Introduction

Organic compounds bearing an amide group are significant building blocks for the synthesis of pharmaceuticals and natural products1,2,3,4. As an ambident nucleophile, the peculiar reactivities of nitrogen and oxygen atoms in amide moiety diversify the reaction processes and simultaneously enhance the difficulty of controlling the reaction selectivities. Generally, the reaction chemoselectivities of ambident amide compounds can be controlled by the utilization of appropriate catalysts and reagents5,6,7,8,9. For examples, Nishikata’s group has reported an intermolecular cyclization of α-bromoamides and acrylates, with the reactivity of the nitrogen or oxygen nucleophile of the amide group being finely controlled by using a copper catalyst with an appropriate base (Fig. 1a)10. Adopting o-alkynylbenzamides as substrate, Belmont and coworkers demonstrated the divergent synthesis of five-membered heterocyclic isoindolinones and isobenzofuranones by silver-catalyzed cycloisomerization11. However, differing from Belmont’s work, Jiang and colleagues realized the divergent synthesis of six-membered isoquinolinone and iminoisocoumarin derivatives, the reaction of which was controlled by using gold/ligand or platinum catalyst (Fig. 1b)12. Despite these excellent advances, to the best of our knowledge, no document describing the control of the ambident reactivities of o-alkenylbenzamides has been reported till now. In this regard, a new protocol dictating the chemodivergent cycloisomerization of o-alkenylbenzamides, by tuning the differentiation of the N vs O nucleophilic strength, to construct novel heterocyclic skeletons should be highly desirable.

Fig. 1: Strategies for divergent cyclization of amide derivatives.
figure 1

a, b Previous works on cycloisomerization of amide derivatives. c This work, hypervalent iodine reagents mediated intramolecular cycloisomerization and 1,2-aryl migration.

Full size image

Over the past decades, hypervalent iodine reagents (HIRs) have attracted great attention in the field of organic synthesis, owing to their mild, environmentally benign characteristics and chemoselective oxidizing properties in contrast to heavy metal-based oxidants13,14,15,16,17,18,19,20. Inspired by the exclusive rearrangement21,22,23,24,25,26,27,28,29,30,31 reaction mediated by hypervalent iodine reagents and our previous work about constructing isocoumarins derivatives via 1,2-aryl migration tactics32, we envisioned that the reaction of hypervalent iodine reagents and o-alkenylbenzamides might trigger 1,2-aryl migration processes33,34,35,36,37,38,39,40,41,42 and thus bring about an opportunity to differentiate nucleophilic sites. Herein, we present our results in controlling the reactivity of o-alkenylbenzamides by hypervalent iodine species generated in situ, which realized the divergent synthesis of isoquinolinones43,44,45,46 and iminoisocoumarins47,48,49,50 derivatives (Fig. 1c). It should be noted that there are no precedents on hypervalent iodine controlled chemoselectivity in the cyclization of o- alkenylbenzamides, and the work described here highlights a one-pot transformation involving an exclusive cascade sequence of chemoselective cyclization, 1,2-aryl migration and elimination processes.

Results and discussion

Our initial studies focused on the optimization of reaction conditions. At the outset, o-alkenylbenzamide 1a was selected as the model substrate to examine the cycloisomerization reaction, and the results were listed in Table 1. When substrate 1a was treated with iodosobenzene (PhIO) in MeOH, combined with BF3·OEt2 (0.2 equiv) as a Lewis acid, we found the reaction displayed a completely distinct N-attack cyclization mode to furnish isoquinolinone 2a in 56% yield. After screening the effect acidic or basic of activators (entries 2–7), we found that TMSOTf was the most efficient catalyst for this transformation (Table 1, entry 5). Further investigation on reaction temperature revealed that the best outcome was obtained when the reaction was operated at reflux temperature (Table 1, entry 8). On the basis of the above screening results, the most optimal conditions for converting 1a to isoquinolinone 2a were concluded to be: substrate 1a with PhIO (1.5 equiv) and TMSOTf (0.2 equiv) in MeOH at reflux temperature.

Table 1 Optimization of the reaction conditionsa.
Full size table

Encouraged by the above results, we then turned our attention toward exploring the chemodivergent pattern of the protocol for the synthesis of O-cyclized iminoisocoumarin products. To our delight, we discovered that when substrate 1a combined with hydroxy(tosyloxy)iodobenzene (HTIB) in DCE (1,2-dichloroethane) at ambient temperature for 0.5 h, the reaction resulted in the formation of iminoisocoumarin 3a in 44% yield (Table 1, entry 9). With the encouraging result in hand, we came to further optimize the reaction conditions. First, a detailed screening of solvents, hypervalent iodine reagents and temperature was carried out (Supplementary Table S2). The result indicated that the reaction of substrate 1a with HTIB at 80 °C in DCE (1,2-dichloroethane) provided iminoisocoumarin 3a with 55% yield, with the reaction completed within 5 min (Table 1, entry 10). Next, almost identical results were obtained when using active hypervalent iodine species formed in situ from iodosobenzene (PhIO) and 4-toluenesulfonic acid (Table 1, entry 11)51,52. Furthermore, sulfonic acids with different structures were further tested (Table 1, entries 12–14), and 2,4,6-tris-isopropylbenzene sulfonic acid (S4) was identified as the most efficient promoter. To our delight, better yields were further achieved with the addition of an exogenous Lewis acid (Table 1, entries 15–18). Specifically, lithium perchlorate (LiClO4) was found to be the optimal additive as the reaction gave the desired product 3a in 90% isolated yield. Further investigation on the loading of LiClO4 indicated that neither decreasing nor increasing the equivalents were beneficial (Table 1, entries 19–20). Based on the above results, the optimal conditions for the transformation to iminoisocoumarin 3a were finalized to be: PhIO (1.5 equiv), 2,4,6-tris-isopropylbenzene sulfonic acid (S4; 1.5 equiv), LiClO4 (1.5 equiv) in DCE at 80 °C.

With the optimized conditions in hand, we began to explore the general applicability of the divergent transformation, targeting the N-cyclization by using variously substituted N-phenyl-2-alkenylbenzamides with MeOH as a reaction partner34,53,54,55,56,57 and solvent being first studied (Fig. 2). Replacing the methyl group with other alkyl substituent in the substrate did not have a significant impact on the outcome of the transformation, as substrates 1b-f were all converted to the desired products 2b-f with good yields. The method was well applicable to aromatic substituents with either electron-donating groups or electron-withdrawing groups at meta- or para-position of phenyl ring, furnishing the corresponding isoquinolinones 2i-q in good yields (70–86%). Furthermore, altering the benzo-ring backbone to biphenyl and naphthyl were also compatible with the reaction conditions, with the corresponding fused heterocycle 2r and 2s obtained in 82 and 77% yield respectively. Notably, substrates with electron-poor or electron-rich skeletons (1t-v) could be conveniently converted to products 2t-v in a satisfactory yield, proving that the method could also be applied to substrates with benzamide nucleus bearing substituent. It is worth noting that, for the reaction of substrates 1g-h under the standard conditions A, the substituent effect of ortho-position prevents the cyclization, as only trace amounts of the desired product was produced in each case. To our disappointment, when the methyl group in the substrate was replaced by a phenyl group, the reaction of the corresponding substrate gave a complex mixture and no desired product was achieved under the standard conditions.

Fig. 2: Substrate scope study for synthesis of isoquinolinones 2.
figure 2

[a]Reaction conditions: 1 (0.5 mmol), PhIO (1.5 equiv) and TMSOTf (0.2 equiv) in MeOH (5.0 mL) reflux for 0.5–2 h. Isolated yield.

Full size image

Next, we came to explore the chemodivergent synthesis of iminoisocoumarins 3 by subjecting substrates 1 to conditions B (Fig. 3). O-alkenylbenzamides 1b–v bearing different alkyl substituted R1 group and the substituted phenyl ring could smoothly furnish the corresponding iminoisocoumarins 3b–v with sole chemoselectivity in satisfactory to excellent yields (73–92%). Notably, in contrast to the inferior performance of substrate 1g and 1h in N-cyclization mode reaction, iminoisocoumarin product 3g and 3h could be obtained in high yield mediated by the modified Koser’s reagent.

Fig. 3: Substrate scope study for synthesis of iminoisocoumarins 3.
figure 3

[a]Reaction conditions: 1 (0.5 mmol), PhIO (1.5 equiv), 2,4,6-tris-isopropylbenzene sulfonic acid (S4; 1.5 equiv) and LiClO4 (1.5 equiv) in DCE (5.0 mL) at 80 °C for 0.1–0.5 h. Isolated yield.

Full size image

In addition, we carried out some control experiments to ascertain the hypervalent iodine species that is responsible for promoting the transformation (Fig. 4). First, we monitored the reaction process of substrate 1a with PhIO in MeOH in the absence of TMSOTf, and it was found that the reaction did not occur (Fig. 4a). Furthermore, the reaction carried out in the absence of PhIO also completely inhibited the transformation of substrate 1a into product 2a (Fig. 4b). The two results suggested that both hypervalent iodine reagent and Lewis acid are key reagents that enabled the transformation to occur. Next, when LiClO4 was solely applied to the transformation of 3a, no reaction occurred either (Fig. 4c). On the basis of this result as well as the outcome of the initial attempt of using HTIB (Table 1, entry 9), we tentatively infer that LiClO4 could on one hand promote the formation of Koser’s reagent, while on the other hand, coordinate with hydroxy group in the hypervalent iodine specie formed in situ.

Fig. 4: Control experiments.
figure 4

a Control experiment to verify the necessity of hypervalent iodine reagent and Lewis acid. b Control experiment to verify the effect of LiClO4.

Full size image

To gain insight into the mechanism and chemoselectivity of above systems, we performed density functional theory (DFT) calculations on the reaction of substrate 1a under conditions A and conditions B (Fig. 5). As previous work shows, PhI(OMe)2 is generated in situ by PhIO and MeOH54,56,57. It is known that TMSOTf can be present as TMS+ + TfO- in organic solvents58,59 .The calculation shows that the complex IM2-1, which is formed by PhI(OMe)2 and TMSOTf, is thermodynamically favored over reagent 1 by 14.0 kcal/mol (Fig. 5a). The amide-iminol tautomerism is achieved via a concerted intermolecular hydrogen exchange between two molecules of substrate 1a with an activation energy of 20.7 kcal/mol. Then, the nucleophilic attack onto the iodine(III) center of IM2-1 by the olefinic double bond of 1a’ proceeds, leading to the formation of iodonium ion IM2-2. Then, isomerization of the amide occurs via an intermolecular proton shift with the aid of MeOH as a proton shuttle, which has an activation energy of 26.1 kcal/mol; this process is rate-limiting step in the reaction. Then, the nitrogen atom attacks the more substituted carbon atom on three-membered heterocycle via a 5-exo cyclization in TS2-1, which has an energy barrier of 23.0 kcal/mol; this process is the rate-limiting step in the reaction.The nucleophilc attack of oxygen on the three-memebered heterocycle needs to overcome an energy barrier of 35.4 kcal/mol, which is much higher than that required by nitrogen-attack mode. Then, IM2-4 is generated by the hydrogen bonding interaction between IM2-3 and MeOTMS. A facile proton shift process occurs to give IM2-5. The activated carbon atom bonded to iodine(III)-center is nucleophilically attacked by the phenyl ring, giving a phenonium ion IM2-6 via TS2-2, which has an energy barrier of 11.8 kcal/mol relative to IM2-5. Then, ring opening of the three-membered ring in IM2-6 takes place with simultaneous ring expansion to the six-membered ring IM2-7 having a tertial carbocation. Finally, a unimolecular elimination process occurs to yield product 2a. The Gibbs free energy profile of the overall reaction shows that the generation of 2a is highly exergonic by 97.3 kcal/mol.

Fig. 5: DFT-computed potential energy profile for the reaction of 1a under the conditions A and conditions B.
figure 5

Ar = 2,4,6-triisopropyl (standard state: 25 °C, 1 mol/L). (For details, see Supplementary data 4).

Full size image

For another reaction (Fig. 5b), it starts with the coordination of the oxygen to the LiClO4 which results in the formation of a thermodynamically more stable complex IM3-1 by 10.8 kcal/mol. Then, the nucleophilic attack onto IM3-1 by the olefinic double bond of 1a takes place, which leads to iodonium ion IM3-2 with the release of ArSO3-. The isomerization of amide group would not happen as the above reaction, because aprotic solvent 1,2-dichloroethane cannot play a role as proton shuttle. The oxygen of carbonyl attacks the more substituted carbon with an energy barrier of 28.1 kcal/mol, while nitrogen atom accomplishes this process with a higher barrier of 35.2 kcal/mol. After that, IM3-6 is generated by a proton shift with a barrier of 1.4 kcal/mol relative to IM3-4. Subsequently, the phenyl ring migration process occurs with an activation energy of 29.6 kcal/mol, yielding carbocation intermediate IM3-7. Finally, product 3a can be obtained by β- elimination. In this reaction, the rate-limiting step is the phenyl migration process of IM3-6 with an activation energy of 29.6 kcal/mol. The Gibbs free energy profile of the overall reaction shows that the generation of 3a is highly exergonic by 79.5 kcal/mol.

Nitrogen or oxygen-attack mode is supported by Hirshfeld charges analysis of IM2-2 and IM3-2, which is capable of predicting electrophilicity and nucleophilicity (Fig. 6). The most important factor responsible for the divergent reactivity of amide moiety is the different nucleophilicity of nitrogen and oxygen atoms under conditions A and B. By the Hirshfeld charges analysis of IM2-2, we observed that more negative charge (−0.202) was concentrated on the nitrogen atom than that on the oxygen atom (−0.180), indicating that the nitrogen attacking mode is more favorable for IM2-2, thus leading to the formation of N-heterocycle intermediate IM2-3. Furthermore, still by the Hirshfeld charges analysis of IM2-2’ shown below, we found that the negative charge at the oxygen atom of the carbonyl group is -0.364, which means that oxygen has higher nucleophilicity than that of nitrogen of IM2-2’, this would result in the generation of O-heterocycle intermediate IM2-3’. However, by comparing the Gibbs free energy of IM2-3 and that of IM2-3’, we found that IM2-3 (−16.9 kcal/mol) is thermodynamically more stable than IM2-3’ (−14.5 kcal/mol). Thus, N-attack mode is thermodynamically favorable in IM2-2 under reaction condition A. (Fig. 6a)60.

Fig. 6: The Hirshfeld charges analysis of IM2-3 and IM3-2.
figure 6

Relative free energies and electronic energies are given in kcal/mol.

Full size image

Additionally, by the Hirshfeld charges analysis of IM3-2, we observed that more negative charge (−0.309) was concentrated on the oxygen atom of amide group than that on the nitrogen atom (−0.070), which indicates that the O-attack mode is more favorable for IM3-2. And, we respectively calculated the activation energy barrier for O-attack and N-attack in IM3-2. The O-attack pathway needs to overcome an activation energy barrier of 28.1 kcal/mol, while nitrogen atom accomplishes N-attack pathway with a higher barrier of 35.2 kcal/mol. The higher barrier of 35.2 kcal/mol makes it impossible for IM3-2 to undergo an N-attack mode under condition B. The comparation between Hirshfeld charge of the oxygen atom (−0.309) and that of nitrogen atom (−0.070) also illustrate that O-attack mode is more favorable for IM3-2. In summary, O-attack is kinetically favorable under condition B, leading to the generation of O-heterocycle intermediate IM3-3 under condition B. (Fig. 6b).

Based on the aforementioned mechanistic studies and the outcomes of the previous reports36,52,54,61,62, we postulated a plausible mechanism for the formation of 2a and 3a (Fig. 7). For the formation of product 2a (Fig. 7a), PhI(OMe)2 is first generated in situ from the reaction of PhIO with MeOH. Then complex IM2-1 is formed by PhI(OMe)2 and TMSOTf, enabling the electrophilic reaction with isomerized substrate 1a’, leading to the formation of iodonium ion IM2-2. Next, isomerization of the amide occurs via an intermolecular proton shift, and the subsequent nucleophilic attack of the nitrogen atom gives intermediate IM2-4.

Fig. 7: Possible reaction mechanism.
figure 7

a Reaction formation mechanism of isoquinolinone 2a. b Reaction formation mechanism of iminoisocoumarin 3a.

Full size image

Then the deprotonation by the methoxy anion forms intermediate IM2-5. The activated carbon atom bonded to iodine-center is nucleophilically attacked by the phenyl ring, giving a phenonium ion IM2-663,64,65,66,67. Then, ring opening of the cyclopropane moiety in IM2-6 occurs with simultaneous ring expansion to give carbocation IM2-7. Finally, the elimination reaction occurs in IM2-7 to form isoquinolinone 2a. Similarly, with regard to the formation of product 3a (Fig. 7b), the modified Koser’s reagent is first formed from the reaction of PhIO with ArSO3H. It is worth noting that he coordination of the hydroxyl group to LiClO4 results in the formation of a thermodynamically more stable complex IM3-1, which undergoes electrophilic addition with olefin to form intermediate IM3-2. Next, the nucleophilic attack of oxygen atom of the amide moiety onto the benzylic carbon center in IM3-2 produces intermediate IM3-4. After that, deprotonation by the sulfonate ion enables the conversion of iminium IM3-4 to imine IM3-6. Then the aromatic backbone of IM3-6 undergoes 1,2-aryl migration36,53,68,69 to form carbon cation IM3-7. Finally, with the removal of the acidic proton by the hydroxide anion released from the prior step, intermediate IM3-7 is converted to iminoisocoumarin 3a is obtained.

In view of the atomic economy of utilizing catalytic hypervalent iodine species, we further investigated the strategy of combining aryl iodine and exogenous oxidant to generate hypervalent iodine species in situ, with a purpose of demonstrating the economic application of this transformation (Fig. 8)70,71,72,73,74. Gratefully, after screening the conditions of the catalytic conversion, we found that the PhI-catalyzed reaction of substrate 1a occurred smoothly by using mCPBA (meta-chloroperoxybenzoic acid) as the terminal oxidant, combined with acetic acid and stoichiometric amount of TMSOTf in MeOH at reflux temperature, with isoquinolinone 2a achieved in a moderate 56% yield (Fig. 8a). Furthermore, iminoisocoumarin 3a could be obtained in 92% yield when substrate 1a was treated with catalytic amount of PhI (20 mol%), mCPBA (2.0 equiv) in the presence of 2,4,6-tris-isopropylbenzene sulfonic acid (S4; 1.5 equiv) and LiClO4 (0.2 equiv) in DCE at 80 °C (Fig. 8b).

Fig. 8: Organocatalytic transformation of chemodivergent synthesis.
figure 8

Experiment to demonstrate the economic application of this transformation.

Full size image

In summary, we have presented an exclusive chemodivergent cycloisomerization approach for constructing isoquinolinones and iminoisocoumarins skeletons starting from an identical o-alkenylbenzamides derivative. Notably, the divergent synthesis employed an exclusive 1,2-aryl migration/elimination strategy, which is realized by utilizing the different hypervalent iodine species generated in situ. In the reaction processes, different hypervalent iodine species was found to play a crucial role in the selectivity switch from N to O-cyclization, with the PhI(OMe)2 species inducing N-attack and the modified Koser’s reagent favoring the O-attack in the cyclization step. DFT studies demonstrated that nitrogen and oxygen atoms of the intermediates in the two reaction systems have different nucleophilicity and thus produce the selectivity of N or O-attack mode.

Methods

General procedure for synthesis of isoquinolinones 2

To a reaction flask filled with iodosobenzene (1.5 equiv, 0.75 mmol) in MeOH (5.0 mL) was added TMSOTf (20 mol %). The mixture was stirred at reflux temperature for 5 min and then reactant 1 (0.5 mmol) was added. The resulting mixture was kept stirring until TLC indicated the total consumption of substrate 1. Then the reaction mixture was quenched with sat. aq. NaHCO3 (5 mL), and extracted with EtOAc (10 mL × 3). The combined organic layer was dried over anhydrous Na2SO4 and the solvent was removed in vacuo. The residue was purified by flash column chromatography on silica gel (petroleum ether/ethyl acetate 10:1) to afford target product 2. Details including experimental procedures are available in the Supplementary Methods (Supplementary Table S1).

General procedure for synthesis of iminoisocoumarins 3

To a reaction flask filled with iodosobenzene (1.5 equiv, 0.75 mmol) in DCE (5.0 mL) was added S4 (1.5 equiv, 0.75 mmol) and LiClO4 (1.5 equiv, 0.75 mmol). The mixture was stirred at 80 °C for 5 min and then reactant 1 (0.5 mmol) was added. The resulting mixture was kept stirring until TLC indicated the total consumption of substrate 1. Then the mixture was quenched with sat. aq. NaHCO3 (5 mL), and extracted with dichloromethane (10 mL × 3). The combined organic layer was dried over anhydrous Na2SO4 and the solvent was removed in vacuo. The residue was purified by flash column chromatography on silica gel (petroleum ether/ethyl acetate 50:1) to afford target product 3. Details including experimental procedures are available in the Supplementary Methods (Supplementary Table S2).