Organocatalytic Highly Enantioselective Synthesis of Secondary α-Hydroxyphosphonates

Abstract

The first organocatalytic cross aldol reaction of ketones and diethyl formylphosphonate hydrate has been realized by using readily available L-prolinamide as the catalyst. Secondary α-hydroxyphosphonates have been synthesized in high enantioselective (up to >99% ee) and good diastereoselectivity.

As close analogs of α-amino acids, αhydroxyphosphonate derivatives have been shown to be very important enzyme inhibitors.¹ For example, they are inhibitors of renin² or human immunodeficiency virus (HIV) protease and polymerase.³ They also show anti-virus⁴ and anti-cancer activities.⁵ Because of their important biological activities, achieving high enantioselectivity in the synthesis of α-hydroxyphosphonates has been the goal of organic chemists.⁶ Recently, we reported the first prolinecatalyzed asymmetric aldol reaction of α-ketophosphonates for the highly enantioselective synthesis of tertiary α-hydroxyphosphonates (Eq 1).⁷ However, from a biological point of view, secondary α-hydroxyphosphonates seem more significant, as all chiral natural amino acids are secondary amines. The optically enriched forms of secondary α-hydroxyphosphonates are mainly obtained through enzymatic methods,^6a such as kinetic resolution of racemic mixture by bacteria, fungi or

lipases⁸ or through asymmetric reduction of α-ketophosphonate with baker’s yeast or fungi.^6a,9 Only a few chemical methods are available,^6b,c, which include the asymmetric reduction of α-ketophosphonates,¹⁰ asymmetric oxidation of benzylphosphonates¹¹ and diastereoselective addition of dialkyl phosphites to aldehydes (phosphoaldol reaction).12a,b These methods are either not catalytic, use special reagents that are difficult to handle, or have very limited substrate scope. A catalytic method based on the phosphoaldol reaction was also reported,^12c-g but the enantioselectivities obtained were dependent on the substrates. Most recently, Chen and co-workers reoported a vanadium-catalyzed oxidative kinetic resolution for the high enantioselective synthesis of secondary α-hydroxyhosphonates;¹³ nonetheless, the disadvantage of this method is the sacrifice of 50% of the starting material.¹³ Thus, the development of a catalytic highly enantioselective method for the synthesis of secondary α-hydroxyphosphonates is warranted. Herein, we wish to report our preliminary results of a highly enantioselective synthesis of secondary α-hydroxyphosphonates via a prolinamide-catalyzed asymmetric aldol reaction.¹⁴

Synthesizing secondary α-hydroxyphosphonates by using our reported protocol⁷ would require formylphosphonate as the starting material (Eq 1, R¹ = H). However, although diethyl formylphosphonate is a known compound,¹⁵ it is unstable, and all our attempts to react this substance with acetone failed. Then we turned our attentions to its hydrate (4), because it was reported to be more stable and in equilibrium with its formyl form.¹⁶

By using acetone (5a) as the model compound, we screened some readily available L-proline-derivatives (Figure 1) as the catalyst for the cross aldol reaction of diethyl formylphosphonate hydrate (4). The results are summarized in Table 1.

Figure 1.

Catalysts screened for the cross aldol reaction.

Table 1.

Catalyst Screening and Reaction Condition Optimizations^a

entry	catalyst	solvent	temp (°C)	time (h)	yield (%)b	ee (%)c
1	1	acetone	rt	24	—	—
2	2	acetone	rt	24	46	72
3	3	acetone	rt	3	93	84
4	3	CH2Cl2	rt	10	76	84
5	3	DMSO	rt	10	45	79
6	3	DMF	rt	10	55	60
7	3	acetone	0	10	95	92
8d	3	acetone	0	24	95	94
9d	3	acetone	-20	24	62	94
10	ent-3	acetone	0	24	91	92e

^aUnless otherwise indicated, all reactions were performed with freshly prepared diethyl formylphosphonate hydrate (4, 0.5 mmol), acetone (0.2 mL), solvent (0.4 mL) and catalyst (10 mol %) at 0 ° C.

^bYield of isolated product after column chromatography.

^cThe enantiomeric excess was determined by chiral GC analysis with a Chiraldex GTA column.

^dWith 5 mol % of catalyst loading.

^e(R)-6a was obtained as the major enantiomer.

As shown in Table 1, although L-proline is a good catalyst for the cross aldol reaction of α-ketophosphonates,⁷ it failed to catalyze the aldol reaction of 4 (entry 1), presumably because 4 is incompatible with its acidity. In contrast, less acidic L-proline tetrazole (2) and L-prolinamide (3) proved to be good catalysts for the desired reaction. At 10 mol % catalyst loading and room temperature, the aldol product 6a was obtained in 72% (entry 2) and 84% ee (entry 3), respectively. The reaction conditions were further optimized for 3, as it is more reactive and enantioselective. Catalyst 3 is slightly less reactive in CH₂Cl₂ (entry 4), but the enantioselectivity maintains at the same level as in acetone. Other common solvents used for aldol reactions, such as DMSO (entry 5) and DMF (entry 6), proved to be less effective than excessive acetone (entry 3). Nevertheless, lowering the reaction temperature to 0 °C resulted in an increase of the enantioselectivity (to 92% ee, entry 7). It is interesting to note that the catalyst loading can be further reduced to 5 mol %, without affecting the enantioselectivity, although the reaction is a little bit slower (entry 8). Further dropping of the reaction temperature to -20 °C did not improve the enantioselectivity, instead has an adverse effect on the reactivity of the catalyst (entry 9). Similar reaction with D-prolinamide (ent-3) as the catalyst produced comparable results as those of L-prolinamide, except that the opposite enantiomer of the product was obtained as the major one (entry 10).

In order to determine the absolute configuration of the product, the carbonyl group in product 6a was reduced to the methylene group in two steps (Eq 2) to give compound 7, for which the absolute configuration is known.¹⁷ Compound 7 is dextrorotary, which indicates that the newly formed chiral center is S-configured.¹⁷

Further studies were carried out under the optimized conditions (0 °C, 5 mol % catalyst loading, and excessive ketone as solvent if possible) to understand the scope of this new reaction, and the results are summarized in Table 2.

Table 2.

Enantioselective Cross Aldol Reaction of Diethyl Formylphosphonate Hydrate (4) and Various Ketones.^a

entry	R1	R2	product/ yield (%)b	drc	ee (%)d
1	H	H	6a/95	—	94e
2	Me	H	6b/62f	—	93e
	H	Me	6b’ /20f	95:5g	99e
3	Et	H	6c/58f	—	97e
	H	Et	6c’/5f	>95:5g	99e
4	Cl	H	6d/65	—	90e
5	OH	H	6e/75	—	43h
6	OMe	H	6f/77	—	85
7	AcCH2	H	6g/60	—	86
8	—(CH2)2—		6h/ 94	95:5i	>99
9	—(CH2)3—		6i/82	65:35g	93(86)
10j	—(CH2)3—		6i/90k	65:35g	96(95)
11j	—CH2OCH2—		6j/85k	70:30g	97(99)
12j	—CH2SCH2—		6k/88k	70:30g	93(97)l

^aUnless otherwise indicated, all reactions were performed with freshly prepared diethyl formylphosphonate hydrate (4, 0.5 mmol), ketone (0.6 mL) and L-prolinamide (5 mol %) at 0 °C; the absolute configuration is tentatively assigned for 6b-k on the basis of the reaction mechanism.

^bYield of isolated product after column chromatography.

^cDetermined by ¹H NMR.

^dUnless otherwise noted, enantiomeric excess was determined by HPLC analyses with a Chiralpak AD-H column; values in parentheses are for the minor diastereomer.

^eThe enantiomeric excess was determined by chiral GC analysis with a Chiraldex GTA column.

^fYields of individual regioisomer as determined by ¹H NMR.

^ganti/syn ratio as determined by NMR.

^hWith a Chiralcel OJ-H column.

ⁱsyn/anti ratio as determined by NMR.

^jThe reaction was performed with 4 (0.5 mmol), ketone (0.2 mL), CH₂Cl₂ (0.5 mL) and L-prolinamide (10mol %) at -20 °C for 24 h.

^kCombined yield of two inseparable diastereomers.

^lWith a Chiralcel OD-H column.

Besides acetone (entry 1), other unsubstituted aliphatic ketones, such as 2-butanone (5b, entry 2) and 2-pentanone (5c, entry 3), also participate in this reaction. The reaction of 5b and 4 gives rise to a mixture of two inseparable regioisomers (6b and 6b’) in a ratio of 3:1. The major regioisomer (6b) is a kinetic product, which was obtained in 93% ee. The thermodynamic product (6b’) was obtained in a 90% de for anti product and the ee value for this diastereomer is 99% (entry 2). Similarly, the reaction 5c and 4 yielded these two products in about 12:1 ratio, with a 97% ee for the major kinetic product (6c), and a single anti diastereomer for the minor product (6c’) in high enantioselectivity (99% ee, entry 3). However, 3-pentanone reacts very slowly and leads to the decomposition of 4 (data not shown).

Substituted ketones may also be used in this reaction. For example, the aldol reaction of α-chloroacetone (5d) produces only the kinetic regioisomer in 65% yield and 90% ee (entry 4). In contrast, acetol (5e) generates very poor enantioselectivity (43% ee) of the product (entry 5), presumably due to the interference of the free hydroxy group. Once the hydroxy group is methylated, compound5f again generates excellent enantioselectivity for the product (6f, 85% ee, entry 6). Similar enantioselectivity (86% ee) was also obtained for the aldol product (6g) of 2,5-hexanedione.

Some cyclic ketones were also studied. Cyclopentanone yielded the aldol product (6h) in excellent yield (94%) and diastereoselectivity (95:5, entry 8). The major diastereomer, which was determined to be syn by NOE, was obtained essentially as a pure enantiomer (>99% ee). The reaction of cyclohexanone generates a 65:35 anti/syn mixture (6i) in excessive cyclohexanone with 93% and 86% ee for these two diastereomers, respectively (entry 9). Similar reaction using CH₂Cl₂ as solvent with 10 mol % of the catalyst (entry 10) led to improved enantioselectivities of these two diastereomers (96% and 95% ee, respectively). Under these conditions, 4-oxacyclohexanone (5j) and 4-thiacyclohexanone (5k) also perform better. A 70:30 anti/syn mixture was obtained for 6j, with 97% and 99% ee for the major and minor diastereomers, respectively (entry 11). The same anti/syn diastereoselectivity was also obtained for the product of 5k, with 93% and 97% ee, respectively (entry 12). As a potential application of this product, 6k may be readily desulfurized to give product 6l in good yield (Eq 3), with total retention of the stereochemistry. Product 6l can be viewed as the aldol product of 3-pentanone and 4; however, as aforementioned, such a direct aldol reaction is not feasible with the L-prolinamide catalysis at this moment.

To account for the formation of the S-configured α-hydroxyphosphonate as the major enantiomer, we propose a reaction mechanism as shown in Scheme 1, which is similar to other proline derivative-catalyzed cross aldol reactions.¹⁸ The re face attack is disfavored due to the unfavorable steric interaction between the large axial phosphonate group and the axial methyl group (Scheme 1, right). Thus, the si face attack is favored, which leads to the observed product (Scheme 1, left).

Scheme 1.

Proposed Transition State Structures

In summary, we have developed the first organocatalytic cross aldol reaction of ketones and diethyl formylphosphonate hydrate by using readily available L-prolinamide as the catalyst. Secondary α-hydroxyphosphonates have been synthesized in high enantioselectivity (up to >99% ee) and good diastereoselectivity.