Identification of the Pyranonigrin A Biosynthetic Gene Cluster by Genome Mining in Penicillium thymicola IBT 5891

Introduction

Pyranonigrins are a family of antioxidative compounds that are reported to be secondary metabolites produced by Aspergillus niger, all featuring a unique pyrano[2,3-c]pyrrole bicyclic skeleton which is rarely found in nature.^1–4 Of this family, pyranonigrin A (Figure 1) was shown to have the activity as a 1,1-diphenyl-2-picrylhydrazyl radical scavenging reagent.⁵ Previously, a biosynthetic gene cluster from A. niger ATCC 1015 was transcriptionally activated by overexpressing the pathway specific transcriptional regulator, leading to the identification of a new member of this family, pyranonigrin E.⁶ Though the biosynthetic pathway of pyranonigrin E was well-established based on gene knockout studies^6,7, the biosynthesis of pyranonigrin A remains unclear. Recent isotope labeling studies showed that the backbone of pyranonigrin A is derived from four units of acetate and one unit of glycine⁸, suggesting a polyketide synthase (PKS) and nonribosomal peptide synthetase (NRPS) hybrid enzyme may be involved.

Figure 1.

Structures of the natural products containing a pyrano[2,3-c]pyrrole core. Previously, the biosynthetic gene clusters of pyranonigrin E and curvupallide C were characterized from Aspergillus niger and Curvularia pallescens, respectively.^6,7,19

PKS-NRPS are bi-modular megasynthases that make a variety of bioactive natural products in fungi.⁹ The PKS module is typically a highly reducing polyketide synthase that functions iteratively to produce a polyketide chain with mixed degrees of α-methylation and β-reduction during each chain elongation cycle. The NRPS module captures the completed polyketide chain in the form of a thioester and condenses with a specifically activated amino acid (by the adenylation domain) to form an amide bond. This product can then be released by a reductive domain followed by Knoevenagel condensation to yield a pyrrolinone product, or undergo a Dieckmann cyclization to afford a tetramic acid product.⁹ Subsequent oxidative and other tailoring steps can lead to a mature product. Because of the large number of structural diversity that is achievable through the PKS and NRPS modules, fungal PKS-NRPS can make diverse compounds, including cytochalasins, decalins, long-chain N-acyl amides, pyridones, etc.^9–10 Genome sequencing efforts have identified that filamentous fungi encodes a significant number of PKS-NRPS containing gene clusters, many of which have unknown function or are not transcriptionally active under standard cultivation conditions.¹¹ Consequently, there is of significant interest to mine these clusters for the production of encoded natural products, using either genetic methods in the natural host, or heterologous expression techniques in model hosts. ^12–13 Here we demonstrate a cryptic PKS-NRPS pathway in the plant endophyte Penicillium thymicola can produce pyranonigirin A when reconstituted in a heterologous host. This allowed us to perform media optimization with the native host to activate the cluster.

Results and discussion

Recently, we sequenced the genome of Penicillium thymicola IBT 5891 and characterized the biosynthetic gene cluster of penigequinolones.¹⁴ Bioinformatic analysis of the genome revealed that there are a total of 99 secondary metabolite biosynthetic gene clusters predicted by antiSMASH, including 23 PKSs, 17 NRPSs, 6 PKS-NRPSs, and 13 terpenes, which far exceeds the number of reported natural products isolated from this strain to date. In an effort to map the products of these secondary metabolite clusters, we focused on the PKS-NRPS clusters in particular due to structural and functional diversity of the products. One such gene cluster, named pyr cluster, was found on scaffold 581 (Figure 2A) and contains an interesting combination of biosynthesis genes. A putative boundary was assigned based on the likelihood of biosynthetic enzymes encoded. Within the boundary are genes encoding a PKS-NRPS (pyrA), its partner α/β hydrolase (pyrD) that may be involved in cyclization⁷, multiple redox enzymes such as cytochrome P450 (pyrB), two flavin-dependent enzymes (pyrC and orf2), GMC oxidoreductase (orf3), and an inosine-5’-monophosphate dehydrogenase (IMPDH) (orf1). Interestingly, the P. thymicola genome encodes two copies of IMPDHs (a housekeeping copy and orf1). This is similar to the genomes of mycophenolic acid (MPA) producing strains^15–17. MPA is a known inhibitor of IMPDH and previous studies established that the IMPDH encoding gene found in the MPA pathway confers resistance of MPA^15–17. This observation suggests that the compound produced by pyr cluster in P. thymicola may have similar IMPDH-inhibiting bioactivities.

Figure 2.

The pyr cluster is responsible for the biosynthesis of pyranonigrin A. (A) The pyr cluster identified from P. thymicola IBT 5891. orf1 encodes IMPDH, pyrA encodes PKS-NRPS, pyrB encodes P450, pyrC encodes flavin-dependent monooxygenase, pyrD encodes hydrolase, orf2 encodes flavin binding oxidoreductase, orf3 encodes GMC oxidoreductase. (B) Characterization of pyrA in A. nidulans A1145. Shown are chromatograms of extracts from: i) A. nidulans expressing pyrA; ii) A. nidulans containing the control vector. (C) Reconstitution of pyr cluster in A. nidulans A1145. Shown are chromatograms of extracts from: i) A. nidulans expressing the putative pyr cluster; ii) A. nidulans containing control vectors.

BLAST¹⁸ search results indicated that PyrA exhibits moderate sequence similarity to PynA (50% identity/67% similarity), and An18g00520 (61% identity/75% similarity), both of which are from A. niger A1179. All three enzymes contain the same domain organizations. In the PKS modules are ketosynthase (KS), acyltransferase (AT), dehydratase (DH), enoylreductase (ER), ketoreductase (KR), and acyl carrier protein (ACP) domains. In the NRPS domains are the canonical condensation (C), adenylation (A), thiolation (T) domains. The last domain in the PKS-NRPS is a reductase/Dieckmann cyclization (R/DKC) domain. Previously, PynA was shown to be responsible for the biosynthesis of pyranonigrin J, which is a precursor to pyranonigrin E.⁷ An18g00520 was shown to be responsible for the production of tetramic acid compound 1 (structure shown in Figure 2), which is proposed to be a precursor to pyranonigrin S.¹⁹ Based on this information, we proposed that PyrA may similarly produce a tetramic acid compound that is structurally similar to the compounds produced by PynA and An18g00520. To confirm the function of PyrA, we introduced pyrA into the heterologous host A. nidulans A1145 using an episomal vector with that expresses pyrA under the control of gpdA promoter. Following four days of growth, the metabolites were extracted and analyzed by LC-MS. Compared with the host containing a control vector, A. nidulans expressing PyrA showed clear accumulation of a new metabolite with the molecular weight (MW) of 193 (Figure 2B). Structural analysis (Table S3 and Figure S5-S9) revealed that this compound is a tetramic acid cyclized from a tetraketide-glycine adduct (Figure 2B) and is identical to the product of An18g00520. This result indicated that the unknown natural product encoded by pyr cluster shares the same biosynthetic precursor with pyranonigrin S.

As the function of PyrA was characterized, we then turn to identify the natural product encoded by the entire pyr gene cluster. All the genes, orf1-3 and pyrA-D, from the pyr cluster were introduced into the A. nidulans host on three vectors. Analysis of the extract showed that a new metabolite with the MW of 223 was accumulated (Figure 2C), though the yield is relatively low (0.1 mg/L). Structural elucidation was carried out after sufficient amount of this compound was purified. Based on the NMR data (Table S4 and Figure S10-S11), the carbon skeleton of this compound is identical to the known natural product, pyranonigrin A. The optical rotation of this compound is [α]_D²⁶ +20 (c=10.0 mg/dL, MeOH), which is consistent with the reported value of pyranonigrin A ([α]_D²⁵ +38, c=1.0 mg/dL, MeOH)³. Thus, we assigned this new metabolite to be pyranonigrin A. To the best of our knowledge, this is the first time a biosynthetic gene cluster was identified for the biosynthesis of pyranonigrin A. Based on the biosynthetic pathway of pyranonigrin E^6,7, we proposed that three or four genes, instead of all the introduced genes, are sufficient for the production of pyranonigrin A. However, the low yield in the A. nidulans host impaired further functional analysis.

To clarify the minimal number of genes required for biosynthesis of pyranonigrin A, we sought to activate the production of pyranonigirin A in the original host, P. thymicola. When we cultured the strain in the medium reported for high production of Fumiquinazoline F and alantrypinone²⁰, we could not detect the production of pyranonigrin A (Figure 3). We then screened numerous other fungal culture media, but production of pyranonigrin A was still not detected (data not shown). Fortuitously, the production (3 mg/L) of pyranonigrin A was detected when we added starch into the culture medium that we used for production of alantrypinone, an unrelated alkaloid synthesized by this strain (Figure 3). To establish the link between the production of pyranonigrin A and the pyr cluster in P. thymicola, the PKS-NRPS gene pyrA was deleted to yield the mutant strain ΔpyrA by gene replacement (Figure S1). LC-MS analysis showed that deletion of pyrA completely abolished the production of pyranonigrin A (Figure 3). RT-PCR analysis of the wild type strain grown in the presence of starch indicated that only orf1 and pyrA-D were transcribed under conditions for pyranonigrin A production (Figure S2), suggesting that orf2 and orf3 are not involved in the biosynthesis of pyranonigrin A. However, these results cannot rule out the possibility that Orf2 and Orf3 may act as tailoring enzymes to modify pyranonigrin A to more complex structures under other conditions. To test this hypothesis, we expressed Orf2 and Orf3 in Saccharomyces cerevisiae BJ5464 and tested their function through biotransformation using pyranonigrin A as the substrate. LC-MS analysis results indicated that neither Orf2 nor Orf3 could catalyze the conversion of pyranonigrin A to other products (Figure S3). These results strongly suggest that pyranonigrin A is the end product of the pyr cluster and narrowed down the gene cluster to five genes, orf1 (encoding IMPDH) and pyrA-D.

Figure 3.

Production of pyranonigrin A from Penicillium thymicola. i) P. thymicola cultured in the medium supplemented with starch; ii) The ΔpyrA mutant strain cultured in the medium supplemented with starch; iii) P. thymicola cultured in the medium without starch.

IMPDH, encoded by orf1, is a key enzyme involved in the de novo biosynthesis of guanine nucleotides, which catalyzes the oxidation of inosine monophosphate (IMP) to xanthosine monophosphate (XMP).²¹ IMPDH is widely considered to be an attractive target for the discovery of drugs in antiviral, antibacterial and anticancer therapeutic areas.²² Discovery of novel small molecule IMPDH inhibitors may lead to the development of new drugs in these clinical areas with enhanced performance. Genome sequenced filamentous fungi typically contain only a single IMPDH encoding gene (the housekeeping copy). However, the genomes of MPA producing strains and some fungal species belonging to Penicillium subgenus Penicillium contain two copies of IMPDH encoding genes.¹⁶ In the genome of P. thymicola, we found two copies of IMPDH encoding genes including orf1, which is encoded alongside pyrA. A reasonable hypothesis is that the natural product encoded by pyr gene cluster could have potent bioactivity targeting IMPDH as exemplified by the MPA biosynthetic gene cluster. To test this hypothesis, we expressed orf1 in S. cerevisiae and the housekeeping copy of the IMPDH gene from P. thymicola (named as IMPDH-g) in E. coli. Then we purified the enzymes and tested their activities. The results from the in vitro assays demonstrated that both enzymes can catalyze the conversion of IMP to XMP (Figure S4). However, the presence of pyranonigrin A, even at a high concentration (30 μM, 10 times of the enzyme concentration used) in the reaction mixtures, had no effect on the enzymatic activities with Orf1 and IMDPH-g (Figure S4). Thus, we validated that pyranonigrin A has no IMPDH inhibition bioactivity and orf1 is not a self-resistant gene. The emergence of two copies of IMPDH encoding genes in P. thymicola may be due to gene duplication, as reported in other MPA non-producing fungal species.¹⁶

Based on the findings from both in vivo and in vitro studies, we concluded that the pyr cluster encoded the biosynthesis of pyranonigrin A and only four genes, pyrA-D, are required. A comparison with the biosynthetic pathways of other known pyranonigrin family members, such as pyranonigrin E^6,7 and curvupallides¹⁹, can lead to a similar proposal for the biosynthesis of pyranonigrin A (Figure 4). The PKS-NRPS (PyrA) incorporates one unit of acetyl-CoA, three units of malonyl-CoA, and one unit of glycine to assemble the combined backbone. Considering the similarities between the biosynthesis of pyranonigrin A and curvupallides, the R/DKC domain of PyrA is likely to catalyze the product release via Dieckmann cyclization to afford tetramic acid 1. Nevertheless, we cannot exclude the alternative that the hydrolase PyrD functions similarly to PynI, an α/β hydrolase from the biosynthetic pathway of pyranonigrin E, which was proposed to be involved in tetramic acid formation⁷. An epoxidation-mediated cyclization of 1 to form the pyrano[2,3-c]pyrrole core could then be performed by the flavin-dependent monooxygenase PyrC, as a similar reaction was proposed to be catalyzed by its homologue, PynG, in the pyranonigrin E pathway⁷. The only remaining enzyme, P450 PyrB, may catalyze the last two modification steps, dehydrogenation and hydroxylation, to complete the biosynthesis of pyranonigrin A.

Figure 4.

The proposed biosynthetic pathway of pyranonigrin A.

Conclusion

In summary, we have identified the biosynthetic gene cluster of pyranonigrin A from the genome of P. thymicola IBT5891, which was not known to be a producer of pyranonigrin family compounds. A BLAST search in the sequenced fungal genomes using the pyr gene cluster as a query indicated that homologous clusters are present in numerous different fungal species, suggesting the opportunity to discover new natural products belonging to the pyranonigirin family.

Significance.

We identified a polyketide synthase and nonribosomal peptide synthetase (PKS-NRPS) hybrid gene cluster from the genome of Penicillium thymicola through genome mining. Heterologous expression of this cluster leads to the production of pyranonigrin A. A series of experiments established that only four genes are sufficient to biosynthesize pyranonigrin A. Based on the results from the current study, a biosynthetic pathway of pyranonigrin A is proposed.