Introduction

Pyridines and their derivatives are prevalent in a wide variety of pharmaceuticals, agrochemicals, bioactive molecules, chelating ligands, functional materials, and others (Fig. 1A)1,2,3,4,5,6,7. In particular, pyridine is the most common nitrogen heterocycle found in drugs approved by the US Food and Drug Administration, with a presence in 18% of the top-selling agrochemicals8. Accordingly, the late-stage functionalization of pyridine has received continuous attention in both academic and industry sectors9,10,11,12. While the ortho- and para-positional functionalization is straightforward13,14,15,16,17,18,19, functionalizing the meta-position is far more challenging, normally requiring harsh reaction conditions20,21. To address this problem, Yu, Shi, Oestreich, Hartwig, and others have developed transition metal-catalyzed meta-selective pyridine C3–H functionalization reactions with the assistance of a suitable ligand22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37. However, these reactions are burdened by the need for a large excess of pyridine substrates, regioselectivity problems, or high-temperature requirements, thus hindering their practical applications in the late-stage functionalization of pyridine-containing pharmaceuticals. New strategies are thus needed to achieve the highly selective pyridine C–H functionalization under mild conditions.

Fig. 1: Pyridine C3–H functionalization reactions.
figure 1

A Examples of drug molecules containing pyridine motifs. B Regioselective C–H functionalization of pyridines via Zincke imine intermediates. C This work: Radical C3–H functionalization of pyridines via N-DNP Zincke imine intermediates.

Full size image

The utilization of an umpolung strategy to alter the electronic properties of pyridines has emerged as a powerful platform for achieving mild pyridine C3–H functionalization38,39. Noteworthy contributions from researchers such as Wang40,41,42,43,44, Studer45,46 and Yu47 involved the temporary transformation of electron-deficient pyridines into electron-rich dihydropyridines or pyridine radical anion, thus enabling the selective C3–H functionalization of pyridines by a range of electrophiles.

Recently, another elegant approach for the C3-selective pyridine halogenation (Cl, Br, I) under mild conditions has been developed by McNally, Paton, and co-workers, involving the N-Tf-Zincke imine intermediates48. In contrast to direct modification on the pyridine ring, the Zincke imine49,50,51,52,53,54,55, readily generated through the nucleophilic ring-opening of the corresponding Zincke salt by a secondary amine, could be efficiently and regioselectively functionalized with electrophiles and further converted to pyridine via 6π-electrocyclization, thus providing an alternative approach for the late-stage functionalization of pyridine under very mild conditions (Fig. 1B). In addition to the electrophilic functionalization of Zincke intermediate, Greaney and co-workers have more recently realized the regiodivergent arylation of pyridine by utilizing the N-Tf-Zincke intermediate (Fig. 1B)56. They accomplished C4 arylation under Pd catalysis and C3 arylation using diaryliodonium salt. Despite these notable advances, the range of transformations and reaction patterns for the functionalization of pyridine via the Zincke intermediate is still relatively restricted and awaits further exploration.

The radical substitution reaction, based on the radical addition-elimination pathway, represents a fundamental pathway for the transformation of unsaturated compounds57,58,59,60,61,62,63. The Zincke imine intermediates exhibit alternative charge distributions along the polarized alkene chain. For instance, within the DNP-Zincke imine, the Fukui fA0 coefficients and natural charges at C3(0.08, −0.41) and C5 (0.07, −0.36) are significantly higher than those at the C4 position (0.03, −0.10), indicating that the meta-position is more kinetically favorable for attack by electrophilic radical species (Fig. 1B)64,65,66. We envisioned that successfully merging the radical reaction mode with Zincke imine intermediates might unlock synthetic routes and broaden the scope of pyridine C3–H transformations. However, the integration of the Zincke imine intermediate with a radical pathway remains underdeveloped.

Here, we present our efforts on regioselective functionalization of 3(4)-substituted pyridines by harnessing the reactivity of the N−2,4-dinitrophenyl (DNP) Zincke imine intermediate with electrophilic chalcogen radical species (Fig. 1C), including arylthiolation, alkylthiolation, trifluoromethylthiolation, thiocyanation, and arylselenylation. Notably, the synthetically challenging C3-H fluorination was also achieved through a two-electron pathway. The use of classic N-DNP-Zincke imine intermediate plays a crucial role in the success of these transformations, distinguishing it from previously established methods.

Results

Optimization studies

We commenced our study with a model reaction, employing N-activated 3-phenylpyridine (1) as the substrate, N-(phenylthio) phthalimide (Phth–SPh, 2a) as the thiolation reagent, and pyrrolidine as the nucleophilic ring-opening reagent in CH3CN at 80 °C for 16 h. To realize such transformation, the ideal N-activating group should efficiently initiate a ring-opening process, resulting in the generation of Zincke imine upon nucleophilic attack by pyrrolidine. In addition, it should automatically be removed after the ring-closing of the thiolated Zincke imine. With these considerations in mind, we initially conducted a systematic examination of a series of electron-withdrawing activating groups, including triflyl (Tf), cyano (CN), nitroso (NO), 3-nitrophenyl and 2,4-dinitrophenyl (DNP) groups. Of these, only the DNP activator produced the thiolation product 3a in 35% yield (Fig. 2A). Subsequent investigations on other types of thiolation reagents did not give better results than 2a (Fig. 2B). Although 2a is known to participate in thiolation reactions to provide a sulfur cation or an electrophilic thiyl radical, the absence of a Lewis acid in our system may prevent the generation of the sulfur cation67,68. Moreover, a conventional sulfur cation-donating reagent, phenylsulfenyl chloride, failed to give 3a. These observations indicated that the thiolation reaction may not proceed through a two-electron pathway in our protocol. Therefore, 2a was proposed to react with Zincke imine through a radical mechanism. Exploring a variety of secondary amines revealed that pyrrolidine remains the optimal nucleophile, as depicted in Fig. 2C. Subsequently, by meticulously examining various reaction parameters (for details, see Supplementary Tables 1–8), including temperatures, additives, and solvents, the yield was eventually optimized to 80% by using H2O as an additive and methyl tert-butyl ether (MTBE) as a co-solvent (Fig. 2D). It is worth noting here that the use of H2O as an additive is pivotal to secure the high yields. Based on our control experiments (Supplementary Figs. 21–24), we attributed the positive effect of H2O to its dual roles, that is, improving Zincke salt dissolution and promoting the conversion of unproductive streptocyanine to effective Zincke imine.

Fig. 2: Optimization of the reaction conditions[a].
figure 2

A Optimization of N-activating groups. B Variation of sulfur reagent. C Optimization of secondary amine. D Control experiment. [a] Overall isolated yield. [b] Reaction conditions: 1 (0.2 mmol), 2a (0.4 mmol), pyrrolidine (0.4 mmol), CH3CN (0.1 M), 80 °C, 16 h. [c] Reaction conditions: 1a (0.2 mmol), 2 (0.4 mmol), pyrrolidine (0.4 mmol), CH3CN (0.1 M), 80 °C, 16 h. [d] Reaction conditions:1a (0.2 mmol), 2a (0.4 mmol), secondary amine (0.4 mmol), CH3CN (0.1 M), 80 °C, 16 h. [e] Reaction conditions: 1a (0.2 mmol), 2a (0.4 mmol), pyrrolidine (0.2 mmol), H2O (2.0 mmol, 10.0 eq.), CH3CN (0.1 M), MTBE (0.4 M), 80 °C, 16 h. “n.d.” stands for not detected.

Full size image

Substrates scope studies

With the optimal conditions in hand, we then investigated the generality of this approach (Fig. 3). This transformation displayed good functional group tolerance and broad substrate scope. N-(2,4-dinitrophenyl) pyridinium salts with various aromatic substitutions at the C3-position of pyridine scaffold were found to be particularly effective, including ortho, meta, and para substituted phenyl rings with various electron-donating (e.g., alkyl, OMe, phenyl and vinyl) or -withdrawing groups (e.g., F, Cl, CF3, CN and Ac). This allowed for the preparation of diverse C3-phenylthiolated pyridine derivatives with moderate to good yields (3b–3q). Notably, in the case of the pyridinium salts bearing the naphthyl or the heteroaryl groups, the thiolation still occurred at the C3-position of the pyridine ring in moderate to good yields (3r–3 v). In addition, alkyl groups at the C3 position of the pyridine were well tolerated, giving 3w–3z in synthetically useful yields. Notably, in the case of the pyridinium salts bearing the cyclohexenyl group, the thiolation still selectively occurred at the C3 position in 40% yield (3aa). 4-Substituted pyridinium salts with two meta-C‒H bonds were also suitable substrates. In most cases, reactions proceeded smoothly to afford the thiolated products in moderate yields (3ab–3af, 30–66%) with nearly monoselectivity, while the 4-cyclopropylpyridinium salt yielded the bisthiolated product 3ag in 37% yield. Disubstituted pyridinium salts, such as 3,4-diphenylpyridinium salt, can be used as reaction substrates to obtain the desired product 3ah in 50% yield. The isoquinolinium salt was also tolerated in this transformation, giving the corresponding product 3ai in 62% yield. Furthermore, fused heterocyclic pyridinium salts, such as thieno[3,2-c]pyridinium salt and 1-methyl-1H-pyrrolo[2,3-c]pyridinium salt, also reacted smoothly to afford the corresponding thiolated products 3aj and 3ak, with yields of 50% and 25%, respectively. However, it is worth noting that this protocol also has limitations in pyridine substrates. The unsuccessful substrates, including those that were difficult to synthesize and those that yielded very low products, are compiled in Supplementary Fig. 85.

Fig. 3: Substrate scope and synthetic applications[a].
figure 3

A Phenylthiolation. B Bioactive molecules and Drug-Like fragments. C Gram-scale experiment. D Synthetic transformation. E Extending to other types of functionalization. [a] Overall isolated yields. [b] Reaction conditions: 1a (0.2 mmol), 2a (2.0 eq.), pyrrolidine (1.0 eq.), H2O (2.0 mmol, 10.0 eq.), CH3CN (0.1 M), MTBE (0.4 M), 80 °C, 16 h. [c] N-deactivated starting materials. [d] m-CPBA (1.0 eq.), DCM (0.2 M), 0 °C, 8 h. [e] m-CPBA (2.0 eq.), DCM (0.2 M), 80 °C, 16 h.

Full size image

The established protocol was also applicable to the late-stage functionalization of pyridine-containing drugs and functional molecules (Fig. 3B). For instance, abiraterone acetate, a hormone therapy drug, could be meta-selectively functionalized with a phenylthio group in moderate yield (3al). Other pyridine derivatives containing drug molecule fragments could also be transformed into the corresponding thiolated derivatives (3am–3ap) in moderate to good yields. In the case of functional molecules containing multiple pyridine rings, the thiolation selectively occurred at the position with less steric hindrance, as demonstrated in the formation of 3aq.

The practicality of this method was further highlighted in the gram scale synthesis of 3a in 67% yield (Fig. 3C). Both one-pot and two-step protocols worked equally well. Notably, the by-products 1-chloro-2,4-dinitrobenzene (DNP–Cl) and phthalimide were successfully recycled for the next thiolation. The recovered DNP–Cl was employed to activate pyridine and participated in the subsequent thiolation reaction to give 3a in 70% yield. The installed thioether moiety could be readily oxidized to corresponding sulfoxide 3ar and sulfone 3as (Fig. 3D). In addition to phenylthiyl group, this protocol also proved to be robust to install a wide variety of functionalities on C3 position of pyridine, such as trifluoromethylthiyl (3at), cyclohexylthiyl (3au), thiocyanate (3av) and phenylseleno (3aw) groups (Fig. 3E).

Mechanistic studies

A series of control experiments were conducted to elucidate the reaction mechanism underlying the C3–H thiolation of pyridine (Fig. 4). To verify our hypothesis that the thiolation may proceed through a radical pathway, radical inhibition experiments were initially performed69. It was observed that the C–H thiolation was significantly suppressed by adding radical inhibitors such as 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO), 2,6-di-tert-butyl-4-methylphenol (BHT), or 1,1-diphenylethylene (Fig. 4A). Key radical adducts were also captured, as revealed by high-resolution mass spectrometry (HRMS). When diphenyl disulfide instead of 2a was used as the thiolating reagent, 3a was not produced under the standard conditions. Intriguingly, the introduction of a radical initiator, such as Phth–Br or TBHP, initiated the reaction, leading to the formation of 3a in a yield ranging from 20% to 40%. These experimental findings strongly indicate that the phenylthiolation reaction operates through a radical addition pathway. Furthermore, complete inhibition of arylthiolation, trifluoromethyl thiolation, thiocyanation, and phenyl selenylation was observed upon the addition of the radical inhibition reagent 1,1- diphenylethylene, suggesting that these reactions also follow a radical addition pathway. Additionally, no primary kinetic isotope effect (KIE) was observed from either parallel or competitive reaction between 1a and 1a-D, indicating that the C3–H bond cleavage of Zincke imine is not the rate-limiting step (Fig. 4B). To verify the key intermediate, we pre-synthesized and reacted Zincke imine (4a), streptocyanine (5a) and Zincke aldehyde (6a) under identical conditions. While 4a produced the target product, 5a and 6a were unreactive but became active with the addition of 2,4-dinitroaniline (DNP–NH2) and H2O. These experiments confirmed N-DNP-Zincke imine 4a as the effective intermediate, with H2O playing a crucial role in its reversible formation. In the presence of a π-conjugated olefinic fragment at the C3 position (1aab), a N-deletion arene 8a was obtained in 45% yield, along with a trace amount of the target product 3aaa (Fig. 4C), excluding 1,2-dihydropyridine as a possible intermediate to react with phenylthiol radical70. These results serve as strong supporting evidence for the hypothetical mechanism involving a ring-opening/radical thiolation/ring-closing cascade pathway.

Fig. 4: Mechanistic studies.
figure 4

A Radical trapping experiments. B KIE experiments. C Identification of key intermediate.

Full size image

DFT calculations

Subsequently, density functional theory (DFT) calculations at the M06-2X/6-311 G(d, p)(SMD, CH3CN)// M06-2X/6-31 G(d)(SMD, CH3CN) level of theory (Fig. 5) were conducted to gain a deeper understanding of the C–H thiolation mechanism. The reaction begins with the aromatic nucleophilic substitution of DNP–Cl with 3-phenylpyridine to form Zincke salt 1a via transition state TS1. A subsequent nucleophilic addition of the pyrrolidine to 1a generates 1,2-dihydropyridine intermediate IM3, which undergoes a ring-opening reaction to give the active Zincke imine intermediate IM4. In the following step, the IM4 either reacts with 2a via TS4-1 to form IM5-1 or reacts with phenylthiol radical to form IM5 via TS4. The calculation results show that a significantly higher energy barrier is required for the interaction of IM4 with 2a as compared with phenylthiol radical (46.6 kcal/mol versus 30.2 kcal/mol), suggesting that the radical addition pathway is more feasible in comparison to the electrophilic process. This is well consistent with the radical trapping experimental results. Subsequently, the hydrogen atom transfer process (HAT) occurs between the phthalimide nitrogen-centered radical and IM5 to generate IM7 via TS5(ΔG≠ = 3.6 kcal/mol). This step is kinetically more favorable and is in good agreement with the observed small KIE value between 1a and 1a-D. Finally, the ring closing reaction can easily proceed via TS6 with a barrier of 4.8 kcal/mol to produce IM8, upon which the pyrrolidine and DNP-Cl are readily released to generate product 3a. From an energy viewpoint, the electrophilic radical addition is the rate-limiting step (ΔG≠ = 13.1 kcal/mol) of the overall reaction. The formation of 3a is irreversible and exergonic by −90.5 kcal/mol at 353 K. The exothermicity of the overall reaction is the driving force for the success of this multistep one-pot process.

Fig. 5
figure 5

DFT studies (free energies in kcal mol−1).

Full size image

Finally, the energy of Frontier Molecular Orbitals (FMOs) of 4a, 2a, phenylthiol radical as well as the N-Tf-Zincke imine were calculated to understand the different reaction reactivity (Fig. 6). The highest occupied orbital (HOMO) of N-DNP zincke imine (4a) are found at −6.30 eV, while the HOMO energy of N-Tf Zincke imine is −6.74 eV. The singly occupied molecular orbital (SOMO) energy level of the phenylthiol radical is found at −3.74 eV. The relatively lower energy gap between the SOMO of phenylthiol radical and HOMO of N-DNP Zincke imine might account for the excellent thiolation reactivity observed for N-DNP Zincke salt (vide supra). In addition, the lowest unoccupied orbital (LUMO + 2) of the Phth–SPh (2a) is found at −0.17 eV, the large energy difference (6.13 eV) is not conducive to the interaction between 2a and 4a. This comparative result further supports the radical addition pathway.

Fig. 6
figure 6

FMOs analysis by DFT. Energies are given in eV.

Full size image

Notably, the challenging meta-selective C–H fluorination also occurred under slightly modified conditions, with excellent functional group tolerance and moderate yields (Fig. 7). Radical trapping experiments indicated that the fluorination process does not involve a radical pathway. Instead, the fluorination might occur via electrophilic addition of NSFI with Zincke imine. Additionally, the energy of Frontier Molecular Orbitals (FMOs) of NFSI was calculated as −1.03 eV. The relatively lower energy gap between FMOs of NFSI and N-DNP Zincke imine indicates a higher reactivity of N-DNP Zincke imine compared to N-Tf Zincke imine (Supplementary Fig. 83). To the best of our knowledge, the selective C3-H pyridine fluorination has never been demonstrated in the literature71.

Fig. 7: Substrate scope of fluorination reaction[a].
figure 7

[a] Yields were determined by 19F NMR analysis with fluobenzene as an internal standard. [b] Reaction conditions: I. 1 (0.2 mmol), N-(pyridin-4-ylmethyl) ethanamine (0.3 mmol), MeOH (0.2 M), −10 °C, 15 min. II. NFSI (2.0 eq.), CH3CN (0.1 M), 80 °C, 16 h.

Full size image

In summary, we have achieved the C3-selective functionalization of pyridines including arylthiolation, trifluoromethylthiolation, alkyl thiolation, thiocyanation, selenylation and fluorination involving N-DNP Zincke imine intermediate. Mechanistic studies, including radical trapping experiments, KIE experiments, and DFT calculations, indicated that thiolation follows a tandem ring-opening, radical addition-elimination, and aromatized ring-closing sequence, executed in a one-pot and single operation manner. The key to this success lies in the utilization of the highly electron-deficient activator N-DNP group. This straightforward operational protocol, featuring exclusive C3-selectivity in the diverse transformation of pyridine, demonstrates its applicability in the late-stage functionalization of drug-like pyridine-containing compounds, holding a promising application prospect.

Methods

Computational details

All calculations were performed using Gaussian 09 program package, employing the M06-2X density functional with the 6-31 G(d) basis set. Geometries were optimized in CH3CN solvent and characterized by frequency analysis at 353 K. The self-consistent reaction field (SCRF) method based on the universal solvation model SMD was adopted to evaluate the effect of solvent. The intrinsic reaction coordinate (IRC) path was traced to check the energy profiles connecting each transition state to two associated minima of the proposed mechanism. Single-point energies for all stationary points were calculated at the M06-2X/ 6-311 G(d,p)(SMD, CH3CN) level of theory.

General procedure for the synthesis of Zincke Salts

An oven-dried 50 mL pressure tube with a magnetic stirring bar was charged with pyridine (5.0 mmol), 1-halo-2,4-dinitrobenzene (7.0 mmol), and acetone or EtOH (8.0 mL) in the air. The resulting mixture was then stirred at 70 °C for 16 h. Then, the solvent was removed. The reaction mixture was washed with a small amount of diethyl ether and finally dried under vacuum to get the Zincke salts.

General procedure for the synthesis of phenylthiolated products

An oven-dried 25 mL Schlenk tube equipped with a magnetic stirring bar was charged with corresponding N−2,4-dinitrophenyl pyridinium chloride (0.2 mmol), N-(Phenylthio) phthalimide (Phth-SPh) (0.4 mmol, 2.0 equiv.), pyrrolidine (0.2 mmol, 1 equiv.), H2O (2.0 mmol, 10 equiv.), methyl tert-butyl ether (MTBE) (0.5 mL), and CH3CN (2.0 mL) in the air. The resulting solution was stirred at 80 °C (heated by heating plate with magnetic stirrer) for 16 h. After cooling to room temperature, the mixture was diluted with dichloromethane and filtered through a short pad of celite, the volatiles were removed under vacuum and the residue was purified by preparative thin layer chromatography (silica gel, petroleum ether/ethyl acetate 20:1 to 10:1) to give pure product 3a-3an. Further experimental details are provided in the Supplementary Information.