Structural biology of DOCK-family guanine nucleotide exchange factors

Abstract

DOCK proteins are a family of multi-domain guanine nucleotide exchange factors (GEFs) that activate the RHO GTPases CDC42 and RAC1, thereby regulating several RHO GTPase-dependent cellular processes. DOCK proteins are characterized by the catalytic DHR2 domain (DOCK^DHR2 ), and a phosphatidylinositol(3,4,5)P₃ -binding DHR1 domain (DOCK^DHR1 ) that targets DOCK proteins to plasma membranes. DOCK-family GEFs are divided into four subfamilies (A to D) differing in their specificities for CDC42 and RAC1, and the composition of accessory signalling domains. Additionally, the DOCK-A and DOCK-B subfamilies are constitutively associated with ELMO proteins that auto-inhibit DOCK GEF activity. We review structural studies that have provided mechanistic insights into DOCK-protein functions. These studies revealed how a conserved nucleotide sensor in DOCK^DHR2 catalyses nucleotide exchange, the basis for how different DOCK proteins activate specifically CDC42 and RAC1, and sometimes both, and how up-stream regulators relieve the ELMO-mediated auto-inhibition. We conclude by presenting a model for full-length DOCK9 of the DOCK-D subfamily. The involvement of DOCK GEFs in a range of diseases highlights the importance of gaining structural insights into these proteins to better understand and specifically target them.

Keywords: CDC42; DOCK proteins; RAC1; guanine nucleotide exchange factors.

Fig. 1

DOCK proteins belong to four subfamilies. (A) Schematic of DOCK subfamily architecture. Subfamilies: DOCK‐A (DOCK1,2,5), DOCK‐B (DOCK3,4), DOCK‐C (DOCK6‐8) and DOCK‐D (DOCK9‐11). (B) Surface representation of the DOCK1^DHR1 domain. DOCK1^DHR1 adopts a C2‐like domain architecture that binds PtdIns(3,4,5)P₃ by means of a positively charged pocket defined by surface loops L1 and L2 [26] (PDB: 3L4C). Figures were generated using chimerax [65].

Fig. 2

Structure of DOCK9^DHR2:CDC42. DOCK proteins catalyse nucleotide exchange through a nucleotide sensor. (A) The catalytic DHR2 domain is composed of three lobes, with lobes B and C contacting the RHO GTPase CDC42 directly. (B, C) Details of the nucleotide sensor in the α10 helix of lobe C clamping open switch 1 of CDC42 and protruding into the GDP‐Mg²⁺‐binding site. This directly occludes Mg²⁺ binding. GDP‐Mg²⁺ is modelled into the nucleotide‐free CDC42 based on structures of DOCK9^DHR2:CDC42:GDP and DOCK9^DHR2:CDC42:GTP‐Mg²⁺. Figure based on [33] (PDB: 2WM9, 2WMN, 2WMO).

Fig. 3

Schematic of the DOCK GEF catalytic mechanism. (A) GDP bound to the nucleotide‐binding site of Rho GTPase. (B) DOCK^DHR2‐mediated release of GDP occurs via the motion of Cys18 and Phe28, disrupting contacts to GDP and exclusion of Mg²⁺ mediated by the catalytic valine (Val1951 of human DOCK9). (C) The binding of GTP‐Mg²⁺ to DOCK^DHR2‐GTPase promotes conformational changes that trigger the discharge of the activated GTPase. Adapted from [33]. Numbered residues refer to CDC42.

Fig. 4

DHR2 domains of DOCK proteins form conserved homodimeric assemblies. Gallery of DOCK^DHR2:GTPase complexes. (A) DOCK9^DHR2:CDC42 ([33] (PDB: 2WM9)), (B) DOCK2^DHR2:RAC1 ([37] (PDB: 2YIN)), (C) DOCK10^DHR2:CDC42 ([43] (PDB: 6TKY)) and (D) DOCK10^DHR2:RAC3 ([43] (PDB: 6TM1)). Numbers in parentheses refer to the two protomers of the DOCK dimers.

Fig. 5

The catalytic valine of the nucleotide sensor is invariant in the DOCK GEF family. Multiple‐sequence alignment of human DOCK proteins, Caenorhabditis elegans CED‐5, Drosophila melanogaster DOCK1 and Arabidopsis thaliana SPIKE1. Figure generated using jalview [66]. The catalytic valine (Val1951 of Homo sapiens DOCK9) of the nucleotide sensor is invariant.

Fig. 6

DOCK specificities for CDC42 and RAC through position 56 of CDC42 and RAC. (A) Superimposition of DOCK9^DHR2:CDC42 ([33] (PDB: 2WM9)), DOCK10^DHR2:CDC42 ([43] (PDB: 6TKY)) and DOCK7^DHR2:CDC42 ([34] (PDB: 6AJ4). (B) The larger Trp56 in RAC is accommodated by the smaller and rotated Asn1532 side chain in DOCK2 and the rotation of Gln2062 in DOCK10. Superimposition of DOCK9^DHR2:CDC42 ([33] (PDB: 2WM9)), DOCK2^DHR2:RAC1 ([37] (PDB: 2YIN)) and DOCK10^DHR2:RAC3 ([43] (PDB: 6TM1)). (C) Superimposition of apo‐DOCK10^DHR2 onto lobes A and B of DOCK10^DHR2 of the DOCK10^DHR2:RAC3 complex shows the relative shift of lobe C to a more open state in apo‐DOCK.

Fig. 7

Cryo‐EM structure of DOCK2:ELMO1:RAC1. (A) Orthogonal views of the complex. (B) As in (A), but DOCK2:RAC1 and ELMO1 are separated for clarity and only one DOCK2:ELMO1:RAC protomer is shown. Figure based on [29] (PDB: 6TGC).

Fig. 8

The binary DOCK2:ELMO1 complex adopts an auto‐inhibited conformation. (A) Cryo‐EM structure of binary DOCK2:ELMO1 with ELMO1 adopting a closed auto‐inhibited state. (B, C) Comparison between the open and closed conformation of ELMO1 explains steric inhibition of RAC1 binding to DOCK2^DHR2. (B) DOCK2:ELMO1:RAC1 with DOCK2 shown as a molecular surface, ELMO1 in the open, active conformation and RAC1 as a cartoon. (C) DOCK2:ELMO1:RAC1 with DOCK2 shown as a molecular surface, ELMO1 shown in the open (up) active conformation (ternary complex) and in the closed (down) auto‐inhibited state (binary complex), RAC1 and DOCK2 Protomer 2 as a cartoon. Rotation of ELMO1^NTD about the hinge helix allows interconversion between open active and closed auto‐inhibited. Figure based on [29] (PDB: 6TGB, 6TGC).

Fig. 9

Model of the dimeric DOCK9:CDC42 complex. The model was based on the alphafold2 prediction of human DOCK9 [49] and the crystal structure of DOCK9^DHR2:CDC42 [33] (PDB: 2WM9).