Phosphorylation of Rab5a Protein by Protein Kinase Cϵ Is Crucial for T-cell Migration*

Background: Rab5a GTPase plays important roles in intracellular transport and cell signaling. Results: T-cell stimulation through the integrin LFA-1 or the chemokine receptor CXCR4 induces PKCϵ-dependent phosphorylation of Rab5a at Thr-7, which is crucial for cytoskeleton remodeling and cell migration. Conclusion: PKCϵ-Rab5a-Rac1 axis regulates T-cell motility. Significance: The study provides novel insights into the role of Rab5a in the adaptive immune response. Rab GTPases control membrane traffic and receptor-mediated endocytosis. Within this context, Rab5a plays an important role in the spatial regulation of intracellular transport and signal transduction processes. Here, we report a previously uncharacterized role for Rab5a in the regulation of T-cell motility. We show that Rab5a physically associates with protein kinase Cϵ (PKCϵ) in migrating T-cells. After stimulation of T-cells through the integrin LFA-1 or the chemokine receptor CXCR4, Rab5a is phosphorylated on an N-terminal Thr-7 site by PKCϵ. Both Rab5a and PKCϵ dynamically interact at the centrosomal region of migrating cells, and PKCϵ-mediated phosphorylation on Thr-7 regulates Rab5a trafficking to the cell leading edge. Furthermore, we demonstrate that Rab5a Thr-7 phosphorylation is functionally necessary for Rac1 activation, actin rearrangement, and T-cell motility. We present a novel mechanism by which a PKCϵ-Rab5a-Rac1 axis regulates cytoskeleton remodeling and T-cell migration, both of which are central for the adaptive immune response.

Leukocyte trafficking is a key element of an immune response that is critically dependent on a multifunctional molecular array of integrin-mediated adhesions and chemokine signals. Interaction of the ␣L␤2 integrin leukocyte function associated-antigen-1 (LFA-1), expressed on T-cells, with its ligand intercellular adhesion molecule-1 (ICAM-1), expressed on the surface of high endothelial venules and other cell types, promotes firm adhesion of T-cells to blood vessel walls and their migration through endothelial junctions into secondary lymphoid organs and sites of inflammation (1)(2)(3). In addition, chemokines (for example the stromal cell-derived factor-1␣ (SDF-1␣)) and their interactions with specific receptors on T-cells (such as C-X-C chemokine receptor type 4 (CXCR4)) direct T-cells to arrest on post-capillary venules at sites of infection/injury and secondary lymphoid organs (4). This multistep process of T-cell migration is precisely regulated by a plethora of signaling molecules including kinases, adaptors, and motor proteins that ultimately result in the reorganization of cytoskeletal systems and provide mechanical force to propel the cell forward. Emerging evidence suggests an important role for endocytic transport and vesicular traffic in the process of cytoskeletal remodeling and integrin-mediated cell motility (5)(6)(7)(8).
Small GTPases of the Rab family are known to function as molecular switches that regulate a variety of cellular processes including proliferation, differentiation, signal transduction, and cytoskeletal reorganization (9 -11). In particular, they control intracellular vesicle transport, such as receptor-mediated endocytosis, exocytosis, and receptor recycling (12)(13)(14), all of which are important for cell migration. Rab GTPases cycle between a GDP-bound inactive state and a GTP-bound active state. Inactive GDP-bound Rab proteins typically reside in the cytosol, whereas GTP-bound active Rab proteins localize on intracellular membranes where effector targets reside. Thus, the Rab family of proteins typically require turnover of GTP/ GDP to function. This reversible control of Rab GTPase activity via GDP/GTP exchange is crucial to their function as regulators of vesicular trafficking (15). One member of the Rab GTPase family, Rab5a, is known to regulate endocytosis, endosomal dynamics, tethering, and fusion of early endosomes. Rab5a con-tributes to cytoskeletal remodeling in part by regulating the endocytic trafficking of Rac, a member of the Rho GTPase subfamily and an important component of the actin cytoskeletal system (16 -18).
The protein kinase C (PKC) family comprises a group of highly related multifunctional serine/threonine kinases that phosphorylate a wide variety of protein targets and play important roles in cellular signaling pathways including those involved in cell migration (19). More than 10 isoforms of PKC have been identified in mammalian tissues, each of which plays a distinct role in the processing and modulation of a variety of physiologic and pathologic responses to external signals. There is substantial evidence that members of the PKC family regulate integrin-mediated cell spreading and migration (20 -23). For example, we have established that signaling through the integrin LFA-1 in T-cells results in the association of PKC␤ with the centrosome and microtubule cytoskeleton, a process that is essential for cell migration (21). In addition, another PKC family member PKC⑀ has been identified as a regulator of ␤1 integrin-dependent cell migration (23)(24)(25).
Although extensive progress has been made to understand T-cell signal transduction pathways, the potential involvement of Rab5a in T-cell migration has not been explored to the similar extent. Moreover, the precise contribution of the PKC isoforms in LFA-1-mediated downstream signaling has remained unclear. In the current study we demonstrate that T-cell stimulation through the integrin LFA-1 or the chemokine receptor CXCR4 triggers PKC⑀-dependent phosphorylation of Rab5a at threonine 7 (Thr-7), which is crucial for Rac1 activation, actin cytoskeleton remodeling, and T-cell migration.

EXPERIMENTAL PROCEDURES
Cells and Reagents-Primary human peripheral blood lymphocyte (PBL) T-cells were isolated from healthy volunteers by standard methods and cultured with phytohemagglutinin and IL-2 as described previously (26). The human T-cell line HuT-78 was obtained from the American Type Culture collection (ATCC) and cultured as described (27)(28)(29)(30). Motility-inducing anti-LFA-1 (clone SPV-L7) was from Monosan (Cambridge Biosciences). Human recombinant ICAM-1-Fc was from R&D Systems. Recombinant human IL-2 and SDF-1␣ were from Peprotech. Phytohemagglutinin, goat anti-human IgG (Fc-specific), goat anti-mouse IgG, mouse anti-␣-tubulin, phalloidin-TRITC, and protein A-agarose beads were from Sigma. Mouse anti-PKC⑀ and mouse anti-Rac1 antibodies were from BD Biosciences. Mouse anti-Rab5a was from Synaptic Systems GmbH. Rabbit anti-Rab5a and anti-PKC⑀ were from Santa Cruz Biotechnology, Inc. HRP-conjugated anti-rabbit and anti-mouse antibodies were from Cell Signaling Technology. The PKC inhibitor bisindolylmaleimide I was from Merck Millipore. Alexa Fluor 488-and 568-conjugated secondary antibodies, Hoechst 33258, and Alexa Fluor 568-transferrin conjugate were purchased from Molecular Probes Invitrogen.
Induction of LFA-1-mediated T-cell Migration-We used our well characterized migration-triggering model system where T-cells are stimulated through the LFA-1 receptor via crosslinking with an immobilized anti-LFA-1 antibody or with the physiological ligand ICAM-1 (21, 26 -29). These migrating T-cells display dynamic protrusions of the leading edge, flexible oscillatory shape changes of the cell body, and detachment of the trailing edge concomitant with polarized or localized distribution of signaling molecules and cytoskeletal rearrangements. Briefly, 96-well tissue culture plates (flat bottom, Nunc TM ), 8-well LabTek chamber slides (Nalge Nunc Inc.), or round coverglass, depending on the particular assay type, were coated overnight at 4°C with 5 g/ml goat anti-mouse IgG in sterile phosphate-buffered saline (PBS). After washing with PBS, the surface was then incubated with a cross-linking anti-LFA-1 antibody (SPV-L7) for 2 h at 37°C. ICAM-1-coated surfaces were prepared by adding 5 g/ml anti-Fc-specific goat antihuman IgG to the culture plates/chamber slides and incubating overnight at 4°C followed by 1 g/ml rICAM-1-Fc for 2 h at 37°C. After washing the surfaces with PBS, HuT-78 or PBL T-cells were loaded into the coated wells and incubated in 5% CO 2 at 37°C for various time points as indicated in "Results" and figure legends of the particular experiment. Migration assays on ICAM-1 also contained 5 mM MgCl 2 and 1.5 mM EGTA in the cell culture medium to induce the high affinity form of the LFA-1 receptor on T cells (11).
Plasmid Constructs and Electroporation-Plasmid vector encoding Rab5a fused to green fluorescent protein (Rab5a-GFP), Rab5aQ79L-GFP, and Rab5aS34N-GFP were generous gifts from M. Zerial, Max Planck Institute of Molecular Cell Biology and Genetics. PKC⑀-GFP and kinase-dead (KD) PKC⑀-GFP were kind gifts from D. Romberger, Department of Internal Medicine, University of Nebraska. Lifeact-Ruby (31) was a generous gift from R. Wedlich-Soldner, Max Planck Institute of Biochemistry. Site-directed mutagenesis of Thr-7, Ser-84, and Ser-123 residues of Rab5a to alanine or glutamic acid was performed by Bio S&T Inc., Montreal, Quebec. Plasmid vectors expressing wild-type or Thr-7 mutant Rab5a as a photoactivable GFP fusion (Rab5a-paGFP or Rab5aT7A-paGFP constructs, respectively) were made by replacing the eGFP cDNA with paGFP cDNA (32), a generous gift from J. Lippincott-Schwartz, Cell Biology and Metabolism Branch, National Institutes of Health. The eGFP cDNA of Rab5aT7A-GFP and PKC⑀-GFP constructs were replaced with the mCherry cDNA (a kind gift from R. Tsien, University of California, San Diego) to generate the corresponding cherry-tagged proteins. HuT-78 T-cells were nucleofected with plasmid constructs using the Amaxa Nucleofector TM system according to the manufacturer's recommendations.
RNA Interference-To deplete PKC⑀ expression in primary PBL T-cells, a mixture of four siRNA duplexes (Dharmacon ON-TARGETplus SMARTpool siRNA Reagents, Thermo Fisher Scientific) targeted against PKC⑀ or nonspecific control siRNAs was used. Cells (5 ϫ 10 6 ) were nucleofected with 1000 nM siRNA using the Amaxa Nucleofector TM system according to the manufacturer's instructions and then harvested after 72 h.
Production of Phospho-Thr-7-specific Rab5a Antibody-A polyclonal antibody against Thr-7-phosphorylated Rab5a was generated by immunizing two rabbits with a synthetic peptide comprising amino acids 2-12 (ASRGApTRPNGP, where pT is phosphorylated Thr) of human Rab5a (NeoMPS PolyPeptide). The total serum was then purified by affinity chromatography on phosphopeptide-coupled columns (Amino-link, Pierce) according to manufacturer's instructions. Briefly, 1 mg of Thr(P)-7 peptide or (OH)Thr-7 peptide was coupled to separate amino-link columns. A total of 0.5 ml of serum was first applied to the (OH)Thr-7-peptide-coupled column to remove nonspecific binding against the (OH)Thr-7 residue of Rab5a. The pre-cleared serum was then purified with a Thr(P)-7-peptide-coupled column and tested by ELISA for specificity and sensitivity.
Cell Lysis, Co-immunoprecipitation, and Western Immunoblotting-Cell lysis was performed as described previously (27) with minor modifications. Briefly, cells were washed with icecold PBS and lysed in the lysis buffer containing Triton X-100 (1%) and protease inhibitors phenylmethylsulfonyl fluoride (2 mM), leupeptin (10 g/ml), and aprotinin (10 g/ml). The protein content of the cell lysates was determined by the Bradford or BCA assay. The immunoprecipitation was performed as described previously (28,33). Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) of the cellular lysates or immunoprecipitates and subsequent Western immunoblotting using indicated antibodies were performed as described (28). Where indicated, affinity-purified or crude anti-phospho Rab5a(Thr-7) antiserum (1:2000 dilution in blocking buffer) was preincubated with 20 g/ml non-phosphorylated Thr-7 peptide for 1 h at room temperature before incubation on the membranes. The immunoreactive bands were visualized using the LumiGLO chemiluminescent detection system (Cell signaling Technology) and subsequent exposure to CL-XPosure TM light-sensitive film (Thermo Scientific). Densitometric analyses of the Western blots were performed by using ImageJ software.
In Vitro Kinase Assay-Rab5a was cloned into the GST expression vector pGEX-6p1, over-expressed in BL21(DE3)pLysS cells, and purified by using glutathione-Sepharose beads (GE Healthcare) according to the manufacturer's instructions. PKC⑀-dependent phosphorylation was measured by incorporation of 32 P i from [␥-32 P]ATP. Purified recombinant Rab5a-GST fusion protein, GST alone, or myelin basic protein (MBP; 200 ng) was incubated in 50 l of kinase assay buffer (40 mM Tris, pH 7.5, 40 mM MgCl 2 , 0.2 mM HEPES, pH 7.4, 0.2 mM DTT, 0.0002% Triton X-100, 0.3 g/ml BSA) containing 1 M ATP, 2 Ci [ 32 P]ATP, 1 M phorbol 12,13-dibutyrate, and 160 M phosphatidylserine. After 20 min of incubation at 30°C, the reaction was stopped by adding 5ϫ SDS-PAGE sample buffer (250 mM Tris, pH 11.5, 10% SDS, 50% glycerol, and 25% ␤-mercaptoethanol) and analyzed by SDS-PAGE followed by autoradiography. In vitro kinase assays were also carried out as above using nonradioactive ATP (Sigma) in place of [␥-32 P]ATP. After allowing the kinase reaction to proceed, SDS-PAGE sample buffer was added, and the samples were resolved on SDS-PAGE gels and probed by Western blotting with the phospho-T7 Rab5a antiserum.
Confocal Imaging and Photoactivation-For confocal imaging and analysis, cells were seeded to rest or migrate on coverslips as described above and then fixed with 3% (w/v) paraformaldehyde in PBS (27). T-cells were permeabilized with 0.3% (v/v) Triton X-100 in PBS and blocked with 5% w/v BSA in PBS for 30 min. After blocking, cells were incubated with primary antibodies for 1 h at room temperature. After washing, cells were incubated with Alexa Fluor 488-or 568-conjugated secondary antibody for 1 h at room temperature. Cells were also stained with Hoechst to visualize nuclei or phalloidin-TRITC to visualize F-actin. After washing, cells were mounted in fluorescence mounting medium (Dako) and stored at 4°C. Fluorescence microscopy was performed using a confocal microscope LSM 510 with a Plan-Apochromat differential interference contrast 63ϫ oil objective and 1.4 numerical aperture (Carl Zeiss, Inc.). Images were analyzed using the LSM Imaging software (Carl Zeiss).
For photoactivation and confocal live-cell imaging, cells expressing photoactivable fluorescently labeled proteins were stimulated to migrate on coverslips as described earlier and then placed onto a heated chamber with the internal temperature set at 37°C (PerkinElmer Life Sciences). Photoactivation was performed with a 405-nm laser using the photobleaching function of LSM Imaging software (Carl Zeiss Inc.) in a timelapse mode. Generally, 1 pulse of the 405-nm laser was sufficient to activate paGFP so that it produced very bright fluorescence emission that was detected by excitation at 488 nm using a 500 -530-nm band pass filter. At least 20 different microscopic fields were observed for each sample.
High Content Analysis-A high content analysis protocol for T-cell morphology analysis has been optimized and established in our laboratory as described (29,34). Briefly, cells were seeded in triplicate on 96-well flat bottom plates precoated with either poly-L-lysine or anti-LFA-1 for 2 h. After washing, cells were fixed by incubating them for 20 min with 3% (w/v) paraformaldehyde in PBS. Attached cells were then stained for F-actin using phalloidin-TRITC, and the was nucleus stained using Hoechst. Plates were scanned (9 randomly selected fields/well at 20ϫ) using an automated microscope IN Cell Analyzer 1000 (GE Healthcare), and the acquired images were automatically analyzed by IN Cell Investigator software (Version 1.6) using multitarget analysis bio-application module (GE Healthcare).
Transferrin Internalization-Cells were serum-starved for 1 h and stimulated to migrate on anti-LFA-1 as described above before incubating with Alexa Fluor 568-transferrin conjugate for 30 min on ice. Cells were then rinsed twice in ice-cold PBS and transferred to 37°C incubator for 10 min to allow internalization before being fixed with 3% (w/v) paraformaldehyde and imaged.
Transwell Migration Assay-Transwell chambers (5-m pores; Corning Costar) were precoated with 5 g/ml rICAM-1-Fc at 4°C overnight and blocked with 5% (w/v) BSA for 1 h at 37°C. Serum-starved T-cells were loaded in triplicate in the upper chambers and allowed to migrate toward 50 ng/ml SDF-1␣-enriched serum-free medium in the lower wells at 37°C. After 4 h, migrated cells in the lower wells were fixed and stained with Hoechst. IN Cell Analyzer 1000 (GE Healthcare) was employed to perform whole well cell counts. Data from at least three independent experiments were pooled by normalizing the counts with corresponding controls.
Determination of Rac1 Activity in Cells (PBD Binding Assay)-The capacity of Rac1-GTP to bind to GST-PBD (p21-activated kinase binding domain) beads was used to analyze the activity of this GTPase. Serum-starved T-cells (2 ϫ 10 6 ) were stimulated with or without immobilized ICAM-1-Fc or SDF-1␣ for 10 min and lysed in 500 l of lysis buffer as described (27). Rac1 activity in the cellular lysates was determined using Rac1 activation assay kit according to the manufacturer's instructions (Millipore). Briefly, cleared lysates were incubated overnight at 4°C with glutathione-Sepharose 4B beads coupled to GST-PBD. After extensive washing, beads were boiled in SDS-PAGE loading buffer, and the amount of bound Rac1-GTP was detected by Western immunoblotting with an anti-Rac1 antibody. In all cases, whole-cell lysates were also analyzed for total Rac1 for normalization purposes. Bands were visualized and quantified as described above, and the Rac1-GTP levels were normalized to the input levels of total Rac1.
Statistical Analysis-The data are expressed as mean Ϯ S.E. For comparison of two groups, p values were calculated by twotailed unpaired Student's t test. In all cases p values Ͻ0.05 were considered to be statistically significant.

PKC⑀ Localizes to Endosomes in Migrating T-cells and Associates with Rab5a-We investigated the subcellular distribution of PKC⑀ in resting and migrating T-cells by transfecting
the HuT-78 cell line with a plasmid construct encoding PKC⑀-GFP. After allowing the transfected cells to polarize on an anti-LFA-1 cross-linking antibody, we monitored the subcellular distribution of the PKC⑀-GFP expression by confocal microscopy. We noted that PKC⑀-GFP had a distinct localization, as it was distributed around the neck/centrosomal region in migrating cells and also displayed a vesicular pattern that was reminiscent of endosomes (Fig. 1A). To investigate whether PKC⑀ was localized to endosomes in T-cells, we transfected HuT-78 cells with a PKC⑀-cherry expression vector in combination with a Rab5a-GFP construct that served as a marker of endosomes. Co-localization of PKC⑀ and Rab5a was clearly evident at the neck region and throughout the cell body of migrating HuT-78 T-cells, as demonstrated by confocal microscopy and analysis ( Fig. 1, B and C). Notably, both proteins also partially co-localized in resting T-cells (Fig. 1D).
We next examined the interactions between PKC⑀ and Rab5a by live cell imaging of migrating T-cells. For this purpose, we expressed Rab5a as a photoactivatable GFP fusion protein (Rab5a-paGFP) in HuT-78 T-cells that also co-expressed PKC⑀-cherry (31). Time-lapse confocal microscopy revealed transient interactions between Rab5a-paGFP-positive endosomes and the PKC⑀-cherry expressing compartments around the neck/centrosomal region of migrating T-cells (supplemental Video S1). A similar localization pattern of Rab5a-containing vesicles has been reported in cultured human U2OS cells (35). Therefore, we focused more closely on the centrosomal region of migrating cells by selecting a population of Rab5a-paGFP positive vesicles and followed their kinetics over time concomitant with PKC⑀-cherry expression (supplemental Video S2). Upon irradiation, Rab5a-paGFP present around the centrosomal area was photo-activated, and in the following 30-s period, Rab5a-positive endosomes fused with a PKC⑀cherry compartment before separating (Fig. 1E, white circle). Next, we applied irradiation aiming at a "single point" corresponding to a PKC⑀-positive vesicle, and photo-activation of Rab5a-paGFP was visualized within the confines of this structure. The two proteins interacted with each other for a further 60 s and then disengaged (supplemental Video S3). These observations of real-time protein-protein interactions in T-cells after LFA-1 stimulation imply that PKC⑀ and Rab5a dynamically interact during the migratory process. Importantly, endogenous PKC⑀ co-localization with endogenous Rab5a was also detected in primary human PBL T-cells migrating on ICAM-1, which is the physiological ligand for the LFA-1 integrin (Fig. 1F). To quantify the protein-protein interaction of endogenous PKC⑀ with endogenous Rab5a, we performed coimmunoprecipitation studies utilizing primary human PBL T-cells. We observed significantly increased levels (Ͼ7-fold) of PKC⑀-associated Rab5a in LFA-1/ICAM-1-stimulated migrating PBL T-cells as compared with that of resting cells (Fig. 1G). Interestingly, the association of Rab5a with PKC⑀ was not influenced by the levels of GTP on Rab5a because constitutively active GTP-bound Rab5a (Q79L mutant) and dominant-negative GDP-bound Rab5a (S34N mutant) both associated with the endogenous PKC⑀ in migrating T-cells to similar extents (Fig.  1H). Hence, the association of Rab5a with PKC⑀ is a nucleotideindependent interaction. Taken together, these studies confirm that PKC⑀ localizes to endosomes in migrating T-cells and physically interacts with Rab5a after LFA-1/ICAM-1 stimulation.
PKC⑀ Phosphorylates Rab5a on Threonine 7-We next investigated whether PKC⑀ could phosphorylate Rab5a. An in vitro kinase assay was performed where active recombinant PKC⑀ was incubated with a purified Rab5a-GST fusion protein in the presence of radio-labeled ATP ([ 32 P]ATP). Incubation of recombinant PKC⑀ with GST alone or with MBP in the presence of [ 32 P]ATP was also used as negative or positive control, respectively. Incorporation of [ 32 P]ATP into Rab5a, GST, or MBP was measured by SDS-PAGE followed by autoradiography. This analysis clearly indicated that PKC⑀ was capable of phosphorylating Rab5a in vitro, whereas GST alone was not phosphorylated ( Fig. 2A). In contrast, MBP was highly phosphorylated when incubated with PKC⑀ in the presence of [ 32 P]ATP ( Fig. 2A).
Members of the PKC family generally target serine/threonine residues that lie within (S/T)X(K/R), (K/R)XX(S/T), or (L/R)X(S/T) motifs, where X indicates any amino acid (36). Using an oriented peptide library, Nishikawa et al. (37) also reported that PKC⑀ and other novel PKC isozymes prefer hydrophobic amino acids at positions ϩ2, ϩ3, and ϩ4. Based on these criteria, we identified three potential sites on human Rab5a, namely Thr-7, Ser-84, and Ser-123, that may be phosphorylated by PKC⑀ (Fig. 2B). These three sites on human Rab5a are evolutionarily conserved, as they are present on Rab5a from all other species (supplemental Fig. S1). Thr-7 lies within the N terminus of Rab5a, a region previously shown to be critical for its functionality, whereas Ser-84 and Ser-123 fall within the highly conserved GTP/GDP binding switch regions of small GTPases (38 -40).
We used a number of approaches to determine that Rab5a was phosphorylated on Thr-7 by PKC⑀ and that phosphorylation of this site (but not at Ser-84 or Ser-123) was crucial for T-cell migration. First, we demonstrated that transfecting HuT-78 T-cells with a kinase-dead PKC⑀ construct perturbed

PKC⑀-Rab5a-Rac1 Axis in T-cell Motility
their ability to migrate through ICAM-1-coated filters toward the chemokine SDF-1␣ in transwell chambers (Fig. 2C). Second, we generated mutant constructs of Rab5a by replacing Thr-7 (Rab5a-T7A), Ser-84 (Rab5a-S84A), or Ser-123 (Rab5a-S123A) residues with alanine, transfected the constructs into HuT-78 T-cells, and analyzed the ability of the cells to migrate toward SDF-1␣ through transwell filters coated with ICAM-1. Expression of the Rab5a-T7A construct in HuT-78 T-cells perturbed migration in comparison to cells transfected with Rab5a-WT (Fig. 2D). In contrast, mutation of the Ser-84 or Ser-123 sites on Rab5a to alanine did not significantly affect T-cell transwell migration (Fig. 2D). We also addressed the effect of mutating the Thr-7 site with a glutamic acid residue to mimic constitutive phosphorylation of Rab5a (Rab5a-T7E) and analyzed its effect on T-cell migration. We found that the expression of this construct in T-cells also perturbed T-cell migration but not to the same extent as the non-phosphorylated Rab5a-T7A mutant (Fig. 2E). Collectively, these experiments indicate that PKC⑀ and phosphorylation/dephosphorylation of Rab5a are independently required for T-cell motility.
To address whether the Thr-7 residue of Rab5a is indeed phosphorylated, we developed an antibody against the N-terminal phosphorylated Thr-7 site as described under "Experimental Procedures" above. The specificity of the antibody was
Next, we examined Rab5a phosphorylation at Thr-7 induced via LFA-1 signaling in cells. PBL T-cells were incubated for a range of time points on an immobilized ICAM-1-coated surface, lysed, and analyzed by Western immunoblotting using the anti-phospho-Rab5a(Thr-7) antiserum. Thr-7 phosphorylation of Rab5a was barely detectable in resting T-cells, which was significantly increased after LFA-1/ICAM-1 stimulation, reaching maximum levels within the time frame of 10 -30 min, and then declined to basal level at 2 h (Fig. 4A). Similar results were obtained when PBL T-cells were stimulated with either SDF-1␣ or with the PKC activator phorbol 12-myristate 13-acetate (PMA; Fig. 4B), further suggesting that Rab5a Thr-7 phosphorylation was PKC-dependent.
To confirm the role of PKC in LFA-1-induced Rab5a Thr-7 phosphorylation, we pretreated PBL T-cells for 30 min with a 10 M concentration of the PKC inhibitor bisindolylmaleimide I. Western immunoblot analysis of T-cells stimulated with or without LFA-1/ICAM-1 showed that bisindolylmaleimide I pretreatment significantly inhibited Rab5a Thr-7 phosphorylation (Fig. 4C). Furthermore, specific knockdown (Ͼ75%) of PKC⑀ in T-cells using a siRNA approach (Fig. 4D) significantly inhibited LFA-1-mediated Thr-7 phosphorylation of Rab5a (Fig. 4E). These data clearly indicate that PKC⑀ is the specific PKC isoform mediating Thr-7 phosphorylation of Rab5a.
T7A Mutation Partially Inhibits Rab5a Targeting to Endosomal Compartments-Because mutation of Thr-7 to alanine on Rab5a affected T-cell migration, we investigated the mechanistic basis of this migratory defect. Truncation mutants of Rab5a lacking the first 14 -22 N-terminal residues have been shown to disrupt Rab5a functionalities (40 -43). However, these mutants remain prenylated, membrane-associated, and functionally interact with guanine nucleotides (40,41,44). In addition, functional aspects of protein folding and the crystal structure also remain unaffected (44,45). Therefore, the T7A mutation would not be predicted to interfere with the GTPase activity of Rab5a and should not affect its three-dimensional conformation. Hence, we investigated whether mutation of Thr-7 on Rab5a affected its subcellular distribution. Expression of wild-type Rab5a-GFP in HuT-78 T-cells demonstrated punctate and globular structures evenly distributed throughout the whole cell body, center, and the periphery (Fig. 5A). Although the expression pattern of Rab5a-S84A-GFP or Rab5a-Ser-123-GFP resembled that of wild type, Rab5a-T7A-GFP appeared to be more cytosolic and less punctate throughout the uropod (Fig. 5A). This distinctive pattern of Rab5a distribution was also verified by co-expressing both wild-type Rab5a-GFP and mutant Rab5aT7A-cherry constructs in T-cells (data not shown). Some punctate Rab5a-T7A-GFP-associated structures were still associated with endocytic vesicles labeled by internalized Alexa Fluor 568-transferrin-conjugate (Fig.  5B). These studies suggest that phosphorylation of the Thr-7 site partially influences Rab5a subcellular localization and that its mutation does not totally disrupt Rab5a targeting to the transferrin receptor-positive endosomal compartment. Consistent with the finding that the T7E mutant did not affect T-cell migration as much as the T7A mutant (Fig. 2E), the phos- . Characterization of anti-phospho Rab5a(Thr-7) antibody. A, serial dilutions (1/100 to 1/12,800) of the affinity-purified anti-phospho-Rab5a(Thr-7) antibody was titrated against phospho Thr-7-containing peptide ((p)T7 peptide) and analyzed by ELISA. Preimmune serum and non-phospho (OH)Thr-7-containing peptides ((OH)T7 peptide) were used as controls. B, various dilutions of the purified pRab5a(Thr-7) antibody ranging from 1:1 to 1:1000 were analyzed by immuno dot-blot assay (IB) using Thr(P)-7 peptide. (OH)Thr-7 peptide or pRab5a(Thr-7) antibody preincubated with the Thr(P)-7 peptide was used as the control. C, purified GST, GST-conjugated wild-type Rab5a (Rab5a-WT), or mutant Rab5a-T7A protein was separately incubated in the presence or absence of PKC⑀. The products were analyzed by Western blot using the affinity-purified anti-pRab5a(Thr-7) or anti-Rab5a (loading control). Densitometry analysis of the Western blots was performed and presented (mean Ϯ S.E.). *, p Ͻ 0.01. Data represents three independent experiments.
phomimetic Rab5a-T7E-GFP transfected cells expressed many small vesicular structures throughout the whole cell body colocalizing with transferrin and resembling the Rab5a-WT-like expression pattern (Fig. 5B).
We next employed photoactivation in combination with time-lapse microscopy to directly visualize and compare the dynamics of wild-type Rab5a and Rab5a-T7A at the centrosomal region in transfected cells. After transfection with Rab5a-paGFP or Rab5a-T7A-paGFP constructs, HuT-78 cells were seeded on coverslips coated with a cross-linking anti-LFA-1 antibody for 1 h to allow extended polarization. A pulse of 405-nm laser light aimed around the centrosomal region of the polarized cell resulted in immediate photoactivation of paGFP fusion proteins within the confines of this area. In cells expressing wild-type Rab5a-paGFP, a patch of photoactivated endosomes at the centrosomal region could be seen to rapidly disperse from this area over the following 5 min (Fig. 6A, supplemental Video S4). Conversely, the dispersal rate of photoactivated Rab5a-T7A from the centrosomal region was relatively slow (Fig. 6A, supplemental Video S5). Quantifying a total of 30 Rab5a-paGFP or 40 Rab5a-T7A-paGFP-expressing cells from 3 independent experiments showed that the dispersal rate of Rab5a in phospho-deficient mutant cells was significantly slower than wild-type Rab5a-paGFP-expressing cells in the first 100 s (Fig. 6, B and C). When the dynamics of the photoacti-vated population of Rab5a-paGFP was tracked for a longer period of time, fluorescence was rapidly dimmed to background levels (Fig. 6C), and this was also accompanied by the appearance of budding vesicles at the leading edge of the polarized cell. This observation was in accordance with a previous study suggesting the half-life of Rab5a-positive early endosomes to be in the range of 2 min (47). In a parallel experiment, we also observed that a majority (ϳ80%) of Rab5a-paGFP-expressing vesicles continuously redistributed from large structures in the cell centrosomal region, to allow the recycling of Rab5a onto budding endosomes at the leading edge (Fig. 7A, Supplemental Video S6). On the other hand, although the pool of Rab5a-T7A-associated endosomes gradually lost its fluorescence signal (Fig. 7A, supplemental Video S7), the number of cells with emerging and budding vesicles at the leading edge was significantly reduced to 33 Ϯ 7% (Fig. 7B).
Rab5aT7A Disrupts Actin Rearrangement during Migration-By means of time-lapse microscopy, we observed that T-cells co-expressing Rab5a-GFP and Lifeact-ruby, a 17-amino acid peptide that stains filamentous actin networks (48), displayed highly motile lamellipodia structures extending in a vibrant umbrella-shaped fast protrusion followed almost immediately by active ruffling and retraction in a highly dynamic cycle in the course of moving forward (supplemental Video S8). We investigated if loss of Thr-7 phosphorylation of Rab5a could interfere with actin remodeling at lamellipodia during migration. For this purpose, HuT-78 T-cells expressing wild-type Rab5a, Rab5a-T7A, or Rab5a-T7E were incubated on a cross-linked anti-LFA-1 antibody for 2 h and fixed. These cells were then stained with phalloidin-TRITC, which binds filamentous actin, and Hoechst, which stains the nucleus (Fig. 8A). Looking closely at the leading edge, we observed that in 25% of cells expressing Rab5a-T7A, filamentous actin bundles were disrupted and condensed around the plasma membrane accompanied by asymmetrically shaped lamellipodia and/or irregular ruffling (Fig. 8B, supplemental Video S9). Such disrupted lamellipodia phenotypes were present only in 12% of the wild-type Rab5a or 5% of the Rab5a-T7E-expressing and -migrating T-cells (Fig. 8B). As migrating T-cells rapidly change shape and the lamellipodia are highly dynamic, it was difficult to compare and quantify cell-to-cell differences. Therefore, these measurements reflect gross differences observed at the lamellipodia of only those crawling T-cells that displayed typical amoeboid phenotypes.
We employed objective quantification of cytoskeletal change using a previously reported cell-based method of high content analysis that entailed population analysis of the morphology of cells expressing GFP fusion protein (34). Utilizing this approach, we observed that although cells expressing the Rab5a-T7A-GFP mutant were polarized after anti-LFA-1 stimulation, the cell polarization index based on the actin cytoskeleton was significantly reduced (p ϭ 0.003) as compared with cells expressing wild-type Rab5a-GFP or Rab5a-T7E-GFP (Fig.  8C). These results suggest that within the population of Rab5a-T7A-GFP-expressing T-cells, a relatively higher proportion of cells with less elongated or irregularly shaped cytoskeleton is present, implying that they have deficiencies in their ability to reorganize their cytoskeleton in response to a migratory stimulus.
A number of previous studies have established Rac1 as the main driving force in the formation of actin protrusions that are essential for cell migration (47,48). A crucial role of Rab5a in the activation of Rac1 has also been demonstrated (17,18). T-cells expressing GFP conjugated wild-type Rab5a (Rab5a-WT) or mutant Rab5a-T7A, Rab5a-S84A, or Rab5a-S123A after stimulation by incubating on an anti-LFA-1 coated surface for 1 h. B, HuT-78 cells were transfected with plasmid constructs expressing GFP conjugated wild-type Rab5a (Rab5a-WT), mutant Rab5a-T7A, or Rab5a-T7E and stimulated via LFA-1 as above. Cells were treated with AlexaFluorா 568-labeled transferrin (red) to allow internalization for 10 min and imaged. At least 20 microscopic fields per slide prepared from three independent experiments were visualized, and a representative data is shown. Scale bar, 5 m.
These findings prompted us to investigate if Thr-7 phosphorylation of Rab5a is important for Rac1 activation in migrating T-cells. To address this question, we transfected HuT-78 T-cells with plasmid constructs encoding wild-type, T7A, or T7E forms of Rab5a, and then cells were incubated on ICAM-1 to trigger LFA-1-induced migration. The levels of Rac1 activation in these cells were determined by a PBD binding assay. Consistent with the above observed role of Rab5a in actin remodeling and T-cell polarity, significantly high Rac1 activity (ϳ2-fold) was detected in T-cells expressing Rab5a-WT or Rab5a-T7E after LFA-1 stimulation in this migration-triggering model system (Fig. 8D). In contrast, expression of Rab5a-T7A perturbed this increase in Rac1 activation after LFA-1/ ICAM-1 stimulation (Fig. 8D). Similar results were observed when these cells were stimulated with SDF-1␣ (data not shown). Taken together, these results clearly demonstrate that PKC⑀-mediated Thr-7 phosphorylation of Rab5a is a necessary step in the regulation of T-cell cytoskeletal reorganization and migration.

DISCUSSION
In this study we establish a novel mechanism involving a PKC⑀-Rab5a-Rac1 axis in the regulation of actin cytoskeletal remodeling and T-cell migration. Specifically, we uncover a key biological function of a previously undescribed phosphorylation site on Rab5a at Thr-7. We show that Rab5a interacts with PKC⑀ in migrating T-cells in a nucleotide-independent manner and that Rab5a is phosphorylated on Thr-7 by PKC⑀ in response to integrin or chemokine stimulation. We also demonstrate that phosphorylation of this site on Rab5a is required for T-cell migration, as ectopic expression of a phospho-deficient T7A mutant of Rab5a resulted in a significant reduction in the ability of T-cells to migrate across ICAM-1-coated membranes toward SDF-1␣. This defect in T-cell motility appears to be due to inefficient trafficking of endosomes toward the leading edge of the cell and correlates with impaired Rac1 activation and actin cytoskeletal remodeling in the lamellipodia, events that are crucial for cell migration.
Rab5 exists as three closely related molecules (Rab5a, Rab5b, and Rab5c) in mammalian cells (40,49), sharing 88.7% of protein sequence identity (supplemental Fig. S2). Although there is evidence to suggest that Rab5a is involved in chemokine receptor recycling, cell motility, actin organization, Rac distribution, and lamellipodia formation in various cell types (14,16,17,50,51), its role in the process of T-cell migration has not previously been reported. During migration, the leading edge of the T-cell actively produces filamentous actin-rich lamellipodia that likely facilitate the interaction of LFA-1 with its counter-ligand ICAM-1 on endothelial cells (52). In doing so, T-cells search for a suitable extravasation site on the venular surface that facilitates rapid directional changes that are characteristic of migrat- ing T-cells. The LFA-1 signal in T-cells has been shown to transiently activate Rac1 and induce actin polymerization (53). Here, we found that signaling through LFA-1 or CXCR4 results in Thr-7 phosphorylation of Rab5a, which in turn is critical for Rac1 activation, subsequent actin remodeling, and T-cell motility. This is consistent with previous findings where Rab5 was shown to regulate trafficking of Rac and its guanine nucleotide exchange factor in T-cell lymphoma invasion and metastasis 1 (Tiam1) (16,17,54). Whether Tiam1 or other guanine nucleotide exchange factors regulate Rac1 activation downstream of Rab5a or PKC⑀ observed in this study requires further investigation.
PKC isoforms have been implicated in T-cell signal transduction including the regulation of migration. For example, our group has demonstrated a role for PKC␤ and PKC␦ in regulating the microtubule cytoskeleton and active T-cell locomotion induced by LFA-1 integrin or CD44 receptors (21,55,56). PKC␦ has also been reported to induce integrin-mediated Thr-566 phosphorylation of phospholipase D and direct cell spreading and migration in COS-7 cells (57). In addition, several studies have suggested the involvement of PKC⑀ in cell migration in many other model systems. Overexpression of PKC⑀ resulted in a highly motile and invasive phenotype in various cancer models (58,59), and its disruption caused inactivation of Rho family GTPases (60). It has also been shown that PKC⑀ can bind directly to both filamentous and globular forms of actin via its actin binding motif (61,62) or interact indirectly via integrins and scaffolding proteins such as receptors for activated protein kinase C (also known as RACK) (63,64). In murine embryonic fibroblasts, PKC⑀ was shown to phosphorylate vimentin and regulate ␤1 integrin recycling contributing to cell motility (24,25). Thus, our identification of Rab5a phosphorylation on the N terminus at Thr-7 represents a novel regulatory mechanism by which PKC⑀ integrates extracellular signals via integrins and chemokine receptors and coordinates signal transduction in T-cell migration. Of note, our studies do not exclude the possibility that PKC⑀ may also phosphorylate other Ser/Thr residues on Rab5a that could contribute to diverse lymphocyte functions.
The GTP/GDP binding switch regions of Rab proteins are the main features for GTPase activity and are the most conserved structures among family members (38,39,65). On the other hand, the N-and C-terminal regions of Rab proteins are variable with the most divergent sequences lying at the extreme N terminus (40,49). Although the isoprenylated C-terminal motif is required for membrane targeting (66,67), functions of the extreme N-terminal region are still unknown. Given that deletion of the first 14 -22 amino acid residues at the N-terminal domain alone is enough to disrupt Rab5a functionalities (42), the N-terminal region is not only an important element for Rab5a function; it might also act as the domain that specifies isoform functionality. Taking it a step further, we have shown that when phosphorylation of Rab5a on the N-terminal Thr-7 site is disrupted, T-cell migration through ICAM-1-coated membranes toward SDF-1␣ is significantly repressed. By replacing Thr-7 with glutamic acid (T7E), which mimics constitutive phosphorylation, we were able to show that this mutant also partially inhibits T-cell migration. Although we do not have a clear explanation for this observation, it is possible that Rab5a dynamically cycles between phosphorylated and dephosphorylated states to function in T-cell migration. This is consistent with the observed dynamic interactions between PKC⑀ and Rab5a (Fig. 1E, supplemental Video S1). It may be that replacement of Thr-7 with T7E could alter the structure or folding of the Rab5a protein necessary for its function in T-cell migration. Future studies on the kinetics of Rab5a recycling in the context of phosphorylation events in T-cells will further clarify this.
The level of Rab5 dynamically fluctuates on individual early endosomes linked by fusion and fission events into a network in time. Although the bulk of early endosomes are not subjected to rapid turnover, the half-life of Rab5-containing early endosomes has been estimated to be in the range of 2 min (46). Moreover, the amount of Rab5 associated with a given endosome varies considerably over time (46). Utilizing a photoactivation-based imaging approach to track the dynamics of Rab5a after LFA-1 stimulation in motile T-cells, we observed that Rab5a-paGFP-expressing vesicles continuously redistribute from the cell centrosomal region (the site of PKC⑀-Rab5a interaction) to the leading edge. However, the dispersal rate of the phospho-deficient mutant Rab5a-T7A-paGFP was significantly slower than the wild-type Rab5a-paGFP in the first 100 s, FIGURE 8. Involvement of Rab5a Thr-7 phosphorylation in the actin cytoskeleton rearrangement at the lamellipodia and Rac1 activation in crawling T-cells. A, HuT-78 T-cells expressing wild-type Rab5a (Rab5a-WT), Rab5a-T7A, or Rab5a-T7E were incubated on anti-LFA-1 coated coverslips for 2 h, stained with phalloidin-TRITC (red) and Hoechst (blue), and imaged. Scale bars, 5 m. B, percentage of cells showing disrupted lamellipodia (as shown in A) were quantified and presented. Phenotypes were scored blindly by three individuals, and data represent the mean Ϯ S.E., n Ͼ 100 cells for each samples. C, cells were seeded on poly-L-lysine (PLL)-or anti-LFA-1 (LFA-1)-coated 96-well plates for 2 h and stained for F-actin and nucleus. Images were automatically acquired and analyzed using an InCell Analyzer TM by selecting only GFP fusion-expressing cells. Results are representative of at least three independent experiments and are presented as the mean Ϯ S.E. D, serum-starved cells expressing Rab5a-WT, Rab5a-T7A, or Rab5T7E were stimulated with or without ICAM-1 for 30 min. Rac1 activation was measured by PAK-1 PBD pulldown assays. The graph shows densitometric analysis of Rac-GTP levels with respect to input levels of total Rac1 for 3 independent experiments. *, p Ͻ 0.01. and fluorescence dimmed to almost background levels by 250 s. Thus, the dynamic turnover of Rab5a seems to be essential to allow its recycling onto budding endosomes in the cell periphery. In support of this concept, earlier studies suggest that the driving force for cell migration is a polarized endocytic/exocytic cycle that delivers adhesion receptors and proteins from the rear of the cell after focal adhesion disassembly to the leading edge of the migrating cell for extension and substrate attachment. Such events are required for driving the cell forward (68,69). This could also be due to extraction of Rab5a from the membrane via guanine nucleotide dissociation inhibitors (15,70,71). Based on these observations, it can be speculated that Thr-7 phosphorylation may be important in the regulation of GDI-mediated extraction of Rab5a from the endosomal membrane.
The recycling of integrins, which involves their endocytosis from cell surface and recycling back to the plasma membrane, is important for their activities as well as for the regulation of their surface levels (72)(73)(74). Previous studies have shown clathrindependent or -independent endocytosis of integrins at the leading edge of migrating cells (75). In neutrophils, LFA-1 is associated with detergent-resistant membrane microdomains and is internalized in a filipin-sensitive and dynamin-dependent manner (76). Consistent with these findings, we observed that intracellular vesicles expressing Rab5a-T7A-paGFP were less efficient in redistributing from the centrosomal region to the leading edge. Another Rab5 GTPase family member, Rab5c, has recently been reported to enhance ␤1-integrin recycling in EGF-induced cancer invasion (74). In addition, LFA-1 recycling has been shown to be independently regulated by Rap2 GTPase as well as G␣ q/11 proteins in T-cells (72,73). Surprisingly, internalization of LFA-1 was not found to be inhibited by a dominant-negative Rab5 construct (N133I) in a Chinese hamster ovary cell model (76). What becomes obvious from all of these studies is that different intracellular pools exist for Rab5a, and its Thr-7 phosphorylation and redistribution may be an important mechanism in the process of cytoskeletal remodeling and lymphocyte migration. It should be noted that although we examined the role of Rab5a mutants in T-cells that still express endogenous Rab5a, it would be interesting to examine these mutants in cells where endogenous Rab5a has been depleted. However, performing such experiments is technically very challenging in T-cells.
In summary, this study provides a new mechanistic insight into the regulation of T-cell migration by a PKC⑀-Rab5a-Rac1 axis. PKC⑀-dependent phosphorylation of Rab5a on Thr-7 regulates its intracellular dynamics and Rac1 activation, which in turn controls actin cytoskeletal remodeling at the lamellipodia and cell migration (Fig. 9). Based on our data, we propose that the Thr-7 phosphorylation of Rab5a might be a missing piece of the structure-function puzzle of this 24-kDa protein and hence may provide new information regarding downstream pathways. Given that Rab5a is central to many fundamental cell processes, this new finding has profound biological implications across cellular organisms. Our results also highlight the concept that LFA-1 not only acts as an adhesion receptor, a role for which it has been known for many years, but also plays an important role in downstream protein-protein interactions and consequent signaling.