Highly Enantioselective Catalytic Synthesis of Functionalized Chiral Diazoacetoacetates

The Michael reaction is one of the most general and versatile methods for carbon-carbon bond formation,¹ and its Mukaiyama-Michael variant provides an efficient strategy for the addition of silyl enol ethers to α,β-unsaturated carbonyl compounds.² Catalytic asymmetric reactions with broad variations in α,β-unsaturated carbonyl compounds and chiral catalyst (Lewis acid and Brønsted acid) are well documented,^3,4 and the enantioenriched 1,5-dicarbonyl compounds formed from these reactions have proven to be useful building blocks. However, there has been limited variation in the silyl enol ethers used in these reactions, and none of them have incorporated multiple functional groups.

We have recently reported condensation reactions of methyl 3-(trialkylsilanoxy)-2-diazo-3-butenoates (1) in Mukaiyama-aldol,⁵ Mukaiyama-Michael,⁶ and Mannich⁵ processes (Scheme 1) in our efforts to construct functionalized diazo compounds. These reactions are especially facile owing to the stabilization afforded by the diazo functional group to the intermediate formed by electrophilic addition (E⁺ + 1 → 5). The resulting multifunctional diazoacetoacetates have proven to be valuable building blocks for the efficient synthesis of functionally complex organic compounds.^5,7 However, attempts to construct chiral multifunctional diazoacetoacetates have been only moderately successful with the only example being the asymmetric catalytic Mukaiyama-aldol reactions of a limited array of aromatic aldehydes with 1 catalyzed by AgF/(R)-BINAP.⁸ We now report the first examples of a broadly applicable, highly enantioselective synthesis of chiral γ-functionalized diazoacetoacetates by catalytic Mukaiyama-Michael addition reactions of 3-(tert- butyldimethylsilyloxy)-2-diazo-3- butenoate 1.

Scheme 1.

Diazoacetoacetate synthesis via condensation reactions of 1.

A survey of chiral Lewis acids for the direct Mukaiya–maaldol or Mukaiyama Michael reactions of 1 with α,β-unsaturated carbonyl compounds showed limited reactivity and low enantioselectivity. The success of Evans’ N-oxazolidinone derivatized α,β-unsaturated carbonyl compounds in chiral Lewis acid catalyzed asymmetric reactions⁹ prompted us to use 6, but even with copper(II) triflate and chiral box and pybox ligands no reaction between 6 and 1 was observed. Since the oxazolidinone basicity of 6 was too strong to effect activation of the α,β-unsaturated carbonyl unit for electrophilic addition, we turned to the less basic α,β-unsaturated 2-acyl imidazole 7,¹⁰ and in a reaction with 1 catalyzed by copper(II) triflate ligated with (S,S)-^tBu-box L₁ (Table 1, entry 5), the Mukaiyama-Michael condensation product 8 was formed in 66% yield but with only 10% ee. In a screening of potential Lewis acids (Table 1), scandium(III) triflate, a preferred catalyst for Mukaiyama-aldol reactions,^2c,11 was ineffective for addition to 1a (entry 1). In contrast, the mild Lewis acids, Ni(OTf)₂, Zn(OTf)₂, and Mg(OTf)₂, combined with L₁, offered moderate enantioselectivity with moderate to low product yields (entries 2~4), but Cu(SbF₆)₂¹² proved to be the most active and effective, giving the desired product in 77% yield with 46% ee (entry 7). By reducing the temperature to −78°C with this copper(II) catalyst, enantioselectvity was improved to 54% ee (entry 8). Because Cu(SbF₆)₂ in combination with L₁ exhibited the highest reactivity in these reactions, this catalytic system was selected for further elaboration.

Table 1.

Selection of Lewis acid for the enantioselective Mukaiyama-Michael addition.^[a]

entry	Lewis acid MXn	yield (%)[b]	ee (%)[c]
1	Sc(OTf)3	<5	0
2	Ni(OTf)2	30	54
3	Zn(OTf)2	42	39
4	Mg(OTf)2	26	47
5	Cu(OTf)2	66	10
6	CuOTf	31	26
7[d]	Cu(SbF6)2	77	46
8[d]	Cu(SbF6)2	43	54

^[a]Reactions were performed with 0.25 mmol 7a, Lewis acid (10 mol%) and L₁ (12 mol%) in 1.5 mL of DCM, 1.5 equiv of 1a in 0.5 mL of DCM was added over 30 min to the reaction mixture at 0°C (except entry 8 which occurred at −78°C) under a N₂ atmosphere. The reaction solution was stirred overnight at 0°C.

^[b]Isolated yield of 8 after chromatography.

^[c]Determined by chiral HPLC (AD-H, Hexane/i-PrOH = 50:50, flow rate 1.0 mL/min, 254 nm, t_r1= 9.0 min, t_r2 =11.1 min).

^[d]Cu(SbF₆)₂ was formed in situ from CuCl₂ and AgSbF₆ under a N₂ atmosphere (ref. 12).

Optimization of this Mukaiyama-Michael transformation was effected on 7a by initially changing the ester alkyl group and silyl ether group of 1(1a–1d). Compared to the TBS group of 1a (Table 2, entry 1), the TMS analog 1b exhibited higher reactivity but had a much lower enantiomeric excess (entry 2). However, although no significant change in %ee was observed with the tert-butyl ester of 1c, a dramatic improvement in enantioselectivity was achieved when the ester alkyl group was changed from methyl to benzyl 1d; % ee improved to 93% (entry 4), and this vinyldiazoester was used in efforts to achieve further optimization.

Table 2.

Optimization of reactant, ligand, and reaction conditions for the enantioselective Mukaiyama-Michael addition.^[a]

entry	1	ligand	solvent	additive	yield (%)[b]	ee (%)[c]
1	1a	L1	DCM	-	43	54
2	1b	L1	DCM	-	64	12
3	1c	L1	DCM	-	33	50
4	1d	L1	DCM	-	42	93
5	1d	L2	DCM	-	30	65
6	1d	L3	DCM	-	33	76
7	1d	L4	DCM	-	40	87
8	1d	L5	DCM	-	34	88
9	1d	L6	DCM	-	<5	ND
10	1d	L1	DCM	HFIP (1.0 eq)	67	83
11	1d	L1	DCM	4Å (0.1g)	44	94
12	1d	L1	THF	4Å (0.1g)	<5	ND
13	1d	L1	Toluene	4Å (0.1g)	35	90
14[d]	1d	L1	DCM	HFIP (1.0 eq)	78	93
15[e]	1d	L1	DCM	HFIP & 4Å	81	94

^[a]Reactions were performed as described in Table 1. The chiral catalyst was prepared according to ref 12.

^[b]Isolated yield of 8 after chromatography.

^[c]Determined by chiral HPLC (See supporting information).

^[d]30 mol% Catalyst was used.

^[e]Catalyst was prepared in a glove box, and the reaction was run for three days.

A survey of chiral box and pybox ligands showed comparable reactivity with the box ligands L₁–L₃ as well as L₄ and L₅ (Table 2), but %ee values were considerably lower with L₂ and L₃, compared with L₁. Although L₄ and L₅ gave product yields and %ee values that were comparable to those of L₁, there was no obvious advantage to their use. Pybox L₆ exhibited very low reactivity, and the %ee value from the use of this ligand was not determined.

Changing the solvent from dichloromethane to THF completely shut down the reaction, but reaction in toluene afforded a %ee value comparable to that in DCM; however, the reaction rate was slower in toluene. With hexafluoroisopropyl alcohol (HFIP) as an additive^3b or using 30 mol% catalyst instead of 10 mol%, a significantly improved yield of 8 was obtained (up to 78% yield with 94% ee, entries 10 and 14 of Table 2), and the result was reproducable by adding 4Å molecular sieve (entry 11).¹³ Having established optimum conditions with [Cu(II) (S,S)-^t-Bu-box](SbF₆)₂, efforts were undertaken to reduce the amount of catalyst required to obtain high product yields: an 81% yield of 8 with 94% ee was obtained with 10 mol% catalyst, which was prepared in glove box and allowed to undergo reactionover a 3 day period (entry 15).

Using these optimized conditions, reactions with a diverse set of α, β-unsaturated 2-acyl imidazoles were examined with Michael donor 1d (Table 3). Aryl and alkyl substitutions all gave high yields and high to excellent enantioselectivity. Those with electron-donating substituents showed higher reactivity and selectivity compared to those with electron-withdrawing substituents (Table 3, entries 1~6 vs 7). α,β-Unsaturated 2-acyl imidazoles with aromatic heterocyclic and naphthyl substituents exhibited comparable reactivities and high ee values (entries 11 and 12). As expected from the electronic effects of aryl substituents, higher enantioselectivity was achieved with the meta-nitro-substituted 7h than with para-nitro-substituted 7g. Surprisingly, the %ee from the reaction with 7m (R = ^t-Bu, entry 13) was greater than that from 7n (R = cyclohexyl, entry 14)

Table 3.

Catalytic enantioselective Mukaiyama-Michael addition of vinyldiazoacetate 1d with representative Michael acceptors.^[a]

entry	R1 (7)	product 8	yield (%)[b]	ee (%)[c]
1	4-ClC6H4 (7a)	8a	81	94
2	4-BrC6H4 (7b)	8b	83	91
3	4-FC6H4 (7c)	8c	76	93
4	4-MeC6H4 (7d)	8d	77	91
5	C6H5 (7e)	8e	75	91
6	4-MeOC6H4 (7f)	8f	88	96
7	4-NO2C6H4 (7g)	8g	62	86
8	3- NO2C6H4 (7h)	8h	68	94
9	2-ClC6H4 (7i)	8i	75	95
10	2,6-2ClC6H3 (7j)	8j	65	91
11	2-furanyl (7k)	8k	80	80
12	2-naphthyl (7l)	8l	72	91
13	t-Bu (7m)	8m	62	95
14	cyclohexyl (7n)	8n	79	81

^[a] Reactions were carried out on a 0.25 mmol scale in DCM with 1.0 equiv. of HFIP, 4Å molecular sieve (0.1 g) and 10 mol% of catalyst which was prepared in situ according to ref 12. 1.5 equiv of 1d in 0.5 mL of DCM was added over 30 min to the reaction mixture at −78°C under a N₂ atmosphere in a dry ice-acetone bath. The reaction solution was stirred for three days at this temperature.

^[b]Isolated yield of 8 after chromatography.

^[c]Determined by chiral HPLC (See supporting information).

The absolute configuration of the generated stereocenter in 8 was determined by converting the Michael addition product (Scheme 2) to a reported chiral diester having a known absolute configuration that had been formed by desymmetrization of the substituted glutaric anhydride with chiral oxazolidinones.¹⁴ Cleavage of the diazoacetoacetate to the diazoacetate and carboxylic acid, a well-known and widely used transformation,¹⁵ followed by esterification of the resulting carboxylic acid with chiral (S)-1-(1-naphthyl)ethanol formed the β-substituted esters (9) in high yield without loss of chirality. Methylation of the imidazole functional group according to the reported procedure¹⁶ smoothly removed the imidazole and produced chiral diesters 10 in good yield. Comparing the NMR data of compound 10e with reported data for the known (1S,3S)-10e and (1S,3R)-10e^14a Confirmed that the product formed from the Mukaiyama-Michael reaction of 1d with 7e is indeed (5S)-8e.

Scheme 2.

Synthesis of chiral 3-substituted pentanedioic acid esters.

In summary, we have developed a catalytic, highly enantioselective Mukaiyama-Michael addition of 3-(trialkylsilanoxy)-2-diazo-3-butenoate to α,β-unsaturated 2-acyl imidazoles with a chiral copper(II) Lewis acid. This methodology offers access to a broad selection of highly functionalized chiral diazoacetoacetates that can be conveniently transformed to chiral diester compounds whose asymmetric center is chemically differentiated solely by different alkyl ester groups. The further utility of these Michael addition products is under investigation.

Experimental Section

The copper catalyst was prepared in glove box according to the Evans’ procedure:¹² CuCl₂ (0.025 mmol), and chiral ligand (0.030 mmol) in DCM (0.5 mL) were stirred for 2 hours in an oven-dried flask, then AgSbF₆ (0.050 mmol) in DCM (0.5 mL) was added dropwise, and this solution was stirred for another 3 hours in the absence of light. The resulting green catalyst suspension was filtered with cotton, and the solution was added to the oven-dried reaction flask, which contained 4Å molecular sieves (100 mg) and the Michael acceptor (0.25 mmol). The reaction flask was then sealed with a rubber stopper before being removed from the glove box. The temperature of the reaction was lowed to −78°C with dry ice-acetone bath, the additive (0.25 mmol) was introduced at this moment followed by dropwise addition of the diazo compound (0.38 mmol) in DCM (0.5 mL) by syringe. The reaction was carried out at this temperature for three days, then quenched with saturated NH₄Cl and purified by flash chromatography on silica gel (eluent: hexanes: EtOAc = 5:1 to 2:1) to give the pure products.