Photocatalytic streamlined dual-functional group transfer from cyanopyridine to internal alkynes

Abstract

The streamlined dual-functional group transfer (streamlined dual-FGT) strategy represents an efficient and sustainable approach for difunctionalization reactions, where all atoms or functional groups from the starting materials are fully incorporated into the final products without generating by-products. Pyridine and nitrile functionalities are prevalent and highly valued structural motifs found in a myriad of natural products, pharmaceuticals, agrochemicals, and polymers. The simultaneous incorporation of these groups via the streamlined dual-FGT strategy is thus of considerable significance in synthetic chemistry. Herein, we report a regioselective pyridylcyanation of internal alkynes enabled by an oxalate-based photocatalytic system, employing cyanopyridine as a streamlined dual-functional group transfer reagent. Mechanistic investigations using time-resolved spectroscopy reveal that the transformation proceeds through a photoinduced regioselective radical addition of the persistent cyanopyridine radical anion to alkynes, followed by the cooperative release and re-addition of the cyanide ion (CN⁻).

The streamlined dual-functional group transfer strategy represents an efficient and sustainable approach for difunctionalization reactions, where all atoms or functional groups from the starting materials are fully incorporated into the final products. Herein, the authors report a regioselective pyridylcyanation of internal alkynes enabled by an oxalate-based photocatalytic system, employing cyanopyridine as a streamlined dual-functional group transfer reagent.

Introduction

In the realm of modern organic synthesis, the development of highly selective, step- and atom-economical, and environmentally sustainable transformations remains a central objective^1,2. Functional group transfer reactions have emerged as powerful tools, enabling the direct transfer of specific atoms or functional groups from one molecule to another^3–7. The difunctionalization of alkenes^8–12 and alkynes^13–15 has proven particularly valuable, offering an efficient approach to introduce two functional groups in a single step, thereby enhancing molecular complexity. Among these strategies, three-component difunctionalization reactions employing two mono-functional group transfer reagents (mono-FGTRs) are well-established, primarily due to the broad availability and structural diversity of these reagents. However, their utility is often limited by suboptimal atom economy and the generation of considerable by-products^16–18. Recent years have witnessed a growing interest in the development of bridged dual-functional group transfer reagents (bridged dual-FGTRs), aiming to enhance reaction efficiency and atom economy by transferring two functional groups from a single reagent. Nevertheless, their broader application remains challenging due to limited reagent accessibility and by-products formation^19–25. In contrast, difunctionalization reactions employing streamlined dual-functional group transfer reagents (streamlined dual-FGTRs) achieve 100% atom economy with no by-products formation^26,27, offering a more sustainable alternative that aligns closely with the principles of green chemistry (Fig. 1a).

Fig. 1. Project overview.

a Strategies for difunctionalization reactions. b Cyanopyridine as a FGTR in photocatalysis. FG Functional Group, FGTR Functional Group Transfer Reagent.

Pyridine^28–30 and nitrile^31,32 groups are prevalent structural motifs in pharmaceuticals, natural products, agrochemicals, and polymers. In medicinal chemistry, pyridine derivatives exhibit various important pharmacological activities such as antimicrobial, antimalarial, antidiabetic, and anti-inflammatory^28,33, while the nitrile group is an excellent ligand for protein binding due to its high polarity, small size, linear geometry, metabolic stability, and capacity to accept hydrogen bonds³⁴. Both groups also exhibit remarkable synthetic versatility: pyridines can be readily reduced to piperidines³⁵, while nitriles can be transformed into a range of functional groups such as amines, amides, carboxylic acids/esters, aldehydes, and ketones^36,37. Consequently, the development of efficient methods for the simultaneous incorporation of both pyridine and nitrile groups into organic molecules is of significant value. Recent advances in photochemistry have established cyanopyridine as a practical mono-FGTR for the synthesis of pyridine derivatives (Fig. 1b). Pioneering work by MacMillan and co-workers demonstrated its utility in photocatalytic direct α-amino and allylic C–H pyridination^38,39. It has also been adopted in three-component 1,2-difunctionalization of alkenes, enabling the rapid assembly of structurally diverse pyridinated products^40–46. In such reactions, cyanopyridine typically undergoes single-electron reduction to form a persistent radical anion, which then couples with the in situ transient radical intermediate, allowing for the incorporation of the pyridine moiety while liberating the cyano group as a cyanide by-product^47–49. However, the direct introduction of nitrile groups remains challenging due to the hazardous nature of conventional cyanide sources, such as KCN, NaCN, CuCN, Zn(CN)₂, and K₄[Fe(CN)₆], which often release toxic hydrogen cyanide (HCN) during reactions. Therefore, strategies that enable the capture and incorporation of the cyano by-product into the final product hold significant promise. Such approaches could effectively transform cyanopyridine into a dual-FGTR, offering a more efficient, greener, and safer route for constructing complex molecules containing both pyridine and nitrile functionalities.

Alkynes are highly versatile building blocks in organic synthesis, and their reductive difunctionalization offers a powerful and cost-efficient approach to increasing molecular complexity. Building on our recent studies on the photocatalytic reductive hydroalkylation, arylalkenylation, and hydrocarboxylation of aryl alkynes via alkyne radical anions generated through an oxalate-based photocatalytic system^50,51, we aimed to develop a difunctionalization reaction involving cyanopyridine through a streamlined dual-FGT strategy. However, several challenges remain: (1) The inherently low reactivity of alkynes makes the one-step construction of two vicinal C−C bonds without transition-metal catalysis particularly challenging^52–54; (2) Achieving high regioselectivity in the difunctionalization of asymmetric alkynes remains a significant hurdle. Herein, we report the use of commercially available cyanopyridine as a streamlined dual-FGTR, enabling the regioselective pyridylcyanation of internal alkynes under visible-light irradiation (Fig. 1b). This method enables the direct construction of two vicinal C−C bonds on internal alkynes, providing an efficient strategy for the rapid assembly of pharmaceutical motifs and value-added products.

Results and discussion

Optimization of the reaction conditions

Inspired by Wen’s work, methyl phenylpropiolate 1a featuring an aryl and electron-withdrawing group (EWG) was employed as the model alkyne substrate⁵⁵. The proposed streamlined dual-FGT with 4-cyanopyridine 2a was investigated under 30 W blue LED irradiation at room temperature (Table 1). Following extensive optimization, the pyridylcyanation product 3aa was obtained in a 91% isolated yield with a 1.17:1 diastereomeric ratio (d.r.) using 2,4,5,6-tetrakis(diphenylamino)-1,3-dicyanobenzene (4DPAIPN) as a photocatalyst, in the presence of oxalic acid (H₂C₂O₄) and 1,1,3,3-tetramethylguanidine (TMG) in DMSO (Table 1, entry 1). The structure of 3aa was unambiguously confirmed through X-ray crystallographic analysis. Unlike conventional transition-metal catalysis^56–59, our method overrides the inherent electronic bias, selectively installing the pyridyl group at the α-position of methyl phenylpropiolate 1a, leading to the highly regioselective product 3aa⁵⁵. Comparative evaluation of various photocatalysts, including [Ru(bpy)₃](PF₆)₂, [Ir(ppy)₂(dtbbpy)](PF₆), [Ir(dF(CF₃)ppy)₂(dtbbpy)](PF₆), and 2,4,5,6-tetrakis(carbazol-9-yl)-1,3-dicyanobenzene (4CzIPN), revealed significantly decreased yields of 3aa (entries 2−5), likely due to the less favorable reduction potentials of their corresponding radical anions compared to 4DPAIPN^•−, which facilitates the key single-electron transfer step⁵¹. Solvent screening showed that N,N-dimethylformamide (DMF) failed to initiate the reaction, while acetonitrile (MeCN) produced 3aa in a modest 49% yield (entries 6 and 7). Replacing the H₂C₂O₄/TMG mixture with Na₂C₂O₄ or K₂C₂O₄ did not yield 3aa (entry 8), whereas tetrabutylammonium oxalate ((ⁿBu₄N)₂C₂O₄) provided 3aa in 56% yield (entry 9). These outcomes are likely due to the distinct solubility profiles of these oxalates in DMSO. Further evaluation of alternative reductants revealed that neither conventional sacrificial amine-based systems (e.g., Et₃N or N,N-diisopropylethylamine (DIPEA)) nor substitution of oxalic acid with formic acid could produce 3aa (entries 10 and 11). These observations demonstrated the indispensable role of CO₂ radical anion (CO₂^•−), generated through single-electron transfer (SET) oxidation of oxalate, in facilitating this transformation. Moreover, we tested green LED (λ_max = 525 nm) and red LED (λ_max = 640 nm) light source under standard reaction conditions. Green light irradiation yielded 3aa in 47% yield, while red light failed to initiate the reaction (entries 12 and 13). This aligns with 4DPAIPN’s UV–vis absorption spectrum showing significant absorption in the blue light region, a weaker tail in the green light region, and negligible absorption in the red light region (Supplementary Fig. 6). Control experiments confirmed that no reaction occurred in the absence of light or photocatalyst (entries 14 and 15), underscoring their essential roles. Additionally, a sensitivity assessment based on reaction condition variations, commonly conducted to verify reproducibility⁶⁰, demonstrated that the transformation is oxygen-sensitive but generally robust to other perturbations (Supplementary Tables 1 and 2).

Table 1.

Optimization of the reaction conditions^a

Entry	Variations from standard conditions	Yield of 3aa (%)b
1	none	94 (91)c
2	[Ru(bpy)3](PF6)2 instead of 4DPAIPN	36
3	[Ir(ppy)2(dtbbpy)](PF6) instead of 4DPAIPN	78
4	[Ir(dF(CF3)ppy)2(dtbbpy)](PF6) instead of 4DPAIPN	54
5	4CzIPN instead of 4DPAIPN	27
6	DMF instead of DMSO	trace
7	MeCN instead of DMSO	49
8	Na2C2O4 or K2C2O4 instead of H2C2O4/TMG	trace
9	(nBu4N)2C2O4 instead of H2C2O4/TMG	56
10	Et3N or DIPEA instead of H2C2O4/TMG	0
11	HCO2H instead of H2C2O4	0
12	green LEDs (λ = 525 nm) instead of blue LEDs	47
13	red LEDs (λ = 640 nm) instead of blue LEDs	0
14	heat to 80 °C in the dark	0
15	no 4DPAIPN	0

LED light-emitting diode, TMG 1,1,3,3-tetramethylguanidine, DIPEAN,N-diisopropylethylamine.

^astandard conditions: 1a (0.2 mmol), 2a (0.5 mmol), H₂C₂O₄ (0.5 mmol), TMG (0.5 mmol), 4DPAIPN (1 mol%), DMSO (2 mL), argon atmosphere, 30 W blue LEDs (λ_max = 456 nm), rt, 90 min.

^byield was determined by ¹H NMR spectroscopy analysis with diphenylacetonitrile (DPA) as an internal standard.

^cisolated yield in parentheses with a 1.17:1 diastereomeric ratio (d.r.).

Substrate scope

With optimal reaction conditions established, the substrate scope was subsequently investigated. The yields depicted in Fig. 2 are isolated yields, with the d.r. varying between 1:1 and 2:1 (see the characterization of products in Supplementary Information for details). The scope of internal alkynes 1 was systematically evaluated first. To our delight, a wide range of ethyl 3-arylpropiolates 1 were viable substrates, delivering the desired products in good to excellent yields. Both electron-donating (e.g., alkyl, alkoxyl, and phenyl) and electron-withdrawing (e.g., fluoro, chloro, alkynyl, trifluoromethyl, and trifluoromethoxyl) substituents on the benzene moiety (1b–1j) were tolerated without a substantial impact on the reaction efficiency, affording the corresponding products in high yields (71–87%). Notably, our protocol exhibited remarkable chem- and regio-selectivity for arylpropiolate moiety, leading to the desired products 3ea and 3fa in 73% and 76% yield, respectively, while preserving the triple bonds of terminal alkyne (3ea) and diphenylacetylene (3fa) for subsequent transformations. These results demonstrated that both the aryl and ester groups on the alkynes were critical for driving the reaction with high regioselectivity. In addition, sterically demanding disubstituted and cyclic phenylpropiolate substrates with heteroatom substituents (1k–1n) were viable under standard conditions, affording the corresponding products in high yields. Other (hetero)aryl groups, including naphthyl, thiophene, and pyridyl (1o−1t), also reacted smoothly, albeit with slightly lower yields compared to phenylpropiolate substrates.

Fig. 2. Substrate scope for the streamlined dual-FGT reaction.

Reaction conditions: 1 (0.2 mmol), 2 (0.5 mmol), H₂C₂O₄ (0.5 mmol), TMG (0.5 mmol), 4DPAIPN (1 mol%), DMSO (2 mL), argon atmosphere, 30 W blue LEDs (λ_max = 456 nm), rt, 90 min. Isolated yields with a diastereomeric ratio (d.r.) ranging from 1:1 to 2:1. The ratio was determined by analysis of the crude ¹H NMR spectra. Refer to the characterization of products in Supplementary Information for details.

Moreover, the reaction was successfully applied to a series of phenylpropiolate esters bearing different functionalities such as phenyl (3a’a), pyridine (3b’a), pyrazole (3c’a), thiophene (3d’a and 3e’a), ether, chloride (3f’a), silane (3g’a), terminal alkyne (3h’a), as well as ketal (3i’a). This protocol not only tolerated a wide range of functional groups but also retained stereocenters without racemization. For instance, the complex chiral substrate 1i’ was compatible with our mild reaction conditions, yielding product 3i’a in 75% yield. The scope of cyanopyridine 2 was also evaluated using methyl phenylpropiolate 1a as the alkyne substrate. As shown in Fig. 2, cyanopyridine substrates with diverse alkyl substituent groups, such as methyl (3ab, 3ac, and 3aj), benzyl (3ad), and sterically hindered tert-butyl (3ae), gave the desired products in high yields. Additionally, radical-sensitive strained cyclopropyl and cyclobutyl groups were tolerated under our reaction conditions (3af and 3ag). Good yields of products were also obtained from the corresponding substrates possessing different functional motifs, such as cyclopentene (2h) and 1,1-difluorocyclohexane (2i). Unfortunately, neither 2-cyanopyridine (2k) nor 3-cyanopyridine (2l) was effective in our reaction system due to their more negative reduction potentials: 2-cyanopyridine (E_p/2(2k/2k^•−) = −1.87 V vs SCE) and 3-cyanopyridine (E_p/2(2 l/2 l^•−) = −1.92 V vs SCE), compared to that of 4-cyanopyridine (E_p/2(2a/2a^•−) = −1.70 V vs SCE) (Supplementary Fig. 5). This led to preferential reduction of alkyne 1a (E_p/2(1a/1a^•−) = −1.78 V vs SCE) instead of 2k or 2l, preventing the crucial radical addition step and ultimately resulting in a complex reaction mixture. Overall, the broad applicability of both arylpropiolate esters 1 and 4-cyanopyridines 2, bearing diverse substitution patterns and functional groups, underscores the potential of our methodology for late-stage functionalization of pharmaceuticals and natural products.

Flow synthesis and product transformations

To demonstrate the synthetic utility of this method, a gram-scale reaction of 1a with 2a was performed using an in-house circulating-flow reactor^51,61, yielding 1.97 g of product 3aa in 87% yield (Fig. 3a). Compared to the batch reaction, our circulating-flow synthesis not only resolved the issues of low light penetration resulting from the poor solubility of oxalates but also significantly enhanced efficiency by cutting the reaction time to just 60 min. The usage of the nitrile and pyridine motifs were demonstrated by several transformations of 3aa, as shown in Fig. 3b. Treatment of 3aa with NaBH₄ in the presence of CoCl₂ led to γ-lactam 5 through chemo-selective reduction of the nitrile group followed by lactamization⁶². Reduction of the ester group of 3aa with LiBH₄, followed by lactonization under HCl/MeOH conditions, furnished lactone 6⁶³. Alternatively, 3aa was reduced by LiAlH₄ to form a 1,4-amino alcohol, which was then treated with triphosgene to afford the oxazolidinone 7^64,65. In addition to these reductive transformations, oxidation of 3aa with DDQ efficiently afforded the all-carbon tetrasubstituted alkene 8 with exclusive E-selectivity⁶⁶. These efficient constructions of pharmaceutical motifs from the product 3aa highlights the rich potential of this protocol in the preparation of bioactive molecules.

Fig. 3. Gram-scale synthesis and product transformations.

a Gram-scale reaction using a circulating-flow reactor. b Further transformations of the product 3aa.

Mechanistic studies

To gain insight into the reaction mechanisms, several control experiments were performed. A light on/off experiment showed that continuous exposure to visible-light was essential for the reaction to occur (Supplementary Fig. 4). Radical trapping experiments with common radical scavengers, such as 2,2,6,6-tetramethyl-piperidine-1-oxyl (TEMPO) and butylated hydroxytoluene (BHT), yielded no desired products, indicating the involvement of radical intermediates in the reaction (see the radical inhibition experiments in Supplementary Information for details). To further elucidate the underlying photocatalytic mechanisms, we recorded time-resolved spectra in conjunction with steady-state spectral analysis (Fig. 4 and Supplementary Figs. 6–12) for photocatalyst 4DPAIPN (PC), both alone and in mixtures with the substrates (1a, 2a, H₂C₂O₄, and/or C₂O₄²⁻) under photoexcitation.

Fig. 4. Reaction mechanism study with time-resolved transient absorption spectroscopy.

a Steady-state emission spectra for ¹PC^* at difference concentrations of C₂O₄²⁻. b Transient absorption spectra for ³PC^*. c Decay traces at 450 nm for ³PC^* at different concentrations of C₂O₄²⁻. d Decay traces at 530 nm for ³PC^* alone, and ³PC^* in the presence of 100 mM 1a, 250 mM 2a, or 250 mM H₂C₂O₄. e Stern–Volmer plots for ¹PC^* + C₂O₄²⁻ (blue) and ³PC^* + C₂O₄²⁻ (red). f Transient absorption spectra for PC^•− + 250 mM 2a. g Decay traces at 520 nm for PC^•− at different concentrations of 2a. h Stern–Volmer plots for PC^•−+ 1a (red) and PC^•− + 2a (blue). i Transient absorption spectra for CO₂^•− + 5 mM 1a. j Kinetics traces at 420 nm for CO₂^•− + 1a at different concentrations. k Kinetics traces at 420 nm for CO₂^•− + 5 mM 1a at different concentrations of 2a. l Stern–Volmer plots for CO₂^•− + 1a (red) and CO₂^•− + 2a (blue).

Electron transfer quenching studies of ¹PC*/³PC* by C₂O₄²⁻

The steady-state fluorescence quenching experiments reveal that the singlet excited state of PC (¹PC*) can be efficiently quenched by C₂O₄²⁻. From the Stern–Volmer plot (Fig. 4a, e) quenching rate constant (k_q) of 5.4 × 10⁹M⁻¹ s⁻¹ is derived, indicating that the quenching of ¹PC* by C₂O₄²⁻ is highly efficient and approaches diffusion-controlled limits. Furthermore, ns time-resolved spectroscopy demonstrates that the triplet state of PC (³PC*) is also efficiently quenched by C₂O₄²⁻ with a rate constant of 3.4 × 10⁸M⁻¹ s⁻¹ (Fig. 4b, c, and e). In contrast, the presence of 1a, 2a, and/or H₂C₂O₄ neither affects the emission intensity of ¹PC* nor the evolution of ³PC* (Fig. 4d and Supplementary Fig. 8), indicating that these substrates do not quench ¹PC* nor ³PC*. These results demonstrate that the photocatalytic cycle is initiated by the quenching of both ¹PC* and ³PC* by C₂O₄²⁻. Additionally, ns-TA spectra reveal that the reaction between ¹PC*/³PC* and C₂O₄²⁻ efficiently generates PC^•−. This radical anion is characterized by two positive bands around 420 nm and around 520 nm, along with a broad, featureless absorption extending up to 750 nm (Fig. 4f and Supplementary Fig. 9). The formation of PC^•− provides direct evidence for an electron transfer quenching mechanism, thermodynamically favorable as confirmed by ΔG⁰ calculations (Supplementary Table 4).

Monitoring the quenching of PC^•− by 1a and 2a

The electron transfer from C₂O₄²⁻ to ¹PC*/³PC* results in the formation of PC^•− and C₂O₄^•−. The open-shell C₂O₄^•− intermediate is expected to rapidly extrude CO₂, producing the highly reductive CO₂ radical anion (CO₂^•−) (E_1/2(CO₂/ CO₂^•−) = −2.22 V vs SCE)⁶⁷. After elucidating that the photocatalytic cycle begins with electron transfer from C₂O₄²⁻ to ¹PC*/³PC* and subsequently identifying PC^•− as the observed spectral species, we explored the possible reactions for PC^•− + 1a and PC^•− + 2a, respectively. As depicted in Supplementary Fig. 10, PC^•− remains stable on the millisecond timescale. In contrast, upon adding 2a, the decay of PC^•− was significantly accelerated, exhibiting a mono-exponential behavior (Fig. 4f,g). Linear fitting of the measured pseudo-first-order rate constants against 2a concentration yields a second-order rate constant of 7.4 × 10⁵M⁻¹ s⁻¹ (Fig. 4h).

The complete quenching of PC^•− by 2a along with the complete recovery of ground-state bleach of PC at 470 nm (Fig. 4f) suggests that the quenching corresponds to the regeneration of PC via electron transfer from PC^•− to 2a. The TD-DFT calculated spectrum for the reduction species of 2a (2a^•−) exhibits minimal absorption in the detected spectral region, accounting for the transient spectral evolution back to zero baseline (Supplementary Fig. 12). Note that the reported reduction potentials for PC (E_p/2(PC/PC^•−) = −1.65 V vs SCE)⁶⁸ and 2a (E_p/2(2a/2a^•−) = −1.70 V vs SCE). The comparable reduction potential indicates that the driving force for the electron transfer from PC^•− to 2a is weak, which accounts for the relatively modest rate of 7.4 × 10⁵M⁻¹ s⁻¹ for the reaction of 2a with PC^•−.

The coexisting CO₂^•− in the solution also reacts with 1a (as will be discussed later), thereby complicating the analysis of PC^•− + 1a reaction. To clarify the possible reaction between PC^•− + 1a, we designed a control experiment in which PC^•− is generated by reducing ¹PC*/³PC* with DIPEA and then quenched by 2a or 1a (Supplementary Fig. 11). In this way, a second-order rate constant of 5.8 × 10⁵ M⁻¹ s⁻¹ for the reaction PC^•−+ 2a was obtained, consistent with the value of 7.4 × 10⁵ M⁻¹ s⁻¹ measured in PC + C₂O₄²⁻ + 2a system (Fig. 4h). Furthermore, it was found that PC^•− can also be quenched by 1a via an electron transfer mechanism with a second-order rate constant of 1.1 × 10⁵ M⁻¹ s⁻¹ (Fig. 4h and Supplementary Fig. 11).

Monitoring the reduction of 1a and 2a with CO₂^•−

As shown in Fig. 4i, introducing additional 1a (5 mM) to the PC/C₂O₄²⁻ system significantly changes the transient spectra. Within 15 μs, the intensity of the negative band around 470 nm decreases, while the intensity of the positive band around 420 nm increase, together with the buildup of a broad band around 620 nm. These changes correspond to the reaction between PC^•−/CO₂^•− + 1a. Given the rate of PC^•−+ 1a (1.1 × 10⁵ M⁻¹ s⁻¹) and the added concentration of 1a (5 mM), the reaction of PC^•− + 1a is expected to occur on the millisecond timescale, ruling out its involvement in the initial 15 μs spectral evolution. Thus, the observed changes are attributed to the reaction CO₂^•− + 1a through electron transfer, forming reduced 1a (1a^•−) and CO₂. TD-DFT calculation indicates that 1a^•− exhibits a strong, broad absorption from 400 to 800 nm, with a peak around 515 nm (Supplementary Fig. 12), supporting the assignment of spectral evolution within 15 μs to 1a^•− formation. Hence, the spectrum at 15 μs reflects the combined contributions of the positive absorption signals of PC^•− and 1a^•−, along with the ground-state bleaching of PC. By monitoring 1a^•− formation at 420 nm, a second-order rate constant for CO₂^•− + 1a of 2.8 × 10⁷ M⁻¹ s⁻¹ is determined through linear fitting of the measured pseudo-first-order rate constants versus 1a concentration (Fig. 4j, l). In addition, 2a is also anticipated to quench the highly reductive CO₂^•− through electron transfer. As shown in Fig. 4k, the kinetic curves for the formation of 1a^•− show that, with the increasing of added 2a concentration, the amount of formed 1a^•− as reflected by the plateau gradually decreases, while the overall formation process accelerates. This confirms that 2a can effectively compete with 1a for the quenching reaction with CO₂^•−. By applying the differential rate equation for parallel reactions to analyze changes in 420 nm kinetics with varying concentrations of 2a, we obtained a second-order rate constant of 3.5 × 10⁸ M⁻¹ s⁻¹ for CO₂^•− + 2a (Fig. 4k, l, and Supplementary Fig. 13), which is more competitive than CO₂^•− + 1a.

Proposed reaction mechanism

Based on kinetic data (Fig. 4h, l), the quenching of PC^•− and CO₂^•− by 2a is predominant over that by 1a owing to the significantly higher rate constants of 2a compared to 1a (5–7 times higher for PC^•− and 10 times higher for CO₂^•−) and the 2.5-fold higher concentration of 2a in the reaction system. Thus, the decay of PC^•− and CO₂^•− primarily corresponds to the formation of 2a^•−, while much less amount of 1a^•− is produced. Without significant formation of 1a^•−, the potential radical coupling of 2a^•− with 1a^•− to form the product 3aa, is unlikely to be the dominant pathway. Furthermore, the calculated energies of the two products from the radical coupling process do not support the experimentally observed regioselectivity, which favors α-pyridylation over the β-pyridylation product (Supplementary Fig. 14). This apparent contradiction further reinforces the conclusion that the radical coupling mechanism shown in Supplementary Fig. 14 is not the primary pathway. On the other hand, the abundant formation of 2a^•− most likely triggers a radical addition process as illustrated in Fig. 5. Irradiation of PC with blue LEDs generates the excited ¹PC*/³PC*, followed by a single electron transfer (SET) process with C₂O₄²⁻ to yield PC^•− and CO₂^•−, alongside the release of CO₂. Both CO₂^•− and PC^•− are quenched by 2a to yield 2a^•−, with PC^•− returning to PC, completing the catalytic cycle. Subsequently, 2a^•− undergoes regioselective radical addition with 1a to yield α-pyridylation intermediate I. This is supported by DFT calculations, which demonstrate that the pathway leading to the α-pyridylation intermediate I is more favorable than that leading to the β-pyridylation intermediate II, both in terms of the stability of the reaction intermediates and the reaction kinetic barrier. This regioselectivity aligns well with prior literature precedent, where the crucial radical additions to arylpropiolates proceeded through α-position attack to form more stabilized benzylic-type radical intermediates^69–71.

Fig. 5. Plausible photocatalytic mechanisms.

A plausible reaction mechanism for the photocatalytic streamlined dual-functional group transfer.

Given that the quenching efficiency of PC^•− by the substrates 1a and 2a is much lower than that of CO₂^•−, and considering the continuous irradiation of the system, PC^•− is expected to be abundant in the reaction mixture. This allows the vinyl radical intermediate III, generated from the α-pyridylation intermediate I by the release of the cyanide ion (CN⁻), to accept an electron from PC^•−, producing the alkenyl anion intermediate IV and regenerating PC. Theoretical calculations confirm the thermodynamic feasibility of this SET process (Supplementary Fig. 16 and Table 5), while deuterium-labeling experiments using D₂C₂O₄ and/or D₂O provide indirect support through the significant deuterium incorporation (up to 85%) at the β-position of the ester in 3aa (see the deuterium labeling experiments in Supplementary Information for details)^72,73. Following protonation of intermediate IV, a facile Michael addition of the CN⁻ to intermediate V and subsequent protonation yield the dual-FGT product 3aa.

Discussion

In summary, we have developed a streamlined dual-FGT reaction under visible-light irradiation, enabling the complete incorporation of cyanopyridine into internal alkynes with excellent regioselectivity. Mechanistic studies using time-resolved spectroscopy revealed that the reaction proceeds via a photoinduced regioselective radical addition of the persistent cyanopyridine radical anion to alkynes, followed by a cooperative release and re-addition of the cyanide ion (CN⁻) to furnish the pyridylcyanated products. This protocol operates under mild conditions, avoids the use of expensive transition metal catalysts, and employs cyanopyridine as a clean and safe cyanation reagent. Its excellent step- and atom-economy, along with broad functional group tolerance, make it highly promising for late-stage functionalization of bioactive molecules. Moreover, the successful demonstration of large-scale circulating-flow synthesis and subsequent product transformation underscores the practicality of this method for the efficient production of value-added pharmaceutical molecules. Beyond its synthetic utility, this work also highlights the power of time-resolved spectroscopy in elucidating photochemical reaction mechanisms and is expected to inspire further advancements in the development of green and sustainable dual-FGT reactions.

Methods

General procedure D

To a 10 mL microwave vial equipped with a magnetic stir bar, arylpropiolate ester 1 (0.2 mmol, 1.0 equiv.), cyanopyridine 2 (0.5 mmol, 2.5 equiv.), oxalic acid (0.5 mmol, 2.5 equiv.), TMG (0.5 mmol, 2.5 equiv.), 4DPAIPN (0.002 mmol, 0.01 equiv.) and DMSO (2.0 mL) were sequentially added. The reaction mixture was degassed three times via freeze-pump-thaw. Then the reactor was placed at a distance (app. 5 cm) from two 30 W 456 nm light-emitting diode (LED) lamps, stirred and irradiated at room temperature with a fan for 90 min under argon atmosphere. After completion of the reaction (by TLC analysis), it was quenched with deionized water (10 mL), and extracted with EtOAc (3 × 10 mL). The combined organic layers were washed with brine (20 mL), dried over Na₂SO₄, filtered, and concentrated by rotary evaporation. The residue was purified by column chromatography using hexane/EtOAc/Et₃N (10:1:0.05 to 5:1:0.05) to afford the desired pyridylcyanated product 3.

Author contributions

J.W. and H.S. supervised the project and coordinated the collaboration. X.T. conceived the research and performed the experiments. J.J. and Y.L. conducted mechanistic investigations using time-resolved spectroscopy and theoretical calculations. H.Y., G.W., and D.T. participated in starting material preparation and flow analysis. M.Y. carried out HRMS measurements and data analysis. X.T., J.J., H.T.A., H.S., and J.W. wrote the manuscript. All authors contributed to the discussion of the manuscript.

Peer review

Peer review information

Nature Communications thanks Ahmad Masarwa, Peng Yang and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The X-ray crystallographic coordinates for structure reported in this article have been deposited at the Cambridge Crystallographic Data Center (CCDC), under deposition numbers CCDC 2414276. These data can be obtained free of charge from the CCDC via www.ccdc.cam.ac.uk/data_request/cif. All data to support the conclusions are available in the main text or the Supplementary Information. The data that support the findings of this study are available from the corresponding authors upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-64029-2.