Enantioselective, Decarboxylative (3+2)-Cycloaddition of Azomethine Ylides and Chromone-3-Carboxylic Acids

Herein, we describe the synthesis of a variety of chiral hybrid pyrrolidine-chromanone polycyclic derivatives. A convenient (3+2)-annulation of azomethine ylides with chromone-3-carboxylic acid realized under Brønsted base catalysis produced highly functionalized products in high yields with good stereoselectivities through asymmetric, intermolecular, and decarboxylative (3+2)-cyclization.


Results
Initially, the quinine-catalyzed cycloaddition between chromen-4-one 5 and imine 3a was attempted (Table 1, entry 1). Disappointingly, no reaction was observed. Therefore, the activation of 2a through the introduction of the carboxylic acid moiety was attempted. We were pleased to observe that the devised decarboxylative cycloaddition with chromone-3-carboxylic acid 2a proceeded efficiently and with concomitant decarboxylation (Table 1, entry 2). Importantly, both the efficiency and the diastereoselectivity of the process were excellent. However, the enantioselectivity Scheme 2. Synthetic objective of the present work.
Herein, we present our studies on decarboxylative (3+2)-cycloaddition between chromone-3-carboxylic acids 2 (acting as electron-poor dipolarophiles) and azomethine ylides 3 or 4 derived from salicylaldehydes and appropriate amines (acting as a dipol) proceeding under mild, basic conditions [43,44]. Our studies demonstrate that the presence of a carboxylic acid group is beneficial for the process, providing an alternative method for the preparation of hybrid molecules 1 containing chromenopyrrolidine units with a quaternary stereogenic center, in some cases.

Results
Initially, the quinine-catalyzed cycloaddition between chromen-4-one 5 and imine 3a was attempted (Table 1, entry 1). Disappointingly, no reaction was observed. Therefore, the activation of 2a through the introduction of the carboxylic acid moiety was attempted. We were pleased to observe that the devised decarboxylative cycloaddition with chromone-3carboxylic acid 2a proceeded efficiently and with concomitant decarboxylation (Table 1, entry 2). Importantly, both the efficiency and the diastereoselectivity of the process were excellent. However, the enantioselectivity required further optimization. Consequently, in the first step, five different catalysts were tested (Table 1, entries 2-6). Optimization studies indicated that bifunctional cinchona alkaloid derivatives 6b-e (Table 1, entries 3-6) were better when compared to simple alkaloids such as quinine 6a (Table 1, entry 2). It was found that the presence of a strong H-bonding unit in the catalyst structure was beneficial for the enantiomeric ratio. Unfortunately, a significant decrease in the diastereoselectivity of the cycloaddition was noted. Subsequently, solvent screening using 6c as the catalyst was performed (Table 1, entries 7-10). As a consequence, CHCl 3 was identified as the best solvent for the optimized cascade (Table 1, entry 8). The temperature screening (Table 1, entries 8, 11, 12) indicated that the enantioselectivity of the process can be slightly enhanced by lowering the reaction temperature to 0 • C (Table 1, entry 11). Further reduction in the temperature to −20 • C did not affect the enantiomeric excess (Table 1, entry 12). Importantly, both the amount of solvent used and the catalyst loading can be reduced without any effect on the reaction outcome (Table 1, entries 12-16).  8,11,12) indicated that the enantioselectivity of the process can be slightly enhanced by lowering the reaction temperature to 0 °C (Table 1, entry 11). Further reduction in the temperature to −20 °C did not affect the enantiomeric excess (Table 1, entry 12). Importantly, both the amount of solvent used and the catalyst loading can be reduced without any effect on the reaction outcome (Table 1, entries 12-16). 1 90:10 a All reactions were performed in a 0.10 mmol scale using 2a or 5 (1.0 equiv.) and 3a (1.0 equiv.) in the presence of the corresponding catalyst 6a-e (20 mol%) in the solvent (0.2 mL) for 24h. b Reaction was performed in CHCl3 (0.4 mL). c Reaction was performed in CHCl3 (0.1 mL). d Reaction was performed using 10 mol % of 6c. e Reaction was performed using 6c (5 mol %).

No.
Cat Having accomplished the optimization studies, the scope of the methodology with regard to both reaction partners was studied (Schemes 3 and 4). Initially, various chromone-3-carboxylic acids 2 were reacted with the imine 3a under optimized reaction conditions (Scheme 3). In some of the cases (Scheme 3, products 1b,f) a longer reaction time and higher amounts of catalyst were required in order to achieve full conversion. To our delight, the decarboxylative (3+2)-cycloaddition proceeded efficiently, providing chromenopyrroles 1a-h in good to high yields. Moreover, the diastereoselectivity of the developed reaction increased in all of the cases. In terms of enantioselectivity, the cycloaddition was found to be unbiased towards the electronic properties of substituents on the aromatic ring in acids 2a-h, and it remained at a similar level to the model reaction. conditions (Scheme 3). In some of the cases (Scheme 3, products 1b,f) a longer reaction time and higher amounts of catalyst were required in order to achieve full conversion. To our delight, the decarboxylative (3+2)-cycloaddition proceeded efficiently, providing chromenopyrroles 1a-h in good to high yields. Moreover, the diastereoselectivity of the developed reaction increased in all of the cases. In terms of enantioselectivity, the cycloaddition was found to be unbiased towards the electronic properties of substituents on the aromatic ring in acids 2a-h, and it remained at a similar level to the model reaction. In the second part of the scope studies, the possibility of employing various diethyl iminomalonates 3a-e in the devised strategy was tested (Scheme 4). It turned out that the application of imines 3a-e had significant influences on both the yield and the enantioselectivity of the methodology. The reaction efficiency decreased compared to that shown in the first part of the optimization studies, and instead of obtaining the product quantitatively, yields in the order of 70-80% were obtained. Moreover, when the branching of the alkyl chain in 3d was introduced, enantioselectivity lowered to 75:25 er. Gratifyingly, strongly electron-withdrawing substituent (NO2) 3e was tolerated in this reaction, and the desired chromenopyrrole 1l was afforded with good yield. The diastereoselectivity of the methodology remained at a good level; in addition for the reaction with the tert-butyl substituent, it was as high as 20:1 dr. The possibility to replace the diethyl iminomalonates 3 scaffold with imines 4 bearing γ-lactone rings was also evaluated, and studies were initiated with the goal of finding optimal reaction conditions. In the first step, catalyst screening was performed, and gratifyingly, it was found that cinchona alkaloids 6a promote cycloaddition ( Table 2, entries 1-5). The reaction was terminated within 24 h, and chromenopyrrole 1 was obtained as a mixture of two diastereoisomers which differed in configuration on C-3 stereogenic centers with yields within the range of 70-90%. In particular, the application of bifunctional catalysts 6b-e, bearing either a thiourea or squaramide moiety, led to a significant increase in the reaction enantioselectivity; however, its diastereoselectivity remained low. Among all catalysts tested, derivative 6c proved optimal. With the best catalyst identified, solvent screening was initiated (Table 2, entries 7-10); however, none of the tested solvents were found to improve the reaction outcome, and therefore, further In the second part of the scope studies, the possibility of employing various diethyl iminomalonates 3a-e in the devised strategy was tested (Scheme 4). It turned out that the application of imines 3a-e had significant influences on both the yield and the enantioselectivity of the methodology. The reaction efficiency decreased compared to that shown in the first part of the optimization studies, and instead of obtaining the product quantitatively, yields in the order of 70-80% were obtained. Moreover, when the branching of the alkyl chain in 3d was introduced, enantioselectivity lowered to 75:25 er. Gratifyingly, strongly electron-withdrawing substituent (NO 2 ) 3e was tolerated in this reaction, and the desired chromenopyrrole 1l was afforded with good yield. The diastereoselectivity of the methodology remained at a good level; in addition for the reaction with the tert-butyl substituent, it was as high as 20:1 dr.
The possibility to replace the diethyl iminomalonates 3 scaffold with imines 4 bearing γ-lactone rings was also evaluated, and studies were initiated with the goal of finding optimal reaction conditions. In the first step, catalyst screening was performed, and gratifyingly, it was found that cinchona alkaloids 6a promote cycloaddition ( Table 2, entries 1-5). The reaction was terminated within 24 h, and chromenopyrrole 1 was obtained as a mixture of two diastereoisomers which differed in configuration on C-3 stereogenic centers with yields within the range of 70-90%. In particular, the application of bifunctional catalysts 6b-e, bearing either a thiourea or squaramide moiety, led to a significant increase in the reaction enantioselectivity; however, its diastereoselectivity remained low. Among all catalysts tested, derivative 6c proved optimal. With the best catalyst identified, solvent screening was initiated (Table 2, entries 7-10); however, none of the tested solvents were found to improve the reaction outcome, and therefore, further optimization was carried out using CH 2 Cl 2 (Table 2, entry 5). In order to obtain better diastereoselectivity, the temperature was lowered to 0 • C (Table 2, entry 11). Carrying out the reaction at −20 • C did not provide any product. In the next part of the optimization studies, the influence of the catalyst amount and the concentration of the reaction were evaluated ( Table 2, entries 13-16), leading to the identification of the optimal reaction parameters ( Table 2, entry 13). Finally, the reaction time was extended from 24 to 48 h, which increased the reaction yield from 71 to 87%.
Having established the best reaction conditions, the scope of the methodology was studied (Schemes 5 and 6). At the beginning, various chromone-3-carboxylic acids 2 containing either electron-withdrawing or donating substituents on the aromatic ring were tested in the reaction (Scheme 5). It was found that the enantioselectivity of the cascade remained at a similar level as for the model reaction; however, its efficiency decreased. Target products 1m-s were obtained in yields within the range of 56-70% as a mixture of two diastereoisomers which differed in their configuration on the C-3 stereogenic center (only the main isomer was presented on Schemes 5 and 6). In this section, the studies also indicated that the position of the substituent in chromone-3-carboxylic acids 2 had no pronounced influence on the stereochemical reaction outcome, and the introduction of two substituents on the aromatic ring was also possible (Scheme 5, product 1s).
Importantly, the incorporation of different functional groups on the aromatic ring of iminodihydrofuran-2-ones 3 was also performed (Scheme 6). To our delight, apart from the example with methyl group 4b, the annulative strategy took place with excellent yields. Moreover, target products 1t-x were afforded with no significant influences either on the diastereoselectivity or the enantioselectivity of the reaction. In all cases, the desired chromenopyrroles 1t-x were obtained as two diastereoisomers at a ratio of around 2:1. For the major diastereoisomer, the enantiomeric ratio was kept around 90:10 er, yet for the minor diastereoisomer, the enantiomeric excess remained at an average level.  Having established the best reaction conditions, the scope of the methodology was studied (Schemes 5 and 6). At the beginning, various chromone-3-carboxylic acids 2 containing either electron-withdrawing or donating substituents on the aromatic ring were tested in the reaction (Scheme 5). It was found that the enantioselectivity of the cascade remained at a similar level as for the model reaction; however, its efficiency decreased. Target products 1m-s were obtained in yields within the range of 56-70% as a mixture of two diastereoisomers which differed in their configuration on the C-3 stereogenic center (only the main isomer was presented on Schemes 5 and 6). In this section, the studies also indicated that the position of the substituent in chromone-3carboxylic acids 2 had no pronounced influence on the stereochemical reaction outcome, and the introduction of two substituents on the aromatic ring was also possible (Scheme 5, product 1s).
Importantly, the incorporation of different functional groups on the aromatic ring of iminodihydrofuran-2-ones 3 was also performed (Scheme 6). To our delight, apart from the example with methyl group 4b, the annulative strategy took place with excellent yields. Moreover, target products 1t-x were afforded with no significant influences either on the diastereoselectivity or the enantioselectivity of the reaction. In all cases, the desired chromenopyrroles 1t-x were obtained as two diastereoisomers at a ratio of around 2:1. For the major diastereoisomer, the enantiomeric ratio was kept around 90:10 er, yet for the minor diastereoisomer, the enantiomeric excess remained at an average level. The absolute configurations of the major stereoisomer of chromone 1m were unambiguously assigned by single-crystal X-ray analysis (Scheme 7) [45]. The absolute configurations of the remaining polycyclic products 1a-x were assigned by analogy. Given these configurational assignments, the reaction mechanism explaining the observed stereochemistry of the products was proposed (Scheme 7). The reaction was initiated The absolute configurations of the major stereoisomer of chromone 1m were unambiguously assigned by single-crystal X-ray analysis (Scheme 7) [45]. The absolute configurations of the remaining polycyclic products 1a-x were assigned by analogy. Given these configurational assignments, the reaction mechanism explaining the observed stereochemistry of the products was proposed (Scheme 7). The reaction was initiated through the deprotonation of 4a by the Brønsted base catalyst 6c to give the corresponding azomethine ylide that participated in a (3+2)-cycloaddition. Importantly, it was postulated that, in this reaction, 6c acted as a bifunctional catalyst. Firstly, a Brønsted base moiety in 6c deprotonated the starting imine 4a to form the corresponding ion pair. Secondly, the H-bonding unit of 6c recognized the chromone-3-carboxylic acid 2a. The subsequent cycloaddition of ylide with chromone-3-carboxylic acid yielded 1m. The decarboxylation of 7 is the key step of the reaction, allowing for the removal of the activating group. The protonation of the enolate 8 thus obtained yielded the desired chromanones 1m. Scheme 6. Enantioselective cycloaddition of chromone-3-carboxylic acids 2a and iminodihydrofuran-2-one 4b-e-scope of iminodihydrofuran-2-one 4 to 1t-x.
The absolute configurations of the major stereoisomer of chromone 1m were unambiguously assigned by single-crystal X-ray analysis (Scheme 7) [45]. The absolute configurations of the remaining polycyclic products 1a-x were assigned by analogy. Given these configurational assignments, the reaction mechanism explaining the observed stereochemistry of the products was proposed (Scheme 7). The reaction was initiated through the deprotonation of 4a by the Brønsted base catalyst 6c to give the corresponding azomethine ylide that participated in a (3+2)-cycloaddition. Importantly, it was postulated that, in this reaction, 6c acted as a bifunctional catalyst. Firstly, a Brønsted base moiety in 6c deprotonated the starting imine 4a to form the corresponding ion pair. Secondly, the H-bonding unit of 6c recognized the chromone-3-carboxylic acid 2a. The subsequent cycloaddition of ylide with chromone-3-carboxylic acid yielded 1m. The decarboxylation of 7 is the key step of the reaction, allowing for the removal of the activating group. The protonation of the enolate 8 thus obtained yielded the desired chromanones 1m. To demonstrate the usefulness of cycloadducts 1 the annulation of 1k with formaldehyde was attempted (Scheme 8). It was found that under acidic conditions in the presence of an aq. solution of formaldehyde, product 9a bearing an oxazine moiety was obtained in 95% yield. Notably, the reaction proceeded with full preservation of the stereochemical information introduced at the decarboxylative cycloaddition step, as 9a was obtained as a single diastereoisomer. To demonstrate the usefulness of cycloadducts 1 the annulation of 1k with formaldehyde was attempted (Scheme 8). It was found that under acidic conditions in the presence of an aq. solution of formaldehyde, product 9a bearing an oxazine moiety was obtained in 95% yield. Notably, the reaction proceeded with full preservation of the stereochemical information introduced at the decarboxylative cycloaddition step, as 9a was obtained as a single diastereoisomer. Scheme 8. Synthesis of oxazine 9a from 1k.

General Methods
NMR spectra were acquired on a Bruker Ultra Shield 700 instrument (Bruker Corporation, Billerica, MA, USA), running at 700 MHz for 1 H and 176 MHz for 13 C, respectively. Chemical shifts (δ) were reported in ppm relative to residual solvent signals Scheme 8. Synthesis of oxazine 9a from 1k.

General Methods
NMR spectra were acquired on a Bruker Ultra Shield 700 instrument (Bruker Corporation, Billerica, MA, USA), running at 700 MHz for 1 H and 176 MHz for 13 C, respectively. Chemical shifts (δ) were reported in ppm relative to residual solvent signals (CDCl 3 : 7.26 ppm for 1 H NMR and 77.16 ppm for 13 C NMR). Mass spectra were recorded on a Bruker Maxis Impact spectrometer using electrospray (ES+) ionization (referenced to the mass of the charged species). Analytical thin layer chromatography (TLC) was performed using pre-coated aluminium-backed plates (Merck Kieselgel 60 F254) and visualized by ultraviolet irradiation. Unless otherwise noted, analytical-grade solvents and commercially available reagents were used without further purification. For flash chromatography (FC), silica gel (Silica gel 60, 230-400 mesh, Merck, Darmstadt, Germany) was used. The enantiomeric ratio (er) of the products were determined by chiral stationary phase HPLC by Ultra Performance Convergence Chromatography (UPCC), using Daicel Chiralpak IA, IB, IC, and IG columns as chiral stationary phases. Diethyl iminomalonates 3 and iminodihydrofuran-2-one 4 were prepared from the corresponding 2-hydroxyaldehyde following the literature procedure [46]. Chromone-3-carboxylic acids 2 were prepared from the corresponding 2-hydroxyacetophenones following the literature procedure [47]. An ordinary screw-cap vial was charged with a magnetic stirring bar, the corresponding chromone-3-carboxylic acid 2 (0.1 mmol, 1 equiv), CHCl 3 (0.2 mL), catalyst 6c (0.005 mmol, 0.05 equiv), and the corresponding 2-hydroxyarylideneaminomalonates 3 (0.1 mmol, 1 equiv). The reaction mixture was stirred at 0 • C and monitored by 1 H NMR spectroscopy. After the complete consumption of the chromone-3-carboxylic acid 2, the mixture was directly subjected to FC on silica gel (hexane:ethyl acetate 5:1) to provide the desired products 1.