Biosynthesis of Polyketides

BY T. J. SIMPSON

1 Introduction

This chapter follows the format of the previous report with the inclusion of an extra section on compounds of mixed polyketide–terpenoid origin, and covers the literature from late 1976 until the end of 1978. 13C-Labelling studies now greatly outnumber those using 14C, although 13C-methods are still almost exclusively limited to studies with micro-organisms. An interesting and potentially very useful development has been in the use of 2H-labelling, the label being detected either directly by 2H n.m.r., or indirectly through its coupling to 13C in the 13C n.m.r. spectra of metabolites enriched from doubly labelled [2H, 13C] precursors. These methods may be of even more use in the study of terpenoid biosynthesis. 3H-Labelling in conjunction with 3H n.m.r. should also be of use, but it has been applied in only one major study to date. There has been an encouraging number of papers describing the incorporation of larger molecules to obtain details of the later stages of polyketide biosynthesis; and close examination of the fate of 13C-label through secondary incorporation routes has shed light on the overall metabolic pathways operating in many organisms. An excellent short review of the fundamentals of polyketide biosynthesis has appeared.

2 Fatty Acids, Polyacetylenes, and Prostaglandins

In a series of related papers, from Cornforth and co-workers, the precise stereochemical course of individual reactions in the biosynthesis of fatty acids was investigated. Incubation of chiral acetates with either chicken-liver or bakers' yeast fatty-acid synthetase showed that in palmitic acid a higher proportion of tritium was retained from (S)-[2-14C, 3H1]acetyl-CoA than from the (R) isomer, indicating an overall stereospecificity in the formation of fatty acids from chiral acetate. The discrimination between the (R) and (S) isomers was small because the small hydrogen isotope effect, operative in the acetyl-CoA carboxylase reaction, led to nearly equal proportions of (R)- and (S)-tritiated malonyl-CoA species. A partial and non-specific exchange of hydrogen catalysed by the synthetase was also observed, and this further reduced the net retention of tritium. Malonyl thiol esters stereospecifically labelled with tritium at C-2 were prepared, and incubation of these chiral malonates with the purified yeast synthetase showed that (2S)-malonate retained 51% of the original tritium whereas the (2R)-isomer retained only 23%. Comparison of these results with those from chiral acetates indicates that carboxylation of acetyl-CoA occurs with retention of configuration. It is known that reduction of the 3-ketoacyl product (1) from condensation of acetyl-CoA with malonyl-CoA is stereospecific, giving rise to the (3R)-hydroxyacyl intermediate (2) which in turn dehydrates to give exclusively the trans-2-enoyl derivative (3). Dehydration of (2R,3R)-3-hydroxy[2-3H1]butyryl thioester by yeast fatty acid synthetase proceeded with retention of tritium to give the trans-2-enoyl derivative, by means of a syn elimination of the elements of water. Assignment of the syn stereochemistry for the dehydration, together with the findings that tritium is retained preferentially from (2S)-[2-3H1]malonate, means that the condensation reaction in fatty acid biosynthesis must proceed with inversion of configuration at C-2 of malonate. Thus the overall stereochemistry of the process is as shown in Scheme 1.

Lynen has examined the selective inhibition of fatty acid synthetase to obtain information on the condensation step in fatty acid biosynthesis. The synthetase reacts with three moles of iodoacetamide, resulting in inhibition of condensation activity, but greatly increasing the malonyl-CoA decarboxylase activity which is normally low. To account for this, a mechanism is proposed (Scheme 2) in which binding of iodoacetamide to the peripheral thiol groups induces the same change in enzyme conformation as does acetyl-CoA, permitting binding of malonate to the active site as usual. Now, however, instead of concerted loss of CO2 and transfer of malonate to the acyl residue [path (a)], simple decarboxylation results as shown in path (b). Further reaction of the inhibited synthetase with three moles of N-ethylmaleimide results in inhibition of the decarboxylase activity also.

The biosynthesis of cyclopentenyl fatty acids from 2-cyclopentenyl carboxylic (aleprolic) acid (4) was tested in seeds and leaves of the Flacourtiaceae viz. Caloncoba echinata and Hydnocarpus anthelminthica, and in various preparations of other higher plants. Only tissues of the Flacourtiacae, where the cyclopentenyl fatty acids occur naturally, were able to accept aleprolic acid as a starter for fatty acid synthesis. The labelling pattern of both straight chain and cyclic fatty acids synthesized after incubation of Flacourtiaceae seeds with [1-14C]acetate indicated initial de novo synthesis of C16 fatty acids in each case, followed by chain elongation to higher homologues. Cyclopentenylglycine, which is found in the tissues of Flacourtiaceae, where it is believed to be formed from acetate and glutamate, serves as a precursor for the cyclic fatty acids, transamination and decarboxylation converting it to aleprolic acid.

Multibranched fatty acids, e.g. 2,4,6,8-tetramethyldecanoic acid (5), are the major fatty acids produced by goose uropygial gland. The enzyme systems from this gland and from goose liver have both been shown to be equally capable of producing both branched and non-branched fatty acids, so that the production of methyl substituted fatty acids is not an inherent property of the enzyme system but simply reflects the availability of methylmalonyl-CoA in the uropygial gland. Radiclonic acid (6) is a multi-branched fatty acid produced by a Pencillium sp. Incorporation of [13C]acetate and methionine confirmed its biosynthesis by the usual fungal pathway via methylation of an intermediate polyketide chain and not from propionate.

Incorporation of octanoic acids tritiated at C-5, C-6, C-7, and C-8 into lipoic acid (7) by cultures of E. coli indicated complete tritium retention at C-5, C-7, and C-8 thus excluding unsaturated intermediates. Fifty per cent of the tritium at C-6, however, was lost, and a further experiment with (6R)- and (6S)-[6-3H1]-octanoate indicated that sulphur was introduced at C-6 with loss of the 6-pro-R hydrogen and inversion of configuration (Scheme 3).

Polyacetylenes with eight carbon atoms are produced exclusively in fungal cultures. Feedings of [18-14C]crepenynate (8), [18-14C]hydroxyester (10), and [10-3H]-10-hydroxydehydromatricaria ester (11) established the origin of the metabolites (12)—(15) from C-18 to C-11 of crepenynate in cultures of Agrocybe dura, Psilocybe merdaria, and Daedalea juniperina as shown in Scheme 4. Junipal (15) is one of only three fungal thiophen-acetylenes. They are mostly associated with the Compositae and their derivation from polyacetylenes by the addition of the elements of hydrogen sulphide is probable.

Biosynthetic studies on mycomycin (16), one of the few natural allenes, have been thwarted by its instability and the consequent difficulty of purifying it. However it has been found possible to isolate the alkali isomerization product (17) by h.p.l.c., and feedings of [9-14C]- and [18-14C]-labelled crepenynate (8) and its triacetylenic derivative (9) to cultures of Resinicium bicolor indicate that mycomycin is derived from C-5 to C-17 of crepenynate as shown in Scheme 5. Drosophilin C (19) is formed along with its allenic isomer, drosophilin D (20), by Drosophila subatrata and both are converted by alkali into the stable readily purified methyl ester (21). The triacetylene (9) was not incorporated into drosophilin C (19) but the f18-14C]diacetylene (18) was (Scheme 6). Thus it appears that the terminal ethynyl group in (19) does not always arise, as it must do in mycomycin and other acetylenes with odd numbers of carbon atoms, by elimination of C-18 of crepenynate by decarboxylation or an equivalent process.

A large number of C10 poly acetylenes has been isolated from Polyporus anthracophilus. When a mixture of E,E- and 2E,8Z-[1-14C]matricaria esters (22) were fed to cultures of P. anthracophilus, one or both of these esters were specifically incorporated into E,E- and Z,Z-matricarianol (23), and dimethyl E,E- deca-2,8-diene-4,6-diyne-1, 10-dioate (24), to provide evidence that in this organism matricaria esters are intermediates on the route to the many other polyacetylenic metabolites.

A major development in the prostaglandin field has been the discovery of prostacyclin (PGX or PGI2; 27). Prostacyclin, in contrast to the thromboxanes which cause blood clot formation, prevents clot formation and causes preformed clots to disperse. Prostacyclin is formed from PGG2 (25); polarization of the endoperoxide in the opposite sense to that required for thromboxane formation could lead to participation of the 5,6-double bond leading to the carbonium ion (26) which by loss of the proton from C-6 gives prostacyclin (Scheme 7.) Reviews of the formation, synthesis, and biological properties of prostaglandins and other metabolites of arachidonic acid have appeared in the recent chemical literature.

3 Tetraketides

Previous studies have demonstrated the polyketide character of barnol (32) and the origin of the alkyl groups: one of the ring methyls is derived from the C1-pool, and the other from reduction of the terminal carboxy-group; and the ethyl group is formed by methylation of an original acetate methyl group. Further details of the biosynthetic pathway have been elucidated by feeding phenolic substances to Penicillum baarnense. Neither orsellenic acid nor 5-methylorsellenic acid were metabolized. However, administration of 14C-labelled 5-methylorcylaldehyde (28) led to the isolation of two radioactive phenols, identified as 4,5,6-trimethyl-resorcinol (29) and 4,5,6-trimethylpyrogallol (30). Similarly [14C]orcylaldehyde was converted to 4,5-dimethylresorcinol, and a cell-free homogenate from P. baarnense converted (29) to (30). Finally 2,4-dihydroxy-6-ethyl-5-methylbenz- aldehyde (31) was incorporated to the extent of 24% into barnol. These results indicate the biosynthetic sequence shown in Scheme 8 and reveal that both ring and side-chain methyls must be introduced before cyclization of the polyketide, and that reduction of the terminal carboxyl precedes ring hydroxylation. Methylation of a polyketide methyl, rather than a methylene, is very unusual. Mosbach's suggestion that methylation occurred via a quinone-methide intermediate is ruled out by the above results. Another possibility is that methylation occurs on a pentaketide precursor followed by the loss of the 'starter' unit. Stellatin (33), a recently isolated metabolite of Aspergillus variecolor, has an apparently similar bis-C-methylated tetraketide skeleton to barnol, and pentaketide metabolites have been isolated from A. variecolor. Experiments to establish the presence of a 'starter' group in these molecules would be of interest.

3H n.m.r. has been used to determine directly the regio- and stereoselectivity of labelling in penicillic acid (35) biosynthesis in Pencillium cyclopium from [3H]acetate, [3H]malonate, and [3,5-3H]orsellinic acid (34). This reveals specific labelling of the 5-methylene with the 3H mainly trans to the C-methyl; in the case of the acetate feed there is a partial loss of label from C-5 relative to C-3, and from C-3 relative to the C-methyl. In addition the C-methyl is not labelled by malonate, revealing a clear acetate 'starter' effect, nor is it labelled by the orsellinic acid (so that label from the orsellinic acid is not being incorporated through prior degradation to acetate). The previously postulated intermediate 2,5-dihydroxy-3-methoxytoluene (36) has been isolated from penicillic acid producing cultures of Penicillium baarnense.

Incorporations of [13C]acetate and methionine into the fungal α-pyrones rosellisin (37), a metabolite of Hypomyces rosellus, and coarcatin (38), a metabolite of Chaetomium coarcatus reveal that both are tetraketide in origin, with the extra carbons being derived from the C1-pool. The structurally similar nectriapyrone (39) was isolated from Gyrostroma missouriense and on the basis of incorporation of activity from [2-14C]mevalonic acid it was said to be a monoterpenoid. This assumption has since been repeated several times in the literature. No degradations to establish the specificity of labelling were reported, however, and nectriapyrone is almost certainly a tetraketide with the 'extra' methyls deriving from the C1-pool.

4 Pentaketides

The detection of 2H through its coupling to 13C in the 13C n.m.r. spectra of metabolites derived from doubly labelled [2H, 13C] precursors offers the possibility of establishing the integrity of C—H bonds during the course of a biosynthetic pathway. This has been demonstrated by Staunton who proposes a new method for the detection of chain starter units by this means. On incorporation of [2-2H3,2-13C] acetate into terrein (40) by cultures of Aspergillus terreus the presence of 2H on C-1, C-3, and C-8 could be inferred by their lowered intensity compared to C-5 (also enriched by 13C acetate but not carrying 2H). In addition the C-3 signal showed 2H-13C coupling satellites. On redetermining the 13C n.m.r. spectrum with deuterium noise-decoupling a singlet at 17.95 p.p.m. (0.81 p.p.m. upfield of the normal chemical shift value for C-1) was assigned to molecules trisubstituted with 2H at this position (the normal chemical shift difference for isotopic substitution is ca. 0.3 p.p.m. for each deuterium) thus confirming that C-1 is a chain starter unit.29 There was also a doublet ([FORMULA OMITTED] 123 Hz) centred at 18.22 p.p.m. (0.55 p.p.m. upfield from normal) corresponding to molecules labelled with CHD2, and the presence of an enriched CH3 singlet (in the 1H noise decoupled spectrum) shows that there is considerable exchange of hydrogen from the methyl group during biosynthesis. The 2H noise decoupled spectrum also showed singlets at 125.3 and 124.8 for C-3 and C-8 each carrying one 2H only.

Previous studies suggested that sclerin (43) a metabolite of Sclerotinia sclerotiorum was formed by condensation of two preformed polyketide chains. The principal difficulty with this proposal was that it necessitated introduction of a methyl from the C1-pool on to the methyl, rather than the usual methylene, of a polyketide chain (see barnol, above). To overcome this Staunton suggested that sclerin was formed via a novel structural reorganization of the known co- metabolite, scleritonin A (41), Scheme 9. This has been confirmed by incorporation of both (41) and (42), which are probably readily interconverted in vivo. The incorporations were low, 0.01 and 0.45% respectively, presumably due to poor cell permeability, but degradation confirmed their specific incorporation. Interestingly, label from the isocoumarin (44) was incorporated with much greater efficiency, but (44) was clearly shown to be incorporated only through prior degradation to [2-14C] acetate, a salutary reminder of the need for rigorous proof of specific incorporation. Incorporation of [13C]acetate and [13C]malonate into sclerin was reported independently earlier. As required by the above mechanism, incorporation of [13C]malonate in the presence of unlabelled acetate showed a clear acetate starter effect for C-12 only. However the authors then interpreted this as evidence for a two-chain pathway, with loss of the starter acetate unit from one of the polyketides.