|Home | About | Journals | Submit | Contact Us | Français|
A 1,2,3,4-tetrahydro-9a,4a-(iminoethano)-9H-carbazole (4) is a central structural feature of the Strychnos alkaloid minfiensine (1) and akuammiline alkaloids such as vincorine (5) and echitamine (6). A cascade catalytic asymmetric Heck-iminium cyclization was developed that rapidly provides 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazoles in high enantiomeric purity. Two sequences were developed for advancing 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazole 27 to (+)-minfiensine. In our first-generation approach, a reductive Heck cyclization was employed to form the fifth ring of (+)-minfiensine. In a second more concise total synthesis, an intramolecular palladium-catalyzed ketone enolate vinyl iodide coupling was employed to construct the final ring of (+)-minfiensine. This second-generation total synthesis of enantiopure (+)-minfiensine was accomplished in 6.5% overall yield and 15 steps from 1,2-cyclohexanedione and anisidine 13. A distinctive feature of this sequence is the use of palladium-catalyzed reactions to form all carbon–carbon bonds in the transformation of these simple precursors to (+)-minfiensine.
Minfiensine (1), a secoiridoid indole alkaloid, was isolated from the African plant Strychnos minfiensis by Massiot and co-workers in 1989.1 Its structure was assigned using 1H and 13C NMR spectroscopy (C–H and H–H COSY, NOESY), with long-range coupling of the C15 methine hydrogen to the C17 allylic methylene and C19 vinylic hydrogens being particularly diagnostic. Although a variety of Strychnos alkaloids having a molecular formula of C19H22N2O have been isolated (e.g., 1–3, Figure 1), the structure of minfiensine has no precedent.2 No details of its biosynthesis are known, but its genesis from a corynantheine precursor is likely.
The 1,2,3,4-tetrahydro-9a,4a-(iminoethano)-9H-carbazole (4) moiety is the defining structural feature of minfiensine (1) (Figure 2). Although rare in Strychnos alkaloids, this tetracyclic ring system is found in a number of akuammiline alkaloids such as vincorine (5) and echitamine (6).3 Extracts containing akuammiline alkaloids are used throughout the world in the practice of traditional medicine, with some heightened interest in the medicinal potential of pure akuammiline alkaloids being seen in recent years.4 Dihydro-9a,4a-(iminoethano)-9H-carbazoles (4 having a 1,2- or 2,3-double bond) could serve as versatile platforms for constructing alkaloids of the types illustrated in Figure 2. Stitching an ethylideneethano unit between the pyrrolidine nitrogen and C3 of such a precursor would generate the ring system of minfiensine, whereas inserting such a unit between the pyrrolidine nitrogen and C2 would generate the ring system of vincorine and congeners.
In this article, we report the development of an efficient catalytic enantioselective synthesis of 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazoles by combining a palladium-catalyzed asymmetric Heck cyclization with an intramolecular iminium ion cyclization. We also describe the details of our use of such an intermediate to complete the first total synthesis of minfiensine (1)5 and our development of an improved end play that allows a notably efficient and concise catalytic asymmetric construction of this structurally unique Strychnos alkaloid to be realized.
We envisaged minfiensine (1), as well as alkaloids of the vincorine family, to be accessible from 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazole 7 (Scheme 1). The central transformation in our total synthesis plan is catalytic enantioselective formation of tetracyclic intermediate 7 from cyclohexadienyl carbamate 10. In this pivotal conversion, an intramolecular catalytic asymmetric Heck reaction would generate dihydrocarbazole 9 and form the critical quaternary carbon stereocenter.6,7 Selective in situ protonation of the enecarbamate double bond of diene product 9 would generate N-acyloxyiminium ion 8, which was expected to be trapped by the tethered amine to form dihydro(iminoethano)carbazole 7. One attractive feature of this sequence was the expected ready construction of cyclohexadienyl carbamate 10 from aniline 11 and 1,2-cyclohexanedione (12).
The sequence developed to prepare the cyclohexadienyl aryl triflate cyclization precursor 21 is summarized in Scheme 2. Although acid-catalyzed condensation of 1,2-cyclohexanedione (12) with anisidine 138 was low-yielding under all the conditions that we examined, the related condensation with morpholine enamine 149 in the presence of 1 equiv of p-toluenesulfonic acid (p-TsOH) in warm benzene provided 2-anilino-2-cyclohexenone 15 in high yield.10 Attempts to protect the nitrogen of this intermediate prior to elaboration to the cyclohexadiene were complicated by competitive reaction of the ketone. For example, deprotonation of intermediate 15 with a strong base (lithium, sodium or potassium bis(trimethylsilyl)amide, or NaH), followed by addition of ClCO2Me, yielded mixtures of carbamate 16 and products resulting from reaction of the ketone enolate on oxygen or carbon.5 We eventually discovered that the aniline nitrogen could be selectively masked by reaction of the lithium salt of aniline 15 with Mander’s reagent.11 The optimum condition identified involved dropwise addition of 3 equiv of lithium bis(trimethylsilyl)amide (LHMDS) to a THF solution of aminocyclohexenone 15 and excess methyl cyanoformate at −78 °C, in which case crystalline carbamate 16 was formed in 89% yield. When 1.5 or 2.0 equiv of LHMDS was used, significant amounts of cyclohexenone 15 were recovered.12 Addition of sodium bis(trimethylsilyl)amide (NaHMDS) to a THF solution of carbamate enone 16 and the Comins’ triflating reagent 17 (5-chloro-2-[N,N-bis(trifluoromethylsulfonyl)amino]pyridine)13 at −78 °C provided cyclohexadienyl triflate 18 in 82% yield. A β-aminoethyl side chain was then introduced in 71% yield by Suzuki cross-coupling of triflate 18 with the alkylborane generated in situ by hydroboration of enecarbamate 1914 with 9-borabicyclo[3.3.1]nonane (9-BBN).15 Finally, silyl-protected phenol 20 was transformed in one step and 85–95% yield to dienyl aryl triflate 21 upon reaction at room temperature with triflating reagent 17 in DMF in the presence of CsF and Cs2CO3.16 In order to obtain reproducibly high yields in this transformation, it was critical to use freshly dried cesium salts. Comins’ reagent was used rather than N-phenyltriflamide (PhNTf2) to ensure a rapid triflation of the cesium phenoxide and to facilitate purification of the aryl triflate product. Without these precautions, the triflation reaction was sluggish and irreproducible, with cyclic carbamate 22 being occasionally formed as a major byproduct. By the sequence summarized in Scheme 2, Heck cyclization precursor 21 was prepared in five steps and 46% overall yield from aniline 13 and 2-morpholino-2-cyclohexenone (14).
The asymmetric Heck cyclization of dienyl aryl triflate 21 was initially examined using Pd(OAc)2/BINAP as the precatalyst (Scheme 3). Under these conditions, intramolecular Heck reaction of 21 proceeded readily at 80 °C in acetonitrile (or toluene) in the presence of 1,2,2,6,6-pentamethylpiperidine (PMP) to give dihydrocarbazole 23 (ca. 60% yield).17 The conjugated cyclohexadiene product is undoubtedly produced by Pd–H-mediated isomerization of the initially formed Heck product (24 → 23). Use of a large excess of PMP or the stronger base 1,8-bis(dimethylamino)naphthalene, which has been reported to minimize double-bond migration,18 did not prevent the formation of the 1,3-cyclohexadiene product.
Chiral (phosphinoaryl)oxazolines have been shown to be effective ligands for asymmetric Heck reactions and to exhibit a low propensity for promoting double-bond migration.19 The Pd[(phosphinoaryl)oxazoline]-catalyzed Heck cyclization of dienyl aryl triflate 21 was initially explored using the commercially available isopropyl-substituted PHOX ligand 26a (Scheme 4). The Heck reaction of triflate 21 using Pd(OAc)2 (20 mol %), ligand 26a, and PMP in toluene provided the cross-conjugated dihydrocarbazole 25 in 79% yield (92% yield based on consumed 21) and 88% ee after heating at 100 °C for 70 h.20 The conjugated alkene isomer 23 was not observed under these reaction conditions. Improved enantioselectivity (96% ee) was achieved when the reaction was conducted in acetonitrile at 85 °C; however, alkene isomerization occurred to a limited extent (ca. 10%) in this more polar solvent. Optimal results were obtained using the tert-butyl-substituted PHOX ligand 26b,19a in which case dihydrocarbazole 25 was produced in 75–87% yield and 99% ee; dihydrocarbazole 23, the product of double-bond migration, was not detected. The only disadvantage of the palladium–phosphinooxazoline catalyst is the long reaction time (ca. 70 h). In order to increase the rate of this transformation, we investigated the effects of microwave heating.21 Larhed, Hallberg, and co-workers reported significant acceleration of intermolecular asymmetric Heck reactions using microwave heating, although enantioselectivities were reduced somewhat.22 Under microwave heating, the Heck reaction of 21 was complete in only 30–45 min at 170 °C, with no erosion of yield or enantiomeric purity of dihydrocarbazole 25 being observed (87% yield, 99% ee). This microwave-promoted Heck cyclization was carried out reproducibly on synthetically useful scales (ca. 500–600 mg). Furthermore, the shorter reaction times allowed the catalyst loading to be reduced to 10 mol % of Pd.
The tandem reaction sequence to form 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazole 27 was achieved in 75% overall yield from dienyl aryl triflate 21 by adding excess trifluoroacetic acid (TFA, 10 equiv) to the crude Heck reaction product after the reaction vessel was removed from the microwave reactor and cooled to 0 °C. A large excess of TFA was required in this one-pot sequence because excess PMP and the basic phosphine ligand were both present in the crude Heck reaction product. The 4aR,9aR absolute configuration of dihydro(iminoethano)-carbazole 27 was not established at this point, but at a later stage of the synthesis (vide infra).
From the outset, we envisaged constructing minfiensine (1) from a 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazole intermediate such as 27 by the sequence outlined retrosynthetically in Scheme 5. The central step in this elaboration would be the formation of pentacyclic ester 28 from tetracyclic precursor 30 by a Heck cyclization–carbon monoxide insertion sequence.6d Assembling the cyclization precursor 30 with the α-orientation of the allylic oxygen substituent was anticipated to favor the desired cascade reaction sequence by avoiding the possibility that pentacyclic palladium alkyl intermediate 29 could decompose by syn-β-hydride elimination. Either ketone 31 or α-epoxide 32 were seen as plausible precursors of dienyl iodide intermediate 30.
If dihydro(iminoethano)carbazole 27 could be epoxidized stereoselectively from the α-face, a concise route to Heck cyclization precursor 30 could likely be developed. For this reason, we examined this possibility even though a preliminary examination of molecular models of tetracyclic alkene 27 did not suggest a significant steric bias for approach of an oxidant from either alkene face. To our delight, standard epoxidation of alkene 27 with m-chloroperoxybenzoic acid (m-CPBA) proceeded with 10:1 stereoselectivity to give, after separation on silica gel, the α-epoxide 34 in 87% yield; the β-epoxide stereoisomer was also isolated in 9% yield (Scheme 6). The relative configuration of epoxide 34 was assigned on the basis of analogy to the major (8:1) epoxide stereoisomer 35 produced by m-CPBA epoxidation of methoxy congener 33. In this latter case, the epoxide was obtained as single crystals, allowing the relative configuration of epoxide 35 to be secured by X-ray analysis.23
To examine what was responsible for the 10:1 selectivity observed in the epoxidation of tetracyclic alkene 27, this intermediate was modeled computationally. This study showed a 1.2–2.5 kcal/mol preference for the cyclohexene ring of 27 to exist in the half-chair conformation illustrated in Figure 3.24 Assuming for steric reasons that C–O bond formation is more advanced at alkene carbon b, torsional effects would favor epoxidation from the alkene stereoface anti to the indoline bridge, as depicted in Figure 3.25
Without success, we explored initially the possibility of converting tetracyclic epoxide 34 to allylic alcohol 36 using lithium amide bases at various temperatures (Scheme 6).26 For example, reaction of epoxide 34 with lithium diethylamide in THF at −40 °C led to rapid loss of the methyl carbamate group. Cleavage of the methyl carbamate group occurred more slowly when lithium diisopropylamide (LDA) was used as a base; however, an allylic alcohol product was never isolated, even after reactions conducted in THF were heated at 45 °C for several hours.
To avoid base-promoted side reactions, we turned to the organoselenide method developed by Sharpless and Lauer.27 As expected, epoxide 34 reacted regioselectively with sodium phenylselenide to generate a single β-hydroxy selenide product. Oxidation of this intermediate to the selenoxide, followed by heating at 60 °C, delivered allylic alcohol 36 in 50% yield from epoxide precursor 34. Protection of alcohol 36 as a benzyl ether proceeded uneventfully under standard conditions to give tetracyclic allylic ether 37.
To our surprise, attempts to cleave the Boc group of intermediate 37 under standard acidic conditions [trifluoroacetic acid (TFA), CH2Cl2, 0 °C] led to the formation of indole 38. This fragmentation was rapid, occurring instantaneously at 0 °C to provide indole aldehyde 38 in 92% yield. 1H and COSY NMR spectra confirmed the presence of an unsaturated aldehyde, with a large vinylic coupling constant (15.7 Hz) indicating the E geometry for the 4-oxo-2-butenyl side chain. Because of the facility of this rearrangement under acidic conditions, other methods were briefly investigated for removing the Boc group. Attempted thermolytic fragmentation upon conventional heating in dimethylsulfoxide (DMSO, 120–150 °C) or in a microwave reactor did not afford the desired secondary amine product. Likewise, attempted deprotection of 37 with ceric ammonium nitrate (CAN)28 provided only recovered starting material and indole aldehyde 38.
A plausible sequence for formation of indole 38 from tetrahydro(iminoethano)carbozle precursor 37 is suggested in Scheme 7. In the presence of acid, aminal 37 should be in equilibrium with tricyclic N-acyloxyiminium ion 39.29 In a process that undoubtedly would be exothermic, fragmentation of bond a of this intermediate would generate indole oxocarbenium ion 40. This latter intermediate would be expected to be trapped with trifluoroacetate to form allylic trifluoroacetate 41 and its allylic isomers. Upon aqueous workup, 41 and/or its E allylic isomer would deliver indole 38.
The sequence that was ultimately developed for advancing epoxide 34 to Heck cyclization precursor 48 is summarized in Scheme 8. As the alkoxycarbenium ion potentially generated by fragmentation of the cyclohexane ring of epoxide 34 would be less stable than that produced from allylic ether 37 (Scheme 7), exchange of the Boc group of epoxy carbamate 34 for an allyloxycarbonyl (Alloc) protecting group was straightforward. Rearrangement of the epoxide functionality of Alloc product 43 and triethylsilyl (TES) protection of the resultant allylic alcohol provided allylic silyl ether 45 in 68% overall yield from Boc-protected epoxide 34. Removal of the Alloc group of intermediate 45 using Pd(PPh3)4 and excess pyrrolidine proceeded cleanly to yield tetracyclic amine 46. The Heck cyclization precursor 48 was then secured in 96% yield by reaction of secondary amine 46 with 2 equiv of allylic tosylate 4730 in hot MeCN in the presence of K2CO3.
We explored briefly a second sequence for elaborating dihydro(iminoethano)carbazole 27 to an appropriate precursor for forming the fifth ring of minfiensine (Scheme 9). Toward that end, tetracyclic alkene 27 was allowed to react with borane–methyl sulfide complex at room temperature, followed by oxidation with tetrapropylammonium perruthenate/N-methylmorpholine-N-oxide (TPAP/NMO) to provide a chromatographically separable mixture of ketones 49 (63% yield) and 50 (21% yield). As has been reported several times previously in hydroborations of structurally related alkenes, reaction of tetracyclic alkene 27 with BH3 placed boron preferentially at the more hindered neopentylic position.31,32 Although the minor ketone isomer was not of immediate interest (vide infra), it did allow the absolute configuration of the product 27 of tandem asymmetric Heck–iminium ion cyclization to finally be established. Thus, reduction of ketone 50 with sodium borohydride provided a 1:1 mixture of separable alcohol products. Reaction of the lower Rf epimer with p-bromobenzoyl chloride yielded crystalline ester 51. Single-crystal X-ray analysis of this derivative allowed its relative and absolute configuration to be determined,33 proving the sense of stereoinduction in the asymmetric Heck cyclization of 21 to give (R)-dihydrocarbazole 25 (Scheme 4).
In an attempt to process the major ketone isomer to a suitable precursor for forming the final ring of minfiensine, tetracyclic ketone 49 was dehydrogenated by reaction with 2-iodoxybenzoic acid (IBX) (Scheme 9).34 Subsequent cleavage of the Boc group with TFA provided cyclohexenone 53 in 40% yield for the two steps. To our surprise, attempted N-allylation of secondary amine 53 with allylic tosylate 4730 yielded dienone 54 as the major product. The formation of this dihydroindolone likely arises from deprotonation of intermediate 53 at the γ-position to give a dienolate anion, which β-eliminates the aniline carbamate conjugate base.35 As an alternate route to an appropriate precursor for closing the final ring of minfiensine had been developed (see Scheme 8), this strategy was pursued no further.
We turned to examine forming the final ring of minfiensine and simultaneously installing the remaining carbon atom by the Heck cyclization–carbonylation sequence posited in Scheme 5. To this end, tetracyclic vinyl iodide 48 was subjected to a number of standard conditions that had been used by others to carry out cascade intramolecular Heck reaction–carbonylation reactions (Scheme 10).7d A variety of palladium precatalysts [Pd(OAc)2, PdCl2(PPh3)2, Pd(PPh3)4], ligands [Ph3P, 1,3-bis-(diphenylphosphoryl)propane, and 1,4-bis(diphenylphosphoryl)butane], bases (triethylamine, K2CO3), cosolvents (DMF, MeCN, toluene), and CO pressures (1 atm to 150 psi) were explored in various combinations. However, under no reaction condition that we examined was the desired pentacyclic ester 55 detected. Besides recovered dienyl iodide 48, secondary amine 46,36 the α,β-unsaturated ester 56 resulting from carbonylation of the vinyl iodide side chain, and a pentacyclic product 57, whose structure was revealed after additional experimentation (vide infra), were isolated from these experiments.
Our inability to accomplish a cascade Heck cyclization–carbonylation reaction prompted us to examine the pivotal ring-forming step under reductive conditions.7d In this context, the intramolecular Heck reaction of vinyl iodide 48 was carried out in the presence of sodium formate with the expectation that pentacycle 58 would result (Scheme 11). However, reaction of alkenyl iodide 48 with 10 mol % of Pd(OAc)2, 1,4-bis(diphenylphosphoryl) butane (dppb), Et3N, and NaO2CH in acetonitrile did not produce pentacyclic product 58, but rather pentacyclic isomer 57 and the product 46 of N-dealkylation. The presence of sodium formate had no apparent effect on this reaction, with pentacyclic allylic alcohol 57 being isolated in 48% yield in a reaction carried out in the absence of sodium formate and triethylamine. Removal of the TES group from pentacyclic product 57 provided the crystalline alcohol 59, whose structure was established by single-crystal X-ray analysis. Remarkably, pentacyclic product 57 produced by Pd(0)-catalyzed cyclization of tetracyclic dienyl iodide 48 is the product of formal C–H insertion of the alkenylpalladium side chain at the allylic carbon of the cyclohexene ring.
Insertion of an alkenylpalladium species into a C–H σ-bond is not expected under Heck reaction conditions.7d,37 We surmised that the formation of product 57 could result from the lability of the aminal functionality of the pyrrolidinoindoline moeity.30 This lability provides a straightforward pathway for migration of the cyclohexene double bond (Scheme 12). Thus, ring opening of tetracyclic aminal 48 to give iminium ion 60, likely promoted by some protic or Lewis acidic species (illustrated by H+ in Scheme 12), could readily lead to the formation of dienamine 61. Protonation of this intermediate at the terminal carbon would provide conjugated iminium cation 62, which upon ring closure would deliver isomer 63 of the original Heck cyclization substrate. Standard 5-exo-Heck cyclization of this intemediate would lead directly to the observed tetracyclic product 57.
To gain experimental evidence for the proposed double-bond migration, a series of NMR studies were carried out with cyclization substrate 48. First, we examined whether the alkene isomerization would take place in the absence of a palladium species. Accordingly, dienyl iodide 48 and dppb (20 mol %) were heated in d3-acetonitrile at 75 °C. After 12 h, ca. 25% of diene 48 had isomerized to isomer 63.38 This experiment showed that double-bond migration could take place in the absence of palladium. However, the rate of this isomerization was slower than the rate at which pentacyclic product 57 was formed during the Heck reaction. Accordingly, it was reasoned that the double-bond migration was proceeding by an additional pathway. As a Pd(II) species could be present and these are well-known to promote alkene isomerization,39 dienyl iodide 48 was exposed to 12 mol % of PdCl2(MeCN)2 in d3-acetonitrile at 75 °C. After 40 min, isomerization to diene isomer 63 was ca. 80% complete. As the enoxysilane double-bond isomer was not observed, we believe that a Pd(II) species formed in situ is functioning as a Lewis acid to activate the aminal functionality toward ring cleavage and facilitate double-bond migration in the conversion of 48 → 57 (Scheme 11).
High catalysis rates are often observed with the phosphine-free Heck reaction conditions initially reported by Jeffery (inorganic bases and tetraalkylammonium halides).40,41 In the hope that Heck cyclization might be faster than double-bond isomerization, we turned to examine the reaction of dienyl iodide 48 under these conditions.42 After some optimization, we found that reductive Heck cyclization of dienyl iodide 48 to form pentacyclic product 58 (80% yield) took place in DMF within 30 min at 80 °C in the presence of 1 mol % of Pd(OAc)2, K2CO3 (5 equiv), (n-Bu)4NCl (2.5 equiv), and NaO2CH (1.2 equiv). Under these conditions, pentacyclic isomer 57 was not detected, and the secondary amine 46 resulting from deallylation was produced to only a limited extent (ca. 10% yield). After removal of the TES group from Heck product 58, single-crystal X-ray analysis of alcohol 64 confirmed that the minfiensine ring system had been assembled (Scheme 13).
To advance tetracyclic alcohol 64 to minfiensine (1) required that a double bond and a one-carbon side chain be installed in the cyclohexane ring. These elaborations were accomplished as follows (Scheme 14). Oxidation of alcohol 64 with Dess–Martin periodinane43 provided ketone 65. The lithium enolate of this ketone upon reaction with methyl cyanoformate11 gave β-keto ester 66, which existed nearly exclusively as the enol tautomer, in 70% yield over the two steps. Upon reaction with excess NaBH4 in THF/MeOH at 0 °C, β-keto ester 66 was converted to largely one β-hydroxy ester product 67; the relative configuration of the alcohol and ester substituents of this intermediate was not determined, as these stereocenters are cleared in the subsequent step. After more traditional methods to dehydrate β-hydroxy ester 67 to α,β-unsaturated ester 68, such as reaction with methanesulfonyl chloride or triflic anhydride and triethylamine, met with failure, this conversion was accomplished by a two-step sequence. The sterically hindered secondary alcohol was first activated by reaction with benzoyl triflate44 and pyridine in CH2Cl2 at 60 °C (sealed tube). Exposure of the resulting benzoate to 2 equiv of potassium bis(trimethylsilyl)amide (KHMDS) in THF at −78 °C then delivered enoate 68 in 83% yield for the two steps. Lithium aluminum hydride reduction of α,β-unsaturated ester 68 provided allylic alcohol derivative 69, which upon reaction with NaOH in MeOH/H2O at 100 °C gave (+)-minfiensine (1) in 85% yield for the two steps. Synthetic (+)-minfiensine (1) displayed 1H and 13C NMR spectral data identical to that reported for the natural product.1 The optical rotation of synthetic (+)-minfiensine (1) was [α]D +125 (c 0.82, CHCl3), which compared well to the rotation of +134 reported under identical conditions for the natural isolate.1
In our first-generation total synthesis, facile ring opening of the aminal linkage of the pyrrolidinoindoline fragment (see Scheme 6) significantly hindered elaboration of (iminoethano)carbazole 37 to (+)-minfiensine. To side step these obstacles, a second-generation approach should obviate the possibility of double-bond migration during formation of the final ring and minimize the likelihood of fragmentation of the 3,4-dihydro-9a,4a-(iminoethano)-9H-carbazole ring system by avoiding intermediates having an electron-donating heteroatom substituent adjacent to the quaternary benzylic carbon. Constructing the final ring by an intramolecular Pd(0)-catalyzed coupling of a ketone enolate and a vinyl iodide side chain, as posited in Scheme 15 (70 → 71→ 72), should accomplish these objectives. To our knowledge, this ring-forming tactic was first used by Piers in the total synthesis of triquinanes.45 In recent years, the use of palladium-catalyzed intramolecular α-vinylation of ketone enolates to assemble complex alkaloid ring systems has been demonstrated by insightful studies from the Cook46 and Bonjoch47 laboratories.48 An additional attractive feature of this strategy was the anticipated ease with which ketone 72 could be elaborated to (+)-minfiensine because this ketone can enolize in only one direction; enolization at the allylic α-carbon would generate a high-energy, anti-Bredt enolate.
Elaboration of the cascade asymmetric Heck/iminium ion cyclization product 27 to tetracyclic ketone vinyl iodide 70 is summarized in Scheme 16. As discussed earlier, hydroboration/oxidation of the double bond of tetracyclic intermediate 27 with BH3 preferentially functionalizes the neopentylic carbon to provide ketone 49 after oxidation (see Scheme 9). Accordingly, we examined hydroboration of this alkene with the sterically larger reagents catecholborane, thexylborane, and 9-borabicyclo-3.3.1]nonane (9-BBN) (Scheme 16). Neither catecholborane nor thexylborane proved sufficiently reactive, returning only unchanged starting material. Similar results were obtained with 9-BBN in refluxing THF. However, hydroboration with 9-BBN did take place rapidly in THF at 100 °C in a microwave reactor. After oxidation of the crude product with TPAP/NMO, ketones 50 and 49 were isolated in a 2.5:1 ratio and a combined yield of 88%. Although regioselection of this hydroboration was only modest, ketone 50 was obtained in 63% yield by this two-step sequence. The Boc-protecting group was readily removed by allowing ketone 50 to react with excess TFA at room temperature. Alkylation of the resulting secondary amine with allylic bromide 7343 then provided the ketone vinyl iodide cyclization substrate 70 in 65% yield for the two steps.
With tetracyclic iodoketone 70 in hand, we turned to forming the final ring of (+)-minfiensine by palladium-catalyzed intramolecular enolate/vinyl iodide coupling (Scheme 17). Exposure of this iodoketone to catalytic Pd(PPh3)4 and KOt-Bu (1.5–4 equiv) in various solvents (THF, DMF, DMF/t-BuOH)46 gave none of the pentacyclic product 72. Elimination of HI to form the corresponding tetracyclic propargylic amine was the major reaction under these conditions, a competing pathway identified by Piers in his seminal report.45a Cook employed weaker bases, typically K2CO3, to realize palladium-catalyzed enolate α-vinylation reactions in the total synthesis of a number of alkaloids.47 In our initial examination of these conditions, ketone 70 was exposed to 10 mol % of PdCl2(dppf)2 •CH2Cl2 and excess K2CO3 in either DMF or MeCN [dppf = 1,1′-bis(diphenylphosphino)ferrocene]. Alkyne formation was suppressed; however, the only product detected, 74, under these conditions arose from loss of the (Z)-2-iodo-2-butenyl side chain.46c Deallylation was also the major pathway when the identical reaction was carried out in 9:1 DMF/water; nonetheless, in this case, pentacyclic ketone 72 was formed in 30–35% yield. A similar low yield of pentacyclic product 72 was realized using Pd(OAc)2/Ph3P instead of PdCl2(dppf). Increasing the percentage of water in the solvent from 10 to 50% (using either DMF or THF as the cosolvent) had no noticeable effect on the outcome of the reaction; competitive formation of secondary amine 74 remained problematic. When performing the reaction in 1:1 MeOH/water using PdCl2(dppf) and K2CO3, the yield of pentacyclic ketone 72 increased to 45–50%, and the formation of the deallylation byproduct 74 decreased to trace levels.49 The most significant improvement in yield was registered when the reaction was conducted in MeOH using 10 mol % of PdCl2(dppf) and 4 equiv of K2CO3 at 70 °C. Under these conditions, starting material was consumed within 1 h and tetracyclic ketone 72 was isolated in 74% yield.
Tetracyclic ketone 72 is ideally constituted for direct elaboration to (+)-minfiensine (Scheme 17). Straightforward conversion of ketone 72 to the corresponding enol triflate 75, followed by palladium-catalyzed carbonylation of this intermediate in MeOH, provided α,β-unsaturated ester 68 in 77% yield for the two steps. This product, [α]D − 124 (c 1.05, CHCl3), displayed spectroscopic properties identical to those of tetracyclic enoate 68 prepared during our first-generation synthesis. As described in more detail in Scheme 14, α,β-unsaturated ester 68 can be converted in two steps and 85% overall yield to (+)-minfiensine, [α]D +125 (c 0.82, CHCl3).
A catalytic asymmetric Heck–iminium ion cyclization sequence was developed that provides either enantiomer of differentially protected 1,2,3,4-tetrahydro-9a,4a-(iminoethano)-9H-carbazole 27 in six steps from 1,2-cyclohexanedione (32% overall yield). Notable in this sequence is the use of microwave heating to accelerate the catalytic asymmetric Heck cyclization of dienyl aryl triflate 21, which even though the reaction temperature is 170 °C provides dihydrocarbazole 25 in 99% ee (Scheme 4).
Two sequences were developed for advancing (iminoethano)-9H-carbazole 27 to (+)-minfiensine. In a first-generation approach, a reductive Heck cyclization was employed to form the fifth ring of (+)-minfiensine. Side reactions resulting from facile ring opening of the pyrrolidinoindoline subunit and our inability to carry out a cascade Heck cyclization–carbonylation sequence combined to compromise the efficiency of this inaugural sequence. A second-generation synthesis side stepped these issues by employing an intramolecular palladium-catalyzed ketone enolate vinyl iodide coupling (70 → 72, Scheme 17) to construct the final ring of (+)-minfiensine. Using this approach, the cyclization precursor 70 was assembled from the product 27 of cascade asymmetric Heck–iminium ion cyclization in four steps rather than the seven steps required to prepare the related intermediate 48 in our first-generation synthesis. In addition, the functionality present in pentacyclic product 72 produced by enolate α-vinylation allowed the final elaboration to (+)- minfiensine to be shortened to four transformations. This secondgeneration total synthesis of enantiopure (+)-minfiensine was accomplished in 6.5% overall yield and 15 steps from 1,2-cyclohexanedione (12) and aniline 13. A distinctive feature of this concise sequence is the use of palladium-catalyzed reactions to form all the carbon–carbon bonds in the transformation of these simple precursors to (+)-minfiensine.
Financial support from the NIH NIGMS (GM-30859), and postdoctoral fellowship support for A.B.D. (CA94471) and A.D.W. (CA108197) from the NIH National Cancer Institute, and P.G.H. from Merck-Europe is gratefully acknowledged. We thank Dr. Joseph Ziller for X-ray analysis of compounds 35, 51, 54, and 64, and Professor Georges Massiot for copies of NMR spectra of natural (+)-minfiensine.
Supporting Information Available: Experimental procedures, spectroscopic and analytical data (1H, 13C, and HPLC) for new compounds, and CIF files of crystallographic information. This material is available free of charge via the Internet at http://pubs.acs.org.