In Vitro Biochemical Study of CYP51-Mediated Azole Resistance in Aspergillus fumigatus

Abstract

The incidence of triazole-resistant Aspergillus infections is increasing worldwide, often mediated through mutations in the CYP51A amino acid sequence. New classes of azole-based drugs are required to combat the increasing resistance to existing triazole therapeutics. In this study, a CYP51 reconstitution assay is described consisting of eburicol, purified recombinant Aspergillus fumigatus CPR1 (AfCPR1), and Escherichia coli membrane suspensions containing recombinant A. fumigatus CYP51 proteins, allowing in vitro screening of azole antifungals. Azole-CYP51 studies determining the 50% inhibitory concentration (IC₅₀) showed that A. fumigatus CYP51B (Af51B IC₅₀, 0.50 μM) was 34-fold more susceptible to inhibition by fluconazole than A. fumigatus CYP51A (Af51A IC₅₀, 17 μM) and that Af51A and Af51B were equally susceptible to inhibition by voriconazole, itraconazole, and posaconazole (IC₅₀s of 0.16 to 0.38 μM). Af51A-G54W and Af51A-M220K enzymes were 11- and 15-fold less susceptible to inhibition by itraconazole and 30- and 8-fold less susceptible to inhibition by posaconazole than wild-type Af51A, confirming the azole-resistant phenotype of these two Af51A mutations. Susceptibility to voriconazole of Af51A-G54W and Af51A-M220K was only marginally lower than that of wild-type Af51A. Susceptibility of Af51A-L98H to inhibition by voriconazole, itraconazole, and posaconazole was only marginally lower (less than 2-fold) than that of wild-type Af51A. However, Af51A-L98H retained 5 to 8% residual activity in the presence of 32 μM triazole, which could confer azole resistance in A. fumigatus strains that harbor the Af51A-L98H mutation. The AfCPR1/Af51 assay system demonstrated the biochemical basis for the increased azole resistance of A. fumigatus strains harboring G54W, L98H, and M220K Af51A point mutations.

INTRODUCTION

Since the late 1990s, azole resistance in Aspergillus fumigatus clinical isolates has been increasing around the world. The global ARTEMIS surveillance program found that 5.8% of A. fumigatus clinical isolates showed elevated MICs of one or more triazoles (1), while the SCARE program in Europe found that 3.4% of A. fumigatus clinical isolates were azole resistant (2); however, the incidence of azole-resistant isolates varied between the 22 medical centers (0 to 26%). In the Netherlands, ∼10% of all clinical isolates are now itraconazole resistant (2), compared to Manchester, where 23% and 31% of isolates were azole resistant in 2008 and 2009 (3).

Azole resistance in A. fumigatus is often mediated through development of point mutations in the Af51A gene. The five A. fumigatus CYP51A positions, or hot spots, most frequently undergoing mutations responsible for conferring azole resistance are glycine-54, leucine-98, glycine-138, methionine-220, and glycine-448 (4). G54, G138, M220, and G448 CYP51A point mutations are thought to have arisen during triazole therapy of patients in the clinic (5, 6), while TR₃₄/L98H and TR₄₆/Y121F/T289A may have arisen in the environment in the Netherlands in response to the use of agricultural triazole fungicides (7–10).

Therefore, new classes of azole-based drugs are required to combat the emerging resistance observed in the clinic against current triazole therapeutics, along with an in vitro screening assay system for assessing the potency of new drug candidates against azole-resistant CYP51A mutant isoforms and understanding the mode of resistance caused by a specific CYP51A mutation. To construct such an assay system, AfCPR1 was expressed, purified, and characterized as the redox partner of Af51A isoforms expressed in Escherichia coli membranes. Determinations of the 50% inhibitory concentration (IC₅₀) of azole using this system demonstrated a direct biochemical basis for the observed increased azole resistance of A. fumigatus strains harboring G54W, L98H, and M220K point mutations in Af51A. In addition, AfCPR1 was an effective redox partner for other CYP51 enzymes.

MATERIALS AND METHODS

Gene cloning.

The AfCPR1 gene (UniProtKB accession number Q4WM67) was synthesized with codon optimization for expression in Escherichia coli by GeneCust (Dudelange, Luxembourg). A six-codon extension (CATCACCATCACCATCAC) encoding six histidine residues was inserted prior to the stop codon. AfCPR1 was cloned into the pCWori⁺ expression vector using NdeI and HindIII restriction sites (11). This process was repeated for the AfCPR2 gene (Q4X224). The A. fumigatus CYP51A (Af51A) and A. fumigatus CYP51B (Af51B) genes were cloned as previously described into the pSPORT expression vector (12), with codon usage optimized for expression in E. coli. The pSPORT:Af51A construct was used as the template for the synthesis of the Af51A-G54W, Af51A-L98H, and Af51A-M220K point mutations by GeneCust (Dudelange, Luxembourg). Gene integrities were verified by DNA sequencing.

Phylogenetic analysis of fungal CPRs.

Alignments of selected Aspergillus and fungal cytochrome P450 reductase (CPR) protein sequences were constructed using ClustalX version 1.8 (http://www.clustal.org/) and TreeviewX (https://code.google.com/p/treeviewx/). The CPR sequences used for phylogenetic analysis are listed in Table S1 in the supplemental material. NCBI-BLAST2 (http://blast.ncbi.nlm.nih.gov/) was used to calculate percentage sequence identities between the CPRs.

Heterologous protein expression in E. coli.

The pCWori⁺:AfCPR1 and pCWori⁺:AfCPR2 constructs in DH5α E. coli cells were cultured in 1-liter volumes of Terrific Broth supplemented with 0.1 mg ml⁻¹ sodium ampicillin, 0.1 mg ml⁻¹ ferrous sulfate, 0.1 mg ml⁻¹ riboflavin, and 1 mM IPTG grown at 37°C and 200 rpm for 5 h followed by expression at 28°C and 200 rpm for 18 h. E. coli cells were harvested and then lysed by ultrasonication using a Branson (Danbury, CT, USA) digital sonifier in 0.1 M potassium phosphate (pH 6.8) and 25% (wt/vol) glycerol followed by centrifugation (21,000 × g for 30 min at 4°C). The membrane fraction was recovered by ultracentrifugation (134,000 × g for 90 min at 4°C) followed by suspension in 0.1 M potassium phosphate (pH 6.8) and 25% (wt/vol) glycerol. Recombinant AfCPR proteins were recovered by solubilization of the membrane fraction with 2% (wt/vol) sodium cholate (2 h at 4°C) followed by ultracentrifugation (134,000 × g for 90 min at 4°C) and then purified by affinity chromatography using nickel-nitrilotriacetic acid (Ni²⁺-NTA) agarose (13), except that 0.1 M potassium phosphate (pH 6.8) and 25% (wt/vol) glycerol (with and without 1 M NaCl) were used as washing buffers. AfCPR proteins were eluted with 1% (wt/vol) l-histidine in 0.1 M potassium phosphate (pH 6.8) and 25% (wt/vol) glycerol followed by dialysis against 25 mM potassium phosphate (pH 7.5) and 10% (wt/vol) glycerol prior to characterization studies.

A. fumigatus CYP51A and CYP51B were expressed in E coli and purified as previously described (12). In addition to 2% sodium cholate, Emulgen 913 (2%) and Tween 80 (2%) were also used to solubilize different batches of E. coli membranes containing Af51A prior to purification by Ni²⁺-NTA agarose chromatography. Protein purity was assessed by SDS-polyacrylamide gel electrophoresis followed by staining with Coomassie brilliant blue R-250. The membrane fractions of the Af51A, Af51B, Af51A-G54W, Af51A-L98H, and Af51A-M220K E. coli clones were isolated as described for AfCPR1 after 18 h expression (27°C, 170 rpm) and suspended in 0.1 M potassium phosphate (pH 7.5) and 25% (wt/vol) glycerol.

Spectral characterization of proteins.

The absolute absorbance spectrum of the AfCPR1 protein was measured between 700 and 300 nm (light path, 4.5 mm) with the CPR concentration determined from the absorbance maximum of the second flavin spectral peak (∼445 to 470 nm) characteristic of the oxidized enzyme using an extinction coefficient of 21.4 mM⁻¹ cm⁻¹ (14). Reduction of 5 μM AfCPR1 by NADPH was monitored by measuring the absolute spectrum in the presence of 4, 40, and 200 μM NADPH. The reduction of cytochrome c catalyzed by cytochrome P450 reductases in the presence of NADPH was used for the kinetic characterization of AfCPR1 using an extinction coefficient at 500 nm of 21 mM⁻¹ cm⁻¹ (15). The assay system contained 0.2 M Tris-HCl (pH 7.8), 2.5 mM glucose-6-phosphate, 3 U ml⁻¹ glucose-6-phosphate dehydrogenase, and 1.7 nM AfCPR1 at 22°C, in which the cytochrome c or NADPH concentrations were varied and the concentration of the second substrate remained fixed at 50 μM. In the reference cuvettes, the NADPH was replaced by an equivalent volume of 0.2% (wt/vol) sodium bicarbonate. Velocities were expressed as nanomoles of reduced cytochrome c produced per nanomole of AfCPR1 per second. The kinetic parameters K_m and V_max were determined by nonlinear regression using the Michaelis-Menten equation (ProFit 6.1.12; QuantumSoft, Zurich, Switzerland).

Cytochrome P450 concentrations were determined by reduced carbon monoxide difference spectroscopy (16) using an extinction coefficient of 91 mM⁻¹ cm⁻¹ (17) for the absorbance difference between 448 and 490 nm. In addition, the Soret peak at 418 to 420 nm in the absolute spectrum of purified Af51B and Af51A was also used to determine P450 concentrations using an extinction coefficient of 125 mM⁻¹ cm⁻¹ (18). Spectral determinations were made using quartz semimicrocuvettes and a Hitachi (San Jose, CA) U-3310 UV-visible (UV-Vis) spectrophotometer.

CYP51 reconstitution assays.

Purified CYP51 enzymes used included heterologously expressed Af51A (UniProtKB accession number Q4WNT5), Af51B (Q96W81), Homo sapiens CYP51 (Q16850), and Candida albicans CYP51 (P10613) (12, 19). In addition, E. coli membrane preparations of Af51A, Af51B, Af51A-G54W, Af51A-L98H, and Af51A-M220K expression clones were also utilized. The CYP51 reconstitution assay system (20) for purified CYP51 enzymes contained 1 μM CYP51, 2 μM AfCPR1, 50 μM lanosterol or 50 μM eburicol, 50 μM dilauryl phosphatidylcholine (DLPC), 4% (wt/vol) 2-hydroxypropyl-β-cyclodextrin (HPCD), 0.4 mg ml⁻¹ isocitrate dehydrogenase, 25 mM trisodium isocitrate, 50 mM NaCl, 5 mM MgCl₂, and 40 mM MOPS (pH ∼7.2). For membrane preparations, 0.5 μM CYP51 was used with 1 μM AfCPR1 and the DLPC was omitted. Assay mixtures were incubated at 37°C for 10 min prior to initiation with 4 mM β-NADPHNa₄ followed by shaking for 4 min (H. sapiens CYP51), 15 min (C. albicans CYP51, Af51B, and Af51A membranes) or 60 min (purified Af51A and Af51B proteins) at 37°C. Sterol metabolites were recovered by extraction with ethyl acetate followed by derivatization with N,O-bis(trimethylsilyl)trifluoroacetamide, tetramethylsilane, and pyridine prior to analysis by gas chromatography-mass spectrometry (21).

Azole IC₅₀ determinations.

IC₅₀ determinations for fluconazole, voriconazole, itraconazole, and posaconazole were performed in duplicate using the CYP51 reconstitution assay described above (20) with E. coli membrane suspensions of A. fumigatus CYP51 proteins and 50 μM eburicol as the substrate. Azole antifungal agents were added in 2.5 μl dimethyl sulfoxide prior to incubation at 37°C and addition of β-NADPHNa₄.

Chemicals.

All chemicals, including azole antifungal agents, were obtained from Sigma Chemical Company (Poole, United Kingdom). Eburicol was supplied by W. David Nes (Texas Tech University, Lubbock, TX). Growth media, sodium ampicillin, and IPTG were obtained from Foremedium Ltd. (Hunstanton, United Kingdom). The Ni²⁺-NTA agarose affinity chromatography matrix was obtained from Qiagen (Crawley, United Kingdom).

RESULTS

CPR phylogenicity.

AfCPR1 clusters next to CPRs from other higher ascomycetes (see Fig. S1A in the supplemental material) with sequence identities of 86, 76, 71, 70, 68 and 63% for Penicillium digitatum, Trichophyton rubrum, Botryotinia fuckeliana, Mycosphaerella graminicola, Fusarium oxysporum, and Blumeria graminis CPRs, respectively. AfCPR1 shared 46, 37, 43, and 44% sequence identity with the CPRs from C. albicans, H. sapiens, Saccharomyces cerevisiae, and Malassezia globosa and only 42% sequence identity with AfCPR2. In contrast, AfCPR2 had low sequence identity with the other fungal CPRs (35 to 41%) and only 33% sequence identity with H. sapiens CPR. Phylogenetic analysis of the Aspergillus CPRs (see Fig. S1B in the supplemental material) clearly shows the presence of two CPR isoenzymes in Aspergillus species. AfCPR1 is located next to other CPR1 enzymes with sequence identities ranging from 84% with Aspergillus ruber CPR1 to 98% with Aspergillus fischeri CPR1. Similarly AfCPR2 is located next to other CPR2 enzymes; however, the sequence identities were more varied than those obtained for the CPR1 enzymes, ranging this time from 55% with Aspergillus nidulans CPR2 to 95% with A. fischeri CPR2. No CPR2 open reading frames could be found in the Aspergillus clavatus or Aspergillus terreus genome database (http://www.broadinstitute.org/annotation/genome/aspergillus_group/MultiHome.html) when Aspergillus oryzae CPR2 and AfCPR2 were used as templates, suggesting either that a technical oversight occurred or that these two Aspergillus species lack a CPR2 gene.

Heterologous expression and purification of recombinant CPR and CYP51 proteins.

Purification of cholate solubilized AfCPR1 by Ni²⁺-NTA agarose chromatography resulted in the recovery of ∼190 nmol of AfCPR1 per liter of cell culture and a 56 μM AfCPR1 stock solution after dialysis. In contrast, no AfCPR2 was recovered when identical expression and purification conditions were used for the full-length enzyme or for two N-terminally modified versions of AfCPR2 (residues 1 to 12 replaced by MALLLAVF [22] or residues 1 to 28 replaced by MAQLDTLD). Further modification/optimization of the N-terminal anchor region of AfCPR2 is required to facilitate expression. Chromatography of Af51B on Ni²⁺-NTA agarose yielded a stock 31 μM Af51B solution after dialysis. Chromatography of Af51A on Ni²⁺-NTA agarose resulted in recovery of ∼10% of the Af51A when the Ni²⁺-NTA agarose matrix was eluted with the 1% (wt/vol) l-histidine buffer. The remaining ∼90% Af51A was recovered only by elution with 2% (wt/vol) sodium cholate or 2% (wt/vol) Emulgen 913 added to the 1% (wt/vol) l-histidine buffer. This suggested that the primary interaction between Af51A and the column matrix was hydrophobic and not through the C-terminal six-histidine tag. After dialysis, the stock Af51A solution recovered was 28 μM. SDS-polyacrylamide gel electrophoresis confirmed AfCPR1, Af51A, and Af51B purities were greater than 90% when assessed by the staining intensity of protein bands. Apparent molecular weights of 78,500 for AfCPR1 and 58,000 ± 2,000 for both Af51 proteins were obtained, which were close to the predicted values of 77,606, 58,888, and 59,751 for AfCPR1, Af51A, and Af51B, respectively, including the six-histidine C-terminal extensions. E. coli membrane suspensions isolated from the pSPORT:Af51A, pSPORT:Af51A-G54W, pSPORT:Af51A-L98H, pSPORT:Af51A-M220K, and pSPORT:Af51B expression clones contained 16, 10, 4, 6, and 30 μM P450, respectively, as determined by reduced carbon monoxide difference spectroscopy.

Spectral properties of recombinant AfCPR1, Af51A, and Af51B proteins.

The AfCPR1 absolute spectrum (Fig. 1A) was typical for a native CPR isolated in the partially reduced, air-stable, blue semiquinone form (23) with a broad semiquinone peak between 580 and 600 nm in addition to the two main flavin peaks at 385 and 465 nm. This was reflected in the pale green color of AfCPR1 instead of the yellow color of the fully oxidized form previously obtained for truncated S. cerevisiae CPR (24). The spectral intensity of the second flavin peak (∼465 nm) of AfCPR1 fell during reduction with 4, 40, and 200 μM NADPH, as previously observed for H. sapiens CPR (23), Phanerochaete chrysosporium CPR (25), and C. albicans CPR (26). The blue semiquinone peak at ∼590 nm also fell in intensity at the higher NADPH concentrations of 40 and 200 μM (Fig. 1A) due to further reduction to the hydroquinone form (27). AfCPR1 obeyed Michaelis-Menten kinetics with respect to both cytochrome c and NADPH (see Fig. 2 in the supplemental material), yielding K_m values for cytochrome c and NADPH of 3.29 ± 0.32 μM and 0.57 ± 0.04 μM. The observed V_max turnover number was 28.5 ± 1.5 s⁻¹. These values were similar to those reported for other CPRs (24, 25, 28, 29).

FIG 1.

Spectral characterization of purified proteins. (A) The absolute spectrum of 5 μM AfCPR1 and absolute spectra of AfCPR1 reduced by 4, 40, and 200 μM NADPH were determined. (B) CYP51 absolute spectra of 2.5 μM purified Af51A (solid line) and Af51B (dashed line) were determined in the resting oxidized state; the heme Soret peak is highlighted (inset). The cuvette path length was 4.5 mm.

FIG 2.

Carbon monoxide difference spectroscopy of wild-type Af51A and Af51B. Sequential spectral determinations were made every 45 s after saturation with carbon monoxide followed by the addition of sodium dithionite to the sample cuvette (path length, 10 mm).

Absolute spectra of purified Af51A and Af51B (Fig. 1B) in the resting oxidized state were typical of cytochrome P450 proteins isolated predominantly in the ferric low-spin state (13). Af51A had a red-shifted heme Soret peak at 420 nm relative to 418 nm for Af51B (Fig. 1B inset). Af51A and Af51B had similar α, β, and δ spectral peaks at 570, 538, and 360 nm. The stability of the reduced carbon monoxide adduct for Af51A (Fig. 2) was poor for both the purified protein and E. coli membrane suspension, with the transient red-shifted heme Soret peak at ∼450 nm progressively declining over 3 min with a commensurate increase in the absorbance peak at 421 nm, indicative of the inactive P420 complex with carbon monoxide. However, for Af51B, the reduced carbon monoxide adduct was quickly formed (reaching a maximum after ∼2 min) and was stable for at least 10 min. The reduced carbon monoxide difference spectra for Af51A-G54W, Af51A-L98H, and Af51A-M220K membrane preparations (see Fig. S3 in the supplemental material) were more stable than that for Af51A.

Characterization of AfCPR1 as a redox partner of CYP51.

Using the CYP51 reconstitution assay (20), AfCPR1 readily supported CYP51 catalysis (Table 1) using C. albicans and H. sapiens CYP51 proteins. Velocities of 7.1 and 23.1 min⁻¹ for lanosterol with C. albicans CYP51 and H. sapiens CYP51 and 9.8 min⁻¹ for eburicol with C. albicans CYP51 indicated that AfCPR1 had been isolated in a fully functional state.

TABLE 1.

AfCPR1 support for CYP51 enzymes^a

Partner CYP51	Yield (nmol liter−1)	CYP51 turnover no. (min−1)
		Lanosterol	Eburicol
H. sapiens CYP51 proteinb	645c	23.1 ± 1.7	ND
C. albicans CYP51 proteinb	270c	7.1 ± 0.8	9.8 ± 0.5
Af51A proteinb	17d	0.034 ± 0.004	0.24 ± 0.03
Af51B proteinb	134d	0.06 ± 0.02	0.26 ± 0.04
Af51A membranes	63	Tre	2.14 ± 0.49
Af51B membranes	197	0.14 ± 0.03	2.54 ± 0.40
Af51A-G54W	65	ND	0.69 ± 0.21
Af51A-L98H	25	ND	4.44 ± 1.01
Af51A-M220K	31	ND	2.16 ± 0.58

^aAfCPR1 was used to complement CYP51 enzymes (2:1) in the CYP51 reconstitution assay for 1 μM CYP51 protein or 0.5 μM CYP51 membrane preparations. CYP51 concentrations were determined by reduced carbon monoxide difference spectroscopy. ND, not determined.

^bNi²⁺-NTA agarose-purified CYP51 protein.

^cFrom the work of Warrilow et al. (19).

^dFrom the work of Warrilow et al. (12).

^eSearching for the mass ion of TMS-derivatized C-14-demethylated lanosterol indicated that trace amounts of the product were present.

Purified Af51A and Af51B enzymes gave low turnover numbers with both eburicol (0.24 and 0.26 min⁻¹) and lanosterol (0.034 and 0.06 min⁻¹) using AfCPR1 as the redox partner (Table 1), albeit with a 4- to 7-fold catalytic preference for eburicol over lanosterol. Using Af51A and Af51B membrane preparations resulted in 10-fold-higher CYP51 turnover numbers and confirmed the substrate preference for eburicol (Table 1). This suggested that cholate solubilization and subsequent purification of Af51A and Af51B had subtly altered the enzyme structure (possibly by cholate adhesion to the apoprotein), resulting in lower-than-expected catalytic activities. The preference for eburicol over lanosterol reflects the higher levels seen in whole-cell sterols of A. fumigatus (30), with lanosterol being converted to eburicol through C-24 methylation catalyzed by an S-adenosylmethionine-sterol-C-methyltransferase. Using Emulgen 913 or Tween 80 as an alternate detergent to solubilize Af51A from E. coli membranes followed by purification on Ni²⁺-NTA agarose with 2% (wt/vol) Emulgen 913 to elute Af51A still resulted in purified enzyme samples that displayed 10- to 15-fold-lower catalytic activity than Af51A membrane suspensions (data not shown). Therefore, Af51A and Af51B appear most catalytically efficient in the E. coli membrane fraction isolated from the expression clones. For this reason, E. coli membrane suspensions of Af51A, Af51B, Af51A-G54W, Af51A-L98H, and Af51A-M220K were used for comparative IC₅₀ determinations with fluconazole, voriconazole, itraconazole, and posaconazole. However, one disadvantage of using membrane CYP51 preparations was that the increased turbidity of the solutions prevented reproducible azole titration curve binding spectra from being obtained that were accurate enough to determine reliable K_d (dissociation constant) values for each azole.

The three Af51A point mutations did not cause dramatic changes in the catalytic efficiency of the Af51A enzyme. Eburicol turnover with Af51A-G54W was only 3-fold lower than that of Af51A, and Af51A-L98H turnover was only twice that of Af51A, while Af51A-M220K had a turnover similar to that of the wild-type enzyme (Table 1).

Azole IC₅₀ determinations.

IC₅₀s for fluconazole were determined with Af51A and Af51B (Fig. 3A). Af51B was sensitive to fluconazole (IC₅₀ = 0.5 μM), while Af51A was relatively insensitive to fluconazole, with a 34-fold-higher IC₅₀ of 17 μM, confirming the previous observation that Af51A and not Af51B confers fluconazole tolerance (31). This is in contrast to only a 3-fold difference in K_d values for fluconazole between Af51A (K_d ≈ 12 μM) and Af51B (K_d ≈ 4 μM) determined using Ni²⁺-NTA-purified proteins (12), suggesting that detergent solubilization of Af51A and Af51B may have altered the catalytic properties of the enzymes. IC₅₀ determinations for the Af51A point mutations G54W, L98H, and M220K were limited to voriconazole, itraconazole, and posaconazole because wild-type Af51A was relatively insensitive to fluconazole.

FIG 3.

Azole IC₅₀ determinations. CYP51 reconstitution assays were performed using a CYP51-to-AfCPR1 ratio of 1:2 for 0.5 μM CYP51 membrane preparations for Af51A (●), Af51B (○), Af51A-G54W (▲), Af51A-L98H (×), and Af51A-M220K (⊙). Azoles used included fluconazole (A), voriconazole (B), itraconazole (C), and posaconazole (D). Values are means and standard deviations (see Table 1 for actual CYP51 turnover numbers).

The observed IC₅₀s for voriconazole, itraconazole, and posaconazole were similar for both Af51A and Af51B membrane suspensions (Table 2), suggesting little difference in binding affinities for these three azoles. It was shown previously that purified wild-type Af51A has 8-, 33-, and 37-fold-higher K_d values (lower affinity) for voriconazole, itraconazole, and posaconazole than purified wild-type Af51B (12). Therefore, detergent solubilization and purification of these two CYP51 enzymes (especially Af51A) may have altered the azole binding and catalytic properties.

TABLE 2.

Azole IC₅₀ determinations for A. fumigatus CYP51 enzymes^a

CYP51	IC50 (μM)
	Fluconazole	Voriconazole	Itraconazole	Posaconazole
Af51A	17	0.38	0.21	0.19
Af51B	0.50	0.33	0.16	0.21
Af51A-G54W	ND	0.80	2.28	5.78
Af51A-L98H	ND	0.65	0.34	0.24
Af51A-M220K	ND	0.84	3.07	1.50

^aAfCPR1 was used to complement CYP51 (2:1) in the CYP51 reconstitution assay containing 0.5 μM CYP51 membrane preparations. ND, not determined.

The Af51A-G54W IC₅₀s for itraconazole and posaconazole (Fig. 3) were 11- and 30-fold higher than those for wild-type Af51A, indicative of significantly lower binding affinities for itraconazole and posaconazole and suggesting reduced susceptibility to these two azole antifungals. In addition, significant residual CYP51 activities of 13 and 18% were apparent for Af51A-G54W in the presence of 32 μM itraconazole and posaconazole. In contrast, susceptibility to voriconazole was only marginally reduced, with just a 2-fold increase in IC₅₀ (Table 2).

Likewise, the Af51A-M220K IC₅₀s for itraconazole and posaconazole (Fig. 3) were 15- and 8-fold higher than that for wild-type Af51A, indicative of substantially lower binding affinities for itraconazole and posaconazole and suggesting reduced susceptibility to these two azoles. Again, significant residual CYP51 activities of 9 and 6% were observed at 32 μM itraconazole and posaconazole. Susceptibility to voriconazole was only marginally reduced, with a 2-fold increase in IC₅₀ (Table 2).

The IC₅₀s of Af51A-L98H (no TR₃₄) for voriconazole, itraconazole, and posaconazole were only marginally higher (1.7-, 1.6-, and 1.3-fold) than that of the wild-type Af51A (Table 2), indicating little change in the binding affinity for these three azoles. However, the L98H mutation retained significant residual CYP51 activity at 32 μM triazole of 6.9%, 7.6%, and 5.4% for voriconazole, itraconazole, and posaconazole. These residual CYP51 activities at high azole concentrations could contribute to the reduction of azole susceptibility of A. fumigatus strains that harbor the Af51A-L98H mutation.

DISCUSSION

Our original aim was to express and purify both AfCPR1 and AfCPR2 and compare their catalytic properties as redox partners of Af51B, Af51A, and Af51A point mutations associated with triazole resistance, in addition to developing an in vitro AfCPR/Af51 assay system for screening the effectiveness of azole-based drug candidates as CYP51 inhibitors. In animals and yeasts, a single CPR interacts with and reduces the wide and divergent range of CYPs present within each organism. However, the occurrence of multiple CPR isoenzymes in some plant and fungal species may reflect the diversity of CYP enzymes and their roles in primary and secondary metabolism (32–34), with multiple CPR isoenzymes being associated with organisms with a high number of CYP enzymes. Previous phylogenetic analysis suggested that multiple independent CPR duplication events occurred in fungi to give rise to the present CPR isoenzymes (34). The sequence diversity of AfCPR1 and AfCPR2 also suggests that the two AfCPR isoenzymes did not arise as a result of a relatively recent gene duplication event and that these two isoenzymes may have different in vivo functions. Differential regulation of CPR isoenzymes has been observed in the plants arabidopsis (35) and parsley (36). Failure to express AfCPR2 in E. coli was disappointing, as this prevented a detailed comparison of the catalytic properties of AfCPR1 and AfCPR2. Instead, a CYP51 reconstitution assay system was developed utilizing AfCPR1 as the redox partner of Af51A and Af51B to investigate the effectiveness of azole antifungals at inhibiting the 14α-demethylation of eburicol.

Af51A point mutations often associated with azole resistance in A. fumigatus include those at positions G54, L98, and M220 (4). Point mutations at G54 are associated with reduced susceptibility toward itraconazole and posaconazole without loss of susceptibility to voriconazole (37), reflected in significantly increased MICs for itraconazole and posaconazole (Table 3) (38, 39). The itraconazole/posaconazole resistance phenotype has been verified by site-directed mutagenesis (40). IC₅₀ determinations for Af51A-G54W confirm the observed azole resistance phenotype with 11- and 30-fold-increased IC₅₀s for itraconazole and posaconazole (Table 2) over that of the wild-type enzyme and significant residual CYP51 activity at high azole concentrations. Susceptibility to voriconazole, however, was only marginally reduced. Point mutations at M220 are associated with reduced susceptibility toward itraconazole and posaconazole and variable susceptibility toward voriconazole (37), reflected in the observed MICs (Table 3). The M220K mutation caused a 4-fold reduction in voriconazole susceptibility and 128- and 516-fold reductions in itraconazole and posaconazole susceptibilities compared to those of azole-sensitive A. fumigatus strains that contain no Af51A mutations (39). The azole resistance phenotype has been confirmed by site-directed mutagenesis (41). IC₅₀ determinations with Af51A-M220K confirmed the small reduction in voriconazole susceptibility and the large reduction in itraconazole and posaconazole susceptibilities observed by 15- and 8-fold increases in IC₅₀s over the wild-type Af51A enzyme (Table 2) accompanied by significant residual CYP51 activity at high azole concentrations. Complementation studies in S. cerevisiae confirmed that G54W and M220K point mutations in Af51A reduced susceptibility to itraconazole and posaconazole (42) (Table 3) with relatively small reductions in voriconazole susceptibility.

TABLE 3.

Azole MICs for Af51A mutations in A. fumigatus and S. cerevisiae

Strain	Af51A	MIC (μg ml−1)			Reference
		Voriconazole	Itraconazole	Posaconazole
A. fumigatus
CM237	Wild type	0.5	0.25	0.06	42
akuBKU80	Wild type	0.25	0.5	0.25	44
CM2266	G54W	0.71	>8	16	38
Clinical isolate	G54W	0.25	>16	>16	39
CM237-TRL98H	TR34/L98H	4	8	0.5–1	10
akuBKU80-TR	TR34	2	1	0.5	44
akuBKU80-L98H	L98H	1	1	0.5	44
akuBKU80-TRL98H	TR34/L98H	2	>16	0.5	44
Clinical isolates (n = 86)	TR34/L98H	2–16	4–>16	0.25–1	39
CM2159	M220K	1.2	>8	2	41
Clinical isolate	M220K	2	>16	>16	39
S. cerevisiaea
DSY3961	Wild type	0.000154	0.000875	0.000475	42
DSY3961	G54W	0.000126	3.40	1.85	42
DSY3961	M220K	0.00269	2.51	0.216	42

^aComplementation studies were carried out in S. cerevisiae DSY3961 using pYES2/CT:Af51A constructs.

The L98H point mutation accompanied by a 34-bp tandem repeat in the Af51A promoter is associated with reduced susceptibility to voriconazole, itraconazole, and posaconazole (multiazole resistance) (6, 43), resulting in increased MICs (Table 3). However, site-directed mutagenesis studies have shown that the full multiazole-resistant phenotype can be obtained only if both the 34-bp tandem repeat and the L98H mutation are present (44) (Table 3). The L98H mutation on its own caused only 2- to 4-fold reductions in azole susceptibility (44). IC₅₀ determinations with Af51A-L98H confirmed that the L98H mutation on its own yielded only marginal reductions in azole susceptibility (1.3- to 1.7-fold increases in IC₅₀s) compared to the wild-type Af51A enzyme (Table 2). However, L98H conferred 5 to 8% residual CYP51 activity at high azole concentrations, which could contribute to azole resistance in vivo. Snelders et al. (44) found that reintroduction of the tandem repeat (TR₃₄) caused a further 16-fold reduction in itraconazole susceptibility but no further reductions in voriconazole or posaconazole susceptibility. The TR₃₄ tandem repeat modification of the Af51A promoter region also caused a 5- to 8-fold increase in Af51A expression levels (10, 45). Therefore, the reduced azole susceptibility phenotype of TR₃₄/L98H appears to be a combination of residual CYP51 activity at high azole concentrations and overexpression of Af51A-L98H. Even though all A. fumigatus strains that harbor the Af51A TR₃₄/L98H mutation display reduced susceptibilities to voriconazole, itraconazole, and posaconazole, the actual sensitivities displayed vary quite widely between individual strains (39). Susceptibility to voriconazole was 4- to 32-fold lower, that to itraconazole was 32- to 128-fold lower, and that to posaconazole was 8- to 32-fold lower than that of azole-sensitive strains that did not harbor an Af51A point mutation, suggesting that additional azole resistance mechanisms may also be present in some of the TR₃₄/L98H strains.

In silico modeling studies of Af51A (37, 44) have provided possible explanations for the observed azole resistance phenotypes exhibited by mutations at G54, L98, and M220. Mutations at G54 and M220 are located in the immediate vicinity of the opening of one of the ligand access channels and are likely to have a direct effect on the docking of azole molecules, especially those with long side chains that interact with amino acid residues lining the access channel, such as itraconazole and posaconazole. This would explain the observed higher IC₅₀s (lower azole binding affinities) for itraconazole and posaconazole with Af51A-G54W and Af51A-M220K. The TR₃₄/L98H mutation appears to have a more indirect effect with the Leu-98-His change altering the flexibility of the BC and IH loops of the Af51A protein causing the displacement of Tyr-121 and Tyr-107 side chains important for triazole binding so that the diameter of both ligand access channels are narrowed and preventing optimal azole ligand binding. Our Af51A-L98H IC₅₀ results suggest that the L98H mutation only marginally weakens the binding affinities for voriconazole, itraconazole, and posaconazole. However, at high azole concentrations, Af51A-L98H retains 5 to 8% residual CYP51 activity compared with the wild-type Af51A enzyme, which could confer reduced azole susceptibility on A. fumigatus strains that harbor the Af51A-L98H mutation.

In summary, we have successfully expressed A. fumigatus CPR1 in E. coli and isolated the purified protein in a fully functional (native) form that catalyzed the 14α-demethylation of lanosterol and eburicol in conjunction with four different CYP51 enzymes and envisage AfCPR1 being a useful redox partner for other filamentous fungi and yeast CYP51 and other CYP enzymes in future catalysis and IC₅₀ studies. We have successfully utilized AfCPR1 as a redox partner with Af51A in CYP51 reconstitution assays to screen the effectiveness of triazole antifungals against Af51A mutant isoforms, demonstrating that the Af51A point mutations G54W, L98H, and M220K confer azole tolerance during CYP51 catalysis consistent with their previously reported azole susceptibility phenotypes. Clinical applications include assessing the likely impact of new emergent Af51A mutations, such as TR₄₆/Y121F/T289A (8), on the azole susceptibility of A. fumigatus strains that harbor the mutations. Also, the AfCPR1/Af51A reconstitution assay system will facilitate preliminary in vitro screening and assessment of next-generation azole-based drug candidates against azole-resistant CYP51A isoforms.

The frequency of azole-resistant A. fumigatus isolates without Af51A mutations has increased significantly in recent years (6, 46). Therefore, azole resistance among A. fumigatus clinical strains cannot be effectively addressed without a full understanding of the non-CYP51-mediated azole resistance mechanisms in tandem with Af51A- and Af51B-mediated azole resistance. We are presently optimizing the protein isolation and purification procedures to overcome the reduced/impaired catalytic function observed with the pure Af51A and Af51B enzymes in preparation for a more comprehensive study of the biochemical basis for CYP51A- and CYP51B-mediated azole resistance in A. fumigatus.