Proc. Nat. Acad. Sci. USA Vol. 68, No. ;i, pp. 987-991, Mlay 1971

Demonstration of Imino Acids as Products of the Reactions Catalyzed by D- and L- Oxidases

EDMUND W. HAFNER AND DANIEL WELLNER Department of Biochemistry, Cornell University Medical College, New York, N.Y. 10021 Communicated by Alton Meister, February 23, 1971

ABSTRACT It had long been thought, but never dem- from the enzyme as the primary oxidation product. Other in- onstrated, that imino acids are formed in the reactions terpretations, however, are not (see catalyzed by D- and L-amino acid oxidases (EC 1.4.3.3 and excluded Discussion be- 1.4.3.2). The formation of imino acids is now shown low). directly by allowing the amino acid oxidase reaction to Coffey, Neims, and Hellerman (9) recently observed that proceed in the presence of NaBH4, when the imino acid is when a mixture of D-amino acid oxidase and ['4C]D- reduced to the corresponding racemic amino acid. Thus, was treated with sodium borohydride, the alanine moiety when NaBH4 is added to a mixture of D-amino acid oxidase and D-alanine, a significant amount of L-alanine is formed. became covalently attached to the enzyme; they subsequently Analogous results are obtained using L-amino acid oxidase found that the labeled gave e-N-(1-carboxyethyl)-L- and L-. Since D-amino acid oxidase is active in the on hydrolysis (10). Studies in this laboratory have con- presence of NaBH4, L-alanine continues to be formed un- firmed this finding and have also shown that i- and D-amino til most of the D-isomer is oxidized by the enzyme. This reaction provides a new method for inverting the con- acid oxidases labeled in this manner are still enzymatically figuration ofan amino acid. active (see Discussion). We thought that these results might be When NaBH4 is added to a system containing D-amino explained in terms of the formation of an imino acid. If a free acid oxidase plus D-alanine and L-lysine, free c-N-(1-carb- imino acid were released as the product of the amino acid oxi- oxyethyl)-L-lysine is formed. When bovine serum albumin dase reaction, it would be expected that its reduction by boro- is substituted for L-lysine, the same compound results upon acid hydrolysis. It is concluded that the amino acid hydride would yield a racemic amino acid. Thus, a significant oxidase reaction produces a free imino acid, which may amount of ialanine should be found when sodium borohy- be reduced by NaBH4 to a racemic amino acid or may dride is added to a system containing D-amino acid oxidase and form Schiff's bases by reaction with the -amino groups D-alanine. Experiments designed to demonstrate this reaction of and of free lysine. and to elucidate the interaction between this postulated prod- The formation of a-imino acids as transient intermediates uct and the amino groups of proteins are described in this in the oxidative deamination of a-amino acids was postulated report. by Knoop (1) before Krebs' discovery of amino acid oxidases MATERIALS (amino acid: oxygen oxidoreductases, deaminating, EC 1.4.3.2 and 1.4.3.3) in animal tissues (2). Although such intermediates Crystalline D-amino acid oxidase was prepared from hog kid- are extremely unstable in aqueous solution and have never ney (11) and crystalline iamino acid oxidase from Crotalus been isolated from enzymatic reaction mixtures, their forma- adamanteus venom (12). Bovine serum albumin was from tion by amino acid oxidases is supported by some indirect Armour; catalase (EC 1.11.1.6) and FAD from Sigma; evidence. For example, evidence exists against an alternative D-alanine and D- from Calbiochem; [1-14C]D-alanine mechanism of oxidation suggested by Dakin (3) involving the and [3H]NaBH4, from New England Nuclear. H-10 Antifoam formation of an a,jB-unsaturated intermediate (for a review was from Dow Chemical. All other reagents were analytical see ref. 4). More directly, it has been shown by Radhakrishnan grade. The D-alanine was treated with iamino acid oxidase and Meister (5) that in the reversal of the D-amino acid oxi- and reisolated on a Dowex-50 column. Subsequent analysis dase reaction, the formation of D- from Al-pyrroline-2- by reaction with D-amino acid oxidase showed that the treated carboxylate, which may be regarded as a substituted a-imino preparation contained, at most, 0.004% ialanine. acid, is considerably faster than that of other amino acids from the corresponding a-keto acid and ammonia. Spectro- METHODS photometric evidence was obtained by Taborsky (6) and by Determination of L-alanine Pitt (7) for the formation of an intermediate absorbing at 300 After the addition of HCl to the reaction mixture (Table 1) to nm in the course of oxidation of ityrosine by snake venom destroy NaBH4, the pH was adjusted to 8.1-8.3 with 3 N i-amino acid oxidase. The data presented, however, do not NaOH and the mixture was recharged with 1 mg of D-amino permit an unequivocal conclusion concerning the nature of acid oxidase, 0.12 gmol of FAD, and 2 mg of bovine serum this intermediate. Recently, Yagi et al. (8) reported evidence albumin. It was then incubated for at least 17 hr at 370C with for the formation of a product, less basic than either the start- vigorous shaking in dark test tubes. The mixture was depro- ing amino acid or ammonia, in the course of oxidation of D- teinized by the addition of 0.8 ml of 25% trichloroacetic acid alanine and other amino acids by D-amino acid oxidase. They followed by centrifugation, adjusted to pH 2.2 with 3 N interpreted their data to mean that an imino acid is released NaOH, and analyzed on a Beckman Model 120C amino acid 987 Downloaded by guest on September 26, 2021 988 Biochemistry: Hafner and Weilner Proc. Nat. Acad. Sci. USA 68 (1971)

analyzer. In a number of experiments, the L configuration TABLE 1. Formation of Lalanine from D-alanine in the of the alanine was verified by demonstrating its disappearance presence of D-amino acid oxidase and sodium borohydride after treatment with -amino acid oxidase. In reaction mixtures containing benzoate, acidification L-Alanine with 0.5 ml of 3 N HC1 resulted in precipitation of benzoic formed Yield acid. This was filtered through Whatman no. 3 paper and Expt. Component omitted (umol) (%) the filtrate was extracted three times with 5 ml of ether. 1 None (NaBH4 added at 10 0.75 5.0 After removal of the residual ether with a stream of nitrogen, see)* the solution was treated with D-amino acid oxidase and 2 None (NaBH4 added at 0 0.71 4.7 analyzed as described above. time)t 3 D-Amino acid oxidase 0.005 ... Determination of e-N-(1-carboxyethyl)-L-lysine 4 NaBH4 <0.003 ... The protein solutions were mixed with an equal volume of 5 D-Alanine (replaced by so- <0.003 ... concentrated HCl, sealed under vacuum, and hydrolyzed at dium pyruvate + NH4Cl)t 110'C for 24 hr (Expts. 1 and 2, Table 6) or for 53 hr (Expt. 3). Analysis was carried out according to Spackman, Stein, * The complete system contained D-amino acid oxidase (1 mg), and Moore (13) with a Beckman 120C amino acid analyzer, FAD (0.12 iimol), bovine serum albumin (2 mg), sodium pyro- using PA28 resin, 0.2 N sodium citrate buffer, pH 3.20, a phosphate buffer, pH 8.3 (50,mol), and D-alanine (15 jsmol, added rate of 70 The buffer last) in a final volume of 1.5 ml. After 10 sec, 53 pmol of NaBH4 temperature of 550C, and a flow ml/hr. (freshly dissolved) in 0.2 ml of water was added. The reaction, change was delayed until methionine was eluted. The new carried out with stirring in a water-jacketed cup maintained at component [identified by Hellerman and Coffey (10) as c-N- 370C and with air blowing over the surface, was stopped after (1-carboxyethyl)-ilysine] appeared 12 min after methionine. 5 min by the addition of 0.2 ml of 3 N HCl. The L-alanine formed Correction was made for small amounts of alloisoleucine eluted was determined as described under Methods. in this position when control samples of protein were analyzed t Borohydride was added together with D-alanine. In other under identifical conditions. The same peak was observed experiments it was added 10 see later. without hydrolysis when the amino acid oxidase reaction was t 10 umol of each. carried out in the presence of free ilysine. Calculations were made using a ninhydrin color value of 1.07 leucine equivalents (10). TABLE 2. Test for imino acid accumulation in the absence of borohydride Radioactivity measurements '4C-labeled compounds were plated on aluminum planchets Time of Time of and counted in a Nuclear Chicago model D47 gas-flow NaBH4 Na benzoate L-Alanine counter fitted with a micromil window, or were dissolved in addition addition formed Yield scintillation fluid (14) and counted in A Nuclear Chicago Expt. (see) (see) (umol) (%) Unilux II scintillation counter. H-labeled compounds were 1 30 30 0.023 0.15 measured by scintillation counting. 2 30 31 0.027 0.18 3 31 30 0.022 0.15 RESULTS 4 30 60 0.387 2.6 Experiments showing ialanine production by reaction of 5 0 0 0.035 0.23 D-amino acid oxidase, D-alanine, and NaBH4 are presented in Table 1. A control (Expt. 3) showing that no racemization is Conditions were as in Table 1 except that 0.2 ml of 2.85 M caused by the borohydride treatment is included. Essentially sodium benzoate was added at the times indicated, measured no ialanine was formed when pyruvate and ammonia were from the moment of addition of D-alanine to the reaction mixture. substituted for D-alanine. It can be seen that almost as much In Expt. 1, sodium benzoate and sodium borohydride were mixed i-alanine was formed when the borohydride was added to- together before addition. In Expt. 5, these reagents were previ- gether with the D-alanine as when it was added 10 see later, ously mixed with the D-alanine. which shows that the enzyme continues to turn over in the presence of NaBH4. When similar experiments (results not lated during the first 30 see of incubation, or that the rate of shown) were carried out with iamino acid oxidase and i hydrolysis of any imino acid present was much greater than leucine, D-leucine was formed. In order to measure in the rate of its reduction by borohydride. In order to see how much imino acid accumulates the rates of these two we oxidized in which the relative reactions, [1_14C]D- absence of borohydride, we performed experiments alanine in the presence of borohydride and measured the ratio benzoate, a potent inhibitor of D-amino acid oxidase, was the -alanine to the of imino acid reduced (indicated by formed) added together with borohydride, 30 see after mixing the pyruvate (or lactate) produced. The results (Table 3) show enzyme with its substrate. As can be seen in Table 2, the small amount of ialanine found (Expt. 1) was no greater than in a at zero control experiment in which both reagents were added * Although the yield of L-alanine was only 2.6%, it was possible time (Expt. 5). When the benzoate was added 1 see after the to convert 78% of the D-alanine to the L-isomer in another ex- borohydride, again very little ialanine was found (Expt. 2). periment [conditions as in Table 3, 1 hr reaction time]. Suit- However, a considerable amount of -alanine was found when able reaction conditions may produce even higher yields. This the benzoate addition was delayed for 30 see (Expt. 4).* These reaction thus provides a new method for inverting the configura- data indicate either that very little imino acid had accumu- tion of an amino acid. Downloaded by guest on September 26, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Imino Acids as Reaction Products 989

TABLE 3. Relative rates of imino acid hydrolysis and reduction TABLE 5. e-N-(1-Carboxyethyl)-L-lysine (CEL) in hydrolysates in the presence of NaBH4 of reductively labeled protein Total (2 X Amount "Pyruvate" alanine i-Alanine ialanine)/ D-Amino Bovine Time of fraction fraction fraction Amount acid serum NaBH4 moles CEL/ (cpm; umol) (cpm; ,Amol) (cpm; umol) "pyruvate") oxidase albumin addition CEL 100,000 g Expt. (mg) (mg) (sec) (smol) of oxidase 44,300; 0. 168 2.3 X 106; 8.70 790, 000; 3.0 36 1 1.44 0 30 0.047 3.3 2 1.44 6.0 30 0.636 44 Reaction conditions were as described in Table 1, except that 3 2.88 0 15 0.059 2.1 9 umol of [1-14C]D-alanine (2.38 X 106 cpm) was used, 700 units of catalase and 0.01 ml of H-10 antifoam were added, and O was The reactions were carried out in a final volume of 1 ml con- blown over the surface of the solution. The reaction was started taining 0.12 of FAD, 5 of D-alanine, 25 umol of sodium by adding the enzyme to a reaction mixture containing all the jMmol I&mol D- other components, including NaBH4. After 5 the reaction pyrophosphate buffer, pH 8.3, and the indicated amounts of min, amino acid oxidase and bovine serum albumin. Other conditions mixture was acidified with 0.5 ml of 25% trichloroacetic acid were as in Table 1. The protein was hydrolyzed and analyzed for and a 1-ml aliquot was passed through a column of Dowex 5OW- e-N-(1-carboxyethyl)-L-lysine as described under Methods. X2 resin (2 ml, H + form). The column was washed with water and the radioactivity of the combined effluent and washes was determined ("pyruvate" fraction). After elution of the alanine with 2 N NH4OH and evaporation to dryness, the residue was could not be determined if any D-amino acid was formed (in redissolved in water and counted (total alanine fraction). The L- addition to the L-amino acid) by the borohydride reduction. isomer was measured by treating a portion of the alanine fraction To investigate this question, we incubated D-methionine with with D-amino acid oxidase until all of the D-isomer was oxidized D-amino acid oxidase and [3H]NaBH4. A portion of the and counting the residual radioactivity in the alanine reisolated methionine isolated from this system was treated with D- by the column procedure described above. amino acid oxidase while another portion was treated with L-amino acid oxidase. In each case, the tritium released into that, under the conditions used, reduction is about 36 times the water and that remaining associated with the amino acid faster than hydrolysis. It may therefore be concluded that very fraction was measured. It was found that 48% of the radio- little imino acid accumulates in the absence of borohydride. activity was released from the amino acid by treatment with The possibility must be considered that the imino acid is not D-amino acid oxidase and 49% by treatment with L-amino acid released but is reduced while still bound at the active center of oxidase. This indicates that the methionine formed by boro- the enzyme. In view of the high stereospecificity of this en- hydride reduction was racemic. Similar results (unpublished) zyme, one might expect that if this were the case, reduction were obtained when we used D-alanine. would yield primarily or exclusively a D-amino acid. Indeed, Hellerman and Coffey (10) reported the finding of e-N-(1- amino acids formed by this enzyme through reductive amina- carboxyethyl)-i-lysine after acid hydrolysis of D-amino acid tion of a-keto acids have been shown to possess the D con- oxidase treated with D-alanine and sodium borohydride. They figuration (5). Furthermore, it was recently shown by Hoch- found the same derivative in similar experiments in which reiter and Schellenberg (15) that i- is formed in L-amino acid oxidase and L-alanine were used. We have car- significant excess over D-glutamic acid by borohydride reduc- ried out the following experiments in an effort to clarify the tion of the corresponding imino acid bound to L-glutamate de- significance of these results in the light of the data presented hydrogenase. In the experiments described in Table 1 above, it above. When the reductive binding of radioactive substrate was measured as described by Hellerman and Coffey (10), either by radioactivity measurements on the precipitated pro- TABLE 4. Protein labeling with [1-"4C]D-alanine, NaBH4, and D-amino acid oxidase TABLE 6. Formation of e-N-(1-carboxyethyl)-L-lysine (CEL) Substrate bound to protein from D-alanine and L-lysine in the presence of D-amino acid n-Amino Bovine oxidase and NaBH4 acid serum moles/ oxidase albumin 100,000 g CEL found Expt. (mg) (mg) cpm X 10-4 umol of oxidase Expt. Component omitted (Amol) 1 1.0 0 0.3 0.01 1 1 None* 0.10 2 2.2 5.0 8.2 0.27 12 2 L-Lysine <0.003 3 0 1.0 0.017 0.0006 ... 3 D-Amino acid oxidase <0.003 4 D-AlMrnine <0.003 The reaction mixtures contained 0.24 umol of FAD, 100 ,mol of sodium pyrophosphate buffer, pH 8.3, 25 pmol of [1_14C]D- * The complete system contained 0.1 mg of D-amino acid alanine (347 cpm/nmol), and enzyme and bovine serum albumin oxidase, 2 mg of bovine serum albumin, 0.12 Mmol of FAD, 5 usmol as indicated, in a final volume of 2.5 ml. The alanine, added last, of D-alanine, 60 usmol of sodium pyrophosphate buffer, pH 8.3, was followed after 10 sec by approximately 50 pmol of solid and 1.5 .mol of L-lysine. Sodium borohydride (26.4 umol in 0.1 NaBH4. After 30 min at 37°C, the solutions were heated to ml) was added 20 sec after D-alanine was added to the remaining 100°C for 5 min, cooled, and dialyzed against water for several components. After 60 min of incubation with shaking at 37°C, days with frequent changes. The protein was then transferred to the solution was dialyzed and the dialysate was analyzed for e-N- aluminum planchets, dried under an infrared lamp, and counted. (1-carboxyethyl)-ilysine. Downloaded by guest on September 26, 2021 990 Biochemistry: Hafner and Wellner Proc. Nat. Acad. Sci. USA 68 (1971)

tein or by chromatographic determination of the new peak alternative interpietation of the data of Yagi et at. (8) would that appeared on the amino acid analyzer near methionine, be that the intermediate they observed is the carbinolamine we found up to 3.3 mol of the alanine derivative bound per indicated in Scheme 1. Such a compound would not be ex- 100,000 g of D-amino acid oxidase (Tables 4 and 5, Expt. 1). In pected to react with borohydride. It has been suggested that the presence of bovine serum albumin, a considerably larger the presence of an a-hydroxy group may considerably reduce amount of the derivative was observed (Tables 4 and 5, Expt. the basicity of the amino group (16), which would account for 2). Since, in Expt. 2 of Table 5, the amount of the derivative the drop in pH observed by Yagi et at. (8). was greater than the amount of lysine present in the enzyme Our conclusion that the imino acid is released as a free (10), we conclude that lysine residues in the bovine serum product is supported by the racemic nature of the amino acid albumin had reacted. It is of interest that no derivative was formed by borohydride reduction. It is also consistent with the found in similar experiments in which the borohydride was fact that this product can react with the lysine amino groups added at zero time, although under such conditions ialanine of another protein, such as bovine serum albumin. The finding continues to be formed (Table 1, Expt. 2). The same deriva- that free lysine also reacts to form a Schiff's base with the tive as obtained from an acid hydrolysate of reductively product of the oxidase reaction indicates that no special labeled D-amino acid oxidase was also obtained without reactivity or protein environment is needed for an amino hydrolysis from the dialysate of a reaction mixture containing group to react. free ilysine, D-amino acid oxidase, D-alanine, and sodium From the data available it may be concluded that the borohydride (see Table 6). lysine residues in D- and iamino acid oxidases that are la- beled by borohydride in the presence of substrate are not at the DISCUSSION active center of these enzymes and do not participate in the The results presented show that the product of the amino acid oxidative reaction. This is supported by the following find- oxidase reaction is a short-lived compound capable of being ings: (a) It was found by Wellner and Zelazo (quoted in ref. reduced to amino acid by sodium borohydride. This represents 17) that sodium borohydride does not inactivate iamino acid the most direct evidence available that this product is an imino oxidase, either in the presence or absence of substrate. This acid. Our conclusions may be summarized by the reactions was confirmed by Massey et al. (18), who also obtained the indicated in Scheme 1 (in which the compounds are repre- same result with D-amino acid oxidase. (b) When borohydride qheme 1 R R R R H-C-NH2 amino acid oxidase l=NH H-C-NH2 + NH2-C-H COOH COOH COOH COOH (racemic) (optically active) H20 R'-NH, R R I HO-C-NH2 R'-NH-C-NH2 COOH COOH I NH3 NH, R R R N C==O R'-N=C R'-Nl-COOH COO"HI COOH COOH sented without regard to their state of ionization). We con- and substrate are added simultaneously to D-amino acid clude that the imino acid is rapidly hydrolyzed to the cor- oxidase, no labeling of the enzyme results, although the en- responding keto acid and ammonia, presumably through the zyme continues to catalyze substrate oxidation under these intermediate formation of a carbinolamine. The extreme conditions (Table 1). This finding may be understood in teims lability of the imino acid is demonstrated by the fact that no of Scheme 1 presented above if the rate of imino acid reduction appreciable amount of accumulated imino acid was trapped by is much faster than that of Schiff's base formation. (c) It was sodium borohydride treatment when the oxidase reaction was found by DeLuca (quoted in ref. 19) that D-amino acid simultaneously stopped with benzoate (see Table 2). Since the oxidase labeled with substrate in the presence of borohydride rate of reduction of the imino acid is about 36 times that of its does not lose its label during subsequent turnover. We have hydrolysis (Table 3), any imino acid present would have been made similar findings (unpublished) with iamino oxidase and largely reduced to the corresponding racemic amino acid. [1-'4C]r-leucine. If the labeled lysine residues were a part of These data are not in agreement with the conclusions of Yagi the active site, either the labeled enzyme should be inactive, or et al. (8), who have postulated the accumulation of imino acid the bound substrate should be removed in a subsequent turn- on the basis of a transient drop in pH in the course of oxidation over. The finding that only a few lysine residues in the enzyme of D-alanine and D-leucrne by D-amino acid oxidase. Although molecule become labeled may be a reflection of the steady it is clear that more work is needed to resolve this problem, an state existing at the time of borohydride addition, resulting Downloaded by guest on September 26, 2021 Proc. Nat. Acad. Sci. USA 68 (1971) Imino Acids as Reaction Products 991

from the simultaneous formation and hydrolysis of substrate- 4. Meister, A., D. Wellner, and S. J. Scott, J. Nat. Canc. Inst., 24, 31 (1960). lysine Schiff's bases. The possibility must also be considered 5. Radhakrishnan, A. N., and A. Meister, J. Biol. Chem., 233, that some lysine residues are more reactive than others and 444 (1958). that the borohydride reaction may provide a means of identi- 6. Taborsky, G., Yale J. Biol. Med., 27, 267 (1955). fying such residues. 7. Pitt, B. M., J. Amer. Chem. Soc., 80, 3799 (1958). It was shown by Massey et al. (18) that in the presence of 8. Yagi, K., M. Nishikimi, N. Ohishi, and A. Takai, Biochim. Biophys. Acta, 212, 243 (1970). sodium borohydride the flavin of both D- and iamino acid 9. Coffey, D. S., A. H. Neims, and L. Hellerman, J. Biol. oxidases is reduced to a new species, identified as the "3,4- Chem., 240, 4058 (1935). dihydro" derivative of FAD (20). The borohydride-reduced 10. Hellerman, L., and D. S. Coffey, J. Biol. Chem., 242, 582 enzyme was further reduced in the presence of substrate and (1967). and R. Bennett, Biochim. Biophys. was the assays. The fact that the 11. Massey, V., G. Palmer, still active in standard Ada, 48, 1 (1961). enzyme was active was presumed to be due to reoxidation of 12. Wellner, D., and A. Meister, J. Biol. Chem., 235, 2013 the flavin to FAD under the assay conditions. However, our (1960). finding that D-amino acid oxidase is enzymatically active in 13. Spackman, D. H., W. H. Stein, and S. Moore, Anal. Chem., the presence of borohydride suggests that the borohydride- 30, 1190 (1958). 14. Jeffay, H., and J. Alvarez, Anal. Chem., 33, 612 (1961). reduced form of FAD may be active as a coenzyme throughout 15. Hochreiter, M. C., and K. A. Schellenberg, J. Amer. Chem. the catalytic cycle. Soc., 91, 6530 (1969). 16. Hine, J., and C. Y. Yeh, J. Amer. Chem. Soc., 89, 2669 (1967). This work was supported in part by grant AM-12068 from the 17. Wellner, D., Annu. Rev. Biochem., 36, 669 (1967). National Institutes of Health, U.S. Public Health Service. 18. Massey, V., B. Curti, F. MIller, and S. G. Mayhew, J. Biol. Chem., 243, 1329 (1968). 19. Neims, A. H., and L. Hellerman, Annu. Rev. Biochem., 39, 1. Knoop, F., Z. Physiol. Chem., 67, 489 (1910). 867 (1970). 2. Krebs, H. A., Kin. Wochschr., 11, 1744 (1932). 20. Maller, F., V. Massey, C. Heizmann, P. Hemmerich, J.-M. 3. Dakin, H. D., J. Biol. Chem., 67, 341 (1926). Lhoste, and D. C. Gould, Eur. J. Biochem., 9, 392 (1969). Downloaded by guest on September 26, 2021