Diastereodivergent Access to cis,cis- and cis,trans-Angular Triquinanes: Total Synthesis of (±)-Arnicenone Chia-Hao Liu, Hung-Tse Huang, Han-Jung Li, Shubham A. Dighe, Deng-Lian Hou, Li-Yun Wu, Yu-Wen Huang Abstract Angular triquinanes constitute an important class of polycyclic natural products characterized by compact, angularly fused ring systems and multiple contiguous stereocenters. Herein, we report a diastereodivergent strategy for stereocontrolled access to both cis,trans- and cis,cis-angular triquinanes from common allene precursors. The cis,trans-fused triquinanes were constructed through a Lewis acid-promoted intramolecular [3+2] Danheiser annulation, whereas the complementary cis,cis frameworks were accessed via photochemical [2+2] cycloaddition followed by oxidative ring expansion. Notably, substitution at the R1 position of the allene precursor reversed the diastereoselectivity of the [3+2] annulation, providing direct access to a cis,cis-angular triquinane. DFT calculations indicate that the initial spirocyclization governs both the reactivity and stereochemical outcome of the [3+2] pathway, with substrate-dependent steric interactions accounting for the observed diastereodivergence. The synthetic utility of this strategy was demonstrated through the total synthesis of (±)-arnicenone via the [2+2] cycloaddition/oxidative ring-expansion pathway, while an alternative synthesis from the cis,cis-[3+2] Danheiser annulation product further highlights the complementary nature of the two approaches. Together, these results establish a unified platform for stereodivergent access to angular triquinanes and their application to complex natural product synthesis. [ja6c09019_0014.webp] Angular triquinanes − (1 and 2) constitute a prominent structural motif in natural products, particularly within terpenoid and sesquiterpenoid families (). Their compact, angularly fused tricyclic frameworkcomprising three contiguous five-membered ringsfeatures multiple stereogenic centers and significant conformational constraint. This dense stereochemical architecture renders angular triquinanes both synthetically challenging and biologically intriguing. Among these, arnicenone − (3) exemplifies the structural complexity associated with this class of molecules. Other representative members include isocomene , (4), pentalenene , (5), retigeranic acid , (6), and waihoensene , (7), which collectively illustrate the structural diversity of angular triquinane natural products. These molecules differ in their oxidation states, substitution patterns, and ring fusion architectures, yet all possess highly congested polycyclic frameworks featuring multiple contiguous stereogenic centers. Their remarkable structural complexity has inspired the development of numerous synthetic strategies and continues to serve as an important driving force in angular triquinane synthesis. Considerable efforts have been devoted to the construction of cis,cis-triquinane skeletons (1). Representative strategies include assembly of the tricyclic core from preformed diquinane intermediates, the Pauson–Khand reaction (a), − ring expansion of cyclobutane derivatives (10) (b), and various pericyclic transformations such as Nazarov reaction (c). , More broadly, the construction of highly congested and stereochemically constrained molecular architectures remains a longstanding challenge in synthetic chemistry, as exemplified by trans-fused 5,5 ring systems, strained bridged polycycles, and atropisomeric systems. − Within this context, angular triquinanes present a distinct synthetic challenge because their compact polycyclic frameworks require precise control over multiple stereogenic centers and ring-junction geometries. While naturally occurring angular triquinanes predominantly exhibit cis,cis ring-junction stereochemistry, the corresponding cis,trans frameworks remain largely unexplored. Despite recent advances in the construction of cis,cis-angular triquinane skeletons, methods enabling diastereodivergent − access to both stereochemical manifolds from closely related precursors remain limited. Such strategies are highly desirable because they would provide access not only to naturally occurring frameworks but also to complementary stereochemical space beyond that represented in nature. In 1981, Danheiser and co-workers reported the first intermolecular [3+2] annulation of allenylsilanes with alkenes, which has since emerged as a useful strategy for the construction of cis-fused 5/5 ring systems. − More than four decades later, Li and co-workers developed the first intramolecular variant of this transformation, providing an efficient approach to trans-fused 5/5 ring systems and demonstrating its synthetic utility in the synthesis of β-funebrene (d). Li and co-workers subsequently extended this strategy to a Type II intramolecular [3+2] annulation, further broadening the synthetic scope of this transformation and enabling its application to the synthesis of strepsesquitriol (e). The synthetic utility of the Danheiser annulation was further demonstrated by Zhai and co-workers in their total synthesis of (−)-conidiogenone, in which the annulation was employed to construct a 5,5,5-cis,cis-angular tricyclic framework. Inspired by these advances, we envisioned that the intramolecular [3+2] Danheiser annulation could provide stereocontrolled access to angular triquinanes with distinct ring-junction stereochemistries. Specifically, cis,trans-angular triquinanes (19) were targeted via a [3+2] Danheiser annulation strategy from allenic precursor (18), leveraging its efficiency in constructing five-membered carbocycles with well-defined stereochemical outcomes (f). In contrast, cis,cis-angular triquinanes (21) were designed to arise from a [2+2] cycloaddition followed by oxidative ring expansion, − offering a mechanistically orthogonal pathway to the same carbon framework. Herein, we report a diastereodivergent strategy for the synthesis of angular triquinanes that enables stereocontrolled access to both cis,trans- and cis,cis-fused systems. The synthetic utility of this approach is further demonstrated through its application to the total synthesis of (±)-arnicenone (3). A series of different allenes were prepared as outlined in . The diketone intermediates 24(a-d) were obtained efficiently via Stork-Danheiser transposition with Grignard (23) and sequential acidic workup in 71–82% yield. These diketones 24(a-d) then underwent regioselective acetylide addition and acylation to deliver 25(a-k) in 29–79% yield. The stage was set for subsequent SN2′ allenation proceeding cleanly to afford the desired allenes 26(a-n) in 57–91% yield, establishing unified intermediates poised for divergent cyclization. [3+2] Danheiser Annulation With the allenes in hand, we first examined the intramolecular [3+2] Danheiser annulation (). Optimization of the Lewis acid, solvent, and reaction temperature identified conditions that efficiently furnished the angular triquinane products with generally high diastereoselectivity. Substrates bearing R1 = H readily underwent the annulation to afford the corresponding angular triquinanes 28(a–l), generally favoring the cis,trans-diastereomer. The cis,cis/cis,trans ratios ranged from 1:2 to exclusive formation of the cis,trans-isomer. Notably, the cyclohexenone-derived substrates 28(j–l) furnished the cis,trans-isomer exclusively. Variation at R2 was well tolerated, with substrates bearing R2 = H (28(g–i)) providing the desired products in moderate to good yields (45–82%). In contrast, increasing the steric demand at R3 resulted in diminished reactivity. Substrates bearing R3 = Et (28f and 28l) preferentially afforded the cis,trans-isomer, albeit in lower yield (33% for 28l), whereas further increasing the size of R3 to n-Bu or i-Pr completely suppressed the desired [3+2] annulation under the optimized conditions. The silyl substituent was also varied to examine its influence on the annulation, with TMS, TES, TBS, TIPS, and DMPS groups being evaluated. In addition, an H,H-substituted allene lacking a silyl group and an H,DMPS-substituted allene were examined, but neither underwent the desired [3+2] annulation under the optimized conditions. The lack of reactivity of the nonsilylated allene is consistent with the proposed Sakurai-type cyclization and the stabilizing β-silyl effect, whereas the failure of the H,DMPS-substituted substrate suggests that an appropriate allene substitution pattern is also required for efficient cyclization. The practicality of the [3+2] Danheiser annulation was further demonstrated by a gram-scale reaction, which furnished 28a in 58% yield (see Supporting Information for experimental details). Interestingly, introduction of a methyl substituent at R1 markedly altered both the reactivity and stereochemical outcome of the annulation. Substrates bearing R1 = Me exhibited substantially reduced reactivity. Nevertheless, the cyclopentenone-derived substrate underwent cyclization to afford 28m in 10% yield with reversed diastereoselectivity, favoring the cis,cis-isomer (cis,cis/cis,trans = 4:1). Desilylation of the allene precursor to give 29 was observed as the predominant competing pathway for these less reactive substrates. The substrate-dependent reversal of diastereoselectivity observed for 28m prompted further investigation of the stereochemical origin of the [3+2] annulation by DFT calculations (vide infra). Notably, despite its modest yield, cis,cis-28m could be further elaborated to (±)-arnicenone (3), demonstrating the synthetic utility of the [3+2] annulation product (vide infra). [2+2] Cycloaddition and Ring Expansion To access the complementary cis,cis-fused framework, we next evaluated a [2+2] cycloaddition strategy (). Cycloaddition of the corresponding allene precursors (27) afforded cyclobutane intermediates 30, which likely exploit the inherent rigidity of the four-membered ring to control the stereochemical outcome, delivering cis,cis-5,4,5- or cis,cis-6,4,5-tricyclic skeletons. These intermediates (30) subsequently underwent oxidative ring expansion to generate the cis,cis-angular triquinane core and the corresponding cis,cis-6,5,5-tricyclic frameworks, respectively. Mechanistically, the distinct stereochemical outcome arises from an orthogonal reaction manifold: the [2+2] pathway establishes an all-cis relationship during cyclobutane formation and preserves it through stereoselective oxidative expansion. Thus, selective choice of annulation mode enables predictable control over ring-junction geometry. After optimization, the sequence proceeded in good overall yield with exclusive diastereoselectivity. In general, the [2+2] cycloaddition exhibited broad substrate scope. In contrast to the [3+2] Danheiser annulation (), which favored substrates bearing R1 = H, the [2+2] cycloaddition pathway tolerated both R1 = H and Me. Substituents at R2 were also well accommodated (H or Me), while variation at R3 (Me, Et, and iPr) was likewise compatible. TMS, TES, and TIPS substituents were all effective under the cycloaddition conditions. Notably, allenes bearing a terminal geminal diproton (30l, 30m, and 30n) also participated efficiently, affording the desired products in 62–85% yield. The practicality of the [2+2] cycloaddition was further demonstrated by a gram-scale reaction, which furnished 30a in 55% yield (see Supporting Information for experimental details). To further examine the effect of linker size on the two complementary cyclization pathways, we also investigated an allene precursor designed to generate a six-membered bottom ring. Under the optimized [2+2] cycloaddition conditions, however, the corresponding substrate failed to afford the desired cycloadduct. In contrast, the same substrate underwent the intramolecular [3+2] Danheiser annulation to furnish the cis,trans-5,5,6-tricyclic product in 30% yield (see Supporting Information for substrate structure and experimental details). These results demonstrate that extension of the linker to generate a six-membered bottom ring is tolerated in the [3+2] pathway, albeit with reduced efficiency, whereas the corresponding [2+2] cycloaddition is unsuccessful under the optimized conditions. Total Synthesis of (±)-Arnicenone With the [2+2] cycloaddition-derived cis,cis-5,4,5-tricyclic intermediate 30n in hand, we next investigated its elaboration toward (±)-arnicenone (3) (). To this end, 30n was subjected to epoxidation with mCPBA to afford epoxide 31 in 87% yield with an α/β ratio of 1:4. The α-isomer was unambiguously confirmed by X-ray diffraction (XRD) analysis. Both diastereomers were then subjected to LiBr-promoted ring expansion in the presence of HMPA at elevated temperature (130 °C), furnishing the 1,4-diketone (33). However, further attempts to convert 33 into the corresponding α-methylated product 34 were unsuccessful. Given the unsuccessful α-methylation of 33, we turned to an alternative strategy involving reduction of ketone 30n, followed by hydroxyl-directed epoxidation of the exo-olefin with mCPBA, which proceeded smoothly to afford epoxide 38 with exclusive β-selectivity (). Treatment of this intermediate with LiBr in the presence of HMPA in toluene at 130 °C promoted rearrangement to furnish the cis,cis-angular triquinane core in 99% yield, with a diastereomeric ratio of 1:5 favoring the desired 3-hydroxyketone regioisomer (40) over the 4-hydroxyketone (39). Oxidation of 40 with IBX then delivered the corresponding 1,3-diketone (41) in 98% yield. Regioselective conversion of the carbonyl group into the dithiane using 1,2-ethanedithiol and BF3·Et2O proceeded efficiently to afford the protected intermediate (42) in 86% yield. The identity of 42 was unambiguously confirmed by X-ray diffraction (XRD) analysis. Wittig olefination employing CH3PPh3Br and tBuOK installed the exocyclic alkene (43), and subsequent TsOH-mediated olefin isomerization furnished the product (44) in 65% yield over two steps. Removal of the thioacetal with MeI in the presence of CaCO3 provided ketone (35) in 80% yield. Formation of the silyl enol ether followed by Pd­(OAc)2-mediated oxidation (Saegusa–Ito reaction) delivered the enone intermediate (36) in 76% yield. Finally, conjugate methyl addition using MeLi/CuI furnished the target molecule (±)-arnicenone (3) in 78% yield (16 steps, 9.6% overall yield). To further demonstrate the synthetic utility of the [3+2] Danheiser annulation, we also established an alternative route to (±)-arnicenone from cis,cis-28m (). Despite its modest yield (10%) in the annulation, 28m possesses the requisite cis,cis-angular triquinane framework for direct elaboration toward the natural product. Desilylation of 28m followed by Saegusa–Ito reaction furnished enone 36, which underwent conjugate methylation to afford (±)-arnicenone (3) (7 steps, 2.1% overall yield). Thus, the cis,cis-angular triquinane generated directly through the [3+2] annulation provides an alternative entry to the natural product. In parallel, the synthetic utility of the cis,trans-[3+2] annulation product was examined through its elaboration toward the corresponding epi-arnicenone analogue. The sequence was successfully advanced to the late-stage epi-arnicenone precursor enone; however, the final conjugate methylation could not be achieved under the conditions examined (see Supporting Information for details). DFT Calculations of [3+2] Danheiser Annulation The identity of R1 (Me vs H) proved crucial in controlling both the reactivity and stereochemical outcome of the [3+2] annulation. Because the present study was directed toward angular triquinane frameworks relevant to natural product synthesis, the investigation of substitution at R1 was focused on R1 = H and R1 = Me; larger substituents at this position were not examined. Substrates bearing R1 = H generally favored formation of the cis,trans-isomer, whereas introduction of R1 = Me resulted in diminished reactivity and reversed the diastereoselectivity, with 28m favoring the cis,cis-isomer (cis,cis/cis,trans = 4:1). Notably, preferential formation of a cis,cis-angular tricyclic framework was also observed by Zhai and co-workers in their Danheiser annulation en route to (−)-conidiogenone. To gain insight into the origin of the stereochemical reversal observed in our system, density functional theory (DFT) calculations were performed. Mechanistically, activation of allene 45 by the Lewis acid TiCl4 initiates a Hosomi–Sakurai-type spirocyclization, generating vinyl cation 46 (). This intermediate subsequently undergoes a 1,2-silyl shift to furnish cation 47. Finally, intramolecular enolate cyclization of 47 delivers the angular triquinane product 48. DFT calculations support the proposal that the initial cyclization of the Lewis acid-bound starting material 45-TiCl 4 to 46 is both the rate-determining and stereochemistry-determining step of the [3+2] annulation. As shown in , for the R1 = H substrate, the trans-forming pathway proceeds through H_TS_ trans _45–46 (20.4 kcal/mol), which is significantly lower in energy than the cis-forming pathway through H_TS_ cis _45–46 (23.9 kcal/mol), corresponding to ΔΔG‡ = 3.5 kcal/mol in favor of the trans pathway. This is consistent with the preferential formation of the cis,trans product. The resulting intermediates, H_ trans _46 (18.6 kcal/mol) and H_ cis _46 (18.0 kcal/mol), are close in energy, indicating that the selectivity is established primarily at the transition-state level. The subsequent 1,2-silyl shift proceeds through H_TS_ trans _46–47 (18.6 kcal/mol) and H_TS_ cis _46–47 (19.0 kcal/mol) to furnish H_ trans _47 (7.6 kcal/mol) and H_ cis _47 (4.3 kcal/mol), respectively; the small energy difference between H_TS_ trans _46–47 and H_TS_ cis _46–47 indicates that this later step does not control the product ratio. Notably, the final product H_48 (cis,cis)­(−27.6 kcal/mol) is substantially lower in energy than H_48 (cis,trans)­(−12.9 kcal/mol), showing that the cis,cis isomer is thermodynamically more stable even though the cis,trans isomer is formed preferentially under kinetic control. In contrast, for the R1 = Me substrate, the relative ordering of the first cyclization transition states is reversed: formation of Me_ cis _46 proceeds through Me_TS_ cis _45–46 with ΔG‡ = 23.3 kcal/mol, whereas formation of Me_ trans _46 occurs through the slightly higher-energy Me_TS_ trans _45–46 with ΔG‡ = 24.4 kcal/mol, giving ΔΔG‡ = 1.1 kcal/mol in favor of the cis-forming pathway. Moreover, the lowest cyclization barrier for R1 = Me remains 2.9 kcal/mol higher than that for R1 = H, accounting for the diminished reactivity of the Me-substituted substrates. Together, these results show that changing R1 from H to Me both raises the barrier for cyclization of 45–TiCl4 and reverses the relative energies of TS_ cis _45–46 and TS_ trans _45–46, thereby accounting for the observed differences in both reactivity and diastereoselectivity. Analysis of the computed energies provides a rationale for the observed stereochemical outcomes (). In the TS_ cis _45–46 transition state, steric repulsion arises between the γ-CH2 group and R2, whereas in the TS_ trans _45–46 transition state, repulsion is primarily between R1 and R2. When R1 = Me, the Me···Me repulsion in the trans arrangement (24.4 kcal/mol) is greater than the γ-CH2···Me (R2) interaction in the cis arrangement (23.3 kcal/mol), thereby favoring formation of the cis,cis isomer. In contrast, when R1 = H, the R1···R2 repulsion (H···Me, 20.4 kcal/mol) is significantly lower than the γ-CH2···Me interaction (23.9 kcal/mol), resulting in a preference for the cis,trans isomer. Thus, the relative steric interactions in the competing transition states account for the observed diastereodivergence, with the cis,trans isomer being strongly favored when R1 = H. Conclusion In conclusion, we have developed a diastereodivergent strategy for the stereocontrolled synthesis of angular triquinanes through complementary [3+2] Danheiser annulation and [2+2] cycloaddition pathways from common allene precursors. The [3+2] annulation generally provides access to cis,trans-angular triquinanes, whereas the [2+2] cycloaddition followed by oxidative ring expansion furnishes the complementary cis,cis-fused tricyclic architectures. DFT calculations revealed that the initial spirocyclization governs both the reactivity and stereochemical outcome of the [3+2] annulation and rationalized the R1-dependent reversal of diastereoselectivity. The practicality of both cyclization pathways was further demonstrated on gram scale. Finally, the synthetic utility of this strategy was demonstrated through the total synthesis of (±)-arnicenone via the [2+2] cycloaddition/oxidative ring-expansion pathway, while an alternative synthesis from the cis,cis-[3+2] Danheiser annulation product 28m further highlighted the complementary nature of the two approaches. Collectively, these results establish a versatile platform for stereodivergent access to angular triquinanes and their application to complex natural product synthesis.