Alanine Metabolism in the Perfused Rat Liver STUDIES WITH 15 N*

We have utilized [ 15 N]alanine or 15 NH 3 as metabolic tracers in order to identify sources of nitrogen for hepatic ureagenesis in a liver perfusion system. Studies were done in the presence and absence of physiologic concentrations of portal venous ammonia in order to test the hypothesis that, when the NH 4 (cid:1) :aspartate ratio is > 1, increased hepatic proteolysis provides cytoplasmic aspartate in order to support ureagenesis. When 1 m M [ 15 N]alanine was the sole nitrogen source, the amino group was incorporated into both nitrogens of urea and both nitrogens of glutamine. However, when studies were done with 1 m M alanine and 0.3 m M NH 4 Cl, alanine failed to provide aspartate at a rate that would have detoxified all administered ammonia. Under these cir-cumstances, the presence of ammonia at a physiologic concentration stimulated hepatic proteolysis. In perfusions with alanine alone, (cid:1) 400 nmol of nitrogen/min/g liver was needed to satisfy the balance between nitrogen intake and nitrogen output. When the model included alanine and NH 4 Cl, 1000 nmol of nitrogen/min/g liver were formed from an intra-hepatic source, presumably proteolysis. In this manner, the internal pool provided the cytoplasmic aspartate that allowed the liver

of the metabolism of alanine-N for urea synthesis. Yang et al. (3) have used tracer infusions of 15 NH 4 Cl in fasting dogs, a model that presents a potential challenge to urea synthesis if the delivery of pre-formed ammonia (largely derived from intestinal urea hydrolysis) exceeds the uptake of amino acids that serve as potential precursors of the aspartate nitrogen that is needed to support urea synthesis. They found that the only [ 15 N]urea isotopomer formed was U mϩ1 (containing one atom of 15 N), and they attributed this to an inability of ammonia to provide nitrogen to aspartate for incorporation into urea (3). A corollary of this observation was the finding that hepatic ammonia uptake was accompanied by the uptake of almost equimolar quantities of alanine, which presumably provided the ␣-amino group to aspartate. Alanine would serve this role following transamination to glutamate and then to aspartate for incorporation into urea. Since alanine is not produced by intestinal metabolism in these fasting dogs, it must be provided by peripheral tissues. One possibility is that proteolysis, perhaps in the liver itself, is a metabolic price that must be paid to compensate the inability of ammonia to serve as a source of nitrogen for cytoplasmic aspartate for incorporation into urea (3). The experiments of Lopez et al. (4) are also germane in this regard. These workers investigated amino acid, ammonia, and urea fluxes across the liver and intestines of fed and postabsorptive rats. In both of these physiological states amino acid uptake exceeded ammonia uptake. Alanine was the principal amino acid extracted by the liver, accounting for 33 and 25% of total hepatic amino acid extraction, respectively, in the fed and postabsorptive animals. In both situations there was a prominent hepatic output of glutamine (in the postabsorptive state glutamine output exceeded that of urea), which these researchers interpreted as a salvage process that conserves nitrogen arising from hepatic amino acid metabolism, especially that of alanine (4).
The process of urea synthesis involves equimolar consumption of NH 4 ϩ and aspartate-N. However, in pathological situations such as cirrhosis (5), cancer (6), renal failure (7), or chronic bacterial hydrolysis of urea and glutamine in the small intestine (3, 5, 8 -10), there is a high rate of production of NH 4 ϩ . Under these conditions, the portal blood does not provide the liver with an equimolar amount of aspartate-N to support the needs of hepatic ureagenesis. Therefore, in this study our aim was to explore the hypothesis that when ammonia in the portal venous system is present at physiologic concentration but nitrogen supply differs from the 1:1 (NH 4 ϩ :aspartate-N) stoichiometry, there is increased hepatic proteolysis to provide cytoplasmic aspartate so that formation of cytosolic argininosuccinate can keep pace with the rapid incorporation of ammonia into mitochondrial carbamyl phosphate.
To test this hypothesis, we have used our previously reported experimental and theoretical methodology that involves the use of 15 N-labeled substrates (11)(12)(13)(14) to explore hepatic nitrogen metabolism and, in particular, to determine the contribution of alanine nitrogen to urea synthesis. We also were able to discriminate between incorporation into urea from the mitochondrial ammonia and cytoplasmic aspartate pools as well as the incorporation of nitrogen into 2-N and 5-N of glutamine. We perfused liver with [ 15 N]alanine in the presence and absence of physiological portal venous concentrations of ammonia. The results show that alanine nitrogen can be used for incorporation into both nitrogens of urea and both nitrogens of glutamine. However, alanine is more effective in providing nitrogen for urea synthesis via cytosolic aspartate than through the mitochondrial ammonia. Similarly, alanine-N was a more effective source of the amino nitrogen of glutamine than of the amide nitrogen. We also found that the presence of physiological concentrations of ammonia increased hepatic alanine uptake and intra-hepatic proteolysis.

MATERIALS AND METHODS
Liver Perfusions-Livers from fed male Harlan Sprague-Dawley rats (weighing about 11-13 g) were perfused in the non-recirculating mode as described by Sies (15). The basic perfusion medium was a Krebs' saline continuously gassed with 95% O 2 , 5% CO 2 and containing lactate (2.1 mM) and pyruvate (0.3 mM) as metabolic fuels. Perfusion flow rate, pH, pCO 2 , and pO 2 (in influent and effluent media) were monitored throughout, and oxygen consumption was calculated. After 20 min of pre-perfusion we changed to a medium that contained, in addition to the lactate and pyruvate, either  15 NH 4 Cl was present at 100% isotopic enrichment. Separate perfusate reservoirs, each containing different media, were used to facilitate changes in perfusions. Samples were taken from the influent and effluent media for chemical and GC-MS analyses. At the end of the perfusions livers were freezeclamped with aluminum tongs precooled in liquid N 2 , and the frozen livers were ground into a fine powder, extracted into perchloric acid, and the extracts used for the analysis of adenine nucleotides by enzymatic techniques (16). Amino acids were determined by high pressure liquid chromatography, utilizing pre-column derivatization with o-phthalaldehyde (17). Ammonia and urea were assayed by standard methods (18,19).
GC-MS Methodology, Determination of 15 N-Labeled Metabolites-GC-MS measurements of 15 N isotopic enrichment were performed on a Hewlett-Packard 5970 MSD and/or 5971 MSD coupled with a 5890 HP-GC, as described previously (12)(13)(14). For measurement of 15 N enrichment in urea and amino acids, samples were prepared as we have described previously (12)(13)(14). Briefly, a 500-l aliquot of effluent or liver extract was purified via an AG-50 (H ϩ ; 100 -200 mesh; 0.5 ϫ 2.5 cm) column and then converted into t-butyldimethylsilyl derivatives. The m/z 231, 232, 233, and 234 of the urea t-butyldimethylsilyl derivative was monitored for singly labeled and doubly labeled urea determination (13,14,20). Isotopic enrichment in citrulline, glutamate, aspartate, and alanine was monitored using ratios of ions at m/z of 443/442, 433/432, 419/418, and 261/260, respectively. Enrichment in [2-15 N]glutamine was determined by monitoring the m/z 259/258 and [5-15 N]glutamine by the difference between m/z 432/431 and m/z 259/258 ratios (20). Doubly labeled glutamine was determined using m/z 433/431 ratio. 15 NH 3 enrichment was measured after conversion of ammonia to norvaline. To this end, we have modified the method previously published by Nieto et al. (21) as follows. After 5 min preincubation with ␣-ketoglutarate to remove any ammonia that may be present in glutamate dehydrogenase, effluent samples (500 l) were incubated for 30 min with 2-oxopentanoic acid to convert ammonia into norvaline as described (21). Incubation was stopped by addition of 1 ml of 4 N HCl, and then samples were passed through an AG-50 (H ϩ ; 100 -200 mesh; 0.5 ϫ 2.5 cm) column, washed with 4 ml of water, and norvaline was eluted with 4 N NH 4 OH (3 ml), and dried down at 60°C under compressed air. Thereafter, samples were reconstituted with 200 l of water:methanol:pyridine (60:32:18), and norvaline was derivatized by addition of 20 l of ethyl chloroformate. This mixture was gently shaken for about 1 min. The derivatized norvaline was extracted with 1 ml of ethyl acetate, dried down with gentle stream of nitrogen gas, and redissolved in 75 l of ethyl acetate for GC-MS analysis. The ions at m/z 144 and 145 were monitored, and m/z 145/144 ratio was used to determine 15 N enrichment in ammonia. With each series of samples a calibration curve of ammonia with a known isotopic enrichment that ranged between 1 and 50 atom % excess was prepared. In almost every preparation, we achieved an excellent agreement between the observed and the expected 15 N enrichment in ammonia with r values better than 0.9 (data not shown).
Data Presentation and Analysis-The formation of 15 N-labeled metabolites was determined by the product of their isotopic enrichment (mol % excess/100) times concentration (nmol/g wet wt) and is expressed as nanomoles of 15 N metabolite per g wet weight (11)(12)(13)(14).
The distribution of [ 15 N]urea mass isotopomers was calculated using the mathematical model we have described previously (14). Briefly, when 15 N-labeled precursor is provided, the urea formed may have a mass of 60 (U m ), 61 (U mϩ1 ), or 62 (U mϩ2 ) depending on whether 0, 1, or 2 15 N atoms of urea are labeled. Let the fractional abundance of 15 N in the mitochondrial ammonia pool be x, then the fractional of 14 N in the same pool is 1 Ϫ x. Similarly, let the fractional abundance of 15 N in the cytoplasmic aspartate pool be y, then the fractional of 14 N in the same pool is 1 Ϫ y. Then the fraction of the urea isotopomer containing no atom of 15 , and the fraction of urea containing 2 atoms of 15 N will be U mϩ2 ϭ xy. Therefore, U m , U mϩ1 , and U mϩ2 sum to unity. This relationship permits us to calculate the fraction of U m , U mϩ1 , and U mϩ2 at any given abundance of 15 N in the mitochondrial ammonia and cytoplasmic aspartate pools, i.e. at any values of x and y, as we have described (14).
Statistical analyses were carried out by the use of Student's t test or analysis of variance test, as appropriate. A p value less than 0.05 was taken as indicating a statistically significant difference.
Materials and Animals-Chemicals were of analytical grade and obtained from Sigma or from Aldrich. Enzymes and cofactors for the analysis of adenine nucleotide and ammonia were obtained from Roche Molecular Biochemicals, and 15

RESULTS
Characterization of the Perfused Livers-Viability of the perfused liver model is verified by the concentration of adenine nucleotides at the end of the 70 min of perfusion (Table I). There were no significant differences between the adenine nucleotide concentrations during the different experimental conditions. These values are similar to those we previously reported in perfused livers (13,14) and to in vivo levels (16). Fig.  1, panels A and B, shows the changes in urea, ammonia, alanine, glutamine, and glutamate in the effluent under the various experimental conditions. Oxygen consumption is also shown. The constancy of oxygen consumption is an indication of the stability of the preparations.
It is apparent that, at all times, the mean uptake of alanine was greater in the presence of ammonia than in its absence. These differences reached statistical significance at 25, 30, 45, 65, and 70 min. Glutamine output was significantly greater in the presence of ammonia than in its absence at all time points except at 25 min. Of course, urea synthesis was always appreciably greater in the presence of ammonia than in its absence. The effluent alanine, glutamine, glutamate, ammonia, and urea represent the major nitrogenous metabolites in these ex- periments (the release of other amino acids was minor). Therefore, we calculated the extent to which these five compounds could account for nitrogen balance across the liver. In the experiments with alanine and ammonia, the combined nitrogenous uptake of these metabolites was about 850 -900 nmol of nitrogen/min/g liver. The output of nitrogen in urea, glutamine, and glutamate was about 1800 -1900 nmol of nitrogen/min/g liver. Thus, approximately, 1000 nmol of nitrogen are produced from endogenous intra-hepatic sources, presumably from proteolysis. In perfusions with alanine alone the uptake of nitrogen (in the form of alanine) averages about 200 nmol/min/g liver, whereas the production of nitrogen (in urea, glutamine and glutamate) averages about 550 -650 nmol of nitrogen/ min/g liver. Again, there is a requirement for a production of nitrogen from endogenous sources of about 350 -450 nmol of nitrogen/min/g liver. This is much less than the requirement of about 1000 nmol of nitrogen/min/g liver in the presence of ammonia. Fate of the 15 N-Labeled Substrates-Livers were perfused with [ 15 N]alanine, with and without 0.3 mM unlabeled NH 4 Cl. Fig. 2 shows that the uptake of [ 15 N]alanine was greater in the presence of NH 4 Cl than in its absence. The uptake reached an early plateau at 40 -45 min and then increased again, after the enrichment of the [ 15 N]alanine was increased from 50 to 100% at 45 min, to reach a new plateau at 60 -70 min. Fig. 3 shows the fraction (% distribution) of the individual urea isotopomers and the total (nanomoles of nitrogen/min/g) output of each isotopomer in the effluent. It is clear that the presence of 15 NH 4 Cl resulted in an immediate and massive production of [ 15 N]urea which consisted of ϳ60 MPE of U mϩ1 and about 20 MPE of U mϩ2 (panel C). There was much less production of [ 15 N]urea from the labeled alanine in the presence of unlabeled ammonia, but again there was ϳ4-fold increased production of [ 15 N]urea in the presence of ammonia (Fig. 3, panel B) than without (Fig. 3, panel A). Most of the labeled urea, i.e. 15-20% of total urea production, was in the form of U mϩ1 versus 2-3% in the form of U mϩ2 (panel B). U m is the remaining portion of urea, i.e. ϳ80% unlabeled.
The 15 N enrichments in citrulline and aspartate are crucial measurements, as we have already shown that these are good indicators of the 15 N enrichment in the two nitrogenous precursor pools for urea synthesis, mitochondrial carbamyl phosphate and cytosolic aspartate, respectively (14). The perfusions with 15 NH 4 Cl (100 MPE) resulted in very substantial labeling of citrulline (about 70 MPE) but much less labeling of aspartate (about 20 MPE) (Fig. 6, panel C). The metabolic response to [ 15 N]alanine perfusions was very different in that aspartate was much more heavily labeled than was citrulline. The degree of labeling of both citrulline and of aspartate was more pronounced in the absence of ammonia (Fig. 6, panel A) than in its presence (Fig. 6, panel B).
Enrichment data for nitrogen-containing metabolites in livers freeze-clamped at the end of the perfusions (70 min) are shown in Fig. 7. It is evident that, when [ 15 N]alanine was the labeled substrate, 15 N was incorporated into both nitrogenous precursors of urea, but the enrichment in aspartate was 2-3fold that in citrulline. The same relationship held with [ 15 N] alanine in the presence of NH 4 Cl except that the incorporation into citrulline was very low (Fig. 7, panel B). With 15 NH 4 Cl as precursor, citrulline became very heavily labeled, reaching an enrichment that was twice as much as aspartate (Fig. 7, panel  C). Measurement of 15 N-labeled metabolites in the liver at the end of 70 min of perfusion provides two important findings. First, the intra-hepatic 15 N enrichment in glutamine, glutamate, aspartate, and citrulline is in excellent agreement with the 15 N enrichment in the same metabolites in the effluent at the end of perfusion (Figs. 4 -6), suggesting labeling in effluent faithfully reflects enrichment of the intra-hepatic compartment. This conclusion agrees with our previous investigation with 15 N-labeled glutamine or ammonia (13,14). The second observation is that [ 15 N]alanine enrichment in the liver extract was about 70 MPE at the end of perfusion even though the [ 15 N]alanine enrichment of the perfusate was ϳ100 MPE (Fig.  7, panel A). Since [ 15 N]alanine was the sole precursor provided to the liver, it follows that approximately one-third of hepatic alanine was derived from unlabeled sources. This calculation agrees with the estimated fraction of nitrogen, presumably derived from proteolysis, that was necessary to compensate the balance between nitrogen uptake and nitrogen output, as indicated above (Fig. 1, panel A). Similarly, the [ 15 N]alanine enrichment in perfusion studies with unlabeled ammonia indi-cates that ϳ65% of alanine was derived from unlabeled sources. This estimate is in agreement with the fraction of nitrogen derived from proteolysis in perfusion with [ 15 N]alanine plus unlabeled ammonia, i.e. ϳ60%, (Fig. 1, panel B). DISCUSSION In this study we employed [ 15 N]alanine as metabolic tracer in order to follow the metabolism of its nitrogen and, in particular, its contribution to urea and glutamine synthesis in the presence or absence of unlabeled NH 4 Cl. We also examined the metabolic fate of nitrogen derived from 15 NH 4 Cl in the presence of unlabeled alanine. Alanine is well recognized as the  (99 MPE). ‚, E, or q, indicate urea isotopomer containing no (U m ), one (U mϩ1 ), or two (U mϩ2 ) 15 N, respectively. The output of the 15 N-labeled isotopomers is the product of 15 N enrichment (mol % excess/ 100) times one-half of total urea nitrogen (nmol nitrogen/min/g), for U mϩ1 , and times total urea nitrogen for U mϩ2 . The output of U m is the product of percent distribution of U m times concentration of total urea nitrogen. Each data point represents the mean Ϯ S.D for four livers. principal glucogenic amino acid (1), but its key role in nitrogen metabolism is less appreciated. Lopez et al. (4) have recently quantified the importance of alanine to hepatic nitrogen metabolism. It is, by far, the principal amino acid removed by the liver in the fed or postabsorptive state. Indeed, it contributes twice as much nitrogen to the liver as does pre-formed ammonia. Much of this alanine-N is formed in the intestine via bacterial urea hydrolysis and enterocyte glutamine utilization (9, 10). Lopez et al. (4) emphasize the role of hepatic glutamine synthesis as an efficient nitrogen sparing mechanism. It is clear, therefore, that it is important to understand the hepatic disposition of alanine-N and its relationship to urea and glutamine synthesis.
Furthermore, alanine is well recognized as the principal amino acid released by skeletal muscle and is taken up by the liver during ingestion of a low protein diet or starvation (1). As such, alanine is a key amino acid precursor for hepatic gluconeogenesis (1,3). Under these conditions, the liver does not receive via the portal blood equimolar aspartate-N for hepatic ureagenesis (3). Therefore, we used liver perfusion with physiological levels of alanine or alanine plus ammonia to explore the hypothesis that, when the nitrogen supply to the portal venous system differs from the 1:1 (NH 4 ϩ :aspartate-N) stoichiometry, there is increased hepatic proteolysis to provide cytoplasmic aspartate so that formation of cytosolic argininosuccinate can keep pace with the rapid incorporation of ammonia into mitochondrial carbamyl phosphate.
We employed the single-pass isolated perfused rat liver because this model preserves the normal lobular microcirculation of the liver and avoids problems of interpretation that may arise from recycling of substrates (such as products of perivenous hepatocytes being recycled to periportal hepatocytes) that occur in isolated hepatocytes or in a recirculating perfusion. The use of [ 15 N]alanine allowed us to use an approach we had already introduced to define the degree to which a nitrogenous substrate provides nitrogen to urea via either aspartate or carbamyl phosphate (13,14). We can similarly define the origins of the two nitrogen atoms of glutamine (14). These data are schematically summarized in Fig. 8, which represents the principal results of these experiments. (i) [ 15 N]Alanine can provide both nitrogens of urea, but it is a much better precursor to urea nitrogen via aspartate than via citrulline. (ii) [ 15 N]Alanine can provide both nitrogens of glutamine, but it is a much better substrate for the provision of the amino than the amide nitrogen. (iii) Addition of NH 4 Cl to perfusions increases the uptake of alanine, both in terms of mass and of 15 N, and increases the output of 15 N products, such as urea and glutamine. (iv) The addition of NH 4 Cl increases the net negative nitrogen balance over that seen with alanine alone. (v) The intra-hepatic 15 N enrichment in glutamine, glutamate, aspartate, and citrulline is in excellent agreement with the 15 N enrichment in the same metabolites in the effluent regardless of 15 N precursor.
These observations are consistent with an inter-related metabolic pattern within the liver. We suggest that alanine is quite limited in its ability to provide ammonia to the mitochondrion for carbamyl phosphate synthesis, and ammonia is somewhat limited in its ability to provide nitrogen to cytoplasmic aspartate for incorporation into argininosuccinate. These proposals are supported by the very much lower 15 N enrichment in citrulline than in aspartate with [ 15 N]alanine as nitrogen donor. The converse is true with 15 NH 4 Cl as labeled precursor (Fig. 6). The low rate at which alanine gives rise to ammonia limits alanine removal. Therefore, when unlabeled NH 4 Cl is included in the [ 15 N]alanine perfusions, we see an increased uptake of alanine (both alanine mass and 15 N) and increased production of 15 N products, principally urea.
The highest rates of urea synthesis are found with NH 4 Cl. However, it is clear that the ability of ammonia to provide nitrogen to carbamyl phosphate synthetase is much greater than its ability to provide nitrogen to cytoplasmic aspartate.
Alanine can provide additional nitrogen to aspartate (by the combined action of alanine aminotransferase and aspartate aminotransferase), but this may be limited by the activity of alanine aminotransferase or, in these experiments, by the equilibrium poise of the enzyme, given that the perfusions are provided with physiological concentrations of pyruvate. Finally, there may be a failure of mitochondrial aspartate to equilibrate with cytoplasmic aspartate, as proposed by Yang et  15 NH 4 Cl (III). The rates of the primary nitrogen output (urea and glutamine) are indicated in circles representing these metabolites along with % distribution of their mass isotopomers. The deficit between nitrogen uptake and nitrogen output is furnished by intra-hepatic proteolysis. For simplicity, this drawing does not differentiate between perivenous hepatocytes (glutamine synthesis) and the periportal hepatocytes (alanine uptake, glutamine metabolism, and urea synthesis). In parentheses are shown the percent enrichment (mol % excess) of 15 N-labeled isotopomers (taken from data in Figs. 3-7, at the end of 70 min of perfusion) of the indicated metabolite. Values for ammonia or alanine uptake and total glutamine or urea output (nmol nitrogen/min/g) are taken from data in Fig. 1, at the end of 70 min of perfusion. Bold arrows indicate primary input/direction of nitrogen metabolite. CP, carbamyl phosphate; GDH, glutamate dehydrogenase.