Biosynthesis of Polyketides

BY T. J. SIMPSON

1 Introduction

Polyketides form a large class of natural products possessing structures of great diversity related by their common formation via the acetate–polymalonate biosynthetic pathway. Acyl units other than acetate, such as propionate, benzoate, and cinnamate, can act as chain-initiating species, and propionate and butyrate as chain elongation units, so that a wide variety of fatty acids, polyacetylenes, single and multiringed phenols, macrolides, flavonoids, and other compounds can be included in this classification.

The literature appearing during 1975 and to mid-1976 is covered in this chapter. An outstanding feature of this period has been the continued growth in 13C n.m.r. methods, in particular the use of doubly labelled 13C-acetate. Analysis of the resultant 13C–13C spin-spin couplings provides a powerful method for determining the manner in which polyketide molecules are assembled on the enzyme surface before the first stable compounds are released, and for probing subsequent molecular rearrangements and cleavage pathways. The scope and methodology of the technique have been discussed in recent reviews.

2 Fatty Acids, Polyacetylenes, and Prostaglandins

Lynen has studied the condensation reaction in fatty acid biosynthesis using dideuteriomalonyl-CoA. No primary iotope effect was observed on the reaction velocity of the yeast-enzyme-catalysed fatty acid synthesis, in which the rate-limiting step is the condensation, or on the condensation itself, studied separately using the β-ketoacyl-(acyl-carrier-protein) synthetase of Escherichia coli. When the condensation was carried out in the presence of tritiated water, no tritium was incorporated into the product. These results exclude condensation mechanisms involving acylation of a malonyl carbanion and indicate a concerted mechanism, (Scheme 1).

2,4,6,8-Tetramethyldecanoic acid (1) is the major fatty acid in the uropygial gland of the goose. Crude cell-free extracts from the gland catalyse the carboxylation of propionyl-CoA but not of acetyl-CoA, whereas more highly purified extracts catalyse both carboxylations. This behaviour was explained by the isolation of a highly specific malonyl-CoA decarboxylase from the crude extracts. Thus acetyl-CoA and methyl-malonyl-CoA are respectively the major chain-primer and elongation agents present in the gland, resulting in the production of the multi-branched fatty acid. Propionate incorporated during chain elongation has been shown to be the branching methyl group donor in biosynthesis of 3- and 13-methylpentacosane (2) and (3), the major cuticular hydrocarbons in the cockroaches Periplaneta americana and P. fulginosa, respectively. In plants, n-alkanes are formed by an elongation of fatty acids followed by decarboxylation, and the 2- and 3-methylalkanes originate from the appropriately branched starter acyl-CoA derived from valine and leucine, whereas in algae the active methyl group from methionine serves as the branching methyl group donor. Cell-free preparations from pea leaves, Pisum sativum, catalysed the decarboxylation of n-dotriacontanoic acid (4), requiring the presence of both ascorbic acid and oxygen, and giving both n-C31 and n-C30 alkanes. Thus decarboxylation and α-oxidation appear to be connected processes; in confirmation, 2-hydroxydotria-contanoic acid, the intermediate in α-oxidation of (4), was converted into the sam two alkanes.

When [1-14C]aleprolic acid (5) was supplied to leaves and seeds of plants belonging to the Flacourtaceae, and also to whole cells of Chlorella vulgaris, cyclopentenyl fatty acids (occurring naturally in seeds and leaves of Flacourtaceae) were synthesized, suggesting that these fatty acids are formed by elongation of aleprolic acid rather than by cyclization of an acyclic fatty acid precursor. Both the availability of aleprolic acid and the ability to use it as a primer for fatty acid synthesis appear as specific characteristics of the Flacourtaceae and thus jointly determine the fatty acid pattern. The main fatty acids in ten strains of acidophilic, thermophilic bacteria isolated from Japanese hot springs are ω-cyclohexyl fatty acids, e.g. (6). Increasing the concentration of glucose in the culture medium increased the production of the cyclohexyl but not the acyclic acids, and from incorporation studies with 14C- and 2H-labelled glucose it was confirmed that the acids are produced by elongation of cyclohexylcarboxylate derived from shikimate, rather than by elongation of cyclohexylpropionate derived from decarboxylation of prephenate. The results are in full agreement with previous studies on cyclohexyl fatty acids from Bacillus acidocaldarius, in which cyclohexylcarboxylate competes with straight- and branched-chain precursors of similar molecular length to determine the fatty acid spectrum. These acids may be the pre- cursors of the n-alkylcyclohexanes present in a number of sediments and crude oils, previously postulated as arising by intramolecular cyclization of unsaturated fatty acids.

Several papers have appeared on the routes to unsaturated fatty acids. Stumpf and co-workers have shown that preparations from saffiower seeds and avocado mesocarp rapidly desaturate stearyl-ACP to oleic acid in the presence of oxygen. Mazliak et al. on the other hand have shown that fractions from a cauliflower homogenate synthesize radioactive oleic acid by an aerobic process from [14C]decanoate, in the presence of ATP, NADPH, Coenzyme A, and oxygen. They proposed a scheme analogous to oleic acid synthesis in anaerobic bacteria, which hardly accounts for the oxygen requirement (Scheme 2): 3-hydroxylauric acid (7) formed from decanoate undergoes β,γ-dehydration to 3-dodecenoic acid which is then elongated to oleic acid (8).

The phleic acids (9) are polyunsaturated acids produced by Mycobacterium phlei, with an unusual distribution of double bonds. When [14C]acetate is incubated with M. phlei, the saturated and unsaturated sections are unequally labelled. [14C]Myristic and [14C]palmitic acids serve as precursors for the phleic acids with m = 12 and 14, respectively. A chain-elongation process involving two acetate units at a time, possibly via crotonate, is postulated; β-hydroxybutyric acid, however, is not incorporated without prior degradation. The biosynthesis of cerulenin (10), an important inhibitor of fatty acid synthetase, has been studied in cultures of Cephalosporium caerulens. The alternate labelling obtained with [l-13C]acetate rules out the possible intermediacy of succinate or glycerol and indicates that the biosynthesis is closely related to that of fatty acids.

α-Linolenic acid (11) has been shown to be formed by desaturation of oleic and linoleic acids in several organs of higher plants and in algae. A chloroplast preparation from Thea sinensis leaves converts linolenic acid and 13-L-hydroxylinolenic acid into cis-3-hexenal, the precursor of 'leaf alcohol'. The biosynthesis of the C8 and C10 acetylenes, diatetryne 2 (14) and diatetryne 3 (15), respectively, via oleate, linoleate, crepenynate (12), and trans-dehydromatricariate (13) has been demonstrated in labelling experiments with the fungus Lepista diemi (Scheme 3). Although the necessary ω-oxidation appears to be possible at both the C18 and C10 stages, the ready incorporation of (13) into both diatetrynes and of diatetryne 3 into diatetryne 2 suggests that chain shortening takes place before ω-oxidation. Incorporation experiments with several species of Compositae confirm that crepenynate is also an intermediate in the biosynthetic conversion of oleic acid into polyacetylenes in higher plants. The more efficient incorporation of the C16 compound (16; n = 5) compared to the C18 compounds (16; n = 7) and (17) into cis-dehydromatricaria ester (13) in Artemisia vulgaris does not support the hypothesis that a direct Baeyer–Villiger type oxidation of a C18 precursor is involved in the biosynthesis of (13). On incorporation of [10-14C, 9,10-3 H]oleate and crepenynate an unexpected loss of tritium was observed. This loss, which was also observed in similar experiments with Lepista diemi, cannot yet be explained.

The endoperoxide pathway to prostaglandins in mammalian systems is now well established. However, Corey and co-workers have shown that the biosynthetic pathway to prostaglandin A2 (PGrA2; 18) in the coral, Plexaura homomalla, does not involve PGrE2 (19), PGrH2 (22), PGG2 (23), or 11-epi- PGE2(20), and hence the epi-endoperoxide, and so differs from that of mammalian systems. Samuelsson's group have detected a labile intermediate of high biological activity in the conversion of arachidonic acid (21) or of the endoperoxide PGG2 (23) into the hemiacetal derivative, thromboxane B2 (25), by washed human platelets. This intermediate, designated thromboxane A2 (24), appears to be identical with rabbit aorta-contracting substance (RCS). An enzyme system has been isolated from the microsomal fractions of disrupted horse and human platelets which mediates the conversion of (22) and (23) into (24), as shown in Scheme 4. Incubation of labelled arachidonic acid with a homogenate of rat stomach led to the isolation of 6-ketoprostaglandin F1α, which exists predominantely in the lactol form (26).

3 β-Polyketomethylene Derivatives

The compounds discussed in this section may be formally considered to be formed via polyketomethylene chains of general type (27), in which reduction of the intermediate β-keto-ester after each condensation step, as in fatty acid synthesis, has not taken place. These compounds are classified, as before, into groups according to the number of C2-units in the intermediate chain, though no distinction has been made between aromatic and non-aromatic compounds; this division is becoming increasingly artifical as interrelationships become clearer.

Tetraketides. — Recent results reported by Lynen extend the enzymic characterization of the biosynthetic pathway to patulin (33) in Penicillium patulum. Two separable enzyme fractions hydroxylate m-cresol (28) to m-hydroxybenzyl alcohol (31) and 2,5-dihydroxytoluene (29), respectively. Time studies of the appearance of activity from [1-14C]acetate into metabolites of the patulin biosynthetic pathway and of the utilization of labelled intermediates show that the methyl hydroxylation of m-cresol is an important reaction on the pathway, whereas the ring hydroxylation appears to be a side-reaction. A further enzyme preparation ring-hydroxylates m-hydroxybenzyl alcohol to gentisyl alcohol (30), but m-hydroxybenzaldehyde (32) was not ring-hydroxylated by any preparation from P. patulum. It was concluded that the main pathway to patulin is via m -hydroxybenzyl alcohol, gentisyl alcohol, and gentisaldehyde (Scheme 5).

Incorporation of 14C-labelled precursors demonstrates the biosynthetic sequence, Scheme 6, leading to penicillic acid (34) in Penicillium cyclopium. [1-14C]Methylorcinol is also incorporated but it may not be on the direct pathway. An enzyme system which carries out the final oxidative ring-opening reaction has been isolated; co-factor requirements indicate a mono-oxygenase system. A similar mechanism for cleavage of aromatic rings via a quinonoid structure may be involved in the biosynthesis of a number of natural products, such as patulin, sulochrin, multicolic acid, and the aflatoxins.

Further analogues of mycophenolic acid (35) were produced by cultures of P. brevicompactum, supplemented with 4,6-dihydroxycoumaran-3-one (36), 5,7-dihydroxyindan-1-one (37), and 2,4-dihydroxyacetophenone (38). Both 3- and 5-trans, trans-farnesyl-2,4-dihydroxyacetophenone were metabolized to the corresponding analogues (39) and (40); it appears that the tendency of the side-chain double bond to be oxidized depends upon its distance from the carbonyl group in the enzyme- substrate complex. Further details have appeared on biosynthetic studies of epoxy- don (41), a major antibiotic metabolite of Phyllostica spp. Incorporations of [13C]-acetate and [14C]-gentisyl alcohol confirm the tetraketide origin of epoxydon via 6-methylsalicylic acid and gentisyl alcohol, known co-metabolites. The remaining co-metabolites (42), (42a) and (42b), may be artefacts as they can be isolated after allowing epoxydon to stand in sterile culture medium (chloride concentration ca. 130 p.p.m.) for several days. The 13C n.m.r. spectra of colletodiol (43) enriched with [13C]acetate in Colletotrichum capsici show enrichments consistent with a biosynthetic pathway via the union of triketide and tetraketide components.

Pentaketides. — Incorporation of [13C]acetate and [13C]methionine by cultures of Sclerotinia sclerotiorum establish that sclerin (44) is derived from five intact acetate units with the introduction of three C-methyl groups from methionine. The results are in accord with a biosynthesis from two separate polyketide chains, though this would involve an aldol condensation and methylation at the methyl group of one of the polyketide chains, both reactions for which there is almost no biosynthetic precedent. To overcome this difficulty, Staunton has proposed an interesting pathway via aromatic ring-cleavage of the known co-metabolite, sclerotinin A (45) (Scheme 7). Feeding of labelled (45) or [13C]malonate might solve this intriguing problem.

Incorporations of the 14C-labelled dihydroisocoumarin (46), and of [1-13C)-and [1,2-13C2]-acetate, into the cyclopentenone terrein (47) by cultures of Aspergillus terreus, indicate a biosynthesis involving contraction of the aromatic ring of (46) with loss of carbon 7, carbons 8 and 9 becoming carbons 6 and 5, respectively, of terrein.

A somewhat different mechanism leads to the chlorine-containing cyclopentenol (49) and the dihydroisocoumarin (48) in Periconia macrospinosa. Incorporations of singly- and doubly-labelled [13C]acetate show that (48) is derived as expected from five intact acetate units assembled as shown in Scheme 8. The cyclopentenol (49) is also derived from five acetates, but only three are intact, consistent with a biosynthesis by ring-contraction, involving fission of the 7–8 bond of an aromatic precursor related to (48) as shown. A previous postulate involved fission of the 4–9 bond. Incorporation of [1-13C]-, [2-13C]- and [1,2-13C)- acetate into the pyrone (50) by cultures of A. melleus also gave an unusual labelling pattern. Detailed examination of the 13C n.m.r. spectrum of [1,2-13C )acetate-enriched (50) revealed the presence of a two-bond 13C-13C coupling (6.2 Hz) between carbons 1 and 7, proving their origin from an originally intact acetate pair. A pathway involving a Favorskii-type rearrangement and decarboxylation of a precursor pentaketide, as indicated, was proposed.

14C-Labelling experiments indicate a polyketide origin for flaviolin (51) in A. niger and for 2,7-dimethoxynaphthazarin (53) in a Streptomyces. Labelled flaviolin and mompain (52) are efficiently incorporated into (53) by the Streptomycete. As 2,7-dimethylflaviolin could not be detected, a pathway in which hydroxylation of flaviolin occurs exclusively before introduction of the O-methyl groups was proposed.

Melanin pigments formed by polymerization of 1,8-dihydroxynaphthalene (1,8-DHN), play an important role in the survival of fungi, being present in structures resistant to extreme conditions and protecting cell walls from the action of bacterial degradative enzymes. The biosynthetic pathway to 1,8-DHN in Verticillium dahliae has been explored (Scheme 9) by isolation of a series of melanin-deficient mutants. The mutant brm-1 (brown microsclerotia) accumulated (+)-scytalone (54) which could be converted into melanin by alm (albino microsclerotia) mutants. When fed to a second brown mutant, brm-2, (+)-scytalone (54) was dehydrated to 1,3,8-trihydroxynaphthalene (55). On feeding (55) to brm-1 cultures, a new metabolite (-)-vermelone (56) accumulated. (-)-Vermelone was dehydrated to 1,8-DHN by another mutant, alm-1. Both 1,8-DHN and (56) are rapidly converted into melanin by alm or brm-2 cultures. It appears that the brm-1 mutant lacks the enzyme activity necessary to convert either of the 3-hydroxytetralones (54) and (56) into their corresponding naphthols, which suggests that the same enzyme might catalyse both dehydratase steps in the conversion of (+)-scytalone (54) into 1,8-DHN. The acetate origin of scytalone has recently been demonstrated in Phialaphora lagerbergii.