Protein kinase activity and endogenous phosphorylation in subfractions of rat liver mitochondria

Rat liver mitochondria were subfractionated into outer membrane, intermembrane and mitoplast (inner membrane and matrix) fractions. Of the recovered protein kinase activity, 80–90% was found in the intermembrane fraction, while the rest was associated with mitoplast. The intermembrane prostimulated kinase was stimulated by cyclic AMP, while the mitoplast enzyme was stimulated by the nucleotide only after treatment with Triton X-100. Extracted protein kinase resolved into three peaks on DEAE-cellulose chromatography. All three peaks were present both in the intermembrane fraction and in mitoplast. One peak corresponded to the catalytic subunit of cyclic AMP-dependent protein kinase, one was a cyclic AMP-independent enzyme, and the third was the cyclic AMP-dependent type II enzyme. The endogenous incorporation of phosphate was particularly high in the outer mitochondrial membrane, and occurred also in the mitoplast fraction. The incorporation in mitoplasts was to a double band of Mr 47 500, and in outer membranes to apparently heterogeneous material of comparatively low molecular weight.     

Protein kinase activities in rat pancreatic islets of Langerhans.

1. Protein kinase activities in homogenates of rat islets of Langerhans were studied. 2. On incubation of homogenates with [gamma-32P]ATP, incorporation of 32P into protein occurred: this phosphorylation was neither increased by cyclic AMP nor decreased by the cyclic AMP-dependent protein kinase inhibitor described by Ashby & Walsh [(1972) J. Biol. Chem. 247, 6637--6642]. 3. On incubation of homogenates with [gamma-32P]ATP and histone as exogenous substrate for phosphorylation, incorporation of 32P into protein was stimulated by cyclic AMP (approx. 2.5-fold) and was inhibited by the cyclic AMP-dependent protein kinase inhibitor. In contrast, when casein was used as exogenous substrate, incorporation of 32P into protein was not stimulated by cyclic AMP, nor was it inhibited by the cyclic AMP-dependent protein kinase inhibitor. 4. DEAE-cellulose ion-exchange chromatography resolved four peaks of protein kinase activity. One species was the free catalytic subunit of cyclic AMP-dependent protein kinase, two species corresponded to 'Type I' and 'Type II' cyclic AMP-dependent protein kinase holoenzymes [see Corbin, Keely & Park (1975) J. Biol. Chem. 250, 218--225], and the fourth species was a cyclic AMP-independent protein kinase. 5. Determination of physical and kinetic properties of the protein kinases showed that the properties of the cyclic AMP-dependent activities were similar to those described in other tissues and were clearly distinct from those of the cyclic AMP-independent protein kinase. 6. The cyclic AMP-independent protein kinase had an s20.w of 5.2S, phosphorylated a serine residue(s) in casein and was not inhibited by the cyclic AMP-dependent protein kinase inhibitor. 7. These studies demonstrate the existence in rat islets of Langerhans of multiple forms of cyclic AMP-dependent protein kinase and also the presence of a cyclic AMP-independent protein kinase distinct from the free catalytic subunit of cyclic AMP-dependent protein kinase. The presence of the cyclic AMP-independent protein kinase may account for the observed characteristics of 32P incorporation into endogenous protein in homogenates of rat islets.     

Characterization of the protein kinase activities of human platelet supernatant and particulate fractions.

After human platelets were lysed by freezing and thawing in the presence of EDTA, about 35% of the total cyclic AMP-dependent protein kinase activity was specifically associated with the particulate fraction. In contrast, Ca2+-activated phospholipid-dependent protein kinase was found exclusively in the soluble fraction. Photoaffinity labelling of the regulatory subunits of cyclic AMP-dependent protein kinase with 8-azido-cyclic [32P]AMP indicated that platelet lysate contained a 4-fold excess of 49 000-Da RI subunits over 55 000-Da RII subunits. The RI and RII subunits were found almost entirely in the particulate and soluble fractions respectively. Chromatography of the soluble fraction on DEAE-cellulose demonstrated a single peak of cyclic AMP-dependent activity with the elution characteristics and regulatory subunits characteristic of the type-II enzyme. A major enzyme peak containing Ca2+-activated phospholipid-dependent protein kinase was eluted before the type-II enzyme, but no type-I cyclic AMP-dependent activity was normally observed in the soluble fraction. The particulate cyclic AMP-dependent protein kinase and associated RI subunits were solubilized by buffers containing 0.1 or 0.5% (w/v) Triton X-100, but not by extraction with 0.5 M-NaCl, indicating that this enzyme is firmly membrane-bound, either as an integral membrane protein or via an anchor protein. DEAE-cellulose chromatography of the Triton X-100 extracts demonstrated the presence of both type-I cyclic AMP-dependent holoenzyme and free RI subunits. These results show that platelets contain three main protein kinase activities detectable with histone substrates, namely a membrane-bound type-I cyclic AMP-dependent enzyme, a soluble type-II cyclic AMP-dependent enzyme and Ca2+-activated phospholipid-dependent protein kinase, which was soluble in lysates containing EDTA.     

Activation of cyclic AMP-dependent protein kinase and stimulation of protein phosphorylation in response to adenosine in C-1300 murine neuroblastoma

DEAE-cellulose chromatography of the 20,000g supernatant fraction of homogenates of C-1300 murine neuroblastoma (clone N2a) yields one major and two minor peaks of cyclic AMP-dependent protein kinase activity. Assessment of the endogenous activation state of the enzyme(s) reveals that the enzyme is fully activated by the treatment of whole cells with adenosine (10 μM) in the presence of the phosphodiesterase inhibitor Ro 20 1724 (0.7 mM). This treatment produces a large elevation in the cyclic AMP content of the cells. The treatment of whole cells with adenosine alone (1–100 μM) or Ro 20 1724 alone (0.1–0.7 mM) produces minimal elevations in cyclic AMP but nevertheless causes significant activations of cyclic AMP-dependent protein kinase. The autophosphorylation of whole homogenates of treated and untreated cells was studied using [γ-³²P] ATP, sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography. Treatments which activate cyclic AMP-dependent protein kinase selectively stimulate the incorporation of ³²P into several proteins. This stimulation is most prominent in the 15,000-dalton protein band. The addition of cyclic AMP to phosphorylation reactions containing homogenate of untreated cells stimulates the phosphorylation of the same protein bands. These results indicate that adenosine may have regulatory functions through its effect on the cyclic AMP: cyclic AMP-dependent protein kinase system.     

Phosphorylation of cardiac sarcolemma by endogenous and exogenous protein kinases.

Sarcolemmal membranes isolated from guinea pig heart ventricles contained endogenous protein kinase activity and protein substrates for this enzyme. Phosphorylation of sarcolemma was modestly stimulated by cyclic AMP with the half-maximal stimulation at 0.5 μm cyclic AMP. The phosphorylation of sarcolemma due to endogenous kinase was dependent on Mg²⁺. The apparent affinity for Mg²⁺ was found to be 1.4 and 0.53 mm in the absence and presence of 1 μm cyclic AMP, respectively. The apparent affinity for ATP was 55 μm. Sarcolemmal membranes were also phosphorylated by exogenous (purified) cyclic AMP-dependent protein kinase(s). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of phosphorylated membranes, followed by slicing and determination of the radioactivity in the gel slices, showed that endogenous protein kinase activity promoted the phosphorylation of specific protein peaks, arbitrarily designated a–g in order of increasing relative mobility (relative molecular weights 125,000, 110,000, 86,000, 58,000, 48,000, 22,000, and 16,000, respectively); peak e (48,000) was the major phosphorylated band. Exogenous protein kinase stimulated the phosphorylation of all peaks. However, the degree of stimulation of the low molecular weight peaks f and g was more marked. Results obtained after treatment of phosphorylated membranes with hydroxylamine at acid pH indicated the absence of any significant amount of acyl phosphate-type incorporation of phosphate. Purified phosphoprotein phosphatase from rabbit liver effected dephosphorylation of previously phosphorylated sarcolemma; this treatment resulted in dephosphorylation of all peaks (a–g). Pretreatment of sarcolemma with trypsin (membrane to trypsin ratio of 100) was found to markedly reduce both the total membrane phosphorylation as well as relative phosphorylation of peaks c, f, and g. On the other hand, pretreatment of sarcolemma with phospholipase c slightly stimulated total membrane phosphorylation with nondiscriminatory enhancement of the phosphorylation of all peaks. Microsomal membrane vesicles (enriched in sarcoplasmic reticulum fragments) isolated from guinea pig heart ventricle also contained endogenous protein kinase activity. Cyclic AMP modestly increased the kinase. Polypeptides of molecular weights 56,000, 22,000, and 16,000 were found to be phosphorylated. Exogenous (purified) cyclic AMP-dependent protein kinase increased the phosphorylation of microsomes and of 22,000 and 16,000 molecular weight polypeptides.     

Intramitochondrial localization of alanine aminotransferase in rat-liver mitochondria: comparison with glutaminase and aspartate aminotransferase

Summary The removal of the outer mitochondrial membrane and hence of constituents of the intermembrane space in rat-liver mitochondria using digitonin showed that phosphate-dependent glutaminase, alanine and aspartate aminotransferase were localized in the mitoplasts. Further fractionation of mitoplasts following their sonication resulted in 90% of glutaminase, 98% of alanine aminotransferase and 48% of aspartate aminotransferase being recovered in the soluble fraction while the remainder of each enzyme was recovered in the sonicated vesicles fraction. These results indicated that glutaminase and alanine aminotransferase were soluble matrix enzymes, the little of each enzyme recovered in the sonicated vesicles fraction being probably due to entrapment in the vesicles. Aspartate aminotransferase had dual localization, in the inner membrane and matrix with the high specific activity in sonicated vesicles confirming its association with the membrane. Activation experiments suggested that the membrane-bound enzyme was localized on the inner side of the inner mitochondrial membrane.     

Comparison of Proteins Involved with Cyclic AMP Metabolism Between Synaptic Membrane and Postsynaptic Density Preparations Isolated from Canine Cerebral Cortex and Cerebellum

Synaptic membrane and postsynaptic density (PSD) fractions isolated from canine cerebral cortex and cerebellum were assayed for the following proteins: adenylate cyclase and phosphodiesterase (PDE) activities against cyclic AMP and cyclic GMP, the regulatory subunit of the cyclic AMP-dependent protein kinase, and the substrate proteins for this kinase. The results were expressed on the basis of both the protein content of the fractions and the number of synapses in the synaptic membrane fractions. The number of synapses on a constant protein content basis was about three times higher in the cerebral cortex synaptic membrane fraction than in the comparable cerebellar fraction. Adenylate cyclase activity was from 3.4 to 5.6 times higher in the cerebral cortex membrane fraction than in the cerebellar membrane fraction based on protein content but only slightly higher based on synapse counts. PSD fractions had no adenylate cyclase activity. The cyclic AMP-PDE activity was from 17 to 27 times higher in the cerebral cortex membrane fraction than in the cerebellar membrane fraction based on protein content, and about five times higher based on synapse counts. By doing PDE histochemistry at the electron microscopy level it was found that all the cerebral cortex PSDs in the isolated fraction contained PDE activity, none being found associated with the broken-up material in the fraction. The amount of the regulatory subunit of the cyclic AMP-dependent protein kinase was about equal in the two fractions based on protein, but about one-third lower in cerebral cortex fraction than in cerebellar fractions. In the cerebral cortex membrane fraction the primary substrate for the cyclic AMP-dependent protein kinase is synapsin I, with much lower amounts in the cerebellar membrane fraction. The PSD fraction from the two sources also showed these differences in synapsin I content. In the cerebellar membrane fraction, the primary substrate for the enzyme is a approximately 245,000 Mr protein not found in the cerebral cortex membrane fraction. The findings that the turnover of cyclic AMP is much higher in cerebral cortex synapses than in cerebellar synapses, and that differences are found between the cerebral cortex and cerebellum with regard to the substrate proteins for the cyclic AMP-dependent protein kinase indicate a divergence in the effect of cyclic AMP between cerebral cortex and cerebellar synapses.     

Creatine kinase of rat heart mitochondria. The demonstration of functional coupling to oxidative phosphorylation in an inner membrane-matrix preparation

To define more clearly the interactions between mitochondrial creatine kinase and the adenine nucleotide translocase, the outer membrane of rat heart mitochondria was removed by digitonin, producing an inner membrane-matrix (mitoplast) preparation. This mitoplast fracton was well-coupled and contained a high specific activity of mitochondrial creatine kinase. Outer membrane permeabilization was documented by the loss of adenylate kinase, a soluble intermembrane enzyme, and by direct antibody inhibition of mitochondrial creatine kinase activity. With this preparation, we documented four important aspects of functional coupling. Kinetic studies showed that oxidative phosphorylation decreased the value of the ternary enzyme-substrate complex dissociation constant for MgATP from 140 to 16 microM. Two approaches were used to document the adenine nucleotide translocase specificity for ADP generated by mitochondrial creatine kinase. Exogenous pyruvate kinase (20 IU/ml) could not readily phosphorylate ADP produced by creatine kinase, since added pyruvate kinase did not markedly inhibit creatine + ATP-stimulated respiration. Additionally, when ADP was produced by mitochondrial creatine kinase, the inhibition of the translocase required 2 nmol of atractyloside/mg of mitoplast protein, while only 1 nmol/mg was necessary when exogenous ADP was added. Finally, the mass action ratio of the mitochondrial creatine kinase reaction exceeded the apparent equilibrium constant when ATP was supplied to the creatine kinase reaction by oxidative phosphorylation. Overall, these results are consistent with much data from intact rat heart mitochondria, and suggest that the outer membrane plays a minor role in the compartmentation of adenine nucleotides. Furthermore, since the removal of the outer membrane does not alter the unique coupling between oxidative phosphorylation and mitochondrial creatine kinase, we suggest that this cooperation is the result of protein-protein proximity at the inner membrane surface.     

Long-chain acyl CoA synthetase,carnitine and β-oxidation in the pea-seed mitochondrion

Mitochondria from pea (Pisum sativum L.) seeds were separated into two fractions, mitoplasts (intact inner membrane) and the outer-membrane fraction. The mitoplasts only oxidised palmitate in the presence of carnitine and added outermembrane fraction. Mitoplasts were able to oxidise palmitoylCoA in the presence of carnitine and added outer-membrane fraction had no effect on this oxidation. It was concluded that a long-chain acylCoA synthetase (EC 6.2.1.3) was located on the outer membrane and that the activity of this enzyme in assays was more than sufficient to account for any observed rate of O₂ uptake during palmitate oxidation by pea mitochondria. The location of carnitine long-chain acyltransferase (carnitine palmitoyl transferase EC 2.3.1.21) would appear to be the mitoplast i.e. the inner mitochondrial membrane, and confirms the previous work at Newcastle.Abbreviation Tris 2-amino-2-(hydroxymethyl)-1,3-propanediol     

Localization of the NADH kinase in the inner membrane of yeast mitochondria

The bulk of NADH kinase of Saccharomyces cerevisiae was recovered in the mitochondrial fraction prepared from spheroplasts. Most of the NADH kinase was localized in the inner membrane fraction, which was separated from other mitochondrial components by the combined swelling, shrinking, and sonication procedure. Treatment of mitoplasts with antiserum against the NADH kinase caused inactivation of the enzyme. On the contrary, no influence was observed upon the same treatment of intact mitochondria. p-Chloromercuribenzoate and eosin-5-maleimide inactivated the enzyme without affecting the matrix ATPase. The NADH kinase was enzymatically iodinated in mitoplasts, but not in the intact mitochondria. These results support the conclusion that NADH kinase is localized and functions at the intermembrane space side of the mitochondrial inner membrane. It is evident that the NADH kinase is encoded by nuclear gene(s) because it is synthesized in the presence of chloramphenicol or acriflavine, and a significant amount of the enzyme was detected in mitochondrial DNA-deficient mutants.     

Phosphorylation of membranes from the rat gastric mucosa

Gastric mucosal membranes derived primarily from parietal cells were found to contain endogenous protein kinase systems as well as several phosphate-accepting substrates. One specific membrane protein with a molecular weight of 88 000 was phosphorylated only in the presence of calcium, while the degree of phosphorylation of three other membrane proteins was similarly increased. The activity of the calcium-dependent protein kinase was found to be totally inhibited in the presence of trifluoperazine, a phenothiazine known to specifically inactivate calmodulin. These results suggest that a calmodulin- and calcium-dependent phosphorylation system may be a component of the parietal cell membrane. Phosphorylation of the membrane proteins was not affected by either cyclic AMP or cyclic GMP. The heat-stable inhibitor protein of cyclic AMP-dependent protein kinase did not inhibit the endogenous protein kinase activity suggesting that the membrane enzyme is not similar to the cytosolic protein kinase. However, the catalytic subunit of the soluble enzyme was capable of phosphorylating a number of membrane proteins indicating that after maximal autophosphorylation of the gastric membranes, phosphate-acceptor sites are still available to the cytosolic cyclic AMP-dependent protein kinase.     

用荧光素磷脂酰乙醇胺直接测定线粒体内膜外表面pH

由磷脂极性头部基团和结合水分子组成的氢键网络有利于质子沿膜表面侧向快速扩散。因而在线粒体氧化磷酸化过程中,与呼吸链电子传递相偶联的跨膜转运质子是否滞留于线粒体内膜外表面即成为一个值得探讨的课题。本文采用荧光素磷脂酰乙醇胺标记于线粒体内膜外表面,首次建立了直接测定线粒体内膜外表面ｐＨ的方法。标记后,线粒体内膜体呼吸控制率,呼吸链电子传递驱动的质子跨膜转移活性及ＡＴＰ合成活性下降了近２８．０％,１１．     

Phospholipid/Ca2+-dependent protein kinase activity in human parathyroid adenoma

We have examined the activities of phospholipid/Ca2+-dependent and cyclic AMP-dependent protein kinases of the parathyroid adenomas and the atrophic glands which were resected from three patients with primary hyperparathyroidism. Phospholipid/Ca2+-dependent protein kinase activity of atrophic parathyroid gland was exclusively present in cytosol fraction (90.7 +/- 12.3%). On the other hand, phospholipid/Ca2+-dependent protein kinase activity of parathyroid adenomas was 66.9 +/- 6.4% in cytosol and 33.1 +/- 6.4% in membrane fraction, suggesting a translocation of the enzyme from the cytosol to the membranes. Cyclic AMP-dependent protein kinase activity appeared to be higher in parathyroid adenoma than in atrophic parathyroid gland in both cytosol and membrane fractions.     

Direct evidence for internalization of mitochondrial aspartate aminotransferase into mitoplasts.

Mitochondrial aspartate aminotransferase, an enzyme localized on the inner face of the inner mitochondrial membrane, is released into the intermembrane space upon addition of a "movement effector" (succinate, fumarate, pyruvate, or glutamate) [Waksman, A., & Rendon, A. (1974) Biochimie 56, 907-924]. After removal of the movement effector, 90% of the released enzyme rebound to mitoplasts. Lubrol fractionation showed that this bound activity was associated with the inner membrane. Internalization was demonstrated by using both enzymatic and molecular approaches. It was found that 70% of the reassociated enzyme became inaccessible from the outside of the mitoplast either to a nonpermeating substrate (NADH), to mild protease hydrolysis, or to recognition by a specific antibody. In contrast, in inside-out vesicles, the enzyme remained accessible to NADH, protease, and antibodies. Latency measurements performed at different temperatures on whole intact mitochondria confirmed the existence of reversible intermembrane movement of the enzyme in situ.     

Reconsideration of the multiple cyclic nucleotide phosphodiesterases in bovine heart.

DEAE-cellulose chromatography demonstrated the presence of three peaks of cyclic nucleotide phosphodiesterase activity in the hearts of cattle during the summer and only two peaks during exposure to freezing temperatures. The hydrolysis of 10^?6M cyclic AMP by peak II, the variable activity, was stimulated 160% by 10^?6M cyclic GMP and was inhibited by chelation of Ca²⁺. Peak II activity was not a distinct enzyme but rather a mixture of activator-dependent phosphodiesterase, phosphodiesterase activator and type II cyclic AMP-dependent protein kinase.     

Properties and cellular distribution of cyclic AMP-dependent protein kinase from thymus

A protein kinase that catalyzes the phosphorylation of histone was partially purified from rat thymus, and the rate of histone phosphorylation was stimulated three- to fourfold by 1 × 10^?6 M adenosine 3′,5′-monophosphate (cyclic AMP). Thymic protein kinase was more active than the enzyme from spleen. Histone fractions f1, f2a, f2b, and f3 were all capable of serving as phosphate acceptors for the thymic protein kinase, and the rate of phosphorylation of each fraction was stimulated by cyclic AMP. The ability of various 3′,5′-mononucleotides to stimulate protein kinase activity was compared. Inosine 3′,5′-monophosphate (cyclic IMP) was the most effective substitute for cyclic AMP. The cellular distribution of cyclic AMP-dependent protein kinase and adenylate cyclase activities in the thymus was determined. Cyclic AMP-dependent protein kinase activity is present in both small thymocytes and residual thymic tissue. The specific activity of protein kinase from residual tissue, both for basal and cyclic AMP-stimulated enzyme, was greater than that of enzyme from small thymocytes. In contrast to this, adenylate cyclase activity is predominately localized in the thymocytes.     

Protein phosphorylation in human peripheral blood lymphocytes. Subcellular distribution and partial characterization of adenosine 3'':5''-cyclic monophosphate-dependent protein kinase.

Cytoplasmic and membrane fractions prepared from human peripheral-blood lymphocytes both contained cyclic AMP-dependent protein kinase activity and endogenous protein kinase substrates. Protein kinase activity in the particulate fractions was not eluted with 0.25 M-NaCl, suggesting that it was not derived from non-specifically absorbed soluble cytoplasmic protein kinase. Nor was the particulate protein kinase activity eluted by treatment with cyclic AMP, suggesting that the catalytic subunit is membrane-bound and arguing against cyclic AMP-induced translocation of particulate activity. Cyclic AMP-dependent protein-phosphorylating activity in the cytoplasmic fraction was highly sensitive to inhibition by Mn2+, and was co-eluted from DEAE-cellulose primarily with type-I rabbit skeletal-muscle kinase. Cyclic AMP-dependent phosphorylating activity in the plasma-membrane fractions was stimulated at low [Mn2+] and inhibited only at high [Mn2+]. When solubilized with Nonidet P-40, plasma-membrane protein kinase was co-eluted from DEAE-cellulose with type-II rabbit muscle kinase. These differences, together with the strong association of the particulate kinases with the particulate fraction, suggest the possibility of compartmentalized protein phosphorylation in intact lymphocytes.     

Effect of the outer mitochondrial membrane fraction on mitoplast respiration

Additions of the fraction of outer mitochondrial membranes to the mitoplast suspension is shown to bring about an increase of the ADP-stimulated respiration rate, indices of respiration control and uncoupled respiration. This effect is not a result of the cytochrome c presence in the fraction of outer membranes. In the glycerol-containing medium which causes dissociation of intermembrane contacts the coupling effect of outer membranes on mitoplast respiration is not revealed. It is concluded that the outer membrane in contact with the inner one takes part in realization of the mitochondrial coupling.     

Reversible in activation of purified (Na+ + K+)-ATPase from human renal tissue by cyclic AMP-dependent protein kinase.

Human renal (Na+ + K+)-ATPase (ATP phosphohydrolase, EC 3.6.1.3) preparations which exhibited a non-linear reaction rate, contained high levels of membrane-bound cyclic AMP-dependent protein kinase, while this latter activity was much less or absent in purified preparations. A non-linear reaction rate was observed in a purified preparation of (Na+ + K+)-ATPase by reconstituting the enzyme into lipid vesicles with cyclic AMP-dependent protein kinase. The addition of cyclic AMP to the ATPase assay of these lipid vesicles inactivated the (Na+ + K+)-ATPase. The cytoplasmic fraction of the cell contained a nondialyzable factor, which prevented (or reversed) the cyclic AMP-mediated inactivation of the enzyme.     

Localization of catalytic and regulatory subunits of cyclic AMP-dependent protein kinases in mitochondria from various rat tissues.

Observation and quantification of the catalytic subunit C of cyclic AMP-dependent protein kinases by immuno-gold electron microscopy suggested a high concentration of cyclic AMP-dependent protein kinases in mitochondria from liver, kidney, heart and skeletal muscle, pancreas, parotid gland and brain cells. The position of gold particles pointed to a localization in the inner membrane/matrix space. A similar distribution was obtained by immunolocalization of the cyclic AMP-dependent protein kinase regulatory subunits RI and RII in liver, pancreas and heart cells. The results indicated the presence of both the type I and the type II cyclic AMP-dependent protein kinases in mitochondria of hepatocytes, and the preferential occurrence of the type I protein kinase in mitochondria from exocrine pancreas and heart muscle. The immunocytochemical results were confirmed by immunochemical determination of cyclic AMP-dependent protein kinase subunits in fractionated tissues. Determinations by e.l.i.s.a. of the C-subunit in parotid gland cell fractions indicated about a 4-fold higher concentration of C-subunit in the mitochondria than in a crude 1200 g supernatant. Immunoblot analysis of subfractions from liver mitochondria supported the localization in situ of cyclic AMP-dependent protein kinases in the inner membrane/matrix space and suggested that the type I enzyme is anchored by its regulatory subunit to the inner membrane. In accordance with the immunoblot data, the specific activity of cyclic AMP-dependent protein kinase measured in the matrix fraction was about twice that measured in whole mitochondria. These findings indicate the importance of cyclic AMP-dependent protein kinases in the regulation of mitochondrial functions.