Introduction

Organic synthesis underpins the evolution and advancement of broad areas of science, from materials to medicine. Arenes are among the most widely used rings in medicine and natural products. The functionalization of arenes is a particularly attractive tool for the production of pharmaceuticals, natural products, and molecular materials1,2,3,4. However, their application has so far been largely focused on C–H functionalization chemistry (peripheral editing), and the precise modification of the aromatic ring skeleton remains largely unexplored (Fig. 1A)5,6,7. Single-atom skeletal editing has become an extremely powerful tool for straightforwardly modifying the core skeleton of organic molecules. Recently, a limited number of single–atom insertion or deletion reactions have been developed to reshape the underlying molecular skeletons8,9,10,11,12,13,14,15,16,17,18,19,20,21. However, the direct modification of valuable core structures by replacing one atom in a ring system without changing the ring size and aromaticity remains elusive22,23,24,25,26,27,28,29,30,31, although it has been recognized as a highly desirable transformation.

Fig. 1: Examples of carbon–nitrogen transmutation and our reaction design.
figure 1

A Molecular editing of aromatic rings. B Examples of the necessary nitrogen atom effect. C Conversion of aryl azides to pyridines. D This study.

Full size image

One of the most studied among N-heteroarenes are pyridines, which serve as a bioisosteric replacement of benzene counterparts within the parent molecules32,33,34. The replacement of carbon with nitrogen in aromatic ring systems can have several important effects on the molecular and physicochemical properties relevant to multiparameter optimization (Fig. 1B). This necessary nitrogen atom effect is a versatile high-impact design element for multiparameter optimization, which has been shown to improve various key pharmacological parameters35. Recently, Burns and Levin independently reported groundbreaking methods for the direct conversion of arenes to pyridines via nitrene internalization (Fig. 1C)36,37. In these process, additional steps for installation and isolation of aryl azides are always requried, which indicated that a selective, and straightforward transformation of diverse arenes into N-heteroarenes remains an important goal38.

A key challenge in this transformation is the stability of the aromatic compounds. Our design overcomes this intrinsic challenge using arenols as substrates. Dearomatization of arenols disrupts the stability of the aromatic ring and promotes subsequent skeletal transformations39,40,41,42,43,44,45,46,47,48. Arenol can also act as a selectivity controlling element in site-selective skeletal editing. Our group’s recent work employing this dearomative strategy to promote ring expansion of arenols inspired us to continue investigating this strategy to more complex skeletal editing transformation49. In this work, we describe the direct carbon-to-nitrogen transmutations in arenols. This reaction involves two stages: ring expansion and contraction (Fig. 1D). In the first stage, the insertion of nitrogen atoms is achieved by azidative dearomatization of an arenol and intramolecular aryl migration. In the second stage, a carbon atom moves out of the ring skeleton through ring-opening, and ring-closing (ANRORC), which ultimately furnishes desired carbon–nitrogen transmutation in polycyclic arenol skeletons50,51,52.

Results

Reaction optimization

We began our investigation using methylphenanthren-9-ol 1a as the reaction partner. (PhO)2POOH, NBS, and N(nBu)4N3 were employed as reagents for the in situ formation of the azido ketone intermediate (see Supplementary Information, section 2.2.2). For optimization, we observed the formation of desired product 1 in 80% yield using FeBr2 and Cy3PO as an effective catalyst–ligand combination in PhCl (Table 1, entry 1). A control experiment revealed that an iron salt was essential for obtaining the desired product (Table 1, entry 2). Other iron salts, including FeCl2, Fe(OTf)2, Fe(OAc)2, and Fe(acac)3, exhibited lower efficiency than inexpensive FeBr2 (Table 1, entries 3 − 6). Furthermore, when other established metal nitrenoid formation catalysts, including copper, rhodium, cobalt, and ruthenium, were used, the desired product was not obtained satisfyingly (Table 1, entries 7–10)53,54. Further optimization showed that this reaction was slightly improved using Cy3PO (Table 1, entry 11). The reaction appeared to be less sensitive to solvents, as replacing the PhCl with either toluene or THF furnished 1 in good yield (Table 1, entries 12 and 13). The yield decreased slightly when 10 mol% FeBr2 and 15 mol% Cy3PO were used (Table 1, entry 14).

Table 1 Screening of reaction conditions.a
Full size table

Substrate scope

Considering the optimal reaction conditions, the substrate scope was determined (Fig. 2). Various arenols, including phenanthrol (1–10), naphthol (11–30), anthranol (33) benzo(a)anthranol (34), and benzo[c]phenanthrol (35), can effectively undergo the desired carbon–nitrogen transmutation. Both electron-rich and electron-deficient aromatic substrates were suitable for the process. It was found that the substituents at the ortho positions of the arenol are significant. When the substituent was an alkyl group, the corresponding arenols underwent atom transmutation smoothly in moderate-to-good yield and chemoselectivity. The presence of a phenyl group or an electron-withdrawing group such as CO2Me at the ortho-position can inhibit this reaction. However, various functional groups, such as ether (4 and 5), acetals (6 and 16), aryl halides (7, 8, 17 and 18), esters (9), trifluoromethyl (10), trimethylsilyl (TMS) (19), alkenes (20), and alkynes (21), were tolerated in this transformation. In addition, several naphthyl-fused rings (22–25) were suitable substrates, affording the desired products in moderate-to-good yields. Heterocyclic moieties such as benzofuran (26), furan (27), dibenzofuran (28), quinoline (29), and phenoxathiine (30) were also compatible. Moreover, fused heteroarenols such as naphtho[1,2-b]thiophene (31) and naphtho[1,2-b]furan (32) can be incorporated, providing pharmaceutically interesting fused-ring skeletons that are non-trivial to prepare. The structures of 3 and 35 were identified using X-ray crystallography.

Fig. 2: Substrate scope of the carbon–nitrogen transmutation in polycyclic arenol.
figure 2

a,b aIsolated yields after chromatography are shown. b Reaction conditions: substrate (0.1 mmol), NBS (0.18 mmol), N(nBu4)N3 (0.3 mmol), (PhO)2POOH (0.075 mmol), FeBr2 (0.015 mmol), and Cy3PO (0.02 mmol) in PhCl (1.0 mL) at 100 °C for 36 h.

Full size image

Synthetic utility

The successful development of the atom transmutation protocol offers a rapid and modular approach to access complex biheteroarene skeleton, a common structural motif found in bioactive compounds (Fig. 3A). Compound 37 could be easily transformed to the iridium complexes 39, which could serve as the red-light-emitting electrochemical cell55,56. Next, the synthetic versatility of the C-to-N transmutation was demonstrated through the preparation of 3,6-disubstituted quinolines 42, which could not be obtained from directly electrophilic substitution of quinolines (Fig. 3B). Specifically, the successful development of the carbon–nitrogen transmutation offers exciting opportunities to devise more complex skeletal editing transformations via combinations of atom insertions and deletions. Benzo[1,4]diazepine 43 can be accessed through a C–H azidation and aryl migration sequence from 11, presently the formal carbon deletion and two nitrogen insertion products of starting naphthol 11a (Fig. 3C).

Fig. 3: Application potential of carbon–nitrogen transmutation.
figure 3

A Applications that allow access to complex biheteroarene skeletons. B Preparation of unconventionally 3,6-substituted quinolines. C Sequential skeletal editing transformations of naphthol.

Full size image

Mechanistic considerations

To elucidate the mechanism of this transformation, series control experiments were first conducted. The reactions without addition of NBS or N(nBu4)N3 failed to produce the desired product (Fig. 4A, equations 1 and 2). And trace amount of 11 was observed when (PhO)2POOH was absent from the reaction mixture (Fig. 4A, equation 3). It’s worth noting that azide ketone 44 could be isolated in 73% yield in the absense of Fe catalyst after 2 h (Fig. 4A, equation 4). These results indicated that the proposed azidative dearomatization of arenol might be involved (Fig. 1D)57. The azide ketone 44 was then tested with and without the addition of the Fe catalyst (Fig. 4B). It was found that the desired product 11 was formed in 40% yield, and 35% yield of 11a was isolated in presence of Fe catalyst, which demonstrated that the proposed azidation is a reversible process via successive single-electron transfer (SET) from Fe(II) to eliminate azide58. On the contrary, only trace amount of product 11 and 11a was observed without Fe catalyst. And a byproduct 45 was detected in 42% yield and recovered 44 in 24%59. These results revealed that the Fe catalyst is not only involved in aryl migration, but is also essential for the ring contraction process60.

Fig. 4: Mechanistic studies.
figure 4

A Control experiments. B Control experiments using azide intermediate showing that Fe catalyst is essential for aryl migration and ring contraction process. C Proposed mechanism.

Full size image

Based on the literature reports and our observations, a plausible mechanism is proposed (Fig. 4C). Initially, the N-bromosuccinimide-mediated dearomatization of the corresponding naphthol of 11a afforded the brominated ketone intermediate A, which subsequently reacted with N(nBu)4N3 to generate azido ketone B. Then iron salt reacted with B can form metal−nitrene species, which would then undergo 1,2-aryl migration to form ring expansion intermediate C. Subsequently, addition of hydroxide anion to the imine group of C induces N,O-hemiketal D. The collapse of D with assistance of iron salt produces ring-opening amino-ketone intermediate E61,62, which undergoes re-cyclization and dehydration to form stable benzoquinoline 11 and release Fe catalyst.

In conclusion, this study proposed a unique strategy that enables straightforward carbon-to-nitrogen transmutations in arenols through a one-pot ring expansion-contraction sequence. This site-selective atom transformation is based on sequentially combining three transformations in one pot using aryl migration and imine transposition as key steps and opens new opportunities for single-atom skeletal edit design. Further preparation of complex biheteroarene skeleton and unconventionally substituted quinoline highlights the potential of this study. This provides an alternative for the development of N-heteroarenes and demonstrates significant potential in materials chemistry.

Methods

General condition for carbon–nitrogen transmutation

Substrate (0.1 mmol), NBS (0.12 mmol), N(nBu4)N3 (0.2 mmol), FeBr2 (0.015 mmol), tricyclohexylphosphine oxide (0.02 mmol), (PhO)2POOH (0.05 mmol), and PhCl (1.0 mL) was successively added to an 10 mL sealed tube equipped with a Teflon-coated magnetic stir bar. The tube then was sealed with a Teflon screw cap and placed on a hotplate pre-heated to 100 °C with vigorous stirring. After 18 h, the reaction was cooled to room temperature and another portion of NBS (0.06 mmol, 0.6 equiv), N(nBu4)N3 (0.1 mmol) and (PhO)2POOH (0.025 mmol) was successively added to the sealed tube. The tube then reacted at 100 °C with vigorous stirring. After 18 h, the reaction was cooled to room temperature. The solvent was evaporated and the residue was directly purified by flash column chromatography on silica gel (petroleum ether/ethyl acetate = 20/1) to give the desired products.