Characterization of a silent azaphilone biosynthesis gene cluster in Aspergillus terreus NIH 2624

Abstract

Filamentous fungal secondary metabolites are an important source of bioactive components. Genome sequencing of Aspergillus terreus revealed many silent secondary metabolite biosynthetic gene clusters presumed to be involved in producing secondary metabolites. Activation of silent gene clusters through overexpressing a pathway-specific regulator is an effective avenue for discovering novel fungal secondary metabolites. Replacement of the native promoter of the pathway-specific activator with the inducible Tet-on system to activate the taz pathway led to the discovery of a series of azaphilone secondary metabolites, among which azaterrilone A (1) was purified and identified for the first time. Genetic deletion of core PKS genes and transcriptional analysis further characterized the taz gene cluster to consist of 16 genes with the NR-PKS and the HR-PKS collaborating in a convergent mode. Based on the putative gene functions and the characterized compounds structural information, a biosynthetic pathway of azaterrilone A (1) was proposed.

1. Introduction

Filamentous fungi have been considered prolific sources of secondary metabolites that display various biological activities and possess great potential for agricultural and pharmaceutical purposes (Keller et al., 2005; Greco et al., 2019; Keller, 2019; Li et al., 2020; Newman and Cragg, 2020). Recently, advances in genome sequencing of filamentous fungi indicate that the number of secondary metabolite genes or gene clusters exceeds the number of secondary metabolites identified so far, representing a potential reservoir of cryptic secondary metabolites. Thus, there is an increasing effort in deciphering the biosynthesis of secondary metabolites that are encoded by silent gene clusters and engineering their biosynthetic pathways to identify novel secondary metabolites and evaluating their biological activities (Keller, 2019; Chiang et al., 2009; Sanchez et al., 2012; Brakhage, 2013; Yaegashi et al., 2014; Guo et al., 2015; Hautbergue et al., 2018).

Among filamentous fungi, Aspergillus species have the promising potential to produce secondary metabolites with intriguing biological activities (Sanchez et al., 2012; Romsdahl and Wang, 2019; Caesar et al., 2020). Notably, Aspergillus terreus has been considered as a remarkable producer of bioactive secondary metabolites, particularly polyketides, such as asperfuranone, asterriquinone, butyrolactone, citreoviridin, sulochrin, terrain, terric acid, terretonin, and especially renowned for the production of the cholesterol-lowering drug, lovastatin (Romsdahl and Wang, 2019; Guo et al., 2012; Guo et al., 2013; Guo et al., 2014; Guo and Wang, 2014). The strain A. terreus NIH 2624 was sequenced in 2005 by the Broad Institute. Analysis using the Secondary Metabolite Unique Regions Finder (SMURF) software revealed 28 polyketide synthase (PKS) genes, 22 non-ribosomal peptide synthase (NRPS) genes, one hybrid PKS-NRPS gene, 2 PKS-like genes, and 15 NRPS-like genes in the genome (Guo and Wang, 2014; Khaldi et al., 2010). Although recent advances in the genome editing of Aspergillus species such as fusion PCR and the establishment of an efficient gene targeting system have greatly expedited secondary metabolite genome mining (Sanchez et al., 2012; Romsdahl and Wang, 2019; Caesar et al., 2020), there are still many secondary metabolite genes or gene clusters in A. terreus whose products remain elusive.

Several approaches have been utilized to activate silent fungal biosynthetic gene clusters (BGCs) to discover unidentified fungal secondary metabolites, including the alteration of culture condition, epigenetic manipulation, co-cultivation with other microorganisms, heterologous expression as well as transcriptional regulators manipulation (Keller, 2019; Brakhage, 2013; Hautbergue et al., 2018; Keller, 2015; Arora et al., 2020; Bergmann et al., 2007). Many studies suggested that transcriptional regulators play a crucial role in regulating fungal secondary metabolites production at pathway-specific, global, or epigenetic levels. Activating pathway-specific regulators is one efficient way to activate silent fungal BGCs (Keller, 2019; Brakhage, 2013; Oakley et al., 2017; Macheleidt et al., 2016; Lyu et al., 2020). Our lab and others have been developing genetic tools in A. terreus to enable the activation of genes (Flipphi et al., 2009; Gressler et al., 2011; Wanka et al., 2016; Sun et al., 2016). One successful genetic activation tool applied to A. terreus is the Tet-on system, which was employed to overexpress the silent NRPS-like gene pgnA by our lab and its product was revealed as (-)-phenguignardic acid (Sun et al., 2016).

Upon successfully developing the Tet-on system to A. terreus, we made efforts to systematically activate secondary metabolite gene clusters containing silent PKS and NRPS genes in A. terreus. For gene clusters with transcription activators located near PKS or NPRS genes, we replaced the native promoters with the Tet-on system. We discovered that overexpression of the transcription factor ATEG_03445 allowed the mutant strain to produce a series of azaphilone type secondary metabolites in A. terreus as observed by LC-MS analysis. Large scale culture and repeated purification efforts allowed us to isolate azaterrilone A (1) at sufficient purity for NMR structural characterization. Azaphilones are a class of fungal polyketide secondary metabolites consisting of a highly oxygenated pyranoquinone bicyclic core and that exhibit a broad spectrum of biological activities, such as antimicrobial, anticancer, cytotoxic, anti-inflammatory, antiviral activities as well as inhibition of dihydrofolate reductase, gp120-CD4 binding, Grb2-SH2 interaction, heat shock protein 90 (Hsp90), HuR-RNA interaction, MDM2-p53 interaction, and Tau aggregation (Gao et al., 2013; Chen et al., 2020; Paranjape et al., 2015; Kaur et al., 2017; Pavesi et al., 2021). There have been 677 azaphilones described from 61 genera of fungi as recently reviewed, and more are being discovered such as Chaetolactam A (Zu et al., 2021). But there are only five main biosynthetic pathways that have been characterized to date (Pavesi et al., 2021). The characterization of this pathway is thus significant for the discovery of azaphilone intermediates and future pathway engineering. Experimental genetic deletion of the two PKS genes located near the putative transcription factor confirmed their involvement in azaphilone biosynthesis. This evidence, combined with bioinformatics and transcriptional analysis, revealed that ATEG_03445 regulates the whole taz gene cluster covering genes from ATEG_03431 to ATEG_03446, a similar gene cluster to the aza gene cluster encoding the azanigerone A biosynthesis in A. niger (Zabala et al., 2012; Yin et al., 2016). Both gene clusters contain an NR-PKS and an HR-PKS collaborating in a convergent mode.

2. Results

2.1. Overexpression of tazR by the doxycycline-dependent Tet-on system activated an azaphilone biosynthesis pathway

To systematically activate silent PKS and NRPS gene clusters with a predicted cluster-specific transcription factor, we used a doxycycline-dependent Tet-on system to overexpress cluster specific transcription factors. Previous bioinformatic analysis of the A. terreus genome indicated an uncharacterized secondary metabolite gene cluster (taz gene cluster) between ATEG_03432 and ATEG_03446 containing two polyketide synthase genes and one transcription factor gene (Yin et al., 2016) (Table 1). The prediction was based on the alignment with its similar gene cluster, including azaA-L and R, in A. niger ATCC 1015 that corresponds to the azanigerone biosynthesis (Zabala et al., 2012; Yin et al., 2016). The prediction indicated that ATEG_03445 is the Zn₂Cys₆ zinc finger transcription factor within this gene cluster, and we anticipated that its expression would activate the whole taz gene cluster. We replaced the promoter of the putative zinc finger regulator gene ATEG_03445 (tazR) with the Tet-on system in a kusA-, pyrG-A. terreus strain using the genetic strategy shown in Fig. S1, and the mutant strain (CW9011, Table S2) was verified by diagnostic PCR (Fig. S2). Subsequently, the CW9011 strain was cultured in liquid LCMM media under 37 °C for 18 h. Then, the Tet-on system mutant strain was induced by adding 50 μg/mL sterilized doxycycline, and then the temperature was lowered to 30 °C. After cultivation for 72 h, the culture broth was extracted by ethyl acetate (EtOAc) and examined by LC-DAD-MS. We observed that overexpression of the transcription factor ATEG_03445 (tazR) allowed the mutant strain to produce a series of secondary metabolites (compounds 1 and 3–8) compared to the wild type (WT) control under LC-DAD-MS examination (Fig. 1). Interrogation of their MS and UV spectra revealed their m/z and UV-Vis absorption as shown in Fig. 2 and Fig. S3.

Table 1.

The taz gene cluster and putative function of the genes.

Gene	A. niger homolog	Putative function	Protein identity (%)
ATEG_03431 (tazF)		Monooxygenase
ATEG_03432 (tazA)	azaA	NR-PKS	50
ATEG_03433 (tazG)	azaG	Dehydrogenase	45
ATEG_03434 (tazM)		Condensation
ATEG_03435 (tazD)		Acetyltransferase
ATEG_03436 (tazK)		MFS transporter
ATEG_03437 (tazC)	azaC	Esterase/lipase	49
ATEG_03438 (tazN)		Dehydrogenase
ATEG_03439 (tazO)		O-methyltransferase
ATEG_03440 (tazE)		Enoyl reductase
ATEG_03441 (tazP)		Hydroxylase
ATEG_03442 (tazL)	azaL	Dehydrogenase	42
ATEG_03443 (tazHJ)	azaH, azaJ	NADB_Rossmann, Enoyl reductase	55
			38
ATEG_03444 (tazI)	azaI	Cytochrome P450	35
ATEG_03445 (tazR)	azaR	Zn2Cys6 regulator	40
ATEG_03446 (tazB)	azaB	HR-PKS	33

Fig. 1.

The DAD traces of extracts from the wild type and mutant strains as detected by UV (200–600 nm).

Fig. 2.

Compound 1 and its ESI-MS and UV-Vis spectra.

To structurally characterize the secondary metabolites, we scaled up the cultivation of the PTet-on-tazR mutant strain and isolated compound 1 by semi-preparative HPLC purification. The molecular formula was established to be C₁₅H₁₆O₅ based on its ¹³C NMR, DEPT, and HR-ESI-MS spectral data, and its structure was elucidated by ¹H NMR, ¹³C NMR, and 2D NMR spectroscopic analyses (Table S3, and Fig. S4). Compound 1 turned out to be an azaphilone type compound, and the name, azaterrilone A, was given. Furthermore, photodiode array analysis showed that additional secondary metabolites observed from the PTet-on-tazR mutant strain possessed highly similar UV-Vis absorption patterns as azaterrilone A (Fig. S3). Therefore, we speculated that these compounds probably are analogs of azaphilone, and the overexpression of tazR activated an azaphilone pathway as predicted. Unfortunately, due to their instability and low yield, we could not identify the other secondary metabolites except for azaterrilone A (1). The compounds’ instability may also explain the peak split of compounds 5, 6, 7, and 8 (Fig. 1).

2.2. Gene deletion analysis revealed that tazA and tazB are required for azaphilone synthesis

The predicted taz gene cluster contains two PKS genes, ATEG_03432 (tazA) and ATEG_03446 (tazB). Hence we hypothesized that tazA and tazB are the PKS genes involved in the biosynthesis of azaterrilone A (1) and the other azaphilone analogs (3–8). Previous heterologous expression of tazA in A. nidulans resulted in the deduction of its product as compound 2 (Chiang et al., 2013), which is the same proposed product of azaA (tazA homolog) in A. niger, providing another support for our hypothesis.

We performed individual gene deletion analysis of tazA and tazB with the simultaneous overexpression of tazR to confirm the hypothesis. Since only one pyrG marker is available in our A. terreus strain, it would have been necessary to recycle the selection marker to produce a mutant with both tazR promoter exchanged and the individual PKS genes deleted. Unfortunately, recycling of the pyrG marker was difficult in this specific A. terreus strain which requires substantial effort. Therefore, we came up with the strategy to circumvent pyrG recycling by replacing the target gene with the PTet-on-tazR construct in the kusA- and pyrG-background (Fig. S1). In this way, we were able to activate the gene cluster and delete the target gene in one step without recycling the pyrG marker (Table S2).

The mutant strains (CW9012 and CW9013) were verified by diagnostic PCR (Fig. S2). Then, the mutants were cultured under the same azaterrilone A producing conditions and induced by doxycycline. Their secondary metabolite profiles were analyzed by LC-DAD-MS. Knocking out the tazA gene with overexpressing the tazR gene simultaneously (ΔtazA::PTet-on-tazR) generated the same metabolite profile as the WT strain (Fig. 1) as all of the azaphilone compounds were not present compared to CW9011 (PTet-on-tazR). This evidence confirmed that tazA is involved in the biosynthesis of the core structure of azaterrilone A (1) and compounds 3–8. While knocking out the tazB gene (ΔtazB::PTet-on-tazR), only compounds 7 and 8 were not present compared to CW9011 (PTet-on-tazR), indicating that these two compounds contain additional polyketide modification. Thus, the tazB participated in the biosynthesis of these two additional products (7 and 8) instead of compounds 1, 3, 4, 5, and 6. Because compounds 7 and 8 possessed the same molecular weight, these two molecules might be isomers.

The fact that disruption of tazB did not affect the function of tazA indicated that they function independently and work in convergence instead of in sequence to synthesize the polyketide product, which is highly parallel to their homologs azaA and azaB in A. niger (Zabala et al., 2012). Future work will focus on identifying compounds 7 and 8, as without identifying final products, the exact function of tazB would remain elusive.

2.3. The taz gene cluster was characterized by RT-qPCR analysis to consist of 16 genes

To determine the extent of transcriptional activation of genes in the taz cluster by tazR overexpression, RT-qPCR was performed on the CW9011 strain (PTet-on-tazR), with or without doxycycline induction. The RT-qPCR primers were designed to amplify a ~ 200 bp segment in each gene for transcription quantitation. The amplified segment was preferred with exon-exon junction spanning to differentiate between amplification of cDNA and potential contaminating genomic DNA by melting curve analysis. The result in Fig. 3 showed that the genes from ATEG_03431 (tazF) to ATEG_03446 (tazB) were significantly upregulated along with tazR overexpression. Therefore, the taz gene cluster was characterized by RT-qPCR analysis to consist of 16 genes, which included an additional tazF gene compared with the previous prediction (Yin et al., 2016). Nonetheless, the confirmation of the taz pathway composition genes requires further experimental gene deletion analysis or heterologous pathway assembly analysis.

Fig. 3.

(A) The taz gene cluster; (B) Transcriptional analysis results by RT-qPCR of the taz gene cluster. By comparing to no-DOX condition (black), all the 16 genes from ATEG_03431 to ATEG_03446 are upregulated upon doxycycline (DOX) induction (grey).

3. Discussion

Aspergillus terreus is an important producer of many bioactive secondary metabolites (Guo and Wang, 2014). More and more efforts are focused on activating silent gene clusters through overexpression of a pathway-specific regulator to mine fungal genomes for novel secondary metabolites (Hautbergue et al., 2018; Oakley et al., 2017; Lyu et al., 2020; Grau et al., 2018). Adopting the doxycycline-dependent Tet-on system has made this strategy possible for activation of silent gene clusters in A. terreus. Our overexpression of tazR allowed the activation of a silent gene cluster consisting of 16 genes demonstrated by RT-qPCR, and the generation of azaphilone type secondary metabolites, including characterized compound, azaterrilone A (1) (Fig. 2). Although we were not able to purify and characterize compounds 3–8, the structures of compounds 3–5 were extrapolated based on their molecular weight, their HPLC retention time relative to compound 1 as well as the homologous pathway of azanigerones in A. niger (Zabala et al., 2012) (Figs. 1 and 4, Fig. S3). The molecular weight difference between compound 5 and compound 1 is about 42 Da, indicating compound 5 has one less acetyl group, which matches its earlier retention time than compound 1. The molecular weight difference between compound 4 and compound 1 as well as compound 3 and compound 5 is about 16 Da, indicating compound 4 and compound 3 have one more hydroxyl group comparing to compound 1 and compound 5, respectively, which matches their early retention time since hydroxyl group makes them more polar. Combining the structural elucidation of compound 1 and the structural extrapolation of compounds 3–5 as well as the putative function of tazA-P predicted by bioinformatic analysis, a speculative biosynthetic pathway for compound 1 was proposed and illustrated in Fig. 4.

Fig. 4.

Proposed biosynthesis pathway for compound 1.

The first step of the pathway begins with the synthesis of the deduced intermediate compound 2 by the NR-PKS TazA, which was already demonstrated by its heterologous expression in A. nidulans, but it didn’t directly identify its product as compound 2 while the actual identified products are quinone type of compounds (Chiang et al., 2013). The activation of tazA in A. terreus adds another evidence to prove its direct product is compound 2. The domain architecture of TazA is composed of SAT-KS-AT-PT-ACP-CMeT-R from the N- to C-terminal. TazA assembles one acetyl-CoA starter unit, five malonyl-CoA units, and catalyzes a series of Claisen condensations, methylation, PT-mediated cyclization, and finally releases compound 2 through the R-domain, indicating it functions in the same way as its homolog AzaA in A. niger. The reduction of compound 2 on its terminal ketone and the following pyran-ring formation to yield compound 3 should occur similarly to the azanigerones pathway, however, the catalyzing enzymes remain unknown. It was proposed that AzaE and AzaH catalyze such reactions for the azanigerones pathway, but the homolog of AzaE cannot be found in the taz gene cluster. Furthermore, although a portion of TazHJ shares significant amino acid similarity to AzaH, its deletion did not affect the accumulation of compound 1 (Fig. S5), indicating another enzyme is responsible for the pyran-ring formation rather than TazHJ. The lack of AzaH homolog in the taz gene cluster highlights significant enzymatic differences between these two biosynthesis pathways. Additionally, its deletion eliminated the production of compounds 7 and 8, suggesting its involvement in producing these two compounds. The conversion of compound 3 to compound 5 experiences dehydration and enoyl reduction. There are no dehydratase homologs predicted within the taz gene cluster. Therefore, the dehydration catalyzing enzyme remains unclear. TazE is predicted as an enoyl reductase, and thus we propose that it catalyzes the enoyl reduction in the conversion of compound 3 to compound 5. TazD and TazM are predicted as an acetyltransferase and condensation enzyme, respectively. Therefore, they might catalyze the acetylation to convert compound 3 to compound 4 and compound 5 to compound 1 correspondingly.

Since the production of compounds 1 and 3–6 are not affected by the deletion of TazB, it is unlikely they are the final products of the taz pathway. Compounds 7 and 8 are likely to be the final pathway products since their biosynthesis involved the function of both tazA and tazB, which was demonstrated by the tazA and tazB individual gene deletion analysis, and they might be stereoisomers due to the same molecular weight. Yin et al. predicted the taz pathway compounds assuming TazB has the same function as AzaB in A. niger (Yin et al., 2016). However, the molecular weight of compounds 7 and 8 did not correspond to this prediction. Therefore, to elucidate the exact function of TazB, the structures of compounds 7 and 8 need to be characterized in the future.

In conclusion, we applied the Tet-on system to A. terreus to activate the taz pathway by overexpressing its pathway-specific transcription factor tazR. Individual gene deletion analysis revealed that the NR-PKS gene tazA and the HR-PKS gene tazB encode the core structure of the pathway compounds, which provides another example for dual PKS-containing gene clusters. Furthermore, the taz gene cluster was characterized by RT-qPCR analysis to consist of 16 genes. The structures of compounds 3–5 were deduced based on the purification and structural characterization of compound 1, as well as the prediction of the putative function of tazA-P by bioinformatic analysis. Hence, we were able to propose a speculative biosynthetic pathway for compounds 1 and 3–5. However, the pathway’s final product’s identity and the function of some genes remained unclear and should be further investigated by combining additional gene activation strategies.

4. Methods and materials

4.1. Strains and molecular genetic manipulations

The primers used in this study are listed in Table S1 in the Supporting Information. The A. terreus wild-type and mutant strains used in this study are listed in Table S2. The strain CW9011 was created by replacing the native promoter of ATEG_03445.1 with the A. fumigatus pyrG gene (AfpyrG) and the Tet-on system (Ptet-on) in the KusA-, pyrG- background of A. terreus. The AfpyrG-Ptet-on fragment was amplified from the previous study strain CW9001 (Sun et al., 2016). Replacement of endogenous promoters with Ptet-on was carried out as shown in Fig. S1 (A). The strains CW9012 and CW9013 were created by replacing the target gene tazA and tazB with the AfpyrG-Ptet-on-tazR fragment, respectively, as shown in Fig. S1 (B). The AfpyrG-Ptet-on-tazR fragment was amplified from the strain CW9011. The construction of fusion PCR products, protoplast generation, and transformation were accomplished by previously described methods (Guo et al., 2012; Chiang et al., 2013; Guo et al., 2013). The scheme of diagnostic PCR was illustrated in Fig. S1, and all transformants were verified by diagnostic PCR (Fig. S2).

4.2. Fermentation and LC-MS analysis

The A. terreus NIH2624 WT strain and mutant strains were inoculated at 37 °C in 25 ml lactose dextrose minimal medium (LCMM) liquid medium (6 g/l NaNO₃, 0.52 g/l MgSO₄·7H₂O, 1.52 g/l KH₂PO₄, 0.52 g/l KCl, 20 g/l lactose, 10 g/l D-glucose, and 1 ml/l of trace element solution (EDTA disodium salt 50 g/l, H₃BO₃ 11.4 g/l, ZnSO₄·7H₂O 22 g/l, MnCl₂·4H₂O 5.06 g/l, CoCl₂·6H₂O 1.61 g/l, CuSO₄·5H₂O 1.57 g/l, FeSO₄·7H₂O 4.99 g/l, and (NH₄)₆Mo₇O₂₄·4H₂O 1.1 g/l) at 1 × 10⁶ spore/ml per 125 ml flask and incubated on the rotator shaker at 180 rpm. After incubation for 18 h, 25 μl sterile doxycycline (50 μg/μl) was added into the culture broth of the A. terreus NIH2624 WT strain and Ptet-on mutant strains, making the final concentration of doxycycline at 50 μg/ml. The A. terreus NIH2624 WT strain and Ptet-on mutant strains, which without treating doxycycline were employed as control models. Subsequently, the incubator temperature was adjusted to 30 °C and the culture medium was collected after 72 h.

The culture medium was further filtered and extracted with equal volume of ethyl acetate (EtOAc). And the pH of the medium was adjusted to approximately 2 by the addition of 6 M HCl and extracted with equal volume of EtOAc again. The EtOAc portion was condensed under reduced pressure to obtain the EtOAc crude extracts and redissolved in 1 ml 20% DMSO in MeOH. Then, the portion (10 μl) was subjected to high performance liquid chromatography-photodiode array detection-mass spectrometry (HPLC-DAD-MS) for data analysis and was accomplished as previous described procedures (Ahuja et al., 2012; Sun et al., 2018).

4.3. Isolation of secondary metabolites

For structural elucidation, the strain CW9011 was cultivated at 37 °C in a total of 1 L LCMM liquid medium (200 ml per 1 L flask) at 1 × 10⁶ spore/ml with shaking at 180 rpm. To induce activation of the tet-on promoter, sterile doxycycline at a final concentration of 50 μg/μl was added into each flask after 18 h of incubation. Then, the incubator temperature was lowered to 30 °C and the culture medium was collected after 72 h of induction. The medium was filtered and extracted with equal volume of ethyl acetate (EtOAc). Then the pH of the medium was adjusted to around 2 by the addition of 6 M HCl and extracted with equal volume of EtOAc again. This EtOAc layer was evaporated to get a crude extract. Further purification was carried out by gradient HPLC on a C₁₈ reversed phase column [Phenomenex, Luna 5 μm C₁₈, 250 × 10 mm] with a flow rate of 5.0 ml/min and measured by a UV detector at 304 nm. The gradient system was MeCN (solvent B) and 5% MeCN/H₂O (solvent A) both containing 0.05% TFA. The gradient condition for HPLC analysis of the crude extract from the Ptet-on_tazR mutant strain was 0–5 min 100–80% A, 5–35 min 80–40% A, 35–38 min 40–0% A, 38–40 min 0–0% A, 40–42 min 0–100% A. Compound 1 (2.1 mg/L) was eluted at about 32.5 min.

4.4. Structural identification

NMR spectra measurements were collected on Varian Mercury Plus 400 spectrometers. Azaterrilone A (1) was isolated as amorphous powder, orange color, possessing the molecular formula deduced to be C₁₅H₁₆O₅. The structure of compound 1 was established on the basis of its MS, ¹H NMR, ¹³C NMR, and 2D NMR spectroscopic data analyses. For ESI-MS and UV-Vis spectra, see Fig. 2; For NMR data and spectra (DMSO-d₆), see Table S3 and Fig. S4.

4.5. Transcription analysis by RT-qPCR

The CW9011 strain was fermented in azaterrilone A producing conditions and induced by doxycycline for mRNA extraction. The strain cultured without doxycycline induction was used as control model. Total mRNA was extracted by using the Qiagen RNeasy Plant Mini Kit. The total mRNA was digested by Recombinant DNase I (ambion by life technologies) to remove DNA contamination. The cDNA library was made from the same amount of mRNA by using TaqMan reverse transcription reagents (T04141) and the oligo DT primer. The primers used for target gene fragment amplification were shown in Table S1. The expression of each gene was analyzed with the ABI 7900HT Fast Real-Time PCR system by following the KAPA SYBR FAST qPCR kit (KK4601) protocol. The experiments were performed in triplicate and the results were illustrated in Fig. 3. The β-tubulin gene atbenA (ATEG_00287.1) was used as control and quantification standard. Relative expression levels were calculated by using the 2^−ΔΔCt method.