Photochemical Organocatalytic Functionalization of Pyridines via Pyridinyl Radicals

Abstract

We report a photochemical method for the functionalization of pyridines with radicals derived from allylic C–H bonds. Overall, two substrates undergo C–H functionalization to form a new C(sp²)–C(sp³) bond. The chemistry harnesses the unique reactivity of pyridinyl radicals, generated upon single-electron reduction of pyridinium ions, which undergo effective coupling with allylic radicals. This novel mechanism enables distinct positional selectivity for pyridine functionalization that diverges from classical Minisci chemistry. Crucial was the identification of a dithiophosphoric acid that masters three catalytic tasks, sequentially acting as a Brønsted acid for pyridine protonation, a single electron transfer (SET) reductant for pyridinium ion reduction, and a hydrogen atom abstractor for the activation of allylic C(sp³)–H bonds. The resulting pyridinyl and allylic radicals then couple with high regioselectivity.

Pyridines and related nitrogen heterocycles are ubiquitous in biologically active molecules,¹ with pyridine being the most abundant heteroaromatic ring in FDA-approved drugs.² Therefore, methods for their regioselective modification are desirable. Contemporary radical chemistry has provided inroads for new pyridine functionalization strategies. The venerable Minisci reaction,³ which is based on the addition of nucleophilic carbon radicals to protonated azines, is a broadly used approach (Figure 1a). However, this process is generally characterized by competing C2/C4 positional selectivity.⁴ Strategies to selectively access C4-alkylated pyridines are available. For example, 4-cyano⁵ and 4-triphenylphosphonium⁶ pyridines are prone to react through an ipso substitution pathway upon single electron transfer (SET) reduction (Figure 1b, path i). Other C4-selective approaches require preinstallation of suitable protecting groups at the pyridine’s nitrogen, followed by a Minisci-type pathway (Figure 1b, path ii).⁷

Figure 1.

Radical pathways for the functionalization of pyridines: (a) Minisci reaction; (b) C4 functionalization via (path i) ipso substitution or (path ii) a Minisci process using blocking groups at the pyridine’s nitrogen; (c) The new reactivity promoted by pyridinyl radicals II generated upon SET reduction of pyridinium ions I.

Herein we disclose a new radical-based pathway for pyridine functionalization that mechanistically diverges from classical Minisci chemistry, enabling distinct positional selectivity (Figure 1c). This direct functionalization of simple pyridines is based on the reactivity of pyridinyl radicals II, generated upon SET reduction of pyridinium ions I formed under acidic conditions. The neutral radical II has been characterized in the past by EPR spectroscopy⁸ and exploited as a one-electron shuttle for the catalytic reduction of carbon dioxide.⁹ Surprisingly, pyridinyl radical II has been largely neglected in synthetic chemistry, since its use was limited to dimerization reactions.¹⁰ We demonstrate here that its reactivity can be leveraged to trap radicals derived from allylic C(sp³)–H bonds with high C4 regioselectivity. The net process couples two unmodified substrates via functionalization of a C–H bond on each substrate.¹¹

This study was motivated by our recent findings on the catalytic reactivity of the dithiophosphoric acid A (structure in Figure 2a).¹² Upon deprotonation, the electron-rich thiolate A^– can drive the formation of photoactive electron donor–acceptor (EDA) complexes.¹³ Visible-light excitation triggers intracomplex SET, leading to the thiyl radical A·, which activates the allylic substrate 2 via a hydrogen atom transfer (HAT) mechanism. We initially sought to translate this photochemical platform to the activation of pyridines 1. Our original idea was that pyridine protonation by the acid catalyst A(14) would form a photoactive ion pair; however, we did not observe the formation of any light-absorbing aggregate (see Supporting Information (SI) section F.2). In contrast, we soon realized (as further detailed below) that the thiolate A^– could directly absorb light¹⁵ to generate a highly reducing excited state [A^–]*, which could undergo SET to pyridinium ion I to afford the pyridinyl radical II (Figure 2a). This redox path also led to the sulfur-centered radical A·, which is prone to C–H abstraction from 2 to deliver allylic radical III.^12,16 Radical coupling of II and III would afford, upon rearomatization, the target functionalized pyridine product 3.

Figure 2.

(a) Proposed mechanism of the direct functionalization of pyridine 1a via the formation of pyridinyl radical II. (b) Optimization studies: reactions were performed on a 0.2 mmol scale at 35 °C for 16 h under illumination by a 365 nm EvoluChem LED spotlight using 10 equiv of 2a. Yields and regioisomeric ratios were determined by ¹H NMR analysis of the crude mixtures. Numbers in parentheses refer to yields of isolated 3a. (c) Cyclic voltammetry studies (scan rate = 100 mV·s^–1). (d) Calculated spin density and NBO analysis of pyridinyl radical II (uB3LYP/6-31G+(d) level of theory). (e) Stern–Volmer quenching studies of the excited anion of catalyst A2 (using A2·Et₃N as the source of A2^–) with increasing amounts of pyridinium·TFA salt I ([A2^–] = 5 mM; [I] = 0.125 to 0.625 mM; excitation wavelength = 350 nm). (f) EPR spectrum of pyridinyl radical II measured after 15 min of irradiation of a 5:1 mixture of 1a and A2 at 77 K. Collidine refers to 2,4,6-collidine (50 mol %); n.d. denotes not detected; Ir-PC = [Ir(dtbbpy)(ppy)₂]PF₆; TFA = trifluoroacetic acid. ^aThe reaction was performed on a 1 mmol scale.

This reasoning, although initially based on an inaccurate mechanistic hypothesis, explains our choice of the model reaction (Figure 2b). Specifically, we reacted simple pyridine (1a) and cyclohexene (2a) in the presence of the dithiophosphoric acid A1 as the catalyst (20 mol %). The allylic substrate 2a was chosen since our previous study¹² demonstrated its propensity to undergo effective HAT from catalyst A to afford cyclohexenyl radical III. The first experiment, conducted in acetone as the solvent under irradiation by blue light (λ = 455 nm), afforded no product formation (entry 1). The lack of reactivity was reconciled with the absence of any visible-light-absorbing aggregate, as inferred by UV–vis spectroscopic analysis of a mixture of catalyst A1 and pyridine 1a. However, these studies established the ability of the deprotonated dithiophosphoric acid (thiolate A1^–) to absorb in the near-UV region (see SI section F.2 for details). Performing the model reaction using an irradiation wavelength of 365 nm led to allylated pyridine 3a in 17% yield with an 8:1 regioisomeric ratio in favor of the C4-functionalized adduct (entry 2).

An optimization campaign identified 2,4,6-collidine (50 mol %) as an effective additive to increase the overall yield, as 3a was formed in 41% yield (Figure 2b, entry 3; see SI section F.8 for a discussion of the role of collidine). The use of the naphthyl-substituted dithiophosphoric acid catalyst A2 improved the results, delivering the allylated pyridine 3a in 67% yield with 6:1 C4 selectivity (entry 4). After purification by column chromatography, 3a was obtained as a single positional isomer. Interestingly, reactivity under blue-light irradiation could be achieved by adding an iridium-based photocatalyst, which afforded a slightly improved yield (entry 5). The process was equally efficient on a 1 mmol scale (entry 6). Control experiments established that the absence of catalyst or light completely inhibited the transformation (entries 7 and 8). A tentative explanation of the rearomatization step is given in SI section F.9.

We then performed investigations to gain mechanistic insights. Electrochemical analysis of pyridine 1a showed no reduction wave when a potential as low as −2.5 V vs Ag⁺/Ag in MeCN was applied (Figure 2c). However, a reversible reduction wave was observed for a preformed pyridinium trifluoroacetate salt, with the reduction occurring at approximately at −0.6 V vs Ag⁺/Ag. The observed well-shaped reversible behavior¹⁷ hinted at a certain kinetic stability of the delocalized pyridinyl radical II. We also evaluated the electronic properties of the key pyridinyl radical II with calculations performed at the uB3LYP/6-31G+(d) level. Greater spin density was found at C4 compared to C2 (Figure 2d), which is in agreement with the spin densities inferred from the EPR hyperfine coupling constants of II.^8c Natural bond orbital (NBO) analysis revealed that the singly occupied molecular orbital (SOMO) is localized preferentially at C4.¹⁸ These results are coherent with the experimentally observed C4 positional selectivity.

We then focused on the photochemical radical generation pathway.¹⁹ Specifically, we investigated the ability of the excited thiolate catalyst A2^– to generate pyridinyl radical II through SET reduction. Upon irradiation of the triethylammonium salt of catalyst A2 (A2·Et₃N, the source of A2^–) at 350 nm, we detected emission centered at 382 nm (Figure 2e; also see SI section F.5). This confirmed that the deprotonated catalyst A2 could access an electronically excited state. Applying the Rehm–Weller formalism,²⁰ the redox potential of the excited thiolate [E(A2·/[A2^–]*)] was estimated as −2.23 V vs Ag⁺/Ag in CH₃CN (see SI section F.3). Therefore, the anion of catalyst A2 becomes a strong reductant in the excited state, and SET to pyridinium ion I (E = −0.6 V vs Ag⁺/Ag in CH₃CN) is highly exergonic. Stern–Volmer quenching studies confirmed that protonated pyridine I (pyridinium chloride) quenched the excited state of A2^– (Figure 2e). In addition, upon UV irradiation of a mixture of pyridine and catalyst A2, low-temperature EPR analysis detected the formation of a carbon-centered radical,^8c consistent with the pyridinyl radical II formed through SET quenching of [A2]* (Figure 2f). Finally, we used transient absorption spectroscopy (TAS) to measure a half-life of ∼5 μs for [A2^–]* (see SI section F.4). Collectively, these studies support the mechanistic pathway proposed in Figure 2a.

Using the conditions described in Figure 2b, entry 4, the generality of the allylation of pyridines was evaluated (Figure 3). Notably, Minisci reactions with allylic radicals are rare,²¹ and the present radical coupling strategy based on the reactivity of pyridinyl radical II can address this gap in synthetic methodology. Ortho-substituted pyridines were converted in good yields and isolated exclusively as C4-allylated products 3b and 3c. Pyridines bearing o-halogen substituents were unsuccessful, instead undergoing dehalogenation (for a list of unsuccessful substrates, see Figure S1). The introduction of m-halogen substituents led to the formation of C4-functionalized pyridines 3d and 3e with high regioselectivity. Electron-withdrawing CF₃ and CN substituents at the meta position were also tolerated, giving the C4 products 3f and 3g in good yields. An alkyl substituent was also accepted (3h). The introduction of a bulkier aryl substituent at the meta position caused a switch of regioselectivity in favor of C6 (3i). Considering the importance of nicotinic acid and its derivatives as lipid-controlling drugs,²² we tested the reactivity of niacin (vitamin B3) and nicotinate esters, nicotinamides, and halogenated derivatives. These substrates afforded products 3j–s in good yields with a positional preference for C6 functionalization. A series of amino acid nicotinates proved to be suitable substrates, showcasing this method’s potential for the functionalization of complex pyridines (entries 3t–w). These results suggest that the greater spin density at C4, which accounts for the positional selectivity with simple pyridines, is overridden in pyridines bearing bulky 3-substitutents, such as those derived from nicotinic acid.¹⁰ Further studies are ongoing to better rationalize the interplay of steric and electronic factors that dictate regioselectivity.

Figure 3.

Photochemical organocatalytic allylation of pyridines and derivatives: (a) scope of pyridines 1; (b) scope of allylic substrates 2. Reactions were performed on a 0.2 mmol scale using 10 equiv of 2. Yields refer to isolated products 3 after purification. Products 3 were obtained as single regioisomers (>20:1 r.r.), unless otherwise stated. When more than one regioisomer was observed, the minor site of functionalization is highlighted by a gray circle. In these cases, the regioisomeric ratio (r.r.) of the crude mixture is specified, and the r.r. after isolation is reported in parentheses. When applicable, d.r. was ∼1:1. ^aYield determined by ¹H NMR analysis. ^b The conditions described in Figure 2b, entry 5 were used. Boc = tert-butoxycarbonyl.

Other azines, including isoquinolines (3x and 3y), pyrimidines (3z–bb), and pyridazine (3cc), also delivered the allylated products with high regioselectivity in modest yields. Next, we turned our attention to the direct functionalization of pharmaceuticals. Three niacin derivatives, i.e., nicoboxil (a rubefacient), etofibrate (a hypolipidemic agent), and picamilon (a dietary supplement), were successfully functionalized in good yields with exclusive C6 regioselectivity (products 3dd–ff), while the ortho-substituted pyridine bisacodyl (a stimulant laxative drug) and pyrimidine-based voriconazole (an antifungal) afforded the C4-allylated adducts 3gg and 3hh, respectively, in synthetically useful yields.

We then evaluated other C(sp³)–H partners that could serve as radical precursors for the functionalization of 2-trifluoromethylpyridine (Figure 3b). Cyclic alkenes with different ring sizes and substitution patterns afforded the desired products with complete C4 positional selectivity. Tetramethylethylene (3nn) and heterocycles, including dihydropyran (3oo) and a protected piperidine (3pp), were also suitable allylic precursors. Finally, the sp³ C–H benzylic precursors tetrahydronaphthalene and acenaphthylene were successfully reacted, affording adducts 3qq and 3rr, respectively.

Further mechanistic studies were undertaken to probe the formation of the key pyridinyl radical II (Figure 4). Both C2- and C4-cyclopropylpyridines 4a and 4b underwent ring opening under the standard conditions to give products 5a and 5b in 51% and 45% yield, respectively. The formation of these products is consistent with the intermediacy of II, which upon ring opening to give highly stabilized benzylic radical IV(23) can undergo radical coupling with allylic radical III.

Figure 4.

Probing the formation of pyridinyl radical II.

In summary, we have developed a new strategy for the radical functionalization of pyridines and related azine hetereoarenes based on the unique reactivity of pyridinyl radicals. These previously neglected intermediates, readily generated upon SET reduction of pyridinium ions, undergo effective radical coupling with allylic radicals. We envisage that this reactivity will offer further opportunities to develop new pyridine functionalization strategies.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.2c12466.

• Details of experimental procedures, full characterization data, and copies of NMR spectra (PDF)

The authors declare no competing financial interest.