Arginine 210 is not a critical residue for the allosteric interactions mediated by binding of cyclic AMP to site A of regulatory (RIalpha) subunit of cyclic AMP-dependent protein kinase.

The guanidinium groups of conserved arginines in the two intrachain cAMP-binding sites of regulatory (R) subunit of cAMP-dependent protein kinase have been implicated in the allosteric interactions by which cAMP binding leads to kinase activation. We have investigated the functional role of Arg-210, the conserved arginine in site A of murine type Iα R subunit, by analyzing the effects of nine different substitutions at this residue on cAMP binding and allosteric properties of bacterially expressed RIα subunits. All substitutions reduced the cAMP binding affinity of site A, but the magnitude of reduction varied from several hundredfold to 106-fold. The differential effects of the different substitutions could not easily be rationalized by interactions with cAMP and might, in part, reflect interactions with other residues in the unoccupied cAMP-binding pocket. None of the Arg-210 substitutions appeared to disrupt the allosteric interaction by which occupation of site A slows dissociation of cAMP from site B, although the effect was difficult to elicit in full with mutations that had strong effects on cAMP binding. The two weakest substitutions, Arg-210 → Ile and Arg-210 → Thr, could be shown to have essentially no effect on the allosteric interaction by which occupation of site A reduces the affinity of R subunit for the catalytic subunit. The weaker mutations had a smaller effect on kinase activation by the suboptimal activator Rp-adenosine cyclic 3′,5′-phosphorothioate than by cAMP, suggesting that the analog largely bypasses interactions with the guanidinium group of Arg-210.

(R) 1 subunit decreases by 5 orders of magnitude or more the affinity of R for the catalytic (C) subunit (reviewed in Ref. 1). Studies with analogs of cAMP implicate the ribose-cyclic phosphate moiety of cAMP in the activation process (2,3). The adenine ring is thought to contribute to the specificity and high affinity of cAMP binding (2,4). A recent crystal structure of a large fragment containing the cAMP-binding domains of R subunit corroborated predictions of an earlier model based on the structure of the catabolite activator protein of Escherichia coli that most of the important contact residues for interaction with the ribose-cyclic phosphate of cAMP are in ␤-strands 6 and 7 of an antiparallel ␤-roll structure forming one face of the cAMP-binding pockets (5,6). It appears from this structure that Glu-201 2 in site A and Glu-325 in site B interact with the 2Ј-hydroxyl group of cAMP and that the guanidinium groups of Arg-210 in site A and Arg-334 in site B interact with the equatorial exocyclic oxygen of the cAMP phosphate group (5). Mutations at these residues or at the conserved Gly residues immediately upstream of the conserved Glu residues markedly reduced the affinities of the mutated sites for cAMP (7)(8)(9)(10). Arg-242, in the long ␣-helix perpendicular to the ␤-roll structure for site A, appears to contribute to the stability of the site A-binding pocket by electrostatic interaction with Glu-201 (5) and is essential for allosteric interaction between sites A and B (11).
R p -phosphorothioate (R p -cAMPS) and dithioate analogs of cAMP have been described as antagonists of wild-type kinase (12)(13)(14). Actually, early studies suggested that R p -cAMPS was a partial agonist for cAMP-dependent protein kinase (15,16), and this view was reinforced by a more recent report showing that R p -cAMPS activates wild-type kinase in the absence of ATP or enzymes with mutant R subunits (Arg-210 3 Lys or Ala-98 3 Ser) in the presence of ATP (17). The apparent importance of the equatorial exocyclic oxygen of cAMP (missing in R p -cAMPS) for full kinase activation led to the suggestion that interaction between this oxygen and a positively charged amino acid side chain was responsible for the cAMP-dependent conformational change in the R subunit underlying the C subunit release (15). From the structure of the cAMP-binding sites, * This work was supported in part by National Institutes of Health, NIDDK Grant DK37583 (to R. A. S.), by grants from the Norwegian Cancer Association (to D. Ø.), and by a NATO collaborative research grant (to R. A. S. and D. Ø.). This paper is submitted by S. S. in partial fulfillment of the requirements for the degree of Doctor of Medicine. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ¶ Supported by a Fellowship from the Oklahoma affiliate of the American Heart Association. ** Student Fellow of the Norwegian Cancer Association. 1 The abbreviations used are: R subunit, regulatory subunit of cAMPdependent protein kinase; C subunit, catalytic subunit of cAMP-dependent protein kinase; R p or S p -cAMPS, R p or S p -adenosine cyclic 3Ј,5Ј-phosphorothioate; Bz-cAMP, N 6 -benzoyl-cAMP; MOPS, 3-(N-morpholino)propanesulfonic acid; bis-tris propane, 1,3-bis[tris(hydroxymethyl)methylamino]propane; dansyl, 5-dimethylaminonaphthalene-1sulfonyl. 2 Because the mouse RI␣ subunit used for these studies has one more amino acid in its amino-terminal domain than does the porcine homolog used by some other investigators, the residue numbers used in this this side chain would have to come from Arg-210 in site A or Arg-334 in site B (5).
To better evaluate the role of Arg-210 in cAMP binding and allosteric functions of site A, we introduced a variety of amino acid substitutions at this site in bacterially expressed recombinant murine RI␣ subunit. Ile-210, Ser-210, and Thr-210 mutations had been identified as spontaneous or mutagen-induced substitutions found in "K a " mutant isolates of S49 mouse lymphoma cells (19) and were transferred from mutant cDNAs to the bacterial expression plasmid. Asn, Asp, Gln, Glu, His, and Lys substitutions were generated by site-directed mutagenesis to assess the role of ionic and/or hydrogen-bonding interactions in the functions of Arg-210. Although the mutations all reduced the binding affinity of site A, the affinities of the mutant sites varied over a range of more than 1,000-fold. Using R subunits with the mutations that least affected cAMP binding, we found that the guanidinium group of Arg-210 is unnecessary for either the allosteric effect of cyclic nucleotide binding that couples the kinetic properties of site B with occupation of site A or the allosteric coupling by which occupation of site A decreases the affinity of R for C subunit.

Materials
Chemicals and Biochemicals- [2,8-3 H]cAMP (30 C i /mmol), [␥-32 P]-ATP (3000 C i /mmol), and Aquassure scintillation mixture were from DuPont NEN; the phosphate acceptor heptapeptide Leu-Arg-Arg-Ala-Ser-Leu-Gly (Kemptide) was synthesized and high performance liquid chromatography-purified by the Molecular Biology Resource Facility of the University of Oklahoma Health Sciences Center; N 6 -benzoyl-cAMP (Bz-cAMP), cGMP, ATP, MOPS, and bis-tris propane were from Sigma; R p -cAMPS was from BioLog Life Sciences Institute; and ultrapure urea and ammonium sulfate were from ICN Pharmaceuticals, Inc. The R p -cAMPS was purified further by high performance liquid chromatography on a reverse-phase C-18 column loaded in 20 mM triethylamine formate (pH 6.7) and eluted with 4% acetonitrile in this buffer. Stock solutions were checked for purity by high performance liquid chromatography and found to contain less than 0.1% contaminating cAMP when used in kinase activation experiments. The mutagenic oligonucleotide described below was synthesized by the Molecular Biology Resource Facility of the University of Oklahoma Health Sciences Center. Other chemicals were reagent grade or better and used without further purification.
Plasmids, Bacteria, and Expression of Kinase Subunits-Full-length nonfusion constructs for bacterial expression of murine C␣ and RI␣ subunits were made in the T7 promoter-containing vector pET-8c (or, for the R subunit plasmid, a variant lacking an EcoRI restriction site) as described previously (20,21). Glu-200, Ile-210, Ser-210, Thr-210, and Asp-324 mutations were introduced into the wild-type R subunit plasmid by cassette replacement using polymerase chain reaction-amplified cDNAs from S49 cell mutants (19). Asn-210, Asp-210, Gln-210, Glu-210, His-210, and Lys-210 mutations were generated by site-directed mutagenesis using the following degenerate oligonucleotide primer in a polymerase chain reaction-based mutagenesis protocol (22): 5Ј-GGAGCTTGGA(A/C/G)A(C/G)CTGGCTTTGA-3Ј. Selected mutations at position 210 were combined with the Asp-324 mutation by exchange of an EcoRI-MluI restriction fragment between appropriate plasmids. The sequences of kinase subunit genes in all plasmid constructs were verified by direct sequencing using a Sequenase II kit (U. S. Biochemical Corp.).
R and C subunits were expressed in E. coli BL21(DE3) and purified to near homogeneity as described previously (11,21). R subunits were rendered cAMP-free by gel filtration in urea solution followed by dial-ysis against R subunit storage buffer as described previously (11). For studies of fluorescence quenching, R subunits were purified further by affinity chromatography on N 6 -aminoethyl-cAMP-Sepharose (23). Samples were loaded in EB; the columns were washed successively with EB, EB plus 1 M sodium chloride, and EB; the R subunits were eluted with urea solution; and the purified subunits were renatured by dialysis against R subunit storage buffer.

Measurements of Site A Affinities by Endogenous Trp Fluorescence
Quenching-R subunits were diluted to 0.2-0.4 M in dissociation buffer without bovine serum albumin but with 10% w/v redistilled glycerol. Fluorescence measurements were made at 20°C with an SLM-Aminco model 4800 Spectrofluorometer using 0.4-cm quartz cuvettes. Samples were excited either at 293 or 300 nm to minimize absorbance of cAMP, and equivalent results were observed for those R subunits tested at both excitation wavelengths. Emission intensities were monitored through Schott WG-345 and Corning 7-54 filters. For titrations, cAMP was increased incrementally by additions of small volumes of stock cAMP solutions, and intensities were corrected for dilution of R subunit. (Final volumes of samples were less than 10% greater than initial volumes.) Assay of Dissociation or Exchange of [ 3 H]cAMP-For studies of the effect of cyclic nucleotide concentrations on site B off-rates, R subunits at about 20 nM in Hepes buffer were incubated with 30 nM [ 3 H]cAMP for 2 h at 30°C and then diluted 30-fold into Hepes buffer containing 150 mM or 3.2 M sodium chloride and unlabeled cAMP, cGMP, or Bz-cAMP to give concentrations indicated in Figs. 3 and 4. Samples taken after various intervals of incubation at 30°C were precipitated with ammonium sulfate and filtered to determine bound radioactivity as described by Døskeland and Øgreid (24). Dissociation half-times for site B were determined graphically from plots of bound radioactivity versus time.
Assays of Protein Kinase Activity-C subunit activity was monitored by the transfer of 32 P i from [␥-32 P]ATP as described previously (8,11).

RESULTS
Initial characterization of mutants with substitutions at Arg-210 suggested that site A function was compromised, but affinities of the mutant sites could not be estimated with our standard filter-binding assay for bound [ 3 H]cAMP. Fig. 1 shows the results from an alternative assay based on the finding that cAMP-dependent quenching of the endogenous Trp fluorescence of R subunit results solely from the interaction of cAMP with site A (11,25). The R subunit concentrations (0.2-0.4 M) were apparently near or above the dissociation constants (K d values) for cAMP binding to site A of the Ile-210 and Thr-210 mutants, but, for the other R subunit preparations, the Trp fluorescence quenching could be titrated with increasing concentrations of cAMP to yield apparent K d values. In terms of relative effects on site A affinity, the mutations could be ordered as follows: (Thr, Ile) Ͻ Lys Ͻ Asn Ͻ His Х Ser Ͻ Asp Ͻ Gln Ͻ Glu. Apparent K d values ranged from about 0.6 M for the Lys-210 mutant R subunit to 1 mM for the Glu-210 mutant R subunit. Fits of the data of Fig. 1 with a cooperative binding model suggested Hill coefficients of about 1.3 for most of the mutants.
Despite numerous attempts to determine site A affinities for the Ile-210 and Thr-210 mutants by [ 3 H]cAMP-binding assays, we were ultimately unsuccessful. The rapid off-rates of the mutant sites compromised retention of labeled nucleotide in the standard filter-binding assay, and data from equilibrium dialysis experiments were ambiguous at best. 3 In attempts to estimate more accurately the effects of the Ile-210 and Thr-210 mutations on site A binding, we combined the position 210 mutations with a Gly-324 3 Asp mutation, which inactivates site B (10, 11). Fig. 2 shows activation curves of kinase reconstituted by incubating these double mutant R subunits with purified C subunit. Fig. 2A compares the effects of the Ile-210/ Asp-324 and Lys-210/Asp-324 mutations with that of the Asp-324 mutation alone on activation by Bz-cAMP. For this experiment, we used relatively low concentrations of R and C subunits to facilitate activation of enzyme with the Lys-210 mutation, and the effect of the Lys-210 mutation on apparent activation constant (K a ) was about 20 times greater than that of the Ile-210 mutation. Fig. 2B shows that the effects of the Ile-210 and Thr-210 mutations on apparent K a values were about the same. For this experiment the ratio of R to C subunits was raised to assure complete reconstitution. Under the conditions of Fig. 2, both A and B, the Ile-210 mutation caused an increase of about 100-fold in apparent K a when compared to the effect of the Asp-324 mutation alone. Equilibrium dialysis experiments with the Ile-210/Asp-324 and Thr-210/Asp-324 double mutant R subunits yielded K d values for [ 3 H]cAMP that were on the order of about 0.5 to 1 M, but the conditions for these experiments (overnight at 4°C) were different from those used for fluorescence quenching, and we were unable to achieve saturation with the Lys-210/Asp-324 mutant R subunit for comparison. 3 Furthermore, we suspect that the site B mutation causes a decrease of about 4-fold in site A affinity based both on our previous data comparing the cAMP binding of wild-type and Asp-324 mutant R subunits (11) and on the observation that the Ile-210/Asp-324 double mutant R subunit gave an apparent K d of 0.3 M by tryptophan fluorescence quenching. 3 Taken together, all of our data are consistent with the Ile-210 and Thr-210 mutations having about an order of magnitude less effect on cAMP-binding affinity than the Lys-210 mutation.
Figs. 3 and 4 present evidence that mutant R subunits with substitutions at Arg-210 retain the allosteric interaction by which occupation of site A retards dissociation of cAMP from site B (9).  -210 and Glu-210) mutations. For wild-type and all the mutant R subunits, site A affinities estimated by the assays of intrachain coupling between sites A and B were substantially lower than those determined by more direct assays of cAMP binding (above).
The apparent discrepancies between cAMP concentrations required to saturate site A (Figs. 1 and 2) and those required to slow cAMP dissociation from site B (Fig. 3) suggested that something more than simple occupation of site A might be required for the allosteric effect on site B. Furthermore, it was unclear from the experiments of Fig. 3 whether or not the stronger mutations disrupted partially the allosteric coupling between sites A and B. In an attempt to understand better how intrachain coupling was related to occupation of site A, we repeated the coupling experiments using either Bz-cAMP or cGMP as the ligand for site A. Bz-cAMP binds to site A with about 3.5-times, and cGMP with about 0.005-times, the affinity of cAMP (4). Fig. 4 shows the data from assays in high salt buffer for R subunits with representative weak (Ile-210), intermediate (Lys-210), or strong (Gln-210) mutations. Consistent with their relative affinities for site A, Bz-cAMP shifted the dose-response curves to the left, and cGMP shifted the curves to the right. The magnitude of the dose-response shift for Bz-cAMP on the Lys-210 mutant subunit was much greater than that on the Ile-210 mutant subunit. Bz-cAMP at 1 mM was able to retard site B dissociation from the Lys mutant subunit to almost the same extent as cAMP in wild-type, Ile-210, or Thr-210 R subunits, and cGMP at 10 mM was able to achieve an equivalent effect for the Ile-210 mutant subunit. Although Bz-cAMP also shifted the apparent dose-responses for intrachain coupling in the Gln-210 mutant R subunit, it did not produce appreciably more coupling than cAMP at the maximum concentrations tested.
To investigate the role of Arg-210 in the allosteric coupling reaction that links occupation of site A with a decrease in the affinity of R for C subunit, we again used R subunits with a functionally inactivated site B. Fig. 5 shows that the Asp-324 mutant R subunit and the Ile-210/Asp-324 or Thr-210/Asp-324 double mutant R subunits all retained high affinity binding for the C subunit in the absence of cAMP. In this experiment the R subunit at various concentrations was incubated briefly with the C subunit in the presence of ATP before adding Kemptide to measure C subunit activity. Because the concentration of the C subunit required for the activity assay is above the apparent K d for the interaction between wild-type R and C subunits, this were preincubated with C subunit for 5 min at 30°C and then mixed with Kemptide and [␥-32 P]ATP for assay of residual protein kinase activity as described under "Experimental Procedures." The final concentration of C subunit was 0.75 nM, and assays were incubated for 10 min at 30°C. R subunit concentrations shown were those in the final assay mixtures.

FIG. 3. The effect of cAMP binding to site A on dissociation of [ 3 H]cAMP from site B in R subunits with mutations at Arg-210.
Dissociation of [ 3 H]cAMP bound to wild-type or mutant R subunits was assayed in the absence or presence of competing unlabeled cAMP to give the total cAMP concentrations indicated as described under "Experimental Procedures." For panel A dissociation was in 0.15 M sodium chloride, and for panel B dissociation was in 3.2 M sodium chloride. Symbols for the different mutant R subunits are as for Fig. 1. Wild-type data are indicated by open diamonds. Because bound cAMP was a significant proportion of the total for the lowest points of the wild-type curves, free cAMP concentrations were calculated for these curves and used for the plots shown. The other plots are to total cAMP Ϸ free cAMP. Dissociation half-times for wild-type R subunits at 1 nM total cAMP (ϳ0.5 nM free) were 4.5 min in low salt and 6 min in high salt (data not shown). assay does not give true K d values but is sensitive to changes in association kinetics. All three R subunit preparations inhibited the C subunit with similar dose responses, suggesting that the basal interaction between R and C subunits was not significantly impaired by the mutations at Arg-210. Fig. 6 shows the results of similar titration experiments performed in the presence of high levels of Bz-cAMP. For these experiments the preincubation and assay times were extended so that the kinase reactions proceeded, for the most part, after the subunits had reached their equilibrium states of association. All three concentrations of Bz-cAMP had similar effects on the interaction of the Asp-324 mutant R subunit with the C subunit, suggesting that the wild-type site A in this subunit remained saturated with the cyclic nucleotide at concentrations at or above 0.6 mM (data for only 15 mM Bz-cAMP is shown). In contrast interaction of the double mutant R subunits with C subunit was weakened progressively as the cyclic nucleotide was increased from 0.6 to 15 mM. At 15 mM Bz-cAMP, the apparent affinity of the Ile-210/Asp-324 R subunit for C subunit was about the same as that for the cyclic nucleotide-saturated Asp-324 mutant R subunit, and that of the Thr-210/Asp-324 R subunit was only slightly higher.
Because the experiments of Fig. 6 suggested that the guanidinium group of Arg-210 was not essential for the coupling between site A and the C subunit-binding site, we decided to reinvestigate the activation potential of R p -cAMPS. Fig. 7 com-pares activation profiles for kinase reconstituted with wildtype or mutant R subunits using either cAMP (Fig. 7A) or R p -cAMPS (Fig. 7B) as an activator. Because the analog slowly hydrolyzes to cAMP during storage, we used a preparation that had been freshly repurified by high performance liquid chromatography; in other experiments we pretreated the analog with cAMP phosphodiesterase and obtained qualitatively similar results. 3 Under the conditions of these assays, cAMP gave essentially full activation with the wild-type, Thr-210, Lys-210, and His-210 enzymes and partial activation with the Gln-210 enzyme. The apparent K a values for all the mutant enzymes were more than 10-fold higher than that for wild-type kinase. Since activation in the mutant enzymes was largely through site B, the differences in apparent K a values were small, but the rank order for all but the Gln-210 mutant was consistent with the binding data of Fig. 1. R p -cAMPS activated all the kinase holoenzymes assayed, but suboptimally. In contrast to the results with cAMP, there was little difference in the apparent K a values for activation of the wild-type and Thr-210 and Lys-210 mutant enzymes with R p -cAMPS. Results similar to those for the Gln-210 mutant were also obtained using an R FIG. 6. The effect of Ile-210 and Thr-210 mutations on the cyclic nucleotide-mediated reduction in affinity of R for C subunit. R and C subunits were mixed in the presence of ATP and Bz-cAMP as described under "Experimental Procedures" and incubated for 1 h at room temperature. Kemptide and [␥-32 P]ATP were then added, and the samples were incubated for an additional hour at 30°C before processing to measure phosphate transfer. C subunit was at 0.33 nM in the reactions, and the final concentrations of Bz-cAMP were 0.6 (q), 3.0 (f), or 15 mM (å, Ç).  7. Activation of wild-type, Thr-210, Lys-210, His-210, and Gln-210 holoenzymes by either cAMP or R p -cAMPS. R and C subunits were reconstituted overnight at 4°C at 75 times their final concentrations as described under "Experimental Procedures." For assay, the reconstituted mixtures were diluted 25-fold, and 10-l portions were added to 20 l of mixture containing Kemptide, [␥-32 P]ATP, and cyclic nucleotide. Final concentrations of R and C subunits were 20 and 3.3 nM, respectively, and incubations were for 15 min at 30°C. R subunits were wild-type (q) or had Thr-210 (Ⅺ), Lys-210 (å), His-210 (f), or Gln-210 mutations (Ç). An additional set of incubations contained C subunit alone (E). subunit in which site A was inactivated by a Glu-200 mutation. 3 The poor activation of these mutant enzymes by the analog suggested that the activation observed with wild-type or the other mutant enzymes required the function of site A as well as of site B.

DISCUSSION
The cAMP-occupied site A-binding pocket is a compact structure stabilized by both intrachain interactions and interactions with the bound cAMP (5). The ribose phosphate moiety of cAMP is apparently anchored by interactions of its 2Ј-hydroxyl group with the ␥-carboxyl group of Glu-201 and the ␣-amino group of Gly-199, and of its equatorial exocyclic oxygen with the guanidinium group of Arg-210 and the ␣-amino group of Ala-203. The guanidinium group of Arg-210 has additional interactions with the backbone carbonyls of Asn-172 and Gly-200 and the ␤-carboxyl group of Asp-171 to stabilize the binding pocket and, perhaps, transmit signals to adjacent regions. In view of the importance of the guanidinium group for these interactions involving Arg-210, we were surprised to find that Ile or Thr substitutions for Arg-210 were relatively benign in their effects on cAMP binding ( Figs. 1 and 2). Ile at position 210 had less effect on site A function than the basic amino acids Lys or His, suggesting that hydrophobic interactions involving the stem of Arg-210 might be important for cAMP binding. These could be either direct interactions with cAMP or, more likely, intrachain interactions stabilizing the cAMP-bound conformation of the binding pocket. The importance of such hydrophobic interactions is also supported by the tighter binding of cAMP to site A of the Thr-210 mutant R subunit than to that of the Ser-210 mutant protein. Substitution of the negatively charged Glu for Arg-210 decreased site A affinity by almost 10 6 -fold, but this effect could not be attributed simply to electrostatic repulsion of cAMP; Asp had an effect that was weaker by nearly 2 orders of magnitude than that of Glu, and Gln had an effect that was stronger than that of Asp. The stronger effects of Glu-210 and Gln-210 mutations than those of Asp-210 and Asn-210 mutations suggest either steric interference between these larger amino acids and cAMP or novel hydrogen-bonding interactions involving the side groups of these amino acids that stabilize the unoccupied conformation of the site A cAMPbinding pocket.
In a previous study we showed that allosteric interaction between sites A and B required Arg-242 and suggested that this interaction involved communication through the 2Ј-hydroxyl group of cAMP and Glu-201 (11). The site A structure shows clearly that Glu-210 and Arg-242 are positioned to effect this interaction and suggests further, that Trp-261 is also involved (5). Consistent with this view of coupling between sites A and B proceeding from interactions at the 2Ј-hydroxyl group of cAMP, the mutations at Arg-210 did not obliterate A-B coupling, and a full coupling response was observed with the Ile-210 and Thr-210 mutant subunits (Fig. 3). The effects of the various residue 210 substitutions on the cAMP-mediated reduction of site B off-rate were related to their effects on site A affinity, but not in a simple linear fashion. Only the Ile-210 and Thr-210 mutants exhibited wild-type levels of coupling with cAMP, and the Lys-210 mutant could be fully coupled with Bz-cAMP (Figs. 3 and 4). For the more severely impaired mutants and the Lys-210 mutant with cAMP or cGMP, the shallow dose-response curves left it unclear whether or not even saturating levels of cyclic nucleotide would elicit coupling to wildtype levels. Wild-type R subunit cAMP concentrations effective for the coupling interaction were about 10-fold higher than expected from the site A-binding affinity, and the apparent discrepancies between site A-binding affinities and effective concentrations for coupling were even greater for the Ile-210, Thr-210, and Lys-210 mutant proteins. Bz-cAMP and cGMP shifted the dose responses for coupling in the directions predicted from their relative affinities for site A, but the magnitudes by which the analogs shifted the curves varied among the different mutants. Taken together, these observations suggest that occupation of site A is necessary but not sufficient for the slowing of site B dissociation. Because of the relationship between the effects of various residue 210 substitutions on cAMPbinding affinities and intrachain coupling, we postulate a conformational equilibrium between two or more forms of cAMPbound R subunit, where higher site A-binding energies and/or longer residence times promote the fully coupled conformation. It remains possible, however, that the stronger mutations interfere directly with the coupling process. Positive cooperativity resulting from interchain interactions between the two sites A in R subunit dimers has been demonstrated in a previous study of wild-type R subunit (26). This could account for the apparent Hill coefficients of about 1.3 in the cAMP-binding curves of Fig. 1 and in the relatively steep titration curves (Hill coefficients ϳ1.3-1.6) for wild-type, Ile-210, and Thr-210 R subunits in Fig. 3A. The slopes for the Ile-210 and Thr-210 mutant subunits in Fig. 3B yielded Hill coefficients closer to one, suggesting that high salt might inhibit this interchain interaction in the mutant subunits.
The allosteric reaction important for kinase activation is that by which cAMP binding to sites A and/or B reduces the affinity of R for C subunit. Assuming that the site A mutations have no effect on the affinity of R for C subunit in the absence of cAMP, the magnitudes of this allosteric effect in different R subunits can be compared by measurements of the affinity of R for C subunit in the presence of saturating cyclic nucleotide. Although we have not been able to measure accurately the affinities of cAMP-free R subunits for ATP-bound C subunit, rapid reassociation assays suggested that both the single and double mutant R subunits used in the present studies retained high affinity binding for C subunit (Fig. 5, and data not shown). Using R subunits with an Asp-324 mutation to prevent cyclic nucleotide binding to site B, we were able to compare the effects of Bz-cAMP occupation of wild-type or mutant site A on the R-C interaction. Several concentrations of cyclic nucleotide were used to approach saturation of wild-type, Ile-210, and Thr-210 sites A. For all three R subunit types, the apparent K d values for C subunit binding in the presence of saturating Bz-cAMP were about 1-2 ϫ 10 Ϫ6 M (Fig. 6) or about 4 orders of magnitude higher than the subnanomolar K d values measured in the absence of cyclic nucleotides (27). 4 Previous studies of R subunits with an Arg-210 3 Lys mutation concluded not only that the guanidinium was essential for high affinity binding to site A (7), but also that it played a role in stabilizing ATP-bound holoenzyme (17,18). Based on the results of experiments testing kinase activation by thio substituted derivatives of cAMP, it was argued that Arg-210 functioned as part of a "molecular switch" where stereospecific competition between the exocyclic oxygens of cAMP and bound ATP for the guanidinium group of Arg-210 would lead to displacement of ATP from the holoenzyme complex and, thereby, promote subunit dissociation through a further cAMP-mediated conformational change. With wild-type holoenzyme in the 4 To measure basal R-C affinities, we reconstituted holoenzyme from a 2,5-dansyl-labeled C subunit and an amino-terminally truncated R subunit (that could not dimerize, but interacted normally with C subunit and cAMP) and used fluorescence polarization to monitor pressureinduced dissociation. We found a K d at atmospheric pressure of about 1 nM in the absence of ATP, and this was shifted to about 10  presence of ATP the S p -, but not the R p -, stereoisomer of cAMPS was an effective agonist. Both isomers were agonists in the absence of ATP. Holoenzyme reconstituted with a Lys-210 mutant R subunit could be activated by R p -cAMPS in both the presence and absence of ATP (17). In our experiments, rather than facilitating activation by R p -cAMPS, the Lys-210 mutation partially suppressed activation with the analog (Fig. 7). Wild-type holoenzyme could be partially activated by R p -cAMPS in the presence of ATP with the level of activation dependent on the experimental conditions, particularly the concentrations of kinase subunits (Fig. 7, and data not shown). Holoenzymes formed with wild-type or mutant R subunits with the weak Ile-210 or Thr-210 substitutions were activated to similar levels by R p -cAMPS, and the Lys-210 mutant enzyme was activated somewhat less (Fig. 7, and data not shown). Holoenzymes with the stronger site A mutations His-210 and Gln-210 exhibited even less activation by R p -cAMPS, thus demonstrating the importance of site A function for activation by this analog. While the apparent activation constants for cAMP were about 10-fold higher for Ile-210 and Thr-210 mutant enzymes than for the wild-type species, the difference in apparent activation constants with R p -cAMPS as activator was only about 2-3-fold. This suggests either that the guanidinium group of Arg-210 is more important for binding of cAMP than of R p -cAMPS or that the guanidinium group actually interferes with binding of R p -cAMPS. The partial agonist activity of R p -cAMPS reported in two early studies (15,16) has been attributed to contamination of the analog with cAMP (12). For our experiments the analog was purified as described for the pure preparations of previous reports (e.g. Ref. 12), and the differences in relative resistances of wild-type and mutant enzymes to activation by cAMP and R p -cAMPS argues against the possibility that the activation seen with the analog resulted from contaminating cAMP. We suspect that our detecting partial agonist activity with the wild-type enzyme results from the combination of relatively low enzyme concentration and high Kemptide concentration in our experiments when compared with those that detected only antagonist activity (e.g. Refs. 12 and 17).
Our results indicate that the guanidinium group of Arg-210, while important for high affinity binding of cAMP to site A, plays little if any role in the allosteric responses provoked by cAMP binding to site A. As discussed above, it is not surprising that Arg-210 is unnecessary for the kinetic coupling of site B dissociation with site A occupation. On the other hand, that substitution of Ile or Thr for Arg-210 had little or no effect on the allosteric response by which cyclic nucleotide binding to site A reduces the affinity of R for C subunit was quite unexpected. It would appear possible that it is the interactions of the exocyclic oxygens of cAMP phosphate with the backbone amides of Ala-203 and Ala-211 (5) rather than those with the guanidinium group of Arg-210 that are important for allosteric transmission. The sulfur substituent in R p -cAMPS reduces the binding affinity to site A of RI subunit by almost 800-fold, where the axial substitution in S p -cAMPS reduces this affinity by only about 5-fold (28), consistent with steric interference between the guanidinium group of Arg-210 and the equatorial sulfur. The suboptimal activation potential of the R p -cAMPS analog may result from both the inability of its reduced binding energy to maintain the R subunit conformation with lowest affinity for C subunit and the failure of its equatorial sulfur substituent to interact effectively with the backbone amide of Ala-203.