QCE1, COE1, COE2, or PET111 conferred this phenotype.
Class III consisted of a single mutant deleted of orf19.102, orthologous to S. cerevisiae AEP1 (ScAEP1). The rate of respiration was slightly less than that seen with the class II mutants and was inhibited about 40% by KCN, which translates into a 98% reduction in ETC activity. The remainder was SHAM sensitive. For all of the mutants, genetic rescue restored the rate and pattern of respiration to those seen with the wild-type control.
Mutants lack various ETC complexes. The deficits implied by respiratory measurements were directly supported by examination of ETC complexes (46,47). All class I mutants exhibited partial or complete loss of complex I, consistent with loss of rotenone-sensitive respiration ( Fig. 2A). Staining of the gels for NADH dehydrogenase activity showed greatly reduced amounts of intact complex I in nue1⌬/⌬ and nue2⌬/⌬ mutants, the presence of partial complexes in goa1⌬/⌬ and nuo4⌬/⌬ mutants, and a complete loss of activity in nuo3⌬/⌬ and mne1 mutants (Fig. 2B). There was no discernible loss of the other ETC complexes ( Fig. 2A). Thus, NUO3, NUO4, NUE1, NUE2, and MNE1 are required for proper expression of NADH:ubiquinone oxidoreductase (complex I).
Class II mutants exhibited one of two patterns. Deletion of QCE1 resulted in loss of complex III with no observable effect on other ETC components (Fig. 2C). In contrast, mutants lacking COE1, COE2, or PET111 were specifically devoid of complex IV (Fig. 2C) and no cytochrome oxidase activity could be detected in these mutants. These results are consistent with the essentially complete loss of KCN-sensitive respiration in these mutants. As with its respiratory profile, the class III AEP1 null mutant was unique in its ETC component profile. The levels of complexes I and III were reduced, and complexes IV and V were absent (Fig. 2C).
The genes encode mitochondrial proteins. The genes of interest were identified based on in silico predictions of mitochondrial localization, and these predictions were   (87). The position of complex III was inferred from its relative size, since no activity stain is available. The relevant genotype is indicated above each lane. Comparable results were obtained in two or more mitochondrial preparations.

Sun et al.
® supported by confocal microscopy and cell fractionation studies. Green fluorescent protein (GFP) fusions were constructed and integrated into the wild-type allele of the corresponding heterozygous deletion mutant. All of the fusion genes allowed growth on glycerol, showing that the fusion proteins were functional. Mitochondria were labeled with the mitochondrion-specific dye MitoTracker Red CMXRos (48) and examined for colocalization of the dye and GFP signal. Cytochrome oxidase subunit Cox8p was fused with GFP and examined as a positive control (49,50). As shown in Fig. 3A, there was a clear visual correspondence between MitoTracker staining and the Cox8-GFP signal. This was substantiated by quantitative analysis of the images (51). A statistically significant Pearson's correlation coefficient and essentially identical Manders coefficients for all of the fluorescence channels (Fig. 3A) fully support the visual correlation. GFP fused to Nuo3, Nuo4, or Aep1 similarly showed a strong correspondence between dye and GFP signals (Fig. 3B, C, and D). For the remaining eight proteins, the GFP signal was not significantly above the background fluorescence of mitochondria. This was true of the Goa1-GFP fusion constructed in these studies, as well as the Goa1p-GFP strain previously examined (8). Fixed and unfixed cells yielded similar results. As an alternative to direct visualization, the distribution of GFP-tagged proteins in various cell fractions was examined. As seen in Fig. 4, each strain contained a fluorescent protein that was strongly enriched in the mitochondrial fraction. The proteins roughly approximated the sizes calculated in the gene-based predictions but, as expected, were somewhat smaller since the samples were not fully denatured in order to preserve GFP fluorescence (52) and since mitochondrial proteins undergo signal cleavage postimport (53). Quantitation of the fluorescence intensities indicated an approximately 250-fold range in expression levels. Those proteins that were not detectable by confocal microscopy were expressed at Ͻ3% of the level of Cox8-GFP, but their enrichment in the mitochondrial fraction suggests that they too were localized to this organelle (Fig. 4).
To assess whether these proteins have a structural role in the affected complexes, purified mitochondria of the GFP-tagged strains were fractionated on blue native gels. For both NUO3 and NUO4 GFP-tagged strains, complex I acquired fluorescence (Fig. 5). Furthermore, consistent with incorporation of GFP into the complex, complex I of these strains was slightly increased in size (Fig. 5). On the basis of their association with complex I, and consistent with the nomenclature of other complex I subunits, the corresponding genes were named NUO3 and NUO4 for (NADH:ubiquinone oxidoreductase). None of the other strains showed fluorescence associated with an ETC component. As the other genes did not appear to encode components of the electron transport chain complexes and were expressed at much lower levels, they apparently Growth rates and yields are reduced in respiratory mutants. The relative contributions of fermentation and respiration to C. albicans metabolism are presently unclear (54,55). Having an assemblage of mutants with various degrees of respiratory capacity provided the opportunity to assess fermentative contributions to growth. In synthetic complete (SC) medium, the rate of growth of complex I mutants was comparable to that of a wild-type strain despite their 65% reduction in respiration (Table 3). However, a 20% to 25% reduction in growth rate was seen when respiration via the ETC was eliminated, as in the mutants lacking complex III or IV. In synthetic defined (SD) medium, which imposes the additional metabolic burden of amino acid biosynthesis, growth rates were reduced 50% in complex I mutants and 75% in the other strains. Growth yields were also compromised. In SC medium, despite maintain-  ing maximal growth rates, complex I mutants exhibited a 60% to 64% reduction in growth yield (Table 3). Mutants with a complete loss of KCN-sensitive respiration showed a 71% to 77% reduction (Table 3). Thus, fermentation alone cannot support maximal growth rates or yields. Influence of respiratory mutations on filamentation varied with the inducing conditions. The relationship of respiration to morphogenesis has been examined for over 40 years, with reduced respiration reported as promoting the yeast-to-hypha transition in some studies (56)(57)(58) and blocking the process in others (8,10,16,59,60). Hence, the panel of mutants was examined for their ability to filament under various conditions.
On medium 199 agar, wild-type cells formed a halo of abundant agar-invading hyphae around the colony center (Fig. 6A). Mutants defective in complex I were still capable of forming hyphae, but these were less abundant and shorter in length than those seen with the wild-type strain (Fig. 6A). Mutants lacking complex III or IV were similarly (or perhaps slightly more) compromised. The aep1Δ mutant, in contrast, was nearly devoid of hyphae. Kinetic analysis of the process showed that filamentation by the mutants was delayed in onset and that hyphal extension occurred at reduced rates (see Fig. S1 in the supplemental material). Under conditions of testing performed in the same media, but in broth suspension, a slightly different pattern was observed. Whereas the wild-type strain formed germ tubes and hyphae, the mutants lacking complex I were restricted to pseudohyphae and yeast (Fig. 6B). But, surprisingly, germ tubes and true hyphae were formed by complex III and IV mutants, although at lower levels of abundance than were seen with the wild type. Again, the aep1⌬ mutant was unable to form hyphae. Extending the incubation time did not result in germ tube formation by complex I mutants.
Reduced oxygen tension has been reported as having either a negative or positive effect on filamentation (57)(58)(59)61). The effect of embedding the cells in YPD-plusserum agar was tested to examine the influence of reduced oxygen tension. Although the mutants failed to filament on the surface of this medium, all formed hyphae when embedded. The extent of filamentation of complex I mutants was difficult to distinguish from the level shown by the wild-type strain, while complex III and IV mutants were compromised only slightly (Fig. 6D). Complementation of the mutations restored a wild-type phenotype under all conditions. Together, these results indicate that respiration is not essential to filamentation and that its influence is dependent on the inducing environment.
Mutants are hypersensitive to fluconazole. Pharmacological inhibition of complexes I, III, IV, and V causes hypersensitivity to fluconazole (22). Fluconazole sensitivity of the respiratory mutants was examined to provide genetic validation of these results. As seen in Fig. 7, all of the respiratory mutants showed hypersensitivity to fluconazole, and the hypersensitivity was lost upon genetic rescue. This further validates the relationship between respiration and fluconazole sensitivity.
Virulence is attenuated in the G. mellonella infection model. Various C. albicans mitochondrial mutants were shown to be attenuated in a mouse model of disseminated disease (6)(7)(8)(9)(10)(11). However, as shown by the filamentation phenotypes, the consequence of respiratory deficiency varies with the growth environment. Therefore, the virulence of the respiratory mutants was assessed in an alternative infection model   (62). Mutants lacking GOA1 are attenuated in mice (8) and were strongly attenuated in G. mellonella as well (Fig. 8). Likewise, virulence of all of the other respiratory mutants was significantly attenuated in this model (Fig. 8). The results indicate that respiration is critical to virulence in the G. mellonella model, as in the mouse model.

DISCUSSION
Mitochondria have a central and critical role in the metabolism and pathobiology of C. albicans (21). Mitochondria perform multiple, diverse processes within the eukaryotic host cell and have evolved in composition and function among various phylogenetic lineages (25,63). The results of this study identify a number of lineage-specific changes in composition and expression of the ETC in the CTG clade of fungi. In addition to GOA1, 24 putative CTG clade-specific genes were identified, seven of which have essential roles in respiration. The current analysis does not distinguish whether these are newly invented genes or orthologs of other fungal proteins that have diverged beyond recognition. Overall, these proteins represent a relatively small fraction of the mitochondrial proteome, which consists of about 1,000 proteins in yeast and humans (64,65), but their importance is demonstrated by their critical role in mitochondrial function. The other 17 genes may function in other mitochondrial processes or may represent false positives, as the sensitivity of the prediction algorithms is around only 40% (39).
The biochemical role of these seven proteins in ETC expression remains to be determined. Two proteins, Nuo3 and Nuo4, were expressed at levels similar to that of Cox8, a subunit of complex IV. This observation, in conjunction with their physical association with complex I during electrophoresis, suggests that they are subunits of the complex. Eukaryotic complex I consists of a core of 14 proteins derived from the bacterial ancestor of mitochondria supplemented with various numbers of supernumerary subunits, e.g., 40 in Y. lipolytica (66) and 46 in mammals (26). Evolutionary addition and deletion of subunits have occurred broadly across phylogenetic lineages, including fungi (26,28,29). Nuo3 and Nuo4 appear to represent another example of evolutionary adaptation of complex I that is specific to the CTG clade. The function of the majority of supernumerary subunits of complex I has yet to be defined (63).
The other respiratory proteins were much less abundant than Nuo3 and Nuo4 and were not observed to associate with any of the ETC complexes. These proteins may participate in expression and assembly of the respiratory complexes, an intricate process requiring proteins that facilitate transcription, mRNA processing, and translation and posttranslational processing (67, 68) of component proteins. Alternatively, these proteins may be minor, nonstoichiometric subunits of the ETC complexes whose association was not apparent due to their low abundance. Goa1 was among these less CTG Clade-Specific Mitochondrial Functions ® abundant proteins, and the signal intensity of the GFP-tagged protein was not sufficiently above mitochondrial background fluorescence to localize it to the mitochondria. A previous report of differential localization of Goa1 appears to indicate a preparation artifact (8).
Beyond lineage-specific adaptations, the sharply different levels of regulation of respiration by C. albicans and S. cerevisiae (30,33) might suggest that considerable functional divergence of orthologous proteins has occurred. Examination of three highly diverged orthologs, MNE1, PET111, and AEP1, suggests that this is partly true. A striking outcome was the absence of complex I in the mne1⌬/⌬ mutant. In S. cerevisiae, Mne1 facilitates excision of the aI5␤ intron, one of seven introns in the COX1 mRNA, and mutants lacking MNE1 are devoid of complex IV (69). In contrast, C. albicans mne1⌬/⌬ mutants had normal levels of complex IV but were devoid of complex I. Interestingly, the seven mitochondrially encoded subunits of complex I contain no introns, suggesting that C. albicans Mne1 (CaMne1) lacks even an intron splicing function. Given that complex I is an ancestral component of the ETC that was subsequently lost in S. cerevisiae and other Saccharomycetales (26,28,29), CaMNE1 likely retains the ancestral function and functional reassignment occurred in S. cerevisiae.
Unlike that of MNE1, the function of PET111 and AEP1 appears to be conserved. In S. cerevisiae, Pet111 is located on the matrix side of the mitochondrial inner membrane, where it binds the 5= untranslated region of COX2 mRNA and facilitates its translation (70). The gene is highly diverged in other yeast species but has retained its orthologous function (71). This appears also to be the case with C. albicans as the corresponding mutant lacked cytochrome oxidase, although the mechanism of that loss was not examined. With regard to AEP1, functional conservation appears likely. C. albicans aep1⌬/⌬ mutants failed to express mitochondrial ATP synthase. ScAEP1 is required specifically for translation of OLI1 mRNA, which encodes subunit 9 of ATP synthase (72). However, in addition to loss of ATP synthase, complex IV was also absent in C. albicans aep1⌬/⌬ mutants and the contents of complexes I and III were reduced. It is not known if these broader changes occur in S. cerevisiae and whether they are secondary to loss of ATP synthase or reflect a more general role of Aep1 in synthesis and assembly of the ETC.
Phenotypic characterization of the respiratory mutants showed that any disruption of respiration had significant biological effects. The yeast-hypha transition is a critical virulence attribute of C. albicans and is influenced by respiration, but various studies have reached different conclusions regarding its role. Reduced respiration in hyphae, the stimulatory effect of ETC inhibitors, and increased filamentation under hypoxic conditions all suggest that a reduction in respiration stimulates the transition to the hyphal form (56,58). Conversely, a positive role for respiration is suggested by studies showing that hypoxic conditions prevent filamentation (57,61) and that phenazines and methylene blue, which short-circuit electron flow through the ETC, are inhibitory (59,60). Genetic studies also support the idea of a positive role as mutants lacking NDH51, GOA1, or COX4 are deficient in filamentation (8,10,16,59). The results reported here showed that the morphological response of respiration-deficient mutants was largely independent of the particular block in respiration and had either no effect or a negative effect depending upon the inducing environment. This was most clearly illustrated by the almost complete lack of filamentation of the mutants on the surface of YPD-plus-serum medium versus formation of abundant hyphae when embedded in the medium. This may reflect differences between the environments with respect to energy metabolism or signal processing (59), but, regardless of the mechanism, the variability of responses in relationship to the environment might account for the disparity in the prior conclusions given the wide range of inducing conditions employed.
Fluconazole is a highly utilized antifungal, and efforts to enhance its efficacy and counter resistance are needed (73). Thus, it was of interest that all of the deletion mutants were hypersensitive to fluconazole irrespective of which respiratory complex was affected. The mechanistic underpinnings of this phenotype remain to be determined; however, the observations of Benhamou et al. (74) may be relevant. Those investigators demonstrated that azoles localize to the mitochondria (74) and that azole derivatives designed to direct localization to the endoplasmic reticulum (ER), where the target of azoles, Erg11, resides, show enhanced antifungal activity (75). If mitochondrial accumulation of azoles is dependent upon mitochondrial membrane potential, then respiratory mutants, because of a reduced mitochondrial membrane potential, may divert less fluconazole to mitochondria, allowing enhanced concentrations at the site of action, and this would be reflected in an enhanced sensitivity to azoles.
Reduced respiration limited growth under optimal conditions and might have a more substantial effect under stress conditions in the host niche. Available studies have shown that mitochondrial function in general (6,7) and electron transport chain components specifically (8,9,11) are essential for C. albicans virulence in mice. This was mirrored in the G. mellonella infection model, where loss of any one component of the electron transport chain attenuated virulence. This suggests that respiration per se is a key requirement for virulence. The results further support the idea of the importance of the mitochondrion in the biology of C. albicans and its relevance as a potential therapeutic target (21).

MATERIALS AND METHODS
Strains and culture conditions. C. albicans strains used are listed in Table S1 in the supplemental material. Strains were routinely cultured on YPD (76) or synthetic complete (SC) medium (United States Biological). Yeast extract-peptone-glucose (YPG) (76) was used to screen for respiration-deficient strains. Carbon source utilization was assessed on yeast nitrogen base (YNB) medium (76) containing 100 mM glucose, 100 mM pyruvate, 3% glycerol, 2% ethanol, or 2% sodium acetate. Media were solidified with 2% agar.
For growth rate determinations, the optical density at 595 nm (OD 595 ) was monitored at 30°C in a Tecan plate reader. Growth rates were calculated in R (77) using the easy linear method implemented in the "growthrates" package (78) with the h value set to encompass approximately two cell doublings. Four to five independent determinations were made for each stain and medium.
Identification of clade-specific mitochondrial proteins. A BLASTP (79) comparison of the predicted proteins from assembly 21 of the C. albicans genome (http://www.candidagenome.org/) with those of S. cerevisiae (http://downloads.yeastgenome.org/sequence/S288C_reference/orf_protein/) and Aspergillus nidulans (http://www.aspergillusgenome.org/) identified a subset of 1,349 nonconserved or poorly conserved C. albicans proteins with an E value of Ͼ10 Ϫ6 . InParanoid (80) and reciprocal BLAST comparison of this subset against the genomes of Candida tropicalis, Candida guilliermondii, Candida lusitaniae, and Lodderomyces elongisporus (https://www.broadinstitute.org/fungal-genome-initiative/comparative -candida-genomic-project) or of C. parapsilosis (http://www.sanger.ac.uk/resources/downloads/fungi/) and Debaryomyces hansenii (http://genome.jgi.doe.gov/) identified 419 orthologous protein families present in all eight species. Putative mitochondrial proteins among the orthologous groups were identified using MitoPred (36), TargetP (37), and Predotar (38). Among 86 of the orthologous groups, mitochondrial localization was indicated for five or more family members by at least one algorithm. These were further analyzed by PsiBLAST (81) comparison of the C. albicans ortholog with the nonredundant database at NCBI to detect more distantly related non-CTG clade orthologs that may have been missed in the initial, less sensitive BLAST screen. Proteins for which the second iteration identified potential orthologs in non-CTG clade species (E value Ͻ10 Ϫ4 ) were removed, leaving 25 putative clade-specific mitochondrial proteins.
Strain constructions. Deletion mutants were constructed in C. albicans strain SN152 essentially as described by Noble and Johnson (82). Details are provided in Text S1 in the supplemental material.
Primers. The primers used in this study are listed in Table S2, Table S3, Table S4, and Table S5. Respiration measurements. Preliminary experiments showed maximal respiration in cells in early log phase, as previously reported (83). Strains that had been cultured to the stationary phase in YPD broth were used to inoculate fresh YPD medium (2.5 ϫ 10 6 cells/ml), and cultures were incubated 4 h at 30°C with vigorous aeration. Aliquots of culture were mixed with prewarmed YPD, and O 2 utilization was measured using a Hansatech Oxygraph system (Hansatech Instruments). Utilization rates were normalized to the OD 595 of the sample. Contributions of ETC and alternative oxidase were assessed by addition of KCN to inhibit cytochrome oxidase or of salicylhydroxamic acid (SHAM) to inhibit alternative oxidase (42). KCN (100 mM)-Tris (pH 8) was added to reach a final concentration of 1 mM. SHAM (500 mM)dimethyl sulfoxide (DMSO) was added to reach a final concentration of 5 mM. DMSO alone had no effect. Rotenone (25 M) was added to assess the contribution of rotenone-insensitive alternative NADH-Q oxidoreductase (45). At least three independent determinations were made for each strain.
Isolation of mitochondria. Mitochondria were isolated by the use of a modification of previously described procedures (84). Details of the protocol are described in Text S1.
Blue native gel electrophoresis. Purified mitochondria were solubilized with 1% n-dodecylmaltoside (85) and electrophoresed on 5% to 15% polyacrylamide gradient gels as described previously (46) except that 0.02% n-dodecylmaltoside was added to the cathode buffer and 0.02% Coomassie G250 was continuously present in the cathode buffer. Gels were destained in 10% acetic acid-100 mM ammonium acetate-50% methanol (86) and stained with Imperial protein stain (Pierce). In-gel enzyme activity assays were performed as described previously (87).
Cell fractionation and SDS-PAGE of GFP-tagged strains. Total cell lysates and gradient-purified mitochondria were prepared from GFP-tagged strains. Total lysates were centrifuged 15 min at 13,000 ϫ g to prepare mitochondrion-free lysates. Samples (100 g protein) were separated on 10% acrylamide-0.1% SDS gels according to Laemmli (88) except that, to prevent denaturation of GFP, samples were not boiled (52). Gels were equilibrated in 100 mM Tris (pH 8.1) twice for 10 min each time, incubated for 5 min in 50 mM Tris (pH 8.1)-100 mM sodium dithionate, and imaged using a GE ImageQuant LAS-4010 imaging system. The use of sodium dithionate reduced flavin-containing proteins and decreased their fluorescence (89). For relative quantitation, mitochondrial fractions containing 10 g protein (NUO3-, NUO4-, and COX8-tagged strains) or 50 g protein (all others) were subjected to electrophoresis on the same gel and imaged. Quantitative analysis of the images was performed with ImageQuant TL ver. 8.1. Mitochondrial extracts were also fractionated by blue native gel electrophoresis and imaged for fluorescence.
Confocal microscopy. GFP-tagged strains were cultured to the stationary phase in YPD at 30°C and inoculated into fresh YPD at a density of approximately 1 ϫ 10 6 cells/ml. After 3 to 4 h of incubation, cells were labeled with MitoTracker Red CMXRos (Molecular Probes), fixed with formaldehyde, and imaged with a Zeiss LSM510 Meta confocal laser scanning microscope equipped with a Plan-Apochromat 100ϫ/1.4 oil objective. Details of the protocol are provided in Text S1.
Colocalization was analyzed as described by Costes et al. (51) using the Coloc 2 algorithm as implemented in the Fiji package of ImageJ (90). Images were processed with a 2-pixel median filter, and background was subtracted with a 1-pixel sliding paraboloid before quantitation. A minimum of 30 cells from at least two independent samples were analyzed, and the Costes significance test was based on 100 randomizations.
Germ tube and induction of hyphae. Germ tube formation was assessed in medium 199 -100 mM morpholinepropanesulfonic acid (MOPS; pH 7.5) inoculated to reach a density of approximately 2.5 ϫ 10 6 cells/ml with cells cultured 48 h in SC medium. To assess filamentation on agar surfaces, stationary-phase cells were diluted to 5 ϫ 10 7 cells/ml and 2 l of suspension was spotted on medium 199 (Gibco)-100 mM MOPS (pH 7.5) or YPD plus 10% fetal calf serum or YNBAGN (59) and incubated at 37°C. The effect of embedded conditions was tested by suspending approximately 50 cells in 25 ml YPD-plus-10% serum agar (held at 47°C) followed by pouring the reaction mixture into petri plates. After solidification, the plates were incubated at 37°C.
Fluconazole sensitivity. Strains were cultured to the stationary phase in SC medium. Ten-fold serial dilutions with densities ranging from 1 ϫ 10 8 to 1 ϫ 10 5 cells/ml were prepared, and 2 l of each was spotted on YPD or YPD with fluconazole (8 g/ml). Plates were incubated at 30°C. G. mellonella survival assay. The survival assay was conducted essentially as described by Fuchs et al. (62,91) using groups of 10 larvae (Vanderhorst, Inc.) (330 Ϯ 20 mg) injected with 5 ϫ 10 6 cells in 10 l PBS or PBS alone. Larvae were incubated at 37°C, and the number of dead larvae was scored daily. Kaplan-Meier analysis was performed, log rank statistics were calculated in R (77) using the package "survival" (92), and survival curves were plotted using "survminer" (93).