Microsatellite Typing and Resistance Mechanism Analysis of Voriconazole-Resistant Aspergillus flavus Isolates in South Korean Hospitals

A recent surveillance study in South Korea revealed that 14% (7/50) of Aspergillus flavus clinical isolates had a voriconazole minimum inhibitory concentration of ≥4 μg/ml. Of seven non-wild-type (non-WT) isolates, six ear isolates from four hospitals shared the same microsatellite genotype.

ABSTRACT

A recent surveillance study in South Korea revealed that 14% (7/50) of Aspergillus flavus clinical isolates had a voriconazole minimum inhibitory concentration of ≥4 μg/ml. Of seven non-wild-type (non-WT) isolates, six ear isolates from four hospitals shared the same microsatellite genotype. None of the non-WT isolates showed cyp51 mutations associated with azole resistance. However, the mean expression levels of efflux pump (MDR2, atrF, and mfs1) and target (cyp51A) genes exhibited significant differences between non-WT and other isolates.

INTRODUCTION

The survival rates of immunocompromised patients with invasive aspergillosis have improved as a result of many factors, one of which is the availability of azole antifungal drugs (1). Voriconazole is one of the preferred agents for first-line treatment (2); thus, the emergence of voriconazole-resistant Aspergillus isolates has impacted the management of aspergillosis. Most invasive aspergillosis cases are caused by Aspergillus fumigatus, isolates of which are frequently resistant to voriconazole (3–5). A. flavus is the second leading cause of aspergillosis (6, 7). A. flavus is usually susceptible to azole antifungals (8); the prevalence of voriconazole-resistant A. flavus isolates is 0% to 4.9% (8–10) based on the Clinical and Laboratory Standards Institute epidemiological cutoff value (ECV, 2 μg/ml) (11). A recent surveillance study in South Korea indicated that 14% of A. flavus isolates had voriconazole MICs above the ECV. In this study, we investigated the genetic relationship and resistance mechanisms of these voriconazole non-wild-type (non-WT) isolates of A. flavus.

A total of 50 A. flavus isolates (28 respiratory tract, 16 ear discharge, and 6 other isolates) obtained from 10 university hospitals from February 2012 through August 2013 were analyzed. All isolates were identified by partial sequencing of the β-tubulin and calmodulin genes (12). The MICs of antifungal agents were determined by the CLSI M38-A2 method (13). To evaluate inter- and intrahospital spread in the university hospitals in South Korea, the isolates were genotyped by microsatellite typing (14). The target genes (cyp51A, cyp51B, and cyp51C) and their promoter regions were sequenced (15, 16). The expression levels of target genes (cyp51A, cyp51B, and cyp51C) and genes encoding efflux pumps (MDR1, MDR2, MDR4, atrF, and mfs1) were quantified by real-time PCR (17).

Based on the CLSI ECVs (11), the rates of non-WT isolates to voriconazole (>2 μg/ml), posaconazole (>0.5 μg/ml), and itraconazole (>1 μg/ml) were 14.0% (7/50), 0.0% (0/50), and 6.0% (3/50), respectively. Intrinsic amphotericin B resistance has been recognized in Aspergillus terreus isolates for many decades, and A. flavus isolates have been known to have reduced susceptibility to amphotericin B (18). In a recent study from Brazil, 87% of A. flavus isolates had an amphotericin B MIC of ≥2 μg/ml (19), whereas we showed that 30% (15/50) had an amphotericin B MIC of ≥2 μg/ml.

Table 1 summarizes the data for seven voriconazole non-WT isolates of A. flavus. Microsatellite typing showed that the six non-WT isolates (E1 to E6) obtained from the ear specimens of six patients with otitis media at four hospitals shared the same microsatellite genotype (type 1); however, one respiratory isolate (R1) from a patient with chronic obstructive pulmonary disease was unique (type 2). Likewise, in two recent studies, none of seven non-WT isolates had missense mutations in cyp51A or cyp51B or alterations in their promoters (20, 21). However, all harbored the same nonsynonymous mutations (M54T/S240A/D254G/N423D) in cyp51C. The seven patients from whom non-WT isolates were obtained in this study had not received azole therapy, suggesting that most azole resistance in A. flavus isolates from South Korean hospitals were environmentally acquired, as seen for A. fumigatus isolates (1).

TABLE 1.

Antifungal susceptibilities, molecular characteristics, and clinical information of seven voriconazole non-wild-type isolates of A. flavus

Isolate no.	Source	Yr of isolation	Hospital	Antifungal MIC (μg/ml)a of:					Amino acid substitutions in 3 target genes and alteration of promoter regionsc			Patient information
				VOR	AMB	ITRA	POSA	Microsatellite typeb	cyp51A	cyp51B	cyp51C	Age (yrs)/sex	Underlying disease(s)d	Previous antifungal exposure
E1	Ear	2013	A	8	4	2	0.5	1	None	None	M54T, S240A, D254G, N423D	45/F	OM	No
E2	Ear	2013	B	8	1	1	0.5	1	None	None	M54T, S240A, D254G, N423D	54/F	OM	No
E3	Ear	2013	C	8	2	2	0.5	1	None	None	M54T, S240A, D254G, N423D	30/F	Cholesteatoma	No
E4	Ear	2012	A	4	0.5	2	0.5	1	None	None	M54T, S240A, D254G, N423D	57/M	OM	No
E5	Ear	2013	A	4	1	1	0.25	1	None	None	M54T, S240A, D254G, N423D	79/F	DM, OM	No
E6	Ear	2013	D	4	1	0.5	0.25	1	None	None	M54T, S240A, D254G, N423D	54/F	OM	No
R1	Respiratory	2012	A	4	1	1	0.25	2	None	None	T34A, M54T, S240A, D254G, N423D	79/M	COPD	No

^aDetermined by CLSI M38-A2 broth microdilution method. VOR, voriconazole; AMB, amphotericin B; ITRA, itraconazole; POSA, posaconazole.

^bBy microsatellite typing, each strain was characterized by using the sizes of the amplified products of nine markers (2A-2B-2C-3A-3B-3C-4A-4B-4C) (14). Type 1 (126-132-146-147-141-156-112-116-152) was shared by 6 ear isolates (E1 to E6), and the one respiratory isolate (R1) was of type 2 (168-122-130-132-171-147-120-136-128).

^cAmino acid substitutions in the three cyp51 genes and alteration of promoter regions were evaluated by sequencing.

^dOM, otitis media; DM, diabetes mellitus; COPD, chronic obstructive pulmonary disease with cor pulmonale.

The microsatellite typing performed in this study, with a panel of nine markers (14), enabled discrimination of all 49 A. flavus isolates tested (Fig. 1). All 49 of the A. flavus isolates exhibited high genetic diversity, yielding 39 distinct genotypes, which reflect the diversity of airborne A. flavus conidia in the hospitals. However, six non-WT isolates from ear cultures were genetically distant from other isolates and clustered into one clonal type (type 1) harboring the same cyp51C mutations (M54T/S240A/D254G/N423D), suggesting the existence of clonal non-WT isolates of A. flavus. However, the cyp51C mutations (M54T/S240A/D254G/N423D) found in all seven of the non-WT isolates were also found in 28.6% (12/42) of the WT isolates.

FIG 1.

Microsatellite typing dendrogram of 49 A. flavus isolates: 15 ear isolates (E1 to E15), 29 respiratory isolates (R1 to R29), and 2 eye discharge (O1, O2), 2 pus (O3, O4), and 1 peritoneal fluid (O5) isolates. By microsatellite typing, 15 isolates shared five genotypes (types 1, 15, 16, 29, and 38) (asterisk), suggesting clonal isolates. The dendrogram is based on a categorical analysis of nine microsatellite markers in combination with unweighted paired-group methodology using arithmetic average (UPGMA) clustering.

Azole resistance mechanisms in clinical A. fumigatus isolates have been investigated extensively, and the main mechanism is the presence of point mutations in the target gene cyp51A (4, 5, 22, 23). In contrast to A. fumigatus, the resistance mechanisms of A. flavus have been examined in a limited number of clinical strains (15, 20, 21, 24). Liu et al. (25) suggested that a T788G (S240A) missense mutation in cyp51C conferred voriconazole resistance on A. flavus isolates. However, in this study, S240A in cyp51C was present in all 49 A. flavus isolates irrespective of their voriconazole susceptibility, in agreement with previous reports (15, 20, 21), which indicates that the S240A mutation seldom contributes to the acquisition of voriconazole resistance in A. flavus isolates. A homology model of non-WT isolates suggested that S196F and N423D missense mutations in cyp51C exhibited major structural and functional effects on cyp51C drug binding (20). In our study, the N423D mutation in cyp51C was also found in 7 non-WT and 15 WT isolates. Overall, our data suggest that mutations resulting in drug target modification may not be important for voriconazole resistance in A. flavus isolates from South Korean hospitals.

To date, three studies investigated the overexpression of target or efflux pump genes in clinical isolates of A. flavus (20, 21, 24). In a study, four of five voriconazole non-WT isolates had cyp51A and MDR1 overexpression (24), while in the other study, one of three non-WT isolates exhibited overexpression of target genes (cyp51A, cyp51B, and cyp51C) and efflux pump genes (MDR1, MDR2, atrF, and mfs1) (20). In a study by Paul et al. (21), two non-WT isolates showed voriconazole-induced overexpression of Cdr1, suggesting a possible role of multidrug efflux pumps. In our study, we analyzed 34 clinical isolates, including 7 non-WT and 27 WT isolates. Of the eight genes tested, the mean expression levels of four exhibited significant differences between non-WT and WT isolates (non-WT versus WT isolates: cyp51A, 2.529 versus 1.789; MDR2, 0.7429 versus 1.037; atrF, 1.129 versus 0.8963; and mfs1, 3.757 versus 2.670, respectively; all P < 0.05) (Fig. 2). Although we did not test Cdr1 overexpression, our results show that overexpression of efflux pump or target genes may explain the high voriconazole MICs of clinical A. flavus isolates from South Korean hospitals, rather than cyp51 mutation.

FIG 2.

Relative expression levels of three target genes (cyp51A, cyp51B, and cyp51C) and drug-efflux-associated genes (MDR1, MDR2, MDR4, artF, and mfs1) in 7 voriconazole non-WT and 27 WT isolates relative to that of a control strain, which was susceptible to all three azoles (set to 1.0). The seven non-WT isolates had higher mean expression levels of cyp51A (2.529 versus 1.789), atrF (1.129 versus 0.8963), and mfs1 (3.757 versus 2.670) than the 27 WT isolates (P = 0.042, 0.0316, and 0.0112, respectively) but a lower expression level of MDR2 (0.7429 versus 1.037; P = 0.0278). The differences in the mean expression levels of cyp51B, cyp51C, MDR1, and MDR4 between non-WT and WT isolates were not statistically significant.