The information of fungicides, including the active ingredients, the distributors, and the concentrations used in this study are listed in Supplementary Table S1. For bioassays, a stock solution of 10 g/liter active ingredient of each fungicide was made in dimethyl sulfoxide (DMSO) and stored at 4°C. The stock solutions were added to molten PDA at various concentrations (Supplementary Table S1) after the medium was cooled to ≈50°C. The final concentration of the solvent DMSO was 0.1% in the medium.

A total of 112 C. truncatum isolates were collected in 2011, 2012, and 2013 from 13 regions in China in a previous study (Diao et al., 2015); Background information of the isolates is provided in Supplementary Table S2. Based on both morphological characteristics and phylogenetic analysis, all 112 isolates were previously determined to be C. truncatum (Damm et al., 2009; Diao et al., 2015). Sensitivity of the 112 C. truncatum field isolates to prochloraz, epoxiconazole, difenoconazole, tebuconazole, and myclobutanil was determined by calculating the 50% inhibitory dose (EC₅₀ value, μg/ml) on fungicide-amended PDA in Petri plates (Supplementary Table S1) as previously reported (Russell, 2004; Fan et al., 2014). Plates were incubated at 28°C for 4 days in darkness. Each combination of isolate and fungicide was represented by three replicate plates, and the experiment was performed twice.

All the 112 field isolates were used in the mutant selection. For each wild-type isolate, 50 agar plugs were placed on PDA containing one of the DMIs at a concentration of EC₉₀. Agar plugs were placed on PDA containing prochloraz at 0.75 μg/ml (the EC₉₀ of most prochloraz-sensitive isolates). After 10–15 days at 28°C in the dark, cultures growing from the plugs were transferred to new PDA plates containing 0.75 μg/ml prochloraz. This step was repeated until there was no significant difference in the linear growth of fast growing sectors on the PDA plates with or without 0.75 μg/ml prochloraz. To obtain mutants with high resistance to prochloraz, the procedure was also performed with six additional concentrations of prochloraz (3, 5, 6, 10, 15, and 20 μg/ml) (Serfling et al., 2007). The same procedure was used to generate mutants resistant to epoxiconazole with 16, 32, 50, 100, 150, 200, 250, 300, and 350 μg/ml and to difenoconazole with 8, 16, 50, 100, 150, 200, 250, 300, and 350 μg/ml. The derived mutants were further examined for their level and stability of resistance to different azoles as described in the following section.

The EC₅₀ values were determined for the mutants and their parents with the three agricultural DMIs, two clinical DMIs, and three non-DMIs listed in Supplementary Table S1 (Russell, 2004; Serfling et al., 2007; Chen et al., 2012). Each treatment had three replicate plates, and the experiment was conducted twice.

The amplified CtCYP51 full-length fragments (primers are listed in Supplementary Table S3) were purified using the EasyPure Quick Gel Extraction Kit (TransGen, Beijing, China) and were cloned into the pEASY-T1 simple plasmid (TransGen, Beijing, China) according to the manufacturer’s recommendations. The vector that inserted into Escherichia coli strain was sequenced (Sunbiotech Co., Beijing, China) using vector primers M13F (5′-ACTGGCCGTCGTTTTAC-3′) and M13R (5′-GTCCTTTGT CGATACTG-3′) (Pang et al., 2013; Fan et al., 2014). DNA sequences were analyzed with DNAMAN5.2.2.0 (Lynnon Biosoft, Quebec, QC, Canada).

For both mutants and their parents, ten mycelial plugs from 4-day-old cultures were added to 60 ml of potato dextrose broth in a 100 ml flask. The flasks were incubated for 24 h on a rotary shaker at 120 rpm. Either prochloraz (0.5 μg/ml), epoxiconazole (30 μg/ml), or difenoconazole (25 μg/ml) were added to the corresponding flasks; other flasks were free of fungicides. After 24 h, mycelia were harvested by vacuum filtration for RNA extraction (Fan et al., 2014). Total RNA was extracted using the SV Total RNA Isolation kit (Promega, Beijing, China), and cDNA was synthesized using the PrimeScrip RT reagent Kit with gDNA Eraser (Takara, Beijing, China) following the manufacturer’s protocol. Three independent experiments were conducted.

Real-time (RT)-PCR was performed with an ABI7500 sequence detection system (Applied Biosystems, United States). Amplifications were conducted using the SYBR Premix Dimer Eraser kit (Takara, Beijing, China). The relative quantities (RQ) of products were calculated using the 2^-ΔΔCt method. The β-tublin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) genes were used as references to normalize the quantification of CtCYP51 expression (Fan et al., 2014). The thermal cycling conditions and primers are listed in Supplementary Table S3.

Bioinformatic analysis was used to investigate the molecular docking of different DMI fungicides with CYP51 proteins. The crystal structure of 5EAB, a CYP51 protein from the yeast Saccharomyces cerevisiae bound with tebuconazole, was retrieved from the Protein Data Bank and used in the current study. Sequence alignment by DNAMAN software indicated that the CYP51 amino acid residues of S. cerevisiae showed about 45% sequence identity with those of Colletotrichum. The crystal structure 5EAB was a suitable template for studying the binding conformation of different DMI fungicides with CYP51 proteins. C. scovillei is also a representative Colletotrichum species in China. Because it was sensitive to all five of the DMIs, the modeling structure of C. scovillei was obtained by using the template of 5EAB and the On-line Swiss-model software². The binding cavity was predicted by a ligand mode for tebuconazole complexed in CYP51 by SYBYL 7.3 software. The position of these different amino acids between C. scovillei and C. truncatum was analyzed in the CYP51 protein; only residues M376 and H373 were in the binding cavity with a distance of less than 10 Å. The Biopolymer-Replace Sequence subset from the SYBYL 7.3 software package was used to produce changes at residues M376 or H373. The Tripos force field with Gasteiger–Marsili charges was used for the energy minimization. Based on the CYP51 structure model of C. scovillei, the modeling of C. truncatum was built by replacing the amino acid at position 376 from methionine to leucine (M376L), at position 373 from histidine to asparagine (H373N), or at both these two positions (M376L/H373N). The molecular conformations of difenoconazole, tebuconazole, and myclobutanil were constructed by Sketch mode and were optimized using the Tripos force field and Gasteiger–Hückel charge. The Surflex-Dock of SYBYL 7.3 was used for the molecular docking modeling (Jain, 2003). The binding cavity was set as “ligand” and the total score was used to evaluate the binding affinity between ligand and protein (Jain, 1996). All molecular modeling between the putative C. scovillei and C. truncatum CYP51 proteins with ligand was conducted on the Silicon Graphics^® (SGI) Fuel Workstation (Silicon Graphics International Corp., Milpitas, CA, United States).

Data were analyzed using DPS software ver. 7.05 (Zhejiang University, Hangzhou, China). Means were separated using Duncan’s multiple range test at P = 0.05. Cross-resistance between two fungicides was analyzed using Spearman’s rank correlation coefficient with log-transformed EC₅₀ values (Chen et al., 2012; Pang et al., 2013).

EC₅₀ values were determined based on the growth of the 112 isolates on PDA containing different concentrations of five commonly used agricultural DMIs. The EC₅₀ value (mean ± SD, μg/ml) was 0.053 ± 0.023 for prochloraz, 1.027 ± 0.644 for difenoconazole, 1.956 ± 0.815 for epoxiconazole, >40 for tebuconazole, and >100 for myclobutanil. The values indicated that C. truncatum isolates were quite sensitive to prochloraz, difenoconazole, and epoxiconazole but not to tebuconazole or myclobutanil. These results suggest differential fungicide sensitivity in C. truncatum. The distributions of EC₅₀ values for prochloraz, epoxiconazole, and difenoconazole were unimodal, thus indicating the absence of azole-resistant subpopulations among the 112 representative isolates (Figure 1).

To evaluate the risk and molecular mechanism in developing resistance to these three effective DMIs in C. truncatum, mutants were generated by exposing C. truncatum cultures to prochloraz, difenoconazole, and epoxiconazole. A total of 11 DMI-resistant mutants were derived from the 112 isolates (Table 1). Two mutants were resistant to prochloraz, three were resistant to epoxiconazole, and six were resistant to difenoconazole. These resistant mutants were all derived from different parental isolates, and no mutants were obtained from other wild-type isolates. Mutation frequencies of C. truncatum to DMIs were high, and reached values of 2 × 10^-4 for prochloraz, 6 × 10^-4 for epoxiconazole, and 1 × 10^-3 for difenoconazole.

Among the 11 DMI-resistant mutants of C. truncatum, cross-resistance to the agricultural azoles prochloraz, epoxiconazole, and difenoconazole could be observed (Table 4). Cross-resistance to other azoles (fluconazole, ketoconazole) could be measured in the 11 DMI-resistant mutants, however, no cross-resistance could be observed with non-DMI fungicides including azoxystrobin, carbendazim, and mancozeb (Table 4).

To explore the genetic mechanism of C. truncatum resistant to the three effective DMIs, the sequence and expression level of CtCYP51 were analyzed in this study. The CtCYP51 gene had a length of 1719 bp, with two introns of 78 and 60 bp, and coded for 526 amino acids. When sequences of CtCYP51 were compared in DMI-resistant mutants and their sensitive parental isolates, no mutation was detected in any of the mutants. After treatment with prochloraz, CtCYP51 expression increased by about 17-fold in PS1, 6-fold in PS2, 52-fold in PR1, and 39-fold in PR2 (Figure 2A), which was about 3–6 fold higher in prochloraz-treated PR1 and PR2 than in PS1 and PS2. After treatment with epoxiconazole, CtCYP51 expression increased by about 12-, 9-, and 4-fold in ES1, ES2, and ES3 and by about 6-, 18-, and 26- fold in ER1, ER2, and ER3 (Figure 2B). Thus, epoxiconazole treatment resulted in about 2–6-fold higher CtCYP51 expression in ER2 and ER3 as compared to ES2 and ES3. After treatment with difenoconazole, CtCYP51 expression increased by about 3- to 35-fold in the six difenoconazole-sensitive isolates but about 7- to 81-fold in the difenoconazole-resistant isolates (Figure 2C), i.e., difenoconazole treatment caused 2- to 6-fold up-regulation in CtCYP51 expression in the difenoconazole-resistant mutants relative to their difenoconazole-sensitive parental isolates.

Based on the result that C. truncatum was only sensitive to prochloraz, epoxiconazole, and difenoconazole, but not to tebuconazole or myclobutanil, the basis for differential DMI sensitivity was investigated. The models were built for the docking of three representative DMI fungicides into C. scovillei or C. truncatum CYP51 binding pocket (Figure 3). The docking scores of these predicted models were shown in Table 5. The model for the wild-type C. scovillei indicated that the Fe – nitrogen coordinate bond between the heme iron active site and the DMI heterocyclic nitrogen atom was at a distance of 2.401, 2.209, and 2.128 Å for difenoconazole, tebuconazole, and myclobutanil, respectively (Figures 3A–C). In addition, there was a hydrogen bond interaction (2.038 Å) between the A306 and –NH group of nitrogen heterocyclic ring of difenoconazole (Figure 3A).

The different CYP51 amino acids between C. scovillei and C. truncatum were analyzed; only residues at position 376 and 373 were in the CYP51 binding cavity that bound to DMIs. The residues M376 and H373 in the structure model of C. scovillei were substituted to the corresponding amino acid L376 and N373 for the modeling of C. truncatum using the SYBYL 7.3 software, respectively. The amino acid change M376L changed the conformation of these three DMI fungicides in the CYP51 binding cavity and thereby led to the loss of the Fe – N coordinate bond (Figures 3D–F). For difenoconazole, the replacement of methionine by leucine generated two strong hydrogen bonds between difenoconazole with V134 (1.963 Å) and T136 (1.859 Å), and the docking score remained high at 7.45 (Figure 3D and Table 5). However, there was only a weak Fe – O coordinate bond (4.412 Å) between the heme iron active site with the O atom of tebuconazole, and a weaker Fe – N coordinate bond (6.819 Å) with the N atom of cyano group of myclobutanil, which decreased the docking score decreased from about 8 to 5 (Figures 3E,F and Table 5). The H373N mutation changed the Fe – N coordinate bond between difenoconazole and the heme iron from 2.401 to 4.205 Å (Figure 3G). As a result, the docking score changed to 8.18 for H373 and to 6.75 for N373 (Table 5). Although two new hydrogen bonds formed between tebuconazole and CYP51 amino acids, 2.011 Å with H309 and 2.425 Å with Y136, the mutation H373N eliminated the strong Fe – N coordinate bond (2.209 Å) between tebuconazole and the heme iron, which decreased the docking score from 8.38 to 4.39 for tebuconazole (Figure 3H and Table 5). The changes in the interaction for myclobutanil were similar to those for tebuconazole (Figure 3I), and decreased the docking score from 8.00 to 3.76 (Table 5).

The double mutations at positions 376 and 373 in the CYP51 of C. truncatum weakened the Fe – N coordinate bond interaction for difenoconazole from 2.401 to 6.341 Å and decreased the docking score from 8.18 to 6.00 (Figure 3J). For tebuconazole and myclobutanil, however, the double mutation M376L/H373N eliminated the Fe – N coordinate bond interactions between the fungicide and the heme iron active site, which caused the docking score decrease to 3.80 for tebuconazole and 3.49 for myclobutanil (Figures 3K,L and Table 5). The changes in the docking score were consistent with the results of the fungicide sensitivity assays, in which C. truncatum showed strong insensitivity to tebuconazole and myclobutanil but high sensitivity to difenoconazole.