Palladium-Catalyzed Annulations of Strained Cyclic Allenes

Abstract

We report Pd-catalyzed annulations of in situ generated strained cyclic allenes. This methodology employs aryl halides and cyclic allene precursors as the reaction partners in order to generate fused heterocyclic products. The annulation proceeds via the formation of two new bonds and an sp³ center. Moreover, both diastereo- and enantioselective variants of this methodology are validated, with the latter ultimately enabling the rapid enantioselective synthesis of a complex hexacyclic product. Studies leveraging transition metal catalysis to intercept cyclic allenes represent a departure from the more common, historical modes of cyclic allene trapping that rely on nucleophiles or cycloaddition partners. As such, this study is expected to fuel the development of reactions that strategically merge transition metal catalysis and transient strained intermediate chemistry for the synthesis of complex scaffolds.

Since the early 1900s, transient strained cyclic intermediates, exemplified by benzyne (1, Figure 1A), have stimulated interest in the chemical community.¹ Although the proposal of benzyne (1) as a reactive intermediate led to significant controversy in the early 20th century, pioneering experiments by Roberts² and Wittig^3,4 in the 1950s validated its existence, despite the high degree of ring strain. Indeed, this ring strain gives rise to unique and synthetically useful reactivity, which in turn has motivated the development of heterocyclic arynes, such as 2⁵ and 3,⁶ and more highly saturated analogs of benzyne (1), such as cyclohexyne (4).^7,8 These in situ generated intermediates have become increasingly prominent in the modern synthetic toolbox, finding applications in the construction of heterocycles,⁹ ligands for catalysis,¹⁰ natural products,^11–13 agrochemicals,¹⁴ and organic materials.¹⁵

Figure 1.

Selected in situ generated strained cyclic intermediates and overview of strained cyclic allene reactivity.

Whereas strained cyclic alkynes have received significant attention from the synthetic community, reactions of a related class of fleeting intermediates, strained cyclic allenes (e.g., 1,2-cyclohexadiene (5)), are relatively less developed.^16,17 This is notable as cyclic allenes and alkynes were both discovered around the same time and share similar strain-promoted reactivity.¹⁸ Moreover, cyclic allenes are advantageous for the synthesis of sp³-rich and stereochemically complex targets, which are becoming increasingly more valuable.¹⁹ However, to date, only three major intermolecular reaction classes have been reported for these intermediates (Figure 1B), the most frequently studied being nucleophilic trappings (e.g., 7 + KOt-Bu → 6) and cycloadditions (e.g., 7 + 8 → 9).¹⁶ More than 50 studies detailing such reactions have been reported, including more recent studies by Guitián,^20,21 West,^22–25 Okano and Mori,^26,27 Schreiber,²⁸ and our laboratory.^29–32 A complementary, albeit rarely studied, approach utilizes transition metal catalysis to intercept the cyclic allene (e.g., 7 + 10 → 11).^20,33 However, this strategy poses an inherent kinetic challenge due to the requirement for reaction between a catalytically generated intermediate and an in situ generated fleeting cyclic allene, both of which are present in low concentration.

Previous efforts to intercept strained cyclic allenes intermolecularly with transition metal catalysis have been limited, with only two examples known in the literature. The first was a single example wherein an in situ generated cyclic allene was intercepted in a Pd-catalyzed [2 + 2 + 2] reaction to generate a mixture of tetralin products, as demonstrated by Guitián and co-workers in 2009.²⁰ Subsequently, our laboratory developed a Ni-catalyzed annulation of cyclic allenes with benzotriazinones, including an asymmetric variant.³³ These studies demonstrate that the combination of transition metal catalysis and cyclic allenes provides an exciting new avenue in strained intermediate chemistry.

With the aim of developing a modular new method for the metal-catalyzed interception of strained cyclic allenes, we designed the transformation depicted in Scheme 1.³⁴ We envisioned that bifunctional annulation partners 12 containing an aryl iodide and a pendant pronucleophile (XH) would first react with a transition metal catalyst, generating oxidative addition intermediates 15. Concurrently, silyl triflates 13 would be converted to strained cyclic allenes 7 via fluoride-mediated desilylation. Cyclic allenes 7 would then undergo migratory insertion into the Ar–M bond of 15, ultimately giving rise to η³-coordinated complexes 16. Cyclization of the pendant pronucleophile would deliver tricycles 14 with regeneration of the catalyst. Herein, we demonstrate the successful implementation of this approach to annulate in situ generated cyclic allenes using Pd catalysis. An array of fused polycyclic products are obtained in synthetically useful yields by varying the cyclic allene precursor and the aryl halide substrate. Stereoselective examples are also demonstrated. These results highlight the value of merging transition metal catalysis with strained cyclic allene chemistry and should enable future reaction development in this emerging area.

Scheme 1.

Envisioned Reaction Design

To validate our proposed reaction design, we studied the reaction of o-iodoaniline derivative 17 with silyl triflate 18 using fluoride-based conditions (Table 1). The conditions we had identified previously for the aforementioned Ni-catalyzed reaction of benzotriazinones with cyclic allenes³³ were deemed a suitable starting point for reaction discovery. Unfortunately, no desired product was observed with a Ni(cod)₂/dppf catalyst system, even at increased temperature (entries 1 and 2). Given the widespread use of palladium in academic and industrial settings, along with its well-established competency in heteroannulation reactions,³⁴ we turned to Pd catalysis. We were gratified to observe that the use of a Pd catalyst enabled the desired reactivity, leading to tetrahydrocarbazole 19a, albeit only in 8% yield (entry 3). Because unreacted silyl triflate 18 was observed under these conditions, we increased the loading of CsF. This led to a significant improvement in the yield, as shown in entry 4.³⁵ After evaluating other ligands (see Supporting Information (SI) for details), we found that employing the bulky monophosphine DavePhos was most effective,³⁶ generating 19a in 71% yield (entry 5). Further optimization led to the use of DMF as solvent, Bu₄NOTf³⁷ as an additive, and a slight excess of silyl triflate 18. Employing these conditions delivered 19a in nearly quantitative yield after 1 h at 80 °C (entry 6).

Table 1.

Reaction Optimization

Entry	Catalyst/Ligand	Solvent	Equiv. CsF	Temperature	Yieldb
1c	Ni(cod)2/dppf	MeCN	2.5	35 °C	0%
2	Ni(cod)2/dppf	MeCN	2.5	80 °C	0%
3	Pd(OAc)2/dppf	MeCN	2.5	80 °C	8%
4	Pd(OAc)2/dppf	MeCN	10	80 °C	51%
5	Pd(OAc)2/DavePhos	MeCN	10	80 °C	71%
6d	Pd(OAc)2/DavePhos	DMF	10	80 °C	97%

^aReaction conditions: 17 (1.0 equiv), 18 (1.0 equiv), catalyst (10 mol %), ligand (20 mol %), CsF (as shown), Na₂CO₃ (3 equiv), solvent (0.1 M), 3 h.

^bYield determined by ¹H NMR analysis using 1,3,5-trimethoxybenzene as an external standard.

^c24 h.

^d18 (1.5 equiv), Bu₄NOTf (5.0 equiv), 1 h.

To assess the generality of this reaction, we tested a range of o-iodoaniline derivatives 20 and silyl triflates 13 (Figure 2). Although our conditions shown in Table 1, entry 6 were generally useful, some modifications were made to obtain optimal yields based on empirical observations. Three different N-substituents were examined, namely tosyl (Ts), tert-butyloxycarbonyl (Boc), and acetyl (Ac), with Ts providing the highest yield (90% of 19a). Electronic and steric perturbations of the annulation partner were also tolerated, as evidenced by the formation of tetrahydrocarbazoles 22–30. Of note, the annulation exhibits good functional group compatibility, leading to products bearing nitriles or esters (26 and 27), as well as aryl bromides, chlorides, or fluorides (28–30).³⁹ Given the importance of heterocycles in medicinal chemistry, we also assessed the compatibility of the reaction with heterocyclic annulation partners and heterocyclic allenes. Gratifyingly, an iodopyridyl annulation partner underwent the desired reaction to afford product 31 in excellent yield. The corresponding bromopyridine substrate could also be employed to give 31, albeit in slightly diminished yield (78%).⁴⁰ Lastly, we varied silyl triflate 13 to enable further modulation of the product structure. Oxa- and aza-derivatives of 13, readily available through known routes,^30,31 were employed to furnish heterocycles 32 and 33, respectively. This highlights the value of heterocyclic allenes in our methodology for the modular assembly of more complex heterocyclic products. Likewise, substituting a 7-membered cyclic allene precursor in place of a 6-membered cyclic allene precursor gave straightforward access to the 6–5–7 ring system seen in 34.

Figure 2.

Scope of the annulation reaction with iodoaniline derivatives 20. Reactions performed with 20 (1.0 equiv) and 13 (1.5 equiv). ^a Reaction was performed using 2-(N,N-dimethylaminomethyl)-1-diphenylphosphanylferrocene as ligand in MeCN at 60 °C for 3 h. ^b ArX = 2-halo-6-methyl-3-N-tosylaminopyridine. ^c Reaction was performed using 2-(N,N-dimethylaminomethyl)-1-diphenylphosphanylferrocene as ligand with CsF (2.5 equiv) in MeCN at 60 °C for 24 h; performed in the absence of Bu₄NOTf. ^d A silyl tosylate was used as the allene precursor (see ref 38).

We also questioned whether other pronucleophiles, beyond iodoaniline derivatives, could be employed in Pd-catalyzed reactions of strained cyclic allenes (Figure 3). With regard to oxygen nucleophiles, we found that benzyl alcohol 35 and benzoic acid 37 were successful in delivering 36 and 38, respectively, under our standard reaction conditions. Using modified conditions with Xantphos as the ligand, we found that amide 39 was a suitable N-pronucleophile, delivering 40 in 71% yield. Lastly, we tested the viability of a C-based pronucleophile by employing diester 41. Using our standard reaction conditions, we obtained tetrahydrofluorene 42, which bears a newly formed quaternary carbon. These examples underscore that the union of strained cyclic allenes and Pd catalysis provides a rapid entryway to the synthesis of structurally diverse polycyclic products, including 6–6–6 and 6–5–6 ring systems bearing ethers, lactones, amides, or all-carbon frameworks.

Figure 3.

Variation of the pronucleophile to furnish structurally diverse products.

The feasibility of diastereo- and enantioselective variants of the methodology was established as shown in Figure 4. In both cases, heterocyclic substrates were utilized to evaluate the strengths and limitations of the metholodogy, while providing access to polycyclic scaffolds. Regarding the diastereoselective annulation, it should be noted that diastereoselective reactions of strained cyclic allenes have been reported in the context of cycloadditions,¹⁶ but there are no prior examples involving transition metal catalysis. Given our success in employing a benzylic alcohol (i.e., 35, Figure 3), we prepared tertiary benzylic alcohol 43. This substrate, rapidly accessed from a commercially available isatin precursor, was treated with 1,2-cyclohexadiene precursor 18 under standard annulation conditions. Tetracycle 44 was obtained as the major product, with d.r. = 5.6:1, thus providing the first example of substrate-guided stereocontrol in a metal-catalyzed reaction of cyclic allenes.⁴¹

Figure 4.

Stereocontrolled annulations and synthetic elaboration of an annulation product.

To probe the feasibility of an enantioselective variant, we tested the reaction of iodopyridine 45 with silyl triflate 46 (Figure 4). Evaluation of chiral ligands led to the identification of Pd₂(dba)₃/Mandyphos as the optimal catalyst system. Moreover, with CH₂Cl₂ as solvent at decreased temperature (3 °C) (see SI for extended optimization details), tricycle (−)-47 was generated in 90% ee.⁴² As catalytic asymmetric reactions of in situ generated strained cyclic intermediates (i.e., arynes, cyclohexynes, cyclic allenes, etc.) are rare, we hope this result will promote further efforts in this area.

Finally, with unique polyheterocyclic product (−)-47 in hand, we questioned if the styrenyl olefin present in the annulation products could be leveraged as a handle for further elaboration. (−)-47 was subjected to oxacyclic allene precursor 48 in the presence of CsF to furnish (+)-49 via a highly diastereoselective [2 + 2] cycloaddition. The olefin in (+)-49, newly introduced in the cyclic allene [2 + 2] reaction, then underwent diastereoselective epoxidation to furnish (+)-50. This short sequence leverages the asymmetric Pd-catalyzed annulation of an azacyclic allene to ultimately generate an enantioenriched hexacyclic compound (i.e., (+)-50) bearing five different fused heterocycles, a highly substituted cyclobutane, and six contiguous stereocenters, one of which is quaternary.

In summary, we have developed Pd-catalyzed annulations of in situ generated strained cyclic allenes. The methodology employs aryl halides and cyclic allene precursors as the reaction partners, ultimately forming two new bonds and generating an array of fused heterocyclic products. Moreover, through the syntheses of 44 and (−)-47 (see Figure 4), we demonstrate stereoselective variants of this methodology. The catalytic asymmetric annulation and resulting swift access to hexacycle (+)-50 highlight some of the opportunities afforded by the combination of strained intermediates and Pd catalysis. As this strategy represents a departure from the more common, historical modes of cyclic allene trapping using nucleophiles or cycloaddition partners, we hope this study will stimulate the future development of reactions that strategically merge transition metal catalysis and transient strained intermediate chemistry for the synthesis of complex scaffolds.⁴³