Biosynthesis

BY R. B. HERBERT

Terpenoid lndole Alkaloids. — Monoterpenes have been the subject of a comprehensive review which includes these alkaloids. Indole alkaloids have been classified according to their biogenesis and another review discusses the application of tritium labelling in biosynthetic studies on these alkaloids as well as several others.

An interesting new technique has been applied to the study of indole alkaloid biosynthesis: the alkaloids in Vinca rosea seedlings were examined after the administration of DL-[2'-14C]tryptophan (ca. 30% incorporation) and the appearance and disappearance of radioactivity noted as a function of time. The technique is thus similar to the widely used method of 14CO2 feeding. The results were in accord with those obtained earlier by precursor feeding. In particular, geissoschizine (3), preakuammicine (5), and tabersonine (7) were confirmed as important ('dynamic') intermediates in the biosynthesis of other alkaloids, whereas ajmalicine (4), catharanthine (8), coronaridine (9), and vindoline (10) appeared as end-products of biosynthesis. An unknown alkaloid was the first to be labelled in the experiment. It showed rapid turnover and is thus an important precursor for other alkaloids. It appeared that it could be (2), lying between vincoside (1) and geissoschizine (3).

The precursor role of tabersonine (7) for the Aspidosperma and Iboga alkaloids was further strengthened by feeding radioactive tabersonine to V rosea and examining the products as for tryptophan. It was suggested, in view of the large differences of incorporation observed for the two classes of alkaloid, that tabersonine might be formed as the racemate, one enantiomer of which [the ( -)-form] would give vindoline (10) and the other [the unknown (+)-form] catharanthine (8) and coronaridine (9).

Stemmadenine (6) did not show the characteristics, in these experiments, of a dynamic intermediate. This could be explained, however, as the consequence of an equilibration with preakuammicine (5) which is enzyme bound, or if stemmadenine is a stabilized form of preakuammicine (5), and is involved in biosynthesis via (5).

The secondine group of alkaloids are interesting biosynthetically as derivatives of (11) or (12), which are regarded as likely intermediates in the biosynthesis of indole alkaloids with rearranged monoterpenoid units. Accordingly, 16,17-dihydrosecodin-17-ol (13) and secodine (14) have been tested as precursors for uleine (15), apparicine (16), vincamine (17), minovine (18), vindoline (10), and catharanthine (8), as well as ajmalicine (4), which has an unrearranged unit. All incorporations were similarly low except for those for secodine into vindoline (10) in Vinca rosea and apparicine (16) in Aspidosperma pyricollum, which were significantly higher. When [carboxy-14C; ar-3H]secodine was used, both of these alkaloids were formed without change in isotope ratio, indicating intact incorporations of the precursor. Degradation of the vindoline showed that the 14C label was located at C-14 and thus, reasonably, the methoxycarbonyl group of secodine (14) was transferred as such into vindoline (10). [carboxy-14C; 19-3H]Secodine was incorporated without change in isotope ratio. On the other hand, [carboxy-14C;3,14,15, 21-3H4]secodine was converted into vindoline with loss of 60% of the tritium label, consistent with its involvement in vindoline biosynthesis via the more highly oxidized (11) or (12). The same precursor was incorporated into apparicine (16) and although the results are regarded as preliminary, the methoxycarbonyl carbon appears to be retained ; tritium was lost (48%) from the piperidine ring, again consistent with involvement of an intermediate similar to that which yields vindoline.

Two new acidic terpenes have been isolated from V. rosea and rigorously characterized as secologanic acid (19) and secologanoside, which was studied as its methylation-acetylation product (21). [14C]Loganic acid (22) was efficiently incorporated into secologanic acid in V. rosea as well as secologanin (20) and loganin (23). A similar efficient conversion of loganin into secologanic acid was recorded. Taking into account the purification of a methyltransferase from V. rosea capable of methylating loganic and secologanic acids, these results suggest a role for these acids in indole alkaloid biosynthesis, and the incorporation of sweroside (24) into the alkaloids may be via secologanic acid (19). Sweroside also serves as a precursor for reserpinine and quinine.

The terpenoid moieties of indole alkaloids have been proved to arise from mevalonate. Although labelled acetate can be specifically incorporated into many terpenoid compounds via mevalonate1 it is not a specific precursor for the indole alkaloids. This has prompted a search for a source other than acetate. When intermediates of glycolysis and the Krebs cycle, leucine, and glycollic acid were tested as precursors for cephaeline (25) in Cephaelis acuminata, randomization of the labels was observed. 17 [2-14C]Glycine, however, was incorporated with specificity; cephaeline (25) was labelled at C-16 but not C-15. In Rauwolfia serpentina, ajmaline (26) was labelled at C-18 but not C-19 and reserpine predominantly in the reserpic acid moiety. The labelled positions showed 15 — 18% of the total activity compared with 20% calculated for utilization of glycine as a C2 unit via mevalonate (Scheme 1). The results for cephaeline were obtained in the summer with four-five-year-old plants. Winter feeding to two-year-old plants, however, gave non-specific incorporation into the terpenoid portions of cephaeline and ipecoside. Similar non-specific incorporation of glycine has been observed for vindoline, ajmalicine, and the Strychnos alkaloids. On the other hand, specific incorporation of [2-14C]glycine into gentianine (27) has been recorded. L-Leucine has been found to be incorporated into vindoline and catharanthine as inefficiently as acetate. Inhibition of protein synthesis in the plant by puromycin caused a measurable but not dramatic increase in leucine incorporation into the alkaloids.

Ergot. — Experiments directed towards solving the outstanding problem of the biosynthesis of the peptide lysergic acid derivatives from lysergic acid have not yet proved definitive, and the recent results for ergotamine and ergometrine represent little advance on those reviewed previously.

[14C]Lysergylalanine (28), ergometrine (29), and its stereoisomer ergometrinine, were poorly incorporated into the α-hydroxyalanine portion of ergotamine (30) in Claviceps paspali, and with considerable randomization of the labels. Therefore, these compounds were not incorporated intact into ergotamine; a similar result was obtained earlier with ergometrine. L-Alaninol and L-alanine served as precursors for ergotamine. Although it appeared that alaninol was a better precursor for ergotamine than alanine this was found to be due to the lower weight of alanine administered which could then be more easily metabolized. It was concluded that alanine or a closely related compound was the precursor of the tX-hydroxyalanine portion of ergotamine.

In confirmation of an earlier result lysergylalanine (28) was found to act as a precursor for ergometrine (29), albeit an inefficient one, in C. paspali; the D-alanyl derivatives were less efficient precursors than the L-alanyl derivatives. The latter were twice as efficient as L-[l-14C]alanine, suggesting some intact conversion into ergometrine. Taking the low incorporation, however, together with the failure to detect lysergylalanine in C. paspali cultures by radioisotope dilution, it seems unlikely that lysergylalanine is a normal intermediate in ergometrine biosynthesis.

DL-[l-14C]Valine, D-, L-, and DL-[l-14C]alanine, and lysergylvaline (31; 14C label as shown) have been studied as precursors for ergocornine (32) and ergocryptine (33), examined mostly as the 3 : 1 mixture ergotoxine. L-Alanine and lysergyl-L-valine were more efficiently utilized than the corresponding D-isomers, but lysergyl-L-valine and DL-valine were incorporated, and the labels randomized to a similar extent, indicating the non-intact incorporation of lysergylvaline. Degradation of the ergotoxine obtained in all the above experiments showed significantly that the hydroxyvaline portion of the ergocornine (32) had a higher specific activity than the valine portion, suggesting that the latter is built into a molecule which leads to ergocornine earlier than hydroxyvaline, and is subject to dilution by more non-labelled pools, i.e. valine-containing peptide intermediates. A preliminary competition experiment between inactive L-valyl-L-proline and L-[l-14C]valine was unsuccessful, however. It is nonetheless an attractive corollary of this idea that the peptide fragments of ergocornine (32), ergocryptine (33), and indeed ergotamine (30) etc. are added on to lysergic acid as complete units; this would explain the failure of attempts to demonstrate the stepwise addition of amino-acids to lysergic acid.

Betalaines. — The Centrospermeae pigments, the betalaines, are derived in a unique pathway from dopa [Scheme 2; shown for betanine (36)]. DL-[1'-14C]Dopa was incorporated into betanine with almost all of the activity located in the three carboxy-groups. Further, as much as 90% of this activity was found in the derived betalinic acid (35). Experiments have now been carried out to locate the label in the betalinic acid moiety of betanidine. Decarboxylation of betanidine (the aglycone of betanine) in ethanol gave (37). Unexpectedly, decarboxylation of the betanidine from the dopa feed resulted in loss of 86 %of the activity, suggesting C-20 rather than C-19 as the labelling site. Decarboxylation in EtOD, however, gave a product in which the C-15 proton was absent but the one at C-17 remained. This is inconsistent with a mechanism for loss of the C-20 carboxy-group, but may be rationalized with loss of the C-19 carboxyfunction (Scheme 3), where C-15 and C-17 of betanidine have become C-17 and C-15, respectively, in (37).

Biological formation of betalinic acid involves ring cleavage of dopa (34) in one of two ways; path a or path b of Scheme 2. Enzymes from sources other than the Centrospermeae will catalyse either ofthese cleavages in several phenols. The reaction to form betalinic acid was examined by incubating L-[3,5-3H2]tyrosine with the pulp taken from the soft centre of young cactus fruits (Opuntia decumbens). A satisfactory incorporation into betanin (36) was recorded (0.03-0.08 %, or 0.06-0.16 %allowing for the expected loss of half the tritium on hydroxylation to dopa, compared with 0.53% for L-[l'-14C]tyrosine) and the majority of the label was confined to the betalinic acid portion. No NIH shift was expected on formation of dopa. Otherwise C-12 (path b) or C-18 (path a) of betanine would have been labelled, resulting from the presence of tritium at C-2 of dopa. The absence of such a shift was proved when exchange of these positions with trifluoroacetic acid caused no tritium loss from betanidine. The dopa was thus labelled as shown in (34), and as cleavage by path a would give betalinic acid (35) with retention of the tritium label (at C-11) whereas path b would result in its loss (from C-17), the former mode of cleavage is indicated. However, as the tritiated tyrosine was incorporated less well than [l'-14C]tyrosine, the operation of path b in addition to path a cannot be excluded at this stage.

Gramine. — The mechanism by which gramine (38) arises from trytophan is still unknown, although it is established that the amino-group, indole nucleus, and β-carbon atom, but not the α-carbon, of tryptophan are incorporated. An earlier experiment with [2'-14C; 2'-3H]tryptophan has been repeated and the finding confirmed that the tritium at this position is retained. A small loss of tritium was observed, suggesting possible modification of the β-carbon atom and tritium retention by a primary isotope effect, but this was excluded when [2'-14C; 2'-2H2]tryptophan was incorporated without deuterium loss.

Indolmycin. — It has been shown that indolmycin (39; corrected stereochemistry) is derived from its co-metabolite in Streptomyces griseus, indolmycenic acid (45; corrected stereochemistry).

The origin of the carbon skeleton of indolmycin has been established recently. Specific incorporations were found for 14C labelled (R,S)-tryptophan (40), (S)-arginine (41), and (S)-methionine (42), as illustrated in Scheme 4. In addition, [l'-14C]- and [2'-14C]-(R,S)-tryptophan and the tryptophan precursors anthranilic acid (as G-3H) and indole (as 2-14C) were efficiently utilized. [Me-14C, Me-3H]Methionine gave indolmycin without change in isotope ratio, indicating incorporation of all three hydrogen atoms of the methyl group but only if no primary isotope effect was operating. (R,S)-[3'-14C; 2'-3H]Tryptophan was incorporated with complete tritium loss whereas (R,S)-[3'-14C;3'-3H]tryptophan gave indolmycin with 52% retention of tritium. A stereospecific loss of hydrogen from C-3' is thus indicated, which may occur in an enolization step prior to methylation of indolepyruvic acid (43), whose formation from tryptophan would result in loss of the tritium from C-2'. Similar results were obtained for indolmycenic acid. The pathway illustrated (Scheme 5) is consistent with the above results and, further, the steps a and b have been carried out using cell-free extracts of S. griseus and tryptophan cannot substitute for indolepyruvic acid in step b. The in vitro reaction b yielded (R)-3'-methylindolepyruvic acid (44), which could be transformed into indolmycin in vivo; the (S)-isomer was not utilized. Finally, and incidentally, (2'R,3'S + 2'S,3'R)-3'-methyltryptophan, but not the (2'R,3'R + 2'S,3'S)-racemate, could act as a precursor for indolmycin. Presumably only (2'S,3'R)-methyltryptophan was incorporated, by way of transamination into (R)-3'-methylindolepyruvic acid.

Gliotoxin. — Gliotoxin (46) is a metabolite of Trichoderma viride and Penicillium terlikowskii. Recently, phenylalanine but not m-tyrosine (in contrast to earlier work) or o-tyrosine or 2,3-dihydroxyphenylalanine has been shown to be incorporated. The incorporation was sufficiently high to allow the use of deuterium-labelled precursors and it could then be shown that all five aromatic hydrogens of phenylalanine are retained in gliotoxin formation. The phenol (49) has been reported as a precursor for gliotoxin, but in view of the above cannot be an obligatory intermediate.

Similar results have been found for bisdethiodi(methylthio)acetylaranotin (47) in Arachniotus aureus. It was rigorously demonstrated that phenylalanine, but not m-tyrosine, was a precursor. Further, DL-[3'-14C;2,6-3H]phenylalanine was incorporated with retention of 80% of the tritium label and L-[2H8]phenylalanine was incorporated into acetylaranotin (48) in Aspergillus terreus with retention of the aromatic and C-3' deuterium atoms. Retention of the latter excludes from acetylaranotin biosynthesis an intermediate of partial structure (50).

Incorporation of phenylalanine into these metabolites would reasonably proceed by a pathway involving 2,3-epoxidation of the aromatic ring, shown in Scheme 6 for gliotoxin (46); additionally, rearrangement of the appropriate 2,3-epoxide would give the oxepin rings found in the aranotins.

It has been found when phenylalanine is administered to T. viride that, in competition with incorporation into gliotoxin which proceeds with retention of both hydrogens at C-3', stereospecific exchange of the pro-3' -(S)-hydrogen occurs.

Pyrrolnitrin. — Pyrrolnitrin (52), a metabolite of Pseudomonas aureofaciens, is known to derive from tryptophan and the o-isomer is a more effective precursor than L-tryptophan. The amino-compound (51) has been isolated from Ps. aureofaciens and is efficiently incorporated into pyrrolnitrin. A biosynthetic pathway (Scheme 7) has been proposed.