Dimerization of DOCK2 Is Essential for DOCK2-Mediated Rac Activation and Lymphocyte Migration

The migratory properties of lymphocytes depend on DOCK2, an atypical Rac activator predominantly expressed in hematopoietic cells. Although DOCK2 does not contain the Dbl homology domain typically found in guanine nucleotide exchange factors (GEFs), DOCK2 mediates the GTP-GDP exchange reaction for Rac via its DOCK homology region (DHR)-2 (also known as CZH2 or Docker) domain. DOCK2 DHR-2 domain is composed of three lobes, and Rac binding site and catalytic center are generated entirely from lobes B and C. On the other hand, lobe A has been implicated in dimer formation, yet its physiological significance remains unknown. Here, we report that lobe A-mediated DOCK2 dimerization is crucial for Rac activation and lymphocyte migration. We found that unlike wild-type DOCK2, DOCK2 mutant lacking lobe A failed to restore motility and polarity when expressed in thymoma cells and primary T cells lacking endogenous expression of DOCK2. Similar results were obtained with the DOCK2 point mutant having a defect in dimerization. Deletion of lobe A from the DHR-2 domain did not affect Rac GEF activity in vitro. However, fluorescence resonance energy transfer analyses revealed that lobe A is required for DOCK2 to activate Rac effectively during cell migration. Our results thus indicate that DOCK2 dimerization is functionally important under the physiological condition where only limited amounts of DOCK2 and Rac are localized to the plasma membrane.


Introduction
Cell migration is a fundamental biological response involving membrane polarization and cytoskeletal dynamics [1]. In response to chemoattractants, filamentous actin (F-actin) polymerizes asymmetrically at the leading edge of the cell, providing the force necessary to extend membrane protrusion in the direction of migration [2]. This morphologic polarity is regulated by Rac, a member of the small GTPases that cycle GDP-bound inactive and GTP-bound active states [3]. Like other small GTPases, conversion of the GDP-bound Rac to the active state is catalyzed by guanine nucleotide exchange factors (GEFs). The classical GEFs share the Dbl-homology (DH) domain to mediate the GTP-GDP exchange reaction [3]. Until recently, DH-domain containing proteins have been considered to be the universal GEFs in eukaryotes. However, accumulating evidence indicates that the evolutionarily conserved DOCK family proteins act as major GEFs in varied biological settings. In mammals, DOCK family proteins consist of 11 members, all of which contain a conserved DOCK homology region (DHR)-2 (also know as CZH2 or Docker) domain and mediate the GTP-GDP exchange reaction for Rho family of small GTPases through this domain [4,5,6].
Lymphocytes are highly motile leukocytes, and play central roles in acquired immune responses to invading pathogens. Lymphocytes differentiate in primary lymphoid organs, and migrate into secondary lymphoid tissues such as the lymph nodes and spleen via the blood. This migration process of lymphocytes is regulated by chemokines such as CCL21, CCL19, CXCL13, and CXCL12, and these chemokines are expressed on stromal cells and vascular or lymphatic endothelial cells [7]. When chemokines bind to their receptors, heterotrimeric G protein is dissociated into aand bc-subunits, which activates a variety of signaling pathways including Rac. In lymphocytes, chemokine-induced Rac activation is critically dependent on DOCK2, an atypical Rac activator predominantly expressed in hematopoietic cells [8,9]. As a result, DOCK2-deficient (DOCK2 2/2 ) lymphocytes exhibit a severe defect in chemotactic responses [9,10,11,12]. Although DOCK2 does not contain the DH domain, DOCK2 catalyzes the GTP-GDP exchange reaction for Rac by means of its DHR-2 domain [4,5]. DOCK2 DHR-2 domain is composed of three lobes (lobes A, B, and C), and Rac binding site and catalytic center are generated entirely from lobes B and C [13]. On the other hand, lobe A has been implicated in dimer formation of DOCK2 and DOCK9 [13,14,15], yet its physiological significance remains unknown. In this study, we examined whether lobe A-mediated DOCK2 dimerization is functionally important for Rac activation and lymphocyte migration.

Results and Discussion
A recent structure of human DOCK2 DHR-2 complexed with Rac indicates that, while Rac binding site and catalytic center are generated entirely from lobes B and C, lobe A mediates dimerization of DHR-2 domain [13]. This was also confirmed in our structural analysis ( Figure 1A) [16]. Consistently, sizeexclusion chromatography revealed that murine DOCK2 DHR-2 mutant lacking lobe A (DHR-2 Dlobe A ) fails to dimerize ( Figure 1B). The valine residue at position 1538 of DOCK2 DHR-2 is known to function as a nucleotide sensor [13]. When this residue was mutated to alanine (designated DHR-2 V1538A ), the Rac GEF activity was almost completely lost ( Figure 1C). However, deletion of lobe A from DOCK2 DHR-2 did not affect the Rac GEF activity in vitro ( Figure 1C), as recently reported [17].
To further examine whether full-length DOCK2 dimerizes via lobe A, we simultaneously expressed FLAG-and green fluorescent protein (GFP)-tagged wild-type (WT) or mutant DOCK2 in HEK-293T cells. Reciprocal immunoprecipitation experiments using the antibody specific for each tag revealed that DOCK2 forms homodimer ( Figure 2A). However, such homodimerization was not induced when DOCK2 mutant lacking lobe A (Dlobe A) was expressed as either form (Figure 2A). Contacts at the human lobe A interface involve aromatic and aliphatic residues such as tyrosine-1315, leucine-1322, and tyrosine-1329 ( Figure 1A). When the corresponding amino acid residues of murine DOCK2 were mutated to alanine, this 3A mutant failed to form homodimer ( Figure 2A). These results indicate that lobe A is also important for dimerization of DOCK2 in cells.
Having found that full-length DOCK2 dimerizes via lobe A in cells, we next examined whether the Dlobe A mutant retains the ability to bind to Rac. For this purpose, extracts of HEK-293T cells expressing FLAG-tagged WT DOCK2 or Dlobe A were pulled down with glutathione-S transferase (GST)-fusion Rac1 in the presence of 10 mM EDTA or 10 mM MgCl 2 plus 30 mM GTPcS. These experiments revealed that Dlobe A as well as WT DOCK2 binds selectively and efficiently to the nucleotide-free Rac1 ( Figure 2B). Consistent with this finding, Rac activation was comparably induced by overexpression of WT DOCK2 and Dlobe A in HEK-293T cells (Fig. 2C). Thus, Dlobe A retains the ability to bind to Rac and to mediate the GTP-GDP exchange reaction for Rac when overexpressed in HEK-293T cells.
To examine the role of DOCK2 dimerization under more physiological conditions, we stably expressed WT or mutant DOCK2 in BW5147abthymoma cells lacking the expression of DOCK2 [9,16]. Although BW5147abcells were unable to migrate efficiently on stromal cells prepared from the lymph nodes, the expression of WT DOCK2, but not the V1538A mutant, significantly improved the motility of these cells ( Figure 3A), indicating that DOCK2 regulates migratory response of BW5147abcells through Rac activation. We then analyzed the motility of BW5147abcells expressing Dlobe A or 3A. The expression level of these mutants was comparable to that of WT DOCK2 in two independent clones ( Figure 3B). In addition, both Dlobe A and 3A showed definite binding to the nucleotide-free Rac1 in pull-down assays ( Figure 3C). These results indicate that the Dlobe A and 3A mutants expressed in BW5147abcells can potentially act as Rac GEFs. Surprisingly, however, the expression of Dlobe A or 3A in BW5147abcells failed to restore the motility ( Figure 3A). BW5147abcells express CXCR4, a chemokine receptor for CXCL12. When BW5147abcells expressing WT DOCK2 were stimulated in suspension with CXCL12, they exhibited polarized morphology with focused distribution of Factin ( Figure 3D,E). However, such morphologic polarity was again not induced in BW5147abcells expressing Dlobe A or 3A ( Figure 3D,E).
Having found that the expression of Dlobe A and 3A was unable to restore motility and polarity in BW5147abcells, we next examined activation of Rac in BW5147abcells by retrovirally transducing Raichu-Rac [18], the fluorescence resonance energy transfer (FRET)-based biosensor to monitor Rac activation at the membrane. Comparison of the FRET efficiency revealed that Rac was activated at the leading edge membrane in the case of BW5147abcells expressing WT DOCK2 ( Figure 4A,B). However, such Rac activation was scarcely found in BW5147abcells expressing Dlobe A as well as those expressing V1538A ( Figure 4A,B). Consistent with this finding, the association between DOCK2 and endogenous Rac was detected in BW5147abcells expressing WT DOCK2, but not those expressing Dlobe A or V1538A ( Figure 4C). These results indicate that lobe A-mediated DOCK2 dimerization is required to bind to and activate Rac effectively in BW5147abcells.
Finally, we wanted to test whether DOCK2 dimerization is also important in primary T cells. For this purpose, we crossed DOCK2-deficient (Dock2 2/2 ) mice with transgenic mice expressing the gene encoding coxsackie-adenovirus receptor (CAR) under the Lck promoter [9,19], and developed the experimental system with which full-length DOCK2 can be readily expressed in Dock2 2/2 T cells by adenoviral transfer ( Figure 5A). When GFPtagged WT DOCK2 was expressed in Dock2 2/2 T cells, the migration speed markedly increased, compared with that of the control expressing GFP alone ( Figure 5B). However, similar to the case of V1538A, the expression of Dlobe A and 3A did not improve the motility of Dock2 2/2 T cells ( Figure 5B).
In this study, we have shown that lobe A-mediated DOCK2 dimerization is important for Rac activation and lymphocyte migration. We found that lobe A is dispensable for the GTP-GDP exchange activity of the DHR-2 domain in vitro, and that DOCK2 mutants having a defect in dimerization retain the ability to bind to the nucleotide-free Rac in pull-down assays. However, the expression of these mutants in BW5147abcells and Dock2 2/ 2 T cells failed to restore Rac activation and migratory responses. Our results thus provide the first evidence for the physiological significance of DOCK2 dimerization in DOCK2-mediated cellular functions. The precise mechanism of how DOCK2 dimerization regulates the efficacy of Rac activation in cells is currently unknown. However, it seems likely that DOCK2 dimerization enhances signal transduction by recruiting Rac to the same signaling complex and by increasing local concentration of active Rac at plasma membrane, as was suggested for protein dimerization in other systems [20]. Interestingly, it has been reported that the DH domains of the classical GEFs also form oligomers, and the mutations causing a defect in oligomerization impaired their biological activities [21,22,23]. Therefore, oligomerization may be a common feature in GEFs to facilitate their signaling activities in cells where only limited amounts of target GTPases are available.

Ethics Statement
The protocol of animal experiments was approved by the committee of Ethics of Animal Experiments of Kyushu University (Permit Number: A24-016-0). All efforts were made to minimize suffering during the experiments.

Mice
Dock2 2/2 mice and CAR-expressing transgenic mice have been described previously [9,19]. Mice were kept under specific pathogen-free conditions in the animal facility of Kyushu University.

Size-exclusion Chromatography
Proteins were chromatographed on a Superdex 200 10/300 GL column (GE Healthcare, Piscataway, NJ) connected to an AKTA

In vitro GEF Assays
The gene encoding DOCK2 DHR-2 domain or its mutants was cloned into the pET-SUMO vector to express fusion protein [17]. For measurement of the Rac GEF activity, GDP-loaded GST-Rac1 was incubated with BODIPY-FL-GTP (Invitrogen, Carlsbad, CA) for 5 minutes at 30uC. After equilibration, WT DOCK2 DHR2 (DHR-2 WT ) or its mutants (DHR-2 Dlobe A and DHR-2 V1538A ) were added to the mixtures, and the change of BODIPY-FL fluorescence (excitation = 488 nm, emission = 514 nm) was monitored at 30uC using a XS-N spectrofluorimeter (Molecular Devices, Sunnyvale, CA).  Plasmids, Transfection, and Cell Culture The genes encoding FLAG-or hemagglutinin (HA)-tagged DOCK2 and its mutants were created with pBJ1 vector [24]. The genes designed to express GFP-or FLAG-tagged DOCK2 and its mutants under the ubiquitin promoter were subcloned into pENTR3C (Invitrogen) before use. These plasmids were transfected by electroporation or with polyethylenimine to BW5147abthymoma cells [25] or HEK-293T cells (RIKEN BioResource Center; RCB2202), respectively. The adenoviral vector pAd-DOCK2 was generated by specific recombination of pENTR3C-DOCK2 with pAd/PL-DEST (Invitrogen). The pAd-DOCK2 with or without mutations was transfected into HEK-293A cells (Invitrogen) to amplify the recombinant adenovirus. The resultant virus was purified with cesium chloride centrifugation, and was used to infect CAR-expressing Dock2 2/2 T cells. Cells were then stimulated with immobilized anti-CD3 (20 mg/ml; eBioscience, San Diego, CA) and anti-CD28 (2 mg/ml; eBioscience) antibodies in RPMI 1640 medium supplemented with 10% fetal calf serum, IL-2 (100 U/ml; PeproTech, Rocky Hills, NJ), and IL-7 (20 ng/ml; PeproTech). After 24 hours of incubation, viable cells were recovered using Lympholyte-M (Cedarlane, Ontario, Canada) and suspended in Leibovitz L-15 medium (Invitrogen) for migration assays.

Migration Assays
Stromal cells were prepared from the peripheral lymph nodes as previously described [16] and stimulated with tumor necrosis factor-a (10 ng/mL; PeproTech) the day before assays. Primary T cells or BW5147abcells expressing WT or mutant DOCK2 were placed on a monolayer of stromal cells in the presence or absence of CCL21 (100 nM; R&D Systems, Minneapolis, MN). After 4 hours of incubation, phase-contrast images were obtained every 30 seconds for 20 minutes at 37uC on an IX-81 inverted microscope (Olympus, Tokyo, Japan). Migration of individual cells was tracked using the MetaMorph imaging software (Molecular Devices), and the migration speed was calculated by dividing the total path length by the total assay time.

FRET
The retroviral vector pMXs was used to create the plasmid encoding Raichu-Rac [18]. This plasmid was transfected into Platinum-E packaging cells [26] to generate the recombinant retrovirus. Retroviral infection was performed as previously described [27]. FRET (excitation 440 nm/emission peak 527 nm) and CFP (excitation 440 nm/emission peak 475 nm) images were obtained by using an Olympus IX-81 inverted microscope. After recordings were made, ratio images of FRET/ CFP were created with MetaMorph and were used to represent the efficiency of the FRET.