Gas chromatography (GC) was performed using a GL Science GC-353 gas chromatograph (GL Science Inc., Japan) equipped with capillary columns (DB-WAX, 0.25 μm, 0.25 mm × 30 m, Agilent Technologies, USA; TC-1, 0.25 μm, 0.25 mm × 30 m, GL Science Inc.; CP-Chirasil-DEX CB, 0.25 μm, 0.25 mm ×

25 m, Varian Inc., USA; Gamma DEX 225, 0.25 μm, 0.25 mm × 30 m, Sigma-Aldrich Inc., USA). Ethyl pyruvate (Figure 1, 1a), diatomaceous earth (granular), sodium hydrogen L-glutamate monohydrate, magnesium sulfate heptahydrate, glycerol, and hipolypepton were purchased from Wako Pure Chemical Industries Ltd. (Japan). Bacto™ malt extract, Bacto™ yeast extract, Bacto™ brain heart infusion, and Bacto™ tryptic soy broth were purchased from Becton Dickinson and Co. (USA). Ethyl lactate (2a), ethyl 3-methyl-2-oxobutanoate (1f), ethyl 2-oxo-4-phenylbutanoate (1h), ethyl 2-hydroxy-4-phenylbutanoate (2h), and beef extract were purchased from Sigma-Aldrich Inc. Ethyl benzoylformate (1g), ethyl 2-methylacetoacetate (1j), and ethyl mandelate (2g) were obtained from Tokyo Chemical Industry, Co., Ltd. (Japan). Ethyl 2-oxobutanoate (1b), ethyl 2-oxopentanoate (1c), ethyl 2-oxohexanoate (1d), ethyl 2-oxoheptanoate (1e), 2-chlorobenzoylformamide (1i), 2-chloromandelamide (2i), α-hydroxy esters (2b-f), and ethyl 3-hydroxy-2-methylbutanoate (2j) were prepared as described previously [4] [7] [19]. All other chemicals used in this study were of analytical grade and commercially available.

Agromyces ramosus NBRC13899, Agromyces cerinus subsp. cerinus NBRC15780, Agromyces luteolus NBRC16235, Agromyces rhizospherae NBRC16236, Agromyces brachium NBRC16238, Agromyces albus NBRC103057, Agromyces iriomotensis NBRC106452, Agromyces subtropicus NBRC106454, Agromyces marinus NBRC109019, Agromyces soli NBRC109063, Agromyces flavus NBRC109064, Agromyces tropicus NBRC109073, Agromyces ulmi NBRC109083, Agromyces neolithicus NBRC109084, Agromyces humatus NBRC109085, Gordonia amarae NBRC15530, Gordonia hydrophobica NBRC16057, Gordonia rhizosphera NBRC16068, Gordonia polysprenivorans NBRC16320, Gordonia alkanivorans NBRC16433, Gordonia terrae NBRC100016, Gordonia namibiensis NBRC108229, Gordonia sihwensis NBRC108236, Gordonia westaflica NBRC108237, Gordonia paraffinivorans NBRC108238, Gordonia soli NBRC108243, Gordonia malaquae NBRC108250, Gordonia alkaliphila NBRC109776, Gordonia jinhuaensis NBRC110001, Gordonia defluvi NBRC110695, and Gordonia sinesedis NBRC110696 were purchased from the National Institute of Technology and Evaluation, Biological Resource Center (NBRC; Japan). These strains were maintained at 25˚C in NBRC-recommended media (227, 230, 253, and 802) solidified with 1.5% (w/v) agar. The 227 medium contained 4.0 g of Bacto™ yeast extract, 10.0 g of Bacto™ malt extract, and 4.0 g of D-glucose per liter of distilled water (pH 7.3). The 230 medium contained 30.0 g of Bacto™ tryptic soy broth per liter of distilled water (pH 7.3). The 253 medium contained 37.0 g of Bacto™ brain heart infusion per liter of distilled water (pH 7.3). The 802 medium contained 2.0 g of hipolypepton, 2.0 g of Bacto™ yeast extract, and 1.0 g of magnesium sulfate heptahydrate per liter of distilled water (pH 7.0). The 15 Agromyces strains were grown in 227, 230, 253, and 802 media for 8 days at 25˚C with aerobic shaking in baffled flasks in the dark, and the 16 Gordonia strains were grown in 227, 230, 231, and 802 media for 8 days at 25˚C with aerobic shaking in baffled flasks in the dark. The actinomycetes were harvested by filtration on filter paper (Whatman No. 4) or glass fiber filters (ADVANTEC GA-55) in vacuo and washed with saline (0.85% NaCl aq.). The harvested cells were immediately used for reduction after washing with the saline.

Wet saline-washed wet actinomycete cells (0.5 g, dry weight approximately 0.15 g) were resuspended in a test tube (f 30 mm × 200 mm) containing 20 mL of saline. The substrate (0.15 mmol; 7.5 mM) and additive (5.0 mmol; 250 mM) were then added, and the reaction mixture was incubated aerobically (with reciprocated shaking at 120 rpm) at 25˚C. A portion (0.5 mL) of the mixture was applied to a short diatomaceous earth column (f 10 mm × 30 mm), extracted with diethyl ether (5.0 mL), and then concentrated under reduced pressure.

The production of alcohols (Figure 1, 2a-j) was measured using GC with a DB-WAX capillary column (100 kPa He at 110˚C: 1a, 3.78 min; 2a, 4.75 min; 1b, 4.73 min; 2b, 5.92 min; 1f, 4.54 min; 2f, 6.41 min; 120˚C: 1c, 4.84 min; 2c, 6.45 min; 1j, 5.54 min; 2j-anti from, 7.62 min; 2j-syn form, 8.13 min; 150˚C: 1d, 3.83 min; 2d, 4.68 min; 1e, 4.78 min; 2e, 6.07 min; 180˚C: 1g, 9.01 min; 2g, 12.08 min) or a TC-1 capillary column (100 kPa He at 140˚C: 1h, 10.02 min; 2h, 10.96 min; 170˚C: 1i, 6.85 min; 2i, 8.34 min). The enantiomeric excess (e.e.) of the product was measured using a GC instrument equipped with an optically active CP-Chirasil-DEX CB (2a-e, 2g-h, and 2j) or a Gamma DEX 225 capillary column (2f and 2i). The e.e. was calculated using the following formula: e.e. (%) = {(R − S)/(R + S)} × 100, where R and S are the respective peak areas of the isomer in GC analyses. The absolute configurations of the α- and β-hydroxy esters (2a-h and 2j), and the α-hydroxy amide (2i) were identified by comparing their retention times as determined by the GC analyses with those of authentic samples [4] [7] [19].

To determine the most suitable medium for liquid culture, 15 Agromyces and 16 Gordonia strains were cultivated in several culture media, after which the wet weight of the cells was measured. All Agromyces strains grew poorly in 227, 253, and 802 media, even after 8 days of culture, and the resulting wet cell weights were 0.2 g or less (see Table 1). However, two strains, A. soli NBRC109063 and A. humatus NBRC109085, yielded more than 0.4 g of wet cells/100mL of culture in the 230 medium, even though the recommended medium for NBRC109063

The actinomycetes were grown in liquid medium (100 mL) at 25˚C for 8 days with aerobic reciprocating shaking (100/min) in baffled 500-mL flask in the dark.

and NBRC109085 strain is 802 medium. These results suggest that the peptone contained in the culture medium is more likely to be assimilated from plants than from animals in liquid cultures of Agromyces strains. As in the case of the genus Agromyces, Gordonia strains also had poor growth in various liquid media. Only two strains, G. hydrophobica NBRC16057 and G. malaquae NBRC108250, yielded more than 0.4 g of wet cells/100mL of the liquid culture in the 230 medium (see Table 2).

Therefore, we investigated the potential ability of two Agromyces strains (NBRC109063 and NBRC109085) and two Gordonia strains (NBRC16057 and NBRC108250) to act as biocatalysts for the asymmetric reduction of carbonyl compounds.

Two actinomycete strains (A. soli NBRC109063 and A. humatus NBRC109085) cultivated in the 230 medium were tested for their ability to reduce several carbonyl compounds (1a-j) (Figure 1). The results of the microbial reductions are summarized in Table 3 and Table 4. Both strains could reduce aliphatic and aromatic α-keto esters (1a-h) and an aromatic α-keto amide (1i). Among the reduction of nine substrates tested, there were compounds in which the conversion rate was over 99% (2c, 2d, and 2f); however, the reduction of most substrates showed only low values to moderate conversion rates. Moreover, the

The actinomycetes were grown in liquid medium (100 mL) at 25˚C for 8 days with aerobic reciprocating shaking (100/min) in baffled 500-mL flask in the dark.

stereoselectivity of the product in the reduction of each substrate also showed low values except for the reduction of ethyl pyruvate (1a). We tried to improve the conversion rate and the stereoselectivity of the alcohols produced by two Agromyces strains by introducing additives (glycerol and sodium hydrogen

L-glutamate). The introduction of glycerol contributed little to the improvement of conversion rates and stereoselectivity of produced alcohols. The addition of L-glutamate also showed no significant improvement in the reduction of substrates; the reduction of only one substrate (from 1d to 2d) improved the conversion rate of both NBRC109063 and NBRC109065 wet cells to >99% and the stereoselectivity to >99% e.e. (S). As shown in Table 5, reduction of ethyl 2-methylacetoacetate (1j), one of β-keto esters, by the wet cells of NBRC109063 strain showed high values of syn/anti ratios and enantioselectivity. However, the conversion rate from 1j to 2j was low and was not improved by the additives. Furthermore, reduction of 1j by the wet NBRC109065 cells resulted in a low syn/anti ratio and low conversion rate and did not improve with the addition of additives.

Two Gordonia strains (G. hydrophobica NBRC16057 and G. malaquae NBRC108250) cultivated in 230 medium were tested for their ability to reduce the keto esters and keto amide (Table 6 and Table 7). The Gordonia strains could reduce aliphatic and aromatic α-keto esters and an aromatic α-keto amide. However, as with the reduction by two Agromyces strains, the substrate was reduced only at low to moderate conversion rates, and the stereoselectivity of the products was low except for the reduction of 1h by the G. hydrophobica NBRC16057 strain. Therefore, the effects of additives on these microbial reduction reactions were also examined. In the reductions by two Gordonia strains, NBRC16057 and NBRC108250, the introduction of glycerol did not improve the conversion rate or stereoselectivity. On the other hand, the addition of L-glutamate improved the conversion rate and the stereoselectivity of the product. In particular, the reductions of 1a, 1c, 1d, and 1g by the NBRC16057 wet cells improved the conversion rates to >99%, and 1d and 1e were reduced to the

---: Not detected.

(S)-hydroxy esters (>99% e.e.) stereospecifically.

As shown in Table 8, in the reduction of β-keto ester (1j) by G. malaquae NBRC108250 wet cells, the conversion rate and syn/anti ratio of the produced β-hydroxy ester (2j) were low and were not improved by the introduction of additives. In contrast, the reduction of 1j by the wet cells of the NBRC16057 strain stereospecifically produced the corresponding alcohol syn-(2R, 3S)-2j in high conversion rate (81%). Furthermore, when L-glutamate was introduced as an additive to the reduction of 1j, the stereoselectivity (syn/anti ratio and e.e.) of the product remained high and the conversion rate was improved to 98%. More specifically, in this reaction, the substrate was reduced to only one of the four theoretically possible stereoisomers. We reported in a previous paper that reduction by the wet cells of Dactylosporangium sp. NBRC101730 strain specifically reduces 1j to the corresponding β-hydroxy ester having anti-(2S, 3S) form [15]. In other words, we succeeded in the specific preparation of syn-(2R, 3S) and anti-(2S, 3S) forms of 2j by using two strains from different actinomycete genera, G. hydrophobica NBRC16057 strain and Dactylosporangium sp. NBRC101730 strain as shown in Figure 2.

In this study, the introduction of glycerol could not improve the conversion rate of the reduction, while the addition of L-glutamate improved the conversion rates. These results are likely due to the fact that the enzyme (keto ester reductase) involved in the reduction of the substrate depends on NADPH, and the coenzyme is excessively supplied during oxidative degradation of the added L-glutamate. In future work, we aim to purify and characterize the reductase from actinomycete whole cells and elucidate the reaction mechanism at the enzyme molecular level.

---: Not detected.