From structure to the dynamic regulation of a molecular switch: A journey over 3 decades

Abstract

It is difficult to imagine where the signaling community would be today without the Protein Data Bank. This visionary resource, established in the 1970s, has been an essential partner for sharing information between academics and industry for over 3 decades. We describe here the history of our journey with the protein kinases using cAMP-dependent protein kinase as a prototype. We summarize what we have learned since the first structure, published in 1991, why our journey is still ongoing, and why it has been essential to share our structural information. For regulation of kinase activity, we focus on the cAMP-binding protein kinase regulatory subunits. By exploring full-length macromolecular complexes, we discovered not only allostery but also an essential motif originally attributed to crystal packing. Massive genomic data on disease mutations allows us to now revisit crystal packing as a treasure chest of possible protein:protein interfaces where the biological significance and disease relevance can be validated. It provides a new window into exploring dynamic intrinsically disordered regions that previously were deleted, ignored, or attributed to crystal packing. Merging of crystallography with cryo-electron microscopy, cryo-electron tomography, NMR, and millisecond molecular dynamics simulations is opening a new world for the signaling community where those structure coordinates, deposited in the Protein Data Bank, are just a starting point!

Keywords: allostery; cAMP; cAMP-dependent protein kinase (PKA); catalytic subunit; crystallography; dynamics; intrinsically disordered regions; protein kinases; protein structure; regulatory subunit.

Figure 1

Evolution of the kinase domain (1991–2021). The first structure of the PKA C-subunit in 1991 defined the fold of a fully active kinase and the docking of a high-affinity pseudosubstrate peptide derived from the heat-stable protein kinase inhibitor (2, 3). The next structure in 1993 defined the intricate way in which ATP was docked into the active site cleft (4). The full appreciation of the dynamics of the C-subunit and its evolution as a dynamic molecular switch unfolded over the next 3 decades. Discovery of R-Spine, 2006 (9); Discovery of AGC-tail as a conserved feature of the AGC family, 2007 (22); Discovery of the C-spine, 2008 (10); Identification of conserved surface pockets in the kinome, 2009 (51); Elucidation of committed steps in catalysis by NMR, 2011 (11); TiBS review of spines, 2013 (52); Deciphering protein kinase regulation, 2013 (53); Community maps, 2014 (14); Violin model of allostery, 2015 (54); Entropy-driven Allostery revealed by NMR, 2017 (12); Zooming in on protons with Neutron Diffraction, 2019 (55); Dynamics of PKA:peptide complexes, 2019 (56); IUBMB review, 2019 (57).

Figure 2

ATP binding as a driver of kinase dynamics. The PKA C-subunit not only defined a novel fold but also defined a novel ATP-binding site that was distinct from the Rossmann fold. The importance of the hydrophobic shell that embraces the nucleotide unfolded over many decades. A, the AT- binding site defined in 1993 (4) showed the intricate network of hydrogen bonding and electrostatic interactions that bury the entire nucleotide in a closed conformation. B, the second metal ion is essential for the synergistic high-affinity binding of ATP in the protein kinase inhibitor complex and for phosphotransfer (7, 8, 13). C, expanded hydrophobic shell that surrounds the ATP includes the C-spine (tan) (10), the R-spine (red) (9), and the shell (blue) (53) as well as additional residues from the N-lobe and C-tail (58). This shell defines the entropy-driven allostery that drives the phosphor transfer mechanism (12, 52, 54). D, conserved motifs of the kinase core. Left, glycine-rich loop (G-Loop) is a conserved feature of the protein kinases. Middle, DFG motif. D binds to the second metal ion in the ATP complex, and F is part of the R-Spine (RS3). G provides flexibility and is part of β strand 9 that is then followed by the activation loop. Right, HRD Motif. This motif contains an R-spine residue (RS1) and defines a key allosteric node. In some kinases such as PKA the H is conserved as a Tyr. The H/Y residue reaches across and stabilizes the catalytic loop, the R reaches out to the activation loop phosphorylation site (pT197 in PKA), and the D is the catalytic base that is positioned at the site of phosphor transfer.

Figure 3

Isoform diversity of the catalytic subunit. The functional and spatial nonredundancy of the Cα and Cβ isoforms expands the size of the PKA subfamily. Our growing recognition that each splice variant is localized differently and committed to a specific function has the potential to greatly expand the size of the kinome. A, the N-terminal tail of the C-subunit is encoded by Exons 1, 2, and 3. B, splice variants of the PKA Cα and Cβ subunits (20). Several oncogenic fusion proteins have been identified where exon 1 of Cα and/or Cβ is replaced by another domain such as the J Domain of DNAJB1 (59) or the N terminus of the ATPase 1 (60). C, N-terminal and C-terminal tails wrapped around both lobes of the conserved kinase core. D, the C-terminal tail also contains many of the residues that differs between Cα1 and Cβ1 (indicated as blue dots). Like the N-terminal tail, the C-tail reaches around both lobes of the kinase core. In the ATP-bound closed conformation the C-tail is firmly anchored to the core through hydrophobic interactions of F³²⁷ and Y³³⁰ with the adenine ring of ATP and through the hydrophobic motif (F³⁴⁷ and F³⁵⁰) that anchors the C terminus to the αC-Helix. E, in the open apo conformation a portion of the C-Tail becomes disordered and the G-Loop assumes an open conformation.

Figure 4

Evolution of the cyclic nucleotide binding (CNB) domain: Although the cAMP-binding sites were captured in the crystal structure of a deletion mutant of the RIα subunit (25), the full domain architecture and in particular the N3A motif was not recognized or appreciated as a functional part of the eukaryotic CNB domain until the LSP analysis was done (9).A, the dynamic features of the CNB domain and in particular the correlated motions of the helical domains are highlighted and further defined in the movie (see Supporting information). The N3A motif that includes the αA and αN helices are in dark red, the B/C helix is teal, and the phosphate-binding motif that is the signature motif of this domain and binds to the phosphate of cAMP is in red. B, the key features of the N3A motif are summarized. Two CNC mutations R144S and S145G/D are also indicated.

Figure 5

Crystal packing in the RIα (91–379) structure. When the first structure of a monomeric RIα was solved the importance of the N3A motif as a defining feature of eukaryotic CNB domains was not recognized. In this structure of RIα (91–379) the N-terminal linker region (residues 92–112) that includes the inhibitor site is disordered and the packing of the N3A motifs between two protomers (N3A:N3A’) was assumed to be an artifact of crystal packing (25). A, crystal packing of the two N3A motifs. B, dimer interface between the two N3A motifs would leave the two residues that were shown to be drivers of CNC solvent exposed in the monomer. C, the interactions between the CNB domains are localized to CNB-A. The black arrows indicate the portion of the linker that precedes the N3A motif that is ordered by crystal packing.

Figure 6

Structure of the full-length RIα dimer. The structure of the full-length RIα dimer highlighted the physiological importance of the N3A dimer interface and provided an explanation for the CNC mutations (26). A, the packing of the two N3A motifs in each CNB-A domain is a defining feature of the full-length dimer. B, the dimer interface is extended in this structure to include residues 108 to 118 and highlights the importance of the segment that precedes the αN helix. C, organization of the full-length RIα dimer. Although the DD domain and the linker are present in this protein, these regions are disordered so the cross talk between these residues and the CNB-A domain is not captured. However, the key interaction between the two CNB-A domains explains the compact nature of the RIα dimer, in contrast to the extended conformation of RIIα and RIIβ. It also begins to explain the significance of the CNC mutations. The black arrows indicate the extended portion of the linker that precedes the N3A motif that is ordered and part of the extended dimer interface. The red arrows indicate intrinsically disordered linker region that includes the inhibitor site (yellow oval).

Figure 7

Propensity for further polymerization is captured in the A211D CNC mutant.A, the packing of two N3A motifs in A211D mutant, which is the same feature as the full-length WT dimer. B, the potential of the extended N3A motif is captured in this A211D mutant. In this mutant the N-terminal extension that is fused to the N helix in the CNB-A domain mediates one dimer interface as shown in Figure 7; however, the segment that extends from the A helix and includes β strands 1 and 2 of the β subdomain mediates a different dimer interface with another protomer. C, the asymmetric unit in this structure includes four dimers and shows how a polymer can form that has features that resemble the Aβ amyloid fibers (42).

Figure 8

Role of the N3A motifs in holoenzymes. Although the N3A motifs can contribute to the organization of RIα holoenzymes, they are never seen as a dimer interface in either the free RIIα/RIIβ subunits (61, 62) or in the RIIβ holoenzymes (47, 48) where they appear to be poised to potentially interact with other proteins. A, an RIα holoenzyme structure shows how a highly complex and dynamic interface can be mediated by the N3A motif (44), and this organization is conserved when the RIα holoenzyme is formed with the oncogenic C-subunit that is fused to the J-domain of DNA-JB1 (45). B, this organization is also seen with an ACRDYS mutant of the RIα subunit where the C-terminal 14 residues are deleted rendering the holoenzyme resistant to activation by cAMP (J. Bruystens, J. Wu, J. Del Rio and S.S. Taylor, unpublished results). C, in the RIIβ holoenzyme the N3A motif in CNB-A interacts with its own CNB-B domain but otherwise is exposed to solvent (47). In D through E we see how the linker region from the inhibitor site to the N3A motif becomes ordered in each of these holoenzymes. D, in RIα ordering of the linker requires ATP (44). E, ACRDYS mutations that delete the C terminus also influence the requirement for ATP (J. Bruystens, J. Wu, J. Del Rio and S.S. Taylor, unpublished results). F, in the RIIβ holoenzyme ATP is not required for ordering of this N-linker region (44). The inhibitor site (P-site residue) is indicated by the blue arrow.

Figure 9

The kinase domain is packaged in an inactive R₂C₂holoenzyme. The activity of the catalytic subunit is trapped in an inactive state by cAMP-binding regulatory subunits. The full-length RIIβ holoenzyme captures the complex isoform-specific allosteric cross talk between the domains. Top, the cross talk between the R:C protomers is captured in a crystal lattice (47). The β4-β5 loop of one CNB-A domain (black) is packed against the C-terminal tail of the opposite C-subunit (tan) thereby assembling a complete ATP-binding pocket in the absence of nucleotide. The R₂C₂ complex shows an axis of symmetry (yellow dot) that cannot be captured in a monomeric R:C complex. Bottom, the complex assembly of the full-length protein, captured with cryo-EM (48), shows how the linker region that joins the N-terminal dimerization/docking domain to the cyclic nucleotide binding (CNB) domains is woven between the two protomers placing the A kinase anchoring proteins (AKAP) docking surface on the same side as the myristyl groups that are attached to the N terminus of the C-subunit. This surface is positioned to interact with membranes by utilizing a polyvalent mechanism that includes the myristyl groups at the N terminus of the C-subunits (49).