Extensive biochemical and genetic evidence has shown that the yeast Ras1 and Ras2 proteins are controlling elements of adenylate cyclase and that, in their absence, cAMP synthesis in vivo is insufficient for viability (Toda et al., 1985; Field et al., 1988). It has been proposed that the Ras proteins interact with the N-terminal inhibitory domain of adenylate cyclase to relieve its inhibition on the catalytic domain (Heideman et al., 1987). Other results have implicated the C-terminal last 100 amino acid residues of adenylate cyclase (Yamawaki-Kataoka et al., 1989) and a domain N-terminally adjacent to the catalytic domain (Uno et al., 1987) as important for interaction with Ras. The activity of the Ras proteins is controlled by the guanine nucleotide exchange factor Cdc25 (Camonis et al., 1986) and the GTPase activating proteins Ira1 (Tanaka et al., 1989) and Ira2 (Tanaka et al., 1990). cAMP controls the activity of protein kinase A (PKA), of which the catalytic subunits are encoded by the TPK1, TPK2 and TPK3 genes (Toda et al., 1987a), and the regulatory subunits by the BCY1 gene (Toda et al., 1987b).

Biochemical and genetic work has identified several targets of the cAMP pathway. High activity of the pathway causes low levels of trehalose and glycogen, low heat resistance and a reduced expression level of 'STRE'-controlled genes, such as CTT1 (catalase) and SSA3 (Hsp70). Low activity of the pathway causes the opposite phenotype (for reviews, see Broach and Deschenes, 1990; Thevelein, 1992; 1994; Tatchell, 1993; Ruis and Schuller, 1995). This is particularly evident during exponential growth on glucose, as trehalose and glycogen levels, heat resistance and the expression of 'STRE'-controlled genes are low in this condition in wild-type cells. There is considerable evidence that these PKA targets are also controlled by one or more cAMP-independent pathways (Belazzi et al., 1991; Hirimburegama et al., 1992; Durnez et al., 1994; Pernambuco et al., 1996; Crauwels et al., 1997). Whether PKA is directly involved in this alternative regulation is not clear. Several biochemical targets of PKA control have been identified in yeast, e.g. trehalase (Uno et al., 1983), fructose-1,6-bisphosphatase (Müller and Holzer, 1981), glycogen phosphorylase (Lin et al., 1995) and phosphatidylserine synthase (Kinney and Carman, 1988).

Only two triggers of the cAMP pathway are well established in yeast (Thevelein, 1991). The addition of glucose to derepressed yeast cells, but not to glucose-repressed cells, triggers a transient increase in the cAMP level within 1 min (van der Plaat, 1974). Intracellular acidification triggers an equally rapid but more pronounced and longer lasting increase in the cAMP level (Caspani et al., 1985). Glucose does not act through an intracellular acidification effect (Thevelein et al., 1987a). Both phenomena of cAMP signalling are specifically controlled by the low-affinity Pde1 phosphodiesterase, as opposed to the basal cAMP level in the cells, which is specifically controlled by the high-affinity Pde2 phosphodiesterase (Ma et al., 1999). Recent work has also identified a transient lag phase-associated increase in the basal cAMP level (Ma et al., 1997).

Although genetic evidence indicates an involvement of the Ras proteins as signal transmitters for glucose- and intracellular acidification-induced cAMP signalling (Mbonyi et al., 1988), recent work has shown that only intracellular acidification enhances the GTP content on the Ras proteins. The addition of glucose does not affect the ratio of GTP/GDP bound to the Ras proteins (Colombo et al., 1998). On the other hand, the G-protein encoded by the GPA2 gene (Nakafuku et al., 1988) is required for glucose-induced but not for acidification-induced cAMP signalling (Colombo et al., 1998). Recent work has also identified a protein, called Gpr1, with structural and functional homology to G-protein-coupled receptors, of which the C-terminus binds to Gpa2. Gpr1 is also required for glucose-induced activation of cAMP synthesis, suggesting that glucose sensing for activation of the cAMP pathway might be carried out by a G-protein-coupled receptor system consisting of Gpr1-Gpa2 (Kraakman et al., 1999).

During our studies of the glucose-induced cAMP signal in glucose repression mutants (Argüelles et al., 1990) and in hexokinase and glucokinase deletion mutants in the same genetic background, we have discovered that, in this background, a mutation was present that nearly abolished both glucose- and acidification-induced cAMP signalling. These strains were kindly provided by K. D. Entian (Frankfurt, Germany). The corresponding wild-type strains containing this mutation had been used for many years as wild-type laboratory strains, and one of these was used to construct the CEN.PK series of wild-type strains (Randez-Gil et al., 1997). In the present work, we have identified this mutation as a K1876M substitution in adenylate cyclase. We show that the mutation does not reduce the maximal Mn²⁺-dependent activity of the enzyme in vitro, but rather affects its GTP/Mg²⁺-dependent activation. In vivo, the K1876M mutation in adenylate cyclase affects all PKA targets investigated, mainly during the transition from gluconeogenic/respirative to fermentative growth.

FIGURES
 

Fig. 1 . Intracellular cAMP level as a function of time after the addition of 100 mM glucose (A and C...

Fig. 2 . Intracellular cAMP level after the addition of 2,4-dinitrophenol to cells of an lcr1 strain...

Fig. 3 . Intracellular cAMP level as a function of time after the addition of 100 mM glucose (A) or 2...

Fig. 4 . Adenylate cyclase activity measured in isolated plasma membranes of the wild-type strain (W...

Fig. 5 . Glucose-induced mobilization of trehalose (A) and glycogen (B) and glucose-induced loss of ...

Fig. 6 . Growth and glucose level in the medium (A), trehalose (B) and glycogen (C) content and heat...

Fig. 7 . Growth on YPD in the presence of acetic acid. Wild-type strain W303-1A (filled symbols), is...

Fig. 8 . Suppression by the cyr1 ^met1876 (lcr1) mutation of the effect of Ras2^val19 and Gpa2^val132 on...

RESULTS

Discovery and segregation of the mutation

The addition of glucose to derepressed yeast cells triggers a rapid, transient increase in the cAMP level within a few minutes (van der Plaat, 1974), while intracellular acidification, for instance by the addition of protonophores, causes a much higher and longer lasting cAMP increase (Caspani et al., 1985). For our studies of the mechanisms underlying these agonist-induced cAMP increases, several mutant strains have kindly been provided by K. D. Entian (Frankfurt, Germany). Remarkably, all strains derived from a common genetic origin, including the 'wild-type' derivatives, lacked the cAMP increase induced by both glucose and the protonophore 2,4-dinitrophenol. This is shown in Fig. 1A and B for the 'wild-type' derivative ENY.cat80-7A. In the same figure, the glucose- and acidification-induced cAMP responses in the wild-type strain SP1 are shown for comparison. The mutation is apparently also present in a new standard laboratory strain, CEN.PK2-1C, that has been derived from this background (Randez-Gil et al., 1997). This strain also lacks glucose- and acidification-induced cAMP responses (results not shown).

Genomic mapping of the mutation

The lcr1 mutation is allelic with CYR1

A Ty element in the promoter of the lcr1 allele is not responsible for the absence of the cAMP responses

Southern blot analysis with EcoRI-digested DNA of the four strains in a tetrad (strains MV7117A-D) showed that the lcr1 descendants had a different restriction pattern compared with the CYR1 descendants. Instead of a single 3.5 kb band in the wild-type strains, two bands of 3 kb and 8 kb were observed (using a 2.6 kb SphI-EcoRI probe, covering part of the promoter and a 405 bp-long initial part of the wild-type CYR1 gene). Southern blot analysis of the parental strains (SP1, ENY.cat80-7A and ENY.cat80-8A) and the descendants of another tetrad (strains MV7106A-D) confirmed that this aberrant restriction pattern was linked with the lcr1 mutation (results not shown). Based on this result and on additional results obtained with other restriction enzymes and other probes (not shown), we concluded that an insertion of unknown length was present 5' upstream of the PvuII site (position 142) in the promoter of the lcr1 mutant allele. We cloned the promoter with the unknown insert from an lcr1 mutant strain using plasmid eviction (see Experimental procedures). Sequence analysis showed that the insertion was a Ty transposable element (Boeke and Sandmeyer, 1991) and that it ended at position 555 in the promoter of the lcr1 mutant allele (results not shown). Two previous reports have dealt with the insertion of a Ty element into the promoter and into the regions 4 to +82 and +124 to +268 of the CYR1 gene (Iida, 1988; Lenzen et al., 1987, respectively). In the latter case, it caused a strong reduction in adenylate cyclase activity, slow growth and a multistress-resistant phenotype.

Construction of chimeric genes indicates that a mutation located in the 3'-third of the lcr1 allele is responsible for the phenotype

Identification of the mutation as a A5627T substitution

Sequencing of the 3'-terminal fragment revealed a substitution of adenine for thymine at position 5627. Sequencing of the other two fragments confirmed that there was no mutation present. The A5627T mutation was confirmed in all lcr1 derivatives and the wild-type adenine nucleotide in all wild-type derivatives of the following tetrads: 7117A-D, 7103A-D and 7106A-D. DNA sequence analysis of the area around nucleotide 5627 in four of the chimeric constructs showed that the WWW construct contained an adenine at position 5627, whereas the MMM, WWM and WMM constructs contained a thymine at this position.

Introduction of the A5627T mutation into a wild-type strain, cAMP responses in vivo and adenylate cyclase activity in vitro

We have introduced the A5627T mutation in the CYR1 gene of the wild-type strain W303-1A by replacement of the C-terminal third of the wild-type gene by the corresponding part of the cyr1 ^met1876 (lcr1) allele (see Experimental procedures). The presence of only the A5627T mutation in the CYR1 gene of the W303-1A strain was confirmed by sequencing. The resulting strain GD1 lacked both the glucose- and acidification-induced cAMP responses, although a small increase was still observed with glucose (Fig. 3A and B). During the investigation of cAMP responses in the original and tetrad-derived lcr1 strains, similar small increases in the cAMP level after the addition of glucose were occasionally observed in some strains (results not shown). These results indicated that the K1876M mutation in adenylate cyclase is able to largely eliminate the cAMP responses by itself. Given the previous results that linked this phenotype to the C-terminal third of adenylate cyclase, it shows that the K1876M mutation alone is responsible for the elimination of the cAMP responses in the lcr1 mutant strains.

Phenotype of a strain lacking the glucose- and acidification-induced cAMP responses

Two isogenic strains, W303-1A and GD1, were now available that allowed us to investigate (without side-effects possibly caused by mixed genetic backgrounds, the presence of suppressor mutations or multicopy suppressor genes) the physiological relevance of agonist-induced cAMP signalling. Glucose- and intracellular acidification-triggered activation of trehalase are considered to be mediated by the cAMP increases caused by these agonists (Thevelein, 1991). In both cases, the extent of trehalase activation was reduced in the cyr1 ^met1876 (lcr1) mutant (Fig. 3C and D). It was not abolished, indicating the involvement of additional factors or the presence of undetected local cAMP increases. Recently, we have shown that the lag phase of growth in yeast is associated with a transient increase in the basal cAMP level. Also, this transient cAMP increase was abolished in the cyr1 ^met1876 (lcr1) mutant (Fig. 3E). This might be the reason for the slightly longer lag phase of the mutant. In addition to the rapid mobilization of trehalose, triggered by the activation of trehalase, mobilization of glycogen and rapid loss of stress resistance are also important characteristics of the adaptation from gluconeogenic/respirative growth to fermentative growth. These effects are also thought to be mediated by a cAMP-triggered protein phosphorylation cascade that occurs after the addition of glucose (Thevelein, 1988). We now show that, for each of these three characteristics, a significant delay occurs in the cyr1 met1876 (lcr1) mutant (Fig. 5). This is consistent with a role for glucose-induced cAMP signalling in the stimulation of the transition to fermentative growth.

Growth of the isogenic strains W303-1A and GD1 on different carbon sources showed a slightly, but reproducibly longer lag phase and higher final cell density on rapidly fermented sugars for the cyr1 ^met1876 (lcr1) mutant (glucose: Fig. 6A; fructose: results not shown) and a slightly shorter lag phase and lower final cell density on non-fermentable carbon sources (glycerol, ethanol: results not shown). On galactose, there was no difference between the two strains. These results would be consistent with a stimulatory role for glucose-induced cAMP signalling in the initiation of fermentative growth. Although the difference under our growth conditions was small (e.g. Fig. 6A), it cannot be excluded that, under certain conditions in the natural environment, this difference would be more pronounced and of selective advantage. The slightly longer lag phase of the cyr1 ^met1876 (lcr1) mutant correlated with a slightly slower consumption of glucose in the medium (Fig. 6A).

It is also well known that, during diauxic growth of yeast cells on glucose, a general fluctuation occurs in PKA-controlled properties. The trehalose and glycogen content and heat resistance drop during the initiation of fermentative growth to reach a minimum during mid-exponential phase, after which they increase again to reach a maximum during growth on ethanol and the subsequent stationary phase (for reviews, see Broach and Deschenes, 1990; Thevelein, 1992; 1994). Figure 6B-D shows that the initial and final levels of trehalose, glycogen and heat resistance were elevated in the cyr1 ^met1876 (lcr1) mutant, while the difference between the two strains was small at early/mid-exponential phase. More important, however, is the observation that, in the strain lacking glucose-induced cAMP signalling, the general fluctuation pattern in the three parameters during diauxic growth was not fundamentally affected. It was even more pronounced in the mutant than in the wild type. This indicates that glucose-induced cAMP signalling is not responsible for this general fluctuation in PKA-controlled properties.

Suppression of Ras2^val19 and Gpa2^val132 by the cyr1^met1876 (lcr1) mutation

The K1876M mutation in adenylate cyclase largely abolishes agonist-induced cAMP signalling

Our results show that a K1876M mutation in adenylate cyclase largely abolishes the cAMP responses in vivo upon addition of glucose or intracellular acidification, without significantly reducing the basal cAMP level measured in vivo. Also, the Mn²⁺-dependent activity of adenylate cyclase in vitro is not significantly affected, whereas GTP/Mg²⁺-dependent activity is strongly reduced. Hence, both in vivo and in vitro data indicate that the cyr1 ^met1876 (lcr1) mutation does not cause a general non-specific drop in the catalytic activity of adenylate cyclase, but rather that agonist stimulation of adenylate cyclase is specifically absent in the mutant enzyme. The most straightforward explanation of our observations is that the mutation specifically affects the regulation of the enzyme by the proteins responsible for agonist-induced activation. If this interpretation is correct, the data would tend to indicate that, in the case of glucose control and control by intracellular acidification, the proteins involved act on a similar site of adenylate cyclase or that the lysine at position 1876 plays a key role in the subsequent response of the adenylate cyclase catalytic domain to this interaction. This interpretation is supported by the finding that the effect of both Ras2^val19 and Gpa2^val132 can be counteracted by the K1876M mutation. An alternative explanation is that the K1876M mutation causes a conformation mimicking constitutive feedback inhibition of adenylate cyclase by PKA. The enzyme is known to be under strong feedback inhibition by PKA (Nikawa et al., 1987). The magnitude and persistence of the glucose-induced cAMP signal correlate negatively with PKA activity (Mbonyi et al., 1990), and high PKA activity abolishes both the glucose- and the acidification-induced cAMP responses (Ma et al., 1999). There are two arguments, however, against this alternative hypothesis. The first is that strains with constitutively high PKA activity (caused by a deletion of the BCY1-encoded regulatory subunit) have a lower cAMP level than the basal level in the cyr1 ^met1876 (lcr1) strains. The second is that the cyr1 ^met1876 (lcr1) mutation does not cause a general reduction in catalytic activity of adenylate cyclase but a reduction specifically in Mg²⁺/GTP-dependent activity.

Physiological significance of agonist-induced cAMP signalling

Up to now, it has been difficult to assess the physiological significance of glucose-induced cAMP signalling because of the lack of strains that specifically affected agonist-induced activation of cAMP synthesis rather than the basal capacity of cAMP synthesis. The results with the cyr1 ^met1876 (lcr1) strain GD1 show that the absence of glucose-induced cAMP signalling clearly affects the rapid changes in several PKA targets that are associated with the transition from gluconeogenic/respirative growth to fermentative growth. The activation of trehalase was reduced. Mobilization of trehalose and glycogen and loss of stress resistance were significantly delayed. These results support the conclusion that glucose-induced cAMP signalling plays a role in stimulating the adaptation to growth on glucose. They fit with the previous observations that deletion of either Gpa2 or Gpr1, which are both required for glucose-induced cAMP signalling, also delays glucose-induced changes in PKA targets (Colombo et al., 1998; Kraakman et al., 1999).

The effect of the cyr1 ^met1876 (lcr1) mutation was only quantitative and largely limited to the transition phase to fermentative growth. During diauxic growth, there was only a little difference from the wild-type strain. This fits with the presence of the cyr1 ^met1876 (lcr1) mutation in laboratory strains considered to be wild-type strains. The absence or, at least, the strong reduction in glucose-induced cAMP signalling does not eliminate the establishment of the typical PKA-controlled phenotype in glucose-growing cells. Moreover, the typical fluctuation in these properties during diauxic growth was even accentuated in the cyr1 ^met1876 (lcr1) mutant. This indicates that glucose-induced cAMP signalling is not responsible for the difference in these properties between cells growing on glucose and cells deprived of glucose. This is in agreement with the observation that, in gpa2 and gpr1 mutants also, the general fluctuation in PKA-controlled properties during diauxic growth is still present (Colombo et al., 1998; Kraakman et al., 1999). This conclusion also fits with the previous proposal that another glucose-dependent pathway, called the fermentable growth medium-induced (or FGM) pathway, is responsible for establishing the striking differences in PKA-controlled properties (Thevelein, 1994).

The cyr1 ^met1876 (lcr1) strain GD1 also displayed a small but reproducible increase in the length of the lag phase, which might be related to the absence of the transient increase in the basal cAMP level during the lag phase. This cAMP increase has only recently been linked specifically to the lag phase of growth, whereas it was previously considered to reflect the level of glucose in the medium (Ma et al., 1997). It might play a role in stimulating exit from stationary phase. A possible role of the cAMP pathway in outgrowth from stationary phase has been suggested (Tatchell, 1993).

Up to now, the physiological significance of acidification-induced cAMP signalling has been rather obscure. Its role might be limited to carbon-starved cells that are known to consume storage carbohydrates (Lillie and Pringle, 1980) and in which acidification is known to trigger the mobilization of storage carbohydrates (Berke and Rothstein, 1957). In carbon-starved cells of the cyr1 ^met1876 (lcr1) mutant strains, 2,4-dinitrophenol-induced trehalose mobilization was retarded compared with the wild-type strain, in agreement with a role for the acidification-induced cAMP increase in the stimulation of trehalose mobilization. Under the conditions of diauxic growth and absence of carbon starvation, as performed in our experiments, the changes caused by the cyr1 ^met1876 (lcr1) mutation in the PKA-controlled properties probably resulted from the absence of glucose-induced rather than acidification-induced cAMP signalling.

The cyr1^met1876 (lcr1) allele as a tool for studying the signalling function of the cAMP pathway

Conclusions

Strains and growth media

S. cerevisiae strains used in this study are shown in Table 1. Tetrads MV7103A-D and MV7106A-D were derived from a cross between SP1 and ENY.cat80-7A. Tetrad MV7117A-D was derived from a cross between SP1 and MV7103B. The presence of the can1 marker was not checked in these tetrads. MV7143 is a diploid strain obtained by crossing MV7117A and Be333. Strains MV7159 and MV7161 were obtained by transformation of MV7117C and MV7117D, respectively, with plasmid YCplac33YACE1. Strain GD1 was constructed by insertion of the vector YIplac211(URA3) containing an XbaI-BamHI fragment of the cyr1 ^met1876 (lcr1) mutant allele and cut with ClaI into strain W303-1A. Subsequently, the strain was plated on a 5-fluoro-orotic acid plate, and spontaneous ura revertants were screened for the presence of the A5627T mutation in the CYR1 gene.

Mapping strains obtained from the Yeast Genetic Stock Center were X4119-19C, X4119-15D, STX145-13D, STX145-15D, STX66-4A, STX82-3A, STX146-19A, STX153-10C, STX84-5 A, STX75-3C, STX147-9B, STX147-4C, STX83-17D, X4120-19D, X4126-6D, STX77-6C, STX155-3C and STX155-9B.

Plasmids

Plasmid pUC19* was constructed from pUC19 (Yanisch-Perron et al., 1985) by deletion of the SacI-SmaI fragment of the multicloning site. Plasmid pYACE1 (Feger et al., 1991) was kindly provided by O. Fasano (Palermo) and contains the CYR1 gene on an 8.47 kb Sau3A-Sau3A fragment in plasmid YEp26 (Broach et al., 1979). Plasmids YCplac33, YIplac128, YIplac204 and YIplac211 have been described by Gietz and Sugino (1988). Plasmid YCplac33YACE1 was constructed by insertion of the CYR1 gene on an 8.5 kb SphI-BamHI fragment derived from pYACE1 into YCplac33.

The YCplac33 plasmids containing different chimeric constructs of the wild-type CYR1 and lcr1 mutant allele were made by replacement of the SphI-Asp718, Asp718-NcoI or NcoI-BamHI fragment in the YCplac33YACE1 plasmid ('pWWW') by the corresponding fragments obtained from parts of the lcr1 mutant allele cloned in the vector YIplac211. These parts were a 5.2 kb SphI-PstI, a 4.9 kb PstI-NcoI and a 1.4 kb NcoI-SnaBI fragment. They were recovered separately by plasmid eviction. The chimeric constructs were first made in the pUC19* vector, cut out with SphI and BamHI and cloned into the SphI-BamHI sites of the multicloning site of vector YCplac33.

For insertion of a plasmid of the YIplac series at different positions in the CYR1 promoter of the wild-type strain SP1, the following plasmids were constructed and restriction sites used: YIplac211 + 2.2 kb HindIII-EcoRI fragment of CYR1, HindIII (1790); YIplac204 + 2 kb HincII-EcoRI fragment of CYR1, HincII (1590); YIplac128 + 1.25 kb AccI-EcoRI fragment of CYR1, AccI (840); and YIplac211 + 0.93 kb DraI-EcoRI fragment of CYR1, DraI (526).

Recombinant DNA methods and genetic analysis

Growth measurements and determination of heat shock resistance

Biochemical determinations

For determination of cAMP content, cells were grown on YPGlycerol until exponential phase, harvested and resuspended in 25 mM Mes buffer (pH 6). They were preincubated for 10 min before the addition of 100 mM glucose or 2 mM 2,4-dinitrophenol. Extraction and determination of cAMP were performed as described previously (Thevelein et al., 1987b).

For trehalose and glycogen determination, cells were collected by filtration, washed once with cold water, weighed and frozen in liquid nitrogen. The pellets were resuspended in 0.5 ml of 0.25 M Na₂CO₃ per 50 mg of cells and boiled at 95°C for 20 min. Samples were taken for trehalose determination, and the remainder was boiled for another 60 min. The samples for trehalose determination were spun down, and 10 l of the clear supernatant was used. The samples for glycogen determination were mixed well and used directly (10 l). All samples were neutralized by the addition of 5 l of 1 N acetic acid. For trehalose determination, 5 l of buffer (300 mM sodium acetate, 30 mM CaCl₂, pH 5.4) and 20 l of Humicola trehalase (360 U ml ¹) (Neves et al., 1994) were added and incubated for 45 min at 40°C. For glycogen determination, 5 l of buffer (400 mM sodium acetate, pH 4.7) and 20 l of amyloglucosidase (0.25 U) from Aspergillus niger (Boehringer Mannheim) were added and incubated for 2 h at 37°C. For all samples, the glucose liberated was measured in 30 l of cleared supernatant using the glucose oxidase/peroxidase method.

Adenylate cyclase activity was assayed in purified yeast plasma membranes. Preparation of crude membranes was performed essentially as described by Mintzer and Field (1995) with minor modifications. Adenylate cyclase was assayed in a reaction mixture containing 20 mM Mes, pH 6.2, 0.1 mM MgCl₂, 0.1 mM EGTA, 1 mM 2-mercaptoethanol, 0.25 mM IBMX, 0.1 mg ml ¹ BSA, 0.25 mM cAMP, 1 mM [-³²P]-ATP (final specific activity 100-150 c.p.m. pmol ¹), 20 mM phosphocreatine, 0.25 mg ml ¹ creatine phosphokinase, 0.1 mM dithiothreitol (DTT), in the presence of 2.5 mM MnCl₂ or 2.5 mM MgCl₂ or 2.5 mM MgCl₂ and 100 M Gpp(NH)p. The reaction was initiated by addition of the reaction mixture to the membranes (20-60 g of protein) in a final volume of 100 l. The tubes were immediately transferred to a 30°C water bath and incubated for 20 min. The reaction was stopped by the addition of 0.8 ml of 'stopping solution' containing 10 mM Tris-HCl, pH 7.5, 0.175 mM cAMP, 5 mM ATP and 0.25% SDS. Cyclic [³H]-AMP ( 20 000 c.p.m.) was added to monitor sample recovery. [³²P]-cAMP produced was determined as described by Salomon et al. (1974) with minor modifications.