Sterol content analysis suggests altered eburicol 14α-demethylase (CYP51) activity in isolates of Mycosphaerella graminicola adapted to azole fungicides
Correspondence: Hans J. Cools, Department of Plant Pathology and Microbiology, Centre for Sustainable Pest and Disease Management, Rothamsted Research, Harpenden, Hertfordshire AL5 2JQ, UK. Tel.: +44 1582 763 133; fax: +44 1582 760 981; e-mail: firstname.lastname@example.org
The recent decline in the effectiveness of some azole fungicides in controlling the wheat pathogen Mycosphaerella graminicola has been associated with mutations in the CYP51 gene encoding the azole target, the eburicol 14α-demethylase (CYP51), an essential enzyme of the ergosterol biosynthesis pathway. In this study, analysis of the sterol content of M. graminicola isolates carrying different variants of the CYP51 gene has revealed quantitative differences in sterol intermediates, particularly the CYP51 substrate eburicol. Together with CYP51 gene expression studies, these data suggest that mutations in the CYP51 gene impact on the activity of the CYP51 protein.
Azole (imidazole and triazole) fungicides interrupt biosynthesis of ergosterol, an essential sterol of fungal cell membranes, by inhibiting the cytochrome P450 eburicol 14α-demethylase (CYP51) (Yoshida & Aoyama, 1987). This inhibition prevents demethylation of eburicol, the principal CYP51 substrate in most filamentous fungi, resulting in a depletion of ergosterol and a concomitant accumulation of nonfunctional 14α-methyl sterols, including eburicol. The accumulation of 14α-methyl sterols and the depletion of ergosterol itself, perturbs membrane function and prevents further fungal growth.
Azoles have been the leading agents for the control of fungal pathogens of humans, animals and plants for over 30 years. Resistance to this class of chemistry is now common in human pathogens, including Candida spp. (Kontoyiannis & Lewis, 2002) and Aspergillus fumigatus (Warnock et al., 1999), and is a serious problem in clinical antifungal therapy. As a consequence, the molecular basis of azole resistance is best understood in these pathogens. Common mechanisms of azole resistance in Candida albicans are alteration of the CYP51 target protein (Sanglard et al., 1998; Marichal et al., 1999) and enhanced efflux of azole outside the cell mediated by the over expression of energy-dependent efflux proteins (Sanglard et al., 1995; Perea et al., 2001). Altered ergosterol biosynthesis, although a less common mechanism of azole resistance, has been reported for C. albicans (Nolte et al., 1997). These isolates, cross resistant to fluconazole and amphotericin B, have reduced levels of ergosterol and accumulate sterols lacking a C5(6) double bond, a result of null mutations in the gene (erg3) encoding the sterol C5(6) desaturase (Kelly et al., 1995, 1996). However, these isolates grew poorly in vivo (Nolte et al., 1997), probably as a result of using 14α-methylated sterols rather than ergosterol in cell membrane synthesis.
Azole resistance in plant pathogens has developed more slowly, and rarely leads to a loss of field efficacy. Recently, however, there has been a notable decline in the effectiveness of a number of azoles in controlling Mycosphaerella graminicola so that some compounds no longer control the disease (Clark, 2006). This fungus causes Septoria leaf blotch, the most economically important foliar disease of winter wheat in North West Europe. Studies suggest alterations in the target protein, caused by mutations in the CYP51 gene, are responsible for this recent decline in sensitivity in M. graminicola populations (Cools & Fraaije, 2008). Some CYP51 substitutions identified in M. graminicola isolates, less sensitive to azoles, are at equivalent residues to those found in resistant strains of C. albicans, for example Y137F (Cools et al., 2005; Leroux et al., 2007), whereas others, including V136A (Leroux et al., 2007), A379G (Leroux et al., 2007) and I381V (Cools et al., 2005; Fraaije et al., 2007; Leroux et al., 2007) are unique to M. graminicola.
While studies have demonstrated clear correlations between CYP51 alterations and altered azole sensitivity in M. graminicola (Fraaije et al., 2007; Leroux et al., 2007), the effects of these changes on azole affinity or CYP51 activity have yet to be elucidated. In this study, we established the impact of alterations in the CYP51 protein of M. graminicola on sterol biosynthesis by analysing the sterol content of M. graminicola field isolates carrying different variants of the CYP51 gene.
Materials and methods
Culturing of M. graminicola isolates
Isolates (Table 1) were grown in triplicate on yeast peptone dextrose (YPD) agar (ForMedium, Norwich, UK) at 15 °C for 7 days. Spores from each culture were harvested and diluted to a concentration of 4.5 × 105 spores mL−1 in 100 mL YPD broth (ForMedium) and incubated for 72 h at 21 °C, with shaking at 220 r.p.m. Replicate fungal cultures were collected by vacuum filtration through 8-μM pore nitrocellulose filters (Millipore, Billerica), snap frozen in liquid nitrogen and freeze dried before free sterol and RNA extractions.
Table 1. Isolates of Mycosphaerella graminicola used for sterol content analysis
|IPO323||North Brabant, the Netherlands; 1981||0.071||0.491||0.382||None (wild type)|
|Berdun 17||Aragon, Spain; 2006||0.066||0.319||0.097||None (wild type)|
|R6-31||Hertfordshire, UK; 2006||0.958||1.36||1.86||Y137F|
|Berdun 13||Aragon, Spain; 2006||ND||ND||ND||Y137F|
|Berdun 10||Aragon, Spain; 2006||0.081||0.273||0.121||L50S|
|Berdun 4||Aragon, Spain; 2006||0.121||0.562||0.293||G460D|
|V103.9||Kent, UK; 2006||1.26||0.492||2.19||L50S, V136A, Y461H|
|V115.3||Kent, UK; 2006||0.817||4.12||0.704||L50S, I381V, Y461H|
|V209.10||Kent, UK; 2006||1.825||5.99||0.657||L50S, I381V, Y461H|
|V209.9||Kent, UK; 2006||0.735||0.712||2.48||V136A, S188N, ΔY459/G460|
|V212.2||Kent, UK; 2006||1.28||4.41||2.58||V136A, S188N, ΔY459/G460|
|V214.7||Kent, UK; 2006||1.66||9.17||0.107||L50S, S188N, A379G, I381V, ΔY459/G460, N513K|
|V202.1||Kent, UK; 2006||1.05||10.1||0.104||L50S, S188N, A379G, I381V, ΔY459/G460, N513K|
Azole sensitivity testing
Sensitivity assays were modified from Fraaije et al. (2007). A 100-μL aliquot of 2 × potato dextrose broth (PDB; Sigma-Aldrich Inc., St. Louis) amended with decreasing concentrations of epoxiconazole (50, 16.7, 5.6, 1.9, 0.62, 0.21, 0.069, 0.023, 0.008, 0.0025 and 0.0008 mg L−1), tebuconazole (50, 20, 8, 3.2, 1.3, 0.51, 0.2, 0.082, 0.033, 0.013 and 0.0052 mg L−1) or prochloraz (10, 3.3, 1.1, 0.37, 0.12, 0.041, 0.014, 0.0046, 0.0015, 0.0005 and 0.00017 mg L−1) was added to wells of flat-bottomed microtitre plates (TPP 92696 test plates, Trasadingen, Switzerland). After 7 days of growth at 15 °C on YPD to ensure yeast-like growth, isolates were suspended in 5 mL of sterile distilled water. Aliquots of 100 μL of isolate spore suspensions (1 × 105 spores mL−1) were added to each well. Plates were incubated for 4 days at 23 °C, and growth measured at A630 nm using a Fluorstar Optima microplate reader (BMG Labtech GmbH, Offenberg, Germany). Fungicide sensitivities were determined as 50% effective concentration (EC50) using a dose–response relationship.
CYP51 genes of M. graminicola isolates IPO323, Berdun 17, R6-31, Berdun 13, Berdun 10, Berdun 4, V103.9, V115.3, V209.10, V209.9, V212.2, V214.7 and V202.1 (Table 1; NCBI GenBank accessions: FJ196446–FJ196457) were amplified and sequenced using primers, PCR conditions and protocols outlined in Cools et al. (2005). Amplifications were carried out using Phusion high-fidelity DNA polymerase to ensure the validity of sequences (Finnzymes, Espoo, Finland).
Total sterol extraction and derivatization
Sterol extraction and analysis was adapted from Hey et al. (2006). Fifty milligrams of lyophilized tissue from each biological replicate was extracted with 5 mL of 2 : 1 (v/v) chloroform : methanol containing 50 μg dihydrocholesterol (internal standard) at 55 °C, for 30 min. The cooled mixture was filtered and the solvent removed under a stream of gaseous nitrogen. To the lipid residue obtained, 1 mL toluene and 2 mL sodium methoxide (0.5 M) in methanol were added and the sample heated at 60 °C for 30 min. Once cooled, 2 mL of 14% boron trifluoride in methanol was added and the sample heated at 60 °C for a further 10 min. Once cooled, 2–3 mL diethyl ether and 5 mL deionized water were added. The ether layer was removed to a clean vial and the aqueous phase re-extracted with a further 2–3 mL diethyl ether and both extracts combined. After backwashing with 5 mL water, the organic phase was then dried, overnight, over anhydrous sodium sulphate and filtered through glass wool. The filtrate was collected in a fresh vial and the solvent was removed under a stream of gaseous nitrogen. The dry residue was resuspended in 300 μL toluene and silylated with 200 μL N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA) at 50 °C for 10 min. All samples (isolates/treatment combinations) were prepared and analysed in triplicate. Reagents were obtained from Sigma-Aldrich (Gillingham, Dorset, UK) unless otherwise stated.
GC-MS analysis was carried out on a Hewlett Packard (Palo Alto, CA) 5890 Series Gas Chromatograph coupled to a 5970 Series Mass Selective Detector and fitted with 7373 automatic liquid sampler and Zebron ZB5 column with 5 m Guardian (Phenomenex, Macclesfield, UK; 30 m × 0.25 mm internal diameter × 0.25 μm film). One-microlitre splitless injections were given at an inlet temperature of 280 °C and 105 kPa head pressure. The oven was set to an initial temperature of 100 °C for 2 min), then ramped at 10 °C min−1 to 320 °C and held for 6 min. The GC interface was 290 °C and electron ionization+mass spectra, 41–750 amu, were acquired from 13 to 30 min, with the threshold set to 150 and a 1.15 scan s−1 scan rate. Data analysis was conducted in the QuanLynx module of MassLynx (Waters, Manchester, UK), following conversion of the data files from their native (HP) format to Masslynx files using the mass transit file conversion utility program (Palisade, Newfield, NY). Peak areas were normalized to the internal standard.
To eliminate the contribution from coeluting hydrocarbons, reconstituted total ion chromatogram (TIC) peak areas were calculated by dividing the integrated extracted ion chromatogram peak areas of individual sterols by the quantification of the ion's contribution to the TIC, allowing direct comparison of the relative concentrations, within and between samples, to be made (equal reconstituted peak areas representing approximately equal concentrations).
For ergosterol, lanosterol, cholesterol and dihydrocholesterol, identifications were confirmed using authentic standards. For all other sterols, putative identifications were made on the basis of literature relative retention times combined with matching the experimentally obtained spectra to those in the National Institute of Standards and Technology (NIST) Spectral Library.
Levels of CYP51 transcript were measured with real-time reverse transcription (RT)-PCR. Roughly 30-mg powdered freeze dried fungal material from each biological replicate was placed in a 2-mL screw top tube with a 3 mm stainless steel ball and run through a FastPrep shaker at 4.5 ms−1 for 30 s (FP120, Bio101/Savant, MPBiomedicals). One millilitre of TRIZOL reagent (Invitrogen, Paisley, UK) was then added and run through the FastPrep once again. TRIZOL was used according to the manufacturer's protocol, and further RNA purification was achieved with an overnight 4-M lithium chloride precipitation.
Reverse transcription of the first strand of cDNA was carried out with the Superscript III RT-PCR kit (Invitrogen) following the manufacturer's protocol using oligo d(T) primers and 5 μg total RNA per reaction. A one in 10 dilution of this reaction was used as the template for the real-time PCR analysis.
Real-time PCR reactions were carried out with Platinum SYBR green PCR mix with ROX (Invitrogen) and a modified version of the supplier's protocol, using 5 μL cDNA solution in a 25-μL reaction and CYP51-specific primers from Cools et al. (2007). The cycling conditions used were 50 °C for 2 min, 95 °C for 2 min, and then 40 replicates of 95 °C for 15 s, 58 °C for 30 s and 72 °C for 36 s. Each reaction was carried out in triplicate, in the ABI PRISM 7500 sequence detection system (Applied Biosystems, Foster City, CA). Relative quantification values were calculated with the method using the constitutively expressed housekeeping β-tubulin gene as an endogenous control, and gene expression levels in the ‘wild-type’ isolate IPO323 untreated sample were used as calibrators. Transcript abundance was measured in triplicate for each isolate.
A linear model and anova were used to estimate the effect of each amino acid alteration on the relative efficiency of the demethylation step calculated as the logarithm of the product of eburicol content (E) and CYP51 transcript abundance (C). Not all combinations of alterations were present. Therefore, it was possible to ascribe the variation in calculated enzyme efficiency to alterations in more than one way. The regression fit provided by all subsets of the mutations was used to determine economical descriptions of the data by effects of individual mutations. Model selection took into account Mallow's Cp, the proportion of the variance accounted for by the model and F-tests of the effect of deleting a factor from the model. Calculations were performed in genstat v10 (VSN International, Oxford).
Isolate azole sensitivities
Correlations between azole sensitivities and CYP51 alterations were consistent with previous studies (Cools et al., 2005; Fraaije et al., 2007; Leroux et al., 2007). Isolates carrying substitution I381V were resistant to tebuconazole and considered sensitive to prochloraz, whereas isolates carrying substitution V136A were resistant to prochloraz and, with the exception of V212.2, more sensitive than isolates carrying I381V to tebuconazole. EC50 values for isolate R6-31, carrying substitution Y137F, were higher to all azoles than for wild-type isolates or those carrying L50S or G460D alone (Table 1).
Identification of sterols
Ergosterol, lanosterol, cholesterol and dihydrocholesterol were identified with authentic standards. Ergosta-7,22-dienol, fungisterol, epiergosterol, 4,4-dimethylfecosterol and eburicol were identified according to the retention time and mass spectra data published or present in the NIST MS database (Table 2). However, due to lack of comparisons, other sterols could not be identified with certainty. These are designated as unknown sterols. It was possible to identify unknown sterols 1 and 2 as desmethyl sterols through MS (data not shown). Mass spectra of ergosterol and its stereoisomer have closely related fragmentation therefore the stereochemistry of the isomer is unknown. Studies by Joseph-Horne et al. (1996) of M. graminicola predicted the isomer to have an α-hydroxy moiety at C-3 rather than the usual β-hydroxy, and as no further evidence can be presented here, the same tentative designation is maintained.
Table 2. Designation of sterols by relative retention times and authentic samples
| ||R2||0.9991||0.9993||0.9962|| |
Sterol content related to CYP51 sequence
Quantitative differences in the level of different sterols were detected, particularly eburicol levels, although these could not be directly related to predicted CYP51 amino acid sequence. There were no significant differences in ergosterol content between isolates (P=0.05), suggesting that all isolates are able to produce functional levels of ergosterol in vitro (Table 3).
Table 3. Sterol content of Mycosphaerella graminicola isolates
|Unknown sterol 1||0.018||0.042*||0.029||0.016||0.046*||0.019||0.019||0.023||0.017||0.018||0.019||0.018||0.015|
|Unknown sterol 2||0.047*||0.041||0.029||0.050*||0.032||0.020*||0.02||0.029||0.024||0.018*||0.023||0.051*||0.033|
There were no differences in the type of sterols present, suggesting that sterol biosynthesis is similar in each of these isolates (Table 3). As in earlier studies by Joseph-Horne et al. (1996), 14α-methylfecosterol or 14α-methyl-3,6-diol were not detected in any isolate.
CYP51 expression and relationship with eburicol content
Expression of CYP51 was analysed from the same samples as sterol content. CYP51 transcript was detected in all isolates, with the lowest level in IPO323 to which relative expression in all other isolates was calibrated. The highest levels of CYP51 transcript were detected in isolates V103.9 and V209.9 (Table 4). CYP51 expression levels could not be directly linked to eburicol levels (Table 4).
Table 4. Grouping of Mycosphaerella graminicola isolates according to the CYP51 efficiency measure, 1/C×E
|Berdun 17||None||3.50E−03||2.59 (± 2.75)||110.23|
|IPO323||None||0.034||1 (± 0.10)||29.02|
|Berdun 10||L50S||4.30E−03||3.54 (± 0.71)||65.61|
|Berdun 4||G460D||0.042||1.48 (± 0.23)||16.21|
|Berdun 13||Y137F||0.256||1.34 (± 0.58)||2.91|
|R6-31||Y137F||0.067||5.12 (± 0.84)||2.93|
|V115.3||L50S, I381V, Y461H||0.195||4.37 (± 0.96)||1.17|
|V209.10||L50S, I381V, Y461H||0.036||15.71 (± 3.33)||1.77|
|V202.1||L50S, S188N, A379G, I381V, ΔY459/G460, N513K||0.06||6.78 (± 2.80)||2.46|
|V214.7||L50S, S188N, A379G, I381V, ΔY459/G460, N513K||0.017||15.08 (± 2.01)||3.91|
|V103.9||L50S, V136A, Y461H||0.079||23.07 (± 12.53)||0.55|
|V209.9||V136A, S188N, ΔY459/G460||0.149||26.2 (± 8.15)||0.26|
|V212.2||V136A, S188N, ΔY459/G460||0.278||5.84 (± 0.57)||0.62|
Mycosphaerella graminicola isolates grouped according to their CYP51 variant efficiency 1/C×E (where C=CYP51 expression, E=eburicol content). Highest values were obtained for the wild-type isolates Berdun 17 and IPO323, and isolates carrying single alterations L50S (Berdun 10) and G460D (Berdun 4). Lowest values were obtained for isolates carrying the CYP51 alteration, V136A (isolates V103.9, V209.9 and V212.2), which had high levels of both CYP51 gene expression and eburicol content (Table 4). SE of CYP51 expression was large in two of the isolates (V103.9 and Berdun 17), but this did not affect the overall 1/C×E classifications.
Statistical analysis of the impact of CYP51 alterations
Three alterations I381V, V136A and Y137F together explained 44% of the variation in the log of eburicol content significant at P=0.08, 0.01 and 0.03, respectively. I381V and V136A together explained 61% of the variation in log transcript abundance, with individual significance at P=0.07 and 0.002. Y137F, I381V and V136A also accounted for 89.3% of the variance for 1/C×E data. All three terms were significant at P<0.001 if dropped individually from the model, indicating that changes in eburicol demethylation efficiency are mainly due to these substitutions. There was no statistical evidence that any of the alterations present compensate for one another. No nonadditive effects of protein alterations on this efficiency measure were observed.
Total lipid extraction and derivatization followed by GC-MS was used to analyse the sterol content of M. graminicola isolates carrying different variants of the CYP51 gene predicted to encode amino acid alterations associated with altered azole sensitivity. Previous experiments (data not shown) had shown a significant increase in eburicol, the substrate of CYP51, in isolates treated with azole, confirming that azole fungicides inhibit eburicol demethylation in M. graminicola.
It was hypothesized that CYP51 changes may affect intrinsic enzyme activity, resulting in eburicol accumulation. The sterol content of 13 M. graminicola isolates with a total of eight different combinations of CYP51 alterations was analysed. Quantitative variations in sterol content between isolates were detected, but their differences did not appear to be related to CYP51 amino acid sequence. Specifically, increased eburicol content could not be directly correlated with CYP51 alterations and there was as much variation in eburicol content between isolates carrying the same CYP51 variant as between isolates with different CYP51s. Azole resistance can be caused by CYP51 overexpression, the most notable case being in azole-resistant Penicillium digitatum isolates, which constitutively overexpressed CYP51 100-fold (Hamamoto et al., 2000). However, in this study, the observed variation in CYP51 expression was considerably lower and could not be directly correlated with either azole sensitivity or CYP51 sequence, evidence that CYP51 sequence, not expression level, is the most important factor in determining azole resistance in M. graminicola.
Using a measurement of eburicol demethylation efficiency, based on a combined measure of CYP51 expression levels and eburicol content (1/C×E), isolates with the same CYP51 grouped together. Furthermore, statistical analyses revealed CYP51 substitution V136A had the largest effect on the measure of efficiency, with substantial effects also produced by I381V and Y137F. Those variants with substitutions L50S and G460D appear to encode proteins with unaltered efficiency. This is consistent with the impact of these amino acid changes on azole sensitivity and the location of these alterations in the predicted protein (Cools & Fraaije, 2008). Although these data do not clarify the effects of the CYP51 alterations on the biochemical properties of the enzyme, they clearly indicate that changes in activity have occurred. 1/C×E also suggests that the accumulation of eburicol, caused by reduced enzyme activity, can be accommodated by increased gene expression.
It is possible that CYP51 amino acid alterations selected by azole treatment are reducing the intrinsic activity of CYP51 particularly as accumulated eburicol does not appear to be converted to the toxic sterol intermediate 14α-methyl-3,6-diol in M. graminicola. Studies of Ustilago maydis, also detected higher levels of 14α-methylated sterols, including eburicol, in isolates with reduced azole sensitivity with normal levels of ergosterol, suggesting leaky CYP51 activity (Joseph-Horne et al., 1995). Further studies are underway to determine the precise impact of individual CYP51 amino acid alterations on intrinsic enzyme activity and azole binding.
Work carried out by T.P.B. was part of a CASE postgraduate studentship funded by the Biotechnology and Biological Sciences Research Council (BBSRC) of the United Kingdom, BASF and the Home Grown Cereals Authority (HGCA). Part of this work was supported by BBSRC project BB/E02257X/1. Rothamsted Research receives grant-aided support from the BBSRC. The authors would like to thank Sandy Hey for technical assistance and Kim Hammond-Kosack for critical reading of the manuscript.