Purines are important structural motifs in biological and pharmaceutical chemistry because of their antivirus and antitumor activities [1-6]. Purine derivatives with the aryl group on the N9 position (Fig. 1) are potential anxiolytic and antidepressant agents [7-10], potential antituberculosis drugs [11] and enterovirus inhibitors [12]. Thus, the development of green and novel synthetic methodologies for the straightforward construction of structurally diverse 9-arylpurines is currently of great importance in the pharmaceutical chemistry.

Traditionally, there are three general routes for the direct N9-arylation of purines using activated arenes via a C–N cross-coupling reaction (Scheme 1). Arylboronic acids, diaryliodonium salts and aryl halides were used to achieve the arylation of various purine derivatives [13-15]. In 2003, Bakkestuen and Gundersen developed a regioselective purine N9-arylation reaction under mild conditions employing arylboronic acid in the presence of Cu(OAc)₂ and phenanthroline (Phen) (Scheme 1a) [16]. In 2011, Niu and co-workers reported an efficient and novel protocol of copper-catalyzed arylation using diaryliodonium salts as the aryl source (Scheme 1b) [17]. After that, Larsen and Ulven described a CuBr and 4, 7-bis(2-hydroxyethylamino)-1, 10-phenanthroline (BHPhen) co-catalyzed arylation approach of purine skeletons with aryl halides (Scheme 1c) [18]. Despite these distinguished advances, some of these synthetic strategies suffer from certain limitations such as long reaction times, high reaction temperatures and/or harsh reaction conditions. Consequently, the development of an eco-friendly and novel synthetic protocol for the preparation of structurally diverse 9-arylpurine derivatives using simple arylation reagents is highly desirable.

The recent development of visible light photocatalysis provides a mild and powerful tool for organic synthesis [19-24]. Romero and co-workers introduced aromatic C(sp²)–H amination via an acridinium photoredox catalyst with 2, 2, 6, 6-tetramethylpiperidine-1-oxyl (TEMPO) as a co-catalyst [25]. Later, Niu and co-workers reported an oxidant-free strategy for C(sp²)–N bond formation using a dual catalytic system combining an acridinium photooxidant with a cobalt complex [26]. Herein, we report a simple and mild metal-free 9-arylpurine synthesis via C(sp²)–H activation of the arene induced by visible light using air as the oxidant (Scheme 1d).

Initially, we optimized the conditions using 6-chloropurine (1a) and mesitylene (2a) as model substrates, and the results are summarized in Table 1. At first, we chose commercially available organic dyes for the desired transformation. Using methylene blue or Eosin Y did not provide the desired product (Table 1, entries 1 and 2). Gratifyingly, the use of 9-mesityl-10-methylacridinium perchlorate (Acr⁺–Mes ClO₄^–) gave the desired product 3a in 92% yield (Table 1, entry 3). 9-Mesityl-10-phenylacridinium tetrafluoroborate (Mes–Acr⁺–Ph BF₄^–) was also tried, but it was less effective (Table 1, entry 4). Different commonly used photocatalysts were then screened (Table 1, entries 5–7). To our delight, we found that 2, 4, 6-triphenylpyrylium tetrafluoroborate (TPT) was an efficient photocatalyst providing our C–N coupled product 3a in 92% yield as well. However, because of the relatively low stability of TPT [27, 28], we chose Acr⁺–Mes ClO₄^– as the photocatalyst. The subsequent screening of various solvents indicated that this reaction was not effective in polar aprotic solvents (entries 12–14). Increasing or decreasing the catalyst amount was not beneficial (entries 15 and 16). Furthermore, in the absence of air, no target product was detected (entry 17). Additionally, the control experiments indicated that the photocatalyst and visible light are indispensable for this transformation (Table 1, entries 18 and 19). In the end, the optimal reaction results were obtained using 3 mol% of Acr⁺–Mes ClO₄^– in 2 mL CH₃CN, under air, at room temperature, and with blue light irradiation (Table 1, entry 3).

After getting the optimized reaction conditions, the scope of this C–N cross-coupling reaction was investigated by reacting 6-chloropurine (1a) with a variety of arenes. As shown in Scheme 2, many arenes were suitable for this transformation (3a–3j). This reaction occurred mainly on the N9 position of the purines, but also on N7 position, like 3b and 3b', 3c and 3c'. With other substates like 3d–3l, only a trace amount of N7-aryl products were produced. Because of the steric hindrance between chlorine on N6 position and the methyl group on mesitylene, we only get N9-substituted product 3a. Delightfully, 9, 9-dimethyl-9H-fluorene and diphenyl oxide were reactive (3k, 3l). In addition, halo, nitro or amino substituents on the phenyl ring were not tolerated under the reaction conditions.

Next, the coupling of mesitylene with a series of purine derivatives under the optimized conditions was carried out (Scheme 3). With halo substituents on the purine ring, the target compound was produced in ≥90% yields (4b–4d). In addition, the structure of 4d was certified by single-crystal X-ray diffraction analysis. The electron-donating groups were also well tolerated (4e, 4f). Unsubstituted purine afforded a separable regioisomeric mixture with a total yield of 69% (4h). Unfortunately, adenine did not provide the desired product in this catalytic system (4i). It is worth mentioning that when protected adenine 3j was used, one Boc group was probably lost during the reaction to provide both 4j and 4j' with a total yield of 72%. To our delight, benzimidazole and methyl 1, 2, 4-triazole-3-carboxylate were also efficiently N-arylated (4k, 4l).

In addition, we conducted a gram-scale experiment to demonstrate the utility of this method (Scheme 4). The expected product was formed in 69% yield under the optimized conditions, demonstrating the procedure's potential application in industrial processes.

In order to gain more insight into the reaction mechanism, a few control experiments were carried out (Scheme 5). When radical scavengers like 1, 1-diphenylethylene or 2, 6-di-tert-butyl-4-methyl phenol (BHT) were added, only a trace amount of product was detected by GC-MS, which indicates that a radical pathway might be involved.

Based on the above results and previous reports [25, 26, 29-32], a plausible mechanism is proposed (Scheme 6). The photocatalyst Acr⁺−Mes ClO₄⁻ is excited by blue LEDs to generate its excited state [Acr⁺−Mes ClO₄⁻]*, which then undergoes a single electron transfer (SET) process with mesitylene 2a to generate mesitylene radical cation A and Acr−Mes radical, which is oxidized by O₂ to generate superoxide O₂^•− and complete the photocatalyst cycle. Then, O₂^•− undergoes a hydrogen atom transfer (HAT) with A and 1a to form intermediate B and HOO^•. Subsequently, B is oxidized by HOO^• to form C via an SET pathway. Finally, deprotonation of C produces the desired product 3a, along with the formation of H₂O₂, which was detected by a starch-iodide experiment (see Supporting information for details) [29, 33]. Furthermore, fluorescence quenching study and cyclic voltammetry experiments (see Supporting information for details) support an SET process [34] between the excited state [Acr⁺−Mes ClO₄⁻]* and mesitylene.

In conclusion, we have reported a highly atom economical method for the synthesis of 9-aryl purine derivatives. Using an acridinium photooxidant under blue light irradiation, arene radical cations are generated and form C(sp²)–N bond with the N9–H in purines. Additionally, other nitrogen heterocycles, benzimidazole and methyl 1, 2, 4-triazole-3-carboxylate, were also used as the nitrogen source. These compounds are important structural motifs in biological and pharmaceutical fields. Compared with previous methods, this metal-free and external oxidant-free purine N9-arylation protocol which occurs under ambient atmosphere while at room temperature, is more attractive and more eco-friendly.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by Basic and Applied Basic Research Foundation of Guangdong Province in China (No. 2021A1515012349).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.04.065.