Molecular evolutionary lines in the formation of indole alkaloids derived from secologanin

Generation of the strictosidine → stemmadenine → ellipticine/olivacine and the strictosidine → vincadifformine → rhazinilam/rhazinal molecular evolutionary lines indicated the multistep bioorganic interactions in the frame of indole alkaloids derived from secologanin. Narrow structural elements, detailed chemotaxonomic data and standard organic reaction mechanisms, together with computer search of the DNP databases contributed considerably to the knowledge of the chemical background of this important class of natural products and detected several long-range electronic connections between the nitrogen atoms in these molecules.


Introduction
The molecular history of natural products is written in their structure.The study of their biogenesis should detect it.2][3][4] The main elements of this work are isolation and structure determination of the natural products, use of isotope-labelled compounds and isolated enzymes, and recently genomic methods as well.Elaboration of synthetic and biomimetic chemical reactions provided also useful support to this work.However, many of these methods are slow, time, material and energy consuming.Often it is difficult or even impossible to prove if reactions discovered or applied under abiotic conditions do really run in cells or organisms.Recently, the construction of large chemical data bases offer an alternative approach.The living cell may be considered as a chemical reactor in which the natural products are formed in parallel and/or consecutive steps from the educt-precursors.Isolated compounds from the same cells, species, genera and/or families may be guide-posts in this process.And in the case when highly reactive or sensitive intermediates could not be isolated in intact form, they can be postulated according to the standard organic reaction mechanisms.2][3][4] More than 2500 isolated representatives of them are formed mainly in three plant families Apocynaceae (APO), Loganiaceae (LOG) and Rubiaceae (RUB) from two building blocks secologanin 1 and tryptamine 2 through a single precursor strictosidine 3 (in this paper they are considered as indole alkaloids; however, sesqui-and dimer derivatives are temporarily disregarded).The narrow relatedness of the structures, the presence or absence of common structural elements, ring systems, functional groups, sesquimer and dimer alkaloids and stable centers of chirality 5,6 give important contributions to this work.And in addition, the chemotaxonomical data provide a strong support as well.It was supposed that these circumstances give a strong internal consistency to the whole group of indole alkaloids, where the results obtained in the different subgroups should mutually strengthen each others.
The aim of the present paper is to show that rather long molecular evolutionary lines can be detected by taking into consideration the three-dimensional structures, organic chemical reaction mechanisms and the appropriate chemotaxonomical data.The necessary chemical and biological data were taken from the Dictionary of Natural Products 7 (in the following: DNP) occasionally completed by the data of the Chemical Abstracts on line and the Beilstein CrossFire databases.In the structural formulae and throughout the text, the biogenetic numbering system shown in structure of strictosidine 3 was consequently used. 8The only exceptions are the building blocks 1 and 2 which are numbered as indicated in their own formulae.

The biogenetic-type system of indole alkaloids
In our previous study, 9,10 a biogenetic-type system of indole alkaloids (Scheme 2) was built up on the base of the three fundamental skeletons out of which secologanin is represented in these alkaloids.(In the schemes the number of the alkaloids in case can be given only approximatively, as it is changing by continuous renewal of the database.)Type I skeleton 5 contains the carbon frame in its original, type II 6 and type III 7 in a rearranged form. 11The generation starts with strictosidine corresponding to vincosan 8 i. e. to the basic skeleton of the type I α indole alkaloids in which C-3 (analogous to C-7 in secologanin) is attached to α position (C-2) of the indole ring.After deglucosylation, the hypothetic aglucone (represented as all-oxo aglucone 4 in scheme 1) undergoes several types of cyclizations and subsequent transformations.The main types of cyclization (scheme 3) take place between the nucleophilic N-4 centre and one of the four electrophilic centres C-17, C-19, C-21 and C-22 of the secologanin subunit, and affords the vallesiachotaman 9, malindan 10, corynanthean 11 and vincoside lactam 12 skeletons, respectively. 12An additional cyclization between C-16 and C-2 is directed, after subsequent 1,2-rearrangement of C-3 to C-7 (β position of the indole ring) to the formation of strychnan skeleton 13, which is the main type of the I β A alkaloids.Cleavage of the C-3-C-7 bond in 13 gives the hypothetic monoseco stemmadenine + intermediate in which the way is opened partially by subsequent izomerization and recyclization toward the aspidospermatan alkaloids (I β B), partially by further fragmentation to the bisseco secodine + intermediate.This likewise hypothetic structure affords by further isomerization and two types of cyclizations the fused (isoplumeran II β and plumeran III β) and the bridged (ibogan II and isoibogan III) alkaloids.Finally, the isoeburnan II α and eburnan III α skeletons are formed from the appropriate β alkaloids by 1,2-rearrangement of C-3 or C-21, respectively, to centre C-2 (i.e. α position of the indole ring).

Generation of line strictosidine 1 → stemmadenine 22E/O → ellipticine/olivacine 37/47
The study of this molecular evolutionary line was stimulated by the fact that the target alkaloids ellipticine 37 (in scheme 6) and olivacine 47 (in scheme 6) belong to a special group of indole alkaloids in which the type I skeleton 5 is intact, but the two carbon atoms C-5 and C-6 of the side-chain of the tryptamine subunit are removed, connection of the two subunits is strongly modified, and the two parent alkaloids are structural isomers, i. e. a methyl group is attached in 37 to the benzene and in 47 to the pyridine ring.The biogenetic numbering clearly shows these differences. 13Namely, the relation of the secologanin subunit and the N-4 atom corresponds in 37 to the corynanthean skeleton (11), in 47 to the malindan skeleton (10)   Derivation of ellipticine 37 is more straightforward.Enzymic deglucosylation of strictosidine 3 results in a multicomponent mixture of aglucones, which are represented also in Scheme 4 by the all-oxo form 4. 14 One of these aglucones may be 14 which after elimination of water gives the alkaloid cathenamine 15, and further by hydrogenation ajmalicine 16.Both of them belong to the corynanthean skeleton 11, and were isolated from natural sources.These types of alkaloids can be oxidized to different further alkaloids depending on the oxidation method. 15One of these products may be the ajmalicine hydroxyindolenine 17E isolated from Catharanthus roseus tissue culture.(In this and the following formulae of the line going to ellipticen alkaloids (E), R 1 =Me (C-18), R 2 =21-H.)The DNP database contains several other hydroxyindolenine derivatives.The next isolated compound in this line is preakuammicine 20E obtained from Vinca rosea (=Cantharanthus roseus).Its formation from 17E can be interpreted by cyclization of C-16 to C-2 in 18E, followed by a 1,2-rearrangement of C-3 to C-7 in 19E with elimination of water, and finally by reduction of the formyl group.18E and 19E are starting points of a large number of oxindole and strychnan alkaloids, respectively.Preakuammicine 20E is an important point in the further evolution.The compound contains long-range interaction between the electron-releasing N-4 and the electron-attracting N-1 atoms (indicated by curved arrows in 20E).Its further polarization by proton results in cleavage of the bond C-3-C-7.The positively charged and presumably highly reactive stemmadenine + A 21E structure was not isolated, but its reduced form stemmadenine 22E (in scheme 5) was obtained from natural sources.Stemmadenine + A can be izomerized by proton into stemmadenine + B, in which recyclization in direction C-21 → C-7 rather than C-3 → C-7 affords the aspidospermatan (I β B) alkaloids (shown in scheme 2).Stemmadenine 22E was isolated, among other species, from Rhazya stricta together with strictosidine and certain secodine derivatives (see later).This fact shows that the key compounds toward the evolution of ellipticine (and olivacine) were found in the same or related plant species.The former is realized in the ulean 31, the latter in the ellipticen 32E alkaloids.As these processes were analyzed previously, 10 at this place only the formation of the ellipticen alkaloids is briefly shown in the upper part of Scheme 6.
In the line going to the ellipticen alkaloids, deprotonation of the cyclization intermediate C 30E (=33, R 1 =Me (C-18), R 2 =21-H) gives already the tetracyclic structure 34.The final steps toward ellipticen alkaloids 37 involve three types of tautomerization, elimination of methanol and carbondioxide (de(methoxycarbonylation), actually probably ester hydrolysis and decarboxylation) and dehydrogenation through 35 and 36.The driving force of the cascade of these reactions is evidently their increasing π-delocalization.The stepwise character of dehydrogenation could be demonstrated by isolation of the partially hydrogenated derivatives in 36.The closing oxidative demethylation (36 → 37) with elimination of C-5 propably runs in the tertiary immonium salt, which is supported by isolation of N-methyl-3,14-dihydroellipticine and N-methylellipticine, according to 36 and 37, respectively.
An additional remark should be added.Among the isolated ellipticen alkaloids there are two compounds in which the carbon atom of the ligands is on a higher oxidation state by one, i. e. 18-hydroxyellipticene and 17-oxoellipticene.Their isolation from natural sources suggests that in both cases one reductive step falls out in their formation.In case of 17-oxoellipticine, this step is the reduction of the formyl group at C-16 (step 19E → 20E), and in case of 18hydroxyellipticene, the 20,21-double bond was not saturated (step 15 → 16).In this latter case, in the subsequent structures from 18E on, a 18,19,20,21-conjugated double bond system is present, which can be partially hydrated in a later state (probably before the final cyclization, i. e. step 33 → 34) of the molecular evolution.
As it was mentioned previously, 8 alkaloids of the isomeric olivacen skeleton (32O in scheme 5) were also isolated from natural sources.When the molecular evolution of ellipticine and olivacine is paralelly followed stepwise back to strictosidine 3, it might be concluded that the extra methyl group in these alkaloids should have been transposed formally from C-19 to C-20, i. e. it should be C-21.Of course, such transposition cannot be rationalized in the frame of the present biogenetic principles.However, such a skeleton might be formed, when at the very beginning of the molecular evolution, the primary cyclization of N-4 takes place at C-19 rather than C-21 (see scheme 3).The skeleton formed by that way corresponds to the malindan (10)  rather, than the corynanthean (11) structure.Really, this skeleton exists in isodihydrocadambine (39 in scheme 4) and several other alkaloids (altogether nearly 40 natural products developed from it). 39 is a glucoside, like strictosidine, and unlike ajmalicine, which means that the malindan skeleton can be formed before deglucosylation.The fact that 39 was isolated from Anthocephalus cadamba of the Rubiaceae family together with strictosidine 3, and its C-18 has a hydroxy group, suggests that it could be generated from strictosidine by epoxidation of the 18,19-double bond (38) followed by a nucleofilic attack of N-4 to 39 (lower part of scheme 4).These circumstances together with the additional fact that both ellipticen and olivacen alkaloids were isolated from the same genera Aspidosperma and Ochrosia (though not from the same species) (Apocynaceae) allows to take up mainly analogous common steps in their formation.These are shown in schemes 4 (steps 1 → 38-42 → 17O-21O), in scheme 5 (steps 21O-26O → 30O → 32O) and in scheme 6 (steps 30O → 43-47).(In the formulae of the line going to the olivacen alkaloids (O), R 1 =21-H, R 2 =Me (C-18).)However, in the line going to olivacine, alkaloid groups analogous to the strychnan, aspidosospermatan, stemmadenine and secodine alkaloids as well as the type II and III alkaloids were not found as natural products.

Generation of line strictosidine 1 → vincadifformine 58 → rhazinilam/rhazinal 64/73
In this line a different challenge was presented to our studies.In the large and diversified group of the plumeran type (III β) alkaloids (more than 350 natural products) (Scheme 2) there is a small set of compounds, in which the bond C-2-C-7 of the tryptamine subunit is cleaved.The nine representatives of it (rhazinilam 64 ( in scheme 10), leuconolam and related compounds) have a quaternary carbon atom in their secologanin subunit, which suggested the presence of type III skeleton 7. Neither of these alkaloids has a methoxycarbonyl group (=C-22) attached to C-16, however, one of them, rhazinal 73 has a formyl group at C-5.It may be supposed that its carbon atom corresponds to C-22, however, its unusual position presented some difficulties for the interpretation of its generation.The first task was to derive the parent structure of these alkaloids (according to 64), followed by a second one to interpret the origin and unusual position of the formyl group in rhazinal 73.
The general molecular evolution going to the plumeran alkaloids is given in scheme 7. The starting point is the charged stemmadenin + A 21E.If its 17-hydroxy group were transformed into a good living group (e. g. by acetylation, protonation or attachment to an enzyme), a fragmentation could start (according to the curved arrows in 21E), secodine + A 48 be formed and subsequently isomerized into secodine + B 49 in a deprotonation-reprotonation process.The driving force of this fragmentation is extension of the conjugation and the probable positive (or at least slightly negative) entropy.In secodine B 50B generated by deprotonation of 49, the conjugated double bonds can undergo a Diels-Alder reaction indicated by curved arrows, and immediately 18,19-didehydrotabersonine 51, the first member of the plumeran series should be formed.The whole matrix of transformations was analyzed previously 6 , and summarized briefly in Scheme 8. Unfortunately, 48 and 49, as well as 50A-D are hypothetic, probably highly reactive intermediates, and their existence could not be demonstrated directly.However, several partially hydrogenated secodine alkaloids were isolated, among other species, from Rhazya stricta.In 50A-D, types of conjugated double bond systems are found which can undergo Diels-Alder reactions through highly polarized transition states indicated by curved arrows, and afford the first representatives of the isoplumeran, plumeran, ibogan and isoibogan alkaloids, respectively.The actual mechanism of these transformations are likewise unknown.However, by the supposed way, a large number of isolated alkaloids (probably more than 500) could be derived from the indicated structures (mainly from 51, and except the isoibogan structure having only 5 alkaloids isolated till now), several of which contain unsaturation exactly at the sites and levels required by the supposed Diels-Alder reactions.
For the further molecular evolution toward rhazinilam 63, it is necessary to cleave bond C-7-C-21.In scheme 9 a further long-range interaction is shown by curved arrows in tabersonine and eburenin derivatives 52 and 53, respectively, which will probably be effective in the presence of proton.In this further key fragmentation the charged intermediates 54 and 55 should be formed, which are again hypothetic structures, however, their hydrogenated representatives in the vincadine 56 and quebrachamine 57 derivatives (14 and 20 alkaloids, respectively) are well known.Some quebrachamine derivatives were isolated even from Rhazya stricta.The fact that the methoxycarbonyl group is absent in rhazinilam 64 and related alkaloids suggests a de(methoxycarbonylation) somewhere in the molecular evolutionary line.Unfortunately it is not known if it takes place before or/and after the fragmentation.In the demethoxycarbonylated derivatives, the original double bond at C-2-C-16 is automatically moved to N-1-C-2 position conjugated to the benzene ring.In addition, the double bonds at position 18,19 and 14,15 of 51 and their derivatives, can partially be removed later by hydrogenation or other usual transformations of the olefinic elements.In the charged intermediates 54 and 55, the positive charge can temporarily be neutralized by a nucleofile from the proper solvent.It should be added that in the presence of ligand C-22, a further alternative is opened for the molecular evolution from 54 to 65 (see later).
Formation of rhazinilam 64 is shown in the lower part of scheme 10.The starting point would be vincadifformine 58, which can be formed by hydrogenation from the first plumeran compound 51.After de(methoxycarbonylation), in eburenin 59 the bond C-7-C-21 should be cleaved with formation of the charged intermediate 60.The C-2-C-7 double bond of it could be oxidized into a vicinal dihydroxy group in 61.This type of reaction is known in the indole alkaloids having an intact pyrrol ring, and several analogous alkaloids are found in the DNP database.By removal of a proton accompanied by a 1,2-rearrangement, C-7 can migrate from C-2 to C-21 with simultaneous formation of a nine-membered ring in 62 characteristic for the derivatives of rhazinilam.Dehydration in 62 with formation of a double bond in conjugation to the benzene ring provides already a natural product 63, and further dehydrogenation completes the formation of rhazinalam 64 itself.Further simple transformations not shown in the scheme afford the leuconolam derivatives.
However, there is a second, more challenging problem.It is highly improbable that in rhazinal 73 a functionalized carbon ligand (C-22) would disappear from position C-16, and a similar one appear at C-5 from a source which would not be secologanin.A more satisfying solution would be the internal transposition of C-22 from C-16 to C-5, i. e. by cleavage of bond C-16-C-22 und construction of a new one C-5-C-22.The first suggestion to such a solution came from the observation that unlike other derivatives of rhazinilam 64, rhazinal 73 was isolated from a malayan Kopsia species.The typical alkaloids of the Kopsia genus are the kopsan alkaloids (more than 40 compounds), and their characteristic structural element is a onecarbon bridge (C-22) between C-6 and C-16.In scheme 9, an analogous possibility is opened in the charged tabersonine intermediate 54 in which 5-H is activated by the immonium character of N-4, and a one-carbon bridge (C-22) might be constructed between C-5 and C-16 in 65.This analogy suggested the derivation of rhazinal 73 along the line applied in the formation of rhazinilam 63, with intercalation of the bridge-making and bridge-braking steps necessary for transposition of C-22 from C-16 to C-5.According to the upper part of scheme 10, the development of rhazinal 73 would start likewise with vincadifformine 58.However, its fragmentation should run without de(methoxycarbonylation), and in 66 the structural conditions of the bridge formation are already given.Cyclization to 67 might formally be catalyzed by proton.The next two steps 67 → 68 → 69 are analogous to steps 60 → 61 → 62 in the generation of rhazinilam.However, in 69 the amide carbonyl is a part of a β-dicarbonyl system and the bond C-16-C-22 could be cleaved by a nucleophile.Further steps 70 → 71 → 72 (dehydration and dehydrogenation) are again analogous to steps 62 → 63 → 64.Were the nucleophile a hydride anion in step 69 → 70, C-22 would get immediately the formyl oxidation level.Were the nucleophile a methanolate ion, the C-22 would be part of a methoxycarbonyl group which should be reduced directly, or through the hydroxymethyl level and partial reoxidation to the formyl group in 73.
Unfortunately, rhazinal is a single compound without other known tightly related structures and was isolated only from an undefined "malayan" species.However, the chemistry in its formation as proposed above shows further analogies observed in the generation of other kopsan alkaloids or isolated from other Kopsia species.Such analogies are the bridge formation mentioned above (step 66 → 67) and the rearrangement of a 1,2-dihydroxy system (61 → 62 and 68 → 69) in the derivation of the melodan alkaloids (isolated partly from a malayan Kopsia species), 16 and the cleavage of the β-dicarbonyl system (step 69 → 70) in the formation of chanofruticosan subgroup of the kopsan alkaloids (isolated from other Kopsia species). 17

Conclusions
The conclusions are summarized in Table 1.It should be remarked that the alkaloids shown in the previous schemes were generally isolated from several species, however, Table 1 gives only the most characteristic one or two sources.1.The molecular evolutionary line from strictosidine 3 to rhazinilam 37 could be derived with high reliability, as both the starting educt 3 and the end product 37 were isolated from the same species Rhazya stricta, and many the intermediates or their narrow derivatives were isolated likewise from Rhazya stricta and/or Catharanthus roseus (=Vinca rosea).2. Rhazinal 73 as a single compound of its own type was isolated only from a malayan Kopsia species but from Rhazya stricta not yet.However, its formyl group at C-5 was successfully derived from the methoxycarbonyl group at 16 in a bridge-making-bridge-braking reaction sequence.3. The molecular evolution of ellipticine from secologanin was abundantly supported by isolated natural products or their tightly related derivatives.This derivation was enhanced by the fact that ellipticen and olivacen alkaloids were isolated from common Ochrosia and Aspidosperma genera (though different species).species.e Ochrosia species and Aspidosperma species.f Kopsia species.g Anthocephalus cadamba 4. Though both ellipticine and olivacine are type I indole alkaloids, they belong to different primary cyclization patterns (common corynanthean and special malindan, respectively).However, their minute structural difference consists only in the mutual exchange of a methyl group and a hydrogen atom.Therefore, as a consequence of the tight structural similarities, the complete molecular evolutionary line could be derived with high reliability, though in the olivacine line only the identity of the final product and the chemotaxonomic relation of the starting precursor isodihydrocadamine with strictosidine could be established.This result could not have been obtained without the high potency of the computer-assisted searching of the large databases.5. Several long-range bond-braking-bond-making interactions between the two nitrogen atoms N-1 and N-4 have a key role in the construction of molecular evolutionary lines of indole alkaloids derived from secologanin.6.Some plant species (e. g.Kopsia species) construct only a few very special types of alkaloids, however, Rhazya stricta, Catharanthus roseus and several other species provide natural products in nearly all classes of indole alkaloids.

Scheme 2 .
Figures indicate approximate number of isolated alkaloids

Scheme 8 .
Scheme 8. Formation and transformation of the secodine structure.