Total Syntheses of (−)-Kopsifoline D and (−)-Deoxoapodine: Divergent Total Synthesis via Late-Stage Key Strategic Bond Formation

Abstract

Divergent total syntheses of (−)-kopsifoline D and (−)-deoxoapodine are detailed from a common pentacyclic intermediate 15, enlisting the late-stage formation of two different key strategic bonds (C21–C3 and C21–O–C6) unique to their hexacyclic ring systems that are complementary to its prior use in the total syntheses of kopsinine (C21–C2 bond formation) and (+)-fendleridine (C21–O–C19 bond formation). The combined efforts represent the total syntheses of members of four classes of natural products from a common intermediate functionalized for late-stage formation of four different key strategic bonds uniquely embedded in each natural product core structure. Key to the first reported total synthesis of a kopsifoline that is detailed herein was the development of a transannular enamide alkylation for late-stage formation of the C21–C3 bond with direct introduction of the reactive indolenine C2 oxidation state from a penultimate C21 functionalized Aspidosperma-like pentacyclic intermediate. Central to the assemblage of the underlying Apidosperma skeleton is a powerful intramolecular [4 + 2]/[3 + 2] cycloaddition cascade of a 1,3,4-oxadiazole that provided the functionalized pentacyclic ring system 15 in a single step in which the C3 methyl ester found in the natural products served as a key 1,3,4-oxadiazole substituent, activating it for participation in the initiating Diels–Alder reaction and stabilizing the intermediate 1,3-dipole.

Introduction

The kopsifolines and their unique core hexacyclic ring system were first disclosed in 2003 when Kam and Choo reported the initial members of this new alkaloid class (Figure 1).¹ Kopsifoline A–F (1–6) were isolated from the leaf extracts of a previously unencountered Malaysian Kopsia species later identified as K. fruticosa (Ker) A. DC., and their structures were established spectroscopically.¹ Subsequent exploration of K. singapurensis led to a second isolation of kopsifoline A (1) and characterization of a seventh kopsifoline (singaporentine A, 7).² Related to the Aspidosperma alkaloids, their more complex hexacyclic core structure incorporates a previously unprecedented C3–C21 carbon–carbon bond linking the terminal carbon (C21) of the C5 ethyl substituent to C3 bearing a methoxycarbonyl group such that the stereochemically rich central six-membered ring incorporates five or six stereogenic centers, three or four of which are quaternary.

Figure 1.

Kopsifolines.

In efforts that served to assign the natural product absolute stereochemistry, we recently disclosed a total synthesis of (+)-fendleridine (9, aspidoalbidine), enlisting an intermediate in which the C5 ethyl substituent bears an oxidized terminal C21 methyl group. Closure of this alcohol onto C19 (C21–O–C19 bond formation) via trap of an intermediate iminium ion provided the fendleridine tetrahydrofuran ring and completed the assemblage of its hexacyclic ring system (Figure 2).³⁻⁶ This same intermediate, albeit in the enantiomeric series and by virtue of conversion of the C5 ethyl group primary alcohol to a methyl dithiocarbonate, was enlisted for the total synthesis of kopsinine (10),^7,8 featuring a diastereoselective SmI₂-mediated free radical transannular conjugate addition reaction for formation of the bicyclo[2,2,2]octane core central to its hexacyclic ring system with C21–C2 bond formation. This late-stage C21–C2 bond formation not only complemented prior Diels–Alder approaches to its bicyclo[2,2,2]octane core,⁹ but it also represented the first synthetic approach that directly provided kopsinine from the underlying pentacyclic Aspidosperma alkaloid skeleton (Figure 2).

Figure 2.

Key strategic bond disconnections for approaching the natural products from a common precursor.

Herein, we report full details of the use of this same key intermediate in initial exploratory efforts that provided the parent kopsifoline skeleton (named (−)-kopsifoline H, 8) and the extension of these studies to the first total synthesis of a naturally occurring kopsifoline, (−)-kopsifoline D (4).¹⁰ By mimicking what is the likely biosynthesis of the kopsifolines,¹ a late-stage C21–C3 bond formation via a transannular enamide alkylation of an alcohol-derived C21 iodide within the Aspidosperma alkaloid skeleton was developed and implemented to complete the assemblage of the kopsifoline skeleton (eq 1) in a route that directly provides the natural product in its final indolenine oxidation state. The studies further established the kopsifoline absolute configuration as enantiomeric to the Aspidosperma alkaloids including fendleridine, but analogous to that found within kopsinine.

1

An additional attribute of the approach is that the same key intermediate was also used herein to access natural (−)-deoxoapodine (11),¹¹ originally isolated from Tabernae armeniaca, by virtue of C21–O–C6 bond formation. These efforts represent only the second total synthesis of the natural product⁵ and the first to provide 11 in optically active form, confirming the anticipated absolute configuration assignment.^11c Thus, a common Aspidosperma-like pentacyclic intermediate, bearing a terminally functionalized C5 ethyl substituent (primary alcohol), was used in the divergent total synthesis¹² of a suite of alkaloids, entailing linkage of the C21 primary alcohol oxygen to C19 (fendleridine)³ and C6 (deoxoapodine) or through linkage of C21 itself to C2 (kopsinine)⁷ and C3 (kopsifoline D) using the C21 functionality to conduct complementary nucleophilic or electrophilic C–C bond-forming reactions (Figure 3). The combined efforts represent the total syntheses of members of four different classes of natural products from a common intermediate deliberately functionalized for late-stage formation of four different key strategic bonds¹³ embedded in each unique core structure.

Figure 3.

Divergent total syntheses of a series of naturally occurring alkaloids containing deep-seated structural differences from a common cascade cycloaddition product.

Results and Discussion

The basis of the approach and key to the assemblage of the underlying Aspidosperma skeleton is a powerful intramolecular [4 + 2]/[3 + 2] cycloaddition cascade of a 1,3,4-oxadiazole that provides the fully functionalized pentacyclic ring system in a single step (Scheme 1).¹⁴ Assembled in two steps from N-benzyltryptamine and subsequent N-acylation of the amino-1,3,4-oxadiazole 12 with 4-(2-tert-butyldimethylsilyloxy)pent-4-enoic acid (13), thermal cyclization of 14 (230 °C, o-dichlorobenzene (o-DCB)) provided the key pentacyclic skeleton 15 (71%) as a single diastereomer. As disclosed in our initial work,³ the initiating intramolecular [4 + 2] cycloaddition¹⁵ is followed by a retro Diels–Alder loss of N₂ to provide an intermediate 1,3-dipole¹⁴ stabilized by the two substituents on the carbonyl ylide. In turn, the stabilized 1,3-dipole undergoes a subsequent diastereoselective [3 + 2] cycloaddition¹⁶ with the tethered indole exclusively through an endo transition state and with an intrinsic regioselectivity that is reinforced by the linking tether to provide the cascade cycloadduct 15. Inherent in the approach, the C3 methyl ester found in the kopsifolines, deoxoapodine, and kopsinine serves as a key 1,3,4-oxadiazole substituent, activating it for participation in the initiating Diels–Alder reaction and stabilizing the intermediate 1,3-dipole. As detailed in our preceding studies albeit with improvements in the originally disclosed conversions,⁷ diastereoselective reductive cleavage of the oxido bridge in 15 (NaCNBH₃, 88%) with exclusive convex face hydride reduction of an intermediate iminium ion flanked by two quaternary centers and subsequent methyl dithiocarbonate formation (NaH, CS₂; MeI; 91%) provided 17(7) in superb conversions.

Scheme 1.

Kopsifoline H

As our studies began, we first focused on the synthesis of 8, the parent kopsifoline skeleton. Although 8 has not yet been isolated as a naturally occurring substance, lacking a Δ^6,7-double bond and containing a saturated indoline, it possesses the parent hexacyclic skeleton and represents the tetrahydro derivative of kopsifoline D. Barton–McCombie deoxygenation¹⁷ of methyl dithiocarbonate 17 (Bu₃SnH, cat. AIBN, toluene, reflux, 1 h, 96%) cleanly provided 18 as a single diastereomer arising from exclusive convex face hydrogen atom reduction of the intermediate stabilized radical and inversion of the C3 stereochemistry (Scheme 2). Silyl ether cleavage of 18 (Bu₄NF, tetrahydrofuran (THF), −78 to 25 °C, 1 h, 92%) provided the primary alcohol 19 that in turn was converted to the iodide 20 (Et₃N, CH₃SO₂Cl, THF, −78 °C, 1 h then NaI, acetone, −78 to 50 °C, 12 h, 81%), setting the stage for studies on its cyclization to the kopsifoline skeleton. After exploration of several alternatives, treatment of 20 with KO^tBu in THF (3 equiv, −78 to 0 °C, 1 h) proved to be highly effective and afforded 21 (95%), whose structure and stereochemistry were confirmed with a single-crystal X-ray structure determination.^18a Initial, albeit limited, attempts to promote the cyclization of the intermediate mesylate (0%) or corresponding tosylate (32–36%) were not as productive. Chromatographic separation of the enantiomers of 21 (α = 1.2, 10% i-PrOH/hexane) was carried out on a semipreparative Daicel ChiralCel AD column, providing (+)-21 and ent-(−)-21. Treatment of each enantiomer of 21 with Lawesson’s reagent¹⁹ (1.1 equiv, toluene, 100 °C, 0.5 h) furnished the thiolactam 22 (89%). The absolute configuration of (+)-22 was unambiguously established by an X-ray structure determination conducted on this intermediate containing a heavy atom (S).^18b Concomitant reduction of the thioamide and removal of the N-benzyl group was accomplished by the treatment of (+)-22 with Raney-nickel (Ra–Ni, H₂, EtOH, 80 °C, 0.5 h) to furnish (−)-8 (70%), which we have come to refer to as kopsifoline H, designating its hydrogenated (H) state.

Scheme 2. Synthesis of Kopsifoline H.

Biomimetic C21–C3 Bond Formation: 6,7-Dihydrokopsifoline D

The additional key question addressed in initial studies was whether a late-stage C21–C3 bond formation via a transannular enamide alkylation of an alcohol-derived C21 electrophile was feasible for formation of the kopsifoline core, permitting the direct introduction of the correct C2 indolenine oxidation state of the natural products (see eq 1). These efforts began with 25, originally prepared in prior studies⁷ and bearing a readily removable indoline N-carboxybenzyl (Cbz) group. In efforts that impact the regioselectivity of the Chugaev elimination,²⁰ conversion of 16 to the corresponding Cbz carbamate 23, followed by an improved methyl dithiocarbonate formation (NaH, CS₂ then MeI, THF, 0 to 25 °C, 3 h, 93%) provided 24 (Scheme 3). The intermediate xanthate 24 underwent clean thermal elimination under mild reaction conditions (toluene, 100 °C bath, 48 h or 150 °C bath, 36 h), providing good yields (60%) of 25. Here, the indoline Cbz carbamate (vs N-benzyl) activates C2–H for xanthate syn elimination, favoring formation of the more substituted and stable olefin.⁷

Scheme 3. Synthesis of 6,7-Dihydrokopsifoline D.

Silyl ether cleavage in 25 (3 equiv Bu₄NF, THF, 25 °C, 1 h, 98%) and conversion of the primary alcohol 26 to the primary iodide 27 (Et₃N, CH₃SO₂Cl, THF then NaI, acetone, 50 °C, 12 h, 80%) followed by reduction of the amide 27 to the tertiary amine 28 (5 equiv BH₃·THF, THF, 0 °C, 1.5 h, 79%) set the stage for examination of the transannular cyclization. After optimization of the conditions, we were delighted to find that the treatment of 28 with BF₃·OEt₂ and dimethyl sulfide (Me₂S) in CH₂Cl₂ (25 °C, 15 h)²¹ proceeded smoothly not only to promote Cbz deprotection but also to afford 30 directly in 75% yield as a single diastereomer, providing 6,7-dihydrokopsifoline D. Shorter reaction times led to detection and isolation of the intermediate Cbz deprotection product 29 that itself undergoes slow cyclization to the indolenine 30 simply upon standing at room temperature. Notably, the corresponding carbinol amine derived from water addition to C2 of 30 could also be isolated if the chromatographic purification (SiO₂) of 30 was carried out without the presence of Et₃N (ethyl acetate (EtOAc) vs 2% Et₃N in EtOAc). Reduction of 30 (NaBH₄, EtOH, 0 to 25 °C) provided 8, confirming the structure of 30 and providing an alternative synthesis of kopsifoline H.

Finally, limited efforts to convert 8 to 30 by oxidation (MnO₂, 2,3-dichloro-5,6-dicyano-1,4-benzoquinone, Me₂⁺S–Cl, KMnO₄) to the C2 imine have not yet proved productive (Scheme 3), suggesting that direct indolenine introduction from intermediates such as 28 may constitute a more effective approach.

Total Synthesis of (−)-Kopsifoline D

Adoption of the approach for the total synthesis of kopsifoline D required installation of the Δ^6,7-double bond. Toward this end, α-phenylselenation of 25 was achieved by treatment of the lactam enolate (lithium diisopropylamide, −78 °C) with phenylselenyl chloride (PhSeCl, 1 equiv) to provide 31 (81%, Scheme 4). The mixture of isomeric phenylselenides (8:1, β:α) was treated with m-chloroperoxybenzoic acid (m-CPBA, 1.5 equiv) in CH₂Cl₂ at −78 °C to yield the corresponding selenoxides, and their subsequent in situ syn elimination proceeded smoothly (−78 to 25 °C) to afford 32 (90%). The enantiomers of 32 were easily separated (α = 1.48, 20% i-PrOH/hexane) on a semipreparative Chiralcel OD column, affording natural (+)-32 and ent-(−)-32. Clean 1,2-reductive removal of the lactam carbonyl was best accomplished by O-methylation of (+)-32 and subsequent reduction of the resulting methoxyiminium ion (3 equiv Me₃OBF₄, 2,6-t-Bu₂Py, CH₂Cl₂, 25 °C; NaBH₄, MeOH, 25 °C, 52%)²² to afford natural (−)-33. Nearly as effective, direct lactam reduction of (+)-32 by treatment with diisobutylaluminum hydride (Dibal-H, 5 equiv, THF, −20 °C, 20 min) also afforded (−)-33 (48%) in a single step. Alternatively, treatment of (+)-32 with Lawesson’s reagent¹⁹ (1.1 equiv, toluene, 100 °C, 0.5 h, 65%), subsequent S-methylation of the intermediate thioamide (3 equiv Me₃OBF₄, 2,6-t-Bu₂Py, CH₂Cl₂, 25 °C) and reduction of the resulting S-methyliminium ion (10 equiv NaBH₄, MeOH, 0 °C, 45 min) also provided 33 (40% overall), albeit in lower overall conversions. Silyl ether cleavage (Bu₄NF, THF, 25 °C, 1 h, 98%) followed by conversion of the primary alcohol (−)-34 to the primary iodide (−)-35 (Et₃N, CH₃SO₂Cl, THF, −78 °C then NaI, acetone, 90 °C bath temperature, 12 h, 68%) set the stage for the key transannular cyclization. Cbz deprotection and subsequent intramolecular ring closure of 35 (BF₃·OEt₂, Me₂S, CH₂Cl₂, 25 °C, 12 h followed by treatment with Et₃N, EtOAc, 25 °C, 12 h) provided (−)-kopsifoline D (4) in superb conversion (79%) and identical in all respects compared to those of the authentic material¹ except for its distinctly different optical rotation ([α]_D −69 (c 0.08, CHCl₃) vs reported [α]_D −27 (c 0.09, CHCl₃)^1b). Use of the intermediate mesylate (0%) and corresponding tosylate (22%) was less effective than the primary iodide 35, and its intramolecular ring closure to provide 4 was slower than that providing 30, requiring a longer reaction time and deliberate base treatment (Et₃N) for completion. Initial concerns over the discrepancy in the optical rotations between our synthetic 4 versus that reported for natural 4 led us to explore several origins of the distinctions and entailed repeated preparations of synthetic material. In the course of these efforts, the water addition product to kopsifoline D (kopsifoline D hydrate, 4a) was also isolated and characterized (Scheme 4). Unexpectedly, the optical rotation of this synthetic hydrate, which could be deliberately prepared by simply stirring a solution of 4 in THF–H₂O (95%, 25 °C, 3 h), perfectly matched that reported for kopsifoline D ([α]_D −26 (c 0.08, CHCl₃) vs reported [α]_D −27 (c 0.09, CHCl₃)^1b). This suggests that, while the spectroscopic characterization of natural kopsifoline D was effectively conducted with 4, the optical rotation itself was measured on a sample that had undergone hydration. The ease of kopsifoline D hydration also suggests that this corresponding hydrate is a natural product as well and that it likely will be isolated and characterized in the years ahead. The unambiguous assignment of the absolute configuration of 4 was accomplished with a single-crystal X-ray structure determination conducted on the natural enantiomer of the primary alcohol derived from 23,^18c which was correlated with (+)-32 (see Supporting Information [SI]).

Scheme 4. Total Synthesis of (−)-Kopsifoline D.

Total Synthesis of (−)-Deoxoapodine

With the natural enantiomer (+)-32 in hand, our efforts turned to the divergent total synthesis of (−)-deoxoapodine (Scheme 5). Complementary to the approach to kopsifoline D, silyl ether deprotection prior to lactam carbonyl removal was anticipated to permit the requisite C21–O–C6 strategic bond formation by conjugate addition of the primary alcohol. Silyl ether deprotection upon treatment of (+)-32 with Bu₄NF (3 equiv) in THF at −5 °C (6 h) under basic conditions smoothly provided only the primary alcohol conjugate addition product (−)-36 (70%) as a single diastereomer without the detection or isolation of the intermediate alcohol. Higher reaction temperatures (0–25 °C) led to generation of an additional minor diastereomer of the conjugate addition product, whereas lower reaction temperatures (−40 °C) resulted in little deprotection and cyclization. In contrast, silyl ether deprotection under acidic conditions conducted by treatment of (+)-32 with HF·pyridine (3 equiv, THF, 0 °C, 2 h) afforded only the deprotected primary alcohol (84–95%) with little or no cyclization even under prolonged reaction times (10 h). Reductive removal of the lactam carbonyl of (−)-36 upon treatment with BH₃·THF (THF, 0 °C, 1.5 h) provided (−)-37 (70%) and subsequent cleavage of the Cbz group (BF₃·OEt₂, Me₂S, 25 °C, 5 h, 82%)²¹ afforded natural (−)-deoxoapodine (11, [α]_D −522 (c 0.17, CHCl₃) vs [α]_D −432 (c 0.76, CHCl₃)^11a and [α]_D −593 (CHCl₃)^11b), which proved identical in all respects to those of the natural product. By virtue of the crystallographic assignment of the absolute configuration of intermediates in route to 11 ((+)-23^18c and (+)-32), the correlation serves to unambiguously confirm the anticipated absolute configuration for the natural product.^11c

Scheme 5. Total Synthesis of (−)-Deoxoapodine.

Conclusions

Divergent total syntheses of (−)-kopsifoline D and (−)-deoxoapodine that unambiguously established their absolute stereochemistry are detailed from a common pentacyclic intermediate 15 with the late-stage formation of two different key strategic bonds unique to their hexacyclic ring systems, complementing its prior use in the total syntheses of kopsinine and (+)-fendleridine. The combined efforts represent the total syntheses of members of four different classes of natural products from a common intermediate functionalized for late-stage formation of four different key strategic bonds uniquely embedded in each natural product core structure. The basis of the approach and central to the assemblage of the underlying skeleton is a powerful intramolecular [4 + 2]/[3 + 2] cycloaddition cascade of a 1,3,4-oxadiazole.^14,23−32 This reaction provided the C21 functionalized pentacyclic ring system 15 in a single step in which the C3 methyl ester found in the natural products served as a key 1,3,4-oxadiazole substituent, activating it for participation in the initiating Diels–Alder reaction and stabilizing the intermediate 1,3-dipole.

Supporting Information Available

Full experimental details and copies of ¹H/¹³C NMR spectra. This material is available free of charge via the Internet at http://pubs.acs.org.

The authors declare no competing financial interest.

Funding Statement

National Institutes of Health, United States