One amino acid change produces a high affinity cGMP-binding site in cAMP-dependent protein kinase.

Discrimination between cAMP and cGMP is a critical feature of cAMP- and cGMP-dependent protein kinases. An alanine/threonine difference in the cyclic nucleotide-binding sites has been proposed to provide a structural basis for this functional distinction. Site-directed mutagenesis of this alanine to a threonine in a cAMP-binding site of cAMP kinase produced a mutant with markedly increased cGMP affinity as determined by cGMP binding and protein kinase activation assays. Studies of other mutants at this position support the role of the threonine hydroxyl group as the component that enhances cGMP binding affinity.


Discrimination
between CAMP and cGMP is a critical feature of CAMP-and cGMP-dependent protein kinases.
An alaninelthreonine difference in the cyclic nucleotide-binding sites has been proposed to provide a structural basis for this functional distinction.
Sitedirected mutagenesis of this alanine to a threonine in a CAMP-binding site of CAMP kinase produced a mutant with markedly increased cGMP affinity as determined by cGMP binding and protein kinase activation assays. Studies of other mutants at this position support the role of the threonine hydroxyl group as the component that enhances cGMP binding affinity.
CAMP and cGMP regulate many cellular processes in eukaryotes by activating CAMP-and cGMP-dependent protein kinases (CAMP and cGMP kinases)' (1). Although both cyclic nucleotides can activate either enzyme, CAMP kinase has 200fold greater affinity for CAMP than for cGMP whereas cGMP kinase has the inverse selectivity (2, 3). Structures of the two CAMP-binding sites in each of the mammalian CAMP kinase isozymes types Ia! and 11~~ have been predicted based on homologies with the bacterial CAMP-binding catabolite gene activator protein (4) for which the crystal structure is known (5) (Fig. 1, top). In these models, key arginine, glutamic acid, and glycine residues make contacts with the cyclic nucleotide ribose phosphate moiety. Mutagenesis of these residues has confirmed their importance in the CAMP-binding sites (6-10). Mammalian cGMP kinase binding sites have also been predicted to maintain the same interactions (11). The CAMP and cGMP kinase binding site models are distinguished by an alanine/threonine difference that has been predicted to be critical for determining cAMP/cGMP binding specificity (11). The threonine hydroxyl group in cGMP kinase binding sites is proposed to hydrogen-bond with the guanine e-amino group of cGMP (Fig. 1, Vol. 265, No. 27, Issue of September 25, pp. 16031-16034, 1990 0 1990 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S. A. tance of this alanine/threonine difference in determining cAMP/cGMP binding selectivity was studied using site-directed mutagenesis to alter one of the two CAMP-binding sites in the regulatory (R) subunit of CAMP kinase. for 30 min on ice in the dark, exposed to UV light for 30 min, subjected to electrophoresis on a 10% sodium dodecyl sulfatepolyacrylamide gel, and autoradiographed (23). Native R is the purified bovine lung type In R subunit. coli strain TGl. Bacterial extracts were fractionated by anion exchange and molecular sieve chromatography.

Mutagenesis
The partially purified wild-type and A334T proteins co-eluted with bovine lung type Icu R subunit by molecular sieve chromatography,' consistent with the recombinant R subunit preparations being full length and dimeric, although both exhibited a slightly higher electrophoretic mobility than the native R subunit (Fig. 2). Biphasic [,'H]cAMP dissociation curves were detected for both the wild-type and A334T R subunits at 25 "C ( Fig. 3, top). The steeper slope represented dissociation from the fast site whereas the shallower slope represented the slow site (14). The dissociation behavior of ["HIcAMP from the slow site of A334T was indistinguishable from that of the wildtype R subunit (Table I). Dissociation from the fast site was also very similar (t,,? in min f SE., n = 3: wild type = 3.8 + 0.3; A334T = 3.4 f 0.2). These values were within the ranges reported for the slow (X-70 min) and fast (3-9 min) sites of type I R subunit (8-10, 18).
The effect of the alanine to threonine mutation on cyclic nucleotide binding selectivity became apparent when ["HI cGMP dissociation rates were measured for the wild-type and A334T R subunits (Fig. 3, bottom)  S.E., n = 3, and were determined from results like those in Fig. 3. K, values are -C S.E., n = 4, and are calculated from results like those in Fig. 4 f 1 S.E., n = 3). These results are consistent with a very high affinity binding of cGMP to the A334T slow site. A334T may surpass cGMP kinase in this respect for two reasons. First, the CAMP-binding sites of CAMP kinase may be inherently better designed for cyclic nucleotide binding, and the addition of a strong positive cGMP binding determinant in A334T might make it a more ideal cGMP site than either site in cGMP kinase. Second, interaction between the CAMP kinase R and C subunits is known to cause at least a 200-fold increase in the exchange rate of CAMP (19). This same phenomenon may also occur with cGMP kinase, but while cyclic nucleotide dissociation from the R subunit can be measured in the absence of C subunit interaction, this cannot be effectively accomplished with cGMP kinase, which has fused regulatory and catalytic components.
Protein kinase activation constants (&) were also measured as another indication of relative cyclic nucleotide binding affinities. An ll-12-fold molar excess of either wild-type or A334T R subunits was required to inhibit 85% of native bovine heart CAMP kinase C subunit activity. The resultant inactive holoenzymes were used to determine the K,, values for CAMP and cGMP (Fig. 4). K, determinations were not greatly affected by the excess R subunit since a 5-fold increase of R subunit over the standard amount elevated the apparent K, for CAMP or cGMP by only 25%. Wild-type and A334T holoenzymes had similar K,, values for CAMP, which were within the reported range of 0.008-0.06 PM (8,10,[20][21][22]. In contrast, cGMP K, values indicated that A334T had a 16-fold higher affinity for cGMP (Table I). Although this difference was substantial, it was less than the 115-fold difference between their slow site cGMP dissociation rates. The K, represents an average of cyclic nucleotide affinities for the slow (mutated) and fast (nonmutated) sites (21) and could account, in part, for this discrepancy. Thus, the lower K, for cGMP of A334T relative to wild-type R subunit was largely due to a marked decrease in the rate of dissociation of this cyclic nucleotide from the slow site of the mutated protein.
The binding affinity of cGMP for CAMP kinase depended upon the amino acid that occupied position 334 (Table I). Substitution of a serine (A334S) produced a slow site dissociation rate for cGMP that was 50-fold slower than that of the wild-type slow site, but half that of A334T. The intermediate cGMP binding activity was also reflected in the K, value of this mutant. Since serine also has a hydroxyl group, its interaction with cGMP may be similar to that of threonine, albeit with a slightly reduced affinity. Like A334T, A334S did not show an appreciable change in its interaction with CAMP relative to the wild-type protein. On the other hand, substitution of Ala-334 with glycine (A334G) was deleterious to both cGMP and CAMP binding. This may be due to the flexible nature of glycine in the peptide backbone, which could allow conformational distortion of the slow site. Decreased CAMP interaction of this mutant extended to the fast site, even though the mutation was only in the slow site (fast site tlh in min + S.E., n = 3: wild type = 3.8 + 0.32; A334G = 1.3 + 0.15). This observation was consistent with other studies which demonstrated that mutations in one R subunit binding site affect the CAMP binding properties of the nonmutated site (S-10).
Other studies of CAMP kinase CAMP-binding site mutants have identified amino acid residues that, when changed, either decreased or altogether eliminated CAMP binding (6-10, 22). In contrast, the mutations at Ala-334 provide the first examples of either an increase in cyclic nucleotide interaction or an alteration in the specificity of CAMP kinase in a predictable amino acid-dependent manner. Site-directed mutagenesis of this amino acid residue further supports the importance of hydrogen bonding potential near the 2-position of the purine moiety for determining cAMP/cGMP binding specificity of cyclic nucleotide-dependent protein kinases. The alanine/ threonine difference described here was an apparently critical step during the evolutionary divergence of these two kinases. Together with changes in the catalytic domains, the mutation would explain the emergence of separate physiological functions for CAMP and cGMP kinases.