The radical cation mediated cleavage of catharanthine leading to the vinblastine type alkaloids: implications for total synthesis and drug design

Abstract

The landmark synthesis of vinblastine (and analogues) by Boger et al. involves a tandem [4 + 2]/[3 + 2] cycloaddition that resulted in the construction of vindoline followed by FeCl₃ promoted, radical cation mediated, catharanthine fragmentation/vindoline coupling that yields anhydrovinblastine (after NaBH₄ reduction of the iminium ion) or vinblastine directly (by virtue of exposure of the first formed anhydrovinblastine to ferric oxalate-NaBH₄, air, 0 °C). It is important to emphasise that this inspired and versatile innovation is essentially a one-pot vinblastine synthesis resulting from exposure of catharanthine/vindoline to a FeCl₃/Fe₂ (oxalate)₃ combination. The mechanistic hypothesis of Kutney, in which the C(16)–C(21) cleavage in catharanthine leading to the azabenzfulvene is triggered by tertiary amine oxidation to the radical cation, has been subjected to a formidable challenge by Boger who has suggested, based on careful experimentation involving a number of judiciously selected simplified analogues, that indole C(2)–C(3) oxidative radical cation generation rather than tertiary amine radical cation formation is the key event that renders fragmentation/coupling possible. This review will attempt to assess the evidence for this novel and intriguing alternative mechanistic proposal and the practical implications of this FeCl₃/Fe₂ (oxalate)₃ combination. A concise history of the azabenzfulvene intermediate will first be presented by way of introduction.

---

Maryam Sadat Alehashem

Maryam Sadat Alehashem was born in May 26, 1987 in IRAN. She graduated from University of Malaya (UM) with a Masters degree in Science and Technology. She worked in the Global Oil and Gas Company in Tehran, Iran. In 2015 she started working towards her PhD in organic synthesis (radical cation mediated heterocyclisation) under the supervision of Prof. Dr Noel F. Thomas in the Department of Chemistry University of Malaya.

Chuan-Gee Lim

Lim Chuan Gee received his BSc from Monash University in 1985 and his MSc from The University of New South Wales in 1990. He joined the then Standards and Industrial Research Institute of Malaysia in 1991 which was then corporatized as SIRIM Berhad in 1996.

Noel F. Thomas

Noel F. Thomas was born in Balham, South London in 1956. After graduating from the University of Salford (B.Sc. Hons 1980), he worked under Dr Alan H. Davidson for his PhD which he obtained in 1985, from the University of Wales Institute of Science and Technology (now Cardiff University). After postdoctoral adventures at the University of Southampton, England and Duke and Trinity Universities, USA, he began his independent research career in the School of Pharmacy and Pharmacology, University of Bath (1990 to 1995). Since 2004, Noel Thomas has taught organic chemistry and conducted research in the Department of Chemistry, University of Malaya developing methods for the synthesis of indolines, bisindolines and indolostilbenes by oxidative generation and manipulation of ortho amidostilbene radical cations. He was promoted to full professor in 2013.

---

Introduction

In a major development, Potier et al.¹ applied his modification of the Polonovski reaction^2–5 to the synthesis of alkaloids of the vinblastine type by exploiting a stereoelectronically permissible fragmentation of the catharanthine N-oxide, a process that leads to the electrophilic conjugated imine (an azabenzfulvene). This method of achieving C(16)-C(21) cleavage via the Polonovski Potier reaction¹ may be fruitfully compared with the single electron transfer oxidative cleavage (by means of FeCl₃) of this crucial bond, which it would appear results in a similar azabenzfulvene intermediate according to Kutney's proposal.⁶ In these processes a high degree of stereochemical control in the installation of the C-16′ stereocentre is critical. A recent detailed examination of the FeCl₃ oxidative cleavage by Boger,⁷ in the course of his vinblastine and related syntheses,^8–10 has cast doubt on Kutney's mechanistic proposal culminating in an equally provocative alternative mechanistic interpretation. Although Boger has recently produced an elegant summary¹¹ of his explorations in the chemistry of vinca alkaloids, vinblastine and related natural products, including methodological development and structure–function properties, our particular interest in this review is the FeCl₃ promoted, radical cation mediated fragmentation/coupling and its implications in the light of recent interest in iron reagents in organic syntheses¹² and the expanding of field of radical cation mediated transformations.^13,14 An assessment of the mechanistic insights of Kutney, Sundberg and of course Boger, will be critical to this endeavour.

Discussion

In the 1950's, the Madagascar periwinkle plant was investigated in the hope that molecules may be discovered that could treat the life threatening condition known as diabetes. This hope was not to be realized but what Noble et al. discovered,¹⁵ almost simultaneously with the Svoboda group,¹⁶ would turn out to be no less important. A molecule was discovered that not only demonstrated high levels of cytotoxicity against a range of cancers but, did so by a novel mechanism.^11,17,18 The molecule, now widely known as vinblastine, inhibits spindle formation during mitosis thus inducing apoptosis. Vinblastine is one of the most important anticancer drugs as well as one of the most expensive and therefore much effort has been devoted to total and partial syntheses of this drug. Our understanding of vinblastine biosynthesis has advanced over the years.^17,18 Kutney established, as a result of careful incubation studies and HPLC^19–22 analysis, that the N-oxide (7), Scheme 2, was not a substrate for the enzymes involved in the process. A more reasonable hypothesis is that a peroxyindolenine intermediate (2) (Scheme 1), already implicated in earlier studies, is involved leading to the dihydropyridinium intermediate (4) whose importance has already been established. It should be noted that recently a basic peroxidase^17,18 presumably responsible for the generation of the peroxyindoleine (2), Scheme 1, and named catharanthine (C) roseus peroxidase, has been purified and characterized from C roseus leaves. The fragmentation of the peroxyindoleine²³ gives rise to the “catharanthyl” electrophile which is then attacked through the C(15) position of the vindoline nucleophile see (3) resulting in the indolo-indoline, vinblastine (6) Scheme 1. After prior conversion of catharanthine to the rigid ‘Ibogane-type” N-oxide (7).^1,24,25 Potier treated the latter with trifluoroacetic anhydride at −50 °C in the presence of vindoline (3). This resulted after work up in anhydrovinblastine (11),^1,26 as the major product (Scheme 2). For a comprehensive and insightful discussion of the isolation, biosynthesis and total synthesis of bisindoline alkaloids of the aspidosperma-cleavamine type (e.g. vinblastine), the indicated reviews^17,18,22 are highly recommended.

Scheme 1 Proposed biosynthesis of anhydrovinblastine and vinblastine.

Scheme 2 Potier's “biomimetically” patterned synthesis exploiting the Potier–Polonoski reaction.

It should be noted that the formation of the azabenzfulvene (9) (Scheme 2), is mechanistically significant. The stability of the conjugated N⁺-C(21) iminium ion (9) rules out the alternative C(6)–C(5) fragmentation (as a substantial contributor). Nucleophilic attack by vindoline occurs predominantly but not exclusively from the α-face of (9) as this conformation (rather than (10)) predominates at −50 °C.^25,27,28 One may be left with the impression that formation of the azabenzfulvene intermediate requires the prior cleavage of the C(16)–C(21) bond, however the brilliant construction of vinblastine by Fukuyama suggests that this need not be the case. Inspired by the seminal investigations of Schill et al.,²⁹ in which 16-methoxycarbonyl-15,20-dihydro-3,N_b-secocleavamine was coupled with (−)-vindoline via a chloroindolenine intermediate, Fukuyama prepared the 11-membered macrocyclic indole²⁷ (16) in eleven steps from the highly functionalized building blocks (12) and (13). This (16) on treatment with t-BuOCl at 0 °C (ref. 27) followed by exposure of the isomerised chloroindoleine (18) to TFA²⁷ in the presence of vindoline (3), effected a remarkably stereoselective (note the C(16′) centre) coupling. The intermediate (20) required only a few more steps to secure (+)-vinblastine.²⁷ Overall (16) was converted to vinblastine in more than forty steps which included a highly original construction of (−)-vindoline. This synthesis, elegantly executed, dispensed with the rigid ibogane framework in favor of the generation of the macrocyclic azabenzfulvene precursor. The Fukuyama synthesis is a veritable treasure trove of novel transformations and rewards careful study (Scheme 3).

Scheme 3 Fukuyama's “macrocylic azabenzfulvene” approach to vinblastine.

This outstanding “non-ibogane fragmentation” approach notwithstanding, the recent report⁷ of the oxidatively generated radical cation controlled fragmentation of catharanthine as a prelude to vinblastine synthesis, its mechanistic and synthetic aspects and structure–activity implications are of such importance, that these investigations must now be the focus of our attention starting with the landmark report of Kutney et al.³⁰ and related papers.^31–33 The coupling of vindoline to catharanthine remains attractive to many as it offers a less hazardous and shorter alternative to total synthetic approaches to the vinblastine molecule. It is common place these days to speak of a “crisis” with regard to total synthesis with increasing skepticism about the value of such explorations and obstacles to continued funding and publication.³⁴ Some suggestions for dealing with this perplexing situation have been provided by Mulzer.³⁴ However examination of recent developments in vinblastine synthesis, and analogues, for example, suggest that rumors of the “death” of total synthesis are greatly exaggerated. We will have more to say about this later in our discussion. Recognizing that anhydrovinblastine is a possible precursor for vinblastine and vincristine, Kutney et al. coupled catharanthine with vindoline in the presence of FeCl₃. Although he considered a concerted mechanism to account for the exclusive generation of the natural configuration at C(16′) of anhydrovinblastine, we will examine in some detail his stepwise mechanism which also requires prior generation of the tertiary amine radical cation. The Kutney mechanism is depicted below (Scheme 4).

Scheme 4 Kutney's mechanistic hypothesis; the radical cation mediated catharanthine fragmentation/vindoline coupling.

Several observations may be made about the mechanism in Scheme 4.

(1) Two Fe³⁺ oxidations are required. The first generates the tertiary amine cation radical (21). The second converts radical cation (22), resulting from fragmentation of the C(16)–C(21) bond, to the resonance stabilized cation (23) and (24) or (25).

Major reviews on the synthetic utility of Fe³⁺ promoted transformations involving radicals and radical cations have appeared.¹³ We reported a FeCl₃ promoted syntheses of a restrytisol analogue³⁵ and later, bisindoline³⁶ and indolostilbene dimers³⁷ via stilbene radical cations. Our systematic study of anodically generated stilbene radical cations and their dimersation chemistry has also appeared.³⁸ Yoon and Ischay have published a fascinating microreview on radical ions³⁹ which includes an assessment of transition metal photocatalysis as a synthetically useful method for generating these reactive intermediates. An even more comprehensive review by MacMillan et al. highlighting the extraordinary versatility of (Ru(bpy)₃²⁺)¹⁴ in visible light photocatalytic radical cation generation was published in 2013.

(2) The Kutney proposal requires the generation of a cation (23) adjacent to a δ⁺ ester carbonyl carbon. This cation is of course resonance stabilized as the formation of the dicationic azabenzfulvene (24) indicates.

(3) There is ample precedent for the generation of tertiary amine radical cations that could be compared to (21) (Scheme 4). A few examples are discussed below:

Pandey et al. reported a photoinduced electron transfer α-functionalisation of a tertiary amine⁴⁰ prior to intramolecular trapping of the iminium ion (Scheme 5). The radical cation (28) is a key intermediate. The mechanistic insights of Rueping et al.⁴¹ in a rather different setting reward careful examination as they illuminate certain aspects of the Pandey and related transformations. A similar process, this time exploiting electrochemical oxidation as the key step in a Kopsidine synthesis,⁴² was reported by Kam et al. (Scheme 6). In both of these examples, the generation of the amine radical cation and formation of the iminium ion required two separate oxidation steps.

Scheme 5 α-Functionalization of a tertiary amine via Pandey's photoinduced electron transfer.

Scheme 6 Kam's electrohemical single electron transfer oxidation approach to Kopsidine.

The photoinduced electron transfer process exploited by Santamaria⁴³ in his Criocerin synthesis is particularly significant as it required oxidation of the tertiary amine in the presence of the indole C(2)–C(3) double bond. The iminium ion (41), generated via the radical cation (40), is vinylogously trapped thus providing the final product. This problem of regio – or chemoselective single electron transfer oxidation will be revisited when the Boger chemistry is discussed.

The Kutney proposal revisited

In Boger's landmark publication,⁷ the suspicion that the mechanistic details revolving around the presumed azabenzfulvene intermediate may be more complex than first realized was initiated by a number of observations, principally, that catharanthine was recovered from the FeCl₃ reaction mixture in the absence of vindoline to a degree that seemed inconsistent with the reaction conditions (NaBH₄, work up) and with the presumed involvement of the azabenzfulvene. Boger had employed modified conditions for Kutney's original transformation, which were (FeCl₃, 25 °C, 2 h, CH₃CH₂OH, 0.1 N HCl – CF₃CH₂OH), then NaBH₄ or FeSO₄ instead of FeCl₃ (with slightly lower conversions)⁴⁴ for the anhydrovinblastine synthesis. The catharanthine/vinblastine coupling proceeds with exclusive inversion of stereochemistry at C(16). This stereochemical outcome might well suggest a concerted mechanism, an alternative Kutney himself considered.³⁰ On the other hand the stepwise azabenzfulvene mechanism would appear to be supported by Sundberg's observation that under Kutney's conditions N-methyl catharanthine (42, Fig. 1) is consumed,⁴⁵ yet does not couple with vindoline most of which was recovered. On the other hand N-methyl catharanthine is recovered unchanged in the absence of vindoline which implies firstly, a catalytic role for vindoline and secondly, that the neutral azabenzfulvene may be the crucial intermediate ((25), Scheme 4). Boger et al. has reported,⁴⁴ in contrast to Sundberg,⁴⁵ that N-methyl catharanthine (42) is consumed (but not coupled) under Kutney's conditions both in the presence and absence of vindoline. Sundberg⁴⁵ embarked upon a meticulous investigation that employed creative design of analogues of catharanthine and an examination of their oxidation profile which will now be discussed. The Sundberg group synthesized catharanthine analogues in which the [2,3] fusion found in the natural compound (1) was replaced by [2,1] (43) and [3,2] (44) fusion as shown in Fig. 1

Fig. 1 The Sundberg natural (1) and (42) and unnatural catharanthine analogues (43) and (44).

Sundberg's objective was to apply the Polonovsky–Potier and Kutney fragmentation protocols to catharanthine (1) and the analogues (43) and (44). He made the intriguing discovery that when (43) was converted to the corresponding amine oxide (45) and subjected to TFAA at low temperature in the presence of vindoline (Potier conditions),¹ the product mixture consisted of mainly unreacted vindoline plus two diastereomeric coupling products (analogous to anhydrovinblastine) (10–15% yield each). The formation of these coupling products is best explained⁴⁵ as shown in Scheme 8.

Scheme 7 Santamaria's single electron transfer oxidation (photoinduced) to Criocerin.

Scheme 8 The subjection of the unnatural catharanthine N-oxide to the Polonovski–Potier conditions.

In an attempt to account for the remainder of the N-oxide (45), it was exposed to Potier's conditions¹ (in the absence of vindoline) followed by NaBH₄ (and separately, NaBD₄) work up. The unnatural catharanthine analogue (43) was recovered the product of reduction of (45), but no deuterium was incorporated. These results naturally lead to the unexpected conclusion that in the absence of vindoline, fragmentation of the N-oxide does not occur (Scheme 9).

Scheme 9 The fate of (45) under Polonovski–Potier condition's followed by a reductive work up (vindoline absent).

Sundberg drew the further conclusion that vindoline may catalyse the fragmentation of the unnatural catharanthine analogue (43). A similar observation has been made in connection with catharanthine – N-oxide. This raises the question of the possibility of a vindoline effect in the case of Fe³⁺ oxidation of catharanthine to which we shall now turn.

Fe³⁺ promoted N-oxide fragmentation; the vindoline effect

When Sundberg et al. subjected the unnatural [3,2] catharathine analogue (44) to Fe³⁺ (i.e. FeCl₃·6H₂O) oxidation with the NaBH₄ work up (and in the presence of vindoline), vindoline was recovered virtually quantitatively.⁴⁵ The major products were the allylic alcohols (57) and (58) Notice that the formation of (56) (57) and (58) is consistent with the formation of the amine radical cation (52).

The fragmentation/capture path that accounts for the products in accordance with Kutney's hypothesis, is shown below and appears to be unaffected by the absence of vindoline (Scheme 10).

Scheme 10 The exposure of the unnatural catharathine analogue (44) to Fe³⁺/NaBH₄.

The strange behavior of catharanthine

When catharanthine (1) (with the natural [2,3] fusion) – Fig. 1, was subjected (in the absence of vindoline) to the same Fe³⁺ oxidation conditions (with NaBH₄ work up) as the [3,2] analogues (Scheme 10) and the [2.1] analogues, products analogous to those depicted in Scheme 10 were not observed. Indeed Sundberg et al. found “almost no reaction^” with vindoline absent whereas greater than 90% fragmentation and concomitant coupling occurred with vindoline present as expected based on Scheme 4. Only 10% conversion of catharanthine was observed in the absence of vindoline. The remarkable persistence (recovery) of catharanthine under these conditions (FeCl₃ vindoline, NaBH₄ work up) demands an explanation. Furthermore in the words of Sundberg⁴⁵ “the implication is that vindoline participates in the rate-determining step for Fe³⁺ oxidation of catharanthine”.

To see what this could mean for the Kutney azabenzfulvene hypothesis, we must now turn to Boger's seminal investigation placing the above discussion in context.

The Boger synthesis of vinblastine

To summarise Boger's achievement it is important to remember that the extensive study of the FeCl₃/catharanthine fragmentation had a wider objective. He developed powerful and versatile methodology that exploited a remarkable tandem [(4 + 2)/(3 + 2)] cycloaddition that provided access to (−) and (+) vindoline. These discoveries were reported in a series of papers^46–51 and laid the foundation for his synthesis of vinblastine and analogues by means of a modification of Kutney's oxidative protocol⁶ to include a stereoselective C(20) oxidation^8,10 (Scheme 11). Boger's first generation approach is depicted in Scheme 11. With the oxadiazole (60) installed, an inverse electron demand Diels–Alder reaction with the tethered benzylenolether gave rise, after nitrogen extrusion, to the 1,3-dipole⁴⁶ (62). This subsequently engages the indole double bond, the dipolarophile, leading to (63).

Scheme 11 Boger's first generation vinblastin and anhydrovinblastin syntheses.

The transformation of (63) to (3) may be summarised as involving α-oxygenation in the form of OTIPS, (see asterisked carbon in 63) – by means of LDA, bis(trimethylsilyl)peroxide (TMSO)₂ followed by triisopropyl trifluoromethanesulphonate (TIPSOTf),⁴⁶ thiolactam formation (Lawesson's reagent), desulphurization, oxybridge cleavage via the iminium ion which was subsequently reduced (H₂/PtO₂) and elimination of the α- OTIPS via the α-OH (PPh₃ – DEAD).⁴⁶ Boger exploited the Kutney FeCl₃ promoted catharanthine fragmentation that enabled stereoselective installation of the C(16′) stereocentre culminating in anhydrovinblastine (5). An ingenious modification that combined FeCl₃ with Fe₂(OX)₃ (ref. 8 and 46) resulted in the stereoselective oxidation of the C(20′) olefin in anhydrovinblastine to produce vinblastine (6). It should be noted that access to (−)-vindoline in this synthesis was expedited by a resolution of the key intermediate⁴⁶ (63). Boger's asymmetric synthesis of vindoline en route to vinblastine⁹ and pre-clinically relevant analogues^11,52–54 will be discussed later in the conclusion where we also produce a summary statement on the azabenzfulvene question.

Suspecting that the Kutney mechanism for the fragmentation of catharanthine via the tertiary amine radical cation may require a more careful examination and the observation of the remarkable persistence of catharanthine in the absence of vindoline (as we have previously described). Boger set about, in an imaginative and systematic study, to answer the following questions:

(1) Will electron rich aromatic substrates structurally simpler than vindoline participate in the catharanthine fragmentation/coupling reaction?

Boger found that the substrates below participate in the reaction with good to excellent yields⁷ Altogether seven substrates were examined although only a few of these are shown below (Scheme 12).

Scheme 12 Boger's model experiment I: the FeCl₃ promoted catharanthine fragmentation/coupling with aromatic substrates. All transformations were carried out as follows (i), FeCl₃, 25 °C/20 h, 0.05 N HCI/CF₃CH₂OH (ii) NaBH₄.

It was found that each transformation produced not only a single diastereoisomer but also a single regioisomer. It was noted that less electron rich substrates e.g. methoxybenzene, 3-methoxythioanisole, and neutral substrates for example benzene and electron deficient substrates like methylbenzoate failed to couple with catharanthine. The site of substitution for I–IV (Scheme 12) and is consistent not with only the involvement of a catharanthine stabilised cation ((23), Scheme 4) but also with a radical mechanism. The lack of any coupling for even moderately nucleophilic systems like anisole and benzene is rather unexpected on the basis of the presumed involvement of the azabenzfulvene intermediate. Paramethoxybenzene rings are known to capture electrophiles intramolecularly, generated in the course of FeCl₃ promoted stilbene dimerisation³⁸

(2) Is the tertiary amine in catharanthine the site of oxidation as had been previously assumed?

Boger designed a series of simplified analogues all lacking the classical azabicyclo[2.2.2]octane or quinuclidine ring system.⁷ He studied the coupling of simple indole derivatives with alternatively vindoline and 3-dimethylamino anisole. These studies were conducted using modified⁴⁴ Kutney conditions (Scheme 13) of which only a selection of the substrates examined by Boger are shown below.⁷ These results are significant for a number of reasons. Firstly the tertiary amine (quinuclidine) appears to be unnecessary from the perspective of a successful coupling. Secondly, if a neutral azabenzfulvene⁴⁵ is thought to be necessary (Scheme 4), this would require a free NH, but Boger's results demonstrate successful coupling of even N-methylated indole derivatives.

Scheme 13 Boger's model experiment II: the successful coupling of indole systems lacking the quinuclidine ring system. All transformations were carried out as follows FeCl₃, (5 equiv.) 25 °C/aq 0.05 N HCI/CF₃CH₂OH.

The above investigation was extended to include the higher order indole systems (75) and (76), Fig. 2, shown below:

Fig. 2 [6.5.7] and [6.5.8] indolic systems employed in an extended study (see Scheme 13).

Thirdly, the possible inhibition of tertiary amine oxidation as a result of protonation under acidic (Kutney's) conditions is now irrelevant since the results summarized in Scheme 13 strongly suggest that oxidative coupling is successful even in the absence of the tertiary amine and that a free indolic NH is not essential. In the light of the above, doubt must now be cast on the necessity for the formation of either a neutral or protonated azabenzfulvene ((21), Scheme 4). Having dispensed with the tertiary amine moiety (but see p. 38, Scheme 23), the generality of the coupling was investigated by use of indole (6.5.5) systems.

(1) Will indolic (6.5.5) systems demonstrate the same coupling tendency as the (6.5.6) systems?

The answer is clearly yes based on the results below (Scheme 14).

Scheme 14 Examination of the coupling of the [6.5.5] indole system (76) with various nucleophiles. All transformations were carried out as follows FeCl₃, (5 equiv) 25° aq. 0.05 N HCI CF₃CH₂OH.

In comparing the results in Scheme 12 and 14, it is difficult to avoid the conclusion that an identical mechanistic pathway must apply to both sets of reactions. Boger et al. further observed that the absence of any products resulting from capture by nucleophiles such as water, (as the reactions were performed in acidic aqueous solution at room temperature) or capture by chloride ion, points to a pathway that does not involve an “indole-derived, neutral, or cationic azabenzfulvene”.⁷ Attention must now shift from tertiary amine oxidation to oxidation of the indole double bond. Lindstrom described the radical cation mediated FeCl₃ oxidation of methyl indole-3-acetate^7,55 (with concomitant dipropylamine trapping), which must proceed through a C(2)–C(3) indole radical cation (see Scheme 15). Readers will recall the work of Santamaria (Scheme 7) in which under photo-induced electron transfer conditions, the tertiary amine moiety in an indolopiperidine substrate underwent selective oxidation to the tertiary amine radical cation, with the indolic olefin unaffected.

Scheme 15 The FeCl₃ promoted oxidation/dipropylamine trapping of indole-3-acetic acid.

In the light of the above and the chemistry depicted in Scheme 15, Boger proposed a very different mechanistic interpretation to the FeCl₃ promoted catharanthine/vindoline coupling.

The Boger proposal: single electron oxidation of the indole moiety and its implications

The Boger group have opted for the oxidation of the indole ring to generate the C(2)–C(3) cation⁷ radical (87). The formation of (87) triggers the strategic C(16)–C(21) cleavage assisted by the lone pair on the tertiary nitrogen. The key steps are shown in Scheme 16.

Scheme 16 The Boger modification of the Kutney hypothesis: indole C(2)–C(3) radical cation formation.

Lone pair assisted fragmentation of the radical cation (88) gives rise to the allyl radical (89), an indolyl C(3)-radical, (Scheme 16) which is set up, via its alternative resonance form (90) – an electrophilic radical, to attack the vindoline partner at the C(15) position ((91)–(92) Scheme 17) to yield, after oxidation of the delocalised radical and deprotonation, the expected anhydrovinblastin.

Scheme 17 The Boger modification of the Kutney hypothesis: coupling of the C(16) catharanthine radical to vindoline.

The transformations shown in Scheme 16 and 17 are extremely thought provoking and some comment would appear to be necessary.

(1) The generation of indole radical cations, the corresponding C(3) indolyl radicals and their synthetic utility have ample precedent. The few examples shown below are instructive. The photochemically induced Diels–Alder reaction between indole and cyclohexadiene was described by Steckhan et al.⁵⁶ as proceeding via ((99) Scheme 18). Additional examples, some of them recent, of the creative exploitation of the cation radical Diels–Alder may be found in the accompanying papers.^56–59

Scheme 18 Steckhan's photochemically induced indole cation radical Diels–Alder reaction.

The indole C(2)–(3) radical cations are also implicated in Takayama's meso-chimonanthine⁶⁰ synthesis promoted by iodobenzene bistrifluoroacetate (PIFA).

It is noteworthy that dimensation proceeds via the indolyl (or pyrrolo-indolyl) radical (105) generated from the indolyl radical cation (104).⁶⁰ It should be noted that the methylated pyrrolloindoline dimers (106), (107), and (108) (R′ = Me) are obtained by RedAl reduction⁶⁰ of the corresponding carbamates. We have analysed this reaction in the context of our own studies.⁶¹ Analogous C(3) indolyl radicals are generated in the course of our FeCl₃ promoted one pot C-(2) phenyl substituted indoline/bisindoline synthesis.^36,62 Our observations (Scheme 20) would appear to be consistent with Takayama's chemistry (Scheme 19).

Scheme 19 Takayama's radical cation mediated Chimonanthine synthesis.

The anodic oxidation of stilbenes to the corresponding radical cations has been extensively studied by Eberson (1969) and Steckhan (1978). If the position of the cation is influenced by electron releasing substituents at the para position of the benzene ring then 3,4-dimethoxy groups would satisfy this condition (110), Scheme 20), in which case (110) must be a significant resonance contributor. Nucleophilic attack by the amide should produce the C-3 indolyl radical as shown in (111), Scheme 20. The formation of the dimer (113) is consistent with the prior formation of intermediates (110) and (111). This mechanistic interpretation also helps to explain the Takayama chemistry in Scheme 19 (where the carbenium ion is stabilised by the lone pair on the indoline nitrogen ((104) Scheme 19). It is worth noting that when the 3,4-dimethoxy substitution in (109) is changed to 3,5-dimethoxy,³⁷ (see (114), Scheme 21) which lacks a paramethoxy substituent, completely different products are obtained in which the bisindoline analogous to (113 Scheme 20) is not observed. The three products (116), (117), (118) (Scheme 21), are all formed, not unreasonably, via a different radical cation i.e. ((115) Scheme 21). Full mechanistic and stereochemical details are provided in the full papers.^36,37

Scheme 20 The acetamido stilbene radical cation approach to indoline and bisindoline synthesis.

Scheme 21 Effect of 3,5-dimethoxy substitution in (114) on the radical cation and dimensation products.

Crystallographic data for the bisindoline (113) and an extended study of the dimerization has been presented.⁶² The unexpected catalytic effect benzophenone on the FeCl₃ promoted oxidation coupling/cyclisation has also been described.⁶¹

After this important diversion, we can now examine further implications of the Boger chemistry (Scheme 16 and 17).

(2) Boger et al. has shown that for the coupling reaction ((91) Scheme 17) C10′ electron-donating and neutral substituents on the catharanthine ring promote the reaction. By contrast, electron-withdrawing substituents progressively produce lower yields and lower reaction rates.

Consistent with the electrophilic nature of the catharanthine derived C(16) radical that couples to vindoline ((91) Scheme 17), is the observation that when Boger treated ethyl iodoacetate with Et₃B, the electrophilic radical so generated coupled to vindoline in an analogous manner⁷ (Scheme 22).

Scheme 22 Boger's demonstration of the capture of an electriphilic radical at C (15) of vindoline – a model study.

(3) A third significant implication arising from Boger's re-examination of Kutney's chemistry is that the generation of the captodatively stabilised radical ((90), Scheme 16) is reversible which would account for the fact that in the course of the oxidation and in the absence of vindoline, catharanthine is often recovered from the reaction mixture. The persistence of catharanthine is hard to explain if the reaction proceeds via an azabenzfulvene intermediate (compare 24, Scheme 4 with (89) or (90) Scheme 16).

(4) Yet another important suggestion with stereochemical implications is the possibility of a “weak” two centre one electron bond between C(16) and C(21) of (90) (Scheme 16). Boger has suggested that this interaction may stabilize a conformation⁷ in which the more sterically accessible α-face directs approach by the vindoline partner producing the desired exclusive selectivity at the new C(16′) stereocentre in vinblastine. All these observations strengthen the Boger mechanistic interpretation. It would be interesting however to re-examine the question of the extent to which protonation of the tertiary amine in catharanthine inhibits conversion to the radical cation. The free energy of formation of the protonated amine is an important issue in a conformationally locked system in which an azabicyclo [2.2.2] octane is fused to an indoloazepine. In other words the protonation of the tertiary amine could be reversible. What would happen if we took this possibility (as yet unproven) into account and modified the second oxidation step in Kutney's original proposal (Scheme 4)?

We can see that for a reversible protonation of the tertiary amine, oxidation of free amine could lead to the Kutney radical cation (21). Fragmentation would give rise to the captodative radical⁶³ (22), but this is the same intermediate generated according to Boger's novel proposal. All that is required is that we dispense with Kutney's second Fe³⁺ promoted oxidation to the dicationic intermediate (23) in favour of Boger's late stage post coupling oxidation step ((92) to (93) Scheme 17). The reader would recall that Sundberg accounted for the rearrangement products obtained by FeCl₃ oxidation (in 0.1 M glycine solution) of the unnatural catharanthine [3,2] analogue (44) (with vindoline present) by the not unreasonable suggestion that tertiary amine radical cation formation had occurred (Scheme 10). The Ockham's razor principle notwithstanding an alternative mechanism involving indole C(2)–C(3) oxidation (if applied to (44) Scheme 10) cannot be completely ruled out in the Sundberg case.⁴⁵

Two questions remain to be addressed. (1) How do the mechanistic insights described above fit in to the wider scheme of Boger's versatile asymmetric syntheses of vinblastine and analogues? And (2), what are the pharmacological implications? We address these questions in the final section.

Conclusion and pharmacological developments

How would we answer the question, is the azabenzfulvene intermediate essential? The answer would appear to be no! since Boger's thoughtfully designed experiments lead to this conclusion. Whether Kutney's original tertiary amine oxidation must also be dispensed with is a more subtle issue that must wait further investigation. Although the intriguing behaviour of Sundberg's “unnatural” analogue ((44) in Scheme 10), and the possibility of reversible tertiary amine protonation, might appear to leave open the tertiary amine oxidation question, it must be acknowledged that Boger's design of simplified (non-quinuclidine incorporating) analogues does suggest that the tertiary amine in catharanthine is not required and the addition of tertiary amines to the reaction mixture e.g. N-methyl catharanthine and even quinuclidine have no impact on the course of the reaction.⁷ Boger's explorations, an outstanding example of creativity and systematic experimental design, have given us invaluable insights into the FeCl₃/catharanthine fragmentation/vindoline coupling at time when the chemistry of Fe based reagents has been the subject of a major review.¹² While not all mechanistic questions have been answered beyond all reasonable doubt, Boger's achievement has major implications for the science of organic synthesis (in particular total synthesis) at a time when its importance and its value to society is increasingly being questioned. In summarizing and bringing up to date the Boger chemistry, he exploited the Teoc protected aspartic acid (122)⁹ which was transformed into the tethered oxadiazole (123), chosen to ensure excellent facial control in the [3 + 2] cycloaddition step (123) to (124) Scheme 24, see also ((59) to (60) to (63), Scheme 11) and taking full advantage of the existing asymmetric centre located α to the oxadiazole amide nitrogen. It should be noted that the shortening of the dienophile tether to three atoms in (124), Scheme 24 (compared to the original four see (60) Scheme 11) was designed not only to enhance the reactivity of the dienophile⁹ but also to promote a more diastereoselective [4 + 2] cycloaddition that must precede the [3 + 2] cycloaddition step. This tactical combination of the [4 + 2] followed by the [3 + 2] cycloadditions (see Scheme 11), yielded the pentacyclic core densely and appropriately functionalised around the C ring. Although the oxadiazole tether in this synthesis, in contrast to the first generation approach described in Scheme 11, addresses effectively the stereochemical outcome of the 3 + 2 cycloaddition thereby setting up the correct B, C ring junction stereochemistry (Scheme 24), this meant that unnatural DE pyrrolidizine system was obtained (see (124) or (125)). It was therefore necessary for Boger to devise an effective ring expansion sequence to generate the desired D, E indolizidine system (e.g. 131). To this end, the γ-lactam (ring E) carbonyl was reductively excised i.e. (124) to (125). Cleavage of the C(8)–MOM-ether (125) under acidic conditions provided the primary alcohol (126) which was subsequently exposed to SO₃-pyridine/silica Et₃N to provide oxybridged intermediate (127) (obtained by trapping of the N⁺-C(7) iminium ion by the C(16) hydroxyl group), which was in turn readily converted to the tosylate (128).

Scheme 23 The Boger and “modified” Kutney hypotheses compared and contrasted.

Scheme 24 Boger's asymmetric synthesis of vindolin en route to vinblastine via a protected aspatic acid and an ingenious rearrangement.

Exposure of (128) to mild basic conditions yielded the ring expanded D,E indolizidinone (131) via oxybridge cleavage/imminium ion formation, trapping by water, ring opening/cleavage of the N–C(7) bond, and intramolecular S_N² displacement of the primary C(8) tosylate by the developing pyrrolidine nitrogen anion (130). Only a few more steps were needed to complete the asymmetric synthesis of vindoline. This arresting total synthetic achievement is even more impressive when it is combined with the biomimetic catharanthine fragmentation/vindoline coupling. We would again like to emphasise the combined effect of the Fe(III) promoted coupling of vindoline to catharanthine and the innovative Fe(III)-mediated hydrogen atom initiated free radical alkene oxidation for which there was no prior precedent, allows for stereoselective installation of the C(20′) alcohol in a single operation thus providing vinblastine^7,44 (Scheme 25).

Scheme 25 Boger's one pot Fe3⁺ catharanthine fragmentation/coupling/C (20) – oxidation.

In demonstrating the versatility of the above methodology and its implications for pre-clinical drug development, a large number of analogues were tested for activity against L1210 (ref. 44) (a mouse leukemia cell line) and HCT 116 (ref. 44) (a sensitive colon tumour cell line) These analogues were further tested against (HCT 116/VM46), a vinblastine resistant colon tumour cell line test that reveals the susceptibility of the analogues to multidrug resistance. A few examples are shown below (Scheme 26).

Scheme 26 Application of the oxadiazole [4 + 2]/[3 + 2] cycloaddition to the construction of vinblastine analogues with C(4)-acetoxy, C(4)-hydroxy and C(4)-deoxy moieties.

From this selection from Boger's extensive study, it is apparent that whereas loss of the 4-acetoxy group in (134) (compared to vinblastine) results in a tenfold loss of potency,⁴⁴ “hydrolysis” of the 4-acetoxy to yield the 4-hydroxy derivative (136) produces an analogue that is as potent as vinblastine. The results, as shown in Boger's⁴⁴ extended list of analogues, clearly indicate that the C(6)–C(7) double bond is critical. None of these compounds demonstrate superior activity against the resistant cell line (HCT116/VM46). Rather the opposite would appear to be true. When analogues with deep seated structural changes (Fig. 3), accessible by Boger's versatile methodology, were evaluated against HCT116,¹¹ the following results, of which only a selection are presented, are shown below. Notice in particular the changes in the DE ring when compared with vinblastine.

Fig. 3 A selection of vinblastin analogues with deep seated structural changes and prepared by total synthesis (IC₅₀) values are included which are based on the HCT/116 cell line.

As a further powerful demonstration of creativity brought to bear on the problem of drug resistance, Boger modified the C(20′) oxidation step critical to the vinblastine synthesis (Scheme 27) by inclusion of cesium azide in the reaction mixture.^44,64 This effectively installs stereospecifically the C(20′) azido function which underwent further transformations to yield a number of novel carbamate analogues. These analogues^11,52–54 were not only remarkably potent with respect to HCT116 but also with respect to the Pgp over expressing vinblastine resistant cell line HCT116/VM46 (Table 1).

Scheme 27 Creativity pre-clinically rewarded and vinblastin drug resistance overcome by virtue of C(20) azidation (see Table 1).

Table 1 Disubstituted C20′ urea derivatives of vinblastine^a

Compound	IC50(nM) HCT116	HCT116/VM46
a HCT116/VM46 (resistant human colon cancer cell line, Pgp overexpression).
Vinblastine (1)	6.8	600
R = pyrrolidine	0.72	50
R = thiomorpholine	0.88	50
R = tetrahydroisoquinoline	0.56	8.7
R = isoindoline	0.60	7.5
R = 5-MeO-isoindoline	0.69	8.7

---

Human colon cancer	IC50 (nM) vinblastine	146
HCT 116	6.8	0.16
HCT 116/VM46	600	3.5

---

Although it is often said that organic synthesis (in particular total synthesis) is a “mature’’ science, this belief is misconceived. The insightful comments of V. K Aggarwal and reflections on his own outstanding contributions to stereoselective synthesis (e.g. organoboron chemistry) in the same article, should be noted.⁶⁵ A captivating exchange on the merits of synthetic biology versus synthetic chemistry between a biologist (J. D Keasling) and two organic chemists (A. Mendoza and P. S. Baran) has also appeared.⁶⁶ For an interesting survey that relates synthetic creativity to scalability see the article “Natural Product Synthesis in the Age of Scalability” by P. S Baran et al.⁶⁷ Baran's own ingenious solutions to many problems encountered in several total syntheses are described in this survey.

We hope our modest effort in this review will contribute to the excitement of new discoveries (often the reward for risk taking) and a spirit of optimism about the future of organic synthesis.

Acknowledgements

The authors would like to thank University Malaya for the Postgraduate Research Grant (PPP)-Research PG178-2015A and University of Malaya and the Ministry of Higher Education (MOHE) Malaysia (HIR-005) for financial support.

References

1. N. Langlois, F. Gueritte, Y. Langlois and P. Potier, J. Am. Chem. Soc., 1976, 98, 7017–7024.
2. L. Kurti and B. Czakó, Strategic applications of named reactions in organic synthesis, Elsevier Academic press 356, 2005.
3. M. Polonovski and M. Polonovski, Bull. Soc. Chim. Fr., 1927, 41, 1190–1208.
4. M. Polonoski, Bull. Soc. Chim. Fr., 1930, 39, 1–39.
5. H. Morita and J. Kobayashi, J. Org. Chem., 2002, 67, 5378–5381.
6. A. E. G. J. Vukovic, J. P. Kutney and M. Misawa, Tetrahedron, 1988, 44, 325–331.
7. H. Gotoh, J. E. Sears, A. Eschenmoser and D. L. Boger, J. Am. Chem. Soc., 2012, 134, 13240–13243.
8. H. Ishikawa, D. A. Colby and D. L. Boger, J. Am. Chem. Soc., 2008, 130, 420–421.
9. Y. Sasaki, D. Kato and D. L. Boger, J. Am. Chem. Soc., 2010, 132, 13533–13544.
10. D. Kato, Y. Sasaki and D. L. Boger, J. Am. Chem. Soc., 2010, 132, 3685–3687.
11. J. E. Sears and D. L. Boger, Acc. Chem. Res., 2015, 48, 653–662.
12. I. Bauer and H.-J. Knölker, Chem. Rev., 2015, 115(9), 3170–3387.
13. D. Crich, Handbook of Reagents for Organic Synthesis, Reagents for Radical and Radical Ion Chemistry, John Wiley & Sons, 2013.
14. C. K. Prier, D. A. Rankic and D. W. MacMillan, Chem. Rev., 2013, 113, 5322–5363.
15. R. Noble, C. Beer and J. Cutts, Ann. N. Y. Acad. Sci., 1958, 76, 882–894.
16. G. H. Svoboda, N. Neuss and M. Gorman, J. Am. Pharm. Assoc., 1959, 48, 659–666.
17. T.-S. Kam and Y.-M. Choo, Alkaloids: Chemistry and Pharmacology, 2006, 63, 181–337.
18. G. BlaskÓ and G. A. Cordell, in The Alkaloids: Chemistry and Pharmacology, ed. B. Arnold and S. Matthew, Academic Press, San Diego, 1990, vol. 37, pp. 1–76.
19. J. P. Kutney, L. S. Choi, T. Honda, N. G. Lewis, T. Sato, K. L. Stuart and B. R. Worth, Helv. Chim. Acta, 1982, 65, 2088–2101.
20. J. P. Kutney, Nat. Prod. Rep., 1990, 7, 85–103.
21. K. Stuart, J. Kutney, T. Honda and B. Worth, Heterocycles, 1978, 9, 1419–1427.
22. The Alkaloids: Antitumor Bisindole Alkaloids from Catharanthus roseus (L.), ed. A. Brossi and M. Suffiness, Academic Press, San Diego, 1990, vol. 37.
23. P. M. Dewick, Medicinal Natural Products A Biosynthetic Approach, John Wiley & Sons, Chichester, New York, 3rd edn, 2009.
24. P. Mangeney, R. Zo Andriamialisoa, N. Langlois, Y. Langlois and P. Potier, J. Am. Chem. Soc., 1979, 101, 2243–2245.
25. P. Potier, J. Nat. Prod., 1980, 43, 72–86.
26. R. J. Sundberg, K. G. Gadamasetti and P. J. Hunt, Tetrahedron, 1992, 48, 277–296.
27. S. Yokoshima, T. Ueda, S. Kobayashi, A. Sato, T. Kuboyama, H. Tokuyama and T. Fukuyama, J. Am. Chem. Soc., 2002, 124, 2137–2139.
28. K. Nicolaou and S. A. Snyder, in Classics in Total Synthesis II: More Targets, Strategies, Wiley-VCH, Weinheim, Germany, 2003, vol. 2, pp. 505–530.
29. G. Schill, C. U. Priester, U. F. Windhövel and H. Fritz, Tetrahedron, 1987, 43, 3765–3786.
30. A. E. G. J. Vukovic, J. P. Kutney and M. Misawa, Tetrahedron, 1988, 44, 325–331.
31. P. Bather, J. L. Smith and R. Norman, J. Chem. Soc. C, 1971, 3060–3068.
32. C. Audeh and J. L. Smith, J. Chem. Soc. B, 1970, 1280–1285.
33. C. Audeh and J. L. Smith, J. Chem. Soc. B, 1971, 1745–1747.
34. J. Mulzer, Nat. Prod. Rep., 2014, 31, 595–603.
35. N. Thomas, K. Lee, T. Paraidathathu, J. Weber, K. Awang, D. Rondeau and P. Richomme, Tetrahedron, 2002, 58, 7201–7206.
36. N. F. Thomas, S. S. Velu, J.-F. F. Weber, K. Lee, A. H. A. Hadi, P. Richomme, D. Rondeau, I. Noorbatcha and K. Awang, Tetrahedron, 2004, 60, 11733–11742.
37. K. Ahmad, N. F. Thomas, M. R. Mukhtar, I. Noorbatcha, J.-F. F. Weber, M. A. Nafiah, S. S. Velu, K. Takeya, H. Morita and C.-G. Lim, Tetrahedron, 2009, 65, 1504–1516.
38. F.-J. Hong, Y.-Y. Low, K.-W. Chong, N. F. Thomas and T.-S. Kam, J. Org. Chem., 2014, 79, 4528–4543.
39. M. A. Ischay and T. P. Yoon, Eur. J. Org. Chem., 2012, 3359–3372.
40. G. Pandey, P. Y. Reddy and U. Bhalerao, Tetrahedron Lett., 1991, 32, 5147–5150.
41. S. Zhu, A. Das, L. Bui, H. Zhou, D. P. Curran and M. Rueping, J. Am. Chem. Soc., 2013, 135, 1823–1829.
42. G.-H. Tan, T.-M. Lim and T.-S. Kam, Tetrahedron Lett., 1995, 36, 1327–1330.
43. J. Santamaria, Pure Appl. Chem., 1995, 67, 141–147.
44. H. Ishikawa, D. A. Colby, S. Seto, P. Va, A. Tam, H. Kakei, T. J. Rayl, I. Hwang and D. L. Boger, J. Am. Chem. Soc., 2009, 131, 4904–4916.
45. R. J. Sundberg, J. Hong, S. Q. Smith, M. Sabat and I. Tabakovic, Tetrahedron, 1998, 54, 6259–6292.
46. H. Ishikawa, G. I. Elliott, J. Velcicky, Y. Choi and D. L. Boger, J. Am. Chem. Soc., 2006, 128, 10596–10612.
47. Z. Q. Yuan, H. Ishikawa and D. L. Boger, Org. Lett., 2005, 7, 741–744.
48. G. I. Elliott, J. R. Fuchs, B. S. Blagg, H. Ishikawa, H. Tao, Z.-Q. Yuan and D. L. Boger, J. Am. Chem. Soc., 2006, 128, 10589–10595.
49. G. D. Wilkie, G. I. Elliott, B. S. Blagg, S. E. Wolkenberg, D. R. Soenen, M. M. Miller, S. Pollack and D. L. Boger, J. Am. Chem. Soc., 2002, 124, 11292–11294.
50. D. L. Boger, Tetrahedron, 1983, 39, 2869–2939.
51. D. L. Boger, Chem. Rev., 1986, 86, 781–793.
52. E. K. Leggans, K. K. Duncan, T. J. Barker, K. D. Schleicher and D. L. Boger, J. Med. Chem., 2012, 56, 628–639.
53. T. J. Barker, K. K. Duncan, K. Otrubova and D. L. Boger, ACS Med. Chem. Lett., 2013, 4, 985–988.
54. K. D. Schleicher, Y. Sasaki, A. Tam, D. Kato, K. K. Duncan and D. L. Boger, J. Med. Chem., 2013, 56, 483–495.
55. J. Bergman, S. Bergman and J.-O. Lindström, Tetrahedron Lett., 1989, 30, 5337–5340.
56. A. Gieseler, E. Steckhan, O. Wiest and F. Knoch, J. Org. Chem., 1991, 56, 1405–1411.
57. S. Lin, M. A. Ischay, C. G. Fry and T. P. Yoon, J. Am. Chem. Soc., 2011, 133, 19350–19353.
58. S. L. Drew, A. L. Lawrence and M. S. Sherburn, Angew. Chem., Int. Ed., 2013, 52, 4221–4224.
59. B. Harirchian and N. L. Bauld, J. Am. Chem. Soc., 1989, 111, 1826–1828.
60. H. Ishikawa, H. Takayama and N. Aimi, Tetrahedron Lett., 2002, 43, 5637–5639.
61. N. F. Thomas, C.-H. Kee, A. Ariffin, K. Awang, J.-F. F. Weber, C.-G. Lim, M. R. Mukhtar and A. H. A. Hadi, Heterocycles, 2008, 75, 1097–1108.
62. C. H. Kee, A. Ariffin, K. Awang, I. Noorbatcha, K. Takeya, H. Morita, C. G. Lim and N. F. Thomas, Molecules, 2011, 16, 7267–7287.
63. H. G. Viehe, Z. Janousek, R. Merenyi and L. Stella, Accounts Chem. Res., 1985, 18, 148–154.
64. E. K. Leggans, T. J. Barker, K. K. Duncan and D. L. Boger, Org. Lett., 2012, 14, 1428–1431.
65. V. K. Aggarwal, Angew. Chem., Int. Ed., 2015, 54, 4150–4151.
66. J. D. Keasling, A. Mendoza and P. S. Baran, Nature, 2012, 492, 188–189.
67. C. A. Kuttruff, M. D. Eastgate and P. S. Baran, Nat. Prod. Rep., 2014, 31, 419–432.

---