Direct evidence that the protein kinase catalytic subunit mediates the effects of cAMP on tyrosine aminotransferase synthesis.

The effect of purified beef heart cAMP-dependent protein kinase catalytic subunit on tyrosine aminotransferase activity in intact cultured rat H35 hepatoma cells was directly tested by micro-injection using human red blood cell ghosts as vehicles. Although the micro-injection procedure itself produced temporary fluctuations in protein synthesis and in tyrosine aminotransferase activity in H35 cells, after a recovery period of 8-12 h, these parameters returned to normal in parallel with restoration of full inducibility of the aminotransferase by both 8-Br-cAMP and dexamethasone. Eight to sixteen hours after fusion of H35 cells with unloaded ghosts, ghosts loaded with bovine serum albumin or mock-loaded with the partially purified protein kinase catalytic subunit, no significant change in the activity of the aminotransferase was detected. In contrast, fusion with ghosts loaded with the catalytic subunit at concentrations between 0.1-2 mg/ml caused reproducible 2-3-fold increases in enzyme activity. Homogeneous preparations of the catalytic subunit exhibited even greater potency as an inducer. The effect was both time- and concentration-dependent and was abolished by inactivation of the catalytic subunit with N-ethylmaleimide prior to loading. The partially purified inhibitor of protein kinase from beef heart, while not affecting basal tyrosine aminotransferase activity, selectively inhibited the ability of 8-Br-cAMP but not that of dexamethasone to stimulate the activity of this enzyme. In addition, micro-injection of the pure regulatory subunit of the kinase blocked the response of the aminotransferase to low concentrations of 8-Br-cAMP. These results provide strong support for the proposition that the catalytic subunit of protein kinase mediates the effects of cAMP on the synthesis of tyrosine aminotransferase.

The effect of purified beef heart CAMP-dependent protein kinase catalytic subunit on tyrosine aminotransferase activity in intact cultured rat H35 hepatoma cells was directly tested by micro-injection using human red blood cell ghosts as vehicles. Although the micro-injection procedure itself produced temporary fluctuations in protein synthesis and in tyrosine aminotransferase activity in H35 cells, afier a recovery period of 8-12 h, these parameters returned to normal in parallel with restoration of full inducibility of the aminotransferase by both 8-Br-CAMP and dexamethasone.
Eight to sixteen hours after fusion of H35 cells with unloaded ghosts, ghosts loaded with bovine serum albumin or mock-loaded with the partially purified protein kinase catalytic subunit, no significant change in the activity of the aminotransferase was detected. In contrast, fusion with ghosts loaded with the catalytic subunit at concentrations between 0.1-2 mg/ml caused reproducible 2-3-fold increases in enzyme activity. Homogeneous preparations of the catalytic subunit exhibited even greater potency as an inducer. The effect was both time-and concentration-dependent and was abolished by inactivation of the catalytic subunit with N-ethylmaleimide prior to loading.
The partially purified inhibitor of protein kinase from beef heart, while not affecting basal tyrosine aminotransferase activity, selectively inhibited the ability of 8-Br-CAMP but not that of dexamethasone to stimulate the activity of this enzyme. In addition, micro-injection of the pure regulatory subunit of the kinase blocked the response of the aminotransferase to low concentrations of 8-Br-CAMP. These results provide strong support for the proposition that the catalytic subunit of protein kinase mediates the effects of CAMP on the synthesis of tyrosine aminotransferase.
Although the role of CAMP-dependent protein kinase is now well established in the control of glycogen metabolism (1, 21, the role of this enzyme as a mediator of the effects of cAMP on selective protein synthesis in eukaryotic cells remains uncertain (3,4). Considerable indirect, correlative data have been obtained which are consistent with the kinase acting as a mediator of cAMP action on the synthesis of selected proteins (3-6) but no direct, causal evidence has been * This work was supported by Grant AM 28545 from the United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article with 18 U.S.C. Section 1734 solely to indicate this fact, must therefore be hereby marked "aduertisenent" in accordance +Present address, Department of Pharmacology, University of Minnesota, Minneapolis, MN 55455.
8 To whom all inquiries should be sent.
reported. Studies with variant cells, resistant to the cytotoxic effects of cAMP derivatives as a result of alterations in kinase subunits, have implicated the kinase in the control of cell death, steroidogenesis and phosphodiesterase activity (7-9). The last two processes appear to involve selective increases in protein synthesis although this has not been rigorously established as yet. Maller and Krebs (10) provided direct evidence that the catalytic subunit of CAMP-dependent protein kinase is responsible for the meiotic arrest of Xenopus laevis oocytes caused by persistently elevated levels of CAMP. Theoretically, these studies provided a means by which the role of protein kinase could be tested in any system but to date such an approach has been limited to mechanical injection of individual cells (11)(12)(13) where responses can be readily detected at the single cell level. Such an approach, of course, cannot be used in cases where the response being monitored requires thousands of cells in order to be detected. We have been studying just such a case, in a cultured rat hepatoma cell line, where cAMP derivatives and adrenal steroids selectively stimulate, by apparently different mechanisms, the synthesis of tyrosine aminotransferase (EC 2.6.1.5) (14, 15). Although a sensitive enzyme assay exists (16), at least 50-100 X lo3 cells are needed to obtain reliable values for basal aminotransferase activity.
With the advent of a feasible procedure for simultaneously micro-injecting millions of cells with foreign proteins using as vehicles, protein-loaded RBC' ghosts, as pioneered by Mekada et al. (17) and Schlegel and Rechsteiner (18), a means was provided to directly test the role of protein kinase as a mediator of the induction of tyrosine aminotransferase by CAMP. We have made use of this approach and have succeeded in micro-injecting protein kinase subunits into rat hepatoma cells. The results obtained provide direct evidence for a mediatory role of the catalytic subunit of CAMP-dependent protein kinase in controlling the synthesis of the aminotransferase. grown and maintained in monolayer culture using serum-containing Dulbecco's MEM as described previously (5,14). Cells were used at 70-90% confluency for fusion and are referred to as confluent monolayers in the text. The procedures for harvest of cells and lysis were as previously reported (5).
Purification and Labeling of Proteins-The C subunit of CAMPdependent protein kinase was purified from beef heart by a combination of the procedures described by Beavo et al. (21) and Corbin et al. (22). This procedure yielded a preparation that exhibited an average purity of 60-80% based upon sodium dodecyl sulfate-gel electrophoresis on slab gels (23). The yield of catalytic subunit was typically 10-30 units/kg assayed against histone HIIA (Sigma) by the procedure of McPherson et al. (24). There was no significant effect of cAMP on the activity of the purified preparations of the catalytic subunit. The average specific activity of the different preparations pmol of 32P transfered to HIIA/min/mg of protein at 30 "C. The used was 3.5 +_ 1.3 units/mg of protein ( n = 8). A unit of activity is 1.0 catalytic subunit was inactivated (98.4% activity was lost) by incubation in 300 PM N-ethylmaleimide for 4 h at 22 "C (25). Homogeneous catalytic subunit from beef heart and type I1 regulatory subunit (R subunit) from rabbit skeletal muscle were kindly provided by Dr. Jackie Corbin of Vanderbilt University. Homogeneous catalytic subunit from beef heart was also generously supplied by Dr. Edmund Fischer, University of Washington..
The heat stable inhibitor of protein kinase was partially purified from beef heart by modifications of the procedure of Whitehouse et al. (26). The material used in most experiments was 1530% pure as judged by sodium dodecyl sulfate-gel electrophoresis. The average yield of the inhibitor was 6,000 units/kg and the average specific activity of the three different preparations was 166,000 units/mg of protein. The activity of the protein kinase inhibitor was assayed with the standard HIIA phosphorylation procedure (24) using highly purified kinase catalytic subunit preparations (average of 4.6 units/mg of protein). Under the conditions used, the suppression of HIIA phosphorylation was proportional to the amount of inhibitor added. The unit of activity is that amount of inhibitor which depresses phosphorylation of HIIA by 15 pmol/min at an activity of the catalytic subunit of 100 pmol/min. Homogeneous inhibitor from rabbit skeletal muscle was kindly provided by Dr. Daniel Friedman of Vanderbilt University.
BSA was labeled with '"1 (27) and purified by chromatography on Bio-Gel P-10. The average specific radioactivity achieved was 1.79 X 10"' cpm/mg. BSA was labeled with FITC by the procedure of Mekada et al. (28) and separated from unreacted FITC by chromatography on Sephadex G25. This preparation exhibited an absorbance ratio (495 nm/280 nm) of 2.0 (28).
Tyrosine aminotransferase activity was measured in 20,000 X g supernatant fractions of H35 cell lysates by a modification of the method of Diamondstone (16) as published previously (14). A unit is defined as pg of product formed in 10 min at 37 "C. Protein content in all cases was determined by the method of Lowry et al. (29). Activity is expressed either as units/mg of protein, net increase (value for treated groupvalue for basal group) or fold increase (value for treated group i value for basal group).
Effects of fusion on protein synthesis were measured by the incorporation of ('H)leucine into hot trichloroacetic acid-insoluble material as described previously (14). The relative rate of synthesis of the aminotransferase was determined by incorporation of ("S)methionine during a 30-min interval. Radioactivity in tyrosine aminotransferase was determined by immunoprecipitation using specific rabbit antibodies while that in total soluble protein was measured by precipitation with trichloroacetic acid as described previously (14).
Red Blood Cell Ghost Preparation and Loading of Proteins-Fresh human blood (5-10 ml) was obtained by venipuncture from healthy donors. After centrifugation at 1500 X g at 4 "C and removal of plasma and the buffy coat, the packed RBC were washed 3 times with phosphate-buffered saline (137 mM NaC1, 2.7 mM KCl, 8.1 mM K2HP04, 1.5 mM Na2HP04, pH 7.2). An aliquot (0.15 ml) of the washed, packed RBC were mixed with 1.35 to 2.85 ml of PBS (generally 1.35 ml) containing the protein to be loaded at concentrations UP to 5 mg/ml. The procedure used for loading foreign proteins into RBC ghosts was based on the gradual dialysis method of Mekada et al. (17). The sample was placed inside boiled dialysis tubing (6 mm against 500-750 ml of 6-fold diluted rPBS (137 mM KCl, 2. 7 mM NaCI, diameter, Spectrapor) and dialyzed for 45 min at room temperature 8.1 mM Na2HP0.,, 1.5 mM K,HPO,, 4 mM MgC12, pH 7.2). The bag was then transferred to 500-750 ml of isotonic PBS and dialysis continued for 45 min at room temperature.
The loaded RBC ghosts were collected by centrifugation at 3000 X g and washed 4 times with 3 ml of cold isotonic PBS. The washed RBC ghosts were then suspended in 1.5-2.0 ml of cold glucose-free HBSS and stored at 4 "C until use. The average number of ghosts generated by this procedure was 4 X IOX/ml (with 1.5 ml final volume) as determined by a hemacytometer.
Fusion of Loaded Ghosts with H35 Cells-The procedure which has yielded the most consistent micro-injection results was that of Mercer et al. (30). H35 cell monolayers were washed once with glucose-free HBSS at 37 "C. Glucose-free HBSS containing phytohemagglutinin (Sigma) at 50 pg/ml was added to the monolayer (30) and swirled for 2 min. RBC ghosts were then added, usually 200 pl/ dish of H35 cells (-4 X IO6 cells at. 70-90% confluency = a ratio of 201). After swirling, the dishes were incubated for 60 min at 37 "C. The medium was then removed and the monolayer washed once with glucose-free HBSS.
To induce fusion, 0.5 ml of freshly prepared PEG-8000 (Sigma) at 22 or 44% (w/v) in 150 mM NaCl, 5 mM MnCL, 20 mM Tris-HC1, pH 7.4 was rapidly added dropwise around the dish. After incubation for 1 min, the PEG was diluted by addition of 8 mi of serum-free MEM and incubation was continued for 30 min at room temperature. The free MEM, 3 times with MEM containing serum (5% fetal calf and 5% medium was removed and the monolayer washed once with serumdonor calQ and incubated in serum-containing MEM with 5 pg/ml gentamycin at 37 "C prior to harvest and assays. In some cases, 8-Br-cAMP (2 mM) or dexamethasone (1 PM) was added 3 or 4 h prior to harvest. Untreated cells refer to dishes containing cells at comparable levels of confluency which have been kept in serum-containing MEM throughout the period that companion dishes were subjected to, and recovering from the process of fusion.

RESULTS
Loading of RBC ghosts-Using ('251)BSA as a model protein to test the effectiveness of loading of RBC ghosts, we obtained results very similar to those reported by Yamaizumi et al. (31). Linear loading of human RBC ghosts could be achieved by the procedure described under "Experimental Procedures" at concentrations of ('251)BSA in the dialysis bag up to at least 1 mg/ml (data not shown). Given the specific radioactivity of the labeled BSA, the radioactivity observed in the supernatant after lysis of the loaded ghosts with Hz0 and the number of ghosts in the sample, one can calculate that -10' molecules of BSA were loaded per ghost on the average a t 1 mg/ml in the dialysis bag. The radioactivity present in the supernatant fraction after the fourth wash of the loaded ghosts with isotonic PBS was t 2 % of that found in the supernatant fraction after lysis with H20.
As shown in Fig. 1, essentially all of the RBC ghosts (>99% of all the original RBC) exhibit some degree of fluorescence when loaded with FITC-BSA. There is noticeable variation in both the amount of BSA loaded and the morphology of the ghosts as has been seen by others (17). If ghosts were mockloaded with FITC-BSA (dialysis only against isotonic PBS), no detectable fluorescence was found associated with the ghosts after the usual washing procedure.
Storage of ghosts loaded with ('*'I)BSA at 4 "C for up to 24 h produced iess than 510% loss of labeled BSA. Exposure of loaded ghosts to trypsin did not lead to loss of (lzsI)-BSA until obvious hemolysis was observed. All of these observations are consistent with the conclusion that the gradual dialysis procedure resulted in the entrapment of substantial amounts of an exogenous protein such as BSA in the RBC ghosts. Based on results with ('251)BSA, the overall efficiency of loading averaged about 4% (3.7 n = 13) which represents somewhat less than 50% of the expected value for complete equilibration (150 p1 ghosts and 1. 35 ml of protein-containing solution).
Fusion Procedure-Using RBC ghosts loaded with ('2sI)BSA, only modest micro-injection of the ('"1)BSA into H35 cells was observed in the absence of phytohemagglutinin (data not shown). When the plant lectin was present at concentrations between 10 and 100 pg/ml, the amount of

VIS
(I2'I)BSA introduced was increased 5-7-fold in agreement with Mercer et al. (30). The maximum response was achieved between 50 and 100 pg/ml and, since lectins have been reported to affect tyrosine aminotransferase activity (32), we chose to use the lower concentration.
Various concentrations of PEG of different average chain lengths were tested for their efficiency as fusogens, as monitored by trichloroacetic acid-insoluble (I2'I) present in H35 cell lysates a t 12 h after fusion with ("'1)BSA-loaded ghosts. To summarize several different experiments, 44% PEG of chain length 8000 gave the greatest yield of micro-injected ("'1)BSA. Higher concentrations were not used for fusion because of obvious severe cytotoxic effects as detected by cell detachment from the substrate. By diluting the PEG 1Bfold after 1 min, the cytotoxic effects of 44% could be reduced such that cell detachment did not occur. However, as shown later, the response of H35 cells to inducers was reduced for several h after fusion even with these modifications of the original procedure of Mercer et al. (30). In later experiments, lower concentrations of fusogen were used and, although recovery of responsiveness was more rapid, the maximal response of tyrosine aminotransferase to micro-injected catalytic subunit was comparable down to 22% PEG.
Using these conditions with 44% PEG and ghosts loaded at a single concentration of the protein (2 mg/ml) in the dialysis bag, the amount of ("'1)BSA micro-injected into H35 cells varied linearly with the ratio of ghosts to cells (data not shown). At the highest number of ghosts added (200 pl/dish, -20 ghosts/H35 cell), we calculated from the acid-insoluble radioactivity in cell lysates that over 2 X 10" molecules of ('2'I)BSA were micro-injected into each H35 cell on the average. From the amount of BSA loaded, it was calculated that the efficiency of micro-injection in this experiment was about 4-5%. This is similar to values reported by others using the same or a different fusogen with different types of cells (18,31).
Electrophoresis of H35 cell lysates prepared 6 h after fusion of H35 cells with ghosts containing (I2'I)BSA revealed only a single radioactive component which co-migrated with the ("'1)BSA used for loading (data not shown). These results are consistent with intracellular delivery of the labeled BSA and uv suggest that significant quantities of the protein are still present in an undegraded form several h after fusion. Similar conclusions have been drawn by Neff et al. (33). Use of ghosts loaded with FITC-BSA has shown that 90% or more H35 cells exhibit significant fluorescence as seen in a fluorescence microscope (data not shown).
Recovery Period-Both PEG and Sendai virus are known to exert cytotoxic effects on cultured cells (18,30). We chose to use PEG since it is easier to obtain and because it has been reported to be superior to Sendai virus in fusing RBC ghosts to monolayer cultures (30). One unexpected manifestation of the use of PEG, however, was rapid elevation of tyrosine aminotransferase activity shortly after fusion ( Table I). Some increase was seen with all concentrations of PEG tested and was independent of the presence of RBC ghosts whether unloaded or loaded with BSA. Thus, the effect was not due to the injection of either BSA or hemoglobin.2 After 8-12 h, the aminotransferase activity returned to values typical for H35 cells not subjected to the fusion process (generally 15-30 units/mg of protein). The precise basis for this increase in enzyme activity has not been investigated but it appears to be the result of exposure to PEG, the process of fusion itself, or both, since phytohemagglutinin alone had no reproducible effect on enzyme activity in our hands.
The ability of CAMP derivatives to induce tyrosine aminotransferase was correspondingly diminished during the period when basal enzyme activities were elevated after fusion (data not shown). A similar period of refractoriness to the effects of dexamethasone was also observed (see below). Full responsiveness to these external inducers was restored after the 8-12-h period when basal tyrosine aminotransferase activities had been re-established. A likely explanation for the sharply reduced response to inducers was the significant inhibition of protein synthesis caused by the process of fusion (see below).
Although the inhibition of protein synthesis was less severe with lower percentages of PEG, the degree of fusion was also reduced (data not shown). The best compromise of fusion versus toxicity appeared to be offered by 22% PEG-8000 and 'Some hemoglobin remains in the loaded ghosts since they are distinctly red.  Effect of fusion conditions a n d various RBC ghost preparations on tyrosine aminotransferase activity in H35 cells Confluent monolayers of H35 cells were fused with or without human RBC ghosts prepared by dialysis against hypotonic rPBS alone; mock-loaded by dialysis against isotonic PBS containing BSA (1 mg/ml); BSA-loaded by dialysis against hypotonic rPBS containing BSA. Fusion was accomplished with 4 4 % PEG-8000 added following a 60-min incubation of H35 cells f appropriate RBC ghosts with 50 pg/ml phytohemogglutinin. After extensive washing with serum-containing medium, the cells were allowed to recover in serum-containing medium before harvest, lysis and assays of enzyme activity. The group not subjected to fusion was kept in serum-containing medium most of the experiments were performed with this concentration of fusogen. However, during the early phases of the work, 44% PEG was used to ensure maximum micro-injection of protein kinase subunits.

Effects of Protein Kinase Catalytic
Subunit-The C subunit of the CAMP-dependent protein kinase was purified to 6040% homogeneity from beef heart and loaded into human RBC ghosts under the same conditions as described above for BSA. The activity of the C subunit in the dialysis bag was assayed after each loading process and recovery averaged 66% with a range from 25 to 122%.
As shown in Table 11, fusion of H35 cells with RBC ghosts loaded with the C subunit caused a significant increase in aminotransferase activity over that observed with ghosts subjected to mock-loading with this kinase preparation. Since mock-loaded RBC ghosts did not cause any increase in enzyme activity, as was true for ghosts loaded only with BSA, the latter was used in all subsequent experiments to conserve purified C subunit.
At 13 h after fusion with the C subunit, the activity of tyrosine aminotransferase was elevated nearly 3-fold and the increase was identical to that achieved by 2 mM 8-Br-CAMP added to cells micro-injected with BSA.3 It should be noted that the C subunit did not provoke exactly the same degree of response as cAMP derivatives in every experiment, but it has generated 2-4-fold increases in aminotransferase activity in all experiments in which it has been used except one. In that experiment, the C subunit inexplicably lost >98% of its activity during loading and, upon micro-injection, this preparation did not elevate aminotransferase activity (data not shown).
Addition of free C subunit (5-25 pg) to cells subjected to The absolute degree of increase in aminotransferase activity with cAMP (5, 14). This increase largely depends on the basal activity the C subunit varies from experiment to experiment as is also true for which also varies. The range of increases seen with the C subunit was between 15 and 60 units/mg of protein. Ability of micro-injected protein kinase catalytic subunit to increase tyrosine aminotransferase activity in H35 cells Confluent monolayers of H35 cells were fused with human RBC ghosts prepared by dialysis against three different purified beef heart C subunit preparations (6040% pure) in isotonic PBS (mock-loaded) or hypotonic rPBS (loaded). The concentrations of C subunit in the while BSA was loaded at 1 mg/ml. The ratio of RBC ghosts to H35 dialysis bag were 0.4, 1.9, and 3.1 mg/ml in the three experiments cells was 20:l in Experiment 1 and 1 0 1 in Experiments 2 and 3. Fusion was achieved with 44% PEG and, after washing with serum-containing medium, the injected H35 cells were allowed to incubate for the intervals shown in serum-containing medium before harvest, lysis and added to ceUs micro-injected with BSA 4 h prior to harvest. Values assays of enzyme activity. Where indicated, 2 mM 8-Br-CAMP was

zase Role in Enzyme Induction
shown are mean f S. E. with number of observations in parentheses. fusion conditions without RBC ghosts produced occasional increases in aminotransferase activity. However, the response was erratic, correlated poorly with the amount of subunit added, and was less effective than that observed when the subunit was introduced by way of RBC ghosts (data not shown).
Once reproducible effects of the C subunit could be obtained using 4 4 % PEG, lower concentrations of the fusogen were tested. The response to the micro-injected kinase subunit appears to be somewhat more rapid with 22% PEG and was greater at both time points than with 4 4 % PEG. In contrast to 44% PEG, with 22% PEG, the increase in aminotransferase activity at 12 h was almost as great as at 16 h. The response to the C subunit at 20-24 h after fusion is similar to that at 16 h but does diminish past this time (data not shown).
Concentrations of PEG below 22% were less toxic to H35 cells and the ability of the C subunit to increase the activity of tyrosine aminotransferase in cells subjected to fusion at this concentration was slightly greater at 4 and 8 h after fusion. However, the maximum response was only about 70% of that obtained with 22% PEG and it diminished after 12 h in contrast to that observed with higher concentrations of the fusogen which showed additional increases. These results suggest that the efficiency of fusion at concentrations of PEG below 22% was too low to be practical with the concentrations of protein available to load ghosts.
Analysis of the incorporation of (3H)leucine into hot and cold trichloroacetic acid-insoluble material showed that the process of fusion with 22% PEG caused substantial inhibition of protein synthesis, as suspected (Fig. 2). The rate of recovery of protein synthesis after fusion correlated reasonably well with the return of inducibility of the aminotransferase by dexamethasone and with the response to the micro-injected C subunit. These results, coupled with the fact that the maximum effect of the kinase subunit on the aminotransferase is seen at about 16 h after fusion with 22% PEG (Fig. 4), strongly suggest that the depression in protein synthesis is likely to be responsible for the reduced responsiveness to steroids, cAMP to fusion relative to that in companion cells not subjected to fusion. No significant difference was detected between the two groups of cells used for fusion with regard to total acid-insoluble radioactivity. Each value is the mean of four (DEX, PK-C) or eight (protein synthesis) observations and the S. E. values were within 10%. The specific activity of the aminotransferase in cells micro-injected with BSA and to which no dexamethasone was added was 38.6 f 3.7 at 4 h, 32.3 f 0.6 at 8 h, and 29.1 f 2.4 at 12 h after fusion. derivatives, and the C subunit. This explanation seems quite plausible since steroids and CAMP elevate tyrosine aminotransferase activity by stimulating its rate of synthesis (3, 14,

15).
Inactivation of the C subunit (>98% loss of activity) with N-ethylmaleimide prior to loading abolished the ability of a C subunit preparation to stimulate aminotransferase activity (Table 111). The untreated C subunit in this experiment exhibited the usual recovery after the loading process and produced a typical 2-4-fold increase in the activity of the aminotransferase. These results are consistent with the suggestion that the C subunit is the active component present in the micro-injected preparation.
If this conclusion is correct, then the response to the C subunit should not only be dependent on time, but also on the amount of the subunit micro-injected. This proposition was tested in two different ways: first, by loading allquots of RBC ghosts with different concentrations of the C subunit while using a fured (-2O:l) ratio of ghosts to H35 cells during fusion, and, second, by increasing the ratio of ghosts to H35 cells during fusion while using two fured concentrations of the C subunit during loading. As shown in Fig. 3, the response of the aminotransferase proved to be dependent upon the amount of C subunit micro-injected with either approach.
The maximum response appears to be reached around 1 mg/ mI of the highly purified kinase preparation used during loading and at 200 pl of ghosts/dish (-2O:l ratio to H35 cell).

FIG.
3. Dependence of the response to micro-injected C subunit on the concentration of subunit loaded and on the multiplicity of loaded RBC ghosts. Each dish of confluent H35 cells was subjected to fusion at 22% PEG with 200 p1 of RBC ghosts loaded with the indicated concentrations of partially purified C subunit (left), and with the indicated volumes of RBC ghosts loaded with either 0.8 or 1.6 m g / d partially purified C subunit (right). The recovery period was 16 h at which time cells were harvested for assays as described under "Experimental Procedures." Each value is the mean of four observations obtained in one (right) or two different experiments (left) and the S. E. values were within 5%. The specific activity o f the aminotransferase in cells micro-injected with BSA was 12.8 f 1.1 (left) and 19.6 k 1.9 (right).
In order to rule out the possibility that some minor contaminant in the partially purified C subunit preparations was responsible for the effects on the aminotransferase, it was necessary to micro-inject pure C subunit. As shown in Fig. 4, the pure C subunit evoked a response virtually identical to that produced by 8-Br-CAMP. The maximum effect was achieved at a -2O:l ratio of ghosts to cells using ghosts loaded at 0.5 mg/ml which represents somewhat less C subunit than was needed with preparations which were only 60-8076 pure. The response at 200 pl of ghosts is at, or very close to the maximum effect expected since 2 mM 8-Br-CAMP is known to produce a maximum response in these cells (5,34). These results strongly suggest that the response of the aminotransferase to the partially purified preparations of the kinase must be due to the C subunit itself and not to some contaminant.
Addition of 8-Br-CAMP to cells micro-injected with the C subunit did not increase enzyme activity beyond that pro- duced by the cyclic nucleotide in BSA-injected cells. In contrast, addition of dexamethasone to cells micro-injected with C subunit led to increases significantly greater than was observed after addition of the steroid to cells micro-injected with BSA (data not shown). Such results are to be expected if the C subunit is the mediator of cAMP action in this system since these two agents induce the aminotransferase by apparently different mechanisms (14, 15).

Direct
Micro-injection of the C subunit significantly stimulated the relative rate of synthesis of tyrosine aminotransferase without causing any notable change in the rate of overall protein synthesis (Table IV). These results are consistent with reports that derivatives of cAMP exert their effects on the aminotransferase by selectively stimulating its synthesis (3, 14, 15). Dexamethasone, added to cells micro-injected with BSA 4 h prior to harvest, stimulated the rate of aminotransferase synthesis to a considerably greater extent than did micro-injected C subunit. This is consistent with the fact that adrenal steroids invariably produce a substantially higher degree of stimulation of enzyme synthesis than do cAMP derivatives (3, 14, 15).
Effects of the Inhibitor of Protein Kinase-If the C subunit of protein kinase is in fact responsible for the effects of cAMP on tyrosine aminotransferase, then micro-injection of the heat-stable inhibitor protein should block the ability of 8-Br-cAMP but not that of dexamethasone to induce the aminotransferase. In two different experiments with the inhibitor prepared from beef heart (1530% pure), 64-82% inhibition of the response of the aminotransferase to 8-Br-CAMP was observed with only a minor reduction in the effects of dexamethasone (Table V). Of equal importance, neither inhibitor preparation altered basal aminotransferase activity. Given the Stimulation of tyrosine aminotransferase synthesis by microinjected C subunit Confluent H35 cells were fused with human RBC ghosts loaded with BSA (I mg/ml) or partially purified (-SO-SO%) C subunit (1.7 mg/ml) from beef heart. Fusion was achieved with 22% PEG and the recovery period was 16 h. Medium containing one-fourth the usual concentration of methionine was added to all dishes 4 h prior to harvest and 1 p~ dexamethasone was added to some dishes at the same time. Thirty min prior to harvest, 100 pCi of (35S)methionine (1.13 Ci/pmol) was added to all dishes. Cells were harvested for assays of enzyme activity, soluble protein content and incorporation of 36S into hot trichloroacetic acid-insoluble material (total soluble protein) and into tyrosine aminotransferase by immunoprecipitation as described under "Experimental Procedures." Values shown are mean f S. E. of four senarate observations. short half-life of this enzyme (14,15), nonspecific inhibition of protein synthesis would be expected to substantially depress aminotransferase activity. As was the case with the effects of the C subunit, the ability of the inhibitor to block induction by 8-Br-CAMP was dependent on the concentration of protein injected (Fig. 5). Once again, there was no significant affect on basal aminotransferase activity and induction by dexamethasone was virtually unaffected by any concentration of the inhibitor protein. A preparation of the inhibitor protein purified to homogeneity has been tested in one experiment and it also led to virtually complete extinction of the effects of 8-Br-CAMP (data not shown).

Effects of the Regulatory Subunit of Protein Kinase-
Given the results presented above, one would predict that the pure R subunit of CAMP-dependent protein kinase should not affect basal tyrosine aminotransferase activity as indeed proved to be the case (Table VI). On the other hand, the R subunit should be able to inhibit the ability of suboptimal concentrations oE 8-Br-CAMP to cause induction by titrating the intracellular cyclic nucleotide. This also proved to be true as the R subunit, injected at about 40% of the maximally effective concentration of C subunit, caused 76% inhibition of the response to 0.1 mM 8-Br-CAMP. As was also expected, an optimal concentration of the CAMP derivative (2 mM) surmounted the inhibitory effect of the R subunit (data not shown).

DISCUSSION
From the data presented in this report, it is probable that the effects of CAMP on tyrosine aminotransferase synthesis are, in fact, mediated by the catalytic subunit of CAMPdependent protein kinase. This conclusion is supported by several observations: 1) both partially purified and homogeneous C subunit cause time-and concentration-dependent increases in aminotransferase activity; 2) the maximum response to pure C subunit occurs at about 0.5 mg/ml while that to the partially purified subunit (60-80% pure) required somewhat higher concentrations; 3) the partially purified inhibitor protein from beef heart blocked the ability of 8-Br-CAMP to induce the aminotransferase but did not significantly affect either basal enzyme activity or the ability of dexamethasone   to act as an inducer; 4) the pure R subunit inhibited the ability of low concentrations of 8-Br-CAMP to induce the aminotransferase but this inhibition was surmounted by addition of excess 8-Br-CAMP. All of these results are to be expected if the C subunit is a participant in the pathway by which the synthesis of tyrosine aminotransferase is regulated by CAMP.
Although such a conclusion was suggested by extensive indirect results (5,34), a direct test of the role of the C subunit was essential to establish its validity unequivocally. This is  (14), in contrast to its effect on enzymes controlling glycogen or fatty acid metabolism (1, 2). Given that control of selective protein synthesis in Escherichia coli involves interaction of the cAMP binding protein with genetic elements (3, 4), a direct effect of the R subunit in the case of tyrosine aminotransferase was not a trivial possibility. The fact that the C subunit itself induces the aminotransferase, coupled with the inhibition of the effects of low concentrations of 8-Br-CAMP by the R subunit, provides a compelling argument against the cAMP binding protein as an agent of induction. Low levels of 8-Br-CAMP should be titrated by the R subunit injected, but this effect should be surmountable by excess 8-Br-CAMP as was the case. Even if the R subunit injected contained some noncovalently bound CAMP, a possibility which is difficult to eliminate completely, it is likely to have been degraded during the 12-h recovery period. H35 cells contain low levels of cAMP apparently because they are able to degrade the nucleotide at an exceedingly rapid rate (34). Thus, any bound cAMP would have largely disappeared allowing the R subunit to interact with the 8-Br-CAMP which was present inside the H35 cell.
Based on the content of the C subunit determined either by its purification to homogeneity from several t,issues (35), or by comparison of its activity in H35 cell extracts with that in rabbit skeletal muscle (36), one can calculate that the average H35 cell contains approximately 275 X lo3 molecules of endogenous C subunit. Assuming that the efficiency of loading (-4%) and micro-injection (-5%) of BSA is analogous to that of the C subunit, and given that the maximum response to the kinase subunit occurs at about 0.5 mg/ml in the dialysis bag, one can estimate that at least 1 X 10" molecules of the C subunit must be injected on the average into each W35 cell to produce a full re~ponse.~ Since there is substantial variation in the amount of protein loaded into each ghost (see Fig. I

),
the fact that an apparent excess of C subunit (i.e. beyond the endogenous level) is required is not surprising since multiple fusion events would be needed to deliver a maximum amount of the subunit into cells which fused with ghosts loaded with below average amounts of this protein. Furthermore, there is some loss of catalytic activity during loading (average = 34% loss) which would reduce the estimate to -680 X 10" molecules of C subunit/cell. Finally, some degradation of the injected C subunit can be expected to occur during the 12-16-h recovery period (33) which should bring the estimate down still further.
Average microgram of C subunit added per dish for maximal response, 25.6 pg X 0.05, 2280 ng/dish injected. Each dish contains -4 X lo6 cells, -70 fg of C subunit/cell. 6.02 X 10"' molecules C subunit/41,000 g = 1.47 X 10' molecules/fg X 70 fg/cell 1 x 10" molecules C subunit/cell. by guest on March 24, 2020 http://www.jbc.org/ Downloaded from Although these calculations are only approximate, they provide additional assurance that the observed response can be realistically ascribed to the C subunit injected since the amount required is reasonably close to the level of endogenous C subunit. It is known from results reported previously by us (5,34) and Liu (36) that all of the endogenous C subunit must be activated to fully induce the aminotransferase (4).
Similar calculations for the inhibitor protein lead to an estimated 115 X lo3 molecules required to be injected/cell for a maximum response which is less than the amount of C subunit required and even less than the level of endogenous C subunit. However, the estimate of inhibitor injected is less precise since only moderately purified material was used. In addition, smaller proteins are both loaded more efficiently (31) and degraded less rapidly (33) than larger proteins. Thus, it is reasonable to believe that the effective concentration of the inhibitor protein/H35 cell has been substantially underestimated compared to that of the C subunit. It is of interest to note that the response to the injected C subunit (see Fig. 3) does not appear to go beyond that produced by optimal concentrations of 8-Br-CAMP. Assuming the lack of any substance in the C subunit preparations which artificially restricts the response, this result implies that the degree of response is dictated by the level of the putative substrate($. Such a conclusion seems reasonable given that the C subunit must interact with several different proteins. At the present time, there is no indication as to what the nature of the presumed intracellular protein substrate responsible for control of aminotransferase synthesis might be. The precise site at which cAMP regulates the synthesis of tyrosine aminotransferase has not been worked out and remains controversial (37). It is possible that one of the kinase substrates previously identified is involved in this system and exhibits multiple functions similar to phosphorylase kinase (1, 2). Alternatively, it could be an as yet unidentified protein similar to ones identified by two-dimensional gel electrophoresis (38). The possibility exists that a phosphatase inhibitor like inhibitor 1 (12) could be the substrate and, in this case, a constitutive, CAMP-independent kinase would have to tonically phosphorylate the substrate involved. Tyrosine aminotransferase itself has been reported to be a substrate for the CAMPdependent kinase in vitro (39), although there is no evidence that such is the case in vivo. Indeed, all three inducers increase phosphorylation of the aminotransferase in vivo (40) even though steroids and insulin do not activate the CAMP-dependent kinase (41). Thus, there is reason to question any role of phosphorylation of the aminotransferase in regulating its synthesis.
The explanations for the inhibition of protein synthesis and the elevation of aminotransferase activity generated by the process of fusion are not known. Whatever the mechanisms prove to be, after 8-12 h, H35 cells return to normal with respect to overall protein synthesis and inducibility of tyrosine aminotransferase. Attempts to fuse ghosts with monolayer cells using Sendai virus have proved to be less effective than with PEG in our hands, and the virus is also known to be cytotoxic (18,30). In any event, the present method should have widespread applicability to any cultured cell system where the role of protein kinase in vivo is uncertain and needs to be tested directly. The results presented in this report provide, to our knowledge, the fist direct evidence for mediation by the C subunit of an effect of cAMP on protein synthesis in eukaryotic cells.