Positions and Numbers of FKS Mutations in Candida albicans Selectively Influence In Vitro and In Vivo Susceptibilities to Echinocandin Treatment

Abstract

Candidemia is the fourth most common kind of microbial bloodstream infection, with Candida albicans being the most common causative species. Echinocandins are employed as the first-line treatment for invasive candidiasis until the fungal species is determined and confirmed by clinical diagnosis. Echinocandins block the FKS glucan synthases responsible for embedding β-(1,3)-d-glucan in the cell wall. The increasing use of these drugs has led to the emergence of antifungal resistance, and elevated MICs have been associated with single-residue substitutions in specific hot spot regions of FKS1 and FKS2. Here, we show for the first time the caspofungin-mediated in vivo selection of a double mutation within one allele of the FKS1 hot spot 1 in a clinical isolate. We created a set of isogenic mutants and used a hematogenous murine model to evaluate the in vivo outcomes of echinocandin treatment. Heterozygous and homozygous double mutations significantly enhance the in vivo resistance of C. albicans compared with the resistance seen with heterozygous single mutations. The various FKS1 hot spot mutations differ in the degree of their MIC increase, substance-dependent in vivo response, and impact on virulence. Our results demonstrate that echinocandin EUCAST breakpoint definitions correlate with the in vivo response when a standard dosing regimen is used but cannot predict the in vivo response after a dose escalation. Moreover, patients colonized by a C. albicans strain with multiple mutations in FKS1 have a higher risk for therapeutic failure.

INTRODUCTION

Fungal infections have emerged over the last few decades as a consequence of increasing cohorts of at-risk individuals. Candida albicans is the most important opportunistic fungal pathogen in humans (1). Mortality rates from C. albicans infection are estimated to be as high as 45% (2), partly due to delayed or inaccurate diagnosis and inappropriate antifungal therapies.

Echinocandin drugs are recommended as the first-line treatment for invasive candidemia (3, 4). Anidulafungin (ANI), caspofungin (CAS), and micafungin (MICA) are lipopeptidic antifungal agents that inhibit the synthesis of the fungal wall component β-(1,3)-d-glucan by noncompetitively blocking the β-(1,3)-d-glucan synthase. Drug resistance is emerging in patients with C. albicans, C. glabrata, C. tropicalis, and C. krusei infections (5–7), which complicates disease monitoring and patient management (8). Presently, echinocandin resistance in C. albicans occurs with an incidence of <1% (9). Single-residue substitutions located in two hot spot regions in each of the genes FKS1 and FKS2 (10), but not FKS3, have been associated with elevated echinocandin MICs (11) in Candida species.

So far in C. albicans, only amino acid substitutions within FKS1 encoding the β-(1,3)-d-glucan synthase complex were found in echinocandin-resistant isolates (12). The most frequent amino acid changes reported were at positions F641, L644, S645, R647, D648, P649 (10), W1358, and R1361 (13) within hot spot 1 (HS1) (amino acid positions 641 to 649, FLTLSLRDP) (12) and hot spot 2 (HS2) (amino acid positions 1357 to 1364, DWIRRYTL) (14), respectively. HS1 and HS2 alterations leading to changes in F641, S645, and R1361 were associated with pronounced MIC elevations. Other HS mutations (in L644, R647, D648, P649, and W1358) confer a discrete MIC increase (14). C. albicans is a diploid fungus carrying a pair of each chromosome and at least two alleles of each gene locus (15). An amino acid substitution in one allele (heterozygous) affects only half of the total cellular glucan synthase pool that can be targeted by echinocandins (5), while substitutions occurring in both alleles (homozygous) affect the whole glucan synthase pool.

Here, we identify and report the first in vivo-selected HS1 FKS1 double mutation (R647R/G and P649P/L) in one allele in C. albicans after long-term caspofungin (CAS) therapy (including a detailed case report and therapeutic regimes). The treatment response of a clinical wild-type (WT) strain and the corresponding sequential heterozygous derivative mutants (R647R/G and P649P/L) are compared using a well-validated murine model of disseminated candidiasis. ANI, CAS, or MICA was applied at a standard dose (designated as AUC₁₀₀, where AUC is the area under the concentration-time curve) or elevated dose (AUC₅₀₀ [five times the standard dose]) and compared with each other and the placebo control group.

The impacts of heterozygous single (R647R/G or P649P/L) and double mutations (R647R/G and P649P/L), as well as homozygous double mutations (R647G and P649L), in FKS1 on virulence and in vitro and in vivo susceptibilities were tested in isogenic (identical genetic background) mutant strains.

The hypothesis addressed in this study was that heterozygous double mutations, as well as homozygous double mutations within one allele, enhance the in vivo therapeutic resistance of C. albicans strains even with a dose escalation. We were able to (i) confirm this hypothesis, (ii) show significant differences in the activities of the three echinocandins, and (iii) demonstrate that the reported single nucleotide polymorphisms (SNPs) have no impact on virulence.

MATERIALS AND METHODS

Patient case.

A 27-year-old female patient was reported to have chronic mucocutaneous candidiasis (CMC), suffering from persistent and recurrent infections of the skin, nails, and oral mucosa. In 2005, a C. albicans (azole-resistant isolate) infection was successfully treated with 12 months of CAS (50 mg/day) (16). In 2007, the patient presented with a relapse of CMC. Caspofungin (50 mg/day) was resumed in combination with terbinafine (250 mg/day) and topical ciclopirox cream for another 12 months, resulting in a marked improvement.

During 2007 to 2010, various azole regimens consisting of voriconazole (VOR) (2 × 200 mg/day) or posaconazole (POS) (2 × 800 mg/day) were attempted, with partial effects. In March 2010, the patient began retreatment with intravenous (i.v.) CAS (50 mg/day) for 2 years. Onychomycosis improved slightly (Fig. 1C). In 2012, while still on CAS, a fulminant infection appeared with perioral/periorbital erythematous and erosive lesions accompanied by plaques of pustules at the forehead that were clinically significant (Fig. 1A), in addition to nail-destructive onychomycosis (Fig. 1B). In vitro susceptibility testing according to Etest and EUCAST classified C. albicans as CAS resistant (Table 1). Antifungal treatment was immediately changed to ANI (100 mg/day), and the infections improved significantly, with full healing achieved by October 2012.

FIG 1.

Presentation of the clinical features of the female 27-year-old patient. (A) At the right side of the forehead and the lips, erythematous plaques topped with crusts and pustules are visible. (B) Paronychia and destruction of the nail by C. albicans. (C) Periodic acid-Schiff stain showing parakeratosis with multiple hyphae (arrow) present in the cornified layers.

TABLE 1.

In vitro susceptibilities for clinical and mutant isolates included in the study and for quality-control strains, as determined by EUCAST or Etest^a

Isolate type and mutant or strain	EUCAST MIC (mg/liter) (susceptibility classification) forb:					Etest MIC (mg/liter) (susceptibility classification) forc:
	ANI	MICA	VOR	ISA	FLU	CAS	AMB
Clinical isolates or QCd strain
SCL 1130.04e	<0.008 (S)	<0.008 (S)	0.5 (R)	0.06	>16 (R)	0.125 (S)	0.5 (S)
RRCL 110.1f	0.03 (S)	>1 (R)	4 (R)	0.5	>16 (R)	2 (R)	0.5 (S)
ATCC 22019g	0.125 (I)	>1 (I)	<0.03 (S)	<0.03	2 (S)	0.5 (S)	0.38 (S)
Mutant isolates or QC strain
SSWT SC5314h	0.015 (S)	<0.008 (S)	<0.03 (S)	<0.03	0.25 (S)	0.19 (S)	0.25 (S)
RRMH2h	0.06 (R)	0.03 (R)i	<0.03 (S)	<0.03	0.25 (S)	0.75 (R)	0.19 (S)
RSMH1h	0.015 (S)	0.03 (R)j	<0.03 (S)	<0.03	<0.125 (S)	0.25 (S)	0.25 (S)
SRMH1h	0.015 (S)	0.008 (S)	<0.03 (S)	<0.03	0.25 (S)	0.25 (S)	0.25 (S)
RRMHO2h	0.06 (R)	1 (R)k	<0.03 (S)	<0.03	<0.125 (S)	1 (R)	0.38 (S)
ATCC 6258g	0.06 (S)	0.125	0.125 (S)	<0.03	16 (R)	0.38 (I)	0.75 (S)

^aSusceptibility classifications were performed according to the EUCAST breakpoints (BP), except for CAS and Etest, for which CLSI BP were adopted, as recommended by the manufacturer.

^bANI, anidulafungin; MICA, micafungin; VOR, voriconazole; ISA, isavuconazole; FLU, fluconazole; S, susceptible; R, resistant; I, intermediate.

^cCAS, caspofungin; AMB, amphotericin B.

^dQC, quality control.

^eThe clinical isolates 952.04 and 5104.04 revealed the same MICs within ±1 dilution step difference; these strains do not carry coding FKS1 HS1 or HS2 mutations.

^fThe clinical isolate 111.12 revealed the same MIC within ±1 dilution step difference; these strains carry the mutations R647R/G and P649P/L.

^gQuality-control strains.

^hSC5314 is the wild-type parental strain; mutations carried by the strains are as follows: RR_MH2, R647R/G and P649P/L; RS_MH1, R647R/G; SR_MH1, P649P/L; and RR_MHO2, R647G and P649L.

ⁱMicafungin trailing 0.03 to 1 but stays at <50% endpoint in this range.

^jMicafungin creeping increase 0.03 to 0.25.

^kMicafungin trailing 0.06 to 0.25 and just >50% endpoint in this range.

Random amplified polymorphic DNA typing of clinical isolates.

Random amplified polymorphic DNA (RAPD) typing of the clinical isolates (n = 5) was performed using the published primer pairs M13, OPA18, OPE18, CA2, RSD10, and RSD12 together with already published PCR conditions (17–20). For DNA extractions, all strains were cultured on Sabouraud 2% glucose agar for 24 h at 37°C. DNA was extracted from pure cultures using the UltraClean microbial DNA isolation kit (Mo Bio Laboratories, Inc.) according to manufacturer's instructions. The DNA concentration of the DNA extract was determined using NanoVue (GE Healthcare Life Sciences), according to the manufacturer's instructions. Genomic DNA extracts were diluted to 50 ng/μl for PCR. A volume of 18 μl of all PCR products was loaded together with 3 μl of orange (6×) gel loading dye (New England BioLabs, Inc.) on 1.8% agar (Rotigarose; Roth) gel with ethidium bromide at a concentration of 0.5 μg/ml. As a reference, the SimplyLoad 100-bp extended range DNA ladder (Lonza Cologne GmbH) was used. The electrophoresis conditions were 3 h at 80 V and 100 mA. The electrophoresis gels were visualized in a GelDoc EZ system (Bio-Rad). The typing assay was performed twice and double-blinded by two different investigators. The C. albicans strains ATCC 90028, ATCC 64550, and ATCC 64548 were used as population controls.

In vitro susceptibility testing.

The MICs for azoles (voriconazole, isavuconazole, and fluconazole) and echinocandins (ANI and MICA) were tested for the echinocandin-susceptible strains isolated in 2004 (C. albicans S_CL 1130.04, S_CL 952.04, S_CL 5104.04, and S_CL 111.12) and the echinocandin-resistant (RR_CL 110.12) strain carrying the mutations R647R/G and P649P/L, the parental strain, C. albicans SS_WT SC5314, the isogenic mutants, and the mutants C. albicans RR_MH2 (R647R/G and P649P/L), RS_MH1 (R647R/G), SR_MH1 (P649P/L), and RR_MHO2 (R647G and P649L), according to EUCAST methodology (EDEF 7.2) (21) (see Table 1). In addition, CAS and amphotericin B were tested using CLSI methodology and Etest (bioMérieux). C. krusei strain ATCC 6258 or Candida parapsilosis strain ATCC 22019 (21, 22) was used as a quality-control strain.

Sequencing of FKS1 gene regions hot spot 1 and hot spot 2.

Genomic DNA was extracted using the UltraClean microbial DNA isolation kit (Mo Bio Laboratories, Inc.), according to the manufacturer's instructions. Sequencing was performed using the same primers and conditions as those published by Garcia-Effron et al. (23). In short, the Candida universal FKS1 primers 1HS1F (AAT GGG CTG GTG CTC AAC AT) and 1HS1F (CCT TCA ATT TCA GAT GGA ACT TGA TG) were used for amplifying FKS1 HS1, while for FKS1 HS2, the primers 1HS2F (AAG ATT GGT GCT GGT ATG GG) and 1HS2R (TAA TGG TGC TTG CCA ATG AG) were used. As a PCR master mix, the Qiagen LongRange PCR kit and conditions were as follows: initial denaturation at 96°C for 4 min, followed by 40 cycles of 94°C for 30 s, 55° for 30 s, 74°C for 90 s, and a final elongation step of 74°C for 4 min. For cleaning the PCR products, ExoSAP-IT was used. For sequencing, the BigDye Terminator version 3.1 cycle sequencing kit was used in combination with the 3500 Genetic Analyzer (Applied Biosystems), according to the manufacturer's instructions.

Construction of isogenic C. albicans FKS1 mutants.

The generated isogenic mutant plasmids used are shown in Table S1 in the supplemental material, and the primers used are shown in Table S2 in the supplemental material. An FKS1 deletion cassette was constructed using the SAT1 flipper technology (24) with the primers FKS1_55 and FKS1KO_53 for the upstream homologous region and FKS1_35 and FKS1_33 for the downstream homologous region. The homologous regions were cloned into pSFS2a (24) using the restriction sites ApaI/XhoI and SacII/SacI, respectively. The resulting plasmid was linearized with PvuI and transformed into C. albicans SC5314 (25) to replace the FKS1 coding sequence and get a heterozygous FKS1/fks1 knockout (see Table S3 in the supplemental material). The SAT1 marker was recycled by overnight cultivation in maltose-containing medium.

For reintegration of the FKS1 gene at its endogenous locus, the FKS1 coding sequence was cloned into pSFS2a-FKS1dr (pSFS2a containing the downstream homologous region of the deletion cassette) with the primers FKS1_55 and FKS1clo_53 and using the restriction sites ApaI/XhoI. The resulting plasmid (pSFS2a-FKS1reint) was linearized with PvuI and transformed into C. albicans strain CA-MT424 (FKS1/fks1). Correct integration was checked by colony PCR.

For the mutagenesis of FKS1, PCRs were performed using pSFS2a-FKS1reint with the primers FKS1_55 and the corresponding mutagenesis reverse primer, as well as with the corresponding mutagenesis forward primer and FKS1_33, respectively (see Table S3 in the supplemental material). The purified PCR products were mixed, and the whole reintegration cassette was amplified using the primers FKS1_55_2 and FKS1_33_2. The cassette was transformed into CA-MT424 to integrate the mutated FKS1 allele at the deleted locus. Correct integration was checked by colony PCR, and the SAT1 marker was recycled.

For the construction of homozygous mutants, the PCRs were performed with primers FKS1_55 and the corresponding mutagenesis reverse primer, as well as with the corresponding mutagenesis forward primer and FKS1clo_53, respectively. The purified PCR products were mixed, the mutated FKS1 coding sequence was amplified using the primers FKS1_55 and FKS1clo_53; this was cloned into pSFS2a-FKS1dr as described above for the reintegration plasmid. Prior to transformation, the plasmids were linearized with PvuI. For the construction of the homozygous double mutant, pSFS2a-FKS1_2mut_reint was used to replace the wild-type allele in C. albicans strain CA-MT441 (see Table S2 in the supplemental material). Correct integration was checked by colony PCR, and the SAT1 marker was recycled. For the construction of the homozygous single mutants, the corresponding plasmids containing mutated FKS1 alleles were used for integration at the deleted alleles of C. albicans strains CA-MT457 and CA-MT458, respectively. In these two strains, the wild-type allele in CA-MT424 had been directly replaced before by the corresponding mutated allele. The presence of the introduced point mutations was confirmed by sequencing the corresponding region in all mutants.

Hematogenous mouse model. (i) Pilot study for inoculum determination.

To evaluate the in vivo virulence of isogenic mutant strains, a total of 24 NRMI mice were challenged in groups of three with four isogenic mutant strains (C. albicans SS_MH2, RS_MH1, SR_MH1, and SS_MHO2; detailed information is shown in Table 1; see also Table S1 in the supplemental material) and two inoculum amounts (5 × 10⁵ CFU/mouse and 5 × 10⁶/mouse). The aim was to achieve sufficient infection without causing mortality. Using a C. albicans cell concentration for infection of 5 × 10⁶/mouse, nine out of 12 mice died. Challenging the mice with a dose of 5 × 10⁵ CFU/mouse led to one out of three mice per group dying, and the kidney burden in the surviving mice was found to be between 10⁴ and 10⁵ CFU/ml kidney homogenate. Hence, an inoculum of 1 × 10⁵/mouse was used for the treatment studies.

(ii) In vivo treatment studies.

Evaluation of in vivo resistances of C. albicans clinical isolates (S_CL and RR_CL strains; Tables 1 and 2) against ANI, CAS, MICA, and the placebo (0.09% NaCl) was performed in a hematogenous mouse model. Eighty-four NMRI mice (weight, 26.0 g to 30.0 g; Harlan Scandinavia, Allerød, Denmark) were kept with free access to food and water. On day 0, the mice were challenged by intravenous injection (200 μl administered with a 25-gauge syringe) of a C. albicans (5 × 10⁵ CFU/ml; final concentration per mouse, 1 × 10⁵) suspension. The mice were challenged with either of the two strains S_CL (n = 42 mice) or RR_CL (n = 42 mice) and treated by the intraperitoneal (i.p.) route on days 1 to 3 with 0.5 ml of either ANI, CAS, or MICA at standard (AUC₁₀₀) and high-dose (AUC₅₀₀) or with a placebo (see below).

TABLE 2.

Overview of FKS mutations found in clinical isolates gained between 2004 (x.04) and 2012 (x.012) from the index/case patient

Isolate	FKS1 hot spot 1 mutation by positiona:						FKS1 hot spot 2 mutation by positiona:
	1641 (A)	1653 (A)	1662 (T)	1929 (A)	1939 (A)	1946 (C)	4215 (C)	4230 (T)
SCL 1130.04	A/T	A/G	T/C	T	A	C	C/T	T/C
952.04	A/T	A/G	T/C	T	A	C	C/T	T/C
5104.04	A/T	A/G	T/C	T	A	C	C/T	T/C
RRCL 110.12	A/T	A/G	T/C	T	A/G	C/T	C/T	T/C
111.12	A/T	A/G	T/C	T	A/G	C/T	C/T	T/C
aab	P547	T551	I554	T643	R647R/Gc	P649P/Ld	I1375	A1410

^aNucleic acid mutation. Boxheads indicate position (wild-type nucleic acid).

^bRegular type, amino acid change: nucleic acid change does not result in a change of amino acid (regular type); bold type, wild-type amino acid/position/heterozygous mutation.

^cAmino acid change, arginine (R) to glycine (G); codon change, AGA to GGA.

^dAmino acid change, proline (P) to leucine (L); codon change, CCU to CUU.

The treatment studies of isogenic mutant strains and their parental strain were divided into three animal experiments (per antifungal compound tested), as a total of 312 NMRI mice (weight between 26.0 g and 30.0 g) were used. The inoculation of mice and treatment scheme were performed as outlined above, and the dosing was as described below. The treatment groups consisted of six mice and control groups of eight mice. As a backup, four additional mice were available to replace mice that were inoculated outside the tail vein and therefore excluded; if the backup mice were not needed, they were allocated to the control groups (hence, some control groups included nine mice).

The echinocandin treatment doses used in the murine model were the standard dose (AUC₁₀₀), which was the dose that resulted in human equivalent exposure and was calculated using our echinocandin pharmacokinetic-pharmacodynamic studies in this exact mouse model reported previously (26), ANI₁₀₀ (2.16 mg/kg of body weight), CAS₁₀₀ (0.76 mg/kg), and MICA₁₀₀ (2.71 mg/kg). The high dose (AUC₅₀₀) was five times the human-equivalent exposure dose and was included in order to examine if a dose escalation would be useful for overcoming the resistance of C. albicans.

For determining the fungal burden (CFU/ml tissue homogenate), kidneys were aseptically removed at day 4. The organ weights were determined, and the kidneys were placed in sterile physiological saline (750 μl per pair of kidneys); the organs were stored at −80°C prior to counting CFU. The fungal burden was determined using the spot technique, which involved plating two 20-μl spots of 10-fold dilutions of tissue homogenate (homogenization performed with a homogenizer [RW 16 Basic; IKA Labortechnik, Bie & Berntsen, Copenhagen, Denmark]). The CFU count was expressed as the log₁₀ of CFU/ml kidney homogenate. The lower limit of detectable CFU was 25 CFU/ml tissue homogenate. All murine experiments reported in this study were approved by the Danish Animal Experimentation Committee under the Ministry of Justice (2009/561-1637).

Statistical analyses.

To check for normal distribution of the fungal burden data, the Shapiro-Wilk normality test was performed. The Kruskal-Wallis test was then applied, as the data were found to be not normally distributed. Outliers (maximum one per control group with <100 CFU/ml kidney homogenate) in the control groups (each eight placebo-treated mice) were removed for statistical significance calculation, as these mice did not successfully establish a manifest C. albicans infection or clear the infection. The outliers were identified by Prism GraphPad version 5. To guarantee the transparency of all data and to provide the readers with a complete image, the outliers are included in all tables and figures. In all analyses, P values of ≤0.05 were regarded as statistically significant.

Filamentation ability tests.

All studied C. albicans strains were grown overnight on Sabouraud 2% glucose agar (Carl-Roth) at 37°C. The cells were harvested in 0.9% NaCl. The cell number was adjusted to 5 × 10⁵ cells/ml using the Neubauer CE chamber hemocytometer. A volume of 500 μl was centrifuged at 13,000 rpm for 5 min in a Mikro 20 (Hettrich) centrifuge, the supernatant was discarded, and the cell pellet was suspended in 500 μl fetal bovine serum (FBS) (Gibco, Invitrogen). FBS with cells was transferred into a Nikon BioStation IM (Nikon) cell incubator and monitoring system. The cells were evaluated by microscopy after 0.5 h, 1 h, 2 h, and 3 h incubation at 37°C in the Nikon BioStation IM using BioStation software version 2.1. In addition, for a long-time (4 h) comparison of the germ tube formation experiment, movies were captured for SC5314, RR_MH2, S_CL, and RR_CL in four individual experiments per strain. The images were captured every 10 min at ×200 and ×400 magnification.

Chitin content measurement by FACS analyses.

All studied C. albicans strains were overnight grown on Sabouraud 2% glucose agar at 37°C. The cells were harvested in 0.9% NaCl. The cell number was adjusted to 5 × 10⁶ cells/ml using the Neubauer CE chamber hemocytometer. The cells were inoculated into fresh Sabouraud 2% glucose agar and were harvested after 4 h of incubation at 37°C with shaking at 200 rpm. The samples were fixed in 10% (vol/vol) neutral-buffered formalin for 15 min. The cells were washed with sterile 0.9% NaCl and stained with 25 μg/ml calcofluor white (Sigma-Aldrich) for 15 min at room temperature. The cells were washed twice with 0.9% NaCl to remove excessive stain, and they were filtered using a CellTrics filter (5 μm pore size; Partec) to remove cell clumps. The cells were prepared using the standard procedure for fluorescence-activated cell sorter (FACS) analysis. The measurement of 10,000 events was carried out using the FACSVerse flow cytometer (BD Biosciences). The results were given as the fluorescence intensity (calcofluor-A). Unstained SS_WT (SC5314) was used as the reference strain for gaiting the cell population. For statistical analyses, the minimum, median, and maximal fluorescence signals were determined. SS_WT stained with calcofluor white served as a wild-type reference. The fluorescence peaks of the clinical isolates (strains S_CL and RR_CL) were compared to the fluorescence peak of the SS_WT population. All FACS experiments were performed independently in triplicate.

RESULTS

In vivo emergence of heterozygous double mutants of C. albicans and CAS medication.

Five clinical isolates obtained between 2004 and 2012 from a female patient suffering from chronic mucocutaneous candidiasis (CMC) (case report available in Materials and Methods) were tested for their in vitro susceptibilities against ANI, CAS, MICA, fluconazole (FLU), isavuconazole, voriconazole (VOR), and amphotericin B. The isolates from 2004 were found to be FLU and VOR resistant but echinocandin susceptible according to the EUCAST breakpoints (using ANI as a marker for ANI and CAS susceptibility, as is currently recommended) and CAS susceptible according to the CLSI breakpoints (22, 27). In contrast, the isolates gained in 2012 were micafungin resistant and showed an elevated ANI MIC compared to that of the initial isolate but not exceeding the EUCAST breakpoint (MIC, 0.03 mg/liter; ANI breakpoint, susceptible [S] ≤ 0.03 mg/liter); they were therefore classified as anidulafungin and caspofungin susceptible using the EUCAST breakpoint, with anidulafungin as a marker for caspofungin. In contrast, these isolates were classified as caspofungin resistant using the Etest and CLSI breakpoints. Detailed data on the MICs are given in Table 1.

To investigate the underlying molecular resistance mechanism, FKS1 HS1 and HS2 were sequenced. Notably, strains isolated in 2004 carried four silent mutations (no amino acid [aa] change, also called synonymous substitution), one within the FKS1 HS1 A1929T (T643) and five outside the FKS1 HS regions, located at nucleic acid positions A1641A/T (P547), A1653A/G (T551), T1662T/C (I554), C4125C/T (I1375), and T4230T/C (A1410). In addition, two missense (i.e., having an aa change, also called nonsynonymous) mutations in the same allele (heterozygous) were found at nucleic acid positions A1939A/G and C1946C/T, leading to the residue changes R647R/G and P649P/L, respectively (Table 2). Identical silent mutations found in FKS1 HS1 and HS2 were shared by all isolates, strongly suggesting their clonal origin and in vivo development of resistance. To verify this hypothesis, randomly amplified polymorphic DNA typing (RAPD typing) was performed with five markers (M13, CA2, OPA18, OPE18, and RSD10). As displayed in Fig. 2, the RAPD genotype was identical for the clinical isolates. The data herein support the development of acquired CAS resistance in C. albicans during a 6-year treatment period with CAS.

FIG 2.

Genotyping of clinical isolates of C. albicans isolated between 2004 and 2012 using RAPD. PCR products for the five markers, M13 (A), CA2 (B), OPA18 (C), OPE18 (D), and RSD10 (E) (17–20) were separated on 1.8% agar gel for a run time of 3 h at 80 V and 100 mA. Electrophoresis gels were stained with ethidium bromide. Lane 0, marker; lane 1, isolate 952/04; lane 2, isolate 1130/04; lane 3, isolate 5104/04; lane 4, isolate 110/12; and lane 5, isolate 111/12.

To evaluate the impact of the double mutations on the therapeutic response in vivo, the in vitro echinocandin-resistant clinical isolate carrying the R647R/G and P649P/L mutations (RR_CL) and the susceptible wild-type clinical isolate (S_CL) were compared in a murine hematogenous candidiasis model. ANI, CAS, and MICA were administered at two doses (AUC₁₀₀ and AUC₅₀₀). The fungal burden in the kidney homogenates of animals infected with the susceptible wild-type strain were significantly reduced (Fig. 3). Comparing the control mouse group with each of the treatment groups, the reduction in kidney burden was found to be highly significant in all treated groups (P < 0.01), with the exception of that for MICA₁₀₀, for which the reduction was less pronounced, albeit still statistically significant (P < 0.05) (see Table S4 in the supplemental material).

FIG 3.

Evaluation of in vivo susceptibilities of the two clinical strains 1130.04 (S_CL, no mutation in FKS1) and 110.12 (RR_CL, heterozygous double mutation in FKS1 at positions R647R/G and P649P/L) and high dose (AUC₅₀₀) or standard dose (AUC₁₀₀) of anidulafungin (ANI), caspofungin (CAS), or micafungin (MICA). The treatment groups consisted of six mice and control groups of eight mice. Infected mice treated with placebo served as a control (C). CFU per ml kidney homogenate is depicted at the end of treatment. The limit of detection (LoD) and median with interquartile range are indicated. Statistical significance was calculated using the Kruskal-Wallis test and Dunnett's multiple comparison test: *, P < 0.5; **, P < 0.01; and ***, P < 0.001. Outliers (maximum of one outlier per control group) were removed for calculation only (CFU ≤ 100 in the placebo-treated groups).

In contrast, the reduction in kidney burden failed to reach statistical significance (between the different treatment groups and between the placebo with the treatment groups) in mice infected with RR_CL (MICs of ≤0.03 μg/ml [ANI], >1 μg/ml [MICA], and >2 μg/ml [CAS]). By comparing the median values of different treatments with those of the placebo groups, it is obvious that the fungal burden is lowest in ANI₅₀₀ and CAS₅₀₀, followed by ANI₁₀₀ and CAS₁₀₀ and the placebo control (see Table S5 in the supplemental material). Significant differences in the CFU/g kidney within the control groups of mice infected with either S_CL or RR_CL were observed, as well as differences between those receiving standard and high-dose medications (see Table S4 in the supplemental material). By comparing the median values of S_CL (MICs of ≤0.008 μg/ml [ANI and MICA] and 0.125 μg/ml [CAS]) and RR_CL, a reduction in that for RR_CL is obvious even if not statistically significant (Fig. 3; see also Table S5 in the supplemental material). These in vivo data fully support and correlate with the in vitro data, as ANI was found to be the most active echinocandin against RR_CL in vivo.

Influence of heterozygous single and double mutations and homozygous double mutations on virulence and in vivo treatment response.

To evaluate the influence of both mutations individually and in combination, both in vitro and in vivo, isogenic mutants (identical genetic background) were generated (see Table S1 in the supplemental material). An overview of the in vitro susceptibility results of the parental strain and isogenic mutants is given in Table 1. The parental strain and SR_MH1 (heterozygous P649P/L) were classified as susceptible in vitro against all echinocandins. RS_MH1 (heterozygous R647R/G) was classified as resistant to MICA only, since it was only one dilution step above the breakpoint. Strain RR_MH2 (heterozygous double mutation; P649P/L and R647R/G) and RR_MHO2 (homozygous double mutation; P649L and R647G) were classified as panechinocandin resistant in vitro. Notably, the ANI and MICA MICs were only one dilution above the respective breakpoints for the heterozygous double mutant RR_MH2. But the homozygous double mutant RR_MHO2 had a MICA MIC that was six dilutions above the breakpoint. Thus, a stepwise increase in MICs and development from single-echinocandin to multiple-echinocandin resistance was observed comparing single to double codon alterations and hetero- to homozygous mutations, respectively. Comparing the clinical heterozygous double mutant RR_CL with the heterozygous isogenic mutant RR_MH2, it is obvious that these two mutations apparently affect ANI susceptibility to a much lesser extent than they affect susceptibility to MICA and CAS, as the MICs for MICA and CAS were several times higher. However, caspofungin testing was performed using the commercial Etest and therefore may not be directly comparable (28), and an interpretation of the MICA susceptibility data should be done with caution, as both isolates showed a trailing phenotype, the implication of which has not been elucidated for the echinocandins (Table 1).

To evaluate the differences in the in vitro responses toward the different echinocandins in vivo, a murine model was again used (Fig. 4). A statistically significant reduction in the fungal burdens in kidneys was observed for ANI, CAS, and MICA (AUC₁₀₀ and AUC₅₀₀) and mutant strains compared to those of the placebo-treated controls. The median of fungal burden per treatment arm and placebo group prove that the parental strain (SS_WT) and the two strains carrying a heterozygous single mutation have the best in vivo treatment responses (see Table S6 in the supplemental material). This is in agreement with in vitro data, as all strains were categorized as ANI susceptible (Table 1). Dose escalation markedly reduced the fungal burden of the heterozygous double mutant RR_MH2, whereas the burden reduction for the homozygous double mutant RR_MHO2 was less prominent (see Table S6). All CAS-treated groups showed statistically significant lower fungal burdens than the mice in the placebo control groups. By comparing the median fungal burden, the treatment efficacy against the heterozygous double mutant RR_MH2 was limited at the standard dose but improved with dose escalation (see Table S6). While in vitro resistance to CAS was confirmed in vivo for the RR_MH2 mutant, resistance in RR_MHO2 was not observed in vivo. Based on their fungal burdens, SS_WT, RS_MH1, and SR_MH1 were found to be CAS susceptible both in vivo and in vitro (Table 1 and Fig. 4).

FIG 4.

In vivo susceptibilities of isogenic mutants derived from parental strain (SS_CL). The parental strain and the four isogenic mutant strains (RR_MH2, RS_MH1, SR_MH1, and RR_MHO2) were evaluated for high (AUC₅₀₀) and standard (AUC₁₀₀) doses of anidulafungin (ANI) (A), caspofungin (CAS) (B), and micafungin (MICA) (C) and compared to placebo control. The treatment groups consisted of six mice and the control groups consisted of eight mice. The CFU counts per ml kidney homogenate are depicted after the end of treatment. The limit of detection (LoD) and median with interquartile range are indicated in each graph. The statistic is based on the Kruskal-Wallis test and Dunnett's multiple comparison tests, with statistical significances given as P < 0.5 (*), P < 0.01 (**), and P < 0.001 (***); outliers (maximum of one per control group) were removed for calculation only (CFU ≤ 100 in the placebo-treated groups). The strains carry the following mutations in their FKS gene: RR_MH2 heterozygous double mutation (R647R/G and P649P/L), RS_MH1 (R647R/G) and SR_MH1 (P649P/L) heterozygous single mutations, and RR_MHO2 homozygous double mutation (R647G and P649L).

The micafungin-treated groups displayed statistically significant lower fungal burdens than those of the placebo-treated mice (P < 0.05; see Fig. 4). The two double mutants RR_MH2 and RR_MHO2 showed only limited responses against MICA₁₀₀ treatment. In contrast to RR_MHO2, RR_MH2 responded to dose escalation. Both strains were resistant toward MICA in vitro. For all other strains, standard dosing was sufficient to clear the infections (Fig. 4), even though RS_MH1 was found to be MICA resistant according to the EUCAST breakpoints (Table 1). However, we would like to mention that the clinical isolates and the mutants constructed in the laboratory represent different background strains, and therefore, comparisons have to be made with caution.

The clinical isolates (S_CL and RR_CL) were found to differ in virulence in terms of kidney CFU count and treatment responses (Fig. 3). On the contrary, isogenic mutants that exclusively carry FKS1 HS1 mutations did not vary in their virulence (Fig. 4). The major difference between RR_MH2 and RR_CL was that RR_MH2 responded to dose escalation of ANI, CAS, and MICA, whereas RR_CL did not. Thus, a clear trend between standard and escalated dosing was found for RR_MH2 and RR_MHO2 and ANI and MICA (Fig. 4; see also Table S5 in the supplemental material). In contrast to the clinical isolates, no differences in virulence were observed in a comparison of the parental strain and the isogenic mutants (Fig. 4; see also Table S6 in the supplemental material).

Fluorescence-activated cell sorting (FACS) analyses on the chitin content of clinical isolates (S_CL and RR_CL), isogenic mutants (SR_MH2, RS_MH2, RR_MH2, and RR_MHO2), and SS_WT (used as a reference) resulted in highly similar fluorescence signals for all cell populations (see Fig. S1 and S2 in the supplemental material). In vitro germination tests in fetal bovine serum showed that both clinical isolates S_CL and RR_CL showed decreased filamentation efficacy. After 1 h at 37°C, about 60% of all cells formed germ tubes, while all isogenic mutants (RS_MH1, SR_MH1, RR_MH2, and RR_MHO2) behaved like the parental wild-type strain SC5314 (after 1 h at 37°C, 100% of all cells formed germ tubes) (Fig. 5). A decreased filamentation ability was most pronounced in RR_CL.

FIG 5.

Filamentation ability of the clinical strains S_CL (A) and RR_CL (B) and the corresponding laboratory strains SS_WT (C) and RR_MH2 (D) in fetal bovine serum after 60 min at 37°C. Delayed germination of approximately 40% of all Candida albicans cells is observed in panels A and B, but 100% of all cells in panels C and D are germinated.

For the first time, it was demonstrated that these SNPs in FKS1 HS1 do not necessarily have an impact on the (i) chitin content, (ii) filamentation ability, and (iii) virulence of C. albicans strains when only these SNPs are integrated into a wild-type strain.

DISCUSSION

In the current study, a female patient suffering from CMC due to MICA- and CAS-resistant C. albicans (Table 1) failed to be cured by long-term CAS medication but was cured by ANI. To determine whether strain RR_CL evolved from S_CL or replaced S_CL under selection pressure of CAS, RAPD genotyping of the clinical isolates was performed. The efficacy of RAPD as a typing method for clinical C. albicans isolates was previously demonstrated (17, 18). RAPD revealed a clonal relationship between the isolates (Fig. 2), which was additionally supported by the presence of six identical synonymous substitutions, one of which was located in the FKS1 HS1 (Table 2). Based on this data, we conclude that echinocandin resistance of C. albicans emerged in the patient.

Molecular analyses demonstrated that RR_CL carried the amino acid substitutions R647R/G and P649P/L in FKS1 HS1 in the same allele (Table 2); the occurrence of such a double amino acid substitution in FKS1 HS1 is new, and its impact on in vitro and in vivo echinocandin susceptibility is demonstrated for the first time.

Recently, it was shown that isolated heterozygous (29) and homozygous (23) point mutations at the amino acid position P649 refer to echinocandin resistance. In contrast to the isolates studied by Arendrup et al. (29) and Garcia-Effron et al. (23), our clinical isolate carries a heterozygous P649P/L substitution that is accompanied by a second amino acid substitution (R647R/G). Also, the isolated homozygous amino acid substitution R647G was published previously by Dannaoui et al. (30). This mutation was found to influence the MICs of only CAS and MICA but not ANI. In general, mutations at positions P649 and P647 in C. albicans were associated with discrete elevations of echinocandin MICs compared with those of FKS1 HS1 alterations involving codons S645 and F641 (22). In our study, only the simultaneous presence of both mutations leads to high-level MICs for CAS and MICA (Table 1). Infections caused by isolates carrying an FKS1 mutation at either position S645 or F641 were not cleared by standard doses (10 mg/kg) of MICA and CAS (31, 32). This is similar to our data, as strain RR_CL was demonstrated in vivo to be resistant in a hematogenous murine model against all echinocandins at a standard dose (Fig. 3). However, in our case study, CMC was cured successfully with ANI treatment. This discrepancy may be explained by various factors. First, the hematogenous infection in the animal model may be more difficult to cure than the less-severe superficial infection in this patient. Second, the lower virulence and growth rate of the clinical isolate may lead to a slower response to treatment, as a certain number of multiplication rounds are needed for activity. Thus, efficacy might have been obtained if the treatment duration had been extended beyond the standard 3 days in the mouse model. Moreover, comparing the in vitro susceptibility of RR_CL with that of the corresponding isogenic mutant strain RR_MH2, it is obvious that ANI susceptibility is less influenced by these mutations than is the case for CAS and MICA. The MICs of the clinical isolate are remarkably higher for CAS and MICA. This finding correlated with the clinical response, as the patient failed caspofungin but not anidulafungin treatment. Since strain RR_MH2 displays pronounced trailing when tested according to the EUCAST breakpoints, however, the micafungin MIC data interpretation may be affected.

A fundamental finding was that the clinical isolate RR_CL and the corresponding RR_MH2 differ in their in vivo responses, particularly with dose escalation (AUC₅₀₀) (Fig. 3 and 4). The efficacy in mice challenged with RR_CL was not improved with high doses of ANI, MICA, and CAS, in contrast to what was observed for mice challenged with the isogenic mutant RR_MH2. Such differences may be explained by the different genetic backgrounds of the clinical and the laboratory-derived mutant strains or by additional molecular mechanisms that were acquired by RR_CL during long-term exposure to CAS. Such additional adaptations associated with echinocandin resistance in C. albicans were published previously; among them are the overexpression of HSP90 (33), the overexpression of Cdr2p ATP-binding cassette (ABC) transporter (34), and elevated chitin levels in fungal cell wall (35). Besides the FKS1 HS1 mutations, additional simultaneously operating mechanisms might contribute to the altered echinocandin resistance of the clinical C. albicans strain RR_CL. To fully address the question of alternative mechanisms that enhance echinocandin resistance, a whole-genome comparison and RNA sequencing will be performed on the clinical isolates S_CL and RR_CL.

Interestingly, others have reported that some FKS1 mutations are associated with reduced virulence and fitness in clinical and laboratory-derived mutant strains (35–37). In the current study, the fungal loads of tissue homogenates of mice infected with either the parental strain SC5314 or isogenic mutant strains were identical, suggesting that mutations do not affect virulence or fitness in vivo. Notably, the clinical isolates S_CL and RR_CL show reduced filamentation in vitro, in contrast to those of the parental strain SC5314 and the derived isogenic mutants (Fig. 1 and 5). Ben-Ami et al. (36) explained the loss of fitness in homozygous FKS1 HS mutants (F641S, S645F, and S645P) by a reduced maximum catalytic capacity of glucan synthase complex and increased cell wall chitin that comes along with thickened cell walls hampered by filamentation capacities (37). The isogenic mutants investigated herein (carrying either R647R/G and/or P649P/L or R649G and P649L in FKS1 HS1) lacked any change in cell wall chitin (see Fig. S1 in the supplemental material). Therefore, mutations at positions R647R/G and/or P649P/L seem not to influence the filamentation of C. albicans, adding to the previous observations that various resistance mutations may or may not influence virulence and fitness (38). The impact of the position of the amino acid substitution in the FKS1 gene on the fitness of C. albicans needs to be further investigated using competition experiments with bar-coded strains. Moreover, increased or decreased fitness of organisms may depend on the immune status of the host, as significant differences between immunocompetent and neutropenic mice were found (37).

Slater et al. (31) reported that infections due to homozygous FKS S645 mutants were not cleared with MICA at standard or elevated (AUC₄₀₀) doses. The finding of Slater et al. (31) might be explained by the fact that S645 is the most dominant resistance mutation described yet, and in comparison to our study dose, it was escalated only four times. In our study, infections with RR_MHO2 failed treatment with MICA₁₀₀, ANI₁₀₀, CAS₁₀₀, MICA₅₀₀, and ANI₅₀₀ (Fig. 4), even though a dose response was observed, as the CFU count decreased for escalated doses (AUC₅₀₀). Compared with strains in the study of Slater et al. (31), our strains carry weaker mutations and therefore show a partial dose-dependent response. However, only CAS₅₀₀ was able to clear the infection with RR_MHO2 below the detection level. Hence, dose escalation may not be sufficient to fully overcome infections with double-mutant strains.

Translated into clinical practice, our data suggest that the EUCAST MICs of echinocandins interpreted by the associated EUCAST breakpoints correlate with in vivo responses using a standard dosing regimen, but they cannot predict the in vivo response to dose escalation according to the murine model applied. Of note, EUCAST and CLSI have abstained from setting caspofungin epidemiological cutoff values for Candida species because of unacceptable high variation in the MIC ranges obtained over time and between centers (27). Also, commercial systems, such as Etest, do not overcome these variation problems (28). Therefore, it is currently not recommended to perform caspofungin susceptibility testing in routine laboratories; instead, anidulafungin and micafungin should be tested and reported as echinocandin markers (27). Nevertheless, caspofungin susceptibility testing is performed for research purposes to evaluate the degree of increase in the MIC. The presence of simultaneous mutations within the FKS1, as well as the history of long-term echinocandin therapy and the occurrence of homozygous FKS1 mutations, should be treated with caution in terms of ongoing treatment with ANI, CAS, or MICA. The presence of an FKS1 HS mutation in a clinical isolate does not necessarily imply that the isolate is less virulent than a wild-type C. albicans strain. C. albicans strains carrying either a heterozygous/homozygous double mutation in an allele or homozygous single mutations are more likely to be therapy refractory in vivo not only at standard doses but also with dose escalation.

In conclusion, the adoption of EUCAST breakpoints for the clinical strains resulted in a classification that correlated with the clinical response in the patient. In the animal model, the clinical strains with heterozygous double mutations within the FKS1 gene and the laboratory-generated strain with a homozygous double mutation failed to respond to dose escalation. C. albicans strains carrying heterozygous double mutations have higher in vitro MICs for echinocandins than do the parental and heterozygous single-mutation strains. Neither the heterozygous single mutations (R647R/G and P649P/L), the heterozygous double mutations (R647R/G and P649P/L), nor the homozygous double mutations (R647G and P649L) in the FKS1 HS1 had an impact on (i) the virulence of isogenic C. albicans with respect to CFU/ml kidney homogenate, (ii) the filamentation capacity, and (iii) chitin content. Therefore, any loss of fitness and/or virulence cannot be provided based on the presence of FKS1 HS mutations, but it probably depends on the nature and position of the specific alteration. The various FKS1 HS mutations differ in MIC elevations, in their response to ANI, CAS, and MICA, and in their impact on virulence.