MrPKS2 synthesized T4HN in S. cerevisiae
To identify the products of MrPKS1 and MrPKS2, the intron-free MrPks1 and MrPks2 open reading frames were expressed in S. cerevisiae BJ5464-NpgA, a host adopted to reconstitute fungal polyketide biosynthetic pathways.(Yu et al., 2017; Xie et al., 2018; Chen et al., 2019; Tang et al., 2019; Wang et al., 2019) In previous study, heterologous expression of MrPks2 in A. nidulans showed no product peak.(Zeng et al., 2018) In contrast, the yeast strain successfully produced the major product of MrPKS2 (Fig. 1A). The mass-to-charge ratio (m/z) ([M-H]− ions) of this product was 205.0142 atomic mass unit (amu), consistent to a molecular formula of C10H6O5 (calcd. m/z 205.0136, mass error 2.4 ppm). This formula and the HRMSMS spectrum were in accordance with flaviolin that can be derived from spontaneous oxidization of T4HN (Fig. 1B). Similarly, treatment of C. lagenarium with the DHN pathway inhibitor, tricyclazole, resulted in the accumulation of flaviolin instead of T4HN.(Vagstad et al., 2012) Thus, our result shows that MrPKS2 is a T4HN synthase likely belonging to the penta-ketide class with ClPKS1 and PfmaE.(Fujii et al., 2000; Watanabe and Ebizuka, 2004; Zhang et al., 2017)
MrPKS1 synthesized new octaketides in S. cerevisiae
Different with MrPks2, the recombinant yeast strain expressing MrPks1 afforded three product peaks, 1–3 (Fig. 1A). The positive mode HRESIMS spectra of the 1–3 displayed [M + H]+ ions at m/z 319.0784, 303.0871 and 301.0727, corresponding to the molecular formula of C16H14O7 (mass error − 10.30 ppm), C16H14O6 (mass error 0.66 ppm) and C16H12O6 (mass error 4.98 ppm), respectively (Fig. 1C). As the major product of MrPKS1, 1 showed m/z value in agreement with the mass of an octaketide with new structure. Its structure was elucidated by NMR (Fig. S1 and Table S1). The 1H NMR data for compound 1 (Table S1-1) revealed the presence of three aromatic proton signals [δH 6.19 (1H, s), 6.39 (1H, s), 6.39 (1H, s)] and one singlet methyl signal (δH 2.17, s). The 13C NMR spectrum (SI Table S1-2), with the aid of HSQC, showed 16 resonances ascribed to a methyl, two methylenes, three methines, and ten quaternary carbons. Analysis of 1D and 2D NMR data (Fig. S1 and Table S1) showed high similarities between 1 and YWA1,(Fujii et al., 2001) except for the extra acetyl group. The position of the acetyl group was determined at C-11 according to the HMBC correlations from H2-11 and H3-13 to the ketone carbonyl at δC 205.0. All of the NMR signals and correlations are consistent with the structure of 1 (Fig. 1C), which is equivalent to attaching one more keto-unit to the structure of YWA1, the heptaketide in DHN melanin pathway. YWA1 can be reversibly convertible to 2-malonyl-T4HN. Similarly, 1 may also be converted from a similar intermediate (Fig. 1C).
MrPKS1 is the first reported natural T4HN synthase producing octaketide. MrPKS1 maintained precise control of cyclization regioselectivity. Octaketides that contain alternative cyclization patterns was not detected. Similar compounds except for YWA1 have also been reported, i.e., the intermediate (5) in the biosynthesis of pre-bikaverin by PKS4 from Giberella fujikuroi (Fig. 2C).(Ma et al., 2007; Newman et al., 2014) The compound 5 has one more keto-unit attached to the end of 1. It is highly possible that the poly-β-keto processing pathway of 1 is identical to those during YWA1 and pre-bikaverin synthesis catalyzed by A. nidulans wA synthase and G. fujikuroi PKS4 synthase, respectively. In brief, cyclization of the first benzene ring is facilitated by the aldol condensation reaction between C-2 and C-7, which requires synthesis of the full length polyketide chain. This is followed by Claisen-like condensation between C-10 and C-1 to generate the bicyclic T4HN scaffold (Fig. 2C).(Ma et al., 2007; Newman et al., 2014) Attack of C-13 carbonyl by the nucleophilic C-9 phenol forms the tricyclic naphthopyrone.
The highly similar high resolution tandem mass (HRMS/MS) fragmentation pattern of 1–3 demonstrated their structure similarity (Fig. 2B and Table S2). Similar to 1, the structure of 2 can be elucidated by the NMR spectra and the HRMS/MS spectrum (Fig. 2B, Fig. S1 and Table S1). In addition to the same aromatic ring skeleton with compound 1, the presence of 2-(2-hydroxypropyl)-γ-pyrone group can be confirmed in compound 2 according to the HMBC correlations from H2-11 to C-2 and C-3, from H-3 to C-4 and C-4a, associated with 1H-1H COSY correlations from H-12 to H3-13 and H2-11 (more details in supporting information). Therefore, the structure of compound 2 was determined as shown in Fig. 1.
According to the structures of 1 and 2, the structure of 3 can be proposed. The hydroxyl group at C13 position of 1 is easily dehydrated to form 3, which can be demonstrated by the presence of 301.0728 fragment ion in the high resolution tandem MS (HRMS/MS) spectra of 1 (mass error − 5.31 ppm) (Fig. 2B). This ion was 18 amu less than the parent ion of 1, and was in accordance to the m/z of 3 (301.0727, [M-H]+). The carbonyl group of 3 at C15 position was reduced to hydroxyl group to form 2. This was corresponding to the near identical fragmentation pattern in HRMS/MS spectra of 2 and 3, except that the 2’s spectra had the − 18 amu fragment ion (m/z 285.0753, the loss of hydroxyl group at C15).
The reactivity of MrPKS1 products
The time course study showed that 1 can be detected after 24 h fermentation with the MrPks1 expressed, reached to a plateau at around 36 h and remained a similar level until 48 h (Fig. 2A). Photos taken at the same periods were in accordance with the product profiles, which showed color difference at 36 hrs and 48 hrs with yellow to brownish yellow color (Fig. 2B). The time course profiles also showed the emergence of minor products, 2 and 3. The dehydration from 1 to 3 and the reduction from 3 to 2 were likely to occur spontaneously. The crude extract of MrPKS1-expressing yeast showed increased amount of 2 and 3 after 1 week (Fig. S2A-B); the purified 1 product in DMSO also gave rise to 2 and 3 (Fig. S2C-D).
Expression of MrPks1 in A. nidulans A1145 resulted in the product 4 in previous study(Zeng et al., 2018) as well as in ours (Fig. 2C and Fig. S3). Trace amount of 4 was also detected in the yeast crude extract expressing MrPks1 with the same retention time and HRMS/MS spectrum as those detected in A. nidulans. The control strains of A. nidulans and yeast both lack this product peak, which confirmed the origin of 4 to be MrPKS1. We also conducted the time course study of MrPKS1 expressed in A. nidulans (Fig. S3), which showed the occurrence of both 1 and 4 after 3-d fermentation. However, the 1 disappeared after 5 d in A. nidulans but still accumulated in yeast as in Fig. 1. In a word, these results showed that 1 is closer to the authentic product of MrPKS1. Although the detailed routes from 1 to the other forms are still unknown, the high reactivity of MrPKS1’s products can be demonstrated.
Synteni analysis of loci encoding the T4HN synthases
Fungal genomes prefer to encode the genes that collaborate to synthesize a secondary metabolite in adjacent loci, so called biosynthetic gene cluster (BGC). Thus the gene composition of BGC provides insights into the biosynthesis pathway, and the gene rearrangement gives clues to evolution process. To better understand the divergence of Pks1 and Pks2 pathways, we performed phylogenetic and syntenic analysis of the homolog BGCs in ascomycete fungi that has been reported to biosynthesize DHN melanin. 19 BGCs containing homologous T4HN synthases were submitted to CORASON.(Navarro-Munoz et al., 2020) Sequences from 15 species were aligned and used to construct the phylogenetic tree with bootstrap values > 88% except for PfmaE (Fig. 3). Since the BGCs of 6 Metarhizium species are highly conserved, three representative species, the broad insecticidal M. robertsii and M. anisopliae as well as the host-specific M. acridum, were selected in this analysis. In addition to the T4HN synthases, homologs of tailoring enzymes were also searched with e-value cut-off of 10− 1. In accordance to the previous study using ketosynthase domain sequences to construct the tree,(Zeng et al., 2018) the PKS1s and PKS2s from the Metarhizium species formed a clade with the taxonomically close Hypocreales fungus, Fusarium gramineanum, with high confidence (96% bootstrap value) (Fig. 3). This clade was divided into two well-supported sub-clades corresponding to PKS1 and PKS2, respectively (Fig. 3). The T4HN synthase from F. gramineanum is more similar to PKS1, suggesting a closer relationship of PKS1 to the common ancestor PKS. Basal to this clade was the clade of the reference BGCs of Alb1 and wA from the Eurotiomycetidae A. fumigatus and A. nidulans. The Alb1 BGC is composed of seven genes encoding proteins involved in DHN-melanin biosynthesis: from downstream to upstream were the T4HN synthase (Alb1), SCD (Arp1), THNR (Arp2) and serine hydrolase (Ayg1) that required for DHN synthesis, as well as two oxidases/laccases (Abr1 and Abr2) and a transcriptional factor.(Tsai et al., 1999) Comparison of the genomic context of PKS1 and PKS2 to that of Alb1 indicates that they separately “inherited” different tailoring enzymes after the duplication event: PKS2 took THNR and SCD while PKS1 adopted the laccase/oxidase (MLac1) that is homologous to Abr2. The serine hydrolase was no longer needed since PKS2 can directly synthesize T4HN. The “remnant” of Abr1 gene homolog in the BGC from M. anisopliae suggests the gradual loss of this gene. Further examination of the other remote clades/BGCs showed that Abr2 (MLac1) and THNR were two genes that frequently occurred, which also indicates their common origin and departure in Metarhizium species. Deletion of the abr2 gene in A. fumigatus changed the gray-green conidial pigment to a brown color compared with wild-type conidia,(Sugareva et al., 2006) suggesting that Abr2 may be the laccase that polymerize DHN. Thus it is reasonable that the PKS2 BGC “cross-talk” with the PKS1 BGC and use the MLac1 to conduct the polymerization.
In addition to the proteins present in the Alb1 BGC, the Pks1 BGC also includes a gene encoding a protein with an EthD domain (Pfam07110); this domain is involved in the degradation of ethyl tert-butyl ether.(Zeng et al., 2017) Previous study has shown that in the genomic context of MrPKS1, only MrPKS1, MrEthD and MLac1 appear to involve in the dark greenish color of M. robertsii: the mutants with MrPKS1 disrupted produced light red conidia; while the mutants with MrEthD or MLac1 disrupted had dark red conidia.(Zeng et al., 2017) It is very likely that MLac1 can polymerize the phenolic product of PKS1, while EthD participate/aid this process. However, whether these three genes are sufficient to the biosynthesis of melanin/pigment (i.e., as a DHN bypass similar to the wA BGC), or in another case the 1 intermediate convergently resulted in DHN still remains a question.
The biosynthesis of DHN by the MrPks2 gene cluster
Based on the synteni analysis, we can hypothesize that the MrPks2 gene cluster is capable to biosynthesize DHN, which could be further polymerized into melanin through the “cross-talk” with the MLac1 encoded in the MrPks1 gene cluster. To test this hypothesis, the intron-free MrTHNR and MrSCD open reading frames were both cloned from the reverse-transcribed cDNA; this cDNA was prepared during cuticle infection.(Chen et al., 2015) Sequencing of the cloned fragments confirmed the presence of 2 introns in both MrTHNR and MrSCD, resulted in 109 bp and 150 bp length shorter compared to the genes encoded in genomic DNA (Fig. S4). Then the MrTHNR and MrSCD were co-expressed with MrPks2 separately or collectively in S. cerevisiae BJ5464-NpgA. Strains expressing MrPks2 and MrTHNR yielded only trace amount of flaviolin; instead, produced a major product with m/z ([M-H]− ion) of 193.0494 that is correspondent to scytalone (C10H10O4, calcd. m/z 193.0506, mass error − 6.22 ppm) (Fig. 4). The structure of this product was further elucidated by analyzing the NMR spectroscopic data of the purified compounds (Fig. S5, Table S4). The expected product, DHN, was also detected in yeast expressing MrPks2, MrTHNR and MrSCD, after 36 h and 48 h fermentation (Fig. 4). The HPLC-HRMS graphs clearly showed the presence of a product peak with [M-H]− of 159.0430 (molecular formula C10H8O2, calcd. m/z 159.0446, mass error 2.4 ppm). The HRMS/MS spectra (Fig. 4C) and the identical retention time in HPLC graph (Fig. S5) showed that this product is highly possible to be DHN. Therefore, the BGC of MrPKS2 is fully functional to generate DHN and participates in the melanin biosynthesis; the reason why expressing MrPks2 in A. nidulans did not show detectable product(Zeng et al., 2018) may be because the T4HN produced by MrPKS2 was utilized by the host melanin pathways.
The reason that the disruption of MrPks2 gene did not cause color change to Metarhizium species is due to its low expression during saprophytic growth. According to RT-PCR (reverse transcription polymerase chain reaction) result (Fig. S4), the genes MrTHNR and MrSCD were strictly regulated and were only transcribed during infection. This is also the reason why blocking the DHN biosynthesis did not cause physiological change to the Metarhizium spp. However, the MrPks2 gene showed transcription on regular fungal media (SDY and PDA), which may be the origin of the brownish color when the MrPks1 was knocked out from the genome.
The last missing puzzle of the DHN melanin pathway in M. robertsii was the oxidative enzyme that polymerizes DHN. According to the synteni analysis, MLac1 in MrPKS1 BGC can be one candidate. However, co-expressing the MrPKS2 BGC with MLac1 did not produce new product peak after 36-h or 48-h fermentation (Fig. 4A). When expressing MrPks2, the yeast fermentation was dark brown, in accordance with the color of flaviolin. Sequential addition of MrTHNR and MrSCD to the pathway lightened the color to be reddish-brown. The color of the fermentation (Fig. 4B) did not change significantly when the MLac1 functioned. To further verify this result, DHN compound was fed to the yeast expressing Mlac1, which displayed similar product profile with the yeast control expressing empty plasmid (Fig. S6). Thus, the DHN product could not be transformed by MLac1 in yeast.
The expression of the MrPks1 gene clusters
The pigment pathway following MrPKS1 is more complicated than that of MrPks2. According to the synteni analysis and previous study, there are at least two possibilities: 1) the product 1 merges into the DHN pathway through genes in MrPks2 BGC; 2) 1 diverged into a different pathway utilizing MLac1 and MrEthD. Based on the RT-PCR analysis (Fig. S4), the first hypothesis is not likely because that MrTHNR and MrSCD were not activated on regular incubation conditions. As expected, the combinatorial expression of MrPks1 with MrTHNR and/or MrSCD failed to further modify 1 (Fig. S7).
To test the second hypothesis, we co-expressed MrPks1 with combinations of MLac1 or MrEthD genes both cloned from the cDNA of M. robertsii (Fig. 5). MrLac1 did not accept 1–3 as substrate (Fig. 5). Addition of MrEthD eliminated 1, the major product of MrPKS1, while significantly increased the yield of 2 and 3 (Fig. 5D). No new products were detected. The color of the fermentation did not change significantly when adding MrEthD and/or MLac1 to the pathway. After individually fed to the yeast expressing MrEthD, 1 disappeared from the product profile without new product (Fig. S7). This result showed that 1 was the substrate of MrEthD and was transformed into a form that cannot be extracted by organic solvent. According to previous study, MLac1 also contributes to the pigmentation of Metarhizium fungi.(Fang et al., 2010; Zeng et al., 2017) It is reasonable to hypothesize that Mlac1 can polymerize the activated form of 1 by MrEthD. However, further addition of MLac1 to MrPKS1 and MrEthD did not produce new product peak, nor changed the color of the yeast fermentation in yeast (Fig. 5).
Biological function of the MrPKS1 and MrPKS2 products
The intrinsic reason why the pathogenic fungi developed two different melanin biosynthesis pathways is an interesting topic in understanding the fungal adaptation to host. The polymerized DHN melanin has been extensively proved to protect fungal cells from biotic and abiotic stresses, especially important to pathogenic fungi. In this study, the small molecule intermediates, i.e., DHN and T4HN (flaviolin), also can protect the yeast cells against UV-B radiation to different extend (Fig. 6). The yeast producing T4HN (expressing MrPks2) showed strongest UV-B radiation resistance than the others under 30-min (0.36 J cm− 2) dose, while the yeast expressing MrPks2 along with MrTHNR showed weaker protection effect. Similarly, the yeast expressing MrPks1 or MrPks1 and MrEthD also had stronger resistance to UV-B radiation, showing the ability of the intermediate products of MrPks1 BGC to protect cell against this stress. In contrast, the MrPks1 or MrPks2 pathway products did not show protective affect to yeast cells under heat stress (Fig. S8). This experiment was repeated three times with two different clones each time and similar trends were observed.