3' :5'-Cyclic Monophosphate Alfred G

Proceedings of the National Academy of Sciences Vol. 67, No. 1, pp. 305-312, September 1970

A Protein Binding Assay for Adenosine 3' :5'-Cyclic Monophosphate Alfred G. Gilman

LABORATORY OF BIOCHEMICAL GENETICS, NATIONAL HEART AND LUNG INSTITUTE, NATIONAL INSTITUTES OF HEALTH, BETHESDA, MARYLAND 20014 Communicated by Marshall Nirenberg, June 24, 1970 Abstract. A simple and sensitive assay for adenosine 3':5'-cyclic monophosphate (cAMP) has been developed that is based on competition for protein binding of the nucleotide, presumably to a cAMP-dependent protein kinase. The nucleotide-protein complex is adsorbed on a cellulose ester filter. Assay conditions are such that a binding constant approaching 10-9 M\1 is obtained, and the assay is thus sensitive to 0.05-0.10 pmol of cAMP.

While assays for adenosine 3' :5'-cyclic monophosphate (cAMIP) are proliferat- ing almost as rapidly as are newly discovered actions of the nucleotide, the tremendous current interest in the biological role of cAMP dictates the need for a sensitive, accurate, and readily performed procedure for estimation of its cellular levels. The methods currently availablel-8 do not have the extreme sensitivity required by the low tissue levels of the compound, or else they are laborious to perform, or both. The present study was initiated with the thought that the binding of [5H]- cAMP to a cAMP-dependent protein kinase9-12 would form the basis for a sensitive assay, if a simple means could be found to isolate the protein-nucleotide complex. Further impetus was -provided by the discovery of Walsh et al.3 that a heat-stable protein, an inhibitor of the cAMP-dependent protein kinase,14"15 increased the affinity of the cyclic nucleotide for the enzyme. 'A protein kinase from muscle was chosen for investigation because of its favorable binding constant for cAMP.16 It was readily determined that cAMP-dependent protein kinase and cAMP-binding activities from muscle extracts could be quantitatively adsorbed on cellulose ester (Millipore) filters. The binding of cAMP to specific proteins from Escherichia coli and the adrenal cortex has recently been studied on Millipore filters.17'18 This simple means for estimating cAMP binding may be utilized for assay of unknown quantities of the cyclic nucleotide in deproteinized tissue extracts. As little as 0.05-0.10 pmol of cAMP can be detected; thus, less than milligram quantities of many tissues are sufficient for assay. Materials and Methods. cAMP-dependent protein kinase: Two purification schemes have been employed-a more extensive one to characterize the protein and a simplified method for routine use. Procedures are patterned after those of Walsh et at.9 and Miyamoto et al."' Fresh bovine muscle was prepared as described" through the ammonium sulfate precipitation step. This fraction, from 250 g of muscle, was applied to a column of DEAE-cellulose (Whatinan DE 11, 1 meq/g; 32 X 2.6 cm), previously equilibratcd with 5 mMI potassium phosphate, pH 7, and tile columni was 305 Downloaded by guest on September 30, 2021 306 BIOCHEMISTRY: A. G. GILMAN PROc. N. A. S.

s A

30 I -300'7 *2R43E II - -430 I FIG. 1. DEAE-cellulose chroma- K20 200 I tography of protein kinase and cAMP xOOZEg O DO < binding activities: gradient elution lo- HO'Ewasperformed4> it as described under c. k Methods. Binding activity was as- B o sayed at pH 4 with 40 nM cAMP. 4 I l81E Fractions were 20 ml. (A): 0, pro- s03 . \ 6 ° tein kinase activity; --- potassium E , -; phosphatei gradient. (B): 0, cAMP l X4o binding activity; --- optical density C,,, 0~~~~~~at 280 nm. I0 20 30 40 50 60 70 Fraction washed with the same buffer. Elution was with a linear gradient of potassium phos- l)hate, pH 7, from 5 to 400 mM, as shown in Fig. 1. Two prominent peaks of kinase and binding activity are seen.i6 Fractions 39-51, the second major peak of kinase and binding activity (DEAE II), were pooled and dialyzed against 5 mM potassium phosphate, pH 7. A portion of this preparation was concentrated by precipitation with ammonium sulfate (0.33 g/ml), dialyzed, and applied to a column of Sephadex G-200 (2.5 X 100 cm), equilibrated and eluted with 5 mM potassium phosphate, pH 7. For routine purposes a satisfactory preparation was obtained by applying the ammonium sulfate fraction1' to a column of D)EAE-cellulose aiid eluting the early peak(s) of activity with 100 mM potassium phosphate, pH 7. The last l)eak (D)EAE II) was theft collected with 300 mM potassium phosphate and was dialyzed against 5 mM potassium phosphate, pH 7. Such a preparation was utilized in this work and had an enzymatic activity of 24 pmol of 32P/,gg protein per 10 mil, and bound 0.3 l)nol of cAMP/Jug protein under standard assay conditions. Up to 200 /Ag protein of this l)relparatioii could be applied to a single Millipore filter with resultant quantitative removal of kinase, binding activity, and protein. When larger amounts were filtered, the kinase, binding activity, and protein were removed to a very similar extent. Both kinase and binding activity are stable for several months at -400C. Ptotein-kinase inhibitor: The inhibitor preparation was modeled after that of Appleman et al.', Bovine muscle was homogenized in 10 mM Tris chloride, pH 7.5 and was boiled for 10 min. After removal of particulate material by filtration, activity was precipitated with 1/9 volume of 50% trichloroacetic acid. The precipitate was collected at 15,000 X g and dissolved in water, and pH was adjusted to 7 with 1 N NaOH. This fraction was dialyzed against distilled water at room temperature, and the precipitate which formed was discarded. The preparation was used at this stage of purity, although it was verified that the kinase-inhibitory activity fractionated on Sephadex G-75 as previously described." Cyclic 3' :5'-nucleotide phosphodiesterase: This enzyme was purified from bovine cardiac muscle through the dialyzed acid-supernatant fraction of Butcher and Suther- land.19 Protein kinase assay: cAMP-dependent protein kinase was assayed essentially as described,ii although phosphorylated protein was precipitated with 8% trichloroacetic acid, and precipitates were collected on glass fiber filters (Millipore). cAMP binding and cAMP assay: The standard binding reaction was conducted in a volume of 50 or 200 ,l in 50 mM sodium acetate/acetic acid, pH 4.0 and incubated for longer than 60 min at 0C. The only other components of the incubation were [3H]- cAMP (Schwarz BioResearch, 16.3 Ci/mmole), sufficient binding protein to bind less than 30% of the nucleotide, and, where indicated, a maximally effective concentration of Downloaded by guest on September 30, 2021 VOL. 67, 1970 ASSAY FOR CYCLIC AMP 307

the protein-kinase inhibitor preparation. Reactions were initiated by addition of binding protein. At equilibrium, the mixtures were diluted to 1 ml with cold 20 mM potassium phosphate, pH 6; 4-5 min later they were passed through a 24-mm cellulose ester (Millipore) filter (0.45 Am) previously rinsed with the same buffer. The filter was then washed with 10 ml of this buffer and placed in a counting vial with 1 ml of Cellosolve, in which the filter readily dissolves. A scintillation mixture of toluene-Cellosolve (3:1) plus fluors was utilized, and efficiency was approximately 30% (10,750 cpm/pmol). Bowud counts were independent of filtration speed and the volume of rinse from 3 to 20 ml. The blank in the absence of binding protein was 20-50 cpm. For the assay of cAMP, [(H]cAMP was utilized at 10-20 nM (0.5-1.0 pmol/50 Al) in the presence of the inhibitor or 40 nM in its absence. These are saturating concentrations of cAMP, and the effect of added unknown or standard cAMP solutions could thus be evaluated from a linear, and nearly theoretical, decrease in the total bound [3H]- cAMP. Protein-kinase inhibitor assay: The inhibitor preparation may be assayed in the cAMP-dependent protein kinase reaction or by its enhancement of [3H]cAMP binding at submaximal concentrations of the nucleotide. Tissue extracts: Tissue samples were homogenized in 1 ml of 5% trichloroacetic acid, and supernatants were extracted 5 times with 2 volumes of ether after the addition of 0.1 ml of 1 N HCl. The eltracts were then dried and redissolved in 50 mM sodium acetate, pH 4. Results. The choice of the DEAE peak (Fig. 1) of binding and protein kinase activity to use in this assay (DEAE II) was made because of its apparent greater purity and because of a slightly greater affinity for cAl\IP (data not shown). The first peak of activity (DEAE I) had the advantage of little de- pendence of binding affinity and total binding sites on pH. An additional small peak(s) of binding and kinase activity between DEAE peaks I and II may be seen in Fig. 1; it was not investigated further. Proof that the binding protein is in fact the cAl\IP-dependent protein kinase must await further purification of the protein. However, cochromatography of the two activities on DEAE-cellulose and Sephadex G-200 (data not shown) supports the hypothesis that they are properties of a single molecule. Cochro- matography on Sephadex G-200 was also apparent when the binding protein was labeled with [3H ]cAMP before gel filtration. During initial purification (Table 1), the specific activity of the protein kinase increased somewhat more than that for binding, perhaps because of removal of inhibitors of kinase activity" or additional cAMP-binding proteins. However, Sephadex G-200 chromatography yielded Quantitatively similar enrichment of the two activities from the DEAE II fraction.20 The possible complexity of the relationship between the nucleotide-binding and kinase activities is indicated in a recent report by Gill and TABLE 1. Protein kinase and cAMP-binding activity purification. Protein kinase (pmol/'ug cAMP binding* Fraction protein/10 mm) (pmol/pg protein) 15,000 X g supernatant 0.58 0.012 pH 4.8 supernatant 0.62 0.011 (NH4)2 S04 precipitate 1.94 0.031 DEAE I 2.75 0.036 DEAE II 17.7 0.191 Sephadex G-200 41.0 0.495 * Assayed at pH 4.0 with 40 nM cAMP. Downloaded by guest on September 30, 2021 308 BIOCHEMISTRY: A. G. GILMAN PROC. N. A. S.

Garren,18 who have independently discovered a procedure similar to the one reported here for assessing binding activity. The effect of pH on the binding of sub- and supra-maximal concentrations of cAMP to the protein is shown in Fig. 2. A very high affinity is apparent at pH 4, and, in addition, an effect of pH on the total number of available sites is evident. This latter fact would seem to indicate heterogeneity of binding sites on one (or more than one) protein in the system. The presence of the inhibitor fraction did not change the pattern of results, although more total binding sites are available in its presence. An explanation for this phenomenon will be offered below. The inhibitor preparation itself did not bind cAMP to the filter. A similar decrease in binding affinity and no change in total binding sites are seen from pH 5 to 7 in phosphate buffer (data not shown), although affinity was higher in phosphate than in acetate at overlapping points. No differences were seen with Na+ or K+ as the monovalent cation. Data from which the binding constant may be evaluated are presented in Fig. 3. While the binding constant is 10-20 nM at pH 6 (phosphate), a

0 2~~~~~~~~~~

~~~~060~ ~ ~ ~ ~ ~~< P1

0 40 45 50 55 -10 -08 -06 -0.4 -0.2 0-0802 04 06 10 pH I/nM cAMP FIG. 2. FIG 3.

FIG. 2. Effect of pH on cAMP binding activity: binding protein was present at 2 ,ug/'2(0 Ml and the inhibitor fraction was absent (circles) or present (squares) at 45 jsg/200 ad. Total cAMP was 1 pmol (open symbols) or 28 pmol (solid symbols). Buffers were 50 mM sodium acetate-acetic acid. FIG. 3. Estimation of cAMP binding constant: binding was determined at pH 6 (50 mMI potassium phosphate) (0) and at pH 4 in the absence ((D) or the presence (A) of 45 ,g inhibitor fraction protein/200 jy1. Binding protein concentration was 2 pg/200 Ml. marked increase in affinity (Ka = 2-3 nM) is seen at pH 4. The inhibitor prep aration increased the affinity approximately 2 times. The kinetics of the attainment of equilibrium at 00C are shown in Fig. 4A. A 60-mm incubation was sufficient at all concentrations examined. In the absence of the inhibitor fraction, a slow decline (I5%/hr) in cAMP bound was evident at 20 nM (data not shown) and 40 nM concentrations. This effect was markedly accentuated30oat C and appeared to be due to protein denaturation. The addition of the inhibitor preparation resulted inl complete stability of the plateau at 0aC. Downloaded by guest on September 30, 2021 VOL. 67, 1970 ASSAY FaR CYCLIC AMP 309

1.2 A1. 0 ~ ~ ~~~40nM .n~n~~o

0oPXm aM<-'/08 0 2;AA- 20 nM Inhibitor 40 A 082 e~°° ° Q0.8- / <40.60 VI_~0 - 0 i 0.3 5 nM E Z o/.-",~ lo5MImm Mg E00.2(

0306012050240 ~~~~~0.10 160 120 180240'360 Time (mm) Time (min) FIG. 4A. Kinetics of establishment FIG. 4B. Release of bound [PHI- of cAMP-binding equilibrium: 200-1l cAMP: 2.3 ug/200 ;l of binding pro- aliquots of a reaction mixture were fil- tein was labeled' for 90 min at pH 4 tered without dilution at the indicated and 00C with ['H] cAMP at the fol- times. Conditions used were 50 mM lowing concentrations: 0, 5 nM; *, acetate pH4 at00Cwith5nM(0),40 5 nM + 10 mM Mg++; ID, 40 nM; nlM (81), and 20 nM cAMP plus 45 pug A, 30 nM + 45 pg inhibitor fraction inhibitor fraction protein/200 ,ul (A). protein/200 Al. At "zero time" un- Binding protein was present at 2.3 labeled cAMP was added to achieve a Ag/200,jl. final concentration of 0.25 mM, and 200-As aliquots were filtered without dilution at the indicated times. The reverse reaction was studied both by dilution (data not shown) and by the addition of a large excess of unlabeled cAMP (Fig. 4B) with similar results. In both cases, in the absence of the inhibitor, approximately 10-20% of the total bound [3H]cAMP was released with great rapidity (<1 min) and was relatively independent of the extent of dilution. This mysterious phenomenon was not seen when the inhibitor fraction was present and may be the explanation for the greater number of binding sites seen when the inhibitor is included in the standard assay (where reaction mixtures are diluted prior to filtration). The reverse reaction was first order, with a half-time of approximately 7 hr at 00C, in the absence of the inhibitor and was somewhat slower in its presence. While Mg++ was not required for binding, and had little effect on the binding constant at pH 4, it did significantly increase the forward and reverse reaction rates. The latter effect is shown in Fig. 4B. Table 2 demonstrates the effect of nucleotides and related compounds on cAMP binding. As might be expected, other 3':5'-cyclic nucleotides are most effective. However, mammalian tissue levels of cGMP are not sufficiently high to interfere.21'22 Of great pragmatic significance is the fact that ATP had only TABLE 2. Effect of nucleotides and related compounds on cAMP binding. MM Concentration at % Inhibition Compound 50% inhibition* Compound at 1 mM* CIMP 0.3 UTP 30 cGMP 5.0 CTP 28 CUMP 10 5'-AMP 21 CCMP 30 ADP 18 GTP 700 Adenosine 0 ATP 1000 Theophylline 0 * cAMP concentration = 40 nM; binding protein = 2 ug/200 Al. Downloaded by guest on September 30, 2021 310 BIOCHEMISTRY: A. G. GILMAN PROC. N. A. S.

a 50% inhibitory effect at 1 mM and virtually no effect at 0.1 mM. Effects of these nucleotides in tissue extracts thus should not be seen if such extracts are assayed at 10- to 50-fold dilutions (depending on the tissue). Similar effects of nucleotides were observed when examined with 20 nM cAM\1P in the presence of the inhibitor preparation. Standard curves obtained with saturating concentrations of [3H]cAMP are presented in Fig. 5. Total picomoles of cAMP plotted on the abscissa represents

6000- 4000 < FIG. 5. Standard curves for cAMP assay: all 2000 v reactions were carried out at pH 4 and 00C in a volume of 50 ,l. [OH] cAMP added/tube was 0, 0.5 pmol; 8, 1.0 pmol; A, 2.0 pmol. 14 ,ug 600\ protein of the inhibitor fraction was present at the D0 two lower cAMP concentrations, and binding pro- E tein was added at 0.5, 1.0, and 2.0 yg for the three 20 conditions respectively. Known quantities of cAMP were added to achieve the total (labeled l0om . plus unlabeled) indicated content of cAMP/tube. 0.S 1.0 2.0 30 5,~07010i 20 Total pmoles cAMP/tube

the sum of [3H]cAM1P and unlabeled standard added/tube. Thus, a variety of curves can be generated, depending on the sensitivity required. The plots are straight lines on a logarithmic plot as a result of the assay being run at satura- tion, and they are very close to the theoretical slope predicted. With the most sensitive curve shown, a significant dilution of total cpm bound is obtained with the addition of 0.10 pmol. Reduction of the volume and/or the amount of t'H]cAMP added by a factor of 2 (which would still be saturating) will result in an assay sensitive to 0.05 pmol/tube, and one in which the specific activity of the [5H]cAMP is becoming the limiting factor. Represeptative assay data are shown in Table 3, and tissue levels obtained are in reasonable agreement with those in the literature.23 More importantly, the TABLE 3. cAMP levels in mouse liver and brain. cAMP Tissue Sample and conditions (pimol/mg wet wt) Liver* 2 mg 1.0 10 mg 0.9 10 mg + 1 pmole/mg cAMP 2.0 10 mg + phosphodiesterase 0.0 Braint '' mg 2.1 10 mg 2.1 10 mg + 1 pmole/mg cANIP 3.2 Brain* 15 sec decapitation 2 mg 6.8 10 mg 6.3 10 mg + 1 pmole/mg cAMP 7.3 10 mg + phosphodiesterase 0.0 Brain* 120 see decapitation 5 mg 18 * Tissues were removed from decapitated animals and frozen in liquid nitrogen. t Mice were killed by immersion in liquid nitrogen and frozen brain tissue was dissected. Downloaded by guest on September 30, 2021 VOL. 67, 1970 ASSAY FOR CYCLIC AMP 311

independence of the values from the amount of tissue analyzed (with final tissue dilutions in the assay of 25- to 250-times) and the results of cyclic 3':5'-nucleotide phosphodiesterase treatment indicate the specificity of the procedure. Finally, known amounts of cAMP added at the time of homogenization were quantitatively recovered. Discussion. The advantages of the procedure described above seem significant. The sensitivity of the assay is high, tissue purification seems unnecessary, and the actual assay is very simple to perform. Assay conditions are such that destruction of cAMP is not a factor. Preparation of the two protein fractions required appears to present no problems, and a preparation of each from 1 kg of muscle yields sufficient material for hundreds of assays. All work required to obtain a functional assay can be performed within a few days. Several further advantages arise from the assay being performed at saturating concentrations of [3H]cAMP. Thus, standard curves are linear throughout and are nearly theoretically predictable. Such potential variables as total volume become less critical, and potential interference from competing materials is minimized. There are also advantages to be gained from the inhibitor preparation, even when the modest increase in sensitivity it provides is not required. The increased stability of the equilibrium plateau and the altered characteristics of the reverse reaction are favorable conditions which probably contribute to the improved and excellent replication seen with inhibitor. Routine use of this preparation is thus recommended, despite the lack of appeal in using this crude fraction. While there was no apparent need to purify tissue extracts to remove interfering materials in the tissues investigated, it is possible that some tissues will prove (when appropriate controls are performed) to require such treatment. Exogenous interfering compounds may also be added. In such cases where purification (and loss) of cAMP from tissue is necessary, the addition of [3H]- cAMP (or other label) to the tissue extract, as in other assays, allows use of this methodology. The recent discovery24 of protein kinases more sensitive to cGMP suggests that a similar assay may be possible for this nucleotide. However, examination of the relative effectiveness of cGMP and cAMP to stimulate these enzymes, and the relative tissue levels of the two nucleotides, suggests that their separa- tion from each other will be essential prior to such an assay. The author is grateful to Drs. N. W. Seeds, S. Wilson, and M. W. Nireiiberg (ill whose laboratory this work was conducted) for helpful advice and criticism. 1 Butcher, R. W., R. J. Ho, H. C. Meng, and E. W. Sutherland, J. Biol. Chcm., 240, 4515 (1965). 2 Kakiuchi, S., and T. W. Rall, Mol. Pharmacol., 4, 367 (1968). 3 Breckenridge, B. M., Proc. Nat. Acad. Sci. USA, 52, 1580 (1964). 4Goldberg, N. D., J. Larner, H. Sasko, and A. G. O'Toole, Anal. Biochem., 28, 523 (1969). 5 Aurbach, G. D., and B. A. Houston, J. Biol. Chem., 243, 5935 (1968). 6 Brooker, G., L. J. Thomas, Jr., and M. M. Appleman, Biochemistry, 7, 4177 (1968). 7Steiner, A. L., D. M. Kipnis, R. Utiger, and C. Parker, Proc. Nat. Acad. Sci. USA, 64, 367 (1969). 8 Wastila, W. B., J. T. Stull, S. E. Mayer, and D. A. Walsh, Fed. Proc., 29, 479 (1970). 9 Walsh, D. A., J. P. Perkins, and E. G. Krebs, J. Biol. Chem., 243, 3763 (1968). 10 Langan, T. A., Science, 162, 579 (1968). 1 Miyamoto, E., J. F. Kuo, and P. Greengard, J. Biol. Chem., 244, 6395 (1969). Downloaded by guest on September 30, 2021 312 BIOCHEMISTRY: A. G. GILMAN PROC. N. A. S.

12Kuo, J. F., and P. Greengard, Proc. Nat. Acad. Sci. USA, 64, 1349 (1969). 13 Walsh, D. A., personal communication. 14 Posner, J. B., K. E. Hammermeister, G. E. Bratvold, and E. G. Krebs, Biochemistry, 3, 1040 (1964). "IAppleman, M. M., L. Birnbaumer, and H. N. Torres, Arch. Biochern. Biophys., 116, 39 (1966). 16 Reimann, E. M., and D. A. Walsh, Fed. Proc., 29, 601 (1970). 17 Kuwano, M., and D. Schlessinger, Proc. Nat. Acad. Sci. USA, 66, 146 (1970). '8Gill, G. N., and L. D. Garren, Biochem. Biophys. Res. Commun., 39, 335 (1970). 19 Butcher, R. W., and E. W. Sutherland, J. Biol. Chem., 237, 1244 (1962). 10 Calculation of overall purification of the individual fractions is dependent on an estimation of their relative abundance in the crude extract. 21 Goldberg, N. D., S. B. Dietz, and A. G. O'Toole, J. Biol. Chem., 244, 4458 (1969). 22 Ishikawa, E., S. Ishikawa, J. W. Davis, and E. W. Sutherland, J. Biol. Chem., 244, 6371 (1969). 23 The "basal" value of 2.1 pmol/mg for brain may be high and appears to be explicable in terms of a generous proportion of cerebellar tissue in these samples. T. W. Rall, personal communication. 24 Kuo, J. F., and P. Greengard, J. Biol. Chem., 245, 2493 (1970). Downloaded by guest on September 30, 2021