Combining silver catalysis and organocatalysis: a sequential Michael addition/hydroalkoxylation one-pot approach to annulated coumarins.

A highly stereoselective one-pot procedure for the synthesis of five-membered annulated hydroxycoumarins has been developed. By merging primary amine catalysis with silver catalysis, a series of functionalized coumarin derivatives were obtained in good yields (up to 91%) and good to excellent enantioselectivities (up to 99% ee) via a Michael addition/hydroalkoxylation reaction. Depending on the substituents on the enynone, the synthesis of annulated six-membered rings is also feasible.

Secondary metabolites from phytobiochemical pathways fulfill various life-sustaining roles in plants. For example, coumarins, which originate from the shikimic acid pathway, are vital for the regulation of oxidative stress, hormonal regulation, and plant protection (Figure 1).[1] Interestingly, biological activity is not limited to plants only, as shown by warfarin and phenprocoumon, which belong to the class of vitamin K antagonists. Both inhibit the enzyme vitamin K epoxide reductase, thus preventing blood clotting in humans and animals.[2] As a result, these anticoagulants have found wide application as pharmaceuticals or rodenticides over the years. Although the unprotected 4-hydroxyl group is necessary for anticoagulant activity, other biologically active natural products have been discovered in which the oxygen is embedded in an annulated ring structure, as found in coumestrol and frutinone A.[3,4] The former interacts with estrogen receptors ERα and -β in humans, while the latter is a potent inhibitor of CYP1A2.

Figure 1

Bioactive coumarin derivatives.

Recently, the combination of transition metals with organocatalysts has emerged as a versatile one-pot strategy for the synthesis of valuable chiral entities, especially in the context of sequential catalysis.[5] However, most reported procedures mainly rely on expensive metal complexes, such as gold, palladium, and iridium. Silver is a comparably cheaper metal and can be employed as an alternative to facilitate these sequential transformations. Silver salts of chiral organic molecules have been used as binary catalytic systems in many asymmetric transformations, but the sequential catalysis employing silver and organocatalysts is less explored.[6,7] Owing to the wide applicability of coumarin derivatives and knowing the potential of sequential catalysis,[8] we envisaged the combination of silver salts with chiral primary amines for the one-pot sequential Michael addition/hydroalkoxylation of 4-hydroxycoumarins 1 with enynones 2 (Scheme 1).

Scheme 1

Intended Strategy

Most of the organocatalytic asymmetric transformations involving 4-hydroxycoumarins focus on Michael additions to common electrophiles, such as simple enones, which undergo electrophilic activation in the presence of primary amines.[9,10] In contrast, enynones 2 have not been used in this context so far. These enynones are challenging Michael acceptors because both the β- and δ-position are prone to nucleophilic addition after electrophilic activation.

To achieve our goal, we started the investigation by optimizing the organocatalytic Michael addition of 4-hydroxycoumarin 1a to enynone 2a using cinchona-derived primary amines.[11] The reaction of 1a with 2a in CH2Cl2 at room temperature in the presence of 20 mol % 9-amino(9-deoxy)epi-quinine A and 40 mol % TFA afforded the desired product 3a within 16 h in 63% yield and 68% ee (Table 1, entry 1). To circumvent this low asymmetric induction, we replaced the catalyst A with primary amines derived from cinchonidine B, quinidine C, and cinchonine D, but no beneficial effect was observed (entries 2–4). In contrast, the choice of solvent had a noticeable effect on the yield and enantiomeric excess, revealing THF as the most suitable solvent (entry 6).

Table 1

Optimization Studies on the Michael Additiona

entry	cat.	additive	solvent	t (h)	yield (%)b	ee (%)c
1	A	TFA	DCM	16	63	68
2	B	TFA	DCM	20	74	59
3	C	TFA	DCM	19	61	–68
4	D	TFA	DCM	21	81	–60
5	A	TFA	CHCl3	24	84	72
6	A	TFA	THF	16	85	78
7	A	TFA	MTBE	16	82	66
8	A	(S)-mandelic acid	THF	16	85	78
9	A	(S)-N-Boc-ala	THF	24	95	82
10	A	(S)-N-Boc-phe	THF	24	65	81
11	A	(S)-N-Boc-leu	THF	24	94	80
12	A	(S)-N-Boc-val	THF	24	48	80
13d	A	(S)-N-Boc-ala	THF	4	94	76
14e	A	(S)-N-Boc-ala	THF	26	95	86
15f	A	(S)-N-Boc-ala	THF	96	88	57
16e,g	A	(S)-N-Boc-ala	THF	72	88	75

Reaction conditions: 0.5 mmol of 1a, 0.6 mmol of 2a, 20 mol % of catalyst, 40 mol % additive, 1.0 mL solvent, rt.

Yield of isolated 3a after flash column chromatography.

Enantiomeric excess was determined by HPLC analysis on a chiral stationary phase of the O-acetylated derivative 3a′.

Reaction was carried out at 50 °C.

Reaction was carried out at 4 °C.

Reaction was carried out at −16 °C.

Reaction was carried out with 0.1 mol % of A and 20 mol % (S)-N-Boc alanine.

We questioned whether the enantiomeric excess could be increased if chiral acidic additives, especially Boc-protected amino acids, were employed instead of TFA, as shown in a seminal publication by Melchiorre et al.[12] This would change the mechanism of activation and stereoinduction from pure iminium activation to a mixed activation mode in which the chiral protonated iminium ion is coordinated by a chiral anion, a concept known as asymmetric counteranion directed catalysis (ACDC).[13] As anticipated, we observed a slight increase in enantiomeric excess for the majority of chiral additives with comparable yields (entries 8–12). With (S)-N- Boc-alanine in hand as the best additive, we focused on the influence of the temperature on the reaction. Naturally, a higher temperature resulted in a faster but less selective reaction (entry 13), while at 4 °C a negligible increase in reaction time and yield was observed, albeit with better enantiomeric excess (entry 14). However, a lower temperature had a deleterious effect, leading to longer reaction times and lower enantioselectivities (entry 15). A similar impact was observed when the catalyst loading was decreased to 10 mol %; thus, 20 mol % had to be used (entry 16).

With the optimized conditions for the Michael addition in hand, we shifted our focus to the cycloisomerization reaction (Table 2). We envisioned that phosphine Au(I) catalysts, which are known to activate internal alkynes, would be an optimal choice for this reaction. However, the initial reaction conditions gave rise to a complex mixture of different products, most likely 6-endo-, 5-exo-, and other unidentified products (entry 1). In contrast, a number of Ag(I) salts gave the 5-exo product in excellent yields within 1 h in the absence of gold catalysts (entries 2–9). Ag2CO3 turned out to be the best catalyst, giving the desired product 4a in 97% yield within 40 min. In addition, we also tested other metal sources which act as carbophilic Lewis acids, but the reaction seemed to be limited to silver salts only (entries 9–10).

Table 2

Optimization of the Cycloisomerization of 4aa

entry	catalyst	solvent	t (min)	yield (%)b
1	PPh3AuCl/AgNTf2	toluene	60	–c
2	AgNTf2	toluene	30	79
3	AgNTf2	THF	>240	–c
4	AgNO3	toluene	40	91
5	Ag2CO3	toluene	40	97
6	AgOAc	toluene	40	94
7	AgOTf	toluene	50	91
8	AgSbF6	toluene	30	91
9	CuI	toluene	>1 d	traces
10	PtCl2	toluene	>1 d	traces
11	Ag2CO3	toluene/THF 4:1	4 h	94
12d	Ag2CO3	toluene	40	97

Reaction conditions: 0.13 mmol of 3a, 10 mol % of catalyst, 1.3 mL solvent, rt.

Yield of isolated 4a after flash column chromatography.

Complicated mixture of products which could not be separated.

In the presence of 20 mol % A and 40 mol % (S)-N-Boc alanine.

Regrettably, further studies revealed that THF, which is used during the Michael addition, is inappropriate for the subsequent cyclization because, similar to the initial reaction conditions, a mixture of products was obtained (entry 3). Thus, the reaction must be performed either in a mixture of toluene and THF (entry 11) or with the solvents changed prior to the addition of Ag2CO3. To compensate for this inconvenience, there seemed to be no notable deactivation of the silver catalyst in the presence of the amine catalyst (entry 12), as there was no notable decrease in yield or increase in reaction time when the reaction was performed in the presence of amine catalyst A and (S)-N-Boc alanine. This is a decisive advantage compared to gold-catalyzed reactions, in which the presence of free amines or basic moieties deactivate the gold catalyst, and strong acidic additives such as TFA or harsher reaction conditions have to be employed to retrieve the active gold species.[14]

With these optimized conditions in hand, we tested the substrate scope of the one-pot Michael addition/cycloisomerization (Table 3). In the case of aryl-substituted enynones good yields (54–91%) and excellent enantioselectivities were obtained (73–99% ee) irrespective of electronic and steric effects (4a–s), though bulky substituents normally resulted with an increased reaction time in the cyclization step. Hydroxycoumarins bearing different substituents were also tolerated (4m–s). In all cases with aryl substituents on the enynone the 5-exo-products were obtained, which can be verified by 4J-coupling of the olefinic proton (around −2 Hz) compared to the 3J-coupling of the endo-product (around 4 Hz). In contrast, enynones with aliphatic substituents led to the formation of 6-endo-products with comparable ee values but lower yields due to a less selective ring formation (5b–c). In the case of a terminal alkyne the 5-exo-product was obtained exclusively, albeit with slightly lower enantioselectivity values (4t). Interestingly, we did not observe isomerization of the 5-exo-products to furans under the applied reaction conditions.

Table 3

Substrate Scope for the Sequential Catalysisa

product	R1	R2	yield (%)b,c	ee (%)d,e
4a	H	Ph	84 [52]	88 [94]
4b	H	4-F-C6H4	76 [55]	89 [94]
4c	H	4-Br-C6H4	76 [56]	85 [89]
4d	H	4-F3C-C6H4	75	93
4e	H	2,3-CH2OCH2-C6H3	67 [47]	81 [98]
4f	H	3-Me-C6H4	80 [53]	89 [97]
4g	H	3-MeO-C6H4	82	93
4h	H	2-naphthyl	79	92
4i	H	2-Cl-C6H4	81	94
4j	H	1-naphthyl	75	80
4k	H	2-furanyl	78	99
4l	H	2-thienyl	76 [59]	77 [96]
4m	6-Me	4-Br-C6H4	54	92
4n	6,7-CH2OCH2-	4-F-C6H4	56	94
4o	7-MeO	1-naphthyl	74	94
4p	6-Cl	3-MeO-C6H4	60	98
4q	6-Cl	Ph	91	73
4r	7-MeO	Ph	58	93
4s	6,7-CH2OCH2-	Ph	72	83
4t	H	H	84	70
5b	H	butyl	52	90
5c	H	cyclopentyl	32	89

Reaction conditions: 0.8 mmol of enynone, 0.5 mmol of hydroxycoumarin, 20 mol % of catalyst, 40 mol % (S)-N-Boc alanine, 1.0 mL of THF, 4 °C, 24–48 h; after completion, removal of THF, addition of 5.0 mL of toluene, 10 mol % Ag2CO3, rt, 1–24 h.

Yield of isolated 4 or 5 after flash column chromatography.

In brackets, yield after one recrystallization from n-pentane/ethyl acetate.

Enantiomeric excess was determined by HPLC analysis on a chiral stationary phase.

In brackets, enantiomeric excess after one recrystallization from n-pentane/ethyl acetate.

The proposed structure of the products, including the absolute configuration, could be assigned by X-ray crystal structure analysis of (S)-4g (Figure 2).[15] To further demonstrate the practicability of our new protocol, we carried out a larger scale synthesis of 4g on a 4 mmol scale. We obtained the same yield (82%, 1.24 g) and a better stereoselectivity of 96% ee.

Figure 2

X-ray crystal structure of (S)-4g.

A plausible mechanism for the reported sequential catalysis is depicted in Scheme 2. Upon condensation with the primary amine A and interactions of two molecules of (S)-N-Boc alanine, the enynone 2 forms a LUMO-activated chiral iminium ion. Similar to recent DFT calculations by Melchiorre et al., the two molecules of (S)-N-Boc alanine should play a pivotal role in the reactivity and selectivity of this supramolecular catalytic assembly.[16] One of the counteranions will interact with the protonated quinuclidine moiety of the primary amine catalyst by hydrogen bonding, thus shielding the Si-face of the iminium ion. This represents the stereochemical defining element responsible for π-facial discrimination. The second counteranion acts as a mediator in a network of hydrogen bonds between the iminium proton and 4-hydroxy-coumarin. Thereby the nucleophile becomes activated while being set up to the Re-face for the subsequent attack on the iminium ion. The nucleophilic attack will yield intermediate 3 after hydrolysis, which will then enter the second catalytic cycle. This cycle is initiated by coordination of Ag(I) to the alkyne moiety and electrophilic activation that allows for the hydroalkoxylation of the triple bond by attack of the nucleophilic hydroxy group. Similar to Au(I)-catalyzed cycloisomerizations, the trans-specific addition should follow Markovnikov’s rule and electronic factors. Thus, depending on the substituent on the alkyne, 5-exo-dig and 6-endo-dig ring formations are observed (see Supporting Information for a more detailed explanation). The products are obtained after regeneration of the silver catalyst and proton transfer.

Scheme 2

Proposed Catalytic Mechanism

In conclusion, we have developed a convenient one-pot sequential Michael addition/hydroalkoxylation by merging silver catalysis with primary amine catalysts. The combination gives rise to pharmaceutically interesting annulated coumarins in good yields and excellent enantioselectivities. Further investigations on the application of sequential catalysis by silver catalysis and organocatalysis are in progress in our laboratories.