Compensatory Regulation of RIα Protein Levels in Protein Kinase A Mutant Mice

The cAMP-dependent protein kinase holoenzyme is assembled from regulatory (R) and catalytic (C) subunits that are expressed in tissue-specific patterns. Despite the dispersion of the R and C subunit genes to different chromosomal loci, mechanisms exist that coordinately regulate the intracellular levels of R and C protein such that cAMP-dependent regulation is preserved. We have created null mutations in the RIβ and RIIβ regulatory subunit genes in mice, and find that both result in an increase in the level of RIα protein in tissues that normally express the β isoforms. Examination of RIα mRNA levels and the rates of RIα protein synthesis in wild type and RIIβ mutant mice reveals that the mechanism of this biochemical compensation by RIα does not involve transcriptional or translational control. These in vivo findings are consistent with observations made in cell culture, where we demonstrate that the overexpression of Cα in NIH 3T3 cells results in increased RIα protein without increases in the rate of RIα synthesis or the level of RIα mRNA. Pulse-chase experiments reveal a 4-5-fold increase in the half-life of RIα protein as it becomes incorporated into the holoenzyme. Compensation by RIα stabilization may represent an important biological mechanism that safeguards cells from unregulated catalytic subunit activity.


From the Department of Pharmacology, University of Washington, Seattle, Washington 98195-7750
The cAMP-dependent protein kinase holoenzyme is assembled from regulatory (R) and catalytic (C) subunits that are expressed in tissue-specific patterns. Despite the dispersion of the R and C subunit genes to different chromosomal loci, mechanisms exist that coordinately regulate the intracellular levels of R and C protein such that cAMP-dependent regulation is preserved. We have created null mutations in the RI␤ and RII␤ regulatory subunit genes in mice, and find that both result in an increase in the level of RI␣ protein in tissues that normally express the ␤ isoforms. Examination of RI␣ mRNA levels and the rates of RI␣ protein synthesis in wild type and RII␤ mutant mice reveals that the mechanism of this biochemical compensation by RI␣ does not involve transcriptional or translational control. These in vivo findings are consistent with observations made in cell culture, where we demonstrate that the overexpression of C␣ in NIH 3T3 cells results in increased RI␣ protein without increases in the rate of RI␣ synthesis or the level of RI␣ mRNA. Pulse-chase experiments reveal a 4 -5-fold increase in the half-life of RI␣ protein as it becomes incorporated into the holoenzyme. Compensation by RI␣ stabilization may represent an important biological mechanism that safeguards cells from unregulated catalytic subunit activity.
The cAMP-dependent protein kinase (PKA) 1 is a key regulatory enzyme responsible for the intracellular transduction of a variety of extracellular signals and for the maintenance of numerous aspects of cellular homeostasis (1). The holoenzyme is composed of a regulatory (R) subunit dimer complexed with two catalytic (C) subunits. Two molecules of cAMP bind to each R subunit causing release of enzymatically active C subunits, which then modify the activity of target proteins by reversible phosphorylation of serine or threonine residues located within an appropriate consensus sequence (2).
Four R subunit isoforms and two C subunit isoforms of PKA have been characterized in the mouse (3). They are highly conserved among mammals, encoded by unique genes located on separate chromosomes, and show unique patterns of gene expression. The ␣-isoforms are expressed ubiquitously while ␤ isoforms show more restricted patterns of expression. RI␤ is induced relatively late in development and is highly expressed in neural tissues (4 -6). RII␤ is expressed during embryogenesis in mouse brain, spinal cord, and liver (7). In adult mice RII␤ protein is most abundant in brain and brown and white adipose tissue, with lower expression in testis and ovary (8). C␤ is most abundant in the brain, but lower levels of C␤ mRNA are found in all tissues examined (9).
PKA holoenzymes can be separated by ion-exchange chromatography and analysis of a variety of mammalian tissues has revealed significant differences in the ratio of type I (RI-containing) to type II (RII-containing) holoenzyme (10). In rats and mice, brain and adipose tissue contain principally the type II holoenzyme, while heart and liver contain mainly type I. The ratio of type I to type II holoenzyme in individual tissues also varies across species. While mouse and rat hearts possess mainly the type I holoenzyme, beef and guinea pig hearts have principally the type II holoenzyme, with human and rabbit hearts showing equivalent amounts of both holoenzymes (11).
The type I to type II holoenzyme ratios can also change dramatically during cell development. Differentiation of Friend erythroleukemic cells results in a large increase in total PKA activity and a shift from equimolar amounts of type I and type II holoenzyme to a majority of RII␤-containing holoenzyme (12). A similar selective increase in the RII␤ regulatory subunit occurs in differentiating ovarian follicles treated with estradiol and follicle-stimulating hormone (13). Selective increases in the RI␣ regulatory subunit and the type I holoenzyme occur during the differentiation of L6 myoblasts, which also show increases in total PKA activity (14). A similar phenomenon has been observed during the differentiation of 3T3-L1 cells (15).
Although the ratio of type I to type II holoenzyme varies in different cell types and stages of differentiation, total R and C subunit levels are thought to be equivalent in a variety of tissues (16). How this extremely tight coordination of R and C subunits is achieved in all tissues remains to be determined; however, experiments performed in cell cultures have revealed one potential mechanism (17,18). The ubiquitous RI␣ subunit has been shown to be unstable when not associated with the C subunit in the type I holoenzyme. In Kin Ϫ cells that lack detectable C subunit, RI␣ subunits are rapidly degraded and the steady-state level of RI␣ is reduced (17,19). In contrast, overexpression of the C subunit in NIH 3T3 cells elicits a coordinate increase in RI␣ protein (18).
In this report we show that loss of RI␤ or RII␤ in genedisrupted mice results in biochemical compensation by RI␣ with no change in RI␣ mRNA levels. We demonstrate in cell culture that this compensation is due to a decrease in the turnover rate of RI␣ protein when it associates with the C subunit. The capacity of RI␣ to compensate for changes in C subunit expression provides a mechanism to protect cells from unregulated C subunit activity during developmental and hormonally induced changes in PKA subunits.

EXPERIMENTAL PROCEDURES
Mice-Generation of RI␤ and RII␤ mutant mice has been described (8,20). Both mutant and wild type mice used in the experiments were age-matched and maintained on the same mixed C57BL/6 ϫ 129Sv/J genetic background.
Cell Culture-Wild type mouse NIH 3T3 fibroblasts and C␣3T3 cells (NIH 3T3 cells stably transfected with a plasmid containing the zincinducible metallothionein promoter driving expression of the mouse C␣ subunit) were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS). Exponentially growing cells in 10 cm plates were treated for 24 h with 90 M zinc sulfate in DMEM containing 10% FBS and then harvested as described previously (18).
Western Blot Analysis-Brain and white adipose tissue were isolated from RI␤ and RII␤ mutant and wild type animals, immediately placed in liquid nitrogen, and stored at Ϫ70°C. Samples were thawed into homogenization buffer (250 mM sucrose, 100 mM NaPO 4, pH 7.0, 150 mM NaCl, 1 mM EDTA, 4 mM EGTA, 4 mM dithiothreitol, 0.5% Triton X-100, 2 g/ml leupeptin, 3 g/ml aprotinin, 0.2 mg/ml soybean trypsin inhibitor, 1 mM AEBSF), sonicated, and centrifuged at 16,000 ϫ g, and the supernatant was collected and assayed for protein concentration using a Bradford assay (Bio-Rad). Total protein (40 g) was run on 10% polyacrylamide gels and transferred to nitrocellulose membranes. Blots were then blocked overnight and probed with affinity-purified polyclonal antibodies to RI␣, C␣, or RII␤. Blots were then washed and incubated with horseradish peroxidase-conjugated secondary antibodies and visualized using the Amersham ECL TM system.
Translation Rate Determination-Wild type NIH 3T3 cells and C␣3T3 cells were treated with 90 M zinc sulfate for 24 h. Cells were then washed twice in labeling media (Hanks' balanced salt solution, 5% NaHCO 3, 1% bovine serum albumin, 25 mM Hepes, pH 7.2, 100 units/ml penicillin, 100 g/ml streptomycin) and then incubated for 1 h at 37°C with 200 Ci/ml EXPRE 35 S 35 S protein-labeling mix (Dupont NEN). After 1 h, cells were harvested by washing twice in cold phosphatebuffered saline (20 mM NaPO 4 , pH 7.0, 150 mM NaCl) followed by addition of lysis buffer (250 mM sucrose, 25 mM Tris, pH 7.2, 25 mM NaCl, 5 mM MgCl 2, 1 mM AEBSF, 1% Triton X-100, 1% sodium deoxycholate). Plates were then scraped, transferred to Eppendorf tubes, sonicated, and spun for 1 h at 100,000 ϫ g. Supernatants were recovered and stored at Ϫ70°C. To determine 35 S-incorporation into total protein, 2 l from each sample was spotted onto Whatman GF/C filters, and protein was precipitated in 10% trichloroacetic acid, followed by three washes in 3% trichloroacetic acid/1% sodium pyrophosphate. Filters were then dried and counted in liquid scintillation fluid. Samples containing equivalent total radioactivity were brought to a final volume of 100 l in a lysis buffer containing 100 mM NaCl and 40 M cAMP and incubated for 2.5 h with affinity-purified polyclonal anti-RI␣ antibodies followed by 30 min with 3 l of a 10% suspension of Protein A-Insoluble (Sigma). Reactions were then overlaid on a cushion of lysis buffer containing 1 M sucrose and centrifuged to pellet the immunoprecipitates, which were stored at Ϫ70°C. Pellets were resuspended and run on 10% SDS-polyacrylamide gel electrophoresis (PAGE) gels. Gels were fixed for 30 min in 10% methanol, 5% acetic acid, followed by a 30-min incubation in Amplify TM . Gels were then dried and exposed to XAR TM Kodak film for 24 h. For determination of RI␣ translation rates in adipocytes, wild type and RII␤ mutant mice were sacrificed, and white adipose tissue from uterine fat pads was weighed and immediately minced using fine razor blades, and then placed in scintillation vials containing 1 ml of adipocyte media (100 mM NaCl, 6 mM KCl, 1 mM MgSO 4 , 1 mM NaH 2 PO 4 , 12 mM Hepes, pH 7.2, 2.5 mM CaCl 2 , 1 mg/ml glucose, 1% bovine serum albumin, 33.6 mg/liter NaHCO 3 ) in a 37°C rotating water bath at 40 rpm. After placing all of the fat pads in culture, 1 mg/ml collagenase was added to each vial and incubated for 1 h to dissociate the cells. At the end of 1 h, cells were washed 4 times in 5 volumes of adipocyte media to remove the collagenase and resuspended in 1 ml of adipocyte media containing 200 Ci/ml EXPRE 35 S 35 S protein-labeling mix and placed in a 40 rpm rotating water bath at 37°C. At the end of 1 h, cells were washed 4 times in 5 volumes of adipocyte media without bovine serum albumin, and the pellets were immediately frozen at Ϫ80°C. Immunoprecipitation and analysis of RI␣ protein were performed as described above.
Pulse-chase Experiments-After labeling of NIH 3T3 and C␣3T3 cells for 1 h with 200 Ci/ml EXPRE 35 S 35 S protein-labeling mix, duplicate 10 cm plates were washed twice in DMEM, 10% fetal bovine serum and then incubated in DMEM, 10% fetal bovine serum plus 90 M zinc sulfate containing 4 mM L-methionine. At each time point the cells were harvested and processed as described above. For the immunoprecipitation reactions, equivalent amounts of total protein were loaded rather than equivalent counts. HPLC Analysis and Protein Kinase Activity-HPLC analysis was performed as described (6). Wild type and RII␤ mutant mice were sacrificed, and uterine fat pads were immediately isolated, weighed, and stored at Ϫ70°C. Fat pads were homogenized in buffer (20 mM Tris, pH 7.6, 0.1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 10 mM dithiothreitol, 5 mM magnesium acetate, 250 mM sucrose, 1 g/ml leupeptin, 3 g/ml aprotinin, 100 g/ml soybean trypsin inhibitor, 0.5 mM AEBSF, 100 M ATP) and centrifuged for 30 min at 16,000 ϫ g, and the supernatants were assayed for protein concentration using a Bradford assay (Bio-Rad). Samples diluted with homogenization buffer to a final concentration of 1-2 mg/ml were loaded onto a DEAE/HPLC column and eluted using a linear NaCl gradient from 0 mM to 250 mM. Fractions were collected and assayed for kinase activity in the presence and absence of 5 M cAMP with Leu-Arg-Arg-Ala-Ser-Leu-Gly (Kemptide) as a substrate (21).
Solution Hybridization-The method used for measuring total amounts of RI␣ and C␣ mRNA has been described (9). Briefly, total nucleic acid samples isolated by proteinase K digestion and phenol/ chloroform extraction were incubated with a single-stranded [ 32 P]CTPlabeled RNA probe at 70°C for 16 h. Following hybridization, samples were digested with RNase A and T1, precipitated in 10% trichloroacetic acid, and filtered onto Whatman GF/C filters. The amount of RNaseresistant probe was determined by liquid scintillation counting. RI␣and C␣-specific mRNA in each sample was determined by comparison to a standard curve constructed with known amounts of M13 DNA containing the sense strand of the RI␣ and C␣ cDNAs. The results, calculated as picograms of RNA hybridized per g of total nucleic acid, were converted to molecules/cell by assuming 6 pg of DNA/cell.

Compensatory Increase in RI␣ in Cerebral Cortex and Hippocampus of RI␤ Null Mutant Mice-
We have previously reported that targeted disruption of the neural-specific RI␤ gene in mice results in deficiencies in hippocampal long term potentiation and long term depression (20,22). Western blots using protein extracts from the cerebral cortex and hippocampus of RI␤ mutant mice were compared with age-matched controls to quantitate changes in RI isoforms. This analysis demonstrated a compensatory increase in RI␣ protein in both tissues (Fig. 1), whereas no changes were observed in C or RII isoforms (data not shown). RI␣ protein levels were determined by densitometry of Western blots from wild type and RI␤ mutant protein extracts. Densitometry analysis revealed an approximate 40% increase in RI␣ protein in both the cerebral cortex and hippocampus of RI␤ mutant mice (Table I). In order to address whether the increase in RI␣ protein was due to an elevation in transcription from the RI␣ gene, solution hybridization exper- iments were performed using total nucleic acid isolated from cortex and hippocampus of wild type and RI␤ mutant mice. This analysis revealed no change in RI␣ mRNA levels in mutant tissues ( Table I).
Disruption of RII␤ Leads to Increased Levels of RI␣ in White Adipose Tissue-The RII␤ regulatory subunit is highly expressed in both white adipose tissue (WAT) and brown adipose tissue in mice. A targeted disruption of the RII␤ gene has been created that displays marked alterations in both WAT and brown adipose tissue metabolism (8). In order to address potential compensation by other regulatory subunits in mice carrying a null mutation in the RII␤ gene, Western blots were performed on WAT from wild type and RII␤ mutant mice. RII␤ mutant mice showed a complete loss of the RII␤ protein ( Fig.  2A). Separate Western blots examining the levels of RI␣ and C␣ revealed a 3-4-fold increase in RI␣ protein in RII␤ mutant WAT, while C␣ protein was reduced by approximately 43% (Fig. 2B). RI␣ mRNA levels in total nucleic acid samples from WAT of wild type and RII␤ mutant mice were identical (Table II).
Assembly of Type I Holoenzyme in RII␤ Null Mutants-The large increase in RI␣ protein observed in WAT from RII␤ mutant mice suggests that the RI␣ subunit has replaced RII␤ and formed a type I holoenzyme. HPLC analysis of WAT obtained from wild type mice revealed that the majority of PKA activity was associated with the type II holoenzyme together with a small free C subunit peak (Fig. 3). In contrast, WAT from RII␤ mutant mice contained only type I holoenzyme. Western blots using protein from HPLC fractions containing the type I holoenzyme peak confirmed the presence of RI␣ and C␣ in these fractions (data not shown). Peak activity fractions were also assayed in the presence of the heat-stable PKA inhibitor, PKI, which confirmed that all the kinase activity was PKA-dependent.
The Rate of Translation of RI␣ Protein-In order to address the mechanism of RI␣ compensation in RII␤ mutant mice, pulse-labeling experiments were performed in primary cultures of white adipocytes from wild type and RII␤ mutant mice. No significant difference was observed in the rate of translation of RI␣ protein between wild type and RII␤ mutant mice after a 1-h pulse (Fig. 4B). Western blots from the same extracts used to perform the pulse-labeling experiments confirmed that RI␣ protein was substantially increased in RII␤ mutant mice (Fig.  4A). This implies that the increased RI␣ protein must be due to stabilization of the protein.
Altered RI␣ Stability in a Cell Culture Model of RI␣ Compensation-Loss of either RI␤ or RII␤ would result in an excess of C subunit over R unless a compensatory mechanism exists to maintain the R/C balance. The observed increase in RI␣ appears to be a response to this imbalance. We used a cell line stably transfected with the C subunit to characterize further the mechanism of RI␣ compensation under conditions where the C subunit is expressed in excess of R. We have previously shown that when C is overexpressed in NIH 3T3 cells there is a specific increase in RI␣ with no change in RII subunits (18). This increase in RI␣ resulted in the appearance of new type I holoenzyme, analogous to the results shown in Fig. 3 for RII␤ mutant WAT (18,23). We therefore used these C␣-overexpressing 3T3 cells for metabolic labeling studies to determine the mechanism of RI␣ compensation. Wild type NIH 3T3 fibroblasts and C␣3T3 cells stably expressing a zinc-inducible expression vector for the mouse C␣ catalytic subunit (18) were treated with 90 M zinc sulfate for 24 h and subsequently analyzed by Western blotting for RI␣ and C␣. Zinc treatment of C␣3T3 fibroblasts resulted in a 27-fold increase in C␣ protein compared with values obtained from wild type 3T3 cells (Table  III). As previously observed (18), an elevation in RI␣ protein was also seen upon overexpression of C␣ (Fig. 5A). Western blot analysis of a range of protein dilutions from C␣3T3 cells and wild type 3T3 cells showed a 4-fold increase in RI␣ protein (Table III).
Solution hybridization experiments demonstrated that mRNA levels for RI␣ remained constant despite the elevation in RI␣ protein (Table III). The increase in RI␣ protein could be due to either an elevated rate of translation or a stabilization of RI␣ protein. In order to determine the mechanism, zinc-treated NIH 3T3 and C␣3T3 cells were labeled for 1 h with [ 35 S]methionine followed by immunoprecipitation of RI␣ protein. The rate of synthesis of RI␣ protein in wild type NIH 3T3 and C␣3T3 cells was equivalent (Fig. 5B). Western blot analysis from the same extracts confirmed that C subunit protein was indeed overexpressed in the zinc-treated C␣3T3 cells as ex-  , n ϭ 3) and RII␤ mutant (Ϫ/Ϫ, n ϭ 3) WAT using an antibody to RII␤. B, Western blot comparing wild type (ϩ/ϩ, n ϭ 4) and RII␤ mutant (Ϫ/Ϫ, n ϭ 4) WAT using antibodies to RI␣ and C␣. 40 g of total protein from WAT homogenates were run in each lane. pected (data not shown). Pulse-chase experiments were performed to determine the half-life of RI␣ protein in wild type NIH 3T3 cells and C␣3T3 cells. The half-life of RI␣ in wild type NIH 3T3 cells was approximately 3.5 h as measured by immunoprecipitation of 35 S-labeled RI␣ protein from cell extracts after a cold chase with unlabeled methionine. In contrast, the half-life of RI␣ protein in zinc-treated C␣3T3 cells was 13.5 h (Fig. 5C). This represents a 4-fold increase in the half-life of the RI␣ protein upon overexpression of C␣ and is in good agreement with the 4-fold increase in RI␣ protein observed in this experiment. DISCUSSION The ability of mammalian cells to assemble and regulate multisubunit protein complexes usually relies on some type of autoregulatory loop. Whereas bacteria frequently solve the problem of coordinate regulation by utilizing multigene operons controlled by a single promoter, in higher animals the genes encoding interacting subunits are dispersed, frequently onto different chromosomes. Nevertheless, expression from these genes generally results in stoichiometric levels of protein subunits. The problem of coordinate regulation becomes crucial when an excess of one of the subunits might lead to undesirable biological effects in the cell. The PKA holoenzyme system is an example of such a case in which an excess of catalytic subunit could result in unwanted biological effects and reduce the ability of the cell to regulate activity by cAMP. Four separate regulatory and two catalytic subunits are produced, sometimes within the same cell, and yet most tissues are able to maintain an equimolar ratio of R and C subunits (16).
In this report we have artificially perturbed the expression of RI␤ and RII␤ subunits using targeted gene disruption in mice and examined the compensatory mechanisms that regulate R/C subunit balance in neurons and adipose tissue. In neurons of RI␤ mutant mice, levels of RI␣ increase and at least partially replace the lost RI␤ subunit. In adipose tissue from RII␤ mutant mice, we find a dramatic compensation by RI␣ and only a modest loss in total C subunit. In both cases, the increase in RI␣ protein is due to stabilization by incorporation into holoenzyme. Since it was not possible to quantitate the changes in RI␣ half-life by pulse-chase experiments in whole animals, we have used a cell culture model system in which the overexpression of exogenous C subunit elicits an increase in RI␣ very similar in magnitude to that observed in WAT from RII␤ mutants. Pulsechase experiments in the cell culture system demonstrate a 4-fold increase in RI␣ half-life when it is incorporated into holoenzyme and stabilized by interaction with C subunit.
Previous studies have shown that the R and C subunits are stabilized against proteolysis when assembled as a holoenzyme. Stabilization of RI␣ through binding to the C subunit has been demonstrated in S49 mouse lymphoma cells (17). Kin Ϫ cells, which lack detectable C subunit, show a 10-fold increase in the turnover rate of RI␣ protein and a significant decrease in steady-state RI␣ levels when compared with wild type S49 cells (17,19). However, when wild type S49 cells are treated with agents that raise cAMP and separate the R and C subunits, the RI␣ protein is destabilized to the same extent observed in Kin Ϫ cells. The C subunit is also exposed to degradative pathways when released from the holoenzyme complex. Chronic activation of LLC-PK cells with cAMP can lead to the loss of more

FIG. 4. Pulse-labeling analysis of RI␣ synthesis in adipocytes.
Adipocytes from wild type (ϩ/ϩ, n ϭ 2) and RII␤ mutant (Ϫ/Ϫ, n ϭ 2) WAT were isolated and pulse-labeled for 1 h as described under "Experimental Procedures." A, Western blot analysis of the cell homogenates used for immunoprecipitation of RI␣ in panel B. B, each cell pellet was homogenized, and samples containing equivalent total trichloroacetic acid-precipitable counts were used to immunoprecipitate RI␣ protein with a polyclonal affinity-purified RI␣ antibody. Immunoprecipitates were run on SDS-PAGE gels and analyzed by autoradiography to assess the level of newly synthesized RI␣.  3. HPLC profile of PKA from wild type and RII␤ null mutant WAT. 2 mg of total protein from WAT homogenates from wild type (top) and RII␤ mutant (bottom ) mice was resolved by HPLC/ionexchange chromatography, and proteins were eluted with a linear salt gradient. Individual fractions were assayed for PKA activity using Kemptide as the substrate (closed circles). Fractions containing peak kinase activity were also assayed in the presence of 5 M PKI peptide to demonstrate that the kinase activity was PKA-specific (open circles). Both panels show HPLC profiles from one wild type and one RII␤ mutant mouse and are representative of three independent experiments run on different mice all with similar results. than 75% of the cell's complement of C subunit within 12.5 h (24).
What are the rules governing the assembly of type I and type II holoenzymes in vivo? Experiments in cell culture have shown that C subunits preferentially assemble with RII subunits rather than RI subunits (25,26). NIH 3T3 cells and wild type WAT express both RI and RII subunits. However, when holoenzymes are separated by ion-exchange chromatography only the type II holoenzyme is observed (8,23). When NIH 3T3 cells are programmed to overexpress exogenous C subunit, the formation of new type I holoenzyme occurs (18,23), suggesting that there is an ordered assembly of first type II and then type I holoenzyme. In contrast, overexpression of RI subunits does not alter the amount of type II holoenzyme nor does it result in increased formation of type I holoenzyme. This suggests that the total amount of free C subunit is rate-limiting with respect to formation of first type II and then type I holoenzyme (6,25).
Emerging from these studies is an appreciation of the cell's capacity to maintain cAMP-mediated control of C subunit activity and the important role played in this process by RI␣. A simple model describing the dynamic assembly of R and C subunits is depicted in Fig. 6 using the example of WAT from wild type and RII␤ mutant mice. In adipocytes, the RII␤ subunits preferentially associate with C, leaving a pool of free RI␣ that is rapidly degraded. Type I holoenzyme is only formed when the level of C subunits exceeds the level of RII subunits (in this case caused by the loss of RII␤). In this situation RI␣ can successfully compete for binding to the pool of free C subunits and is therefore stabilized in a holoenzyme complex. Preferential binding of RII subunits to C probably does not arise because of intrinsic differences between RI and RII subunits in their affinity for C, as these affinities have been shown to be quite similar (27). We propose that the phenomenon occurs as a result of a lower K a for cAMP-dependent activation of the RI holoenzyme compared with the RII holoenzyme. Free RI subunits have been shown by numerous investigators to have a higher affinity for cAMP than do RII subunits. Published values for the K d of RI-cAMP binding range from 0.1 (28) to 1 nM (29). In contrast, higher K d values for RII-cAMP binding are consistently reported, ranging from 4 (30) to 6 nM (31). We have shown that the apparent K a for cAMP activation of RI␣ holoenzyme is about 4-fold lower than that for RII␤ holoenzyme when measured in cell extracts (8). Given the enhanced sensitivity to activation of RI-containing holoenzyme, we predict that C subunits would shift preferentially to the RII-containing holoenzyme complex until the RII binding capacity of the cell is FIG. 6. Model for RI␣ compensation in RII␤ mutant mice. In wild type WAT, RI␣ protein is synthesized but is unable to compete with the RII␤ subunit for catalytic subunits and is thus rapidly degraded with a half-life of approximately 3 h. In RII␤ mutant WAT, the absence of the RII␤ protein results in a large increase in free catalytic subunits that associate with RI␣ to form new type I holoenzyme. The RI␣ protein is stabilized approximately 4-fold with a half-life of approximately 14 h. The loss in total catalytic subunit results from the lower affinity interaction between RI␣ and C␣ at physiological concentrations of cAMP resulting in increased degradation of the free catalytic subunit. Duplicate plates were treated with zinc sulfate and then labeled with 35 S as described under "Experimental Procedures." Lysates containing equivalent total trichloroacetic acid-precipitable counts were incubated with a polyclonal affinity-purified RI␣ antibody and immunoprecipitated with Protein A-Insoluble. Immunoprecipitated RI␣ was run on SDS-PAGE and analyzed by autoradiography. Control lysates were incubated with a polyclonal affinity-purified conalbumin antibody. C, pulse-chase analysis of RI␣ in wild type (left) and C␣3T3 (right) cells. Five sets of duplicate 10-cm plates of wild type and C␣3T3 cells were treated with zinc sulfate and labeled as above. Plates were then chased for a total of 24 h in DMEM plus 10% FBS containing 4 mM L-methionine and 90 M zinc sulfate. Lysates were made at 0, 3, 6, 12, and 24 h, and RI␣ was immunoprecipitated and analyzed as above with the exception that lysates containing equivalent amounts of total protein rather than equivalent trichloroacetic acid-precipitable counts were used for the immunoprecipitations.

saturated.
When the concentration of free C subunit increases due to the loss of RI␤ or RII␤, RI␣ rapidly responds to this perturbation via protein stabilization in a holoenzyme complex, thus protecting the cell from unregulated C subunit activity and rescuing the C subunit from rapid proteolysis. This biochemical adaptation provides a very effective mechanism for regulating the ratio of type II to type I holoenzyme formed in a given tissue and for maintaining regulation when C subunit levels change.
Modulation of RI␣ turnover rate may represent an important biological mechanism for maintaining equivalent amounts of R and C subunits. Loss of this ability to maintain cAMP-dependent regulation of C subunit activity during the process of cellular differentiation could have catastrophic consequences, a phenomenon that we have recently observed in mutant mice lacking RI␣ altogether. 2 RI␣ null mutants display early embryonic lethality with severe developmental abnormalities.