The Late Endosomal Transporter CD222 Directs the Spatial Distribution and Activity of Lck Free

T cell signaling molecules have to be tightly controlled spatially and temporally to ensure proper T cell activation without hyperresponsiveness (1). A fundamental step during the onset of T cell signaling is the recruitment of the lymphocyte-specific protein tyrosine kinase (Lck) to the immunological synapse (IS) and the subsequent phosphorylation of the ITAMs of the CD3 ζ-chain (CD3ζ). ZAP70 can thereafter bind to the phosphorylated ITAMs of the TCR complex where it is phosphorylated by Lck to initiate further downstream signaling (2). For fine-tuning of T cell activation active and inactive Lck pools must be kept at equilibrium. It has been shown that Lck activity is not directly linked with TCR engagement (3, 4). Lck rather exists in several conformational and activity states: 1) an unphosphorylated “primed” form; 2) an active form phosphorylated at the tyrosine 394 (pY394); 3) an inactive “closed“ form phosphorylated at the COOH-terminal tyrosine 505 (pY505); and 4) a double phosphorylated form (pY394 + pY505) that is thought to possess an equal kinase activity as single phosphorylated Lck at Y394 (3). Hence, active Lck is constitutively present in resting T cells, and it is its coordinated intracellular distribution that is crucial to balance T cell activity and nonreactivity (5–7). The antagonistic activities of the phosphatase CD45 and the kinase Csk control Lck’s mode of action (2, 8, 9). CD45 plays a pivotal role in the activation of Lck via dephosphorylation of Y505 (10, 11), but it is not clear where the interaction takes place and how it is regulated in space and time.

The late endosomal transmembrane molecule CD222, also known as the cation-independent mannose 6-phosphate/insulin-like growth factor 2 receptor, is a multifunctional broadly expressed regulator of protein trafficking (12). It binds mannose 6-phoshate–bearing proteins including acid hydrolases and TGF-β (13), the insulin-like growth factor 2 (14) and plasminogen (15). On delivering its cargo, CD222 traffics between the trans-Golgi network (TGN), endosomes, and the plasma membrane (16). CD222 is barely expressed on the surface of resting T cell but has been reported to be upregulated at the plasma membrane upon T cell activation in rodents (17). In this study, we show that CD222 is also upregulated on the surface of human T cells upon stimulation. Remarkably, we found that CD222 transports Lck intracellularly and directs its distribution to CD45 at the plasma membrane. The loss of CD222 causes a decrease in signal transduction and subsequently a profound blockade of T cell effector functions.

The mAb to CD222, unlabeled and conjugated with AF647 (MEM-238), CD45, both unlabeled (MEM-28) and conjugated with Pacific Orange (HI30), and CD4, unlabeled (MEM-241), were purchased from EXBIO (Prague, Czech Republic); the mAb to CD3 (MEM-57), CD59 (MEM-43/5), pTag (H902), and α-fetoprotein (AFP-12) were provided by Dr. V. Hořejší (Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Prague, Czech Republic). The anti–TCRβ-chain mAb C305 was a gift from Dr. A. Weiss (University of California, San Francisco, San Francisco, CA). The stimulatory CD3 mAb OKT3 was from Ortho Pharmaceuticals (Raritan, NJ). The CD28 mAb Leu-28, anti–phospho-CD3ζ/CD247 (Y142) mAb K25-407.69, and anti–phospho-linker of activated T cells (LAT; Y171) were from BD Biosciences (San Jose, CA). The HRP-conjugated anti-phosphotyrosine mAb 4G10 was from Merck Millipore (Billerica, MA). The anti-GAPDH mAb 14C10, anti–phospho-ZAP70 Ab (Y319)/Syk (Y352), anti-ZAP70 mAb 99F2 and L1E5, anti–phospho-p44/42 MAPK Ab ERK1/2 (Y202/Y204), anti-p44/42 MAPK Ab ERK1/2, anti–phospho-Src mAb (Y416), anti–non-phospho-Src mAb (Y416), anti–phospho-Lck Ab (Y505), anti-Lck mAb 73A5, and the rabbit mAb to early endosome Ag 1 (EEA1) C45B10 were from Cell Signaling Technology (Danvers, MA). The anti-Lck mAb H-95 was from Santa Cruz Biotechnology (Santa Cruz, CA). Biotinylation of unconjugated mAb was performed in our laboratory. As secondary reagents we used HRP-conjugated goat anti-mouse and anti-rabbit IgG (Sigma-Aldrich, St. Louis, MO), streptavidin-HRP conjugate (GE Healthcare, Buckinghamshire, U.K.), and goat anti-mouse IgG+IgM (H+L)-FITC conjugate (An der Grub, Kaumberg, Austria).

The T cell line Jurkat E6.1 and the B cell line Raji were from the American Type Culture Collection. The Lck-deficient Jurkat T cell line JCaM 1.6 (18) was from the European Collection of Cell Culture (Salisbury, U.K.). The immortalized CD222 negative mouse fibroblasts were provided by E. Wagner (Spanish National Cancer Research Centre, Madrid, Spain). The Jurkat IL-2 Luc T cells were stably transfected with a luciferase expressing IL-2 promoter reporter construct as described previously (19). The Jurkat NFAT Luc cells containing an artificial promoter region with four NFAT binding sites were created in our laboratory. Human PBMC were isolated from purchased buffy coats of healthy adult volunteers (Rotes Kreuz, Vienna, Austria) by density gradient centrifugation on Lymphoprep (Nycomed). All primary human T cells as well as mouse and human T cell lines were maintained in RPMI 1640 medium, the HEK-293 cells in DMEM, all supplemented with 10% heat-inactivated FCS (Sigma-Aldrich), 2 mmol/L l-glutamine (Life Technologies, Carlsbad, CA), 100 μg/ml penicillin (Life Technologies), and 100 μg/ml streptomycin (Life Technologies). Cells were grown in a humidified atmosphere at 37°C and 5% CO₂ and passaged every 2–3 d.

Primary human T cells isolated by CD14-depletion from PBMC were cultured for 2 d to separate non- from adherent cells. T cell lines and primary human T cells (5 × 10⁴) were seeded in 96-well plates and kept unstimulated or stimulated by: 1) plate-bound CD3 mAb OKT3 (1 μg/ml) alone or together with soluble CD28 mAb Leu-28 (0.5 μg/ml); 2) CD3 Ab-coupled beads; 3) staphylococcal enterotoxin E (SEE)–loaded Raji B cells; or 4) PMA (5 ng/ml) plus ionomycin (1 μmol/l). After indicated time intervals, the cells were harvested and analyzed via flow cytometry, immunoblotting, microscopy, or functional cellular assays. To measure proliferation, before stimulation, the cells were resuspended in PBS, incubated with CFSE (0.5 μmol/l; Molecular Probes, Eugene, OR) at 37°C for 10 min, and thereafter washed twice with medium.

For immunoprecipitation of CD222, 2–20 × 10⁷ cells were lysed in lysis buffer (20 mmol/l Tris and 140 mmol/l NaCl [pH 7.5]), containing 1% lauryl maltoside or 1% Nonidet P-40 as detergent, complete protease inhibitor mixture (Roche, Basel, Switzerland), and benzonase (Novagen, Merck Millipore) for 30 min on ice. Lysates were incubated for 2 h at 4°C with cyanogen bromide–activated beads coupled with CD222 mAb MEM-238 or isotype control mAb AFP-12. The beads were washed three times with ice-cold lysis buffer, and the bound proteins were eluted with 1.2× nonreducing Laemmli sample buffer at 95°C for 5 min.

For immunoprecipitation of Lck, the Lck-deficient Jurkat T cell clone JCaM 1.6 was transduced with a lentiviral expression construct for Lck fused to the One-Strep-tag epitope (Lck-OST). Lysates of resting or sodium pervanadate–stimulated cells were subjected to precipitation with Streptactin–Sepharose beads. Biotin was used to elute bait proteins from the capture beads. Biotin-blocked beads served as a negative control.

For the analysis of the CD45–Lck interaction at the plasma membrane, JCaM 1.6 T cells expressing Lck-GFP were silenced for CD222. Then, the silenced and control cells were surface labeled with CD45 mAb on ice, washed, and lysed with lysis buffer containing detergent 1% Nonidet P-40. The mAb complexes were then precipitated by using protein A/G beads (Santa Cruz Biotechnology).

For specific immunoprecipitation of cell surface CD222, unstimulated PBMC were labeled with mAb against CD222, CD45 (positive control), or AFP-12 (negative control). Proteins were cross-linked with 1 mM DSP (Pierce, Thermo Scientific Fisher, Rockford, IL), and cells were lysed with detergent 1% lauryl maltoside on ice. The Ab–protein complexes were fished with protein A/G beads overnight. Beads were washed three times, and proteins were eluted with Laemmli sample buffer.

The immunoprecipitates were analyzed by Western blotting. Briefly, proteins were separated by SDS-PAGE, followed by transfer of the proteins to Immobilon polyvinylidene difluoride membranes (Merck Millipore). The membranes were blocked with 5% nonfat milk (Maresi, Vienna, Austria) in TBS-Tween and incubated with specific Ab dilutions for detection of indicated proteins. For visualization, blots were developed either with HRP secondary conjugates and a HRP substrate (Biozym, Hessisch Oldendorf, Germany) or with secondary fluorescence Abs, followed by detection at either the Fuji LAS-400 imager (Fuji, Tokyo, Japan) or the Odyssey infrared imaging system (LiCor Biosciences, Lincoln, NE). Band intensities were quantified via Multi Gauge Software Version 3.1 (Fuji), Image Studio Version 2.0, or ImageJ.

For mass spectrometric analysis, CD222 coimmunoprecipitates and corresponding controls were digested in-solution as described previously (20), and samples were prepared, according to the filter aided sample preparation procedure following analysis via liquid chromatography–mass spectrometry (LC-MS). Lck coimmunoprecipitates were separated on 4–12% gradient Bis-Tris NuPAGE gels (Invitrogen Life Technologies), the gels were washed in distilled water, lightly stained with Colloidal Blue (Invitrogen Life Technologies), and subjected to Gel-LC-MS as described previously (21).

Stable knockdown cell lines were produced via lentiviral introduction of short hairpin (sh)RNA specific for the CD222 mRNA. The following CD222 silencing sequences were used for cloning into the vector pLKOpuro1 (a gift from S. Steward, Washington University School of Medicine, St. Louis, MO): shCD222/P1 at position 6588 with the sequence 5′-GCCCAACGATCAGCACTTCttcaagagaGAAGTGCTGATCGTTGGGC-3′ and shCD222/P2 at position 4525 with the sequence 5′-GAGCAACGAGCATGATGACttcaagagaGTCATCATGCTCGTTGCTC-3′. The MISSION nontarget shRNA control vector (pLKOpuro1; Sigma-Aldrich) was used as a negative control (shCTR). Virus particles were generated via transfection of HEK-293 cells with the silencing constructs and the lentiviral envelope coding plasmid pMD2.G and the packaging vector psPAX 2 (both helping vectors were constructed by D. Trono, Ecole Polytechnique Federal de Lausanne, and obtained from Addgene, Cambridge, MA) as described previously (22). After 48 h, the virus particles were harvested, filtered, and used for the target cell line infections in the presence of 5 μg/ml polybrene (Sigma-Aldrich). The next day, the cells were washed with and maintained in medium containing 1 μg/ml puromycin (Sigma-Aldrich) to select transduced cells. CD222 knockdown cells were used from 10 d to 2 mo postinfection for all experiments.

Lck constructs were prepared in the retroviral expression vector pBMN-Z for transfection of the Phoenix amphotropic virus producer cell line (both provided by G. Nolan, Stanford University School of Medicine, Stanford, CA) (15). The target cells were transduced one to three times with the viral supernatants depending on the efficacy of transduction and maintained afterward in normal culture medium.

Cys2His2-based zinc fingers were created via the Context Dependent Assembly method provided by the free software tool ZiFiT to identify and target the genomic sequence of CD222 specifically (23). The zinc finger constructs targeting exons 18 and 22 were cloned into pMLM290/pMLM292 and pMSM800/pMLM802 (both from Addgene) to generate zinc finger nuclease expression constructs. These constructs were used for the transfection of Jurkat T cells via electroporation. For this, 2 × 10⁶ cells were resuspended in 100 μl Cytomix (120 mM KCl, 0.15 mM CaCl₂, 5 mM MgCl₂, 10 mM K₂HPO₄, 25 mM HEPES [pH 7.6], 2 mM EGTA, 5 mM Glutathion, 1.25% DMSO [v/v], and 100 mM sucrose), mixed with the plasmids (each 0.5 μg), transferred in electroporation cuvettes, and transfected with the Amaxa System program S-18 (Amaxa, Lonza, Basel, Switzerland). Cells were immediately removed from the cuvettes, shortly spun down, and cultured in 12-well plates containing prewarmed medium. After cell recovery and expansion, CD222-negative cells were repeatedly sorted with a FACSAria (BD Biosciences), expanded in Jurkat T cell–conditioned medium, and cultured in RPMI 1640 medium until analysis.

Primary T cells or T cell lines were let to adhere on adhesion slides (Superior-Marienfeld Laboratory Glassware, Lauda-Köningshofen, Germany), fixed, and permeabilized with 4% formaldehyde and 0.1% saponin, respectively, and stained with the following Ab: the rabbit anti-Lck Ab H-95 and the rabbit anti-human mAb EEA1 (C45B10), followed by a goat anti-rabbit AF488 Ab, the anti-Lck mouse Ab (73A5), followed by anti-mouse AF488 Ab, the AF647-conjugated CD222 mAb MEM-238, and the Pacific Orange-conjugated CD45 mAb HI30 (all diluted 1:50) in PBS containing 2% BSA. For stimulation of T cells before staining, cells were incubated with SEE-loaded Raji B cells at 37°C for the indicated time points. Mouse fibroblasts were either cultured in 8-well chamber slides (Nunc, Thermo Scientific Fisher) until confluency or trypsinized and cytospun onto glass slides prior to fixation and staining as described above. For each cell, one vertical and one horizontal cut was made through the cell, and the Lck distribution was analyzed using the ZEN 2011 blue edition software (Zeiss, Jena, Germany). Nuclei were stained with DAPI. The slides were washed with 1× PBS and mounted with mounting medium for fluorescence analysis (Vectashield; Vector Laboratories, Burlingame, CA). Isotype-matched controls were included. Pictures were captured with a confocal laser scanning microscopy (CLSM) 700 (Zeiss).

For the analysis of cell surface Ags, cells were washed and resuspended in staining buffer (1× PBS containing 1% BSA and 0.02% NaN₃). Subsequently, the cells were incubated for 20 min at 4°C with a fluorochrome-conjugated mAb. For indirect immunofluorescence staining, the cells were additionally incubated with secondary conjugates for 20 min at 4°C. Before analysis, the cells were washed with staining buffer. Flow cytometry was performed using a LSR II (BD Biosciences) and analyzed with FlowJo version 8.8.6 for Macintosh or FlowJo version 7.2.5 for Windows.

To assess the IL-2 promoter activity and NFAT binding to DNA, IL-2 luciferase- and NFAT-reporter Jurkat T cells were assayed, respectively. Briefly, the reporter cells (2 × 10⁵) were either kept unstimulated or were stimulated for 6 h in triplicates in CD3 mAb OKT-3–coated (1 μg/ml) 96-well plates with or without soluble CD28 mAb Leu-28. The cells were washed with PBS and subsequently lysed in 100 μl luciferase lysis reagent (Promega, Madison, WI) for 30 min on ice. Seventy-five microliters of lysate were transferred into white 96-well microplates, and luminescence was measured after addition of the reaction reagent at a Mithras LB940 Elisa Reader (Berthold Technologies, Bad Wildbach, Germany). The protein concentration was determined via Bradford assay (Bio-Rad, Hercules, CA), and the luminescence intensity was normalized to the protein content of each sample.

Culture supernatants (30 μl) of primary T cells were harvested after 24 h of stimulation and analyzed via the Luminex xMAP suspension array technology. Briefly, the cells were stimulated with plate-bound OKT3 (1 or 5 μg/ml) and soluble Leu-28 (2 μg/ml) or kept unstimulated. Standard curves were generated using recombinant cytokines (R&D Systems, Minneapolis, MN). Experiments were performed at least twice with each sample in triplicates. Data are expressed as mean values of triplicates.

Cells (1 × 10⁶) were washed, resuspended in 100 μl RPMI 1640 medium containing 1 μmol/l Indo-1 AM (Invitrogen Life Technologies) and incubated for 30 min at 37°C. The cells were washed with medium, incubated in 1 ml medium for next 30 min at 37°C, and rested thereafter on ice until data acquisition at a LSR II flow cytometer. For the analysis of intracellular calcium flux, 300 μl Indo-1–loaded cells were prewarmed for 5 min at 37°C, the baseline response was recorded for 30 s, before the cells were stimulated with a 1:300 dilution of the hybridoma supernatant of the anti-TCR mAb C305 for 3 min to analyze the increase in calcium mobilization. The indo-violet/indo-blue ratio was calculated with the FlowJo software and plotted against the time. Ionomycin was used to check the overall responsiveness and cell viability.

Jurkat T cells or primary T cells (1.5 × 10⁶) were starved for 1 h in RPMI 1640 medium containing 1% FCS at 37°C, 5% CO₂ and 80% humidity. Cells were rested on ice for 10 min, stimulatory mAb were added, allowed to bind for 5 min, and then, the cells were incubated for several time points at 37°C. The cells were quickly washed with ice-cold lysis buffer without detergent and lysed in complete lysis buffer containing 1% detergent Nonidet P-40 (Promega), a protease inhibitor mixture (Roche), sodium orthovanadate (Sigma-Aldrich), sodium fluoride, and benzonase for 30 min on ice. Lysates were sampled with 4× reducing or nonreducing sample buffer and analyzed by Western blot analysis.

For real-time PCR analysis, primary human T cells, both resting and stimulated for 1 d with plate-bound CD3 mAb OKT3 (1 μg/ml) together with soluble CD28 mAb Leu-28 (0.5 μg/ml), were lysed in TRIzol (Invitrogen Life Technologies), and RNA was extracted according to the manufacturer’s instructions. cDNA was synthesized from 500 ng total RNA with SuperScript VILO cDNA Synthesis Kit (Invitrogen Life Technologies). Gene expression was measured by the 2^−ΔΔCT method (24), based on quantitative real-time PCR using the CFX96 Real-Time PCR system (Bio-Rad) with TaqMan primer sets for human CD222, and YWHAZ as an endogenous control. For conventional PCR, 20 ng genomic DNA purified by standard isopropanol precipitation was used as a template of PCR amplification with Taq polymerase (40 cycles at 95°C for 30 s, 61°C for 30 s, and 72°C for 30 s) and with primer sets for human CD222 bridging an exon–exon junction, and intronless gene endosialin, as an endogenous control (TaqMan Gene Expression Assay; ABI, Life Technologies).

Cells (2–3 × 10⁷) were lysed for 30 min on ice in lysis buffer containing 1% detergent Brij-58 and protease inhibitors. Lysates were adjusted with 80% sucrose solution to 40% sucrose in 1 ml, placed on a 1 ml 60% sucrose layer and overlaid by 20, 10, and 5% sucrose fractions (all 1 ml containing 0.5% detergent Brij-58 and protease inhibitors). Gradients were ultracentrifuged for 16–18 h at 150,000 × g at 4°C. Fractions (500 μl) were taken from the top to the bottom and denatured at 95°C with sample buffer and analyzed by SDS-PAGE, followed by Western blot analysis.

The GE Healthcare AKTA FPLC system was used for the gel filtration experiments. Cell lysates were prepared by solubilization with 1% detergent Triton X-100 (Promega). Samples were loaded at 4°C in a volume of 500 μl onto a Superose 6 HR 10/300 GL column (GE Healthcare) equilibrated with Triton X-100 (0.5%) in PBS. The absorbance at 280 nm was monitored, and 500-μl fractions were collected and analyzed by immunoblotting. The following standards of molecular mass (all from GE Healthcare or Sigma-Aldrich) were used: Blue dextran (2000 kDa), thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), BSA (66 kDa), and carbonic anhydrase (29 kDa).

Cells (6 × 10⁷) were incubated in 2 ml hypotonic buffer (42 mmol/l potassium chloride, 5 mmol/l magnesium chloride, 10 mmol/l HEPES, and protease inhibitor mixture [pH 7.5]) for 30 min on ice and thereafter passaged nine times through a 30-g needle. Cell fragments were centrifuged twice at 300 × g, and the resulting supernatant was centrifuged for 40 min at 35,000 × g at 4°C. The cytosolic supernatant was collected and the pellet containing membranous cellular parts was solubilized in extraction buffer (1× lysis buffer containing 1% Triton X-100, 0.1% NaDodSO₄, 0.5% sodium deoxycholate, and the protease inhibitor mixture) for 30 min on ice and centrifuged at 300 × g to remove debris. The fractions were analyzed via Western blotting.

Data analysis was performed in GraphPad Prism5 (GraphPad Software, La Jolla, CA). Different experiments were analyzed by one-sample or standard Student t test or two-way ANOVA, followed by the Bonferroni posttest. A p value < 0.05 was considered as significant (*). Statistical analyses were performed in GraphPad Prism. In general, data are expressed as mean ± SEM (or mean ± SD, when indicated).

CD222 is upregulated on the cell surface of rodent T cells upon T cell activation (17). Therefore, we hypothesized that this molecule might play a role in T cell activation. We first investigated the expression of CD222 on human lymphocytes. Isolated human PBMC were labeled with CFSE and analyzed for surface expression of CD222 via flow cytometry using a specific CD222 mAb. Upon cross-linking of the TCR complex with an activatory CD3 mAb, CD222 was upregulated on the cell surface on days 2, 3, and 4 (Fig. 1A). However, analysis of CD222 mRNA levels in resting and CD3/CD28 mAb–stimulated T cells revealed no significant increase in CD222 expression upon activation (Fig. 1C). This suggests the regulation of CD222 via subcellular distribution and not via gene expression. Regardless of the cell type, ∼90% of CD222 have been reported to be localized in the TGN and endosomal compartments at steady state (25). Therefore, the majority of CD222 is kept intracellular and only a small proportion is displayed, dependent on the cellular condition, at the cell surface.

Investigation of the T cell proliferation kinetics showed that CD222 upregulation on the plasma membrane preceded T cell division cycles. Both stimulated, yet not blasting T cells and blasting T cells displayed more CD222 on the surface compared with unstimulated T cells (Fig. 1B). Notably, the majority of proliferating cells was positive for CD222, which pointed toward a possible involvement of CD222 in T cell activation and/or differentiation.

To investigate the putative role of CD222 in T cell activation, primary human PBMC were silenced for the expression of CD222 via the lentiviral-based introduction of shRNA. Total CD222 protein content was analyzed for shCTR and CD222-silenced (shCD222/P1) primary T cells via Western blotting and revealed a reduction by ∼75% (Fig. 1D). Subsequent flow cytometric analysis revealed that CD222 surface expression was downregulated by ∼50% on shCD222/P1 T cells compared with shCTR T cells (Fig. 1E). To assess their effector functions, both shCTR and shCD222/P1 T cells were stimulated with low or high concentrations of plate-bound CD3 mAb (1 and 5 μg/ml) in combination with costimulatory soluble CD28 mAb (2 μg/ml). Secreted cytokine levels were measured via the Luminex system. Secretion of TNF-α, IL-2, and IFN-γ was reduced by >50% in shCD222/P1; higher concentrations of CD3 mAb could not rescue the response in shCD222 cells (Fig. 1F). For further experiments, the wild-type (wt) Jurkat T cell line as well as the IL-2 and NFAT luciferase reporter Jurkat T cell lines were used because the overall CD222 silencing efficiency was mostly moderate in primary T cells in our hands, whereas in Jurkat T cells, the efficiency of silencing was much higher. Furthermore, CD222 is a very stable molecule. It takes ∼7–10 d upon silencing for the complete degradation of CD222 so that the functional effects become observable. As primary T cells are very sensitive to prolonged in vitro culture, we used the T cell lines for further experiments.

To determine whether the cytokines were less synthesized or the transport/secretion was hampered upon CD222 downregulation, we used human luciferase reporter T cell lines: first, CD222- and control-silenced Jurkat T cells that expressed luciferase under the control of the IL-2 promoter were stimulated via CD3 or CD3/CD28, and the lysates were analyzed for luciferase activity. Silencing of CD222 at two distinct positions (shCD222/P1 and shCD222/P2) led to a reduction of the luciferase activity upon T cell stimulation via CD3 (Fig. 2A). Even activation of the costimulatory pathway via CD28 ligation, which quantitatively enhances TCR-induced signals (26), could not rescue the IL-2 promoter activity in the reporter cells silenced at the first position (P1) and resulted in a significant luciferase reduction. Although silencing at the second position (P2) did not reduce CD222 transcripts as effective as P1 (Fig. 2A, inset), the corresponding cells still displayed a reduced IL-2 promoter activity, although not significant. Second, we used human reporter Jurkat T cells that expressed the luciferase gene driven by an artificially designed promoter region containing multiple binding sites for the transcription factor NFAT (27). The NFAT-driven luciferase expression was significantly reduced in shCD222/P1 cells compared with shCTR cells upon stimulation with CD3/CD28 mAb (Fig. 2B). These data revealed that CD222 knockdown affected cytokine production at the transcriptional level.

To analyze whether decreased Ca²⁺ mobilization was responsible for reduced NFAT promoter activity, we measured intracellular Ca²⁺ fluxes by flow cytometry with the Ca²⁺ sensitive dye Indo-1. To enable an equal Lck expression in control- and CD222-silenced T cells, we used the Lck-deficient Jurkat T cells JCaM 1.6 that we reconstituted before silencing with a GFP-tagged Lck followed by expressing gating. We found that silencing of CD222 at P1 and P2 led to a significant reduction in the cytosolic Ca²⁺ mobilization capacity compared with control-silenced Lck-GFP⁺ JCaM 1.6 T cells (Fig. 2C). Detailed analysis of Ca²⁺ flux experiments showed that the peak values of the 390/495 ratio and the overall mean values of shCD222/P1-silenced cells were significantly reduced, whereas the baseline values remained unchanged (Fig. 2D). Silencing of CD222 at the less efficient P2 site influenced the Ca²⁺ flux to a lesser extent, which might be explained via a dose-dependent mode of action of CD222. Indeed, this assumption was supported by other experiments: titration of virus, carrying the silencing construct, correlated with lower expression of CD222 and reduced Ca²⁺ mobilization in Jurkat T cells (Fig. 2E) suggesting a direct correlation between the total amount of the CD222 protein and the T cell response to TCR stimuli. Furthermore, the artificial overexpression of CD222 led to significantly higher intracellular Ca²⁺ flux in Jurkat T cells and CD222-overexpressing cells reacted faster than cells expressing the control vector (Fig. 2F). However, it has to be mentioned that cells highly overexpressing CD222 are more susceptible for apoptosis compared with control cells (28). Furthermore, high amounts of CD222 lead to a decreased cellular growth rate (29). Therefore, just cells slightly overexpressing CD222 survived and grew adequately. Finally, to verify the data obtained via RNA interference, genomic CD222 was targeted and cut via sequence-specific zinc finger nucleases in Jurkat T cells (Supplemental Fig. 1A). Analysis of three independently sorted CD222 knockout populations (C20, C21, and C25; Supplemental Fig. 1B), revealed that each of them displayed a reduction in intracellular Ca²⁺ fluxes compared with CTR cells (Supplemental Fig. 1C). Importantly, this phenotype was rescued in knockout cells that were stably transduced with recombinant human CD222 (C20 cells reconstituted with CD222; Supplemental Fig. 1D–F). Taken together, these data imply that CD222 is an essential regulator of the TCR signaling cascade.

To untangle the mode of operation of CD222 in TCR signaling, we stimulated control- and CD222-silenced Jurkat T cells with the TCR-specific mAb C305 and evaluated the tyrosine phosphorylation in the cell lysates via Western blot analysis: several molecules were less phosphorylated in CD222-silenced cells, particularly in molecular mass regions corresponding to the TCR proximal signaling molecules ZAP70, LAT, and CD3 (Fig. 3A), suggesting a very early effect of CD222 on T cell signal transduction. To verify this, we used specific mAb in a multicolor fluorescence multiplexing approach. Indeed, the earliest T cell signaling proteins CD3ζ and ZAP70 displayed lower tyrosine phosphorylation upon CD222 downregulation. Accordingly, all ensuing signaling molecules, such as LAT and ERK, were less phosphorylated in shCD222 cells (Fig. 3B). Statistical evaluation of multiple experiments analyzing pCD3ζ and pERK1/2 revealed that the phosphorylation of both (i.e., early and late signaling molecules) was significantly reduced upon CD222 knockdown (Fig. 3C). In addition, Lck's serine 59, a substrate of active ERK1/2 (30), was less phosphorylated in CD222 knockdown cells, which possibly could underlie an insufficient signal amplification to sustain T cell activation (Supplemental Fig. 2A). Supporting evidence for the implication of CD222 in TCR-proximal signaling was provided via CLSM: upon 15-min stimulation of JCaM 1.6 T cells with superantigen SEE-loaded Raji B cells, CD222 redistributed from the cell surface and/or intracellular pericentrosomal compartments (31) to the interface of the IS in ∼70% of JCaM 1.6 T cells that were in contact to Raji B cells (n = 24; Fig. 3D).

To unravel possible interaction partners of CD222, we pulled down CD222 with specific mAb-coupled beads after cell lysis of resting Jurkat T cells with lauryl maltoside, and analyzed coprecipitated proteins via LC-MS. We found, along with known CD222 binding proteins, various T cell–specific proteins. Among these proteins we found Lck and CD45, molecules implicated in the regulation of T cell signaling (Table I). To verify the protein–protein interactions, we precipitated CD222 from lysates of resting Jurkat T cells via CD222 mAb-coated beads and analyzed coimmunoprecipitated proteins by Western blotting. We detected both, Lck and CD45, in the CD222 coimmunoprecipitates (Fig. 4A). Moreover, we detected the Lck-CD222 interaction also by a reverse approach performed with resting JCaM 1.6 T cells, expressing Lck fused to the One-Strep-Tag epitope (Lck-OST). Therein, CD222 was reproducibly found in the isolated Lck complexes via LC-MS analysis (Mascot protein score = 486.34). The interaction of CD222 and Lck also was confirmed by coimmunoprecipitation studies with JCaM 1.6 T cells, followed by immunoblotting (Fig. 4B). A truncated nonfunctional mutant of Lck encompassing only the first 10 aa of Lck (LckN10) did not coimmunoprecipitate with CD222 (Supplemental Fig. 2B), suggesting a specific protein–protein interaction. Furthermore, we verified the interaction of CD222 and Lck in nonstimulated primary human T cells isolated from peripheral blood (Fig. 4C). Immunofluorescence-based in situ localization via CLSM revealed that Lck colocalized with CD222 in resting Jurkat T cells (Fig. 4D). However, the majority of interactions seemed to take place in intracellular compartments: Upon conditions allowing the specific immunoprecipitation of cell surface-bound CD222, no Lck was coimmunoprecipitated from lysates of primary T cells (Fig. 4E). Furthermore, CD222 and Lck colocalized with the EEA1 in resting Jurkat T cells (Supplemental Fig. 3). Moreover, when we stimulated JCaM 1.6 T cells that ectopically expressed Lck-GFP with SEE-loaded Raji B cells, we found a coordinated intracellular accumulation of Lck and CD222 submembrane of the active IS (Fig. 4F), suggesting that the cytoplasmic interaction of CD222 and Lck prevails also upon T cell activation. Analysis of consecutive z-stacks of the depicted cells confirmed that CD222 colocalized with Lck almost exclusively within the cell with a Pearson’s coefficient of ∼0.3, whereas the Pearson’s coefficient was practically zero at the plasma membrane edges (Fig. 4F, right panel). These findings suggest that the interaction between CD222 and Lck occurs preferentially in cytoplasmic vesicles and that CD222 is involved in the initiation of the early signaling cascade through acting on Lck.

Lck has been reported to exist in resting T cells in four different pools, each constituting ∼25% of total Lck and constantly maintaining at this equilibrium (3, 4) with CD45 being an important regulator of the Lck phosphorylation status. Because we found CD222 to interact with both Lck and CD45, we hypothesized that CD222 might be a mediator involved in the regulation of Lck phosphorylation. To test this, we investigated the tyrosine phosphorylation of Lck normalized to total Lck in shCD222/P1, shCD222/P2, and shCTR Jurkat T cells via LiCor infrared fluorescence Western blot analysis (Fig. 5A, 5C). Corresponding CD222 surface expressions were analyzed via flow cytometry (Fig. 5B). Lck was significantly more phosphorylated at Y505 in shCD222/P1 cells compared with shCTR cells (Fig. 5A, 5C). Silencing with the second construct (shCD222/P2) as well increased Lck phosphorylation at Y505 (Fig. 5C), although not significantly, which correlated with the less effective silencing at this position (Fig. 5B). Lck phosphorylation at Y394 was not changed (Fig. 5A, 5D), whereas the amount of Lck not phosphorylated at Y394 (non-pY394) significantly increased upon CD222 knockdown (Fig. 5A, 5E). These findings suggest that silencing of CD222 causes a shift in Lck pools toward the less active form solely phosphorylated at the inhibitory Y505.

To pinpoint whether the increased phosphorylation at Y505 was the cause of a blockade in signaling upon CD222 silencing, we used a mutated form of Lck that was made up of a phenylalanine instead of the tyrosine at the position 505 and consequently could not be phosphorylated at this site (32). Lck^wt and the mutated form of Lck (Lck^Y505mut) had been ectopically expressed in Lck-deficient JCaM 1.6 T cells before the cells were silenced for CD222 and intracellular calcium mobilization was measured via flow cytometry. JCaM 1.6 T cells expressing Lck^wt showed a reduced calcium flux upon CD222 compared with control silencing. However, silencing of CD222 did not affect the calcium flux in JCaM 1.6 T cells carrying the mutated form of Lck (Fig. 5F), suggesting that the altered phosphorylation of Lck upon CD222-silencing caused the less reactive T cell phenotype.

To test whether CD222 was responsible for the efficient transport of Lck, we analyzed the distribution and membrane localization of Lck via immunofluorescent staining and CLSM of resting control and CD222 knockdown Jurkat T cells. The typical homogenous surface distribution of Lck was disrupted in CD222 knockdown cells (Fig. 6A). Approximately 60–70% of shCD222/P1 and shCD222/P2 Jurkat T cells showed a diminished Lck surface distribution when counted manually. Surface plots of Lck staining showed a punctuated distribution in CD222-silenced cells with Lck-free gaps among Lck-rich membrane areas (Fig. 6B), although the total Lck content in crude cell lysates of unstimulated Jurkat T cells was even higher in CD222-silenced cells as detected by immunoblotting (Supplemental Fig. 2C). Lck is posttranslationally modified by palmitoylation and myristoylation at its N terminus, which directs and anchors the molecule into lipid rafts (33, 34)—organizational platforms of cellular membranes important for T cell signaling (35–38). We investigated total cellular lipid raft fractions in resting shCD222/P1 and shCTR Jurkat T cells by means of sucrose-based density gradient flotation analysis. The comparison of lipid raft- with nonraft fractions did not reveal any significant changes upon CD222 knockdown (Supplemental Fig. 4A). The size exclusion chromatography analysis on a Superose 6 column revealed in shCD222 cells a slight shift of both Lck and CD45 towards higher molecular mass fractions, which also contained EEA1 (Supplemental Fig. 4B) corresponding to endosomal membranes (39).

To analyze the intracellular distribution of Lck in more detail, we evaluated the surface/cytoplasm ratio of Lck. Lck-deficient JCaM 1.6 T cells were retrovirally transduced to ectopically express GFP-tagged Lck (Lck-GFP) to visualize the molecule in situ without staining (Fig. 6C, upper panel). CD222 silencing at P1 led to a significant reduction of surface-localized Lck, as indicated by the decrease in the surface/cytoplasm ratio (Fig. 6C). shCD222/P2 cells also displayed a significantly decreased surface/cytoplasm ratio but again less pronounced than shCD222/P1. This phenomenon was highly specific because the transport of the Lck mutant variant bearing only the first 10 aa (LckN10-GFP), encompassing myristoylation and palmitoylation sites indispensable for the cell membrane targeting (Fig. 6D, upper panel) and lipid raft partitioning, was unaffected (Fig. 6D). In addition, the GFP-tagged Lck lacking the first 10 aa (LckΔN10-GFP), a mutant variant that in contrast cannot be inserted into the inner leaflet of the plasma membrane (Fig. 6E, upper panel) was distributed evenly within the cell of all cell types investigated, as can be seen by the ratio around 1 (Fig. 6E). When human Lck-GFP was retrovirally introduced in CD222-negative mouse fibroblasts, we observed similar patterns: Lck was not sufficiently transported to the surface, whereas in cells coexpressing human recombinant CD222, the surface localization of Lck-GFP increased (Fig. 6F). These findings indicate that CD222 does not influence lipid raft partitioning of Lck but rather directs its cellular distribution.

Finally, we investigated the interaction of Lck and CD45. Unstimulated JCaM 1.6 T cells ectopically expressing Lck-GFP were stained for CD45 and analyzed for the colocalization of CD45 with Lck in shCTR and CD222-silenced cells (shCD222/P1 and /P2; Fig. 7A). The overall Pearson’s coefficient, a measure of colocalization, of the CD45-Lck interaction was low, as reported previously (4), but markedly decreased upon silencing of CD222 (Fig. 7B). In silenced cells, Lck was apparently retained in intracellular membranes and/or accumulated freely in the cytoplasm. By biochemical separation of total cellular membranes from cytosolic cell contents, we found a significant increase in cytosolic Lck upon CD222 silencing (Fig. 7C). To test whether the CD45–Lck interaction occurred at the plasma membrane, we labeled the surface of living cells with a CD45 mAb on ice, lysed the cells, immunoprecipitated cell surface CD45, and analyzed the eluates for coimmunoprecipitated Lck (Fig. 7D). The interaction of cell surface CD45 with Lck was reduced by ∼40 and 20% in shCD222/P1 and shCD222/P2 knockdown cells, respectively (Fig. 7D). In contrast, LckN10, the non-functional mutant variant of Lck consisting of the 10 N-terminal amino acids responsible for plasma membrane anchorage, did not coimmunoprecipitate with surface CD45 (Fig. 7E). These data suggest that CD222 controls the transport of the kinase Lck to the plasma membrane and subsequently its interaction with CD45, the key activatory phosphatase of Lck.

Lck is described to exist in four different phosphorylation states, constantly maintained at equilibrium, with ∼40–50% of constitutive active Lck within resting T cells (3, 40). As active Lck represents a high risk for a T cell to get permanently activated resulting in hyperreactive adaptive immune reactions in an organism, the spatial arrangement of Lck has to be accurately organized in time and space. Limiting the access of possible kinase substrates or orchestrating the interaction with regulatory proteins can prevent overwhelming or self-directed immune responses. On the basis of our data, we propose a novel function for the late endosomal transporter CD222 in the early TCR signaling cascade. CD222 is essential for the efficient recruitment of Lck to the plasma membrane, the place where it interacts with its phosphatase CD45. The loss of CD222 leads to accumulation of Lck in its inactive form, which results in diminished T cell activation and effector functions.

Because Lck represents a very powerful kinase that is involved in various T cell signaling pathways (41–44), its transport and regulation has been investigated extensively (45–49). One important molecule implicated in the transport of Lck is the myelin and lymphocyte protein (MAL). Upon knockdown of MAL, Lck was stuck in the TGN, without significant increase in cytosolic Lck. Furthermore, MAL was essential for lipid raft partitioning of Lck, whereas the transport to nonraft membranes was unhampered (45). In contrast, knockdown of CD222 hampered the efficient transport of Lck to the plasma membrane with an apparent accumulation of Lck within intracellular membranes and freely floating in the cytoplasm but without any significant effect on lipid raft partitioning. It has been shown previously via single molecule microscopy and bleaching experiments that single Lck molecules recolonize the cell membrane directly from the cytoplasm, rather than via two-dimensional transversal diffusion (49), which may explain the high amount of non–membrane-bound cytosolic Lck. Lck has been reported previously to be present in endosomes (47, 50). Our confocal microscopy and also gel filtration fractionation experiments revealed an accumulation of Lck and CD222 in the EEA1-positive fractions corresponding to endosomal membranes. Furthermore, via mass spectrometry analysis we identified Rab11, a marker of endosomal membranes critically involved in protein trafficking, to interact with CD222. This strongly suggests that CD222 coordinates the endosome-dependent transport of Lck that was shown to be essential for T cell activation (48). Uncoordinated 119 protein has been shown to activate endosomal Rab11 for the plasma membrane targeting of Lck (48) and activation of Fyn (51). However, unlike MAL and uncoordinated 119 protein, CD222 does not contribute to the IS, but rather builds an early vesicular platform submembrane of the IS. Although several consecutive devices obviously coordinate Lck transport (52–54), the contribution of CD222 seems to be very specific as human CD222 can perform its action on human Lck even in mouse fibroblasts (i.e., irrespective of other T cell–specific molecules). Moreover, the absence of MAL in fibroblasts (55) suggests that CD222 operates upstream of MAL.

To our knowledge, the role of CD222 in T cell signal transduction has yet not been investigated in detail, but CD222 has been shown to enhance TCR-induced signaling via CD26 internalization (56). However, we found no evidence for the contribution of CD26 to the effect of CD222 on Lck because we have observed the effect of CD222 knockdown on T cell activation in Jurkat T cells that lack CD26 expression (57). Interestingly, ligation of the common costimulatory molecule CD28, which also quantitatively enhances T cell responses to TCR stimuli (26), did not rescue, but even resulted in a more pronounced phenotype induced by silencing of CD222.

The loss of CD222 in resting T cells results in accumulation of Lck single phosphorylated at residue Y505 that represents the closed inactive form of Lck (7, 58). Consequently, the homeostasis of the four proposed conformational forms of Lck (3) is shifted toward the inhibitory closed conformation of Lck. Recently, it has been proposed that the intracellular distribution of Lck is regulated via its phosphorylation state and the corresponding conformations (4, 6): the open active form (pY394) is thought to induce clustering, whereas the closed inactive form (pY505) prevents the formation of clusters, which has been shown to be lipid independent (4). Putting these data together one might suggest that the CD222 knockdown, which results in a 2-fold increase of inhibitory Lck pY505, does prevent Lck from clustering and T cells from getting activated. The finding that the hyperactive mutated form Lck^Y505mut triggers the activation of the secondary messenger Ca²⁺ also in the absence of CD222 supports this model. On the basis of our data, CD222 mediates the shift of Lck toward the active dephosphorylated form by complexing it with CD45. We pinpoint the CD222-driven CD45–Lck interaction to the cell surface, which indicates, first, that dephosphorylation of Y505 via CD45 takes place at the cell membrane, and second, that Csk-induced phosphorylation of Y505 (59) can occur within the cell.

Because we found CD222 to interact with both the kinase Lck and its phosphatase CD45, the question aroused whether the transport of CD45 might be as well miss-regulated upon silencing of CD222. However, several facts speak against the simultaneous transport deregulation of Lck and CD45 in the CD222 knockdown phenotype. First, we found just a minor downregulation of CD45 at the cell surface. Second, CD45 is not only a positive but also a negative regulator of T cell activation and differentiation (59, 60). Intermediate levels of CD45 surprisingly produce hyperreactive T cells (59), which we could not detect at all in our experimental setup. Third, CD45 dephosphorylates not only Lck but also CD3ζ (4, 59–61). However, the phosphorylation of CD3ζ was unaffected in resting, and notably, even significantly reduced in stimulated CD222 knockdown cells. Taking these facts into consideration, we conclude that the primary function of CD222 in T cell activation is to act as intermediary in the interaction of CD45 with Lck resulting in the correct dephosphorylation of Lck at the inhibitory Y505.

We show in this study that the interaction between CD222 and Lck occurs preferentially inside the cell and disbands at the cell surface, where Lck is “handed over” to CD45. This indicates that CD222 is involved in very early signaling events. However, we do not have at the moment an explanation for CD222´s upregulation on the surface of proliferating and differentiating T cells—in the later phase of T cell activation, shortly before cell division is initiated. One can envisage several possible reasons of the increased CD222 surface accumulation during later phases of T cell activation: 1) It might be necessary to fully confer the capacity of cells to respond. 2) It might result from higher protein transport activity mediated by CD222. Because upon T cell activation the turnover of Lck is enhanced (50), the fusion of CD222-containing transport vesicles with the plasma membrane might result in elevated CD222 on the cell surface. 3) It might be important for the later signaling events as a negative feedback loop to abrogate T cell activation. The latter could be accomplished via regulating the bioavailability of the mitogenic insulin-like growth factor 2 that is both postnatally expressed in thymic epithelial cells (62) and upregulated in diseases (28), namely via internalization and subsequent degradation in lysosomes (12, 63). 4) Furthermore, CD222’s surface upregulation might play a role in T cell differentiation from the double negative to the double positive stage as CD222 Ab was reported to block T cell ontogeny (64). Notably, Lck knockout mice also display a similar reduction in double-positive thymocytes in T lymphocyte development (65) as the one caused by the CD222 Ab treatment. However, because of the lethality of the CD222 knockout in mice (66–68), the function of CD222 in T cell differentiation has never been shown in vivo. 5) An involvement of CD222 in T cell migration can also not be excluded as it has been shown that CD222 negatively regulates cell adhesion via uPAR and integrins (22). 6) Alternatively, it could be the result of a better accessibility of the CD222 mAb caused by altered interactions that unmask the epitope.

Irrespective of the pathways that are amenable in the upregulation of cell surface CD222 upon T cell activation, we show in this study a central function of the endosomal transporter CD222 in the initiation of T cell signal transduction through controlling Lck distribution: CD222 targets Lck to CD45 at the plasma membrane and thereby maintains the equilibrium of Lck phosphorylation at the steady state.

We thank Peter Steinberger (Institute of Immunology, Medical University of Vienna) and Thomas Decker (Max F. Perutz Laboratories, Department of Genetics, Microbiology and Immunobiology, University of Vienna) for productive discussions and constructive ideas; Erwin Wagner (Spanish National Cancer Research Centre, Madrid, Spain) for providing mouse fibroblasts, Václav Hořejší (Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Prague, Czech Republic), Nicole Boucheron (Division of Immunobiology, Institute of Immunology, Center for Pathophysiology, Infectiology and Immunology, Medical University of Vienna, Austria), and Arthur Weiss (Department of Medicine and Department of Microbiology and Immunology, Rosalind Russell-Ephraim P. Engleman Medical Research Center for Arthritis, and Howard Hughes Medical Institute, University of California, San Francisco, San Francisco, CA) for providing Ab; Reinhard Fässler (Department of Molecular Medicine, Max Planck Institute of Biochemistry, Martinsried, Germany) for experimental and intellectual help with mass spectrometric analysis; S. Steward (Washington University School of Medicine, St. Louis, MO); G. Nolan (Baxter Laboratory in Stem Cell Biology, Department of Microbiology and Immunology, Stanford University, Stanford, California) for providing cloning vectors and cell lines; and Cristiana Soldani, Anna Elisa Trovato and Chiuhui Mary Wang (Humanitas Clinical and Research Center, Rozzano, Milan, Italy) for practical help with immunofluorescence analysis. We are grateful to the imaging and flow cytometry core facilities of the Medical University of Vienna.

The authors have no financial conflicts of interest.