Dok-4 Is a Novel Negative Regulator of T Cell Activation¹

Adaptor molecules are proteins containing modular binding domains that are devoid of intrinsic enzymatic activity, but that serve as essential scaffolds for the formation of multimolecular signaling complexes in a variety of cell lineages, notably in T cells (1). Dok⁶ (downstream of tyrosine kinase) (6) proteins are a growing family of adaptor molecules, containing an amino-terminal pleckstrin homology (PH) domain, a central phosphotyrosine-binding (PTB) domain, and a carboxyl-terminal part with several tyrosine phosphorylation sites leading to interactions with SH2 domains. The roles of the first two identified Dok family members, Dok-1 and Dok-2, have been assessed in the T cell lineage (2). These proteins are tyrosine phosphorylated upon TCR triggering or the engagement of the costimulatory molecules (3, 4, 5). They are potent inhibitors of T cell activation via their binding to RasGAP, which regulates Ras activation, but also probably via their association with Src and Tec protein tyrosine kinase (PTK) family members (6, 7).

Overexpression studies involving Dok-1 or Dok-2 molecules provided evidence that these adaptor proteins inhibit phospholipase C (PLC)γ1 and Ras-ERK/MAPK signals and can prevent IL-2 promoter activation (4, 7). Knockdown of Dok-1 and Dok-2 in human primary T cells reveals their negative role on T cell functions (3). Upon TCR triggering, T cells from mice lacking both Dok-1 and Dok-2 displayed increased IL-2 production, as well as proliferation (8).

The additional five members of this Dok-related family have been identified and named from Dok-3 to Dok-7 (9, 10, 11, 12, 13, 14, 15, 16). Among these members, only the gene encoding for Dok-4 is expressed in resting T cells (14). Based on analysis of phylogenetic trees and exon/intron structure of Dok family members, Dok-4 is part of a group that is distinct from that including Dok-1 and Dok-2. In nonimmune cells, Dok-4 appears to be a substrate of c-Ret or insulin receptor tyrosine kinase and Src protein tyrosine kinase family members (11, 12). Functional studies of Dok-4 have yielded conflicting results, with two reports claiming that it enhanced Ret or insulin receptor-mediated ERK activation (11, 12) and one other report finding that it inhibited Ret and Src PTK-mediated activation of the ERK substrate, Elk-1 (13). These apparent discrepancies may relate to differences in cell types, stimuli, and readouts used in the various studies. In Ret signaling, Dok-4 leads to activation of the small G protein Rap1, which induces a sustained activation of ERK in nerve cells (17). Therefore, it would be useful to extend the studies of Dok-4 to other cell types that expressed it, including T cells, where its function remains unknown. Here, we investigated if Dok-4 could be a substrate of the PTKs involved during T cell activation. We report that Dok-4 accumulates in an intracellular vesicular compartment that is recruited to the polarized MTOC (microtubule organizing center) during T cell activation and that it requires its PH domain to adopt this particular localization. The PH domain was also necessary for Dok-4 ability to inhibit ERK activation and IL-2 promoter activity in T cells.

Jurkat JA16 T cell subclone and Raji B cell lymphoma were grown in RPMI 1640 medium supplemented with 10% FCS, 2 mM l-glutamine, and 1 mM sodium pyruvate. L cells expressing human B7.1 (B7.1 L cells) were cultured in DMEM supplemented with 10% FCS, 2 mM l-glutamine, and 1 mM sodium pyruvate (18). Hut-78 T cells were grown in RPMI 1640 medium supplemented with 10% FCS (3).

PBMC from healthy donors (Etablissement Français du Sang, Marseille, France) were isolated on Ficoll-Hypaque (Pharmacia) gradients before purification of the CD4⁺ T cell subset using the naive CD4⁺ T cell isolation kit (Miltenyi Biotec).

The construct βDNA4HADok4 (Dok-4), corresponding to the wild-type full-length human Dok-4 cDNA tagged with HA epitope in the 5′ end, was generated by subcloning of pGEM-T easy vector containing HA tagged Dok-4, described previously (14), into the βDNA4 vector, using the restriction site NotI. The βDNA4vector corresponds to the pCDNA3 vector (Invitrogen) where its promoter region has been replaced by the promoter region of chicken β-actin (7). Expression plasmid βDNA4HA-ΔPHDok4 (ΔPHDok4), containing the human Dok-4 cDNA tagged with HA epitope in the 5′ end lacking the PH domain (deletion of aa 1–119), was generated by PCR amplification of βDNA4HADok4 using the following primers: sense primer 5′-atgtacccatacgacgtcccagactacgctagtctgggagaacctgacctc-3′ (containing HA epitope), antisense primer 5′-tcactgggatggggtcttggcctcac-3′.

GFP fusion protein expression vectors Dok4-GFP and ΔPHDok4-GFP were obtained by removing HA epitope and stop codon from βDNA4HADok4 and βDNA4HA-ΔPHDok4, respectively, and introducing cDNAs into the BamHI/KpnI sites of the pEGFP-N1 vector (Clontech).

The PH domain-tagged GFP (PHDok4-GFP) (aa 1–122) was amplified from βDNA4HADok4 with the following primers: sense primer 5′-ccgaattccatggcgaccaatttcagtg-3′ and antisense primer 5′-cggtggatcccgtcccagactgatgtcgttga-3′. The PCR product was subcloned into BamHI/KpnI sites of pEGFP-N1 expression vector. The expression vector Dok4PTB*-GFP encodes for a Dok4-GFP fusion protein containing point mutations into the PTB domain of Dok-4, corresponding to the substitution of arginine 185 and 186 in alanine (R185/186A). These mutations were previously described for Dok-1 (19). They were obtained using the QuickChange system (Stratagene) as from the Dok4-GFP expression vector, with the following primers: sense 5′-ccctctgctcactggccgcctatggccggg-3′ and antisense 5′-gcatcccggccataggcggccagtgagcag-3′ (corresponding mutations R185/186A are underlined). The expression vector Dok1-GFP encoding for a Dok1-GFP fusion protein was generated by PCR amplification from βDNA4HADok1 using the following primers: sense 5′-agatctcgagctatggacggggctgtgatggag-3′ and anti-sense 5′-ggtcaagtctgagggttccaccgtcgacggtaccgcg-3′. The PCR products were sequenced and cloned in pEGFP-N1 XhoI/KpnI.

The promoter assay plasmids pIL-2-Luc composed of IL-2 promoter, fused to firefly luciferase reporter gene, and pβ-actin-Rluc composed of β-actin promoter, fused with Renilla luciferase gene, were previously reported (20).

For small interfering RNA (siRNA), the pH1shDNA plasmids were derived from the pH1-XhoI plasmid (a descendant of pBlueScript KS+) and contained a XhoI-flanked fragment containing the human H1 promoter amplified by PCR from human blood mononuclear cell genomic DNA, the template small hairpin DNA (shDNA) sequences encoding siRNAs. The hairpins contained the 19-nt sense sequence of the target transcript, which was separated by a 9-nt loop from the 19-nt antisense sequence of the target mRNA and followed by five thymidines as a termination signal, as previously described (21). All constructs were verified by sequence analysis. Two sequences for the hairpin RNA were shown: 5′-ccccagacagatcgcttcaatgttcaagagacattgaagcgatctgtctgtttttggaaa-3′ (19 nt corresponding to bp 403–421), which reduces the expression of Dok-4 (>50% in immunoblot analysis, data not shown) and corresponds to pBChH1-RNAiDok4; and the second sequence 5′-ccccattactcgtatccctgcattcaagagatgcagggatacgagtaatgtttttggaaa-3′ (19 nt corresponding to bp 760- 778), which does not reduce the expression of Dok-4. This latter plasmid was used as a control in the RNA interference (RNAi) experiments (pBChH1-Control).

CD3 mAbs 289, OKT3, and CD28 mAb 248 have been already reported (18). Polyclonal anti-Dok1 Abs have been described previously (20). Anti-Dok2 mAb was purchased from BD Transduction Laboratories. Polyclonal anti-Dok4 Abs used in Western blot experiments were purchased from Abgent or described previously (13). Polyclonal anti-Dok4 Abs used in immunoprecipitation experiments were described previously (13). Anti-phosphotyrosine (PY) 4G10 mAb was purchased from Millipore. Polyclonal anti-GFP Abs and anti-α-tubulin mAb were purchased from Abcam. Anti-γ-tubulin mAb and anti-Rap1 Abs were purchased from Santa Cruz Biotechnology. Anti-phospho-ERK, anti-ERK, anti-phospho-PLCγ1, anti-PLCγ1, anti-phospho-JNK, anti-phospho-p38 polyclonal Abs were purchased from Cell Signaling Technology. Superantigen Staphylococcus enterotoxin E (SEE), CellTracker Orange CMTMR (5-(and-6)-(((4-chloromethyl)-benzoyl-amino)tetramethyl-rhodamine), and ionomycin were purchased, respectively, from Toxin Technology, Molecular Probes, and Calbiochem (VWR International). PMA and poly-l-lysine were purchased from Sigma-Aldrich.

To distinguish Raji B cells from Jurkat T cells, Raji cells were preincubated in RPMI 1640 with 10% FCS containing 10 μM CellTracker Orange CMTMR for 30 min at 37°C, washed, and resuspended (5 × 10⁶ cells/ml) in RPMI 1640 with 50 mM HEPES, as indicated. Raji cells were then incubated for 20 min with or without 5 μg/ml SEE. Transfected Jurkat cells were mixed at a 2:1 ratio with Raji cells pulsed with or without SEE, and then incubated at 37°C for 45 min. After stimulation, the cells were deposed onto poly-l-lysine-coated coverslips, let sediment for 3 min, and then centrifuged at 300 rpm for 1 min. The conjugates were fixed for 5 min in methanol. As indicated, immunofluorescence staining was performed. Cells were permeabilized in PBS-0.1% Triton X-100 for 10 min and saturated in PBS-5% BSA for 20 min. The staining with the appropriate Abs (at the dilution 1/500 in PBS-5% BSA) was performed for 20 min, using goat anti-mouse Alexa 594 as secondary Ab (Molecular Probes). Slides were mounted with fluorescent mounting medium (Dako). Images were taken and processed using a confocal microscope (Leica TCS NT confocal microscope).

Jurkat cells (10 × 10⁶) were stimulated at 37°C in RPMI 1640 with 50 mM HEPES. Stimulations were conducted for the indicated times with Raji cells (ratio 2:1) pulsed or not for 20 min at 37°C with SEE at 5 μg/ml, CD3 mAb 289 at 10 μg/ml, CD28 mAb CD28.2 at 10 μg/ml, or CD28 ligands expressed at the surface of murine L cells (L-B7.1, ratio 2:1). The cells were pelleted in a microcentrifuge and lysed in buffer containing 50 mM HEPES (pH 7.4), 1% Nonidet P-40, 150 mM NaCl, 20 mM NaF, 20 mM iodoacetamide, 1 mM PMSF, 1 μg/ml protease inhibitors (protease inhibitors cocktail; Sigma-Aldrich), and 1 mM Na₃VO₄ for 10 min at 4°C, then centrifuged at 13,000 rpm for 10 min at 4°C.

For fractionation experiments, the pellets were lysed using the ProteoExtract subcellular proteome extraction kit (Calbiochem/VWR International), according to the manufacturer’s instructions. The efficiency of subcellular fractionation from the Jurkat T cell lysates has been investigated by SDS-PAGE and immunoblotting of selected marker proteins. The CD28 molecule was exclusively found in the membrane fraction, the heat shock protein-90 was detected in the cytosol and nucleus fractions, the myc protein was detected in the nucleus fraction, and cytoskeleton compounds such as actin, vimentin, and α/γ-tubulin were present in both cytoskeleton and cytosol fractions (data not shown).

For immunoprecipitation, lysates were clarified and incubated with purified anti-GFP or anti-Dok4 polyclonal Abs coupled to protein A-Sepharose for 2 h at 4°C. Immunoprecipitates were washed twice in 1 ml of lysis buffer and then boiled in reducing SDS gel sample buffer for 5 min. Samples were resolved by standard 10% SDS-polyacrylamide gels.

For immunoblotting, membranes were blocked and probed with specific Abs. Blots were then incubated with the appropriate secondary Abs, anti-rabbit IgG or anti-mouse IgG (Amersham Biosciences), both conjugated with HRP. Immunoreactive bands were visualized by ECL (Amersham Biosciences). Protein bands on Western blots were quantified with the ImageJ software.

Rap activity was determined as described previously using a GST-RalGDS-RBD fusion protein (22). Briefly, Jurkat cells (10 × 10⁶) were stimulated as indicated with 10 μg/ml CD3 mAb (clone 289), lysed in standard Nonidet P-40 buffer, and clarified by centrifugation at maximal speed in an Eppendorf centrifuge for 10 min at 4°C. Two micrograms of purified GST-RalGDS-RBD coupled to glutathione-Sepharose beads was added to the supernatant and incubated at 4°C for 60 min. Beads were washed four times in lysis buffer and denaturated with SDS.

Jurkat cells (10 × 10⁶) were electroporated at 960 μF and 250 V using a Bio-Rad Gene Pulser with 12 μg of pIL-2-Luc or NF-AT-Luc plasmid, 1 μg of pβ-actin-Rluc, and 15 μg of the other expression plasmids. Stimulations with APC were conducted for 8 h with Raji cells (ratio 1:1) pulsed for 20 min at 37°C with SEE at 10 ng/ml. Stimulation with Abs were conducted for 16 h, using CD28 mAb (clone 248; ascites fluid at a 1/400 dilution) and CD3 mAb (clone 289; ascites fluid at a 1/4000 dilution). Stimulations with drugs were conducted for 16 h, using PMA (50 ng/ml) and ionomycin (2 μg/ml). Following cell lysis, proteins were quantified by Bradford reagent (Bio-Rad), and 30 μl of cell lysates were then subjected to dual luciferase reporter assay according to the manufacturer’s instructions (Promega). Results were corrected by the activity of firefly luciferase standardized by that of Renilla luciferase and quantification of proteins in lysates.

CD4⁺ PBL-T cells were transfected with the Amaxa Nucleofector technology according to the manufacturer’s instructions and were used 18 h after transfection. CD4⁺ T cells (10⁵ cells/well) were plated in Cellstar U-bottom 96-well culture plates (Greiner Bio-One) and left unstimulated or stimulated with CD3 mAb (OKT3 at 0.5 μg/ml) plus CD28 mAb (CD28.2 at 0.5 μg/ml) or PMA at 1 ng/ml plus ionomycin at 1 μg/ml. After 24 h of culture, T cells were incubated with PE-conjugated CD69 mAb from BD Biosciences. The detection of CD69 expression was done by flow cytometry. After 72 h of culture, wells were pulsed with 1 μCi of [³H]thymidine (Amersham) for the remaining 24 h and then harvested onto glass-fiber filters. Thymidine incorporation was measured in a direct beta counter (Matrix 9600; Packard Instruments).

We have previously reported DOK-4 gene expression in resting T cells (14). An anti-Dok4 polyclonal antiserum (see Materials and Methods) was produced to determine the levels of Dok-4 protein in cells including T cells. COS cells transiently transfected or not with an expression vector containing the Dok-4 cDNA demonstrated the specificity of this Ab. As shown in Fig. 1,A, strong expression of a 37-kDa protein, consistent with the expected molecular mass of Dok-4, was detected in lysates from Dok-4-transfected compared with control COS cells. The weak signal detected in nontransfected COS-7 cells was thought to correspond to endogenous Dok-4 expression, which had been previously reported in several epithelial cell lines (13). Dok-4 protein was also detected in peripheral blood T cells (PBL-T) and in the Jurkat T cell line, confirming previous mRNA expression data (14). To determine the subcellular localization of Dok-4 in T cells, three different fractions of Jurkat T cell lysates corresponding to cytosol, membranes, or cytoskeleton were analyzed by anti-Dok-4 immunoblotting (Fig. 1,B, top panel). Endogenous Dok-4 was detected only in the cytosol and cytoskeleton fractions and was absent from membranes. This distribution pattern paralleled that of actin, vimentin, and α/γ-tubulin that are present in both cytoskeleton and cytosol fractions (data not shown). Unfortunately, using different polyclonal anti-Dok4 Abs, we were not able to detect the endogenous Dok-4 protein in T cells by confocal microscopy. To determine the intracellular localization of Dok-4, we produced a Dok4-GFP fusion protein, which we expressed in Jurkat T cells. Anti-GFP immunoblots confirmed that the transiently expressed Dok4-GFP was largely contained in the cytosol and in the cytoskeleton fractions like the endogenous protein (Fig. 1,B, top vs bottom panel). Under confocal microscopy, Dok4-GFP was primarily localized in a condensed area of Jurkat T cell cytoplasm (Fig. 1,C). Since Dok-4 contains a PH domain that might be recruited by phosphoinositides, localization experiments were duplicated in a T cell line with a normal PI3K activation, Hut-78. Consistent with the previously reported insensitivity of the Dok-4 PH domain to the activation of PI3K (23), a similar distribution of Dok4-GFP was observed in both Jurkat and Hut-78 T cells (Fig. 1,C). Notably, the level of Dok4-GFP overexpression in these experiments was moderate, as documented by comparison of the 62-kDa Dok4-GFP fusion protein vs the 37-kDa endogenous Dok-4 signal on the same immunoblot (Fig. 1 D). These observations confirm that Dok-4 is expressed in specific compartments in T cells, prompting further investigation into the effect of T cell activation upon Dok-4.

Other members of Dok family such as Dok-1 are substrates of PTKs in T cells (4, 7, 24). In Jurkat T cells, Dok-1 is prominently tyrosine phosphorylated upon engagement of CD28 by its B7-1 (CD80) ligand, but poorly phosphorylated upon TCR stimulation with CD3 mAbs (25). Therefore, we examined the tyrosine phosphorylation state of Dok-4 under TCR or CD28 stimulation. Jurkat cells were stimulated with CD3 mAbs or with L cells expressing B7-1 ligand. Lysates obtained from these cells were subjected to immunoprecipitation with anti-Dok4 (Fig. 2^,A, top panels) or anti-Dok-1 (Fig. 2^,A, bottom panels) Abs followed by anti-phosphotyrosine immunoblotting. As previously described, Dok-1 was more prominently phosphorylated upon CD28 stimulation than TCR stimulation (Fig. 2^,A, bottom panels). In contrast, Dok-4 was tyrosine phosphorylated only when the TCR/CD3 complex was stimulated, although detection of this phosphorylation required long exposure of the immunoblots. TCR-mediated phosphorylation was also detectable in Hut-78 T cells (supplemental Fig. 1).⁷ To determine the phosphorylation state of Dok-4 during TCR-mediated cell-cell contacts, Jurkat T cells expressing Dok4-GFP or Dok1-GFP were activated by SEE-pulsed Raji B cells as APCs. Cell lysates were subjected to immunoprecipitation with anti-GFP Abs followed by anti-phosphotyrosine immunoblotting. Basal tyrosine phosphorylation of both overexpressed Dok-1 and Dok-4 proteins was detectable in unstimulated Jurkat cells (Fig. 2,B). As previously described (25), tyrosine phosphorylation of Dok-1 increased more prominently in Jurkat T cells in the absence than in the presence of a TCR stimulus (Fig. 2,B, bottom panels). In contrast to Dok-1, tyrosine phosphorylation of Dok-4 was most strongly increased upon presentation of SEE by Raji B cells (Fig. 2 B, top panels). Dok-4 phosphorylation was still detectable after 20 min of contact with SEE-pulsed Raji B cells (supplemental Fig. 2). Thus, these data suggest that the Dok-4 protein is phosphorylated upon TCR engagement.

Several intracytoplasmic PTK substrates are recruited to the T cell plasma membrane in the context of a T cell-APC contact (26, 27). The cell contact zone has been termed the “immunological synapse” (IS) (28). To determine whether Dok-4 is recruited to the IS, we analyzed by confocal microscopy cell contacts between Jurkat T cells expressing Dok4-GFP and SEE-pulsed Raji B cells labeled in red with a CellTracker dye (Fig. 3,A). Twenty minutes after contact, Dok-4 protein was located in an intracytoplasmic compartment opposite to the IS. More prolonged contacts (45 min) resulted in the relocalization of Dok-4 in close proximity to the IS. This change in Dok-4 localization did not occur in the absence of Ag, suggesting that it was induced by TCR stimulation. This late distribution pattern evoked similarities with a polarized MTOC, which has been described before in this position (29). To determine whether Dok-4 was located within the MTOC, the centrosomal protein γ-tubulin, a key factor in microtubule nucleation, was used as a marker (30). In support of Dok-4 localization in the MTOC, we found that, in cell fractionation experiments, Dok-4 and γ-tubulin presented the same distribution pattern (Fig. 3,B). To confirm this, confocal analysis was performed in cell conjugates formed between Dok4-GFP-transfected Jurkat T cells and SEE-pulsed Raji B cells. In unstimulated T cells, γ-tubulin was condensed in a cytoplasmic dot-like structure typical of a MTOC, whereas Dok-4 was in a distinct compartment in close proximity to the MTOC (Fig. 3,C, top panels). Upon contact with SEE-pulsed Raji B cells (time of contact, 20–45 min), a polarized MTOC was detected in the vicinity of the IS on both T cell and APC side (Fig. 3,C, middle and bottom panels). During the early phase of cell contact formation, Dok-4 was preferentially found opposite to the IS and MTOC in a membrane or submembrane compartment (Fig. 3,C, middle panels). Later on, Dok-4 relocalized within the MTOC itself (Fig. 3,C, bottom panels). Upon synapse formation, tyrosine phosphorylation of Dok-4 was consistently detectable until Dok-4 joined the polarized MTOC, at which point Dok-4 was no longer phosphorylated (supplemental Fig. 2 and Fig. 3,D). These results indicate that Dok-4 is specifically shuttled during T cell activation, being first excluded from the IS, but then moving back to its vicinity into the polarized MTOC after a phase of transient recruitment at the plasma membrane (Fig. 3, C and D). Nocodazole treatment (a microtubule-disrupting drug) confirmed that Dok-4 migration to the vicinity of the IS was dependent on microtubule network formation (supplemental Fig. 3).

Dok-4 contains two protein-protein interaction domains, the PH and PTB domains (11, 12, 14). Both domains are important for Dok-4 plasma membrane localization in epithelial cells (13). To test the role of these domains on Dok-4 localization in T cells, we generated different mutants of Dok-4 fused to GFP, lacking either the PH domain (ΔPHDok4-GFP) or the PTB and the carboxyl-terminal part (PHDok4-GFP). We also developed a PTB mutant of Dok-4 containing a substitution of critical residues of the PTB domain (Dok4PTB*-GFP); these arginine to alanine mutations are usually able to abolish the binding functions of PTB domains as described for the PTB domain of Dok-1 (19). The localization of the different Dok-4 mutants in unstimulated Jurkat T cells or in contact with SEE-pulsed Raji B cells was determined by confocal microscopy. Deletion of the PH domain (ΔPHDok4-GFP) resulted in a diffuse cytosolic distribution in stimulated or unstimulated Jurkat cells (Fig. 3,D, top panels). The isolated PH domain of Dok-4 (PHDok4-GFP) was localized in a cytosolic and membrane fraction, and it was not recruited to the MTOC during IS formation (Fig. 3,D, middle panels). Point mutations in the PTB domain (Dok4PTB*-GFP) did not alter the distribution of Dok-4 (Fig. 3 D, bottom panels). These results indicate that the PH domain of Dok-4 is necessary for basal localization of Dok-4 and its recruitment in the MTOC compartment during T cell/APC junction.

To determine the role of the Dok-4 molecule in signaling events downstream of TCR stimulation, we analyzed ERK1/2 activation in Jurkat cells upon contact with SEE-pulsed Raji B cells. Different research groups reported that Dok-4 was involved in ERK activation and downstream transcription factors such as Elk-1 (11, 12, 13, 17). However, studies found Dok-4 to be an activator of ERK pathway in Chinese hamster ovary cells or nerve cells (11, 12, 17) and other studies found that Dok-4 inhibited the activation of Elk-1 in Caco-2 epithelial cells (13, 31). To resolve this issue in T cells, we developed an alternative approach to knock down Dok-4 expression using RNA interference (RNAi). As shown in Fig. 4,A, transfection with a plasmid containing a Dok-4 RNAi sequence resulted in almost complete inhibition of the endogenous expression of Dok-4 in Jurkat T cells after 1 day, whereas transfection with an empty vector or a construct containing an inefficient Dok-4 RNAi sequence did not affect Dok-4 expression. Expression of Dok-1 and Dok-2 was not affected in Dok-4 RNAi-treated cells (Fig. 4,A, middle and bottom panels), confirming the specificity of Dok-4 RNAi. In Jurkat cells stimulated by Raji B cells, ERK phosphorylation was detected only upon presentation of SEE (Fig. 4,B, top panels). In Dok-4 RNAi-treated Jurkat cells, TCR engagement resulted in enhanced phosphorylation of ERK as compared with control RNAi-treated cells. In contrast, tyrosine phosphorylation of PLCγ1 was not altered by Dok-4 RNAi (Fig. 4, B and C, bottom panels). The other MAPK pathways involved in T cell activation were also examined and relative protein phosphorylation level was calculated (supplemental Fig. 4). This showed that JNK phosphorylation was mildly altered by Dok-4 RNAi (JNK phosphorylation is decreased at 30 min), whereas surprisingly p38 MAPK phosphorylation was decreased in the presence of down-regulated Dok-4 (Fig. 4 C, center panels).

In complement to the Dok-4 RNAi approach, we examined the effect of Dok-4 overexpression on MAPK pathways in Jurkat T cells. As shown in Fig. 5,A, overexpression of Dok-4 markedly decreased ERK phosphorylation induced by CD3 plus CD28 engagement (relative phosphorylation intensity is shown in supplemental Fig. 5). Consistent with Dok-4 RNAi experiments (Fig. 4,C), Dok-4 overexpression appeared to slightly increase JNK and p38 MAPK phosphorylation under CD3 plus CD28 costimulation (Fig. 5,A and supplemental Fig. 5). In contrast to wild-type Dok-4, overexpression of the PH-deleted mutant of Dok-4 induced an increase of ERK activation and it had little effect in general on activation of the other MAPK pathways (Fig. 5,A, right side and supplemental Fig. 5). Taken together, these results suggest that, in T cells, Dok-4 negatively regulates ERK activation, but positively regulates p38 MAPK activation. Because Dok-4 has been reported to activate the Ras-related GTPase Rap1 in nerve cells (17) and because Rap1 negatively regulates ERK activation in T cells (32), we investigated the impact of Dok-4 overexpression on Rap1 activation in T cells. Rap1 activity assays were performed on cell lysates of Jurkat T cells transfected with wild-type Dok-4 or the PH-deleted mutant of Dok-4 (Fig. 5 B). Rap1 was activated upon CD3 stimulation and this activation was enhanced in the presence of Dok-4 overexpression. In contrast, overexpression of the PH-deleted mutant of Dok-4 blocked Rap1 activation in TCR-activated T cells. Thus, activation of Rap1 by Dok-4 could plausibly account for our observation of Dok-4-mediated inhibition of ERK in T cells.

To further determine the role of Dok-4 on T cell activation, we examined IL-2 promoter activation. We had previously shown that SEE-pulsed Raji B cells induced Jurkat cells to transcribe a reporter luciferase gene under the control of an IL-2 promoter sequence (33). In RNAi-treated Jurkat cells, a decrease of ∼90% in Dok-4 protein expression was achieved (Fig. 6,A, right panels) and was accompanied by increased activity of the IL-2 promoter upon contact with SEE-pulsed Raji B cells (Fig. 6,A, left panel). These results were consistent with the observed effect of Dok-4 RNAi on ERK activation (Fig. 4, B and C). Conversely, overexpressing Dok-4 in Jurkat cells (Fig. 6,B) reduced IL-2 promoter activity induced by CD3 plus CD28 costimulation. However, Dok-4 overexpression had no impact on IL-2 promoter activity induced by a stimulus (PMA plus ionomycin treatment) that bypasses the early steps in TCR signaling. In contrast to wild-type Dok-4, overexpression of the PH-deleted mutant of Dok-4 did not inhibit IL-2 promoter activation (Fig. 6 B, left panel). Anti-HA epitope tag immunoblotting confirmed that similar levels of wild-type Dok-4 (HADok4) and PH-deleted mutant (HAΔPHDok4) were obtained (data not shown). Therefore, deletion of the PH domain abolished the repressive action of Dok-4 on IL-2 promoter activity.

Primary CD4⁺ PBL-T cells isolated from human peripheral blood were also examined. Dok-4 siRNA sequences were introduced in these cells by nucleofection technology, resulting in an ∼60% reduction in Dok-4 expression after 24 h (Fig. 6,C, right panel). Dok-4 inhibition resulted in an increase of CD69 cell surface expression (Fig. 6,C, left panel) and of T cell proliferation (Fig. 6 D) in response to CD3 plus CD28 costimulation. Altogether, results show that Dok-4 is a negative regulator of ERK phosphorylation, IL-2 promoter activation, and proliferation in T cells.

In this report, we demonstrate that a novel Dok family member, Dok-4, is expressed in specific cytoplasmic compartments in T cells, is phosphorylated in response to TCR engagement, and negatively regulates T cell activation.

Having previously shown that Dok-4 mRNA is expressed in resting T cells (14), we have now confirmed that Dok-4 protein is expressed in these cells and that it undergoes tyrosine phosphorylation upon TCR stimulation (Fig. 2 and supplemental Fig. 1). Of note, as compared with phosphorylation of Dok-1 and Dok-2, which is quite robust following T cell stimulation (3), tyrosine phosphorylation of Dok-4 was more subtle and was consistently more difficult to detect. In Dok-1 and Dok-2, most sites of tyrosine phosphorylation appear to be contained within the C-terminal region. In contrast, most tyrosine residues within the C-terminal region of Dok-4 sequence are predicted to be poor substrates for PTKs (data not shown). Indeed, the deletion of this C-terminal region did not alter the tyrosine phosphorylation of Dok-4, suggesting that some tyrosine residues in the PH or PTB domain are accessible to the PTKs (13). Among the PTKs so far shown to phosphorylate Dok-4 (11, 12, 13), only members of the Src family, such as Fyn, are involved in TCR signaling events.

Subcellular localization might be a limiting factor for tyrosine phosphorylation of Dok-4 in T cells. Dok-4 was primarily located at the vicinity of the MTOC, as identified by γ-tubulin staining (Fig. 3,C). Both γ-tubulin and endogenous Dok-4 accumulated in cytoskeletal fractions (Fig. 3,B). Following superantigen presentation, Dok-4 was initially located at the distal pole of the immunological synapse. Later on, Dok-4 accumulated at the polarized MTOC in the vicinity of the immunological synapse (Fig. 3). Dok-4 was also transiently detected at the plasma membrane (Fig. 3 C), which could permit the phosphorylation of Dok-4 by the Src kinase members present in the inner leaflet of the plasma membrane (supplemental Fig. 2). However, the exact identity of the subcellular compartments containing the Dok-4 adaptor molecule remains under investigation.

Following a prolonged cell contact, Dok-4 was located at the polarized MTOC (Fig. 3,C). Several signaling molecules involved in T cell activation, such as protein kinases (like PKCβ, FAK, or Pyk-2), colocalize with the MTOC (34, 35). To address the mechanisms responsible for targeting Dok-4 to these intracellular compartments, we examined the role of its PH domain. PH domains are generally considered as phosphoinositide-binding modules and have often been implicated in recruitment of proteins to the plasma membrane (36). Some molecules like the four-phosphate adaptor proteins 1 and 2 (FAPP1 and FAPP2) accumulate via their PH domains in intracellular compartments such as the trans-Golgi network (37). The PH domain of Dok-4 is important to localize Dok-4 at the plasma membrane of the epithelial cells (13). Here, we show that, in T cells, Dok-4 localizes to a unique intracellular compartment through the action of its PH domain (Fig. 3,D). However, an isolated Dok-4 PH domain alone could not accumulate in the same intracellular compartments, suggesting that cooperation with another region of Dok-4 is required. This is not surprising given evidence that other PH domains require the cooperative action of another domain for subcellular targeting (38, 39, 40). Indeed, Dok-4 itself requires both its PH and PTB domains to localize at the membrane in epithelial cells (13). However, using a point mutant of the PTB domain, we show that the localization of Dok-4 to its unique intracellular compartment in T cells appears to be independent of PTB domain function (Fig. 3 E). Since specific phosphoinositides control the correct location and timing of many trafficking events (41) and since PH domains are generally recognizing a variety of phosphoinositides (42), defining the phosphoinositide-binding specificity of the Dok-4 PH domain will be important to identify its role in mediating subcellular targeting of Dok-4.

Our studies show that, in addition to its role in subcellular targeting, the PH domain is also essential to mediate the inhibitory effect of Dok-4 on T cell activation. It has previously been suggested that Dok-4 regulates the ERK signaling pathway (11, 12, 13), but conflicting results exist as to its stimulatory vs inhibitory impact. Using a Dok-4 RNAi strategy as well as an overexpression approach, we showed that Dok-4 negatively regulates the activation of ERK without altering other key signaling events such as the tyrosine phosphorylation of PLCγ1 or the activation of JNK. How Dok-4 targets the ERK pathway is unclear because, unlike Dok-1/2, Dok-4 does not bind p120RasGAP, an inhibitor of the Ras/ERK pathway (11). However, as had previously been shown in nerve cells stimulated through Ret (17), we found that Dok-4 activates Rap1 downstream of TCR stimulation in T cells (Fig. 5 B). In T cells, Rap1 has been generally implicated in the regulation of cell adhesion (43), but it has also been reported to antagonize ERK activation in anergic T cells, thus contributing to repression of IL-2 transcription in these cells (32, 44). To further dissect the inhibitory mechanisms involved in Dok-4 action, it will be essential to identify the partner molecules for Dok-4 in T cells, which are currently unknown.

Our findings, taken together with previous studies of Dok-1 and Dok-2 (3, 4, 7, 8), demonstrate that the three Dok-related adaptors expressed in resting T cells (Dok-1, Dok-2, and Dok-4) are negative regulators of T cell activation, although they likely perform this function through different mechanisms.

We thank Dr. Daniel Isnardon for providing technical assistance for the confocal microscopy, and Alexandre David, Dr. Frédérique Michel, Dr. Philippe Pierre, and Dr. Francisco Sanchez-Madrid for providing some reagents. We are grateful to Dr. Andrès Alcover, Dr. Yves Collette, and Dr. Francisco Sanchez-Madrid for helpful discussions and to Marie-Claire and Peter Gerhards for the correction of the manuscript.

The authors have no financial conflicts of interest.