A CTG Clade Candida Yeast Genetically Engineered for the Genotype-Phenotype Characterization of Azole Antifungal Resistance in Human-Pathogenic Yeasts

ABSTRACT

A strain of the opportunistic pathogenic yeast Candida lusitaniae was genetically modified for use as a cellular model for assessing by allele replacement the impact of lanosterol C14α-demethylase ERG11 mutations on azole resistance. Candida lusitaniae was chosen because it is susceptible to azole antifungals, it belongs to the CTG clade of yeast, which includes most of the Candida species pathogenic for humans, and it is haploid and easily amenable to genetic transformation and molecular modeling. In this work, allelic replacement is targeted at the ERG11 locus by the reconstitution of a functional auxotrophic marker in the 3′ intergenic region of ERG11. Homologous and heterologous ERG11 alleles are expressed from the resident ERG11 promoter of C. lusitaniae, allowing accurate comparison of the phenotypic change in azole susceptibility. As a proof of concept, we successfully expressed in C. lusitaniae different ERG11 alleles, either bearing or not bearing mutations retrieved from a clinical context, from two phylogenetically distant yeasts, C. albicans and Kluyveromyces marxianus. Candida lusitaniae constitutes a high-fidelity expression system, giving specific Erg11p-dependent fluconazole MICs very close to those observed with the ERG11 donor strain. This work led us to characterize the phenotypic effect of two kinds of mutation: mutation conferring decreased fluconazole susceptibility in a species-specific manner and mutation conferring fluconazole resistance in several yeast species. In particular, a missense mutation affecting amino acid K143 of Erg11p in Candida species, and the equivalent position K151 in K. marxianus, plays a critical role in fluconazole resistance.

INTRODUCTION

Azole antifungals are the most widely used antimicrobial agents for the treatment of superficial and invasive fungal infections and for prophylaxis and empirical treatment in clinical wards, notably in oncohematology units (1). Frequent exposure to azole antifungals may contribute to the selection of resistance, both in filamentous fungi (2) and in Candida yeasts (3). The cellular target of azole antifungals is Erg11p (synonym CYP51 for the filamentous fungi), the lanosterol 14α-demethylase of the ergosterol biosynthetic pathway. By considering their frequency, three main mechanisms generally support resistance to azoles in Candida yeast (3, 4): point mutations of the ERG11 gene, which decrease the binding affinity of the target protein to azole antifungals; overexpression of ERG11, which increases the intracellular concentration of the target protein; and overexpression of efflux membrane transporters, which lowers the intracellular concentration of the antifungal drug. In C. albicans, the two last mechanisms may arise from gene duplication by aneuploidization (5) or from point mutations in the transcription factors UPC2 and TAC1, which regulate expression of the relevant gene ERG11 and CDR efflux pumps, respectively (6, 7). In addition, mutations arising by single-nucleotide polymorphism (SNP) of the ERG11 gene are expected to occur frequently. More than 150 missense mutations within the coding region of ERG11 were described so far in different yeast species (8), but very few of them were demonstrated to be directly responsible for the resistance. The impact of a missense mutation on azole resistance can be demonstrated either by purifying the mutated Erg11 protein and measuring the kinetic parameters of the enzymatic activity in the presence and in the absence of azole inhibitors (9, 10) or by heterologous expression of the ERG11 allele in a yeast host model, generally Saccharomyces cerevisiae (9, 10, 11). However, while most Candida species are susceptible to azole antifungals, S. cerevisiae is much less susceptible, particularly to fluconazole. A genetically modified S. cerevisiae strain lacking the Pdr5p ABC efflux transporter was developed to enhance its susceptibility to fluconazole for heterologous expression experiments (12). Other issues may arise from the use of a 2μ-derived multicopy plasmid for expressing ERG11 alleles in S. cerevisiae and from driving its expression with strong promoters, such as GAL10 or ADH1 (12), which can contribute to overexpress ERG11 and to decrease azole susceptibility (13). Finally, S. cerevisiae is very distant from the CTG clade of yeasts that decode the CUG codon as a serine instead of a leucine (14, 15, 16), sometimes making site-directed mutagenesis necessary to render the Candida ERG11 allele compatible with the genetic code of S. cerevisiae (10).

In this work, we created a genetically engineered strain of Candida lusitaniae (teleomorph Clavispora lusitaniae) to characterize the effect of missense point mutations of the ERG11 gene at the phenotypic level. Compared to other biological systems, C. lusitaniae has the advantages of belonging to the CTG clade of Candida species, of being haploid, and of being susceptible to azole antifungals with no trailing growth, allowing detection of very small MIC variations. Conceptually, the system relies on the replacement of the resident ERG11 allele of C. lusitaniae by any other ERG11 allele. Replacement is targeted at the ERG11 locus by the reconstitution of a functional auxotrophic marker in the 3′-untranslated region of the ERG11 gene, and expression of the new allele is driven by the C. lusitaniae resident ERG11 promoter. Our system was challenged by expression in C. lusitaniae-susceptible and -resistant ERG11 alleles from C. lusitaniae and C. albicans and from a clinical isolate of Kluyveromyces marxianus cross-resistant to all azole antifungals and which bore two missense mutations leading to E123Q and K151E amino acid substitutions (17). Both homologous and heterologous expression in C. lusitaniae allowed us to resolve unequivocally the link between ERG11 mutations and azole resistance.

RESULTS

Phylogenetic analysis of fungal pathogens based on lanosterol C14α-demethylase.

Before using Candida lusitaniae as an alternative yeast species for heterologous expression of ERG11, we verified its position in a phylogenetic tree made with a total of 52 protein sequences identified as fungal lanosterol C14α-demethylase and that of the oomycete Saprolegnia parasitica, a parasite of fish close to the fungal kingdom, used to root the tree. The selection included 25 Ascomycota yeasts (2 Taphrinomycota, 23 Saccharomycotina), 21 Pezizomycotina hyphomycetes, of which 17 were Eurotiomycetes and 4 were Sordariomycetes, 4 Basidiomycota yeasts, and 2 Mucormycota. This selection contains species pathogenic for plants and animals, including humans, except 13 ascomycetous yeast species that are nonpathogenic. The tree thus obtained (Fig. 1) overlaps the phylogeny of the fungal species generally observed with other genetic markers (18, 19). Most of the human pathogens, including opportunistic fungi, of the phylum Ascomycota are distributed over two subphyla, the Pezizomycotina and the Saccharomycotina. The Pezizomycotina branch separates in two lineages of pathogens corresponding to the Sordariomycetes (Fusarium species) and the Eurotiomycetes (Aspergillus; dermatophytes and dimorphic fungi). The Saccharomycotina branch supports two lineages. One, also known as the CTG clade, contains the higher number of opportunistic pathogenic yeast of the genus Candida, including C. albicans and C. lusitaniae. These yeasts decode the CUG codon as a serine instead of a leucine. The other lineage contains almost nonpathogenic yeasts, except C. glabrata and K. marxianus, and is represented by S. cerevisiae. The phylogenetic position of C. lusitaniae close to the node that separates the two lineages of yeasts makes it attractive for use as a versatile heterologous expression system among pathogenic yeasts. An EMBOSS Needle (EMBL-EBI) pairwise alignment was performed to evaluate the distance between the Erg11 protein of C. lusitaniae and those of C. albicans (70.5% identity) and K. marxianus (64.2% identity), which were further used in this study.

FIG 1.

Maximum likelihood tree of pathogenic filamentous fungi and pathogenic and nonpathogenic yeasts using a Clustal Omega alignment of amino acid sequences of lanosterol 14-alpha-demethylase. Bootstrap values: thick branch, 90 to 100%; medium branch, 70 to 89%; thin branch, lower than 70%. Asterisks indicate species that are not considered to be pathogenic. Arrows show the position of the yeast species used in this study.

Engineering a C. lusitaniae strain for ERG11 allele replacement.

The ERG11 gene (CLUG_04932) was located on the supercontig 6 of the genome of the C. lusitaniae strain ATCC 42720 (GenBank accession number NW_003101573). The intronless open reading frame of 1,578 nucleotides encodes a predicted protein of 525 amino acids. Synteny with C. albicans SC5314 was conserved. At 1,055 bp downstream of C. lusitaniae ERG11 is a gene encoding a putative homoserine kinase (THR1 gene), and at 2,040 bp upstream an open reading frame (ORF) with unknown function is present. The intergenic region between ERG11 and THR1 was large enough to integrate a genetic marker. As a recipient strain we used C. lusitaniae CBS 6936 ura3Δ, a mutant strain bearing a deletion of 990 bp covering the totality of the URA3 gene. CBS 6936 is the type strain of the species Clavispora lusitaniae, the teleomorph of C. lusitaniae. The strain is susceptible to all antifungals, and our team developed several molecular tools to modify its genome. The release of the genome of the strain CBS 6936 (GenBank accession number LYUB02000000 [20]) allowed us to verify that the genomic map surrounding ERG11 (GenBank accession number OVF10151.1) was comparable to that of ATCC 42720. In a first step, URA3 was integrated within the ERG11-THR1 intergenic region (Fig. 2A). For that, we used a construction made of the URA3 gene with its own 5′- and 3′-untranslated region (UTR) flanked on each side by large DNA fragments (around 500 bp) homologous to the ERG11-THR1 intergenic region of integration. The linear DNA cassette was transformed by electroporation into the ura3Δ recipient strain, and the selected prototroph colonies were verified for correct targeted integration of the cassette by PCR and nucleotide sequencing, which occurred in 8 transformants over 12 analyzed (75% of targeted integration events). One transformant was selected (strain E11URA3), and in a second step, the promoter and two-thirds of the 5′ part of the URA3 coding region were removed from strain E11URA3. For that, we constructed a deletion cassette made of the 3′-UTR ERG11 region directly fused to the 3′ coding part of URA3 (Fig. 2B) to transform strain E11URA3. Replacement of the wild URA3 copy by integration of the truncated ura3 cassette occurred in 100% of the clones analyzed, conferring resistance to the toxic analog 5-fluoroorotic acid (5FOA) (21). The molecular organization was verified by nucleotide sequencing and Southern blot analysis of EcoRV-digested genomic DNA with a URA3 probe (Fig. 2C). One strain harboring a nonfunctional part of the URA3 gene in its ERG11-THR1 intergenic region was selected and named E11ura3Δ5′ (deleted in the 5′ end of URA3) and was further used in this study for all ERG11 allele replacement experiments. We verified that the molecular modifications performed in strain E11ura3Δ5′ did not change its susceptibility to antifungals: in particular, the MIC of fluconazole was 0.5 μg/ml, identical to that of wild-type strain 6936.

FIG 2.

Molecular modifications sequentially achieved to obtain the recipient strain E11ura3Δ5′. (A) Transformation of the 6936 ura3Δ strain with a DNA cassette bearing the genetic marker URA3 (white box) flanked by DNA sequences homologous to the 3′-UTR region of the ERG11 locus (gray boxes). Crossing lines symbolize recombination events that allow the incoming linear DNA cassette to be integrated into the recipient chromosome to give the E11URA3 strain. (B) Transformation of the E11URA3 strain with a DNA cassette bearing a nonfunctional URA3 truncated of its promoter and of the NH2 part of the ORF (white box), resulting after recombination in the strain E11ura3Δ5′. (C) Southern blot analysis of the different strains used and constructed in this study after digestion of the genomic DNA with EcoRV and hybridization with URA3 probes. DNA fragment size is given in kilobase pairs. WT, wild-type strain 6936; EV, EcoRV site; ATG, initiation codon; TAA, stop codon; UTR, untranslated regulatory region; IG, intergenic region.

ERG11 allele replacement.

ERG11 allele replacement was targeted by the reconstitution of a functional URA3 marker in the ERG11-THR1 intergenic region. The promoter region of the C. lusitaniae ERG11 gene, the ORF of the new allele ERG11 to be tested with its own 3′ UTR region, and the promoter and 5′ coding part of URA3 were fused in vitro as a single fragment and transformed into the E11ura3Δ5′ strain (Fig. 3). The 5′ coding part of the URA3 gene was designed to be redundant over 300 bp with the resident 3′ coding part, so that a crossing over can take place allowing integration of the DNA cassette and the concomitant reconstitution of a functional URA3 marker. Recovery of prototroph transformant yeast cells indicated that the DNA cassette had integrated in the ERG11 region. Molecular verifications by nucleotide sequencing showed that the second recombination event necessary for integration of the cassette took place in the promoter region of the C. lusitaniae ERG11 gene in 100% of the transformants tested. Before measuring the fluconazole MIC by the CLSI microplate method in RPMI medium lacking uracil, we verified that the strain E11ura3Δ5′, expressing different ERG11 alleles, had no growth defect attributable to decreased expression of the URA3 gene compared to the growth of the reference strain 6936.

FIG 3.

Strategy used to replace the resident ERG11 allele of C. lusitaniae by any other ERG11 allele bearing or not bearing a mutation. The E11ura3Δ5′ recipient strain was transformed by a DNA cassette made of the promoter region of the C. lusitaniae ERG11 gene (gray box, 5′-UTR), the ORF, and the 3′-UTR region of any new ERG11 allele (black box) and the URA3 complementary information (white box), allowing reconstitution and expression of a functional URA3 allele downstream of ERG11. *, mutation; abbreviations are defined in the legend to Fig. 2.

Heterologous expression of C. albicans ERG11 allele in C. lusitaniae.

Our genetic model was first challenged with the ERG11 wild-type (CaERG11wt) alleles A and B of the diploid C. albicans SC5314. This strain, commonly used in research laboratories, is susceptible to fluconazole but exhibits a strong trailing growth that can make MIC determination difficult (Fig. 4A). Using Etest, the fluconazole MIC inhibiting 50% of the growth of C. albicans SC5314 was 0.094 μg/ml. When expressed in the E11ura3Δ5′ strain of C. lusitaniae, both CaERG11wt alleles A and B conferred susceptibility with a MIC of 0.047 μg/ml (as shown for allele A in Fig. 4B), that is to say a MIC comparable to that observed for C. albicans SC5314 and 10 times lower than that measured for C. lusitaniae E11ura3Δ5′ (0.5 μg/ml). This result shows the tight correlation between the Erg11 protein expressed and the fluconazole MIC conferred to C. lusitaniae E11ura3Δ5′ and suggests that susceptibility to fluconazole mainly depends on Erg11p. Moreover, while a strong residual growth of C. albicans SC5314 could be observed within the inhibition ellipse, growth of C. lusitaniae E11ura3Δ5′ expressing CaERG11wt was fully inhibited within the ellipse zone. As already mentioned, this growth inhibition was not the consequence of a growth defect of the E11ura3Δ5′ strain in RPMI lacking uracil but rather the result of the susceptibility of C. albicans Erg11p to fluconazole inhibition. The CLSI method was preferred to Etest in the rest of the study because of the occurrence of fluconazole hypersusceptibility with some ERG11 alleles. With the CLSI method, the CaERG11wt alleles A and B conferred susceptibility comparable to that of C. lusitaniae E11ura3Δ5′ with a MIC of 0.0625 μg/ml (Table 1).

FIG 4.

Etest determination of fluconazole susceptibility after 24 h of incubation on RPMI medium. (A) C. albicans SC5314. (B) C. lusitaniae E11ura3Δ5′ strain expressing the ERG11 allele A of C. albicans SC5314.

TABLE 1.

Susceptibility of C. lusitaniae E11ura3Δ5′ strains expressing different ERG11 alleles to fluconazole

ERG11 donor strain	ERG11 allele/mutation	FLC MICa (μg/ml)	MIC variation factorb	S/Rc
C. albicans SC5314	WT allele A	0.0625	Ref	S
C. albicans SC5314	WT allele B	0.0625	1	S
C. albicans SC5314	G464S	0.25	4	S
C. albicans SC5314	Y132H	0.0625	1	S
C. albicans SC5314	G464S + Y132H	0.25	4	S
C. lusitaniae 6936	G460S	0.5	1	S
C. lusitaniae 6936	Y132H	0.5	1	S
K. marxianus PAZ	E123Q + K151E	64	Ref	R
K. marxianus PAZ	E123Q + K151	0.0078	1/8,200	S
K. marxianus PAZ	E123 + K151E	32	0.5	R
K. marxianus PAZ	E123 + K151	0.0078	1/8,200	S
C. lusitaniae 6936	E115 + K143 (WT)	0.5	Ref	S
C. lusitaniae 6936	E115Q + K143	0.5	1	S
C. lusitaniae 6936	E115 + K143E	8	16	R
C. lusitaniae 6936	E115Q + K143E	16	32	R
C. albicans SC5314	E115 + K143E	8	128	R

^aMICs were measured by the microplate dilution CLSI method and defined as the minimal concentration of FLC inhibiting 90% of the growth of the strain compared to its growth in drug-free medium.

^bRef corresponds to the FLC MIC inhibiting growth of the C. lusitaniae strain E11ura3Δ5′ expressing the unmodified ERG11 wild-type allele A of C. albicans SC51314, the clinical allele of K. marxianus PAZ, or the wild-type allele of C. lusitaniae. MIC variation is expressed as a multiplication factor corresponding to the MIC conferred by a modified allele divided by the MIC of the reference allele from the same donor strain.

^cS, susceptible (MIC of ≤2 μg/ml); R, resistant (MIC of >4 μg/ml).

Our model was further challenged with three different ERG11 alleles of C. albicans SC5314 in which nonsynonymous mutations were introduced in vitro. We chose two mutations previously described in C. albicans to support fluconazole MIC increase. The mutation Y132H (tyrosine at position 132 replaced by histidine) is located in hot spot I of the protein, a helix-rich segment important for substrate entry, and the mutation G464S (glycine at position 464 replaced by a serine) in the hot spot III region, important for heme fixation (22).

Both mutations were introduced separately and in combination in the CaERG11wt allele A, and the modified alleles were expressed in C. lusitaniae E11ura3Δ5′. The Y132H replacement did not result in any fluconazole MIC increase, with the strain being as susceptible to fluconazole (MIC of 0.0625 μg/ml) as the one expressing the unmodified CaERG11wt allele. The G464S replacement gave a 4-fold increase in fluconazole MIC, and the combination of both mutations Y132H and G464S yielded the same MIC as that for the single G464S replacement. As a control, we introduced the mutations Y132H and G464S separately in a C. lusitaniae ERG11 wild-type allele (ClERG11wt) and expressed them in E11ura3Δ5′, but no change of the fluconazole MIC was observed compared to that of the wild-type allele (fluconazole MIC of 0.5 μg/ml). The expression level of the wild-type and mutated ERG11 alleles in E11ura3Δ5′ was verified by reverse transcription-quantitative PCR (RT-qPCR) (Fig. 5A and C). In the case of C. albicans, the amount of mRNA of each mutated allele varied significantly (P < 0.05) compared to that of the wild-type allele, but the amplitude of variation did not exceed a factor of 2.5 (e.g., for the Y132H and G464S alleles). Moreover, a 4.3-fold difference in mRNA amounts between the CaERG11 G464S allele and the CaERG11 Y123H + G464S allele has no effect on the fluconazole MIC (0.5 μg/ml for both alleles). We therefore considered that the wild-type and mutated ERG11 alleles were expressed at a similar level in C. lusitaniae E11ura3Δ5′ and that small mRNA variations had no influence on fluconazole MIC.

FIG 5.

Normalized expression levels of different ERG11 alleles in the C. lusitaniae E11ura3Δ5′ strain grown in RPMI medium. (A) C. albicans ERG11 allele expression in fold variation relative to the expression level of the reference SC5314 WT ERG11 allele A in C. lusitaniae. (B) K. marxianus ERG11 allele expression in fold variation relative to the expression level of the reference K. marxianus PAZ clinical resistant ERG11 allele in C. lusitaniae. (C) C. lusitaniae ERG11 allele expression in fold variation relative to the expression level of the reference 6936 WT ERG11 allele in C. lusitaniae. Expression of reference alleles (described in Table 1) in the C. lusitaniae E11ura3Δ5′ strain was set to 1. ERG11 allele expression levels were quantified and normalized relative to the housekeeping gene ACT1. Values are presented with SEM (standard errors of the means). *, P < 0.05 compared to the respective reference ERG11 allele.

Heterologous expression of K. marxianus ERG11 alleles from a clinical fluconazole-resistant isolate in C. lusitaniae.

Kluyveromyces marxianus (anamorph Candida kefyr) is a yeast species emerging in oncohematology. We recently described a K. marxianus clinical isolate which was resistant to azole antifungals (17). Nucleotide sequencing of ERG11 revealed two missense mutations leading to the amino acid substitutions E123Q and K151E compared to the Erg11p sequence of the fluconazole-susceptible K. marxianus reference strain CBS 6556. Expression in C. lusitaniae E11ura3Δ5′ of the K. marxianus ERG11 allele from the resistant clinical isolate (KmERG11ci) bearing both E123Q and K151E mutations conferred fluconazole resistance with a MIC as high as 64 μg/ml (Table 1). We then engineered different KmERG11ci alleles by removing each of the mutations separately and then both at the same time. Expression of these alleles in C. lusitaniae E11ura3Δ5′ showed that the sole mutation K151E was necessary to confer fluconazole resistance of up to 32 μg/ml. Surprisingly, expression of a KmERG11ci allele bearing only the E123Q mutation or no mutation conferred hypersusceptibility to fluconazole (fluconazole MIC of 0.0078 μg/ml), that is to say a MIC nearly 8,200 times lower than that conferred by the KmERG11ci allele bearing the two mutations and 64 times lower than that observed for C. lusitaniae E11ura3Δ5′ expressing its own wild-type ERG11 allele (MIC of 0.5 μg/ml).

The E123Q and K151E mutations of KmERG11ci would be equivalent to E115Q and K143E in ERG11 alleles of Candida species. As a control, the mutations E115Q and K143E were introduced in combination and separately in the C. lusitaniae ERG11 wild-type allele and expressed in the E11ura3Δ5′ strain. This confirmed that the K143E mutation supported a 16-fold fluconazole MIC increase from 0.5 to 8 μg/ml and that the mutation E115Q did not influence the MIC compared to the wild-type allele (MIC of 0.5 μg/ml). However, we observed repeatedly that when K151/143E was combined with E123/115Q in the ERG11 alleles of both K. marxianus and C. lusitaniae, the resulting ERG11 allele conferred a 2-fold higher fluconazole MIC to the C. lusitaniae E11ura3Δ5′ strain than the allele bearing the sole K151/143E mutation. Finally, we introduced the K143E mutation in the CaERG11wt allele and showed that it conferred a 128-fold increase of fluconazole MIC (from 0.0625 to 8 μg/ml) when expressed in C. lusitaniae E11ura3Δ5′. We verified that wild-type and mutated ERG11 alleles were expressed at a similar level in C. lusitaniae by reverse transcription-quantitative PCR (Fig. 5A, B, and C).

DISCUSSION

Allelic variability and missense mutations of ERG11 constitute one of the molecular mechanisms by which fungi can resist the cellular action of azoles, i.e., by decreasing the binding affinity of the enzyme to the antifungal drugs (8, 9, 22, 23). In this work, we genetically engineered a strain of C. lusitaniae, a haploid opportunistic pathogenic yeast belonging to the CTG clade of Candida yeasts and susceptible to all antifungals (24), for the homologous and heterologous expression of different wild-type and mutated ERG11 alleles. As the growth of an erg11 null mutant of C. lusitaniae is strongly impaired, the recipient strain used in our experiments harbors its own resident ERG11 allele until it is replaced by the incoming new ERG11 allele. Both the selection of yeast transformants and the expected targeting of the new ERG11 allele are achieved by reconstituting an entire functional metabolic gene, URA3 here, in the 3′ intergenic region of ERG11. Owing to this strategy, ERG11 allelic replacement occurred in 100% of the prototroph Ura⁺ yeast transformants. This efficiency was expected because events leading to nonhomologous recombination cannot be selected, as the genetic marker used in transformation experiments is nonfunctional until it recombines with its complementary part integrated near ERG11. The system ensures that any introduced ERG11 allele is expressed from the same wild-type resident C. lusitaniae ERG11 promoter.

The C. lusitaniae E11ura3Δ5′ strain thus obtained was first challenged by the successful heterologous expression of wild-type ERG11 alleles of C. albicans SC5314. This clinical isolate is susceptible to fluconazole but exhibits a strong trailing growth in the presence of fluconazole concentrations above the MIC. Assembly 22 of the C. albicans SC5314 genome sequence now makes it possible to distinguish the nucleotide sequence of both alleles at any particular genetic locus (25). In the SC5314 strain, the A and B alleles of ERG11 display 10 SNP, of which two are nonsynonymous (T348A and A383C), resulting in D116E and K128T amino acid changes. As already reported in other isolates (11, 26, 27), these changes were not associated with azole resistance. When expressed individually in C. lusitaniae E11ura3Δ5′, both CaERG11 alleles conferred full susceptibility to fluconazole with the same MIC as that observed with the SC5314 strain (0.0625 μg/ml). Surprisingly, this fluconazole MIC was about 10 times lower than that conferred by the allele of C. lusitaniae (MIC of 0.5 μg/ml), suggesting that the CaErg11 protein is intrinsically more susceptible to fluconazole inhibition. Hence, C. lusitaniae can be considered a performing expression system for the genotype-phenotype characterization of ERG11 mutations, as fluconazole MIC is only dependent on the Erg11 protein expressed and not on the cellular environment in which it is expressed. Trailing growth, interpreted as tolerance to fluconazole, was not observed in C. lusitaniae E11ura3Δ5′ expressing CaERG11. This was not surprising as, in our experience, trailing growth in the presence of fluconazole was not observed with C. lusitaniae wild-type and clinical isolates. This shows that tolerance is not dependent on Erg11 protein but rather on other factors, such as pH, Hsp90 acetylation, or Vps21-dependent membrane trafficking, as previously reported for C. albicans (for a review, see reference 28).

Two amino acid substitutions, Y132H and G464S, then were introduced in the C. albicans ERG11 allele and expressed in C. lusitaniae E11ura3Δ5′. Both mutations were previously shown to be responsible for a 4-fold increase in fluconazole MIC by heterologous expression of C. albicans ERG11 alleles in S. cerevisiae (11). In our experiments, only the G464S substitution was confirmed to confer a 4-fold fluconazole MIC increase (from 0.0625 to 0.25 μg/ml). The Y132H substitution had no effect, and the lack of synergy between the Y132H and G464S mutation contrasts with a previous work (11). However, the expression systems cannot be compared: one used a pdr5 mutant of S. cerevisiae as the host and a 2μ multicopy plasmid to express a heterologous ERG11 allele under a galactose-inducible promoter (11), and the other, in this study, used a fluconazole-susceptible Candida species where the heterologous ERG11 allele is introduced as a single copy expressed from the C. lusitaniae ERG11 endogenous promoter. Surprisingly, in spite of using the same clonal genetic background of strain E11ura3Δ5′, RT-qPCR assays showed variations in amounts of ERG11 mRNA detected, possibly due to different mRNA half-lives within the cell or to slight variations of UPC2 expression, the transcription factor that regulates ERG11 (29). However, the variations detected, less than 3-fold, are not expected to have a significant effect on fluconazole MIC or to be correlated to a specific resistance pattern, as previously reported in several studies with C. albicans (30, 31, 32, 33). Furthermore, both the Y132H and G464S CaERG11 alleles were expressed at the same level in C. lusitaniae E11ura3Δ5′, indicating that the lack of phenotype with Y132H only was probably not derived from a decreased expression level of CaERG11. Alternatively, heterologous expression may lead to posttranslational modifications, enhanced degradation, or protein sequestration that can account for the lack of phenotype. However, when introduced in the C. lusitaniae ERG11 allele, the mutations Y132H and G460S (equivalent to C. albicans G464S) had no effect on the fluconazole MIC, showing that the impact of a mutation on resistance cannot necessarily be extrapolated from one species to another. Most variations in susceptibility to azole antifungals are due to conformational changes of the protein, which themselves depend on specific interactions between several key amino acids that can differ from one species to another.

We then explored whether C. lusitaniae could be used to express ERG11 alleles from non-CTG clade pathogenic yeasts, i.e., from the Saccharomycetaceae. We previously described a Kluyveromyces marxianus clinical isolate cross-resistant to all azole antifungals (17). The ERG11 allele of this isolate harbored two nonsynonymous mutations leading to E123Q and K151E substitutions. Expression in C. lusitaniae E11ura3Δ5′ of site-directed mutagenized ERG11 alleles of K. marxianus, C. lusitaniae, and C. albicans provided a new demonstration that substitution at the K151 (Saccharomycetaceae)/K143 (CTG clade) position of Erg11p confers resistance to fluconazole (MIC increased by factors of 32, 16, and 128, respectively), as previously reported in K143R mutants of C. albicans (23) and C. tropicalis (34). Recently, crystal structures of the S. cerevisiae (35) and C. albicans (36) sterol C14α-demethylases have been resolved. The ionized lysine residue at position 151/143, located in the αC region of Erg11, forms an H-bond with the heme ring D of the catalytic pocket (35, 36). Alteration of this bond could be sufficient to hamper the fixation of azole antifungals to the sixth coordination site of iron. Interestingly, we show here that anionic lysine can be replaced by cationic glutamic acid and not only by anionic arginine, as previously reported (23, 34), to confer high resistance levels. In contrast, the E123Q substitution (E115Q for the CTG clade) had no role in resistance, although a synergistic effect (2× MIC) was always observed when combined with K143/151E. Surprisingly, expression of the K. marxianus allele without K151E substitution conferred hypersusceptibility to C. lusitaniae to 0.006 μg/ml, while fluconazole MICs reported for the clinical isolates of C. lusitaniae generally vary from 0.19 to 4 μg/ml (24).

Besides modifying susceptibility to azole antifungals, the impact of ERG11 mutations on the biology of the fungal cell remains poorly investigated. At the Erg11p level, some amino acid substitutions decrease the binding affinity of the enzyme to fluconazole (37), while others tend to increase binding affinity to lanosterol (38). In some cases, mutations can decrease the activity of fluconazole-resistant Erg11p, which then could contribute to slight accumulation of 14-methyl fecosterol in the yeast cell, as observed in some Y132F variants of C. parapsilosis (39). However, no correlation could be drawn between a specific sterol pattern, azole susceptibility, ERG11 variability, and ERG11 expression level in a collection of 39 fluconazole-susceptible and -resistant isolates of C. parapsilosis (39). On the other hand, considerable sterol variations were described in a C. glabrata clinical isolate harboring a G315D substitution in Erg11p. This mutation nullified the function of Erg11p and allowed viability of the cells in spite of massive accumulation of 14α-methylated sterols and complete lack of ergosterol, with concomitant acquisition of amphotericin B resistance (40). This type of mutation probably cannot be expressed in C. lusitaniae, since a null erg11 mutant is unviable. In the present work, the MIC of amphotericin B for strain E11ura3Δ5′ expressing either its own allele or different ERG11 alleles is the same (1 μg/ml; data not shown), suggesting few variations, if any, in ergosterol content.

We demonstrated in this study that C. lusitaniae can be successfully used as a heterologous expression system to assess the impact of different mutations of the ERG11 gene on fluconazole susceptibility. We showed that some mutations modulating susceptibility to fluconazole are species specific, while the K143/151E mutation has ubiquitous effects conferring high levels of resistance to fluconazole when present in ERG11 alleles of different yeasts species. The expression system we have developed is close to the clinical context and MIC determination outcomes: C. lusitaniae is pathogenic and susceptible to antifungals, including azoles, and allele replacement is performed at the chromosomal level as a single copy expressed from the endogenous UPC2-regulated promoter (unpublished data) of ERG11. In this study, the versatility of C. lusitaniae as an expression system was tested with ERG11 alleles of different yeast species, i.e., C. albicans, from the CTG clade, and K. marxianus, an emerging pathogen of the Saccharomycetaceae family. We believe that our biological system could facilitate structure-function studies, since the crystal structure of a fungal Erg11p has been very recently reported for S. cerevisiae (35) and C. albicans (36). This opens the way to the systematic functional characterization of the interactions between important amino acids of the protein and different substrates of the enzyme, i.e., sterols and cofactors, and antagonist molecules such as azoles.

MATERIALS AND METHODS

Yeast strains and culture conditions.

The Candida lusitaniae wild-type strain 6936 (Centraalbureau voor Schimmelcultures, Utrecht, The Netherlands) was used for gene cloning and as a susceptible reference strain for antifungal agent susceptibility tests. The auxotrophic mutant 6936 ura3Δ (41), bearing a deletion of 990 bp in the orotidine 5′-phosphate decarboxylase gene, was used for transformation experiments. Candida albicans SC5314 (ATCC MYA-2876) and the Kluyveromyces marxianus (anamorph Candida kefyr) clinical isolate PAZ (17) were used as ERG11 donor strains for allelic replacement (Table 2).

TABLE 2.

Origin and susceptibility to fluconazole of the strains used in this study

Strain	Origin	Genotype	FLC MICa (μg/ml)	S/Rb	Reference or source
C. lusitaniae CBS 6936	Citrus peel	Wild type	0.5	S	20
C. lusitaniae 6936 ura3Δ	CBS 6936	ura3Δ	0.5	S	41
C. lusitaniae E11ura3Δ5′	6936 ura3Δ	ura3Δ	0.5	S	This study
C. albicans SC5314	Clinical	Wild type	0.0625	S	46
K. marxianus PAZ	Clinical	Wild type	>256	R	17

^aMIC determined by the microplate dilution CLSI method.

^bS, susceptible (MIC of ≤2 μg/ml); R, resistant (MIC of >4 μg/ml).

Yeasts were cultivated in liquid YPD medium (1% yeast extract, 2% peptone, 2% glucose) at 35°C under agitation (250 rpm). In transformation experiments, selection was achieved on synthetic YNBS medium (0.67% yeast nitrogen base without amino acids [Difco Laboratories], 2% glucose, 1 M sorbitol) and, when needed, 25 μg/ml uracil and 0.8 mg/ml 5-fluorootic acid (5FOA). For antifungal susceptibility tests, we used RPMI 1640 (Sigma). Solid media were obtained with 2% agar (Sigma).

Antifungal agents and susceptibility tests.

Stock solutions of 2 mg/ml fluconazole (ICN Biomedicals Inc.) and 100 mg/ml 5-fluoroorotic acid were prepared in sterile distilled water and dimethyl sulfoxide (DMSO), respectively. Microdilution assays were performed according to CLSI standards (42) in RPMI 1640, pH 7.0, buffered with 0.165 M 4-morpholinepropanesulfonic acid (MOPS). Yeast suspensions were diluted with RPMI 1640 medium at a final cell suspension of 10³ cells/ml. The 96-well plates were incubated for 48 h at 35°C, and growth was measured with an automated microtiter reader at 450 nm. All experiments were done at least in triplicate, and growth variations did not exceed 10%. For each ERG11 allele expressed in C. lusitaniae, fluconazole MIC was determined for seven prototroph clones randomly selected after transformation. Fluconazole susceptibility was also determined using Etest (bioMérieux, France) with concentration ranges from 0.016 to 256 μg/ml. Etest strips were deposited onto solid RPMI medium inoculated with the yeast strains at the cellular density recommended by the supplier.

DNA extraction, PCR amplifications, and nucleotide sequencing.

Genomic DNA was extracted by a glass bead method for yeast cell disruption as previously described (41). The high-fidelity DNA polymerase Pfu Turbo (Stratagene) was used to amplify DNA fragments by PCR for cloning steps and for overlapping PCR. Routine PCRs were performed with Hot-Start Taq DNA polymerase (Qiagen) as recommended by the supplier. All primers used in this study (see Table S1 in supplemental material) were synthesized by Eurofins MWG Operon (Europe). When required, PCR products were sequenced using the BigDye Terminator v3.1 kit (Applied Biosystems) by Eurofins MWG Operon.

Molecular constructions.

For the DNA cassette constructions, the different DNA fragments to be assembled were generated by PCR using primers having 15-bp overlap homology at their ends. The DNA fragments then were recombined in vitro and cloned in a pUC19 plasmid vector using the In-Fusion HD cloning kit (Clontech). Recombinant plasmid DNA was selected by transforming the Stellar competent Escherichia coli cells provided with the kit and plating on LB solid medium supplemented with 50 μg/ml ampicillin according to the supplier's recommendations. DNA cassettes were recovered from the recombinant plasmids by PCR amplification with the relevant primers, purified through QIAquick columns (Qiagen), and used for yeast transformation. All steps of the molecular constructions performed in this work are detailed in Fig. S1 in the supplemental material in connection to the oligonucleotide information presented in Table S1.

Yeast transformations.

Auxotrophic strains were transformed by an electroporation procedure previously described (43), slightly modified by adding 1 M sorbitol to the lithium acetate buffer, using 1 to 2 μg of transforming DNA per experiment. Prototrophic transformants were selected on YNBS selective medium after 3 days of incubation at 35°C.

Southern blotting.

Approximately 10 μg of C. lusitaniae DNA was digested with the appropriate restriction enzyme, separated by electrophoresis in a 1% agarose gel, and transferred onto nylon membranes (Hybond N+; Roche Molecular Biochemicals). Hybridization was carried out with digoxigenin (DIG)-labeled probes synthesized with a PCR DIG probe synthesis kit (Roche Molecular Biochemicals), as specified by the supplier.

RT-qPCR.

Total RNA was extracted with a guanidinium thiocyanate-phenol-chloroform extraction procedure (44) using TRI Reagent (Molecular Research Center, Inc.) as described by the manufacturer. Briefly, 10⁷ cells were treated with 1 ml TRI Reagent, transferred to a screw-cap 1.8-ml tube containing 450 μl of 0.5-mm acid-washed glass beads, and disrupted with 2 cycles of 40 s at 5 m/s (6,800 rpm) in a Precellys 24 tissue homogenizer at 4°C. The recovered aqueous phase was extracted successively with 5:1 (vol/vol) acidic phenol-chloroform, pH 5.1, and 24:1 (vol/vol) chloroform-isoamyl alcohol and precipitated with isopropanol. Total RNA was solubilized in RNase-free water and quantified with an Agilent RNA 6000 Nano chip.

Biological triplicates were prepared by growing yeast strains to mid-exponential growth phase (1.5 × 10⁷ cells/ml) in 5 ml of RPMI medium. The expression of ERG11 was determined by RT-qPCR in a one-step real-time PCR system (GoTaq 1-Step RT-qPCR; Promega) and the CFX96 real-time PCR detection system (Bio-Rad Laboratories). Fold changes in gene expression relative to that of control wild-type strain 6936 were determined from ACT1-normalized expression levels by the 2^−ΔΔCT method. The primers used for this assay are described in the supplemental material (Table S1).

Databases and software used in this study.

The amino acid sequences of the lanosterol C14α-demethylase proteins of different fungi were retrieved from the National Center for Biotechnology Information (NCBI), GenBank, and UniProt. The names of species, strains, and accession numbers are given in Table S2 in supplemental material. Multiple-sequence alignment and molecular phylogeny were performed using the programs embedded in Seaview v4.5.3 (45). Alignments were carried out with Clustal Omega, and phylogenetic trees were computed using the PhyML v3.1 algorithm (maximum likelihood) with a bootstrap repetition of 100. Fluorescence data from RT-qPCR were collected and analyzed, including statistical tests, with CFX Manager software, version 3.1 (Bio-Rad).