Isocoumarins, including 3,4-dihydroisocoumarin, are naturally abundant flavonoid scaffolds, that exhibit diverse pharmacological activities (Saeed, 2016; Saddiqa et al., 2017). In particular, 3-aryl isocoumarins possess unique and potent biological functions including antifungal (Saeed, 2003), anti-diabetic (Kim et al., 2017), immunomodulatory (Matsuda et al., 1998), antiallergic (Yoshikawa et al., 1992, 1994, 1996), and antimicrobial (Yoshikawa et al., 1992, 1994, 1996) activities, attracting the attention of synthetic and medicinal chemists. In medicinal chemistry, pyridine is considered to be the most efficient and popular bioisosteres with a phenyl moiety to improve biological activities and PK/PD profiles (Foye, 2008; Gaikwad et al., 2012). In this regard, azaisocoumarins, including 3-aryl-5-azaisocoumarin, are potential privileged scaffolds for developing drug candidates to treat a diverse range of diseases. Hence, there has been significant interest in the development of an efficient and novel synthetic route to 3-aryl-5-azaisocoumarin.

Despite the continual development of synthetic strategies for 3-aryl-isocoumarin analogs because of their interesting biological activities (Pal et al., 2011; Saddiqa et al., 2017), only a few methods and examples have been reported for the synthesis of 3-aryl-5-azaisocoumarin (Hellal et al., 2008; Li et al., 2010; Begouin and Maria-João, 2011; Panda et al., 2011; Park et al., 2011; Singh et al., 2018). To date, the synthesis of 3-aryl-5-azaisocoumarins has mostly been accomplished in the same manner as that of 3-aryl-isocoumarins, i.e., via construction of the 2-pyrone moiety in the final step via the intramolecular lactonization of 2-alkylnyl nicotinate derivatives. Because lactonization between the activated alkyne and ortho-carboxylate can occur by the 6-endo-dig or 5-exo-dig cyclization sequence (pathways a and b, respectively, in Figure 1), selectivity against methylenefuranone skeleton is a major challenge in these 2-pyrone formation routes. For instance, Li et al. reported the synthesis of 3-aryl-5-azaisocoumarin the Pinnick-type oxidative lactonization of 2-(phenylethynyl)-nicotinaldehyde in 2010 (Figure 1a) (Li et al., 2010). The transition metal-mediated intramolecular lactonizations of 2-alkynyl nicotinates or 2-alkynyl nicotinaldehyde are popular approaches to obtain 5-azaisocoumarins by construction of the 2-pyrone moiety. Queiroz et al. attempted to make the 2-pyrone moiety of 3-aryl-5-azaisocoumarin via the one-pot Sonogashira coupling/transition metal catalyzed alkyne activation cascade (Figure 1b) (Begouin and Maria-João, 2011). These two methods produced 5-exo-cyclized furanone as a major product. Youn et al. adopted NHC-catalyzed alkyne activation for the synthesis of 5-azaisocoumarins, which selectively afforded a 6-endo or 5-exo cyclized product with appropriate reaction conditions (Figure 1c) (Park et al., 2011). 3-Butyl-5-azaisocoumarin was selectively synthesized via 6-endo cyclization mode from the corresponding precursor. However, for the arylethynyl precursor, the major product was the 5-exo-cyclized lactone instead of 3-aryl-5-azaisocoumarin. Several successful examples of 3-aryl-5-azaisocoumarin synthesis via selective formation of the 2-pyrone moiety have been reported. The Bihel (Figure 1d) (Hellal et al., 2008) and Sarkar groups (Figure 1e) (Panda et al., 2011) reported 3-aryl-5-azaisocoumarin synthesis by transition metal-catalyzed 6-endo-lactonization from ethynylnicotinate or ethynylnicotinaldehyde precursors, respectively. In addition, selective 6-endo-lactonization toward pyrano[4,3-b]quinolin-1-one, a phenyl-fused azaisocoumarin, was accomplished under metal/additive-free domino reaction conditions (Figure 1f) (Singh et al., 2018). In addition to the selectivity problem, previous 2-pyrone formation approaches are inefficient for the installation of substituents on pyridine moiety of 5-azaisocoumarins.

Recently, the metal-catalyzed 6-endo-dig cycloisomerization of N-propargyl enaminone derivatives has been widely used for the synthesis of pyridine-fused complex heterocycles, including azaanthraquinone, and steroidal pyridine (Abbiati et al., 2003; Yan et al., 2007; Jiang et al., 2010; Fei et al., 2011). In addition, intensive efforts to establish efficient synthetic methods for pyridine-fused coumarins and natural products using the metal-catalyzed cycloisomerization of N-propargyl enaminone intermediates have been reported (Ahn et al., 2019; Yoon and Han, 2019). In these studies, the pyridine moiety was constructed in a highly 6-endo-selective manner, which can be explained by σ-bond formation between the nucleophilic enolizable α-position and exo-position of the electrophilic alkyne-metal complex and subsequent spontaneous oxidation to form a stable aromatic pyridine ring (Abbiati et al., 2003; Yan et al., 2007; Cacchi et al., 2008; Jiang et al., 2010; Fei et al., 2011; Mikušek et al., 2016; Lyubov'N et al., 2018; Ahn et al., 2019). It was anticipated that construction of the pyridine moiety via cycloisomerization of N-propargylamine intermediates in the last-stage would be an efficient strategy for the synthesis of 3-aryl-5-azaisocoumarins. As part of our effort to develop a novel synthetic approach toward privileged 3-aryl-5-azaisocoumarin scaffolds, we investigated the metal-catalyzed cycloisomerization of N-pyranonyl propargylamines.

Herein, we describe a concise synthesis of 3-aryl-5-azaisocoumarins via AgOTf-catalyzed cycloisomerization, which to the best of our knowledge, has not been reported to date.

Our research on the development of a novel synthetic route toward 3-aryl-5-azaisocoumarin commenced with the preparation of cycloisomerization precursors. As shown in Scheme 1, N-pyranonyl propargylamine precursors (3a–3k) were prepared according to literature procedures with slight modifications (Dong et al., 2011). The known hydroxy-2-pyrone 1 (Prasad et al., 1995) was treated with tosyl chloride and triethylamine to afford tosylate 2 in 92% yield. Nucleophilic aromatic substitution of pyrone 2 with commercially available hydrochloride salts of propargylamine or diverse known substituted propargylamine (Experimental details: see Supplementary Materials) readily afforded the corresponding N-pyranonyl propargylamines 3a–3h in moderate yields (48–74%). However, 3-bromopropargyl- and 1-phenylpropargyl analogs (3j–3k) were not obtained. The nitrophenyl substituted analog 3i was used for the next reaction without further purification due to inseparable impurities.

With the cycloisomerization precursors 3a–3i in hand, the cycloisomerization conditions were examined using the simple 3a as a model substrate (Table 1). Initially, cycloisomerization of 3a was conducted under thermal rearrangement conditions. When 3a was heated at 120°C in DMSO for 10 h, the desired 5-azaisocoumarin 4a was obtained in poor yield (15%, Table 1, entry 1). The reaction was performed at higher temperature (220°C) in triethylene glycol (TEG) (Han, 2017), affording 5-azaisocoumarin 4a in a slightly higher yield (28%, entry 2). The structure of 4a was confirmed by comparison of its NMR spectrum with previously reported data (Table S1) (Hellal et al., 2008). Recently, many research groups have continuously reported the efficiency of various transition-metal Lewis acid catalysts for alkyne activation in cycloisomerization reaction (Abbiati et al., 2003; Yan et al., 2007; Cacchi et al., 2008; Saito et al., 2009; Jiang et al., 2010; Fei et al., 2011; Mikušek et al., 2016; Vessally et al., 2016; Nizami and Hua, 2017; Sakthivel et al., 2017; Gianni et al., 2018; Lyubov'N et al., 2018; Mancuso et al., 2018; Ahn et al., 2019). To increase the product yield, cycloisomerization was performed in the presence of various metal catalysts, including Ag(I), Au(I), Cu(I), Cu(II), and In(III). Based on our previous work (Ahn et al., 2019), the cycloisomerization of 3a was performed with 0.2 equiv. of AgSbF₆ (entry 3), affording 4a in a higher yield (38%) than that without the catalyst after heating in DMSO at 120°C for 7 h. Encouraged by this yield improvement with the Ag(I) catalyst, various counter-anions was screened (entries 4–7) given that the alkyne-activating Lewis acidity of metal ions such as Ag(I) depends on the nature of the counter-anion (Yamamoto, 2000; Lu et al., 2017). When Ag₂O was employed, the yield of 4a did not notably improve over that obtained with AgSbF₆ (38%, entry 4). However, the yields increased slightly when AgCl (50%, entry 5) or AgNO₃ (53%, entry 6) were used. The best result was obtained when the cycloisomerization was performed with AgOTf, furnishing 4a in 66% yield (entry 7). Moreover, the reaction proceeded to completion much faster (2 h) than with AgSbF₆. Although regioisomeric byproducts were not isolated as expected, the optimal yield for the cycloisomerization of N-pyranonyl propargylamine precursors 3a was slightly lower than that reported in a previous study on pyridocoumarins (Ahn et al., 2019), likely because of the greater degree of decomposition of the acid-labile pyrone moiety. When the reaction was conducted with AuCl, the yield was reduced compared to that observed in the thermal aza-Claisen rearrangement in TEG (24%, entry 8). Furthermore, the desired cycloisomerization product 4a was not obtained when using InCl₃ (entry 9). Cu(I) and Cu(II) catalysts afforded 5-azaisocoumarin 4a in better yields than with Au(I) or In(III) catalysts (33%−62%; entries 10–14). Although CuI afforded 4a in the best yield among the copper catalysts tested herein (62%, entry 11), the yield of 4a and reaction rate did not show any notable improvements over that obtained with AgOTf. Other polar solvents including DMF (entry 15), MeCN (entry 16), and ionic liquid systems (entries 17–18), resulted in lower reaction rates and yields of 4a. In particular, the use of an ionic liquid hampered the cycloisomerization, contrary to previous reports (Corma et al., 2005; Neatu et al., 2009; Mancuso et al., 2019). The lower yields obtained in ionic liquids was attributed to the inhibition of the interaction between the alkyne and metal catalyst by the highly polar solvent.

Next, the scope and limitations of the newly developed Ag(I)-catalyzed 3-aryl-5-azaisocoumarin synthesis were examined (Table 2). We attempted to synthesize the 3-aryl-5-azaisocoumarins from corresponding substituted alkynes (3b–3i). Methyl (3b–3c; entry 1–2) and phenyl substituted (3d–3f; entry 3–5) propargylamines generated the corresponding 3-aryl-5-azaisocoumarins (4b–4f) without any notable yield decreases. However, cycloisomerization under standard condition of enynylamine precursors (3g–3h; entry 6–7) resulted in an unidentifiable mixture of byproducts, whereas only substrates were recovered when the reactions were conducted at ambient temperature. These results were attributed to preference of the metal-alkyne complex for reaction with the electron-rich alkene rather than enolizable α-position. Although the cycloisomerization reaction of triisopropylsilyl (TIPS) propargylamine (3i) was allowed to proceed for 24 h, only substrate 3i was recovered in 88% yield (entry 8). This implied that the bulky silyl substituents of alkyne can interfere with formation of the metal-alkyne complex.

In conclusion, an alternative synthetic route for 3-aryl-5-azaisocoumarins was developed which includes novel approach to install pyridine moieties on a readily available phenyl-2-pyrone precursor via regioselective metal-catalyzed cycloisomerization. Through an intensive preliminary investigation, AgOTf was determined to be an adequate alkyne-activation catalyst and was applicable in rearrangement of N-pyranonyl propargylamines bearing methyl or phenyl substituents to C-6 or C-8 substituted 3-aryl-5-azaisocoumarin. Derivatization of the pyridine moiety was accomplished efficiently via three-step reaction using the common starting material hydroxy-2-pyrone 1 and known propargylamines. Considering that 3-phenyl-5-azaisocoumarin 4a, which was synthesized previously as a regioisomeric mixture using conventional lactonization strategies, was obtained regioselectively in moderate yield, this novel step-economical procedure can be widely utilized in derivatization of C-3, C-6, and C-8 of 5-azaisocoumarins. Further studies on the synthetic applications of the newly developed method are currently underway.

All datasets generated for this study are included in the article/Supplementary Material.

YH conceptualized the work. JY and CL synthesized all compound and performed m.p. R_f, NMR, IR, LRMS, and HRMS analysis. The manuscript was written by CL and YH. All authors approved the manuscript in its final form for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.