Methylene Insertion into Nitrogen‐Heteroatom Single Bonds of 1,2‐Azoles via a Zinc Carbenoid: An Alternative Tool for Skeletal Editing

Abstract

The nitrogen‐heteroatom single bonds of 1,2‐azoles and isoxazolines underwent methylene insertion in the presence of CH₂I₂ (6 equiv.) and diethylzinc (3 equiv.) to produce a wide variety of the ring‐expanded six‐membered heterocycles. Density functional theory calculations suggest that the methylene insertion proceeds via cleavage of nitrogen‐heteroatom single bonds followed by ring closure.

Methylene insertions of a zinc carbenoid into nitrogen‐heteroatom single bonds of 1,2‐azoles and cyclic oximes are demonstrated. The reaction produces a wide variety of six‐membered heterocycles from easily preparable substrates. Furthermore, density functional theory calculations reveal the reaction mechanism, where cleavage of nitrogen‐heteroatom single bonds and following ring‐closure are essential for the ring‐expansion.

1. Introduction

Skeletal editing is the precise modification of molecular skeletons to facilitate the rapid diversification of complex molecular architectures.^{[ 1 ]} This strategy for modifying arenes and heterocycles involves three main transformations: ring expansion, ring contraction, and atom exchange. In particular, ring expansion with a single carbon insertion, one of the skeletal editing techniques, has been achieved through [2+1]‐cycloaddition of a carbon‐carbon double bond with a carbene or metal carbenoid followed by isomerization.^{[ 2 ]}

On the other hand, ring expansion by insertion of a carbon atom into a single bond is still less common. Diazo compounds are known as the most widely used carbene or transitional metal carbenoid precursors^{[ 3 ]} and the corresponding reactive intermediates insert into various types of single bonds, including C─C,^{[ 4 ]} C─O,^{[ 5 ]} C─Si,^{[ 6 ]} and C─S^{[ 7 ]} bonds in cyclic compounds. Nitrogen‐heteroatom single bonds in heteroarenes also undergo thering expansion with a single carbon insertion. Rhodium carbenoids insert into the N─X (X = N, O, S) single bonds of 1,2‐azoles^{[ 8 ]} to give the corresponding six‐membered products with a single carbon atom inserted (Scheme 1‐A). Although these strategies using diazo precursors lead to various transformations of the compounds, the diazo compounds have structural limitations to stabilize themselves.

Scheme 1.

Methylene insertion into nitrogen‐heteroatom single bonds of 1,2‐azoles.

Diazirines are alternative practical carbene precursors. The stability of diazirines allows their application in photoaffinity probes.^{[ 9 ]} Although the free carbenes generated from diazirines by UV light have been reported to insert C─H, N─H, O─H, and Si─H single bonds,^{[ 10 ]} the reactivity of diazirines with other types of single bonds remains unknown. Recently, Levin et al. reported ring expansion from pyrazoles to pyrimidines via benzyl carbenes generated from chlorodiazirines (Scheme 1‐B).^{[ 11 ]}

Zinc carbenoids are well‐known reagents for cyclopropanation of carbon–carbon double bonds, called Simmons–Smith cyclopropanation. Especially, Furukawa et al. reported the reliable and reproducible iodomethylzinc reagent (EtZnCH₂I) prepared from diethylzinc (Et₂Zn) and CH₂I₂.^{[ 12 ]} Further development of zinc carbenoid reagents has improved their reactivity with unreactive olefines^{[ 13 ]} and enabled their application in asymmetric synthesis.^{[ 14 ]} However, zinc carbenoid insertion into single bonds in cyclic compounds has not been reported.

We have developed direct functionalization of isoxazoles to access multi‐functionalized or heteroarene‐fused isoxazoles.^{[ 15 ]} During the way of investigation, we found an unexpected methylene insertion of a zinc carbenoid into the N─O single bond of isoxazole‐fused azaborines despite the presence of a vinyl group in a molecule.^{[ 15f ]} In this study, we demonstrate methylene insertion into nitrogen‐heteroatom (N─X: X = N, O, S) single bonds of 1,2‐azoles I via a zinc carbenoid (EtZnCH₂I) to afford the corresponding ring‐expanded products II (Scheme 1‐C). Furthermore, we examined the mechanism of this ring expansion by density functional theory (DFT) calculations to elucidate the unique reactivity of the iodomethylzinc reagent. Since the resulting 2H‐1,3‐oxazines are labile under acidic or heating conditions, their general synthetic methods have not been established.^{[ 16} , ^{17 ]}

2. Results and Discussion

2.1. Methylene Insertion into N─O Single Bonds of Isoxazoles

We first examined the methylene insertion into 3,5‐diphenylisoxazole (1a) (Table  1 ). Treatment of 1a with excess amount of CH₂I₂ (10 equiv.) and Et₂Zn (5 equiv.) in CH₂Cl₂ gave the corresponding ring‐expanded product 2a in 63% yield (entry 1). Although the reduction of the amount of CH₂I₂ and Et₂Zn to one‐fifth resulted in the poor yield of 2a (5%: entry 2), the use of CH₂I₂ (4 equiv.) and Et₂Zn (2 equiv.) gave the product 2a in moderate yield (56%, entry 3). Next, solvent effects on the methylene insertion were examined (entries 4–7). Although oxazine 2a was not obtained in THF and ethyl acetate (entries 4 and 5, respectively), oxazine 2a was generated in moderate yield in toluene (42%, entry 6) and in good yield in 1,2‐dichloroethane (62%, entry 7). Diluted condition (in 0.05 m) increased the yield of 2a (68%, entry 8). When the reaction was carried out at 10 °C, oxazine 2a was obtained in the better yield (86%, entry 9). However, longer reaction time did not improve the yield of 2a but afforded several unidentified by‐products (entry 10). Finally, the use of CH₂I₂ (6 equiv.) and Et₂Zn (3 equiv.) gave the best result: oxazine 2a was obtained in 91% yield with 6% recovery of 1a (entry 11). To clarify the importance of zinc species, control conditions were examined by using several types of zinc species. However, no other zinc species than Et₂Zn gave oxazine 2a (Table S1, Supporting Information). Needless to say, 1a was not consumed in the absence of Et₂Zn (entry 12).

Table 1.

Optimization of methylene insertion into N─O single bond of isoxazole 1a.

Entry a)	X	Y	Solvent	Temp [°C]	2a [%] b)	Recovered 1a [%] b)
1	10	5	CH2Cl2	0 to r.t.	63	9
2	2	1	CH2Cl2	0 to r.t.	5	22
3	4	2	CH2Cl2	0 to r.t.	56	0
4	4	2	THF	0 to r.t.	0	84
5	4	2	EtOAc	0 to r.t.	0	quant.
6	4	2	toluene	0 to r.t.	42	0
7	4	2	DCE	0 to r.t.	62	0
8 c)	4	2	DCE	0 to r.t.	68	7
9 c)	4	2	DCE	10	86	9
10 c , d)	4	2	DCE	10	55	trace
11 c)	6	3	DCE	10	91	6
12 c)	6	0	DCE	10	0	quant.

^a) Reaction conditions: 1a (0.20 mmol), CH₂I₂ (X equiv.), Et₂Zn (Y equiv.), solvent (2 mL), 4 h;

^b) NMR yield using dibromomethane as an internal standard;

^c) DCE (4 mL, 0.05 m) was used;

^d) reaction time: 12 h;

Ph = phenyl, THF = tetrahydrofuran, EtOAc = ethyl acetate, DCE = 1,2‐dichloroethane.

With the optimized conditions (Table 1, entry 11), we examined the methylene insertion into various di‐ or tri‐substituted isoxazoles 1 (Table  2 ). Although oxazine 2a was labile on silica gel, its decomposition was suppressed by deactivation of silica gel using triethylamine (1% v/v in the eluent), and oxazine 2a was isolated in 80% yield by preparative thin‐layer chromatography. The reaction was applicable to a 1.0 mmol scale reaction, and 2a was obtained without a significant decrease in yield (76%). With the established procedures, isoxazoles having aryl groups gave oxazines 2b–d in good yields. Isoxazole 1e having a sterically hindered 2‐tolyl group at the C‐3 position (R¹) gave oxazine 2e in low yield (36%). Although alkyl groups such as cyclohexyl (1f) and tert‐butyl (1 g) were tolerated to give the corresponding oxazines 2f and 2 g in 26% and 62% yields, respectively, an electron withdrawing group such as an ethoxy carbonyl group gave oxazine 2 h in high yield (85%). Aryl groups at the C‐5 position gave the products 2i‐l in moderate yields (42–52%). Although tert‐butyl group and ester group were also tolerated at the C‐5 position, oxazine 2n having an ethyl ester group was obtained in lower yield (2h: 85% vs 2n: 42%). We carefully examined the difference in yields between products 2 h and 2n. A total of three experiments were performed for each, and the average yields were 81% and 41%, respectively (for each of yields, see Supporting Information). These results indicate that the difference in yields is due to the substituent effects. Furthermore, silyl substituents as R³ gave oxazines 2o and 2p in high yields. Trisubstituted isoxazoles 1q and 1r gave the corresponding oxazines 2q and 2r in 56% and 39% yields, respectively, without affecting the bromo and ester substituents in the molecule. Benzoisoxazoles 1s and 1t were also converted to the corresponding bicyclic oxazines in moderate to high yield (2s: 61%, 2t: 88%). Furthermore, we applied the methylene insertion to the late‐stage skeletal editing of drug molecules. Zonisamide (1u),^{[ 18 ]} a drug approved by the FDA in 2000, afforded the product 2u in 17% yield. Benzioxazole 1v,^{[ 19 ]} a histone deacetylase (HDAC) inhibitor also underwent the methylene insertion selectively into the N─O bond of isoxazole in the presence of a 1,2,3‐triazole ring in the molecule, giving the corresponding product 2v in 16% yield. As for unsuccessful experiments,^{[ 20 ]} isoxazoles with nonsubstituted (1w) and monosubstituted at the C‐3 (1x) or C‐5 (1y) position did not afford the corresponding oxazines 2w‐y, suggesting that substituents R¹ and R³ strongly affect this reaction (for further unsuccessful examples, see: Figure S1, Supporting Information). The DFT calculations suggest the possibility of oxazine formation, although the presence or absence of substituents affects the activation barrier to N‐alkylation. These results suggest that unsubstituted isoxazoles can lead to ring expansion products, but the products formed may be unstable. In fact, these oxazine derivatives have not been reported. Furthermore, complex mixtures were also obtained with mono 3‐ or mono 5‐substitutions. Substrates that were not successful are listed in Figure S1 (Supporting Information).

Table 2.

Scope and limitations of isoxazoles. ^a)

^a) Reaction conditions: substrate 1 (0.20 mmol, 1.0 equiv.), CH₂I₂ (6.0 equiv.), Et₂Zn (3.0 equiv.), DCE (4.0 mL), 10 °C, 4 h;

^b) 1.0 mmol of 1a was used;

^c) Average yield in three trials;

^d) CH₂I₂ (3.0 equiv.) and Et₂Zn (1.5 equiv.) were used;

Me = methyl, Cy = cyclohexyl, t‐Bu = tert‐butyl, TMS = trimethylsilyl, TIPS = triisopropylsilyl.

2.2. Methylene Insertion into N─X Single Bonds of 1,2‐Azoles and Cyclic Oximes

We next investigated ring‐expansion of other 1,2‐azoles and related heterocycles (Table  3 ). When the 3,5‐diphenylisothiazole 3a was exposed to the optimized conditions, the ring‐expanded product 4a was generated in 35% yield with 65% recovery of the starting material 3a. We also examined methylene insertion of EtZnCH₂I into N─N single bonds of pyrazoles. Although N‐tosylated pyrazole 3b gave dihydropyrimidine 4b in 19% yield, pyrazoles having other substituents (R = H, acetyl, tert‐butoxycarbonyl, methanesulfonyl) did not give the corresponding products. Indazoles having methoxy (3c), benzyloxy (3d), or 2,4,6‐trimethylbenzoyloxy (3e) groups afforded the corresponding bicyclic dihydropyrimidines 4c–e in 31–39% yields. Increasing the reaction temperature or the amount of the iodomethylzinc reagent did not improve the reaction yield. On the other hand, unsubstituted indazole (R = H) gave N‐methyindazole in 74% yield via methylene insertion into N─H single bond (Scheme S1, Supporting Information). Furthermore, the current methylene insertion mediated by zinc carbenoid was applicable to five‐membered cyclic oximes. Indeed, diphenyl dihydrooxazine 4f was produced in 80% yield from diphenylisoxazoline 3f. The ring‐expanded products 4 g and 4 h were also obtained albeit in 30% and 57% yields, respectively. The low yields of 4 g and 4 h are probably due to the instability of dihydro oxazine structures which were readily converted to β‐keto alcohols via hydrolysis. In fact, the hydrolyzed products 5i and 5j were obtained from 3i and 3j, respectively, and the desired 4i and 4j were not observed. In the same manner, the methylene insertion into the six‐membered oxime 3k was attempted, resulting in only γ‐keto alcohol 5k in 74% yield.

Table 3.

Investigation of methylene insertion to other 1,2‐azoles 4a‐e and cyclic oximes 4f‐k. ^a)

^a) Reaction conditions: substrate 3 (0.20 mmol, 1.0 equiv.), CH₂I₂ (1.20 mmol, 6.0 equiv.), Et₂Zn (0.60 mmol, 3.0 equiv.), 1,2‐dichloroethane (4 mL), 10 °C, 4 h;

^b) Reaction conditions: substrate 3 (0.20 mmol, 1.0 equiv.), CH₂I₂ (0.60 mmol, 3.0 equiv.), Et₂Zn (0.30 mmol, 1.5 equiv.), 1,2‐dichloroethane (4 mL), 10 °C, 4 h. Ts = tosyl, Bn = benzyl, Mes = mesityl.

2.3. Mechanistic Investigation

To clarify the mechanism of the methylene insertion, we conducted DFT calculations on the ring‐expansion of isoxazole 1a and cyclic oxime 3f. All calculations were performed by the Gaussian 16 program at the level of B3LYP‐D3/LANL2DZ for I and Zn and 6–31G(d,p) for other elements in 1,2‐dichloroethane (polarizable continuum model, PCM). MeZnCH₂I^{[ 21 ]} was used as a model of zinc carbenoid for the calculations. We first examined the possibility of concerted insertion of zinc carbenoid into the N─O single bond, similar to the well‐known mechanism of Simmons–Smith reaction.^{[ 22 ]} However, no desired transition state structures were obtained, and the candidates converged to N‐ or O‐alkylated structures. Therefore, we focused on the stepwise path (Figure 1–a). Isoxazole 1a and MeZnCH₂I formed the complex Int1, and S_N2‐like N‐alkylation occurred between the nitrogen of isoxazole 1a and MeZnCH₂I, leading to the intermediate Int2 via the transition state TS1. The energy barrier required for this N‐alkylation (ΔG ^‡ _1a→TS1) is 14.5 kcal mol⁻¹ (also see entry 1 in Figure 1–b). After the N‐alkylation, the zinc moiety (IZnMe) was eliminated from the intermediate Int2 to afford the ylide intermediate Int3, followed by the N─O single bond cleavage to give the intermediate Int4 through the transition state TS2. The energy barrier required for the N─O single bond cleavage (ΔG ^‡ _Int3→TS2) is 0.5 kcal mol⁻¹, and this small energy barrier agrees with the calculation by Khlebnikov's group.^{[ 23 ]} Finally, electrocyclization of the intermediate Int4 occurred through the transition state TS3 with an energy barrier (ΔG ^‡ _Int4→TS3 = 7.8 kcal mol⁻¹) to yield oxazine 2a. These calculations suggest that the initial step from 1a to TS1 is a rate‐determining step of the entire reaction and that the coordination to zinc in Int1 would promote this N‐alkylation step. This was supported by the control experiment where no reaction was observed in the absence of diethylzinc (Table 1, entry 12).

Figure 1.

a) Gibbs free energy profile of methylene insertion into the N─O single bond of isoxazole 1a; b) the summary of activation energies and product yield of each 1,2‐azole and c) Gibbs free energy profile of methylene insertion into the N─O single bond of cyclic oxime 3f via MeZnCH₂I. All calculations were conducted at the level of B3LYP‐D3/LANL2DZ for Zn, I, and 6–31G(d,p) for other elements in 1,2‐dichloroethane (PCM).

To compare the reactivities of 1,2‐azoles, we calculated the energy profiles of isothiazole 3a and N‐tosyl‐3,5‐diphenylpyrazole 3b (for the whole energy profiles, see Figures S2 and S3, Supporting Information). The activation energies in the steps TS1, TS2, and TS3, and product yields for 1,2‐azoles 1a, 3a, and 3b are summarized in Figure 1b. The rate‐determining step of each substrate is the same N‐alkylation step. The activation energies required for isothiazole 3a and pyrazole 3b were higher than that of isoxazole 1a (1a: ΔG ^‡ _TS1 = 14.5 kcal mol⁻¹ vs 3a: ΔG ^‡ _TS1 = 25.2 kcal mol⁻¹ and 3b: ΔG ^‡ _TS1 = 29.9 kcal mol⁻¹). These results are consistent with the lower yields of products 4a (35%) and 4b (19%) compared to 2a (80%). The activation energies required for the following ring‐opening step (ΔG ^‡ _TS2) were also higher than that of isoxazole 1a. In contrast to the first and second steps, the less activation energies were needed for the final ring‐closing steps required less activation energy (1a: ΔG ^‡ _TS3 = 7.1 kcal mol⁻¹ vs 3a: ΔG ^‡ _TS3 = 1.6 kcal mol⁻¹ and 3b: ΔG ^‡ _TS3 = 6.9 kcal mol⁻¹).

We next demonstrated the DFT calculation on the insertion of five‐membered oxime 3f (Figure 1c). Oxime 3f has a chiral center, therefore (R)‐isoxazoline 3f was employed in this calculation. As with the case of isoxazole 1a, oxime 3f underwent the N‐alkylation via S_N2‐like transition state TS4 with higher activation energy (1a: ΔG ^‡ _TS1 = 14.5 kcal mol⁻¹ vs 3f: ΔG ^‡ _TS4 = 26.0 kcal mol⁻¹) after coordination to the zinc species (Int5) and afforded the intermediate Int6. After the conformational change from Int6 to Int7, the intermediate Int7 underwent N─O single bond cleavage via the transition state TS5 with the activation energy (ΔG ^‡ _TS5) of 9.1 kcal mol⁻¹. Finally, the intermediate Int8 cyclized to afford the product 4f via the transition state TS6 and subsequent elimination of the zinc moiety. The reaction barrier (ΔG ^‡ _TS6) was 4.9 kcal mol⁻¹, a suitable activation energy for the formation of the product 4f. In contrast to the mechanisms for 1,2‐azoles, the zinc species (IZnMe) was involved in all steps after N‐alkylation. Therefore, the zinc species would be essential to sustain structures of the intermediates from Int6 to Int8.

3. Conclusion

In conclusion, we have demonstrated the methylene insertion by a zinc carbenoid (EtZnCH₂I) into nitrogen‐heteroatom single bonds in 1,2‐azoles and cyclic oximes. This transformation enables rapid access to synthetically challenging ring‐expanded heterocycles from the readily available 1,2‐azole scaffolds. The DFT calculations provided insight into the reaction mechanism of the methylene insertion by the zinc carbenoid and the difference in reactivity toward 1,2‐azoles and cyclic oximes. Since ring expansion with a single carbon insertion is an important skeletal editing technique, the current methylene insertion into nitrogen‐heteroatom single bonds in 1,2‐azoles and cyclic oximes provides an effective tool for scaffold hopping of heterocycles, especially in medicinal chemistry.^{[ 1 ]} Further extensions of methylene insertion with zinc carbenoids are currently in progress.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Tsuda M., Morita T., Morita Y., Takaya J., Nakamura H., Methylene Insertion into Nitrogen‐Heteroatom Single Bonds of 1,2‐Azoles via a Zinc Carbenoid: An Alternative Tool for Skeletal Editing. Adv. Sci. 2024, 11, 2307563. 10.1002/advs.202307563

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.