Carnitine acetyltransferases are required for growth on non-fermentable carbon sources but not for pathogenesis in Candida albicans

The polymorphic yeast Candida albicans is the most common fungal pathogen of humans and it can infect virtually any tissue and organ within the body. The most serious manifestation, disseminated haematogenous candidiasis, has a mortality rate of ∼40 % (Wenzel, 1995; Wisplinghoff et al., 2004). C. albicans is also, however, a ubiquitous and harmless part of our normal microbiota. The transition from commensal to pathogen generally results from changes in the host innate immune system. Therefore, the interaction between this fungal pathogen and phagocytic cells is a key to disease development.

We focus on the interaction between macrophages and C. albicans. In vitro, C. albicans is able to survive phagocytosis by differentiating into elongated hyphae that eventually disrupt the macrophage. An extensive microarray analysis of the Candida–macrophage interaction revealed a very complex programme of gene expression in response to phagocytosis in this fungus; a substantial part of this programme was the upregulation of specific metabolic pathways and stress response systems (Lorenz et al., 2004). Expression analysis of neutrophil–C. albicans interactions revealed a similarly complex response, but one that was quite different in terms of the pathways and processes induced (Fradin et al., 2003; Rubin-Bejerano et al., 2003). Thus it is clear that C. albicans elaborates a very sophisticated response to immune cell contact.

The transcript profiles of phagocytosed cells resemble those of cells starved for carbon, with particular induction of pathways to utilize alternative or non-fermentable carbon sources (Lorenz et al., 2004). Like other fungi, C. albicans uses glucose as the preferred carbon source, but can utilize a variety of non-fermentable carbon sources such as ethanol, acetate, citrate, lactate and oleate. These alternative carbon sources are assimilated through three main metabolic pathways: β-oxidation, the glyoxylate cycle and gluconeogenesis. We and others have shown that disabling these pathways through the mutation of key enzymes causes varying degrees of attenuation of virulence in a mouse model of disease (Barelle et al., 2006; Lorenz & Fink, 2001; Piekarska et al., 2006; Ramírez & Lorenz, 2007). These studies make it clear that carbon starvation is a relevant and important stress encountered by C. albicans during infection; as a result we are interested in further defining metabolic requirements in vivo.

A key intermediate in alternative carbon metabolism is acetyl-CoA, which is produced by the breakdown of ethanol, fatty acids and some amino acids, and is used by the glyoxylate and TCA cycles. Some of these processes are compartmentalized in the fungal cell, occurring in the peroxisome, mitochondria or cytosol, necessitating intracellular transport of acetyl-CoA. This is facilitated by carnitine acetyltransferases (CATs) that catalyse the transfer of the acetyl group from acetyl-CoA to carnitine followed by the transport of acetylcarnitine across the membrane. This process is referred to as the carnitine shuttle (Swiegers et al., 2001) and is important during growth on non-fermentable carbon sources. Saccharomyces cerevisiae has three CAT genes, YAT1, YAT2 and CAT2 (Swiegers et al., 2001). C. albicans also has three predicted CAT genes, CTN1, CTN2 and CTN3 (Prigneau et al., 2004); these genes are significantly upregulated by macrophage phagocytosis (Lorenz et al., 2004; Prigneau et al., 2004).

Here we describe the characterization of the CTN genes in C. albicans. To this end, we generated deletion mutants in the CTN1, CTN2 and CTN3 genes and tested their growth in vitro and in vivo. Δctn1 and Δctn2 mutants had an overlapping spectrum of growth defects in the presence of alternative carbon sources: the Δctn1 strain could not utilize acetate, ethanol or citrate while the Δctn2 mutant did not grow on acetate, ethanol or oleate and grew poorly on citrate. The Δctn3 mutant had no obvious phenotype, as previously reported (Prigneau et al., 2004). These phenotypes are more severe than the S. cerevisiae CAT mutants, reinforcing recent findings that there are significant differences in carbon metabolism between the two species. We verified that all three genes were induced in the macrophage, as well as on some non-fermentable carbon sources. We also showed that C. albicans CTN1 and CTN3 complement the corresponding S. cerevisiae mutants. None of these mutants, however, affected virulence in a mouse model of disseminated candidiasis. These findings extend our understanding of alternative carbon source utilization pathways in C. albicans and their role in pathogenesis.

Strain construction.
Strains are listed in Table 1. The C. albicans Δctn1, Δctn2 and Δctn3 mutants were constructed using the SAT1-flipper cassette (Reuß et al., 2004). Briefly, 600 bp fragments flanking the gene of interest were cloned between KpnI and XhoI (5') or SacII and SacI (3') sites in pSFS1. The resulting plasmid was linearized with KpnI and SacI and the disruption cassette was used to transform strain SC5314 by electroporation (Reuß et al., 2004) with selection on YPD plates containing 100 µg nourseothricin ml^–1. Correct integration was confirmed by PCR, and excision of the Nour^R cassette was induced by growth in YCB-BSA at 30 °C for 24 h. Proper excision was confirmed by PCR. The second allele was deleted in the same manner. Mutants were complemented by reintroducing the ORF at its native locus. The upstream fragment of the disruption construct was replaced with the complete ORF; the transforming and selecting were the same as for the disruptions. The double mutant Δctn1 Δctn3 was constructed by sequentially deleting the two genes.

Table 1. Strains

The S. cerevisiae mutants Δyat1, Δyat2, Δcat2 and Δcit2 were obtained from the haploid collection made by the Saccharomyces Genome Deletion Project (Winzeler et al., 1999). Δyat1 Δcit2, Δyat2 Δcit2 and Δcat2 Δcit2 double mutants were constructed by introducing Δcit2 : : Nour^R into Δyat1, Δyat2 and Δcat2 strains using a PCR disruption approach (Wach et al., 1994) using the Nour resistance cassette from pAG25 (Goldstein & McCusker, 1999). Correct double mutants were confirmed by PCR.

Plasmid construction.
To make S. cerevisiae overexpression plasmids, the ScYAT1, ScYAT2 and ScCAT2 ORFs were PCR amplified and cloned between EcoRI and HindIII sites in pRS426-GPD, a high-copy vector with the constitutive GPD1 promoter (Mumberg et al., 1995); CaCTN1 and CaCTN3 were cloned between BamHI and SalI sites; CaCTN2 was cloned between BamHI and XhoI sites. Plasmids used in this study are listed in Table 2.

Table 2. Plasmids

In vitro growth assays.
To test the growth of C. albicans strains on different carbon sources, the strains were grown in YPD to mid-exponential phase. Cells were collected by centrifugation and washed with water. They were transferred to a 96-well plate at OD₆₀₀ ∼0.5 and serially diluted fivefold. They were spotted using a multiprong pin replicator onto solid YNB medium containing 2 % glucose, potassium acetate, ethanol, sodium citrate, lactate or oleate as the sole carbon source, and incubated at 30 °C for 2–7 days, depending on the carbon source.

For the S. cerevisiae complementation assays, mutant strains Δyat1 Δcit2, Δyat2 Δcit2 and Δcat2 Δcit2 were transformed with the pRS426-GPD plasmids. Transformants were patched to YNB medium containing 2 % glucose and amino acids to satisfy growth requirements. Plates were incubated at 30 °C for 3 days and replica plated to media containing 2 % glucose, or 2 % ethanol plus 10 µg carnitine ml^–1, and incubated at 30 °C for 3–10 days.

Northern analysis.
SC5314 cells were grown in YNB with 2 % glucose overnight, collected by centrifugation, washed twice with water, resuspended in YNB with 2 % glucose, acetate, oleate or lactate and grown for 1 h. Cells were collected by centrifugation and frozen in dry ice-ethanol. Total RNA was extracted using the hot acid phenol method as described by Ausubel et al. (1993). Fifteen nanograms of each sample was separated on a 1 % MOPS-agarose gel with formaldehyde and then transferred to a nylon membrane. Gene-specific probes were amplified by PCR and were labelled with [³²P]dCTP using the RadPrime DNA labelling system (Invitrogen). Probes were purified using Roche Quick Spin columns. Blots were incubated in pre-hybridization solution containing 5xSSC (1xSSC is 0.15 M NaCl, 0.015 M sodium citrate), 50 % formamide, 5x Denhardt's solution, 0.1 % SDS and 100 µg ml^–1 single-stranded DNA for 2 h at 42 °C followed by hybridization overnight. rRNA was used as a loading control.

RT-PCR.
Cells (4x10⁷) of the murine macrophage line J774A were plated in a 750 ml tissue culture flask in RPMI with 10 % FBS and incubated overnight at 37 °C with 5 % CO₂. Overnight cultures of strain SC5314 were diluted 1 : 50 in YPD and grown for ∼4 h at 30 °C. Cultures were collected by centrifugation, washed in PBS and counted using a haemocytometer; 10⁸ C. albicans cells were added to the flasks containing macrophages (to give a C. albicans : macrophage ratio of approx. 2 : 1) and incubated for 1 h. For control cultures, 2x10⁸ C. albicans cells were inoculated to tissue culture flasks in 50 ml RPMI plus 10 % serum. Cells were collected by scraping the flasks with rubber scrapers in ice-cold water and transferring to 50 ml conical tubes. Cells were centrifuged at 3000 r.p.m. at 4 °C, washed twice with ice-cold water, transferred to microfuge tubes, and frozen on dry ice. Total RNA was isolated by hot phenol extraction (Ausubel et al., 1993). First-strand cDNA was synthesized according to the Invitrogen SuperScriptII RT protocol. The cDNA product was diluted 1 : 10 twice and 2 µl of each dilution was used as template in the PCR reaction. Primers were designed to amplify a ∼300 bp internal fragment of CTN1, CTN2 or CTN2, or 18S rRNA as a control.

Filamentation assay.
C. albicans strains from fresh YPD plates were struck to single colonies on medium 199 (pH 7.0), or YP medium containing 3 % glycerol or 10 % serum. Plates were incubated at 37 °C for 4 days. Colony morphology was examined under a Zeiss dissecting scope. Images were assembled in Adobe Photoshop.

In vivo virulence assays.
Mouse virulence assays were carried out as described previously (Ramírez & Lorenz, 2007). Adult, female, outbred ICR mice (21–25 g) were obtained from Harlan. Cultures of C. albicans were grown in YPD and serially passaged twice by overnight growth at 30 °C. Cells were collected by centrifugation, washed and resuspended in PBS. Mice were infected through tail vein injections with 10⁶ C. albicans yeast-form cells in 100 µl PBS. The group size was 10 mice per strain. Infected mice were monitored for signs of infection and euthanized when moribund according to approved protocols. Survival data were analysed with Prism3 (Graphpad Software) using the log rank test. All animal assays were conducted in accordance with protocols approved by the University of Texas Health Science Center Animal Welfare Committee.

Carbon metabolic pathways are compartmentalized into several organelles in eukaryotes, including the mitochondria and peroxisome, as well as the cytosol, and several shuttles facilitate the intracellular transport of various intermediates. Acetyl-CoA, the most important of these intermediates, is transported by three CATs. Fig. 1 shows the organization of these pathways inside the cell and the carnitine shuttle that serves to transport the key intermediate acetyl-CoA between the compartments.

(21K): Fig. 1. Schematic representation of the compartmentalization of alternative carbon sources, metabolic pathways and the carnitine shuttle. The C. albicans CATs Ctn1p, Ctn2p and Ctn3p are shown with their predicted localization and function. The figure is adapted from Swiegers et al. (2001).

Mutant strain construction
There are three genes coding for predicted CATs in C. albicans: CTN1 (YAT1, CAT1, orf19.4551), CTN2 (CAT2, orf19.4591) and CTN3 (YAT2, CAT3, orf19.2809). All three are upregulated significantly following macrophage phagocytosis (by 20.4-, 7.2- and 3.9-fold, respectively; Lorenz et al., 2004). Ctn1p is predicted to be associated with the outer mitochondrial membrane and is 58 % similar to S. cerevisiae YAT1. Ctn3p has two peroxisome-targeting signal motifs; otherwise it is 53 % similar to the cytosolic ScYat2p. CTN2 has two predicted translation initiation sites, with a mitochondrial targeting signal after the first methionine, and a peroxisome-targeting signal after the second methionine, suggesting a dual localization. This is a similar architecture to the yeast CAT2 gene, whose product is both peroxisomal and mitochondrial (Elgersma et al., 1995). These proteins are 63 % similar to each other (Prigneau et al., 2004).

To disrupt these three genes, we used the SAT1 flipper approach (Reuß et al., 2004) to precisely remove the ORFs of CTN1, CTN2 and CTN3 (see Methods). The SAT1 cassette (encoding resistance to nourseothricin) was excised and the second allele was disrupted in the same way. All three mutants were complemented by inserting one copy of the ORF back at its original locus as described (Reuß et al., 2004; see Methods). We also constructed a Δctn1 Δctn3 double mutant as part of the process of making a strain lacking all three CTN genes (we have not tested the Δctn1 Δctn2 or Δctn2 Δctn3 combinations). To date we have been unable to construct this triple mutant strain, which may indicate that this is an essential gene family. All mutants described here are homozygous deletions unless stated otherwise (see Table 1).

CTN mutants do not grow on certain non-fermentable carbon sources
We used a spot dilution assay to test the ability of the CAT mutant strains to assimilate different carbon sources. Strains were grown to exponential phase, washed and serially diluted, then spotted using a pin replicator to minimal YNB medium containing 2 % glucose, potassium acetate, ethanol, sodium citrate, lactate or the monounsaturated fatty acid oleate as the sole carbon source (Fig. 2). We included a Δfox2 mutant, lacking the β-oxidation gene, as a control; this mutant has pleiotropic defects on several carbon sources (Ramírez & Lorenz, 2007; Piekarska et al., 2006).

(53K): Fig. 2. Growth assay on different carbon sources. C. albicans wild-type (SC5314), Δfox2 (MRC6), Δctn1 (MLC48), Δctn3 (MLC50), Δctn1 Δctn3 (MLC53), Δctn1 : : CTN1 (MLC58), Δctn3 : : CTN3 (MLC60), Δctn2 (HZC30) and Δctn2 : : CTN2 (HZC32) strains were grown to mid-exponential phase and serially diluted 1 : 5 in a 96-well plate. The dilutions were applied using a multiprong replicator onto minimal YNB medium containing the indicated carbon sources at 2 %.

As shown in Fig. 2, the Δctn1 mutant was unable to grow on ethanol, potassium acetate and sodium citrate, although its growth on oleate and lactate was indistinguishable from the wild-type strain in this plate assay. The Δctn3 mutant did not show defects in growth on any of the carbon sources tested; this mutant was previously reported to have no phenotype in the presence of glucose, acetate or methanol (Prigneau et al., 2004). The double Δctn1 Δctn3 mutant had the same phenotype as the single Δctn1 mutant. The Δctn2 mutant failed to grow on media containing acetate, ethanol, or oleate and was significantly compromised on citrate. Growth was restored in the complemented strains. We also tested a second, independent mutant for each gene and observed identical results (data not shown).

These findings are in contrast to S. cerevisiae, in which the Δyat1, Δyat2 and Δcat2 mutants do not have any growth defect on non-fermentable carbon sources unless the peroxisomal citrate synthase CIT2 is also mutated (van Roermund et al., 1995; Swiegers et al., 2001). Though these proteins and metabolic pathways are very highly conserved, several similar phenotypic discrepancies have been recently described (Piekarska et al., 2006; Ramírez & Lorenz 2007; Sexton et al., 2007; M. A. Ramírez & M. C. Lorenz, unpublished observations), suggesting that there has been significant adaptation of the function or regulation of carbon metabolism in these species.

CTN genes are upregulated in macrophages and on non-fermentable carbon sources
To address the regulation of these genes in C. albicans, we assayed gene expression following phagocytosis and in the presence of various carbon sources. All three CTN genes have been shown to be induced following macrophage phagocytosis by microarray (Lorenz et al., 2004) and Northern analysis (Prigneau et al., 2004). CTN3, additionally, was repressed by glucose and induced by multiple carbon sources including glycerol, oleate, serum, acetate, ethanol and methanol (Prigneau et al., 2004). Given the homology between these genes, there is some concern that cross-hybridization could complicate this analysis, so we re-examined gene expression, testing for specificity using our knockout strains.

We collected wild-type C. albicans (SC5314) 1 h after macrophage phagocytosis and assayed gene expression by RT-PCR (Fig. 3a). CTN1 and CTN3 were strongly upregulated, as shown previously (Lorenz et al., 2004; Prigneau et al., 2004), though we were unable to demonstrate induction of CTN2 (which was the least induced CAT by microarray). The RT-PCR primers did not amplify a band from the corresponding mutant strains, demonstrating specificity (data not shown).

(40K): Fig. 3. Expression pattern of CTN1, CTN3 and CTN2. (a) Wild-type (SC5314) cells were added to flasks containing macrophages (108 cells, giving a 2 : 1 C. albicans : macrophage ratio) in RPMI plus 10 % serum and incubated for 1 h. Control cells were incubated in the same medium without macrophages. Cells were harvested and RNA extracted. The expression of these genes was determined by RT-PCR. For each gene/condition, three 1 : 10 serial dilutions of the RNA (indicated by the gradient sign) were amplified (see Methods). 18S rRNA was amplified as a control. (b) SC5314 was grown to mid-exponential phase in minimal YNB medium with glucose, collected by centrifugation, washed with water, and shifted to medium containing 2 % glucose (Glu), potassium acetate (Ace), oleate (Ole) and lactate (Lac) for 1 h. RNA was extracted from these cells and gene expression was determined by Northern analysis.

We next examined the expression pattern of these genes in different carbon sources by Northern blot analysis (Fig. 3b). None of the CAT-encoding genes were detectable in glucose- or lactate-grown cells; CTN1 and CTN3 were induced in acetate-grown cells and further induced moderately (CTN3) or dramatically (CTN1) in oleate. CTN2 was detectable only after growth on the fatty acid oleate (Fig. 3b). The probes are specific to each gene and there was no cross-hybridization (data not shown). In summary, all three CAT genes were upregulated on non-fermentable carbon sources, most strongly in cells exposed to oleate.

The induction of the CAT genes by non-fermentable carbon sources is similar to that of the S. cerevisiae homologues. In S. cerevisiae YAT1 was repressed by glucose and induced by ethanol (Schmalix & Bandlow, 1993). CAT2 was induced tenfold in oleate as compared to glucose (Atomi et al., 1993).

CTN1 and CTN3 complement S. cerevisiae mutants
The C. albicans CTN genes are annotated based on their sequence homology to S. cerevisiae YAT1, YAT2 and CAT2. To determine the validity of these assignments, we tested whether the C. albicans CTN genes can complement the S. cerevisiae mutant phenotypes. Since in S. cerevisiae, mutation in CAT genes alone does not yield a growth defect on non-fermentable carbon sources, we introduced the Δcit2 mutation into Δyat1, Δyat2 and Δcat2 strains (see Methods). Indeed, as previously described (Swiegers et al., 2001), the double mutants were unable to grow on ethanol as sole carbon source (Fig. 4 and data not shown). The double mutant strains were then transformed with high-copy plasmids containing the CAT genes from S. cerevisiae or C. albicans under the control of a constitutive GPD1 promoter.

(79K): Fig. 4. Complementation of the S. cerevisiae mutants. S. cerevisiae strains Δyat1 Δcit2, Δyat2 Δcit2 and Δcat2 Δcit2 were transformed with pRS426-GPD-ScYAT1, ScYAT2, ScCAT2, CaCTN1, CaCTN3 and CaCTN2 respectively. Transformants were patched to minimal YNB medium containing 2 % glucose and incubated at 30 °C for 3 days before replica plating to minimal YNB medium containing 2 % glucose, or 2 % ethanol plus carnitine (10 mg l–1), and incubated at 30 °C for 3 (glucose) or 10 (ethanol) days.

As shown in Fig. 4, the Δcit2 Δcat double mutant strains transformed with empty vector were unable to grow on ethanol as the sole carbon source, and overexpression of the S. cerevisiae genes complements these phenotypes. CaCTN1 restores growth on ethanol in both the Δyat1 and Δyat2 mutants, while CaCTN3 complements only Δyat2. We were unable to complement Δcat2 with any of the C. albicans genes, and the CaCTN2 gene did not function in S. cerevisiae. This can occasionally be the result of the altered codon usage in C. albicans, but we found no CTG codons in CTN1 or CTN2. CTN3, which is functional in the S. cerevisiae Δyat2 mutant, has five CTG codons. Thus, this analysis confirms that there is some conservation of function in this gene family but again suggests that there are some significant differences in carbon metabolism between these related species.

CTN mutants undergo hyphal morphogenesis on solid media
In a previous study, deletion of CTN3 was found to disrupt filamentation on solid media (Prigneau et al., 2004). We tested whether the deletion of the other two genes also affects colony morphology under the same conditions. The wild-type and Δctn1, Δctn2 and Δctn3 mutant strains were incubated on solid YP-glycerol, YP-serum or medium 199. After 4 days of growth at 37 °C, all four strains formed long peripheral hyphae (Fig. 5 and data not shown), with no difference between the wild-type and the mutant strains under the conditions tested. In contrast to the earlier report for Δctn3, we found that none of these mutants were affected in their filamentation ability. This difference in phenotype may be due to the URA3 deletion marker used in the earlier study; chromosomal position effects of this gene have been found to alter several phenotypes, including hyphal differentiation (Bain et al., 2001; Lay et al., 1998).

(58K): Fig. 5. Peripheral filaments of C. albicans. wild-type (SC5314), Δctn1 (MLC48), Δctn3 (MLC50) and Δctn2 (HZC30) strains from fresh YPD plates were struck to single colonies on YP medium containing 3 % glycerol or 10 % serum. Plates were incubated at 37 °C for 4 days before photographing under a Zeiss dissecting scope.

CTN mutants are not attenuated in virulence in the mouse model
Previous work has demonstrated that perturbations in alternative carbon source utilization pathways led to mild to significant attenuation in virulence in a mouse model of haematogenously disseminated candidiasis. These include β-oxidation of fatty acids via the Δfox2 mutation (Ramírez & Lorenz, 2007; Piekarska et al., 2006), the glyoxylate cycle via the Δicl1 mutation (Ramírez & Lorenz, 2007; Lorenz & Fink, 2001), and gluconeogenesis via the Δfbp1 (Ramírez & Lorenz, 2007) or Δpck1 (Barelle et al., 2006) mutations. To determine if CATs play any role in survival and pathogenesis in vivo, we tested the Δctn mutants in a standard mouse tail vein injection model. Briefly, 10⁶ C. albicans cells (yeast form) were injected into outbred adult female mice. Infected mice were monitored for signs of infection and euthanized when moribund according to approved protocols.

Mice infected with wild-type strains had a mean time to death (MTD) of 3.9 days (Fig. 6). Mice infected with the mutants had a comparable death rate, with the MTD for Δctn1, Δctn2 and Δctn3 being 4.8, 4.0 and 4.2 days respectively (Fig. 6a, b, c). The double mutant Δctn1 Δctn3 behaved similarly (Fig. 6d). We also tested an independent strain for each mutant and found the same results. Thus, deletions in the CTN genes have no obvious effect on in vivo survival and virulence even though they caused defects in carbon utilization in vitro that were similar to other avirulent mutants.

(18K): Fig. 6. Virulence test in the mouse model of disseminated candidiasis. A total of 106 cells of each strain were injected via the tail vein. All assays used the wild-type strain SC5314 as a control. (a) Δctn1 (MLC48) vs Δctn1 : : CTN1 (MLC58); (b) Δctn2 (HZC30) vs Δctn2 : : CTN2 (HZC32); (c) Δctn3 (MLC50) vs Δctn3 : : CTN3 (MLC60); (d) Δctn1 Δctn3 double mutant (MLC53).

In this study, we have examined the function and regulation of the three predicted C. albicans CAT genes, CTN1, CTN2 and CTN3. Mutations in two of the three genes (CTN1, CTN2) confer growth defects in the presence of non-fermentable carbon sources and all three genes are induced under these conditions. However, strains lacking one or two of these genes retain full virulence, marking a discrepancy between in vitro carbon utilization and in vivo virulence phenotypes, in contrast to other recent reports for metabolic genes (Piekarska et al., 2006; Ramírez & Lorenz, 2007).

Both the C. albicans Δctn1 and Δctn2 mutant strains were unable to grow on media containing ethanol or acetate. Both of these compounds are converted to acetyl-CoA and transported into the mitochondria via the carnitine shuttle, where they can be assimilated via the TCA and/or glyoxylate cycles. Ctn2p is predicted to be in both the peroxisome and the mitochondrial matrix while Ctn1p is associated with the mitochondrial outer membrane (see Fig. 1). Combined with these localizations, our data suggest that acetyl-CoA is converted to acetylcarnitine on the mitochondrial surface by Ctn1p, transported into the matrix, and converted back to acetyl-CoA by Ctn2p before incorporation into the TCA cycle. During growth on fatty acids, the acetyl-CoA produced by peroxisomal β-oxidation is converted to acetylcarnitine by Ctn2p, and could move from the peroxisome to the mitochondria (via the cytosol) with no need for Ctn1p and, indeed, Δctn1 mutants can use oleate as a carbon source. The weak growth of Δctn2 mutants on citrate is potentially explained by the presence of a mitochondrial citrate transport protein; C. albicans has an uncharacterized homologue of the well-studied S. cerevisiae Ctn1p protein (Kaplan, et al., 1996), thus citrate can probably enter the mitochondria in the absence of CATs. The phenotypes we observed clearly indicate a need for CAT activity during citrate metabolism, however.

As previously reported, Δctn3 strains do not have carbon utilization phenotypes that we could detect (Prigneau et al., 2004). CTN1 complemented deletions in both S. cerevisiae YAT1 and YAT2, while CTN3 complemented only the Δyat2 mutation. This demonstrates a basic conservation of function of these proteins, though this is not absolute, since none of these could restore growth to a S. cerevisiae Δcat2 mutant.

In an earlier study, CTN3 was found to affect filamentous growth, as Δctn3 strains were unable to form hyphae in several inducing conditions (Prigneau et al., 2004). In contrast, we report here that deletion of the CTN genes did not disrupt formation of hyphae. While there may be some differences in strain background, both studies used SC5314-derived strains. In the study of Prigneau et al. (2004), the CTN3 mutant strains were constructed using the URA blaster method (Fonzi & Irwin, 1993). This discrepancy may be related to URA3 position effects, which have been subsequently recognized to cause numerous phenotypes, including morphogenetic, that are unrelated to the deleted gene (Bain et al., 2001; Brand et al., 2004; Lay et al., 1998). The current study did not use ura3 auxotrophic strains; instead the mutations were constructed in a prototrophic strain using a dominant drug-resistance marker, nourseothricin (Reuß et al., 2004).

Although the CTN mutant strains had dramatic growth defect on non-fermentable carbon sources in vitro, there was no attenuation of virulence in the systemic infection model. We found this result surprising, given that the in vitro phenotypes of the Δctn strains are similar to those of published mutants with known virulence defects, such as Δicl1, Δfbp1, Δpck1 and Δfox2 (Lorenz & Fink, 2001; Piekarska et al., 2006; Ramírez & Lorenz, 2007; Barelle et al., 2006). This discrepancy may result from greater expression of the CTN genes in the host environment, mitigating effects of single gene deletions on overall CAT activity. Alternatively, C. albicans could import simple carbon molecules, such as acetate or lactate, from the host that do not require intracellular transport for utilization or that make use of other transport systems, such as the mitochondrial succinate–fumarate shuttle or the citrate transporter. In a co-culture assay with murine macrophage cell line J774A, all the deletion strains were able to form hyphae and escape from macrophage engulfment at the same rates as the wild-type control (data not shown). This has been observed previously, even in strains with in vivo virulence defects (Lorenz & Fink, 2001; Ramírez & Lorenz, 2007) and we believe that the short duration of the macrophage assay does not allow sufficient time to reveal starvation-related defects.

In contrast to the Δctn1 and Δctn2 mutations, there is no phenotype associated with the deletion of the S. cerevisiae CAT genes (YAT1, YAT2, CAT2) (Kispal et al., 1993; Schmalix & Bandlow, 1993; Swiegers et al., 2001). Indeed, any of these single mutants will grow on non-fermentable carbon sources unless the CIT2 gene, encoding a peroxisomal isoform of citrate synthase, is also deleted. This indicates a flexibility not present in C. albicans, an idea reinforced by the presence of three CIT isoforms in S. cerevisiae, and only a single gene in C. albicans (CIT1, highly homologous to all three S. cerevisiae proteins). This redundancy likely allows S. cerevisiae to run much of the glyoxylate cycle in either the peroxisome or mitochondria and increases the spectrum of compounds that could be shuttled between compartments, thereby reducing the importance of acetyl-CoA transport under most circumstances. Why this is the case is not clear, but it is tempting to speculate that it improves the efficiency of ethanol utilization, a process more important in the highly fermentative S. cerevisiae than in other yeasts.

Though these carbon metabolic pathways and enzymes are highly conserved, there have been several previous examples in which gene deletions do not have the same phenotype in these related species. Snf1p, a kinase involved in glucose repression, is essential in C. albicans, but not in S. cerevisiae (Petter et al., 1997). In addition, mutations of the C. albicans fox2 gene, encoding an enzyme of β-oxidation, has several pleiotropic phenotypes not seen in the S. cerevisiae mutant (Piekarska et al., 2006; Ramírez & Lorenz, 2007). It is worth considering whether the more restricted localization of carbon metabolic pathways in C. albicans that we propose here underlies this pleiotropy. The adaptation of metabolic networks to suit the in vivo niche of C. albicans is likely to be an important aspect of the biology of this pathogen.

We thank A. Carman and M. Ramírez for assistance with the mouse virulence assays and A. Carman and K. Morano for comments on the manuscript. This work was supported in part by NIH awards 1R01AI075091-01 and 5T32DE015355-05.

Edited by: J. M. Becker