Junko
Yaegashi
a,
Jillian
Romsdahl
a,
Yi-Ming
Chiang
ab and
Clay C. C.
Wang
*ac
aDepartment of Pharmacology and Pharmaceutical Sciences, School of Pharmacy, University of Southern California, Los Angeles, California 90089, USA. E-mail: clayw@usc.edu
bGraduate Institute of Pharmaceutical Science, Chia Nan University of Pharmacy and Science, Tainan 71710, Taiwan
cDepartment of Chemistry, College of Letters, Arts and Sciences, University of Southern California, Los Angeles, California 90089, USA
First published on 6th August 2015
Meroterpenoids are a class of secondary metabolites that are produced from polyketide and terpenoid precursors. 15-Deoxyoxalicine B (1) belongs to one structural group consisting of a unique pyridinyl-α-pyrone polyketide subunit and a diterpenoid subunit connected through a characteristic asymmetric spiro carbon atom. An understanding of the genes involved in the biosynthesis of this class of compounds should provide a means to facilitate engineering of second-generation molecules and increasing production of first-generation compounds. We found that the filamentous fungus Penicillium canescens produces 15-deoxyoxalicine B (1). Using targeted gene deletions, we have identified a cluster of 12 responsible contiguous genes. This gene cluster includes one polyketide synthase gene which we have designated olcA. Chemical analysis of wild-type and gene deletion mutant extracts enabled us to isolate and characterize 7 additional metabolites that are either intermediates or shunt products of the biosynthetic pathway. Two of the compounds identified have not been reported previously. Our data have allowed us to propose a biosynthetic pathway for 15-deoxyoxalicine B (1).
Although many meroterpenoids have been isolated and characterized, the genes involved in their biosynthesis have been revealed in only a few instances in fungi,12–17 while the majority of them have yet to be elucidated. Examples of such compounds are oxalicines A and B. They were first isolated from Penicillium oxalicum by Ubillas et al.18,19 and were the first of a rare skeletal class of diterpenic meroterpenoids (Fig. 1). Since then, other related compounds such as 15-deoxyoxalicines A, B and decaturins A–F have been isolated from P. decaturense,20P. thiersii,21 and/or P. oxalicum,22 and many of them were shown to have antiinsectan activity against the fall armyworm (Spodoptera frugiperda). These compounds form a structurally unique class of natural products, because their basic structure is composed of the following two subunits: a pyridinyl-α-pyrone polyketide subunit (Fig. 1, blue), and a diterpenoid subunit (Fig. 1, red). This pyridinyl-α-pyrone polyketide subunit itself is rare among natural products, having only been found in anibine, a plant metabolite,23 and pyripyropenes, a group of potent acyl-CoA cholesterol acyltransferase inhibitors isolated from Aspergillus fumigatus.12 The early steps of pyripyropene A biosynthesis has been determined previously,12 and the formation of the pyridinyl-α-pyrone polyketide subunit is most likely highly similar to that of oxalicines and decaturins. These steps involve the CoA ligase Pyr1 and PKS Pyr2 for the construction of 4-hydroxy-6-(3-pyridinyl)-2H-pyran-2-one (HPPO, Scheme 1), followed by the prenyltransferase Pyr6, FAD-dependent monooxygenase Pyr5, and terpene cyclase Pyr4 for the production of deacetyl-pyripyropene E, a precursor of pyripyropene A.12 However, instead of the incorporation of a farnesyl group in pyripyropene A biosynthesis, oxalicine and decaturin biosynthesis incorporate a geranylgeranyl group to produce the diterpenoid PKS hybrid. Moreover, the two subunits of the oxalicines and decaturins are connected through a unique and characteristic asymmetric spiro carbon atom, a feature that is lacking in pyripyropenes.
Fig. 1 Structurally related fungal meroterpenoids. The polyketide portion is shown in blue, and the diterpenic terpenoid portion in red. Compounds 1–6 were isolated in this study. |
In recent years, the Joint Genome Institute (JGI) has released a number of complete genome sequences for various Penicillium species, making it feasible to perform in silico analysis of potential SM gene clusters. Genome sequence data analysis shows that this species harbors many core SM biosynthesis genes, suggesting that it has the potential to produce far more SMs than is currently known. Taking advantage of the fact that early steps of pyripyropene A biosynthesis have been determined, we attempted to identify the biosynthetic genes for meroterpenoids that are structurally related to pyripyropene A in Penicillium species by genome mining. We found that P. canescens was the only Penicillium species genome-sequenced by JGI that harbored a complete set of genes homologous to genes in the pyripyropene A biosynthetic gene cluster, suggesting its potential to produce pyripyropene-like meroterpenoids. Interestingly, these genes were surrounded by genes encoding additional potential tailoring enzymes, suggesting that the final product of this biosynthetic gene cluster would be different from pyripyropene A. We grew P. canescens on various media and found that cultivation on Czapek's media induced the production of a compound that we isolated and identified as 15-deoxyoxalicine B (1). We then developed an efficient gene targeting system for P. canescens, and this allowed us to identify and characterize a gene cluster containing 12 contiguous genes that are involved in the biosynthesis of 15-deoxyoxalicine B (1). Several of the gene deletant strains accumulated chemically stable intermediates or shunt products in sufficient amounts for full structural characterization by spectroscopic methods. These strains were cultivated in large-scale, and we were able to isolate a total of 7 related compounds (2–8). Two of these compounds (7 and 8, Scheme 1) have not been reported previously. Combined with further bioinformatics analysis, we have proposed a biosynthetic pathway for 15-deoxyoxalicine B.
Gene designation | Protein IDa | A. fumigatus homologs (Afu6gxxxxx) | Similarity/identity (%) | Putative function |
---|---|---|---|---|
a Protein IDs as designated in JGI database. b These genes are predicted to be outside the gene cluster. | ||||
410805b | Cytoskeletal protein adducin | |||
olcB | 333321 | Cytochrome P450 CYP3/CYP5/CYP6/CYP9 subfamilies | ||
olcC | 351326 | Geranylgeranyl pyrophosphate synthase | ||
olcD | 437321 | 13950 (pyr4) | 57/41 | Integral membrane protein (terpene cyclase) |
olcE | 351329 | 13970 (pyr5) | 74/60 | FAD-dependent monooxygenase |
olcF | 367480 | Short chain dehydrogenase | ||
olcG | 393266 | Cytochrome P450 CYP3/CYP5/CYP6/CYP9 subfamilies | ||
olcH | 410812 | 13980 (pyr6) | 68/52 | Prenyltransferase |
olcA | 400488 | 13930 (pyr2) | 59/42 | PKS |
olcI | 437327 | 13920 (pyr1) | 71/58 | CoA ligase |
olcJ | 333335 | Cytochrome P450 CYP3/CYP5/CYP6/CYP9 subfamilies | ||
olcK | 367485 | Hydroxylase | ||
olcL | 351342 | Predicted transporter (major facilitator superfamily) | ||
367486b | Hypothetical protein |
Fig. 2 HPLC profiles of extracts from (A) parental strain and (B) olcAΔ strain as detected by UV-vis at 254 nm and mass spectrometry in positive mode of extracted ion chromatogram (EIC) at m/z = 504. |
Next, we set out to identify additional genes involved in the biosynthesis of 1. This process is facilitated by the fact that fungal secondary metabolite biosynthesis genes are usually clustered, and as mentioned previously, several pyr gene homologs were identified nearby (Fig. 3A and Table 1). In addition to these genes, we found that olcA is surrounded by genes that encode additional putative tailoring enzymes. We individually deleted 13 additional genes surrounding olcA. The 13 deletants were cultivated under 15-deoxyoxalicine B-producing conditions and their SM profiles were examined by LC-DAD-MS (Fig. 3B). Deletion of genes corresponding to protein IDs 333321, 437321, 351329, 367480, 393266, 410812, 437327, 333335, 367485, and 351342 resulted in complete elimination of 1. Deletion of protein ID 351326 greatly diminished production of 1. SM profiles remained unchanged after deletion of protein ID 410805 and 367486, indicating that these genes are not involved in the biosynthesis of 1 and that we have established the borders of the gene cluster. We now designate the genes surrounding olcA that are involved in the biosynthesis of 1 as olcB–olcL (Fig. 3A and Table 1).
Compound 7 has a molecular formula of C30H39NO4 as determined by its 13C-NMR and HRESIMS data, representing 12 indices of hydrogen deficiency (IHD). The 1H, 13C, and gHSQC NMR data of 7 (Tables S5 and S6 and Fig. S6, S7 and S12†) exhibited signals for five methyl groups [δH 0.76, 0.87, 0.93, 1.03, and 1.66 (each 3H, s)], one carbinol methine group [δH 3.12 (1H, dd, J = 12.0 and 4.8 Hz, H-27); δC 79.9 (C-27)], one downfield methylene [δH 3.23 and 3.26 (each 1H, d, J = 16.2 Hz, H2-15); δC 24.0 (C-15)], and the pyridinyl-α-pyrone moiety [δH 6.65 (1H, s, H-12), 7.54 (1H, dd, J = 8.4 and 4.8 Hz, H-5), 8.23 (1H, dt, J = 8.4 and 1.8 Hz, H-4), 8.59 (1H, dd, J = 4.8 and 1.8 Hz, H-6), and 8.98 (1H, d, J = 1.8 Hz, H-2)]. Besides the pyridinyl-α-pyrone moiety, there are two additional quaternary sp2 carbons [δC 129.2 (C-16) and 138.3 (C-14)] (Table S6†). Considering that there are only 12 IHD in compound 7 with one pyridinyl-α-pyrone moiety and one olefin, compound 7 should contain a tricyclic diterpenoid moiety. Comparing the 1H and 13C NMR data of 7 with those of decaturin E (Tables S5 and S6†),22 the main differences are that there is no olefinic proton in the diterpenoid moiety at ∼5.7 ppm and no spiro carbon at ∼100.0 ppm in compound 7, suggesting the existence of a tetrasubstituted olefin and absence of the spiro. In the gHMBC spectrum, the long-range 1H–13C correlations between H2-15 and three quaternary carbons (C-10, C-11, and C-14, Fig. S1a and S8–S11†) further connected the pyridinyl-α-pyrone and tricyclic diterpenoid moieties. Taken together, compound 7 was assigned to be a precursor before spiro formation (Scheme 1). Detailed analysis of gCOSY, gHSQC, and gHMBC NMR data (Fig. S1a and S8–S13†) allowed complete assignment of 7 and corroborated our proposed structure. We named compound 7 predecaturin E.
Compound 8, which we named decaturin G, has a molecular formula of C30H35NO5. The 1H and 13C NMR spectra (Tables S4–S6 and Fig. S14 and S15†) indicated its structural resemblance to decaturin D (5), which was also isolated from the olcJ deletion strain. One main difference was the presence of an allylic carbinol methine proton [δH 4.46 (1H, br t, J = 4.8 Hz, H-18)] in 8. The fact that H-19 is a doublet of doublets in 5, and a doublet in 8 indicated that 8 is an 18-hydroxy derivative of 5. A key NOE correlation between H-18 and Heq-25 indicated H-18 is an equatorial proton located at the α phase (Fig. S2†). gCOSY, gHSQC, and gHMBC NMR data (Fig. S1b and S16–S20†) allowed complete assignment of 8.
On the basis of the structure similarity of the polyketide subunit with pyripyropenes, we performed bioinformatics analysis to find homologs of the responsible genes in Penicillium species. Using the genome sequence database provided by JGI, we found that P. canescens harbors homologs of all the genes in the pyripyropene A biosynthetic gene cluster. We then proceeded to perform a series of targeted gene deletions including genes encoding additional putative tailoring enzymes to identify the genes involved in the biosynthetic pathway of 1. We showed that the pathway involves at least 12 genes in a single cluster. We have isolated and characterized 7 additional intermediates, 2 of which have not been reported previously. Further bioinformatics analysis together with the intermediates identified from the gene deletion strains has allowed us to propose a biosynthetic pathway for 1 (Scheme 1A).
Compound 1 has a polyketide subunit as a α-pyrone with an attached pyridine ring, very similar to pyripyropene A. This structural similarity has allowed us to propose the early steps of the biosynthetic pathway of 1 (Scheme 1A, blue). Although we did not detect intermediates from our LC-MS analysis of strains carrying deletions of olcI and olcA, these genes have high sequence similarity to genes involved in the first steps of the biosynthesis of pyripyropene A, pyr1 and pyr2, respectively. This finding suggests that CoA ligase OlcI catalyzes the formation of nicotinyl-CoA. PKS OlcA then uses this nicotinyl-CoA as a starter unit to which it catalyzes the condensation of two malonyl-CoA molecules to form HPPO.
Unlike pyripyropene A in which a farnesyl pyrophosphate (FPP) is attached to HPPO in the subsequent step, 1 is linked with geranylgeranyl pyrophosphate (GGPP). The deletion of olcC, a gene for a GGPP synthase, significantly decreased the production of 1, suggesting that a large portion of the GGPP used here seems to be generated by OlcC, however, the deletion did not completely eliminate the production of 1. RT-PCR analysis confirmed the complete inactivation of olcC in the deletant strain (Fig. S5†). Bioinformatics analysis of the P. canescens genome showed 5 additional genes encoding GGPP synthases, which may provide the GGPP necessary to produce 1 in the olcC deletant strain. This olcC gene is interesting because, based on further bioinformatics analysis, this is the only GGPP synthase in P. canescens that is located in sufficient proximity to a PKS to be a part of a biosynthetic gene cluster. In contrast, the pyripyropene A gene cluster in A. fumigatus is located on a completely separate chromosome from the FPP synthase.
In pyripyropene A biosynthesis, the attachment of FPP to HPPO to form farnesyl-HPPO is catalyzed by the prenyltransferase Pyr6. The homolog of Pyr6 in P. canescens is OlcH (68% protein sequence similarity), and this enzyme is the most likely to catalyze the attachment of GGPP to HPPO to form geranylgeranyl–HPPO.
The next steps consist of the epoxidation and cyclization of the terpenoid subunit. It was demonstrated in the pyripyropene biosynthetic pathway that the FAD-dependent monooxygenase (FMO) Pyr5 catalyzes the epoxidation step, and the integral membrane protein Pyr4 functions as the terpene cyclase. The olc gene cluster also contains homologs of these two enzymes, OlcE (57% similarity) and OlcD (74% similarity). From this information, we propose that the FMO OlcE catalyzes the epoxidation of geranyl-geranyl–HPPO and OlcD catalyzes the cyclization of the terpenoid component, resulting in the formation of the tricyclic terpene moiety seen in predecaturin E (7).
Deletion of olcG resulted in the accumulation of 7, suggesting that OlcG, a putative cytochrome P450, is the next enzyme in the biosynthetic pathway. We propose that OlcG catalyzes the allylic oxidation of compound 7, which is followed by spirocylization with concomitant loss of one molecule of water to form decaturin E. Although our deletion strains did not produce decaturin E, we isolated and identified decaturin D (5) and decaturin G (8) from olcJ deletion strains. This result suggests that in the absence of OlcJ, decaturin E may be shunted to a pathway in which it is oxidized to a ketone, possibly by OlcF (see below), to form 5, which undergoes further allylic oxidation to yield 8 (Scheme 1B).
The next steps involve the rearrangement of the diterpenic subunit leading to the formation of the hemiacetal seen in decaturin C (4) possibly via the 29-hydroxyl-27-one intermediate. Since compounds 5 and 8, which accumulated in the olcJ deletion strain, both lack the 29-hydroxyl group, we propose that decaturin E is the substrate of the putative cytochrome P450 OlcJ which hydroxylates it at the C-29 position to form decaturin F (6). Conserved domain analysis showed OlcF is a putative short chain dehydrogenase, and deletion of olcF resulted in the accumulation of 6. OlcF may catalyze the oxidation of 6 to generate the 29-hydroxyl-27-one intermediate, and subsequent hemiacetal formation will lead to the formation of 4.
Deletion of olcK and olcL both resulted in similar SM profiles, showing the accumulation of 4. This discovery suggests that OlcK and OlcL are downstream enzymes of 4. olcB deletants accumulated decaturin A (3), which has an added hydroxyl group to 4. These data suggest that both OlcK and OlcL are involved in the biotransformation of 4 to 3. Conserved domain analysis of OlcK showed that this enzyme belongs to the 2-oxoglutarate-Fe(II) oxygenase superfamily and has a 54% protein sequence similarity to Fum3, a fumonisin C-5 hydroxylase in Fusarium verticillioides.31 Interestingly, this enzyme superfamily includes peroxisomal enzymes. OlcL, on the other hand, is a putative MFS transporter. Computer analysis by TMHMM (http://www.cbs.dtu.dk/services/TMHMM/) indicated that it is a highly hydrophobic protein with 14 transmembrane helices. Analysis of Pex19 (peroxisome biogenesis factor 19) binding sequences (http://www.peroxisomedb.org/) in the OlcL protein revealed one putative Pex19 binding site between amino acids 132 and 143, within the 2nd transmembrane helix. This finding suggests that OlcL may be inserted in the peroxisomal membrane via the import receptor Pex19. On the basis of these analyses, although speculative, we hypothesize that OlcK may be a peroxisomal enzyme that catalyzes the hydroxylation of 4 once it is shuttled into the peroxisome by the MFS transporter OlcL. However, localization studies will be necessary to test our hypothesis.
In the final step of 15-deoxyoxalicine B biosynthesis, the oxidative rearrangement32 of 3 could occur via either a 32- or 33-hydroxyl intermediate (Scheme 2). This reaction is catalyzed by a predicted cytochrome P450, OlcB, to yield 1. In both olcK and olcL deletants, in addition to the production of 4, we identified the production of 15-deoxyoxalicine A (2). On the basis of the function of OlcB, it is most likely that in the absence of OlcK and/or OlcL, 4 is accumulated and can be catalyzed by OlcB to yield 2 in a shunt pathway (Scheme 1B).
It is of note that the proposed 15-deoxyoxalicine B biosynthetic gene cluster does not contain a putative pathway-specific transcriptional activator gene. This is also the case for the previously reported pyripyropene biosynthetic gene cluster. This may indicate that the regulation of biosynthesis of these compounds are occurring at a more global level.
HPLC-MS was carried out using a ThermoFinnigan LCQ Advantage ion trap mass spectrometer with a RP C18 column (Alltech Prevail C18 3 mm 2.1 × 100 mm) at a flow rate of 125 μl min−1. The solvent gradient for HPLC-DAD-MS was 95% MeCN/H2O (solvent B) in 5% MeCN/H2O (solvent A), both containing 0.05% formic acid, as follows: 0% solvent B from 0 to 5 min, 0–100% solvent B from 5 min to 35 min, 100–0% solvent B from 40 to 45 min, and re-equilibration with 0% solvent B from 45 to 50 min.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5sc01965f |
This journal is © The Royal Society of Chemistry 2015 |