A mutation affecting the catalytic subunit of cyclic AMP-dependent protein kinase in CHO cells.

A mutant of Chinese hamster ovary (CHO) cells has been selected which is unresponsive to the growth inhibitory effects of cholera toxin or other agents which elevate intracellular cyclic AMP levels. Crude extracts of the mutant CHO cells exhibit altered cyclic AMPdependent protein kinase (ATP protein phosphotransferase; EC 2.7.1.37) activity when compared with the parent. With histone as substrate, the cyclic AMP-dependent protein kinase of mutant cells exhibits an approximate K, for activation by cyclic AMP loo-fold greater than that of the enzyme from wild type cells, while cyclic AMP binding activity is not appreciably altered. The DE52-cellulose elution profile of the mutant protein kinase activity is markedly different from that observed for the wild type. As determined by DE52-cellulose chromatography, the mutant has more type I protein kinase activity which is stimulated only by high levels of cyclic AMP, whereas type II kinase activity is very low or nonexistent. The approximate K, for cyclic AMP activation of mutant type I protein kinase (K, -1 PM) is significantly altered relative to the parent type I protein kinase (Ka 2 nM). The basis of these changes appears to be in the catalytic subunit since the mutant type I protein kinase and the mutantfree catalytic subunit show a decreased affinity for the ATP substrate when compared to the wild type. Further, the mutant-free catalytic subunit has a different protein substrate specificity from the wild type. We conclude that the mutant cells have an altered catalytic subunit of cyclic AMP-dependent protein kinase. and flattened. It also slows cell growth (3,4), decreases agglutinability by plant lectins (5, 6), increases collagen synthesis (7), and in CHO! cells induces hamster endogenous virus (8). When CHO cells are exposed to cholera toxin, whic’h stimulates adenylate cyclase and, thereby, increases cyclic AMP levels, the same cell shape change and growth inhibition occurs as previously reported with cyclic AMP treatment (9). This increase in cyclic AMP level is correlated with activation of cyclic AMP-dependent protein kinase (10, 1 l), suggesting that the activity of cyclic AMP may be mediated by phosphorylation of cellular components. In this paper, we report the biochemical characterization of a CHO mutant which is resistant to the growth inhibitory effects of cholera toxin. This is one of a series of independent mutants isolated by a variety of means, which do not show any growth inhibition or morphological change when exposed to 8-bromo-cyclic AMP and are relatively resistant to a variety of other agents which raise intracellular cyclic AMP levels.’ In order to elucidate the defect in the mutant, cyclic AMPdependent protein kinase in wild type and mutant cells have been compared. We find the mutant CHO cells have a defective cyclic AMP-dependent protein kinase and that the defect resides in the catalytic subunit of this enzyme. The existence of this mutant provides strong direct evidence that the growth inhibitory and morphological effects of cyclic AMP in CHO cells are mediated by cyclic AMP-dependent protein kinase.

A mutant of Chinese hamster ovary (CHO) cells has been selected which is unresponsive to the growth inhibitory effects of cholera toxin or other agents which elevate intracellular cyclic AMP levels. Crude extracts of the mutant CHO cells exhibit altered cyclic AMPdependent protein kinase (ATP protein phosphotransferase; EC 2.7.1.37) activity when compared with the parent.
With histone as substrate, the cyclic AMP-dependent protein kinase of mutant cells exhibits an approximate K, for activation by cyclic AMP loo-fold greater than that of the enzyme from wild type cells, while cyclic AMP binding activity is not appreciably altered.
The DE52-cellulose elution profile of the mutant protein kinase activity is markedly different from that observed for the wild type. As determined by DE52-cellulose chromatography, the mutant has more type I protein kinase activity which is stimulated only by high levels of cyclic AMP, whereas type II kinase activity is very low or nonexistent.
The approximate K, for cyclic AMP activation of mutant type I protein kinase (K, -1 PM) is significantly altered relative to the parent type I protein kinase (Ka -2 nM). The basis of these changes appears to be in the catalytic subunit since the mutant type I protein kinase and the mutantfree catalytic subunit show a decreased affinity for the ATP substrate when compared to the wild type. Further, the mutant-free catalytic subunit has a different protein substrate specificity from the wild type. We conclude that the mutant cells have an altered catalytic subunit of cyclic AMP-dependent protein kinase.
When CHO cells are exposed to cholera toxin, whic'h stimulates adenylate cyclase and, thereby, increases cyclic AMP levels, the same cell shape change and growth inhibition occurs as previously reported with cyclic AMP treatment (9). This increase in cyclic AMP level is correlated with activation of cyclic AMP-dependent protein kinase (10, 1 l), suggesting that the activity of cyclic AMP may be mediated by phosphorylation of cellular components.
In this paper, we report the biochemical characterization of a CHO mutant which is resistant to the growth inhibitory effects of cholera toxin. This is one of a series of independent mutants isolated by a variety of means, which do not show any growth inhibition or morphological change when exposed to 8-bromo-cyclic AMP and are relatively resistant to a variety of other agents which raise intracellular cyclic AMP levels.' In order to elucidate the defect in the mutant, cyclic AMPdependent protein kinase in wild type and mutant cells have been compared. We find the mutant CHO cells have a defective cyclic AMP-dependent protein kinase and that the defect resides in the catalytic subunit of this enzyme. The existence of this mutant provides strong direct evidence that the growth inhibitory and morphological effects of cyclic AMP in CHO cells are mediated by cyclic AMP-dependent protein kinase.

MATERIALS AND METHODS
The molecular basis for the action of cyclic AMP in mammalian tissues is not completely understood. Kuo and Greengard (1) and others have suggested that all of the effects of cyclic AMP are mediated by cyclic AMP-dependent protein kinases, but direct effects of cyclic AMP on transcription as occur in prokaryotes (for review see Ref. 2) and other still undefined mechanisms have not been excluded. We have begun a genetic analysis of somatic cell mutants unable to respond to cyclic AMP in an effort to understand the mechanism of cyclic AMP action in cultured mammalian cells. Many aspects of the behavior of cultured Chinese hamster ovary cells (CHO) and other fibroblastic cells are controlled by the intracellular concentration of cyclic AMP. This nucleotide regulates cell shape, causing cells to become elongated Isolation of Cholera Toxin-resistant Mutants-The CHO cell line 10001 was a subclone of a CHO Pro-line kindly sent to us bv Dr. L. Siminovitch, and originally derived by Puck et al. (12). After ethylmethanesulfonate mutagenesis in supension at 37°C (150 pg/ml for 7 h), cells were allowed to grow for 2 weeks under nonselective conditions. At this time, 2.0 x 10' cells were washed and suspended at 2 x 10" cells/ml in serum-free medium containing 1 pg/ml of cholera toxin (Schwarz/Mann).
After 3 h, the medium was made 10% in fetal bovine serum and cells were incubated in suspension for 5 days. Cholera toxin slows CHO cell growth but does not quantitatively kill cells. Growing cells were washed and retreated with cholera toxin as described above for 5 more days. At this time, cells were plated on 150-mm dishes in 100 rig/ml of cholera toxin and clones were grown and scored for size and morphology.
Strain 10215  results in pronounced activation of the adenylate cyclase activity of both cell types. It is readily apparent that the mutant adenylate cyclase activity is significantly greater than that of the parent at all cholera toxin concentrations used. The intracellular level of cyclic AMP of the mutant is also markedly elevated over that of the parent both in the basal state and after stimulation by cholera toxin (Fig. 1B). Since the mutant CHO cells do not respond to cholera toxin with either a change in shape or inhibition of growth (Table I), these data are consistent with a block in CAMP action at or beyond the level of protein kinase.

Protein
Kinase and Cyclic AMP Binding Activity in Crude Extracts-When the kinase activity of the 50,000 X g supernatant is measured by phosphorylation of histone, the activation of kinase by cyclic AMP is markedly altered in the mutant (Fig. 2). Maximal activation occurs with 1 X 10e7 M cyclic AMP for the parent and 1 x 1O-4 M cyclic AMP for the mutant with apparent K, values of 2 X lo-' M and 1 X lOA" M, respectively.
However, as shown in Fig. 3 the cyclic AMP binding curves determined with crude 50,000 X g supernatant are similar in the two cell types.
These results suggest that in the mutant cyclic AMP binds normally to the regulatory subunit of the protein kinase but that the dissociation of the regulatory and catalytic subunits is abnormal.
To characterize the defect, the cyclic AMPdependent protein kinases were partially purified by DE52cellulose chromatography.  supernatant prepared by 50,000 x g centrifugation of homogenates for 15 min. Cyclic AMP binding was assayed as described under "Materials and Methods" with 150 to 300 pg of supernatant protein/tube. dependent protein kinase from wild type CHO cells after DEAE-cellulose chromatography. Two cyclic AMP-dependent protein kinases are eluted from the column with increasing ionic strength as expected from the data of others (20, 21). Type I cyclic AMP-dependent protein kinase elutes between 0.06 and 0.09 M NaCl and type II cyclic AMP-dependent protein kinase elutes between 0.125 and 0.15 M NaCl. The addition of 1 pM cyclic AMP to the assay mix stimulated type I and type II protein kinase activities approximately 2-fold. Both kinase peaks show specific cyclic ["HIAMP binding (Fig.  4B). The cyclic AMP-independent protein kinase activity observed in the fractions prior to the gradient is apparently due to free catalytic subunit since this activity is completely inhibited by the protein kinase inhibitor protein (data not shown).
The elution pattern of the mutant kinase activity is markedly different from that observed for the wild type (Fig. 5). The first peak of kinase activity (type I) elutes at about 0.075 M NaCl. This activity is stimulated approximately 3-fold by 10 PM cyclic AMP. In the region where type II protein kinase was expected only a very small amount of cyclic AMP-independent protein kinase was detected. Two distinct peaks of cyclic AMP binding were observed with Kd values similar to that of type I and type II cyclic AMP-dependent protein kinases from the wild type (Kd = 50 nM).
Characterization ofDE52-fractionated Protein KinaseActiuities-Type I protein kinase of the mutant is dissociated by incubation with histone (data not shown). This property is similar to that of the type I protein kinase of the wild type type I protein kinases (Panel A) and free catalytic subunits (Panel B). A, the protein kinase (type I) preparations were obtained by DE52-cellulose chromatography as shown in Figs. 4 and 5 and protein kinase activity was determined in the presence of 10 pM cyclic AMP with the indicated concentration of ATP. B, catalytic subunits were isolated as follows. Cells (6 X 10") were homogenized in Buffer I and centrifuged, and the 50,000 x g supernatant was incubated for 1 h at 4°C with 10 pM cyclic AMP to obtain binding equilibrium and then for 4 min at 30°C to facilitate the dissociation of the catalytic subunit from the cyclic AMP regulatory subunit complex. The dissociated preparation was applied to a DE52 column (0.6 x 13 cm). The free catalytic subunit was eluted with 10 ml of Buffer I and assayed for phosphorylation of histone with the indicated concentration of ATP. results depicted in Fig. 6A show that the mutant type I protein kinase exhibits a markedly altered ATP concentration curve as compared to the type I protein kinase of the wild type. This is apparently due to a decreased affinity for the ATP substrate.
Characterization of the Free Catalytic Subunit-The abnormal affinity for ATP, as noted with crude extracts and with DE52-purified type I cyclic AMP-dependent protein kinase, suggests that the catalytic subunit of the mutant is defective. To test this hypothesis the catalytic subunit from both wild type and mutant cells was obtained as described under "Materials and Methods." The catalytic subunit isolated from the mutant shows the same alteration in its ATP dependence as noted with type I protein kinase (Fig. 6B). One possible explanation for this finding is that the altered ATP requirement results from increased ATPase activity present in the mutant preparation.
Studies carried out to assess this possibility show that neither catalytic subunit preparation significantly hydrolyzed 10 pM [y-'"PIATP." The mutant free catalytic subunit also exhibits an altered protein substrate specificity when compared to the wild type free catalytic subunits. When assayed with 0.5 mM [y-"'P]ATP mutant and wild type free catalytic subunits catalyze the phosphorylation of histone fzb to the same extent (Fig. 7A). However, as shown in Fig. 7B  mutant do not differ in the optimal pH (6.5) or the optimal temperature (37°C) of the phosphotransferase reaction with histone IIA as substrate. Both catalytic subunits are completely inhibited by the protein kinase inhibitor protein (Sigma Chemical) (data not shown).
Sucrose Density Gradient Sedimentation of DE52-fractionated Protein Kinase Actiuities-To determine whether any catalytic subunit was bound to the R subunit in the mutant, type I and type II cyclic AMP-dependent protein kinase activities isolated by elution from DE52 columns were sedimented through a 5 to 20% W/V sucrose gradient. As shown in Fig. 8A type I protein kinase of the wild type sediments as a peak of holoenzyme showing cyclic AMP-dependent protein kinase activity and cyclic AMP binding activity with a sedimentation coefficient (s2oJ of 6. The peak of cyclic AMPindependent kinase activity with an ~20,~~ of 3.2 corresponds to the free catalytic subunit apparently dissociated during preparation. Type II protein kinase of the wild type sediments as a peak of holoenzyme with an sZo+ of 6.9. A second peak of cyclic AMP binding activity with an spg+, of 5 corresponds to the presence of regulatory subunit not bound to catalytic subunit (Fig. 8B). When type I and type II protein kinases of the wild type are dissociated by incubation for 2 h with 10 pM cyclic AMP, single peaks of protein kinase activity from type I and type II were observed corresponding to free catalytic subunit with an szo,w of 3.2 (Fig. 8, C and D).
Type I cyclic AMP-dependent protein kinase from the mutant sediments essentially as a single peak with cyclic AMP-dependent protein kinase activity and cyclic AMP binding activity with an sZO,,,, of 6 which is similar to that of wild A, PKI initially obtained by elution from a DE52 column (pool of Fractions 36 to 41) was applied to the gradient and sedimented as described. B, PKII, which was isolated by elution from a DE52 column (pool of Fractions 49 to 60), was applied to the gradient and centrifuged as described. C, PKI was incubated for 2 h with 10 pM cyclic AMP and then sedimented through a sucrose gradient containing 10 pM cyclic AMP. D, PKII was dissociated by incubation for 2 h with 10 pM cyclic AMP and then centrifuged through a sucrose gradient containing 10 PM cyclic AMP. type type I protein kinase (Fig. 9A). The binding activity from the region of the DEAE column corresponding to the type II protein kinase sediments with an s~O,~, of 5 which is the same as that of regulatory subunit not bound to catalytic subunit (Fig. 9B). There is only a very small peak of holoenzyme with cyclic AMP-dependent protein kinase activity and an SZO,~, of 6.9. We conclude that the cyclic AMP binding activity eluted from DE52 in the mutant type II region is due in large measure to the presence of free regulatory subunit and not to an inactive holoenzyme of type II protein kinase which binds cyclic AMP but cannot be activated.
Mutant type I protein kinase was dissociated by incubation with 10 PM cyclic AMP and this preparation sedimented through a sucrose gradient also containing 10 pM cyclic AMP (Fig. 9C). A broad peak of protein kinase activity was observed. This peak has a sedimentation coefficient of s20,ur of 5. As shown in Fig. 9D when mutant type II protein kinase was dissociated with 10 pM cyclic AMP and sedimented through a cyclic AMP containing gradient in a similar manner, a small The altered sedimentation coefficient noted for the catalytic moiety of mutant type I protein kinase (Fig. 9C) suggested that either the mutant catalytic subunits exist as a dimer or that the mutant type I cyclic AMP-dependent protein kinase may not be completely dissociated by cyclic AMP. When free catalytic subunits isolated from either wild type or mutant type I protein kinase were sedimented through a similar 5 to 20% w/v sucrose gradient, both exhibited an SZ,,,~~ of 4 ( Fig.  10). This result indicates that the altered sZO,+ of the protein kinase activity presented in Fig. 9C is due to the incomplete dissociation of mutant type I cyclic AMP-dependent protein kinase by cyclic AMP.

DISCUSSION
In animal cells increasing cyclic AMP levels by stimulation of adenylate cyclase is thought to activate cyclic AMP-dependent protein kinase, which in turn phosphorylates specific substrates and mediates the biological activity of the stimulating agent (24). Selection of mutants which are defective at various steps in cyclic AMP action provides a powerful tool to evaluate critically the mechanism by which this nucleotide regulates cell behavior.
Cyclic AMP-dependent protein kinase is normally activated by binding of cyclic AMP to a regulatory subunit. This results in dissociation of the regulatory and catalytic subunits (25,26) and activates the catalytic subunit. It has been suggested for the protein kinases of a number of mammalian tissues that the catalytic subunits are identical but the regulatory subunits differ (27).
Protein kinase mutants have been isolated in mouse lymphoma cells (28,29) and in neuroblastoma cells (30) on the basis of their resistance to killing by cyclic AMP. Most of these mutants have abnormal regulatory subunits which cannot bind cyclic AMP normally and, hence, the catalytic subunits cannot be activated. Borman et al. (31) have reported the isolation of a CHO variant which does not respond morphologically to dibutyryl cyclic AMP but the biochemical nature of the defect in this cell line has not been reported.
We have recently described the isolation of 11 independent mutants of CHO cells resistant to the growth inhibitory effect of CAMP." In this work, we describe the detailed characterization of one of these mutants (10215) selected on the basis of resistance to cholera toxin. The growth and morphology phenotype of this mutant is reviewed in Table I.
Crude extracts of the mutant cells have normal cyclic AMP binding activity (K,,~-8 x lo-' M) (Table II) but decreased cyclic AMP-dependent protein kinase activity. The mutant extracts require about loo-fold higher cyclic AMP concentrations for activation than wild type extracts. When the mutant protein kinase was chromatographed on DEAE-cellulose, normal cyclic AMP binding activity was observed in the type I and type II regions, suggesting that both regulatory subunits are present and normal. However, mutant type I protein kinase had a loo-fold increase in the apparent K, for cyclic AMP activation of protein kinase despite normal cyclic AMP binding (Table II). Furthermore, very little kinase activity was found in the type II region despite normal cyclic AMP binding ( Fig. 5 and Table II). Sucrose density gradient studies showed that the cyclic AMP binding noted in the PKII region of the mutant was accounted for almost entirely by free regulatory subunit (Fig. 8). The normal cyclic AMP binding, the altered K, of type I protein kinase and the absence of type II protein kinase all suggested a defect in the catalytic subunit. To investigate this hypothesis, the mutant catalytic subunit was dissociated from RI by a high concentration of cyclic  AMP, and isolated free of RI. The mutant catalytic subunit was found to have an altered affinity for ATP (Fig. 6B) and an altered protein substrate specificity (Fig. 7). The substrate specificity results show that the mutant catalytic subunit is as effective as that of the wild type in phosphorylating the basic proteins histone IIA and fZb. However, the acidic protein phosvitin is not phosphorylated by the mutant catalytic subunit. Casein is a mixture of both acidic and basic proteins, and the mutant subunit is about 40% as effective as the wild type in catalyzing the phosphotransferase reaction with this substrate. Since the apparent affinity of the mutant catalytic subunit is not altered, it is conceivable that certain proteins within the mixture are phosphorylated normally, while other, possibly acidic, proteins within the casein mixture are not phosphorylated at all. These results indicate that the mutant catalytic subunit may be defective in its ability to phosphorylate certain endogenous substrates even under conditions where it might be dissociated from the regulatory subunit R,.
Taken together these data indicate that the catalytic subunit of mutant protein kinases is abnormal. We propose that in the mutant, this altered catalytic subunit binds more tightly than normal to the regulatory subunit RI, but can be dissociated by high cyclic AMP concentrations.
In this model, the failure to find type II cyclic AMP-dependent protein kinase could be due either to a reduced affinity of R, for the mutant catalytic subunit or to the failure of RP to compete with R, for of cyclic AMP mediated effect seen in Mutants in which one gene is altered will have a recognizable phenotype only when the mutant product is dominant.
In the case of our protein kinase mutant in which an abnormal catalytic subunit binds tightly to the RI regulatory subunit, a wild type catalytic subunit which is dissociable from the regulatory subunits by cyclic AMP would tend to be excluded from the complex between regulatory and catalytic subunits.
In support of this hypothesis, we have found that hybrids between mutant and wild type cells, in which there is reason to believe that there is an ample amount of wild type subunit, express the mutant phenotype indicating that cyclic AMP resistance is dominant.'