Appl Microbiol Biotechnol (1992) 37:599-603

Applied Microbiology Biotechnology © Springer-Verlag 1992

Transformation of Nt-CBZ-L-lysine to CBZ-L-oxylysine using L-amino acid oxidase from Providencia alcalifaciens and L-2-hydroxy-isocaproate dehydrogenase from Lactobacillus confusus Ronald L. Hanson, Kenneth S. Bembenek, Ramesh N. Patel, and Laszlo J. Szarka Department of Microbial Technology, Bristol-Myers Squibb, New Brunswick, NJ 08903, USA Received 22 November 1991/Accepted 9 April 1992

Summary. Biotransformations were developed to oxidize Nc-carbobenzoxy(CBZ)-L-lysine and to reduce the product keto acid to L-CBZ-oxylysine. Lysyl oxidase (Llysine: 02 oxidoreductase, EC 1.4.3.14) from Trichoderma viride was relatively specific for L-lysine and had very low activity with Ne-substituted derivatives. L-Amino acid oxidase (L-amino acid: O2 oxidoreductase [deaminating], EC 1.4.3.2) f r o m Crotalus adamanteus venom had low activity with L-lysine but high activity with Ne-formyl-, t-butyoxycarbonyl(BOC)-, acetyl-, trifluoroacetyl-, or CBZ-L-lysine. L-2-Hydroxyisocaproate dehydrogenase (EC 1.1.1.-) from Lactobacillus confusus catalyzed the reduction by N A D H of the keto acids f r o m Nc-acetyl-, trifluoroacetyl-, formyl- and CBZ-L-lysine but was inactive with the products f r o m oxidation of L-lysine, L-lysine methyl ester, L-lysine ethyl ester or Nc-t-BOC-L-lysine. Providencia alcalifaciens (SC9036, A T C C 13159) was a good microbial substitute for the snake venom oxidase and also provided catalase (H202: H202 oxidoreductase EC 1.11.1.6). Ne-CBZ-LLysine was converted to CBZ-L-oxylysine in 95% yield with 98.50/0 optical purity :by oxidation using P. alcalifaciens cells followed by reduction of the keto acid using L-2-hydroxyisocaproate dehydrogenase. N A D H was regenerated using formate dehydrogenase (formate: N A D oxidoreductase, EC 1.2.1.2) from Candida boidinii. The Providencia oxidase was localized in the particulate fraction and catalase activity was predominantly in the soluble fraction of sonicated cells. The p H optima and kinetic constants were determined for the reactions.

converting enzyme (ACE) inhibitor being developed for treatment of hypertension (Fig. 1) (Karenewsky et al. 1988). The precursor of this intermediate, L-oxylysine, has also been used to prepare an analog of the antibiotic, butirosin (Haskell et al. 1973), and for synthesis of a piperidine alkaloid (Aketa et al. 1976). CBZ-L-Oxylysine has been prepared f r o m L-lysine by diazotization with sodium nitrite and sulfuric acid followed by treatment of the product L-oxylysine with benzyl chloroformate (Karenewsky et al. 1988). As an alternative approach, we have explored the enzymic oxidation of Llysine or CBZ-L-lysine to the keto acid followed by reduction using a dehydrogenase with appropriate enantioselectivity.

Materials and methods Enzyme assays. L-Amino acid oxidase was monitored by coupling the H202 evolved in the reaction to the oxidation of o-dianisidine catalyzed by horseradish peroxidase (Berezov and Lukasheva 1988). The reaction solution contained in 1 ml: 50 mM potassium phosphate, pH 7.4, 1 mM L-lysine or lysine derivative, 0.2 mM odianisidine, and 10gg [2.17 units (U)] horseradish peroxidase. The reaction was started by the addition of L-amino acid oxidase, and the increase in absorbance at 460 nm was monitored (absorbance = 11.3 mM- 1 cm- 1). All continuous spectrophotometric assays were performed at 25° C. The reaction solution for L-2-hydroxyisocaproate dehydrogenase (L-HIC dehydrogenase) coupled to L-amino acid oxidase conNHZ

NH3+

Introduction As requirements for optical purity of pharmaceuticals become more stringent, enzymic methods for production of chiral synthons are receiving increased attention. Carbobenzoxy(CBZ)-L-oxylysine is an intermediate needed for synthesis of ceronapril, a new angiotensin Correspondence to." R. L. Hanson

L-Z-oxylysine

Ceronapril

Fig. 1. Structures of carbobenzoxy(CBZ)-L-oxylysine and ceronapril

600 tained in 1 ml: 0.1 M potassium phosphate, pH 7.4, 1 mM lysine or lysine derivative, 2000 U bovine liver catalase, and 1.7 U L-HIC dehydrogenase. Reactions were started by the addition of L-amino acid oxidase and the absorbance decrease at 340 nm was monitored. The catalase assay contained in 1.0 ml:0.1 M potassium phosphate, pH 7.4, and 0.06% H202. Absorbance decrease after addition of enzyme was monitored at 240nm (absorbance =0.0436 mM -1 cm-1). Protein was determined by the dye-binding method of Bradford (1976), using bovine serum albumin as standard.

phate buffer, pH 7.4, and stored frozen at - 18 ° C until used for biotransformation.

Enzyme localization. Cells (1.40g, wet weight) were collected from a 17-h shake flask culture, and sonicated in 15 ml of 50 mM potassium phosphate, pH 7.4. Debris was removed by centrifugation for 10 rain at 12000g. The extract supernatant was centrifuged for 1 h at 101000 g to give a supernatant and pellet fraction. The pellet was resuspended in 2 ml of 50raM potassium phosphate, pH 7.4. Oxidation of CBZ-L-lysine was measured by incubating 5 mM CBZ-L-lysine with 0.6 ml of extract or 101000 g pellet or supernatant fractions in 1.5 ml containing 0.1 M potassium phosphate, pH 7.4, and 3000 U catalase for 16 h. The amount of 6-CBZ-amino-2-oxohexanoic acid produced by 1 mg Crotalus adamanteus venom L-amino acid oxidase under these conditions was used as a standard for 100% conversion in the HPLC assay.

Analytical methods. H P L C analysis of CBZ-L-lysine transformations were performed with a Hewlett-Packard, Palo Alto, California, USA, hypersil C18 2 0 × 4 . 6 c m column, with 5 ~tm particle size. The column temperature was 40 ° C, the mobile phase was 37°7o methanol and 63o/o water containing 0.05°7o H3PO4, the flow rate was 1 ml/min, the detection wave length was 215 nm and the injection volume was 5 ~tl. Retention times were 9.8 min for 6CBZ-amino-2-oxohexanoic acid, 13.7 min for CBZ-L-oxylysine, and 23.3 min for CBZ-L-lysine, with the keto acid peak skewed toward higher retention times. Samples were boiled for 2 min, centrifuged and filtered before HPLC analysis. Optical purity of CBZ-oxylysine was determined by derivation and separation of diastereomers by gas chromatography (Jemal and Cohen 1987).

Materials. L-HIC dehydrogenase and formate dehydrogenase were gifts from Dr. H. Scht~tte, Gesellschaft ft~r Bioechnologische Forschung, Braunschweig, FRG. Commercial sources were: lysyl oxidase (Yamasa Shoyu, Choshi, Chiba, Japan); L-amino acid oxidase Type 1 from Crotalus adamanteus (Sigma, St. Louis, Mo., USA); polyethylene glycol (PEG)-2000-NADH (Braunschweiger Biotechnologie, Braunschweig, FRG).

Results and discussion

Microbial growth conditions. Providencia alcalifaciens SC9036 was obtained from the Squibb culture collection and is ATCC strain 13159. It was grown on medium described by Szwajcer et al. (1982) containing 1% peptone, 0.2% casein hydrolysate, 0.2% yeast extract and 0.6% NaC1 at pH 7.2-7.4. Growth was at 37 ° C, and 100 rpm in shake flasks. A 200-ml overnight culture was used to inoculate a 15-1 tank containing the same medium at 37 ° C, stirred at 200 rpm, and aerated at 20 I/min. After 11 h, cells were harvested by centrifugation, washed with 50mM potassium phos-

Substrate specificities The enzymatic approach to synthesis of CBZ-L-oxylysine is shown in Fig. 2. C. adamanteus (rattlesnake) venom L-amino acid oxidase has been reported to oxidize CBZ-L-lysine to the corresponding keto acid (Meister 1954) and enzyme screening studies showed that L-HIC dehydrogenase from Lactobacillus confusus converted

o Z H N @ o

H

+

H20

NH2

I

/

O2

L-aminoacidoxidase

catalase o ZHN@

oo2 formate

OH

+ NH 3 +

H202

O

formate dehydrogenase

(

NADH

L-2-hydroxyisocaproate dehydrogenase

NAD*

O

ZHN~~."~OH OH

Fig. 2. Scheme for enzymatic synthesis of CBZ-L-oxylysine

601 Table 1. Substrate specificities of L-amino acid oxidases and L-2-hydroxyisocaproate dehydrogenase

Enzyme

Substrate

Specific activity (U/mg oxidase)

Trichoderma viride lysyl oxidase

L-Lysine L-Oxylysine Ne-Acetyl-L-lysine Ne-Formyl-L-lysine Ne-t-BOC-L-Lysine Ne-Trifluoroacetyl-L-lysine Ne-CBZ-L-Lysine (0.5 mM) Ne-CBZ-L-Lysine methyl ester L-Lysine methyl ester L-Lysine ethyl ester Lys-pro-NHz

1.663 0.000 0.003 0.033 0.000 0.027 0.003 0.000 0.847 1.083 0.000

Crotalus adamanteus L-amino acid oxidase

L-Lysine L-Oxylysine Ne-Acetyl-L-lysine Ne-Formyl-L-lysine Ne-t-BOC-L-Lysine Ne-Trifluoroacetyl-L-lysine Ne-CBZ-L-Lysine Ne-CBZ-L-Lysine methyl ester L-Lysine methyl ester L-Lysine ethyl ester Lys-pro-NH2

0.000 0.001 0.405 0.338 0.254 0.406 0.405 0.006 0.001 0.000 0.001

T. viridine lysyl oxidase coupled to L-2-hydroxyisocaproate dehydrogenase

L-Lysine

0.000

L-Lysine ethyl ester L-Lysine methyl ester

0.000 0.000

C. adamanteus L-amino acid oxidase coupled to L-2-hydroxyisocaproate dehydrogenase

Ne-Acetyl-L-lysine Ne-Formyl-L-lysine Ne-t-BOC-L-Lysine Ne-Trifluoroacetyl-L-lysine Ne-CBZ-L-L ysine

0.062 0.120 0.000 0.028 0.120

BOC, butoxycarbonyl; CBZ, carbobenzoxy; U, units the keto acid to CBZ-L-oxylysine (R. S. Robinson and M. G. Doremus, unpublished Squibb work). Lysyl 2oxidase f r o m Trichoderma viride was tested as a microbial alternative to the snake venom enzyme. Substrate specificity studies (Table 1) showed that the activity of the enzyme is greatly reduced with e-substituted lysine derivatives. Activity with Ne-formyl-, acetyl-, trifluoroacetyl-, t-butoxycarbonyl(BOC)-, or CBZ-L-lysine was less than 2°70 of the activity with lysine. There was no activity with L-oxylysine. On the other hand, the snake venom oxidase was active with the e-substituted lysine derivatives but had very little activity with L-lysine. The methyl and ethyl esters of lysine were substrates for the T. viride oxidase but were not utilized by the snake ven o m enzyme. The product f r o m lysine oxidation by T. viride lysyl oxidase was not a substrate for L-HIC dehydrogenase (Table 1) or for several other dehydrogenases that were screened at p H 7.4 or 9.0. With snake venom L-amino acid oxidase, Ne-formyl-, acetyl-, CBZ-, or trifluoroacetyl-L-lysine were good substrates when the oxidation was coupled to L-HIC dehydrogenase, but e-t-BOC-L-lysine was not an effective substrate. The product of lysine oxidation by the T. viride enzyme has been shown to cyclize to a Schiff base (Kusakabe et al. 1980) and this

may interfere with utilization by L-HIC dehydrogenase. o

+H~o

II H 2 N - - (CH2)4-- C -- C O O H •

©,

COOH

A 1-piperideine-2-carboxylate Alternatively, although L-HIC dehydrogenase shows rather broad specificity (Schtitte et al. 1984) it may require a substituent on the e-amino of lysine for activity. Synthesis o f CBZ-L-oxylysine H P L C analysis showed that 0.3 U / m l of T. viride lysyl oxidase coupled to 0.8 U / m l of L-HIC dehydrogenase was able to convert 1 m g / m l of CBZ-L-lysine to 1 rag/ ml of CBZ-L-oxylysine after 48 h in a reaction mixture that also contained 1 mM NAD, 0.2 M sodium formate, 0.7 U / m l of formate dehydrogenase, and 1250 U / m l of catalase in 0.1 M potassium phosphate at p H 8.0. In an effort to find a more active microbial CBZ-L-lysine oxidase activity to substitute for the snake venom enzyme, several strains known to possess L-amino acid oxidase were screened. Four strains of Proteus and Providencia

602 t-

A215

20.0

13.687

144-

O r'Q

15.0

E

107-

-5

10.0

E >:,

70-

5.0

> o