Heterologous pathway assembly reveals molecular steps of fungal terreic acid biosynthesis

Terreic acid is a potential anticancer drug as it inhibits Bruton’s tyrosine kinase; however, its biosynthetic molecular steps remain unclear. In this work, the individual reactions of terreic acid biosynthesis were determined by stepwise pathway assembly in a heterologous host, Pichia pastoris, on the basis of previous knockout studies in a native host, Aspergillus terreus. Polyketide synthase AtX was found to catalyze the formation of partially reduced polyketide 6-methylsalicylic acid, followed by 3-methylcatechol synthesis by salicylate 1-monooxygenase AtA-mediated decarboxylative hydroxylation of 6-methylsalicylic acid. Our results show that cytochrome P450 monooxygenase AtE hydroxylates 3-methylcatechol, thus producing the next product, 3-methyl-1,2,4-benzenetriol. A smaller putative cytochrome P450 monooxygenase, AtG, assists with this step. Then, AtD causes epoxidation and hydroxyl oxidation of 3-methyl-1,2,4-benzenetriol and produces a compound terremutin, via which the previously unknown function of AtD was identified as cyclooxygenation. The final step involves an oxidation reaction of a hydroxyl group by a glucose-methanol-choline oxidoreductase, AtC, which leads to the final product: terreic acid. Functions of AtD and AtG were determined for the first time. All the genes were reanalyzed and all intermediates and final products were isolated and identified. Our model fully defines the molecular steps and corrects previous results from the literature.


Results
Cloning and intron identification of TA biosynthetic genes. The mRNAs of atA, atE, atG, atD, and atC were obtained, reversely transcribed to cDNA and sequenced. The protein-coding sequence and introns of each gene were then identified ( Supplementary Fig. S1), thus correcting the previous results of genome shotgun sequencing (GenBank accession No. CH476602.1). The newly identified sequences were deposited in the The proposed biosynthetic pathway of terreic acid (TA) in A. terreus. The scheme was referred to ref. 7 . Hypothetical compounds are boxed. Proteins and intermediates for TA biosynthesis are shown. Proteins with an unknown function are marked with a question mark, and proteins whose functions needed to be verified are marked with an asterisk. database, and details for each gene are summarized in Table 1. The full-length protein-coding sequence of atA is 1405 bp with an intron of 64 bp (Table 1), which codes for a protein of 466 amino acid residues (aa) and is different from that of the previously predicted sequence (gene locus: ATEG_06272). The full-length coding sequence of atE is 1850 bp [with four introns: 64, 61, 64, and 53 bp (Table 1)] coding for a 535 aa protein. Moreover, the identified atE was found to have 329 more base pairs in the 5′-flanking region as compared to that of the previously predicted sequence (gene locus: ATEG_06277). Repeated cloning and sequencing experiments showed that the atD (gene locus: ATEG_06276) previously predicted by whole-genome shotgun sequencing misses a cytosine nucleoside at position 1163. Thus, the coding sequence of atD is 1160 bp [with two introns 65 and 54 bp ( Table 1)] encoding a 320 aa protein. The full-length coding sequence of atG is 583 bp [with two introns, 50 and 59 bp (Table 1)] coding for a 157 aa protein. The identified atE has 94 more base pairs in the 5′-flanking region as compared to that of previously predicted atE (gene locus: ATEG_06280). The full-length coding region of atC is 2277 bp [with eight introns: 52, 57, 58, 56, 50, 73, 60, and 50 bp (Table 1)] coding for a 606 aa protein, which is different from that of previously predicted atC (gene locus: ATEG_06274). The genes after intron removal were cloned from their cDNA and then applied to construction of expression plasmids.
Identification of functions of AtX, AtA, and AtE. In our preliminary study, the expression of phosphopantetheinyl transferases encoded by Aspergillus nidulans npgA in P. pastoris turned inactive apo-ACP into active holo-ACP, leading to activation of polyketide synthase AtX and production of 6-MSA 10 . In this case, npgA and atX were reconstructed in a single plasmid (pPICβ-npgA-atX) to leave more selective markers for screening of various expression strains. Accordingly, strain GS-NX expressing atX and npgA under the control of promoter P AOX1 was constructed. Relevant products were extracted and analyzed by high-performance liquid chromatography (HPLC) after methanol induction for 48 h. Strain GS-NX produced 6-MSA (2) (193.6 mg/L) as compared to the wild-type GS115 strain (negative control; Fig. 2A-C). We next introduced atA or atE into strain GS-NX using the same promoter system. Analysis of products showed that the expression of atA resulted in a strain (GS-NXA) capable of producing 3-methylcatechol (5) (61.0 mg/L), whereas nothing changed after the expression of atE in GS-NX ( Fig. 2D-F). The product of 3-methylcatechol was identified by liquid chromatography with mass spectrometry (LC-MS) and 1 H nuclear magnetic resonance (NMR) analysis ( Supplementary Fig. 2) in a comparison with other experimental results 11 . These findings proved that AtA but not AtE catalyzes the decarboxylative hydroxylation after AtX in TA biosynthesis, in agreement with gene knockout results in a native strain 7 . We then coexpressed atE and atA in GS-NX under the control of promoter P AOX1 and generated strain GS-NXAE. A comparison strain, GS-NXAG, was also created by expressing atG in GS-NXA because both AtE and AtG contain a conserved putative P450 monooxygenase domain (Table 1). Of note, strain GS-NXAE was found to produce two specific products (compound 6, 77.8 mg/L, and compound 7, unstable [titer not determined]; Fig. 2H and E) whereas GS-NXAG does not produce any new compound (still generates compound 5, 63.3 mg/L) as compared to GS-NXA ( Fig. 2G and E). Next, compound 6 was isolated and analyzed for its chemical structure. Because no published NMR data are available for compound 6, one-dimensional (1D) and 2D NMR spectroscopy ( 1 H NMR, 13 C NMR, heteronuclear multiple bond correlation [HMBC], and heteronuclear single quantum correlation [HSQC]) enabled us to identify its structure as 3-methyl-1,2,4-benzenetriol (m/z 140, Supplementary  Fig. S3), which has not been previously isolated and identified 7 . Moreover, compound 6 was partially converted to a new compound, 7 (m/z 140 by LC-MS, ~12 min) with broth pH decreasing during strain GS-NXAE culture ( Supplementary Fig. S4). Moreover, preparative-HPLC-purified compound 7 immediately transformed into 6 according to analytical HPLC assay ( Supplementary Fig. S4). These results revealed that 6 and 7 have identical molecular weights and easily convert into each other; these data helped us to predict compound 7 as a tautomer of 6. It is probably 3,4-dihydroxy-2-methylcyclohexa-2,5-diene-1-one, which is unstable and could not be purified for chemical structure analysis. Identification of functions of AtD, AtG, and AtC. Guo et al. knocked out atG and atD in native A. terreus, and each procedure caused a loss of the TA-biosynthetic ability 7 . Nevertheless, because no intermediates or shunt products were identified, functions of atG and atD could not be confirmed. Therefore, strains GS-NXAEG and GS-NXAED were constructed, and the heterologous expression results confirmed that AtD but not AtG works in this reaction, in contrast to the previous prediction of AtG function 7 ( Fig. 3A-C). The product was identified as terremutin (3,49.0 mg/L) by LC-MS and 1 H NMR analysis ( Supplementary Fig. S5) and was consistent with the literature data 7 . Moreover, to test whether AtG assists AtD at this step, we then introduced atG into strain GS-NXAED to generate GS-NXAED-G transformants, which contain all other intact biosynthetic genes of strain GS-NXAED. Several GS-NXAED-G strains were selected randomly, and none of them produced a higher titer of terremutin (3) than GS-NXAED did (Supplementary Table S1). Therefore, AtG did not assist AtD in this reaction. Strains GS-NXAEDC and GS-NXAEGDC were constructed by introducing atD and atC simultaneously into GS-NXAE and GS-NXAEG, respectively. As compared to GS-NXAED, a specific product was produced by both strain GS-NXAEDC (0.9 mg/L) and strain GS-NXAEGDC (5.8 mg/L) at retention time 25.8 min (Fig. 4A,B,E).
The new product had the same retention time and ultraviolet (UV) absorption spectrum as did the TA standard (1), and feeding the TA standard into the extracted sample enhanced the compound absorption peak as expected.
Besides, LC-MS results indicated m/z of 154 for this compound, in line with TA's m/z ( Supplementary Fig. S6). These results revealed that the newly produced compound was probably the final product: TA. Nonetheless, because production of this newly generated compound was very low in strains GS-NXAEDC and GS-NXAEGDC in shake flask culture, it was not easy to obtain enough of this compound for 1 H NMR analysis. Consequently, the strain was fermented in a 5 L bioreactor and the target compound was purified. 1 H NMR data (Supplementary Fig. S6) were in agreement with other results 7 , confirming that AtC catalyzes transformation of terremutin (3) into TA (1). As shown in Fig. 4B and E, higher TA production was observed in strain GS-NXAEGDC compared with GS-NXAEDC. Nevertheless, given that strains GS-NXAEDC and GS-NXAEGDC were constructed separately, they may contain different biosynthetic gene copies, which affected the concentration of intermediates and final products. Thus, we introduced atG into the GS-NXAEDC strain to obtain GS-NXAEDC-G transformants and compared the TA production levels. Three GS-NXAED-G strains were selected for culture randomly and none of them produced a higher titer of TA (1) than GS-NXAEDC did (Supplementary Table S2). This finding indicated that AtG neither catalyzed nor assisted AtC in catalyzing the final reaction.
AtG boosted AtE catalysis. In the native strain of A. terreus, accumulation of intermediates or shunt products was not detected after a knockout of atG. Additionally, AtG has a putative function of cytochrome P450 monooxygenase, which might work with the other cytochrome P450 monooxygenase: AtE. Accordingly, we introduced atG into the GS-NXAE strain to generate GS-NXAE-G transformants. Three strains were selected randomly for analysis of production of compounds 6 and 7. Both compounds were highly accumulated in GS-NXAE-G strains compared with GS-NXAE, especially 6 after 24 h methanol induction and 7 after 48 h methanol induction (Fig. 5). Levels of 6 were higher even in GS-NXAED-G and GS-NXAEDC-G transformants than in their parent strains GS-NXAED and GS-NXAEDC (Supplementary Table S3). Thus, we may conclude that AtG assists AtE but not AtD or AtC in the TA-biosynthetic pathway. Accordingly, the molecular steps for TA biosynthesis were clarified (Fig. 6), and AtX, AtA, AtE/AtG, AtD, and AtC were found to function stepwise in this process. Nevertheless, how AtG works with AtE to improve the reaction step still kept unknown. To test if AtG interacts with AtE and thereby forms a protein complex, we then conducted a yeast two-hybrid (Y2H) assay 12 on both proteins. However, the interaction between AtG and AtE was not observed ( Supplementary Fig. S7), indicating that they probably not form protein complex and work in other way that needs further deep work to clarify.

Discussion
To determine the biosynthetic molecular steps for TA (1) in A. terreus, stepwise pathway assembly of TA was performed here in a heterologous host: methylotrophic yeast P. pastoris. Coding sequences and introns of biosynthetic genes atA, atE, atD, atG, and atC were identified by reverse transcription, and our findings corrected the results previously submitted to databases. As expected, P. pastoris expresses AtA, AtE, AtD, AtG, and AtC correctly according to western blotting analysis ( Supplementary Fig. S8). Conserved domain analysis indicated that the identified AtA contains a FAD-dependent salicylate 1-monooxygenase domain catalyzing the conversion of salicylate to catechol. Coexpression of atX and PPTase-encoding gene npgA produced 6-MSA (2), and introducing atA next led to the biosynthesis of 3-methylcatechol (5) by a decarboxylative hydroxylation reaction. Quick BLASTp results on AtA matched several functional uncharacterized salicylate 1-hydroxylases/ 1-monooxygenases. As reported elsewhere, some salicylate 1-monooxygenase NahG (1-hydroxylating, decarboxylating, EC 1.14.13.1) was identified in Pseudomonas putida 13 . A hydrolase OpS4 (UniProtKB accession No. J4VWM7) catalyzing orsellinic acid to 6-methyl-1,2,4-benzenetriol by decarboxylative hydroxylation was also identified in Beauveria bassiana 14 . Nevertheless, AtA shares only 34% identity with NahG and 25% identity with OpS4. BLAST results indicates that AtA also shares 32% identity with an A. nidulans salicylate 1-monooxygenase, SalA, which was characterized biologically but not chemically (Table 1) 15 . Moreover, the oxidation behavior of AtA is similar to that of three other reported FAD-dependent monooxygenases, TropB in tropolone biosynthesis 16 , SorbC in sorbicillinoid biosynthesis 17 , and AzaH in azaphilone biosynthesis 18 , where they perform oxidative dearomatization of their specific substrates. AtA may hydroxylate 6-MSA (2) to form an unstable intermediate, which would easily undergo decarboxylation to generate 3-methylcatechol (5). For biosynthesis of TA in this case, sorbicillinoids such as sorbicillactones 17 and azaphilones such as rubropunctatin 18 require only ring oxidation. Nonetheless, tropolone biosynthesis requires oxidation of both the ring itself and the ring methyl group of a polyketide aldehyde 16 , in contrast to the biosynthesis of citrinin, where only the ring methyl is oxidized 19 .
Recently, a 6-MSA decarboxylase, PatG, was identified in patulin biosynthesis and was found to catalyze the first biosynthetic step, namely, decarboxylation but not hydroxylation of 6-MSA to form m-cresol 20 . Thus, these results finally confirmed AtA as a 6-MSA 1-monooxygenase but not the previously predicted 6-MSA decarboxylase 7 . Quick BLASTp of AtE revealed good identity to cytochrome P450 monooxygenases, among which, PatI and PatH (Table 1) in the biosynthesis of patulin have been chemically identified 21 . Introduction of atE next led to production of 3-methyl-1,2,4-benzenetriol (6) via hydroxylation of 3-methylcatechol, also in agreement with the putative function of cytochrome P450 monooxygenase AtE 7 . Furthermore, 6 easily converted to a specific compound 7 during an active culture phase with decreasing pH, and 7 quickly converted back to 6 after purification, allowing us to infer that it is 3,4-dihydroxy-2-methylcyclohexa-2,5-diene-1-one (7), a tautomer of 6 whose chemical structure could not be identified precisely. This uncertainty did not affect the biosynthetic pathway analysis in this case. The most uncertain molecular step for this biosynthetic pathway is the conversion of 3-methyl-1,2,4-benzenetriol (6) to terremutin (3) as reported in another work 7 . By experimental gene identification in our study, AtG was confirmed as a protein only 157 aa long. Although AtG contains a conserved cytochrome P450 monooxygenase domain, it shows low identity to other proteins in BLAST results. Particularly, AtG shows much smaller molecular weight than the proteins from BLAST results. Therefore, we predicted that AtG may not work for terremutin production, and introduction of atG into one of our strains finally proved this conclusion. By contrast, introduction of atD into one of our strains successfully generated terremutin. A conserved-domain analysis in AtD suggested that this protein contains a cupin_2 domain, and Quick BLASTp search yielded ~20 hypothetical homologues with undefined function (from filamentous fungi) with high identity to AtD, including a putative cupin protein (identity of 70%) 22 and hypothetical dioxygenase PatJ (identity of 62%) 9 (Table 1). Considering its identified function in this case, it may be designated as an epoxidase with epoxidation functions. On the other hand, AtD showed no significant similarity with some reported epoxidation-mediated epoxidases or cytochrome P450 enzymes, e.g., those participating in the biosynthesis of squalene 23 , lasalocid 24 , mycinamicin 25 , and FD-891 26 . Its enzymatic mechanism and specific biosynthetic roles in other fungi will be an interesting topic for future research. Introduction of atC into one of our strains next produced the final product, TA (1), proving that AtC functions at this step, in line with the gene knockout results 7 . Conserved-domain analysis indicated that AtC is a GMC oxidoreductase that matches many homologues. Nevertheless, only the versicolorin B synthase (VBS) from Aspergillus parasiticus 27,28 and that from Aspergillus flavus 29 (Table 1) have been characterized.
Fungal cytochrome P450 monooxygenases usually have versatile biocatalytic activities 30,31 . Elsewhere, a knockout of atA in native A. terreus caused potent accumulation of 6-MSA (2) whereas a knockout of atG did not 7 , meaning that the reaction of 6-MSA (2) producing 3-methylcatechol (5) is independent of AtG. To further clarify the specific function of atG, it was then introduced here into the terremutin-producing or TA-producing strains. These experiments yielded no new compound or improvement of terremutin or TA biosynthesis, suggesting that AtG does not function at these two steps. Another study showed that a knockout of atE in native A. terreus causes strong accumulation of shunt product 4 of 3-methylcatechol (5), but a knockout of atG does not 7 , meaning that atE but not atG plays the essential role in this reaction. Our finding that strain GS-NXAE but not GS-NXAG (genes are transcribed and proteins are expressed correctly, Supplementary Figs S8 and S9) Figure 5. Introduction of atG into strain GS-NXAE improved the biosynthesis of compounds 6 and 7. Three resulting strains (GS-NXAE-G #18, #33, and #36) were selected randomly and tested. Gene copies of atG in each strain were not determined, and the production levels of the two compounds differed among the three GS-NXAE-G strains. Compound 6 was quantified properly, but compound 7 was quantified only as the relative HPLC peak area (the highest titer of GS-NXAE-G #18 after 48 h induction was set to 100%). One-way analysis of variance (ANOVA) was employed to determine significant production differences of compounds 6 and 7 between GS-NXAE and each GS-NXAE-G strain. The P-value was used to check the significance, and it was significant at P < 0.05. OriginPro 8.0 (OriginLab Corporation, USA) was used for ANOVA. **P < 0.01 at 24 h; ## P < 0.01 at 48 h. Detailed P-value for each run was shown in Supplementary Table S8.
Scientific REPoRTS | (2018) 8:2116 | DOI:10.1038/s41598-018-20514-x produces 3-methyl-1,2,4-benzenetriol (6) confirmed this notion. Of note, when atG was introduced into the 3-methyl-1,2,4-benzenetriol (6)-producing strain (GS-NXAE), it highly improved the biosynthesis of this compound. BLAST searches revealed that AtE and AtG share very low identity. As opposed to AtG, many homologues of AtE (with relatively high identity) were found by Quick BLASTp. Thus, AtG could be a putative cytochrome P450 monooxygenase assisting AtE at the hydroxylation step. Fungal cytochrome P450 usually contains four kinds of conserved motifs 30,31 , and sequence analysis revealed that AtG possesses PER and EXXR motifs whereas AtE contains only PER motifs. To date, a vast number of cytochromes P450 classified into ~400 families have been identified in >2500 fungal species [30][31][32] . Nonetheless, cooperation of the two types of cytochrome P450 monooxygenase at a single biocatalytic step has seldom been reported. Our protein-protein interaction analysis by Y2H assay preliminarily showed that AtG and AtE did not combine with each other, while how these enzymes (with widely divergent molecular weights) work together at this catalytic step is still an interesting topic for a future study.
After these efforts, we finally clarified the molecular steps in the TA biosynthetic pathway. These data show a fundamental pathway for biosynthesis of TA derivatives, which can be screened for anticancer pharmaceuticals. Moreover, because 6-MSA is an abundant primary intermediate in fungal secondary metabolism 9,20,21,33 , the results we reported in this work may be useful for analysis of the biosynthetic mechanism for other 6-MSAderived bioactive compounds. In addition, the successful heterologous expression proved that P. pastoris is a good chassis organism maintaining correct bioactivity of fungal proteins, and these properties certainly facilitate heterologous biosynthesis of fungal secondary metabolites. Moreover, with a short culture phase, clean metabolic background, and easy genetic manipulations [34][35][36][37] , this host may be a good choice for either biosynthetic analysis or improvement of production of fungal secondary metabolites.

Methods
Strains, plasmids, media, and culture conditions. Genes for TA biosynthesis were cloned from the at cluster of A. terreus NIH2624. Escherichia coli TOP10 served as a storage host for plasmids. P. pastoris GS115 was used as the basic host for heterologous expression of TA-biosynthetic genes. Vectors pAG32 (kindly provided by Prof. Saurabh Joshi in University of California, San Diego) 38   extract, 1% tryptone, and 0.5% NaCl. P. pastoris was cultivated at 30 °C in the YPD medium consisting of 1% yeast extract, 2% tryptone, and 2% glucose for seed preparation, and then cultivated in the minimal medium (MM) composed of 1.34% YNB (Sigma) and methanol for protein expression and compound biosynthesis. Methanol was added to 0.5% (v/v) every 24 h as a carbon source and inducer.

Molecular biological techniques.
For PCR experiments, standard protocols were applied with a PCR amplification kit (TaKaRa, Cat. # R011). Fungal RNA was extracted by means of the RNAsimple Total RNA Kit (TIANGEN Cat. # DP419). Plasmid DNA was isolated from E. coli using the TIANprep Rapid Mini Plasmid Kit (TIANGEN Cat. # DP105-03). DNA fragments separated in an agarose gel were extracted with the Universal DNA Purification Kit (TIANGEN Cat. # DP214-03). Multiple fragments were assembled via the ClonExpress TM II One Step Cloning Kit (Vazyme Biotech Co., Ltd., China). Strains P. pastoris GS115 and E. coli TOP10 and yeast vectors pPICZ B and pPIC3.5 K were purchased from Invitrogen. Transformation of yeast cells and screening of transformants were executed according to Pichia protocols 39 . Yeast two-hybrid (Y2H) assay were described in detail in supplementary data file (Supplementary Fig. S7).
Identification of introns of genes within the at cluster. The mRNA sequences of genes within the at cluster are already predicted in GenBank (GenBank accession No. CH476602.1), but many of them are different from the prediction results of the SoftBerry software. To confirm the exact positions of introns and express correct enzymes in P. pastoris for TA biosynthesis, cDNA for each gene was obtained and analyzed by reverse transcription of RNA. An A. terreus strain was cultivated at 28 °C and 120 rpm in the PDB medium for 7 days, and total RNA was then extracted. A series of primers (Supplementary Table S4) for each gene were used to amplify cDNA of each gene, and the intron positions were then confirmed after DNA sequencing.

Construction of the GS-NX strain.
In our previous study, we successfully implemented 6-MSA biosynthesis in an engineered P. pastoris carrying Aspergillus nidulans PPTase-encoding gene npgA and A. terreus 6-MSAS-encoding gene atX 10 . Given that several genes need to be expressed in P. pastoris and selective markers were limited, npgA and atX were then inserted into one plasmid in this case. The npgA and atX expression cassettes with the AOX1 promoter (P AOX1 ) and AOX1 terminator were amplified from plasmids pPIC3.5K-npgA and pPICZ B-atX 10 , respectively. Two pairs of primers TT-AOX-F/TT-HIS4-R and Amp-AOX-F/AOX-TT-R were employed in PCR, and DNA fragment 1 (2353 bp) and fragment 2 (6785 bp) were obtained. Moreover, the selective marker HIS4 was amplified from plasmid pPIC3.5 K with primers TT-HIS4-F and ori-HIS4-R (fragment 3). Replicon ori and a selective marker-ampicillin resistance gene AmpR with the AmpR promoter-(fragment 4) were amplified together from plasmid pPIC3.5 K with primers HIS4-ori-F and AOX-Amp-R. After that, fragments 1, 2, 3, and 4 were assembled, leading to the expression plasmid pPICβ-npgA-atX. It was transfected into E. coli TOP10. After PCR verification with primers 5AOX1 and 3AOX1 and DNA sequencing, the correct plasmid was linearized by means of BspEI and transfected into wild-type P. pastoris GS115 by electroporation. The histidine auxotroph was used for screening of transformants for those positive for GS115-NpgA-AtX (GS-NX). The strains were then verified by genotyping PCRs (Supplementary Fig. S10).

Construction of strains GS-NXA, GS-NXE, GS-NXAE, and GS-NXAG.
Genes atA and atE were obtained by means of primers ZB-atA-F and ZB-atA-his6-R or ZB-atA-F and ZB-atE-his6-R from A. terreus cDNA, respectively. They were then ligated to the pPICZ B vector digested with EcoRI and XhoI via seamless assembly, leading to expression plasmids pPICZ B-atA and pPICZ B-atE. The plasmids were transfected into E. coli TOP10 and positive transformants with correct plasmids were identified by colony PCR with primers 5AOX1 and 3AOX1 and DNA sequencing. The GAP promoter as an integration locus was amplified from P. pastoris genomic DNA with primers ZB(BglII)-GAP-F and GAP-AOX-R. It was then inserted into pPICZ B-atA digested with BglII to obtain plasmid pPICZ B-atA-GAP. The atE gene containing the promoter and terminator was amplified from pPICZ B-atE with primers ZB-BglII-AOX-F and TT-GAP-R, and next inserted into pPICZ B-atA-GAP digested with BglII, yielding expression plasmid pPICZ B-atA-GAP-atE. The correct plasmids were then identified by colony PCR with primers 5AOX1 and 3AOX1 and DNA sequencing. After that, pPICZ B-atA and pPICZ B-atE were linearized with PmeI, and pPICZ B-atA-GAP-atE was linearized with AvrII, and transfected into strain GS-NX by electroporation. Zeocin at a final concentration of 100 μg/mL served for selection of positive transformants of GS115-NpgA-AtX-AtA (GS-NXA), GS115-NpgA-AtX-AtE (GS115-NXE), and GS115-NpgA-AtX-AtA-AtE (GS-NXAE). For construction of GS115-NpgA-AtX-AtA-AtG (GS-NXAG), the atG gene with a flanking sequence was first cloned by means of primers of 3.5k-AOX-atG-F and 3.5K-his6-atG-R. It was then inserted into pPIC3.5 K digested with EcoRI and BamHI, thereby producing expression plasmid pPIC3.5K-atG-his6. The plasmid was next linearized with SalI and transfected into strain GS-NXA by electroporation to generate strain GS115-NpgA-AtX-AtA-AtG (GS-NXAG). The strains were verified by genotyping PCRs (Supplementary Figs S11-S13).

Construction of strains GS-NXAEG, GS-NXAED, GS-NXAEGD.
The GAP promoter as an integration locus was amplified from P. pastoris genomic DNA with primers 3-pGGAP-F and 3-pGGAP-R. After that, it was digested with SacI and SpeI and ligated into the same sites of opened vector pAG32 to obtain vector pAGG (Hyg r ). The atG gene was cloned from A. terreus cDNA using primers ZB-atG-F and ZB-atG-R and inserted into pPICZ B digested with EcoRI and XhoI. The atD gene was cloned from A. terreus cDNA by means of primers pAG-atD-F and pAG-atD-his6-R and inserted into the pAGG plasmid digested with SalI and BamHI. Thus, expression plasmids pPICZ B-atG and pAGG-atD were obtained. The correct plasmids were then identified by colony PCR with primers 5AOX1 and 3AOX1 and DNA sequencing. Considering the selection marker and His tag, the atG containing a promoter and terminator was amplified from pPICZ B-atG with primers Scientific REPoRTS | (2018) 8:2116 | DOI:10.1038/s41598-018-20514-x Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information files). Genes re-annotated are also deposited in GenBank and the assigned accession numbers are provided in this published article.
Ethical approval and informed consent. We declare that this paper does not report any data collected from humans or animals.