Specificity Features of the ATP Binding Site
The ATP binding site of protein kinases consists of two subsites with very different properties: the triphosphoryl subsite contains most of the invari-antly conserved residues and is thus entirely conserved. All residues with sidechain contacts to the ATP molecule are either charged or electrophilic. The contacts of the triphosphoryl group to the enzyme form an extensive electrophilic network that involves the conserved sidechains either directly or via metal ions, and also involves several backbone atoms from the gly-cine-rich loop. Both the interactions with conserved sidechains and the loop interactions, because of the high structural homology of the protein kinases, seem likely to be conserved. Structures from other protein kinases with ATP or ATP analogues, however, showed that these contacts are not always identical. This is especially true for the metal ions, and the contacts to the gly-cine-rich loop (Russo et al. 1996; Hubbard 1997; Aubol et al. 2003; Nolen et al. 2003). Apparently, although at least within each of the subfamilies of the Ser/Thr and the Tyr-specific kinases amino acid residues in contact with the triphosphoryl group are strictly conserved, structural or electrical differences appear to exist and thus might contribute to the selectivity of suitable kinase inhibitors. In contrast to this highly conserved triphosphoryl subsite, the adenosine binding site contains only one invariant residue, the homologue of Val57, located at the C-terminus of the glycine-rich motif in PKA, and one very conserved residue, the homologue of Ala70 from b-strand 3. All other residues with sidechain contacts to ATP are variable but conservatively exchanged in the kinase family. Charged residues exist only in the ri-bose-hydroxyl subsite, Glu (corresponding to Glu127 of PKA), and Asp residues are common here. Apart from that, few residues are hydrophilic, such as Thr183 (PKA) and most are hydrophobic. The conserved polar interactions of adenosine in the binding site are again with backbone atoms, in PKA from N1 of the adenosine moiety to the amide of Val123 and from N6 to the carbonyl of Glu121. A functional reason for the variability of residues in the ATP site is not obvious. As all kinases bind ATP, differences might affect ATP binding and adenosine diphosphate (ADP) release rates, and could conceivably affect catalytic rates as well. As kinase (down)regulation often is accompanied by significant structural changes (Engh and Bossemeyer 2001; Huse and Kuriyan 2002), a specific role of ATP-site residues in inactivated conformations might also exist. A correspondence between the rotamer conformation of a residue on the enzyme surface and the sidechain change of a residue in the adenosine binding site has been shown recently with mutants of PKA. Gln181, usually in sidechain contact with hydrophilic residues on the surface, changed its conformation and partly obstructed the adenosine binding site after exchange of Val123, a variable hinge region residue, to alanine (Gafiel et al. 2003). As will be discussed in more detail later, a corre sponding conformational change of such a Gln residue has been observed in the structure of PDK1 with UCN01 (Komander et al. 2003). PDK1 has alanine in the Val123 (PKA) position too. Exchange of Gln181 in PKA to the much more conserved lysine in the position of Gln181 resulted in a fixed surface orientation (Gafiel et al. 2003). This clearly demonstrates secondary functional consequences of single residue exchanges that require compensation of some sort. Thus, the high variability of residues which line the aden-osine subsite may represent functional differences or may simply reflect less stringent evolutionary constraints in this region; in either case, it offers interesting opportunities to selectively target protein kinases with small molecule inhibitors by making use of the small differences in the individual shape and electronic environment in each protein kinase ATP pocket.
The sequence similarity of kinases within the same branch of kinases, such as the AGC kinases, leads to cross selectivity of protein kinase inhibitors. Obviously, closely related kinases appear to have a higher conservation of the kinase fold. A good example is the structures of active PKB (Yang et al. 2002), showing an ATP binding site which is almost indistinguishable from that of PKA. Mutagenesis of PKA and exchange of residues in the ATP pocket to PKB-specific residues emphasizes these similarities even further (Gafiel et al. 2003). Correspondingly, commercially available protein kinase inhibitors have very similar activities against both kinases (Davies et al. 2000). This is true for several protein kinase inhibitors that target AGC kinases. Though certainly a predicament for the design and development of selective protein kinase inhibitors, it can be turned into an advantage to achieve structural information about inhibitor/target interactions even if the primary target is difficult to deal with. PKA has been used as an ersatz kinase to study the binding mode of various AGC kinase inhibitors, mostly in cases were the actual target was structurally not available. For a long time PKA was the only AGC kinase with a known crystal structure. In the last 2 years, the crystal structures of several other members of the AGC branch have been solved, such as PKB/Akt (Yang et al. 2002), PDK1 (Biondi et al. 2002), Aurora (Bayliss et al. 2003) and Grk2 (Lodowski et al. 2003).
Was this article helpful?