-
REVIEW /
SYNTHGSE
Cross-activation: overriding cAMP/cGMP selectivities of protein kinases in tissues1 HANGJIANG,JOHNB. SHABB,AND JACKIE D.
CORBIN~
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
Department of Molecular Physiology and Biophysics, Vanderbilt University School of Medicine, Nashville, TN 3 7232-0615, U.S.A. Received March 25, 1992 JIANG,H., SHABB,J. B., and CORBIN,J. D. 1992. Cross-activation: overriding cAMP/cGMP selectivities of protein kinases in tissues. Biochem. Cell Biol. 70: 1283-1289. CAMP-and cGMP-dependent protein kinases are homologous proteins and are predicted to exhibit very similar threedimensional structures. Their cyclic nucleotide binding domains share a high degree of amino acid sequence identity. CAMP- and cGMP-dependent protein kinases are activated relatively specifically by cAMP and cGMP, respectively; and a single alanine-threonine difference between CAMP- and cGMP-binding domains partially accounts for this specificity. Thus, it would be expected that cAMP and cGMP mediate separate physiological effects. However, owing in part to the lack of absolute specificity of either enzyme and to the relatively high level of cAMP or cGMP in certain tissues, it is also possible that either cyclic nucleotide could cross-activate the other kinase. Increases in either cAMP or cGMP cause pig coronary artery relaxation. However, only cGMP-dependent protein kinase specific cyclic nucleotide analogues are very effective in causing relaxation, and cAMP elevation in arteries treated with isoproterenol or forskolin activates cGMP-dependent protein kinase, in addition to CAMP-dependent protein kinase. Conversely, increases in either cAMP or cGMP cause C1- secretion in T-84 colon carcinoma cells, and the cGMP level in T-84 cells can be elevated sufficiently by bacterial enterotoxin to activate CAMP-dependent protein kinase. These results imply specific regulation of CAMP-and cGMP-dependent protein kinases by the respective cyclic nucleotides, but either cyclic nucleotide is able to cross-activate the other kinase in certain tissues. Key words: cGMP, CAMP, smooth muscle relaxation, protein phosphorylation. JIANG,H., SHABB,J. B., et CORBIN,J. D. 1992. Cross-activation: overriding cAMP/cGMP selectivities of protein kinases in tissues. Biochem. Cell Biol. 70 : 1283-1289. Les proteine kinases dipendantes du cAMP et du cGMP sont des proteines homologues et elles devraient posstder des structures tridimensionnelles trks semblables. Leurs domaines de liaison des nucleotides cycliques ont des sequences d'acides aminks presque identiques. La proteine kinase dependante du cAMP et la protkine kinase dependante du cGMP sont respectivement activees de facon assez specifique par le cAMP et le cGMP; cette specificite s'explique partiellement par une seule difference alanine-thrkonine entre les domaines de liaison du cAMP et du cGMP. Ainsi, le CAMP et le cGMP pourraient exercer des effets physiologiques distincts. Toutefois, Ctant donne le manque de specificit6 absolue des deux enzymes et le taux relativement eleve de cAMP et de cGMP dans certains tissus, il est Cgalement possible que chacun des nucleotides cycliques active l'une ou I'autre kinase. Des augmentations du cAMP ou du cGMP provoquent la relaxation des artkres coronaires chez le porc. Cependant, seuls les analogues des nucltotides cycliques spkcifiques de la proteine kinase dependante du cGMP sont trks efficaces pour causer la relaxation et l'klevation du cAMP dans les artkres traitkes avec 17isoprotCrCnolou la forskoline active la proteine kinase dependante du cGMP en plus de la proteine kinase dependante du CAMP.Reciproquement, des augmentations du CAMPou du cGMP entrainent la sicretion de C1 dans les cellules T-84 du carcinome du c6lon et le taux de cGMP dans les cellules T-84 peut 2tre suffisamment eleve par 17entCrotoxinebactkrienne pour activer la proteine kinase dependante du CAMP. Ces resultats impliquent la regulation specifique de la proteine kinase dependante du cAMP et de la proteine kinase dipendante du cGMP par leurs nucleotides cycliques respectifs, mais chacun des nucleotides cycliques est capable d'activer l'autre kinase dans certains tissus. Mots elks : cGMP, CAMP, relaxation des tissus, phosphorylation des proteines. [Traduit par la redaction]
Cross-talk is a popular expression that refers to cross-over between two or more second messenger pathways. For example, an elevation in cell calcium levels might cause a decrease in cAMP through activation of the calciumstimulated phosphodiesterase (Mattson and Spaziani 1986). Cross-activation is defined as a special form of cross-talk in which a ligand that is highly specific for a particular recepABBREVIATIONS: cAK, CAMP-dependent protein kinase; cGK, cGMP-dependent protein kinase; R and C subunits, regulatory and catalytic subunits of CAMP-dependent protein kinase; CAP, catabolite gene activator protein; BPDE, cGMP-binding phosphodiesterase. his review is based on a talk and abstract presented at the 34th Annual Meeting of the Canadian Federation of Biological Societies. ' ~ u t h o rto whom all correspondence should be addressed. Printed in Canada / lmprimh au Canada
tor interacts with another receptor that is highly specific for a different ligand under physiological conditions. This review will focus on cAMP and cGMP cross-activation of the cyclic nucleotide dependent protein kinases. To better understand how cross-activation of protein kinase can occur in vitro or in vivo, one should first consider effector specificity of cyclic nucleotide dependent protein kinases. It has been assumed that cAK is quite specifically activated by its effector, CAMP,whereas cGMP specifically activates cGK (Lincoln and Corbin 1983). This phenomenon has been seen in many tissues, where the activation state of cAK or cGK is usually associated with the elevation of its corresponding cyclic nucleotide, cAMP or cGMP (Beavo et al. 1974; Ogreid and Doskeland 1982; Lincoln 1983; Lincoln and Corbin 1983; Fiscus, e t al. 1985). The cyclic nucleotide specificity of protein kinases serves as the
1284
BIOCHEM. CELL BIOL. VOL. 70. 1992
CAMP-DEPENDENT PROTEIN KINASES
DIM^
INH
1
CAMP
1
I
CAMP
SITE 2 SITE 1 (FAST SITE) (SLOW SITE)
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
cGMP-DEPENDENT PROTEIN KINASES
CGMP
I CGMP I
CATALYTIC
I
CATABOLITE GENE ACTIVATOR PROTEIN FIG. 1. Linear domain structures of some cyclic nucleotide binding proteins. DIM, dimerization domain; INH, inhibitory domain; DNA, DNA binding domain.
nucleotide by certain hormones might cause the crossactivation of the other protein kinase. Moreover, the specificity of the protein kinases may be lessened by certain modifications of the kinases. For example, it has been shown that autophosphorylation of cGK (type Ia) improved its affinity for cAMP activation by 6- to 10-fold (Foster et al. 1981; Landgraf et al. 1986). In general, though effector specificity of protein kinases provides a mechanism for defined and precise regulation, cross-activation of these enzymes would permit some flexibility in the regulation of enzyme activities, and it might reflect an opportunistic mechanism of nature in using an additional second messenger to regulate a pathway that is already in place. This review is divided into two parts. The first part deals with the effector specificities of cyclic nucleotide dependent protein kinases, based on the comparison of cAMP and cGMP binding domains. The second part provides background information and some results to suggest that cross-activation of cyclic nucleotide dependent protein kinases can actually occur in certain intact tissues under physiological situations.
FIG. 2. Schematic representation of the structure of a generic cyclic nucleotide binding site of cAK or cGK, modeled by analogy with the crystal structure of CAP containing bound CAMP. Solid black oval, cAMP or cGMP. Invariant amino acids are indicated. The STA residue is a S in cGK, T in CAP, and A in cAK binding sites. (From Shabb and Corbin 1992.)
means for precise regulation, allowing the fluctuation of only one second messenger to turn a specific regulatory pathway on or off. However, effector specificity of protein kinases may not be absolute for several reasons. cAK and cGK are homologous proteins, with structural similarities between cAMP binding sites of cAK and cGMP binding sites of cGK (Weber et al. 1987, 1989). The similarities allow CAMP activation of cGK or cGMP activation of cAK in vitro. Since the level of one cyclic nucleotide in certain cells is severalfold higher than that of the other one (Francis et al. 1988; Forte et al. 1992), the elevation of this higher basal level of cyclic
Specificity of cyclic nucleotide dependent protein kinases It has been proposed that many effects of CAMP and cGMP are mediated by cAK and cGK, respectively, though other receptors also exist (Krebs 1972; Mackenzie 1982; Shabb and Corbin 1992). Although either cyclic nucleotide can cross-activate the other enzyme in vitro, cAK exhibits > 50-fold higher affinity for cAMP than for cGMP, whereas the affinity of cGK for cGMP is > 50-fold higher than that of cAMP (Doskeland et al. 1983; Corbin et al. 1986; Landgraf et al. 1986). Despite the fact that the two enzymes specifically and preferentially bind different cyclic nucleotides, which is the key issue to be addressed, mammalian cAK and cGK show a high degree of amino acid sequence similarity. The free C subunit of cAK is homologous to the catalytic domain (fused to a regulatory domain) of cGK, and both of them also show homology with all other protein kinases (Corbin et al. 1978; Hanks et al. 1988). As shown by a comparison of linear domain structures (Fig. 1) for cyclic nucleotide binding proteins, each of the two intrasubunit cyclic nucleotide binding sites of cAK and cGK shares amino acid sequence homology with the cAMP binding domain of the bacterial CAP (Weber et al. 1982, 1987; Takio et al. 1984). Recently, the cyclic nucleotide gated
REVIEW / SYNTHESE
CAMP-BINDINGSlTE (CAW
cGMP-BINDING SlTE
(cGK)
dNH'
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
Thr
0 NH YNH2 NH2+
0
FIG. 3. Predicted protein - cyclic nucleotide contacts for cAK and cGK. The sites shown represent either the slow or fast site of the respective kinase.
cation channels have also been proposed to be within this same family of proteins (Kaupp et al. 1989; Ludwig et al. 1990; Dhallan et al. 1990; Altenhofen et al. 1991). Because of this sequence homology and the known crystal structure of the CAP protein containing bound cAMP (McKay and Steitz 1981; McKay et al. 1982; Weber and Steitz 1987), modeling of cyclic nucleotide binding sites in cAK, cGK, and the rod photoreceptor cGMP-gated ion channel has been achieved (Weber et al. 1987, 1989; Kumar and Weber 1992). The models have extended the research in this field by allowing predictions of functional importance of specific amino acid residues tht could be subsequently tested by site-directed mutagenesis (e.g., Bubis et al. 1988a, 1988b; Kuno et al. 1988). Figure 2 shows the predicted tertiary structure of the fast site of cGK, which is composed of three a helices (Aa, Ba, and Ca) and one antiparallel /3-roll structure. The cGMP molecule is located at the bottom of the internal hydrophobic pocket formed by the 0-barrel and C a helix. This pattern of secondary structure and tertiary folding is retained in all of the cyclic nucleotide binding domain models of cAK and cGK. In fact, no gross differences in the overall structures are predicted to account for the cyclic nucleotide specificities towards their corresponding kinases (Weber et al. 1989). Since the cAMP and cGMP molecules differ mainly in structure of the appended groups on the purine ring, it has been suggested that the differences in affinity between CAMP and cGMP binding to the corresponding kinases could be due to different amino acid contacts with these different groups (Weber et al. 1989). Thus, the amino acids that are very close to the bound cyclic nucleotide, forming the binding pocket, have been examined for features that might explain the specificity of cyclic nucleotide binding. Several specific contacts between cyclic nucleotides and the binding pockets are predicted to be the same for cAK and cGK, based on amino acid sequence comparison of CAP, cAK, and cGK cyclic nucleotide binding domains. As a matter of fact, three amino acids in the binding pockets of cAK and cGK are found to be invariant. According to the models, an arginine, glycine, and glutamic acid make contacts with the ribose and phosphate moieties of the cyclic nucleotides (Fig. 3) (Weber et al. 1989). Results of mutagenesis of these residues in cAK (Steinberg et al. 1987;
Ogreid et al. 1988; Bubis et al. 1988a; Woodford et al. 1989) and cyclic nucleotide analogue studies in both cAK and cGK (Yagura and Miller 1981; De Wit et al. 1984; Corbin et al. 1986; Butt et al. 1990) are consistent with the importance of these residues in interaction with either CAMPor cGMP. The binding of cAMP or cGMP in the corresponding binding pocket requires the conserved ionic interaction between one of the exocyclic phosphate oxygens of the cyclic nucleotide and the arginine. In addition, a series of hydrogen bonds are formed between the ribose phosphate moiety of the cyclic nucleotide and the surrounding amino acid residues in the binding pocket. For example, the ribose 2'OH forms hydrogen bonds with the conserved Glu and Gly (Weber and Steitz 1987). Disruption of these hydrogen bonds by either mutation at the invariant residues or by using cyclic nucleotide analogues is thought to explain the loss or deterioration of cyclic nucleotide binding. However, it has been predicted, based on the CAP model, studies of CAMPanalogue binding, and amino acid sequence comparisons of the binding sites, that one residue in the binding pocket is critical for determining cAMP versus cGMP binding specificity of cAK and cGK. This residue is an Ala in all CAMP binding sites of RI and RII subunits, but the corresponding residue is a Thr in all cGMP binding domains (Fig. 3) (Weber et al. 1989). It has also been suggested from cGMP analogue studies that cGMP is bound in the syn conformation, which would allow a hydrogen bond to form between the 2-NHz group of guanine and the OH group of Thr in both cGK binding sites (Weber et al. 1989). This hydrogen bond cannot occur in the cAMP binding sites of cAK, since there is no such OH group on Ala corresponding to the position of Thr in the cGMP binding pocket. Studies by Shabb et al. using mutagenesis suggest that the threonine OH group is the binding site component which enhances cGMP binding affinity (Shabb et al. 1990, 1991). The wild-type and mutant bovine RIa subunits have been expressed in Escherichia coli and the holoenzyrnes have been formed by recombination of the purified C subunits and the expressed R subunits. A site 1 (slow site) mutant of cAK, in which Ala-334 is changed to Thr using oligonucleotidedirected mutagenesis, has a marked increase in cGMP
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
1286
BIOCHEM. CELL BIOL. VOL. 70, 1992
FIG. 4. Activation of cGK (m) and cAK (0) by isoproterenol in pig coronary arteries. Tissue segments were incubated with various concentrations of isoproterenol for 4 min at 37°C before preparation of extracts and assay of cGK and cAK activities. Data are calculated as fold stimulation of the control (no isoproterenol). Results are mean SE from four separate experiments done in duplicate. *, p < 0.05 (one-tailed t-test); **, p < 0.01 (one-tailed t-test); and ***, p < 0.01 (two-tailed t-test), comparing control and isoproterenol-treated tissues. (From Jiang et al. 1992.)
*
affinity but no change in cAMP affinity, as determined by cyclic nucleotide binding and protein kinase activation assays. The corresponding site 2 (fast site) mutant, created by changing Ala-210 to Thr, exhibits an increased affinity for cGMP and a decreased affinity for CAMP. A double mutant containing changes in both the slow and fast sites has an improved cGMP versus cAMP selectivity of -200 fold in comparison with that of wild-type cAK, making the double mutant about equal to native cGK regarding cGMP binding affinity. Thus, a protein kinase with high affinity for cGMP is produced from cAK by making a single amino acid change from Ala to Thr in each of the two cAMP binding pockets. Therefore, this single mutation of Ala to Thr, which requires only a single base change, is a key evolutionary development in the divergence of cAMP and cGMP binding domains (Weber et al. 1989). The Ala-Thr difference also explains cAMP versus cGMP binding selectivity of the related cyclic nucleotide gated ion channels, which contain a Thr at the corresponding position and have higher affinity for cGMP than for cAMP (Altenhofen et al. 1991). Cross-activation of cyclic nucleotide-dependent protein kinases Although cAK and cGK are rather specifically activated by their respective cyclic nucleotides, owing mainly to the Ala-Thr difference in their cyclic nucleotide binding pockets, the specificity might not be sufficient to eliminate crossactivation in some cells. For example, elevation of cAMP in certain tissues might actually cross-activate cGK to produce the same effect as elevation of cGMP. Smooth muscle is a likely tissue in which cross-activation may occur. It is well known that elevation of either cGMP or cAMP can cause smooth muscle relaxation (Hardman 1984; Lincoln and Corbin 1983; Hofmann et al. 1992), but it is uncertain if one or both of the cyclic nucleotide dependent
protein kinases can mediate these cyclic nucleotide actions. Recently, Francis et al. have examined this question by comparing the potencies of cyclic nucleotide analogues in relaxing pig coronary arteries and guinea pig tracheal smooth muscle with their abilities to activate the purified cAK and cGK. cGK-specific cyclic nucleotide analogues, but not cAKspecific analogues, cause muscle relaxation, suggesting that cAK is not the mediator of cAMP effects in these tissues (Francis et al. 1988). One possible explanation is that protein phosphorylation catalyzed by cGK, but not by cAK, mediates smooth muscle relaxation caused by elevation of either cGMP or CAMP. This would also explain the recent observation that 8-Br-CAMPcauses activation of cAK in rat vas deferens without producing smooth muscle relaxation (Hei et al. 1991). Lincoln et al. have reported that restoration of cGK, but not of cAK, to cGK-depleted aortic smooth muscle cells restores the calcium-lowering effects elicited by cGMP or CAMP.These workers have also hypothesized that cGK mediates the relaxing effects of both cyclic nucleotides (Cornwell and Lincoln 1989; Lincoln et al. 1990; Lincoln and Cornwell 1991). Considering all of the findings mentioned above, it is possible that elevation of cAMP in smooth muscle may cross-activate cGK to produce relaxation. cAMP levels are - 5-fold higher than cGMP levels in some vascular smooth muscle tissues (Francis et al. 1988), and autophosphorylation of cGK improves its affinity for cAMP by 6- to 10-fold (Foster et al. 1981; Landgraf et al. 1986), making it more readily activated by cAMP than is the unphosphorylated form of cGK. Therefore, under some conditions, small elevations of cellular cAMP could cross-activate cGK, even though the affinity of unphosphorylated cGK (type ICY or Ip) for cAMP is -50-fold lower than the affinity of this enzyme for cGMP. Although the idea of cross-activation has been around for many years, the direct experimental evidence showing the occurrence of cross-activation of cGK by cAMP in an intact tissue has been obtained only recently. The activity ratios of cAK and cGK have been measured in extracts of pig coronary arteries after treating the tissues with CAMP-and cGMP-elevating agents (Jiang et al. 1992). This study has been facilitated by the development of a substrate (BPDEtide) that is phosphorylated better by cGK than by cAK (Colbran et al. 1992). The peptide is based on the amino acid sequence of the phosphorylation site of the bovine lung BPDE (Thomas et al. 1990; Colbran et al. 1992). The use of BPDEtide has improved the specificity of the assay for detecting the activation of cGK, as compared with the heptapeptide substrate, which has been previously used as a relatively specific substrate for cGK versus cAK (Glass and Krebs 1982). Using this assay, it is revealed (Fig. 4) that the CAMP-elevatingagent isoproterenol produces a significant increase in the cGK activity ratio in addition to its activation of cAK. Thus, cross-activation of cGK by physiological elevation of intracellular cAMP could be a mechanism by which CAMP-elevatingagents induce smooth muscle relaxation (Fig. 5) (Jiang et al. 1992). To our knowledge, the cAMP cross-activation of cGK is the first demonstrated example of which cellular concentrations of one second messenger act on the highly selective receptor site(s) of another second messenger. Cyclic nucleotide cross-activation may also occur in a converse manner. It has recently been shown that the cGMP levels in T-84 colon carcinoma cells can be markedly elevated
REVIEW / SYNTHESE
cyclic nucleotide protein
ADP
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
or 1 EDRF
various
agents
-
tower ca2+
RELAXATION FIG. 5. Proposed mechanism for cGMP and cAMP in smooth muscle relaxation. by bacterial enterotoxin t o cause cAK activation, which stimulates C1- secretion (Forte et al. 1992). Whether or not the cross-activation of cAK by enterotoxin-elevated cGMP in T-84 cells is a physiological event as well as a pathological event remains t o be demonstrated. Recently, another possible example of cyclic nucleotide cross-activation has been reported. The olfactory ion channel protein, which has specificity for cGMP over CAMP in vitro (Altenhofen et al. 1991), has been suggested to be activated by cAMP in the olfactory epithelium (Lancet and Pace 1987; Breer et al. 1990). Cyclic nucleotide gated ion channels are within the same family of proteins as cGK, cAK, and CAP (Shabb and Corbin 1992) (Fig. 1). More studies are needed to establish if the olfactory ion channel protein is regulated in vivo by CAMP, cGMP, or both. It is suggested that cross-activation extends beyond the cyclic nucleotide receptors. The concept of specificity of various hormones for their target receptors has been well established. However, this specificity may not be absolute or necessary under certain circumstances. The cyclic nucleotide cross-activation is an example which suggests the need for further exploration of this phenomenon. For reasons that are not well understood in all instances, nature has evolved specific regulation of cAK and cGK by cAMP and cGMP, respectively. In many tissues, the specificity of these enzymes is essential for avoiding degeneracy of distinct signal transduction pathways. There could be an alternative explanation for the evolution of a protein kinase with a high affinity for one cyclic nucleotide coupled with a low affinity for the other cyclic nucleotide. Perhaps this is due to the fact that in some cells both CAMP and cGMP regulate the same kinase, such as cGK, and in this case cAMP is present at a much higher concentration than is cGMP. In other words, the natural selection process could have designed the binding sites of cGK with affinities appropriate for cellular regulation by either cGMP or CAMP. In summary, although specific effects of cAMP or cGMP are mediated by its respective protein kinase, either cAMP or cGMP can also cross-activate the other kinase in certain tissues.
,
Acknowledgement This work was supported in part by research grants DK-40029 and GM-41269 from the National Institute of Health. Altenhofen, W., Ludwig, J., Eismann, E., Kraus, W., Bonigk. W., and Kaupp, U.B. 1991. Control of ligand specificity in cyclic nucleotide-gated channels from rod photoreceptors and olfactory epithelium. Proc. Natl. Acad. Sci. U.S.A. 88: 9868-9872. Beavo, J.A., Bechtel, P.J., and Krebs, E.G. 1974. Activation of protein kinase by physiological concentrations of cyclic AMP: kinetics at high enzyme concentrations. Proc. Natl. Acad. Sci. U.S.A. 71: 3580-3583. Breer, H., Boekhoff, I., and Tareilus, E. 1990. Rapid kinetics of second messenger formation in olfactory transduction. Nature (London), 345: 65-68. Bubis, J., Neitzel, J.J., Saraswat, L.D., and Taylor, S.S. 1988a. A point mutation abolishes binding of cAMP to site A in the regulatory subunit of CAMP-dependentprotein kinase. J. Biol. Chem. 263: 9668-9673. Bubis, J., Saraswat, L.D., and Taylor, S.S. 1988b. Tyrosine-371 contributes to the positive cooperativitybetween the two cAMP binding sites in the regulatory subunit of CAMP-dependent protein kinase I. Biochemistry, 27: 1570-1576. Butt, E., van Bemmelen, M., Fischer, L., Walter, U., and Jastorff, B. 1990. Inhibition of cGMP-dependent protein kinase by (Rp)guanosine 3',5'-monophosphorothioates. FEBS Lett. 263: 47-50. Colbran, J.L., Francis, S.H., Leach, A.B., Thomas, M.K., Jiang, H., McAllister, L.M., and Corbin, J.D. 1992. A phenylalanine in peptide substrates provides for selectivity between cGMP- and CAMP-dependent protein kinases. J. Biol. Chem. 267: 9589-9594. Corbin, J.D., Sugden, P.H., West, L., Flockhart, D.A., Lincoln. T.M. and McCarthy, D. 1978. Studies on the properties and mode of action of the purified regulatory subunit of bovine heart adenosine 3 ' :5' -monophosphate-dependent protein kinase. J. Biol. Chem. 253: 3997-4003. Corbin, J.D., Ogreid, D., Miller, J.P., Suva, R.H., Jastorff, B., and Doskeland, S.O. 1986. Studies of cGMP analog specificity and function of the two intrasubunit binding sites of cGMPdependent protein kinase. J. Biol. Chem. 261: 1208-1214. Cornwell, T.L., and Lincoln, T.M. 1989. Regulation of intracellular c a 2 + levels in cultured vascular smooth muscle cells. J. Biol. Chem. 264: 1146-1 155.
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
1288
BIOCHEM. CELL BIOL.
De Wit, R.J.W., Hekstra, D., Jastorff, B., Stec, W.J., Baraniak, J., Van Driel, R., and Van Haastert, P.J.M. 1984. Inhibitory action of certain cyclophosphate derivatives of cAMP on CAMPdependent protein kinases. Eur. J. Biochem. 142: 255-260. Dhallan, R.S., Yau, K.-Y., Schrader, K.A., and Reed, R.R. 1990. Primary structure and functional expression of a cyclic nucleotide-activated channel from olfactory neurons. Nature (London), 347: 184-187. Doskeland, S.O., Ogreid, D., Ekanger, R., Sturm, P.A., Miller, J.P., and Suva, R.H. 1983. Mapping of the two intrachain cyclic nucleotide binding sites of adenosine cyclic 3' ,5'-phosphate dependent protein kinase I. Biochemistry. 22: 1094-1 101. Fiscus, R.R., Rapoport, R.M., Waldman, S.A., and Murad, F. 1985. Atriopeptin I1 elevates cyclic GMP, activates cyclic GMPdependent protein kinase and causes relaxation in rat thoracic aorta. Biochim. Biophys. Acta, 846: 179-184. Forte, L.R., Thorne, P.K., Eber, S.L., Krause, W.J., Freeman, R.H., Francis, S.H., and Corbin, J.D. 1992. Stimulation of intestinal C1- transport by heat-stable enterotoxin: activation of CAMP-dependent protein kinase by cGMP. Am. J. Physiol. 263: C607-C615. Foster, J.L., Guttman, J., and Rosen, O.M. 1981. Autophosphorylation of cGMP-dependent protein kinase. J. Biol. Chem. 256: 5029-5036. Francis, S.H., Noblett, B.D., Todd, B.W., Wells, J.N., and Corbin, J.D. 1988. Relaxation of vascular and tracheal smooth muscle by cyclic nucleotide analogs that preferentially activate purified cGMP-dependent protein kinase. Mol. Pharmacol. 34: 506-517. Glass, D.B., and Krebs, E.G. 1982. Phosphorylation by guanosine 3 ' ,5 ' -monophosphate-dependent protein kinase of synthetic peptide analogs of a site phosphorylated in histone H2b. J. Biol. Chem. 257: 1196-1200. Hanks, S.K., Quinn, A.M., and Hunter, T. 1988. The protein kinase family: conserved features and deduced phylogeny of the catalytic domains. Science (Washington, D.C.), 241: 42-52. Hardman, J.G. 1984. Cyclic nucleotides and regulation of vascular smooth muscle. J. Cardiovasc. Pharmacol. 6: 5639-5645. Hei, Y.J., MacDonell, K.L., McNeill, J.H., and Diamond, J. 1991. Lack of correlation between activation of cyclic AMP-dependent protein kinase and inhibition of contraction of rat vas deferens by cyclic AMP analogs. Mol. Pharmacol. 39: 233-238. Hofmann, F., Dostmann, W., Keilbach, A., Landgraf, W., and Ruth, P. 1992. Structure and physiological role of cGMPdependent protein kinase. Biochim. Biophys. Acta, 1135: 51-60. Jiang, H., Colbran, J.L., Francis, S.H., and Corbin, J.D. 1992. Direct evidence for cross-activation of cGMP-dependent protein kinase by cAMP in pig coronary arteries. J . Biol. Chem. 267: 1015-1019. Kaupp, U.B., Niidome, T., Tanabe, T., Terada, S., Bonigk, W., Stuhmer, W., Cook, N.L., Kangawa, K., Matuso, H., Hirose, T., Miyata, T., and Numa, S. 1989. Primary structure and functional expression from com~lementaryDNA of the rod photoreceptor~cyclicGMP-gated channel. ~ a t u r (London), e 342: 762-766. Krebs, E.G. 1972. Protein kinases. Curr. Top. Cell. Regul. 5: 99-133. Kumar, V.D., and Weber, I.T. 1992. Molecular model of the cyclic GMP-binding domain of the cyclic GMP-gated ion channel. Biochemistry, 31: 4643-4649. Kuno, T., Shuntoh, H., Sakaue, M., Saijoh, K., Taketa, T., Fukuda, K., and Tanaka, C. 1988. Site-directed mutagenesis of the CAMP-binding sites of the recombinant type I regulatory subunit of CAMP-dependentprotein kinase. Biochem. Biophys. Res. Commun. 153: 1244-1250. Lancet, D., and Pace, U. 1987. The molecular basis of odor recognition. Trends Biochem. Sci. 12: 63-66. Landgraf, W., Hullin, R., Gobel, C., and Hofmann, F. 1986. Phosphorylation of cGMP-dependent protein kinase increases the affinity for cyclic AMP. Eur. J. Biochem. 154: 113-117.
VOL. 70,
1992
Lincoln, T.M. 1983. Effects of nitroprusside and 8-bromo-cGMP on the contractile activity of the rat aorta. J. Pharmacol. Exp. Ther. 224: 100-107. Lincoln, T.M., and Corbin, J.D. 1983. Characterization and biological role of the cGMP-dependent protein kinase. Adv. Cyclic Nucleotide Res. 15: 139-192. Lincoln, T.M., and Cornwell, T.L. 1991. Towards an understanding of the mechanism of action of cyclic AMP and cyclic GMP in smooth muscle relaxation. Blood Vessels, 28: 129-137. Lincoln, T.M., Cornwell, T.L., and Taylor, A.E. 1990. cGMPdependent protein kinase mediates the reduction of c a 2 + by cAMP in vascular smooth muscle cells. Am. J. Physiol. 258: C399-C407. Ludwig, J., Margalit, T., Eismann, E., Lancet, D., and Kaupp, U.B. 1990. Primary structure of CAMP-gated channel from bovine olfactory epithelium. FEBS Lett. 270: 24-29. Mackenzie, C.W. 1982. Bovine lung cyclic GMP-dependent protein kinase exhibits two types of specific cyclic GMP-binding sites. J. Biol. Chem. 257: 5589-5593. Mattson, M.P., and Spaziani, E. 1986. Regulation of crab Y-organ steroidogenesis in vitro: evidence that ecdysteroid production increases through activation of CAMP-phosphodiesterase by calcium-calmodulin. Mol. Cell. Endocrinol. 48: 135-151. McKay, D.B., and Steitz, T.A. 1981. Structure of catabolite gene activator protein at 2.9 A resolution suggests binding to lefthanded B-DNA. Nature (London), 290: 744-749. McKay, D.B., Weber, I.T., and Steitz, T.A. 1982. Structure of catabolite gene activator protein at 2.9-A resolution. J. Biol. Chem. 257: 9518-9524. Ogreid, D., and Doskeland, S.O. 1982. Activation of protein kinase isoenzymes under near physiological conditions: evidence that both types (A and B) of cAMP binding sites are involved in the activation of protein kinase by cAMP and 8-N3-CAMP.FEBS Lett. 150: 161-166. Ogreid, D., Doskeland, S.O., Gorman, K.B., and Steinberg, R.A. 1988. Mutations that prevent cyclic nucleotide binding to binding sites A or B of type I cyclic AMP-dependent protein kinase. J. Biol. Chem. 263: 17 397 - 17 404. Shabb, J.B., and Corbin, J.D. 1992. Cyclic nucleotide-binding domains in proteins having diverse functions. J. Biol. Chem. 267: 5723-5726. Shabb, J.B., Ng, L., and Corbin, J.D. 1990. One amino acid change produces a high affinity cGMP-binding site in CAMPdependent protein kinase. J. Biol. Chem. 265: 16 031 - 16 034. Shabb, J.B., Buzzeo, B.D., Ng, L., and Corbin, J.D. 1991. Mutating protein kinase CAMP-bindingsites into cGMP-binding sites. J . Biol. Chem. 266: 24 320 - 24 326. Steinberg, R.A., Russell, J.L., Murphy, C.S., and Yphantis, D.A. 1987. Activation of type I cyclic AMP-dependent protein kinases with defective cyclic AMP-binding sites. J. Biol. Chem. 262: 2665-2671. Takio, K., Wade, R.D., Smith, S.B., Krebs, E.G., Walsh. K.A., and Titani, K. 1984. Guanosine cyclic 3',5'-phosphate dependent protein kinase, a chimeric protein homologous with two separate protein families. Biochemistry, 23: 4207-4218. Thomas, M.K., Francis, S.F., and Corbin, J.D. 1990. Substrateand kinase-directed regulation of phosphorylation of a cGMPbinding phosphodiesterase by cGMP. J. Biol. Chem. 265: 14 971 - 14 978. Weber, I.T., and Steitz, T.A. 1987. Structure of a complex of c~tabolitegene activator protein and cyclic AMP refined at 2.5 A resolution. J. Mol. Biol. 198: 31 1-326. Weber, I.T., Takio, K., Titani, K., and Steitz, T.A. 1982. The CAMP-binding domains of the regulatory subunit of CAMPdependent protein kinase and the catabolite gene activator protein are homologous. Proc. Natl. Acad. Sci. U.S.A. 79: 7679-7683. Weber, I.T., Steitz, T.A., Bubis, J., and Taylor, S.S. 1987. Predicted structures of cAMP binding domains of type I and type I1 regulatory subunits of CAMP-dependent protein kinase. Biochemistry, 26: 343-351.
REVIEW /
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
Weber, I.T.,Shabb, J.B., and Corbin, J.D. 1989. Predicted structures of the cGMP binding domains of the cGMP-dependent protein kinase: a key alanine/threonine difference in evolutionary divergence of CAMPand cGMP binding sites. Biochemistry, 28: 6122-6127. Woodford, T.A., Correfl, L.A., McKnight, G.S., and Corbin, J.D. 1989. Expression and characterization of mutant forms of the
SYNTHBSE
1289
type I regulatory subunit of CAMP-dependent protein kinase. J. Biol. Chem. 264: 13 321 - 13 328. Yagura, T.S., and Miller, J.P. 1981. Mapping adenosine cyclic 3'3'-phosphate binding sites on type I and type I1 adenosine cyclic 3 ' ,5 '-phosphate dependent protein kinases using ribose ring and cyclic phosphate ring analogues of adenosine cyclic 3',5'-phosphate. Biochemistry, 20: 879-887.
REVIEW /
SYNTHGSE
Cross-activation: overriding cAMP/cGMP selectivities of protein kinases in tissues1 HANGJIANG,JOHNB. SHABB,AND JACKIE D.
CORBIN~
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
Department of Molecular Physiology and Biophysics, Vanderbilt University School of Medicine, Nashville, TN 3 7232-0615, U.S.A. Received March 25, 1992 JIANG,H., SHABB,J. B., and CORBIN,J. D. 1992. Cross-activation: overriding cAMP/cGMP selectivities of protein kinases in tissues. Biochem. Cell Biol. 70: 1283-1289. CAMP-and cGMP-dependent protein kinases are homologous proteins and are predicted to exhibit very similar threedimensional structures. Their cyclic nucleotide binding domains share a high degree of amino acid sequence identity. CAMP- and cGMP-dependent protein kinases are activated relatively specifically by cAMP and cGMP, respectively; and a single alanine-threonine difference between CAMP- and cGMP-binding domains partially accounts for this specificity. Thus, it would be expected that cAMP and cGMP mediate separate physiological effects. However, owing in part to the lack of absolute specificity of either enzyme and to the relatively high level of cAMP or cGMP in certain tissues, it is also possible that either cyclic nucleotide could cross-activate the other kinase. Increases in either cAMP or cGMP cause pig coronary artery relaxation. However, only cGMP-dependent protein kinase specific cyclic nucleotide analogues are very effective in causing relaxation, and cAMP elevation in arteries treated with isoproterenol or forskolin activates cGMP-dependent protein kinase, in addition to CAMP-dependent protein kinase. Conversely, increases in either cAMP or cGMP cause C1- secretion in T-84 colon carcinoma cells, and the cGMP level in T-84 cells can be elevated sufficiently by bacterial enterotoxin to activate CAMP-dependent protein kinase. These results imply specific regulation of CAMP-and cGMP-dependent protein kinases by the respective cyclic nucleotides, but either cyclic nucleotide is able to cross-activate the other kinase in certain tissues. Key words: cGMP, CAMP, smooth muscle relaxation, protein phosphorylation. JIANG,H., SHABB,J. B., et CORBIN,J. D. 1992. Cross-activation: overriding cAMP/cGMP selectivities of protein kinases in tissues. Biochem. Cell Biol. 70 : 1283-1289. Les proteine kinases dipendantes du cAMP et du cGMP sont des proteines homologues et elles devraient posstder des structures tridimensionnelles trks semblables. Leurs domaines de liaison des nucleotides cycliques ont des sequences d'acides aminks presque identiques. La proteine kinase dependante du cAMP et la protkine kinase dependante du cGMP sont respectivement activees de facon assez specifique par le cAMP et le cGMP; cette specificite s'explique partiellement par une seule difference alanine-thrkonine entre les domaines de liaison du cAMP et du cGMP. Ainsi, le CAMP et le cGMP pourraient exercer des effets physiologiques distincts. Toutefois, Ctant donne le manque de specificit6 absolue des deux enzymes et le taux relativement eleve de cAMP et de cGMP dans certains tissus, il est Cgalement possible que chacun des nucleotides cycliques active l'une ou I'autre kinase. Des augmentations du cAMP ou du cGMP provoquent la relaxation des artkres coronaires chez le porc. Cependant, seuls les analogues des nucltotides cycliques spkcifiques de la proteine kinase dependante du cGMP sont trks efficaces pour causer la relaxation et l'klevation du cAMP dans les artkres traitkes avec 17isoprotCrCnolou la forskoline active la proteine kinase dependante du cGMP en plus de la proteine kinase dependante du CAMP.Reciproquement, des augmentations du CAMPou du cGMP entrainent la sicretion de C1 dans les cellules T-84 du carcinome du c6lon et le taux de cGMP dans les cellules T-84 peut 2tre suffisamment eleve par 17entCrotoxinebactkrienne pour activer la proteine kinase dependante du CAMP. Ces resultats impliquent la regulation specifique de la proteine kinase dependante du cAMP et de la proteine kinase dipendante du cGMP par leurs nucleotides cycliques respectifs, mais chacun des nucleotides cycliques est capable d'activer l'autre kinase dans certains tissus. Mots elks : cGMP, CAMP, relaxation des tissus, phosphorylation des proteines. [Traduit par la redaction]
Cross-talk is a popular expression that refers to cross-over between two or more second messenger pathways. For example, an elevation in cell calcium levels might cause a decrease in cAMP through activation of the calciumstimulated phosphodiesterase (Mattson and Spaziani 1986). Cross-activation is defined as a special form of cross-talk in which a ligand that is highly specific for a particular recepABBREVIATIONS: cAK, CAMP-dependent protein kinase; cGK, cGMP-dependent protein kinase; R and C subunits, regulatory and catalytic subunits of CAMP-dependent protein kinase; CAP, catabolite gene activator protein; BPDE, cGMP-binding phosphodiesterase. his review is based on a talk and abstract presented at the 34th Annual Meeting of the Canadian Federation of Biological Societies. ' ~ u t h o rto whom all correspondence should be addressed. Printed in Canada / lmprimh au Canada
tor interacts with another receptor that is highly specific for a different ligand under physiological conditions. This review will focus on cAMP and cGMP cross-activation of the cyclic nucleotide dependent protein kinases. To better understand how cross-activation of protein kinase can occur in vitro or in vivo, one should first consider effector specificity of cyclic nucleotide dependent protein kinases. It has been assumed that cAK is quite specifically activated by its effector, CAMP,whereas cGMP specifically activates cGK (Lincoln and Corbin 1983). This phenomenon has been seen in many tissues, where the activation state of cAK or cGK is usually associated with the elevation of its corresponding cyclic nucleotide, cAMP or cGMP (Beavo et al. 1974; Ogreid and Doskeland 1982; Lincoln 1983; Lincoln and Corbin 1983; Fiscus, e t al. 1985). The cyclic nucleotide specificity of protein kinases serves as the
1284
BIOCHEM. CELL BIOL. VOL. 70. 1992
CAMP-DEPENDENT PROTEIN KINASES
DIM^
INH
1
CAMP
1
I
CAMP
SITE 2 SITE 1 (FAST SITE) (SLOW SITE)
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
cGMP-DEPENDENT PROTEIN KINASES
CGMP
I CGMP I
CATALYTIC
I
CATABOLITE GENE ACTIVATOR PROTEIN FIG. 1. Linear domain structures of some cyclic nucleotide binding proteins. DIM, dimerization domain; INH, inhibitory domain; DNA, DNA binding domain.
nucleotide by certain hormones might cause the crossactivation of the other protein kinase. Moreover, the specificity of the protein kinases may be lessened by certain modifications of the kinases. For example, it has been shown that autophosphorylation of cGK (type Ia) improved its affinity for cAMP activation by 6- to 10-fold (Foster et al. 1981; Landgraf et al. 1986). In general, though effector specificity of protein kinases provides a mechanism for defined and precise regulation, cross-activation of these enzymes would permit some flexibility in the regulation of enzyme activities, and it might reflect an opportunistic mechanism of nature in using an additional second messenger to regulate a pathway that is already in place. This review is divided into two parts. The first part deals with the effector specificities of cyclic nucleotide dependent protein kinases, based on the comparison of cAMP and cGMP binding domains. The second part provides background information and some results to suggest that cross-activation of cyclic nucleotide dependent protein kinases can actually occur in certain intact tissues under physiological situations.
FIG. 2. Schematic representation of the structure of a generic cyclic nucleotide binding site of cAK or cGK, modeled by analogy with the crystal structure of CAP containing bound CAMP. Solid black oval, cAMP or cGMP. Invariant amino acids are indicated. The STA residue is a S in cGK, T in CAP, and A in cAK binding sites. (From Shabb and Corbin 1992.)
means for precise regulation, allowing the fluctuation of only one second messenger to turn a specific regulatory pathway on or off. However, effector specificity of protein kinases may not be absolute for several reasons. cAK and cGK are homologous proteins, with structural similarities between cAMP binding sites of cAK and cGMP binding sites of cGK (Weber et al. 1987, 1989). The similarities allow CAMP activation of cGK or cGMP activation of cAK in vitro. Since the level of one cyclic nucleotide in certain cells is severalfold higher than that of the other one (Francis et al. 1988; Forte et al. 1992), the elevation of this higher basal level of cyclic
Specificity of cyclic nucleotide dependent protein kinases It has been proposed that many effects of CAMP and cGMP are mediated by cAK and cGK, respectively, though other receptors also exist (Krebs 1972; Mackenzie 1982; Shabb and Corbin 1992). Although either cyclic nucleotide can cross-activate the other enzyme in vitro, cAK exhibits > 50-fold higher affinity for cAMP than for cGMP, whereas the affinity of cGK for cGMP is > 50-fold higher than that of cAMP (Doskeland et al. 1983; Corbin et al. 1986; Landgraf et al. 1986). Despite the fact that the two enzymes specifically and preferentially bind different cyclic nucleotides, which is the key issue to be addressed, mammalian cAK and cGK show a high degree of amino acid sequence similarity. The free C subunit of cAK is homologous to the catalytic domain (fused to a regulatory domain) of cGK, and both of them also show homology with all other protein kinases (Corbin et al. 1978; Hanks et al. 1988). As shown by a comparison of linear domain structures (Fig. 1) for cyclic nucleotide binding proteins, each of the two intrasubunit cyclic nucleotide binding sites of cAK and cGK shares amino acid sequence homology with the cAMP binding domain of the bacterial CAP (Weber et al. 1982, 1987; Takio et al. 1984). Recently, the cyclic nucleotide gated
REVIEW / SYNTHESE
CAMP-BINDINGSlTE (CAW
cGMP-BINDING SlTE
(cGK)
dNH'
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
Thr
0 NH YNH2 NH2+
0
FIG. 3. Predicted protein - cyclic nucleotide contacts for cAK and cGK. The sites shown represent either the slow or fast site of the respective kinase.
cation channels have also been proposed to be within this same family of proteins (Kaupp et al. 1989; Ludwig et al. 1990; Dhallan et al. 1990; Altenhofen et al. 1991). Because of this sequence homology and the known crystal structure of the CAP protein containing bound cAMP (McKay and Steitz 1981; McKay et al. 1982; Weber and Steitz 1987), modeling of cyclic nucleotide binding sites in cAK, cGK, and the rod photoreceptor cGMP-gated ion channel has been achieved (Weber et al. 1987, 1989; Kumar and Weber 1992). The models have extended the research in this field by allowing predictions of functional importance of specific amino acid residues tht could be subsequently tested by site-directed mutagenesis (e.g., Bubis et al. 1988a, 1988b; Kuno et al. 1988). Figure 2 shows the predicted tertiary structure of the fast site of cGK, which is composed of three a helices (Aa, Ba, and Ca) and one antiparallel /3-roll structure. The cGMP molecule is located at the bottom of the internal hydrophobic pocket formed by the 0-barrel and C a helix. This pattern of secondary structure and tertiary folding is retained in all of the cyclic nucleotide binding domain models of cAK and cGK. In fact, no gross differences in the overall structures are predicted to account for the cyclic nucleotide specificities towards their corresponding kinases (Weber et al. 1989). Since the cAMP and cGMP molecules differ mainly in structure of the appended groups on the purine ring, it has been suggested that the differences in affinity between CAMP and cGMP binding to the corresponding kinases could be due to different amino acid contacts with these different groups (Weber et al. 1989). Thus, the amino acids that are very close to the bound cyclic nucleotide, forming the binding pocket, have been examined for features that might explain the specificity of cyclic nucleotide binding. Several specific contacts between cyclic nucleotides and the binding pockets are predicted to be the same for cAK and cGK, based on amino acid sequence comparison of CAP, cAK, and cGK cyclic nucleotide binding domains. As a matter of fact, three amino acids in the binding pockets of cAK and cGK are found to be invariant. According to the models, an arginine, glycine, and glutamic acid make contacts with the ribose and phosphate moieties of the cyclic nucleotides (Fig. 3) (Weber et al. 1989). Results of mutagenesis of these residues in cAK (Steinberg et al. 1987;
Ogreid et al. 1988; Bubis et al. 1988a; Woodford et al. 1989) and cyclic nucleotide analogue studies in both cAK and cGK (Yagura and Miller 1981; De Wit et al. 1984; Corbin et al. 1986; Butt et al. 1990) are consistent with the importance of these residues in interaction with either CAMPor cGMP. The binding of cAMP or cGMP in the corresponding binding pocket requires the conserved ionic interaction between one of the exocyclic phosphate oxygens of the cyclic nucleotide and the arginine. In addition, a series of hydrogen bonds are formed between the ribose phosphate moiety of the cyclic nucleotide and the surrounding amino acid residues in the binding pocket. For example, the ribose 2'OH forms hydrogen bonds with the conserved Glu and Gly (Weber and Steitz 1987). Disruption of these hydrogen bonds by either mutation at the invariant residues or by using cyclic nucleotide analogues is thought to explain the loss or deterioration of cyclic nucleotide binding. However, it has been predicted, based on the CAP model, studies of CAMPanalogue binding, and amino acid sequence comparisons of the binding sites, that one residue in the binding pocket is critical for determining cAMP versus cGMP binding specificity of cAK and cGK. This residue is an Ala in all CAMP binding sites of RI and RII subunits, but the corresponding residue is a Thr in all cGMP binding domains (Fig. 3) (Weber et al. 1989). It has also been suggested from cGMP analogue studies that cGMP is bound in the syn conformation, which would allow a hydrogen bond to form between the 2-NHz group of guanine and the OH group of Thr in both cGK binding sites (Weber et al. 1989). This hydrogen bond cannot occur in the cAMP binding sites of cAK, since there is no such OH group on Ala corresponding to the position of Thr in the cGMP binding pocket. Studies by Shabb et al. using mutagenesis suggest that the threonine OH group is the binding site component which enhances cGMP binding affinity (Shabb et al. 1990, 1991). The wild-type and mutant bovine RIa subunits have been expressed in Escherichia coli and the holoenzyrnes have been formed by recombination of the purified C subunits and the expressed R subunits. A site 1 (slow site) mutant of cAK, in which Ala-334 is changed to Thr using oligonucleotidedirected mutagenesis, has a marked increase in cGMP
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
1286
BIOCHEM. CELL BIOL. VOL. 70, 1992
FIG. 4. Activation of cGK (m) and cAK (0) by isoproterenol in pig coronary arteries. Tissue segments were incubated with various concentrations of isoproterenol for 4 min at 37°C before preparation of extracts and assay of cGK and cAK activities. Data are calculated as fold stimulation of the control (no isoproterenol). Results are mean SE from four separate experiments done in duplicate. *, p < 0.05 (one-tailed t-test); **, p < 0.01 (one-tailed t-test); and ***, p < 0.01 (two-tailed t-test), comparing control and isoproterenol-treated tissues. (From Jiang et al. 1992.)
*
affinity but no change in cAMP affinity, as determined by cyclic nucleotide binding and protein kinase activation assays. The corresponding site 2 (fast site) mutant, created by changing Ala-210 to Thr, exhibits an increased affinity for cGMP and a decreased affinity for CAMP. A double mutant containing changes in both the slow and fast sites has an improved cGMP versus cAMP selectivity of -200 fold in comparison with that of wild-type cAK, making the double mutant about equal to native cGK regarding cGMP binding affinity. Thus, a protein kinase with high affinity for cGMP is produced from cAK by making a single amino acid change from Ala to Thr in each of the two cAMP binding pockets. Therefore, this single mutation of Ala to Thr, which requires only a single base change, is a key evolutionary development in the divergence of cAMP and cGMP binding domains (Weber et al. 1989). The Ala-Thr difference also explains cAMP versus cGMP binding selectivity of the related cyclic nucleotide gated ion channels, which contain a Thr at the corresponding position and have higher affinity for cGMP than for cAMP (Altenhofen et al. 1991). Cross-activation of cyclic nucleotide-dependent protein kinases Although cAK and cGK are rather specifically activated by their respective cyclic nucleotides, owing mainly to the Ala-Thr difference in their cyclic nucleotide binding pockets, the specificity might not be sufficient to eliminate crossactivation in some cells. For example, elevation of cAMP in certain tissues might actually cross-activate cGK to produce the same effect as elevation of cGMP. Smooth muscle is a likely tissue in which cross-activation may occur. It is well known that elevation of either cGMP or cAMP can cause smooth muscle relaxation (Hardman 1984; Lincoln and Corbin 1983; Hofmann et al. 1992), but it is uncertain if one or both of the cyclic nucleotide dependent
protein kinases can mediate these cyclic nucleotide actions. Recently, Francis et al. have examined this question by comparing the potencies of cyclic nucleotide analogues in relaxing pig coronary arteries and guinea pig tracheal smooth muscle with their abilities to activate the purified cAK and cGK. cGK-specific cyclic nucleotide analogues, but not cAKspecific analogues, cause muscle relaxation, suggesting that cAK is not the mediator of cAMP effects in these tissues (Francis et al. 1988). One possible explanation is that protein phosphorylation catalyzed by cGK, but not by cAK, mediates smooth muscle relaxation caused by elevation of either cGMP or CAMP. This would also explain the recent observation that 8-Br-CAMPcauses activation of cAK in rat vas deferens without producing smooth muscle relaxation (Hei et al. 1991). Lincoln et al. have reported that restoration of cGK, but not of cAK, to cGK-depleted aortic smooth muscle cells restores the calcium-lowering effects elicited by cGMP or CAMP.These workers have also hypothesized that cGK mediates the relaxing effects of both cyclic nucleotides (Cornwell and Lincoln 1989; Lincoln et al. 1990; Lincoln and Cornwell 1991). Considering all of the findings mentioned above, it is possible that elevation of cAMP in smooth muscle may cross-activate cGK to produce relaxation. cAMP levels are - 5-fold higher than cGMP levels in some vascular smooth muscle tissues (Francis et al. 1988), and autophosphorylation of cGK improves its affinity for cAMP by 6- to 10-fold (Foster et al. 1981; Landgraf et al. 1986), making it more readily activated by cAMP than is the unphosphorylated form of cGK. Therefore, under some conditions, small elevations of cellular cAMP could cross-activate cGK, even though the affinity of unphosphorylated cGK (type ICY or Ip) for cAMP is -50-fold lower than the affinity of this enzyme for cGMP. Although the idea of cross-activation has been around for many years, the direct experimental evidence showing the occurrence of cross-activation of cGK by cAMP in an intact tissue has been obtained only recently. The activity ratios of cAK and cGK have been measured in extracts of pig coronary arteries after treating the tissues with CAMP-and cGMP-elevating agents (Jiang et al. 1992). This study has been facilitated by the development of a substrate (BPDEtide) that is phosphorylated better by cGK than by cAK (Colbran et al. 1992). The peptide is based on the amino acid sequence of the phosphorylation site of the bovine lung BPDE (Thomas et al. 1990; Colbran et al. 1992). The use of BPDEtide has improved the specificity of the assay for detecting the activation of cGK, as compared with the heptapeptide substrate, which has been previously used as a relatively specific substrate for cGK versus cAK (Glass and Krebs 1982). Using this assay, it is revealed (Fig. 4) that the CAMP-elevatingagent isoproterenol produces a significant increase in the cGK activity ratio in addition to its activation of cAK. Thus, cross-activation of cGK by physiological elevation of intracellular cAMP could be a mechanism by which CAMP-elevatingagents induce smooth muscle relaxation (Fig. 5) (Jiang et al. 1992). To our knowledge, the cAMP cross-activation of cGK is the first demonstrated example of which cellular concentrations of one second messenger act on the highly selective receptor site(s) of another second messenger. Cyclic nucleotide cross-activation may also occur in a converse manner. It has recently been shown that the cGMP levels in T-84 colon carcinoma cells can be markedly elevated
REVIEW / SYNTHESE
cyclic nucleotide protein
ADP
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
or 1 EDRF
various
agents
-
tower ca2+
RELAXATION FIG. 5. Proposed mechanism for cGMP and cAMP in smooth muscle relaxation. by bacterial enterotoxin t o cause cAK activation, which stimulates C1- secretion (Forte et al. 1992). Whether or not the cross-activation of cAK by enterotoxin-elevated cGMP in T-84 cells is a physiological event as well as a pathological event remains t o be demonstrated. Recently, another possible example of cyclic nucleotide cross-activation has been reported. The olfactory ion channel protein, which has specificity for cGMP over CAMP in vitro (Altenhofen et al. 1991), has been suggested to be activated by cAMP in the olfactory epithelium (Lancet and Pace 1987; Breer et al. 1990). Cyclic nucleotide gated ion channels are within the same family of proteins as cGK, cAK, and CAP (Shabb and Corbin 1992) (Fig. 1). More studies are needed to establish if the olfactory ion channel protein is regulated in vivo by CAMP, cGMP, or both. It is suggested that cross-activation extends beyond the cyclic nucleotide receptors. The concept of specificity of various hormones for their target receptors has been well established. However, this specificity may not be absolute or necessary under certain circumstances. The cyclic nucleotide cross-activation is an example which suggests the need for further exploration of this phenomenon. For reasons that are not well understood in all instances, nature has evolved specific regulation of cAK and cGK by cAMP and cGMP, respectively. In many tissues, the specificity of these enzymes is essential for avoiding degeneracy of distinct signal transduction pathways. There could be an alternative explanation for the evolution of a protein kinase with a high affinity for one cyclic nucleotide coupled with a low affinity for the other cyclic nucleotide. Perhaps this is due to the fact that in some cells both CAMP and cGMP regulate the same kinase, such as cGK, and in this case cAMP is present at a much higher concentration than is cGMP. In other words, the natural selection process could have designed the binding sites of cGK with affinities appropriate for cellular regulation by either cGMP or CAMP. In summary, although specific effects of cAMP or cGMP are mediated by its respective protein kinase, either cAMP or cGMP can also cross-activate the other kinase in certain tissues.
,
Acknowledgement This work was supported in part by research grants DK-40029 and GM-41269 from the National Institute of Health. Altenhofen, W., Ludwig, J., Eismann, E., Kraus, W., Bonigk. W., and Kaupp, U.B. 1991. Control of ligand specificity in cyclic nucleotide-gated channels from rod photoreceptors and olfactory epithelium. Proc. Natl. Acad. Sci. U.S.A. 88: 9868-9872. Beavo, J.A., Bechtel, P.J., and Krebs, E.G. 1974. Activation of protein kinase by physiological concentrations of cyclic AMP: kinetics at high enzyme concentrations. Proc. Natl. Acad. Sci. U.S.A. 71: 3580-3583. Breer, H., Boekhoff, I., and Tareilus, E. 1990. Rapid kinetics of second messenger formation in olfactory transduction. Nature (London), 345: 65-68. Bubis, J., Neitzel, J.J., Saraswat, L.D., and Taylor, S.S. 1988a. A point mutation abolishes binding of cAMP to site A in the regulatory subunit of CAMP-dependentprotein kinase. J. Biol. Chem. 263: 9668-9673. Bubis, J., Saraswat, L.D., and Taylor, S.S. 1988b. Tyrosine-371 contributes to the positive cooperativitybetween the two cAMP binding sites in the regulatory subunit of CAMP-dependent protein kinase I. Biochemistry, 27: 1570-1576. Butt, E., van Bemmelen, M., Fischer, L., Walter, U., and Jastorff, B. 1990. Inhibition of cGMP-dependent protein kinase by (Rp)guanosine 3',5'-monophosphorothioates. FEBS Lett. 263: 47-50. Colbran, J.L., Francis, S.H., Leach, A.B., Thomas, M.K., Jiang, H., McAllister, L.M., and Corbin, J.D. 1992. A phenylalanine in peptide substrates provides for selectivity between cGMP- and CAMP-dependent protein kinases. J. Biol. Chem. 267: 9589-9594. Corbin, J.D., Sugden, P.H., West, L., Flockhart, D.A., Lincoln. T.M. and McCarthy, D. 1978. Studies on the properties and mode of action of the purified regulatory subunit of bovine heart adenosine 3 ' :5' -monophosphate-dependent protein kinase. J. Biol. Chem. 253: 3997-4003. Corbin, J.D., Ogreid, D., Miller, J.P., Suva, R.H., Jastorff, B., and Doskeland, S.O. 1986. Studies of cGMP analog specificity and function of the two intrasubunit binding sites of cGMPdependent protein kinase. J. Biol. Chem. 261: 1208-1214. Cornwell, T.L., and Lincoln, T.M. 1989. Regulation of intracellular c a 2 + levels in cultured vascular smooth muscle cells. J. Biol. Chem. 264: 1146-1 155.
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
1288
BIOCHEM. CELL BIOL.
De Wit, R.J.W., Hekstra, D., Jastorff, B., Stec, W.J., Baraniak, J., Van Driel, R., and Van Haastert, P.J.M. 1984. Inhibitory action of certain cyclophosphate derivatives of cAMP on CAMPdependent protein kinases. Eur. J. Biochem. 142: 255-260. Dhallan, R.S., Yau, K.-Y., Schrader, K.A., and Reed, R.R. 1990. Primary structure and functional expression of a cyclic nucleotide-activated channel from olfactory neurons. Nature (London), 347: 184-187. Doskeland, S.O., Ogreid, D., Ekanger, R., Sturm, P.A., Miller, J.P., and Suva, R.H. 1983. Mapping of the two intrachain cyclic nucleotide binding sites of adenosine cyclic 3' ,5'-phosphate dependent protein kinase I. Biochemistry. 22: 1094-1 101. Fiscus, R.R., Rapoport, R.M., Waldman, S.A., and Murad, F. 1985. Atriopeptin I1 elevates cyclic GMP, activates cyclic GMPdependent protein kinase and causes relaxation in rat thoracic aorta. Biochim. Biophys. Acta, 846: 179-184. Forte, L.R., Thorne, P.K., Eber, S.L., Krause, W.J., Freeman, R.H., Francis, S.H., and Corbin, J.D. 1992. Stimulation of intestinal C1- transport by heat-stable enterotoxin: activation of CAMP-dependent protein kinase by cGMP. Am. J. Physiol. 263: C607-C615. Foster, J.L., Guttman, J., and Rosen, O.M. 1981. Autophosphorylation of cGMP-dependent protein kinase. J. Biol. Chem. 256: 5029-5036. Francis, S.H., Noblett, B.D., Todd, B.W., Wells, J.N., and Corbin, J.D. 1988. Relaxation of vascular and tracheal smooth muscle by cyclic nucleotide analogs that preferentially activate purified cGMP-dependent protein kinase. Mol. Pharmacol. 34: 506-517. Glass, D.B., and Krebs, E.G. 1982. Phosphorylation by guanosine 3 ' ,5 ' -monophosphate-dependent protein kinase of synthetic peptide analogs of a site phosphorylated in histone H2b. J. Biol. Chem. 257: 1196-1200. Hanks, S.K., Quinn, A.M., and Hunter, T. 1988. The protein kinase family: conserved features and deduced phylogeny of the catalytic domains. Science (Washington, D.C.), 241: 42-52. Hardman, J.G. 1984. Cyclic nucleotides and regulation of vascular smooth muscle. J. Cardiovasc. Pharmacol. 6: 5639-5645. Hei, Y.J., MacDonell, K.L., McNeill, J.H., and Diamond, J. 1991. Lack of correlation between activation of cyclic AMP-dependent protein kinase and inhibition of contraction of rat vas deferens by cyclic AMP analogs. Mol. Pharmacol. 39: 233-238. Hofmann, F., Dostmann, W., Keilbach, A., Landgraf, W., and Ruth, P. 1992. Structure and physiological role of cGMPdependent protein kinase. Biochim. Biophys. Acta, 1135: 51-60. Jiang, H., Colbran, J.L., Francis, S.H., and Corbin, J.D. 1992. Direct evidence for cross-activation of cGMP-dependent protein kinase by cAMP in pig coronary arteries. J . Biol. Chem. 267: 1015-1019. Kaupp, U.B., Niidome, T., Tanabe, T., Terada, S., Bonigk, W., Stuhmer, W., Cook, N.L., Kangawa, K., Matuso, H., Hirose, T., Miyata, T., and Numa, S. 1989. Primary structure and functional expression from com~lementaryDNA of the rod photoreceptor~cyclicGMP-gated channel. ~ a t u r (London), e 342: 762-766. Krebs, E.G. 1972. Protein kinases. Curr. Top. Cell. Regul. 5: 99-133. Kumar, V.D., and Weber, I.T. 1992. Molecular model of the cyclic GMP-binding domain of the cyclic GMP-gated ion channel. Biochemistry, 31: 4643-4649. Kuno, T., Shuntoh, H., Sakaue, M., Saijoh, K., Taketa, T., Fukuda, K., and Tanaka, C. 1988. Site-directed mutagenesis of the CAMP-binding sites of the recombinant type I regulatory subunit of CAMP-dependentprotein kinase. Biochem. Biophys. Res. Commun. 153: 1244-1250. Lancet, D., and Pace, U. 1987. The molecular basis of odor recognition. Trends Biochem. Sci. 12: 63-66. Landgraf, W., Hullin, R., Gobel, C., and Hofmann, F. 1986. Phosphorylation of cGMP-dependent protein kinase increases the affinity for cyclic AMP. Eur. J. Biochem. 154: 113-117.
VOL. 70,
1992
Lincoln, T.M. 1983. Effects of nitroprusside and 8-bromo-cGMP on the contractile activity of the rat aorta. J. Pharmacol. Exp. Ther. 224: 100-107. Lincoln, T.M., and Corbin, J.D. 1983. Characterization and biological role of the cGMP-dependent protein kinase. Adv. Cyclic Nucleotide Res. 15: 139-192. Lincoln, T.M., and Cornwell, T.L. 1991. Towards an understanding of the mechanism of action of cyclic AMP and cyclic GMP in smooth muscle relaxation. Blood Vessels, 28: 129-137. Lincoln, T.M., Cornwell, T.L., and Taylor, A.E. 1990. cGMPdependent protein kinase mediates the reduction of c a 2 + by cAMP in vascular smooth muscle cells. Am. J. Physiol. 258: C399-C407. Ludwig, J., Margalit, T., Eismann, E., Lancet, D., and Kaupp, U.B. 1990. Primary structure of CAMP-gated channel from bovine olfactory epithelium. FEBS Lett. 270: 24-29. Mackenzie, C.W. 1982. Bovine lung cyclic GMP-dependent protein kinase exhibits two types of specific cyclic GMP-binding sites. J. Biol. Chem. 257: 5589-5593. Mattson, M.P., and Spaziani, E. 1986. Regulation of crab Y-organ steroidogenesis in vitro: evidence that ecdysteroid production increases through activation of CAMP-phosphodiesterase by calcium-calmodulin. Mol. Cell. Endocrinol. 48: 135-151. McKay, D.B., and Steitz, T.A. 1981. Structure of catabolite gene activator protein at 2.9 A resolution suggests binding to lefthanded B-DNA. Nature (London), 290: 744-749. McKay, D.B., Weber, I.T., and Steitz, T.A. 1982. Structure of catabolite gene activator protein at 2.9-A resolution. J. Biol. Chem. 257: 9518-9524. Ogreid, D., and Doskeland, S.O. 1982. Activation of protein kinase isoenzymes under near physiological conditions: evidence that both types (A and B) of cAMP binding sites are involved in the activation of protein kinase by cAMP and 8-N3-CAMP.FEBS Lett. 150: 161-166. Ogreid, D., Doskeland, S.O., Gorman, K.B., and Steinberg, R.A. 1988. Mutations that prevent cyclic nucleotide binding to binding sites A or B of type I cyclic AMP-dependent protein kinase. J. Biol. Chem. 263: 17 397 - 17 404. Shabb, J.B., and Corbin, J.D. 1992. Cyclic nucleotide-binding domains in proteins having diverse functions. J. Biol. Chem. 267: 5723-5726. Shabb, J.B., Ng, L., and Corbin, J.D. 1990. One amino acid change produces a high affinity cGMP-binding site in CAMPdependent protein kinase. J. Biol. Chem. 265: 16 031 - 16 034. Shabb, J.B., Buzzeo, B.D., Ng, L., and Corbin, J.D. 1991. Mutating protein kinase CAMP-bindingsites into cGMP-binding sites. J . Biol. Chem. 266: 24 320 - 24 326. Steinberg, R.A., Russell, J.L., Murphy, C.S., and Yphantis, D.A. 1987. Activation of type I cyclic AMP-dependent protein kinases with defective cyclic AMP-binding sites. J. Biol. Chem. 262: 2665-2671. Takio, K., Wade, R.D., Smith, S.B., Krebs, E.G., Walsh. K.A., and Titani, K. 1984. Guanosine cyclic 3',5'-phosphate dependent protein kinase, a chimeric protein homologous with two separate protein families. Biochemistry, 23: 4207-4218. Thomas, M.K., Francis, S.F., and Corbin, J.D. 1990. Substrateand kinase-directed regulation of phosphorylation of a cGMPbinding phosphodiesterase by cGMP. J. Biol. Chem. 265: 14 971 - 14 978. Weber, I.T., and Steitz, T.A. 1987. Structure of a complex of c~tabolitegene activator protein and cyclic AMP refined at 2.5 A resolution. J. Mol. Biol. 198: 31 1-326. Weber, I.T., Takio, K., Titani, K., and Steitz, T.A. 1982. The CAMP-binding domains of the regulatory subunit of CAMPdependent protein kinase and the catabolite gene activator protein are homologous. Proc. Natl. Acad. Sci. U.S.A. 79: 7679-7683. Weber, I.T., Steitz, T.A., Bubis, J., and Taylor, S.S. 1987. Predicted structures of cAMP binding domains of type I and type I1 regulatory subunits of CAMP-dependent protein kinase. Biochemistry, 26: 343-351.
REVIEW /
Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by 74.12.82.5 on 11/19/14 For personal use only.
Weber, I.T.,Shabb, J.B., and Corbin, J.D. 1989. Predicted structures of the cGMP binding domains of the cGMP-dependent protein kinase: a key alanine/threonine difference in evolutionary divergence of CAMPand cGMP binding sites. Biochemistry, 28: 6122-6127. Woodford, T.A., Correfl, L.A., McKnight, G.S., and Corbin, J.D. 1989. Expression and characterization of mutant forms of the
SYNTHBSE
1289
type I regulatory subunit of CAMP-dependent protein kinase. J. Biol. Chem. 264: 13 321 - 13 328. Yagura, T.S., and Miller, J.P. 1981. Mapping adenosine cyclic 3'3'-phosphate binding sites on type I and type I1 adenosine cyclic 3 ' ,5 '-phosphate dependent protein kinases using ribose ring and cyclic phosphate ring analogues of adenosine cyclic 3',5'-phosphate. Biochemistry, 20: 879-887.
