Parallel control of hepatic proteolysis by phenylalanine and phenylpyruvate through independent inhibitory sites at the plasma membrane.

Intracellular protein degradation in the rat hepatocyte is regulated by 7 amino acids of which Leu, Gln, and Tyr play major roles. Although Phe is known to inhibit proteolysis as effectively as Tyr at high concentrations, it has not been regarded as an active regulator because of its rapid hydroxylation to Tyr. We now show that proteolytic responses to Phe during liver perfusion differ strikingly from effects of the multiphasic regulators Leu, Gln, and Tyr in eliciting mirror image responses at half-normal and normal plasma concentrations. Since response curves to phenylpyruvate and Phe were identical, we considered the possibility that phenylpyruvate mediated its anomalous inhibition intracellularly. However, when phenylpyruvate was produced from phenyllactate intracellularly at a rate providing the same rate of transamination (and intracellular concentration) as that derived from the uptake of phenylpyruvate, no response was obtained. Hence, the effect of phenylpyruvate was not initiated within the cell but more likely from the plasma membrane. Comparable evidence for Phe is less direct. Recent findings indicate that recognition sites for Leu and Gln are located at the plasma membrane. Since Phe augments the concerted inhibition by Leu and Gln at 4-fold normal levels, Phe is probably recognized in close proximity to them. However, the failure of phenylpyruvate to substitute for Phe in this interaction suggests that proteolytic inhibition by phenylpyruvate and Phe is mediated through similar, although independent, plasma membrane sites.

Intracellular protein degradation in the rat hepatocyte is regulated by 7 amino acids of which Leu, Gln, and Tyr play major roles. Although Phe is known to inhibit proteolysis as effectively as Tyr at high concentrations, it has not been regarded as an active regulator because of its rapid hydroxylation to Tyr. We now show that proteolytic responses to Phe during liver perfusion differ strikingly from effects of the multiphasic regulators Leu, Gln, and Tyr in eliciting mirror image responses at half-normal and normal plasma concentrations. Since response curves to phenylpyruvate and Phe were identical, we considered the possibility that phenylpyruvate mediated its anomalous inhibition intracellularly. However, when phenylpyruvate was produced from phenyllactate intracellularly at a rate providing the same rate of transamination (and intracellular concentration) as that derived from the uptake of phenylpyruvate, no response was obtained. Hence, the effect of phenylpyruvate was not initiated within the cell but more likely from the plasma membrane. Comparable evidence for Phe is less direct. Recent findings indicate that recognition sites for Leu and Gln are located at the plasma membrane. Since Phe augments the concerted inhibition by Leu and Gln at 4-fold normal levels, Phe is probably recognized in close proximity to them. However, the failure of phenylpyruvate to substitute for Phe in this interaction suggests that proteolytic inhibition by phenylpyruvate and Phe is mediated through similar, although independent, plasma membrane sites.
The intracellular degradation of long-lived proteins in the perfused rat liver is primarily controlled by the multiphasic action of 7 regulatory amino acids: leucine, glutamine, tyrosine, proline, methionine, histidine, and tryptophan (Ref. cently, typical effects of leucine were reported in the isolated rat hepatocyte (6), demonstrating that the multiphasic response is a general feature of amino acid regulation in a single population of cells. Except for differences in degree of inhibition, these features include a primary inhibition a t halfnormal plasma amino acid concentrations, a zonal loss of inhibition within a narrow range a t normal concentrations, and a secondary inhibition a t high physiologic levels (2, 4). The zonal loss can be explained by the lack of plasma alanine, which is specifically required for expression of inhibition by the regulatory amino acids at normal concentrations (7, 8).
A basic description of the multiphasic response is reasonably complete except for one point. Since tyrosine exhibits the same multiphasic features as leucine (4) and is closely related to phenylalanine metabolically, tyrosine was initially considered the active regulator and representative of the aromatic group. Although phenylalanine is as inhibitory as tyrosine a t 4~' normal plasma concentrations (Z), nothing is known of its activity a t normal levels. Because of the effectiveness of phenylalanine hydroxylase in converting small increases in plasma phenylalanine to tyrosine during a single transhepatic passage (9), it seemed unlikely that the two amino acids would regulate proteolysis in tandem through the same process. Hence, we have considered the possibility that liver might contain more than one regulatory mechanism. To this end, the present study was undertaken to compare the respective roles of phenylalanine and tyrosine and their keto acids in the control of hepatic proteolysis.

EXPERIMENTAL PROCEDURES
Animals-Male rats of the Lewis strain (Harlan Sprague-Dawley, Indianapolis, IN) were used as liver donors for perfusion experiments and hepatocyte preparations. The animals, weighing 120-140 g at the time of perfusion, were routinely maintained on commercial laboratory chow and water ad libitum in a light-controlled room (light off, 1900 to 0700 h). For the determination of protein degradation, liver protein was labeled in vivo with ~-[U-'~C]valine (Du Pont-New England Nuclear, 250 mCi/mmol); 4 pCi in 0.25 ml of 0.85% NaCl were injected intraperitoneally, 24 and 17 h before perfusion.
(nonrecirculating) mode as described earlier (10). Unless stated Liver Perfusion-Livers were perfused in situ in the single-pass otherwise, all experiments were carried out at approximately 1100 h. When proteolysis was determined, the single-pass phase lasted 40 min. Perfusate flow was then switched to a second stage cyclic perfusion, with 18 p~ cycloheximide added, for the measurement of labeled valine release. Finally, the specific radioactivity of released valine was determined in a third stage cyclic perfusion in which exogenous, unlabeled valine was eliminated and the specific activity of released valine allowed to reach equilibrium (11).
The perfusion medium consisted of Krebs-Ringer bicarbonate buffer, 3% bovine plasma albumin (fraction V, Pentex, ICN Biomedicals, Inc., Costa Mesa, CA), 10 mM glucose, and freshly washed bovine erythrocytes (27%, v/v); its preparation has been described in detail elsewhere (10,11). All additions to the medium were prepared in 0.85% NaCl with pH adjusted to 7.4 by dilute NaOH.
Measurement of Protein Degradation-As detailed in a previous report ( l l ) , rates of long-lived protein degradation in livers, previously labeled in uiuo with ~-[U-'~C]valine, were determined from the cumulative release of labeled valine in the second stage perfusion flask described above. In the presence of cycloheximide, rates of label release between 5 and 15 min have been shown to reflect the breakdown of long-lived protein that occurs in the liver during the initial single-pass perfusion (10,11); uncertainties arising from the degradation of short-lived and endocytosed protein are virtually eliminated by the use previously labeled livers as described (11).
The total accumulation of free ['4C]valine in liver and perfusate was computed a t each point of sampling between 5 and 15 min (11); proteolytic rates then were calculated by least squares regression after correction for the specific radioactivity of released valine (see above). Values were expressed as nanomoles of valine min"/liver (100 g body weight). The average valine content of liver protein for a 100-g rat has been shown to be 465 pmol (10).
isolated Hepatocytes-Suspensions of hepatocytes were prepared according to Seglen (12) and purified by Percoll centrifugation (13); 90% or more of the cells excluded trypan blue. The conditions of incubation are given in the legends of Figs. 2 and 3.
Radiolabeled Phenylpyruvate and Phenyllactate- Analytical Procedures-For the determination of phenylpyruvate, plasma was deproteinized with an equal volume of 15% sulfosalicylic acid (in liver, 1 volume of tissue to 9 volumes of acid) and the keto acid derivatized with o-phenylenediamine (16). Analysis was carried out by reverse phase high performance liquid chromatography according to Livesey and Edwards (16). In measuring rates of phenylpyruvate transamination, label in phenylalanine and tyrosine was determined together. Samples of perfusate plasma or medium from hepatocyte incubations were deproteinized in 0.3 N perchloric acid. The supernatants then were passed through small columns of Dowex 50 and the amino acids eluted with 5 N NH4OH. Where phenylalanine and tyrosine were measured in plasma and liver, the tissues were deproteinized with sulfosalicylic acid as described above and the amino acids determined by the dansylation method of Tapuhi et al. (17). Radioactivity incorporated into liver protein was determined as described by Khairallah and Mortimore (18). Samples with 14C were counted in Liquiscint (National Diagnostics, Inc.) with a Beckman LS 7800 liquid scintillation spectrometer; results were expressed as disintegrations/min.

RESULTS
Proteolytic Dose Responses to Tyrosine, Phenylalanine, and Their Derivatives-As depicted in Fig. 1A tyrosine exhibited dose-dependent effects typical of the multiphasic responses reported previously for leucine, glutamine, and the regulatory group (4). In addition, responses to the keto acid p-hydroxyphenylpyruvate were closely analogous to those of a-ketoisocaproate in evoking inhibition equal to the corresponding amino acid at levels higher than l x while failing to inhibit at the lowest concentration (2). The lack of inhibition a t 0.5X suggests that tyrosine's primary inhibition is not mediated through its keto acid despite the fact it is actively transaminated (19). Surprisingly, phenylalanine revealed a dose response curve quite unlike that of tyrosine (Fig. 1B); no effect was seen a t 0.5X, but maximal inhibition was achieved at Ix. Effects of phenylpyruvate were identical to those of phenylalanine.
Hepatic Uptake and Metabolic Fate of Phenylpyruvate-It is clear from the dissimilar proteolytic responses a t 0.5 and l x (Fig. 1, A and B

TABLE I
Uptake and intracellular fate of phenylpyruvate in the perfused rat liver Livers from normal fed rats were perfused in the single-pass mode for 5 min with 0.3 mM [l-'4C]phenylpyruvate in the presence of 10 mM glucose; based on preliminary experiments, net uptake attains equilibrium by 5 min and was computed from the transhepatic difference in plasma phenylpyruvate multiplied by the plasma flow. At the end of perfusion, samples of affluent and effluent perfusate plasma and of the liver for later analysis were taken, immediately frozen in liquid NS, and stored a t -20 "C as described under "Experimental Procedures." CO? was collected from the perfusate during the perfusion and its radioactivity determined according to Mortimore et al. (4). Fluxes were calculated from the accumulated counts of a given constituent, divided by the specific radioactivity of plasma phenylpyruvate. Results are means f S.E. of three experiments.  (9,19), the responses to phenylalanine cannot be explained by its conversion to tyrosine, and the effects of phenylalanine and phenylpyruvate must be conveyed directly through themselves or through metabolites of the keto acid. We initially evaluated the latter by measuring the hepatic uptake and metabolism of [l-'4C]phenylpyruvate under steady state conditions. As shown in Table I, ~5 5 % of the phenylpyruvate taken up was transaminated to phenyl-alanine (and to tyrosine after hydroxylation), whereas 43% was reduced to phenyllactate; approximately 4% was decarboxylated. To exclude the remote possibility that the keto acid influx was saturated at the concentration tested (300 p~) , we determined rates of transamination in isolated hepatocytes and found them to be directly proportional to added phenylpyruvate from 0 to 500 p~ (Fig. 2). These data are consistent with a high K , for phenylpyruvate influx. Similar results have been reported for a-ketoisocaproate transport in isolated hepatocytes (20,21).
It is clear from results in Table I1 that the inhibitory activity of phenylpyruvate cannot be explained by its transamination t o phenylalanine or tyrosine as both intracellular pools remained unchanged over the effective range of keto acid concentration. We cannot, however, rule out the possibility that phenylpyruvate has other intracellular actions. Although the ineffectiveness of phenyllactate appears to exclude it as a mediator (Fig. lB), this could be explained by low rates of conversion of the lactate to the keto acid if the latter is active in the cell. On the other hand, if phenylpyruvate is readily produced from phenyllactate, the reaction could provide a useful way to assess the intracellular effectiveness of phenylpyruvate.
IntracellularlExtracellular Distribution of Phenylpyruuate-To evaluate the feasibility of this approach, attempts were made to determine the intracellular concentration of   Fig. 3B. After 10-20 s (a period required to clear the perfusion tubing of preexisting medium) label in perfusate plasma abruptly declined, falling off exponentially with halflives generally less than 10 s. During the chase, 95% or more of the initial label was eluted with uniform kinetics; similar results were obtained for p-hydroxyphenylpyruvate (see Table  111). Half-lives for the release of phenylalanine and tyrosine produced from phenylpyruvate during the pulse were significantly greater, ranging from 23 to 50 s (not shown).
In initial experiments we found that the elution of sucrose conformed to a single-compartment model in which the rate of plasma flow and the half-life of elution are inversely related (Fig. 3A). Under steady state conditions of perfusion, the volume of the compartment (eluted space) could be computed as follows, Eluted space (ml/g) = plasma flow (ml/s) x half-life ( s ) 0.693 X liver weight (9) 0%. 1) where 0.693 = In 2. Control experiments revealed that the keto acids do not equilibrate with erythrocyte water for periods up to 60 s. Thus it was not necessary to consider erythrocyte volume in the equation.
The findings in Table I11 show that the phenylpyruvate and p-hydroxyphenylpyruvate spaces in liver are approxi-   For most experiments (A-D) a single liver was employed for three to four separate determinations; the order was varied to avoid errors that might have arisen from sequence effects or duration of perfusion. No difference was found between the two keto acid spaces, but the pooled ratio of keto acid to sucrose space of 0.802 was significantly different from unity ( p < 0.001). It is of interest that the same inverse relationship between half-life and plasma flow was observed with phenylpyruvate (where flow was varied) as with sucrose in Fig. 3A.
The reason for the decrease in the apparent extracellular space for the keto acids is not known. Cellular uptake and metabolism of the labeled keto acids during the wash-out can be excluded by the lack of correlation between individual space measurements and flow rate. In addition, the 10 mM unlabeled chase would have diluted their specific activities to negligible values, thus minimizing the loss of label. It is conceivable, though, that the negative charge of the keto acids might have restricted their diffusion into the water layer at the cell surface since membrane glycoproteins bear a similar charge. On the assumption that the proportionality of the decrease would remain when perfusion was stopped, we computed an apparent space for liver samples of 0.128 ml/g. This was based on the pooled ratio of 0.802 multiplied by 0.159, the extracellular space of the normal unperfused liver minus erythrocyte volume (22).
In two control experiments in which livers were perfused with phenylpyruvate at an average transhepatic concentration of 198 FM, the total recovery of the keto acid was 29.1 k 1.8 nmol/g. Taking 0.128 ml/g as the extracellular space for phenylpyruvate, we computed the mean intracellular pool to be 3.8 f 1.8 nmol/g or about 8 nmol/ml of intracellular water.
Production of Intracellular Phenylpyruvate from Phenyllactate-The possibility was considered that reactions promoted by aromatic a-keto acid reductase and to a lesser extent lactate dehydrogenase, which together catalyze the reduction of the keto acid (23), might be reversed by external phenyllactate. Based on in vitro experiments, a ratio of reduced to oxidized substrate of 16:l or greater appears to be needed for the reversal (23). The preceding results have shown that intracellular pools of phenylpyruvate are small relative to plasma levels. Thus it seemed possible that only moderate additions of phenyllactate would be required.
As in the experiments of Fig. 2, the isolated hepatocyte rather than the perfused liver was employed to evaluate phenylpyruvate production, since the process is enzymatic and does not involve proteolytic regulation. a-Keto acid formation was monitored from the accumulation of labeled amino acids in the presence of cycloheximide, a method that provides a close approximation of transamination (compare Table I and Fig. 2). The results in Fig. 4 show that phenylpyruvate is produced at appreciable rates with comparatively small additions of phenyllactate. The overall flux at 300 p~, computed from the double-reciprocal plot (inset), was 28 nmol min"/108 cells, a value similar to the rate of transamination in Fig. 2 with 300 p~ phenylpyruvate (32 nmol min"/lOR cells). Since loR cells approximate 1 g of perfused liver', both values may be compared with the transamination rate of 30 nmol min"/g in livers perfused with 300 p~ phenylpyruvate (Table I).
These findings indicate that concentrations of phenylpyruvate at sites of transamination, which are probably cytosolic in distribution, are virtually the same with 300 p~ phenyllactate as with 300 FM phenylpyruvate. The failure to see lower rates with phenyllactate can be explained by reversal of the

Additivity of the inhibitory effects of phenylalanine and phenylpyruvate (PP) on proteolysis in the perfused rat liver
The experiments were carried out under the conditions of Fig. 1 with redox flux and the loss of phenylpyruvate utilization by its reduction to phenyllactate when phenylpyruvate is added ( Table I). Since phenylpyruvate production from the inward transport of phenyllactate can be assumed to occur throughout the cytoplasm, the intracellular distribution of phenylpyruvate from this source should generally coincide with that derived from the inward transport of phenylpyruvate. Although phenylpyruvate levels in the internal vicinity of transporters might be higher at 300 p~ than those generated from phenyllactate at 300 pM, this would not be the case at 75 pM phenylpyruvate where its uptake was 50% lower (Table I and Fig. 2) than the enzymatic production of phenylpyruvate at 300 p~ phenyllactate (Fig. 4), and inhibition was maximal (Fig. 1B). Thus the total lack of effectiveness of phenyllactate up to 300 p~ (Fig. 1B) appears to exclude as possible mediators all sites of phenylpyruvate recognition within the cell as well as metabolites arising from its decarboxylation. Moreover, it clearly points to the plasma membrane as the initial site of proteolytic regulation. Concerted Interactions between Regulatory Amino and Keto Acids-As for the location of the site of phenylalanine regulation, the close agreement between the amino and keto acid dose responses in Fig. 1B suggests that they may be recognized at a common site or, alt-rnatively, at independent plasma membrane sites operating in parallel. The possibility that phenylalanine acts through an intracellular locus seems improbable, since we have shown that the intracellular pool does not increase when the external level is raised from zero to 1X.' The fact that 38 pM phenylalanine and 38 pM phenylpyruvate are additive, evoking significantly greater inhibition combined than when given separately (Table IV), suggests that the putative site, or sites, can react with more than one regulator. The additivity is not complete, though, as the combined effect amounts to only 60% of the inhibition obtained with either agent alone a t 75 p~. While the finding is of interest, it does not favor one possibility over the other.
The question was partly resolved by testing the effect of combining the aromatic keto and amino acids with the two major regulators, leucine and glutamine. Past studies have revealed that the inhibitory effect of tyrosine at 4X is not additive to effects of leucine and glutamine individually (2, 11) but does add to the combined effect when it is less than maximal (11). Maximal additivity between leucine and glutamine is found in synchronously fed rats (8) but not in animals fed commercial pellets ad libitum (2, 11). The fundamental nature of these concerted effects is not known. However, on the basis of evidence in the accompanying paper ' G . E. Mortimore, unpublished data. Livers were perfused in the single-pass mode for the determination of proteolytic rates. 24-h starved rats rather than fed rats were used because their responses at 4X levels are generally less variable. 4X additions of a standard, complete plasma amino acid (AA) mixture (4) or single amino acids from the mixture were added as listed below. In the case of the a-keto acids, the amounts added were the molar equivalents of their parent amino acids. The results are means f S.E. of two to six perfusions. Because of the small concerted effects of Tyr and Phe, liver-to-liver variability was minimized by pairing; four to five paired experiments, each on a different day, were carried out in each of the last four groups. (24), we have proposed that they derive from interactions between sites of regulation for leucine and glutamine at the plasma membrane.
Results in Table V confirm the foregoing additive effects of leucine and glutamine and, in addition, show that tyrosine and phenylalanine each increases the inhibition to the maximal level obtained with the complete amino acid mixture. The increases are typically small since the expected combined effect of leucine and glutamine is greater than 80% of the maximum (2). Nevertheless they are real and hence relevant to the question of where phenylalanine acts. The fact that phenylpyruvate and p-hydroxyphenylpyruvate are both inactive when substituted for their respective amino acids indicates that the amino and keto acids are recognized at independent sites. The findings thus exclude the common site hypothesis in favor of independent parallel sites (see first paragraph of this section).

DISCUSSION
The striking dissimilarity in proteolytic dose responses that was observed between two closely related amino acids, phenylalanine and tyrosine, raises new questions concerning the control of long-lived protein degradation in the hepatic parenchymal cell. Although the characteristic multiphasic response to tyrosine has been described previously (4), the finding that phenylalanine, the precursor of tyrosine, evokes a much simpler single-phase inhibition at concentrations below the normal plasma level was unexpected, and its implications are not fully understood. Nevertheless, it is clear from the shape of the curves in Fig. 1, A and B, that the 2 amino acids act through different mechanisms.
The single-phase responses to phenylalanine and phenylpyruvate are curiously reminiscent of earlier effects of leucine and a-ketoisocaproate in skeletal and cardiac muscle (25- 27) and of leucine in cultured kidney cells (28). It is not unreasonable therefore to suggest that the phenylalanine response in hepatocytes is the manifestation of a basic cellular control mechanism that coexists with the multiphasic action of tyrosine. But regardless of its nature, it is likely to remain unex-

Control of Hepatic Proteolysis by
Phenylalanine and Phenylpyruvate 22065 pressed in the presence of plasma amino acids as suggested by the following results. When alanine is deleted from a complete amino acid mixture a t normal (Ix) concentrations, proteolytic rates accelerate to near-maximal values as the multiphasic regulatory amino acids lose their inhibitory effectiveness through the loss of alanine's coregulatory activity (7, 8). If phenylalanine were as effective under these conditions as it is by itself, the above accelerating effect would be significantly diminished. The fact that a decrease is not seen implies that phenylalanine's inhibition is overridden by a dominant stimulatory response to the loss of alanine.
In view of the present findings with phenylpyruvate, it is appropriate to review effects of leucine and a-ketoisocaproate on protein degradation in muscle and to compare the responses in the two tissues. Apart from the aforementioned similarity in dose responses between leucine and phenylalanine and their respective keto acids, studies with rat skeletal muscle have indicated that leucine's action is achieved through products of leucine catabolism, a conclusion based on the fact that its effect is abolished by the transaminase inhibitor cycloserine (25,26). In the perfused rat heart, however, Chua and co-workers (27) found that with 10 mM pyruvate it is possible to suppress the decarboxylation of aketoisocaproate by 95% without affecting its inhibitory effectiveness. Taken together, the results support the possibility that a-ketoisocaproate rather than leucine is the mediator; a n important question, though, is the cellular site of recognition. This was not addressed in the above studies, but the evidence is not inconsistent with a plasma membrane location. It is known, for example, that a-ketoisocaproate accumulates in the medium when skeletal or cardiac muscle is perfused with leucine (29). Unless the keto acid was removed by medium replacement, inhibition could have been achieved from the outside.
By contrast, responses of the hepatocyte to leucine and its keto acid as well as to tyrosine and p-hydroxyphenylpyruvate (Fig. 1A) are mediated through a multiphasic mechanism. Because leucine is poorly transaminated in liver ( 2 ) , its inhibitory effects, particularly at low concentrations (0.25-0.5X), can occur only through leucine itself (2). And since the keto acid does not inhibit at the lowest concentration, it must act through an independent mechanism (2); a similar conclusion may be drawn for the action ofp-hydroxyphenylpyruvate (Fig.  1A). Recent studies have shown that the leucine analogue isovaleryl-L-carnitine can mimic leucine's multiphasic response curve (5) by reacting at the recognition site, or sites, for leucine (24). Moreover, they appear to be located at the plasma membrane in close proximity to similar sites for glutamine, tyrosine, and other regulatory amino acids (24). Such an association could provide a molecular basis for the complex concerted actions of these amino acids which have been documented previously (2,8,11).
Although the exact correspondence between dose responses t o phenylalanine and phenylpyruvate suggest, at first glance, that they share a single regulatory mechanism, such a view is not compatible with the differences we observed in the way they interacted with leucine and glutamine (Table V). Since the character of the response is a function of the regulatory site, not the regulator, the responses must have been generated through similar but independent mechanisms. The partial additivity can be explained by a limited ability of phenylalanine to react with the phenylpyruvate sites. However, it is not likely that the cross-reactivity is reversed, i.e. that the keto acid reacts with the amino acid site. If it were, one would not have observed the aforementioned differences in concerted effects.
The final question concerns the location of the regulatory site for phenylalanine. Although evidence in support of the plasma membrane is more indirect than it is for phenylpyruvate, nevertheless it is the most probable choice. Because recognition sites for leucine and glutamine are very likely at the plasma membrane (24), a similar location would provide a simpler model for explaining concerted interactions between the 3 amino acids than one in which the sites are widely separated. The same argument can be used for locating the site for tyrosine. In this instance, though, there is added relevance in the fact that it evokes multiphasic responses. Since the external concentrations at which the sharp inflections occur are the same in multiples (or fractions) of its normal plasma concentration as those of leucine and glutamine (4,8), an externally facing locus provides a reasonable explanation for the effects.