Proteomics Identification of Sorting Nexin 27 as a Diacylglycerol Kinase ζ-associated Protein

Diacylglycerol kinase ζ is a member of the diacylglycerol kinase family of enzymes, which generate phosphatidic acid through diacylglycerol phosphorylation. In addition to the catalytic and cysteine-rich domains found in all diacylglycerol kinases, diacylglycerol kinase ζ has a MARCKS domain as well as a C-terminal region containing four ankyrin repeats and a PDZ-binding motif. Previous reports demonstrated that diacylglycerol kinase ζ interaction with several proteins is an important mechanism for modulating the localization and activity of this enzyme. Here we used a proteomics approach to search for novel diacylglycerol kinase ζ-interacting proteins and identified sorting nexin 27 (SNX27), a recently described member of a protein family involved in intracellular trafficking, which has a PDZ domain in addition to the phox homology domain characteristic of SNX proteins. Co-immunoprecipitation studies and two-hybrid analysis confirmed physical, PDZ-dependent association between SNX27 and diacylglycerol kinase ζ. Because diacylglycerol kinase ζ is expressed abundantly in T lymphocytes, we characterized SNX27 expression and subcellular localization in these cells. SNX27 co-localized with transferrin receptor-positive vesicles, pointing to its participation in T cell endocytic recycling. Expression of deletion mutants revealed that in addition to the phox homology domain the SNX27 PDZ domain contributed to vesicle localization of this protein, suggesting that interaction with diacylglycerol kinase ζ regulates SNX27 localization. Analysis of cells with RNA interference-mediated knockdown of diacylglycerol kinase ζ showed accelerated transferrin receptor exit from the lymphocyte endocytic recycling compartment back to the plasma membrane, further confirming diacylglycerol kinase ζ-dependent control of vesicle trafficking. These data support a previously unreported role for diacylglycerol kinase ζ in the modulation of membrane trafficking, which may also help to define SNX27 function.

Intracellular membrane traffic requires a complex molecular machinery with a plethora of small GTPases, adaptors, and coat components that must be assembled and disassembled in different steps to ensure correct vesicle formation. Membrane lipids are key components in this process; many proteins involved in vesicle formation have lipid-binding domains, and modulation of lipid-modifying enzymes profoundly alters secretion and/or endocytosis (1).
Diacylglycerol (DAG) 1 is a lipid with important functions in membrane trafficking. When generated in restricted membrane regions, the characteristic negative curvature of DAG promotes the membrane constriction essential for fission and the instability required for fusion (2)(3)(4)(5). In addition to modifying membrane characteristics, DAG binds to and activates various proteins needed for vesicle formation, such as protein kinase D and ADP-ribosylation factor (Arf) GTPase-activating protein (6,7). In addition to DAG, phosphatidic acid (PA) also confers the negative curvature that facilitates vesicle fission or fusion (3, 8 -10). PA can also bind to and activate enzymes that participate in membrane trafficking, such as coatomer, Arf, N-ethylmaleimide-sensitive factor, kinesin, phosphatidylinositol-4-phosphate 5-kinase, and Arf6 GTPase-activating protein (11)(12)(13)(14)(15).
The diacylglycerol kinase (DGK) family is an evolutionarily conserved family of lipid kinases that phosphorylate DAG to produce PA (16). All members of the family have at least two N-terminal cysteine-rich domains (C1) and a conserved catalytic domain. These proteins have additional functional domains that allow their classification into five subgroups (I-V). Their structural diversity, distinct tissue expression, and specific intracellular localization confer on each DGK isoform the ability to regulate different DAG and PA pools and thus to participate in diverse signaling complexes (17). DGK belongs to the type IV DGK family, characterized by a MARCKS (myristoylated alanine-rich protein kinase C (PKC) kinase substrate) homology domain as well as a C-terminal region with four ankyrin (Ank) repeats and an ETAV sequence (18). These four amino acids can bind the class I PDZ (postsynaptic density protein, disc-large, and zonula occludens-1) domain, thus constituting a PDZ-binding motif (PDZbm) (19). DGK is expressed ubiquitously and is associated with cell cycle regulation, cytoskeletal reorganization, and modulation of the immune response among other functions (20 -24).
DGK is expressed abundantly in T lymphocytes; studies using GFP-coupled DGK chimeras in live T cells demonstrated receptor-dependent membrane translocation of this enzyme (25).This analysis indicated that the C-terminal region of the protein confers specificity for protein translocation, suggesting the importance of regulation based on proteinprotein interactions. Accordingly recent studies showed DGK interaction with several proteins such as PKC, Rac, syntrophins, leptin receptor, and Src (23, 26 -29), indicating that DGK functions may be largely dependent on the formation of distinct protein complexes.
Using a mass spectrometry-based analysis of DGK-interacting proteins, we identified sorting nexin 27 (SNX27), a member of the SNX family of proteins involved in membrane traffic and protein sorting (30,31). The association between these proteins was direct and was mediated by interaction of the SNX27 PDZ domain with the C terminus of DGK. SNX27 was expressed in hematopoietic cells and localized to the endocytic recycling system of T lymphocytes. Finally we traced transferrin receptor (TfR) recycling, which was accelerated in cells with diminished DGK levels. Our results identify SNX27 as a new DGK-binding protein and unveil a function for SNX27⅐DGK complex in the control of protein trafficking of T cells.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Orthovanadate, PMSF, poly(DL-lysine) and Igepal CA-630 were from Sigma. Leupeptin and aprotinin were purchased from Roche Applied Science, wortmannin was from Calbiochem, and transferrin tetramethylrhodamine (Tf-Rhod) was from Molecular Probes (Leiden, The Netherlands). Rabbit polyclonal anti-DGK antibody (Ab) raised against a C-terminal peptide was a generous gift of Dr. Kaoru Goto (Department of Anatomy and Cell Biology, Yamagata University School of Medicine, Yamagata, Japan) (32). Rabbit polyclonal anti-DGK Ab raised against an N-terminal peptide was a generous gift of Dr. M. K. Topham (University of Utah, Salt Lake City, UT) (33). To generate the polyclonal anti-SNX27, the N-terminal 265-residue fragment of SNX27, which contains PDZ and phox homology (PX) domains, was produced as a GST fusion protein in bacteria. The fragment was cleaved with thrombin and used to immunize rabbits with Freund's adjuvant (Invitrogen). To affinity purify antibodies to SNX27, serum from immunized rabbits was incubated with the antigen coupled to cyanogen bromide-activated Sepharose (Amersham Biosciences). Bound antibody was eluted with Immuno-Pure IgG elution buffer (Pierce), neutralized with PBS, pH 7.4, and then dialyzed against the same solution. We used the following mouse monoclonal antibodies: anti-hemagglutinin (anti-HA) (Babco, Richmond, CA), -Myc (Cell Signaling Technology, Danvers, MA), -GST (Santa Cruz Biotechnology, Santa Cruz, CA), -GFP (Roche Applied Science), -tubulin (Sigma), -early endosomal antigen 1 (EEA1), -GM130 and -SNX2 (BD Transduction Laboratories), and -CD63 (On-cogene Research, San Diego, CA). Mouse anti-human TfR and rabbit anti-Rab11 were from Zymed Laboratories Inc.. The monoclonal Ab to human LAMP1 developed by J. Thomas August and James E. K. Hildreth was obtained from the Developmental Studies Hybridoma Bank (developed under the auspices of the NICHD, National Institutes of Health and maintained by the Department of Biological Sciences, University of Iowa, Iowa City, IA). Cy3-and Cy5-conjugated antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA), and Alexa 488-conjugated Ab was from Molecular Probes.
Plasmids and DNA Constructs-The pcDNA3MycDGK, GFP-DGK, and HA-DGKCT constructs were described previously (25). For generation of GST fusion proteins, pcDNA3MycDGK was BglIIdigested, blunted, and KpnI-digested, and the 3.9-kb fragment was subcloned in the pEBG eukaryotic GST vector digested with KpnI/ClaI (GST-DGKFL). The construct encoding the C-terminal region of the protein (GST-DGKCT, including the four ankyrin repeats and the PDZbm) was excised from pGEM-T with NotI and then subcloned in pEBG vector digested with NotI. The human Myc-tagged SNX27b full length (Myc-SNX27bFL) was described previously (34). Myc-tagged deletion mutants (Myc-SNX27b⌬RA/-SNX27b⌬PX/-SNX27bRA) were generated from the full-length SNX27b using PCR and subcloned into pDMyc-neo vector, which is a modified version of the pCIneo vector (Stratagene) (35), with the same restriction enzyme sites as above. All constructs were confirmed by sequencing. The coding region corresponding to mouse SNX27a cloned into the pSPORT1 vector was obtained from Open Biosystems (Clone number 6431126, GenBank TM accession number BC053495); pSPORT1-SNX27a was digested with SalI/BamHI, and the 1.7-kb fragment was subcloned in pEGFP-C2 digested with SalI/BamHI. To generate GFP-SNX27a⌬PDZ, the sequence 336 TCCGAG 341 in GFP-SNX27a was mutated to GTCGAC to generate a SalI site. Site-directed mutagenesis was performed using the QuikChange mutagenesis kit (Stratagene). The plasmid was cleaved with SalI, and the resulting 6.4-kbp fragment lacking the PDZ domain was then religated.
For yeast two-hybrid interaction assays, four DGK constructs were subcloned in pGBKT7 vector fused to GAL4BD as bait, the SNX27 N-terminal region (containing the PDZ and PX domains) was subcloned into pGADT7 vector fused to GAL4AD, and pGBKT7DGKFL was generated by digesting pcDNA3MycDGK with EcoRI, and the 3.4-kb fragment was subcloned into EcoRI-digested pGBKT7. pGBKT7DGK⌬Ank was generated from pGBKT7DGKFL digested with SacI and religated. To generate pGBKT7DGK⌬PDZbm, GFP-DGK⌬PDZbm was digested with NcoI, blunted, and digested with EcoRI/AflII; the 2.9-kb fragment was subcloned in pGBKT7 digested with EcoRI/SmaI. The pGBKT7CT construct, including the four ankyrin repeats and PDZbm, was generated by PCR from GFP-DGKFL with appropriate primers (AnkPDZ1, 5Ј-GAATTCGCACTGC-CCCAAGGTGAAG-3Ј; AnkPDZ, 5Ј-GTCGACTACACAGCTGTCTC-CTGGTCC-3Ј), including two restriction sites, EcoRI and SalI. The 430-bp PCR product was subcloned in the pGEM-T Easy vector and then excised with EcoRI/SalI for subcloning in EcoRI/SalI-digested pGBKT7. To generate pGADT7SNX27⌬RA, pCIneoSNX27 was digested with XhoI/XbaI, and the 730-bp fragment containing the Nterminal portion of the protein was subcloned in pCDNA3 digested with XhoI/XbaI (pCDNA3SNX27⌬RA). pCDNA3SNX27⌬RA was digested with XbaI, blunted, and EcoRI-digested; the 730-bp fragment was subcloned in pGADT7 digested with EcoRI/SmaI.
Cell Lines and Transient Transfection-The rat basophilic leukemia mast cell line was provided by Dr. S. Corbalá n García (Departamento de Bioquímica y Biología Molecular, Universidad de Murcia, Murcia, Spain). The following cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA): K562 human chronic myelogenous leukemia, the CTLL2 mouse cytotoxic cell line, the EL4 mouse T lymphoma, Jurkat human acute T cell leukemia, and HEK293/HEK293T human embryonic cell lines. Thymocytes were obtained from BALB/c mice following standard protocols, and dendritic cells were from Dr. C. F. Ardavín (Centro Nacional de Biotecnología/CSIC, Madrid, Spain). Jurkat, HEK293, and HEK293T cell lines were cultured in Dulbecco's modified Eagle's medium (BioWhittaker, Walkersville, MD) supplemented with 10% FCS (Sigma) and 2 mM glutamine (37°C, 5% CO 2 ). Jurkat cells in logarithmic growth phase were transfected (1.2 ϫ 10 7 in 400 ml of complete medium) with 20 g of plasmid DNA by electroporation with a Gene Pulser (Bio-Rad; 270 V, 975 microfarads); cells were immediately transferred to 10 ml of complete medium and assayed after 24 h. HEK293T and HEK293 cells were transfected using Jet-PEI reagent (PolyPlus Transfection, Illkirch, France) and Lipofectamine Plus (Invitrogen), respectively.
Coomassie Blue-stained bands were excised manually from gels, deposited in 96-well plates, and processed automatically in an Investigator ProGest protein digestion station (Genomics Solutions, Cambridgeshire, UK) where samples were in-gel-reduced, alkylated with iodoacetamide, and trypsin-digested (37). Resulting peptides were analyzed by MALDI-TOF MS; 0.3 l of matrix solution (5 mg/ml 2,5-dihydrobenzoic acid in 33% (v/v) aqueous acetonitrile and 0.1% (v/v) trifluoroacetic acid) was added to an AnchorChip MALDI target (Bruker Daltonics GmbH, Bremen, Germany) and allowed to dry at room temperature. A 0.3-l aliquot of each peptide mixture was then deposited onto matrix surfaces and dried at room temperature.
MALDI mass spectra were acquired automatically on a Bruker Reflex IV MALDI-TOF mass spectrometer (Bruker Daltonics) by Flex-Control 1.1 software and processed by Xtof 5.1.1 software to analyze raw data. Each spectrum was calibrated internally with two trypsin autolysis reference ions, specifically 842.510-and 2211.105-Da peptides, to reach a typical mass measurement accuracy of Ϯ30 ppm in the 800 -3000 m/z range. All known contaminants were excluded during the process. The parameters used to analyze data were a signal-to-noise threshold of 20 and resolution higher than 4000.
Database Search-For protein identification, tryptic peptide masses were transferred to the BioTools 2.0 interface (Bruker Daltonics) to search in the National Center for Biotechnology non-redundant (NCBInr) database using Mascot software (Matrix Science, London, UK). Search parameters were set as follows: carbamidomethyl cysteine as fixed modification by the treatment with iodoacetamide, oxidized methionines as variable modification, peptide mass tolerance of 80 ppm, and one missed cleavage site. In all protein identifications, the probability Mowse scores were greater than the minimum score fixed as significant (78 in all cases) with a p value less than 0.05.
Yeast Two-hybrid Assay-DGK constructs were subcloned in pG-BKT7 vector fused to GAL4BD as bait, and SNX27⌬RA was sub-cloned into pGADT7 vector fused to GAL4AD. The interaction assay was developed according to the manufacturer's protocols (Clontech). The AH109 yeast strain was co-transformed by the LiAc method with pGADT7SNX27⌬RA with each of the DGK constructs or with control empty vector. To select co-transformed yeast, cells were plated on SD medium lacking leucine and tryptophan. Growing colonies were replated on high stringency SD medium lacking leucine, tryptophan, alanine, and histidine plus 5-bromo-4-chloro-3-indolyl-␣-D-galactopyranoside (X-␣-gal) to confirm interacting proteins.
Subcellular Fractionation-We performed subcellular fractionation of Jurkat cells as described previously (38). Briefly 3 ϫ 10 7 Jurkat cells in logarithmic growth phase were washed twice with PBS at 4°C and harvested by centrifugation. Cells were resuspended in homogenization buffer A (250 mM sucrose, 20 mM Hepes, pH 7.4, 1 mM EDTA, leupeptin, pepstatin, aprotinin, 50 mM NaF, 50 mM glycerophosphate, 1 mM orthovanadate, and 1 mM PMSF) and then disrupted using a 23-gauge needle. Whole cells and nuclei were removed by centrifugation (800 ϫ g, 10 min, 4°C). All subsequent manipulations were performed at 4°C. The postnuclear supernatant was centrifuged (20,000 ϫ g, 20 min, 4°C), and the high density microsome fraction was pelleted from the resulting supernatant by centrifugation (45,000 ϫ g, 30 min, 4°C). Low density microsomes were collected from the resulting supernatant by further centrifugation (180,000 ϫ g, 90 min, 4°C). The supernatant from this last centrifugation contained the cytosolic fraction. The pellet resulting from the 20,000 ϫ g centrifugation contained the crude plasma membrane; it was collected and resuspended in buffer A, overlaid on 1 ml of 35% sucrose prepared in buffer A, and then centrifuged (100,000 ϫ g, 1 h). The purified plasma membrane was collected from the top of the interphase, mixed with buffer B (buffer A without sucrose), and concentrated by centrifugation (108,000 ϫ g, 40 min, 4°C). All pellets were resuspended in the same volume of buffer A using a 25-gauge needle. Samples were analyzed by SDS-PAGE, loading the same volume for each fraction (the cytosol sample represented 1 ⁄25 of the total cytosol).
Immunoprecipitation and Western Blot-Jurkat or HEK293 cells, transiently transfected with selected plasmids, were lysed in Nonidet P-40 buffer and cleared by centrifugation. Protein lysates (400 g) were incubated with the indicated antibodies (2 l, 1 h, 4°C) followed by G protein coupled to Sepharose (1 h, 4°C). Immunoprecipitated complexes were washed, and proteins were separated by 7.5% SDS-PAGE, transferred to nitrocellulose membrane, incubated with appropriate antibodies, and developed using the ECL detection kit (Amersham Biosciences). For immunoprecipitation of endogenous DGK, Jurkat cells were lysed in buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, and protease inhibitors. Protein lysates (1 mg) were precleared for 30 min using 50 l of blocked G protein-Sepharose, and the supernatants were incubated with anti-DGK Ab overnight at 4°C. The immunoprecipitated proteins were collected using G protein coupled to Sepharose (2 h, 4°C) and washed three times with washing buffer. Immunoprecipitated complexes were separated by 10% SDS-PAGE, transferred to nitrocellulose membrane, incubated with appropriate antibodies, and developed using the ECL detection kit (Amersham Biosciences).
Immunofluorescence and Confocal Microscopy-At 24 h posttransfection, cells were transferred to poly(DL-lysine)-coated coverslips and allowed to attach for 30 min. Where indicated, attached cells were treated with wortmannin (10 M, 1 h, 37°C) or serum-starved for 30 min and then incubated with Tf-Rhod (20 g/ml, 15-30 min, 37°C). Cells were then fixed in cold methanol and washed with PBS. Primary antibodies (diluted 1:100 in PBS with 3% FCS) were incubated (1 h, 37°C) and washed with PBS; the same procedure was followed for secondary antibodies. Cells mounted on glass slides were imaged with an Olympus Fluoview FV-1000 laser-scanning confocal microscope. A 488 nm krypton-argon laser line was used to record images of GFP-coupled proteins and Alexa 488 staining, 543 nm helium-neon was used for Cy3 and Tf-Rhod, and 633 nm helium-neon was used for Cy5. Images were processed using Adobe Photoshop software.
Small Interfering RNA of DGK-A 64-bp double strand DNA oligonucleotide encompassing an interfering 21-nucleotide sequence of human DGK (2290 -2310) and a hairpin structure were cloned in the pSUPER (pSUPER-RNAiDGK) and the pSUPERGFP (pSUPERGFP-RNAiDGK) vectors (Oligoengine). We previously used this sequence to down-regulate DGK in HEK293 cells (20). The pSUPER constructs were transfected in Jurkat cells by electroporation. Cell samples were collected from days 1 to 5 post-transfection, and DGK levels were evaluated by Western blot (WB) to select the optimal time (96 h) of DGK down-regulation. In the case of the pSUPERGFP construct, GFP-positive Jurkat cells were isolated by sorting (Altra Hypersor, Beckman Coulter). pSUPER empty vectors or pSUPER constructs coding scrambled small interfering RNA were used as controls.
Transferrin Recycling-Recycling experiments were as described previously (39 -41) with some modifications. Briefly cells were incubated in serum-free medium containing 0.1% BSA (1 h, 37°C) to remove residual transferrin and then transferred to 4°C. After 10 min, cells were seeded on poly(DL-lysine)-coated coverslips where samples were labeled (25 min) with 20 g/ml Tf-Rhod in the assay medium. Cells were then washed twice with cold assay medium to remove uninternalized Tf-Rhod, and one aliquot was maintained at 4°C in cold serum-supplemented medium as control. The remaining cells were incubated in serum-supplemented medium at 37°C for different time periods to measure Tf-Rhod recycling as the loss of cell-associated fluorescence. At indicated times, samples were fixed with cold methanol, and cells with pericentriolar Tf-Rhod were counted by confocal microscopy analysis (TCS-NT, Leica, Wetzlar, Germany). z series optical sections (1.25 m) were recorded. Four contiguous optical sections were stacked using ImageJ software and contained all the three-dimensional fluorescence information. All images were acquired with invariable adjusted laser beam and photomultipliers. We calculated the percentage of cells with pericentriolar Tf-Rhod relative to total cell number (registered in bright field images).

Identification of SNX27 by MS Analysis of DGK-associated
Proteins-To identify DGK-interacting proteins, we generated plasmids encoding full-length (FL) DGK or a C-terminal (CT) region of the protein (encompassing the Ank repeats and the PDZbm) fused to GST (Fig. 1A). HEK293T cells were transfected with each construct or with empty vector, and expression of the fusion proteins was assessed by WB. Both recombinant proteins were recognized by GST Ab and showed the predicted molecular weight (Fig. 1B). GST-fused proteins were batch-purified using glutathione-Sepharose beads. After extensive washing, recombinant proteins and their associated proteins were eluted with Laemmli buffer and separated by SDS-PAGE. Coomassie Blue gel staining showed nonspecific bands associated with glutathione-Sepharose in control cells (Fig. 1C, Mock). DGKFL and DGKCT pulled down a ϳ60-kDa double band that was absent in controls (Fig. 1C). Bands were excised from gels and identified by MS (see "Experimental Procedures").
Proteomics analysis of the excised bands with MALDI-TOF showed several peptides matching the SNX27 sequence. SNX27 comprises two variants, a and b, originally identified as products generated by alternative splicing of the mrt1 gene in rat. SNX27a is expressed constitutively in brain and testis, whereas the b isoform is induced in brain following methamphetamine treatment (42). Both isoforms are identical except in their last amino acids, two for the b isoform and 13 for the a isoform (Fig. 1D, bottom, boxed). For band 1 (Fig. 1C), 20 of 28 mass values matched with SNX27a (accession number gi:55960553), covering 37% of the protein sequence (Fig. 1D, left, underlined). For band 2 (Fig. 1C), 24 of 31 mass values searched matched with SNX27b sequence (accession number gi:31742501), spanning 44% of the full-length protein (Fig.  1D, right, underlined). In both cases, experimentally measured peptides masses were compared with the theoretical SNX27 sequences using FindMod software package from ExPASy Proteomics Server (www.expasy.org). This analysis showed that one of the experimental peptides corresponding to band 1 matched exclusively with the SNX27a sequence; this peptide was within the last 13-amino acid sequence specific to this splice variant.
DGK Interacts with SNX27a and -b in HEK293 Cells-SNX27 is the only member of the SNX family that, in addition to the PX domain, has a Ras association (RA) domain and a class I PDZ domain. It is noteworthy that SNX27 was pulled down with DGK constructs with a common C-terminal region containing a PDZbm; this prompted us to analyze the requirement for the DGK PDZbm in this interaction.
HEK293 cells were transfected with empty vector, DGKFL, or a DGK mutant with deletion of the C-terminal PDZbm (DGK⌬PDZbm) fused to a Myc tag ( Fig. 2A). Ectopically expressed proteins were immunoprecipitated with an anti-Myc Ab, and association of endogenous SNX27 was determined by WB. The SNX27 Ab revealed a double band both in lysates and in DGKFL immunoprecipitates but not in DGK⌬PDZbm immunoprecipitates (Fig. 2B). This doublet probably corresponds to the two SNX27 splice variants (42), although the existence of post-translational protein modifications such as phosphorylation cannot be excluded. We further analyzed DGK association with each SNX27 isoform independently by co-immunoprecipitation of the overexpressed proteins. HEK293 cells were co-transfected with GFP-fused SNX27a and with Myc-fused DGKFL, DGK⌬PDZbm, or control empty plasmid (Fig. 2C). Immunoprecipitation with anti-Myc Ab showed association of ectopically expressed SNX27a in DGKFL but not in DGK⌬PDZbm immunoprecipitates (Fig. 2C). Similar data were obtained when GFPfused DGK constructs or control vector was co-transfected with Myc-tagged SNX27b, immunoprecipitated with anti-GFP, and developed using anti-Myc and anti-SNX27 Abs (Fig. 2D); as shown, a doublet corresponding to endoge-nous and ectopically expressed SNX27 was found only when we used anti-SNX27 Ab. These results demonstrate that the DGK PDZbm is indispensable for its interaction with SNX27 isoforms.
To further assess the requirement of the SNX27 conserved domain for interaction with DGK, we co-expressed Myctagged SNX27b constructs bearing different deletions (Fig.  2E) and GFP-tagged DGKFL or GFP control and immunoprecipitated with anti-GFP Ab. SNX27b was not detected in control anti-GFP pellets (not shown). All SNX27 recombinant proteins with an intact PDZ domain co-immunoprecipitated with DGK (Fig. 2E). In contrast, the SNX27 mutant bearing a PDZ domain deletion was not found in DGK immunoprecipitates (Fig. 2F). All together, these results show that DGK interacts, through its C-terminal PDZbm, with the PDZ domain of constitutive and inducible SNX27 forms.
Direct PDZ-mediated Interaction of SNX27 with DGK-Coimmunoprecipitation experiments suggested direct interac-  Fig. 1C). C, HEK293 cells were co-transfected with GFP-SNX27a and the indicated Myc-tagged DGK constructs; after 24 h, cell lysates were immunoprecipitated with anti-Myc Ab. Proteins were resolved, and the indicated proteins were developed in WB using anti-Myc (top) and -GFP Ab (bottom). D, HEK293 cells were co-transfected with Myc-SNX27b and the indicated GFP-DGK constructs; after 24 h, cell extracts were immunoprecipitated with anti-GFP Ab, immunoprecipitates were resolved by SDS-PAGE, and the indicated proteins were developed in WB using anti-GFP (top) and anti-Myc Abs (middle). Transfected and endogenous SNX27 were identified with anti-SNX27 Ab (bottom). In B and D, DGK, but not DGK⌬PDZbm, associated with both endogenous (solid arrowhead) and ectopically expressed SNX27 (open arrowhead). E, HEK293 cells were co-transfected with GFP-DGK and Myc-SNX27 constructs bearing deletions of different domains; after 24 h, we determined the presence of the constructs in DGK immunoprecipitates by WB using anti-GFP (top) and -Myc Abs (bottom). F, HEK293 cells were co-transfected with Myc-DGK and GFP, GFP-SNX27a, or a mutant bearing a deletion of the PDZ domain (GFPSNX27⌬PDZ). SNX27 in anti-Myc immunoprecipitates was assessed as in E. In E and F, only the SNX27 construct with an intact PDZ domain was able to associate with DGK. All experiments were repeated three times with similar results. Input (Lysates) represents 5% of starting material. IP, immunoprecipitate. tion between DGK and SNX27. To rule out the possible participation of an intermediate protein, we analyzed the DGK-SNX27 interaction by yeast two-hybrid assays. Four DGK constructs encompassing different portions of the protein were subcloned in the pGBKT7 vector fused to the GAL4BD, whereas SNX27⌬RA was subcloned in pGADT7 (Fig. 3A, top). All constructs were analyzed for normal protein expression in yeast (not shown). pGADT7SNX27⌬RA was cotransformed in the AH109 yeast strain with each DGK construct or empty pGBKT7 as control (Fig. 3A, bottom). Cells were grown on SD medium lacking leucine and tryptophan (Fig. 3B, left); co-transformed yeast were then plated on highly stringent SD medium lacking leucine, tryptophan, adenine, and histidine plus X-␣-Gal (Fig. 3B, right). Those DGK constructs lacking the complete CT region (⌬Ank) or the PDZbm (⌬PDZbm) were unable to interact with SNX27⌬RA, whereas strong interaction was observed for the constructs encoding the complete protein or the CT region (Fig. 3B, right). These results strongly suggest the direct association of DGK and SNX27 via PDZ interaction.
SNX27 Is Expressed in Cells of Hematopoietic Origin-Previous studies using rat tissues have shown that mRNAs for SNX27a and -b are expressed predominantly in brain and testis (42), but protein expression has not been fully assessed. We investigated expression of endogenous SNX27 in hematopoietic cells. WB analysis in distinct cells of hematopoietic origin demonstrated SNX27 expression in all cell types evaluated (Fig. 4A).
We performed new co-immunoprecipitation assays to determine whether ectopically expressed SNX27 also associated with DGK in Jurkat T lymphocytes as it does in HEK293 cells. Cells were co-transfected with plasmids expressing GFP (control) or GFP-SNX27 and Myc-tagged DGKFL, DGK⌬PDZbm, or empty plasmid. Analysis of Myc immunoprecipitates showed that both endogenous and ectopically expressed SNX27 associated exclusively with DGKFL but not with DGK⌬PDZbm (Fig. 4B). These results confirm that the PDZ-specific interaction between SNX27 and DGK is also found in T lymphocytes.
Analysis of endogenous protein interaction was largely impaired by the lack of suitable antibodies. Although the anti-SNX27 Ab was very efficient in WB, it bound the PDZ region, making immunoprecipitation-based assessment of PDZ-mediated interaction impossible. This is also true for most anti-DGK Abs, which were specific for a C-terminal peptide encompassing the ETAV sequence that mediates SNX27 interaction. Immunoprecipitation experiments were thus performed with an Ab raised against the DGK N-terminal domain; albeit not very efficient, this Ab permitted detection of a weak band with a molecular weight corresponding to that of SNX27 (Fig. 4C). Detailed analysis of the interaction of the endogenous proteins must thus await the generation of improved tools for immunoprecipitation.
As an alternative approach to assess co-localization of endogenous proteins, we examined the subcellular expression pattern of both proteins in different T lymphocyte fractions. The cytosolic fraction contained very low levels of DGK, whereas SNX27 and EEA1 were abundantly found suggesting, as described previously (46), that binding to 3-phosphorylated PI does not sustain recruitment to endosomes. DGK was absent from the plasma membrane fraction where a very faint SNX27 band was observed. SNX27 colocalized with DGK in internal membrane fractions corresponding to high and low density microsomes (Fig. 4D), suggesting interaction between these proteins at these subcellular localizations.
PX and PDZ Domains Are Responsible for SNX27 Vesicular Association in T Lymphocytes-As the fractionation experiments indicated no specific SNX27 location, we examined its exact localization in intact cells. Immunofluorescence (IF) analysis of Jurkat T cells showed that endogenous SNX27 localized to cytosol with accumulation in vesicular structures (Fig. 5A). To analyze the role of SNX27 domains in the subcellular localization of this protein, we generated several deletion mutants, which we transfected into Jurkat T cells and analyzed by IF. The subcellular localization pattern of Myc- FIG. 3. Characterization of the DGK-SNX27 interaction using the yeast two-hybrid system. A, four DGK constructs were subcloned in the pGBKT7 vector fused to GAL4BD. The N-terminal region of SNX27 (containing the PDZ and PX domains) was subcloned in pGADT7 fused to GAL4AD (top). The AH109 yeast strain was cotransformed with pGADT7SNX27⌬RA and the indicated DGK constructs or control empty vector (bottom). B, co-transformed yeast were plated on leucine-and tryptophan-free SD medium (SDϪLW, left). To confirm interacting proteins, co-transformed yeast were grown on highly stringent medium lacking leucine, tryptophan, alanine, and histidine (SDϪAHLW) plus X-␣-Gal (right). tagged SNX27b (MycSNX27FL) and of a mutant lacking the RA domain (MycSNX27⌬RA) was similar to that of the endogenous protein, suggesting that the RA domain is not necessary for vesicular localization (Fig. 5B, top). Vesicle colocalization was lost following deletion of the PX domain (MycSNX27⌬PX) as was also the case when the RA domain was expressed alone (MycSNX27RA) (Fig. 5B, bottom). A fusion protein N-terminally labeled with GFP showed vesicular distribution similar to that of the endogenous protein (Fig. 5C), and vesicle colocalization was partially lost after PDZ deletion (GFPSNX27⌬PDZ) (Fig. 5C). The data confirm that the SNX27

FIG. 4. SNX27 is expressed in cells of hematopoietic origin.
A, different hematopoietic cell types were lysed, separated by SDS-PAGE, and analyzed by WB with anti-SNX27 Ab; anti-tubulin Ab was used as loading control. RBL, rat mast cell line; Jurkat, human acute T cell leukemia; K562, human chronic myelogenous leukemia; CTLL2, mouse cytotoxic cell line; EL4, mouse T lymphoma; Thym., mouse thymocytes; DC, mouse dendritic cells. B, Jurkat T cells were cotransfected with GFP or GFP-SNX27 and either Myc-DGK or a mutant lacking the PDZbm, and cells were processed as in Fig. 2. Myc-DGK but not Myc-DGK⌬PDZbm immunoprecipitates showed association of endogenous (solid arrowhead) and ectopically expressed SNX27 (open arrowhead). C, Jurkat cell lysates were immunoprecipitated with either preimmune (PI) serum or with an anti-DGK Ab to the N-terminal domain of the protein, and endogenous SNX27 was assessed in the immunoprecipitates as described in Fig. 2. Lysates represent 5% of starting material. D, subcellular fractions of Jurkat T cells were purified by high speed centrifugation (see "Experimental Procedures"), and the presence of DGK and SNX27 was determined in WB with appropriate antibodies. IP, immunoprecipitate; HDM, high density microsome fraction; LDM, low density microsome fraction.

FIG. 5. Vesicular targeting of SNX27 in Jurkat T cells is PX and PDZ domain-dependent.
A, Jurkat T cells were fixed and then stained with rabbit anti-SNX27 Ab followed by anti-rabbit IgG-Alexa 488. B, Jurkat T cells were transfected with the indicated Myc-SNX27b constructs; after 24 h, cells were fixed and stained for IF using anti-Myc and anti-mouse IgG-Cy3 antibodies. Myc-SNX27FL and Myc-SNX27⌬RA showed a punctate pattern that accumulated in a specific cytosolic location, whereas Myc-SNX27⌬PX and Myc-SNX27RA did not localize in vesicles. C, Jurkat T cells were transiently transfected with a GFP-SNX27 construct or a mutant with deletion of the PDZ domain (GFP-SNX27⌬PDZ). After 24 h, cells were fixed and analyzed by confocal microscopy. The GFP-SNX27 construct showed the same expression pattern as the endogenous protein; GFP-SNX27⌬PDZ showed less vesicle localization than the FL counterpart. Bars, 3 m.
PX domain is indispensable for its vesicular location, as reported for other SNX family members (46), and demonstrate that vesicle localization of SNX27 is not as efficient without an intact PDZ domain.
SNX27 Is Present in the Recycling Endocytic Pathway in T Lymphocytes-The SNX family members are proposed to participate in various aspects of endocytosis and membrane trafficking. Endocytic pathways control the internalization of extracellular components, a mechanism that is essential in the regulation of diverse signal transduction pathways. The best characterized endocytic process is receptor-mediated endo-cytosis, which involves receptor internalization by clathrincoated vesicles. Receptors are delivered to sorting endosomes, which can follow one of at least three pathways. One is the prodegradative pathway through late endosomes and lysosomes, another is the Golgi network, and the third is the endocytic recycling pathway in which receptors can return to the cell surface directly or through the endocytic recycling compartment (ERC) (30,31).
To elucidate the exact nature of the SNX27-positive vesicles, we performed co-localization analyses of GFP-SNX27 with markers specific for each sorting pathway. We used

FIG. 6. SNX27 is found in sorting endosomes in Jurkat T cells.
Cells were transfected with GFP-SNX27, fixed after 24 h, and processed for labeling of vesicular markers with primary Ab followed by anti-mouse IgG-Cy3 antibodies. A, LAMP1 and CD63 were used to label late endosome/lysosome and immature vesicles (top and middle rows, respectively); GM130 was used to detect cis-Golgi (bottom row). Merged images (third column) are magnified to show detail (fourth column). None of these vesicular markers co-localized with GFP-SNX27. B, EEA1 (top row) or SNX2 Ab (bottom row) were used to stain sorting endosomes. Yellow staining indicates co-localization of GFP-SNX27 and the vesicular markers. Note that GFP-SNX27 co-localized with both markers but that some additional GFP-SNX27-labeled vesicles were negative for sorting endosome staining. Bars, 3 m. were fixed and stained with anti-TfR Ab followed by anti-mouse IgG-Cy3 Ab (top row) or were treated with 20 g/ml Tf-Rhod, then fixed, and analyzed by confocal microscopy (middle row). Untransfected Jurkat T cells were Tf-Rhod-treated, fixed, and then stained with anti-SNX27 and anti-rabbit IgG-Alexa488 Abs (bottom row). Merged images (third column) are magnified to show detail (fourth column). Immunofluorescence analyses showed strong co-localization (yellow) between SNX27 and TfR/Tf-Rhod. B, Jurkat T cells were processed 24 h after GFP-SNX27 transfection for staining with anti-TfR (top) or were treated with Tf-Rhod (bottom) as in A. Untransfected control cells are indicated with an asterisk. Immunofluorescence analysis revealed no specific TfR or Tf-Rhod accumulation in cells overexpressing GFP-SNX27. LAMP1 and CD63 as late endosome/lysosome markers (48,49) (Fig. 6A, top and middle rows) and GM130 as a cis-Golgi marker (50) (Fig. 6A, bottom row). SNX27 did not co-localize with any of these markers (Fig. 6A, third and fourth columns), indicating that it is absent from late endosomes/lysosomes and the Golgi network. In contrast, SNX27 clearly co-localized with proteins of the sorting endosomes, such as EEA1 (51) (Fig. 6B, top row) and SNX2, another sorting nexin reported to associate with sorting endosomes (52) (Fig. 6B, bottom row). Higher magnification revealed that, in addition to co-localization with sorting endosomes, GFP-SNX27 was present in other vesicles (Fig. 6B, fourth column). We thus analyzed SNX27 co-localization with TfR, one of the best characterized proteins in the endocytic recycling pathway (53). SNX27 colocalized extensively with steady-state intracellular TfR, suggesting that SNX27 is enriched in the vesicles that transport this receptor (Fig 7A, top row). The TfR can also be tracked by labeling with Tf-Rhod. SNX27 co-localization with Tf-Rhod (Fig.  7A, middle row) confirmed that SNX27 localizes in the endocytic recycling pathway in T lymphocytes. Results were similar when endogenous SNX27 was analyzed (Fig. 7A, bottom row). Overexpression of some SNXs may perturb endosomal structures, leading to increased TfR staining, thereby artificially affecting receptor trafficking (46). To test whether SNX27 overexpression altered transferrin transport through the endocytic recycling pathway, we compared GFP-SNX27-transfected versus control cells (Fig. 7B). We found no differences in accumulation of TfR (top) or Tf-Rhod (bottom), suggesting that SNX27 overexpression does not alter endosomal structures in this system. FIG. 8. Vesicle association of SNX27 is wortmannin-sensitive and requires DGK interaction. A, Jurkat T cells were transfected with GFP-SNX27; after 24 h, cells incubated with Tf-Rhod were treated with 10 M wortmannin and then fixed for immunofluorescence analysis. Vesicular GFP-SNX27 localization was lost after wortmannin treatment, resulting in diffuse cytosolic staining. Bars, 3 m. B, Jurkat cells were transfected with GFP-SNX27 or GFP-SNX27⌬PDZ; after 24 h, cells were treated with 20 g/ml Tf-Rhod, fixed, and stained with anti-Rab11 followed by anti-rabbit IgG-Cy5 Ab. GFP-SNX27, green, first column; Tf-Rhod, red, second column; Rab11, blue, third column. In merge images (fourth column), yellow staining indicates GFP-SNX27 and Tf-Rhod co-localization, purple staining shows Tf-Rhod and Rab11 co-localization, and cyan indicates GFP-SNX27 and Rab11 co-localization. White merge signal shows GFP-SNX27, Tf-Rhod, and Rab11 co-localization. Bars, 3 m. C, Jurkat cells were co-transfected with GFP-SNX27 and a plasmid encoding the CT region of DGK fused to an HA tag. 24 h after transfection cells were fixed and stained with anti-HA Ab followed by anti-mouse IgG-Cy3. The reduced vesicle localization of SNX27 in cells expressing both proteins is compared with its localization in cells not expressing the HA-DGK construct. (1 and 2 are two different optical z sections.) and PDZ domains in vesicle localization of the protein. SNX27-positive vesicles co-localized extensively with steady-state intracellular TfR, suggesting that SNX27 is enriched in the vesicles that transport this receptor. The products of PI3K may target PX-containing proteins to specific subcellular locations. To study the 3-phosphorylated PI requirement in SNX27 vesicular localization, we examined GFP-tagged SNX27/Tf-Rhod distribution in Jurkat T cells treated with the PI3K inhibitor wortmannin (54). Under this treatment, endosomal structures were profoundly modified, and GFP-SNX27 redistributed promptly to cytosol (Fig. 8A), confirming that SNX27 vesicular association depends on PI3K products.

Phosphoinositide-3-OH Kinase (PI3K) Activity and DGK Interaction Regulate SNX27 Binding to Vesicles in T Lymphocytes-Previous experiments indicated a role for SNX27 PX
Rab11 is a small GTPase localized in the ERC that is essential for TfR transport to the membrane (55,56). To confirm SNX27 participation in the endocytic recycling pathway and to further assess the role of the SNX27 PDZ domain in the subcellular localization of this protein, we determined SNX27 or SNX27⌬PDZ and Rab11 localization in Tf-Rhod-treated cells. GFP-SNX27 clearly co-localized with the Tf-Rhod-and Rab11-positive structures (Fig. 8B). SNX27⌬PDZ and Rab11 fluorescent signal overlap was less pronounced, suggesting that lack of this domain prevented enrichment of SNX27 in TfR-containing vesicles.
These experiments support our observations with deletion mutants, showing that although the PX domain is essential for SNX27 binding to vesicles the PDZ domain is also required for efficient association. To further study DGK participation in SNX27 binding to vesicles, we disrupted interaction with endogenous DGK by overexpressing its C-terminal domain. A construct encompassing the DGK Ank repeats and PDZbm showed no specific localization (Fig. 8C). SNX27 binding to vesicles was diminished in cells expressing this construct, suggesting that disruption of interaction with endogenous DGK affected SNX27 binding to vesicles (Fig. 8C).
DGK Depletion in Jurkat T Cells Accelerates TfR Exit from the ERC-Our studies demonstrated that, in T cells, DGK interacts with SNX27 and that SNX27 localizes in TfR-positive vesicles. Deletion of the PDZ domain altered SNX27 vesicle co-localization as did overexpression of the DGK C-terminal domain. These experiments suggested that SNX27 and DGK interaction promotes enrichment of the former in TfR-positive vesicles. Co-localization of endogenous DGK with SNX27 and/or TfR-positive vesicles was difficult to determine due to the lack of an anti-DGK Ab suitable for IF analysis. In addition, ectopically expressed GFPDGK has a broad expression pattern in T lymphocytes, making it difficult to assess its vesicular localization during TfR trafficking. To assess the functional role of the SNX27-DGK interaction, we therefore evaluated possible DGK function in TfR recycling. We analyzed TfR recycling in T lymphocytes in which DGK expression was knocked down using RNA interference (RNAi).
We transfected Jurkat T cells with plasmids encoding a small hairpin RNAi (pSUPER-RNAiDGK or pSUPERGFP-RNAiDGK) and assessed down-regulation of endogenous protein by WB (Fig. 9A). RNAi efficiently down-regulated DGK levels expressed in T cells (57). Transfected T cells were labeled with FIG. 9. DGK depletion accelerates the kinetics of Tf-Rhod exit from the ERC. A, a 64-bp double strand DNA oligonucleotide encompassing an interfering 21-nucleotide sequence of human DGK (2290 -2310) (left) and a hairpin structure were cloned in pSUPER. pSUPER (control) or pSUPER-RNAiDGK were transfected in Jurkat cells by electroporation. Cell samples were taken 96 h after transfection, and DGK levels were evaluated by WB using anti-DGK Ab. Anti-tubulin Ab was used as loading control. RNAiDGK efficiently decreased DGK expression (right). B, to measure the rate of Tf-Rhod recycling exit from the ERC, Jurkat T cells were transfected with empty plasmid or pSUPER-RNAiDGK; after 96 h, cells were treated with Tf-Rhod at 4°C, washed, and transferred to 37°C to count the number of cells with pericentriolar Tf-Rhod at different times after sample transfer. C, the number of control or DGK-knocked down cells with pericentriolar Tf-Rhod, counted at the times indicated, is expressed as a percentage of total cells. Mean Ϯ S.D. is shown for three independent experiments. Tf-Rhod (see "Experimental Procedures"), and staining was analyzed at different times. Even at time 0 (when cells were transferred from 4 to 37°C), we observed strong pericentriolar Tf-Rhod staining, suggesting that TfR internalization in Jurkat T cells is very rapid compared with other cell types (39). Exit from the pericentriolar compartment was reflected by loss of fluorescence (Fig. 9B). At time 0, pericentriolar staining was similar in controls and cells with down-modulated DGK (Fig.  9, B and C). In contrast, the percentage of pericentriolar Tf-Rhod at 30 and 50 min was lower in cells with diminished DGK compared with controls. Similar data were obtained when we enriched the DGK-knocked down cell population using the GFP version of the RNAi plasmid. Analysis of three independent experiments indicated that down-regulation of DGK protein levels increased the rate of TfR recycling back to the cell surface (Fig. 9, B and C). These results demonstrate that DGK acts as a negative modulator of TfR recycling from ERC to the plasma membrane. DISCUSSION DGK is expressed ubiquitously, and its functions range from cell cycle regulation to cytoskeletal remodeling and/or regulation of immune function (20 -24). DGK localization to specialized membrane regions is a general mechanism for its functional regulation, and the presence of well characterized protein-protein interaction domains suggests that DGK can participate in various signaling pathways. Using a proteomics approach to detect novel DGK-interacting proteins, we identified SNX27. We demonstrated that DGK interacts with the SNX27 PDZ domain, and we describe previously unreported functions for this isoform in the regulation of membrane trafficking.
Videomicroscopy studies in living Jurkat T cells showed that DGK translocates to the plasma membrane in response to activation of an exogenously expressed muscarinic type I receptor (25). These studies showed that the C-terminal region of the protein confers specificity for DGK translocation in response to different receptors, suggesting that membrane localization of DGK would require C-terminal region-mediated interaction with specific scaffold proteins (25). This hypothesis was confirmed by reports that identified DGK-associated proteins, indicating that PDZ-mediated interaction is central to DGK localization to and function in specific membranes (26,33,58). In neurons, for instance, PDZ-mediated interaction with syntrophins regulates DGK translocation to the plasma membrane where it translates receptor signals into the cytoskeletal rearrangements required for neurite outgrowth (23). The identification of SNX27 as a DGK partner reveals a novel PDZ-dependent interaction, suggesting that DGK acts in membrane compartments other than the plasma membrane.
The sorting nexin family comprises 29 distinct proteins in mammals and 10 in yeast with roles ranging from prodegradative sorting and internalization to endosomal recycling and/or endosomal sorting (31). This family is characterized by the presence of an SNX PX domain that targets these proteins to 3-phosphorylated PI-enriched membranes. Our experiments demonstrated vesicular localization for both endogenous and ectopically expressed SNX27 in Jurkat T cells similar to SNX27 localization in adherent cells such as A431 (34), HEK293, and HeLa. 2 We also showed that SNX27 depends on its PX domain to localize in TfR-positive vesicles within the sorting endosomes and the ERC. This PX-dependent association was sensitive to wortmannin, confirming that SNX27 subcellular localization has a strict 3-phosphorylated PI requirement as described for other family members (46). In addition, our results showed that the PDZ domain, unique to this SNX isoform, provides additional regulation for vesicle binding. This suggests that PDZ-mediated interactions of SNX27 with protein partners are important for vesicular localization.
Several SNX proteins also have a C-terminal coiled coil region encoding a Bin/amphiphysin/Rvs (BAR) domain (59), which acts as a dimerization and a membrane-binding region, able to sense membrane curvature (60). The presence of both a BAR and a PX domain suggests a complex SNX localization mechanism in which SNX membrane association would be regulated by the degree of physical membrane curvature. Some SNX family members have additional protein-protein interaction domains, including the SH3 (Src homology domain type 3), RGS (regulators of G protein signaling), or RA domain, all of which may regulate protein localization. Based on the variety of domains in the SNX family, some authors speculate that only those SNXs containing a BAR domain would act as a "real" SNX, whereas the family members with protein-protein interaction domains might function as signaling molecules (31). According to this classification, SNX27 would belong to the latter group as it lacks the BAR domain but has a PDZ and an RA domain. SNX27 could thus act as a signaling platform, bringing proteins to vesicular compartments through PDZ-mediated interaction. Recent experiments identified SNX27 as responsible for the specific recruitment of the 5-hydroxytryptamine (4a) receptor to early endosomes, modulating the signaling pathway triggered by this receptor (34). The association of SNX27 with DGK described here points to a similar scenario in which SNX27 localization at 3-phosphorylated PI-enriched membranes would bring DGK into these compartments to exert its function. DGK localization to these structures would in turn contribute to stabilization of SNX27 in vesicles. Concurring with this hypothesis, we showed that disruption of the SNX27-DGK interaction by overexpression of a DGK C-terminal region results in a loss of vesicle localization similar to that observed for the SNX27 lacking the PDZ domain.
The SNX27⅐DGK complex would provide a mechanism for regulating membrane DAG and PA levels in the ERC compartment. Accordingly we showed that DGK-knocked down cells have enhanced TfR recycling to the membrane, suggesting that DGK modulates the endocytic recycling pathway. Although their precise contribution to recycling kinetics remains to be fully elucidated, DAG and PA have well recognized membrane trafficking functions (2,3,9). Recent experiments showed that depletion of PLD 2 , but not of PLD 1 , inhibit the TfR recycling rate without affecting the internalization rate. HeLa cells with reduced PLD 2 show TfR accumulation in a Rab11positive compartment, suggesting a role for PLD 2 in the endocytic recycling to the plasma membrane (41). Independent observations indicate that Arf6-mediated PLD activation is required for TfR and major histocompatibility complex protein class I recycling. Cells overexpressing Arf6 mutants defective in PLD activation inhibit major histocompatibility complex protein class I recycling to the plasma membrane; in these cells, propranolol treatment to increase PA levels by inhibiting phosphatidic acid phosphohydrolase does not rescue recycling, suggesting a requirement for phosphatidic acid phosphohydrolase-derived DAG rather than PA in this process (14). This is consistent with our results using DGK-depleted cells in which acceleration of TfR recycling might be due to DAG level increase and/or PA depletion. In addition to its functions in membrane dynamics, DGK modulation of DAG levels may have other roles, such as decreasing protein activation/localization in the recycling endosome. Classical PKCs, such as PKC␣ and PKC␤II, move to the ERC following their activation; this translocation has functional consequences because PKC regulates the dynamics of endocytosis and transferrin trafficking through the endocytic recycling pathway (61,62). DGK is both a substrate and a potent modulator of PKC␣; in this setting, SNX27-mediated DGK localization to the endosomal system would provide an additional mechanism for modulating PKC-regulated delivery of molecules to and from the ERC.
The endosomal system acts as an intracellular sorting network with a key role in decisions on the fate of internalized cargo. From the plasma membrane, internalized proteins reach the sorting endosomes where trafficking decisions are taken. Nutrient receptors such as TfR release their ligands and return to the plasma membrane. Signaling receptors can be recycled to the plasma membrane or progress to lysosomes where they are degraded (63). Sorting and recycling endosomes not only regulate traffic to and from the cell surface but also represent a major center for receiving and delivering biosynthetic traffic from the Golgi (31). Within the endosomal system, the ERC has an important function in polarized sorting of the endocytic and secretory pathways, promoting release of inflammatory cytokines to newly forming phagocytic cups in macrophages (64) or exocytosis of E-cadherin in epithelial cells (65). Here we characterized SNX27 as a component of the T lymphocyte endocytic recycling system and identified it as a DGK-interacting protein; this led to detection of a role for this enzyme in the regulation of exit traffic from the ERC. ERC-regulated recycling is necessary for efficient T cell receptor accumulation in the immunological synapse (IS). This recycling coincides with the TfR recycling pathway, and its inhibition leads to a reduction in TCR in the IS with a concomitant reduction in signaling (66,67). T lymphocytes from DGK-deficient mice are hyperresponsive to ex vivo TCR stimulation and produce elevated cytokine levels in response to viral infection (24). Based on this observation and on our findings, we thus speculate that the absence of DGK in these mice contributes to the increase in TCR levels in the IS, enhancing the response.
The importance of endosomal trafficking regulation in T lymphocytes is highlighted by recent findings on the physiology of HIV-1-infected cells. Viral infection often profoundly modifies normal cell function to optimize viral propagation and cell survival. The HIV-1 Nef protein induces a severe reduction in TCR and Lck expression at the immunological synapse by inhibiting endocytosis and recycling (67). The identification of SNX27 as a DGK partner allows us to postulate new functions for this lipid kinase family in regulating the many functions ascribed to the endocytic recycling pathway.