Site‐Selective Carbonylation of Azetidines via Copper‐Catalyzed Difluorocarbene Insertion

Abstract

γ‐Lactams are privileged five‐membered pharmaco‐phores in numerous bioactive compounds, but access to these motifs typically relies on cycloaddition/substitution chemistry involving activated substrates or CO carbonylations under harsh conditions. Here, we report a new route to functionalized γ‐lactams through formal carbonylation of azetidines under nonprecious metal catalysis. The method leverages a copper‐stabilized difluorocarbene to promote site‐selective insertion followed by in situ hydrolysis to unmask the lactam group. In contrast to most difluorocarbene reactions that cause ring cleavage of saturated heterocycles in the presence of heat, the present system operates at a low temperature and retains the integrity of the cyclic structure. Synthesis of various drug‐like lactams and a therapeutic agent for diabetes highlights utility.

We disclose a site‐selective Cu‐catalyzed carbonylation of azetidines promoted by difluorocarbene insertion at low temperature, enabling access to a variety of functionalized γ‐lactams and a therapeutic agent for diabetes (40 examples). This skeletal expansion strategy provides a distinct disconnection approach to synthesize medicinally relevant five‐membered heterocycles.

The [1,2]‐Stevens‐type rearrangement of ylides^{[ 1 ]} is a classical transformation used to construct functionalized heteroatom‐containing scaffolds in organic chemistry. Analogous processes that make use of substituted carbenes during the course of ylide formation offer a convenient way to modify the skeleton of cyclic molecules,^{[ 2} , ³ , ⁴ , ⁵ , ^{6 ]} enabling formal homologation of saturated rings to afford ring‐enlarged heterocycles^{[ 7} , ⁸ , ⁹ , ¹⁰ , ¹¹ , ^{12 ]} or rearranged acyclic entities of increased complexity.^{[ 13} , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ^{21 ]} Contrary to other carbene variants, difluorocarbenes^{[ 22} , ²³ , ^{24 ]} are rarely explored in the context of Stevens‐type rearrangements. Song and co‐workers^{[ 25 ]} recently disclosed a catalyst‐free protocol involving difluorocarbene‐induced rearrangement of tertiary amines carrying α‐electron‐deficient groups to deliver linear formamides at 90 °C under basic conditions (Scheme 1). However, facile ring rupture typically occurs when cyclic amines are employed as substrates due to the presence of Lewis basic species in solution that trigger nucleophilic ring‐opening,^{[ 26} , ^{27 ]} making any attempts at carrying out skeletal ring expansion a challenge.

Scheme 1.

Harnessing difluorocarbene as a carbonyl surrogate for skeletal enlargement of azetidines to γ‐Lactams.

Based on Zhang's seminal study on electrophilic copper‐ligated difluorocarbenes^{[ 28 ]} and our recent discovery that such species are capable of inserting into epoxides to form fluorinated oxetanes without ring deconstruction,^{[ 29 ]} we wondered if an appropriate copper catalyst may be identified to generate an electrophilic difluorocarbenoid that regioselectively reacts with the Lewis basic nitrogen of a cyclic amine (such as an azetidine) through a single‐electron mechanism. The resulting intermediate I could undergo cyclization and elimination of copper to afford the ring‐expanded intermediate II, which is susceptible to hydrolysis^{[ 30 ]} with water to furnish a γ‐lactam. γ‐Lactams are widely found in many bioactive natural products and pharmaceuticals,^{[ 31} , ^{32 ]} but their syntheses mostly rely on cycloadditions,^{[ 33 ]} cyclizations^{[ 34} , ³⁵ , ³⁶ , ^{37 ]} or C─H amidation.^{[ 38} , ^{39 ]}

This catalytic regime not only provides a new route to γ‐lactams [see Supporting Information (SI) Section VII for a survey of existing γ‐lactam synthetic methods] by preserving the cyclic framework of the heterocycle but also offers a practical approach to harness difluorocarbenes as carbonyl surrogates,^{[ 23} , ²⁶ , ²⁷ , ⁴⁰ , ^{41 ]} potentially replacing existing carbonylation systems that rely on poisonous carbon monoxide at high temperatures.^{[ 42 ]} However, competitive nucleophilic attack by trace Lewis basic species must be suppressed in order to inhibit the deconstruction of I into undesired acyclic fragments.^{[ 26} , ^{27 ]} Herein, we describe the successful development of a nonprecious copper‐catalyzed manifold that achieves formal carbonylation of azetidines via site‐selective difluorocarbene insertion at a low temperature.

Optimization studies were conducted using 2‐phenyl azetidine 1a as the model substrate (Table 1). After extensive evaluation of various reaction parameters,^{[ 43 ]} we found that the site‐selective carbonylation of 1a was achieved in the presence of 10 mol% of ligand‐free CuI using commercially available diethyl (bromodifluoromethyl)phosphonate 2 as the difluorocarbene precursor, KF as base and DME as solvent at 0 °C, furnishing γ‐lactam 3a as the sole regioisomer in 76% GC (75% isolated) yield within a day (entry 1). The CO unit was preferentially inserted into the C─N bond at the more substituted carbon,^{[ 29 ]} and small amounts of ring‐opening impurities with no trace of fluorinated products were detected. These conditions differ from our previous study on catalytic difluorocarbene insertions that demanded elevated temperatures to proceed.^{[ 29 ]} Switching CuI to other copper(I) complexes or lowering the catalyst loading to 5 mol % led to slightly diminished yields (entries 2 and 3). Unsurprisingly, excluding the copper catalyst or switching CuI with other transition metal complexes had a detrimental effect on reaction efficiency (entries 4 and 5), underscoring the importance of copper in facilitating the catalytic process. Intriguingly, performing the reaction at −5 °C or room temperature gave lower yields of 3a without compromising regioselectivity (entries 6 and 7). Varying the reaction time or loading of 2 also did not improve results (entries 8 and 9). Replacing 2 with (bromodifluoromethyl)trimethylsilane as an alternative difluorocarbene source negatively impacted carbonylation efficiency (entry 10). Likewise, 3a was generated in appreciably lower yields when the solvent or base was individually changed (entries 11 and 12).

Table 1.

Reaction optimization.

Entry	Deviation from standard conditions	Yield (%) a)
1	None	76 (75) b)
2	CuCI, CuBr or Cu(MeCN)4BF4 instead of Cul	63–64
3	Cul (5 mol%)	64
4	Without Cu catalyst	11
5	FeBr2, Col2, Nil2, or Pd(PPh3)2Cl2 instead of Cul	2–19
6	−5 °C	70
7	RT	45
8	12 h	67
9	2 (1.5 equiv.)	47
10	TMSCF2Br instead of 2	43
11	THF as solvent	19
12	CsF as base	49

^a) Yields were determined by GC using n‐tridecane as internal standard.

^b) Isolated yield.

^c) Ratio of the regioisomeric ring expansion products that arise from net CO insertion into the C─N bonds of 1a as determined by GC and GCMS analysis.

Next, we evaluated the generality of the established reaction conditions by surveying a variety of functionalized azetidines bearing diverse N‐substituents (Table 2). 2‐Aryl azetidines are reported to be potent pharmacophores^{[ 31} , ^{32 ]} and could be readily accessed through facile simple nucleophilic substitution of 1,3‐dichlorides^{[ 44 ]} with primary aliphatic amines derived from abundant feedstocks, advanced drug intermediates and complex bioactive compounds.^{[ 45 ]} Substrates bearing N‐alkyl substituents with functional groups such as aromatic halides (3g, 3h), alkyl halide (3m), heterocycles (3i, 3j, 3l), alkene (3k) as well as base‐ and acid‐sensitive groups such as silyl ether (3n) and acetal (3o) were all tolerated in our copper‐catalyzed skeletal expansion system, furnishing products 3a−3o in up to 97% yield and complete site selectivity across the board. To further test the functional group compatibility of the method, we subjected complex drug‐like azetidine substrates derived from naturally occurring amines, drug precursors and pharmaceuticals to the standard conditions for carbonylation and successfully secured γ‐lactams 3p−3u in 50%–87% yield (Table 2, inset). These examples highlight the excellent chemoselectivity of the transformation as a mild carbonylation strategy for late‐stage functionalization applications within the context of densely functionalized and stereochemically sophisticated azetidines. However, azetidines containing N‐aryl substituents and N‐electron withdrawing substituents (such as benzoyl group) were ineffective substrates.^{[ 46 ]}

Table 2.

Scope of N‐substituents in copper‐catalyzed carbonylation of azetidines. ^a)

^a) As in Table 1 (entry 1), using 2 (0.1 mmol). Yields denote isolated yields.

Structural modifications on the azetidine core carrying different C‐substituents were also investigated (Table 3). In general, ring expansion proceeded smoothly in the presence of an activating C(sp ²)‐ or C(sp)‐hybridized substituent on the 2‐position, delivering γ‐lactams 5a−5r in 45%–99% yield. These included substrates containing electronically and sterically varied (hetero)arenes with electron‐donating or electron‐deficient groups on the ortho‐, meta‐ or para‐ sites (5a−5l). The transformation of 2,3‐disubstituted azetidine was diastereoselective, delivering 5m in 45% yield as a single trans diastereomer. This result is consistent with our previous work involving ring expansions of oxetanes,^{[ 29 ]} which probably occur via six‐membered copper metallacycle intermediates which minimize steric repulsions (see SI Section II). Remarkably, the presence of fully substituted carbon centers on the 2‐position did not interfere with the catalytic process, delivering the desired products 5n−5p bearing sterically congested quaternary centers. Besides aryl groups, substrates with synthetically useful handles such as an alkyne (5q) or alkene (5r) also underwent efficient carbonylation to deliver the expected products. However, less‐activated azetidines that lack 2‐substituents or those that contain 2‐alkyl groups failed to undergo ring expansion (poor conversions).

Table 3.

Scope of C‐substituents in copper‐catalyzed carbonylation of azetidines. ^a)

^a) As in Table 1 (entry 1), using 2 (0.1 mmol).

^b) Reaction conducted at room temperature. Yields denote isolated yields.

N.R., no reaction.

To showcase practicality, we performed the copper‐catalyzed reaction of 1a and 2 on 1.0 mmol scale at higher reaction concentration and successfully isolated 3a in 73% yield with minimal loss of efficiency (Scheme 2). Application to the concise synthesis of a therapeutic agent for diabetes 8 ^{[ 47 ]} was explored. Straightforward formation of the requisite azetidine 7 from commercially available alcohol 6 set the stage for the site‐selective generation of 8 in 67% yield via copper‐catalyzed carbonylation.

Scheme 2.

Synthetic applications.

Mechanistic studies were conducted to obtain insights into the copper‐catalyzed transformation (Scheme 3). Specifically, we speculated that the CO group inserted originated from difluorocarbene insertion followed by defluorinative hydrolysis^{[ 26} , ^{27 ]} of the resulting CF₂ motif by water upon reaction workup. To this end, we repeated the standard catalytic reaction of 1a and 2 and subsequently added 2 equiv. of exogenous ¹⁸O‐labeled water to trigger hydrolysis. Expectedly, ¹⁸O‐labeled 3a was almost exclusively detected by GC‐MS analysis.

Scheme 3.

Mechanistic studies.

We also carried out the reaction in the presence of excess radical scavenger 2,2,6,6‐tetramethyl‐1‐piperidinyloxy (TEMPO), which was found to inhibit the ring expansion. When an enantioenriched azetidine (S)‐1a was treated with the standard copper‐catalyzed conditions, significant stereochemical erosion was observed and the resulting γ‐lactam 3a was obtained as a racemic mixture. These results suggest that transient radical intermediates, which likely arise from homolytic C─N bond rupture, might be formed in solution. The generation of radical species may explain why an activating substituent on the 2‐position of the azetidine substrate is necessary for promoting regioselective C─N bond cleavage at the more substituted carbon to give the more stabilized radical.^{[ 29 ]}

Based on the aforementioned results and previous findings,^{[ 29 ]} we reasoned that a copper‐ligated difluorocarbene III is generated in situ from CuI and 2, which associates with the azetidine and undergoes site‐selective single‐electron C─N bond cleavage and C─N bond formation with the difluorocarbene to afford I. Intramolecular recombination of the alkyl radical with the copper center leads to a six‐membered metallacycle IV, which collapses to give II through reductive elimination. The ensuing hydrolysis with water delivers the final γ‐lactam product, thus completing the formal carbonylation process.

In summary, we showed that difluorocarbenes could be harnessed as carbonyl surrogates to achieve efficient and site‐selective carbonyl insertions into azetidines, enabling access to prized γ‐lactam pharmacophores under mild copper‐catalyzed conditions at low temperature. We expect this new catalytic method to provide a practical synthetic alternative to high‐temperature carbonylations and find applications in heterocycle synthesis^{[ 47 ]} as well as late‐stage skeletal editing^{[ 48} , ⁴⁹ , ⁵⁰ , ^{51 ]} applications in drug discovery campaigns.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Zhou F., Tan T.‐D., Koh M. J., Angew. Chem. Int. Ed. 2025, 64, e202505033. 10.1002/anie.202505033

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.