Dehydrogenative Pd and Ni Catalysis for Total Synthesis

CONSPECTUS:

The development of novel synthetic methods remains a cornerstone in simplifying complex molecule synthesis. Progress in the field of transition metal catalysis has enabled new mechanistic strategies to achieve difficult chemical transformations, increased the value of abundant chemical building blocks, and pushed the boundaries of creative and strategic route design to improve step economy in multistep synthesis. Methodologies to introduce an olefin into saturated molecules continue to be essential transformations because of the plethora of reactions available for alkene functionalization. Of particular importance are dehydrogenation reactions adjacent to electron-withdrawing groups such as carbonyls, which advantageously provide activated olefins that can be regioselectively manipulated. Palladium catalysis occupies a central role in the most widely adopted carbonyl dehydrogenation reactions, but limits to the scope of these protocols persist. In this Account, we describe our group’s contributions to the area of transition-metal-catalyzed dehydrogenation using palladium catalysis and more sustainable and economical nickel catalysis. These metals are used in conjunction with allyl and aryl halides or pseudohalides that serve as oxidants to access a unique mechanistic approach for one-step α,β-dehydrogenation of various electron-withdrawing groups, including ketones, esters, nitriles, amides, carboxylic acids, and electron-deficient heteroarenes. The pivotal reaction parameters that can be modified to influence reaction efficiency are highlighted, including base and oxidant structure as well as ligand and salt additive effects. This discussion is expected to serve as a guide for troubleshooting challenging dehydrogenation reactions and provide insight for future reaction development in this area.

In addition to enabling dehydrogenation reactions, our group’s allyl-Pd and -Ni chemistry can be used for C–C and C–X bond-forming reactions, providing novel disconnections with practical applications for expediting multistep synthesis. These transformations include a telescoped process for ketone α,β-vicinal difunctionalization; an oxidative enone β-functionalization, including β-stannylation, β-silylation, and β-alkylation; and an oxidative cycloalkenylation between unstabilized ketone enolates and unactivated alkenes. These bond-forming methodologies broaden the range of transformations accessible from abundant ketone, enone, and alkene moieties. Both the dehydrogenation and C–C and C–X bond-forming methodologies have been implemented in our group’s total synthesis campaigns to provide step-efficient synthetic routes toward diverse natural products.

Through the lens of multistep synthesis, the utility and robustness of our dehydrogenation and dehydrogenative functionalization methodologies can be better appreciated, and we hope that this Account will inspire practitioners to apply our methodologies to their own synthetic challenges.

Graphical Abstract

INTRODUCTION

The step efficiency and scalability of multistep synthesis are propelled through the development of robust chemical transformations. Methods development provides a means to increase the value of starting materials and enables strategic combinations of steps to simplify synthesis. Transition metal catalysis remains a critical horizon to advance such developments.

The introduction of an olefin adjacent to an electron-withdrawing group is a fundamental transformation in organic chemistry because of the synthetic utility and versatility of the resulting polarized double bond, which is susceptible to a variety of methods for regioselective functionalization. As a result, α,β-unsaturated carbonyls are common building blocks in organic synthesis. Dehydrogenation reactions are therefore commonly employed in natural product and pharmaceutical synthesis, as reactive alkene groups are unveiled from their saturated counterparts.⁵

While many approaches for carbonyl dehydrogenation have been reported, including classical two-step approaches involving α-functionalization and elimination⁶ or one-step approaches using stoichiometric DDQ⁷ or IBX,⁸ transition-metal-catalyzed dehydrogenation has emerged as one of the most widely employed strategies because of its broad applicability in multistep synthesis.⁵ Pioneering studies by Saegusa demonstrated that enoxysilane derivatives of ketones undergo transmetalation to provide Pd enolates that yield enones upon β-hydride elimination.⁹ Relatedly, Tsuji showed that allyl β-keto esters and allyl enol carbonates undergo decarboxylation to generate allyl-Pd enolates for enone formation (Scheme 1).¹⁰ In addition to requiring two steps, these early approaches often necessitate high loadings of Pd for efficient conversion.

Scheme 1.

Pioneering Work in Pd-Catalyzed Dehydrogenation

Recent elegant advances by Stahl utilize a mechanistically distinct Pd-catalyzed α-C–H functionalization approach for aerobic ketone and aldehyde dehydrogenation.¹¹ Dong developed a related Pd-catalyzed α-C–H functionalization approach for carbonyl dehydrogenation that was leveraged for ketone β-functionalization.¹² More recently, Dong demonstrated that a soft enolization strategy for carbonyl dehydrogenation can be achieved using Pd or Pt catalysis.¹³

Despite these advances in Pd-catalyzed dehydrogenation, the scope of carbonyls that can be efficiently oxidized is largely limited to relatively acidic ketones, aldehydes, or specialized substrates. To overcome these limitations, our group developed a dehydrogenation approach that harnesses undesired reactivity in Pd-catalyzed alkyl cross-coupling to achieve a one-step α,β-dehydrogenation of less acidic functional groups.^5c We further expanded the catalysis from Pd to Ni, providing new reactivity that enabled further reaction discovery.

Herein we provide a comprehensive summary of our group’s contributions to transition-metal-catalyzed dehydrogenation using unconventional but readily available allyl- and aryl-based organic oxidants. This Account highlights the key modifiable reaction parameters that impact the reaction efficiency and serves as a guide for practitioners applying this methodology to their own synthetic challenges. Additionally, this Account covers our efforts to extend our dehydrogenation chemistry to oxidative C–C and C–X bond-forming reactions. Practical applications of these methodologies were demonstrated in several of our group’s natural product synthesis campaigns, highlighting the potential of our methodologies to expedite the construction of complex molecules.

ALLYL-PD-CATALYZED DEHYDROGENATION

The general experimental protocol for allyl-metal-catalyzed carbonyl dehydrogenation involves substrate deprotonation with a strong base followed by introduction of a catalyst and allyl oxidant mixture to promote β-C–H cleavage. Our current understanding of the underlying mechanism of this transformation is shown in Scheme 2. First, enolization of a carbonyl compound (A) with a lithium amide base followed by transmetalation with a zinc salt or direct deprotonation with a bis(amido)zinc base generates the zinc enolate (B), which is needed to enter the catalytic cycle. This zinc enolate then undergoes transmetalation with an allyl-Pd or -Ni species (J) to form the corresponding allyl-metal enolate C, which undergoes β-hydride elimination to give the dehydrogenated product E. Allylation byproducts (D) can be formed via reductive elimination of allyl-metal enolate C or by nucleophilic substitution of allyl-metal species J by enolate B. After β-hydride elimination, the resulting allyl-metal hydride (F) undergoes reductive elimination to afford propene gas (G) as a stoichiometric byproduct and low-valent metal species H that completes the catalytic cycle via oxidative addition to allyl electrophile I.

Scheme 2.

Mechanism of Allyl-Pd/Ni-Catalyzed Carbonyl α,β-Dehydrogenation

In our initial reports of allyl-Pd-catalyzed ester dehydrogenation,¹⁴ a parallel kinetic isotope effect (KIE) experiment provided a primary KIE of 2.3. This result suggests that C–H bond cleavage (β-hydride elimination) or C–H bond formation (reductive elimination) is involved in the rate-determining step (RDS) (Scheme 3A). In order to elucidate the RDS, the related intramolecular competition experiment was conducted using 1a-d₁, and a KIE of 1.0 was observed, suggesting that reductive elimination is rate-determining (Scheme 3B).

Scheme 3.

KIE Experiments for Allyl-Pd-Catalyzed Carbonyl Dehydrogenation

Zinc enolates are essential for allyl-Pd-catalyzed dehydrogenation (Scheme 4). When LiTMP was employed as the base, numerous byproducts were observed, such as byproducts of allylation (3) and Claisen condensation (4), arising from direct reaction of the lithium enolate with the allyl oxidant. However, those byproducts were suppressed when ZnCl₂ was added to the reaction mixture to form a less reactive and more functional-group-tolerant zinc enolate that leads to improved selectivity for dehydrogenated product 2a.

Scheme 4.

Zinc Enolates are Crucial for Dehydrogenation

Early studies on the effect of various oxidants in the context of nitrile dehydrogenation revealed that aryl-based oxidants such as bromobenzene (Ox 1) gave lower yields than allyl-based oxidants (Ox 2), indicating that the oxidant structure is important to the reaction efficiency (Scheme 5). Furthermore, an exploration of the different leaving groups of allyl-based oxidants indicated that less electrophilic C–O bond-based oxidants such as allyl pivalate (Ox 5) and allyl acetate (Ox 6) provided the highest yields of dehydrogenation product, likely as a result of their decreased susceptibility to enolate substitution reactions. Interestingly, Dong reported allyl carbonates to be the optimal oxidants for Pt-catalyzed carbonyl dehydrogenation enabled by soft enolization,^13b attesting to the generality of allyl-based oxidants for dehydrogenation.

Scheme 5. Oxidant Structure and Leaving Group Effects^a.

^aThe ¹H NMR yields of 6a are indicated in parentheses and were determined using 1,3,5-trimethoxybenzene as an internal standard.

By modification of the metal enolate and the oxidant structure, the first examples of one-step ester and nitrile dehydrogenation were achieved. Selected examples of the substrate scope are shown in Scheme 6. Our mechanistic approach for dehydrogenation allows for excellent functional group compatibility, including aryl chlorides (2b) and readily oxidizable amines (2d and 6c) and furans (6d). Both di- and trisubstituted alkenes were efficiently prepared. While dehydrogenation of esters exclusively provided (E)-alkenes, nitriles gave mixtures of E and Z isomers. Comparable dehydrogenation methods, such as selenoxide elimination, give similar isomeric ratios.^6e

Scheme 6.

Dehydrogenation of Esters and Nitriles

After establishing allyl-Pd catalysis as an effective mechanistic strategy for the dehydrogenation of less acidic carbonyls, our group explored the dehydrogenation of other important substrates, such as amides,² which are present in proteins, bioactive molecules, and polymers. Furthermore, unsaturated amides are an important class of covalent inhibitors used in drug discovery.¹⁵ Under our optimized ester and nitrile dehydrogenation conditions, model amide substrate 7a underwent inefficient dehydrogenation, providing unsaturated amide 8a in only 26% yield. Modifications of the standard reaction parameters such as oxidant, temperature, time, and equivalents were ineffective.

Structural modifications to the lithium amide base, which may affect the ligand sphere on the reactive allyl-Pd species, were found to influence the dehydrogenation efficiency (Scheme 7). Attempts to use other commercial bases such as LDA (B2) and LiNCy₂ (B3) provided an improvement in yield but with incomplete conversion of the starting amide. Keeping one N-cyclohexyl substituent constant, an arene substituent was introduced to provide additional means of structural modification. While the simple benzene substituent resulted in decreased yield (B4), the introduction of a 2,6-diisopropylbenzene substituent (B5, LiCyan) resulted in complete conversion and 99% yield. Replacing the isopropyl groups with isopropoxy groups (B6) resulted in worse conversion and yield, and further optimization of the alkyl substituent established the order Ph < ^tBu < ⁱPr ≈ Cy. Gratifyingly, decreasing the catalyst loading to 0.25 mol % did not lead to yield depreciation. CyanH is readily synthesized on a preparative scale via reductive amination and recrystallization and can be recovered from dehydrogenation reaction mixtures in up to 99% yield. The identification of LiCyan as an efficient base for allyl-Pd-catalyzed amide dehydrogenation demonstrates that structural modifications to the base is an effective means to tune reactivity.

Scheme 7. Novel Lithium Anilide Base for Amide Dehydrogenation^a.

^aThe conversion of 7a is indicated in parentheses.

With these modified conditions, a variety of amides and lactams were efficiently dehydrogenated. Furthermore, oxidation-prone OH and NH functionalities were well-tolerated by the use of additional equivalents of base, and a selected scope is shown in Scheme 8. Secondary alcohols (7b) and mono-Boc- (7c) or Ts-protected (7d) aliphatic amines remained intact, thereby avoiding additional protection and deprotection concession steps. As mono-N-substituted amides are tolerated as spectator groups (7e), a protecting group can be used to the control chemoselectivity between two amides. Leveraging the difference in acidity and driving force for aromatization allowed the selective dehydrogenation of a PMB-protected lactam in the presence of a dibenzyl amide (7f). Additionally, the use of LiCyan as the base enabled the dehydrogenation of esters bearing free OH and NH functionality, including substrates with unprotected indole (2e), primary alcohol (2f), and unprotected carbamate (NHCbz) functionality (2g).¹⁶

Scheme 8.

Dehydrogenation of Amides and Esters Containing Free OH and NH Functionality

In addition to carboxylic acid derivatives, our dehydrogenation strategy was extended to the dehydrogenation of the parent carboxylic acids,¹⁷ which are naturally abundant and important in many biological processes. Significant challenges are associated with Pd-catalyzed α,β-dehydrogenation of carboxylic acids. The formation of the necessary C-bound Pd enolate for β-hydride elimination is complicated by the tendency for Pd to coordinate to the carboxylate oxygen center. Additionally, Pd is known to effect the decarboxylation of both carboxylic acids and enoic acids.¹⁸ While our optimized conditions for ester, nitrile, and amide dehydrogenation were ineffective for acid dehydrogenation, we found that premixing the bis(amido)zinc base Zn(TMP)₂·2LiCl (generated in situ from LiTMP and ZnCl₂) with excess ZnCl₂ led to remarkably high yields of unsaturated natural and synthetic fatty acids. This transformation is thought to proceed via zinc enediolate 11, and a selected scope is shown in Scheme 9. Interestingly, substrates bearing an olefin did not require the ZnCl₂ additive for full conversion (10a). Additionally, Pd- and base-sensitive functional groups such as alkyl bromides (10b) and epoxides (10c) were tolerated.

Scheme 9. Dehydrogenation of Carboxylic Acids.

^a2.3 equiv of Zn(TMP)₂·2LiCl was used, and no ZnCl₂ was added. The reaction was quenched after 3 h.

While our early dehydrogenation work focused on less acidic substrates that did not have one-step dehydrogenation protocols, we recognized the importance of also developing an allyl-Pd-catalyzed dehydrogenation of ketones, despite the existing reports of one- and two-step approaches.⁵ Our allyl-Pd-catalyzed methodology is complementary to methods that operate under acidic conditions^8,11,12 and provides advantages over basic dehydrogenation methods such as Mukaiyama’s protocol using N-tert-butylsulfinimidoyl chloride,¹⁹ which generates stoichiometric toxic sulfur byproducts that complicate purification. In particular, we were interested in the dehydrogenation of complex polycyclic ketones typically found in terpenoid natural products. While attempts to apply our optimized conditions for dehydrogenation of carboxylic acid derivatives provided poor results, a specific combination of commercial, salt-free Zn(TMP)₂²⁰ as the base and diethyl allyl phosphate²¹ as the oxidant enabled efficient ketone dehydrogenation,²² providing access to cycloalkenones 13a–f, including those bearing acid-sensitive functionality (Scheme 10). Overoxidation to the phenol was observed for 13a, but this could be avoided with controlled stoichiometry. Unfortunately, acyclic ketones were unsuccessful substrates under these conditions. Importantly, the presence of salts (LiCl, ZnCl₂) in the reaction was detrimental, and the effectiveness of diethyl allyl phosphate as the oxidant is thought to be the result of the weak coordinating ability of the diethyl phosphate ion, which may result in longer catalyst lifetimes and the observed slower precipitation of Pd black.

Scheme 10.

Cyclic Ketone Dehydrogenation

NICKEL-CATALYZED DEHYDROGENATION

While we have demonstrated that allyl-Pd-catalyzed dehydrogenation can be applied to a broad range of carbonyls and that modifications of the base, oxidant, and salt additives influence the reactivity, we were motivated to explore whether a more sustainable and cost-effective transformation could be achieved using an earth-abundant first-row transition metal.²³ Furthermore, we wondered what new transformations could be accessed if catalysis could be extended to first-row metals. A broad examination of first-row transition metal salts as catalysts (Mn, Fe, Co, Ni, Cu) revealed that only Ni could participate in allyl-metal-catalyzed dehydrogenation,²⁴ despite the higher energetic barrier for β-hydride elimination with Ni relative to Pd.²⁵ Interestingly, the use of diethyl allyl phosphate as the oxidant was essential for enabling allyl-Ni catalysis, as the use of allyl acetate or allyl methyl carbonate led to rapid precipitation of Ni and no conversion. While polar amide-based solvents are typically employed in Ni catalysis, our enolate chemistry dictated the use of ethereal solvents, which provide poor solubility for Ni(II) precatalysts. Diethyl allyl phosphate is thought to occupy a dual role in allyl-Ni catalysis, acting as both an oxidant and a polar additive that solubilizes Ni to enhance the reaction efficiency.

With this advance, we achieved the first practical report of carbonyl dehydrogenation catalyzed by a first-row transition metal (Scheme 11).³ Enolization with LiCyan and ZnBr₂ was most effective for acyclic ketone (15a), ester (15b), nitrile (15c), and amide (15d) substrates. Doubly activated substrates such as α-aryl esters and nitriles (15b and 15c), which predominantly provided allylation byproducts under Pd catalysis, were well-tolerated using a less electrophilic allyl-Ni species. Cyclic ketones (15e), lactones (15f), and lactams (15g) were also dehydrogenated using Ni, and higher yields were obtained when Zn(TMP)₂ was used as the base, thereby enabling a simplified protocol involving only a single transmetalation.

Scheme 11. Allylnickel-Catalyzed Dehydrogenation.

^a1,4-Dioxane was used as the solvent. ^b1.0 equiv of diethyl allyl phosphate was used.

These exciting findings expanded our horizons in Ni catalysis and motivated us to explore the dehydrogenation of functionalities beyond carbonyls. Because of their importance in pharmaceuticals, electron-deficient heteroarenes were investigated with this catalyst system.²⁶ Functionalization of heteroarenes using Ni is a well-recognized challenge because of the propensity of the Lewis basic N heteroatoms to coordinate to Ni and poison the catalyst. Attempts to employ allyl-Ni catalysis with Zn(TMP)₂ as the base led to poor yields for benzylic dehydrogenation of 16a, with benzylic allylation byproducts accounting for the mass balance. The hetero-benzylic anions generated upon deprotonation bear structural and electronic similarities to azaenolates, which are more nucleophilic than their carbonyl counterparts, and are more prone to nucleophilic substitution with allyl-metal intermediates. This challenge was overcome by employing heteroaryl halides as oxidants, which generate less electrophilic aryl-Ni intermediates that cannot undergo outer-sphere substitution (Scheme 12).

Scheme 12. Thiophene-Based Oxidant for Benzylic Dehydrogenation^a.

^aThe conversion of 16a is indicated in parentheses.

When bromobenzene (Ox 11) was employed as the oxidant, significant amounts of benzylic arylation byproducts were observed, suggesting the need to suppress reductive elimination. As electron-rich arenes participate more slowly in reductive elimination,²⁷ we employed electron-rich heteroaryl bromides such as furans and thiophenes as oxidants (Ox 12–18) and found a dramatic increase in the dehydrogenation yield. The thiophene scaffold was chosen for its increased stability and availability, and of the commercial substituted thiophene electrophiles explored, 2-bromo-5-methylthiophene (Ox 17) has the optimal structural and electronic properties for promoting Ni-catalyzed benzylic dehydrogenation.

This aryl-Ni-catalyzed dehydrogenation was further promoted by tetrabutylammonium iodide (TBAI), which improved the selectivity to provide 17a in 94% yield (Scheme 13A). Additionally, a small electron-rich phosphine ligand, PMe₃, was crucial for promoting high conversion, which is consistent with the idea that increased electron density at the metal center slows reductive elimination, resulting in more efficient β-hydride elimination (Scheme 13B).²⁸ These combined oxidant, additive, and ligand effects represent additional important tunable parameters for improving dehydrogenation reactions and may be harnessed for the development of new oxidative transformations.

Scheme 13.

Additive and Ligands Effects in Benzylic Dehydrogenation

Under the optimized conditions for benzylic dehydrogenation, various 2-alkyl heteroarenes were efficiently oxidized, including pyridines (17b, 17d), pyrimidines (17f), pyrazines (17g), pyridazines (17h), and triazines (17e) (Scheme 14). The importance of the coordinating N heteroatom is highlighted by 17b, in which only the benzylic position adjacent to the nitrogen is dehydrogenated, without overoxidation to the corresponding quinoline despite the thermodynamic driving force for aromatization. In contrast, 5,6-dihydroisoquinoline (17c) cannot be accessed using this methodology because of the lack of an adjacent N heteroatom at the benzylic position. Furthermore, triazine substrate 16e undergoes double dehydrogenation to give the aromatized product 17e because of the presence of a N heteroatom adjacent to both benzylic positions.

Scheme 14. Benzylic Dehydrogenation of Electron-Deficient Heteroarenes.

^a2.1 equiv of Ox 17 was used.

KIE experiments were conducted to probe the rate-determining step of the benzylic dehydrogenation. An intramolecular competition experiment with 16h-d₁ showed a primary KIE of 2.8 (Scheme 15A). The related intermolecular competition experiment between 16h and 16h-d₂ showed a diminished KIE of 1.4 (Scheme 15B). These combined results suggest that neither β-hydride elimination nor reductive elimination is rate-determining²⁹ and contrasts with our previous studies indicating that reductive elimination was rate-determining for allyl-Pd-catalyzed carbonyl dehydrogenation.¹⁴

Scheme 15.

KIE Experiments in Aryl-Ni-Catalyzed Benzylic Dehydrogenation

C–C AND C–X BOND FORMATION

Our group’s allyl-metal catalysis can be used beyond α,β-dehydrogenation reactions. We have successfully exploited this reactivity to develop oxidative C–C and C–X bond-forming reactions, providing step-efficient approaches for rapid conversion of abundant starting materials to highly functionalized products with increased structural complexity.

The aforementioned allyl-Pd-catalyzed ketone dehydrogenation was used to develop a one-pot ketone β-functionalization reaction¹² via telescoping with organocuprate conjugate addition chemistry.³⁰ Organocuprates generated from organolithium, Grignard, or Negishi reagents were directly added to the enone mixtures generated by allyl-Pd-catalyzed dehydrogenation to provide β-functionalized ketones in a single synthetic operation. This telescoped process is made possible by the compatibility of the basic and anhydrous conditions employed in both processes as well as the lack of incompatible byproducts formed from dehydrogenation. Existing one-step dehydrogenation protocols typically employ incompatible acidic conditions and polar aprotic solvents^8,11 and are not amenable to telescoping with organocuprate chemistry. Because of the robustness of both the dehydrogenation and conjugate addition, a broad range of complex ketones were directly β-functionalized with various organocuprates, including those bearing alkenyl (18a), aryl (18b), heteroaryl (18c, 18d), acyl (18e), and functionalized alkyl (18f–i) functionalities (Scheme 16).

Scheme 16. Telescoped β-Functionalization of Ketones.

^aOrganocuprate derived from a Grignard reagent. ^bOrganocuprate derived from an organolithium reagent. ^cOrganocuprate derived from a lithiated aldimine reagent. ^dOrganocuprate derived from a Negishi reagent.

Remarkably, the enolates generated after the telescoped dehydrogenation–conjugate addition process could be intercepted by electrophiles, leading to a one-pot α,β-vicinal difunctionalization of ketones. Methyl (19a, 19d), allyl (19b), and benzyl (19f) halides were viable electrophiles for the three-component coupling (Scheme 17). Furthermore, employing paraformaldehyde as an electrophile provided β-hydroxymethyl ketones (19c, 19e) via an aldol reaction using ZnCl₂ as a crucial promoter. This telescoped approach allows for the rapid, programmable installation of molecular complexity to ubiquitous ketones.

Scheme 17. Telescoped α,β-Vicinal Difunctionalization of Ketones.

^aOrganocuprate derived from a Grignard reagent. ^bOrganocuprate derived from an organolithium reagent.

Leveraging our studies on ketone dehydrogenation, we reasoned that an oxidative β-functionalization of enones would be possible by generating the requisite metal enolate through a 1,4-conjugate addition.⁴ We recognized that enolates generated from conjugate addition to an enone may similarly undergo transmetalation to access allyl-Pd enolates for dehydrogenation. Using Pd catalysis and diethyl allyl phosphate as the oxidant, we identified that stannyllithium and silyllithium nucleophiles effectively underwent conjugate addition and provided reactive lithium enolates that participated in allyl-Pd-catalyzed dehydrogenation to give β-stannyl (21a–c) and β-silyl (21d–f) enones (Scheme 18). Interestingly, vinyl silane product 21f demonstrates that a vinylogous dehydrogenation of lactones is feasible. Direct oxidative C–C bond formation at the enone β-position (21g–i) required the use of mixed triorganozincate nucleophiles generated from organolithium or Grignard species. It is noteworthy that transmetalation occurs more rapidly with the zinc enolate than the β-stannyl-enone product. Unfortunately, the use of more general organocuprate conjugate addition chemistry was not feasible, thereby limiting the nucleophile scope.

Scheme 18. Oxidative β-Stannylation, Silylation, and Alkylation of Enones.

^a1.0 equiv of Bu₃SnZnEt₂Li and 2.3 equiv of diethyl allyl phosphate were used. ^bR₃ZnLi was used. ^cR₃ZnMgCl was used.

To further expand the repertoire of allyl-metal chemistry, a mechanistic modification of the allyl-Ni-catalyzed dehydrogenation process enabled an oxidative cycloalkenylation of ketones.² By incorporation of a migratory insertion step prior to β-hydride elimination, a C–C bond can be directly formed between an unstabilized ketone enolate and a pendent unactivated alkene (Scheme 19). This process precludes the need to preactivate the ketone as the corresponding enoxysilane³¹ or the alkene as the corresponding vinyl halide,³² thereby enabling the use of more abundantly available precursors to enable a step-efficient process for ring formation.³³ To adapt our allyl-Ni-catalyzed ketone dehydrogenation for an oxidative ketone cycloalkenylation, we found that the addition of ZnBr₂ led to a higher-yielding reaction and a broader scope. We speculate that ZnBr₂ changes the aggregation state of the enolate or modifies the ligand sphere on Ni to provide a more efficient cycloalkenylation. The addition of salts may slow down β-hydride elimination such that migratory insertion proceeds first, which is consistent with our observation that ketone dehydrogenation requires a salt-free reaction mixture.²² An additional literature report suggests that ZnBr₂ may abstract halides from Ni centers to generate cationic Ni intermediates,³⁴ helping to improve the rate of olefin insertion.

Scheme 19.

Mechanistic Strategy for Oxidative Cycloalkenylation

To highlight the power of this methodology, we sought to demonstrate the variety of various hetero- and carbocyclic architectures that could be prepared through the oxidative cycloalkenylation (Scheme 20). The formation of fused bicycles was possible, and both five- and six-membered rings could be accessed (23a–i). In contrast, six-membered-ring formation using existing cycloalkenylation strategies involving Pd catalysis and enoxysilanes is challenging.³¹ While this transformation typically required the use of terminal olefins, olefins substituted with an electron-withdrawing group (23a) were tolerated. Monocyclic enones and spirocycles (23d–f) were also prepared from the corresponding acyclic methyl or ethyl ketones bearing pendent olefins. Upon cyclization and oxidation, the resulting olefin isomerizes into conjugation with the carbonyl to give the more thermodynamically stable α,β-enone products. Excitingly, bridged bicycles were accessed despite the possibility of early β-hydride elimination to give undesired α,β-dehydrogenation products. Lower reaction temperatures with longer reaction times improved the selectivity for cycloalkenylation over dehydrogenation (23g). For all substrates, the optimized conditions were selective for the exo mode of cyclization and provided cis-fused bicycles. Additionally, no alkene isomerization was observed for fused and bridged bicycles. Importantly, the alkyl tether requires geminal substitution for efficient cyclization, underscoring the importance of substrate geometry and structural preorganization in cycloalkenylation.

Scheme 20. Allyl-Ni-Catalyzed Oxidative Cycloalkenylation.

^aThe reaction was conducted at 40 °C.

APPLICATIONS IN NATURAL PRODUCT SYNTHESIS

Our group’s dehydrogenation and oxidative C–C and C–X bond-forming reactions have found practical applications in natural product synthesis. These transformations were employed in multistep synthesis campaigns to enable step-efficient conversion of abundant starting materials to complex functionalized products, obviating the need to prepare activated species or isolate potentially unstable intermediates.

For example, the aforementioned allyl-Ni-catalyzed oxidative cycloalkenylation was employed in a strategic two-step sequence to assemble the carbocyclic framework of eudesmane sesquiterpenoid natural products from naturally occurring piperitone (24) (Scheme 21). A Cu-catalyzed conjugate addition installs the pendent olefin in ketone 22j, which undergoes efficient allyl-Ni-catalyzed cycloalkenylation to provide 10-epi-acolamone (23j), a derivative of the natural product 10-epi-juneol (25). Compound 23j was previously prepared from 25 for structure and stereochemistry confirmation of the rare cis-eudesmane natural product class.³⁵ While naturally occurring cis-eudesmanes are less common, epimerization of the cis-decalin to the more thermodynamically stable trans-decalin should be feasible and would provide rapid access to the broader eudesmane family. Previous ground-breaking work by Baran provided a versatile alternative route to the eudesmane natural products.³⁶

Scheme 21.

Cycloalkenylation to Access Eudesmane Sesquiterpenes

Our telescoped ketone dehydrogenation/conjugate addition approach accelerates access to several key β-functionalized ketones (18j–l) previously synthesized en route to biologically active natural products (Scheme 22).³⁷ The existing syntheses of 18j–l required a two-step dehydrogenation process followed by discrete isolation of the enone for the ensuing conjugate addition, resulting in a lengthy and lower-yielding three-step sequence. Our telescoped process demonstrated that these same intermediates can be obtained in a single synthetic operation in higher yield, effectively shortening the syntheses of (−)-hibiscone C,^37a cyclobakuchiol B,^37b and 11-nor-Δ⁸-THC-methyl ester.^37c

Scheme 22.

Telescoped Process Expedites Natural Product Synthesis

The efficient gram-scale synthesis of the pentacyclic triterpenoid natural product justicioside E aglycone (28) was made possible through the use of our oxidative enone functionalization methodology.³⁸ These synthetic studies, in combination with mechanistic and computational investigations, led to structural reassignment of the molecule and provided a better understanding of the skeletal rearrangement responsible for the conversion of oleanane triterpenoids to their related justicane scaffolds (Scheme 23). The synthesis of justicioside E aglycone commenced with enone 20e, which is prepared in two steps from commercial oleanolic acid. Enone 20e was subjected to our oxidative enone β-silylation conditions to provide β-silyl enone 21e in 82% yield on 10 g scale. Silyl enone 21e was converted to the corresponding 1,3-diketone by formal Tamao oxidation, which combined represents a novel two-step approach for accessing 1,3-diketones from readily available enones. This 1,3-diketone was then diastereoselectively reduced in the same pot to afford 1,3-diol 27 using a hydrogen atom transfer (HAT)-mediated approach. A four-step sequence, including a key skeletal rearrangement of the 6,6-ring system of oleananes to the 5,7-ring system of justicanes, provided 28 in eight steps.

Scheme 23.

Synthesis of Justicioside E Aglycone

Our group’s oxidative enone functionalization was also employed to provide efficient synthetic approaches to limonoid natural products. Xylogranatopyridine B (33) is a member of a recently discovered group of limonoid alkaloid natural products characterized by phosphatase inhibitory activity and a pyridine ring embedded within its polycyclic core. A fragment coupling strategy was adopted to construct xylogranatopyridine B, and elaborated stannane 29 was identified as a crucial coupling partner (Scheme 24). Utilizing our allyl-Pd-catalyzed oxidative enone β-stannylation, readily available 6-methylcyclohexenone (20a) was stannylated at an early stage on a large scale to obtain the requisite amounts of β-stannyl enone 21a needed to access stannane 29 over a three-step sequence.⁴ By the use of modified conditions for a Liebeskind pyridine ring synthesis, stannane 29 and oxime benzoate 30 (accessed in four steps from commercial materials) were coupled to form the tetracyclic core of the natural product, which is then oxidized to ketone 31 via a benzylic oxidation. Subjecting ketone 31 to allyl-Pd-catalyzed ketone dehydrogenation efficiently afforded enone 32 on 1 g scale, even in the presence of functionality readily oxidized by many methods (pyridine and furan). Finally, vicinal difunctionalization of the enone over a three-step sequence afforded 33 in a longest linear sequence of 11 steps.

Scheme 24.

Synthesis of Xylogranatopyridine B

After the synthesis of xylogranatopyridine B, our lab undertook the synthesis of the related limonoid alkaloids xylogranatin F (41), xylogranatin G (42), and granatumine A (43), a protein tyrosine phosphatase 1B (PTP1B) inhibitor (Scheme 25).³⁹ Because of the more complex structures of these pentacyclic bislactone limonoid alkaloids, an alternative fragment coupling approach via a Knoevenagel condensation was envisioned. Conversion of commercial 2,6-dimethylcyclo-hexanone (34) to enone 35 using our lab’s allyl-Pd-catalyzed dehydrogenation was the first step to access key fragment 37. Importantly, our ketone dehydrogenation conditions were modified to be more cost-effective, using LDA as base and allyl acetate as the oxidant, with a Pd catalyst loading of only 1 mol %. These modifications broaden the utility of our methodology by providing more scalable access to important synthetic building blocks. Application of known Birch reduction or bromination/elimination protocols to access enone 35 led to variable results and purification challenges on a large scale.⁴⁰ Subsequently, a three-step sequence converts enone 35 to the degraded limonoid (+)-pyroangolensolide, which is further converted to (+)-azedaralide (36) by a site-selective allylic oxidation. Oxidation of 36 affords aldehyde 37 for the convergent fragment coupling. A Knoevenagel condensation between aldehyde 37 and 1,3-diketone 38 (prepared in six steps from commercial α-ionone) followed by spontaneous oxa-6π-electrocyclization and subsequent Luche reduction provided 2H-pyran 39. A pyran to pyridine conversion provided access to the pentacyclic core of the bislactone limonoid alkaloids. Manipulation of the benzylic alcohol in one or two additional steps afforded xylogranatins F (41) and G (42) and granatumine A (43) in a longest linear sequence of 10 steps. The five natural products prepared in this sequence relied on the robustness and scalability of the allyl-Pd-catalyzed dehydrogenation used in the first step to provide the necessary material needed to complete the synthesis.

Scheme 25.

Synthesis of Pyroangolensolide, Azedaralide, Xylogranatins F and G, and Granatumine A

CONCLUSION

Since 2015, our group has developed a new mechanistic approach for Pd-catalyzed dehydrogenation and then extended this strategy to Ni catalysis, providing a more sustainable alternative with opportunities for new reaction development. These methods expanded the scope of one-step carbonyl dehydrogenations to include less acidic carboxylic acid derivatives and were later adapted to achieve the benzylic dehydrogenation of electron-deficient heteroarenes, suggesting the notion that if a substrate can be deprotonated it can be dehydrogenated. Modifications to the base and oxidant as well as the introduction of salt additives and ligands are effective means to tune the reaction efficiency. Novel oxidative C–C and C–X bond-forming reactions were achieved using this reaction manifold, and these methodologies were leveraged for step-economical strategies and syntheses of complex terpenoid and alkaloid natural products. We hope that these transformations will be adopted into a modern organic chemist’s toolbox to expedite the multistep synthesis of complex molecules.

Biographies

David Huang was born in New York City and received his B.A. in Chemistry from Princeton University in 2015, working under the supervision of Prof. Erik Sorensen in the areas of C–H functionalization and base-metal catalysis. He completed his Ph.D. at Yale University with Prof. Timothy Newhouse in 2020, focusing on developing methods using palladium and nickel catalysis with unconventional oxidants. Since 2020 he has been a medicinal chemist at Amgen.

Timothy R. Newhouse was born in New Hampshire and grew up in northern New England. He received his B.A. in Chemistry from Colby College in 2005, where he was mentored by Prof. Dasan M. Thamattoor. He completed his Ph.D. at The Scripps Research Institute with Prof. Phil S. Baran in 2010. At Scripps, he also worked in the laboratories of Prof. Donna G. Blackmond. He then returned to the East Coast for postdoctoral studies with Prof. E. J. Corey at Harvard University. He started at Yale University in the Department of Chemistry in 2013. He was promoted to the rank of Associate Professor in 2018 and is a member of the Interdepartmental Neuroscience Program.