Asymmetric Reduction of 3-Ketoproline Ethyl Ester by Modified Borohydrides and Various Vegetables

ECD spectrum and optical rotation were recorded on a spectropolarimeter (JASCO J-720WI) and Anton-Paar MCP200 Polarimeter. High-resolution mass spectra were obtained from Mass Spectrometry Instruments (MSI), Autoconcept-HRMS with MACH 3Xe data system. The ¹H and ¹³C NMR spectra were registered in CDCl₃ with Joel Resonance ECZ400S [400 MHz (¹H) and 100 MHz (¹³C)] using TMS as the internal standard. Preparative thin layer chromatography(PTLC) was done using silica gel 60 F₂₅₄ pre-coated glass sheets (layer thickness 2 mm, Merck). Analytical TLC was performed on silica gel 60 F₂₅₄, Merck (layer thickness 0.25 mm, Merck) and visualized with UV light and KMnO₄ as the detecting agent. Enantiomeric excesses were determined by HPLC analysis on Agilent 1260 Infinity Quaternary LC equipped with a UV detector and a chiral column Chiralpak AS-H (4.5 mm × 250 mm). (±)-N-boc-3-ketoproline ethyl ester (1) were prepared from ethyl glycine hydrochloride and ethyl acrylate according to the procedure previously described by Williams at al.³⁴. The enantiomeric mixture of 1 (%ee 2.6:1) was determined by chiral column Chiralpak AS-H.

A solution of substrate 1 (100 mg) in methanol, in the presence or absence of anhydrous CaCl₂, MgCl₂, NiCl₂ and AlCl₃ (2 eq) was stirred at room temperature for 30 min. The reaction mixture was cooled at 0 °C, and NaBH₄ (1.2 eq) was slowly added. After being stirred for another 30 min, the bulk of methanol was removed in vacuo. The resulting mixture was poured into saturated NH₄Cl solution and extracted with ethyl acetate (50 mL × 3). The organic layer was dried over anhydride Na₂SO₄ and concentrated in vacuo. Purification of the product by preparative TLC (eluted with n-hexane-ethyl acetate 8:2) provide a mixture of cis- and trans-3-hydroxyproline ethyl ester as summarized in Table 1. Identification and semi-preparative isolation of the enantio-products of alcohols 1a, 1b, 1c and 1d were performed on the chiral column (Chiralpak AS-H). The solvent system was n-hexane : ethyl acetate : methanol (60:35:5), a flow rate of 0.5 mL/min and wavelength of 280 nm. Enantiomeric excess (ee) and diastereomeric excess (de) were determined directly chiral column HPLC, while its absolute configurations were established using CD spectra.

Ketoproline (1): Pale yellow oil, ¹H NMR (400 MHz, CDCl₃, 25 °C): δ (enantiomeric mixtures; 2.6:1) = 4.30 and 4.25 (s, 1H), 4.02 (q, J = 7.2 MHz, 2H), 3.66 (m, 1H), 3.58 (m, 1H), 2.49 (m, 2H), 1.28 and 1.21 (s, 9H) and 1.09 (t, J = 6.8 MHz, 3H). ¹³C NMR (100 MHz, CDCl₃): δ (enantiomeric mixtures; 2.6:1) = 204.7/204.2, 166.3/166.1, 154.1/158.8, 81.0, 65.8/65.4, 62.1, 42.2/41.6, 37.1/36.4, 27.8 and 14.2.

Hydroxyproline (1a): Colorless oil, [α]²⁰_D = −31.59 (c = 0.38 μM, MeOH); ¹H NMR (400 MHz, Acetone-d₆, 25 °C): δ 4.59 (br s, 1H), 4.32 (d, J = 6.8 MHz, 1H), 4.21 (q, J = 6.4 MHz, 2H), 3.63 (m, 1H), 3.48 (m, 1H), 2.09 (m, 1H), 1.99 (m, 1H), 1.40 (s, 9H) and 1.29 (t, J = 6.8 MHz, 3H). ¹³C NMR (100 MHz, Acetone-d₆): δ 170.2, 153.7, 79.1, 71.9, 63.9, 60.2, 43.8, 31.9, 27.7 and 14.0. CD (c = 0.38 μM, MeOH) nm [Δε]: 212 [−0.90], 238 [−0.27], 259 [−0.03], 270 [−0.08], 289 [0.08], 304 [−0.08] and 317 [0.03].

Hydroxyproline (1b): Colorless oil, [α]²⁰_D = +10.35 (c = 0.38 μM, MeOH); ¹H NMR (400 MHz, Acetone-d₆, 25 °C): δ 4.43 (br s, 1H), 4.26 (s, 1H), 4.17 (q, J = 7.6 MHz, 2H), 3.60 (m, 2H), 2.11 (m, 1H), 1.95 (m, 1H), 1.45 (s, 9H) and 1.28 (t, J = 6.8 MHz, 3H). ¹³C NMR (100 MHz, Acetone- d₆): δ 170.8, 153.7, 78.8, 74.4, 68.1, 61.4, 44.6, 32.7, 28.5, and 14.2.

Hydroxyproline (1c): Colorless oil, [α]²⁰_D = +10.53 (c = 0.38 μM, MeOH); ¹H NMR (400 MHz, Acetone-d₆, 25°C): δ 4.57 (br s, 1H), 4.39 (d, J = 6.4 MHz, 1H), 4.23 (q, J = 6.4 MHz, 2H), 3.62 (m, 1H), 3.46 (m, 1H), 2.24 (m, 1H), 1.17 (m, 1H), 1.44 (s, 9H) and 1.27 (t, J = 6.8 MHz, 3H). ¹³C NMR (100 MHz, Acetone-d₆): δ 170.0, 154.0, 79.0, 71.0, 63.5, 60.1, 44.2, 32.5, 27.8 and 13.9.

Hydroxyproline (1d): Colorless oil, [α]²⁰_D = −21.06 (c = 0.38 μM, MeOH); ¹H NMR (400 MHz, Acetone-d₆, 25 °C): δ 4.43 (br s, 1H), 4.18 (q, J = 7.6 MHz, 2H), 4.15 (s, 1H), 3.63 (m, 2H), 2.08 (m, 1H), 1.92 (m, 1H), 1.40 (s, 9H) and 1.27 (t, J = 6.8 MHz, 3H). ¹³C NMR (100 MHz, Acetone-d₆): δ 171.4, 153.5, 79.1, 71.9, 63.9, 60.2, 43.8, 31.9, 27.7 and 14.4. CD (c=0.38 μM, MeOH) nm [Δε]: 216 [−0.8], 258 [−0.25], 288 [−0.09], 302 [−0.16] and 322 [0.07].

Fresh sample of vegetables were rinsed with 5% sodium hypochlorite solution and sterile distilled water, and cut into small pieces (approx. 1 cm long slice) with a sterile knife. In separate experiments, substrate 1 (100 mg) was added to the suspension of vegetable slices (10 g) in 70 mL of distilled water. The mixture was treated by shaking (120 rpm) at room temperature. Finally, the suspension was filtered off, and the vegetables slices were washed with distilled water. The aqueous solution was then extracted with ethyl acetate (3 × 50 mL), and the organic phase was dried with anhydrous Na₂SO₄ and evaporated in vacuo. The resulting residue was applied to a chiral column HPLC (Chiralpak AS-H). The solvent system was n-hexane : ethyl acetate : methanol (60:35:5), a flow rate of 0.5 mL/min and wavelength of 280 nm. Enantiomeric excess (ee) and diastereomeric excess (de) were determined directly from the areas under the curve of the alcohols 1a, 1b, 1c and 1d as summarized in Table 2.

It is known that aldehydes and ketones are rapidly reduced by NaBH₄. In addition, combinations with metal salts have also been reported increasing the yield and selectivity of reduced products of various kinds of prochiral ketones.³⁵ Nevertheless, this method provides a simple inexpensive alternative procedure, which proceeds under mild condition, neutral pH and short reaction time.³⁶ Thus, in this work, we had used NaBH₄ alone and together with several metal salts (CaCl₂, MgCl₂, NiCl₂ and AlBr₃) to reduce (±)-N-boc-3-ketoproline ethyl ester (1) that would lead to the desired product that will be used as precursor in the synthesis of lactacystin analogues.

Initially, ketoproline 1 in methanol was treated with 1.2 molar equivalent NaBH₄, or with 2 molar equivalent of metal salts; CaCl₂, MgCl₂, NiCl₂ and AlBr₃ followed by 2 molar equivalent NaBH₄ reagent. After work-up, the mixtures of alcohol products (Scheme 1) were purified by column chromatography, and followed by chiral column HPLC to determine the ratio of % ee and absolute configuration of the alcohol products (Table 1).

Scheme1.

Reduction products of (1).

The structure of all alcohol products 1a-1d have been interpreted based on the ¹H and ¹³C NMR spectral data, while the configuration of protons H-2 and H-3 at the chiral carbons of alcohols 1a/1c and alcohols 1b/1d as cis and trans, respectively, have been confirmed from the NOE correlation. Moreover, further study using Electronic Circular Dichroism (ECD) spectra data established that configurations of the chiral carbon C-2 in alcohols 1a and 1d were S and R, respectively (Fig. 2). With respect to the ECD spectra and optical rotation data of alcohols 1a and 1d, it could be concluded that configurations of alcohols 1b and 1c were 2R and 2S, respectively.

Figure2.

Experimental CD and UV Spectral of 1a and 1d.

Reduction of 1 with NaBH₄ produced 78.7% of the cis-alcohol and 6.4% of the trans-alcohol in mixtures of enantiomers (64.9% ee and 62.6% ee, respectively) (Table 1). Adding metal chlorides CaCl₂ and MgCl₂ increased the selectivity of reduction towards only the cis-alcohol with 100% de. In the presence of metal chloride NiCl₂, the proportion of the cis-alcohol increased from 92.8% de to 95.5% de. Nevertheless, with metal bromide AlBr₃, the reaction did not go towards completion (cis-alcohol 7.3% yield, trans-alcohol 1.2% yield), and increasing the amount of reagent or prolonging the reaction time had no effect. In this study, using the combination NaBH₄-metal salt decreased the yield of the alcohol products compared to using NaBH₄ alone. Besides that, both (R)- and (S)-enantiomers of 1 reduced to their corresponding cis-alcohols with the same feasibility, as % ee of products were consistent with the enantiomeric ratio of the starting material 1 (ee 2.6:1).

Steric hindrance of the C-3 keto groups in (S)- and (R)-enantiomers of 1 by the C-2 ester group led to Si-face and Re-face attack by NaBH₄ (Fig. 3); consequently, the cis-alcohol (2S,3R)-1a (92.8% de) and (2R,3S)-1c (92.1% de) were formed, respectively (Scheme 1). This effect was strengthened by the addition of metal salts, as the cis-alcohol product was formed in a higher de. However, the opposite result was observed for the combination of NaBH₄ with metal salt AlBr₃ (1a 87.4% de and 1c 83.6% de), in which the selectivity was lower than that when used NaBH₄ alone. Based on its ionic radius, the results of this study displayed that, the reduction of 3-ketoproline ethyl ester (1) with NaBH₄ in the presence of metal chloride did not show a direct correlation between ionic radius and the diastereoselectivity. This is contrary to the results previously described by Taniguchi et al.,³⁷ in which the selectivity reduction decreased along with the increasing of ionic radius. However, the result was similar to that reported by Fraga et al.,²⁷ on the reduction of cyclic β-ketoester. This might due to the different properties of cyclic β-ketoester and straight-chain β-ketoester. The ionic radius of metal and the ratio of the products (cis/trans) were as follows: Ca²⁺ 1.14 Å (100%); Mg²⁺ 0.86 Å (100%); Ni²⁺ 0.74 Å (95.8/4.2); Al³⁺ 0.67 Å (87.4/12.6). The % de and % ee of the reduction products of (±)-1 are shown in Table 1.

Figure3.

The mechanism of action of NaBH₄ (upper) and NaBH₄-Metal salt (bottom) in reducing ketoproline (1).

Whole plant tissue is an excellent choice to substituted isolated enzymes, since the oxido-reductase, cofactor and the cofactor regeneration systems are all within the cells.^38,39 Thus reductants being eco-friendly environmental reducing agents, accessible in all seasons and have the advantage of a being low cost. Therefore, to find an ideal biocatalyst for asymmetric reduction reaction of the (±)-N-boc-3-ketoproline ethyl ester (1), several vegetables as summarized in Table 2 were applied as the biocatalysts.

As shown in Table 2, the incubation of ketoproline 1 with fresh vegetables in water at room temperature gave its corresponding alcohol (Scheme 1) with high product enantioselectivity. In particular, the (S)-enantiomer of 1 was completely reduced affording a single diastereomer, cis-alcohol 1a (100% de), whilst the (R)-enantiomer was reduced to a mixture of diastereomers, cis-alcohol 1c (> 90% de) and trans-alcohol 1d (< 10% de). The absence of trans-alcohol 1b in the former reaction product indicates that the vegetables catalytically reduced (±)-1 in a diastereospecific manner. Furthermore, the low conversion of the (R)-enantiomer shows that those biocatalysts had good enantioselectivity towards the racemic substrates, which the (S)-enantiomer was reduced before the (R)-enantiomer. This case is similar to the reduction of two β-ketoester racemates (ethyl 2-oxo-1-cyclopentane-carboxylate and ethyl 2-oxo-1-cyclohexane-carboxylate) by Daucus carota (carrot), in which one of the enantiomers was reduced faster than the other.^24,40 Interestingly in the reduction of (R)-enantiomer via carrot afforded the trans-alcohol 1d (97.9% de) as the major product and cis-alcohol 1c (2.1% de) as the minor.

The transfer of a hydride (Re- or Si-faces) in the formation of (R)- and (S)-alcohols depends on the orientation of the substrate binds to the enzyme, while enzyme (dehydrogenase) transfers either pro-(R)-hydride or pro-(S)-hydride of the coenzymes (NADPH) depending on the kind of the enzyme.⁴¹ Every plant species have the ability to produce several enzymes, which are able to distinguish between small and large groups, thus hydride is delivered from either the Re- or Si-face in an enantiomeric way.

The formation of diastereomer (2S,3R)-1a in high de indicated that the dehydrogenase pro-(R)-hydride in our selected vegetables was more dominant than its pro-(S)-hydride. Therefore, the dehydrogenase delivered the hydride exclusively from the Si-face of the carbonyl group of both enantiomers to give the (R)-alcohol. However, the selectivity of the dehydrogenase also depends on the size of the carbon chains attached to the carbonyl group.⁴² Fig. 4 shows the possible mechanism action of transferring a hydride from NADPH to the carbonyl group of 1. This mechanism mimics the (R)-2-hydroxyglutarate dehydrogenase catalysis reduction of 2-oxoglutarate to (R)-2-hydroxyglutarate as reported by Martins et al.⁴³ Based on this mechanism, the large steric hindrance around the carbonyl group of the (S)-enantiomer of 1 only allows the hydride to attack from the Si-face, thus only a single diastereomer which is cis-alcohol 1a was detected. On the other hand, the less steric hindrance around the carbonyl group of the (R)-enantiomer of 1 allows the hydride to attack from both the Re- and Si-faces to give the mixture of cis- and trans-alcohols, with the attack from the Re-face is more favorable than that from the Si-face. However, in the reduction of (R)-enantiomer of 1 using carrots, the attack from the Si-face is more favorable and trans-alcohol is found to be the major product. This may be due to that the main hydride from the enzyme in the carrot is the dehydrogenase pro-(R)-hydride.

Figure4.

Mechanism of an NADH-depending dehydrogenase transferring a hydride between substrate and coenzyme.²⁸

In this study, the selectivity reduction of all selected vegetables is used very high. Nevertheless, the long reduction times and low product yields are unfavorable and need to be optimized. In this case, Utsukihara et al.⁴⁴ has described that the problem using plant culture-cells as biocatalysts is that they usually grow very slow and require a large amount of the catalyst. However, according to Matsuo et al.⁴⁵ the addition of sucrose can shorten the reduction time; unfortunately this did not affect the chemical yield and enantioselectivity. The best result in this study was shown by carrot (94.5% yield in two days) and jicama (64.8% yield in three days), while the moderate results were displayed by ginger (56.1% yield in two days), bitter gourd (58.9% yield in seven days) and garlic (59.2% yield in seven days).