Concise Total Synthesis of Lundurines A-C Enabled by Gold Catalysis and a Homodienyl Retro-Ene/Ene Isomerization.

The total synthesis of lundurines A-C has been accomplished in racemic and enantiopure forms in 11-13 and 12-14 steps, respectively, without protection/deprotection of functional groups, by a novel tandem double condensation/Claisen rearrangement, a gold(I)-catalyzed alkyne hydroarylation, a cyclopropanation via formal [3 + 2] cycloaddition/nitrogen extrusion, and a remarkable olefin migration through a vinylcyclopropane retro-ene/ene reaction that streamlines the endgame.

Lundurines A (1), B (2), and C (3) were isolated from Kopsia tenuis,[1] a plant endemic to the north of Borneo, and show interesting cytotoxicity.[2] These alkaloids feature a unique indoline-fused polyhydropyrroloazocine and cyclopropyl moiety fused to the indoline (Figure ). Related alkaloids lacking the cyclopropane ring, such as lapidilectam, lapidilectines, grandilodines, and tenuisines, have also been isolated from plants of the Kopsia genus.[3]

Figure 1

Lundurines A–C and related alkaloids.

The lundurines have recently attracted considerable attention,[4] and the total syntheses of lundurine A and lundurine B have been reported.[4b−4f] However, all previous approaches were lengthy, involving over 20 linear synthetic steps, thus making the synthesis of large quantities of the natural products and/or analogues, for broad biological assays, inconvenient.[4b−4f] We now report the expedient total synthesis of the three members of the lundurine family, including the first total synthesis of racemic and enantiopure lundurine C, by constructing the key lactam intermediate 4 in a single step by a condensation/Claisen rearrangement followed by a gold(I)-catalyzed intramolecular hydroarylation to form the 8-membered ring[5] (Scheme ). In our initial plan, we expected that a transition-metal-catalyzed reaction of a carbene precursor of type I would lead to intramolecular cyclopropanation of the indole nucleus. However, a more effective solution was found using an acid-catalyzed pyrazoline formation. The endgame relied on an unexpectedly facile vinylcyclopropane retro-ene/ene reaction that led to alkene migration, streamlining the culmination of the synthesis.

Scheme 1

Retrosynthetic Approach

For the synthesis of key chiral intermediate 5, we envisioned condensing oxoester 6 with commercially available 5-methoxytryptamine. This should lead to imine 7, which should undergo lactamization to form pyrrolidinones 8- and 8- Ultimately, 8- and 8- could afford 5 through a Claisen rearrangement.[6] For the enantioselective synthesis of 4, we proposed building the C20 stereocenter by enantiodiscrimination through transfer of chirality in the Claisen rearrangement (Scheme ).

Scheme 2

Enantioselective Claisen Rearrangement

Examples of efficient transfer of chirality on flexible systems featuring a “traceless” chiral auxiliary on the allyl fragment, and in the absence of a Lewis acid, are scarce.[7] Nonetheless, we prepared a range of (S)-chiral alcohols by enzymatic resolution of the racemic allylic alcohols,[8] which were converted into the desired chiral oxoesters 6 in a single step. The best transfer of chirality was achieved with R = c-pentyl (89:11 er). It is important to note that the use of basic conditions turned out to be essential to avoid the Pictet–Spengler type reaction that would form tetrahydro β-carbolines.[9] Thus, mixtures of pyridine/toluene or Et3N/toluene proved to be optimal, affording high yields of lactam (R = H: 5a 74%; R = c-pentyl, 5f 84%) (Scheme ).

Scheme 3

Synthesis of (+)-9f and Hydrazone (+)-12

CYLview depiction of the X-ray crystal structure of (+)-12.

Initially, we had expected that the system would be under Curtin–Hammett conditions, as a result of a fast equilibrium between 8- and 8-. However, in a closely related model system, we isolated the E- and Z-pyrrolidinones (2.6:1 ratio), which did not undergo equilibration after being heated at 100 °C in 1:2 toluene–Et3N for several hours.[10] Presumably, the major 8- pyrrolidinone reacts preferentially through a boat-like transition state TS to form (S)-5, whereas the minor isomer 8- reacts through TS (Scheme ).[11]

Aldehyde 5 was immediately homologated into the corresponding alkyne 4 employing the Ohira–Bestmann reagent (4a 88%; 4f 84%, 89:11 er), setting the stage for the key 8-endo-dig gold(I)-catalyzed hydroarylation (Scheme ). This was accomplished with perfect 8-endo selectivity with 5 mol % AuCl (9a 83%; 9f 79%, 89:11 er). Compound 9f was crystallized to obtain virtually enantiopure material (mother liquor, 56%, > 99:1 er). The methyl carbamate at the indole nitrogen was then introduced (10a 80%; 10f 88%), and the exocyclic olefin was converted to the corresponding aldehyde via a dihydroxylation/oxidative cleavage sequence, that was performed in one pot. Although aldehyde 11 may be isolated, it was routinely converted without further purification into tosyl hydrazone 12 ((±)-12, 91% and (+)-12, 79% from 10). The absolute configuration of (+)-12 was determined by single crystal X-ray diffraction, confirming the C20 (S)-configuration of all previous intermediates.

Initial attempts to form 14a by various transition-metal-catalyzed procedures[4f] were unsuccessful. However, 14b was formed through deprotonation of the hydrazone and generation of the corresponding diazo compound, although we were not able to obtain yields higher than 20–25%, the main products being the undesired vinyl-substituted tetracycles 13a–b (Scheme ). Most surprising was the fact that in 14b(12,13) the double bond had migrated to the opposite side of the hexahydroazocine ring. We also isolated pyrazoline 15,[12] which is the first example of a formal [3 + 2] dipolar cycloadduct between a diazocompound and an indole. By performing a formal [3 + 2] cycloaddition of tosyl hydrazone 12 in the presence of BF3·OEt2 as the Lewis acid,[14] we obtained 14a in 79–80% yield. Remarkably, this product of direct cyclopropanation (14a) could be converted in essentially quantitative yield into its isomer 14b by simple heating at 155 °C for 2 h.

Scheme 4

Indole Cyclopropanation and Olefin Isomerization

The puzzling isomerization of 14a most likely proceeds by a homodienyl retro-ene rearrangement[15] via 1,4-diene II, followed by the reverse process to form 14b (Scheme ). This type of transformation has been studied before in cyclic and bicyclic systems, leading irreversibly to skipped dienes.[16] The homodienyl retro-ene rearrangement of bicyclo[5.1.0]octen-2-ene has been reported to take place at 150–170 °C, to furnish 1,4-cyclooctadiene with an activation energy of ca. 33 kcal·mol–1.[16d] The reverse process, the formation of vinyl cyclopropanes from skipped dienes under thermal conditions, has only one precedent in the oxy-homodienyl rearrangement, which requires a temperature of ca. 260 °C (activation energies of 41–43.5 kcal·mol–1).[17] However, according to DFT calculations, the two transition states for the hydrogen shifts in our system have much lower barriers (ca. 29.5 kcal·mol–1) and the formation of a more stable conjugated enaminone drives the equilibrium toward the formation of 14b.[18]

Scheme 5

Migration of the Olefin from 14a to 14bvia Homodienyl Retro-Ene/Ene Rearrangement

Numbers in parentheses correspond to relative free energies in kcal·mol–1 (B3LYP/6-31G(d), solvent = toluene).

Isomers 14a and 14b behave very differently in their reactions with borane. Thus, whereas 14a reacted with excess BH3·SMe2 by exclusive reduction of the lactam to give 16 (56%), 14b led to an unexpected and remarkably inert heptacyclic diborane 17 (Scheme ).

Scheme 6

Reduction of 14a–b with Borane

CYLview depiction of the X-ray crystal structure of (±)-17.

Hydrogenation of 16 using PtO2 as the precatalyst gave lundurine C (3), albeit in a rather low yield (44%), while hydrogenation of the olefin of 14a prior to borane reduction of the lactam was unsuccessful. Gratifyingly, the ready access to 14b led to a considerably more efficient synthesis of 3 and, more importantly, provided an entry to the synthesis of lundurines A (1) and B (2). Hence, the first total synthesis of 3 could be completed in two steps from 14b, by reduction of the enaminone double bond with NaBH3CN[19] in the presence of formic acid to form saturated lactam 18, followed by a second reduction with BH3·SMe2 (Scheme ). Surprisingly, enantiopure lundurine C (3) presented an optical rotation ([αD589 = −1.1 ± 0.6°, CHCl3, c 0.98, 300 K] and [αD589 = −6.2 ± 0.8°, CH2Cl2, c 0.3, 301 K]) differing significantly from the one reported for the natural product [αD589 = −25°, CHCl3, c 0.067],[1,2] although chiral HPLC analysis of our synthetic sample of lundurine C left no doubt with regards to its enantiopurity. Furthermore, we prepared crystalline quaternary ammonium iodide 19, whose absolute configuration was established by X-ray crystallography. The discrepancy in the value of the optical rotation may arise from the very low concentration at which the natural product was measured originally that induced a significant error in the measurement.

Scheme 7

Synthesis of Lundurines A–C

CYLview depictions of the X-ray crystal structures of iodide salt 19 and lundurine A ((−)-1), with absolute configurations.

Lundurines A (1) and B (2) were both prepared in three additional steps from 18, by thiolation/C-sulfinylation–elimination and either oxidation or reduction (Scheme ). Intermediate 18 was first subjected to Lawesson’s reagent to form thiolactam 21, which then reacted with p-toluenesulfinyl chloride, in the presence of Hünig’s base, to generate in situ an α-sulfinyl thiolactam.[20] Upon heating at 80 °C, a Cope-type elimination gave thiolundurine A (22). Oxidation of 22 with m-CPBA at −78 °C produced (−)-lundurine A (1) in 69% yield, while treatment of 22 with iodomethane followed by sodium borohydride gave (−)-lundurine B (2) in 72% yield. Interestingly, unlike stated in the isolation and previous syntheses, racemic and enantiopure 1 are crystalline solids and we have also obtained the crystal structure of this natural product, confirming its absolute configuration and the one of the whole family of natural compounds.

In conclusion, we have developed a unified approach toward the synthesis of lundurines A–C, including the first enantioselective total synthesis of lundurine C, taking advantage of a gold(I)-catalyzed 8-endo-dig selective hydroheteroarylation to build the polyhydroazocine ring. Our synthesis of the lundurines is the shortest and most efficient to date (12–14 steps from known chiral alcohol 20f,[21] 6.6% overall yield for lundurine C and 3% overall yield for lundurines A and B, >99:1 er) and is perfectly suited to the preparation of analogues for biological evaluation as well as its extension to the synthesis of other Kopsia alkaloids. Worthy of note is the implementation of a practical chirality transfer in a complex tandem transformation and the new intramolecular cyclopropanation of indoles by formation of a pyrazoline by formal [3 + 2] cycloaddition in the presence of a Lewis acid. Finally, as often encountered in total synthesis endeavors, serendipity also played a significant role in the discovery of a new transformation in which a double bond migrates by means of a homodienyl retro-ene/ene rearrangement, which streamlined the access to this family of alkaloids.