Leucine, Glutamine, and Tyrosine Reciprocally Modulate the Translation Initiation Factors eIF4F and eIF2B in Perfused Rat Liver*

Leucine, glutamine, and tyrosine, three amino acids playing key modulatory roles in hepatic proteolysis, were evaluated for activation of signaling pathways involved in regulation of liver protein synthesis. Furthermore, because leucine signals to effectors that lie distal to the mammalian target of rapamycin, these downstream factors were selected for study as candidate mediators of amino acid signaling. Using the perfused rat liver as a model system, we observed a 25% stimulation of protein synthesis in response to balanced hyperaminoacidemia, whereas amino acid imbalance due to elevated concentrations of leucine, glutamine, and tyrosine resulted in a protein synthetic depression of roughly 50% compared with normoaminoacidemic controls. The reduction in protein synthesis accompanying amino acid imbalance became manifest at high physiologic concentrations and was dictated by the guanine nucleotide exchange activity of translation initiation factor eIF2B. Paradoxically, this phenomenon occurred concomitantly with assembly of the mRNA cap recognition complex, eIF4F as well as activation of the 70-kDa ribosomal S6 kinase, p70S6k. Dual and reciprocal modulation of eIF4F and eIF2B was leucine-specific because isoleucine, a structural analog, was ineffective in these regards. Thus, we conclude that amino acid imbalance, heralded by leucine, initiates a liver-specific translational failsafe mechanism that deters protein synthesis under unfavorable circumstances despite promotion of the eIF4F complex.

The amino acids represent a class of biologic molecules exerting dynamic and complex influences on highly disparate physiologic processes including pancreatic insulin and glucagon secretion, protein degradation and synthesis, hepatic gluconeogenesis, and sensitization of tissues to the anabolic effects of insulin (1)(2)(3)(4)(5). The effects of amino acids are somewhat enigmatic but often involve the interplay of hormones and other factors intrinsic to the cellular environment. Recently, the branched chain amino acid, leucine, has been demonstrated to modulate pathways of signal transduction and may indeed contribute importantly to the cellular interpretation of integrated signals. With regard to protein homeostasis, several reports now exist supporting the hypothesis that leucine im-pacts protein turnover through mechanisms beyond those of protein synthetic substrates (1)(2)(3)(4).
In an elegant series of experiments, Mortimore et al. (4,5) demonstrated that the amino acids leucine, glutamine, and tyrosine, individually as well as cooperatively and in a manner that is concentration-dependent, attenuate hepatic macroautophagic proteolysis induced by deprivation of amino acids. Furthermore, insulin functions additively and synergistically with these amino acids, thereby enhancing the efficacy of proteolytic inhibition by leucine, glutamine, and/or tyrosine. Inherent in their potency as modulators of protein homeostasis, amino acids generally exert reciprocal control of hepatic protein degradation and synthesis. The latter process is governed primarily at the level of translation initiation through the regulative affinities and activities of several eukaryotic initiation factors (eIFs). 1 Components of the translational apparatus demonstrated to play particularly important roles include the guanine nucleotide exchange factor, eIF2B, the eIF4F heterocomplex, and the 70-kDa 40 S ribosomal protein S6 kinase (p70 S6k ).
The guanine nucleotide exchange factor, eIF2B, is a heteropentameric enzyme that performs a function critical for the successive and cyclic nature of initiation. After a ribosome is loaded onto the mRNA, the initiation complex is disassembled in a process requiring hydrolysis of eIF2-associated GTP. Because the resultant GDP bound to eIF2 dissociates very slowly and because eIF2⅐GTP is necessary for recruitment of Met-tRNA i , the eIF2B-catalyzed exchange of eIF2-bound GDP for GTP is essential for ternary complex formation and subsequent rounds of initiation (6). The activity of eIF2B is negatively affected by phosphorylation of the ␣ subunit of eIF2, which consequently competitively inhibits eIF2-targeted nucleotide exchange. Also, allosteric-like effects as well as direct phosphorylation of eIF2B are implicated in regulating its activity. Amino acid deprivation has been shown to suppress protein synthesis concomitant with inhibition of eIF2B activity. However, the involvement of the phosphorylation state of eIF2␣ remains debatable (7)(8)(9).
Although contentious, the association of eIF4A (an ATP-dependent RNA helicase), eIF4E (an mRNA cap-binding protein), and eIF4G (a scaffolding protein) has been suggested to be the rate-limiting event in the initiation of translation (10). These factors, collectively referred to as eIF4F, bind to the m 7 GTP cap 1 The abbreviations used are: eIF4E, eukaryotic initiation factor 4E; eIF4G, eukaryotic initiation factor 4G; eIF2B, eukaryotic initiation factor 2B; eIF2␣, ␣ subunit of eukaryotic initiation factor 2; 4E-BP1, eIF4E binding protein 1; BCAA, branched chain amino acid; mTOR, mammalian target of rapamycin; PI, phosphatidylinositol; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid; MOPS, 3-(N-morpholino)propanesulfonic acid; MAP, mitogen-activated protein; GSK-3, glycogen synthase kinase 3; ANOVA, analysis of variance. structure of mRNA and facilitate the recruitment of other eIFs as well as the 40 S ribosomal subunit, culminating in the formation of the 48 S preinitiation complex. The assembly of eIF4F is determined by the phosphorylation state of a family of competitive inhibitors of eIF4G, the eIF4E binding proteins (4E-BPs). Hypophosphorylated 4E-BPs exhibit strong affinity for eIF4E and as such, restrict access of eIF4G to eIF4E. Thus, aggregation of integral components of the cap-binding holocomplex is hindered. However, phosphorylation on multiple residues neutralizes the inhibitive properties of 4E-BPs and facilitates eIF4E⅐eIF4G interaction. The signal transduction pathway responsible for 4E-BP phosphorylation is common to p70 S6k , a cell cycle-regulated kinase implicated in expression of mRNAs of the TOP (terminal oligopyrimidine) family (11,12). Deprivation of amino acids results in dephosphosphorylation of 4E-BP1 and p70 S6k in several cell types (7,11); these effects are reversed upon readdition of amino acids, and this reversal is rapamycin-sensitive, underscoring the involvement of the mammalian target of rapamycin (mTOR) in mediation of these signals.
Amino acids influence hepatic protein turnover, at least in part, by reciprocal regulation of protein synthesis and protein degradation. Because a distinct group of amino acids, namely the regulatory group, and in particular, leucine, glutamine, and tyrosine, exert most of the observed inhibition of deprivationinduced proteolysis in the liver, we sought to characterize the modulatory role(s), if any, of these amino acids on hepatic protein synthesis. Furthermore, this study was designed to address the hepatic response to physiological changes in amino acid concentration in an effort to isolate potentially important amino acid signaling events; particularly, those induced by leucine. Thus, the specific effects of leucine, glutamine, and tyrosine on eIF4F assembly, eIF2B activity, p70 S6k activation, and the relative contribution of these events in determination of overall protein synthesis was evaluated.

EXPERIMENTAL PROCEDURES
Animals-Male Sprague-Dawley rats weighing approximately 100 -125 g were maintained on a 12 h light/dark cycle and were provided food (Harlan-Teklad Rodent Chow) and water ad libitum.
Materials-ECL detection reagents and horseradish peroxidase-conjugated sheep anti-mouse and donkey anti-rabbit immunoglobulins were purchased from Amersham Pharmacia Biotech. Polyvinylidene difluoride membranes were acquired from Bio-Rad. Insulin was purchased from Eli Lilly and Co. Rapamycin was purchased from Calbiochem.
Liver Perfusion-Livers were perfused in situ essentially as described previously (13) with the following modifications. Perfusate was delivered at flow rate of 7 ml/min under nonrecirculating conditions. Following an initial 5 min washout, livers were perfused and radiolabeled for 15 min in the presence of 5 mM valine. The amino acid composition differed from that previously reported; the 1ϫ designation is described in the legend of Fig. 1. Addition of this mixture has been shown to yield perfusate concentrations that closely approximate those reported for rat plasma (14). The amino acid compositions of the perfusate utilized throughout this study were multiples of 1ϫ as described in the figures. Moreover, under some circumstances, 10 nM insulin was added to the perfusing medium. For determination of protein synthesis, L- [3, H]valine (NEN Life Science Products) was added at 1 Ci/ml to the perfusate.
Measurement of Protein Synthesis-Rates of protein synthesis were determined essentially as described previously with slight modification (15) by measuring the incorporation of [ 3 H]valine into newly synthesized protein.
Quantitation of eIF4E, 4E-BP1⅐eIF4E, and eIF4G⅐eIF4E Complexes-Quantitation of the respective factors and complexes were performed essentially as defined elsewhere (16). eIF4E, 4E-BP1⅐eIF4E, and eIF4G⅐eIF4E complexes were immunoprecipitated from 10,000 ϫ g supernatants of whole liver homogenate using a mouse anti-eIF4E monoclonal antibody. The antibody was raised against recombinant human eIF4E as described previously (20). The antibody-antigen complex was isolated by incubation with goat anti-mouse Biomag immuno-globulin G beads (PerSeptive Diagnostics). Prior to incubation with antigen-antibody complexes, the beads were washed in 1% nonfat, dry milk in buffer B (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 0.1% ␤-mercaptoethanol, 0.5% Triton X-100, 50 mM NaF, 50 mM ␤-glycerophosphate, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 0.5 mM sodium vanadate). The beads were captured using a magnetic stand, washed twice with buffer B, and washed once with buffer B containing 500 mM NaCl rather than 150 mM. Immune complexes bound to the beads were eluted by resuspension of the beads in SDS sample buffer and then boiling for 5 min. The beads were pelleted by centrifugation, and the supernatants were subjected to SDS-polyacrylamide gel electrophoresis. Separated proteins were then electrophoretically transferred to polyvinylidene difluoride membranes. Following transfer, the membranes were incubated with a mouse monoclonal anti-eIF4E antibody, a rabbit polyclonal anti-4E-BP1 antibody, or a rabbit polyclonal anti-eIF4G antibody overnight at 4°C. The immunoblots were then developed using an ECL Western blotting kit as described previously (20).
Quantitation of Phosphorylated and Unphosphorylated 4E-BP1 in Liver Homogenates-Quantitation of the phosphorylation state of 4E-BP1 was carried out exactly as described previously (16). Briefly, the phosphorylated and unphosphorylated forms of 4E-BP1 were collected by immunoprecipitation of 4E-BP1 from 10,000 ϫ g supernatants of whole liver homogenate. For this purpose a mouse monoclonal anti-4E-BP1 antibody was incubated with the 10,000 ϫ g supernatant. The immunoprecipitated proteins were collected and separated as outlined above. The migration of 4E-BP1 on SDS-polyacrylamide gels is inversely proportional to the degree of phosphorylation of the protein (17,18). Therefore, multiple phosphorylation forms were separable following SDS-polyacrylamide gel electrophoresis as described above.
Quantitation of the Phosphorylated and Unphosphorylated Forms of the ␣ Subunit of eIF2-The proportion of eIF2␣ in phosphorylated and unphosphorylated forms was determined using slab gel isoelectric focusing electrophoresis followed by protein immunoblotting. Aliquots of post-mitochondrial supernatants were heated for 3 min in SDS sample buffer at 100°C, cooled to room temperature, and then mixed with 0.8 volume of isoelectric focusing gel buffer (0.1 g of dithiothreitol, 0.4 g of CHAPS, 5.4 g of urea, and 1 ml of Ampholytes (pH 3.5-9.5 from Pharmacia/LKB) in 6 ml of water). Proteins were resolved and detected using a rabbit monoclonal anti-eIF2␣ antibody as described elsewhere (19) and detected with ECL.
Measurement of eIF2B Activity-Determination of eIF2B activity in liver was performed exactly as outlined elsewhere (20) by measuring the rate of exchange of [ 3 H]GDP, which is present in an exogenous eIF2⅐[ 3 H]GDP complex, for free, nonradiolabeled GDP. Briefly, following excision of the liver, the tissue was rinsed in ice-cold saline, weighed, and homogenized in a Polytron in four volumes of buffer consisting of 20 mM triethanolamine, pH 7.0, 2 mM magnesium acetate, 150 mM potassium chloride, 0.5 mM dithiothreitol, 0.1 mM EDTA, 250 mM sucrose, 5 mM EGTA, and 50 mM ␤-glycerophosphate. Homogenates were then centrifuged for 10 min at 12,000 ϫ g at 4°C. Supernatants were then assayed for guanine nucleotide exchange activity. Essentially, 35 l of a prepared binary complex, which was assembled by incubation of purified eIF2 with 1.3 M [ 3 H]GDP (10.7 Ci/mmol), was combined with a mixture consisting of 35 l of liver homogenate, 87.5 l of water, and 140 l of buffer A (50 mM MOPS, pH 7.4, 209 M GDP, 2 mM magnesium acetate, 100 mM potassium chloride, 1 mM dithiothreitol, and 200 g/ml bovine serum albumin). The reaction was initiated by combination of these reactants and transfer to a 30°C water bath. At five time points (0, 2, 4, 6, and 8 min), a 75-l aliquot was removed and placed into tubes containing 2.5 ml of ice-cold wash buffer (buffer A devoid of bovine serum albumin). The contents were mixed and immediately filtered through a nitrocellulose filter disc. The guanine nucleotide exchange activity was measured as a decrease in eIF2⅐[ 3 H]GDP complex bound to the filters.
Determination of eIF2B⑀ Kinase Activity-The activity of eIF2B⑀ kinase(s) was performed as described previously (21) except that 0.5 g of purified, recombinant eIF2B⑀ was used as substrate in the reaction instead of the purified eIF2B holoenzyme.

RESULTS
The amino acids leucine, glutamine, and tyrosine are known to hinder proteolysis; therefore, the role(s) of these regulatory amino acids in modulating protein synthesis was evaluated in the perfused liver. Perfusion with a complete, 10ϫ amino acid mixture resulted in a 25% elevation in overall protein synthesis compared with livers administered a mixture of amino acids at 1ϫ (Fig. 1A). Intriguingly, perfusion with an imbalanced amino acid mixture comprised of leucine, glutamine, and tyrosine at 10ϫ, whereas all other amino acids were maintained at 1ϫ depressed protein synthesis by almost 50% relative to control. Thus, whereas heightened levels of a total amino acid mixture had an anabolic effect in the liver, equimolar amounts of the three regulatory amino acids impaired protein synthesis. This regulatory amino acid-mediated phenomenon became manifest at approximately 4ϫ concentrations (Fig. 1B). Although raising the concentration of leucine, glutamine, and tyrosine to 4ϫ depressed protein synthesis, this hepatic response was exacerbated by a further increment in amino acid concentration to 10ϫ.
To determine the mechanism(s) by which leucine, glutamine, and tyrosine affects overall protein synthesis, their influence on particular translational control points were evaluated. Surprisingly, raising the concentration of leucine, glutamine, and tyrosine while maintaining all other amino acids at 1ϫ potently disrupted the inhibitory eIF4E⅐4E-BP1 complex ( Fig. 2A) concomitant with hyperphosphorylation of 4E-BP1 (Fig. 2B) and p70 S6k (Fig. 2D). Activation of hyperphosphorylated p70 S6k was evidenced by hyperphosphorylation of endogenous S6 with increasing concentration of leucine, glutamine, and tyrosine (data not shown). Furthermore, eIF4E preferentially associated with eIF4G under these conditions (Fig. 2C). Although doubling the concentrations of the regulatory amino acids (that is, 2ϫ) relative to others exerted little effect on these factors, the phosphorylation states of 4E-BP1 and p70 S6k were marked enhanced at 4ϫ. Collectively, these results suggest that elevations in circulating levels of leucine, glutamine, and tyrosine above 2ϫ promote assembly of eIF4F and activation of p70 S6k despite a simultaneous concentration-dependent inhibition of global translation.
Insulin potently modulates the phosphorylation states and activities of components of the eIF4F system as well as p70 S6k (15,22,23). Therefore, we sought to characterize the role of insulin on these processes under conditions of mild amino acid imbalance. However, in the presence of either 0.5ϫ (which was not statistically different from 1ϫ; not shown) or 2ϫ concentrations of leucine, glutamine, and tyrosine, insulin was without effect on the liver's protein synthetic response (Fig. 3A). Whereas it appears that, in the hepatocyte, the anabolic effects of insulin are secondary to those of amino acids, insulin and amino acids synergistically promote assembly of eIF4F and activation of p70 S6k (12,24). Our findings in the perfused liver corroborate these reports. In combination with insulin, the three regulatory amino acids additively promoted the disunion of the eIF4E⅐4E-BP1 complex and the association of eIF4E with eIF4G (Fig. 3, B and C). Although phosphorylation of p70 S6k was enhanced slightly in the presence of insulin at 0.5ϫ leucine, glutamine, and tyrosine, an additive influence of insulin was masked at higher amino acid concentrations perhaps because p70 S6k was maximally activated in the presence of 2ϫ amino acids (Fig. 3D). In essence, insulin promoted eIF4F assembly within the physiologic realm but remained an impotent determinant of global protein synthesis.
Proteolytic studies conducted in perfused rat liver have revealed that of the seven individual amino acids with inherent regulatory properties, macroautophagic proteolytic inhibition by leucine was unrivaled, although glutamine and tyrosine enhanced this inhibition when added in combination (25,26). These three amino acids appear to reciprocally modulate autophagic proteolysis and the signal transduction pathway(s) leading to eIF4F complex formation and activation of p70 S6k . Therefore, the contribution of leucine to the activating properties of the three regulatory amino acids was examined in the context of eIF4F assembly and p70 S6k activation.
As shown previously, concentrations of the regulatory amino acid trio mimicking the upper physiologic threshold (that is, 4ϫ) inhibited hepatic protein synthetic activity approximately 25% (Fig. 4A). However, replacing leucine with equivalent amounts of the structurally similar BCAA, isoleucine, prevented the observed defect in protein synthesis. Furthermore, a perfusate containing 4ϫ leucine elicited a synthetic response intermediate between that observed with a complete, 1ϫ mixture of amino acids and that of 4ϫ leucine, glutamine, and tyrosine. These data suggest not only that these phenomena are leucine-specific but also that glutamine and tyrosine are only marginally influential in the absence of leucine.
Leucine, glutamine, and tyrosine, when present at 4ϫ concentrations, effectively destabilized the eIF4E⅐4E-BP1 complex (Fig. 4B). However, substitution of leucine with isoleucine in the perfusate was virtually without effect on the stability of the complex, implying that leucine is required for maximal effect.
Moreover, leucine appears to play a dominant role in this process because leucine alone nearly reduplicated the uncou- pling of eIF4E⅐4E-BP1 observed with leucine, glutamine, and tyrosine. Furthermore, eIF4E and eIF4G were conjoined in the presence of the three regulatory amino acids, whereas exchange of leucine with isoleucine in the perfusing medium attenuated formation of this complex (Fig. 4C). Again, leucine alone harbored the bulk of the influence of the regulatory triplet. Finally, optimal hyperphosphorylation of p70 S6k was achieved in the presence of leucine, glutamine, and tyrosine, whereas the combination of isoleucine, glutamine, and tyrosine produced a pattern of phosphorylation similar to that of the control amino acid mixture (Fig. 4D). Phosphorylation of p70 S6k by leucine alone was interjacent to the cumulative effect of leucine, glutamine, and tyrosine and that of control. Thus, it appears that the influences of the three regulatory amino acids seen here are largely attributable to the weighty contribution(s) of leucine. However, the effect of leucine on the phosphorylation states of both 4E-BP1 and p70 S6k , although augmented, remained submaximal, suggesting that the presence of glutamine and tyrosine serves to enhance these leucine-induced responses.
Alterations in eIF4F complex assembly and/or eIF2B activity are often sufficient to account for corresponding changes in total protein synthesis. Because, in this inquiry, protein synthesis and regulatory amino acid concentrations were inversely correlated, and the diminution of total protein synthetic activity was independent of eIF4F assembly, an examination of the activity of eIF2B was undertaken. The guanine nucleotide exchange activity of eIF2B was inversely related to ambient concentrations of leucine, glutamine, and tyrosine. In fact, the percentage of decline in eIF2B activity virtually mirrored the depression in protein synthesis observed at identical regulatory amino acid concentrations (c.f. Fig. 1B and Fig. 5A) and was independent of modified expression of eIF2B subunits (data not shown). Hence, the dose-dependent attenuation of eIF2⅐GTP regeneration was sufficient to account for the suppression of global protein synthesis conduced by leucine, glutamine, and tyrosine. The mechanism triggering these changes in the exchange activity of eIF2B was not attributable to phosphorylation of eIF2␣ because the proportion of the phosphorylated species did not change appreciably as a function of regulatory amino acid concentration (data not shown). As expected, these changes in eIF2B activation were dictated by the superior influence of leucine. The rate of guanine nucleotide exchange diminished in the presence of a 4ϫ mixture of leucine, glutamine, and tyrosine relative to that of a complete, 1ϫ mixture (Fig. 5B). Not unexpectedly, the catalytic activity of eIF2B was unaffected when isoleucine was substituted for leucine in the perfusing medium. Once again, however, the observed leucine-specific depression in eIF2B activity was independent of a change in eIF2␣ phosphorylation (Fig. 5C). Taken together, the findings not only demonstrate that leucine, glutamine, and tyrosine, when present at concentrations of 4ϫ, minimize the rate of eIF2B-mediated guanine nucleotide exchange, but also that leucine is indispensible for this effect.
Because substantial evidence has accumulated to place mTOR downstream of amino acid-induced signals, we sought to address the requirement of mTOR as a mediator of translational regulation by leucine, glutamine, and tyrosine. Rapamycin is a macrolide with immunosuppressive properties that binds with high affinity to endogenous FK506-binding proteins. Although several members of this protein family have been demonstrated to interact with rapamycin, only the FK506binding protein 12⅐rapamycin complex directly interacts and thereby inhibits the kinase activity associated with or intrinsic to mTOR. Thus, rapamycin has proven to be a powerful agent in elucidation of signal transduction pathways downstream of mTOR. Addition of this compound to the perfusing medium had little effect under basal, normoacidemic conditions, suggesting that at concentrations of 1ϫ, the eIF4 system and p70 S6k are largely inactive (Fig. 6). However, whereas 10ϫ concentrations of leucine, glutamine, and tyrosine robustly induced the appearance of 4E-BP1-␥ (Fig. 6B) as well as slower electrophoretic species of p70 S6k (Fig. 6D), rapamycin completely abolished these effects. Moreover, the corresponding proteinprotein interactions governed by the phosphorylation state of 4E-BP1 were similarly rapamycin-sensitive (Fig. 6, A and C). Therefore, it appears that mTOR is a key intermediate in the regulatory amino acid signals impacting eIF4 and p70 S6k .
Consistent with demonstrations of the subtle influence of rapamycin on general protein synthesis (27,28), the macrolide is largely ineffective in the modulation of global protein synthesis at both normal and supraphysiologic amino acid concentrations (Fig. 7A). Furthermore, the reduced activity of eIF2B engendered by amino acid imbalance was unmitigated by rapamycin (means Ϯ S.E. for 1ϫ ϩ Rap versus 10ϫ ϩ Rap were 23.3 Ϯ 1.3 versus 12.9 Ϯ 2.1, respectively), suggesting that FIG. 5. Regulatory amino acids inhibit the activity of eIF2B dose-dependently and leucine-specifically. A, livers were perfused with an imbalanced amino acid mixture containing 0.5ϫ, 2ϫ, 4ϫ, or 10ϫ concentrations of leucine, glutamine, and tyrosine, whereas all remaining amino acids were maintained at 1ϫ. The guanine nucleotide exchange activity of eIF2B was determined as described under "Experimental Procedures." B, livers were perfused as described in A with one of three amino acid mixtures: 1) a complete, 1ϫ mixture of amino acids; 2) an imbalanced mixture containing 4ϫ leucine, glutamine, and tyrosine while all other amino acids were maintained at 1ϫ; or 3) an imbalanced mixture containing 4ϫ isoleucine, glutamine, and tyrosine with all remaining amino acids at 1ϫ. eIF2B activity was determined as described under "Experimental Procedures." C, whole cell extracts were subjected to isoelectric focusing followed by protein immunoblot analysis for the ␣ subunit of eIF2 as described under "Experimental Procedures." The blots and values are the means Ϯ S.E. and are representative of three individual experiments. A, *, p Ͻ 0.05 versus 0.5ϫ; †, p Ͻ 0.05 versus 0.5ϫ. B, *, p Ͻ 0.05 versus 1ϫ; †, p Ͻ 0.05 versus 4ϫ IQY. p values were determined using ANOVA and Tukey post-test comparisons.
although mTOR is likely to be activated by amino acids in the perfusing medium, its activation is unrelated to the impaired eIF2B activity induced by amino acid imbalance. The eIF2Bmediated protein synthetic reduction occurred concomitant with attenuation of kinase activity targeting the ⑀ subunit of eIF2B (Fig. 7B). Although a trend of diminished eIF2B⑀ kinase activity in the presence of rapamycin was noted at 1ϫ amino acid concentrations, this compound did not affect the activity of the eIF2B⑀ kinase(s) at 10ϫ. DISCUSSION This inquiry has demonstrated that a 10-fold increase in the concentration of all amino acids reportedly found in rat plasma gives rise to a 25% elevation of protein synthesis in the perfused liver. Meanwhile, the hepatic response to elevated levels of leucine, glutamine, and tyrosine, potent regulators of macroautophagic proteolysis, is characterized by an eIF2B-mediated reduction in overall protein synthesis with coinstantaneous activation of two events usually associated with enhanced mRNA translation: assembly of eIF4F and phosphorylation of p70 S6k . We propose that this apparent paradox may represent an important circumstance surrounding protein turnover in the liver. Because eIF4F activity can be dissociated from increased global protein synthesis, an essential yet insufficient role for leucine in signaling the protein synthetic appa-ratus is revealed. Restated, although leucine alone is a potent activator of eIF4F and p70 S6k , cohort amino acids house the remaining signal(s) necessary for activation of eIF2B and, therefore, elevated protein synthesis.
Although traditionally eIF4E has been considered to be integrally involved in the regulation of global protein synthesis, contemporary research has revealed that this initiation factor plays an especially important role in the translation of mRNAs with extensively structured 5Ј-untranslated regions; this transcript class includes cyclin D1, Myc, ornithine decarboxylase, and ornithine aminotransferase (29,30). The notion of the strict requirement of global translation for participation of eIF4E has, no doubt, been perpetuated by reports of its presence in limiting quantities. However, this theory remains controversial (31). It has been postulated that incorporation of eIF4E into the initiation heterocomplex, eIF4F, may be responsible for discriminatory mRNA translation (32,33). The demonstration that rapamycin, which abrogates eIF4F assembly (Fig. 6), hinders the insulin-stimulated translation of Myc mRNA while leaving the translation of ␤-actin transcripts unaffected corroborates this hypothesis (34).
The activation of p70 S6k and subsequent phosphorylation of the 40S ribosomal protein S6 have been shown to correlate with increased translation of the TOP-containing transcript family whose nucleotide signature is a stretch of several pyrimidines near the 5Ј cap structure (27,35). These mRNAs encode ribosomal proteins and other components of the translational machinery. Although rapamycin has been shown to inhibit 4E-BP1 phosphorylation, its effect on general translation is slight FIG. 6. Activation of eIF4F and p70 S6k by the regulatory triplet is rapamycin-sensitive. Livers were perfused with either 1ϫ or 10ϫ leucine, glutamine, and tyrosine, whereas all remaining amino acids were maintained at 1ϫ. Perfusion medium either excluded or included rapamycin (100 nM) as indicated. Extracts of perfused liver were immunoprecipitated with a monoclonal anti-eIF4E (A and C) or a monoclonal anti-4E-BP1 antibody (B) as described under "Experimental Procedures." The immunoprecipitates were subjected to protein immunoblot analysis for 4E-BP1 (A and B) and eIF4G (C). Whole cell extracts were also immunoblotted for p70 S6k ; arrows indicate various electrophoretic species as described in the legend of Fig. 2 (27,28). Hence, despite the involvement of both p70 S6k and 4E-BP1 in mRNA translation, the role of these factors under normal circumstances may not be in the regulation of global protein synthesis but rather in the selective expression of a particular subset of proteins. In fact, eIF4F and p70 S6k are dissociable from the regulation of overall protein synthesis during histidine deprivation in L6 myoblasts; rather, the activity of eIF2B determines protein synthetic rate (7). Indeed, in the present study, regulation of general protein synthesis occurs in an eIF4F-independent fashion.
How is a signal generated by leucine, glutamine, and tyrosine, and which signal transducers are involved in propagation of that signal? Others have reported that whereas a combination of leucine, tyrosine, and phenylalanine effectively inhibits hepatic autophagic proteolysis, these amino acids enhance phosphorylation of ribosomal protein S6, the physiological substrate of p70 S6k (36). Moreover, the same investigation revealed that inhibition of proteolysis by amino acids was rapamycin-sensitive, implicating mTOR in mediation of the amino acid-induced proteolytic signal. It is fairly well entrenched that mTOR lies proximal both to 4E-BP1 and p70 S6k and may represent a functional bifurcation point that signals both targets (30,32,(37)(38)(39)(40).
Several studies have now shown that the availability of amino acids can potently affect the phosphorylation states of eIF4E, 4E-BP1, elongation factor 2, and p70 S6k as well as influence eIF4F assembly and eIF2B activity. Moreover, the effects of amino acid withdrawal on eIF4F assembly and activation of p70 S6k are sensitive to rapamycin (11,12,24,41,42). Chinese hamster ovary cells deprived of amino acids exhibit rapid dephosphorylation of p70 S6k and 4E-BP1; this effect can be reversed upon readdition of amino acids to the culture medium. Furthermore, this reversal was reportedly rapamycinand wortmannin-sensitive, implying that mTOR and PI 3-kinase, respectively, are involved in transduction of this signal (11). Although previous investigations have shown that protein kinase B may act upstream of p70 S6k (43,44), it has been excluded as a candidate intermediator (11). In a similar investigation, amino acid deprivation hindered p70 S6k and 4E-BP1 phosphorylation but was without effect on insulin-stimulated tyrosine phosphorylation, phosphotyrosine-associated PI 3-kinase activity, protein kinase B activity, and MAP kinase activity, thus precluding the involvement of proximal insulin-signaling effectors and the Ras/MAP kinase pathway in the amino acid-generated signal (12). Finally, HEK293 cells transiently transfected with a rapamycin-resistant p70 S6k variant were protected from inhibition by amino acid withdrawal (12). Taken as a whole, these data suggest that the amino acid-specific input(s) may impinge directly on mTOR or may initiate an as yet undefined signaling pathway which, at some point, includes mTOR.
Although amino acids activate the path to 4E-BP1 and p70 S6k phosphorylation presumably via mTOR, a second rapamycin-insensitive signal initiated by amino acids (or amino acid imbalance) impacts a distinct component of the translational apparatus. The best characterized mechanism of regulation of the guanine nucleotide exchange activity of eIF2B is that of phosphorylation of the ␣ subunit of eIF2 (reviewed in Ref. 45). However, the ⑀ subunit of eIF2B is the substrate, at least in vitro, for casein kinase I and II (46) as well as glycogen synthase kinase 3 (GSK-3) (47). GSK-3 phosphorylates eIF2B⑀ at Ser 535 in the rat, which is conserved among mammals and is dephosphorylated coordinately with inactivation of GSK-3 in response to insulin (47). Moreover, transfection of Chinese hamster ovary cells overexpressing the human insulin receptor with dominant negative variants of the p85 regulatory subunit of PI 3-kinase or the mammalian homolog of the son of sevenless guanine nucleotide exchange factor, which interfere with PI 3-kinase and MAP kinase signaling, respectively, demonstrates an obligatory role of PI 3-kinase in regulation of eIF2B⑀ dephosphorylation in response to insulin (48). The signal to eIF2B⑀ emanating from PI 3-kinase appears to diverge upstream of mTOR, because rapamycin prevents neither the insulin-induced inactivation of GSK-3 nor the activation of eIF2B activity (48). In this inquiry, rapamycin did not affect the activity of eIF2B⑀ kinase(s), although this activity was reduced by a disproportionate elevation of leucine, glutamine, and tyrosine. Because in perfused liver, 10ϫ concentrations of a balanced amino acid mixture augments protein synthesis (Fig.  1A), reduces the nonpolysomal population of ribosomal particles (49), and enhances eIF2B activity (49), it seems plausible that eIF2B⑀ kinase activity (at least that eIF2B⑀ kinase activity which increases the activity of eIF2B), would either remain undiminished or elevated in response to balanced hyperaminoacidemia. If this is the case, then the signal(s) influencing the activity of eIF2B (perhaps via inactivation of eIF2B⑀ kinase(s)) under the conditions of this investigation are likely due to the phenomenon of imbalance itself (that is, leucine imbalance) and not due to typical amino acid-induced signals. It will be of interest to determine the effects of amino acid imbalance on the regulation and activity of eIF2B in physiologic and pathologic scenarios.
Studies performed in perfused liver have shown that multiphasic inhibition by leucine of deprivation-induced proteolysis can be mimicked by structural analogs such as ␣-hydroxyisocaproate (4), isovaleryl-L-carnitine (26), and 4-amino-6methylhept-2-enoic acid (50), suggesting that the proteolytic signal is generated by neither leucyl-tRNA nor leucine-enriched peptides. Also, p70 S6k is similarly activated in FAO hepatoma cells exposed to equimolar amounts of either leucine or ␣-ketoisocaproate, a leucine metabolite. Furthermore, inhibition of protein synthesis in the perfused liver under conditions of low amino acid concentrations is independent of tRNA charging (13). Poor transamination of leucine in the liver has excluded byproducts of leucine metabolism from serious consideration as mediators of this signal (5,51). A nontransportable multiple antigen peptide derivative constructed by attaching eight leucine residues to a lysine core (Leu 8 -MAP) was effective in suppressing deprivation-induced macroautophagy in isolated hepatocytes with an apparent K m equivalent to that of leucine (52). Furthermore, photoaffinity-labeling experiments revealed that the putative Leu 8 -MAP substrate was a protein of approximately 340,000 M r and was enriched within membrane-fractions, suggesting that this factor may be a plasma membrane receptor (53). In light of these data, it is tempting to speculate that leucine-(and/or regulatory amino acid-) specific signals not only regulate hepatic macroautophagic proteolysis, eIF4F assembly, and p70 S6k but also, owing to a shared rapamycin sensitivity, may do so via a common mechanism.
Do physiological circumstances exist that predetermine aminoacidemia, and does leucine, glutamine, and/or tyrosine play a participatory role in these processes? For over 30 years, hyper branched chain aminoacidemia has been a hallmark serological perturbation intrinsic to the pathology of diabetes (54). Diabetic cachexia results, in part, from net protein catabolism in which proteolysis exceeds protein synthesis, particularly in skeletal muscle. Furthermore, the disequilibration of protein degradation and protein synthesis adversely affects tissue repair; thus, diabetics are especially susceptible to injury or infection (55). Either insufficient endogenous insulin secretion (insulin-dependent diabetes mellitus) or peripheral insulin re-sistance (non-insulin-dependent diabetes mellitus) disables the predominant site of BCAA disposal (that is, skeletal muscle). As a result, this tissue is refractory to insulin-stimulated BCAA uptake. Whereas these amino acids are metabolized quickly in the periphery, low level expression of branched chain amino acid transaminase renders hepatic BCAA metabolism nominal (1). The inability of the liver to clear plasma BCAAs is strikingly manifest in artificial models of diabetes in that hypoinsulinemia hinders the muscle-specific uptake of these amino acids. Because muscle is the primary metabolic sink for BCAAs, and because this mechanism for clearance of BCAAs is inoperative in diabetes, hyper branched chain aminoacidemia ensues under both fasting and fed conditions. Although the concentrations of the BCAAs in diabetic animals are approximately 2ϫ in either the postabsorptive or fed condition, the levels of these amino acids may rise uncontrollably to 6 -7ϫ following ingestion of a protein-enriched meal (56). Moreover, streptozotocin-induced diabetic rats maintained on a diet of 20% protein exhibit serum BCAA concentrations tripling those of controls (57). Although it has been proposed that the elevated levels of circulating BCAAs may serve to counteract the depressed rate of protein synthesis observed in diabetes, we find this possibility excludable in the liver. In fact, leucine, which of the BCAAs has been most acutely implicated in enhancing protein synthesis, inhibits this process in a concentrationdependent manner.
In summary, this investigation has revealed that although a complete amino acid mixture dose-dependently augments protein synthesis in the liver, elevating the levels of leucine, glutamine, and tyrosine relative to other perfusate amino acids depresses this process. Although this protein synthetic attenuation is attributable to a defect in eIF2B-mediated guanine nucleotide exchange, it is separable from activation of eIF4F. Moreover, the effects on protein synthesis, eIF2B, eIF4F, and p70 S6k are predominantly leucine-specific. The unique physiologic scenario engendered in this study demonstrates that the hepatic protein synthetic response requires the integration of both leucine-specific and cryptic amino acid-induced inputs. The aberrant signal(s) generated by leucine imbalance compels translational components to function with highly disparate efficiencies. As a result, the relatively sluggish rate of eIF2Bmediated catalysis determines the overall rate of protein synthesis. Moreover, although leucine alone is insufficient to stimulate protein synthesis, it does activate the translational machinery implicated in expression of two unique mRNA subclasses, suggesting that activation of the eIF4 system and p70 S6k may have functional consequences.