The Biosynthesis of Polyketides

BY T.J. SIMPSON

1 Introduction

This chapter covers the literature appearing between January 1979 and December 1981 and follows the format of the previous report. It has been a particularly active period with a welcome, increasing trend towards studies aimed at elucidating the mechanisms of the intermediate steps in polyketide biosynthesis. The potential of 2H-labelling mentioned previously has been Yrealised and extended in the review period with several studies using both direct methods, ie 2H n.m.r., and indirect methods, ie 2H α-isotope shifts and 2H-13C couplings in 13C n.m.r. using doubly labelled [2H,13C] precursors. These methods have been reviewed. A potentially more useful technique than the α-isotope shift method has been described. In this, 2H is placed β to the reporter 13C nucleus in a doubly labelled precursor; an isotope shift is still observable but the unfavourable relaxation and nOe effects associated with 2H directly attached to 13C are avoided. This has been applied to only one study in the review period, but will clearly find much use. A related and also extremely useful technique makes use of 18O-induced isotope shifts in 13C n.m.r. to detect the biosynthetic origins of oxygen by incorporating doubly labelled [18O,13C precursors or by growing organisms in an atmosphere containing 18O2 and subsequent 13C n.m.r. analysis of the labelled metabolites. The number of studies using advanced intermediates continues to increase and 2H has great potential in this area. A number of books which cover aspects of polyketide biosynthesis have appeared, with Steyn's book on the biosynthesis of mycotoxins being particularly valuable.

2 Fatty Acids

The stereochemical mechanism of enoyl reductase, the enzyme catalysing the final reduction in the cycle of condensation-reduction-dehydration-reduction that lengthens the fatty acid chain by one -CH2CH2- unit at each turn of the cycle on fatty acid synthetase, appears to be species specific. As shown in Scheme 1, the enoyl reductase from yeast converts the enoyl thioester (1) to the acyl thioester (2) by an anti-addition of hydride from NADPH to the si face of the β-carbon with protonation of the α-carbon from thsi face. However with the reductase from both E. coli and Brevicterium ammoniagenes, a syn-addition of hydrogen via a 2-re, 3-si attack occurs, whereas the reductase from rat liver also carries out a-addition, but this time via 2-si, 3-re attack. The stereochemistry of hydride donation from NADPH is related to the stereochemistry of addition, with the pro-4S hydrogen being used for 3-si addition and the pro-4R hydrogen being used for 3-re addition.

Incorporation studies with [2-13C, 2H3] acetate and analysis of the 2H α-isotope shifts in the simultaneously 1H and 2H noise decoupled 13C n.m.r. spectrum has shown that palmitic acid (3) is biosynthesised in the alga Anacystis nidulans with a gradation of 2H retention along the acyl chain as shown in Scheme 2. The results are interpreted as being consistent with a 'post-malonate' exchange process, presumably associated with reversible transfer of the growing acyl chain from the acyl carrier protein to a cysteine residue of β-keto acyl ACP synthetase. Similar results have been obtained using [2H3] acetate in E. coli.

The incorporation of 2H from [2H3] acetate into lipoic acid (4) is consistent with its formation from octanoic acid with the loss of one 2H label from C-8. The 2H incorporated at C-6 of octanoic acid is retained, and since this 2H is incorporated with the L-configuration during fatty acid biosynthesis but is known to have the D-configuration in lipoic acid, an inversion of configuration must occur at C-6 during sulphur insertion. This suggested the involvement of hydroxylated octanoic acids as intermediates. However, feeding studies with 2H-labelled 6-hydroxy-, 8-hydroxy-, and 6,8-dihydroxyoctanoic acids gave negligible incorporations and so direct introduction of sulphur at the saturated carbons of octanoic acid seems likely.

An authoritative review of the structure of fatty acid synthetase has appeared.

3 Tetraketides

Both 2H n.m.r. spectroscopy and β-isotope shifts in 13C n.m.r. have been used to measure the incorporation of 2H from [2H3]- and [1-13C, 2H3]acetates into 6-methyl-salicylic acid (5) by Penicillium griseofulvum. Both methods show that there is a preferential in corporation into the methyl of the acetyl-CoA derived starter unit and significantly more 2H is retained at C-3 than at C-5. It is suggested that the non-uniform incorporation could arise from differing degrees of random exchange during the chain assembly process or, more interestingly, it could reflect the actual mechanisms of cyclisation and aromatisation of the precursor polyketide. More examples will be needed to test the validity of this observation.

Addition of 5-chloroorsellinic acid to growing cultures of Penicillium cyclopium inhibits the biosynthesis of penicillic acid (8) and results in the accumulation of the previously indicated intermediates orsellinic acid (6) and 3-methoxytoluquinol (7) and its corresponding quinone.

The post-gentisaldehyde part of the biosynthetic pathway to patulin (13) has been extensively investigated, using mutant strains of Penicillium urticae. A patulin-minus mutant, J1, accumulates phyllostine (10) and isoepoxydon (11). Another patulin-minus mutant, J2, which is blocked immediately after gentisaldehyde (9), converts both (10) and (11) to patulin in yields of 90% and 60% respectively. They are interconverted by a specific dehydrogenase and cell-free systems have been isolated from the mutants which carry out their interconversion and further transformation. A further mutant, S15, accumulates isopatulin (12) and immobilised cells of the wild-type strain convert phyllostine to isopatulin in good yield. Finally, a cell suspension of mutant P3, blocked between phyllostine and isopatulin, converts isopatulin to patulin. These results are summarised in Scheme 3. The changes in oxidation levels occurring during these transformations are somewhat puzzling and labelling studies would be useful. This and earlier works are summarised in a review which also compares patulin biosynthesis with the pathways leading to a number of other fungal lactones.

Elasnin (14), a novel inhibitor of human granulocyte elastase, has been isolated from Streptornyces noboritoensis. Incorporation of [13C2] acetate has shown it to be derived from twelve acetates but it would appear to be best regarded as a tetraketide, as the most plausible route is extension of a hexanoate starter by three 2-butylmalonate units as indicated in Scheme 4.

[13C2] Acetate is incorporated into zinniol (15) by cultures of Alternaria solani as shown.

4 Pentaketides

Incorporation studies with singly and doubly labelled [13C] acetates have confirmed that the dihydroisocournarin moiety of the important mycotoxin ochratoxin A (16) has a regular poly-ketide origin and so previous proposals of a phenylpropanoid precursor are no longer tenable. Similar results have been reported for mellein (17) in Aspergillus melleus.

Austdiol (19) is a toxin produced by Aspergillus ustus. Incorporation of [13C2]acetate resulted in two 13C-13C couplings being observed for carbons 5,6,7,8, and 9 while carbons 11,3,4,10,12, and 1 showed only one coupling. On feeding [13C] methionine, C-13 was enriched approximately twice as much as C-1 and C-12. This labelling pattern is consistent with a biosynthetic pathway, shown in Scheme 5, where a methionine-derived methyl is oxidised and the polyketide carboxyl is reduced to give the symmetrical dialdehyde (18) as an intermediate.

Full details of 13C studies on aspyrone (20) have appeared. Asperlactone (21) is a co-metabolite of aspyrone in A. melleus and has the same carbon skeleton. Computer-aided resolution enhancement of the 13C n.m.r. spectrum of [13C2] acetate-enriched asperlactone shows a 2-bond 13C-13C coupling between C-2 and C-8. Pathways involving rearrangement of a linear polyketide (22), or rearrangement and cleavage of an aromatic precursor (23), were proposed. However, on incorporation of [2-13C,2H3] acetate and determination of the simultaneously 1H and 2H noise-decoupled 13C n.m.r. spectrum, the resonance for the C-7 methyl showed two isotopically shifted signals, indicating that two acetate-derived hydrogens are retained on C-7, and so intennediates in which this carbon fonns part of an aromatic ring are excluded. On the basis of stereochemical differences between aspyrone and asper-lactone, the epoxide (24) is proposed as a common intermediate; alternative modes of attack by the carboxylate on the epoxide would lead to (20) and (21) as shown in Scheme 6. Oxygen-18 labelling studies should yield further information on this point.

The biosynthesis of diplosporin (25), a toxic metabolite of Diplodia macrospora, has been studied, using 13C-labelled acetates and methionine. The results indicate its derivation from a pentaketide chain, folded as shown in Scheme 7, with C-5 and C-2 derived from the c1-pool. The presence of a methionine-derived carbon atom in a carbocyclic ring is highly unusual; cf biogenesis of tropolones via rearrangement of 3-methylorsellinic acid. The mechanism may proceed via methylation of the pentaketide at either C-4 or C-8 followed by oxidative activation of the newly formed C-methyl to facilitate ring closure. The introduction of a methionine-derived carbon into a heterocyclic ring is also uncommon and the possibility that it is introduced via Q-methylation, cf rotenone, cannot be excluded.

Full details of 13C, 2H, and advanced Precursor studies on sclerin biosynthesis in Sclerotinia sclerotiorum have appeared. These are consistent with formation of sclerin (27) via ring cleavage and reorganisation of the carbon skeleton of sclerotinin A (26). A full paper has appeared on 13C- and 2H-labelling studies on scytalone biosynthesis in ialaphora lagerbergii. To account for the lack of 2H incorporation from acetate on C-2 and C-7 it is suggested that scytalone (28) may be formed via deacylation of a hexaketide-derived naphthol, (29). Such compounds are known - see O-methylasparvenone (46) below. Attempts to incorporate [13C]-malonate to check for a 'starter' effect were unsuccessful in this study. However, [2-13C]malonate has been incorporated with high efficiency into scytalone and no 'starter' effect was observed.

The antifungal metabolite citrinin (34), produced by Penicillium citrinum,has been the subject of intensive study by several research groups and notable use has been made of advanced precursors. The isocoumarin (35), labelled with 14C at C-9, was specifically incorporated into citrinin whereas label from (36) was only incorporated after prior degradation to acetate, indicating that methylation of the polyketide precursor occurs before aromatisation. However, in a study using a novel technique where P. citrinurn was cultured in D2O, incorporation of [1,2-13C2,1H3]-acetate and subsequent 13C n.m.r. analysis allowed the origin of the hydrogens to be elucidated. This indicated that the hydrogen on C-4 of citrinin was acetate-derived and so, although (35) is specifically incorporated,it cannot be an obligatory intermediate on the pathway. (This was confirmed by a 2H-labelling study using [2H3] acetate and 2H n.m.r. 35). This study also revealed a marked difference in the protium content at C-1 and C-3, suggesting that the necessary reductions at these two sites are carried out at markedly different stages in the bio-synthesis. Taken with the non-intermediacy of (35), this indicated that either the lactone (37) (reduction at C-3 but not C-1) or the aldehyde (30) (reduction at C-1 but not C-3) must be the first enzyme-free intermediate on the pathway. Both these compounds were synthesised with a single 2H label on the C-11 methyl and fed to cultures of P. citrinurn. 2H n.m.r. analysis showed that only (30) was incorporated into citrinin. The incorporation efficiency was 6.5%, with a dilution value of ca 62.5. In a further interesting experiment P. citrinum was grown in the presence of ethionine, which is known to inhibit methylation, resulting in suppression of citrinin production. On incorporation of (30) labelled with 2H at C-1 in the presence of ethionine in a replacement medium, a small amount of citrinin was isolated. Now, it was so highly enriched that the specific incorporation of (30) could be demonstrated by 1H n.m.r. The incorporation rate was now 9.5% but the dilution was only 1.25. Parallel work by Scolastico and co-workers, using specifically 14C-labelled precursors,has shown that both (30) and (32) are specifically incorporated;9 ,4o and so the pathway shown in Scheme 8 is indicated for citrinin biosynthesis. Sankawa and co-workers have reported incorporations of [2-13C,2H3]-, [1-13C,18O2]-, and [1-13C,17O]acetates into citrinin. The result are consistent with Scheme 8,and in the 13C n.m.r. spectrum of [1-13C,18O2] acetate-enriched citrinin, isotopically shifted signals are observed for the resonances due to C-3, C-6,and C-8, indicating origin of the attached oxygens from acetate, so that the quinone-methide structure must be formed by elimination of the hemi-acetal hydroxyl from (32).

5 Hexaketides

Incorporations of 13C acetates and methionine and of 14C-labelled advanced precursors into ascochitine (41), a metabolite of the phytotoxic fungus Ascochyta fabae, have been reported. These show its derivation from a single hexaketide chain with introduction of three C1 units from methionine to give a quinone-methide structure related to citrinin. The aldehyde (39) and quinone-methide (40) are specifically incorporated. The specific incorporation of the methyl ester (42) shows that the organism can convert it directly to the enzyme-bound thioester (38). The enol lactone (43) is also specifically incorporated, but 'enzyme trap' experiments show that it is not on the direct pathway, indicating that aldehyde (39) is formed by direct reduction of the thioester. Thus, the pathway shown in Scheme 9 can be proposed.

Incorporations of singly and doubly 13C-labelled acetates into O-methylasparvenone (46), a dihydronaphthalene metabolite of Aspergillus parvulus, indicated a hexaketide origin with the novel acetate-assembly pattern shown in Scheme 10. Incorporation of [2H3] acetate and analysis of the resultant 2H n.m.r. spectrum showed labelling of the 10-methyl, 5-, 2-axial , 3-axial hydrogens and significantly no labelling at C-4. The loss of label from C-4 and its appearance on C-3 can only be explained by an N.I.H. shift, which implies that hydroxylation of a 1,6,8-trihydroxy-naphthalene (44) to the corresponding 1,4,6,8-tetrahydroxy-naphthalene (45) is a necessary step in the biosynthesis of (46). The 10-rnethyl is labelled to less than twice the level of H-5. This is significant as, in a number of 2H-labelling studies, preferential labelling of the acetyl-CoA-derived 'starter' position relative to positions derived from malonyl-CoA is observed. This therefore suggests that the necessary loss of ketide oxygen from C-9 occurs after aromatisatton, allowing loss of label from C-10 relative to C-5 by exchange from an acetyl side-chain and via reduction to and dehydration of the resultant 1'-hydroxyethyl group during conversion to the ethyl side-chain. Support from this comes from a related study on the biosynthesis of the naphthoguinone (47), a metabolite of Hepdersonula toruloidea.