Protein phosphorylation is critical for regulation in eukaryotic cells. The human genome encodes more than 500 protein kinases, making this one of the largest gene families. Although very diverse in how they receive and transmit signals, all protein kinases share a conserved catalytic core. While it is essential to understand how these enzymes function as catalysts, it is equally important to understand how they are regulated, how they function as scaffolds, and how they are localized.
Studying the structure and function of cAMP-dependent protein kinase (PKA), one of the simplest members of the protein kinase family, is our primary focus. Our initial chemical studies led eventually to solving the crystal structure of PKA in 1991; this structure continues to serve as a prototype for the entire family. In parallel, we solved structures of the regulatory subunits, the major receptors for cAMP in mammalian cells. Recent holoenzyme structures revealed for the first time how the catalytic subunit is regulated. Structures of A kinase–anchoring protein (AKAP) motifs bound to the regulatory subunits explain the molecular basis for targeting of PKA. An underlying theme of our laboratory is the iterative use of biophysical, biochemical, and biological approaches to probe PKA structure and function at both the atomic and cellular levels.
Multiple structures of the catalytic (C) subunit not only showed how PKA recognizes substrates and traverses the catalytic cycle but also defined its conformational flexibility. We have defined "open" and "closed" conformations. At the same time, we solved the structure of an ADP:AlF3 complex with a substrate peptide that resembles a transition state. Opening the active-site cleft, not phosphoryl transfer, is the rate-limiting step of catalysis. The crystal structures provide a static view of various conformational states, but understanding how the domains move in solution and how those motions correlate with functional steps of catalysis requires solution methods. In collaboration with David Johnson (University of California, Riverside), we used time-resolved fluorescence anisotropy to measure local motion and discovered an isoform-specific myristylation switch. We could also follow catalysis and holoenzyme activation. In collaboration with Gianluigi Veglia (University of Minnesota), we are using NMR (nuclear magnetic resonance) to map the dynamic properties of PKA comprehensively.
To compare active and inactive kinases, we developed a rapid computational method to show the spatial relatedness of specific residues. In this way we first defined a hydrophobic "regulatory" spine that is disassembled in inactive kinases. Activation, typically by phosphorylation of the activation loop, causes the spine to form. This spatially conserved spine motif consists of noncontiguous residues that come from both the small and large lobes. Further analysis revealed a second "catalytic" spine that is completed by the adenine ring of ATP. The hydrophobic F-helix serves as an anchor for both spines and is the organizing scaffold for the entire molecule.
By searching for residues that are uniquely conserved in AGC kinases, we discovered that the C-terminal tail (C-tail) is a conserved regulatory element of all AGC kinases. The motifs that are conserved within the tail interact uniquely with functional motifs in the kinase core, thus defining the C-tail as a cis-regulatory element. Without the C-tail, AGC kinases are not active. The C-tail also interacts with other proteins, such as the activating kinase, PDK1, and Hsp90.
Solving the structure of a truncated regulatory subunit, RIα, provided a first glimpse of the tandem cAMP-binding domains. The subsequent solution of an RIIβ structure, and most recently RIα, RIIα, and RIIβ holoenzyme complexes, coupled with site-specific and deletion mutagenesis, has allowed us to dissect the dynamic subdomains of the cyclic nucleotide–binding (CNB) domain. This ancient signaling domain, conserved from bacteria to humans, allosterically couples cAMP binding to a DNA-binding motif (CAP), to kinase activation (PKA and PKG), to ion channels (KCN), and to a guanine nucleotide exchange factor (EPAC).
The N-terminal dimerization/docking (D/D) domain serves as a docking site that binds to AKAPs, thereby localizing PKA to specific sites in the cell. In collaboration with Patricia Jennings (University of California, San Diego), we used NMR spectroscopy to solve the structure of the RIα D/D domain, which revealed a novel helix bundle. The crystal structure of RIIα bound to an AKAP peptide was recently solved, as well as crystal structures of the RIα D/D domain free and bound to the AKAP peptide. The dynamic features of the RIα and RIIβ subunits have been mapped by hydrogen/deuterium (H/D) exchange coupled with mass spectrometry, and recently, in collaboration with Giuseppe Melacini (McMaster University, Canada), we described the NMR structure of an RIα deletion mutant.
HoloenzymesWhile structures of R and C subunits revealed the ligand-binding properties of R and the catalytic features of C, holoenzyme complexes were required to understand how C is inhibited by R and how cAMP activates the holoenzyme. Solving the structure of an RIα deletion mutant revealed the molecular features of the complex for the first time. Most striking were the major conformational changes in the CNB domain as it releases cAMP and binds to C. Subsequent holoenzyme structures of RIIα, RIα, and RIIβ show how the two CNB domains become uncoupled and wrap around the large domain of C. The holoenzyme structures also allow us for the first time to gain a mechanistic appreciation of how PKA is activated by cAMP. Finally, they provide templates for identifying novel therapeutic strategies for designing isoform-specific activation/inhibition.
To understand PKA signaling, allostery, and isoform diversity will require structures of the full length tetrameric holoenzymes. SAXS (small-angle x-ray scattering), carried out in collaboration with Donald Blumenthal (University of Utah), has already shown that the isoforms differ substantially. RIα holoenzyme, like free RIα, is Y-shaped; the free RII subunits are extended and rod-shaped. Upon formation of holoenzyme, the RIIα complex remains extended, while the RIIβ holoenzyme assumes a compact globular shape.
Macromolecular Assemblies for PKA Signaling
Using a yeast two-hybrid screen, we identified two novel dual-specific A-kinase–anchoring proteins (D-AKAPs) that bind both RI and RII. D-AKAP1 targets PKA to either mitochondria or endoplasmic reticulum via N-terminal motifs. D-AKAP2 has two RGS domains and is linked via its C-terminal PDZ motif to cotransporters and to Rabs through its RGS domains. We now are exploring the biochemical and physiological roles of these AKAPs at several levels, including x-ray crystallography and H/D exchange.
The importance of PKA signaling at the mitochondria is a special focus as we discover novel mitochondrial PKA substrates such as ChChd3. ChChd3 is localized to the inner membrane space, where it is anchored through its myristylation motif to the outer membrane and through its ChChd (coiled-coil-helix coiled-coil-helix domain) domains to the inner membrane. Knockouts of ChChd3 lead to the formation of mitochondria that are devoid of cristae and to the prediction that ChChd3 may play a role in cristae biogenesis.
In addition to AKAPs, we discovered another mechanism for targeting of PKA. A novel A kinase–interacting protein (AKIP1) binds to the N terminus of the C subunit. By helping to shuttle C into the nucleus, AKIP1 provides a novel targeting role for the C subunit. AKIP is a scaffold for other proteins as well, such as apoptosis-inducing factor (AIF) and Hsp70. It is also reported to be a target for neddylation.