Radical/Ylide-Tunable Pyridinium Salt Enables Photo/Thermocatalytic Divergent [3 + 2] Cycloadditions

Abstract

Pyridinium salts are valuable building blocks in synthetic design, yet most derivatives display only one reactive mode. We report an α-carboxyalkylpyridinium salt that functions as a dual mode reagent, acting either as a photoactivated radical precursor or as a thermally generated pyridinium ylide. The salt is readily prepared from amino acid and permits divergent [3 + 2] cycloadditions by simply changing the energy input (visible light versus heat). Two photoinduced electron transfer pathwaysintermolecular (with the photocatalyst) and intramolecular (within the pyridinium salt)generate different radical intermediates, which subsequently undergo distinct reaction sequences, affording the same γ-butyrolactone. Thermal activation induces decarboxylation to pyridinium ylides, allowing access to tetrahydroindolizines via 1,3-dipolar cycloaddition, which subsequently undergo an unprecedented photocatalytic 1,2-phenyl migration to yield fluorescent indolizines.

Introduction

Pyridinium salts represent privileged scaffolds in medicinal chemistry (e.g., Mestinon, NAD, NADP) and function as versatile synthetic intermediates. Despite their broad utility in synthesis, the development of multimodal pyridinium platformscapable of accessing divergent reaction pathways from a single precursorremains an unresolved challenge. Pyridinium ylide precursors and Katritzky salts are representative synthetic intermediates with distinct functions dictated by their mutually exclusive structure characteristics (Figure a). For instance, the 2,4,6-triphenyl-substituted framework is central to Katritzky radical precursors, , whereas the 2,6-disubstituted ylide precursors remain scarce, − likely due to the difficulty in forming quaternary carbon stereocenters. Moreover, although a highly electron-deficient α-carbon is essential for pyridinium ylide formation, , it also impedes the C–N bond cleavage in Katritzky salts. We therefore sought to determine whether a unified pyridinium architecture could be designed to divergently generate pyridinium ylides and free radicals, thereby enabling access to product scaffolds beyond the reach of conventional ylide and radical precursors.

1.

(a) Structural characteristics of pyridinium ylide precursor and Katritzky salt. (b) Activation mode of Katritzky salt. (c) Structural design and reactivity prediction.

Research on Katritzky salts commenced in the 1980s with pioneering work by Katritzky et al. on their synthesis and applications in electrophilic substitutions. − The renaissance of Katritzky salts began when Watson and coworkers used them as nonhalide alkyl radical precursors in nickelcatalysed Suzuki-Miyaura coupling. − Subsequent photoredox studies by Glorius and others demonstrated that single-electron transfer (SET) from a photocatalyst to a Katritzky salt enables radical generation , (Figure b). Aggarwal’s group in 2018 introduced electron donor–acceptor (EDA) complexes formed between a Katritzky salt and an electron donor, which absorb blue light and undergo direct photoinduced SET from the donor to the Katritzky salt acceptor, yielding alkyl radicals. − However, conventional Katritzky salts are not capable of spontaneous activation in the absence of a photocatalyst or a reductant partner, and they do not provide access to ylide-based reactivity.

To address these limitations, we envisioned integrating an intrinsic electron-donating motif within the Katritzky salt structure to enable intramolecular electron transfer (IET), , thereby eliminating the need for external donors. Inspired by the intermolecular SET observed in dithiocarbamate-Katritzky salt EDA complexes, − we designed the α-carboxyalkylpyridinium salts derived from amino acids (Figure c). We hypothesized that photoexcitation would cleave the C–N bond, releasing α-carbon/carboxylate diradicals − that could undergo [3 + 2] cycloaddition with alkenes to form γbutyrolactones. In contrast, under thermal conditions, decarboxylation was expected to generate pyridinium ylides, enabling 1,3-dipolar cycloadditions to yield sterically hindered indolizinesscaffolds previously inaccessible from classical ylide precursors. −

Results and Discussion

We selected pyridinium salt 1a (derived from alanine) and 2-vinylnaphthalene 2a as model substrates. Upon irradiation of a mixture of 1a (0.3 mmol) and 2a (0.1 mmol) in dichloromethane (1.0 mL) under an argon atmosphere with 10 W 455 nm LEDs at room temperature for 12 h, γ-butyrolactone 3a was obtained in 37% isolated yield (Table , entry 1). Switching to shorter-wavelength light (425 nm) resulted in a lower yield (Table , entry 2). The use of CHCl₃ or DCE as solvent led to similar or slightly diminished yields (Table , entries 3–4). Reducing the reaction concentration improved the yield to 50% (Table , entry 5), probably due to suppression of radical polymerization. Although the reaction could proceed in the absence of a photocatalyst, adding PC-1 (5 mol %) significantly accelerated the process, affording 3a in 99% yield within 3 h (Table , entry 6). Other photocatalysts (PC-2 and PC-3) gave low or trace yields (Table , entries 7–10). No reaction occurred in the dark (Table , entry 13), confirming the photochemical nature of the radical generation step.

1. Optimization of the Reaction Conditions .

entry	solvent	photocatalyst	time (h)	yield (%)
1	DCM	–	12	37
2	DCM	–	12	30
3	CHCl3	–	12	35
4	DCE	–	12	32
5	DCM	–	12	50
6 ,	DCM	PC-1	3	99
7 ,	DCM	PC-2	3	21
8 ,	DCM	PC-2	12	71
9 ,	DCM	PC-3	3	trace
10 ,	DCM	PC-3	12	trace
11 ,	toluene	PC-1	3	trace
12 ,	CHCl3	PC-1	3	57
13 , ,	DCM	PC-1	3	NR

^aReaction conditions: 1a (0.3 mmol), 2a (0.1 mmol) in solvent (1.0 mL) under an argon atmosphere with 10 W 455 nm LED irradiation at room temperature.

^bIsolated yield.

^c10 W 425 nm LED irradiation.

^d2.0 mL solvent.

^e5 mol % photocatalyst.

^fIn the dark.

Under optimized conditions, we evaluated a range of aryl alkenes (Scheme ). Substituted 2-vinylnaphthalenes afforded γ-butyrolactones 3b–e in 88–99% yield. Para-substituted styrenes bearing electron-donating or mildly electron-withdrawing groups furnished products 3f–j in 52–86% yield, whereas 4-methoxycarbonyl styrene exhibited lower efficiency. Both ortho- and meta-substituted styrenes with either donating or withdrawing groups were well tolerated (3l–s), although electron-withdrawing substituents led to more noticeable yield reductions. Notably, a heteroaromatic alkene, which was not reactive in α-halocarboxylic acid cycloadditions, was compatible under this system, delivering 3u in acceptable yield. 1,1 and 1,2-disubstituted alkenes afforded lactones 3v–z in moderate yields.

1. Substrate Scope of Aryl Olefins in Photocatalytic cycloaddition .

^a Katritzky salt 1a (0.3 mmol), styrene 2 (0.1 mmol), and fac-Ir­(ppy)₃ (5 mol %) in DCM (2.0 mL) under an argon atmosphere with 10 W 455 nm LED irradiation at room temperature for 3 h.

Substituent effects influenced both yield and diastereomeric ratio, with styrenes bearing electron-donating groups not only affording γ-butyrolactones in higher yields but also exhibiting slightly superior dr values compared to those with electron-withdrawing groups. The presence of a methyl group at the α-position of styrene significantly enhanced diastereoselectivity (3v), and substitution with a bulkier ethyl group led to γ-disubstituted butyrolactones (3w) in 69% yield with a dr >20:1. 1-Substituted naphthalene delivered 3aa in 53% yield. The pharmaceutically relevant α-asarone formed 3ab in 50% yield. Butadiene underwent regioselective cycloaddition at the exocyclic double bond to give 3ac, with no endocyclic products observed. We then investigated pyridinium salts derived from various amino acids (Scheme ). Salts prepared from leucine, phenylalanine and methionine afforded lactones 3ad-3af in good to excellent yields (78–91%). The use of O-benzyl serine and O-methyl tyrosine provided 3ag and 3ah in 49% and 70% yield, respectively. In contrast, the glycine-derived salt failed to produce the desired cycloadduct. Interestingly, a salt derived from an alanine-containing dipeptide underwent chemoselective cleavage of the amide bond under photoirradiation, yielding lactone 3a′ in 71% yield.

2. Substrate Scope of Pyridinium Salts in Photocatalytic Cycloaddition .

^a Katritzky salt 1 (0.3 mmol), 2-vinylnaphthalene 2a (0.1 mmol) and fac-Ir­(ppy)₃ (5 mol %) in DCM (2.0 mL) under argon an atmosphere with 10 W 455 nm LED irradiation at room temperature for 3 h.

Having established the versatility of α-carboxyalkylpyridinium salts as radical precursors, we next investigated their complementary reactivity as pyridinium ylide precursors. Under thermal conditions, these salts undergo decarboxylation to generate pyridinium ylides that participate in 1,3-dipolar cycloaddition. The thermally induced [3 + 2] cycloaddition of pyridinium salt 1a with a series of electrophilic alkenes proceeded smoothly in CH₂Cl₂/Et₃N (1:1) at 50 °C (Scheme ), affording tetrahydroindolizines 5a–e in 42–80% yields. The pyridinium salts derived from phenylalanine, methyltyrosine, and O-benzylserine were also dual photo- and thermal-responsive, as evidenced by the formation of tetrahydroindolizines 5f, 5g, and 5h. Notably, we discovered an photocatalytic oxidation–rearrangement process. Under photocatalytic conditions, tetrahydroindolizines underwent a-1,2-aryl migration − to yield fluorescent indolizines 6a–c in moderate yields. This rearrangement did not proceed in the dark, confirming its light-dependent nature. Indolizines 6 display intense UV absorption and yellow-to-red emission that matches the peak efficiency band of silicon solar cells. Coupled with their large Stokes shifts (Δλ ≥ 220 nm), indicating their potential value for application in luminescent solar concentrators (LSCs).

3. Thermocatalytic Cycloaddition and Derivatization .

^a Katritzky salt 1a (0.5 mmol) and acrylate 4 (1.5 mmol) in DCM:Et₃N = 1:1 mixed solvent (5.0 mL) in air at 50 °C for 16 h.

^b Tetrahydroindolizine 5 (0.1 mmol) and fac-Ir­(ppy)₃ (2 mol %) in DCM:Et₃N = 1:1 mixed solvent (2.0 mL) under an argon atmosphere at room temperature for 12 h.

Mechanistic Studies

To elucidate the reaction mechanism, we conducted a series of control experiments (Scheme ). Both the photocatalyst-mediated and photocatalyst-free [3 + 2] cycloadditions were completely suppressed by TEMPO (Scheme a,c), with the respective formation of radical adducts 7 and 9. This result supports a radical-based mechanism. When unactivated alkenes were employed as substrates, no formation of γ-butyrolactones was detected regardless of the presence of a photocatalyst (Scheme b,d). However, the catalyst-free system afforded compound 10 in 9% yield. The difference in products indicates a marked divergence in the reaction pathways with and without a photocatalyst. To clarify the reaction pathway of the photocatalyst-free [3 + 2] cycloaddition, we first examined the activation mechanism of the pyridinium salt. No increase in absorbance was detected upon adding 2-vinylnaphthalene to a dichloromethane solution of the pyridinium salt (Scheme h), ruling out activation via an EDA complex. Furthermore, neither dichloromethane nor 2-vinylnaphthalene exhibits appreciable absorption of blue light. These results suggest that the pyridinium salt is likely activated spontaneously in the absence of a photocatalyst. We subsequently confirmed that the carboxyl group is essential for this self-activation (Scheme e,f). Finally, Stern–Volmer quenching studies showed that the emission intensity of pyridinium salt 1a continuously decreased with increasing concentration of 2-vinylnaphthalene, indicating that the pyridinium salt is probably directly photoexcited and undergoes single-electron transfer or energy transfer with 2-vinylnaphthalene (Scheme i). Separately, heating the salt in the absence of any additives led to clean decarboxylation, directly demonstrating its inherent thermal reactivity (Scheme g). For the photoinduced oxidative rearrangement of tetrahydroindolizine, the absence of the photocatalyst, triethylamine, or dichloromethane each led to a significant decrease in yield. When the reaction was conducted in air, the yield dropped to 32%, suggesting that oxygen likely has a detrimental effect on the reaction. Moreover, the use of TEMPO also partially suppressed product formation (see SI, Table S3). Based on these results and literature precedents, we propose the mechanistic pathways outlined in Scheme . The photocatalyst-mediated [3 + 2] cycloaddition likely proceeds via a carbocationic pathway. −

4. Mechanistic Experiments.

5. Proposed Mechanism.

The pyridinium salt first undergoes a single-electron transfer with the excited-state photocatalyst, followed by cleavage of the C–N bond to generate the carboxylate-α-carbon radical IA. Subsequently, IA attacks the terminal position of the alkene to form the more stable intermediate IB, which is further oxidized by the high-valent photocatalyst to yield the carbocation intermediate IC. Finally, deprotonation of IC affords the target product. The carbocationic mechanism accounts for the systematically lower yields observed with aryl substrates bearing electron-withdrawing groups compared to those with electron-donating groups. This is likely because the electron-withdrawing effect hinders the oxidation of the benzyl radical intermediate by the oxidized photocatalyst. This conclusion is further corroborated by the inability of unactivated alkenes to undergo cycloaddition (Scheme b). For the photocatalyst-free [3 + 2] reaction pathway, radical trapping experiments have ruled out the possibility of direct homolysis of the pyridinium salt generating the carboxylate-α-carbon radical. The photoexcited pyridinium salt undergo intramolecular electron transfer to form the diradical intermediate II-D (Path A), and then rapidly convert into the lactone II-E, making it difficult to be trapped by TEMPO. , DFT free energy calculations support this hypothesis. While the formation of a Dewar zwitterion remains possible, the generation of the lactone is thermodynamically more favorable (Scheme b). On the other hand, the excited-state pyridinium salt may act as a photosensitizer to undergo energy transfer with the aryl alkene (Path B). The activated aryl alkene could then abstract a chlorine atom from dichloromethane to form intermediate II–C. This process is supported by quenching studies and radical trapping results. Furthermore, the photocatalyst-free [3 + 2] cycloaddition proceeds efficiently only when chlorinated alkane solvents are used, strongly indicating the involvement of a chlorine atom in the transformation (see SI, Table S1). The neutral oxygen center of lactone II-E attacks the intermediate II–C to obtain II–F.

Based on previous reports, we speculate that the pyridinium salt first decarboxylates thermally to yield the pyridinium ylide intermediate III-A in the thermally induced [3 + 2] cycloaddition (Scheme c). Triethylamine serves to suppress the direct conversion of this ylide into decarboxylated byproducts. Subsequently, the pyridinium ylide undergoes a 1,3-dipolar cycloaddition with the alkene to afford compound 5.

Conclusions

We have developed a unified pyridinium salt platform that integrates radical and ylide reactivities within a single scaffold. This structure can undergo photoinduced intermolecular or intramolecular electron transfer to form distinct intermediates, which subsequently engage with alkenes via divergent radical pathways to yield γ-butyrolactones. Alternatively, thermal activation unlocks a divergent pathway through decarboxylative ylide formation, yielding pentasubstituted indolizines via 1,3-dipolar cycloaddition. These intermediates further undergo an unprecedented photocatalytic 1,2-aryl migration to access fluorescent indolizines. This work establishes a new paradigm for designing multifunctional reagents, with inherent photo- and thermal-activation unlocking previously inaccessible synthetic pathways.

Glossary

Abbreviations

NAD

nicotinamide adenine dinucleotide

NADP

nicotinamide adenine dinucleotide phosphate

SET

single-electron transfer

EDA

electron donor–acceptor

IET

intramolecular electron transfer

PC

photocatalyst

TEMPO

2,2,6,6-tetramethyl-1-piperinedinyloxy

LSCs

luminescent solar concentrators

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01470.

• Experimental procedures; characterization data for all new compounds; NMR spectra; reaction condition screening; cyclic voltammetry (CV) for redox potential determination; radical trapping experiments; UV-Vis and fluorescence spectra (PDF)

∥.

Key Laboratory of Preclinical Study for New Drugs of Gansu Province, School of Basic Medical Sciences & Research Unit of Peptide Science, Lanzhou University, Lanzhou 730000, China

#.

F.S. and Y.K. contributed equally. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Shenzhen Medical Research Fund (D2403008, D2503007, B2502009), Guangdong Basic and Applied Basic Research Foundation (2025A1515011707), National Natural Science Foundation of China (22271317, 22101306), and Medical Innovation and Development Project of Lanzhou University (lzuyxcx-2022-156).

The authors declare no competing financial interest.