Mechanisms of Acquired In Vivo and In Vitro Resistance to Voriconazole by Candida krusei following Exposure to Suboptimal Drug Concentration

Five Candida krusei isolates (susceptible and resistant) recovered from the urine of a kidney transplant patient treated with voriconazole (VRC) 200 mg twice daily for 20 days were studied. Eight unrelated clinical isolates of C. krusei were exposed in vitro to VRC 0.001 μg/ml for 30 days. Development of VRC transient resistance occurred in vivo, and induction of permanent resistance occurred in vitro.

ABSTRACT

Five Candida krusei isolates (susceptible and resistant) recovered from the urine of a kidney transplant patient treated with voriconazole (VRC) 200 mg twice daily for 20 days were studied. Eight unrelated clinical isolates of C. krusei were exposed in vitro to VRC 0.001 μg/ml for 30 days. Development of VRC transient resistance occurred in vivo, and induction of permanent resistance occurred in vitro. Mostly, ABC1 and ERG11 genes were overexpressed, and a homozygous T418C mutation in the ERG11 gene was found.

INTRODUCTION

Candida species are unique in their ability to both colonize and cause invasive disease in the urinary tract (1). Candiduria is asymptomatic in up to 96% of patients and does not systematically require antifungal therapy (2). However, it can spark development of systemic candidiasis in critically ill patients. Candiduria may also be a sign of candidemia. Patients diagnosed with candiduria in the intensive care unit have higher mortality rates than similar patients without candiduria (3). Fluconazole (FLC) is the first choice for treatment because 60% to 80% is excreted unchanged in urine by the kidneys; also, it reaches high concentrations in urine, exceeding plasma levels by up to 20 times (4–7). Other azoles, such as voriconazole (VRC) and posaconazole (POS), achieve minimal urinary excretion, can concentrate well in kidney tissue, and may be effective in Candida renal parenchymal infections (7–9). Liposomal amphotericin B is not recommended for the treatment of lower urinary tract infections caused by Candida species because it is not found in significant concentrations in urine (10). However, amphotericin B bladder irrigation, although controversial, has recently been considered an alternative treatment in patients with cystitis and urinary fungus balls or those who cannot be treated with systemic antifungals (11–13). Two major mechanisms associated with azole resistance are described in C. krusei: the presence of Abc1 and Abc2 efflux pump proteins, encoded by the ABC1 and ABC2 genes (14), and reduced sensitivity of the Erg11 protein, encoded by the ERG11 gene, to azole antifungals (15).

In this study, we describe a case of acquired resistance to VRC in a renal transplant patient with several complications, including candiduria due to C. krusei infection. A 20-year-old woman was submitted to renal transplantation and received immunosuppressive therapy. Twelve days after transplantation, bacteremia due to Escherichia coli developed; the patient was administered intravenous ceftriaxone and ciprofloxacin. On day 18 after transplantation, surgical procedures were performed to preserve the graft. At 25 days after transplantation, the first episode of candiduria occurred. On day 44 posttransplantation, oral VRC 200 mg twice daily was prescribed for 20 days despite its probable low urinary diffusion, associated with low expected efficacy against candiduria. However, the primary concerns were (i) to protect the parenchyma of the kidney transplant from Candida infection and (ii) to treat candiduria, because the patient was a transplant recipient with a high risk of rejecting the transplant. Fluconazole could not be administered due to C. krusei intrinsic resistance, and amphotericin B was not initially considered for treatment because of its nephrotoxicity. At 82 days after renal transplantation, the patient was readmitted to the emergency department with fever; a dissecting aneurysm of the renal artery was diagnosed by computed tomography scan. The kidney graft was resected, and the patient was started on antibiotics and intravenous amphotericin B 1 mg/kg daily for 3 weeks. Since the first episode of candiduria, several C. krusei isolates were recovered from the urine, and five were selected according to their VRC susceptibility profile. They were designated according to the day of isolation, as follows: Ck_B.VRC (isolated before VRC therapy), Ck_D.VRC9 and Ck_D.VRC16 (isolated during VRC therapy, on days 9 and 16), and Ck_A.VRC10 and Ck_A.VRC18 (isolated after VRC discontinuation, on days 10 and 18). VRC, POS, and itraconazole (ITC) susceptibility profiles were determined according to CLSI reference protocols M27-A3 and M27-S4, and the susceptibility profile was attributed according to the same protocols (16, 17). The timeline of isolation and susceptibility profiles of the C. krusei clinical isolates are detailed in Fig. 1a. All of the isolates exhibited the same CKRS-1 electrophoretic pattern (Fig. 1b), determined according to Carlotti et al. (18), and the same allelic CKTNR microsatellite profile, i.e., profile e-f/d, described by Shemer et al. (19). Previously, combination of these two typing methods showed a discriminatory power of 0.995 (F. Grenouillet, unpublished data). Therefore, with an error risk of 0.5% (20), we stated that the same C. krusei strain was colonizing the transplanted patient during this period despite exhibiting distinct antifungal susceptibility profiles. Three unrelated C. krusei strains were used as control, two clinical isolates recovered from other patients and C. krusei strain CBS 573 (18, 19).

FIG 1.

In vivo induction of resistance to voriconazole. (a) Timeline of the renal transplant procedure and antifungal therapy; identity of the C. krusei clinical isolates recovered from the kidney-transplant recipient and their respective susceptibility profiles. (b) Genotyping of C. krusei clinical isolates. M, molecular weight marker. (c) Relative gene expression profile of ABC1, ABC2, and ERG11 genes for the C. krusei clinical isolates. *, P ≤ 0.05.

The transient acquisition of resistance to VRC in vivo was intentionally replicated in vitro in the presence of subinhibitory concentrations of VRC: eight independent C. krusei clinical isolates susceptible to VRC obtained from distinct patients not previously submitted to VRC therapy (D0) were grown in brain heart infusion (BHI) containing VRC 0.001 μg/ml for 30 days (D30). The MICs of VRC, POS, and ITC were determined as described previously (16, 17) and are presented in Table 1. The MICs of VRC for the resistant strains were redetermined in accordance with CLSI M27-A3 and M27-S4 protocols in the absence and presence of FK506 (tacrolimus) 100 μg/ml, a blocker of ATP-dependent efflux pumps (21). The MIC values decreased by 2-fold up to 7-fold in all C. krusei strains assayed, and all reverted from resistant to susceptible phenotypes (Table 2). Analyses of resistance gene expression were performed according to Ricardo et al. (22), except for the SYBR green used in the quantitative real-time PCR mixture (SensiFAST SYBR No-Rox mix 1×, Bioline) and the primer annealing temperature of 60°C. Target genes exhibiting a ≥2-fold increase in the expression level and a P value of ≤0.05 were considered significantly overexpressed. The ERG11 gene (1,890 bp) was amplified by PCR and other sequencing reaction conditions and sequence analyses were according to Ricardo et al. (22). Resistant C. krusei clinical isolates Ck_D.VRC9 and Ck_D.VRC16 presented significant overexpression of ABC1 and ERG11 genes, which came to basal levels in the posttherapy susceptible isolates (Fig. 1c). No mutations in the ERG11 gene were found in these isolates, which is in accordance with the transient phenotype, where gene expression increases only in the presence of the antifungal. A transient resistant phenotype has already been described in Candida albicans due to the presence of instable aneuploidies, which often disappear when the strains grow in the absence of the drug together with loss of resistance (23, 24). A similar phenomenon may occur in the resistant C. krusei clinical isolates; albeit, the ineffective antifungal treatment due to the low concentration of VRC reaching the urinary tract was enough to induce a transient resistance in vivo. Besides, the acquisition of antifungal drug resistance due to short-term exposure or the presence of low doses of antifungals associated with therapeutic protocols has recently been described in Candida tropicalis and Candida auris isolates (25, 26). C. krusei strains incubated in vitro with very low concentrations of VRC revealed different gene expression profiles. All strains, except Ck24_D30, Ck34_D30, and Ck40_D30, exhibited at least one of the resistance genes significantly overexpressed (>2-fold increase; P< 0.05) compared with the respective susceptible strain (D0). The most remarkable increase in both ABC efflux genes was registered in strain Ck21_D30, which also presented the highest MIC value of VRC (Table 1). This fact can explain the highest reduction in the MIC value of VRC in the presence of FK506 (Table 2), and efflux was probably the main mechanism of resistance in this strain. Different mutations in the ERG11 gene sequence were found in these strains (Table 1). The mutations found in strains Ck21_D30, Ck24_D30, Ck32_D30, and Ck42_D30 were previously described in C. albicans (27); in particular, the T418C mutation found in strains Ck24_D30 (homozygous) and Ck42_D30 (heterozygous) was associated with VRC resistance in C. krusei (22) and C. albicans (28, 29). Strain Ck24_D30, which had the lowest expression of resistance genes, was the only strain showing a homozygous T418C mutation translating into Tyr140→His. We can relate this mutation to its resistance phenotype, although other experiments, such as expression of the mutated allele in a susceptible yeast strain, could be performed to confirm this association. Concerning strain Ck42_D30, because the evolution of resistance can occur in a stepwise manner (30), occurrence of the heterozygous T418C mutation may represent an intermediate step toward a more stable resistant phenotype. The other two heterozygous mutations in strains Ck21_D30 and Ck32_D30 (Table 1) translate the amino acid Ala122 to Ser or Val, respectively, located within 12 Å of the Erg11 protein binding site for VRC and FLC (31), thus interfering with azole affinity to the target protein. Strains Ck34_D30 and Ck40_D30 presented no mutation in the ERG11 gene or significant overexpression of efflux pumps encoding genes and ERG11, which may explain the resistance phenotype. Studies by Cuomo and colleagues (32), identified orthologues of genes in a C. krusei clinical isolate that were previously associated with drug resistance in C. albicans. Hence, these orthologue genes may be present in these strains conferring resistance and explaining the reversion of the phenotype in the presence of FK506. The presence of multiple mechanisms of resistance in C. albicans confers crossed resistance to azoles (33); therefore, the same may occur in our resistant C. krusei strains, explaining not only the resistance phenotype to VRC but also the elevated MIC values of POS and ITC.

TABLE 1.

Susceptibility profile; relative expression of ABC1, ABC2, and ERG11 genes; and ERG11 gene mutations of C. krusei strains incubated in vitro with VRC

Strain ID	Biological sample	MIC50 (μg/ml)/susceptibility profilea for:						Gene expression (fold increase [P value])			ERG11 gene sequence mutations
		VRC		POS		ITC		ABC1	ABC2	ERG11	D0	D30
		D0	D30	D0	D30	D0	D30
Ck1	Sputum	0.25/S	2.0/R	0.125	0.5	0.25	1.0	3.67 (0.0045)	b	b
Ck8	Mouth	0.25/S	2.0/R	0.125	1.0	0.5	>16	5.55 (0.0132)	b	b
Ck21	Digestive fistula	0.50/S	8.0/R	0.25	1.0	0.5	>16	8.89 (0.009)	16.02 (0.0028)	b		Heterozygous nonsynonymous G364G/T (A122→S)
Ck24	Urine	0.25/S	4.0/R	0.25	0.25	0.5	>16	b	b	b		Homozygous nonsynonymous T418C (Y140→H)
Ck32	Sputum	0.125/S	2.0/R	0.125	1.0	0.25	>16	2.15 (0.099)	b	2.97 (0.0015)		Heterozygous nonsynonymous C365C/T (A122→V)
Ck34	Stools	0.50/S	4.0/R	0.25	0.5	0.5	>16	b	b	b
Ck40	Sputum	0.25/S	2.0/R	0.25	0.5	0.5	>16	b	b	b
Ck42	Ascites	0.25/S	4.0/R	0.25	0.25	0.5	>16	2.07 (0.0089)	b	2.71 (0.0013)	Heterozygous nonsynonymous C1091T (A364→V)	Heterozygous nonsynonymous C1091T (A364→V) and T418C (Y140→H)

^aS, susceptible; R, resistant.

^bGene expression fold increase <2; P ≥ 0.05.

TABLE 2.

MIC and susceptibility profile to VRC alone and with FK506

Resistance induction	Strain	MIC50 (μg/ml)/susceptibility profile for:
		VRC	VRC + FK506 100 μg/ml
In vivo	CkD.VRC9	4/R	0.125/S
	CkD.VRC16	4/R	0.125/S
In vitro	Ck1D30	2/R	0.25/S
	Ck8D30	2/R	0.06/S
	Ck21D30	8/R	0.06/S
	Ck24D30	4/R	1.0/I
	Ck32D30	2/R	0.125/S
	Ck34D30	4/R	0.125/S
	Ck40D30	2/R	0.06/S
	Ck42D30	4/R	0.25/S

In summary, the development of resistance in C. krusei strains from different backgrounds was studied, even in the presence of very low doses of VRC, both in vivo and in vitro. This shows that candiduria should not be considered an innocent situation, and physicians should pay particular attention to such cases. Different mechanisms of resistance were found, i.e., increased efflux pump expression and the presence of mutations in the ERG11 gene conferring increased resistance to VRC and promoting the high MIC values of POS and ITC.