Agric. Biol. Chem., 55 (8), 2043-2049, 1991 2043

Purification and Properties of NADPH-LinkedL-Sorbose Reductase from Gluconobacter melanogenus N44-1 Teruhide Sugisawa,1 Tatsuo Hoshino and Akiko Fujiwara Department of Applied Microbiology, Nippon Roche Research Center, 200, Kajiwara, Kamakura, Kanagawa 247, Japan Received January 23, 1991

NADPH-LinkedL-sorbose reductase from the cytosol fraction of Gluconobacter melanogenus N44-1, which is a 2-keto-L-gulonic acid producer from L-sorbose or D-sorbitol, was purified about 500-fold by DEAE-Cellulose, Blue Sepharose, and DEAE-Sepharose column chromatographies. The purified enzyme was entirely homogeneous on polyacrylamide gel and SDS gel electrophoresis. NADPH and NADPwere specifically required for the reduction and oxidation of the substrate, respectively. The apparent molecular weight was 60,000 by Sephadex G-200 column chromatography, and that of its subunit was 60,000 by SDS-gel electrophoresis. The pH optimum for the reduction of L-sorbose and D-fructose was 7.0. The pH optimum for the oxidation of D-sorbitol to L-sorbose and D-mannitol to D-fructose was 10.0-10.5.

As previously reported, 1} we have developed decreased, further improvement of 2KGA a fermentation process that produces 2-keto- could be expected. L-gulonic acid (2KGA) from L-sorbose and d- The enzyme reducing L-sorbose (we say sorbitol via L-sorbosone by a mutant derived L-sorbose reductase) is found in various from G. melanogenus IFO 3293, and isolated Gluconobacters as an NADP-linked enzyme the two key enzymes of membrane-bound catalyzing the L-sorbose-D-sorbitol conversion, L-sorbose dehydrogenase2) and NAD(P)-de- but the enzymehas not been isolated. pendent L-sorbosone dehydrogenase3) found As the characterization of the enzyme through the study of the metabolic pathway reducing L-sorbose in a high 2KGAproducer for 2KGAformation.4) (G. melanogenus N44-1) was thought to be When G. melanogenus UV10 and its advantageous to the further improvement of mutants1} were used in 2KGA-fermentation 2KGAyield, we tried to isolate the L-sorbose process, a substantial amount of L-sorbose fed reductase. to the mediumwas consumedto yield carbon This report deals with the purification and dioxide without contributing to 2KGAforma- characterization of NADPH-linkedL-sorbose tion. The CO2 evolved from L-sorbose by the reductase isolated from the cytosol fraction of 2KGA high producer was mainly via the G. melanogenus N44-1 during study of the pentose phosphate pathway.5) This shows that mechanism of 2KGA formation from l- L-sorbose is metabolized to D-sorbitol besides sorbose and D-sorbitol (Fig. 1).4) the oxidation of L-sorbose to L-sorbosone for the main pathway to produce 2KGA. In addition, we showed4) D-sorbitol was produced Materials and Methods from L-sorbose by the enzyme contained in the Chemicals. NADPand reduced NADPwere purchased cytosol fraction of 2KGAproducers. If we from Oriental Yeast Ind., Co.; D-sorbitol, D-fructose, and could obtain a mutant such that the reducing D-mannitol, from Nakarai Chemicals, Ltd. 5-Keto-fructose activity of L-sorbose to D-sorbitol was was purchased from Chemical Dynamics Co.; L-Sorbose,

To whomall correspondence should be address. 2044 T. Sugisawa, T. Hoshino and A. Fujiwara

Fig. 1. The Mechanism of 2KGAProduction from L-Sorbose and D-Sorbitol in Gluconobacter melanogenus.4) Enzymes in the cytoplasmic membrane: SLDH, D-sorbitol dehydrogenase; SDH, L-sorbose dehydrogenase; IDH, L-idonic acid dehydrogenase. Enzymes in the cytosol: SNR, L-sorbosone reductase; SNDH, L-sorbosone dehydrogenase; 2KGAR, 2-keto-L-gulonic acid reductase; SR, L-sorbose reductase (described in this paper). from E. Merck. was washed with 0.85% NaCl twice. From 2 liters of broth, DEAE-Cellulose DE-52 was from Whatman, Ltd. and we obtained 30 g (wet weight) of the cells, which was frozen DEAE-Sepharose CL-6B, Blue Sepharose, and Sephadex at -20°C until use. G-200 were from Pharmacia Fine Chemicals. All other chemicals were reagent grade. Enzymeassay. The enzyme activity was estimated by measuring the decrease (or increase) in absorbance at Microorganism. Gluconobacter melanogenus N44- l , 1} a 340nmdue to NADPH.The complete reaction mixture mutant strain producing large amounts of 2-keto-L-gulonic contained 0.25 mmol of L-sorbose, 0.25 ^mol of NADPH acid from L-sorbose or D-sorbitol and selected through (or NADP) in 50mM potassium phosphate buffer, pH 7.0 repeated mutagenesis of G. melanogenus UV10, was used. (or in 50mM Na2CO3-NaHCO3 buffer, pH 10.0) and enzyme solution in a total volume of 0.51 ml. Medium and Cultivation. The culture mediumconsisted One unit of enzyme activity was defined as the amount of 100g of L-sorbose, 15g of yeast extract, 2.5g of ofenzyme catalyzing the oxidation (or reduction) of 1 /imol MgSO4à"7H2O, 0.5g ofglycerol, 20g ofCaCO3, and 1.5g ofNADPH(or NADP) per minute at 25°C. of antifoam in one liter of deionized water. G. melanogenus N44-1 was transferred from a potato- Protein measurement. Protein was measured with a agar slant to 5ml of mediumin a test tube and cultivat- Bio-Rad protein assay kit (Bio-Rad Laboratories) by ed at 30°C for 2 days. One ml of this culture was trans- measuring the absorbance at 595nmwith catalase as the ferred to 50ml of the same mediumin a 500-ml shaker standard. flask and cultivated at 30°C for 2 days on a rotary shaker (180 rpm). Electrophoresis. A 7.5% polyacrylamide gel and One hundred ml of this culture was used as the inoculum Tris-glycine buffer (pH 8.3) were used for disc gel for a 3-literjar fermentor containing 2 liters of the medium. electrophoresis as described by Davis.6) The jar fermentor was operated at 30°C, 500rpm for SDSgel electrophoresis was done by the methods of agitation, and 0.75 vvmfor aeration. Weber and Osborn7) and of Weber and Kuter.8) After 42 hr of fermentation, the culture was harvested for the collection of cells. The broth was centrifuged at Othermeasurements 1,500 rpm for 10 min for the removal of calcium carbonate Methods of molecular weight measurement of the and then at 8,000rpm for pelleting the cells. The cell cake enzyme and other methods are referred to in the Results NADPH-LinkedL-Sorbose Reductase of Gluconobacter melanogenus 2045 section. Products were identified by gas chromatography, to the buffer and the enzyme was eluted ai high performance liquid chromatography, and thin layer around 0. 1 m NaCl. The specific activity of peal; chromatography. 1] fractions (No. 76-78) were almost the same and the fractions were stored separately ai -20°C Results Purification of the enzyme, summarized in 1. Enzyme purification Table I, was approximately 520-fold with a All operations were done at 4°C to 8°C, yield of 23.9% if the crude enzyme was taken unless otherwise stated. The buffer mentioned to be 100%. in all steps is lOmMpotassium phosphate (pH 7.0). 2. Properties of purified enzyme Step 1: Preparation of cytosolfraction. The To investigate the catalytic properties oi frozen cells (20 g) stored at - 20°C were thawed L-sorbose reductase, the purified enzyme, with and suspended in 80 ml of buffer. The cells were a specific activity of 140.8, was used under the disrupted in the presence of glass beads (0. 1 mm diameter) with a Dyno Mill homogenizer Table I. Purification of l-Sorbose Reductase (Willy, A Bachofen Co., Basle) at 2,000rpm _.._ . Total Total Specific _,.., for 4min. The cells, debris, and glass beads Purification . . . . . Yield activity protein activity were removed by centrifugation at 1,800 x g for Step (unit) (mg) (unit/mg) {/o) lOmin. The supernatant was then centrifuged at 100,000x0 for 60min. The resulting Crudeenzyme 282 1,032 0.27 100 (Cytosol fraction) supernatant was dialyzed against 3 liters of the DEAE-Cellulose 1 50.5 3 1.2 4.83 53.4 same buffer (changed twice) for 18hr and Blue-Sepharose 1 14.8 4. 1 28.0 40.7 is referred to here as the cytosol fraction DEAE-Sepharose 67.6 0.48 140.8 23.9 (65 ml). Step 2: DEAE-Cellulose column chromatog- raphy. The dialyzed enzyme solution (60ml) was put on a DEAE-Cellulose DE-52 column (2.5 x 20cm), which had been equilibrated with the same buffer. An NaCl linear gradient (0 to 0.2m) was added and the enzymeactivity was found in the fractions eluted with about 0.075 m NaCl. The combined active fractions (about 60 ml) were concentrated to 18ml with a membrane filter (Amicon PM-10) and dialyzed against the buffer for 5hr. Step 3: Blue-Sepharose column chromatog- raphy. The dialyzed enzyme solution (20ml) was passed through a Blue-Sepharose CL-6B column (1.5 x 15cm), which had been equili- brated with the buffer. Step 4: DEAE-Sepharose column chromatog- raphy. The active fractions from Step 3 were combined and put on a DEAE-Sepharose CL-6B column (1.0 x 50cm), which had been equilibrated and washed with the buffer. A Fig. 2. Polyacrylamide Gel Electrophoresis ofL-Sorbose linear gradient of NaCl (0-0.15m) was added 2046 T. Sugisawa, T. Hoshino and A. Fujiwara standard assay conditions unless otherwise was lost in Sephadex G-200 column chroma- stated. tography. Acrylamide gel electrophoresis. Polyacryl- The molecular weight of the subunit was amide disc gel electrophoresis of the purified estimated by 10% (w/v) SDS-polyacrylamide enzyme obtained from Step 4 is shown in Fig. gel electrophoresis. The enzyme preparation 2. A single protein band was found, indicating gave a single protein band with a molecular the electrophoretic homogeneity of the prep- weight of60,000 ± 1,000 (Fig. 3). This indicates aration. that L-sorbose reductase from G. melanogenus Molecular weight. The purified L-sorbose is monomeric. reductase was put on a SephadexG-200column Effects of pHon activity and stability of the (10 x 500mm)chromatography and developed enzyme. The enzyme activity for oxidation and with lOOniM potassium phosphate buffer (pH reduction reactions was measured at various 7.0) containing 0.1m NaCl. The molecular pHs. The pH optimum of the reduction of weight of the enzyme was estimated to be L-sorbose was about 7.0 and that of the 58,000± 1,000, but about 22% of the activity oxidation ofD-sorbitol was about 10 (Fig. 4). The enzyme was stable at pH 6.5-7.5 on incubation at 30°C for 30min and was stored with little loss of activity at -20°C for a month or more in 0.01 m potassium phosphate buffer, pH 7.0. Effects of temperature on stability and activity of the enzyme. The effects of temperature on the stability and activity of the enzyme were investigated. Thermostability of the enzyme was tested by a 10-min incubation of the enzyme at various temperatures and then its activity to reduce L-sorbose was tested. The enzymewas stable at up to 50°C-55°C, but

Fig. 3. SDS-Gel Electrophoresis of the Purified NADPH- Linked L-Sorbose Reductase. Purified L-sorbose reductase with a specific activity of 140.8 was used and 3.2/ig of protein was put on. The direction of electrophoresis was left to right. After the protein staining with Coomassie brilliant blue R-250 in 7%acetic acid, the molecular weight of L-sorbose reductase was Fig. 4. Effects of pH on Enzyme Activity. measured by the plotting mobility versus molecular weight The purified enzyme was used and the activity was of the standard marker proteins. The position of L-sorbose measured under the standard assay conditions using the reductase is marked by a closed circle. The standard marker following buffers: Reduction of L-sorbose: -0-, 0.1 m proteins are indicated alphabetically, a, phosphorylase, Mcllvaine buffer; -A-, 0. 1 m phosphate buffer; -å -, mol. wt. 92,500; b, bovine serum albumin, mol. wt. 66,200; 0.2m Tris buffer. Oxidation of D-sorbitol: -O-, 0.1 m c, ovalbumin, mol. wt. 45,000; d, carbonic anhydrase, mol. Mcllvaine buffer; -å¡-, 0.2m Tris buffer; -V-, 0.1 m wt. 31,000; e, soybean trypsin inhibitor, mol. wt. 21,500; NH4OH-NH4C1 buffer; -ft-, 0.2 m Na2CO3-NaHCO3 f, lysozyme, mol. wt. 14,400. buffer. NADPH-LinkedL-Sorbose Reductase of Gluconobacter melanogenus 2047 rapid inactivation occurred whenthe enzyme 0.49 mM quinine (Table II). was treated at over 60°C. The highest activity Substrate specificity. The relative activities was found at 30°C. of the purified enzyme for various substrates Inhibitors. The effects of various metals and were surveyed and are listed in Table III. reagents on L-sorbosereductase activity were L-Sorbose reductase from G. melanogenus studied. The activity was inhibited by Cu2+, N44-1 had a high rate of activity to reduce Fe2+, and Ni2+ and completely inhibited by L-sorbose to D-sorbitol and to oxidize d- sorbitol to L-sorbose in the presence of Table II. Effects of Inhibitors on the Activity NADPH, but NADHdid not work as a OF L-SORBOSE REDUCTASE co factor for this reaction. In addition, D-fructose, 5-keto-D-fructose, D-xylulose, d- Concentration Relative activity" Inhibitor (niM) (%) ribulose, and D-tagatose were found to be substrates for L-sorbose reductase. This enzyme EDTA 0.97 0.49 0 also oxidized D-mannitol and erythritol. Quinine 0.97 0 Identification of reduction and oxidation 0.97 81 7V-Ethylmaleimide 0.97 products. The reaction mixture (1 ml) contain- Sodiumazide 0.97 90 ing 50mM each of L-sorbose, D-fructose and Monoiodoacetate 0.19 96 5-keto-D-fructose, and 0.28mM of NADPH PCMBb 0.97 43 was incubated for 30min in pH 7.0 buffer Na2HAsO4 -7H2O 0.97 90 0.97 103 Sodium fluoroacetate 0.19 (100mMpotassium phosphate) at 30°C with 103 Sodium fluoride 0.38 about 4.4 units of the purified enzyme. On the KCN 101 other hand, the reaction mixture (lml) 102 consisting of 40mM each of D-sorbitol and Relative activity is expressed relative to that without D-mannitol and 0.28mM of NADPwas the inhibitor. incubated for 30min in pH 10.0 buffer (100mM PCMB, />-chloromercuribenzoic acid. Na2CO3-NaHCO3) at 30°C with about 4.4 units of the purified enzyme. After the reaction, as Table III. Substrate Specificity of Purified described in Materials and Methods, the L-SORBOSE REDUCTASE products were identified TLCand GLC(data Relative activity0 (%) not shown). L-Sorbose, D-fructose, and 5-keto- Substrate D-fructose were reduced to D-sorbitol, d- NADPH(pH7.0) NADP (pH 10.0) mannitol, and D-fructose, respectively, in the 0 presence of NADPH. On the other hand, (100) Sorbose 0 Glucose 0 D-sorbitol and D-mannitol were oxidized to Mannitol 0 134 Sorbitol 0 134 Table IV. Apparent Michaelis Constants Fructose 142 0 for Various Substrates Glycerol 0 0 Adonitol 0 0 Substrate pH Km(m) Erythritol 0 65.3 Na-Gluconate 0 L-Sorbosea 7.0 3.28x KT1 Ca-Idonate 0 D-Fructosefl 7.0 1.60X KT1 Xylose 0 D-Sorbitolfl 10.0 1.67X10-1 Xylulose 25.7 D-Mannitolfl 10.0 7.30x l(T3 Ribulose 2.4 NADPH* 7.0 1.llxlO"4 Tagatose 0.8 5-Keto-D-fructose NADPHwas used as co factor for L-sorbose and D-fructose, and NADP for D-sorbitol and d- Eachreaction rate with substrate is expressed as per mannitol. cent of sorbose. The assay was done in the presence of 490mM Not tested. L-sorbose. 2048 T. Sugisawa, T. Hoshino and A. Fujiwara L-sorbose and D-fructose respectively in the Cummins et al., 12'13) three enzymes for the presence ofNADP.However, the oxidation of oxidation of D-sorbitol were found in the D-fructose was not observed in the presence of cell-free extract of A. suboxydans (syn. G. NADP. suboxydans). In addition to a dehydrogenase Effects ofsubstrate concentration on reaction in the membrane fraction,14'15) there were two rate. The effects of substrate concentration on enzymesin the soluble fraction; one catalyzes the enzyme activity were tested. Apparent D-fructose formation in the presence of NAD Michaelis constants for various substrates were and another catalyzes L-sorbose formation in calculated9) and are summarized in Table IV. the presence of NADP.However, the NADP- It was found that the affinity of the enzyme in linked D-sorbitol dehydrogenase has not been the presence of NADP was highest for purified because of its instability. Shaw16) and D-mannitol amongall the carbohydrates tested. Kersters et al., 17) have reported on NADP- linked D-mannitol dehydrogenase activity in Discussion cell extracts of A. suboxydans TACC621(syn. G. suboxydans) and G. suboxydans strain SU, We purified and characterized the NADPH- respectively. Yamada et al.,18) reported the linked L-sorbose reductase of G. melanogenus purification and properties of NADP-linked N44-1, which was one of the reductases 5-keto-D-fructose reductase of G. albidus IFO functioning in the metabolic pathway of2KGA 3250, but the substrate specificity of the enzyme production from D-sorbitol or L-sorbose (Fig. was significantly different from ours, namely 1). This enzymewas apparently different from 5-keto-D-fructose reductase is not active in the other, reductases such as L-sorbosone and catalysis of L-sorbose. 2KGAreductase because it was not active in Comparing the catalytic properties of the the reduction of L-sorbosone and 2KGA.The enzymes described above with ours, we suggest L-sorbose reductase with regards to its function that our L-sorbose reductase is in fact the same can also be called an oxidoreductase because as NADP-linked D-sorbitol dehydrogenase and it catalyzes not only the reduction ofL-sorbose D-mannitol dehydrogenase as described by and D-fructose in the presence ofNADPHbut Widmer, Cummius,Shaw, and Kersters et al. also the oxidation ofD-sorbitol and D-mannitol In the metabolic pathway for the production in the presence ofNADPas illustrated below: of2KGA, our enzyme was thought to function pH 10.0 mainly as a reductase in vivo during the D-Sorbitol +NADP+ < L-Sorbose+ NADPH + H+ fermentation of 2KGA,which means that the (D-mannitol) pH 7.0 (D_Fructose) conversion of L-sorbose to D-sorbitol is the Optimum pHs were around 10.0 for the main function of the enzyme as described by forward reaction and 7.0 for the reverse Tsukada and Permian10) because the reductive reaction. activity of the enzyme was higher than the Here we report a purification scheme of oxidative activity at the pH between 5 and 7 NADP-linked L-sorbose reductase from G. as shown in Fig. 4, which was the pH range melanogenus, an enzyme active in the catalysis suitable for cell growth and 2KGAproduction of D-sorbitol and D-mannitol. Tsukada and as described in a previous paper,1} and in Permian10) have reported that the cell-free addition, the intracellular pH of the acetic acid extract of G. melanogenus IFO 3293 have bacteria is thought to be around 7.0. The activity to convert L-sorbose to D-sorbitol and products formed from L-sorbose by the vise versa in the presence of NADP(H);they reductive function of the enzyme are D-sorbitol postulated that the enzyme is a coenzyme- and NADP. dependent dehydrogenase and that the reaction In the 2KGAfermentation from L-sorbose favors the conversion from L-sorbose to or D-sorbitol, our mutants of G. melanogenus, D-sorbitol. According to Widmer1X) and as previously reported,1} about 40%of both NADPH-LinkedL-Sorbose Reductase of Gluconobacter melanogenus 2049 substrates added was wasted without contrib- 5) M. Shinjoh, Y. Setoguchi, T. Hoshino and A. uting to 2KGA formation. L-Sorbose was Fujiwara, Agric. Biol. Chem., 54, 2257 (1990). 6) B. J. Davis, Ann. N.Y. Acad. Set, 121, 404 (1964). considered to be dissimilated finally to CO2via 7) K. Weber and M. Osborn, /. Biol. Chem., 244, 4406 D-sorbitol and then via the pentose cycle. In (1969). fact, a blocked mutant of L-sorbose reductase 8) K. Weber and D. J. Kuter, /. Biol. Chem., 246, 4504 could not grow on L-sorbose as a sole carbon (1971). 9) H. Lineweaver and D. Bruk, J. Am. Chem. Soc., 56, source. In this context, the L-sorbose reductase 658 (1934). seems to be the first-step enzyme for the 10) Y. Tsukada and D. Perlman, Biotech, and Bioeng., dissimilation of L-sorbose in our microorgan- 14, 799 (1972). isms. ll) C. Widmer, T. E. King and V. H. Cheldelin, J. Bacteriol, 71, 737 (1956). Acknowledgment. The authors thank Mr. M. Tazoe 12) J. T. Cummins, T. E. King and V. H. Cheldelin, /. and Ms. S. Someha for providing us with the blocked Biol. Chem., 224, 323 (1957). mutant of L-sorbose reductase. 13) J. T. Cummins, V. H. Cheldelin and T. E. King, J. Biol. Chem., 226, 301 (1957). 14) E. Shinagawa, K. Matsushita, O. Adachi and M. References Ameyama, Agric. Biol. Chem., 46, 135 (1982). 1) T. Sugisawa, T. Hoshino, S. Masuda, S. Nomura, 15) C. A. Baker and G. W. Claus, FEMS Microbiol. Y. Setoguchi, M. Tazoe, M. Shinjoh, S. Someha and Lett., 18, 123 (1983). A. Fujiwara, Agric. Biol. Chem., 54, 1201 (1990). 16) D. R. D. Shaw, Biochim. Biophys. Ada, 113, 608 2) T. Sugisawa, T. Hoshino and A. Fujiwara, Agric. (1966). Biol. Chem., 55, 363 (1991). 17) K. Kersters, W. A. Wood and J. De Ley, J. Biol. 3) T. Hoshino, T. Sugisawa and A. Fujiwara, Agric. Chem., 240, 965 (1965). Biol. Chem., 55, ,665 (1991). 18) Y. Yamada, K. Aida and T. Uemura, /. Biochem., 4) T. Hoshino, T. Sugisawa, M. Tazoe, M. Shinjoh and 61, 803 (1967). A. Fujiwara, Agric. Biol. Chem., 54, 1211 (1990).