Keywords

1.1 Introduction

The chemical synthesis of structurally complex natural products has been a great challenge in organic chemistry over several decades. Therefore, we have engaged in the total synthesis of C(sp3)-rich natural products that bear multiple quaternary carbon stereocenters (i.e., stereogenic carbon centers having four different carbon substituents) on the carbocyclic framework [1]. These compounds represent formidable synthetic targets [2] and often exhibit superior biological activities than do achiral or “flat” compounds [3].

Brasilicardins A–D (BraA–D, 1–4) (Fig. 1.1), isolated from the pathogenic actinomycete Nocardia brasiliensis IFM 0406, is a novel C(sp3)-rich tricyclic diterpenoid that exhibits potent immunosuppressive activity [4,5,6,7]. Among the brasilicardin family members, BraA (1) is a promising drug lead compound because it exhibits strong immunosuppressive activity (IC50 = 0.057 μg/mL), low toxicity, and a mode of action that differs from that of current clinical drugs, such as tacrolimus (FK-506) and cyclosporine A [8]. Thus, BraA (1) has been extensively studied, particularly for developing a new type of immunosuppressive drug without serious side effects. However, further preclinical investigation of this promising drug candidate has been impeded by its low availability from natural sources. Therefore, the efficient chemical synthesis of 1, as well as its analogs, derivatives, and probe molecules, is required to support further biological studies.

Fig. 1.1
Chemical structures of brasilicardin A, brasilicardin B, brasilicardin C, and brasilicardin D. It is composed of phenanthrene, methyloxane, and methoxybutanoic acid.

Structures of brasilicardins A–D (BraA–D, 1–4)

Full size image

As shown in Fig. 1.1, BraA–D (1–4) share a highly strained anti-syn-anti-fused perhydrophenanthrene terpenoid skeleton (i.e., the ABC-ring; hereafter referred to as anti-syn-anti-fused 6,6,6-tricyclic skeleton) containing two angular quaternary methyl groups with the central ring (i.e., the B-ring) in the boat conformer. Different amino acid and sugar units are connected to this skeleton.

Their characteristic biological properties and novel, complex structures render this family attractive targets by synthetic organic chemists. Several research groups have conducted synthetic studies [9,10,11,12], including the first total synthesis of BraA (1) and BraC (3) by Anada and Hashimoto in 2017 [13]. A semi-synthetic approach for the large-scale production of BraA (1) was also reported in 2021 [14]. We launched synthetic studies on brasilicardins in 2008, aiming to develop an efficient route that can be accessed to all BraA–D (1–4) members from the same late-stage intermediate, while enabling the synthesis of various analogs and substructures for biological testing. We accomplished the total syntheses of 1–4 in 2018, including the first total syntheses of BraB (2) and BraD (4) [15, 16]. In this chapter, we describe our efforts toward the total synthesis of brasilicardins, focusing on how to overcome various synthetic challenges. In addition, several key lessons learned from our 10-year synthetic journey are discussed.

1.2 Previous Synthetic Approaches

The anti-syn-anti-fused 6,6,6-tricyclic skeleton is found in several bioactive natural products (Fig. 1.2a) and is an important intermediate (i.e., transient protosteryl cation 8) in the enzymatic cyclization of squalene in steroidal biosynthesis (Fig. 1.2b) [17]. This unusual and synthetically challenging anti-syn-anti configuration of the tricyclic skeleton has attracted the attention of organic chemists over the last few decades. However, in contrast to the detailed and extensive synthetic investigations performed in the field of classic terpenoids and steroids, the synthesis of such skeletons remains unexplored [9, 13, 18,19,20,21,22,23,24], with only a few total syntheses of natural products 5–7 having been reported [22,23,24]. Representative methods for constructing a skeleton are shown in Fig. 1.3. Because synthetic approaches to this skeleton, regarding equilibrium control, are expected to produce a more thermodynamically stable system (i.e., anti-anti-anti-fused system that does not possess the central ring boat conformer), kinetic control is required to access an anti-syn-anti-fused system. Thus, most previous syntheses have adopted the following two-phase strategy (Fig. 1.3a) [9, 13, 18, 19, 22, 24]. In phase I, a more thermodynamically stable 6,6,6-tricyclic skeleton is constructed. This skeleton includes a tricycle bearing a double bond in the ring juncture (e.g., 10) or one possessing all chair conformers (e.g., 18; Fig. 1.4). In phase II, a stereogenic center at the ring juncture is constructed under kinetic control. As shown in Fig. 1.3a, in the total synthesis of protostenediols (7) [24], Corey and Virgil first constructed a stable 6,6,6-system (10) bearing a double bond at the ring juncture via Robinson annulation (phase I). The crucial generation of the anti-AB fusion was achieved using the allylic diazene rearrangement of the in situ-generated hydrodiazene intermediate 11 (phase II), which resulted in the anti-syn-anti-fused product 12 as a major isomer along with the syn-AB fused isomer 13. The desired isomer 12 was converted to 7. In addition to the above two-phase strategy, transannular Diels–Alder reaction-based [20] and intermolecular/transannular Michael reaction cascade-based approaches [21] have been reported (not shown here).

Fig. 1.2
2 parts. Top. Chemical structures of fusidic acid, isoaplysin, and protostenediols. All are composed of 3 fused cyclohexane rings. Fusidic acid and protostenediols also have a cyclopentane ring. Bottom. Reaction schematic. Squalene is converted to protosteryl cation followed by lanosterol.

Examples of anti-syn-anti-fused 6,6,6-tricyclic systems found in nature. a Selected examples of previously synthesized terpenoids. b Related biosynthetic intermediate

Full size image
Fig. 1.3
Two synthetic approaches to anti-syn-anti-fused 6, 6, 6-tricyclic systems. Corey and Virgil's 2-phase method involves Robinson annulation and conversion steps. Corey and Nishizawas biomimetic approach mimics biological processes with reactions using reagents like mercury 2 trifluoroacetate-N-methoxy methylenechynium chloride, and bromine.

Reported synthetic approaches to anti-syn-anti-fused 6,6,6-tricyclic systems. a Two-phase approach by Corey and Virgil. b Biomimetic approach by Nishizawa et al.

Full size image
Fig. 1.4
Chemical reactions presenting the total synthesis of brasilicardin A by Anada and Hashimoto with M O M O protected cyclohexanone and T B S O as starting materials. Phase 1 converts starting materials into molecule 18 through a 7-step process. Phase 2 converts molecule 18 into brasilicardin A.

Summary of the total synthesis of brasilicardin A by Anada and Hashimoto

Full size image

Although biomimetic approaches often enable access to complex targets in a concise and stereoselective manner, applying this approach to a thermodynamically less stable anti-syn-anti-fused 6,6,6-tricyclic skeleton is adversely affected by low stereoselectivity (Fig. 1.3b) [23]. For example, in the total synthesis of isoaplysin-20 (6), Nishizawa et al. directly constructed an anti-syn-anti-fused 6,6,6-tricyclic system using the biomimetic Hg(OTf)2-mediated polyene cyclization of (E,E,E)-geranyl acetate (14), affording the desired product 16 after subsequent bromination. However, in this biomimetic cyclization, the major product was the stereoisomer 15 with anti-anti-anti-ring juncture, which forced the ring system to adopt a stable chair-chair-chair conformation.

The first total syntheses of BraA (1) and BraC (3) by Anada and Hashimoto are summarized in Fig. 1.4 [13]. In their total synthesis, the anti-syn-anti-fused 6,6,6-tricyclic skeleton was constructed using the Diels–Alder reaction/angular methylation sequence developed by Coltart and Danishefsky [9]. Thus, the Diels–Alder reaction of Wieland–Miescher ketone-derived cyanoenone 17 with siloxydiene proceeded smoothly to yield ketonitrile 18 as the sole isomer. The reductive angular methylation of 18 afforded the desired C-methylation product 19 with high chemo- and stereoselectivities. Incorporating the amino acid moiety was conducted using an anti-selective aldol reaction (20 → 22) using titanium enolate generated from chiral iminoglycinate 21. The stereocontrolled glycosylation of alcohol 23 with disaccharide 24 under Schmidt’s conditions delivered BraA (1) following the removal of the protecting groups. BraC (3) was synthesized using a similar reaction sequence to that of 23 via glycosylation with a monosaccharide.

1.3 Synthetic Challenges and Initial Model Studies

From a synthetic perspective, the total synthesis of structurally complex BraA–D (1–4) members must overcome the following challenges:

  1. (1)

    The development of a stereoselective methodology for a highly strained carbocyclic skeleton with two quaternary carbon stereocenters at the ring junctures (ABC-ring system; Fig. 1.1) represents the most important issue in this synthesis program.

  2. (2)

    In relation to (1), the stereoselective construction of two neighboring quaternary carbon stereocenters must involve stereoselective carbon–carbon forming reactions that proceed in a sterically congested environment.

  3. (3)

    Stereoselective construction of the amino acid component.

  4. (4)

    Regio- and stereoselective glycosylation of the sugar unit.

  5. (5)

    Overall protecting group strategy toward an efficient total synthesis.

With these considerations in mind, our synthetic journey to BraA–D (1–4) began with the development of a stereocontrolled route to the ABC-ring system using a model substrate without hydroxy functionalities on the A-ring. We found that nitriles were the important functional groups in total synthesis. Therefore, we designed a nitrile-based synthetic strategy for the ABC-ring system because of their following advantages and features [25, 26]:

  1. (1)

    Minimal steric demand of the compact cyano group arising from the linear nature of the CN moiety with an A-value of 0.2 kcal mol–1. In comparison, carbonyl and methyl groups have A-values of 0.6–2.0 and 1.74 kcal mol–1, respectively [27].

  2. (2)

    Compared to carbonyl analogs, α-cyano carbanions exhibit an exceptionally high nucleophilicity. This occurs because minimal delocalization into the nitrile group localizes the charge density on the adjacent carbon atom, which results in the enhancement of carbon nucleophilicity.

  3. (3)

    Nitriles are useful and versatile synthetic intermediates for further functionalization and bond-forming reactions.

As a model study, we first investigated the construction of a B-ring bearing two adjacent quaternary carbon stereocenters, which was a central challenge in this journey, and established a synthetic route to the ABC-ring core 32 (Fig. 1.5a) [28]. Thus, α,β-unsaturated lactone 27 bearing an alkanenitrile moiety on the side chain was synthesized from racemic α-ionone (26). Upon treatment with sodium bis(trimethylsilyl)amide (NaHMDS)/hexamethylphosphoric triamide (HMPA), 27 underwent intramolecular endocyclic conjugate addition (Michael addition) to afford the desired product 28 as the major isomer, along with its C8 epimer 29 (28:29 = 90:10). Notably, the cyclization proceeds smoothly even at − 78 ℃ despite the low reactivities of the α-cyano carbanions generated from non-activated simple alkanenitriles as Michael donors [25]. The origin of the high stereocontrol is due to the conjugate addition of the α-cyano carbanion derived from 27 that preferentially proceeded via the transition state TS-A rather than the alternative transition state TS-B to avoid the 1,3-repulsion of the two methyl groups in TS-B (Fig. 1.5a). Then, cyclization product 28 was converted to the (E)-α,β-unsaturated ester 30 having a 1,1-dibromo alkene group. When 30 was exposed to lithium dimethylcopper (Me2CuLi), stereoselective cyclization occurred, thus providing the tricyclic core 32 as a sole isomer (unoptimized 39% yield). This reaction appeared to proceed via the in situ generation of (Z)-vinyl copper intermediate 31 followed by the intramolecular conjugate addition of 31 [29]. Therefore, we developed a synthetic route to the anti-syn-anti-fused 6,6,6-tricyclic skeleton of BraA–D (1–4) using two sequential intramolecular conjugate additions as the key steps. In addition, we recognized the synthetic utility of the α-cyano carbanion for the construction of sterically demanding quaternary carbon stereocenters.

Fig. 1.5
2 reaction schematics. a begins with alpha ionone followed by compounds 27 to 29, T S A, T S B, 30, 31 and 32. b highlights low yield and issues including unoptimized yield, multiple steps, and diastereomers formed in the first step.

Initial model study toward the ABC-ring system of brasilicardins. a Summary of the synthetic routes. b Serious issues in the synthesis

Full size image

However, critical issues remain with this synthesis. Intermediate 33 showed extremely poor reactivity toward various transformations, because 33 exists in a stable lactol form with a diamond-like structure (Fig. 1.5b). Consequently, the cleavage of the carbon–oxygen bond between the C12 and O1 atoms was difficult. For example, homologation reactions, including the Corey–Fuchs reaction and hydride reduction, do not proceed at elevated temperatures. To overcome this difficulty, the alkynyl C2-unit was attached to 33 for the reductive cleavage of the C12–O1 bond (34 → 35). However, this C2-unit was not incorporated into the second cyclization precursor 30, which indicates that it is synthetically inefficient. Second, to overcome this issue, the conversion of the first cyclization product 28 to the second cyclization precursor 30 required 13 steps with a low overall yield (0.39%). These results led us to develop an alternative synthetic strategy, which is explained in the following section.

1.4 Strategy and Retrosynthesis

Because the intramolecular conjugate addition of alkanenitrile to α,β-unsaturated lactone serves as a powerful method for constructing a sterically congested quaternary stereocenter (Fig. 1.5, 27 → 28 + 29), we examined its substrate scope to evaluate its synthetic potential, in parallel with the total synthesis program. We found that this addition proceeded smoothly even in the unfused simple unsaturated lactones (e.g., 37) with certain modifications, where adding a bulky silylating reagent triisopropylsilyl chloride (TIPSCl) for trapping the lactone enolate was required to prevent the unfavored reversible retro-addition (Fig. 1.6, 37 → 38 → 39) [30]. By applying this methodology, we designed an intramolecular conjugate addition of an acyclic α,β-unsaturated ester bearing an alkanenitrile on the side chain as the key technology for this synthesis program (40 → 41 → 42). If this addition occurred with facial discrimination of the rotationally unsaturated ester, the compact cyano group would cause stereoselective cyclization, which would result in the formation of the contiguous quaternary and tertiary stereocenters simultaneously after the hydrolysis of the resulting ketene silyl acetal intermediate 41. In addition, the potentially different reactivities of the sterically demanding cyano group at a quaternary stereocenter and monosubstituted ester group in product 42 would enable their chemoselective transformations. We envisioned that this cyclization would fit into the construction of the A- and B-rings of brasilicardins.

Fig. 1.6
A reaction schematic presents the scientific strategy for the intramolecular conjugate addition of an unfused alpha, beta-unsaturated lactone. Starting with compound 37, the reaction involves treatment with L i H M D S and T I P S C I at negative 78 degrees Celsius in T H F and H M P A, generating an alpha-cyano carbanion species.

Intramolecular conjugate addition of unfused α,β-unsaturated lactone and potential cyclization strategy

Full size image

With this strategy in mind, our retrosynthesis of BraA–D (1–4) is illustrated in Fig. 1.7. Aiming at the detailed structure–activity relationships for deeper understanding the mechanism of action of 1–4 in the future, we utilized strategies in which each ring of the carbocyclic core, amino acid and saccharide units, would be constructed in a stepwise manner. The labile sugar moiety was installed via the regioselective glycosylation of the N-Fmoc-protected aglycons 43 (for BraA and BraC) or 44 (for BraB and BraD) at the final stage of synthesis. These aglycons could be obtained from ester 45 via the construction of the amino acid unit. Thus, we identified that tricyclic core 45 could serve as a central intermediate for unified synthesis. The requisite core 45 was synthesized using an intramolecular conjugate addition-based strategy. Particularly, 45 can be accessed from (E)-α,β-unsaturated ester 46 via intramolecular conjugate addition promoted by Me2CuLi, as described in Sect. 1.3. Dibromide 46 was synthesized from bicyclic cyano ester 47 via the carbon chain elongation of the substituents. The B-ring was constructed via the intramolecular nitrile conjugate addition of (Z)-α,β-unsaturated ester 48, which was derived from cyano ester 49. Subsequently, the formation of the A-ring was conducted by a similar conjugate addition of (E)-α,β-unsaturated ester 50. Therefore, this compound can be accessed in an enantiomerically pure form from commercially available 2,2-dimethylpropane-1,3-diol (51) via Sharpless asymmetric dihydroxylation.

Fig. 1.7
A retrosynthetic analysis of brasilicardins A to D disconnects the sugar unit and amino acid moiety from a common intermediate, and then constructs the core tricyclic skeleton.

Retrosynthetic analysis of brasilicardins A–D

Full size image

1.5 Construction of the A-Ring via Intramolecular Conjugate Addition

For the asymmetric total synthesis of BraA–D (1–4), we initially focused on the synthesis and cyclization of chiral (E)-α,β-unsaturated ester 57 (Fig. 1.8). The mono-TBS protection of 2,2-dimethylpropane-1,3-diol (51) followed by the Swern oxidation of the remaining alcohol 52 afforded the corresponding aldehyde, which was subsequently converted to (E)-α,β-unsaturated ester 53 (E/Z > 99:1) through a one-pot Horner–Wadsworth–Emmons (HWE) olefination [31]. The Sharpless asymmetric dihydroxylation of 53 using the monomeric ligand (DHQ)PHN [32] afforded optically active diol 54 with high enantiopurity (95% ee). The protection of the diol in 54 with methoxymethyl (MOM) groups followed by the LiAlH4 reduction of the ethyl ester and iodination of the resulting alcohol gave primary iodide 55. The alkylation of 55 with carbanion derived from propanenitrile and subsequent removal of the TBS group afforded alcohol 56. This compound was subjected to oxidation using tetrapropylammonium perruthenate and HWE olefination to afford 57 as a substrate for intramolecular conjugate addition.

Fig. 1.8
A synthesis and intramolecular conjugate addition of E-alpha, beta-unsaturated ester 57 involves several steps to form molecule 55 from commercially available starting materials with a good yield. Molecule 56 undergoes an intramolecular conjugate addition reaction to form a cyclic molecule 57 in 88% yield.

Synthesis and intramolecular conjugate addition of (E)-α,β-unsaturated ester 57

Full size image

With 57 available, the construction of the A-ring was examined (Fig. 1.8). The nitrile intramolecular conjugate addition of 57 occurred under similar conditions to 37 (i.e., TIPSCl/LiHMDS) to afford cyano ester 58 as a mixture of three isomers with acceptable diastereoselectivity (60% yield, dr = 82:11:7). Although the reduction of ester 58 with diisobutylaluminum hydride (DIBAL) gave the desired aldehyde 59 (36% yield), this conversion was accompanied by the formation of the over-reduced alcohol 60 as the major product. To access aldehyde 59 in a chemoselective manner, we planned to use an α,β-unsaturated N-methoxy-O-methylamide (commonly known as Weinreb amides) [33] as an alternative Michael acceptor.

The requisite (E)-α,β-unsaturated Weinreb amide 61 was prepared from 56 using a reaction sequence similar to that for 57 (Fig. 1.9). After screening the reaction conditions, when 61 was exposed to NaHMDS at − 78 ℃ in THF, intramolecular conjugate addition proceeded smoothly with complete stereoselectivity, furnishing the desired product 62 in improved yield (93%). The complete stereocontrol of 62 was assumed to arise because of chelation control, where the nucleophilic keteniminate and electrophilic α,β-unsaturated amide were both oriented in the equatorial positions with an antiparallel dipolar arrangement in the transition state TS-C. Despite the known poor reactivity of α,β-unsaturated amides as Michael acceptors [34], the higher reactivity of the Weinreb amide in this Michael addition was suggested to arise from the supposed and stable tetrahedral intermediate 63 formed upon the addition, which would prevent unfavorable reversible retro-conjugate addition. Thus, the Weinreb amide played two important roles in this addition: (1) enhancement of the stereoselectivity and (2) suppression of the retro-addition. Therefore, we unexpectedly found the superior reactivity of α,β-unsaturated Weinreb amides as Michael acceptors.

Fig. 1.9
A chemical reaction presenting the synthesis and intramolecular conjugate addition of E-alpha, beta-unsaturated Weinreb amide 61 starts with methyl E-crotonate 56 and proceeds in 3 steps to form molecule 61 in 96% yield. It then undergoes an intramolecular conjugate addition to form a new 6-membered ring.

Synthesis and intramolecular conjugate addition of (E)-α,β-unsaturated Weinreb amide 61

Full size image

1.6 Construction of the B-Ring

Having realized the power of intramolecular nitrile conjugate addition, we moved on to the next phase of the synthesis, which was the construction of the B-ring (Fig. 1.10). Similar to the case of the A-ring, we decided to use an α,β-unsaturated Weinreb amide as the Michael acceptor for B-ring formation. As expected, the chemoselective reduction of the Weinreb amide in the presence of the cyano group in 62 was accomplished with DIBAL in THF, and the subsequent one-carbon elongation of the resulting aldehyde using Wittig’s reagent afforded enol ether 64. Compound 64 was subsequently converted to alkene 66 using a four-step reaction sequence. The regioselective introduction of a cyano group to 66 was achieved using Co-catalyzed hydrocyanation with TsCN [35] to afford secondary nitrile 68. The oxidation of alcohol 68 gave the corresponding aldehyde, which was olefinated under Ando’s HWE reaction conditions [36, 37] to afford (Z)-α,β-unsaturated ester 69 with exclusive Z-selectivity (Z:E > 99:1). Ester 69 was successfully converted into O-methyl Weinreb amide 70 using magnesium amide [38]. Notably, the elongation of the C2-unit via stepwise Wittig reactions (62 → 65) was necessary, because substitution with various vinyl metal reagents was unsuccessful for 71 and 72.

Fig. 1.10
Chemical reactions present synthetic route for Z alpha, beta-unsaturated Weinreb amide 70 that starts with a M O M O-protected diol and undergoes a 2-step process to install a methoxy group and convert a carbonyl group to O H. It then converts the intermediate to Z-70 through a 2-step process involving DIBAL and cobalt catalyst.

Synthetic route of (Z)-α,β-unsaturated Weinreb amide 70

Full size image

Because the second intramolecular conjugate addition is another crucial step in the synthesis, we carefully investigated the reaction conditions (Fig. 1.11). Upon treatment with NaHMDS in the presence of TIPSCl and HMPA in THF at − 78 ℃, unsaturated Weinreb amide 70 smoothly underwent the cyclization to in situ produce O-silyl N,O-ketene acetal 73. Upon the addition of tetrabutylammonium fluoride (TBAF) to the reaction mixture in one pot, the desired product 74 and its C8,9-diastereomer 75 were quantitatively obtained as an inseparable mixture (74:75 = 45:55). While cyclization proceeded with the sole use of NaHMDS (89% yield, 74:75 = 50:50), the addition of TIPSCl improved the product yields. After exploring the solvents, additives, and reaction temperatures, using Et2O as the solvent in the absence of HMPA resulted in better stereoselectivity (dr = 80:20). For comparison, the intramolecular conjugate addition of the corresponding (E)-isomer 76 afforded 75 under the same conditions, indicating that the alkene geometry affected the stereoselectivity of this process.

Fig. 1.11
An intramolecular conjugate addition reaction of Z-alpha, beta-unsaturated Weinreb amide 70 with N a H M D S in T H F at negative 78 degrees Celsius to form a six-membered ring system 74 in 89% yield.

Intramolecular conjugate addition of 70

Full size image

The presumed transition-state model for the intramolecular conjugate addition of 70 is shown in Fig. 1.12. The desired isomer 74 was probably obtained via transition state TS-D in which the nucleophilic keteniminate and electrophilic unsaturated Weinreb amide unit both occupied the axial direction. Conversely, diastereomer 75 was obtained via transition state TS-E in which both occupied the equatorial direction. Because the energy difference between the two transition states was small, a considerable quantity of 75 was formed. We envisioned that if the methoxy group on the Weinreb amide moiety was replaced with a sterically demanding tert-butoxy group [39], the transition state TS-F, which leads to the desired stereoisomer, would be more favorable than the alternative transition state TS-G to avoid repulsive interaction between the two 1,3-diaxial methyl groups, as well as between the bulky t-butoxy group of the Weinreb amide and methyl group next to the keteniminate.

Fig. 1.12
2 plausible transition-state models for the intramolecular conjugate addition of molecule 70. Model T S D positions the Weinreb amide carbonyl oxygen closer to the alpha-carbon, forming a partial bond. The alpha-carbon's hydrogen moves towards the Weinreb amide nitrogen. Model T S E positions the alpha-carbon closer to the Weinreb amide nitrogen.

Plausible transition-state models for the intramolecular conjugate addition of 70

Full size image

As expected, under similar reaction conditions for the cyclization of 70 (i.e., NaHMDS/TIPSCl in Et2O; TBAF), the intramolecular conjugate addition of O-tert-butyl Weinreb amide 77, which was synthesized from ester 69, resulted in the stereoselective formation of the desired product 79 as a 93:7 inseparable mixture with diastereomer 80 (Fig. 1.13). The two key nitrile conjugate additions (61 → 62 and 77 → 79) were performed reproducibly on a gram scale, thus demonstrating the high synthetic utility of this cyclization.

Fig. 1.13
Chemical reactions presenting the cyclization of O-tert-butyl Weinreb amide 77 with L i O H followed by T F A and P O affords molecule 78 in 88% yield. The bottom right exhibits molecule 78 after an aqueous workup.

Successful cyclization of O-tert-butyl Weinreb amide 77

Full size image

1.7 Stereoselective Synthesis of the ABC-Ring

Having developed a synthetic route to the AB-ring bearing two quaternary stereocenters based on strategic nitrile conjugate additions, our next objective was the third intramolecular conjugate addition, which resulted in the ABC-ring of brasilicardins (Fig. 1.14). Thus, Weinreb amide 79 was converted into dibromoalkene 81 by reduction of the Weinreb amide moiety to an aldehyde followed by Corey–Fuchs olefination, after which the inseparable isomer 80 was separable. After reduction of the cyano group in 81, HWE olefination of the resulting aldehyde furnished the (E)-unsaturated ester 82.

Fig. 1.14
Chemical reactions present the construction of the A B C-ring system of brasilicardins through an intramolecular conjugate addition reaction. It exhibits how Weinreb amide 77 reacts with L D A to form the A B C-ring system of brasilicardins in 79% yield.

Construction of the ABC-ring system of brasilicardins

Full size image

The crucial third intramolecular conjugate addition to construct the C-ring was performed under the optimized reaction conditions. Thus, the exposure of 82 to Me2CuLi in Et2O at − 78 ℃ generated (Z)-vinylcopper species 83 in situ. After the reaction mixture was increased to − 40 ℃ and stirred, the subsequent conjugate addition of 83 proceeded to provide the tricyclic compound 84 by controlling the stereochemistry at the C14 position in 83% yield. The stereochemistry of 84 was confirmed by X-ray crystallographic analysis. Thus, we established a novel method for the stereoselective formation of an anti-syn-anti-fused 6,6,6-tricyclic skeleton (ABC-ring) using sequential triple intramolecular conjugate addition as the key step.

1.8 Construction of the Amino Acid Component of Brasilicardins A and C

After obtaining the tricyclic compound, we focused on the construction of an appropriate amino acid component for the tricyclic skeleton. First, we investigated the installation of an anti-β-methoxy-α-amino acid moiety for BraA (1) and BraC (3). Two possible plans were considered (Fig. 1.15). Although glycine aldol reactions between aldehydes and glycine derivatives or chiral glycine equivalents are an efficient and direct method for constructing such systems (plan A, 86 + 87 → 85), these methods were challenging to apply to the functionalized intermediates of brasilicardins from our investigations using model compounds and advanced substrates. We encountered critical issues, including harsh conditions (6 M aq. HCl, 80 ℃) for the removal of the camphor-derived chiral auxiliary in the use of 88 [40], and the requirement of excess aldehyde in the organocatalyzed asymmetric aldol reaction with 89 [41] that was not suitable for the late-stage installation using the precious aldehyde. Therefore, we elected to build this amino acid moiety using indirect methods via the substitution of azide ions with chiral trans-epoxides 91 and 92 or hydroxy ester 93 (plan B) and first examined the route via trans-epoxy ester 91.

Fig. 1.15
Chemical reactions present strategies for installing an anti-beta-methoxy-alpha-amino acid component into a molecule. Strategy A involves adding the anti-beta-methoxy-alpha-amino acid component directly to the molecule. Strategy B creates a new molecule with the former acid component that can be coupled to the desired target molecule.

Two possible strategies for the installation of an anti-β-methoxy-α-amino acid component

Full size image

The attempted construction of the amino acid moiety using chiral epoxy ester 100 is shown in Fig. 1.16. Tricyclic aldehyde 94, derived from ABC-ring compound 84, was subjected to the aldol reaction using Bu2BOTf/i-Pr2NEt with Evans-type chiral oxazolidinone 95, affording syn-α-chloro-β-hydroxy adduct 96 as a sole isomer [42]. However, this reaction exhibited poor reproducibility and often afforded a considerable quantity of cyclic ether 97 as a side product. Exposure of 96 to NaOMe enabled epoxide formation and esterification to produce trans-epoxy ester 100, which involved the in-situ epimerization of syn-chlorohydrin 98 followed by ring closure of the resulting anti-chlorohydrin 99. When treating 100 with hydrogen azide, regio- and stereoselective substitution of azide ions occurred at the α-position of the ester, affording anti-α-azido-β-hydroxy ester 101. We examined another approach for synthesizing 100 via the asymmetric epoxidation of an (E)-α,β-unsaturated N-acylpyrrole [43] as well; however, the diastereomeric selectivity of this reaction varied between 10:1 and 1:1. Thus, owing to their irreproducibility, we abandoned these approaches and explored the route via the substitution of azide ions with epoxy alcohol 92 (plan B, Fig. 1.15).

Fig. 1.16
Chemical reactions presenting the attempted construction of an anti-beta-methoxy-alpha-amino acid component via trans-epoxy ester 100. The synthesis starts with a protected dihydroxy acid and goes through several steps, but does not yield the desired product.

Attempted construction of anti-β-methoxy-α-amino acid component via trans-epoxy ester 100

Full size image

The requisite chiral 2,3-trans-epoxy alcohol 102 was prepared from 84 in a four-step sequence, including Katsuki–Sharpless asymmetric epoxidation (Fig. 1.17). The C2-azide substitution reaction of 102 was accomplished using NaN3 and B(OMe)3, which were developed in our laboratory [44] to produce the C2-substitution product 104 with high regio- and stereoselectivity as an inseparable mixture with the C3-product 105 (dr > 91:9). The reaction is suggested to proceed via the endo-mode epoxide opening of an intramolecular boron chelate, such as 103. The subsequent treatment of the crude mixture with aqueous NaIO4 affected the oxidative cleavage of 105 to the aldehyde (not shown), furnishing 104 in its pure form following purification using silica gel column chromatography. After a three-step conversion of 104 to mono-alcohol 106, including O-methylation with Meerwein reagent (Me3OBF4), oxidation of the primary alcohol in 106 to carboxylic acid, followed by esterification with HCl in MeOH afforded methyl ester 107 accompanied by the deprotection of both MOM groups. Finally, the conversion of the azide in 107 to an amino group with SnCl2 to afford compound 108, which is the methyl ester of the BraA (1) and BraC (3) aglycon. 1H- and 13C-NMR spectra and optical rotations of 108 were identical to those derived from natural sources [4].

Fig. 1.17
Chemical reactions for the alternative approach for synthesizing an anti-beta-methoxy-alpha-amino acid component display a 2-step process starting with epoxy alcohol 102 to create an intermediate. It then suggests the deprotection of O C H 3 from compound 104 to yield the desired product.

Alternative approach via epoxy alcohol 102 and attempted deprotection of the methyl ester

Full size image

In this study, we also checked whether deprotection of the methyl ester in the functionalized substrate proceeded, because late-stage chemoselective deprotection would be required to achieve total synthesis. Therefore, we examined the deprotection of α-azido methyl ester 109 to carboxylic acid 110 (Fig. 1.17). However, this deprotection was problematic despite testing several methods. In addition, an inefficient 12-step sequence was necessary for the conversion from 84 to 108, primarily because of the oxidation-state adjustment of the amino acid moiety, which led us to investigate an alternative synthetic route via hydroxy ester 93 (cf., Fig. 1.15, plan B).

A successful approach using a hydroxy ester by applying Rama Rao’s procedure [45] is shown in Fig. 1.18. We chose tert-butyl ester as the protecting agent for the carboxylic acid instead of the methyl ester. The half-reduction of ester 84 using DIBAL and subsequent HWE reaction of the resulting aldehyde furnished (E)-unsaturated tert-butyl ester 111. The Sharpless asymmetric dihydroxylation of 111 using the (DHQ) PHN ligand [32] gave diol 112 as a sole diastereomer (dr > 99:1). After the regioselective mono-nosylation of the C17 alcohol in 112 with 4-nitrobenzenesulfonyl chloride (p-NsCl), treatment of the resulting nosylate 113 with NaN3 afforded β-azide 114 with an inversion of the configuration. The O-methylation of the hydroxy group in 114 using Me3OBF4 followed by reduction of the azide and protection of the resulting free amine with a 9-fluorenylmethyloxycarbonyl (Fmoc) group in one pot, afforded the protected aglycon of BraA and BraC (i.e., 115). Lastly, the removal of both MOM groups with HCl in methanol produced diol 116. The stereochemistry of the amino acid moiety was unambiguously confirmed using X-ray crystallography after conversion to the p-bromobenzamide derivative 117. Although it might require more steps than the ideal glycine aldol-based approach, this route was robust and provided reproducibly sufficient material (> 100 mg in one batch) to accomplish the total synthesis.

Fig. 1.18
Chemical reactions present the construction of the amino acid component for brasilicardins A and C. The strategy involves using a chiral glycine enolate and reacting it with aldehyde 41 to form a beta-hydroxy-alpha-amino acid in 97% yield with good diastereoselectivity.

Construction of the amino acid component of brasilicardins A and C

Full size image

1.9 Construction of the Amino Acid Component of Brasilicardins B and D

In contrast, the amino acid components of BraB (2) and BraD (4) were constructed via the Yamada’s asymmetric alkylation [46] as the key step (Fig. 1.19). Thus, ethyl ester 84 was transformed into iodide 118 via reduction of ester with LiAlH4 and iodination of the resulting alcohol. Alkylation of the chiral Schiff base 119, prepared from α-pinene and glycine, with 118 proceeded uneventfully when KHMDS was used as a base, affording imino tert-butylester 120 as a single diastereomer. Contrary to the literature [46], the use of KHMDS as a base yielded superior results to those of LDA. Notably, the asymmetric alkylation of 118 using sultam-derived glycine imine 123 [47] and organocatalytic asymmetric alkylation with 89 [48] afforded inferior results, including a low product yield (11%) in the former method and low diastereoselectivity (dr = 50:50) in the latter method. Compound 120 was converted into diol 122 in a sequential three-step process: (1) hydrolytic removal of the chiral auxiliary, (2) protection of the resulting free amine with a Fmoc group, and (3) removal of both MOM groups. Stereochemistry of the amino acid moiety was verified using a modified version of Mosher’s method [49]. The protected aglycons 116 and 122 were used in the following glycosylation studies.

Fig. 1.19
Chemical reactions present the construction of the amino acid component of brasilicardins B and D. The synthesis starts with an L-threonine derivative and undergoes steps to form a protected alpha-keto acid. The acid is then converted to a chiral amino acid derivative which is deprotected under acidic conditions.

Construction of the amino acid component of brasilicardins B and D

Full size image

1.10 Stereoselective Glycosylation of Disaccharides and Completion of the Total Synthesis of Brasilicardins A and B

The remaining task for completion of the total synthesis was the challenging regioselective glycosylation of aglycons 116 and 122. First, we explored the glycosylation of 116 with BraA (1) as the priority target for this project. Our glycosylation study commenced with the Schmidt trichloroacetimidate glycosylation protocol [50] because of its well-documented success in the synthesis of natural products (Fig. 1.20a). Because the most reliable method for controlling stereoselectivity in 1,2-trans glycosylation is based on neighboring group participation by a 2-O-acyl functionality, we chose the acetyl group as the protecting and stereodirecting group at the C2′-alcohol of the glycosyl donor (cf., 24).

Fig. 1.20
A chemical reaction links sugars together, toward brasilicardin A. Part A attempts to glycosylate brasilicardin A using a sugar unit and the Schmidt protocol. Part B proposes reasons for the low yield of the desired product by mentioning possible side reactions that consume starting materials or lead to undesired products.

Glycosylation study toward brasilicardin A using the Schmidt protocol. a Attempts at Schmidt glycosylation. b Detectable side products and plausible reaction mechanism

Full size image

The treatment of aglycon 116 with peracetyl imidate 24 (2 equiv), which was prepared in nine steps from L-rhamnose [10] with BF3·OEt2 under the standard Schmidt procedure, afforded the desired C2-α-monoglycoside 124; however, significant quantities of the side products, such as C3-glycoside 125, C2-acetate 126, and C3-acetate 127, were also obtained. These side products are formed as follows (Fig. 1.20b). Upon activation of the glycosyl donor 24, a more stable transient acetoxonium ion 128 was formed. Alcohol 116 attacks the anomeric carbon atom, affording glycosides 124 and 125 (path A). In contrast, when 116 attacks the dioxolenium carbon atom of 128, C2- or C3-acetates (126 or 127) are formed via the isomerization of the resulting 1,2-orthoester intermediate 129 (path B) [51]. Thus, the acetyl group found in 126 and 127 was suggested to originate from the 2′-O-acetyl group in glycosyl donor 24. Such side reactions have also been observed in the total synthesis reported by Anada and Hashimoto [13], and acetylated trichloroacetimidate donors tend to promote orthoester formation, particularly in functionalized substrates or slow glycosylation reactions [52].

To suppress orthoester formation and achieve regio- and stereoselective glycosylation of diol 116, we investigated the coupling of 116 with other glycosyl donors, including glycosyl (N-phenyl)trifluoroacetimidate [53], glycosyl sulfide [54], and glycosyl sulfoxide [55], and their activation conditions. Among the various glycosylations that were examined, the less reactive glycosyl fluoride donor 130 afforded the best results (Fig. 1.21). The treatment of 116 and 130 (2 equiv) with Cp2HfCl2/AgOTf [56] afforded the desired C2-α-glycoside 124 (ca. 50%) without the formation of C3-glycoside 125 and acetates 126 and 127. Because the longer reaction time led to concurrent undesired glycosylation at the C3-alcohol, this reaction was stopped before the full consumption of 116, with the recovery of glucosyl acceptor 116 (43%). The removal of tert-butyl group in 124 proceeded smoothly with trifluoroacetic acid (TFA) without the formation of side products. Finally, subsequent treatment with 1,2-ethylenediamine induced simultaneous deprotection of the five O-acetyl and N-Fmoc groups to furnish BraA (1) (6.8% overall yield in 39 linear steps from 51).

Fig. 1.21
Chemical reactions presenting the final steps in the total synthesis of brasilicardins A and B involving glycosylation using a sugar donor to form an intermediate in 52% yield. The bottom left exhibits an alternative pathway for synthesizing brasilicardin B from the same intermediate.

Regio- and stereoselective glycosylation of glycosyl fluoride and completion of the total synthesis of brasilicardins A and B

Full size image

Unexpectedly, confirmation of the identity of the synthetic material was problematic because its 1H-NMR spectrum was strongly dependent on the pH and concentration of the solvent. However, its identification was confirmed by 1H-NMR measurements of a 1:1 mixture of synthetic and natural BraA (1) after both materials were purified using reverse-phase HPLC. Additionally, the fact that the spectral data (13C-NMR, IR, and HRMS) and optical rotation value of the synthesized compound fully matched those of the isolated natural sample supported the successful synthesis of BraA (1). BraB (2) was synthesized from 122 using the same reaction sequence as that used for 1 (6.5% overall yield in 37 linear steps).

1.11 Stereoselective Glycosylation of Monosaccharides and Completion of the Total Synthesis of Brasilicardins C and D

Encouraged by the successful total synthesis of BraA (1) and BraB (2), we next pursued the total synthesis of BraC (3) and BraD (4) bearing a monosaccharide unit. Contrary to our expectations, this was more difficult and we encountered several issues (Fig. 1.22). Although glycosyl fluoride 130 derived from a disaccharide was effective for the regioselective glycosylation with diol 116 for BraA (1) and BraB (2) (cf. Figure 1.21), a similar glycosylation of peracetylated glycosyl fluoride donor 131 derived from L-rhamnose produced the desired C2-α-glycoside 132 in a low yield. In this reaction, bis-glycosylated product 133 and C3-α-glycoside 134 were obtained as the major products. Although other types of glycosyl donors were examined to improve the C2-selectivity, we found that controlling the regioselectivity of the sterically less-hindered monosaccharide glycosyl donor was challenging. Therefore, we decided to temporarily protect the C3-alcohol in diol 116 as an acetate.

Fig. 1.22
Chemical reactions present an attempt to glycosylate a monosaccharide fluoride to synthesize brasilicardin C. The reaction between a fluorinated monosaccharide and a glycosyl acceptor in the presence of C p 2 H f C l 2 and A g O T f afforded the desired glycosylated product in 25% yield.

Attempted glycosylation of monosaccharide fluoride toward brasilicardin C

Full size image

The required C3-protected glycosyl acceptor 136 was synthesized from diol 116 via regioselective silylation with TBSOTf/2,6-lutidine, acetylation of the resulting alcohol 135, and deprotection of the TBS group (Fig. 1.23). We found that the Au(I)-catalyzed glycosylation [57] between glycosyl o-cyclopropylethynylbenzoate donor 137 and acceptor 136 was the most effective protocol for the installation of L-rhamnose in a stereoselective manner, thereby affording α-glycoside 138 in good yield as a single isomer. Other glycosyl donors did not react with 136, probably due to the steric hindrance of the protected alcohol in 136. The final task to achieve the total synthesis was the deprotection of the protecting groups. We performed the conversion of 138 to BraC (3) using the same procedure (TFA; ethylenediamine) as for BraA (1). However, because the sterically demanding neopentyl C3-acetate remained intact under the abovementioned conditions, the resulting C3-acetate 139 was heated with NaOMe/MeOH at 60 ℃, which induced the epimerization of the amino acid moiety to give an epimeric mixture of BraC (i.e., 140, dr = 55:45).

Fig. 1.23
Chemical reactions present a challenge faced during the synthesis of brasilicardin C. The T B S group is attached to a hydroxyl group on the molecule. The T B S group cleaves when the gold-catalyzed glycosylation reaction with the glucosyl donor is performed.

Au-catalyzed glycosylation toward brasilicardin C and the protecting group problem

Full size image

Among the several protecting groups tested for the C3-alcohol, the easily removable methoxyacetyl group [58] afforded the best results, resulting in the total synthesis of BraC (3) (Fig. 1.24). Thus, the requisite protected alcohol 142 was synthesized from 135 via a two-step reaction sequence, including the protection of the C3-alcohol with MeOCH2COCl, followed by the removal of the C2-TBS group. The Au-catalyzed glycosylation of 142 with glycosyl donor 137 proceeded smoothly to afford α-glycoside 143 as a single isomer. Finally, glycosylation product 143 was successfully converted to BraC (3) via the removal of tert-butyl group with TFA and subsequent simultaneous deprotection of the remaining three O-acetyl, O-methoxyacetyl, and N-Fmoc groups using an aqueous lithium hydroxide (12% overall yield in 42 linear steps from 51).

Fig. 1.24
Chemical reactions present the final steps in the total synthesis of brasilicardin C, which starts with intermediate 141. The first step cleaves a tert-butyl ester group using T F A. The second step involves the global deprotection of several protecting groups using an aqueous lithium hydroxide solution.

Total synthesis of brasilicardin C

Full size image

BraD (4) was synthesized from 122 using the same sequence as that used for BraC (3) (14% overall yield in 40 linear steps) (Fig. 1.25).

Fig. 1.25
Chemical reactions present the final steps in the total synthesis of brasilicardin D, which starts with intermediate 145. The first step cleaves the tert-butyl ester group using T F A and C H 2 C l 2 at 0 degrees Celsius. The second step involves a global deprotection of protecting groups using L i O H solution in methanol at 45 degrees Celsius.

Total synthesis of brasilicardin D

Full size image

1.12 Conclusions

The chemical syntheses of C(sp3)-rich natural products with intriguing three-dimensional structures have inspired a number of developments in novel synthetic strategies and organic transformations. In this chapter, we describe our 10-year synthetic efforts toward brasilicardins, which are unique C(sp3)-rich natural products with a terpenoid–amino acid–saccharide(s) hybrid structure, resulting in the complete synthesis of BraA–D. Notable key features of our total synthesis are: (1) the development of a novel nitrile cyclization, i.e., the stereoselective intramolecular conjugate addition of an α,β-unsaturated Weinreb amide bearing an alkanenitrile unit as a nucleophilic site, which enables carbocycle formation and the construction of contiguous quaternary–tertiary carbon stereocenters simultaneously; (2) a conjugate addition-based synthetic strategy for the stereoselective formation of the highly strained anti-syn-anti-fused 6,6,6-tricyclic skeleton (ABC-ring system); (3) the stereoselective construction of the amino acid moiety to the tricyclic core; and (4) regio- and stereoselective formation of the 1,2-trans-glycosidic linkage to the functionalized aglycon using the appropriate glycosyl donors. In addition, we learned the synthetic utility of nitriles in complex natural product synthesis and the previously unknown but interesting high reactivity of α,β-unsaturated Weinreb amide as a Michael acceptor. We believe that the chemistry described here offers a solution to challenging synthetic problems. In addition, it opens a viable chemical avenue for brasilicardin family natural products and their synthetic derivatives to aid the development of new immunosuppressive agents and to gain a fundamental and better understanding of the therapeutic potential of these compounds.