Purine nucleotide cycle. Evidence for the occurrence of the cycle in brain.

Cell-free extracts of rat brain catalyze the reactions of the purine nucleotide cycle. Ammonia is formed during the deamination but not the amination phase of the cycle. The activity of adenylate deaminase in brain is sufficient to account for the maximum rates of ammonia production that have been reported. The activity of glutamate dehydrogenase is not sufficient to account for these rates of ammonia production. The activities of adenylosuccinate synthetase and adenylosuccinase are nearly sufficient to account for the steady state rates of ammonia production observed in brain. Demonstration of the cycle in extracts of brain is complicated by the occurrence of side reactions, in particular those catalyzed by phosphomonoesterase, nucleoside phosphorylase, and guanase.

Cell-free extracts of rat brain catalyze the reactions of the purine nucleotide cycle. Ammonia is formed during the deamination but not the amination phase of the cycle. The activity of adenylate deaminase in brain is sufficient to account for the maximum rates of ammonia production that have been reported. The activity of glutamate dehydrogenase is not sufficient to account for these rates of ammonia production. Objections to this mechanism have been raised on experimental grounds (2, 4) and on theoretical grounds (5). The glutaminase reaction, which converts the amide group of glutamine to ammonia, is another possible source of ammonia. However, glutamine can be ruled out as a major source of ammonia in brain for several reasons. Deamidation of endogenous glutamine can account for only 5% of the ammonia formed by brain slices (6). Administration of methionine sulfoximine, an inhibitor of glutamine synthetase, results in a 5-fold increase in the concentration of ammonia in mouse brain (7). The increase occurs before any fall in glutamine can be observed (8 Evidence is presented below which shows that the reactions of the purine nucleotide cycle occur in extracts of rat brain, and that their activities are such that they can account for the rates of ammonia production observed in vitro and in uiuo.

Preparation of Brain Entract
Male rats weighing 350 to 500 g were decapitated.
The whole brains (1.6 to 1.8 g each) were removed, rinsed free of blood, and homogenized with a TenBroeck homogenizer (an all glass hand-operated homogenizer), in 3 ml/g of brain of T&citrate buffer, pH 7.3 (20 mM Tris base and 6.4 mM citric acid), containing 1 mM dithiothreitol. The extract was centrifuged first at 31,000 x g for 20 min in a Sorvall RC2-B centrifuge, and then at 85,000 x g for 30 min in head number 30 of a Spinco model L preparative centrifuge. The high speed supernatant was freed of endogenous metabolites by passage through a column packed with Sephadex G-25 (coarse grade) which was pre-equilibrated with a solution containing Tris-citrate buffer, pH 7.3 (20 mM Tris base and 6.4 rnM citric acid), 15 rnM potassium phosphate, 150 mM KCI, and 1 mM dithiothreitol.
The pooled protein solution was concentrated by ultrafiltration using an Amicon diaflo membrane (type PM-30). The concentrated brain extract, which contained 29 mg of protein/ml, was used in the cycling experiments.
Brains of three rats were pooled for the determination of enzyme activities.
The brains were treated as described above, except that additional extractions were carried out. The residue from the low speed centrifugation was homogenized again in 2 ml of buffer/g of brain, and centrifuged at 31,000 x g as above. The process of re-extraction was repeated up to six times, and a separate high speed supernatant was prepared from each low speed supernatant.

Cycling Experiment
The An equivalent volume of water was added to the control. The progress of the reactions was followed by removing 0.6-ml samples for spectral analysis at 5.min intervals. The samples were added to a quartz cuvette and the light path was reduced to 1 mm by the insertion of a quartz spacer g-mm thick. Spectral scans were made from 310 nm to 240 nm on a Perkin Elmer model 356 spectrophotometer. The changes in total adenine nucleotide were calculated using the following equations (9): AAdenine nucleotide The spectral scans showed that amination came to a stop after l'% hours, and that deamination began spontaneously. The spectral scans were continued for another 1.5 hours until deamination stopped. Other samples (1.0 ml) were removed from the reaction mixture at various times and deproteinized by addition to an equal volume of 12% perchloric acid. After centrifugation, these samples were neutralized with a mixture of 2 N KOH and 0.5 M triethanolamine, and allowed to stand on ice for 20 min. Insoluble potassium perchlorate was then removed by centrifugation, and the samples were frozen rapidly and stored frozen for later analysis of metabolites.
Enzymes of the purine nucleotide cycle and of various side reactions were assayed before the start of the cycling experiment and after they had been in the reaction mixture for 4 hours at 30". in a final volume of 3.0 ml. The purines and purine nucleosides were assayed by their conversion to uric acid using the sequential addition of the following enzymes: xanthine oxidase for the assay of hypoxanthine plus xanthine; followed by nucleoside phosphorylase for the assay of inosine; followed by adenosine deaminase for the assay of adenosine; followed by guanine deaminase (guanase) for the assay of guanine plus guanosine. The formation of uric acid was measured on the double beam spectrophotometer set to read absorbance at 293 nm minus absorbance at 350 nm.

Enzymatic
Determination of IMP, Adenine Nucleotide, and Guanine Nucleotide The following were preincubated at room temperature: 0.03 ml of 1 M glycine-NaOH buffer, pH 9.0, 0.05 ml of deproteinized extract, and 20 ~(g of snake venom 5'-nucleotidase (specific activity 20 units/mg). Under these conditions, the preparation of snake venom used by us hydrolyzed IMP to inosine; AMP, ADP, and ATP to adenosine; and GMP, GDP, and GTP to guanosine. After 45 min the volume was adjusted to 3 ml with 0.1 M potassium phosphate buffer, pH 7.4, and inosine, adenosine, and guanosine were assayed as described above. This assay yields the sum of nucleotides and nucleosides. The difference between this assay and the assay for nucleosides described above yielded IMP and the sums of both AMP plus ADP plus ATP, and GMP plus GDP plus GTP. AMP was determined according to Williamson and Corkey (ll), except that we used the double beam spectrophotometer set to read absorbance at 400 nm minus absorbance at 340 nm. Protein was determined by the method of Lowry et al. (12).

Enzyme Assays
Each assay was based on the reference cited, but was modified to yield optimum conditions with extracts of rat brain. The enzyme rates were linear as a function of protein concentration over the range quoted.

Adenylate Deaminase
The deamination of AMP to IMP was measured by following the decrease in absorbance at 262.5 nm minus 310 nm. The reaction mixture contained 27 mM imidazole-HCl buffer, pH 7.0, 50 mM LiCl, 2 mM AMP, 2 mM ATP, and high speed supernatant of rat brain, 0.05 to 0.2 mg of protein, in a volume of 0.6 ml. The light path was 0.05 mm (13). of rat brain, 0.5 to 1.0 mg of protein, in a total volume of 0.6 ml. The light path was 1 mm. The presence of adenylate deaminase does not interfere with this assay since AMP and IMP have the same absorbance at 282 nm. 1

Adenosine Deaminase
The deamination of adenosine to inosine was measured by following the decrease in absorbance at 262.5 nm minus 310 nm. The reaction mixture contained 0.1 M potassium phosphate buffer, pH 7.4, 0.5 mM adenosine, and high speed supernatant of brain, 0.05 to 0.1 mg of protein, in a volume of 0.6 ml. The light path was 1 mm (14).

Nucleoside Phosphorylase
The rate of conversion of inosine to hypoxanthine was measured in a coupled assay by following the rate of formation of uric acid as described below. The reaction mixture contained 0.1 M potassium phosphate buffer, pH 7.4, 0.5 mM inosine, xanthine oxidase, 0.02 unit/ml, and high speed supernatant of rat brain, 0.05 to 0.1 mg of protein, in a total volume of 3 ml (15).

Guanase
The rate of conversion of guanine to xanthine was measured in a coupled assay by following the rate of formation of uric acid as described below. The complete reaction mixture contained 0.1 M Tris-HCl buffer, pH 8.8, 0.07 mM guanine, xanthine oxidase, 0.04 unit/ml, and high speed supernatant of rat brain, 0.05 to 0.1 mg of protein, in a total volume of 3 ml (16).

Xanthine Oxidase
The formation of uric acid from hypoxanthine was measured by following the change in absorbance at 293 nm minus 350 nm. The complete reaction mixture contained 0.15 mM hypoxanthine, 0.1 M potassium phosphate buffer, pH 7.4, and 1 to 2 mg of high speed supernatant of rat brain, in a total volume of 3 ml (17).
'These conditions are based on routine assays carried out in our laboratory. of rat brain, 5 to 10 pg of protein, in a total volume of 3 ml. The change in absorbance was measured at 400 nm minus 340 nm (18).

5'-Nucleotidase
The complete reaction mixture contained 1 rnM AMP, 24 mM imidazole-HCI buffer, pH 7.0, 10 rnM MgCI,, and high speed supernatant of rat brain, 1 to 2 mg of protein, in a total volume of 1 ml. The mixture was incubated at 30" for 20 min. Inorganic phosphate was measured according to the method of Mozersky et al. (20). The assays were performed in the presence and in the absence of 1 rnM adenosine oc,P-methylene diphosphonate, a potent inhibitor of 5'-nucleotidase (21). The difference between the rates of hydrolysis of AMP in the presence and absence of adenosine ol,fl-methylene diphosphonate is quoted as the 5'.nucleotidase activity.

p-Nitrophenyl Phosphatase
The reaction mixture contained 5 rnM p-nitrophenyl phosphate, 27 mM imidazole-HCl buffer, pH 7.0, 8.3 rnM MgCl,, and high speed supernatant of rat brain, 0.3 to 0.6 mg of protein, in a total volume of 2 ml. The control lacked substrate.
Aliquots of 0.1 ml were withdrawn at 3-min intervals and added to 0.9 ml of 0.1 M borate buffer, pH 9.5, and the absorbance was measured at 400 nm using a Zeiss PMQ II spectrophotometer (22).

Purine Nucleotide Cycle Enzymes in Rat Muscle
Rat muscle was extracted in 90 mM potassium phosphate buffer, pH 6.5, 180 mM KCI, and a high speed supernatant was prepared as described previously (9

RESULTS
As is shown below, preparations of brain cytoplasm catalyze the conversion of IMP to AMP and of AMP to IMP. The demonstration of these reactions is complicated by the occurrence of various side reactions.
Conversion of IMP to AMP-The formation of adenine nucleotide from IMP begins immediately after the addition of aspartate. Fig. 1 shows that it is accompanied by a concomitant decrease in hypoxanthine compounds. No accumulation of adenylosuccinate was observed. This is not unexpected since in rat brain the specific activity of adenylosuccinase is about five times that of adenylosuccinate synthetase. Nucleotide analyses show that all of the adenine nucleotide formed was converted to ATP during this phase of the cycle. This indicates that AMP reacted with endogenous ATP' to produce ADP via endogenous adenylate kinase. The ADP was in turn phosphorylated to ATP by the action of endogenous pyruvate kinase and added phosphoenolpyruvate.
The nucleoside triphosphate regenerating system is required for the adenylosuccinate synthetase step, insofar as it regenerates GTP from GDP and thereby counters the powerful inhibitory effect of GDP on the enzyme. In addition the regenerating system prevents the deamination of AMP by causing the conversion of AMP to ATP. As soon as phosphoenolpyruvate becomes depleted the regenerating system stops functioning, amination of IMP ceases, and deamination begins spontaneously. For each molecule of IMP aminated to AMP, one molecule of GTP is converted to GDP. The conversion of the AMP and GDP to their respective triphosphates consumes 3 molecules of phosphoenolpyruvate.
In the absence of side reactions, a 3-fold excess of phosphoenolpyruvate over IMP should therefore be sufficient to bring about complete conversion of IMP to ATP. In practice a much larger excess of phosphoenolpyruvate has to be added, because of the high phosphatase activity of the extract, which continually hydrolyzes GTP, ATP, and IMP. In the experiment shown in Fig. 1, the amination came to a halt at 80% of completion and the amount of phosphoenolpyruvate used was 7 times the amount expected from the amount of IMP consumed.
The IMP added to the reaction mixture is also hydrolyzed by phosphatases (reaction I) and the resulting inosine is acted upon by nucleoside phosphorylase (reaction II). Both enzymes are present in the brain extract. During the preincubation period, inosine and hypoxanthine accumulate. The hypoian- inosine + P,cz hypoxanthine + rihose l-phosphate (II) thine formed in reaction II is converted back to IMP by hypoxanthine-guanine phosphoribosyltransferase. This is discussed later. The conversion of IMP to inosine might also be partially reversed by inosine kinase (reaction III) (24), although such activity has not been demonstrated in brain. When the amina-*An amount of ATP equivalent to 2 PM in the final reaction mixture escaped separation by the gel filtration procedure. x, ammonia. A base-line value of 0.08 rnM ammonia was subtracted from al1 points plotted in the figure. inosine + ATP + IMP + ADP (III) tion is started by adding aspartate, the levels of inosine and hypoxanthine drop.
Conversion of AMP to IMP- Fig.  1 shows that amination stops at 1% hours, and that deamination begins spontaneously. The deamination was followed to about 92% of completion. Throughout the amination phase, which occurs during the condition of "energy excess," guanine nucleotides are maintained in the form of the triphosphate. GMP appears in the reaction mixture for the first time when the phosphoenolpyruvate is used up and deamination begins. Shortly after the onset of deamination, total guanine nucleotides begin to decrease. This occurs by the following sequence of reactions: GMP + H,O + guanosine + P, guanosine + P, H guanine + ribose 1-phosphate (V) wu GMP is hydrolyzed to guanosine (reaction IV), and this in turn is converted to guanine by nucleoside phosphorylase (reaction V). The final step is the irreversible deamination of guanine to xanthine by guanase (reaction VI). Guanase is specific for guanine and does not react with either guanosine or GMP. The decrease in guanine nucleotides lags behind the decrease in adenine nucleotides because of the relatively slow rates of reactions IV and V. Xanthine does not react further due to the absence in the brain extract of xanthine oxidase (Table 1). The increase in hypoxanthine compounds agrees well with the decrease in adenine plus guanine nucleotides during the deamination half of the cycle.
In some of the 12 similar experiments that were run the spectra obtained became distorted during the deamination phase. The characteristic peak at 262.5 nm (the wavelength of the difference maximum between AMP and IMP) shifted, sometimes to shorter and other times to longer wavelengths, and the isosbestic point at 250 nm disappeared. As a result, the calculations based on changes at 262.5 nm and 270 nm became increasingly divergent. Thus during deamination, the spectral method for following the interconversion of purine nucleotides was, in some experiments, less accurate than the enzymatic analysis.
The spectral distortions that occur during the deamination phase of some experiments are probably due to side reactions involving guanine nucleotides. These side reactions do not necessarily occur at the same rate in both the complete reaction mixture and the control which lacks aspartate, since phosphoenolpyruvate may not be completely exhausted at the same time in the two systems. Although guanine breakdown is always observed during the deamination phase of the cycle, the occurrence of spectral shifts seems to depend on a difference in the reaction rates of the control and the test cuvettes.
Brain tissue contains hypoxanthine-guanine phosphoribosyltransferase in high activity (25). The enzyme converts the purine bases guanine and hypoxanthine into their respective monophosphates, while the other product, pyrophosphate, is hydrolyzed to orthophosphate by pyrophosphatase (reactions VII and VIII). 5-Phosphoribosyl l-pyrophosphate (PP-ribose-P) is synthesized in the brain from ribose 5-phosphate and ATP (reaction IX). Ribose 5-phosphate is generated from ribose 1-phosphate by the enzyme phosphoribomutase (reaction X) (26). This sequence of reactions together with the reaction catalyzed by nucleoside phosphorylase constitutes the salvage guanine + PP-ribose-P -GMP + PP, + 2 Pi (VII) hypoxanthine + PP-ribose-P H IMP + PP, -2 P, ATP + ribose 5.phosphate u PP-ribose-P + AMP (IX) ribose 1-phosphate tf ribose 5-phosphate w pathway whereby purine bases and nucleosides are reconverted to nucleotides. The brain extracts used by us contained al1 of these enzymes as is demonstrated in the experiment shown in Table II. Hypoxanthine is converted to IMP in the presente of either PP-ribose-P or ribose 5-phosphate plus ATP. Ribose l-phosphate is also effective in generating IMP, but it reacts more slowly. The concentrations of the substrates used in the experiment are similar to those found during the amination phase of the cycling experiment, but the protein concentration was only half. Phosphoenolpyruvate was added to prevent the breakdown of ATP to AMP by endogenous ATPases, which Ammonia Production in Brain  489   TABLE I  TABLE II Enzyme actiuities in particle-free extracts of rat brain and skeletal muscle Brains of three rats were pooled, multiple extractions were made, and maximum enzyme activities were assayed as described under "Experimental Procedure." Activities were also assayed under conditions approximating those of the cycling experiment shown in Fig. 1 would in turn allow the formation of IMP by AMP deaminase. UMP was added to reduce the hydrolysis of the newly synthesized IMP by acting as a competitive substrate of phosphatases. The rates of IMP production which are reported in Table II are probably minimum values since the protection of IMP hydrolysis by UMP may be incomplete. UMP afforded greater protection after 5 min of incubation, probably because more IMP was present at 15 than at 5 min.
Ammonia Production-In the experiments reported here, ammonia comes from the deamination not only of adenine but also of guanine nucleotide.
Ammonia formation was not observed until the nucleotides began to be deaminated. Thereafter the rate of ammonia production closely followed the rate of disappearance of adenine plus guanine nucleotides, and the rate of appearance of IMP plus inosine plus hypoxanthine plus xanthine (Fig. 1). The stoichiometry between ammonia appearance and nucleotide disappearance showed a discrepancy of about 0.06 mM during the deamination phase. This may be due to a systematic error in the analyses.
In a separate experiment, in which endogenous metabolites were not removed from the brain extract by passage through Sephadex, no significant changes in the concentrations of glutamine and glutamate were detected. These substances were measured by standard enzymatic methods (27,28), and a change of 0.05 mM could easily have been detected.
Under conditions which are similar to those that occur during the deamination phase of the cycle, adenosine deaminase and adenylate deaminase activities of the brain extract are similar (Table I). (Under these conditions, adenylate deaminase activity is only a small fraction of its maximum because the enzyme is only partially activated.) This raises the possibility that the adenosine deaminase reaction may contribute substantially to ammonia production in the experiment shown in Fig. 1. However, hydrolysis of 5'-mononucleotides is very slow, and the adenosine concentration was an average of 6 pM during the deamination phase. At this concentration the activity of adenosine deaminase is sufficient to account for about one-third of the adenine containing compounds deaminated.
Activities of Purine Nucleotide Cycle Enzymes in Brain-Adenylate deaminase, adenylosuccinate synthetase, adenylosuccinase, adenylate kinase, and pyruvate kinase activities in extracts of rat brain are shown in Table I. Activities are quoted both under optimal conditions and under conditions comparable to those found during the cycling experiments. Adenylosuccinate synthetase is the least active enzyme of the Ammonia Production in Brain cycle, and under the conditions of the in vitro cycling experiment, it operates at close to its V,,,. Fig. 1 shows that adenylate deaminase operates at an activity that is comparable to that of the synthetase, in other words far below its maximum activity. This is to be expected from the results shown in Table I for a number of reasons. The maximum concentration of AMP during the deamination phase is only 0.2 mM, which is one-tenth that used for the optimum assay. The ATP concentration drops rapidly from 0.15 mM at the beginning of deamination (80 min) to less than 0.01 mM halfway through the deamination, a concentration that is insufficient to activate the enzyme appreciably. Moreover, enzyme concentration uersus activity measurements show that the activity of adenylate deaminase decreases with increasing protein concentration. While the assays quoted in Table I contained a maximum protein concentration of 0.3 mg per ml, the cycling experiment contained 9.7 mg/ml. This 30.fold difference in protein concentration results in a decrease of the specific activity of adenylate deaminase of at least 60%. Another factor in the cycling experiment is the presence of orthophosphate, an inhibitor of the enzyme, which rises from 5 mM at the start to about 13 mM at the onset of deamination.
The rate of deamination is not limited by the adenylate kinase activity of the brain extract. In a separate cycling experiment, addition of exogenous adenylate kinase, in amounts up to four times the endogenous adenylate kinase activity, either at the beginning of the experiment or at the onset of deamination, did not significantly alter the rate of deamination.
No loss of adenylate kinase activity occurred when the complete reaction mixture was incubated at 30" for 4 or 6 hours. Adenylosuccinate synthetase activity decreased by 50% after 4 hours at 30". The other enzyme activities involved in the purine nucleotide cycle or its side reactions remained unaltered, except that adenylosuccinase and nucleoside phosphorylase decreased by 20% during this period. Table I also shows the activities in particle-free extracts of brain of nucleoside phosphorylase, guanase, and p-nitrophenyl phosphatase.
5'.Nucleotidase is virtually absent from the particle-free extract used in the cycling experiment (0.02 pmol/g fresh weight/min at 30"). Assays of the whole homogenate of brain showed the presence of 5'.nucleotidase equivalent to 1.3 pmol/g fresh weight/min at 30". No xanthine oxidase activity was detected in the extracts, in harmony with the observation that hypoxanthine and xanthine but not uric acid accumulated during the deamination phase of the cycle. Activity of Purine Nucleotide Cycle Enzymes in Skeletal Muscle-Activities of the purine nucleotide cycle enzymes in muscle are quoted in Table I for comparison with those in brain. The maximum capacity for the amination of IMP in muscle is about 10 times that found in brain, since in muscle amination is limited by the activity of adenylosuccinase, in contrast to brain where the synthetase is the least active enzyme. The maximum capacity of adenylate deaminase in muscle is 30 times that found in brain.

DISCUSSION
Rates of ammonia production by rat brain in uiuo are related to the level of cerebral activity. An initial rate of 7.5 wmol/g fresh weight/min was observed in rat brain after application of convulsive agents (7,29,30). This rate is within the capacity of adenylate deaminase (Table I) (5.2 pmol/g fresh weight/min at 30", which is equivalent to 8.3 rmol/g/min at 38"),3 but not of glutamate dehydrogenase (Table III, and see below). Ammonia may also be derived from adenosine, guanine, and guanosine, but at a maximum rate of only 2.3 rmol/g/min at 38" (30). Several workers have determined the activity of glutamate dehydrogenase in brain (31-33), but these measurements were all made in the direction of reductive amination of a-ketoglutarate. In order to estimate the maximum rates of ammonia production it is necessary to measure the activity of the enzyme in the direction of oxidative deamination of glutamate. Table III shows the rates of amination and deamination catalyzed by glutamate dehydrogenase of rat brain at several pH values, with either DPN or TPN, and in the presence and absence of the activator ADP. The maximum rate of deamination of glutamate is 2.2 wmol/g fresh weight/min at 30" and pH 8.0, which is equivalent3 to 3.5 pmol/g/min at 38". This is insufficient to account for the maximum rates of ammonia production observed in uiuo. At lower pH values the maximum rates of glutamate deamination are even lower.' Other experimental findings show that the glutamate dehydrogenase reaction can at most account for only 50% of the ammonia produced by brain slices in the absence of added substrates (4, 34). However, these experiments did not prove that any ammonia is actually produced via this reaction. Weil-Malherbe and Gordon (4) found that in the presence of 10 mM 5-bromofuroate, an inhibitor of glutamate dehydrogenase, aerobic ammonia formation by slices of brain cortex continued at 95% of the control rate during the 1st hour of incubation, but that bromofuroate caused an increasing inhibition of ammonia production during the 2nd and 3rd hour. We have found that 5-bromofuroate inhibits brain adenylosuccinate synthetase. Dixon plots (35) show the inhibitor is competitive with respect to IMP, with a K, of approximately 5 mM, and noncompetitive with respect to aspartate with a K, > 15 mM. Bromofuroate also inhibits glutamate dehydrogenase of rat brain. The inhibition is competitive with respect to glutamate, with a K, of 0.9 mM. Although 5-bromofuroate is a weaker inhibitor of adenylosuccinate synthetase, 10 mM 5-bromofuroate causes a 40% inhibition of the brain enzyme when it is assayed in the presence of 80 FM JMP and 4 mM aspartate. Thus, the effect of high concentrations of bromofuroate on ammonia production, which was quoted as evidence in favor of the involvement of glutamate dehydrogenase, can also be interpreted in terms of an inhibition of adenylosuccinate synthetase. Weil-Malherbe and Gordon (4) showed further that n-aspartate and n-glutamate are weak inhibitors of ammonia formation by brain slices after the 1st hour of incubation, and that this inhibition can be reversed by L-aspartate. They interpreted this as an effect on glutamate dehydrogenase, but this observation too is consistent with an inhibition of the adenylosuccinate synthetase step of the purine nucleotide cycle by n-aspartate (36) and possibly of transaminase by high concentrations of n-glutamate.
It has been demonstrated that rat brain dispersions produce 3Corrected from 30" to 38" by multiplying by 1.6 (this correction assumes that the rate increased by a factor of l.S/lO").
' Since the validity of this statement depends on the completeness of our extraction of glutamate dehyrogenase, it is worth pointing out that the maximum rate of reductive amination of a-ketoglutarate observed by us ('22 pmol/g fresh weight of brain/min at 30", Table III  Two rat brains were homogenized with a motor-driven Teflon and glass homogenizer in 9 ml of 10 mM potassium phosphate buffer, pH 7.4, 0.5 mM ADP, and 10 PM EDTA/g of brain. The homogenate was sonicated for 1 min and centrifuged, first at 10,000 x g for 10 min, and then at 165,000 x g for 30 min. The pellets from the low and high speed centrifugations were recombined and rehomogenized in 6 ml of buffer/g of brain. The processes of sonication and centrifugation were repeated as above. A total of three extractions were performed in this manner. Three additional extractions resulted in not more than 10% additional glutamate dehydrogenase activity. Prior to assay, the high speed supernatants so obtained were passed through a column of Sephadex G-25 which had been preequilibrated with the extraction buffer. ADP stabilizes glutamate dehydrogenase in the brain extract, and its activity remains unchanged for at least 12 hours. Activity in the direction of reductive amination was measured by mixing 50 mM Tris-acetate buffer (50 mM with respect to Tris), 0.1 mM EDTA, 83.3 mM NH&l, 6.7 mM cu-ketoglutarate, and 0.1 mM DPNH or 0.2 mM TPNH, in a total volume of 3 ml. The reaction was initiated by adding the brain extract. The enzyme contributed 4 to 8 PM ADP to the assay. The initial velocity was measured by following the change in absorbance at 400 nm minus 340 nm in a Perkin Elmer model 356 double beam spectrophotometer.
Activity in the direction of oxidative deamination was measured by mixing 50 mM Tris-acetate buffer (50 rnsr with respect to Tris), 0.1 mM EDTA, 33 mM sodium glutamate, and 0.6 mM DPN or 0.6 mM TPN, in a total volume of 3 ml. The change in absorbance at 340 nm minus 400 nm was measured as described above, except that the instrument was set to 0.1 absorbance unit full scale. All reactions were run in the absence and presence of 0.5 mM ADP. Enzyme activities are quoted as micromoles/g fresh weight of tissue/ min. The assays were performed at 30".  (Table  I). The steady rate is preceded by an initial burst of ammonia production of about 0.17 nmol/g/ min (calculated from an ammonia production of 5 nmol/g during the first 30 min). This exceeds the amount of adenylosuccinate synthetase measured by us, but not the capacity of adenylate deaminase. The total amount of ammonia produced during the initial burst exceeds the total adenine mononucleotide content of rat brain (3.0 to 3.6 nmol/g fresh weight (30,37)). The activity of adenylosuccinate synthetase can account for a maximum of 0.058 x 30 = 1.7 wmol of ammonia produced. The remainder (5.0 -1.7 = 3.3 pmol) may be produced by largely draining the adenine mononucleotides into IMP, and possibly by the action of glutaminase.
Guinea pig slices produce ammonia at a steady state rate of 0.13 pmol/g/min in the first 2 hours. Thereafter, rates of ammonia formation decrease (6). The activity of adenylosuccinate synthetase in extracts of guinea pig brain is the same as in rat brain5 which means that in these experiments about one-half of the ammonia produced cannot be accounted for in terms of adenylosuccinate synthetase activity. However, the slices were incubated in the absence of substrate for 5 hours, 'V. Schultz, unpublished experiments. and it seems possible that degradative processes contributed to ammonia production. For example, amide nitrogen in proteins (38) and nucleic acids (6) of brain have been shown to break down under similar conditions. It is also possible that under these nonphysiological conditions glutamate oxidation by glutamate dehydrogenase occurred to a significant extent.
Inhibition of ammonia formation during anaerobiosis and by inhibitors of electron transport was previously attributed to the requirement by glutamate dehydrogenase of DPN for ammonia production (6). However, under uncoupling conditions, the level of ATP, an activator of adenylate deaminase, is low and the level of orthophosphate, an inhibitor of the enzyme, is high. The exact experimental conditions are critical. Enhancement of ammonia formation by addition of uncouplers has been observed in a liver system (39) and in brain (34); this can be accounted for by an increase in the level of AMP while maintaining a reasonably high level of ATP.
The results presented above demonstrate that the purine nucleotide cycle occurs in brain, and that it may account for at least one-half the ammonia production observed with brain slices in the absence of added substrates. Measurements are in progress of rates of ammonia production by rat brain i n uiuo.