Visible-light-induced radical switching using benzothiazolyl sulfides for geminal carbon–carbon bond formation

Abstract

We propose a 2-benzothiazolylthio group as a novel auxiliary for radical switching. Using an acridinium-based photocatalyst, carbon radicals are generated from α-C–H bonds in sulfides, which are subsequently coupled with electron-deficient alkenes. This approach enables efficient access to γ,γ-disubstituted α-amino acid derivatives via two sequential photoredox reactions.

---

Carbon–heteroatom bond cleavage under visible light is a reliable method for generating carbon radicals and is generally achieved through photoredox catalysts or the formation of EDA complexes.¹ The functional groups used in this chemistry generate carbon radicals through photoinduced single-electron transfer (SET) followed by their elimination, and are incapable of generating alternative carbon radicals by eliminating neighbouring groups. On the other hand, some functional groups are known to assist in the generation of carbon radicals by breaking carbon–hydrogen bonds on adjacent carbon atoms. In this regard, ethers and amines are representative functionalities,² and a variety of carbon–carbon bonds have been formed at their α-carbons. Therefore, functional groups that are capable of precisely controlling the formation of these two types of carbon radical are expected to diversify derivatives from common precursors.

As a part of our efforts directed toward generating carbon radicals from organosulfur compounds,³ we reported the organo-photocatalytic desulfurative generation of carbon radicals from alkyl benzothiazolyl sulfides (Scheme 1A).^3a Recent relevant studies into alkyl tetrafluoropyridinyl⁴ and alkyl thiazolinium⁵ sulfides revealed that their carbon–sulfur bonds can be photocatalytically cleaved via reductive SET. Carbon radicals have also been generated from sulfides via indirect methods involving homolytic substitution by aryl radicals generated from haloarenes.⁶ Hence, the desulfurative generation of carbon radicals is becoming increasingly common. Meanwhile, carbon radicals have reportedly been generated at the α-carbons of sulfides under visible light on several occasions (Scheme 1B). Most such transformations involve modifying the α-carbon of a dialkyl sulfide (as represented by tetrahydrothiophene), which is achieved via hydrogen atom transfer⁷ or oxidative SET followed by proton transfer.^8,9

Scheme 1 Generating carbon radicals from sulfides under visible light.

Against this background, we hypothesised that a heteroaryl sulfide capable of facilitating the aforementioned two radical-generating processes will function as an auxiliary capable of switching between oxidative and reductive radical-generation pathways (Scheme 1C). While the 4-tetrafluoropyridinylthio and 2-mercaptothiazolinium groups are potential radical-switching functional-group candidates, there appear to be no reports to date that establish α-C–H scission chemistry involving these sulfides. Herein, we introduce the 2-benzothiazolylthio group as a new radical-switching auxiliary and demonstrate its utility through organo-photocatalytic sequential geminal C–C bond formation.

We began by investigating the reaction conditions, and focused on the work of Alfonzo et al.⁹ who demonstrated that sulfides can be α-C–H functionalised via dual organophotoredox and weak Brønsted-base catalysis. They reported the transformation of methyl benzothiazolyl sulfide in low yield as the only alkyl benzothiazolyl sulfide example. Accordingly, we first examined the reaction of ethyl derivative 1a with phenyl acrylate (2a) using their combination of 3,6-di-tert-butyl-9-mesityl-10-phenylacridinium tetrafluoroborate (Mes-^tBu-Acr⁺) as the acridinium catalyst and sodium trifluoroacetate. Contrary to expectations, the desired α-functionalised product 3a was not formed in a mixture of DCM and methanol as the solvent, although 1a was completely consumed. The use of water was found to be crucial for achieving an efficient transformation, with 3a produced in 71% yield in a DCM/water mixture (Table 1, entries 1–3). No reaction was observed when acetonitrile or toluene was used as the solvent (entry 4), while THF, 1,4-dioxane, and DMF delivered complex mixtures of products (for details, see Table S1 in the ESI†). The reaction did not proceed in the absence of the photoredox catalyst, and 9-mesityl-10-methylacridinium tetrafluoroborate (Mes-Acr⁺), as an alternative acridinium catalyst, failed to promote the reaction owing to its decomposition (entries 5 and 6). Other organophotoredox catalysts (for details, see Scheme S1, ESI†), even 2,4,6-triphenylpyrylium tetrafluoroborate (TPP), which has a higher oxidation potential ¹⁰ than Mes-^tBu-Acr⁺,¹¹ exhibited no catalytic activity for this reaction (entry 7). While Brønsted bases strongly promoted the reaction (for details, see Table S2, ESI†), sodium acetate inhibited product formation presumably due to a decarboxylative side-reaction (entries 8 and 9). Meanwhile, the addition of trifluoroacetic acid negatively affected the efficiency of the reaction, which suggests that deprotonation is involved in the reaction mechanism (entry 10). The reaction did not proceed in the absence of light (entry 11), and multiple by-products were obtained in the presence of oxygen (entry 12). It should be noted that the reaction temperature needs to be controlled, with a significantly lower product yield (37%) obtained at lower temperatures (entry 13).

Table 1 Optimising the α-C–H functionalisation of alkyl benzothiazolyl sulfide 1a

Entry	Variation from the standard conditions	Yield (%)
a The reaction was performed with 1a (0.300 mmol), 2a (0.100 mmol), Mes-^tBu-Acr^+ (5 mol%), and CF3CO2Na (0.100 mmol) in DCM/H2O (9/1, 1.0 mL) at 35–40 °C (bath temp.) under blue light. b Standard conditions in ref. 9b.
1	None^a	71
2	MeOH instead of H2O^b	0
3	No H2O, time 24 h	47
4	MeCN or toluene instead of DCM	0
5	No catalyst	0
6	Mes-Acr^+ instead of Mes-^tBu-Acr^+	0
7	TPP instead of ^tBu-Mes-Acr^+	0
8	No CF3CO2Na, time 24 h	27
9	NaOAc instead of CF3CO2Na, time 24 h	Trace
10	TFA instead of CF3CO2Na, time 24 h	<3
11	No light	0
12	Air instead of Ar	0
13	Temp. 30–35 °C	37

---

With the established reaction conditions in hand, we next explored the substrate scope using various alkenes (Scheme 2A). With the exception of the protocol leading to 3d, where the reaction ceased before complete conversion, all reactions were carried out until the acceptor had been completely consumed. Benzothiazolyl sulfide 1a was not only α-C–H functionalised with phenyl acrylate but also with other electron-deficient alkenes, including α-methylene-γ-butyrolactone, N-phenyl maleimide and phenyl vinyl sulfone, which afforded 3b, 3c and 3d in 78%, 35% and 32% yields, respectively. In the reaction leading to 3c, structurally unidentified by-products were also formed. Aryl- and nonconjugated alkenes, such as styrene, 1,1-diphenylethylene, and 1-heptene, were unreactive, with 1a recovered from the reaction mixture. Similarly, internal alkenes such as phenyl cinnamate, phenyl (E)-but-2-enoate, and cyclohexenone also showed no reactivity under the current conditions. Meanwhile, difunctionalised electron-deficient alkenes exhibited excellent reactivities, with 3e and 3f obtained in 98% and 88% yields, respectively. Methyl 2-acetamidoacrylate also proved to be a viable substrate and was converted into amino acid derivative 3g in 28% yield. Although the alkene was completely consumed and no by-products were detected by TLC analysis, the low yield of 3g is likely due to the formation of highly polar compounds resulting from the decomposition of the acetyl group, which could not be isolated. Amino-acid synthesis efficiency was improved through the use of an N-phthalimide or N-Boc₂ analogue as a radical acceptor, which afforded 3h and 3i in 88% and 99% yields, respectively. Furthermore, we successfully reduced the amounts of reagents to 1.2 equiv. (1a) and 0.2 equiv. (CF₃CO₂Na) for reactions involving highly reactive acceptors, which furnished 3b, 3e, 3f, 3h and 3i in yields of 73%, 93%, 95%, 99%, and 98%, respectively.

Scheme 2 Substrate scope for the α-C–H functionalisations of alkyl benzothiazolyl sulfides. Diastereomeric ratios were determined by ¹H NMR spectroscopy. The diastereomeric ratio of compound 3s could not be determined.

We next investigated the effect of the substituent on the sulfide under the improved reaction conditions used for the dehydroalanine derivative (Scheme 2B). Sulfides bearing alkyl substituents provided the corresponding adducts 3j–l in yields of 77–98% regardless of whether the carbon chain is linear or branched. Ester and carbamoyl groups were tolerated under the reaction conditions (3m: 99%, 3n: 76%, 3o: 94%). However, the carboxylic-acid-bearing substrate predominantly underwent decarboxylative addition (3p: 84%), which suggests that the poor result obtained in the presence of sodium acetate instead of CF₃CO₂Na (Table 1, entry 9) is attributable to the decarboxylative degradation of the base additive. Benzothiazolyl methyl sulfide not only afforded 3p but also overreacted product 3p′ in 85% combined yield (3p: 70%, 3p′: 15%). This unexpected overreaction was completely suppressed when the deuterated derivative was used, with 3p-d₂ obtained in 85% yield.¹² Furthermore, the method was applicable to sulfides derived from natural products such as menthol, geraniol, and dehydroepiandrosterone, affording the desired products 3q–s. In the case of the geraniol-derived substrate, a mixture containing an unidentified by-product was obtained; however, mass spectrometry, NMR, and IR analyses confirmed that the major component of the mixture corresponded to the expected product 3r.

We subsequently conducted several mechanistic studies (Scheme S2A–D, ESI†). Radical-trapping and radical-probe experiments confirmed the likelihood of a radical pathway. α-C–H functionalisation was inhibited by the addition of TEMPO. The product obtained from the reaction of 2-((cyclopropylmethyl)thio)benzo[d]thiazole with 2a was subjected to ¹H NMR spectroscopy and mass spectrometry, which revealed the formation of the ring-opening product, suggestive of the formation of a radical intermediate. A competitive kinetic isotope effect (KIE) experiment with equal amounts of benzothiazolyl methyl sulfide and its deuterated derivative revealed a KIE value of 3.0, with a parallel KIE experiment affording a value of 1.5. These results suggest that the rate-determining step may involve deprotonation. Light on/off experiments revealed that 3h was formed during light-on cycles, while the reaction was completely shut down during light-off cycle, which strongly suggests that the reaction is unlikely to proceed through a radical chain mechanism. Furthermore, absorption spectroscopy and fluorescence quenching experiments revealed that 1a (E_p/2 = +1.65 V vs. SCE) quenches the excited photocatalyst , consistent with an oxidative SET pathway (Fig. S3–S5, ESI†). Notably, the reaction proceeded with sulfides showing relatively lower potentials (1o–r, E_p/2 < +1.81 V vs. SCE), affording 3t–w in 6–38% yields, whereas no reaction was observed with more electron-deficient sulfides showing higher potentials (1s–u, E_p/2 > +2.28 V vs. SCE), further supporting the involvement of an oxidative SET pathway (Scheme S2E, ESI†). Accordingly, we propose a reaction mechanism involving oxidative single electron transfer followed by proton transfer, as shown in Scheme S3 (ESI†).

We finally demonstrated the synthetic utility of the developed sulfide-mediated radical-switching chemistry by synthesising γ,γ-disubstituted-α-amino acid derivatives via organo-photocatalytic sequential geminal C–C bond formation. Five alkyl benzothiazolyl sulfides bearing ethyl, isopropyl, cyclohexyl, tert-butoxycarbonyl, and di-Boc-aminomethyl groups were independently added to a dehydroalanine derivative (Scheme 3, step A), after which the obtained adducts were treated under the reaction conditions used to desulfinatively couple electron deficient alkenes^3a (step B). Overall, the two organo-photocatalytic reactions that use alkyl benzothiazolyl sulfides as linchpin reagents furnished ten functionalised amino acids in 35–91% yields over two steps (55 : 45–88 : 12 d.r.). The scalability of this sequential transformation was confirmed by reacting benzothiazolyl ethyl sulfide on a 1.2-mmol scale, which highlights the practical utility of the site-specific derivatisation process (4i: 78% over two steps).

Scheme 3 Syntheses of γ,γ-disubstituted-α-amino acid derivatives via organophotocatalytic sequential geminal C–C bond formation. Step A: methyl 2-(bis(tert-butoxycarbonyl)amino)acrylate (1.0 equiv.), 1 (3.0 equiv.), Mes-^tBu-Acr⁺ (5 mol%), CF₃CO₂Na (1.0 equiv.), DCM/H₂O (9/1), 35–40 °C (bath temp.), blue LEDs, 24 h. Step B: 3 (1.0 equiv.), alkene (1.5 equiv.), 2,4,5,6-tetrakis(diphenylamino)isophthalonitrile (4DPAIPN, 1 mol%), HEH-(OH)₂ (2.0 equiv.), K₂CO₃ (2.0 equiv.), DMSO, blue LEDs, 24 h.

In conclusion, we developed chemistry for the α-C–H functionalisations of alkyl benzothiazolyl sulfides. α-C–H scission in the sulfide was achieved using an acridinium catalyst and CF₃CO₂Na under blue light, thereby establishing the 2-benzothiazolylthio group as a radical-switching auxiliary. The utility of the alkyl benzothiazolyl sulfide as a linchpin reagent was demonstrated through the organo-photocatalytic sequential formation of geminal C–C bonds, which enabled the syntheses of γ,γ-disubstituted amino acid derivatives. This radical-switching protocol will be used to hybridise and functionalise bioactive molecules in the near future.

We thank Mr Reo Takahashi (Shizuoka University) for partially supporting this research. This work was financially supported by JSPS KAKENHI Grant number 22K05092 and the Shorai Foundation for Science and Technology. T. S. conceived and supervised the project. N. H. and K. S. developed the α-C–H functionalisation of alkyl benzothiazolyl sulfides. N. H. and H. H. synthesised γ,γ-disubstituted α-amino acid derivatives. N. H. investigated the reaction mechanism. T. S. and N. H. wrote the manuscript.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. S. Crespi and M. Fagnoni, Chem. Rev., 2020, 120, 9790–9833.
2. (a) I. Robb and J. A. Murphy, Org. Lett., 2024, 26, 2218–2222; (b) J. D. Bell, I. Robb and J. A. Murphy, Chem. Sci., 2022, 13, 12921–12926; (c) J. B. McManus, N. P. R. Onuska, M. S. Jeffreys, N. C. Goodwin and D. A. Nicewicz, Org. Lett., 2020, 22, 679–683; (d) S. Zhu, A. Das, L. Bui, H. Zhou, D. P. Curran and M. Rueping, J. Am. Chem. Soc., 2013, 135, 1823–1829; (e) 1 L. Ruiz Espelt, E. M. Wiensch and T. P. Yoon, J. Org. Chem., 2013, 78, 4107–4114.
3. (a) T. Sengoku, K. Matsune, T. Shimotori, N. Kikuchi, H. Hijikata, S. Nishioka and R. Takahashi, ChemCatChem, 2025, 17, e202401427; (b) T. Sengoku, H. Iwama, T. Shimotori, K. Fujimoto, T. Inuzuka, K. Matsune and H. Yoda, J. Org. Chem., 2023, 88, 12776–12782; (c) T. Sengoku, D. Ogawa, H. Iwama, T. Inuzuka and H. Yoda, Chem. Commun., 2021, 57, 9858–9861.
4. (a) B. A. van der Worp, M. D. Kosobokov, V. V. Levin and A. D. Dilman, Adv. Synth. Catal., 2021, 363, 1152–1158; (b) M. D. Kosobokov, M. O. Zubkov, V. V. Levin, V. A. Kokorekin and A. D. Dilman, Chem. Commun., 2020, 56, 9453–9456.
5. Y. Tian, X.-X. Fu, R.-M. Cheng, J. Feng, M.-H. Shen, C.-F. Zhu and H.-D. Xu, Eur. J. Org. Chem., 2024, e202400344.
6. (a) C. E. Flitcroft, K. A. Jolliffe and C. S. P. McErlean, Chem. – Eur. J., 2023, 29, e202301083; (b) L. Geniller, M. Taillefer, F. Jaroschik and A. Prieto, Adv. Synth. Catal., 2022, 364, 4249–4254; (c) Y. Wang, L.-F. Deng, X. Zhang, Z.-D. Mou and D. Niu, Angew. Chem., Int. Ed., 2021, 60, 2155–2159.
7. (a) A. Bisoyi, A. R. Tripathy, A. Behera and V. R. Yatham, J. Org. Chem., 2024, 89, 12540–12546; (b) D. Liu, A. Hazra, X. Liu, R. Maity, T. Tan and L. Luo, Angew. Chem., Int. Ed., 2024, 63, e202403186; (c) M. Wu, Z. Wu, H. T. Ang, B. Wang, T. Liu, S. Wu, Z. Lei, M. W. Liaw, S. Chanmungkalakul, C.-L. K. Lee, X. Liu, Y. Lu and J. Wu, ACS Catal., 2024, 14, 9364–9373; (d) Y. Li, S. Guo, Q.-H. Li and K. Zheng, Nat. Commun., 2023, 14, 6225.
8. (a) A. S. Anderton, O. J. Knowles, J. A. Rossi-Ashton and D. J. Procter, ACS Catal., 2024, 14, 2395–2401; (b) P. Kanti Roy, S. Okunaka, H. Tokudome and Y. Hitomi, Adv. Synth. Catal., 2023, 365, 4556–4561; (c) M. J. Rourke, C. T. Wang, C. R. Schull and K. A. Scheidt, ACS Catal., 2023, 13, 7987–7994; (d) O. J. Knowles, L. O. Johannissen, G. E. M. Crisenza, S. Hay, D. Leys and D. J. Procter, Angew. Chem., Int. Ed., 2022, 61, e202212158; (e) Z. Tan, S. Zhu, Y. Liu and X. Feng, Angew. Chem., Int. Ed., 2022, 61, e202203374; (f) J. Kim, B. X. Li, R. Y.-C. Huang, J. X. Qiao, W. R. Ewing and D. W. C. MacMillan, J. Am. Chem. Soc., 2020, 142, 21260–21266.
9. (a) E. Alfonzo and S. M. Hande, Org. Lett., 2021, 23, 6115–6120; (b) E. Alfonzo and S. M. Hande, ACS Catal., 2020, 10, 12590–12595.
10. E. Alfonzo, F. S. Alfonso and A. B. Beeler, Org. Lett., 2017, 19, 2989–2992.
11. A. Joshi-Pangu, F. Lévesque, H. G. Roth, S. F. Oliver, L.-C. Campeau, D. Nicewicz and D. A. DiRocco, J. Org. Chem., 2016, 81, 7244–7249.
12. Reactions with benzothiazolyl sulfides bearing phenyl or 2-pyridyl groups instead of alkyl groups were attempted under the current conditions, but no reaction was observed.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details and all relevant data (including NMR spectra and mechanistic-study data). See DOI: https://doi.org/10.1039/d5cc01998b

---