Action of Liver Glutamine Transaminase and L-Amino Acid Oxidase on Several Glutamine Analogs PREPARATION AND PROPERTIES OF THE 4-S, 0, AND NH ANALOGS OF a-KETOGLUTARAMIC ACID*

Abstract The l-glutamine analogs, l-albizziin (l-α-amino-β-ureido-propionic acid), S-carbamyl-l-cysteine, and O-carbamyl-l-serine were found to be substrates for purified rat liver glutamine transaminase, and the α-keto acid product formed in each case was found to cyclize to a lactam analogous in structure to the cyclic form of α-ketoglutaramic acid (2-pyrrolidone-5-hydroxy-5-carboxylic acid). Evidence was obtained that the initial product of transamination of albizziin, S-carbamylcysteine, and O-carbamylserine are the corresponding α-keto acids, which were found to be converted by ω-amidase (followed by spontaneous decarboxylation) to β-aminopyruvate, β-mercaptopyruvate, and β-hydroxypyruvate, respectively. Incubation of the glutamine analogs with l-amino acid oxidase from Crotalus adamanteus venom gave the corresponding cyclic lactam forms; the products obtained from albizziin and S-carbamyl-cysteine were rapidly and irreversibly dehydrated in acid or base to yield 2-imidazolinone-4-carboxylic acid and 2-thiazolinone-4-carboxylic acid, respectively. Neither α-ketoglutaramate nor 2-oxazolidone-4-hydroxy-4-carboxylic acid (which was isolated as the corresponding barium salt) was dehydrated under these conditions.

Previous studies on highly purified rat liver glutamine transaminase showed that this enzyme can catalyze transamination between a wide variety of a-keto acids and glutamine, glutamic acid y-ethyl ester, y-glutamyl methylamide, methionine, and ethionine (1). In the present studies we have examined the activity of the enzyme toward several analogs of glutamine in which the 4-methylene moiety of glutamine is replaced by S, 0, or NH.
In earlier work (2,3) it was shown that the ol-keto analog of glutamine, a-ketoglutaramic acid, exists predominantly in solution in an unreactive cyclic (lactam) form, 2-pyrrolidone-*This work was supported in part by a grant from the National Institutes of Health, Public Health Service. 8500  (7) as modified by Cooper and Meister (1). In some of the experiments transamination was followed by determining the disappearance of glyoxylate by the 2,4-dinitrophenylhydrazine procedure. Aliquots (5 ~1) were withdrawn and added to 0.1 ml of 0.1% 2,4-dinitrophenylhydrazine in 2 N HCl; after incubation at 37" for 20 min, 0.9 ml of 1 N potassium hydroxide was added and the absorbance at 430 nm was read against the blank lacking keto acid.
L-Amidase was assayed as described by Hersh (6). The specific activity of the w-amidase used in these studies was 600 to 750 units per mg. A unit is defined as the amount of enzyme that converts 1 pmole of cY-ketoglutaramate to Lu-ketoglutarate per hour under the conditions of assay. Ammonia was determined using a Conway (8) diffusion apparatus; color was developed with Nessler's reagent or by use of the indophenol procedure (9).
Paper chromatography was carried out by the ascending technique using the solvents given below. a-Keto acids were detected by the o-phenylenediamine technique (10). In this procedure, cu-keto acids may be located on paper chromatograms sprayed with o-phenylenediamine in trichloroacetic acid by their fluorescence under ultraviolet light. The procedure is very sensitive for a number of cr-keto acids; however, glyoxylate does not fluoresce, but quenches under these conditions. The cyclic compounds were detected by the Rydon and Smith procedure (11) as modified by Pan and Dutcher (12). Since Tris reacts under these conditions, the experiments in which chromatographic procedures were used were carried out in bicarbonate or borate buffers.

Transamination between Glyoxylate
and Glutamine Analogs-Rat liver glutamine transaminase was found to be active when glutamine was replaced by certain glutamine analogs. Thus, as indicated in Table I, substantial transamination was observed when the enzyme was incubated with glyoxylate and X-carbamyl-L-cysteine, 0-carbamyl-L-serine, or L-albizziin. Earlier studies (2,3,13) showed that transamination of glutamine with cY-keto acids yields the corresponding amino acids and the a-keto analog of glutamine, cr-ketoglutaramate, which spontaneously cyclizes to form the keto lactam, 2-pyrrolidone-5-hydroxy-5carboxylate. At pH 7.0 more than 99% of the a-keto analog of glutamine exists in the cyclic form, but, as stated above, the compound is a good substrate for o-amidase at values of pH above 7.5 because the rate of interconversion between the open chain and cyclic forms is rapid. Thus, as demonstrated previously (3,13) and in Table I, incubation of glutamine, glyoxylate, and rat liver glutamine transaminase in the presence of purified w-amidase leads to ammonia formation which is almost equivalent to the extent of transamination. In comparable experiments in which S-carbamyl-L-cysteine and glyoxylate were incubated with both the transaminase and the W-amidase, the formation of ammonia was appreciable but considerably less than the extent of transamination. (Relatively short incubation periods were used in experiments with X-carbamylcysteine because this amino acid readily decomposes to cysteine and cyanate at values of pH above 8.0 (14).) The findings suggest that the ol-keto analog of S-carbamyl-L-cysteine expected to be formed by transamination under these conditions is deamidated to the corresponding X-carboxy derivative, which would undergo spontaneous decarboxylation to yield &mercaptopyruvate. In agreement with these expectations, paper chromatographic studies of reaction mixtures identical to those described in Experiment 2 (Table I)  In studies in which 0-carbamyl-L-serine was incubated with glyoxylate in the presence of the transaminase and w-amidase, a small but significant amount of ammonia was formed (Experiment 3, Table I) ; paper chromatographic study of such reaction mixtures revealed the formation of ,&hydroxypyruvate. The latter compound would be expected to be produced bydecarboxylation of the product formed by enzymatic deamidation of the a-keto analog of 0-carbamylserine.
In analogous studies on transamination between glyoxylate and albizziin in the presence of w-amidase (Experiments 4 to 6; Table I), small but significant amounts of ammonia were formed. It is notable that the formation of ammonia was increased when larger amounts of w-amidase were added to the reaction mixture. Paper chromatographic studies revealed the formation of /?aminopyruvate. Deamidation of the cr-keto analog of albizziin would be expected to yield the corresponding /3-N-carboxy derivative, which on spontaneous decarboxylation would give 8501 &aminopyruvate.
While the latter compound has apparently not been previously described, the available data are consistent with its formation under these conditions.
The new compound formed in reaction mixtures containing albizziin, glyoxylate, glutamine transaminase, and w-amidase (but not in controls in which any one of these was omitted) gave a brown spot on paper chromatograms after treatment with the o-phenylenediamine reagent when viewed under ultraviolet light. Additional evidence for the formation of &aminopyruvate was obtained in studies in which albizziin was incubated with pyruvate, glutamine transaminase, and w-amidase under conditions similar to those given in Table I except that 50 units of w-amidase were added and the reaction mixture was incubated for 3 hours. After incubation, the reaction mixture was adjusted to pH 4.0 by addition of 1 N HCl.
The acidified reaction mixture was then treated with 1 M hydrogen peroxide, and after incubation for 10 min at 37" an aliquot was subjected to paper chromatography. These studies revealed the presence of glycine (the expected product of the reaction between ,&aminopyruvate and hydrogen peroxide) ; no glycine was formed in control studies in which reaction mixtures lacking w-amidase were treated with hydrogen peroxide.
The findings described above are consistent with the occurrence of Reactions 1, 5, and 6 indicated in Fig. 1. While the data indicate that the 3 analogs of glutamine studied undergo enzymatic and nonenzymatic reactions analogous to those previously observed with glutamine (2,3,13), it is evident that the analogs are less effectively transaminated than is glutamine, and that the corresponding cr-keto acid analogs are less susceptible to deamidation by w-amidase than is cr-ketoglutaramate. These considerations suggest that the ar-keto analogs of X-carbamylcysteine, 0-carbamylserine, and albizziin, like that of glutamine, undergo spontaneous cyclization. Studies described below demonstrate that this is the case.
y-Cyano-L-a-aminobutyrate was also found to effectively replace L-glutamine in the L-glutamine-glyoxylate transamination system. The rate of transamination (determined by measurement of the rate of glycine formation) under the conditions given in Table I was about 85y0 of that observed with glutamine. y-Cyano-ol-ketobutyrate was formed as determined by the 2,4dinitrophenylhydrazine procedure. Identification of the cr-keto acid by paper chromatography of the corresponding 2,4-dinitrophenylhydrazone was hindered by difficulty in separating the 2,4-dinitrophenylhydrazones of glyoxylate and y-cyano-a-ketobutyrate using the solvent systems previously described (15). However, these a-keto acids can easily be separated by paper chromatography of the corresponding L-y-glutamyl hydrazones (16). Thus, after incubation, the reaction mixtures were placed at 100" for 30 s and then cooled to 37"; 20 ~1 of 0.1 M L-y-glutamyl hydraaide were added and the mixture was incubated at 37" for 30 min. Paper chromatography of the L-y-glutamyl hydrazones led to separation of the derivatives of glyoxylate and y-cyano-ol-ketobutyrate (see below) ; authentic y-cyano-or-ketobutyrate was obtained by oxidation of the corresponding L-amino acid with L-amino acid oxidase in the presence of catalase by the general procedure previously described (17).

Oxidation
of G&amine Analogs by L-Amino Acid Oxidase-As indicated in Table II, the glutamine analogs discussed above are substrates of L-amino acid oxidase.
When the enzymatic oxidation was carried out in the presence of added w-amidase, no additional ammonia was formed in the studies with albizziin, S-carbamylcysteine, and 0-carbamylserine, but (in confirmation of earlier studies (3))) additional ammonia was formed in experi- L-Glutamine. 0-Carbamyl-L-serine. L-Albizziin ments with glutamine. These observations indicate that the cu-keto analog of glutamine is formed and hydrolyzed by w-amidase, but that the cll-keto analogs of albizziin, S-carbamylserine and 0-carbamylserine are not formed to a detectable extent under these conditions. Hafner and Wellner (18) have shown that the initial product formed in the oxidation of an L-amino acid by L-amino acid oxidase is the corresponding imino acid; this species has a half-life of several seconds at pH 8.0 and is extremely reactive.
According to Hafner and Wellner, imino acids react with semicarbazide at pH 8.0 at rates that are several thousand times greater than the rates of reaction of semicarbazide with the analogous cY-keto acids .l These workers have found that more than 95% of the imino acid formed can be trapped as the corresponding a-keto acid semicarbazone when the L-amino acid oxidase reaction is carried out on L-leucine in the presence of 100 mM semicarbazide. We therefore carried out the experiments described in Table III in which L-glutamine, 0-carbamyl-L-serine, and L-albizziin were incubated with L-amino acid oxidase in the presence of semicarbazide at pH 8.4; the absorbance of the cu-keto acid semicarbazone formed was determined at 248 nm. The results ob- d Estimated from the intensities of the spots found for 2-oxazolidone-4-hydroxy-4-carboxylate after paper chromatography and application of the modified Rydon and Smith procedure (11, 12). 6 The a-keto acid was determined as the 2,4-dinitrophenylhydrazone (1). tained with L-glutamine are similar to those obtained by Hafner and Wellnerr in studies on other n-amino acids that are oxidized by L-amino acid oxidase to yield open chain ar-keto acids; thus, a rapid increase in the absorbance at 248 nm was observed and the amounts of semicarbazone formed were virtually identical to those found for ammonia.
On the other hand, the oxidation of n-albizziin under these conditions did not lead to significant semicarbazone formation.
In experiments with O-carbamyl-nserine semicarbazone formation was much lower than ammonia formation.
The findings indicate that the imino acid derived from albizziin cannot be trapped under conditions under which at least 95% of the imino acid derived from glutamine can be trapped as the corresponding semicarbazone (Table III). The data show that about 35yo of the imino acid derived from Ocarbamyl-L-serine can be trapped. An experiment of this type could not be performed with S-carbamyl-L-cysteine since this compound reacts directly with semicarbazide to yield a derivative which absorbs at 248 nm.
The oxidation of the several glutamine analogs by L-amino acid oxidase was also studied in the absence of catalase and in the presence of added hydrogen peroxide (Table IV).
The findings indicate that the formation of product from glutamine and the 3 glutamine analogs was not substantially decreased by omission of catalase. In contrast, cr-keto acid formation from L-methionine in the absence of catalase was only 5% of that observed in the presence of catalase.* In the studies in which hydrogen peroxide was added, there was a marked decrease in the formation of product from n-glutamine, 0-carbamyl-n-serine, and L-methionine; in contrast, addition of hydrogen peroxide did not destroy the product obtained in the oxidation of Lalbizziin and X-carbamyl-L-cysteine.
Since it would be ex-2 The rate of decarboxylation is influenced by the concentration of hydrogen peroxide and may thus increase as the over-all rate of amino acid oxidation is increased.
Relatively rapid oxidation of n-glutamine by n-amino acid oxidase in the absence of catalase gives good yields of succinamic acid (2). petted that hydrogen peroxide decarboxylates the cr-imino acid (as well as the cY-keto acid), these findings are in accord with the view that the a-imino acids formed from r,-albizziin and X-carbamyl-L-cysteine undergo cyclization very rapidly, more rapidly than do the a-amino derivatives derived from glutamine and Ocarbamylserine.
This conclusion is consistent with the observations on the trapping of the cr-imino acids by semicarbazide (Table III).
Preparation and Properties of a-Keto Analogs of Albizziin, S-Carbamylcysteine, and 0-Carbamylserine-The findings described above indicate that the ru-keto analog of albizziin, which is formed in the transaminase reaction exhibits a marked tendency to undergo conversion to a form, presumably a cyclic derivative, which does not possess a reactive a-keto group.
The data also indicate that the G-nine acid formed by the action of L-amino acid oxidase on L-albizziin undergoes conversion to a cyclic form, apparently without prior conversion to the corresponding ru-keto acid. No evidence for the conversion of the cyclic form to the open chain form was obtained.
However, the open chain form of the cY-keto analog of albizziin is probably formed transiently in the transamination reaction; thus, in the studies described above in which n-albizziin was incubated with transaminase, w-amidase, and glyoxylate, ammonia was formed and evidence for /3-aminopyruvate formation was obtained. Similar conclusions appear applicable to the cr-keto analogs of S-carbamylcysteine and 0-carbamylserine. In order to further investigate the properties of these a-keto acids, the products formed by oxidation of the several glutamine analogs with L-amino acid oxidase in the presence of catalase were obtained in protein-free solution (pH 7 to 8) by ultrafiltration of the reaction mixtures.
Studies on these solutions indicated that the initial product of oxidation is in each case analogous in structure to the lactam form of a+ketoglutaramate (see below).
Attempts to isolate the oxidation products as the free acids led in two cases to dehydration of the cyclic products. Thus, upon acidification of the products obtained by oxidation of X-carbamyl-n-cysteine and L-albizziin, dehydration occurred leading to the formation of 2-thiazolinone-4-carboxylic acid and 2-imidazolinone-4-carboxylic acid, respectively. Some of the properties of the ru-keto acid analogs of S-carbamylcysteine, 0-carbamylserine, and albizziin are summarized in Table V; paper chromatographic  data are summarized  in  Table VI. The products obtained in solution after oxidation of these glutamine analogs by L-amino acid oxidase, like that obtained in the oxidation of L-glutamine, did not readily form a 2,4-dinitrophenylhydrazone, nor were these compounds decarboxylated by hydrogen peroxide under conditions in which pyruvic and glyoxylic acids as well as a great many other ar-keto acids are decarboxylated (3). The products obtained by enzymatic oxidation of glutamine and the three glutamine analogs did not react with o-phenylenediamine in a manner typical of cr-keto acids that possess a reactive cu-keto acid moiety.
In contrast, a-ketoglutaramic acid and the compounds obtained by oxidation of X-carbamyl-L-cysteine, 0-carbamyl-L-serine, and L-albizziin all formed a yellow complex on treatment with ferric chloride.
The product obtained by oxidation of 0-carbamyl-n-serine, like cu-ketoglutaramic acid, exhibited only end absorption in the ultraviolet (Fig. 2). Similar end absorption was observed with the initial products of oxidation of S-carbamylcysteine and albizziin.
However, on treatment of these products with alkali or acid, or after passage through a column of Dowex 50 (H+), the compounds derived from S-carbamylcysteine and albizziin ex-  hibited characteristic absorption in the ultraviolet as indicated in Fig. 2. No such changes in absorbance on acidification or on treatment with alkali were observed with the oxidation products obtained from 0-carbamylserine and glutamine. After the products obtained from S-carbamylcysteine and albizziin were acidified they failed to form a colored complex when treated with ferric chloride. Ultraviolet spectral studies of reaction mixtures containing the several glutamine analogs, glyoxylate, and glutamine transaminase indicate that the compounds formed from the glutamine analogs by transamination do not exhibit characteristic ultraviolet absorbance; on acidification of such reaction mixtures ultraviolet absorbance similar to that shown in Fig. 2 was observed only in the experiments with X-carbamyl-L-cysteine and n-albizziin.

%Thiazolinone-&carboxylic
Acid-S-Carbamyl-n-cysteine (164 mg; 1 mmole) was dissolved in 3 ml of water containing 200 mg of dialyzed Crotahs adamanteus venom and 1000 units of beef liver catalase. The solution was aerated at 37" for 24 hours and the protein was then removed by ultrafiltration with a Diaflow UM-2 membrane. The protein-free solution was passed through a column (volume, 2 ml) of Dowex 50 (H+), the effluent was then lyophilized. The white powder obtained was crystallized three times from boiling water and then dried in vucuo over phosphorus pentoxide at 25". The product darkened at 260" and melted between 283 and 286". The yield was 40 mg (29%). (5) the Rydon and Smith method (11,12)  8, 2-oxazolidone+Lhydroxy+carboxylic acid (from 0-carbamyl-n-serine).
-, in 10 mu potassium phosphate (pH7.0);---,inlNNaOH;.*.*,in2NHCl. Acid-L-Albizziin (200 mg; 1.36 mmoles) was oxidized with L-amino acid oxidase and the proteinfree solution was processed as described above. The crystalline product melted at 250" with decomposition; yield 92 mg (56%). CHON 4 4 3 2 Calculated: C 37.5 H 3.13 N 21.9 Found : C 37.4 H 3.29 N 21.9 Hilbert (19) reported a melting point of 261" for this compound; however, a product synthesized by the published procedure by Dr. Ralph Stephani of this laboratory exhibited a melting point of 251", and a mixed melting point between this material and the product obtained enzymatically was 250". The ultraviolet spectrum (Fig. 2) of the enzymatically prepared product was identical to that reported by Ditmer et al. (20). The infrared spectrum of the enzymatically prepared material and of the chemically synthesized product prepared in our laboratory were identical (Fig. 3).
Barium ~-Oxazolidone-4-hydroxy-4-carboxylate~-Carbamyl-~serine (300 mg; 2.0 mmoles) was dissolved in 10 ml of water containing 200 mg of Crotalus adamanteus venom and 1000 units of beef liver catalase; the procedure described above was followed. The free acid obtained after passing the solution of the product through Dowex 50 (H+) was found to be extremely deliquescent, and this product was therefore dissolved in 0.5 ml of water and treated with 0.1 M barium hydroxide until the pH of the solution reached 5.0. After removal of insoluble material by filtration, the barium salt was precipitated by addition of 5 volumes of ethanol at 0". The product was reprecipitated twice from water, washed with ethanol, and dried in vacua over phosphorus pentoxide at 25". Yield, 137 mg (30%).
CJLO~NBa~/2 Calculated: C 22.4 H 1.87 N 6.52 Found : C 21.9 H 2.12 N 6.22 The infrared spectrum of 2-imidazolinone-4-carboxylic acid (Fig. 3) exhibits a peak at 1600 cm-l which may be ascribed to the C=C moiety; this band is also present in the spectrum of 2-thiazolinone-4-carboxylic acid, but is not found in the infrared spectra of 2-pyrrolidone-5-hydroxy-5-carboxylic acid or 2-oxazolidone-4-hydroxy-4-carboxylic acid. The infrared spectra of the last two mentioned compounds are similar between 1800 and 1100 cm-l; these findings offer additional evidence that the product obtained by oxidation of 0-carbamyl-L-serine remains on acidification in the lactam form.
Studies in which the initial products of oxidation by L-amino acid oxidase of S-carbamyl-L-cysteine and L-albizziin were incubated with w-amidase did not lead to the formation of ammonia; similar results were obtained when the w-amidase was incubated with 2-thiazolinone-4-carboxylate, 2-imidazolinone-4-carboxylate, and 2-oxazolidone-4-hydroxy-4-carboxylate.
None of these compounds was found to be a substrate in enzymatic transamination with amino acids under conditions in which a-ketoglutaramate was shown to be active (1). DISCUSSION The findings indicate that L-albizziin, X-carbamyl-L-cysteine, and 0-carbamyl-L-serine are substrates of glutamine transaminase (Fig. 1, Reaction l), and are converted initially to an open chain ar-keto acid (II), which is in each case, like a-ketoglutaramate, a substrate for w-amidase. Thus, in experiments in which the glutamine analogs were incubated with both the transaminase and the w-amidase, we obtained evidence for the formation of ammonia (Table I), and the corresponding @-sub- (2) 2-imidazoline-4carboxylic acid (prepared by chemical synthesis); (3) 2thiazolinone-4-carboxylic acid; (4) 2-oxazolidonel-hydroxy-4carboxylic acid; (6) cr-ketoglutaramic acid (2.pyrrolidoned-hydroxyd-carboxylic acid) ; 0.33% in KBr.
stituted pyruvic acid derivative (Fig. 1, Reactions 5 and 6). The data also indicate that the open chain ac-keto acids (II) are rapidly converted to the lactam form (III) ; conversion of these cyclic forms (except for X=CHz) to the open chain forms (Reaction 2a) was not demonstrated in the present studies. The mechanism by which compounds of Structure III are formed from the corresponding amino acids by L-amino acid oxidase differs from that involved in the transaminase pathway. Thus, the initial enzymatic product of the L-amino acid oxidase reaction has been shown by Hafner and Wellner (18) to be the corresponding imino acid; this compound, which has been estimated to have a half-life of several seconds at pH 8,1 is hydrolyzed (in the case of most amino acids) to yield the corresponding cr-keto acid and ammonia.
The present findings (Table III) suggest that the imino acid formed in the oxidation of albizziin must cyclize very rapidly to Structure III (X=NH), presumably by hydrolysis of the intermediate 2-imidazolinone-4-amino-4-carboxylic acid. Such cyclization is also very rapid when O-carbamyl-nserine is the substrate, although some imino acid formation was detected with this substrate.
In contrast, the results with glutamine indicate a pathway typical of many other amino acids in which imino acid formation occurs; cyclization of the imino acid derived from glutamine would yield Structure III ( is supported by the studies in which w-amidase was added to the reaction mixtures; thus, Reactions 5 and 6 could not be demonstrated in studies in which the glutamine analogs were incubated with L-amino acid oxidase and w-amidase. Reaction 5 can be demonstrated with L-glutamine in the presence