Acute in vivo stimulation of low-Km cyclic AMP phosphodiesterase activity by insulin in rat-liver Golgi fractions

A low-Km phosphodiesterase activity, which is acutely stimulated by insulin in vivo, has been identified in plasma membranes and Golgi fractions prepared from rat liver homogenates in isotonic sucrose. Within seconds after insulin injection (25 micrograms/100 g body weight) cAMP phosphodiesterase activity increases by 30-60% in Golgi fractions and by 25% in plasma membranes; activity in crude particulate and microsomal fractions is unaffected. The increase in activity is short-lived in the light and intermediate Golgi fractions, but persists for at least 10 min in the heavy Golgi fraction. It precedes the translocation of insulin and insulin receptors to these fractions, which is maximal at 5 min. The doses of insulin required for half-maximal and maximal activation are, respectively, 7.5 micrograms/100 g and 25 micrograms/100 g body weight. Golgi-associated cAMP phosphodiesterase activity shows non-linear kinetics; a high-affinity component (Vmax, 13 pmol min-1 mg protein-1; Km, 0.35 microM) is detectable. Insulin treatment increases the Vmax 60-70%, but does not affect the Km. Unlike the low-Km cAMP phosphodiesterase associated with crude particulate fractions, the Golgi-associated activity is not easily extractable by solutions of low or high ionic strength. On analytical sucrose density gradients, low-Km cAMP phosphodiesterase associated with the total particulate fraction equilibrates at lower densities than endoplasmic reticulum and lysosomal markers, but at a higher densities than plasma membrane, Golgi markers and insulin receptors. Insulin treatment increases the specific activity of the enzyme by 20-60% at densities below 1.12 g cm-3, and by 20-40% in the density interval 1.23-1.25 g cm-3. Such treatment also causes a slight, but significant shift in the distribution of phosphodiesterase towards lower densities. It is suggested that Golgi elements or physically similar subcellular structures are a major site of localization of insulin-sensitive cAMP phosphodiesterase in rat liver. However, internalization of the insulin-receptor complex is probably not required for enzyme activation.     

Studies on the heterogeneity of sarcoplasmic reticulum vesicles.

Isolated sarcoplasmic reticulum vesicles from rabbit white muscle were separated into a light (15--20% of total microsomes) and a heavy (80--85%) fraction by density gradient centifugation. The ultrastructure, chemical composition, enzymic activities and localization of membrane components in the vesicles of both fractions were investigated. From the following results it was concluded that both fractions are derived from the membranes of the sarcoplasmic reticulum system of the muscle: (i) The protein pattern of both fractions is essentially the same, except for different ratios of acidic, Ca2+-binding proteins. (ii) The 105000 dalton protein of the light fraction cross-reacts immunologically with the Ca2+-dependent ATPase of the heavy fraction. (iii) Ca2+-dependent ATPase, although of different specific activity, is found in both fractions. After rendering the vesicles leaky, specific activities in both fractions reach the same value. The light fraction was found to consist of "inside-out" vesicles by the following criteria: (i) No Ca2+ accumulation can be measured and the Ca2+-dependent ATPase activity is low and variable. (ii) The rate of trypsin digestion is lower and, compared to the heavy microsomes, a different ratio of degradation products is obtained. (iii) The sarcoplasmic reticulum membrane has a highly asymmetrical lipid distribution. This distribution of aminophospholipids is opposite to that in vesicles of heavy fraction. The light sarcoplasmic reticulum fraction has a higher phospholipid to protein ratio than the heavy one. This is consistent with the possibility that the two fractions derive from different parts of the sarcoplasmic reticulum system.     

A comparison of vesicles derived from terminal cisternae and longitudinal tubules of sarcoplasmic reticulum isolated from rabbit skeletal muscle

Sarcoplasmic reticulum vesicles were separated into heavy (derived from terminal cisternae) and light (derived from longitudinal tubules) fractions, according to Meissner [Biochim. Biophys. Acta, 389, 51-68 (1975)]. The similar Ca2+ sensitivities of phosphoprotein formation, ATPase activity and calcium uptake, and the similar phosphoprotein turnover rates (ATPase/phosphoprotein formation) of both fractions indicate that the same ATPase enzyme is present in the terminal cisternae and longitudinal sarcoplaxmic reticulum. The higher V for Ca2+-activated ATPase activity and calcium uptake in the light fraction correlated with the higher concentration of ATPase enzyme per mg of membrane protein in this fraction. In both the presence and absence of calcium-precipitating anions, the light fraction stored more calcium than the heavy. The Ca2+ dependence of calcium release after addition of EGTA appeared similar in both fractions, but the rate of calcium release was more rapid in the light fraction. These findings suggest that calcium release may occur more rapidly from longitudinal than terminal cisternae portions of the sarcoplasmic reticulum and that calcium release, like calcium uptake, may be mediated by the ATPase enzyme in the sarcoplasmic reticulum membrane. Although the activation energies for Ca2+-activated ATPase activity above and below the transition temperature were significantly different for the heavy and light fractions, their transition temperatures were similar. Partial purification of the ATpase enzyme by deoxycholate treatment modified the activation energies of the light but not the heavy fraction and caused the activation energies to become similar. The phosphoprotein levels of heavy and light vesicles did not become similar after deoxycholate treatment, although gel electrophoretograms indicated both samples contained > 90% ATPase protein. These results indicate the protein-lipid associations in these two fractions may be different.     

Characterization of inositol 1,4,5-trisphosphate receptors and calcium mobilization in a hepatic plasma membrane fraction

The distribution of hepatic binding sites for the calcium-mobilizing second messenger, inositol 1,4,5-trisphosphate (IP3), was analyzed in subcellular fractions of the rat liver by binding studies with [32P]IP3 and compared with the Ca2+ release elicited by IP3 in each fraction. Three major subcellular fractions enriched in plasma membrane, mitochondria, and endoplasmic reticulum were characterized for their 5'-nucleotidase, glucose-6-phosphatase, succinate reductase, and angiotensin II binding activities. The fraction enriched in plasma membrane showed 7- and 20-fold increases in IP3 binding capacity over those enriched in endoplasmic reticulum and mitochondria, respectively, and contained a single class of high-affinity binding sites with Kd of 1.7 +/- 1.0 nM and concentration of 239 +/- 91 fmol/mg protein. IP3 binding reached equilibrium in 30 min at 0 degrees C, and the half-time of dissociation was about 15 min. The specificity of the IP3 binding sites was indicated by their markedly lower affinities for inositol 1-phosphate, phytic acid, fructose 1,6-bisphosphate, 2,3-bisphosphoglycerate, and inositol 1,3,4,5-tetrakisphosphate. The Ca2+-releasing activity of IP3 in the subcellular fractions was monitored with the fluorescent indicator, Fura-2. All three fractions showed ATP-dependent Ca2+ uptake and rapidly released Ca2+ in response in IP3. The fraction enriched in plasma membrane was the most active in this regard, releasing 174 +/- 67 pmol Ca2+/mg of protein compared to 45 +/- 10 and 48 +/- 7 pmol/mg protein for the fractions enriched in endoplasmic reticulum and mitochondria, respectively. These data suggest that the [32P]IP3 binding sites represent specific intracellular receptors through which IP3 mobilizes Ca2+ from a storage site associated (or co-purifying) with the plasma membrane of the rat liver. It is likely that a specialized vesicular system (to which IP3 can bind and trigger the release of Ca2+) is located in close proximity with the plasma membrane and is thus adjacent to the site at which IP3 is produced during stimulation of the hepatocyte by Ca2+-mobilizing hormones.     

Effect of caffeine on the Ca2+-transport function of sarcoplasmic reticulum vesicles in the rat myocardium

The effects of caffeine on active transport of Ca2 by heavy and light fractions of rat myocardial microsomes were investigated with the use of a Ca2+-selective electrode and nephelometry. It was found that under the effect of caffeine (5 mM) the rate of Ca2 transport in the presence of oxalate decreased by 30 to 40%. The caffeine-induced inhibition was prevented by ruthenium and tetracaine, thus suggesting the inhibitor specificity. Since caffeine is a specific blocker of Ca2 transport to the terminal cisterns of the skeletal muscle sarcoplasmic reticulum, it is assumed that the microsomal fraction of rat myocardium contains terminal cistern fragments.     

cAMP phosphodiesterase from phototrophic bacteria Rhodospirillum rubrum]

cAMP phosphodiesterase activity is discovered in supernatant of R. rubrum cell homogenate after centrifugation at 1000 g. The enzyme is highly active (5.62 nmoles/mg of protein per 1 min) at a broad pH range--from 7.0 to 9.0. The enzyme activity is strongly inhibited with caffeine and dithiotreitol and very significantly inhibited by ascorbic acid. The dependence of the enzyme activity on the incubation time and protein and substrate concentrations in the reaction mixture is estimated. cAMP phosphodiesterase is found in soluble fraction and in particule fractions sedimenting at 30 000 g. The enzyme activity is completely absent in washed chromatophores sedimenting at 160 000 g.     

Ca2+ transporting activity of membrane fractions isolated from the post-mitochondrial supernatant of rat liver

The post-mitochondrial supernatant of rat liver contains two vesicular fractions which transport Ca2+ actively. The heavier fraction, sedimenting at 17.500 xg, 20 min, is enriched in plasma membrane markers and apparently contains both a Ca2+ pumping ATPase and a Na+/Ca2+ exchanger. These activities have been attributed to the plasma membrane vesicles. The lighter fraction, sedimenting at 100.000 xg, 60 min, is enriched in endoplasmic reticulum markers, and contains only a Ca2+ pumping ATPase, which can be differentiated from that of the heavier fraction on the basis of the sensitivity to vanadate. The Ca2+ pumping activity of endoplasmic reticulum appears to be regulated by both a cAMP-dependent, and a calmodulin-dependent system. The former system involves a heat-stable protein fraction from the cytosol. The regulation by the cAMP and the calmodulin-dependent systems involves the phosphorylation of several proteins in the endoplasmic reticulum membrane.     

Endogenous phosphorylation of sarcoplasmic reticulum fragments of rabbit fast skeletal muscles

The purified membrane fragments of sarcoplasmic reticulum (SR) of rabbit fast skeletal muscles were found to incorporate 32P from[gamma-32P]ATP in endogenous membrane substrates and in histone H1. The existence of membrane-bound protein kinase of SR was demonstrated by steady state binding of [3H]-cAMP to the SR membranes. The constant of [3H]cAMP binding to the membranes is 2.5 +/- 0.003 x 10(6) M-1, the number of binding sites is 6.1 +/- 0.8 pmol per 1 mg of protein. The endogenous phosphorylation of SR components was inhibited by cAMP and cGMP at concentrations of 10(-7)-10(-6) and depended on Mg2+ and Ca2+. The thermostable protein inhibitor of cAMP-dependent protein kinase inhibited the endogenous phosphorylation of SR membranes by 30-40%. The protein phosphoproduct of SR membranes revealed the properties of a phosphoester. The membrane-bound protein kinase was active towards the exogenous substrate--histone H1. Phosphorylation in the presence of histones was independent of cyclic nucleotides, Mg2+ and Ca2+. Fractionation of 32P-labelled solubilized membranes in polyacrylamide gel in the presence of Na-SDS showed that the radioactivity is bound to protein zones with molecular weights of 95 000 and 6000.     

Characteristics of sarcoplasmic reticulum from slowly glycolysing and from rapidly glycolysing pig skeletal muscle post mortem

The composition and function of fragmented sarcoplasmic reticulum from pig skeletal muscle was examined in the period immediately post mortem. Muscle was defined as being either slowly glycolysing or rapidly glycolysing on the basis of colour, pH and concentrations of glycogen and lactate. The microsomal fraction was separated on a discontinuous gradient of 35, 40 and 45% (w/v) sucrose into heavy and intermediate fractions which sedimented to the interfaces, and a light fraction which remained on the surface of the 35%-sucrose layer. The sarcoplasmic reticulum from rapidly glycolysing muscle had a lower buoyant density than had that from slowly glycolysing muscle. This was reflected in the consistent lack of material in the heavy fraction and a greater proportion in the light fraction. The latter material had significantly lower ratios (w/w) of protein to phospholipid (2.3:1 versus 3.8:1) and of protein to cholesterol (10.4:1 versus 15.6:1). There were no gross differences in phospholipid content or in fatty acid composition of individual phospholipid classes in the membranes from the two types of muscle. Analysis of membrane proteins by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis showed that ATPase (adenosine triphosphatase) was a major component of each fraction and that its contribution to the total protein content of the membrane was greater in rapidly glycolysing muscle, suggesting a loss of non-ATPase proteins. The two fractions of sarcoplasmic reticulum prepared from rapidly glycolysing muscle had approximately one-third the normal activities of Ca(2+) binding and Ca(2+) uptake in the presence of ATP and one-half the passive Ca(2+)-binding capacity in the absence of ATP of the fractions from slowly glycolysing muscle. However, the (Ca(2+)+Mg(2+))-stimulated ATPase activities were similar. Efflux from actively loaded vesicles, after the addition of EDTA, consisted of a rapid and a slow phase. Vesicles from rapidly glycolysing muscle lost 60% of associated Ca(2+) (approx. 0.10mumol of Ca(2+)/mg of protein) during the rapid phase, compared with 30% (approx. 0.17mumol of Ca(2+)/mg of protein) in those from slowly glycolysing muscle. The efflux rate during the slower phase was comparable in both types of vesicles. Analysis of the temperature-dependence of (Ca(2+)+Mg(2+))-stimulated ATPase activity revealed that a high-activation-energy process operating in the temperature range 31-45 degrees C in the intermediate and light fractions from slowly glycolysing muscle was not apparent in vesicles from rapidly glycolysing muscle. Conditions that result in the prolonged activation of glycogenolysis in pig muscle post mortem primarily affect the protein components of the sarcoplasmic-reticular membrane, giving rise to a loss of loosely associated proteins. The function of the membranes observed under these conditions does not appear to be due to enhanced permeability of the membrane to Ca(2+) and may be the result of a defect in the transport of Ca(2+) into the vesicles.     

Aminoacyl-tRNA-synthetase activity of subcellular fractions of cerebral cortex]

The total activity of aminoacyl-tRNA-synthetases of myelin, synaptic membranes, heavy and light synaptosomes, mitochondria and soluble fractions of rat cerebral cortex was studied. It was found that the highest activity of the enzymes is localized in the fractions of synaptic membranes and heavy and light synaptosomes and is practically absent in the myelin fraction. The specific activity of the total aminoacyl-tRNA-synthetase fraction in the soluble fraction is 2 times as low as compared to the synaptic membranes and light and heavy synaptosomes.     

Effects of denervation on the contents of cholesterol and membrane systems involved in muscle contraction in rabbit fast-twitch sarcotubular system.

Denervated fast-twitch rabbit muscles were progressively losing their fresh weight and the yield of sarcotubular protein was increasing. The activity of Ca(2+)-ATPase was affected but very slightly, the basal Mg(2+)-ATPase and the Mg(2+)-ATPase/Ca(2+)-ATPase ratio however increased together with a simultaneous depression of the membrane-bound acetylcholinesterase activity. We did not observe any differences in density properties of sarcotubular fractions between control and denervated muscle. However, a relative enrichment in SM and H fraction could be seen after denervation with small changes in the content of the Ca(2+)-pump protein, increased levels of calsequestrin and cholesterol, mostly in the heavy and the SM fraction. After denervation the binding sites for 3H-PN-200-110 did not show any changes in receptor affinity, but the number of putative Ca(2+)-channels increased twice along with a depression of 3H-ouabain binding sites. We suggest that the denervation of fast-twitch muscle leads to the hypertrophy of the junctional sarcoplasmic reticulum and the T-system. Changes in the cholesterol content, in the number of putative Ca(2+)-channels and in Na+, K(+)-ATPase can affect the muscle contraction.     

Intracellular localization of the caffeine-sensitive form of Ca-dependent ATPase in the sarcoplasmic reticulum

A Ca-selective electrode was used to study the effect of caffeine on different fractions of sarcoplasmic reticulum membranes of rabbit skeletal muscles. Caffeine was found to uncouple Ca2+ transport and ATP hydrolysis in a fraction, which is enriched with fragments of terminal cisterns according to the electron microscopy data. Caffeine does not produce any effect on the light fraction containing no fragments of terminal cisterns. It is concluded that caffeine-sensitive Ca2+-dependent ATPase is localized in terminal cisterns of the sarcoplasmic reticulum.     

Some properties of partially purified preparations of cAMP phosphodiesterase from beef heart]

Beef heart cAMP phosphodiesterase (EC 3.1.4.17) was isolated and partially purified using fractionation by ammonium sulfate and gel filtration on the columns with Sephadex G-200 and Sepharose 6B. This method allowed to preserve the enzyme binding to the low-molecular weight thermostable protein regulator of the phosphodiesterase activity. The enzyme preparation was purified 130--180-fold as compared to the original homogenate. The pH-dependence of the enzyme activity in the imidazole and tris -- buffers for the fraction with maximal activity was carried out. The kinetic analysis of this fraction revealed an abnormal kinetic behaviour with two Km values. The enzyme is represented by four forms differing in their molecular weights and possessing different capacity for activation by Ca2+ and protein regulator. No activation was observed in the forms with higher molecular weights, whereas the activity of the forms with lower molecular weights depended on the presence of Ca2+ and protein regulator. It is assumed that some of the above-described forms are capable of interconversions.     

Blockers of the uncoupling action of caffeine on the Ca2+-transport function of sarcoplasmic reticulum membranes

The influence of caffeine on the efficiency of Ca2+ transport in the presence of oxalate by different fractions of sarcoplasmic reticulum (SR) membranes from rabbit skeletal muscle was studied. It was shown that caffeine (5 mM) decreases 4-fold the value of the Ca/ATP ratio in terminal cisterns of the SR without having any appreciable influence on the efficiency of Ca2+ transport by the light fraction of the SR. The uncoupling effect of caffeine is completely blocked by ruthenium red and by the local anesthetics, tetracaine, procaine, and benzocaine.     

Heterogeneity of rat liver mitochondrial fractions and the effect of tri-iodothyronine on their protein turnover

1. Rat liver mitochondria were separated into heavy, light and fluffy fractions by differential centrifugation under standard conditions. 2. All mitochondrial fractions possessed soluble as well as membrane-bound enzymes typical of mitochondria. 3. The heavy fraction represented the stable mitochondrial structures and the fluffy particles appear to be loosely coupled. 4. The light mitochondrial fraction lacked the ability of coupled phosphorylation. 5. A study of mobility and isoelectric pH indicated a similarity in the basic membrane structure of all the mitochondrial fractions. 6. The turnover rates of proteins in the heavy and fluffy particles were almost identical; however, this rate was rapid for the light mitochondrial fraction. 7. On treatment with 3,3',5-tri-iodo-l-thyronine, succinoxidase activity was maximally stimulated much earlier in the light mitochondrial fraction than in the heavy fraction. The activity of the fluffy particles, however, remained almost unaffected. 8. Malate dehydrogenase activity in all the mitochondrial fractions was stimulated only at 40h after tri-iodothyronine treatment. 9. The pattern of incorporation of dl-[1-(14)C]leucine in vivo in the tri-iodothyronine-treated animals indicated a rapid initial incorporation and high synthetic ability of the light mitochondrial fraction. 10. The turnover pattern of proteins of the mitochondrial fractions from animals receiving repeated doses of tri-iodothyronine was remarkably different from the normal pattern and suggested that preformed soluble protein units may be incorporated in the light mitochondrial fraction during maturation to form the stable heavy mitochondria. 11. The amount of light-mitochondrial proteins decreased by 40% on thyroidectomy and increased by 160% on treatment with tri-iodothyronine. 12. The possible significance of these results is discussed in relation to mitochondrial genesis.     

Effect of digitonin on rat myometrium subcellular membrane fractions

Isopycnic centrifugation experiments using sucrose density gradients showed that in digitonin-treated microsomes the distribution of the plasma membrane (PM) marker 5'-nucleotidase was shifted to higher densities. The treatment also caused similar but less pronounced changes in the distribution of protein, the putative endoplasmic reticulum (ER) marker NADPH-dependent cytochrome c reductase, and the inner mitochondrial marker cytochrome c oxidase. Similar experiments using more purified membrane fractions showed that the digitonin treatment led to a comparable increase in the densities of the fractions N1 and N2 previously described as subfractions of plasma membrane and to considerably less increase in the density of the fraction N3B which is enriched in the endoplasmic reticulum and the inner mitochondrial markers. Digitonin inhibited the ATP-dependent Ca uptake by the N1 fraction in a concentration-dependent manner (I50 = 0.3 mg/mL). Digitonin (0.5 mg/mL) inhibited the ATP-dependent azide-insensitive Ca uptake by all the fractions. The results support the hypothesis that (a) N1 and N2 are subfractions of plasma membrane, and (b) ATP-dependent azide-insensitive Ca uptake in rat myometrium is a property of plasma membranes.     

Binding of epidermal growth factor (EGF) and insulin to human liver microsomes and Golgi fractions

Microsomes and Golgi fractions were isolated from 13 human liver samples without local malignancy. Binding of insulin to microsomes (per cent per 0.5 mg protein) was 14.4 +/- 7.9% with two classes of receptors: K1 = 1.4 nM, R1 = 0.28 pmol/mg; K2 = 8.1 nM, R2 = 0.62 pmol/mg. The binding was insignificantly lower than in rats. Binding of EGF was only 3.4 +/- 1.7% with two classes of receptors: K1 = 1.4 nM, R1 = 0.06 pmol/mg; K2 = 10.8 nM, R2 = 0.22 pmol/mg; the binding was much lower than in rats (26.3 +/- 5.8%). Binding of insulin to Golgi fraction (per cent per 0.1 mg protein) was 5.5 +/- 0.4% with straight line Scatchard plot; Kd = 5.6 nM, Ro = 3.06 pmol/mg; it was only half of that found in rats. In one case of hepatoma, the binding of insulin to microsomes was normal but that of EGF very low.     

Effect of caffeine and glycerin on the Ca transport system of sarcoplasmic reticulum fragments from frog skeletal muscles

Parameters of the Ca2+-ion transport system by a fragmented sarcoplasmic reticulum isolated from phasic and tonic frog skeletal muscles were investigated under the action of caffeine or caffeine in combination with glycerol. No changes were observed in the Ca-transport system of a light fraction of the sarcoplasmic reticulum under the influence of caffeine and caffeine-glycerol combination. Caffeine reduced the value of Ca/ATP and enhanced the outflux of Ca2+-ions from membrane fragments of the caffeine-sensitive sarcoplasmic reticulum fraction of both the muscles; the combined effect of caffeine and glycerol was analogous to the action of caffeine applied alone. It is concluded that the potentiation of muscle contraction in the presence of glycerol is not due to the excess of Ca-release from the sarcoplasmic reticulum caused by this agent.     

Activation of the Ca2+ release channel of skeletal muscle sarcoplasmic reticulum by caffeine and related compounds

The caffeine-sensitive Ca2+ release pathway in skeletal muscle was identified and characterized by studying the release of 45Ca2+ from heavy sarcoplasmic reticulum (SR) vesicles and by incorporating the vesicles or the purified Ca2+ release channel protein complex into planar lipid bilayers. First-order rate constants for 45Ca2+ efflux of 1 s-1 were obtained in the presence of 1-10 microM free Ca2+ or 2 X 10(-9) M free Ca2+ plus 20 mM caffeine. Caffeine- and Ca2+-induced 45Ca2+ release were potentiated by ATP and Mg.ATP, and were both inhibited by Mg2+. Dimethylxanthines were similarly (3,9-dimethylxanthine) or more (1,7-, 1,3-, and 3,7-dimethylxanthine) effective than caffeine in increasing the 45Ca2+ efflux rate. 1,9-Dimethylxanthine and 1,3-dimethyluracil (which lacks the imidazole ring) did not appreciably stimulate 45Ca2+ efflux. Recordings of calcium ion currents through single channels showed that the Ca2+- and ATP-gated SR Ca2+ release channel is activated by addition of caffeine to the cis (cytoplasmic) and not the trans (lumenal) side of the channel in the bilayer. The single channel measurements further revealed that caffeine activated Ca2+ release by increasing the number and duration of open channel events without a change of unit conductance (107 pS in 50 mM Ca2+ trans). These results suggest that caffeine exerts its Ca2+ releasing effects in muscle by activating the high-conductance, ligand-gated Ca2+ release channel of sarcoplasmic reticulum.     

Caffeine promotes survival of cultured sympathetic neurons deprived of nerve growth factor through a cAMP-dependent mechanism.

The effects of caffeine on neuronal survival independent of trophic factor support were examined in developing superior cervical ganglion in vitro. We found that caffeine promoted neuronal survival in the absence of nerve growth-factor (NGF) in a dose-dependent manner (EC50 = 6 mM). Pulse treatment with caffeine or high K+ (40 mM), which caused only a transient increase in intracellular free Ca2+ levels ([Ca2+]i), did not promote survival. In contrast, caffeine potentiated the saving effect of various phosphodiesterase inhibitors including theophylline (EC50 = 3 mM) and 3-isobutyl-1-methylxanthine (EC50 = 0.4 mM). Non-xanthine phosphodiesterase inhibitor Ro 20-1724 potentiated the survival promoting effect of caffeine or IBMX. Indeed, administration of 20 mM caffeine rapidly restored the cAMP level of NGF-deprived neurons to normal (0.34 pmol/well) within 10 min; the level reached a plateau level (0.69 pmol/well) at 10 h. Even after 1 day, the sustained level was maintained in the presence of caffeine. In contrast, noradrenaline and isoproterenol, which cause only a transient increase in cAMP levels, did not support survival. These data, in conjunction with others, suggest that sustained levels of second messengers, including not only the [Ca2+]i but also the cAMP level, would support the survival of superior cervical ganglion cells independent of trophic factor support.