Intramolecular, Visible Light-Mediated Aza Paternò-Büchi Reactions of Unactivated Alkenes

Abstract

Azetidines are of particular interest in medicinal chemistry for their favorable properties, including increased resistance to oxidative metabolism and lower lipophilicity. The recent development of [2+2] reactions has significantly expanded the limited repertoire of methods for azetidine synthesis, but access to more complex architectures still requires further development. Herein we report a visible-light enabled intramolecular [2+2] cycloaddition of unactivated alkenes that proved previously unreactive to access tricyclic azetidines with 3D complex structures and high levels of saturation.

Graphical Abstract

Incorporation of highly substituted, three-dimensional scaffolds into target structures is an essential strategy in medicinal chemistry to facilitate the development of future pharmaceuticals.¹ Current discovery efforts are shifting towards the inclusion of more saturated, complex frameworks to increase the likelihood of identifying promising lead structures.¹ Benefits from higher saturation and higher complexity range from increased solubility to greater specificity, and lower toxicity, which are desirable qualities for a drug.² Azetidines are 4-membered nitrogen-containing heterocycles that recently garnered a lot of attention as medicinally desirable scaffolds in the quest for new lead compounds in drug design and development.³ Specifically, azetidines exhibit favorable pharmacokinetic properties and biological activities including decreased lipophilicity, high polarity, and rigidity, in addition to facilitating the rapid build-up of molecular complexity.^{1, 2, 4–9} Implementation of azetidines as alternatives for 5- and 6-membered nitrogen-containing heterocycles in current potential therapeutic candidates is known to improve their metabolic stability by enhancing their overall structural rigidity, resulting in enhanced resistance to oxidative metabolism.^3–6 For example, 5-HT_2C agonists, 1 and 2 identified by Pfizer, differ only in the ring size of the N-heterocycle, yet the azetidine results in a more potent compound with an EC₅₀ of 12 nM compared to 273 nM (Fig. 1A).⁴

Figure 1.

A. Azetidines as important scaffolds for current drug design and development. B. Currently available synthetic strategies to access functionalized azetidines.

However, despite these desirable characteristics, the incorporation of azetidines in current pharmaceutical lead compounds remains limited. Importantly, less than 1% of all nitrogen-heterocycles included in current FDA-approved pharmaceuticals are functionalized azetidines.¹⁰ This divide is due to a dearth in efficient synthetic methods to access these desirable 4-membered ring nitrogen-containing heterocycles.^11–15

Among the synthetic strategies most often applied are nucleophilic substitutions (3),^{13, 14} strain release of azabicyclobutanes (4),^13–15 and the reduction of lactams (5, Fig. 1B).^{13, 14} However, all of these approaches require the synthesis of specific precursors and are prone to limitations in scope. Arguably, the most efficient and direct strategy towards functionalized azetidines relies on [2+2]-cycloaddition reactions between imines and alkenes, also referred to as aza Paternò-Büchi reactions (6).¹⁶

Additional efficient procedures for these aza Paternò-Büchi reactions have recently been developed, relying on visible light-mediated energy transfer (Fig. 2). These methods harness the triplet energy excited states of either alkenes or oximes, generated upon energy transfer from the triplet excited states of suitable photocatalysts.^17–26 Following this approach, intramolecular aza Paternò-Büchi reactions proceed upon accessing the triplet excited states of activated alkenes (7) to undergo the desired [2+2]-cycloaddition with pendant oximes in up to 99% yield to form azetidines (8) (Fig. 2A).¹⁷An intermolecular variant of the aza Paternò-Büchi was subsequently developed that relies on unactivated alkenes 9 and the triplet excited states of cyclic oximes 10 to promote the formation of azetidines 11 in up to 99% yield (Fig. 2B).¹⁹

Figure 2.

A. Intramolecular methodology using excited alkenes. B. Intermolecular methodology using excited oximes and unactivated alkenes. C. Novel intramolecular methodology using excited oximes and unactivated alkenes.

Herein we expand the repertoire of efficient aza Paternò-Büchi reactions to report the development of an intramolecular transformation of unactivated alkenes by relying on the design principle of oxime sensitization to access complex azetidine scaffolds. Specifically, cyclic oximes react with the unactivated alkene moieties in 12 to result in the formation of tricyclic azetidines 13 to access substrates with opposite substitution patterns in comparison to the previous intermolecular approach. Notably, both metal-based and organic photocatalysts are shown to promote the desired transformation in yields up to 84% and 41% respectively. Additionally, N–O bond cleavage of the tricyclic azetidine initially formed upon [2+2]-cycloaddition is shown to proceed readily to result in the formation of the free azetidine products. To identify optimal conditions for intramolecular aza Paternò-Büchi reactions of unactivated alkenes, we evaluated a range of photocatalysts varying in triplet energies upon irradiation of oxime 14 with blue LED lights (Table 1). Specifically, conversion of oxime 14 with catalytic amounts of photocatalysts displaying a high triplet energy, around 60 kcal mol⁻¹, including Ir(dF(Me)ppy)₂(dtbbpy)] and Ir1·PF₆ resulted in the formation of azetidine 15 in 73% and 78% yield, respectively (entries 1–2, Table 1). Under otherwise identical reaction conditions, conversion of oxime 14 with Ir2 gave rise to azetidine 15 with increased yields of 90% (entry 3, Table 1). Using a catalyst with a lower triplet energy of 58.6 kcal mol⁻¹, fac-[Ir(Fppy)₃, resulted in a decreased yield of 61% for 15 (entry 4, Table 1). Catalysts with triplet energies below 57 kcal mol⁻¹ including [Ir(dFppy)₂(dtbbpy)]PF₆, fac-[Ir(ppy)₃], and fac-[Ir(4’-CF₃-ppy)₃] resulted in diminished formation of azetidine 15 in yields of less than 29% (entries 5–6, Table 1). Organic photocatalysts were also considered under similar conditions, varying the triplet energy of the catalyst used. In particular, utilizing thioxanthone with a higher triplet energy than the optimal iridium catalyst, Ir2, no product was observed (entry 8, Table 1).

Table 1.

Evaluation of photocatalysts in intramolecular, visible light-mediated aza Paternò-Büchi reactions.

Conditions: 0.1 mmol 22 and photocatalyst are dissolved in acetonitrile (0.1 M) and the solution was sparged for 10 min. The sample was irradiated with blue LED lamps 16 hours. Percent yield was determined by NMR with mesitylene as internal standard.

^aRan in toluene for increased solubility of some organic catalysts.

Using organic catalysts with lower triplet energies than thioxanthone, 2-MeOTx and 2CzPN, gave rise to azetidine 15 in 41% yield for both catalysts (entries 9–10, Table 1). However, conversion of oxime 14 relying on 4CzIPN exhibiting a triplet energy between 2-MeOTx and 2CzPN failed to provide the desired azetidine 15 (entry 12, Table 1). Suspecting that the solubility of the organic photocatalyst may be influencing the yield, the solvent was switched to toluene. Under these modified reaction conditions, thioxanthone produced trace amounts of azetidine 15 in 6% yield (entry 12, Table 1). 2-MeOTx did not perform as well in toluene as it did in acetonitrile, resulting in a decreased yield of 26%, while 2CzPN performed similarly, producing 15 in 38% yield (entries 13–14, Table 1). 4CzIPN in toluene produced an increased yield of 19% (entry 15, Table 1). The optimal organic catalyst conditions were identified as 2-MeOTx or 2CzPN in acetonitrile, which is the first instance of organic catalysts being used in a visible light-mediated [2+2] cycloaddition between alkenes and oximes to access azetidine products by exciting the oxime to its triplet state.

With an optimal photocatalyst identified, we subsequently focused on further improving the reaction conditions, with a particular emphasis on concentration (Table 2). Specifically, at 0.2 M a conversion of 98% was observed, however the desired product 15 was formed in diminished yields of only 76% (entry 16, Table 1). In comparison, decreasing the concentration to 0.1 M resulted in the formation of azetidine 15 with improved yields of 84% while complete consumption of oxime 14 was observed (entry 17, Table 1). Notably, decreasing the concentration further to 0.05 M or 0.01 M did not result in an improvement in yield (entries 18–19, Table 1). Without photocatalyst, no formation of azetidine 15 was observed when oxime 14 was irradiated with blue LED light (entry 20, Table 1).

Table 2.

Evaluation of the scope of intramolecular, visible light-mediated aza Paternò-Büchi reactions.

Conditions: 0.25 mmol isoxazoline and Ir2 (1 mol %) are dissolved in acetonitrile (0.1 M) and the solution is sparged for 10 min. The reaction is irradiated with blue LED lamps (427 nm) for 16–20 hours.

^aReaction run at 0.01 M on 0.05 mmol scale, yield by HNMR.

^bReaction was run on a 1 mmol scale using standard reaction conditions.

Utilizing the optimized conditions, the substrate scope for intra molecular aza Paternò-Büchi reactions of oximes incorporating unactivated alkenes was explored. The optimal reaction conditions identified proved to be compatible with a variety of alkene substitution patterns, chain lengths, and oxime backbone substituents, providing access to azetidines in up to 84% yield. The trisubstituted alkene substrate resulted in the formation of tricyclic azetidine 15 in 84% yield, while disubstituted alkenes (19, 20, and 23) produced yields up to 69% yield. The optimal reaction conditions were found to tolerate terminal alkenes, which was surprising as the postulated reaction mechanism for this intramolecular aza Paternò-Büchi reaction relied on the formation of a primary radical. Employing substrates incorporating terminal alkenes allows access to azetidines with methylene groups adjacent to the nitrogen, which were previously only accessible using ethylene gas following the former intermolecular reaction protocol. Terminal alkene 16 resulted in a yield of 27%, however this was improved to 80% by decreasing the concentration of the reaction. Substitution adjacent to the ester moiety resulted in a higher yield of 69% for the dimethyl substrate 17 although a decrease in yield to 9% was observed for azetidine 18, bearing a mono-methyl substituent. Conversely, the crotyl ester resulted in a 68% yield of azetidine 19 with a 1.3:1 mixture of diastereomers. Increasing the size of the substituent to a n-propyl group did not improve the diastereoselectivity, however did produce 20 in 61% yield. Importantly, azetidines incorporating 6-membered ring-lactones were similarly readily accessible, albeit in slightly diminished yields compared to their 5-membered ring analogs. The substrate utilizing a terminal alkene resulted in 21 being formed in 6% yield while the trisubstituted alkene resulted in 22 being formed in an increased 41% yield due to the increased substitution. By decreasing the reaction concentration for 21 and 22, the yields increased to 25% and 71%, respectively. An acrylate was also tolerated under these optimal reaction conditions, resulting in 27% yield of 23. Different substitution patterns on the oxime backbone were amenable, including the unsubstituted oxime 26 resulting in 80% yield, which is comparable to the 87% yield of 15 obtained from the corresponding dimethyl substituted oxime. Monosubstituted oximes were tolerated, however yields varied based on the substituent. Specifically, the n-butyl substituent azetidine 24 was produced in 72% yield, the highest yield for monosubstituted substrates while phenyl (27) and ethyl ether (28) resulted in 58% and 28% yields, respectively. Bulkier disubstituted oxime substituents, specifically diphenyl 25 and piperidine 29, resulted in 22% and 57% yield respectively. Dilution of the reaction for 25 to 0.01 M resulted in an improved yield of 61%. Diluting the reaction for select low yielding substrates resulted in the yields improving. Importantly, the increasing of the yield for terminal alkene 17 demonstrates that terminal alkenes can work just as well as trisubstituted alkenes under more dilute conditions.

We subsequently evaluated the robustness²⁷ of the intramolecular aza Paternò-Büchi reaction by subjecting oxime 14 to the optimized reaction conditions in the presence of one equivalent of various additives (Figure 3A). The addition of equimolar amounts of benzoic acid, chloroalkanes, methanol, and sodium carbonate resulted in azetidine 15 with less than 12% decrease in overall yield, suggesting a high level of functional group tolerance. Notably, the free azetidines are accessible upon N-O bond cleavage. Specifically, under hydrogenolysis relying on catalytic amounts of Pd(OH)₂, spirocyclic azetidine 30 was obtained in 37% yield (Fig. 3B).

Figure 3:

A. Robustness evaluation. B. N–O bond cleavage conditions to access spirocyclic, free NH azetidines.

The previously proposed mechanism for the intermolecular aza Paternò-Büchi reaction utilizing unactivated alkenes and cyclic oximes is suspected to be operative for the intramolecular [2+2] reaction.¹⁹ The triplet-state photoexcited iridium catalyst sensitizes isooxazoline 12 through energy transfer to its triplet state I. The triplet-state isooxazoline I can then undergo C–C bond formation with the proximal alkene carbon to form 1,4-biradical II, which can then undergo intersystem crossing to intermediate III and form azetidine 13 through C–N bond formation.

To conclude, we have reported the development of an intramolecular visible light-enabled [2+2]-cycloaddition reaction between unactivated alkenes and cyclic oximes to access tricyclic azetidines in yields up to 84%. This transformation complements existing reports for aza Paternò-Büchi reactions to expand the repertoire of currently available efficient procedure for the formation of complex azetidine products. Furthermore, this transformation relies on an iridium photocatalyst but can similarly be promoted by an organic photocatalyst to achieve yields of the desired product of up to 41%, resulting in complex, highly saturated scaffolds. Notably, this method gives rise to a reverse substitution pattern in comparison to previous methods developed,^17,19 allowing access to molecules that occupy distinct chemical space. We anticipate this new strategy to be used for the rapid build-up of molecular complexity when incorporating functionalized azetidines in current drug design and development.

Figure 4:

Mechanistic Hypothesis for the intramolecular aza Paternò-Büchi reaction.