Biosynthesis

BY R. B. HERBERT

'Being very anxious to find by experiment some support for this still purely speculative view....'

'It was completely unforeseen and opens to physiology new horizons, distant, but sure'

L. Pasteur

1 Introduction

As this is the fifth of these Reports the Reporter is prompted to look back over the past five years in an attempt to recall the important developments of the period. There has of course been a prodigious amount of experimental work and it can be said fairly accurately that the gross topography of the biosynthesis of almost all the plant bases is now known. In a general sense, and rising out of consideration of the detail of biosynthetic pathways, it is the tracing of the stereochemistry of the biological processes which has proved the most fascinating and stimulating both from an intellectual and from an experimental point of view. As a worthwhile consequence more effort is being expended in attempts to understand the enzymic processes involved in biosynthetic processes. In a different direction a recent development has been the harnessing of plants for the synthesis of unnatural alkaloids. At its most sophisticated this can also provide information on enzyme function.

The major siege that was laid against the redoubtable problem of the biosynthesis of a large group of indole alkaloids, represented by ajmalicine (1), akuammicine (2), and catharanthine (3), had been raised by the beginning of the quinquennium with the discovery that the non-indolic C9-C10 unit of each (indicated by heavy bonding) had a common terpenoid origin and that loganin (4) is a key intermediate. More information followed, and related pathways to the Cinchona and Ipecac alkaloids were delineated. Yet there remains fascinating detail to be uncovered.

The solution to the enigma of the biosynthesis of colchicine (5) is of longer standing. More details on the unexpected but simple pathway to this non-basic alkaloid have come with the publication of full papers.

In the light of established pathways to aporphine alkaloids (see p. 15) study of the biosynthesis of glaucine (6) and related bases in Dicentra eximia might have been expected to yield orthodox results. On the contrary, however, a novel pathway was unearthed which implicates dienone intermediates of quite unexpected structure: (7)/(8) for glaucine (6). There is good evidence that the dienone (7) is also involved in the elaboration of the Erythrina alkaloids, along a pathway which has several remarkable features.

Hidden in the old literature was the solution to the long-standing problem of the biosynthesis of the C2 unit (C-1 and C-9) of anhalonidine (12) and the C1 unit (C-1) of anhalamine (13). Only recently was the original suggestion for the biosynthesis of these cactus alkaloids examined, with positive results. Thus the two acids (10) and (11) were found to be precursors for (12) and (13) respectively. They apparently derive in turn from the phenethylamine (9) and pyruvic acid or glyoxylic acid. Attention should be drawn at this point to the very detailed mapping of phenethylamine biosynthesis in cactus species.

Much was already kndwn of the gross features of Amaryllidaceae alkaloid biosynthesis five years ago and in the ensuing period it has been the revelation of the intricate stereochemistry of the processes involved which has proved most interesting. Newer aspects of this work are discussed in this Report (p. 19).

Structural relationships are not always what they seem: it has recently been demonstrated that the mesembrine alkaloids, superficially related to those of the Amaryllidaceae, arise by a quite different pathway, albeit from the same amino-acids (see p. 22).

Research on the biosynthesis of piperidine alkaloids has been consistently stimulating and interesting. There has been a most successful marriage of hypothesis and experiment which can be traced back over the past five years. As a result of the detailed and sophisticated studies elegant theory now stands on a firm experimental base. In particular, it is the fates of individual tritium and carbon atoms in the conversion of lysine into these alkaloids which has allowed the development of the pathway as it now stands (see p. 5). Of significance here is the demonstration that L-lysine is the preferred progenitor of piperidine alkaloids whereas the D-isomer is converted into pipecolic acid in the same plant (see p. 7).

In this area the biosynthesis of the Lycopodium alkaloids, e.g. lycopodine (14) and cernuine (15), is of further interest. Lycopodine and cernuine were considered to be modified dimers of pelletierine (16) because two intact molecules of precursors like lysine and Δ1-piperideine were used for the construction of a single molecule of the alkaloids. Paradoxically, however, pelletierine (16) only gives rise to one of two C8N units (heavy bonding) of (14) and (15). It is now clear that pelletierine is an obligatory intermediate in lycopodine biosynthesis and a satisfying explanation has been advanced to account for the paradox.

Although structurally very similar to pelletierine (16) the hemlock alkaloid coniine (17) is notably of quite different origins, arising as it does from acetate in a fairly well understood pathway.

The biosynthesis of alkaloids containing a pyrrolidine ring such as hyoscyamine (18) and hygrine (19) is similar to the biosynthesis of those with a piperidine nucleus only in so far as the alkaloids arise from homologous amino-acids. The notable difference is that early N-methylation (of ornithine) is apparently an important reaction in the elaboration of pyrrolidine alkaloids, whereas N -methylation occurs late in the formation of the piperidine bases.

There has been a steady progress in the fitting together of the pieces that make up the pattern of furoquinoline and ergot alkaloid biosynthesis so that the pathways for each of these groups is now fairly clear. Both have been the subject of recent publication (see p. 35 and 27).

In the above, discussion has been curtailed on important topics which appear later in this Report, and the reader is referred to the appropriate sections for a fuller treatment.

2 Piperidine, Pyridine, and Pyrrolidine Alkaloids

Piperidine Alkaloids. — There is now a wealth of detail on the incorporation of lysine into the piperidine nucleus of alkaloids such as sedamine (20), anabasine (21), and A-methylpelletierine (22). Five of the carbon atoms of L-lysine (26) (C-2 to C-6) are incorporated into the piperidine ring of these bases in a manner which does not allow carbons 2 and 6 of lysine to become equivalent, i.e. no symmetrical intermediates are permissible. However, in the conversion of lysine into the piperidine rings of the related bases cernuine (15), lycopodine (14), decodine, and the lupine alkaloids C-2 and C-6 do become equivalent and at least one symmetrical intermediate must therefore be involved. Such an intermediate could be cadaverine (27). Cadaverine is a specific precursor for anabasine (21) and is also incorporated into pseudopelletierine (24) and N -methylpelletierine (22). This is paradoxical, for as stated above N-methylpelletierine (22) and anabasine (21) are derived from lysine in unsymmetrical fashion.

One way in which the unsymmetrical incorporation of lysine could be explained was by invoking mono-N-methyl derivatives as adduced analogously for the biosynthesis of the related pyrrolidine alkaloids. Intermediates which follow lysine on this pathway are ε-N-methyl-lysine and N-methylcadaverine. The former compound was shown to be present in Sedum acre, a plant which also produces sedamine (20). Further, its formation from [6-3H]Iysine and [Me-14C]methionine could also be demonstrated. But the 14C: 3H ratio of the derived ε-N-methyl-lysine was quite different from that of the sedamine (20) isolated in the same experiment. Furthermore, attempts to demonstrate the formation of N-methylcadaverine in two Sedum species and Nicotiana glauca [a plant which produces anabasine (21)] were unsuccessful. More direct evidence comes from feeding DL-[2-14C,N-Me-3H]-N-methyl-lysine to S. acre. The precursor was poorly converted into sedamine with alteration of the isotope ratio, which indicated non-intact incorporation. These results, as well as data on the biosynthesis of anabasine, require rejection of the 'N-methylation' hypothesis.

In further consideration of the biosynthesis of the piperidine alkaloids the question of the significance of the incorporation of cadaverine must be answered. Accordingly further research has been directed to this point and it has been shown that cadaverine is a normal component of S. acre, that it is a specific precursor of sedamine (20), and that it is formed from lysine at the same time as sedamine. It follows then that any scheme for the biosynthesis of the piperidine alkaloids which does not accommodate cadaverine as a normal component is unrealistic.

An eminently reasonable hypothesis which fits all the evidence is shown in Scheme 1; it was anticipated in last year's Report. For those alkaloids derived from lysine without the intervention of a symmetrical intermediate, cadaverine formed by decarboxylation of lysine must remain enzyme-bound and therefore unsymmetrical. Exogenous cadaverine enters the pathway at this point by absorption on to the enzyme to give (29). In order to explain the incorporation of lysine into some alkaloids by way of a symmetrization step it is necessary only to postulate equilibration of bound with unbound cadaverine. The proposal that pyridoxal phosphate is involved in this pathway is more than mechanistically attractive, for L-lysinedecarboxylase (EC 4.1.1.18, L-lysine carboxy-lyase) and diamine oxidase [EC 1.4.3.6, diamine:oxygen oxidoreductase (deaminating)], the two enzymes whose participation in the conversion of lysine into Δ1-piperideine (30) is likely, both require pyridoxal phosphate as a co-factor.

The validity of the above scheme was further explored with cadaverine samples chirally labelled with tritium at C-1. (The samples were obtained by decarboxylation of L-lysine mediated by L-lysinedecarboxylase from Bacillus cadaveris. The absolute configuration of the two materials is unknown and they were accordingly named [1A-3H]- and [1B-3H]-cadaverine.) The labelling pattern of the derived N-methylpelletierine (22) was in accord with stereospecific oxidative deamination to Δ1-piperideine (30) and in agreement with the proposed model. Both cadaverine samples afforded A-methylpelletierine (22) with a label at C-6 but only [1A-3H]cadaverine labelled C-2. (The puzzling loss of 25% of the tritium from non-chirally labelled [1-14C,1-3H]cadaverine on incorporation into N-methylpelletierine is now explained in terms of this model, the tritium loss being exactly as predicted. It seems that subsequent elaboration of N-methylpelletierine (22) to pseudopelletierine (24) is accompanied by preferential tritium retention at C-6 by a primary isotope effect.)

In a notable piece of research it was shown that the L-isomer of lysine was much preferred for anabasine biosynthesis whereas the D-isomer was preferentially utilized for L-pipecolic acid biosynthesis in N. glauca. In a more rigorous study this was confirmed for sedamine (20), N-methylpelletierine (22), N-methyl -allosedridine (25) (in two Sedum species), and anabasine (in N. glauca) and also for pipecolic acid in each of these plants. Thus, in terms of the model, both decarboxylation and oxidative deamination are stereospecific.

This study with the chirally labelled cadaverines brings to light an apparent anomaly. Decarboxylation of L-[2-3H]lysine by the enzyme from B. cadaveris affords [1B-3H]cadaverine. When this material is converted into N-methylpelletierine (22) and N-methylallosedridine in S. sarmentosum the tritium destined for C-2 is lost. On the other hand conversion of lysine into the sedamine in S. acre results in the retention of tritium originally present at C-2. The simplest explanation of this is that the protonation of (28) in the micro-organism and the plants proceeds with opposite stereochemistry.

A final point arises from the fact that the stereochemistry at C-2 is not the same in different piperidine bases. For example (–)-sedamine (20) is 2S whereas (–)-pelletierine [as (23)] is 2R. This indicates that addition of the side chain to Δ1-piperideine (30) can occur on both the re and si faces of the molecule, for later epimerization at C-2 seems to be excluded by tritium retention at this site.

Lobinaline. — Discussion of the biosynthesis of lobinaline (31) was omitted from earlier Reports and so is included here. Both lobinaline and 8-phenyl-lobelol-I (32) are constituents of Lobelia inflata. The structure of lobinaline may be broken visually along the dashed line in (31) into two units of the type seen in (32) and sedamine (20). Sedamine, it is known, is formed by the union of a C6-C2 unit derived from phenylalanine and a cyclic C5N unit originating from lysine without intervention of symmetrical intermediates.

In accord with the dimer hypothesis phenylalanine and lysine were found to be specific precursors for both 'halves' of lobinaline and label appeared at the expected sites with apparently equal distribution of activity between the 'halves' of lobinaline. Further substantiation of the hypothesis came with the specific and efficient incorporation of [4-14C]-2-phenacylpiperidine [as (33)] with equal labelling of C-2 and C-6'.

Slaframine. — Further details are now available on the biosynthesis of slaframine (39), a toxin produced by the mould Rhizoctonia leguminicola. Both DL-[1-14C]- and DL-[6-14C]-lysine afforded labelled slaframine, indicating intact incorporation of all the carbons of the amino-acid. Addition of inactive pipecolic acid (34) to the cultures diluted the lysine label in the derived slaframine, and pipecolic acid was labelled by radioactive lysine. Further carboxyl- and ring-labelled pipecolic acids [as (34)] were both well incorporated into slaframine. A clear indication is thus obtained of the biosynthetic sequence: lysine [right arrow] pipe -colic acid (34) [right arrow] slaframine (39); 2-hydroxymethylpiperidine is not a precursor. Attention is drawn to the discovery of a similar pathway to a metabolite of similar structure also produced by R. leguminicola.

The incorporation of (36), (35), and (37) with increasing efficiency allowed further definition of the pathway to slaframine. The results indicate that a tritium label at C-1 of (36) is retained on formation of (39), and so (36) must follow (35) on the pathway. Further support for this relationship comes from the discovery that a cell-free extract of R. leguminicola would catalyse the NADPH -dependent reduction of (35) to (36). This extract also catalysed the acetyl-CoA-dependent formation of slaframine (39) from (38). The pathway to slaframine so far deduced is illustrated in Scheme 2.

L-Pipecolic acid [as (34)] is derived from D-lysine in several higher plants, rats, and the bacterium Pseudomonas putida. In contrast, recent evidence indicates that L-pipecolic acid arises from L-lysine in the mould R. leguminicola; D-lysine, it appears, is converted into ε-N-acetyl-lysine. Otherwise the evidence is that formation of pipecolic acid is via Δ-piperideine-2-carboxylic acid (40) as expected.

Quinoiizidine Alkaloids. — Analysis of Goehelia pachycarpa shoots after administration of 14C-labelled matridine (41) and matrine (42) has indicated the sequence for alkaloid biosynthesis as: matridine, matrine, matrine N-oxide, sophocarpine (43), sophocarpine N-oxide, and sophoramine (44).

Efficient and specific incorporations of various 14C-labelled matridines into matrine have been recorded.

Securinine. — A unique tetracyclic structure is a feature of the alkaloids of the Securinega genus (family Euphorbiaceae). Study of the biosynthesis of the most commonly occurring base, securinine (47), has revealed an appropriately unusual genesis for rings C and D.

Incorporation of DL-[2-14C]tyrosine was to give specific labelling of C-1, whilst the results with L-[U-14C]tyrosine confirmed the utilization of this amino-acid as the source of a C8 unit for rings c and d. (Tyramine, dopa, and homogentisic acid were much less effective precursors for securinine, and homogentisic acid label was randomized.) The results with phenylalanine exclude it as a precursor and provide yet another example of the non-equivalence of phenylalanine and tyrosine in biosynthesis in higher plants.