Structural Basis for Peptide Binding in Protein Kinase A: Role of Glutamic Acid 203 and Tyrosine 204 in the Peptide-Positioning Loop

For optimal activity the catalytic (C) subunit of cAMP-dependent protein kinase requires a phosphate on Thr-197. This phosphate anchors the activation loop in the proper conformation and contributes to catalytic efficiency by enhancing the phosphoryl transfer rate and increasing the affinity for ATP (1). The crystal structure of the C-subunit bound to ATP and the inhibitor peptide, IP20, highlights the contacts made by the Thr-197 phosphate as well as the role adjacent residues play in contacting the substrate peptide. Glu-203 and Tyr-204 interact with arginines in the consensus sequence of PKA substrates at the P-6 and P-2 positions, respectively. To assess the contribution that each residue makes to peptide recognition, the kinetic properties of three mutant proteins (E203A, Y204A and Y204F) were monitored using multiple peptide substrates. The canonical peptide substrate, Kemptide, as well as a longer nine-residue peptide and corresponding peptides with alanine substitutions at the P-6 and P-2 positions, were used. While the effect of Glu-203 is more localized to the P-6 site, Tyr-204 contributes to global peptide recognition. An aromatic hydrophobic residue is essential for optimal peptide recognition and is conserved throughout the protein kinase family.


Introduction
The predominant regulatory mechanism used by eukaryotic cells to convey a message from external stimuli is phosphorylation, mediated by protein kinases. These messages control regulation of diverse pathways in response to stress, antigen presentation, and development to name a few. Members of the protein kinase family are related through a structurally conserved catalytic core comprised of two lobes. The smaller N-terminal lobe dominated by β-sheets, is responsible for nucleotide binding, while the larger C-terminal lobe made up primarily of αhelices, relays substrate specificity (2). cAMP dependent protein kinase (PKA) is one of the simplest and best understood members of the protein kinase superfamily. It exists as an inactive holoenzyme complex consisting of a regulatory (R) subunit homodimer, and two catalytic (C) subunits. Upon increased levels of intracellular cAMP, each regulatory subunit cooperatively binds two molecules of cAMP inducing a conformational change resulting in the unleashing of the active catalytic subunits (3). The C-subunit is a 350 amino acid (4), 41kD protein with the conserved kinase core represented by residues 40-300. The core is flanked at the N-terminus by a 39 amino acid helical region and a 50 amino acid C-terminal tail, with each flanking region undergoing co-or posttranslational modifications. The simplicity of this molecule and its ability to define the conserved and active kinase core allows it to serve as a model for other enzymes in the family.
Several specific determinants contribute to recognition of substrates and physiological inhibitors by the active C-subunit. Most prominent is the positioning of Arg at the P-6, P-3 and contribute to peptide recognition are located primarily in the large lobe with many of the residues located specifically within the enzyme's activation segment.
The activation segment, broadly defined as residues 184 to 208, lies on the surface of the large lobe and is essential for organizing the entire enzyme's active site. The activation segment contains little secondary structure, yet includes several distinct functional regions. First is the Magnesium Positioning Loop, residues 184-187, which positions the magnesium essential for coordinating the γ phosphate of ATP. Residues 188-192 comprise β-sheet 9, the only element of regular secondary structure. This segment interacts with the A -helix outside the core as well as the essential phosphate on Thr-197 in the activation loop. The activation loop follows with residues 194-197. The activation loop is also a site of regulation for most members of the protein kinase family, where phosphorylation on one or two key Tyr, Thr, or Ser residues is required for optimal activity (7). Based on structural comparisons of several active and inactive protein kinases such as cdk (8)(9)(10), src (11), and hck (11)(12)(13), phosphorylation at these positions appears to change the conformation of the loop and arrange it in a position necessary for optimal activity (14,15). In the C-subunit of PKA the k cat is reduced and the K m for ATP is increased when Thr-197 is not phosphorylated (1). This leads to a 50-fold decrease in catalytic efficiency (k cat /K m ). The activation loop of the kinase resides in the large lobe and the phosphorylated residue in the activation loop of the C-subunit, Thr-197, makes several contacts within the large lobe (Arg-165 and Arg-189), as well as one of the few interactions between the large lobe and the small lobe in the closed conformation (His-87). The next region is the P+1 loop, 198-205.
The properly positioned P+1 loop contains regions that interact with the P+1 hydrophobic residue, as well as the P -2 a nd P -6 arginines of peptide substrates. Examination of a crystallographic molecular model consisting of the C-subunit bound to MgATP and the inhibitor peptide IP20, residues 5-24 of the protein kinase inhibitor (PKI), lends insight into the molecular nature of these interactions (16). The model demonstrates that a hydrophobic pocket is formed, where the side chains of Leu-198, Pro-202, and Leu-205 make the largest contribution to the pocket ( Figure 1A). Also in the P+1 loop Glu-203 forms a hydrogen bond with the P-6 Arg, and Try-204 forms a hydrogen bond with Glu-230 which directly interacts with the P-2 Arg. These residues contribute to recognition nodules where distal parts of the molecule come together. The residues that facilitate recognition of the P-3 Arg, Glu-127 and Try-330, lie outside the activation loop. Finally there are the conserved APE residues, 206-208. These residues serve as an anchor to the large lobe via interaction with Arg-280, another conserved residue in the large lobe.
To further define the role of the P+1 loop in the overall organization of peptide binding, in particular at sites other than the P+1 site, Ala scanning mutagenesis of the entire loop was carried out (17). We focus here on Glu-203 and Tyr-204 for further kinetic analysis. Our goals were to determine whether these residues contribute to localized recognition of the P-2 and P-6 Arg, respectively, or whether they contribute more globally to peptide recognition. An additional mutant, Y204F, w as engineered to assess the contributions of the aromatic ring without the hydrogen bonding to Glu-230. The steady-state kinetic parameters of the mutants were measured using several synthetic peptide substrates. The traditional heptapeptide substrate, Kemptide [LRRASLG], was assayed in addition to a longer nine-residue peptide [GRTGRRNSI]. The longer peptide was modified by substituting an Ala for Arg at the P-2 position or P-6 position.
The results indicate there is a greater communication among the sites of substrate recognition than was previously appreciated by simple examination of the crystal structure and lead us to define this segment more globally as the peptide-positioning loop. Mouse monoclonal anti-myc and anti-HA antibodies (Covance, Princeton, NJ). The C -subunit antibodies were generated as described (19) Antibodies to the phosphorylated Thr-197 were originally generated to the phosphorylated Thr-500 of PKC and were a gift from A. Newton (University of California, San Diego) (20). All peptide substrates were synthesized at the Peptide and Oligonucleotide Facility at the University of California, San Diego on a Milligen 9050 PepSyn peptide synthesizer using standard Fmoc methodology activator and purified by highperformance liquid chromatography. All DNA sequencing was performed with the ABI Prism 310 Genetic Analyzer from PE Applied Biosystems.

Site-Directed Mutagenesis of the PKA Catalytic Subunit. cDNA for the murine PKA
Cα-subunit in the bacterial expression vector pRESTB was used as a template for Kunkel-based site-directed mutagenesis as described previously (21). All mutations were made using the Muta-Gene kit as per the manufacturer's recommendations. DNA sequencing analysis confirmed the presence of the correct mutation.
Expre ssion of Murine PKA Catalytic Subunit. Wild-type and mutant C-subunits were expressed in the E. coli strain BL21(DE3). Cells were grown in YT medium containing 100µg/ml ampicillin at 37°C to an optical density at 600nm of 0.5-0.8, induced with 0.5mM isopropyl-β-D-thiogalactopyranoside (IPTG), incubated for an additional 6 hours at 24°C, collected by centrifugation and stored frozen. Cells were lysed with a French pressure cell (American Instruments) at pressures between 1000 and 1500 psi using 15ml of lysis buffer/L culture. Insoluble material was removed by centrifugation at 25,000xg at 4°C for 45 minutes.
Purification of Catalytic Subunit. Wild type and mutant proteins were purified using phosphocellulose chromatography and Mono S FPLC. Briefly, cells pellets were resuspended in lysis buffer (30mM MES pH 6.5, 1mM EDTA, 50mM KCl, 5mM β-mercaptoethanol), lysed, pelleted, diluted with cold water, and batch bound to P11 resin (1 g resin/L culture) overnight at 4°C. Resin was batch washed in running buffer (30mM MES pH6.5, 1mM EDTA, 5mM βmercaptoethanol) and eluted with running buffer containing potassium phosphate at 0, 50, 90, 250, and 500mM. Wild type C-subunit eluted at 90mM while the mutant proteins typically eluted at 250mM. Elutions were diluted with 3 volumes of cold water and bound to the Mono S 10/10 column. The proteins were eluted in 20mM potassium phosphate pH 6.5, 5mM βmercaptoethanol with a 0-500mM KCl gradient.
Catalytic Activity Assays. The kinetic values for the proteins were obtained by a direct phosphorylation filter-binding assay using [ 32 P]-γ-ATP (22). The assays were performed as described (23). Briefly, the C -subunit (0.25-1.0nM) was incubated in 50mM MOPS (pH 7.0), 0.1M KCl, 10mM MgCl 2 , 1mM DTT, 100µg/mL bovine serum albumin, 2.5µCi of [ 32 P]-γ-ATP, 1mM unlabeled ATP and peptide substrate. To determine the K m (ATP), peptide concentrations were held constant, and the total ATP was varied from 1.0 µM to 2.0 mM. To determine the K m 's for peptide substrate, the ATP concentration was fixed and the peptide substrate varied.
Reactions were initiated with the addition of peptide substrates and incubated at 30°C in a final volume of 50µL. Reactions were terminated with 20µL of 50% acetic acid. Aliquots were spotted on P81 filter disks and washed together in 0.5% phosphoric acid (4 times, 500mL, 10 minutes). Filter disks were rinsed once with acetone, air-dried and counted in 5mL of EcoLume.
Background reactions containing no peptide substrate were subtracted from all data. All reactions were performed in triplicate. Phosphorylation is assessed using antibodies specific for the phosphorylated Thr-197 The activity of these PDK-1 phosphorylated proteins was assayed using the PepTag PKA activity assay. This assay uses a fluorescent-tagged Kemptide substrate, where a change in net charge occurs upon phosphorylation. This change is detected by a shift in its direction of mobility when run on an agarose gel.

Results
Prior Ala scanning mutagenesis studies were performed to assess how the residues in the activation segment contribute to phosphorylation on  replacing the two arginines involved in substrate recognition at the P-2 and P-6 position, were thus used as comparative substrates. As expected, the added contacts offered by the larger substrate results in a 30-fold lower K m compared to Kemptide (Table 2). Alanine substitution at positions 203 and 204 result in 30-fold increases in K m compared to wild type C. These effects are similar to those observed for Kemptide and the alanine mutants. Finally, both alanine mutants have nominal effects on k cat using GRTGRRNSI as a substrate.
The P -2 substituted peptide, GRTGRANSI, displays the highest K m with the wild type enzyme compared to the other peptides ( Table 2). The K m 's are also elevated for both mutant proteins using this peptide, although the largest effect occurs with Y204A. The K m for GRTGRANSI is 2-and 30-fold larger for E203A and Y204A, respectively, than for wild type (Table 2). Although the K m values for the P-2 substituted peptide to wild type and the mutants are higher than those for GRTGRRNSI, the k cat values are similar. The P-6 substituted peptide, GATGRRNSI, shows a similar trend with E203A displaying higher affinity than Y204A. The K m for GATGRRNSI is 6-and 30-fold larger for E203A and Y204A, respectively, than for wild type ( The soluble fraction of these bacterial cell lysates was used as substrate material for phosphorylation by the Thr-197 kinase, 3-phosphoinositide dependent protein kinase-1 (PDK-1).
The success of the phosphorylation reaction was tested using antibodies specific for the phosphorylated form of Thr-197. Figure 2A depicts immunoblots of the material in the PDK-1 reaction. The C -subunit antibody shows that each reaction contained the same amount of Csubunit, and the phospho-Thr-197 antibody indicated that each is phosphorylated with similar efficiency. Any lack of activity relative to wild type will not be due to low expression or poor phosphorylation by PDK-1. Aliquots from the phosphorylation reactions were tested for activity toward Kemptide using a qualitative assay. Figure 2B indicates that even when phosphorylated on Thr-197, the perturbation of Gly-200 and Thr-201 abolishes activity.

Discussion
After carrying out a qualitative screen of alanine mutants made in the activation loop residues (17), two C -subunit mutants were selected for kinetic analysis based on their unusual kinetic parameters. The Y204A mutant showed reduced Kemptide activity in the qualitative PepTag assay, but was able to autophosphorylate when expressed in E. coli. This apparent contradiction led to a closer examination of its kinetic parameters. Replacement of Glu-203 with Ala led to a protein that was active and able to autophosphorylate. The structure suggests that the latter residue is involved in recognition of the P -6 a rginine (16), a residue not found in the Kemptide substrate. Kinetic analysis of these two mutants revealed that the P+1 loop, as well as these two specific residues, plays a global role in organizing the binding of peptide substrates.
Each residue not only contributes to a local site, but also shows more long range effects. This work redefines the P+1 loop and suggests that it should more appropriately be described as the peptide-positioning loop, as its contributions go well beyond recognition of only the P+1 residue.
In addition, the loop contributes either directly or indirectly to recognition of the P-site, the P-2 site and the P-6 site.
The steady-state kinetic parameters presented in this study explain why Y204A displays lower activity compared to wild type and E203A using the Pep-Tag assay. Although both mutants bind Kemptide with equivalent, poor affinity compared to wild type, the larger decrease in k cat /K m , the catalytic efficiency term, for Y204A compared to E203A is due to a l ower turnover number, k cat (Table 1). Indeed, the K i values for Ala-Kemptide are much higher than the K m 's for Kemptide confirming this point.
By studying the longer peptide substrate, GRTGRRNSI, and its derivatives we were able to better evaluate substrate recognition determinants. This peptide contains P -2 and P -6 arginines that can be used to assess the roles of Tyr-204 and Glu-203. While the latter interaction is direct, the former is mediated indirectly via Glu-230 ( Figure 1A). In general, the K m values for GRTGRRNSI are lower for all the enzymes studied, consistent with improved affinities of the longer peptides. Furthermore, the relative changes in K m values for the mutants follow those for the Kemptide K m 's. Alanine substitution at positions 203 and 204 lead to K m increases of 10-fold for Kemptide (Table 1) whereas these substitutions lead in K m increases of between 2 -and 30-fold for GRTGRRNSI and its derivatives ( Table 2). When either mutant protein was assayed with the substrate peptide l acking the corresponding arginine residue (i.e.-P-2 Ala substitution, Y204A; P-6 Ala substitution, E203A), elevations in K m are obtained. For example, the K m for GATGRRNSI is 7 -fold larger than that for GRTGRRNSI with wild type (Table 2). Furthermore, t he K m for GRTGRANSI with is approximately 30-fold larger than that for GRTGRRNSI and wild type ( Table 2). These results demonstrate that both Glu-203 and Tyr-204 contribute to peptide binding to an extent not predicted by the X-ray structure. For instance, the crystal structure clearly illustrates that the hydroxyl group of Tyr-204 interacts with Glu-230, which in turn is in hydrogen bonding distance from the substrate P-2 arginine. In light of the kinetic parameters for the Y204A mutant, it seemed likely that the aromatic ring of the tyrosine might also be contributing to peptide recognition perhaps through an interaction between the P-2 Arg and the π electrons of the aromatic ring. The Y204F mutant protein does indeed demonstrate that the aromatic ring contributes to substrate recognition. The consequence of losing this side chain is seen in the crystal structure of the Y204A mutant protein (28).
Glu-203 and Tyr-204 are involved in peptide recognition of a specific substrate residue, but also contribute to overall substrate recognition indirectly. The direct interactions are clearly shown in the crystal structure, by the proximity of Glu-203 to the P-6 Arg and the network of interactions involving P-2 Arg recognition, termed the P-2 nodule ( Figure 1B, Table 3). In the P-2 nodule Tyr-204 with Arg-133 is aiding the positioning of Glu-230, which interacts with the P-2 Arg. Additionally, Glu-170 in the catalytic loop helps coordinate the P -2 arginine. When the Glu-203 mutant is assayed with Kemptide the results show an increase in K m that is quite larger than expected for a mutant whose substrate determinant was not present. This observation suggests that Glu-203 influences peptide binding beyond the P-6 interaction, perhaps through the P+1 loop. The same holds true for Y204A. This mutation disrupts P -2 Arg binding on the substrate, but a peptide with an alanine substituted at this position increased the K m even more.
Perhaps the P-2 residue itself is contributing to the stability of the enzyme by coordinating those residues that are involved in its binding. Certainly the absence of the P-2 Arg would disrupt the network of the P-2 nodule.
Both Glu-203 and Tyr-204 lie within a sequence that forms a hydrophobic pocket that binds the substrate's P+1 hydrophobic residue. The positioning of this loop appears to be of critical importance. Hydrophobic residues are contributing to the pocket and the other nonhydrophobic residues contribute to the proper positioning of the loop and of the substrate. Table   4 lists the residues that make up the P+1 loop as well as the interactions in which they are involved. The Glu-203 and the Tyr-204 residues on this loop don't contribute to its hydrophobicity with their side chains, instead they are directed away from the hydrophobic pocket. The increased K m for mutant proteins with peptide substrates lacking their corresponding arginine may be due to disruption of the P+1 loop and other interactions it makes. Sequence alignments for members of the kinase family highlight the importance of this P+1 loop in substrate recognition (18). There are distinct sequence differences between Ser/Thr kinases and Tyr kinases in this loop. These changes reflect the need to accommodate the larger substrate tyrosine. The Thr-201 position is conserved as either a ser or thr in ser/thr kinases but is almost exclusively pro in the tyr kinases. The Glu-203 position is conserved in the AGC superfamily of kinases, but not in other Ser/Thr kinases, where as this position is basic in the Tyr kinases. The Tyr at 204 is somewhat conserved in the Ser/Thr kinases with some substitutions of Phe or Trp, but is always aromatic in the Tyr kinases. The inactivity of G200A and T201A even when phosphorylated, further demonstrates the importance of these individual residues as well as the global nature the contributions this loop makes.
The C -subunit is an enzyme that is poised to phosphorylate its substrates.
Crystallographic molecular models show the peptide-positioning loop on the surface of the enzyme, prepared to bind substrate. This is evident in its exhaustive list of substrates, which are involved in wide ranging cellular functions. This characteristic works for the C -subunit because it is also unique as a kinase in its mechanism of regulation. Although there are key sites of phosphorylation that are required for activity, regulation does not occur through the dynamic transfer of a phosphate on/off at the activation loop. Instead regulation occurs through the R subunits, and further through the A -Kinase-Anchoring-Proteins or AKAPs that target the PKA signal to the various parts of the cell. It will be at this level where substrate specificity will next be described.