Structure of cyclin G-associated kinase (GAK) trapped in different conformations using nanobodies

Abstract

GAK (cyclin G-associated kinase) is a key regulator of clathrin-coated vesicle trafficking and plays a central role during development. Additionally, due to the unusually high plasticity of its catalytic domain, it is a frequent 'off-target' of clinical kinase inhibitors associated with respiratory side effects of these drugs. In the present paper, we determined the crystal structure of the GAK catalytic domain alone and in complex with specific single-chain antibodies (nanobodies). GAK is constitutively active and weakly associates in solution. The GAK apo structure revealed a dimeric inactive state of the catalytic domain mediated by an unusual activation segment interaction. Co-crystallization with the nanobody NbGAK_4 trapped GAK in a dimeric arrangement similar to the one observed in the apo structure, whereas NbGAK_1 captured the activation segment of monomeric GAK in a well-ordered conformation, representing features of the active kinase. The presented structural and biochemical data provide insight into the domain plasticity of GAK and demonstrate the utility of nanobodies to gain insight into conformational changes of dynamic molecules. In addition, we present structural data on the binding mode of ATP mimetic inhibitors and enzyme kinetic data, which will support rational inhibitor design of inhibitors to reduce the off-target effect on GAK.

Figure 1. Overview of GAK and its Nb complexes

(A) Dimeric assembly of GAK in the apo structure, which is enabled by an exchange in the extended activation segments. (B) Sedimentation velocity AUC of GAK demonstrates an existence of mainly monomeric kinase in solution with a possibility of a weak dimeric formation. Also see Supplementary Figure S1(B) (http://www.biochemj.org/bj/459/bj4590059add.htm). (C) Ribbon representations of the structures of GAK–NbGAK_1 (left-hand panel) and GAK–NbGAK_4 (right-hand panel), in which the Nbs are coloured yellow and cyan respectively. Secondary structure elements of the kinase are labelled and kinase inhibitors are shown as sphere. (D) Superimposition of GAK and MPSK1 revealing a highly conserved overall topology.

Figure 2. Analysis of specific NbGAK_1 and NbGAK_4 binding to GAK

(A) Overall structures of the two Nbs demonstrate a classical β-sheet immunoglobulin scaffold. Their CDR regions are highlighted and disulfide bridges are indicated. For detailed structural and sequence analysis of the Nb interaction see Supplementary Figure S2 (http://www.biochemj.org/bj/459/bj4590059add.htm). (B) Interaction analysis of GAK with NbGAK_1 and NbGAK_4. Grey curves represent the measured responses, whereas red lines reflect the applied global fit using the 1:1 Langmuir interaction model. (C) Different epitopes of GAK are recognized by two Nbs. NbGAK_1 (yellow) binds to the N-terminal lobe, whereas NbGAK_4 (pale blue) recognizes the helical architectures of the C-terminal lobe of the kinase. For details see Supplementary Figures S3(A) and S3(B) (http://www.biochemj.org/bj/459/bj4590059add.htm).

Figure 3. Nb_GAK1 and Nb_GAK4 capture different activation states of GAK

(A) Arrangement of GAK oligomeric states in the crystals of dimeric apo-GAK (left-hand panel), dimeric GAK–NbGAK_4 (middle panel) and monomeric GAK–NbGAK_1 (right-hand), in which each kinase molecule is coloured wheat and pale blue, and their activation segments are in magenta and cyan. In the apo and NbGAK_4-complexed structures, formation of a homodimer in a head-to-tail fashion is assisted by an activation segment-exchange mechanism, whereas this is not observed in a fully ordered activation segment that prompts a monomeric form in the NbGAK_1-complexed crystals. The disulfide bridge formed between two cysteine residues from the αC observed in the GAK–NbGAK_1 complex is coloured red. (B) Details of intersubunit contacts within the activation segment-exchange region of the GAK–NbGAK_4 structure. (C) Superimposition of GAK from the NbGAK_1-complexed structure and MPSK1 reveal similar topology of the fully ordered activation segment with two highly conserved unique features, the ACSH and the parallel loop within the activation loop. This conformation of GAK also demonstrates a formation of the hydrophobic regulatory spine (R-spine), a characteristic of an active kinase. (D) Conformation of the activation segment confers an active state of GAK. Interactions between the HRD motif residues to both ASCH Glu²¹⁷ and P+1 loop Thr²²³ residues, including communication between the αEF Glu²³⁰ to the C-lobe Arg³⁰³, lock the typically flexible activation segment in place, and this creates a platform for substrate binding. Modelling of the substrate peptide (grey ribbon) and ATP-MgCl₂ (pink stick and ball) from the PKA structure (PDB code 1ATP) into GAK suggests a suitable pocket for accommodation of a substrate peptide for phosphorylation.

Figure 4. Kinase activities of GAK and its binding with inhibitors

(A) Kinase activities of GAK in the absence or presence of the Nbs. Results are means±S.E.M. (B) ΔT_m values for several commercially available kinase inhibitors. (C) SPR binding assays of some inhibitors to GAK. Detailed interactions between Wee1/Chk1 (D) and indirubin E804 (E and F) and the kinase with the insets showing the |F_o|−|F_c| omitted map contoured at 3σ. Note that two modes of binding of indirubin E804 are observed in two different complexes.