Inverse Electron Demand Diels—Alder Reactions of Heterocyclic Azadienes, 1‑Aza-1,3-Butadienes, Cyclopropenone Ketals, and Related Systems. A Retrospective

Abstract

A summary of the investigation and applications of the inverse electron demand Diels—Alder reaction is provided that have been conducted in our laboratory over a period that now spans more than 35 years. The work, which continues to provide solutions to complex synthetic challenges, is presented in the context of more than 70 natural product total syntheses in which the reactions served as a key strategic step in the approach. The studies include the development and use of the cycloaddition reactions of heterocyclic azadienes (1,2,4,5-tetrazines; 1,2,4-, 1,3,5-, and 1,2,3-triazines; 1,2-diazines; and 1,3,4-oxadiazoles), 1-aza-1,3-butadienes, α-pyrones, and cyclopropenone ketals. Their applications illustrate the power of the methodology, often provided concise and nonobvious total syntheses of the targeted natural products, typically were extended to the synthesis of analogues that contain deep-seated structural changes in more comprehensive studies to explore or optimize their biological properties, and highlight a wealth of opportunities not yet tapped.

Graphical Abstract

1. INTRODUCTION

The opportunities to provide solutions to complex biological problems using complex natural products have grown as the methods to identify their biological targets have improved and as the techniques used to study, depict, and directly determine their target bound structures have advanced. In addition to the natural compounds themselves, the examination of key synthetic partial structures, compounds that contain deep-seated structural modifications, and even their unnatural enantiomers provide an added and powerful complement to such studies. Well-designed structural modifications in the natural products may be used to address the structural basis of their interaction with biological targets and to define fundamental relationships between structure, activity, functional reactivity, and biological properties. Natural products incorporate multiple properties and functions integrated into a highly functionalized, compact molecule. In many instances, each structural feature and substituent is thought to contribute to the biological activity. However, we recently highlighted examples where even single atom changes to a series of natural products were found to substantially improve on their biological properties.¹ In this work, a challenging problem with natural products is first to understand the subtle design elements that nature has provided and then to rationally extend them to provide more selective, more efficacious, or more potent compounds.

Central to such studies and a major challenge for natural products is the development of synthetic strategies and new synthetic methodology tailored to their structures that permit their preparation and that of key analogues incorporating the deep-seated structural changes. In this Perspective, we summarize our efforts on the development and applications of inverse electron demand Diels—Alder reactions that innocently began nearly 40 years ago and continue to provide solutions to challenging synthetic problems.² The collective efforts have provided a series of unique, nonobvious total syntheses capable of not only affording the targeted natural products but also key or a comprehensive set of structural analogues. Often, this was used to explore the origin of the biological properties of a natural product and to discover compounds with improved activity.

Because an extraordinary amount of work has been conducted on both the mechanistic and synthetic aspects of the Diels—Alder reaction, a full overview is outside the scope of this Perspective. However, a series of comprehensive treatments and reviews have chronicled aspects of its remarkable evolution and central role in synthesis.³ The Diels—Alder reaction⁴ may be classified into one of three types of π2s + π4s cycloadditions that reflect the interacting frontier molecular orbitals governing the rate, regioselectivity, and diastereoselectivity of the reactions: the normal HOMO_diene-controlled reaction, the neutral reaction, and the inverse electron demand or LUMO_diene-controlled reaction⁵ (Figure 1). Shortly after its discovery, the normal Diels—Alder reaction was quickly integrated into the field because of its ability to generate all-carbon six-membered ring systems with predictable regio- and stereochemical control. In contrast, the inverse electron demand Diels—Alder reaction was slower to emerge. The recognition that conjugated diene systems containing nitrogen are electron deficient led to the investigation of the inverse electron demand Diels—Alder reaction and a demonstration of its utility in the preparation of six-membered heterocyclic ring systems. This was first disclosed in 1959⁶ when Carboni and Lindsey published the synthesis of pyridazines and dihydropyridazines through the reaction of 1,2,4,5-tetrazines with dienophiles. In this work, the recognition of the intrinsic electron-deficient character of the heteroaromatic azadienes led to the demonstrated rate acceleration that accompanies the reversal of the traditional electronic properties of diene/dienophile reaction partners. This work also provided the basis for the examination of additional heterocyclic azadienes. Most notable of these are the early studies of Sauer and Neunhoeffer with six-membered heterocyclic azadienes⁷ and the early examination of oxazoles⁸ and related five-membered heterocyclic azadienes.⁹ Herein, we summarize many of our contributions to the field, consisting of the investigations of heterocyclic and acyclic azadienes, cyclopropenone ketals, and related systems and their participation in inverse electron demand Diels—Alder reactions. Initiated at a time the inverse electron demand Diels—Alder reaction was relatively obscure, these studies now span more than 35 years, and some represent the first examples of the application of the reaction to the total synthesis of complex natural products. In most instances, it was the natural product targets and the underlying biological properties that were first chosen for study and that in turn inspired the methodology discovery, development, or application. However, and for the purpose of this Perspective, the work is presented in the context of a series of natural product total syntheses where the methodology served as a key strategic step in the approach and is organized under the type of electron-deficient diene used. In doing so, we hope to illustrate the potential of the methodology for the preparation of complex molecules and to highlight the wealth of transformations yet to be discovered or exploited.

Figure 1.

Classifications of Diels—Alder reactions.

Our initial successes with the approach were sufficiently stunning (streptonigrin, 1983; lavendamycin, 1985; colchicine, 1986; prodigiosin, 1987; CC-1065, 1988)¹⁰ that they inspired our continued examination of new variations on the reactions throughout my career. It led to the introduction of a powerful strategy for divergent (total) synthesis (1984), consisting of late-stage aryl annulations from a common ketone intermediate, that we now routinely use. A unique transition-state anomeric effect was delineated (1991) that is responsible for the remarkable cycloaddition diastereoselectivity of our 1-azadienes, and we disclosed the first Diels—Alder reactions (1982) that entailed catalytic generation of enamine dienophiles and their in situ reaction with 1,2,4-triazines. The studies led to the discovery of the reversible thermal generation of π-delocalized vinylcarbenes from cyclopropenone ketals (1984) and the development of their ensuing cycloaddition reactions. A powerful tandem intramolecular cycloaddition cascade of 1,3,4-oxadiazoles was introduced (2002) that we presently use extensively. Even today, we continue to discover new, as yet, untapped potential for the methodology that highlights its remarkable scope. Our recent discovery of the first general method for catalysis of the inverse electron demand Diels—Alder reactions of heterocyclic azadienes (HFIP hydrogen bonding, 2016) likely expands their scope far beyond what we highlight herein. Similarly, our exploration of previously untapped (e.g., 1,2,3-triazines, 2011) or presently unknown (e.g., 1,2,3,5-tetrazines, unpublished) heterocyclic azadienes promise to provide new opportunities not yet envisioned by the field. We also like to think that our work, which highlighted the remarkable rates and efficiencies of many heterocyclic azadiene Diels—Alder reactions, helped inspire the tetrazine ligation¹¹ and its related biorthogonal conjugation and labeling technology introduced by others and widely used today.

2. HETEROCYCLIC AZADIENES

1,2,4,5-Tetrazines.

Electron-deficient heterocyclic azadienes participate in well-defined inverse electron demand Diels—Alder reactions with electron-rich dienophiles, providing useful reagents for the preparation of highly substituted and functionalized heterocycles that are not as easily accessed by other means.¹² Among them, 1,2,4,5-tetrazine is the most widely employed heterocyclic azadiene, acting as the 4π-electron component in [4 + 2] cycloadditions with a wide range of dienophiles, due to its remarkable reactivity, surprisingly good stability, synthetic accessibility, and broad synthetic applications (Figure 2).¹³

Figure 2.

Inverse electron demand Diels—Alder reactions of 1,2,4,5-tetrazines with dienophiles.

Since the initial disclosure in 1959,⁶ extensive investigations have defined the scope of 1,2,4,5-tetrazine [4 + 2] cycloaddition reactions. Generally, their inverse electron demand Diels—Alder reactions with (hetero)dienophiles proceed through three steps: (1) a [4 + 2] cycloaddition reaction to form an initial bicyclic cycloadduct; (2) a retro [4 + 2] cycloaddition reaction with loss of nitrogen to afford a dihydro-1,2-diazine intermediate; and (3) final aromatization of the resulting intermediate to provide 1,2-diazine products by elimination of an amine, alcohol, or thiol, which is commonly the rate-limiting step. A wide variety of dienophiles are capable of participation in Diels—Alder reactions with the electron-deficient 1,2,4,5-tetrazines and include the following five general categories:¹³ (1) electron-rich dienophiles, e.g. enamines, enol ethers and acetates, enols and enolates, ketene acetals, ynamines, selected heteroaromatics; (2) activated or strained dienophiles, e.g. norbornenes, benzynes, cyclopropenes, cyclobutenes, allenes, trans-cyclooctenes; (3) electron-neutral dienophiles, e.g. simple olefins, styrenes, alkynes; (4) electron-deficient dienophiles, e.g. acrylates, maleic anhydrides, propiolates; (5) heterodienophiles, e.g. imidates, amidines, thioimidates, hydrazones, imines, cyanamides (Figure 3).

Figure 3.

Representative dienophiles that participate in 1,2,4,5-tetrazine [4 + 2] cycloaddition reactions and reactivity patterns.

On the basis of the nature of inverse electron demand Diels—Alder reactions, electron-rich and strained dienophiles, both of which possess an inherent high lying HOMO, usually participate in 1,2,4,5-tetrazine [4 + 2] cycloaddition reactions at room temperature or even lower, whereas neutral olefinic or typical electron-deficient dienophiles require higher reaction temperatures (50−200 °C) and more extended reaction times. Consistent with their impact on lowering the diene LUMO, the rate of 1,2,4,5-tetrazine participation in [4 + 2] cycloaddition reactions with electron-rich dienophiles increases as the electron-withdrawing capabilities of its substituents increase. For instance, Diels—Alder reactions of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with electron-rich or strained dienophiles are often conducted at room temperature or lower.¹⁴ Moreover, it is sufficiently reactive to undergo reactions with unactivated and electron-deficient dienophiles (Figure 3).^10,13

Predictable tuning of both the reactivity as well as the reaction regioselectivity may be made by adjusting substituents on either the 1,2,4,5-tetrazine or the dienophile. The only mode of cycloaddition observed with 1,2,4,5-tetrazines occurs across C3/C6, and we are not aware of any example of a N1/N4 cycloaddition. Early studies of the 1,2,4,5-tetrazines have been comprehensively reviewed elsewhere,^10b,13 including our own work, and focused on symmetrically substituted derivatives. More recent studies have provided approaches to the synthetically more challenging unsymmetrical 1,2,4,5-tetrazines, expanding their synthetic utility.^11c In our studies, the unsymmetrical 1,2,4,5-tetrazines (3, 5−12) were prepared and shown to participate in well-defined and predictably regioselective cycloaddition reactions with a wide range of electron-rich and unactivated dienophiles, providing the corresponding 1,2-diazines in excellent yields (Figure 4).^14,15 However, there are examples of cycloaddition reactions that proceeded with unexpected regioselectivities. For instance, the [4 + 2] cycloaddition reaction of the unsymmetrical 1,2,4,5-tetrazine 6 with ketene diethyl acetal 13 afforded the single regioisomeric product 14 as anticipated. However, the analogous sulfoxide-substituted 1,2,4,5-tetrazine 10 proceeded with a cycloaddition regioselectivity opposite predictions based on simple zwitterionic models or FMO analysis of the [4 + 2] cycloaddition reaction (Figure 4).¹⁶ Although the origin of the reversed regioselectivity remains undefined, it is possible that this may be attributed to a destabilizing steric or electronic interaction of the dienophile substituents with the larger and more electronegative methylsulfinyl substituent.

Figure 4.

Regioselective Diels—Alder reactions of unsymmetrical 1,2,4,5-tetrazines.

The intermolecular Diels—Alder reactions of 1,2,4,5-tetrazines proceed slowly with unactivated dienophiles and typically require higher reaction temperatures and extended reaction times. For example, vigorous reaction conditions (140 °C, 60 h) are required for [4 + 2] cycloaddition of 1,2,4,5-tetrazine 3 with phenylacetylene (Scheme 1).^14b By contrast, the intramolecular variants of such reactions proceed under much milder conditions (23−80 °C, 6−16 h). 1,2,4,5-Tetrazine 17, possessing a terminal unactivated acetylene linked by a three-carbon tether, underwent near-quantitative conversion to the bicyclic 1,2-diazine 18 with formation of a fused six-membered ring when warmed in dioxane at 80 °C (6 h).^14b Likewise, the tethered alkyne 19 underwent a sequential N-acylation and [4 + 2] cycloaddition with formation of a fused five-membered ring at room temperature (16 h), affording 20 (96%).¹⁵

Scheme 1.

Intramolecular Diels—Alder Reactions of 1,2,4,5-Tetrazines Utilizing Tethered Unactivated Acetylenes

Among the first problems addressed in my academic career, we recognized the powerful synthetic utility that the Diels—Alder reactions of 1,2,4,5-tetrazines possessed for the preparation of highly substituted, functionalized heterocyclic systems not easily accessed by traditional approaches. Initially conducted at a time when the inverse electron demand Diels—Alder reaction was an obscure novelty relative to the more conventional normal Diels—Alder reaction, we began an extensive series of investigations that now span more than 35 years to explore and expand the heterocyclic azadiene reaction scope while implementing their use in the total synthesis of complex biologically active natural products. This began at a time when the total synthesis of a natural product required a team of students and still often concluded with the development of an approach to, but not the completion of, the projected total synthesis. As a result, I still find it stunning that our first applications of the methodology (steptonigrin in 1983, CC-1065 in 1988, and prodigiosin in 1987) each entailed the effort of a single, albeit exceptionally talented, graduate student. The efforts also resulted in the delineation of three general strategies for implementing this powerful transformation for the preparation of highly functionalized pyridines, benzenes, and pyrroles, respectively (Figure 5).

Figure 5.

Three general synthetic strategies utilizing the inverse electron demand Diels—Alder reactions of 1,2,4,5-tetrazines.

Since the products of inverse electron demand Diels—Alder reactions of 1,2,4,5-tetrazines are 1,2,4-triazines or 1,2-diazines that may also serve as 4π components for a second [4 + 2] cycloaddition reaction, it was envisioned that highly substituted or functionalized heterocyclic systems found in complex natural products could be assembled by sequential inverse electron demand Diels—Alder reactions. Central to the approach to the total synthesis of streptonigrin was the implementation of a 1,2,4,5-tetrazine → 1,2,4-triazine → pyridine Diels—Alder strategy for convergent synthesis of the natural product through formation of the central and fully substituted pyridine ring. Similarly, we first developed and then applied a 1,2,4,5-tetrazine → 1,2-diazine → benzene Diels—Alder strategy for the total synthesis of natural products bearing a highly functionalized indole or indoline moiety including PDE-I and II, CC-1065, cis- and trans-trikentrin and the Amaryllidaceae alkaloids. Finally, and starting with its use in a total synthesis of prodigiosin, we developed a 1,2,4,5-tetrazine → 1,2-diazine → pyrrole reaction sequence, including a unique penultimate reductive ring contraction reaction of an electron-deficient 1,2-diazine. The latter strategy proved ideally suited for preparation of the densely functionalized pyrrole cores found in a wide range of additional natural products, including isochrysohermidin, ningalins, roseophilin, and lycogalic acid.

Streptonigrin.

Streptonigrin (30) is a historically important potent antitumor antibiotic that contains central to its structure a fully substituted and highly functionalized pyridine.¹⁷ Our convergent total synthesis of streptonigrin was based on the 1,2,4,5-tetrazine → 1,2,4-triazine → pyridine sequential Diels—Alder strategy for construction of this central pentasubstituted pyridine and assemblage of the full skeleton of the natural product. It was the targeted natural product and our interest in defining the source of its biological properties that inspired this methodology development, permitting the incorporation of structural modifications easily within each key region of the molecule.

The first of two sequential Diels—Alder reactions was the [4 + 2] cycloaddition reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with a quinolyl S-methylthioimidate to provide a substituted 1,2,4-triazine. In studies that led to the development of this reaction, no identifiable products could be isolated from the reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with amidines 25 despite an initial exothermic reaction that was accompanied by the evolution of nitrogen (Figure 6).¹⁸ This suggested that a subsequent undesired Diels—Alder reaction of the resulting dihydro-1,2,4-triazine products 21 with unreacted amidine was occurring faster than the aromatization of the penultimate intermediate 21. Moreover, the imidate 24 was found to provide the desired 1,2,4-triazine product 22, albeit in modest yield. In sharp contrast, the [4 + 2] cycloaddition reaction of S-methylthioimidate 23 with 1,2,4,5-tetrazine 1 provided the 1,2,4-triazine cycloadduct 22 in superb yield (70%) under mild reaction conditions (80 °C, dioxane).¹⁸ Beautifully, the Diels—Alder reaction of quinolyl S-methylthioimidate 26 with 1,2,4,5-tetrazine 1 used in the total synthesis of streptonigrin smoothly afforded the 1,2,4-triazine 27 in an excellent yield (82%) (Scheme 2).¹⁹ In this case, the success of the Diels—Alder reaction proved to be dependent on the nature of heterodienophile and attributable to an optimal combination of nucleophilic character of the C═N dienophile (−NR₂ > −OEt = −SMe) balanced against the leaving group ability of its X-substituent (−SMe > −OEt ≫ −NR₂) such that aromatization of the penultimate intermediate effectively competes with further cycloaddition of the intermediate dihydro-1,2,4-triazine.

Figure 6.

Diels—Alder reactions of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with heterodienophiles.

Scheme 2.

Key Steps in the Total Synthesis of Streptonigrin

The second [4 + 2] cycloaddition reaction was that of the 1,2,4-triazine 27 with the pyrrolidine enamine 28 in which addition occurred selectively across C3/C6 of the 1,2,4-triazine with the nucleophilic carbon of the dienophile attached to C3, providing the fully functionalized pyridine core and the entire tetracyclic framework (29) of streptonigrin (Scheme 2).¹⁹ The work culminated in a convergent total synthesis of streptonigrin, requiring only 17 steps. This proved to be roughly half the number of steps used in contemporary efforts at the time (34 steps (Weinreb)²⁰ and 27 steps (Kende)²¹) and remains one of the most concise of the four total syntheses of streptonigrin disclosed to date,^17,22 highlighting the power of the sequential cycloaddition strategy.

In these studies, we prepared and examined a series of key partial structures and analogues of streptonigrin.²³ These and studies of others have provided insights into the structural features of streptonigrin required for expression of its biological activity. Most recently, and with the samples prepared nearly 30 years earlier, we were able to define its biological target, protein arginine deiminase 4 (PAD4), and elucidate the role of these key structural features. We showed that the quinoline-5,8-dione portion of streptonigrin (A and B rings) is required for enzyme inactivation and forms covalent adducts with the enzyme, that the pyridyl C ring and its substituents can significantly impact potency, and that rings C and D convey isozyme selectivity (vs PAD1−3).²⁴

PDE-I, PDE-II, and (+)-CC-1065.

(+)-CC-1065 (40), an antitumor antibiotic isolated from cultures of Streptomyces zelensis, was shown to possess exceptionally potent cytotoxic activity, broad-spectrum antimicrobial activity, and promising in vivo antitumor activity. It is one of the most potent compounds yet identified, being approximately 1000 times more active than typical cytotoxic agents. Today, it is one of the most widely recognized and best characterized DNA-alkylating agents, although none of these features were defined at the time we initiated our studies. The noncovalent DNA binding selectivity and its DNA alkylation reaction may be attributed to two complementary structural features: the identical central and right-hand subunits of CC-1065 are responsible for the high affinity, sequence-selective DNA minor groove binding, and the 4,4-spirocyclopropylcyclohexa-2,5-dienone unit present in the left-hand segment functions as a selective alkylating agent effectively delivered to double-stranded DNA.

PDE I (38) and PDE II (39), two 3′,5′-cAMP phosphodiesterase inhibitors isolated from Streptomyces strain MD769-C6, possess the identical 1,2-dihydro-3H-pyrrolo[3,2-e]indole structure found in CC-1065. The total synthesis of PDE I, PDE II, and (+)-CC-1065 and its unnatural enantiomer, ent-(−)-CC-1065, was accomplished utilizing a 1,2,4,5-tetrazine → 1,2-diazine → benzene Diels—Alder strategy. We envisioned that the central benzene ring of the core 1,2-dihydro-3H-pyrrolo[3,2-e]indole structure could be constructed by an intramolecular Diels—Alder reaction of an alkyne-1,2-diazine which in turn would be derived from the product of an intermolecular inverse electron demand [4 + 2] cycloaddition reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1).

The potential and scope of the intramolecular alkyne−1,2-diazine Diels—Alder reaction was first examined. The results of the studies indicated that 1,2-diazine and alkyne substitution have only subtle effects on the rate of cycloaddition, whereas the length of the diene/dienophile linker chain (n = 1) and the nature of heteroatoms (X = NCO₂Me) in the chain are the primary determinants of the success of the reaction (Figure 7).²⁵ Notably, such unactivated 1,2-diazines are typically reluctant diene participants in intermolecular Diels—Alder reactions, and the entropic assistance provided by the tethered dienophile is sufficient to overcome the barrier to reaction.

Figure 7.

Intramolecular Diels—Alder reactions of alkyne-1,2-diazines.

The total synthesis of PDE-I, PDE-II,²⁶ and CC-1065²⁷ began with the inverse electron demand Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with 4,4-dimethoxy-3-buten-2-one (31) (Scheme 3). Even with the modest reactivity of dienophile 31, the reaction still only required mild reaction conditions (60 °C, 22 h) and provided 1,2-diazine 32 in good yield (71%).²⁶ Reduction of the methyl ketone 32 with sodium borohydride afforded the lactone 33 directly. Subsequent hydrolysis of the remaining C3 methyl ester followed by acidification led to an unexpectedly facile but welcomed room-temperature decarboxylation. This two-step sequence provided 1,2-diazine 34 and completed differentiation of the C3/C6 methoxycarbonyl groups present in 32. Following attachment of the alkyne side chain, the alkyne-1,2-diazine 35 proved reluctant to undergo the desired intramolecular Diels—Alder reaction necessary for indoline ring construction, perhaps because of the geometrical constraints of the alkyne side chain. Thus, hydrolytic ring opening of the N-alkyloxazinone system with potassium hydroxide followed by benzylic alcohol oxidation and N-acetylation afforded the alkyne-1,2-diazine 36 in which the alkyne side chain was no longer constrained. The intramolecular Diels—Alder reaction of 36 proceeded smoothly and provided the indoline 37 in 87% yield. Overall, the sequential Diels—Alder strategy constituted a novel approach by which to construct the key indoline moiety of the natural products. Additional key steps in the synthesis include the implementation of a Hemetsberger indole annulation to install the fused indole-2-carboxylate and a Lewis acid-catalyzed benzylic hydroperoxide rearrangement for installation of the C4 phenol.²⁸

Scheme 3.

Total Synthesis of PDE-I, PDE-II, and (+)-CC-1065

This approach to CC-1065 was also used for the preparation of an extensive series of analogues bearing deep-seated structural modifications in the natural product with the intention of defining the origin of its characteristic DNA alkylation selectivity, establishing the link between DNA alkylation and the ensuing biological properties, and determining the relationships between structure, chemical reactivity, and biological activity.²⁹ These studies now include not only our work on CC-1065^26,27,30 but also duocarmycins A³¹ and SA³² and yatakemycin³³ where we not only conducted total syntheses of the natural products,³⁴ defining their absolute stereochemistry and correcting a misassigned structure (yatakemycin),³³ but also characterized their DNA alkylation properties.³⁵ In these studies, the DNA alkylation selectivity, rates, reversibility,³⁶ and stereoelectronically controlled reaction regioselectivity³⁷ of both their natural and unnatural enantiomers were defined, and we characterized their isolated adenine N3 adducts (Figure 8).³⁸ In fact, the effective and unprecedented behavior of the unnatural enantiomers was key to understanding the source of their DNA alkylation selectivity, which we showed arises from their noncovalent binding selectivity preferentially in the narrower, deeper AT-rich minor groove (shape selective recognition).³⁹ We also defined the beautiful source of catalysis for the DNA alkylation reaction, which arises from a DNA binding induced conformational change that disrupts the vinylogous amide conjugation stabilizing the alkylation subunit (shape dependent catalysis).⁴⁰ NMR-derived high-resolution structures of the natural products and their unnatural enantiomers bound to a common site in DNA were obtained in collaboration with Walter Chazin (Figure 8),⁴¹ and the studies helped established that the compounds undergo a “target-based activation” (induced conformational change) for DNA alkylation.⁴² More than 135 publications and more than 2000 analogues of the natural products (e.g., CBI) have been disclosed in our efforts that were used to define fundamental relationships between structure, reactivity, and activity,⁴³ and their impact on the DNA alkylation properties and biological activity of the natural products.⁴⁴ Included in these, we recently demonstrated the key role the hydrophobic properties of the compounds play, driving the inherently reversible DNA alkylation reaction (hydrophobic binding-driven-bonding), and established the stunning magnitude of the effect.⁴⁵ Data from more than 30 alkylation subunit analogues was used to establish a predictive parabolic relationship between reactivity and cytotoxic potency that spans a 10⁶ range of reactivity and activity.⁴⁶ This defined the optimal balance between reactivity and stability and provided a fundamental design feature that was used to improve the potency of CC-1065 nearly 10-fold with a single atom exchange.⁴⁷ In retrospect and especially given the time frame when such studies were uncommon, it is remarkable that such striking structural simplifications could be made to this class of natural products while still maintaining or even improving their extraordinary potency or selectivity.⁴⁸ Today, the simplified structures and alkylation subunits that we introduced (e.g., CBI)⁴⁴ are being used in the development of targeted therapies including antibody—drug conjugates,⁴⁹ highly selective sequence selective DNA alkylating agents based on Dervan hairpin polyamides,⁵⁰ and unique prodrug designs subject to tumor-selective free drug release.⁵¹ The powerful and now widely used fluorescent intercalator displacement (FID) assay⁵² for establishing DNA binding selectivity or affinity was introduced in these studies and a convenient and more efficient M13-derived alternative to ³²P-end-labeling of restriction fragments for DNA cleavage studies was developed for use in these studies.⁵³

Figure 8.

(Top) Characteristics of the CC-1065 DNA alkylation reaction. (Bottom) Structures of DNA bound (+)-duocarmycin SA and ent-(−)-duocarmycin SA established by NMR.

cis- and trans-Trikentrin A.

In initial studies of the intramolecular Diels—Alder reactions of alkyne-1,2-diazines, the [4 + 2] cycloaddition required reaction conditions (230 °C) and extended reaction times (12−22 h) that we strove to improve. This led to the examination of intramolecular allene-1,2-diazine Diels—Alder reactions and the development of much milder reaction conditions for complementary cycloadditions.¹⁵ Cis- and trans-trikentrin A (47 and 48), isolated from the marine sponge trikentrion flabelliforme, are highly substituted indoles representative of a class of alkaloids that possess antimicrobial activity.⁵⁴ Their total synthesis was accomplished in only five steps based on the implementation of a sequential 1,2,4,5-tetrazine → allene-1,2-diazine → indole Diels—Alder strategy.⁵⁵ The 1,2-diazine cycloadducts derived from the [4 + 2] cycloaddition of 3,6-bis(thiomethyl)-1,2,4,5-tetrazine (2) were found to serve as direct precursors to substituted allene-1,2-diazines that undergo a subsequent intramolecular Diels—Alder reaction (Scheme 4).¹⁵ Moreover, such allene-1,2-diazines participate in [4 + 2] cycloaddition reactions under substantially milder thermal conditions (120 °C) than the corresponding alkyne-1,2-diazines (230 °C). This may be attributed to both the increased reactivity of an allene dienophile and its entropically more favored adoption of a productive conformation for the intramolecular Diels—Alder reaction. The products of the allene-1,2-diazine intramolecular Diels—Alder reaction were designed to be indoles versus indolines by virtue of a cycloadduct tautomerization and a unique thermal elimination of methanesulfinic acid for aromatization. With direct access to the requisite allene-1,2-diazines from the products of the inverse electron demand Diels—Alder reactions of 3,6-bis(thiomethyl)-1,2,4,5-tetrazine (2), this 1,2,4,5-tetrazine → allene-1,2-diazine → indole sequential Diels—Alder strategy constitutes a general approach to synthesis of highly substituted indoles.

Scheme 4.

Sequential 1,2,4,5-Tetrazine → 1,2-Diazine → Benzene Diels—Alder Strategy Featuring an Intramolecular Diels—Alder Reaction of an Allene-1,2-Diazine

These studies provided the basis of an exceptionally concise total synthesis of cis- and trans-trikentrin A. Treatment of 3,6-bis(thiomethyl)-1,2,4,5-tetrazine (2) with the cis/trans mixture of pyrrolidine enamine 41 at room temperature cleanly provided dihydro-1,2-diazine 42 as a single diastereomer, possessing exclusively the cis-dimethyl stereochemistry regardless of the stereochemical integrity of the starting dienophile (Scheme 5).⁵⁵ This selectivity may be attributed to the kinetically preferred Diels—Alder reaction of the cis-isomer through a less sterically encumbered [4 + 2] cycloaddition transition state. Subsequent acid-catalyzed elimination of pyrrolidine provided stereochemically pure 1,2-diazine 43 in superb yield (85% overall). Following oxidation to bis-sulfone 44 and introduction of a single allene side chain through an easily controlled S_NAr displacement reaction, the intramolecular allene-1,2-diazine Diels—Alder reaction of 45 conducted under mild thermal conditions provided N-acetyl cis-trikentrin A (46). In addition, trans-trikentrin A was prepared by an identical sequence following deliberate epimerization of the methyl substituent in 44 under basic conditions.

Scheme 5.

Total Synthesis of cis- and trans-Trikentrin A

Anhydrolycorinone, Hippadine, and Anhydrolycorinium Chloride.

In efforts that extended the synthetic applications of the 1,2,4,5-tetrazine → 1,2-diazine → benzene Diels—Alder strategy, an intramolecular inverse electron demand Diels—Alder reaction of an unsymmetrical 1,2,4,5-tetrazine was employed in the total synthesis of Amaryllidaceae alkaloids (52−54). Their characteristic pyrrolophenathridine skeleton presents an ideal target for the application of the unsymmetrical 1,2,4,5-tetrazine → alkene-1,2-diazine → indoline sequential Diels—Alder strategy.⁵⁶

The unsymmetrical 1,2,4,5-tetrazine 19 was prepared by clean, selective displacement of one methylthio group in 3,6-bis(thiomethyl)-1,2,4,5-tetrazine (2) with 1-amino-3-butyne (Scheme 6). The first intramolecular [4 + 2] cycloaddition was accomplished smoothly at room temperature in superb yield (96%) under N-acylation conditions and the intermediate N-Boc derivative prior to cyclization was not detected. The cycloadduct 20 was N-Boc deprotected and used directly in the amide coupling reaction with 49 to afford 1,2-diazine 50. The second intramolecular [4 + 2] cycloaddition reaction of 50 and subsequent aromatization via elimination of methanol was achieved under more vigorous conditions (265 °C, 20 h) in quantitative yield to provide anhydrolycorinone precursor 51 bearing the fully functionalized pentacyclic skeleton of the Amaryllidaceae alkaloids.

Scheme 6.

Total Synthesis of Anhydrolycorinone, Hippadine, and Anhydrolycorinium Chloride

OMP.

In 1980, Kornfeld and co-workers described the first and a single example of the reductive ring contraction of a dimethyl 1,2-diazine-3,6-dicarboxylate, prepared by a tetrazine Diels—Alder reaction, effected by treatment with Zn/HOAc in the preparation of a dimethyl pyrrole-2,5-dicarboxylate.⁵⁷ The reaction proceeds through initial reduction to a 1,4-dihydrodiazine, subsequent reductive N—N bond cleavage, and final pyrrole cyclization (Scheme 7). Therefore, and in our work, the inverse electron demand Diels—Alder reactions of 1,2,4,5-tetrazines were projected to also provide access to pyrroles through this reductive ring contraction reaction. In our initial exploration of the scope of this reaction, it proved surprisingly tolerant of additional functionality (e.g., ethers, esters, carbamates and ketones), suggesting widespread utilization in organic synthesis.⁵⁸

Scheme 7.

Reductive Ring Contraction of Electron-Deficient 1,2-Diazines to Pyrroles

Our first demonstration of this 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy provided a simple preparation of octamethylporphyrin (OMP, 58).⁵⁸ The synthesis began with a room temperature inverse electron demand Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with 2-triethylsilyloxy-2-butene (55) to afford 1,2-diazine 56 in excellent yield (87%). Treatment of 56 with Zn/HOAc at room temperature smoothly provided the fully substituted pyrrole 57 that serves as the monomeric building block of OMP (Scheme 8). This study was a prelude to use of this strategy in the total synthesis of a family of polypyrroles possessing a common, characteristic pyrrolylpyr-romethene skeleton.

Scheme 8.

Synthesis of OMP

Prodigiosin.

Prodigiosin (66), a blood-red pigment, was the initial member of this class of natural products and displayed useful antimicrobial and potent cytotoxic activity and immunosuppressive properties that we wanted to explore.⁵⁹ It was envisioned that its central pyrrole B ring could be prepared by this 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy.⁶⁰ The room temperature [4 + 2] cycloaddition of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with 1,1-dimethoxyethylene (94%) followed by reductive ring contraction of the resulting 1,2-diazine 59 provided dimethyl 3-methoxy-pyrrole-2,5-dicarboxylate (60) (Scheme 9). A three-step sequence achieved the effective differentiation of the C2/C5 carboxylates and selective removal of the electronically and sterically more accessible C5 methoxycarbonyl group, cleanly affording methyl 3-methoxypyrrole-2-carboxylate (61). A subsequent intramolecular palladium(II)-promoted 2,2′-bipyrrole oxidative coupling reaction conducted with polymer-supported palladium(II) acetate provided the unsymmetrical 2,2′-bipyrrole AB ring system of prodigiosin.⁶⁰ Both the method developed for the synthesis of the starting unsymmetrical N,N′-dipyrrole carbonyl 63, employing pyrrole-1-carboxylic acid and its in situ generated acid chloride 62,⁶¹ and its use in an intramolecular unsymmetrical 2,2′-bipyrrole coupling through two sequential electrophilic palladations and subsequent reductive elimination were unique at the time of our work (1987) and are still innovative in today’s era of undirected C—H activation reactions for unsymmetrical biaryl couplings. Finally, completion of the prodigiosin synthesis was accomplished through the condensation of aldehyde 65 derived from 64 with 2-methyl-3-pentylpyrrole (Scheme 9). In addition to prodigiosin, desmethoxyprodigiosin (67) and prodigiosene (68) were also prepared by this approach. Their comparisons demonstrated that the sequential removal of the peripheral substituents diminished the in vitro cytotoxic activity (Figure 9).⁶⁰

Scheme 9.

Total Synthesis of Prodigiosin

Figure 9.

Comparison of the cytotoxic activity of prodigiosin, desmethoxyprodigiosin, and prodigiosene.

Isochrysohermidin.

In a unique application of the 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy, we applied this approach to the total synthesis of d,l- and meso-isochrysohermidin (72), which we projected were ideal candidates for interstrand DNA cross-linking. Central to our approach was the implementation of two consecutive inverse electron demand Diels—Alder reactions of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with 1,1,4,4-tetramethoxy-1,3-butadiene (69) followed by a double reductive ring contraction reaction for the construction of a fully functionalized 3,3′-bipyrrole key intermediate.⁶² Treatment of 1 with 69 provided the double Diels—Alder cycloadduct 70 in good yield (65%). In a detailed study of this remarkable transformation, the first of the two [4 + 2] cycloaddition reactions was found to proceed rapidly at room temperature. The second was observed to be slower and required more vigorous, albeit still mild, thermal reaction conditions (60 °C) due to the increased steric hindrance or decreased nucleophilic character of the intermediate dienophile. Although the aromatization steps requiring the loss of methanol proved sluggish, these were promoted by added 4 Å molecular sieves. Double reductive ring contraction of symmetrical 4,4′-bis-1,2-diazine 70 upon treatment with activated zinc in glacial acetic acid smoothly afforded the symmetrical pyrrole dimer 71 in good yield (68%). Significantly, this strategy established the key dimeric skeleton of isochrysohermidin through a simple two-step sequence and constitutes a general approach to highly functionalized 3,3′-bipyrroles (Scheme 10).

Scheme 10.

Total Synthesis of Isochrysohermidin

With the dimeric pyrrole skeleton constructed, the completion of the synthesis of isochrysohermidin proved equally concise. Following N-methylation of 3,3′-bipyrrole 71, effective differentiation of the internal and external methyl esters was accomplished through exhaustive ester hydrolysis, cyclic seven-membered anhydride formation, diazomethane esterification of the remaining terminal carboxylic acids, and deliberate anhydride hydrolysis. A final low temperature ¹O₂ [4 + 2] cycloaddition across each of the pyrroles followed by in situ oxidative decarboxylation with endoperoxide fragmentation⁶³ generated isochrysohermidin as a readily separable mixture of d,l- and meso-isochrysohermidin. It is likely that most readers, like us, will view this as not only a concise total synthesis of isochrysohermidin but also a nonobvious one where the core of a nonaromatic five-membered heterocycle was assembled with a heterocyclic azadiene Diels—Alder reaction.

With material prepared by this efficient eight-step synthesis, both d,l- and meso-isochrysohermidin were shown to exhibit the projected interstrand DNA cross-linking properties through a reversible nucleophilic trap at each of the carbinolamides.⁶² This not only represented the discovery of only the fourth class of natural products to exhibit interstrand DNA cross-linking properties but also defined a new and potentially general approach to the development of synthetic DNA cross-linking agents. For instance, incorporation of a single carbinolamide subunit of isochrysohermidin into distamycin, an A—T-rich duplex DNA minor groove binding agent, was explored (Figure 10).⁶⁴

Figure 10.

Isochrysohermidin—distamycin hybrids.

Roseophilin.

Roseophilin (80), a novel antitumor antibiotic, possesses a topologically unique pentacyclic skeleton that consists of a potentially strained macrocycle incorporated in an ansa-bridged azafulvene conjugated with a heterocyclic ring system composed of a substituted furan and pyrrole.⁵⁹ Our approach to roseophilin featured a 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy for construction of an appropriately functionalized pyrrole ring, a ring-closing olefin metathesis reaction for introduction of the 13-membered macrocycle and a 5-exo-trig acyl radical—alkene cyclization for formation of the cyclopentanone found in the upper tricyclic core structure of roseophilin.⁶⁵ The key inverse electron demand Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with the optically active enol ether 73 proceeded effectively at room temperature to provide 1,2-diazine 74 in excellent yield (91%). Reductive ring contraction of 74 effected by treatment with Zn/TFA gave the highly functionalized pyrrole 75 for further incorporation into the precursor 76. Ring closing metathesis of 76 afforded 77, completing the introduction of the macrocycle. Following the transformation to phenyl selenoester 78, acyl radical—alkene 5-exo-trig cyclization of 78 cleanly formed the strained cyclopentanone and generated 79 as a single disastereomer in 83% yield (Scheme 11). Aside from the cycloaddition studies, it was our introduction of phenyl selenoesters as effective precursors to acyl radicals, their slow competitive reduction relative to alkyl radicals, and our demonstration of their synthetic utility and scope of use in inter- and intramolecular alkene addition reactions that preceded its use in this key step in the total synthesis of roseophilin.⁶⁶

Scheme 11.

Total Synthesis of ent-(−)-Roseophilin

Notably, the studies culminated in the total synthesis of ent-(−)-roseophilin, which along with the concurrently disclosed synthesis of Tius and Harrington⁶⁷ permitted the assignment of the absolute configuration of natural (+)-roseophilin. Remarkably, our unnatural enantiomer was found to be 2−10-fold more potent than the natural enantiomer in cancer cell growth inhibition assays, providing an unexpected conclusion to synthetic efforts that provided the unnatural enantiomer.⁶⁵ It also provides an additional valuable tool for defining the mechanism of action and biological target(s) for not only roseophilin but also the structurally related prodigiosins. Incapable of promoting the metal-dependent DNA damage and cleavage reactions observed with the prodigiosins and championed by some,⁶⁸ the observations suggest the antimicrobial activity, cell growth inhibition activity, and immunosuppressive properties of the prodigiosins and roseophilin are derived from alternative and perhaps independent targets. Thus, although the pyrrolylpyrromethene core of the prodigiosins can generate reactive oxygen species when bound by copper and can also function as a chloride ion symporter,⁶⁹ the recent discovery by Harran that tailored fragments of roseophilin and unique roseophilin/prodigiosin hybrids can selectively antagonize Mcl-1 provides an attractive alternative biological target that entails inhibition of a protein—protein interaction.⁷⁰

Ningalins.

Beginning in 1997, members of a growing ningalin family of marine natural products have been isolated, which possess potent Pgp inhibitory activity, effective multi-drug resistance (MDR) reversal properties and potent cytotoxic activity against sensitive and resistant tumor cell lines.⁷¹ Each of these natural products contains a central highly substituted pyrrole nucleus bearing diverse aryl substituents, which make them ideal targets for the 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy (Figure 11). Both the unique reactivity of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) and the electron-rich character of the required diaryl alkynes combine to provide well matched cycloaddition partners for each of the ningalins and the related natural products.⁷²⁻⁷⁴

Figure 11.

Unified strategy for total synthesis of ningalin natural products.

Ningalin A.

The total synthesis of ningalin A (84), the simplest member of this family of marine natural products, began with the inverse electron demand Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with the electron-rich diaryl acetylene 81 to afford the 1,2-diazine 82 in excellent yield (87%).⁷² The relative effectiveness of this transformation may be attributed to the electron-donating properties of the dienophile aryl alkoxy groups which increase the nucleophilic character of the acetylene and improve what is the typically poor reactivity of unactivated alkynes toward 1,2,4,5-tetrazines. Subsequent reductive ring contraction of the 1,2-diazine by treatment with Zn/HOAc smoothly afforded the pyrrole 83 bearing all requisite functionality and the key skeleton of ningalin A (Scheme 12).

Scheme 12.

Total Synthesis of Ningalin A

Lamellarin O and Lukianol A.

The total synthesis of lamellarin O (89) and lukianol A (90) employed a Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with the alternative diaryl acetylene 85 to assemble the fully substituted 1,2-diazine 86 (85%), which was followed by a Zn-mediated reductive ring contraction reaction to provide the corresponding pyrrole core 87 in a highly efficient two-step sequence.⁷² N-Alkylation of 87 with 2-bromo-4′-methoxyacetophenone gave the pentasubstituted pyrrole 88. Selective hydrolysis of the symmetrical diester 88 with 1.3 equiv of LiOH followed by acid-promoted decarboxylation of the resulting monoacid afforded the appropriately substituted and functionalized pyrrole core found in both lamellarin O and lukianol A (Scheme 13).

Scheme 13.

Total Synthesis of Lamellarin O and Lukianol A

Permethyl Storniamide A.

Similarly, the first of the two key transformations in the total synthesis of permethyl storniamide A (95) relied on the Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with the acetylene 91 in toluene to give 1,2-diazine 92 in superb yield (90%).⁷² The unusual facility with which this [4 + 2] cycloaddition reaction occurs may be attributed to the six methoxy groups donating electron density into the alkyne dienophile. Subsequent zinc-promoted reductive ring contraction of 92 afforded the pyrrole 93, which was further converted to the core structure 94 found in the natural product via N-alkylation (Scheme 14).

Scheme 14.

Total Synthesis of Permethyl Storniamide A

Ningalin B.

Complementary to the studies described above, the approach to ningalin B (100) also employed the 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy. The inverse electron demand Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with the unsymmetrical diaryl acetylene 96 proceeded smoothly and afforded 1,2-diazine 97 in excellent yield (92%).⁷³ Reductive ring contraction of 97 afforded the pyrrole core 98 and the subsequent N-alkylation of 98 followed by MOM deprotection with concomitant lactonization generated lactone 99, containing the full skeleton of ningalin B. The total synthesis of ningalin B was completed by the selective hydrolysis of methyl ester 99, decarboxylation, and global deprotection (Scheme 15).

Scheme 15.

Total Synthesis of Ningalin B

Ningalin D.

The most complex of the ningalin marine natural products is ningalin D (105), incorporating a biphenylene quinone methide superimposed on a now oxidized pentasubstituted pyrrole core. The key element of our synthetic approach to ningalin D was a heterocyclic azadiene Diels—Alder reaction of 1 followed by reductive ring contraction of the resulting 1,2-diazine, affording the fully substituted pyrrole core central to the ningalin D structure. The key [4 + 2] cycloaddition of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with the symmetrical alkyne 101 proceeded smoothly in refluxing toluene to provide the 1,2-diazine 102 in superb yield (87%) and was followed by treatment with Zn/TFA to provide the pyrrole core 103.⁷⁴ Although such reductive ring contraction reactions are typically conducted with Zn/HOAc, we disclosed that the overall reaction is much faster (4−7 vs 24 h for 102, 25 °C) with Zn/TFA and often much cleaner (Scheme 16). Additional highlights of the synthesis include a double Dieckmann cyclization of 104 to close the C and D aryl rings, and a subsequent highly effective and sterically demanding Suzuki coupling of the corresponding C and D phenol triflates for introduction of the F and G aryl rings. Perhaps the most innovative of the additional key steps was a penultimate and unusually effective formal oxidative decarboxylation reaction that entailed Curtius rearrangement of a dicarboxylic acid and subsequent in situ air oxidation to directly provide the biphenylene quinone methide found central to the structure of ningalin D.⁷⁴

Scheme 16.

Total Synthesis of Ningalin D

In the course of our efforts on the ningalins, a large series of structural analogues were prepared by this approach in a concise and efficient manner (5−9 steps, 20−50% overall yield). These studies enabled us to conduct a systematic biological evaluation of the ningalin family and their synthetic analogues.⁷²⁻⁷⁵ Although most possess modest to low inherent cytotoxic activity, many were found capable of reversing the multidrug-resistant (MDR) phenotype, resensitizing resistant tumor cell lines such as HCT116/VM46 to vinblastine and doxorubicin at a lower dose than the prototypical agent verapamil. Notably, each of the active compounds was found to incorporate three hydrophobic domains that characterize the Pgp binding pharmacophore models, but unique in the ningalin series is the lack of a basic amine central to the structure. As such, they not only constitute efficacious compounds exhibiting unusually potent MDR reversal activity for a lead series, but they also depart from structural features considered central to common pharmacophore models of Pgp binding.

Lycogarubrin C and Lycogalic Acid.

Lycogalic acid (also known as chromopyrrolic acid, 113), first isolated from Lycogala epidendrum, has been identified as a common intermediate in the biosynthesis of the indolo[2,3-a]carbazole alkaloids, including rebeccamyin and staurosporine. Complementary to the synthetic studies on the ningalin family, we envisioned that lycogalic acid and its methyl ester, lycogarubin C (112), would be readily accessible through use of a 1,2,4,5-tetrazine → 1,2-diazine → pyrrole Diels—Alder strategy.⁷⁶ An initial approach explored the inverse electron Diels—Alder reaction of dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) with acetylene 106 and provided 1,2-diazine 107 in good yield (65%). However, the reaction was slow, requiring 15 d in refluxing toluene (110 °C). Compared to the productive dialkoxyphenyl acetylenic dienophiles that were effectively used in the total synthesis of ningalins, the lower reactivity of 106 may be attributed to its less electron-rich character. Therefore, the more electron-rich acetylene, 1,2-bis-(tributylstannyl)acetylene (108), was adopted in an alternative approach. The Diels—Alder reaction of 108 with dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1) proceeded smoothly under remarkably mild conditions (45 °C, dioxane, 24 h), providing 1,2-diazine 109 in exceptional yield (97%). Stille coupling of 1,2-diazine 109 with 110 proceeded effectively and twice, affording the same key 4,5-bis(indol-3-yl)-1,2-diazine 107 in good yield (70%). Treatment of 107 with Zn/HOAc cleanly promoted the reductive ring contraction reaction, providing pyrrole 111 containing the core structure of the natural products and completing the 1,2,4,5-tetrazine → 1,2-diazine → pyrrole conversions originally envisioned (Scheme 17).

Scheme 17.

Total Synthesis of Lycogarubrin C and Lycogalic Acid

1,2,4-Triazines.

Similar to 1,2,4,5-tetrazines, 1,2,4-triazines participate effectively as electron-deficient heterocyclic azadienes in inverse electron demand Diels—Alder reactions with electron-rich dienophiles, such as enamines, ynamines, and strained olefins.⁷⁷ The [4 + 2] cycloaddition of the parent 1,2,4-triazine occurs across C3/C6 with the nucleophilic carbon of the electron-rich olefins attached to C3. Loss of N₂ from the initial [4 + 2] cycloadduct followed by aromatization provides substituted pyridines in which the C3 and C4 substituents are derived from the dienophile. The only exceptions to this mode of cycloaddition are ynamines, which preferentially add across C5/N2 to provide pyrimidines.⁷⁸ Typically, the Diels—Alder reactions proceed under mild conditions (25−80 °C), and the rate-limiting step of the overall reaction is often not the [4 + 2] cycloaddition or subsequent loss of nitrogen, but rather the aromatization step.⁷⁹ Early in our studies, we developed a “catalytic” inverse electron demand Diels—Alder reaction of in situ generated enamines with 1,2,4-triazines, which constitutes a novel variant of the conventional Diels—Alder reaction. Catalytic and in situ generation of a pyrrolidine enamine in the presence of a 1,2,4-triazine precedes the ensuing inverse electron demand Diels—Alder reaction which is followed by the immediate loss of nitrogen and finally aromatization with regeneration of the catalytic pyrrolidine, completing a mild, one-flask pyridine annulation (Figure 12).⁸⁰ Additives (e.g., 4 Å MS) used to promote in situ enamine formation often also improved the slow aromatization step such that the catalytic version of the reaction provided higher reaction yields than the stoichiometric reaction. Notably, it also represented one of the first examples (1982) of the enamine activation mode of organo-catalysis and the first employed for a cycloaddition.

Figure 12.

1,2,4-Triazine—enamine [4 + 2] cycloaddition.

Added electron-withdrawing substituents on the 1,2,4-triazine nucleus affect the rate of cycloaddition, influence the mode of cycloaddition (C3/C6 vs C5/N2), and may control or alter the cycloaddition regioselectivity (Figure 13). 3,5,6- Tris(ethoxycarbonyl)-1,2,4-triazine (114) behaves in a manner analogous to 1,2,4-triazine itself, but it is more reactive by virtue of its enhanced electron-deficient character.⁸¹ However, the increased reactivity and the noncomplementary substitution with C3 and C5 electron-withdrawing substituents does result in the occasional observation of lower reaction regioselectivity. The single noncomplementary C3 substitution with a strong electron-withdrawing substituent serves to reverse the [4 + 2] cycloaddition regioselectivity without altering the mode of cycloaddition (C3/C6 vs C5/N2), but it does diminish the rate of reaction. Similarly, the complementary substitution with a single C6 electron-withdrawing substituent increases the rate, regioselectivity, and mode (C3/C6) of cycloaddition.

Figure 13.

1,2,4-Triazine substituent effects.

Streptonigrin.

As highlighted earlier, streptonigrin (30) is a potent antitumor antibiotic that contains a fully substituted and highly functionalized pyridine. Central to our synthetic approach to streptonigrin was a 1,2,4,5-tetrazine → 1,2,4-triazine → pyridine iterative Diels—Alder strategy for construction of this fully functionalized pentasubstituted pyridine core.¹⁹ The first of two Diels—Alder reactions, the [4 + 2] cycloaddition of thioimidate 27 with dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate (1), was presented in the section on 1,2,4,5-tetrazines. The second [4 + 2] cycloaddition reaction was that of the product 1,2,4-triazine 27 with the enamine 28 for the pyridyl C ring construction. In our preliminary investigation of this transformation, 3,5,6-tris(ethoxycarbonyl)-1,2,4-triazine (114) underwent rapid reaction with the pyrrolidine enamine 28 under mild thermal conditions (CHCl₃, 45 °C, 3 h) with addition preferentially occurring across C3/C6 of the 1,2,4-triazine nucleus and the nucleophilic carbon of the dienophile attached to C3 (Figure 14).⁸² This reaction provided the fully substituted pyridine 115 in good yield and regioselectivity (59%, 9:1). However, the Diels—Alder reaction of the pyrrolidine enamine 28 with 1,2,4-triazine 27 under thermal conditions failed to deliver the desired product due in part to the thermal instability of the dienophile. Although the morpholino enamine 116 provided a sluggish addition with 1,2,4-triazine 27 and the desired product was obtained, the vigorous reaction conditions (CHCl₃, 120 °C, 42 h) required for complete reaction reduced the observed regioselectivity (68%, 1:1). All efforts to promote the [4 + 2] cycloaddition of the morpholino or pyrrolidine enamines 116 and 28 with 1,2,4-triazine 27 with conventional Lewis acids were unsuccessful and only led to the consumption of the electron-rich olefin with no evidence of Diels—Alder catalysis. However, it was found that the use of pressure-promoted Diels—Alder reaction conditions were useful in increasing the rate of cycloaddition at mild reaction temperatures (25 °C). As summarized in Figure 14, the pressure-promoted (6.2 kbar, 25 °C) cycloaddition of either enamine 28 or 116 with 1,2,4-triazine 27 provided the desired cycloadducts 29/117 with a preference for the desired regioisomer 29 (1.4:1 and 2.8:1 29/117, respectively) in acceptable yields (58% and 65%). Under the best conditions examined, the desired product 29 bearing the full skeleton with the requisite functionality of streptonigrin was isolated in nearly 50% yield from the reaction of 27 with 28 (6.2 kbar, 25 °C) (Figure 14).¹⁹ Today, this would be a reaction especially well-suited for H-bonding catalysis with HFIP or TFE. Based on the sequential implementation of two inverse electron demand Diels—Alder reactions, our convergent synthesis of streptonigrin (17 steps) remains one of the shortest total synthesis disclosed to date.^17,20−22

Figure 14.

1,2,4-Triazine—enamine cycloaddition in the total synthesis of streptonigrin.

Lavendamycin.

Lavendamycin (122) is a highly substituted and highly functionalized quinolone-5,8-quinone structurally and biosynthetically related to streptonigrin.⁸³ It was envisioned that the central pyridine C ring could be prepared through an inverse electron demand Diels—Alder reaction of 1,2,4-triazine 114 with the pyrrolidine enamine 118.^84,85 Cycloaddition of the enamine 118 occurred across C3/C6 of 3,5,6-tris(ethoxycarbonyl)-1,2,4-triazine (114) with the nucleophilic dienophile carbon attached to C3 and provided the fully substituted 4-arylpyridine 119 central to the structure of lavendamycin in good yield (50%). It is notable that the 1,2,4-triazine triester substitution permitted the reaction to be conducted at room temperature and proceeded with good regioselectivity (6−8:1) (Scheme 18). The completion of the total synthesis of lavendamycin, complementary to four others that emerged near simultaneously,¹⁷ was based on a clever differentiation of the three ethyl esters, an innovative Pd(0)-mediated amine aromatic substitution reaction for closure of the β-carboline (121)⁸⁵ and a late-stage Friedlander condensation. Among these key transformations, the Pd(0)-mediated amination for construction of the β-carboline (121) is one we are especially proud of, having been conducted while I was still a beginning faculty member in a small program. At a time the mechanism of even the Ullmann and Goldberg coupling reactions were unresolved, the reaction was designed to proceed by Pd(0) oxidative insertion into the aryl bromide (120), formation of a six-membered Pd(II) complex with coordination of the C3 pyridyl amine, and subsequent reductive elimination of Pd(0) with nitrogen—carbon bond formation.⁸⁵ This constituted the first example of a palladium-catalyzed preparation of aryl amine with carbon—nitrogen bond formation, demonstrated that reductive elimination of nitrogen with nitrogen—carbon bond formation was possible, defined the viable Pd(0)/Pd(II) catalytic cycle and ultimately provided the foundation for the development of the Buchwald—Hartwig reaction in studies initiated 10 years later.

Scheme 18.

Key Steps in the Total Synthesis of Lavendamycin

Phomazarin.

Phomazarin (125) is the most widely known and extensively studied aza anthraquinone whose original structural assignment was revised twice. Our total synthesis, which served to unambiguously establish the structure of phomazarin, was based on the implementation of a heteroaromatic azadiene Diels—Alder reaction (1,2,4-triazine → pyridine) for preparation of the fully substituted and appropriately functionalized pyridine C ring.⁸⁶ The [4 + 2] cycloaddition reaction of 3,5,6-tris(ethoxycarbonyl)-1,2,4-triazine (114) with 1,1,2-trimethoxyethylene (123) proceeded effectively at room temperature and typically provided a mixture of both unaromatized and aromatized cycloadducts, which was fully converted to the pyridine product 124 at higher reaction temperatures (100 °C, dioxane, 30 min) and by the brief Et₃N treatment (25 °C, 1 h). The pyridine product 124 was obtained in excellent yield (85%) and was ideally functionalized for elaboration to the pyridine C-ring of phomazarin (Figure 15). The dienophile methoxy substituents not only served to enhance its nucleophilic character but also served to introduce the phomazarin C3 and C4 hydroxy groups. In a complementary manner, the three electron-withdrawing ester substituents on the 1,2,4-triazine nucleus increased its [4 + 2] cycloaddition reactivity toward such electron-rich dienophiles and provided the necessary functionality for introduction of the natural product C2 carboxylic acid and C9/C10 quinone carbonyls. Regioselective nucleophilic addition of the A-ring to the C-ring C6 carboxylate followed by Friedel—Crafts closure of the B-ring provided the fully functionalized phomazarin skeleton. Our synthetic work confirmed the latest structural assignment for phomazarin and provided a preliminary assessment of the natural product and closely related synthetic analogues. The biological evaluation in a cancer cell growth inhibition assay revealed that phomazarin was inactive, but that two of its immediate precursors, esters 126 and 127, exhibited potent cytotoxic activity.

Figure 15.

Key step in the total synthesis of phomazarin and biological evaluation of phomazarin and its analogues.

1,3,5-Triazines.

1,3,5-Triazines are electron-deficient six-membered heteroaromatic rings that participate in well-defined [4 + 2] cycloaddition reactions with electron-rich dienophiles, such as ynamines, enamines, N,O-ketene acetals, and amidines (Scheme 19).⁸⁷ Typically, the initial Diels—Alder reaction is rapid and occurs at room temperature, and the slow step in the overall reaction cascade is the retro-Diels—Alder loss of a nitrile followed by aromatization. The addition of electron-withdrawing substituents to the 1,3,5-triazine nucleus serves to accelerate both the rate of the initial Diels—Alder reaction as well as this subsequent retro-Diels—Alder reaction of the cycloadduct.

Scheme 19.

1,3,5-Triazine—Ynamine and 1,3,5-Triazine—Enamine Diels—Alder Reactions

2,4,6-Tris(ethoxycarbonyl)-1,3,5-triazine (128), like 1,3,5-triazine itself, participates in [4 + 2] cycloaddition reactions at room temperature with the more reactive electron-rich dienophiles. The [4 + 2] cycloaddition of 128 with ynamines proceeds at room temperature and is followed by a retro-Diels—Alder reaction (40−100 °C) with loss of ethyl cyanoformate to provide the corresponding substituted pyrimidines.⁸⁸ In contrast, the [4 + 2] cycloaddition reaction of 128 with enamines (25−100 °C) provides cycloadducts that more slowly undergo a retro-Diels—Alder reaction and subsequent aromatization. Although efforts to promote the retro-Diels—Alder reaction/aromatization proved modestly successful, the reaction may be effectively carried out by either subsequent acid treatment (4 N HCl or HOAc) or even by simply conducting the reaction under acidic conditions (CH₂Cl₂−HOAc) that serve to facilitate the retro-Diels—Alder reaction and subsequent aromatization (Scheme 19).⁸⁸

In addition, we found that amidines effectively participate in [4 + 2] cycloaddition reactions with 1,3,5-triazines.⁸⁹ The unique cycloaddition reaction of amidines with 1,3,5-triazines proceeds through in situ amidine to 1,1-diaminoethene tautomerization and its [4 + 2] cycloaddition with the 1,3,5-triazine and is followed by imine generation through loss of ammonia from the Diels—Alder cycloadduct, imine to enamine tautomerization, and retro-Diels—Alder loss of ethyl cyanoformate to provide the product pyrimidine. In our studies, this reaction proved general for a wide range of amidines and can be effectively conducted using either the 1,3,5-triazine or the amidine as the limiting reagent, indicating that the in situ trap of the 1,1-diaminoalkene derived from amidine tautomerization is unusually efficient. Moreover, in addition to the differences in the relative ease of preparation of amidines versus ynamines, especially where R = H, amidines may be employed without deliberate introduction of an amine protecting group and permit the use of cyclic dienophiles not accessible with ynamines (Figure 16).

Figure 16.

1,3,5-Triazine—amidine Diels—Alder reactions. 1,3,5-Triazine and dienophile substituent effects.

Consistent with expectations, addition of electron-withdrawing substituents to the 1,3,5-triazine facilitates cycloaddition, and the more nucleophilic dienophiles derived from amidine tautomerization participate more readily than those derived from thioimidates or imidates.⁸⁹ Moreover, it was shown that the presence of electron-withdrawing substituents serve to accelerate the retro-Diels—Alder reaction of the [4 + 2] cycloadducts. The Diels—Alder cycloadducts of the parent 1,3,5-triazine or 2,4,6-tris(ethoxycarbonyl)-1,3,5-triazine (128) typically undergo the retro-Diels—Alder reaction under mild thermal conditions (40−100 °C). In contrast, the [4 + 2] cycloaddition reaction of ynamines with 2,4,6-tris(methylthio)-1,3,5-triazine (129) provides stable [4 + 2] cycloadducts, which undergo the retro-Diels—Alder reaction to provide the substituted pyrimidines under subsequent thermal (150−230 °C) or acid-catalyzed (p-TsOH, 100 °C) reaction conditions. In addition, the retro-Diels—Alder reaction may be accelerated by treatment with m-CPBA to provide the methanesulfonyl derivatives that undergo in situ retro-Diels—Alder reaction with loss of methanesulfonylcyanate at room temperature (Figure 16).^88,89

Pyrimidoblamic Acid and Bleomycin A₂.

Pyrimidoblamic acid (135) is the heteroaromatic chromophore of bleomycin A₂ (136), a potent and clinically effective antitumor agent that derives its properties through a metal-dependent sequence-selective cleavage of duplex DNA. Our initial approach to the pyrimidoblamic acid subunit was based on two complementary [4 + 2] cycloaddition reactions of 2,4,6-tris(ethoxycarbonyl)-1,3,5-triazine (128) for the construction of a fully substituted pyrimidine nucleus.⁹⁰ The first of these constitutes a Diels—Alder reaction of 1,3,5-triazine 128 with ynamine 130 in which the [4 + 2] cycloaddition occurs at room temperature, and it was the retro-Diels—Alder collapse of the cycloadduct that required the more vigorous thermal conditions (95%, 100 °C) (Scheme 20). Acid-catalyzed debenzylation of the resulting pyrimidine 131 (TfOH, CH₂Cl₂, 40 °C) provided 4-amino-pyridimine 132 in good yield (75%).⁹⁰

Scheme 20.

Total Synthesis of (−)-Pyrimidoblamic Acid and the Key Steps in Total Synthesis of Bleomycin A₂

Alternatively, 132 was derived directly and more conveniently in one step by treatment of 1,3,5-triazine 128 with propionamidine hydrochloride (133) (DMF, 100 °C) in a reaction cascade that proceeded with thermal tautomerization of 133 to 1,1-diaminopropene and its [4 + 2] cycloaddition with 128 (Scheme 20).⁹⁰ The sequential elimination of ammonia, imine to enamine tautomerization, and subsequent retro-Diels—Alder loss of ethylcyanoformate under the reaction conditions provided 132 directly in excellent yield (80%). The reaction was found to proceed best in the polar aprotic solvent DMF under modest thermal reaction conditions (100 °C), and the amidine hydrochloride salts were shown to routinely provide higher yields of the product pyrimidines than the corresponding free base. It is likely that the tautomerization of 133, retro-Diels—Alder reaction and the subsequent aromatization reaction were facilitated by both the presence of HCl derived from the amidine hydrochloride salts and the use of a polar, aprotic solvent under the thermal reaction conditions (>80 °C).

A remarkably simple differentiation of the [4 + 2] cycloadduct ethyl esters was accomplished by treatment with NaBH₄ (0 °C) in which the sterically more accessible and electronically more reactive pyrimidine C2 ester was selectively reduced. Following conversion to the aldehyde and imine formation, diastereoselective addition of the stannous enolate of an Evans’ N-acyloxazolidinone provided predominately the anti imine addition product 134 bearing the absolute configuration of the natural product (anti/syn = 87:13).⁹¹ In this work, the absolute stereochemistry of the C2 benzylic center was unambiguously established and served to confirm the original but unconfirmed assignment of Umezawa.

The additional key strategic elements of our approach to bleomycin A₂ include two highly diastereoseletive glycosylation reactions through the use of diphenyl glycosylphosphates as glycosyl donors in the synthesis of both the disaccharide and the glycosidated β-hydroxyl-l-histidine.⁹² Finally, sequential amide coupling of the glycosidated β-hydroxyl-l-histidine subunit with tetrapeptide S and (−)-pyrimidoblamic acid provided bleomycin A₂ in a highly convergent manner. The intact sulfonium salt and only two protecting groups were incorporated into the final coupling substrates simplifying the final stages of the total synthesis.

(+)-P-3A.

Analogous to our approach to pyrimidoblamic acid, (+)-P-3A (139) and a series of structurally related agents were prepared using the [4 + 2] cycloaddition reaction of 2,4,6-tris(ethoxycarbonyl)-1,3,5-triazine (128) with in situ generated 1,1-diaminoethylene derived from thermal tautomerization of acetamidine (85%).⁹³ The corresponding ynamine [4 + 2] cycloaddition was not examined due to its synthetic inaccessibility. Analogous to the total synthesis of (−)-pyrimidoblamic acid and following strategic differentiation of two ethyl esters and imine formation, the subsequent use of a diastereoselective N-acyloxazolidinone enolate—imine addition reaction provided the stereocontrolled introduction of the pyridimidine C2-acetamido side chain. Finally, without protection of the unreactive arylamine or hindered secondary amine of 137, direct amide coupling of 137 with the dipeptide 138 followed by acid-promoted global deprotection afforded (+)-P-3A (Scheme 21).

Scheme 21.

Total Synthesis of (+)-P-3A

Bleomycin A₂ acts through the sequence-selective cleavage of DNA by a reaction that is both metal-ion and O₂ dependent.⁹⁶ Approximately 70 analogues of the natural product^90,92−94 that probed each subunit and nearly every substituent in the structure were prepared through use of this modular total synthesis of bleomycin A₂.⁹⁵ Our studies revealed that nearly every substituent and functional group in the natural product is integral to its activity.⁹⁵ Key among the studies were those that helped establish that a G triplex-like H-bonding in the minor groove is responsible for the DNA cleavage selectivity,⁹⁷ and that fundamental conformational properties within the linker region of bleomycin contribute to the efficiency of DNA cleavage.⁹⁸ An NMR-derived high resolution structure of DNA bound deglycobleomycin A2 was established in collaboration with JoAnne Stubbe (Figure 17).⁹⁹

Figure 17.

(Top) Functional roles of bleomycin A₂ subunits. (Middle) DNA cleavage selectivity of bleomycin A₂ derived from G-triplex-like H-bonding. (Bottom) Structure of Co^III−OOH bleomycin A₂ bound to DNA determined by NMR.

These studies defined a remarkable integration of the functional and conformational features of the natural product into a compact biologically active compound.⁹⁵ It is the pyrimidine C4 amino group and pyrimidine N2 that H-bonds with G at a cleavage site, providing the anchoring DNA interaction and recognition responsible for the DNA cleavage selectivity.⁹⁷ The combined linker substituents restrict the compound to a now predictable dominant compact conformation, preorganizing the molecule into a single rigid conformation productive for DNA cleavage by the bound complex (Figure 17).^98,100

1,2,3-Triazines.

1,2,3-Triazines, or v-triazines, were unknown until Smolinsky’s synthesis of 4,5,6-triphenyl-1,2,3-triazine through the thermolysis of an azidocyclopropene in the course of his studies on nitrenes in the 1960s.¹⁰¹ 1,2,3-Triazines are electron-deficient, six-membered aromatic heterocycles containing three contiguous nitrogen atoms. They represent the least explored of the isomeric triazines. In part, this may be attributed to the early challenges associated with their preparation combined with their poor cycloaddition scope in initial studies.¹⁰² Igeta provided a convenient synthesis,¹⁰³ entailing the oxidative (NaIO₄) ring expansion of N-aminopyrazoles, but the scope of their cycloaddition reactions remained limited. In 2011, there were only a handful of substituted monocyclic 1,2,3-triazines known at the time we initiated our efforts.^101−104 We conducted a systematic study of the reactivity of 1,2,3-triazines in inverse electron demand Diels—Alder reactions, including an examination of the impact of a C5 substituent (Figure 18).¹⁰⁵ The C5 substituents were found to predictably exhibit a remarkable impact on the 1,2,3-triazine cycloaddition reactivity and further enhanced their intrinsic C4/N1 cycloaddition regioselectivity. Moreover, the impact of the addition of a complementary C5 electron-withdrawing substituent (R = NO₂ > CO₂Me > Ph > H)¹⁰⁵ was of a magnitude that it converted the modest cycloaddition scope of the parent 1,2,3-triazine into a heterocyclic azadiene system with a broad synthetic utility, substantially expanding the range of participating dienophiles and enhancing the rate of productive cycloaddition. Ynamine, enamine, ketene acetal, enol ether, and enol acetate cycloadditions proved effective, the 5-carbomethoxy and especially the 5-nitro-1,2,3-triazine display an even larger scope of productive dienophiles, and a newly discovered amidine or imidate cycloaddition^105,106 was found to proceed with extraordinary rates and efficiencies. These studies were extended to systematic examinations of C5 substituted electron-rich and electron-poor 1,2,3-triazines bearing noncomplementary substitution patterns,¹⁰⁶ providing a comprehensive understanding of a substituent’s impact on the scope, reactivity, and regioselectivity of 1,2,3-triazine cycloaddition reactions. Notably, even the electron-rich 1,2,3-triazines participate in the newly discovered and effective cycloaddition reactions with amidines. Combined, the studies provided an additional new heterocyclic azadiene, complementary to the isomeric 1,2,4-triazines and 1,3,5-triazines, that participates in powerful inverse electron demand Diels—Alder reactions and extends the heterocyclic ring systems accessible with the methodology. Additional observations led to the development of a method for the synthesis of β-aminoenals through ring-opening of the parent 1,2,3-triazine with secondary amines (Figure 18).¹⁰⁷

Figure 18.

1,2,3-Triazine Diels—Alder reactions.

Finally, in the course of recent work that provided a total synthesis of methoxatin, we discovered the first general method for catalysis of the inverse electron demand Diels—Alder reactions of heterocyclic azadienes that entails hydrogen bonding activation by hexafluoroisopropanol (HFIP, Figure 19).¹⁰⁸ This development likely expands the scope of all heterocyclic azadiene cycloadditions far beyond what we highlight herein, analogous to the impact Lewis acid catalysis had on the normal Diels—Alder reaction.

Figure 19.

HFIP-promoted inverse electron demand Diels—Alder reactions of triazines. Yields in parentheses are those obtained with chloroform as solvent.

Pyrimidoblamic Acid and P-3A.

Bleomycin A₂ is the major component of the clinical anticancer drug blenoxane. Although currently used in the clinic,^95,96,109 treatment with bleomycin is often limited by dose-dependent pneumonitis, lung fibrosis, and skin toxicities.¹¹⁰ The discovery of bleomycin-based analogues that lack off-target toxicity would improve their tolerability. Each structural unit of bleomycin plays an important role in the DNA binding and cleavage reaction that is responsible for its biological effects (see Figure 17). From a series of studies, (−)-pyrimidoblamic acid and P-3A were identified as key subunits for the preparation of improved and/or simplified bleomycin analogues.¹¹¹

Three prior total syntheses of (−)-pyrimidoblamic acid had been disclosed, including our own, and typically suffered from an inability to (fully) control the stereochemistry at the benzylic pyrimidine C2 tertiary center.^90,112 Our synthesis, like that of P-3A, utilizing a 1,3,5-triazine cycloaddition (see Scheme 20) relied on a late-stage introduction of this center in a reaction that proceeded with modest diastereoselectivity (87:13). We envisioned second-generation total syntheses of (−)-pyrimidoblamic acid and P-3A in which all necessary stereochemistry would be installed prior to a late-stage 1,2,3-triazine [4 + 2] cycloaddition for introduction of the pyrimidine.¹¹³ This would address the limitation inherent in previous approaches, while also permitting late-stage divergent modification of both the natural product side chain or core heterocycle.

Our second-generation total synthesis of (−)-pyrimidoblamic acid enlisted a [4 + 2] cycloaddition reaction between a trisubstituted 1,2,3-triazine and an appropriately functionalized amidine (Scheme 22). Consistent with our prior studies, the cycloaddition reaction was found to proceed at room temperature or lower, and the amidine was found to add exclusively across C4/N1 (C6/N3) with no evidence of a redirected C5/N2 mode of cycloaddition. Cycloadduct 142 was produced as a single diastereomer with full control of the side chain stereochemistry in 54% yield (5 °C, 14 h then 25 °C, 6 h). The synthesis of (−)-pyrimidoblamic acid was completed in seven steps, involving protection of the secondary amine, hydrolysis, and Curtius rearrangement of the ethyl ester, removal of the acetonide protecting group followed by Jones oxidation, amidation of the resulting carboxylic acid with ammonia and HOBt/EDCI, and a final global deprotection.

Scheme 22.

Total Synthesis of (−)-Pyrimidoblamic Acid

Application of the key 1,2,3-triazine cycloaddition to the synthesis of P-3A might appear straightforward, but it was not clear what the impact of the 1,2,3-triazine C5 substituent might be. A study of the relative cycloaddition reactivity of the C5 substituted and unsubstituted triazines was conducted, and the unsubstituted 1,2,3-triazine 144 was found to be even more reactive than its C5-methyl counterpart, providing the product pyrimidine at a faster rate and in a higher yield, with appreciable conversions observed in as little as 5 min. Importantly and despite the lack of a C5 substituent, 144 showed no evidence of competitive cycloaddition across C5/N2 (Scheme 23).¹¹³

Scheme 23.

Total Synthesis of P-3A

The key cycloaddition reaction of 144 with amidine 140 proceeded smoothly, providing the desired pyrimidine in 76% yield as a single diastereomer (25 °C, CH₃CN, 12 h, Scheme 23).¹¹³ Pyrimidine 145 was elaborated to P-3A in a route modeled on that used for (−)-pyrimidoblamic acid. Amine Boc-protection, saponification of the ethyl ester, Curtius rearrangement, and removal of the acetonide protecting group afforded amino alcohol 147, which was transformed to the primary carboxamide in two steps. Saponification of the tert-butyl ester and coupling with N^Im-Boc-l-His-l-Ala-Ot-Bu provided the fully assembled and protected P-3A, which was converted to the natural product in a global deprotection using TFA—CH₂Cl₂ (3:1).

Dihydrolysergic Acid and Dihydrolysergol.

Among the pharmacologically active ergot alkaloids,¹¹⁴ lysergic acid¹¹⁵ and its semisynthetic derivative lysergic acid diethylamide (LSD) are the most widely recognized members, and several other derivatives are used clinically in the treatment of a range of neurological disorders.¹¹⁶ Due to the wide range of activities exhibited by this class of natural products and the interesting structural features inherent in the conformationally restricted phenethylamine pharmacophore, the ergot alkaloids have been the subject of extensive study over the last half century.¹¹⁷ We reported a total synthesis of dihydrolysergic acid and dihydrolysergol,¹¹⁸ the latter of which had been prepared only by semisynthesis from lysergic acid prior to our work. The synthesis relied on two key steps. In addition to a powerful Diels—Alder reaction of 5-carbomethoxy-1,2,3-triazine with a conjugated enamine,^105b we used an intramolecular Pd(0)-catalyzed indole annulation,¹¹⁹ which we had developed earlier in other studies, for the formation of a tricyclic ketone isomeric with those^117,120 widely used in accessing the ergot alkaloids. By design, the advanced ketone-derived conjugated enamine and a series of late-stage heterocyclic azadiene inverse electron demand [4 + 2] cycloaddition reactions were used for divergent synthesis of ergot alkaloid analogues bearing deep-seated structural changes not readily accessible by conventional approaches.

The synthesis of dihydrolysergic acid and dihydrolysergol began with the preparation of ketone 150 (Scheme 24). This sequence rests on the intramolecular Pd(0)-catalyzed Larock cyclization of 148, forming indole 149 in excellent yield (90%). Our prior studies on the intramolecular Larock annulation targeting the chloropeptins¹²¹ revealed that catalysts such as PdCl₂(PPh₃)₂, Pd(PPh₃)₄, and Pd₂(dba)₃ were effective in the reaction while PdCl₂, PdCl₂(PhCN)₂, or PdCl₂(PPh₃)₂ were not. Pd₂(dba)₃ was first found to be especially successful in a challenging Suzuki coupling reaction in our vancomycin total synthesis, arising from an unusually rapid oxidative addition step with an analogous electron-rich aryl bromide.¹²² Optimization of our conditions produced the desired indole from a hindered o,o′-substituted aryl bromide substrate in 90% yield on gram scale (15% Pd₂(dba)₃, DtBPF, Et₃N, DMF, 130 °C, 3 h). Further elaboration of 149 over four steps provided the tricyclic ketone 150.

Scheme 24.

Synthesis of Tricyclic Ketone 150 en Route to Dihydrolysergic Acid and Dihydrolysergol

The key inverse electron demand Diels—Alder reaction of 150 with 5-carboxmethoxy-1,2,3-triazine (152) (Scheme 25) was found to proceed in excellent yield upon conversion of ketone 150 to enamine 151 (pyrrolidine, 4 Å MS, CHCl₃, 25 °C, 8 h) and subsequent treatment with 152 (0.1 M CHCl₃, 25 °C, 3 h, 75% over two steps).¹¹⁸ This cycloaddition proceeded rapidly to provide pyridine 153 as a single regioisomer at room temperature without detection of intermediates. The exclusive N1/C4 regioselectivity and azadiene reactivity were in accordance with expectations and benefitted from the complementary azadiene substitution (C5-CO₂Me).

Scheme 25.

Total Synthesis of Dihydrolysergic Acid and Dihydrolysergol

Tetracycle 153 was elaborated to the natural products in a sequence involving N-methylation of the pyridine, exhaustive reduction with NaCNBH₃ to provide almost entirely a single diastereomer, indoline deprotection and oxidation to the indole, and final ester hydrolysis (dihydrolysergic acid) or reduction (dihydrolysergol). In addition, the enamine derived from ketone 150 was employed in additional heterocyclic azadiene cycloaddition reactions for the late-stage, divergent preparation of a series of heterocyclic derivatives without optimization. This included the reaction of a series of substituted 1,2,3-triazines, each of which displayed the expected C4/N1 regioselectivity and provided differentially substituted pyridines, 1,3,5-triazine, which provided the pyrimidine, and 3,6-dicarbomethoxy-1,2,4,5-tetrazine, which provided a substituted pyridazine and access to the pyrrole through a zinc-mediated reductive ring contraction (Scheme 26).¹¹⁸

Scheme 26.

Synthesis of Heterocyclic Ergot Alkaloid Analogues

Methoxatin.

In studies that led to the development of the first general method for the catalysis of the inverse electron demand Diels—Alder reactions of heterocyclic azadienes, we targeted the total synthesis of methoxatin, a redox-active quinone that serves as a cofactor for methanol dehydrogenase and is found in a variety of methylotrophic bacteria.¹²³ Although the discovery of methoxatin, also known as pyrroloquinoline quinone (PQQ), led to a series of studies devoted to the discovery of a methoxatin-dependent mammalian enzyme, no such protein has yet been identified.¹²⁴ Nevertheless, PQQ has been shown to play a role in a variety of mammalian processes,¹²⁵ is offered as a dietary supplement, and has been shown to reduce the risk of injury in heart attacks and strokes.¹²⁶ Due to its potentially important role in human health, its putative role as a vitamin, and its unusual structure, PQQ has been the target of a number of increasingly concise total syntheses.¹²⁷ We recently disclosed the shortest total synthesis of methoxatin reported to date that relied on the discovery and development of a powerful hydrogen bonding activation of heterocyclic azadiene cycloaddition reactions.¹⁰⁸ The approach was designed to permit the late-stage divergent synthesis of PQQ analogues through use of a series of complementary heterocyclic azadiene cycloaddition reactions.

The synthesis of methoxatin began with the known ketone 156 (Scheme 27). Treatment with pyrrolidine provided enamine 157, which was used without further purification in the key cycloaddition. Initial optimization studies revealed that the addition of trifluoroacetic acid (1.5 equiv) improved reaction yields, suggesting that acid-promoted aromatization or hydrogen bonding may be contributing to the improvement. Unaromatized cycloadduct was not detected as a stalled intermediate, indicating that the additive was not serving as a catalyst for the final aromatization step, but was accelerating the cycloaddition itself. Systematic studies led to the discovery that HFIP was a remarkably effective solvent or additive for the cycloaddition (Figure 20), promoting the reaction through H-bonding to the heterocyclic azadiene and providing the desired cycloadduct in a stunning 95% yield over two steps (0.1 M HFIP, 60 °C, 24 h).¹⁰⁸

Scheme 27.

Total Synthesis of Methoxatin

Figure 20.

H-bonding effect of solvent on cycloaddition.

Methoxatin was accessed in only three steps from cycloadduct 159. DDQ oxidation (4.0 equiv, 0.004 M C₆H₆, 90 °C, 48 h) followed by treatment with RuO₄ using modified conditions that generated RuO₄ in situ (0.4 equiv of RuO₂, 5.0 equiv of NaIO₄, 0.004 M H₂O—CH₂Cl₂−CH₃CN 1:1:1, 23 °C, 30 min) cleanly provided the desired o-quinone in 63% yield over two steps. Lastly, a global saponification (0.01 M THF−0.5 M aq LiOH, 23 °C, 6 h) provided methoxatin in 94% yield.

Finally, and as designed, a series of representative heterocyclic azadienes were found to undergo cycloaddition with enamine 157 in HFIP without optimization to provide the corresponding methoxatin analogues (Scheme 28).¹⁰⁸

Scheme 28.

Divergent Synthesis of Methoxatin Analogues

1,3,4-Oxadiazoles.

The tandem Diels—Alder/1,3-dipolar cycloaddition reactions of 1,3,4-oxadiazoles have been studied by a limited number of groups, generally employing symmetrical oxadiazoles bearing strong electron-withdrawing substituents in intermolecular reactions.^128,129 These five-membered aromatic heterocycles were first shown by Vasil’ev^9b to undergo inverse electron demand Diels—Alder reactions that trigger a cycloaddition cascade proceeding through a [4 + 2] cycloaddition, loss of nitrogen (N₂) to generate a carbonyl ylide, and a final 1,3-dipolar cycloaddition. While this cycloaddition cascade is potentially powerful and was exemplified several times, including with the preparation of complex norbornadiene structures,¹³⁰ limitations of the reaction precluded its widespread adoption. These limitations include the lack of differentiation between the dienophile and dipolarophile since the ensuing 1,3-dipolar cycloaddition is faster than the initiating Diels—Alder reaction and the necessity for use of symmetrical oxadiazoles and olefins/alkynes (Figure 21). Informed by our study of the reactivity of electron-deficient heterocyclic azadienes and because of our interest in the Vinca alkaloids, we designed and developed an intramolecular [4 + 2]/[3 + 2] cycloaddition cascade inspired by the structure of vindoline, the lower half of vinblastine.¹³¹ Representing the first example of a tandem intramolecular Diels—Alder/1,3-dipolar cycloaddition cascade of 1,3,4-oxadiazoles, this methodology differentiates the dienophile and dipolarophile, extends the scope of oxadiazoles used in such reactions, and provides regio- and stereocontrol over the resultant cycloadducts (Figure 21). We recently reviewed the emerging scope of this intramolecular cycloaddition cascade.¹³²

Figure 21.

1,3,4-Oxadiazole cycloaddition reactions and tandem intramolecular cycloaddition cascade.

Anhydrolycorinone.

Although our work with this hetero-cycle primarily focused on the tandem cycloaddition cascade detailed above, the first use of a 1,3,4-oxadiazole in a natural product total synthesis was exemplified with anhydrolycorinone.¹³³ The basis for the development of this reaction cascade came from the observation that tethered alkyne dienophiles and olefins bearing a leaving group generated high yields of furan products after an initial Diels—Alder cycloaddition (Figure 21). This reactivity was recognized as being complementary to the cycloadditions of oxazoles but entails a more facile loss of N₂ (vs nitrile) in the retro-Diels—Alder aromatization event and was exploited in a concise synthesis of a highly substituted benzene system. The use of a methyl enol ether as the initiating dienophile provided a furan through loss of methanol that was appropriately set up to react further in a second Diels—Alder reaction with a tethered olefin, generating the pentacyclic core of anhydrolycorinone in a single step (Scheme 29). Sequential ester saponification and decarboxylation provided the natural product.

Scheme 29.

Total Synthesis of Anhydrolycorinone

Vindoline, Vinblastine, and Vincristine.

The inspiration for our development of the intramolecular [4 + 2]/[3 + 2] cascade cycloaddition reaction of 1,3,4-oxadiazoles was the dimeric Vinca alkaloids vinblastine and vincristine. Thus, among the first total syntheses to be published utilizing this methodology was that of vindoline, in 2005.¹³⁴ Since then, we have disclosed two additional total syntheses of vindoline, a series of syntheses of related Aspidosperma alkaloids, the divergent total syntheses of four different natural product alkaloid classes from a common cascade cycloadduct, as well as the total syntheses¹³⁵ of vinblastine, vincristine, and an extensive series of analogues including ultrapotent vinblastines.

Vinblastine and vincristine were isolated in 1958 and 1961, respectively, in trace quantities from Catharanthus roseus (L.) G. Don¹³⁶ and are two of the most important and widely recognized antitumor drugs currently in use in the clinic.¹³⁷ Their clinical efficacy has resulted in their use in combination therapies for the treatment of Hodgkin’s disease, testicular cancer, breast cancer, ovarian cancer, head and neck cancer, and non-Hodgkin’s lymphoma (vinblastine) and in curative treatments of childhood lymphocytic leukemia and Hodgkin’s disease (vincristine) and has earned them positions in the World Health Organization’s List of Essential Medicines.

Because of their clinical importance, low natural abundance, and complex dimeric structure, vinblastine and vincristine have remained the subject of extensive investigations, including our own interest in the molecules over the last two decades.^138,139

In the course of these efforts, we comprehensively defined the importance and role of nearly every atom and substituent in vinblastine and vincristine, including the vindoline C4 acetate,¹⁴⁰ C5 ethyl substituent,¹⁴¹ C6−C7 double bond, and the vindoline core structure itself,¹⁴² and have systematically explored the upper catharanthine-derived C20′ ethyl substituent,¹⁴³ C16′ methyl ester,¹⁴⁴ added C10′ or C12′ indole substitutions,¹⁴⁵ and, most recently, the C20′ alcohol.¹⁴⁶ We have shown that the latter is remarkably tolerant to replacement, resulting in the discovery of both ultrapotent analogues of the natural products^146b as well as those that match the potency of the natural products but do not suffer from overexpressed Pgp-derived resistance^146a (Figure 22). Because each of three classes of improved analogues that we discovered^145,146 not only incorporate added structural features not found in the natural products, but also substantially surpass their properties, we like to think of them as prototypical examples of “supernatural” products.¹⁴⁷

Figure 22.

Synthesis of vinblastine. Highlighted regions are those targeted in our medicinal chemistry efforts.

These discoveries were enabled by our development of the powerful intramolecular [4 + 2]/[3 + 2] cascade cycloaddition reaction of 1,3,4-oxadiazoles and its use in concise syntheses of vindoline, the lower subunit of vinblastine. Our first-generation synthesis of (−)-vindoline (11 steps) rested on the cycloaddition cascade of either 161 or 162.^134,148 This reaction established the full pentacyclic core of the natural product, generated six contiguous stereocenters about the newly formed central six-membered ring, and introduced nearly all of the functionality found in vindoline in a single step. Although the key features of this reaction have been detailed in our work, it is important to note that the difference in the facility of the cycloaddition of 161 and 162 arises from the electronic character of the benzyloxy substituent of 161, which decelerates the typically fast 1,3-dipolar cycloaddition. This rate deceleration arises from a destabilizing electrostatic interaction in the transition state between the central oxygen of the 1,3-dipole and the (Z)-OBn substituent. Both cycloadducts were elaborated to vindoline in 7 or 10 steps, providing efficient syntheses of the natural product (Scheme 30).

Scheme 30.

Eleven-Step Total Synthesis of Vindoline

This observation of the importance of the initiating dienophile in the cascade cycloaddition reaction led to the examination of additional functionalized dienophiles and the discovery that an allene could be used to initiate the reaction to provide the desired cascade cycloadduct in exceptional yields (Scheme 31).¹⁴⁹ The initial Diels—Alder reaction affords a cycloadduct that in turn undergoes loss of N₂ to provide an unusual cross-conjugated 1,3-dipole that proved to be extraordinarily effective in the ensuing 1,3-dipolar cycloaddition, providing the cycloadduct as a single diastereomer in 92% yield. Corey dihydroxylation and subsequent Pb-(OAc)₄-mediated diol cleavage followed by lactam α-hydroxylation and in situ silyl ether formation provided an intermediate in our first-generation synthesis of (−)-vindoline, completing a formal total synthesis of the natural product.

Scheme 31.

Total Synthesis of Vindoline through an Allene-Initiated Oxadiazole Cycloaddition Cascade

Finally, while the enantiomers of late-stage intermediates in the syntheses detailed above were directly and chromatographically resolved to provide access to both (−)-vindoline and ent-(+)-vindoline, we also developed a complementary asymmetric total synthesis.¹⁵⁰ This was based on a cycloaddition cascade in which the dienophile linking tether bears a chiral substituent that sets the absolute stereochemistry of all six stereocenters about the central six-membered ring in the resulting cycloadduct (Figure 23). The length of the linking tether was shortened to permit effective control of the cycloaddition facial selectivity, enlisted milder cycloaddition reaction conditions (140 °C, 16 h), and required a ring expansion later in the synthesis. This ring expansion was accomplished by two approaches. The first relied on the hydrolytic ring expansion of a N,O-ketal generated through a unique oxidative deformylation reaction, while the second employed generation of an intermediate aziridinium ion in a thermodynamically controlled ring expansion. Together, these two approaches provided two distinct and concise asymmetric total syntheses of (−)-vindoline that complement our efforts utilizing direct chromatographic chiral resolution.

Figure 23.

Asymmetric total syntheses of vindoline.

Our work culminated in a 12-step total synthesis of vinblastine that featured a biomimetic iron-mediated coupling of vindoline and catharanthine¹⁵¹ and a further iron-mediated hydrogen atom transfer (HAT) olefin functionalization for introduction of the C20′ alcohol.^{135,152−154} These two key reactions have now been leveraged for the synthesis of hundreds of informative analogues of vinblastine.

Related Alkaloids.

Complementary to our work on vindoline, we disclosed the total syntheses of several related Aspidosperma alkaloids.¹⁴⁸ These efforts were conducted in parallel with our work on the methodology itself, allowed for the synthesis of a series of natural products of increasing complexity, and defined the scope and key features of the cycloaddition cascade. These investigations not only culminated in our synthesis of vindoline, but many also served to address longstanding questions on the absolute configuration and characterization of the targeted natural products. The first set of natural products were accessible from a common cycloadduct intermediate, highlighting the use of the methodology in divergent synthesis (Figure 24).

Figure 24.

Divergent total synthesis of Aspidosperma natural products from a common cycloadduct.

Minovine.

Minovine, isolated from Vinca minor L., is implicated in the biosynthetic pathway of alkaloids bearing the vincadifformine core structure.¹⁵⁵ The naturally occurring material was reported to possess an [α]_D = 0 in alcoholic solvents, an unusual feature that had not been addressed in previous racemic syntheses of the natural product. Our synthesis of optically active minovine, proceeding in six steps from cycloadduct 171 derived from the tandem cycloaddition cascade of 170 (74%), addressed this issue (Scheme 32).¹⁵⁶

Scheme 32.

Total Synthesis of Minovine

Optically active minovine was found to exhibit a solvent, but not concentration, dependent optical rotation: natural 172 [α]²³_D −17 (c 0.35, CHCl₃), +16 (c 0.40, MeOH), and 0 ± 3 (c 0.28, EtOH). In addition, studies were carried out to determine whether a reversible Diels—Alder reaction, previously suggested to be responsible for racemization and the observed optical rotation of 0, was possible with the natural product. Attempts at such a reaction showed that it is not observed. Thus, naturally occurring minovine is not racemic, nor does it racemize, but it does exhibit an unusual solvent dependent optical rotation. For our own purposes, the synthesis of minovine allowed us to probe the reactivity of the lactam and oxido bridge present in cycloadduct 171 and was instrumental in our development of effective ring opening and lactam reduction reactions. The discovery that these steps could be conducted in either order expanded the range of targets that could be pursued using the methodology (Scheme 32).

N-Methylaspidospermidine.

The enantiomer of the cycloadduct used in the total synthesis of optically active minovine was used for the first total synthesis of (+)-N-methylaspidospermidine and served to confirm the absolute configuration of the natural product. Isolated from Evatamia peduncularis, Vallesia dichotoma, Aspidosperma quebracho blanco, and Vinca minor L.,¹⁵⁷ (+)-N-methylaspidospermidine and the closely related aspidospermidine possess an absolute configuration opposite that found in vindoline and the Vinca alkaloids. Cycloadduct 171 was elaborated to the target compound in just five steps, provided further insight into the optional order of lactam reduction/oxido bridge opening, and established an approach for the removal of the methyl ester, expanding the scope of future targets (Scheme 33).¹⁴⁸

Scheme 33.

Total Synthesis of (+)-N-Methylaspidospermidine

4-Desacetoxyvindorosine.

This same cycloadduct was used in the concise total synthesis of 4-desacetoxyvindorosine. Although it is a key late-stage biosynthetic intermediate enroute to vindorosine, it had never been fully characterized and no total synthesis of this alkaloid had been reported. Differing from our ultimate target vindoline only by lack of the C16 methoxy and C4 acetoxy substituents, 4-desacetoxyvindorosine allowed investigation of the introduction of the C6−C7 double bond found in both natural products. The strategy chosen for the introduction of this functionality involved α-hydroxylation of the lactam present in 171. Following carbonyl excision, secondary alcohol elimination was expected to provide the Δ^6,7-olefin, following the precedent of Kuehne on similar intermediates.

Initial efforts to use this second transformation on a C7 alcohol focused on the conditions first developed by Kuehne¹⁵⁸ (Ph₃P—CCl₄, Et₃N) and later by Fukuyama¹⁵⁹ (Ph₃P—CCl₄). Although a thorough study was conducted, the reaction proved to be more difficult to implement than anticipated. A sensitivity to moisture was observed, leading to recovery of starting material with the potential intermediacy of an aziridinium ion, and use of excess reagent resulted in competitive in situ chloride displacement or addition (Figure 25). Variations on the reaction conditions provided small improvements but were ultimately ineffective with the more highly functionalized substrates required for vindoline or vindorosine. However, we found that the Mitsunobu reagent¹⁶⁰ Ph₃P—DEAD was uniquely effective, providing high conversions under mild conditions, due to its ability to activate the secondary alcohol for aziridinium ion formation or elimination without release of a nucleophile or the necessity of added base. Compound m was elaborated to (+)- and ent-(−)-4-desacetoxyvindorosine in seven steps (Scheme 34).¹⁴⁸

Figure 25.

Introduction of the Δ^6,7-olefin.

Scheme 34.

Total Synthesis of (+)-4-Desacetoxyvindorosine

4-Desacetoxyvindoline and 4-Desacetoxyvinblastine.

Concurrent with our efforts toward vindoline, a total synthesis of its 4-desacetoxy variant was conducted and illustrated the ease of preparation of simpler derivatives of the natural product. 4-desacetoxyvindoline, a natural product in its own right, is the penultimate biosynthetic precursor to vindoline.¹⁶¹ Prior to our work, no total synthesis of the compound had been reported, and it had only been characterized by MS and ¹H NMR.¹⁶² In addition and since desacetoxyvinblastine, lacking only the vindoline C4 acetoxy group, is also a natural product and had been shown to be equally or more efficacious in vivo than vinblastine itself,¹⁶³ we undertook the synthesis of 177 and 4-desacetoxyvinblastine by employing the tandem cycloaddition cascade of 175 (Scheme 35). This approach proved to be straightforward and used the more reactive disubstituted (vs trisubstituted for vindoline) tethered dienophile in the key cycloaddition cascade that proceeded in excellent yield and with perfect diastereoselectivity (TIPB, 230 °C, 18 h, 83%). A further seven steps provided 4-desacetoxyvindoline (177), which was fully characterized for the first time.¹⁴⁸ It was also further incorporated in 4-desacetoxyvinblastine in a single additional step, permitting its contemporary assessment alongside additional analogues of vinblastine.¹³⁵

Scheme 35.

Total Synthesis of (+)-4-Desacetoxyvindoline

Vindorosine.

A concise total synthesis of (−)- and ent-(+)-vindorosine was developed that allowed us to probe the key cycloaddition reaction of a substrate bearing the trisubstituted dienophile required for the introduction of the C4 acetoxy group.¹⁶⁴ Vindorosine is identical in structure to vindoline, lacking only the C16 methoxy substituent. Its densely functionalized structure and similarity to vindoline have resulted in it being the focus of several beautiful and historically important total syntheses.¹⁶⁵

Our efforts focused on the use of (Z)- and (E)-178, bearing isomeric trisubstituted electron-rich enol ethers as the tethered initiating dienophiles. Trisubstitution of the dienophile generally slows initiation of the intramolecular [4 + 2]/[3 + 2] cycloaddition cascade, although recent studies have defined reaction conditions that permit the use of even more difficult trisubstituted electron-deficient and unactivated alkenes.¹⁴⁰ The exceptions to this generalization are trisubstituted olefins bearing an electron-donating substituent that serves to activate the dienophile for participation in an 1,3,4-oxadiazole inverse electron demand Diels—Alder reaction.

The electron-rich olefins in (Z)- and (E)-178 were anticipated to be ideally suited for initiation of the cycloaddition cascade and introduction of the C4 alkoxy substituent (Figure 26). Moreover, (Z)-178 not only directly provides the natural β-stereochemistry but was also expected to be more effective in the key cycloaddition cascade due to decreased steric interactions in the ensuing 1,3-dipolar cycloaddition transition state (Figure 26). While (E)-178 was shown to undergo the cycloaddition cascade in excellent yield (95%, TIPB, 230 °C, 1 mM, 18 h), (Z)-178 proved more challenging, requiring considerable optimization to provide the desired cycloadduct in good yield (78%, TIPB, 230 °C, 0.1 mM, 60 h).¹⁴⁸

Figure 26.

Cycloaddition cascades en route to vindorosine.

Considerable work was conducted to define the subtleties of the effects responsible for the observations. To summarize these studies, it was observed that the 1,3-dipolar cycloaddition, generally the fast step in the tandem cycloaddition cascade, was now the rate-limiting step in the case of (Z)-178. Two explanations may account for this observation. The first is that the transition state of the ensuing 1,3-dipolar cycloaddition of (Z)-178 suffers from a destabilizing, and thus decelerating, electrostatic interaction of its central oxygen with the OBn substituent of the reacting dienophile (Figure 26). This was confirmed by comparing the reactions of (Z)- and (E)-181, each bearing a methyl group in place of the OBn substituent (Figure 27).

Figure 27.

Model systems for the key cycloaddition cascade.

Both were shown to be equally effective in the cycloaddition cascade, indicating that electronic and not steric features were contributing to the difference in the behavior of (Z)- vs (E)-178. A second but less likely explanation involves the preferred stereochemistry of putative cyclobutene epoxides that may act as reversibly generated intermediates for the highly reactive carbonyl ylides. It is plausible that they may possess epoxide stereochemistries that influence the rate of the ensuing 1,3-dipolar cycloaddition. The enantiomers of cycloadduct 179 were elaborated to (−)- and ent-(+)-vindorosine in seven steps (Scheme 36).¹⁴⁸

Scheme 36.

Total Synthesis of (−)-Vindorosine

Spegazzinine and Aspidospermine.

Finally, we undertook the synthesis of (−)-aspidospermine and (+)-spegazzinine, in which we examined the effect of aryl substitution on the key [4 + 2]/[3 + 2] cascade.¹⁶⁶ Spegazzinine, isolated in 1956 from Aspidosperma chakensis Spegazzini,¹⁶⁷ has been shown to be a potent inhibitor of mitochondrial photophosphorylation and oxidative phosphorylation.¹⁶⁸ Aspidospermine, isolated in the late 1800s from the Aspidosperma quebrancho,¹⁶⁹ exhibits a wide range of biological activities.¹⁷⁰ At the time of our work, spegazzinine had not been targeted by total synthesis, whereas aspidospermine has been one of the most attractive synthetic targets among the Aspidosperma alkaloids.¹⁷¹

Our approach enlisted the cycloaddition cascade of 183, which proceeded in excellent yield to provide 184 as a single diastereomer (71%, o-Cl₂C₆H₄, 180 °C, Scheme 37).¹⁶⁶ (−)-Aspidospermine and (+)-spegazzinine could be accessed from 184 by a sequence of oxido bridge ring opening, lactam carbonyl excision, removal of the C3 methyl ester, benzyl deprotection, acetylation, and in the case of the former, phenol methylation and C3 alcohol removal. This work served to establish the absolute configuration of natural (+)-spegazzinine and unambiguously determine its relative C3 alcohol stereochemistry. As remarkable as it may seem, this synthesis of aspidospermine also was the first not to rely on a late-stage Fisher indole synthesis first introduced by Stork.^171a

Scheme 37.

Total Synthesis of (+)-Spegazzinine and (−)-Aspidospermine

Divergent Total Synthesis: Diverse Natural Product Families from a Common Intermediate.

The cycloadduct derived from the cycloaddition cascade of 187 was used as a common intermediate in the divergent total synthesis of natural products from four different alkaloid classes. These syntheses enlisted four different late-stage strategic bond formations embedded in each natural product core structure (Figure 28).

Figure 28.

Divergent total synthesis of four different alkaloid natural product classes from a common cycloadduct.

Fendleridine.

Fendleridine was isolated in 1964 from Aspidosperma fendleri WOODSON and is the parent compound of the aspidoalbine family of alkaloids.¹⁷² Unique to this class of natural products is the oxidized C19 N,O-ketal embedded in the Aspidosperma alkaloid pentacyclic ring system. Although a handful of total and formal syntheses of fendleridine (189)¹⁷³ and its congener 1-acetylaspidoalbidine¹⁷⁴ were disclosed prior to our work, all provided racemic material and the assignment of the absolute configuration remained to be established.

The synthesis of these compounds was accomplished in 9 and 10 steps, respectively, from cycloadduct 188 and established the absolute configuration of the natural products.¹⁷⁵ In addition to the cycloaddition cascade, which proceeded in excellent yield (71%, o-Cl₂C₆H₄, 180 °C), the synthesis featured a key tetrahydrofuran ring introduction with direct formation of the C19 N,O-ketal (Scheme 38). This transformation, which takes advantage of the inherent C19 oxidation state, was affected in a single step through treatment of 193 with HF/pyridine to generate a stable cyanohydrin. Direct trap of the reversibly generated iminium ion, generated through oxido bridge ring opening, is facilitated by the two quaternary centers flanking C19.

Scheme 38.

Tetrahydrofuran Ring Formation en Route to a Total Synthesis of (+)-Fendleridine

Kopsinine.

Kopsinine was first isolated from Kopsia Longiflora Merr.¹⁷⁶ and differs from the closely related Aspidosperma alkaloids by the addition of a bond joining C21 to C2 to provide a bicyclo[2.2.2]octane system embedded in its hexacyclic core. Our interest in the molecule stemmed from the potential of its access through a late-stage C2−C21 strategic bond formation that would be distinct from previous syntheses of Kopsia alkaloids, all of which employ Diels—Alder reactions of unnatural Aspidosperma-like pentacyclic C2−C5 dienes.¹⁷⁷

Kopsinine was accessed in eight steps from cycloadduct 188.¹⁷⁸ Key to the synthesis was a late-stage C2−C21 bond formation mediated by SmI₂ in 10:1 THF—HMPA, proceeding through a radical-mediated conjugate addition cyclization and subsequent protonation of the further reduced ester enolate from the less hindered convex face to provide the product in excellent yield (75%) and as a single diastereomer (Scheme 39).

Scheme 39.

Strategic C2−C21 Bond Formation in the Total Synthesis of (−)-Kopsinine

Kopsifoline D.

Kopsifolines A—F, first isolated from K. fruticosa (Ker) A. DC. in 2003,¹⁷⁹ constitute the first members of a new family of alkaloids that possesses a unique core hexacyclic ring system and a previously unprecedented C3−C21 bond. The central six-membered ring contains five or six stereogenic centers, of which three or four are quaternary. Our synthesis of kopsifoline D relied on a biomimetic late-stage C21−C3 strategic bond formation via a transannular enamide alkylation of a C21 iodide within the Aspidosperma skeleton, directly introducing the correct C2 oxidation state.¹⁸⁰ The key C21−C3 bond formation was accomplished by treatment of 198 with BF₃·OEt₂ and Me₂S to affect Cbz deprotection, followed by subjection to Et₃N in EtOAc to affect the intramolecular ring closure and provide the natural product (Scheme 40). This not only provided the first total synthesis of a naturally occurring kopsifoline and established the absolute configuration for the alkaloid class, but it remains the only reported total synthesis still today.

Scheme 40.

C3−C21 Bond Formation in the Total Synthesis of (−)-Kopsifoline D

Deoxoapodine.

As part of these studies, we were also able to access (−)-deoxoapodine, isolated from Tabernae armeniaca,¹⁸¹ through a late-stage C21−O—C6 bond formation, providing the first synthesis of the natural product in optically active form. C21−O—C6 bond formation to provide access to deoxoapodine was accomplished by treatment of 200 with Bu₄NF, which resulted in clean silyl ether deprotection with concomitant conjugate addition (single diastereomer) without detection or isolation of the intermediate alcohol (Scheme 41). Carbamate deprotection provided deoxoapodine, identical in all respects to authentic material.¹⁸⁰

Scheme 41.

Tetrahydrofuran Ring Formation En Route to a Total Synthesis of (−)-Deoxoapodine

3. 1-AZA-1,3-BUTADIENES

N-Sulfonyl-1-aza-1,3-butadienes.

At the time we initiated our studies, the Diels—Alder reactions of acyclic azadienes, such as α,β-unsaturated imines, were rare since their intrinsic instability, competitive 1,2-imine addition, or imine tautomerization precludes successful [4 + 2] cycloaddition.^3d Moreover, the cyclic enamine products often proved unstable under the reaction conditions. In our studies, we addressed these limitations by incorporation of suitable substituents on the acyclic 1-aza-1,3-butadienes. The complementary N1 and C3 substitution of α,β-unsaturated imines with electron-withdrawing substituents enhance the intrinsic electron-deficient nature of the azadiene and thus accelerate its rate of reaction with electron-rich dienophiles in inverse electron demand Diels—Alder reactions. In addition, a bulky electron-withdrawing N-1 substituent preferentially decelerates 1,2-imine addition through effective steric shielding and conveys product stability to the [4 + 2] cycloadduct by affording a deactivated enamine. Through such design principles, we established approaches to control and accelerate the 4π participation of acyclic 1-aza-1,3-butadienes in [4 + 2] cycloaddition reactions.¹⁸²

In our studies, N-sulfonyl and N-phosphinyl α,β-unsaturated imines were shown to constitute stable, easily accessible, substituted 1-aza-1,3-butadienes suitable for use in well-behaved Diels—Alder reactions (Figure 29).¹⁸² The reaction exhibits characteristics of a concerted [4 + 2] cycloaddition. The dienophile and diene geometry are conserved in the reaction products, trans-1,2-disubstituted dienophiles react faster than cis-1,2-disubstituted dienophiles, there is little or no solvent effect on the rate of the reaction, and the reaction displays a characteristic pressure-induced endo diastereoselectivity.¹⁸² Significantly, the reaction is not only regiospecific but also endo specific, typically exhibiting unusually high levels of diastereoselectivities (≥20:1) often to the extent that a single cycloadduct is observed. This remarkable diastereoselectivity may be attributed to two complementary features of the reaction: (1) a stabilizing secondary orbital interaction between diene C2 and the dienophile OR substituent that preferentially stabilizes the endo transition state and (2) a transition state anomeric effect in which the nitrogen lone pair and the electron-rich σ bond of the enol ether are trans periplanar in the endo boat transition state.^182a Both effects enhance the characteristic endo diastereoselectivity and both are absent in the alternative exo transition state. The latter effect is also unique to 1-aza-1,3-butadienes and not found in the all-carbon Diels—Alder reaction, accounting for the superior diastereoselectivity.

Figure 29.

N-Sulfonyl-1-aza-1,3-butadiene Diels—Alder reaction.

By virtue of lowering the diene LUMO, both the complementary C3 addition and the noncomplementary C2 or C4 addition of an electron-withdrawing substituent to the N-sulfonyl 1-azadiene substantially accelerate their reaction rate in inverse electron demand Diels—Alder reactions, while maintaining or enhancing the superb regioselectivity and endo diastereoselectivity.¹⁸² Typically, a single [4 + 2] cycloadduct is generated at satisfactory rates at 25 °C or lower and 1-aza-1,3-butadienes bearing complementary C3 electron-withdrawing groups react faster than 1-azadienes bearing noncomplementary C2 and C4 electron-withdrawing groups (Figure 30).

Figure 30.

Complementary and noncomplementary substituent effects.

We have also shown that the thermal reactions of N-sulfonyl-1-aza-1,3-butadienes with enol ether dienophiles bearing chiral auxiliaries provide highly diastereoselective (endo and facial diastereoselection) inverse electron demand Diels—Alder cycloadditions.¹⁸³ This is largely the result of the exquisitely organized [4 + 2] cycloaddition transition state. Vinyl ethers have been shown to preferentially react with dienes through an s-trans conformation. As the transition state model in Figure 31 illustrates, when the dienophile approaches the diene in this s-trans conformation in the highly organized endo boat transition state, the dienophile preferentially adopts a conformation in which the H is oriented toward and its large group (R_L) is oriented away from the diene. The chiral auxiliary comes into proximity to the sterically demanding and electronegative sulfonyl groups, requiring the sulfonyl phenyl group to orient itself on the diene face opposite the approaching dienophile.

Figure 31.

Transition-state model of an asymmetric Diels—Alder cycloaddition of N-sulfonyl-1-aza-1,3-butadienes with enol ethers bearing chiral auxiliaries.

In our studies, 18 optically active enol ethers were examined.¹⁸³ Among them, three new, readily accessible, and previously unexplored γ-lactone and γ-lactam chiral auxiliaries were found to provide remarkable selectivities (up to 49:1 endo/exo and 48:1 facial selectivity). The transition state model for these reactions places the electronegative sulfonyl oxygen on the same face as the approaching dienophile. This directs the electronegative ester, lactone, or lactam carbonyls of the effective dienophiles, although sterically undemanding, away from the sulfonyl group, behaving as effective R_L groups. X-ray crystal structures of the cycloadducts, albeit being ground-state structures of the reaction products, embody each of these features.

The acyclic 1-aza-1,3-butadienes have also been examined in intramolecular Diels—Alder reactions.^182e,184 Similar to observations made in the intermolecular cycloadditions, the reaction displays a strong endo diastereoselectivity (>20:1). Representative of this, the product 203 was derived from cycloaddition of 202 through an anti-endo transition state as depicted in Figure 32. In addition, the entropic assistance provided by the intramolecular nature of such reactions has also allowed the use of less reactive azadienes, including O-alkyl oximes and less reactive electron-neutral or electron-deficient dienophiles, including tethered terminal or 1,2-disubstituted alkenes and alkynes.

Figure 32.

Representative intramolecular 1-aza-1,3-butadiene Diels—Alder reactions.

The inverse electron demand Diels—Alder reactions of acyclic N-sulfonyl-1-aza-1,3-butadienes provide a stabilized highly substituted cyclic enamine, which can be further converted to a fully substituted pyridine (Figure 33). Based on the studies that defined the scope of this powerful transformation, we have applied this strategy in the total synthesis of a series of complex natural products containing a fully substituted pyridine or pyridone ring as a key structural element, including streptonigrone, fredericamycin A, nothapo- dytine, mappicine, camptothecin, rubrolone aglycon, and piericidins A1 and B1.

Figure 33.

Pyridine synthesis based on the inverse electron demand Diels—Alder reactions of acyclic 1-azadienes.

Streptonigrone.

Streptonigrone (211) is a highly substituted and densely functionalized quinolone-5,8-quinone related to the potent antitumor antibiotic streptonigrin that contains a fully substituted pyridone C ring central to its structure.¹⁷ A well-designed inverse electron demand Diels—Alder cycloaddition of a fully functionalized N-sulfonyl-1-aza-1,3-butadiene bearing a complementary C3 electron-withdrawing substituent was anticipated to provide a cycloadduct that could be easily elaborated to the pyridone.¹⁸⁵ The C3 carboxylate incorporated into the azadiene as a lactone serves three strategic functions. First, it further accelerated the rate of diene participation in the LUMO_diene-controlled Diels—Alder reaction and reinforced the inherent cycloaddition regioselectivity. Second, it offered a convenient manner to differentiate and protect the D ring phenol. Finally, it was designed to serve as suitable functionality for the introduction of the pyridone C5 amine. As expected, the inverse electron demand Diels—Alder reaction of 1-azadiene 206 with ketene acetal 207 occurred at room temperature to afford cycloadduct 208 in only 30 min. Treatment of 208 with tBuOK followed by DDQ provided the fully substituted pyridine 210.¹⁸⁵ This unusual aromatization sequence proceeded with intermediate generation of imidate 209 derived from methanesulfonamide deprotonation, loss of sulfene facilitated by vinylogous amide activation of the departing amine, and finally loss of methoxide. Subsequent oxidative aromatization was accomplished upon DDQ treatment of the intermediate imidate, providing the full skeleton 210 of streptonigrone (Scheme 42). As far as we are aware, our total synthesis of streptonigrone represents the only total synthesis disclosed to date.¹⁷

Scheme 42.

Key Step in the Total Synthesis of Streptonigrone

Fredericamycin A.

Fredericamycin A (217), a structurally unique and potent antitumor antibiotic, features an unusual spiro[4.4]nonene (CD ring system) and four aromatic rings including a pyridone (F ring). A key step in our total synthesis of fredericamycin A was the room temperature inverse electron demand Diels—Alder reaction of N-sulfonyl-1-aza-1,3-butadiene 212, bearing a noncomplementary C2 electron-withdrawing group, with 213 for the construction of the appropriately substituted pyridone F ring.¹⁸⁶ The non-complementary addition of the strong electron-withdrawing C2 ethoxycarbonyl group lowers the inherent low lying azadiene LUMO to the extent that even the modestly reactive dienophile 213 was found to react effectively at 25 °C (95%, Scheme 43). The cycloadduct 214 was converted directly to pyridine 215 by simple treatment with DBU. This was followed by a single-step Michael addition/intramolecular acylation upon LDA deprotonation and reaction of the anion with cyclopentenone and subsequent DDQ aromatization for annulation of the DE ring system.¹⁸⁶ Notably, the synthesis of the carbon skeleton of the DEF ring system (216) was accomplished in only four steps.

Scheme 43.

Key Step in the Total Synthesis of Fredericamycin A

Our convergent total synthesis of fredericamycin A featured two additional key transformations, a regiospecific intermolecular chromium carbene benzannulation reaction for AB ring construction and a simple aldol closure for the introduction of the spiro[4.4]nonene CD ring system.^186,187 With this approach, ent-fredericamycin A, the key partial structure 218, constituting the fully functionalized ABCDE ring system,¹⁸⁷ and 219, constituting the fully functionalized DEF ring system,¹⁸⁶ were prepared to address the origin of its biological properties. The important observations were made that both enantiomers of fredericamycin A exhibit potent and indistinguishable cytotoxic activity and that each fragment (ABCDE and DEF) contributes significantly to the biological properties of fredericamycin A.

Piericidin A1 and B1.

An analogous 1-azadiene bearing a noncomplementary C2 electron-withdrawing group was utilized in a total synthesis of piericidin A1 (227) and B1 (228). The piericidins are an important class of biologically active natural products. They are among the most potent inhibitors of the mitochondrial electron transport chain protein NADH-ubiquinone reductase (complex I, K_i = 0.6−1.0 nM) and have contributed extensively to the elucidation of the enzyme properties. It has been suggested that piericidins mimic the structure of enzyme cofactor ubiquinone (coenzyme Q) and competitively bind complex I.¹⁸⁸ In addition, the piericidins were identified as a class of highly selective antitumor agents, displaying a greater selectivity and potency than the comparison standards.¹⁸⁹

Central to our approach to the piericidins was an inverse electron demand Diels—Alder reaction of the N-sulfonyl-1-aza-1,3-butadiene 220, bearing a noncomplementary C2 carboxylate, with tetramethoxyethene (221) followed by Lewis acid promoted aromatization for synthesis of the fully functionalized pyridine core.¹⁹⁰ Treatment of 220 with tetramethoxyethene (221) smoothly afforded the [4 + 2] cycloadduct 222 (64%) at 50 °C, even though 221 is tetrasubstituted and sterically demanding (Scheme 44). The enhanced reactivity of 220 in the Diels—Alder reaction may be attributed to the lower inherent azadiene LUMO derived from the incorporation of a C2 electron-withdrawing group into the azadiene. Efforts to induce aromatization under a variety of basic conditions were unsuccessful, whereas the use of BF₃·OEt₂ cleanly affected this transformation and provided the fully substituted pyridine 223. Reduction of the ethyl ester followed by protection of the resulting primary alcohol provided 224. It was envisioned that directed lithiation of the C4′ position on the pyridine ring followed by a borylation/oxidation sequence would permit incorporation of the remaining C4′ hydroxyl group. Interestingly, treatment of 224 with an excess of butyllithium followed by trimethylborate and peracetic acid unexpectedly afforded the C-silylated product 225, resulting from both the desired C4′ hydroxylation and a competitive reverse brook rearrangement of the benzylic TIPS ether under the strongly basic conditions. The disilylation of 225 followed by Appel reaction of the resulting primary alcohol cleanly provided a surprisingly stable heterobenzylic bromide 226 and the left-hand fragment of piericidins.¹⁹⁰

Scheme 44.

Key Step in the Total Synthesis of Piericidins

Additional key steps in this synthesis include the use of an asymmetric anti-aldol reaction for installation of the C9 and C10 relative and absolute stereochemistry, convergent assemblage of the side chain through formation of the C5−C6 trans double bond with a modified Julia olefination, and a surprisingly effective penultimate benzylic Stille cross-coupling reaction of the pyridyl core with the fully functionalized side chain. Notably, this strategic late-stage, convergent coupling approach was anticipated to permit routine access to the analogues in which each half of the molecule could be systematically modified.¹⁹⁰ Consequently and in addition to the natural products, a series of key analogues were prepared.^190b Their evaluation permitted a scan of the key structural features of the piericidins, providing new insights into the importance and the role of each substituent found in both the pyridyl core as well as the side chain.

Nothapodytine B, (−)-Mappicine, and (+)-Camptothecin.

Camptothecin (238) is the parent member of a clinically important class of DNA topoisomerase I inhibitors that exhibit efficacious antitumor activity.¹⁹¹ Nothapodytine B (236), an oxidized derivative of mappicine (237), has been identified as an antiviral lead with selective activities against HSV-1, HSV-2, and human cytomegalovirus (HCMV). Since all three alkaloids possess the same ABCD ring system, we designed a single unified strategy for their total syntheses.¹⁹² Central to our approach was the implementation of a room temperature, inverse electron demand Diels—Alder reaction of the N-sulfonyl-1-aza-1,3-butadiene 229, bearing a noncomplementary C4 carboxylate, with two different dienophiles for the introduction of the pyridone D ring with assemblage of the full carbon skeletons of either nothapodytine B and (−)-mappicine or (+)-camptothecin (Scheme 45).

Scheme 45.

Key Step in the Total Synthesis of Nothapodytine B, (−)-Mappicine, and (+)-Camptothecin

Treatment of 229 with ketene acetal 207 at room temperature led to the formation of the sensitive [4 + 2] cycloadduct 230. Notably, the deliberate incorporation of the noncomplementary C4 electron-withdrawing substituent resulted in a Diels—Alder cycloaddition that proceeded at 25 °C without altering the inherent [4 + 2] cycloaddition regioselectivity. Subsequent aromatization of the cycloadduct under basic conditions provided the highly substituted pyridine 231 in 64% yield without intermediate isolations.^192b Addition of EtMgBr to 231 in the presence of a tertiary amine proceeded cleanly to give the corresponding ethyl ketone 232 without tertiary alcohol formation by virtue of tertiary amine-promoted ketone enolization. Finally, deprotection of both the benzylic and pyridone methyl ethers with in situ benzylic bromination and subsequent cyclization afforded nothapodytine B. Reduction of nothapodytine B with (S)-BINAL-H provided (−)-mappicine (73%, 99.9% ee).^192b

This strategy was also applied to a total synthesis of camptothecin.^192a The inverse electron demand Diels—Alder cycloaddition of the same 1-azadiene 229 with ketene acetal 235 proceeded at room temperature within 4 h to give the desired [4 + 2] cycloadduct 236. Addition of sodium ethoxide to the sensitive Diels—Alder cycloadduct 236 resulted in sequential elimination and aromatization to form the highly substituted pyridine 237 (70%) under mild conditions without intermediate isolations. Significantly, based on this unified strategy, concise total syntheses of nothapodytine B, (−)-mappicine and (+)-camptothecin were all accomplished in only five to nine steps from a readily available α,β-unsaturated ketone as the precursor to the N-sulfonyl-1-aza-1,3-butadiene 229.

Rubrolone.

Rubrolone (239), a red tropoloalkaloid, possesses the unique azuleno[2.3-c]pyridine-2,5,13-trione aglycon 240 characteristic of a class of structurally related agents. Key to our total synthesis of rubrolone was the implementation of a rare 4π participation of an O-alkyl α,β-unsaturated oxime in an intramolecular Diels—Alder reaction for construction of a tetrasubstituted pyridine and assemblage of the AB ring system.^184,193 In our studies, the intramolecular Diels—Alder reaction of 204 was observed at 170 °C and the [4 + 2] cycloadduct gradually converted to the desired pyridine 205 through in situ elimination of methanol (Scheme 46). No reaction was found to occur at temperatures lower than 140 °C, and a productive rate of reaction was observed at 170−200 °C to provide 205 (70%).¹⁸⁴

Scheme 46.

Key Step in the Total Synthesis of Rubrolone Aglycon

Precedent for this Diels—Alder reaction may be found in the use of α,β-unsaturated N,N-dimethylhydrazones in intermolecular¹⁹⁴ and intramolecular¹⁹⁵ [4 + 2] cycloaddition reactions. However, the previously reported failure of α,β-unsaturated oxime cycloadditions suggested that their use may not prove viable. Therefore, the disclosure that the O-alkyl α,β-unsaturated oxime 204 participates as an effective 4π component of an intramolecular Diels—Alder reaction with an electron-deficient dienophile indicates that the introduction of an alkoxy electron-donating substituent (OR) on the nitrogen atom of the inherently electron-deficient 1-aza-1,3-butadiene system, like the dimethylamino group of the unsaturated dimethylhydrazones (NMe₂), may be sufficient to promote its participation in a normal, HOMO_diene-controlled Diels—Alder reaction.¹⁸⁴

Additional highlights in our total synthesis of rubrolone aglycon include the introduction of the C-ring oxygenated troplone through a room-temperature intermolecular, exo-selective Diels—Alder reaction of the cyclopropenone ketal followed by in situ generation of a norcaradiene and room-temperature electrocyclic rearrangement.¹⁹³ The details of these transformations are elaborated in the ensuing paragraphs.

4. 3-CARBOMETHOXY-2-PYRONES

Much of the work we have described has involved the formation of an electron-rich olefin from an aliphatic carbonyl group followed by inverse electron demand [4 + 2] cycloaddition with a series of electron-deficient azadienes for the late-stage divergent synthesis of a wide range of heteroaromatics. Complementary to this, we developed an approach to a full range of aromatics bearing selectively protected and positionally varied oxygen substituents.¹⁹⁶ This process involves conversion of the carbonyl to an electron-deficient diene, a 3-carbomethoxy-2-pyrone, and its reaction in a series of inverse electron demand Diels—Alder reactions with electron-rich dienophiles bearing oxygen substituents. In this manner, a single aliphatic substrate bearing a carbonyl group provides a point of divergence for a full range of aryl annulations (Figure 34).

Figure 34.

Divergent late-stage aryl annulation.

Although the synthetic utility of α-pyrones in normal Diels—Alder reactions was well established prior to our work, their use in inverse demand reactions was less developed. C3 substitution of the pyrone with an electron-withdrawing group accelerates its participation in inverse electron demand [4 + 2] cycloadditions and enhances the observed regioselectivity. This observation was exploited for the divergent preparation of a full range of oxygen substituted aromatics and was extended to the total synthesis of a number of natural products, including sendaverine,^196a juncusol,¹⁹⁷ 6,7-benzomorphans,^196a and the azafluoranthene alkaloids, including rufescine, norrufescine, and imeluteine (Figure 35).¹⁹⁸ Unlike conventional approaches in which the synthesis typically begins with a precursor containing the aromatic ring, the late-stage divergent aryl annulation from a common precursor ketone permits the rapid and comprehensive examination of a key series of aryl and heteroaromatic natural products and analogues.¹⁹⁸

Figure 35.

3-Carbomethoxy-2-pyrone synthetic applications.

5. CYCLOPROPENONE KETALS

Cyclopropenone Ketal Cycloaddition Reactions.

Structural or electronic features that accelerate a normal Diels—Alder reaction would be expected to decelerate the corresponding inverse electron demand [4 + 2] cycloaddition. The exception to this generalization is the behavior of strained olefins, which react effectively with both electron-rich and electron-deficient dienes. This unique reactivity can be attributed to both their raised HOMO and lowered LUMO energy levels as well as the release of ring strain that accompanies their cycloaddition. In our studies on the total synthesis of tropoloalkaloids, we found that α-pyrones react with cyclopropenone ketals in Diels—Alder reactions, and were able to extend this observation to their reaction with a variety of electron-rich and electron-deficient dienes (Figure 36).¹⁹⁹ In these investigations, we also discovered a remarkable thermal [3 + 4] cycloaddition reaction of cyclopropenone ketals with α-pyrones that provided a complementary approach to tropone annulation.¹⁹⁹ This additional reactivity was found to be derived from a reversible, thermal generation of a π-delocalized singlet vinylcarbene and its 2π participation in a π2s + π4s cycloaddition. The full scope of the cycloaddition reactivity of such thermally generated π-delocalized singlet vinylcarbenes was defined and was found to extend to a range of additional [3 + 2] and [1 + 2] cycloaddition reactions with electron-deficient olefins and carbonyls (Figure 36).^199,200

Figure 36.

Cyclopropenone ketal cycloaddition reactions.

Recently, we disclosed the intramolecular cycloaddition reactions of cyclopropenone ketals.²⁰¹ Such ketals tethered to olefins bearing a single electron-withdrawing substituent were found to undergo [3 + 2] cycloadditions in excellent yields under mild conditions (80 °C) while others displayed a bifurcated reactivity, providing intermediate cyclopropanes derived from an initial endo [1 + 2] cycloaddition of the thermally generated π-delocalized singlet vinylcarbene under mild thermal conditions (80−100 °C) and the [3 + 2] cycloadducts at higher reaction temperatures (170−180 °C, Figure 37). The ease of accessing the [3 + 2] cycloadducts directly under mild conditions was found to be dependent on the nature of the linking chain as well as the nature of the electron-withdrawing substituent. The corresponding intramolecular reactions with electron-rich, electron-deficient, and neutral 1-substituted dienes were also investigated and found to proceed under mild conditions and provide exclusive exo [4 + 2] cycloaddition products in normal, inverse electron demand, and neutral Diels—Alder reactions. These reactions do not depend on reaction conditions, diene substituents, or nature of the linking tethers and proceed without the intervention or trap of the vinylcarbenes.

Figure 37.

Intramolecular cycloaddition reactions.

Colchicine.

The cycloaddition reactions of cyclopropenone ketals first found application in the total synthesis of colchicine.²⁰² The first of the tropolone alkaloids to be identified, colchicine is a historically important antimitotic agent that inhibits cell division through tubulin binding and exhibits antitumor activity. Even today, new therapeutic applications of colchicine are emerging that would be exciting to probe, including its use in the treatment of cerebrovascular ischemia, autoinflammatory diseases, atherosclerosis-associated inflammation, dermatology, and atrial fibrillation.

Our approach to colchicine was complementary to a series of pioneering total syntheses²⁰³ and was based on a remarkable thermal [3 + 4] cycloaddition reaction of a cyclopropenone ketal with Eschenmoser’s α-pyrone, a reaction that proceeds by reversible thermal generation of a π-delocalized singlet vinylcarbene and its 2π participation in a π2s + π4s cycloaddition (Figure 38). More than 35 years after our disclosure of this synthesis, it remains one of the most concise and efficient total synthesis disclosed to date²⁰³ and likely could be further improved with implementation of modern advances in the field.

Figure 38.

Key step in the total synthesis of colchicine.

Azafluoranthene and Tropoloisoquinoline Alkaloids.

In efforts on the total synthesis of the azafluoranthene alkaloids,¹⁹⁸ we envisioned the development of a divergent synthesis of the tropoloisoquinoline alkaloids, including imerubrine, isoimerubrine, grandirubrine, and granditropone.²⁰⁴ These efforts, in which the powerful concept of “divergent total synthesis” was introduced as a synthetic strategy, also inspired our development of methodology for late-stage aromatic annulation (Figure 39).^198,205 Key to the approach to the tropoloisoquinoline alkaloids was the use of a room temperature pressure-promoted [4 + 2] cycloaddition of a cyclopropenone ketal with a common α-pyrone also utilized for the total syntheses of the azafluoranthene alkaloids. Additional highlights include the development of a modified isoquinoline synthesis,^206a the use of a direct synthesis of 2-canyoquionlines via N-tosyl Reissert intermediates,^206b the implementation of an effective α-pyrone synthesis utilizing Meldrum’s acid, and a regioselective hydroxylation of granditropone.

Figure 39.

Divergent synthesis of azafluoranthene and tropoloisoquinoline alkaloids.

Rubrolone.

The [4 + 2] cycloaddition reaction of cyclopropenone ketals was also utilized in a total synthesis of the rubrolone aglycon.^193,207 Rubrolone, isolated from Streptomyces enchinoruber,²⁰⁸ was identified in a single-crystal X-ray structure determination and shown to possess a unique aglycon characteristic of a class of structurally related agents.²⁰⁸ Our approach, which is only one of two total syntheses disclosed to date,²⁰⁹ centered on two key steps. The AB ring system of the natural product was constructed by an intramolecular [4 + 2] reaction of an O-alkyl α,β-unsaturated oxime and a tethered alkyne, and the C ring was introduced with the use of a Diels—Alder reaction between a cyclopropenone ketal and a highly oxygenated AB ring system as a prelude to C-ring tropolone introduction (Scheme 47).

Scheme 47.

Tropolone Introduction and Completion of the Total Synthesis of the Rubrolone Aglycon

The key cyclopropenone ketal Diels—Alder reaction proceeded at room temperature to provide a single diastereomer in superb yield (97%), derived from exclusive cycloaddition through the less sterically encumbered exo transition state (Figure 40). Bromination, selective deprotection of the dimethyl ketal, base-promoted elimination of HBr followed by acid-catalyzed hydrolysis of the mixed ketal provided tropolone 256. This reaction sequence proceeds through mixed ketal hydrolysis, enolization with norcaradiene generation, electrocyclic rearrangement to the cycloheptatrie-none ketal, and tautomerization to 256.

Figure 40.

Exo [4 + 2] cycloaddition of a cyclopropenone ketal en route to the rubrolone aglycon.

6. CONCLUSION

A remarkable series of more than 70 total syntheses of complex natural products are summarized, each of which enlisted a key inverse electron demand Diels—Alder reaction central to the synthetic approach and emerged from our studies spanning nearly 40 years. Underlying these efforts are fundamental studies on the discovery, development, or scope of the cycloaddition reaction of heterocyclic azadienes, 1-aza-1,3-butadienes, cyclopropenone ketals, α-pyrones, and related systems. Perhaps the single most remarkable feature of the studies is the number of key cycloaddition reactions that were conducted at room temperature. In practice, we have found that it is easier to match the complementary diene/dienophile reactivity in the LUMO_diene-controlled cycloadditions than is typically accessible to the normal Diels—Alder reaction. The rate, predictable intrinsic reaction regioselectivity, mode of cycloaddition, and diastereoselectivity of the inverse electron demand Diels—Alder reactions rival or surpass those of normal Diels—Alder reactions. Our recent discovery of the solvent hydrogen bonding catalysis (HFIP or TFE) of heterocyclic azadiene cycloadditions now not only allows the reactions to be conducted under milder conditions with a broader scope than previously realized, but also offers a general fundamental feature for future use analogous to the impact of Lewis acid catalysis on the normal Diels—Alder reaction. The applications of these methods in the total syntheses of the series of natural products highlight the power of the methodology, especially for the synthesis of complex heterocyclic ring systems. In the context of our work, the advent of the synthetic methodology not only permitted the preparation of a series of targeted biologically important natural products but also enabled the divergent syntheses of key partial structures, and analogues incorporating deep-seated structural modifications. This provided the opportunity to gain valuable insights into the structure—function relationships of the natural products, to explore the origin of their biological properties, and to design compounds with improved potency and selectivity. In most instances, it was the targeted natural products that inspired the methodology development or discovery. The development of previously unknown heterocyclic azadienes including the 1,2,3,5-tetrazines, extensions of the recent studies on solvent hydrogen bonding catalysis of the Diels—Alder reaction, alternative tethering strategies for the intramolecular cascade cycloaddition reactions of 1,3,4-oxadiazoles for accessing additional alkaloid scaffolds, key applications of the late-stage divergent aryl annulation strategy, and the total synthesis of additional natural product classes utilizing these approaches are ongoing and will be detailed in due course.

Biographies

Dale Boger received his BSc in Chemistry from the University of Kansas, his PhD in Chemistry from Harvard University with E. J. Corey, and served on the faculty at the University of Kansas (Medicinal Chemistry) and Purdue University (Chemistry) before joining The Scripps Research Institute in 1990 where he is the Richard and Alice Cramer Professor of Chemistry.

Vyom Shukla received his BA in Chemistry from Reed College (2014) where he worked with Professor Alan Shusterman and Professor Summer Gibbs of Oregon Health & Science University on the synthesis and application of photoswitchable fluorophores for super-resolution microscopy. He is currently a PhD candidate at The Scripps Research Institute (2014−present) with Professor Boger, working on the total synthesis and bioorganic investigation of alkaloid natural products.

Jiajun Zhang received his BSc in Chemistry from Peking University, where he conducted research in total synthesis of Schisandraceae triterpenoids and vibsane diterpenoids under the supervision of Professor Zhen Yang and Professor Jiahua Chen. In 2013, he moved to The Scripps Research Institute for graduate studies with Professor Boger, focusing on the new generation total synthesis of vinblastine and previously inaccessible analogues.