Construction of cyclopentane-fused coumarins via DBU-catalyzed [3+2] cycloaddition of 3-homoacyl coumarins with cyclic 1-azadienes.

The metal-free DBU catalyzed [3+2] cycloaddition of 3-homoacyl coumarins with cyclic 1-azadienes proceeded smoothly to furnish the corresponding highly functionalized cyclopentane-fused coumarins with excellent diastereoselectivity and complete chemoselectivity and in good yields under mild conditions. This journal is © The Royal Society of Chemistry.

Coumarins[1] and cyclopentane scaffolds[2] are widely distributed in natural products and display a wide range of biological and pharmacological activities. When combining coumarin skeletons with cyclopentane moieties, the cyclopentane-fused coumarins show interesting biological activities. For example, aflatoxins, which occur naturally, exhibit acute toxicity, teratogenicity, mutagenicity and carcinogenicity (Fig. 1).[3] Herbertenolide, which belongs to the family of sesquiterpenoids, was first isolated from the leafy liverwort Herberta adunca, the extract of which showed significant inhibition against the growth of certain plant pathogenic fungi (Fig. 1).[4] Not surprisingly, the strategies for synthesis of cyclopentane-fused coumarins have attracted much attention.[5]

Fig. 1

Bioactive molecule bearing cyclopentane-fused coumarin.

Recently, the group of Lin developed a 1,3-dipolar precursor 3-homoacyl coumarin, which is an efficient synthon for the construction of cyclopentane-fused coumarins under the catalysis of bases (Scheme 1a and b).[6] However, the partners reacted with 3-homoacyl coumarins were focus on α,β-unsaturated carbonyl compounds and conjugated dienes. The other dipolarophiles, such as aza-dienes, might also be potential candidates for the [3+n] cycloadditions with 3-homoacyl coumarins but never been developed.

Scheme 1

The reaction of 3-homoacyl coumarins with dipolarophiles catalyzed by Brønsted base.

The cyclic 1-azadienes are extensive used dipolarophiles and have been widely involved in a series of cyclization reactions as two-,[7] three-[8] or four[9] member synthons. While the organocatalytic [3+2] cycloaddition of cyclic 1-azadiene as two synthons has rarely been investigated.[7] In 2016, Chen's[7] and Guo's[7] group respectively developed a asymmetric [3+2] annulation reaction of Morita–Baylis–Hillman carbonates with cyclic 1-azadienes catalyzed by Lewis base. Encouraged by these works above and as our continuing efforts on cycloadditions,[10] herein we expected to achieve the first [3+2] cycloaddition reaction of 3-homoacyl coumarins with cyclic 1-azadienes catalyzed by Brønsted base for synthesis of various functionalized cyclopentane-fused coumarins derivatives efficiently (Scheme 1d). However, Huang's group reported a enantioselective 1,4-addition reaction of benzofuran azadiene with 3-homoacyl coumarin, instead of cycloaddition (Scheme 1c).[11] To achieve our assumption in high chemoselectivity would be a challenging work.

In an initial experiment, cyclic 1-azadiene 1a and 6-bromo-3-(2-oxo-2-phenylethyl)-2H-chromen-2-one 2a were employed as the model substrates to carry out the reaction in CH2Cl2 at room temperature in the presence of DABCO. To our delight, the desired [3+2] cycloadduct 3aa was obtained in 56% yield (Table 1, entry 1). Subsequently, several bases were screened and when the use of stronger base (Table 1, entries 2–4), DBU, the reaction gave a higher yield in 12 h, and no 1,4-addition product was observed (Table 1, entry 3). Further screening of several representative solvents, such as THF, toluene, DCE and CH3CN, revealed that the reaction proceeded better in THF with 86% yield (Table 1, entry 5). Therefore, the best reaction conditions were determined as below: DBU, THF and room temperature (Table 1, entry 5).

Screening of the reaction conditionsa

Entry	Base	Solvent	Time (h)	Yieldb (%)	drc
1	DABCO	CH2Cl2	24	56	>20 : 1
2	DMAP	CH2Cl2	24	60	>20 : 1
3	DBU	CH2Cl2	12	78	>20 : 1
4	Et3N	CH2Cl2	24	67	>20 : 1
5	DBU	THF	12	86	>20 : 1
6	DBU	Toluene	12	31	>20 : 1
7	DBU	DCE	12	76	>20 : 1
8	DBU	CH3CN	12	73	>20 : 1

Reactions were carried out with 1a (0.1 mmol), 2a (0.12 mmol), and base (20 mol%) in 2 mL of solvent at rt.

Isolated yields.

Determined by 1H NMR.

Under optimal reaction conditions, the substrate scope of the cyclic 1-azadienes 1 was investigated and the results were summarized in Table 2. As expected, the desired [3+2] cycloadducts 3ba–3qa were obtained in moderate to good yields. Both electron-withdrawing (entries 1–8) and electron-donating substituents (entries 9–14) on the benzene ring were tolerated and the yields of the former were slightly higher than the latter. And either para-, meta- or ortho-substituted phenyl cyclic 1-azadienes 1 could serve as suitable reaction partners, while 1m bearing ortho-methoxyphenyl gave moderate yield (67% yield) due to the steric hindrance (entry 12). Moreover, 2-naphthyl and 2-thienyl substituted substrates 1p and 1q exhibited good reactivities, delivering the desired products 3pa in 72% yield and 3qa in 74% yield, respectively (entries 15–16).

Substrate scope of cyclic 1-azadienes 1a

Entry	R1 in 1	3	Yieldb (%)	drc
1	2-FC6H4 (1b)	3ba	85	>20 : 1
2	3-FC6H4 (1c)	3ca	90	>20 : 1
3	4-FC6H4 (1d)	3da	85	>20 : 1
4	3-ClC6H4 (1e)	3ea	86	>20 : 1
5	4-ClC6H4 (1f)	3fa	80	>20 : 1
6	3-BrC6H4 (1g)	3ga	79	>20 : 1
7	4-BrC6H4 (1h)	3ha	80	>20 : 1
8	4-CNC6H4 (1i)	3ia	75	>20 : 1
9	2-MeC6H4 (1j)	3ja	79	>20 : 1
10	3-MeC6H4 (1k)	3ka	76	>20 : 1
11	4-MeC6H4 (1l)	3la	73	>20 : 1
12	2-OMeC6H4 (1m)	3ma	67	>20 : 1
13	3-OMeC6H4 (1n)	3na	75	>20 : 1
14	4-OMeC6H4 (1o)	3oa	77	>20 : 1
15	2-Naphthyl (1p)	3pa	72	>20 : 1
16	2-Thienyl (1q)	3qa	74	>20 : 1

Reactions were carried out with 1 (0.1 mmol), 2a (0.12 mmol), and DBU (20 mol%) in 2 mL of THF at rt for 12–48 h.

Isolated yields.

Determined by 1H NMR.

Subsequently, we performed the application of cyclic 1-azadiene 1a in DBU-catalyzed [3+2] cycloaddition with a variety of 3-homoacyl coumarins 2 under the optimal conditions (Table 3). And substrates 2 with electron-withdrawing (F, Cl, Br, Table 2, entries 1–5) or electron-donating (Me, MeO, entries 6–9) substituents at 6 or 7 position were all suitable for the cycloaddition, affording the cycloadducts 3aa–3ai in good to excellent yields of 78–94%. Replacing the R1 group with H, the desired product 3aj was obtained in 85% yield (entry 10). Notably, when the R2 were para-substituted phenyl groups, the cycloaddition reactions underwent smoothly to deliver the products 3ak–3am in up to 92% yield, and para-methyl substituted 2m gave a lower yield compared to para-electron-withdrawing substituted 2k and 2l (entries 11–13). The structure of product 3aj was confirmed by its X-ray crystallographic data.[12]

Substrate scope of 3-homoacyl coumarins 2a

Entry	R1/R2	3	Yieldb (%)	drc
1	6-Br/C6H5 (2a)	3aa	86	>20 : 1
2	6-F/C6H5 (2b)	3ab	88	>20 : 1
3	6-Cl/C6H5 (2c)	3ac	78	>20 : 1
4	7-Cl/C6H5 (2d)	3ad	84	>20 : 1
5	7-Br/C6H5 (2e)	3ae	86	>20 : 1
6	6-Me/C6H5 (2f)	3af	87	>20 : 1
7	6-OMe/C6H5 (2g)	3ag	94	>20 : 1
8	7-Me/C6H5 (2h)	3ah	86	>20 : 1
9	7-OMe/C6H5 (2i)	3ai	83	>20 : 1
10	H/C6H5 (2j)	3aj	83	>20 : 1
11	H/4-FC6H4 (2k)	3ak	90	>20 : 1
12	H/4-BrC6H4 (2l)	3al	92	>20 : 1
13	H/4-MeC6H4 (2m)	3am	80	>20 : 1

Reactions were carried out with 1a (0.1 mmol), 2 (0.12 mmol), and DBU (20 mol%) in 2 mL of THF at rt for 12–48 h.

Isolated yields.

Determined by 1H NMR.

To explore the asymmetric variant of this [3+2] cycloaddition reaction of 1a and 2j, a series of commercially available chiral amines were screened, and unfortunately, this reaction did not proceeded in CH2Cl2 and THF. However, when CH3CN was employed as a solvent, this reaction could be catalysed by a few of chiral amines, giving poor enantioselectivities and low to moderate yields (see ESI Table S1†). As shown in Table 3, cinchona catalyst C1 catalyzed the reaction to afford the 3aj in 46% yield with the highest 27.3% ee, and the reaction could be catalyzed by diimidazole catalyst C16 to give the highest 73% yield but poor 11% ee. The subsequent attempts to find the optimal asymmetric reaction conditions failed (Table 4).

Screening of enantioselective reaction conditionsa

Entry	Catalyst	Time (h)	Yieldb (%)	drc	eed (%)
1	C1	120	46	>20 : 1	27.3
2	C2	120	52.4	>20 : 1	9.5
3	C3	120	60	>20 : 1	8
4	C9	120	31	>20 : 1	4.5
5	C16	120	73	>20 : 1	11
6	C18	120	36	>20 : 1	11

Reactions were carried out with 1a (0.1 mmol), 2j (0.12 mmol), and 20 mol% catalyst in 2 mL of CH3CN at rt.

Isolated yields.

Determined by 1H NMR.

Determined by HPLC analysis.

On the basis of the results and previous literature,[6] herein we proposed a plausible mechanism for the [3+2] cycloaddition reaction (Scheme 2), which proceeded via stepwise mechanism with zwitterion.[13] Firstly, 1a is deprotonated to deliver the dienolate intermediate A under basic conditions. Subsequently, the α-carbanion of A attracks the olefinic bond of 2a to form the anion B. Then through cyclization and protonation, the final [3+2] cycloaddition product 3aa is given.

Scheme 2

Plausible reaction mechanism.

As shown in Scheme 3, the saccharin-derived cyclic 1-azadiene 4 was tested under the optimized reaction conditions. Delightfully, the [3+2] cycloadduct 5 could also be easily prepared in 85% yield and >20 : 1 dr. To explore the synthetic utility of this cycloaddition, a gram scale reaction was carried out to obtain the desired cycloadduct 3aa without any loss of yield and diastereoselectivity. The lactone of 3aa was opened under basic condition to give the multisubstituted cyclopentane 6 in 79% yield and >20 : 1 dr (Scheme 3).

Scheme 3

[3+2] cycloaddition of 4 and 2a, gram-scale reaction and further transformation.

Conclusions

In summary, we have successfully developed a DBU catalyzed [3+2] cycloaddition reaction of 3-homoacyl coumarins with cyclic 1-azadienes. The present protocol offers an efficient methodology to synthesize cyclopentane-fused coumarin derivatives with complete chemoselectivity and excellent diastereoselectivity in good yields. Efforts on further investigations of this protocol are underway in our group.

Conflicts of interest

There are no conflicts to declare.