Activation of Rho GTPases by DOCK Exchange Factors Is Mediated by a Nucleotide Sensor

See allHide authors and affiliations

Science  11 Sep 2009:
Vol. 325, Issue 5946, pp. 1398-1402
DOI: 10.1126/science.1174468


Activation of Rho guanosine triphosphatases (GTPases) to the guanine triphosphate (GTP)–bound state is a critical event in their regulation of the cytoskeleton and cell signaling. Members of the DOCK family of guanine nucleotide exchange factors (GEFs) are important activators of Rho GTPases, but the mechanism of activation by their catalytic DHR2 domain is unknown. Through structural analysis of DOCK9-Cdc42 complexes, we identify a nucleotide sensor within the α10 helix of the DHR2 domain that contributes to release of guanine diphosphate (GDP) and then to discharge of the activated GTP-bound Cdc42. Magnesium exclusion, a critical factor in promoting GDP release, is mediated by a conserved valine residue within this sensor, whereas binding of GTP-Mg2+ to the nucleotide-free complex results in magnesium-inducing displacement of the sensor to stimulate discharge of Cdc42-GTP. These studies identify an unusual mechanism of GDP release and define the complete GEF catalytic cycle from GDP dissociation followed by GTP binding and discharge of the activated GTPase.

Rho guanosine triphosphatases (GTPases) are conserved regulators of cell motility, polarity, adhesion, cytoskeletal organization, proliferation, and apoptosis (1, 2). Two distinct families of Rho guanine nucleotide exchange factors (GEFs), the Dbl homology (DH) and DOCK proteins (35), activate Rho GTPases through the exchange of guanosine diphosphate (GDP) for guanosine triphosphate (GTP). In humans, there are ~70 DH-containing Rho GEFs and 11 DOCK family members. DOCK proteins have been implicated in the activation of Rac and Cdc42 in cell migration, morphogenesis, and phagocytosis (6, 7) and as important components of tumor cell movement and invasion (8, 9). In humans, DOCK proteins are organized into four subfamilies, characterized by their differing specificities for Rac and Cdc42, regulatory domains, and associated subunits (6, 7). DOCK A and B subfamilies activate Rac, the DOCK D subfamily is specific for Cdc42, whereas the DOCK C subfamily has dual specificity for Rac and Cdc42 (1013). All contain a GEF catalytic domain named the DHR2, CZH2, or DOCKER domain of ~400 residues situated within the C terminus and a second conserved region, the ~250 residue DHR1 or CZH1 domain. The DHR1 domain of DOCK180 (DOCK1) interacts with phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3] to control membrane localization (14, 15). Whereas many of the upstream signals that activate DOCK proteins remain to be elucidated, it is clear that integrins are important stimulators of DOCK A- and B-induced Rac activation through the adaptors ELMO, Crk, and p130Cas (8, 9, 1618).

Although the structural basis of nucleotide release catalyzed by the DH family of Rho GEFs is well established, the lack of sequence homology between DHR2 and other GEF domains means that the molecular mechanism of GEF activity of DOCK proteins is unknown. Further, the structural events that permit discharge of the activated GTPase when GTP binds to the nucleotide-free GTPase in a GEF-GTPase complex have not been defined. Here, through structural analyses of the DOCK9 DHR2 domain (DOCK9DHR2) in complex with nucleotide-free Cdc42, and in complexes with GDP- and GTP-Cdc42 (table S1) (19), we reveal a complete GEF catalytic cycle.

The catalytic domains of GEFs can form stable complexes with nucleotide-free GTPases. We solved the structure of DOCK9DHR2 complexed with nucleotide-free Cdc42 (Fig. 1). The structure of the DHR2 domain differs from that of other GEF catalytic domains. It is organized into three lobes of roughly equal size (lobes A, B, and C), with the Cdc42 binding site and catalytic center generated entirely from lobes B and C (Fig. 1A, fig. S1). Lobe A is formed from an antiparallel array of five α helices. Through extensive contacts with lobe B, lobe A stabilizes the DHR2 domain. Its α4 and α5 helices generate the conserved dimer interface (fig. S2) (20). Although the dimer interface is not responsible for direct interactions with Cdc42, mutations of α5 that disrupt dimerization partially reduce DOCK9DHR2 GEF activity, most likely due to destabilization of lobe A (fig. S3). Lobe B adopts an unusual architecture of two antiparallel β sheets disposed in a loosely packed orthogonal arrangement, whereas lobe C comprises a four-helix bundle. Helix α10 of lobe C, the most conserved region of DHR2 domains (fig. S1), is interrupted by a seven-residue loop, the α10 insert, that divides the helix into α10N and α10C (Fig. 1).

Fig. 1

Views of nucleotide-free DOCK9DHR2-Cdc42. (A) Overall structure. Lobes A, B, and C colored in blue, cyan, and green, respectively; α10 insert with Val1951 (nucleotide sensor) in yellow. Cdc42 is colored orange with switch 1 and 2 and P loop in red, purple, and blue, respectively. (B) Details of the DOCK9DHR2-Cdc42 switch 1 interactions show that switch 1 mediates extensive contacts with DOCK9DHR2. Figure produced by using PyMOL (

Switch 1 and switch 2 are conserved nucleotide contact regions of GTPases that, on exchange of GDP for GTP, alter their conformation and allow interaction with effector proteins. Interactions of Cdc42 with DOCK9DHR2 are dominated by switch 1 and augmented through contacts of switch 2, the β2/β3 hairpin, and β6/α5 loop with DOCK9DHR2 (Fig. 1 and fig. S4). When in a complex with DOCK9DHR2, switch 1 undergoes a conformational transition that exposes Cdc42’s nucleotide-binding site. Switch 2, however, remains unchanged (2123) (Fig. 2A,B). Extensive interactions between the two proteins stabilize the displaced conformation of switch 1. Through a concerted mechanism, the α10 insert clamps open switch 1 and projects into the GTPase nucleotide-binding site (Fig. 1). Of particular relevance for the exchange reaction, the side chain of a key nucleotide-coordinating residue, Phe28C (Cdc42 residues are denoted with superscript “C”), is displaced into a hydrophobic pocket within lobe B (Fig. 1B).

Fig. 2

Mechanism of DOCK9DHR2-mediated nucleotide release. (A) Comparison of Cdc42 from nucleotide-free DOCK9DHR2-Cdc42 (orange) with isolated Cdc42-GDP in light blue (22). GDP of isolated Cdc42 in light green. Switch 1 is red and cyan in complex and isolated states, respectively. M indicates the Mg2+ ion. (B) Details of conformational changes at the nucleotide-binding site. (C) DOCK9DHR2-Cdc42-GDP ternary structure at the nucleotide-binding site. GDP-Mg2+ and water molecules (W) of ternary complex are in dark green and red. GDP and waters of isolated Cdc42-GDP complex are in light green and pink (22). (D) DOCK9DHR2-Cdc42-GTP ternary structure. GTP-Mg2+ and water molecules (W) of ternary complex are in dark green and red, respectively. Nonhydrolyzable GTP-Mg2+ and waters of isolated Cdc42-GTP complex (23) are in light green and pink, respectively.

To understand how DOCK9DHR2 decreases the affinity of Cdc42 for nucleotide, we determined the structure of a DOCK9DHR2-Cdc42-GDP complex. Incubating preexisting DOCK9DHR2-Cdc42 crystals with GDP generated a ternary complex suitable for structural analysis. Electron density for GDP and residues of the α10 insert were clearly defined (fig. S5A). Binding of GDP to the DOCK9DHR2-Cdc42 complex causes essentially no change in protein conformation (fig. S6), and the nucleotide adopts an identical mode of binding and conformation to that of isolated Cdc42 (22) (Fig. 2C). Importantly, no electron density was visible for Mg2+ coordinated to the β phosphate of GDP. The expected Mg2+-binding position is directly occluded by the nonpolar side chain of Val1951 at the tip of the α10 insert. By comparing the DOCK9DHR2-Cdc42 structures with isolated Cdc42-nucleotide complexes, we defined the molecular basis for DOCK9DHR2-mediated decrease of nucleotide affinity (Fig. 2). First, the conformational transition of switch 1 on binding DOCK9DHR2 exposes the nucleotide-binding site. Specifically, displacement of the switch 1 residue Phe28C disrupts the conserved aromatic-guanine base interaction important in guanine nucleotide binding (24) (Fig. 2B). Second, movement of switch 1 is linked to rotation of the P-loop Cys18C thiol group that disrupts a hydrogen bond with the nucleotide α phosphate (Fig. 2B). Third, by intruding into the GTPase nucleotide-binding site, Val1951 directly occludes nucleotide-coordinated Mg2+ (Fig. 2C). Because magnesium enhances nucleotide affinity by neutralizing the negatively charged phosphate groups, its exclusion profoundly reduces nucleotide affinity (3). The key role of Val1951 is confirmed by the observation that substitution with an alanine residue severely abrogates catalytic activity (fig. S7). Furthermore, Val1951 of DOCK9 is strictly conserved in all DOCK homologs from plants to humans (12).

This mechanism of magnesium exclusion by a direct effect of a residue within the catalytic domain is strikingly different from that found in other GEFs where displacement of the conserved switch 2 Ala residue of the GTPase (Ala59 in Ras and Rho family GTPases) sterically interferes with Mg2+ binding (2528) (Fig. 3). In these GEFs, motion of switch 2 also creates a salt bridge between Glu62 of switch 2 and Lys16 of the P loop, conferring charge repulsion on the nucleotide’s phosphate groups and disrupting their favorable interaction with Lys16. In contrast, DOCK9DHR2 does not alter the positions of either switch 2 or Lys16 of Cdc42 (Fig. 2B,3). Therefore, unlike DH proteins (29), DOCK GEFs would not use either Ala59 or Glu62 of the Rho GTPase for nucleotide exchange. Consistent with this notion, substituting Gly for Ala59C or Ala for Glu62C of Cdc42 did not diminish the nucleotide exchange activity of DOCK9DHR2 toward the GTPase (fig. S8C). This finding suggests that DOCK proteins might activate a broader spectrum of Rho GTPases, for example those lacking Ala59 and Glu62 (fig. S4), than are activated by DH family GEFs.

Fig. 3

Schematic of GEF catalytic reaction mechanisms. (A) GDP bound to nucleotide-binding site of Rho GTPase. Color-coding is as for Fig. 1. (B) DOCK9DHR2-mediated release of GDP occurs via motion of Cys18C and Phe28C, disrupting contacts to GDP, and exclusion of Mg2+ mediate by Val1951. (C) Other GEFs [for example Sos, DH, SopE, and PRONE (2528)] also promote shifts of Cys18C and Phe28C, but GEF-induced motion of switch 2 causes Ala59C to sterically occlude Mg2+ and promotes a salt bridge between Lys16C and Glu62C. (D) Binding of GTP-Mg2+ to DOCK9DHR2-Cdc42 promotes conformational changes that trigger discharge of the activated GTPase.

Nucleotide-depleted Cdc42 interacts tightly with DOCK9DHR2 but is discharged by GTP, yielding the activated GTPase. To understand the molecular basis of this discharge process, we determined the structure of a GTP-bound ternary complex. As observed in the GDP complex, electron density was well resolved for the nucleotide. The coordinating Mg2+ and water molecules were also clearly defined (fig. S5B). The mode of binding of GTP and Mg2+ is identical to that of isolated Cdc42 (Fig. 2D) (23). Association of Mg2+ is incompatible with the closed conformation of the α10 insert, and accordingly the tip of the α10 insert was completely disordered, with no electron density visible for Val1951 and flanking residues (fig. S5B). Displacement of the α10 insert removes the clamp from switch 1 and induces a set of interdependent conformational changes within other regions of the DOCK9DHR2-Cdc42 interface (Fig. 4). These involve translation of the C terminus of α10 (α10C), motion of the neighboring β5/β6 loop of lobe B, and a shift (~1.5 Å) of the α6 helix and β3 strand of lobe B. This reorganization of the complex disrupts DOCK9DHR2 contacts to switch 1 and is indicative of disengagement of Cdc42 from DOCK9DHR2 and its transition to an enhanced affinity for nucleotide. For example, because of motion of α6, the binding pocket for Phe28C is enlarged, partially exposing Phe28C, whereas rotation of the Cys18C thiol group to hydrogen bond with the α phosphate of GTP increases GTP affinity (Fig. 3D, compare Fig. 4, B and C).

Fig. 4

GTP-Mg2+–induced conformational changes of DOCK9DHR2-Cdc42. (A) Superimposition of DOCK9DHR2-Cdc42 and DOCK9DHR2-Cdc42-GTP (superimposed on Cdc42), showing conformational changes of α6 and α10C helices (indicated by arrows) and β3/β4 and β5/β6 loops. Root mean square deviation between equivalent Cα atom is 0.8 Å. M, Mg2+ ion. (B) Surface representation of DOCK9DHR2 with Cdc42 as a ribbons representation. Color-coded as in Fig. 1A. (C) DOCK9DHR2-Cdc42-GTP ternary complex, showing how disordered α10 insert removes clamp-restraining switch 1.

The structure of DOCK9DHR2-Cdc42 reveals the mechanism of GDP dissociation and GTP-stimulated discharge of activated Cdc42 from DOCK9DHR2 (Fig. 3). Critical to these processes is the α10 insert, which functions to open switch 1, induce release of nucleotide, and block discharge of nucleotide-free Cdc42. Activation of Cdc42 is detected by the presence of Mg2+ tightly bound to GTP. This triggers displacement of the α10 insert and propagation of conformational changes to the DOCK9DHR2-Cdc42 interface (Fig. 4). Because GDP does not induce equivalent conformational changes, GTP may specifically promote discharge of the GTP-bound Cdc42 from DOCK9DHR2 to complete the catalytic exchange reaction. Thus, the α10 insert of DOCK proteins acts as sensor of the GDP- and GTP-bound states of Cdc42. This study provides a complete structural description of the steps involved in GTPase activation by a guanine nucleotide exchange factor and may have implications for other GEF mechanisms.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Table S1


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. This work was funded by grants from Cancer Research–UK to D.B. and C.J.M. C.J.M. is a Gibb Life Fellow of Cancer Resesarch–UK, and work was carried out with the support of the Diamond Light Source. We thank staff at the European Synchrotron Radiation Facility for help with data collection, T. Bunney for advice and reagents, and D. Komander for discussions. Accession numbers for coordinates and structure factors are 2wm9, 2wm9sf, 2wmn, 2wmnsf, 2wmo, and 2wmosf for the DOCK9DHR2-Cdc42 complex and -GDP and -GTP complexes, respectively.

Stay Connected to Science

Navigate This Article