|Home | About | Journals | Submit | Contact Us | Français|
Helper and cytotoxic T cells accomplish focused secretion through the clustering of vesicles around the MTOC and translocation of the MTOC to the target contact site. Here, using Jurkat cells and OT-I T cell receptor (TcR) transgenic primary murine CTLs, we show that the dynein-binding proteins NDE1 and dynactin (as represented by p150Glued) form mutually exclusive complexes with dynein, exhibit non-overlapping distributions in target-stimulated cells, and mediate different transport events. When Jurkat cells expressing a dominant negative form of NDE1 (NDE1-EGFP fusion) were activated by SEE-coated Raji cells, NDE1 and dynein failed to accumulate at the immunological synapse (IS) and MTOC translocation was inhibited. Knockdown of NDE1 in Jurkat cells or primary mouse CTLs also inhibited MTOC translocation and CTL-mediated killing. In contrast to NDE1, knockdown of p150Glued, which depleted the alternative dynein-dynactin complex, resulted in impaired accumulation of CTLA-4 and granzyme-B containing intracellular vesicles at the IS, while MTOC translocation was not affected. Depletion of p150Glued in CTLs also inhibited CTL-mediated lysis. We conclude that the NDE1/Lis1 and dynactin complexes separately mediate two key components of T cell focused secretion, namely translocation of the MTOC and lytic granules to the IS, respectively.
T cells typically eliminate pathogens through the cytoskeleton-directed focused secretion of effector molecules (1-3). The importance of secretion to cytotoxic and NK cell function in immunity is seen in primary hemophagocytic and Chediak-Higashi syndromes and in inhibition of cytotoxicity in tumor microenvironments (4-8). Typically, this process takes place in a series of steps beginning with the formation of a specialized T cell: target contact known as the immunological synapse (IS) (9, 10). This is followed by translocation of the microtubule organizing center (MTOC) to the IS which often brings secretory vesicles with it although vesicles can also accumulate after the MTOC has translocated (11-13). Similar mechanisms appear to operate in certain T helper secretory events (14, 15).
At present, the mechanism of MTOC movement towards the synapse is not fully understood and is somewhat controversial. Alternative models of MTOC translocation posit either a dynein- or actin-dependent mechanism for driving MTOC movements. Dynein is a minus end-directed microtubule motor protein that if anchored at the IS could reel in microtubules and pull the MTOC up to the IS (16). Variants of the Dynein-based models either propose that dynein causes microtubules to loop through the IS and continue to slide rearward (11, 17, 18) or that microtubule plus ends depolymerized as they move towards the IS (19), perhaps similar to the model for chromosome-to-pole movements. The actin-based model proposes that microtubules become linked to a patch of newly polymerized actin at center of the IS. As the patch of actin expands to form a peripheral ring, microtubules would be pulled laterally driving the MTOC forwards towards the IS.
Evidence that dynein is involved in MTOC translocation is derived from studies showing that dynein accumulates at the IS following T cell activation, and that siRNA-mediated depletion of the dynein heavy chain blocked MTOC translocation (18). Additionally, in Jurkat cells, reduced expression of ADAP, a scaffolding protein anchored to the IS, led to a loss of dynein at the synapse and failure of MTOC translocation (17). Finally, the small molecule dynein inhibitor, Ciliobrevin, was shown to block MTOC translocation (19).
Dynein is also needed to move secretory vesicles along microtubules towards the MTOC (20). Clustering of vesicles around the MTOC allows their movement en masse with the MTOC as it translocates. On the other hand, after the MTOC has translocated to the IS, vesicles can still move along microtubules from the cell periphery towards the MTOC such that they concentrate at the synapse (12, 21).
That the same dynein motor carries out such distinctly different processes raises the question of how these processes are differentially regulated and coordinated. Dynein is known to form two different complexes, one with NDE1 (Nuclear distribution E; NUD-E) / Lis1 (Lissencephaly 1) (22) and the other with dynactin, a multisubunit complex whose largest subunit is p150Glued (23). We hypothesized that these two different dynein complexes were responsible for different aspects of the secretory process, one for MTOC translocation and the other for vesicle movements.
In this study, we employed the human Jurkat T cell line and OT-I TcR transgenic murine CTLs to examine the role of these two complexes in MTOC translocation, vesicle delivery to the IS, and CTL-mediated target cell lysis. We found that these proteins form mutually exclusive complexes with the dynein intermediate chain (DIC) and differ in their intracellular distributions and transport functions. Upon TcR ligation, the NDE1/Lis1/dynein complex moves to the IS where, similar to dynein and Lis1, it forms a ring-like structure corresponding to the peripheral supramolecular activation cluster (pSMAC) of the IS (10). Depletion of NDE1 or expression of a dominant negative EGFP-NDE1 construct in Jurkat cells prevented dynein accumulation at the IS and blocked MTOC translocation. When NDE1 was depleted in OT-I CTLs, both MTOC translocation and target cell lysis were greatly reduced. In contrast, p150Glued did not form a peripheral ring at the IS and was not required for MTOC translocation in Jurkat cells, but it was needed for clustering of vesicles around the MTOC and for their accumulation at the cSMAC. Furthermore, depletion of p150Glued in OT-I CTLs impaired recruitment of cytotoxic granules to the IS and lysis of target cells.
While NDE1 and p150Glued formed mutually exclusive complexes with the dynein intermediate chain (DIC), we found that both complexes bound to DISC1 (Disrupted in schizophrenia 1). DISC1 is a scaffold protein that in neurons is known to interact with dynein complexes and plays important roles in a variety of nerve cell functions including nuclear movements, neuronal migration, delivery of synaptic vesicles to nerve terminals, cell adhesion, and cell signaling (24-31). In Jurkat cells, DISC1 concentrates at the IS upon TcR engagement and might provide a higher order of regulation for both dynein complexes.
Jurkat T cell line (clone E6-1) and Raji human Burkitt's lymphoma B cell line were obtained from ATCC (Manassas, VA). Cell culture media including RPMI 1640 (cat #31800-014) and Opti-MEM (cat # 31985-070) were obtained from Life Technologies. Colchicine (cat # C9754), mitomycin C (cat # M4287), puromycin (cat # P8833) and bovine serum albumin (cat # A7906) were obtained from Sigma-Aldrich. Heat inactivated fetal bovine serum (FBS) was obtained from Atlanta Biologicals (Flowery Branch, GA, cat # S11550). Cell Tracker Blue was purchased from Invitrogen (cat # C211). Purified Staphylococcus enterotoxin E (SEE) was obtained from Toxin Technology, Inc., Sarasota, Florida (cat # ET404). Ovalbumin fragment (chicken, 257-264 amino acids, SIINKEKL-OH) peptide was purchased from New England Peptide Inc., Gardner, MA (cat # BP10-915).
Jurkat and Raji cells were grown in RPMI 1640 medium supplemented with 10% FBS (v/v). 50 μM β-mercaptoethanol, 24 mM NaHCO3, 1 mM pyruvate, 1 mM glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin. Cells were cultured at 37°C in a 5% CO2 atmosphere.
Antibodies used in this project are rabbit anti-NDE1 antibody (Protein Tech Group, cat #10233-1-AP), rat anti-NDEL1 antibody (a kind gift from Dr. Atsushi Kamiya, John Hopkins University), mouse anti-DIC IgG antibody clone 74.1 (Covance, cat # MMS-400P), mouse anti-DIC IgM antibody clone 70.1 (Sigma-Aldrich, cat # D5167), mouse anti-Lis1 antibody (Sigma-Aldrich, cat # L7391), mouse anti-beta tubulin antibody (Sigma-Aldrich, cat # T4026), mouse anti-p150Glued antibody (BD Transduction Laboratories, cat # 610474), mouse anti-human Vβ8 antibody (BD Pharmingen, cat # 555604), rabbit anti-DISC1 antibody (Boster Biological Technology, cat # PA2023 and Epitomics, cat# S0371), rabbit anti-granzyme B antibody (Abclonal, cat # A-2557), rabbit anti-GFP antibody (Thermo Fisher Scientific, cat # A11122), unconjugated rabbit anti-mouse IgG secondary antibody (Thermo Fisher Scientific, cat # 31190), TRITC conjugated goat anti-mouse IgG-Fc specific antibody (Sigma-Aldrich, cat # T7657), FITC conjugated goat anti-mouse IgG antibody (Sigma-Aldrich, cat # F2012), Alexa Flour 594-conjugated goat anti-rabbit antibody (Thermo Fisher, cat # A-11037), TRITC conjugated phalloidin (Sigma, cat # P1951), HRP conjugated goat anti-mouse IgG antibody (Thermo Fisher Scientific, cat # 31430), HRP conjugated goat anti-rabbit IgG antibody (Santa Cruz Biotechnology, cat # sc-2004).
Total mRNA was isolated from Jurkat cells using GenElute Mammalian Total RNA Miniprep Kit (Sigma, cat # RTN10) and cDNA library was created by reverse transcriptase PCR using MMLV Reverse Transciptase 1st strand cDNA synthesis kit (Epicentre, Madison, WI, cat # MM070110) and stored at −80°C until use.
To create the EGFP-NDE1 construct, NDE1 transcript variant 1 (Accession no. NM_001143979) was amplified from Jurkat cDNA library using the following primers: ‘NDE1_F’ 5’-ATGGAGGACTCCGGAAAGACT-3’, ‘NDE1_R’ 5’-TCAGCAGGAGCTGGACGA-3’. The coding region (CDS) was cloned into pEGFP-C1 plasmid between Kpn1 and Xma1 restriction sites. This led to N-terminal EGFP fused to NDE1 sequence.
To create the NDE1-EGFP construct, the coding sequence from EGFP-NDE1 plasmid was subcloned into pEGFP-N1 plasmid between XhoI and HindIII restriction sites. The plasmid was further modified to disrupt the Kozak consensus sequence at the beginning of EGFP to abolish EGFP only expression from the plasmid. Additionally, a Kozak sequence was added at the beginning of the start codon of NDE1 to enhance its expression in mammalian cells.
To create the NDE1-mEGFP or mEGFP-NDE1 constructs (monomeric EGFP), a point mutation was created at A206K position of EGFP of the preceding plasmid following Stratagene site directed point mutagenesis protocol using the following primers: ‘mEGFPa206k_F’ 5’-CTGAGCACCCAGTCCAAACTGAGCAAAGACCCC-3’ and ‘mEGFPa206k_R’ 5’-GGGGTCTTTGCTCAGTTTGGACTGGGTGCTCAG-3’. The presence of the NDE1 coding region in each plasmid was confirmed by sequencing.
The mouse CTLA-4-YFP construct was a gift from Dr. James Allison, M. D. Anderson Cancer Center, Houston, TX. To obtain CTLA-4-mCherry, the CTLA-4 coding sequence was sub-cloned into pIRES-PURO3-mCherry vector between ClaI and NheI restriction enzyme sites. pIRES-PURO3-mCherry plasmid was kindly provided by Dr. Roger Tsien, University of California-San Diego, CA and was subsequently modified to insert several restriction enzyme sites following standard PCR protocol. mCherry was placed on the 3’ end of CTLA-4 to avoid possible complications with membrane insertion.
For each transfection of plasmids, 15 ×106 wild type Jurkat cells were washed in sterile PBS (3.8 mM KCl, 1.2 mM KH2PO4, 139 mM NaCl, 3.15 mM Na2HPO4, 1 mM MgSO4, pH 7.2), resuspended in 400 μL of serum free Opti-MEM medium, and transferred into a 0.4 cm Gene Pulser cuvette (Bio-Rad, cat # 165-2081). The cells were incubated at room temperature for 5 minutes with 15 μg of appropriate plasmid maxi-prep before electroporation using Bio-Rad Gene Pulser II machine with 950 μF and 250 V. The cells were then resuspended in complete growth medium and incubated with 5% CO2 at 37°C. Selection with G418 or puromycin antibiotic started 24 hours post transfection and continued for two weeks before sorting using FACSaria cell sorter (BD Biosciences).
For protein knock down in Jurkat cells, a pool of siRNA oligos targeting NDE1 transcripts (Sigma, cat # SASI_Hs01_00074363-66), p150Glued (Sigma, cat # SASI_Hs01_00065675-78) or siRNA universal negative control (Sigma, cat # SIC002) were electroporated were introduced into cells by electroporation as described above. To determine the amount of protein expression, 1×106 cells were collected at 0, 24, 48 and 72 hours post-transfection. Cells were then lysed in RIPA buffer, boiled with 2X SDS lysis buffer at 95°C for 5 minutes and expression was analyzed on Western blots. For both NDE1 and p150Glued, the lowest expression was seen at the 48 hour time point and these cells were used for subsequent experiments.
Coverslips were cleaned in ethanol in presence of 10% concentrated potassium hydroxide. They were then coated with 1 mg/mL poly-L-lysine, rinsed in distilled water, air dried, and used immediately or stored at 4°C for future use. Raji cells were washed and incubated with SEE for 1 hour in serum-free RPMI 1640 medium. For identification, Raji cells were labeled with Cell Tracker Blue (CTB) for the last 15 minutes. Raji cells were then paired with Jurkat cells by co-pelleting the cells at 1000 g for 5 minutes. Pellets were then gently resuspended, cells were adhered to polylysine-coated coverslips at 37°C for 10 minutes. Cells on coverslips were fixed in PBS containing 1 mM CaCl2, 5 mM glucose, and 4% freshly prepared depolymerized paraformaldehyde for 30 minutes at room temperature followed by permeabilizing for 15 minutes in a pre-chilled 1:1 acetone-methanol solution on ice. Fixed cells were then blocked with 5% goat serum and 0.5% Tween-20 at room temperature for 30 minutes. Primary and secondary antibody incubation were done sequentially for 1 hour each at room temperature in blocking solution. The antibody concentrations used were as recommended by the manufacturers. The coverslips were mounted on glass slides using ProLong Diamond antifade reagent and sealed with nail polish.
For microtubule depolymerization assays, Jurkat cells were incubated with either DMSO (vehicle) or 10 μM colchicine for 15 minutes before paired with SEE-coated Raji cells.
To prepare cell lysates for immunoprecipitation, 107 cells per sample were pelleted, washed in PBS, and mixed with ice-cold RIPA buffer (100 mM Tris pH 8.0, 50 mM NaCl, 0.25% Triton X-100, 1 mM EDTA, 10 mM NaF, 2 mM Na3VO4, 2x Roche mini EDTA-free protease inhibitor). The lysate was homogenized by pipetting through a 21-gage sterile needle. Cellular debris were separated by centrifugation at 10,000 rpm for 10 min. The supernatant was precleared with Protein A agarose beads for 1 hr and then transferred to a tube containing Protein A agarose beads coupled to antibody and incubated for 3 hours at 4°C. The beads were subsequently pelleted at 1000 g for 3 minutes, washed 3 times with ice-cold RIPA lysis buffer and boiled at 95°C for 5 minutes with 2X SDS-protein loading buffer. Unless otherwise mentioned, 10% of the supernatant of the immunoprecipitation reaction was processed in the same way as the samples and used as ‘input’ on SDS-PAGE analysis.
Precleared whole cell lysate or immuno-precipitation samples were run on 10% polyacrylamide gels (SDS-PAGE). Proteins from SDS-PAGE were wet-transferred to 0.2 μm nitro-cellulose membrane at 100 Volts for 1 hour. The membrane was blocked in 5% nonfat milk in TBS and incubated with primary antibody solution (5% BSA with 0.05% Tween-20 in TBS, TBS: 25 mM Tris, 137 mM NaCl, 2.7 mM KCl) overnight at 4°C shaker followed by washing and 1 hour incubation with corresponding HRP-conjugated secondary antibody solution (5% nonfat milk in TBS) on 50 rpm shaker at room temperature.
Cells were washed and resuspended in PBS containing 5mM glucose, 1mM CaCl2, and 1% FBS to a final concentration of 106 cells/mL and incubated at 37°C for 30 minutes in presence 1μM indo-1-AM (Thermo Fisher Scientific, cat # I1203) or with DMSO vehicle. Cells were washed, resuspended in the PBS buffer, incubated for additional 10 minutes and then transferred to a standard fluorometer cuvette. Measurements were performed in a fluorometer equipped with dual emission detectors, a water-jacketed cuvette holder maintained at 37°C and a magnetic stirrer to keep the cells in suspension (PTI, Birmingham, NJ). Fluorescence was excited at 354 nm and the emission collected at 404 and 485 nm. After baseline measurements were obtained 500 ng/mL of anti-Vβ8 antibody was then added followed by 500 ng/mL of unconjugated rabbit anti-mouse IgG secondary antibody. At the end of each trace, ionomycin was added to 2 μM followed by 0.1 mg/mL digitonin and then 2 mM EGTA together with 6 mM Tris base.
The target sequences for knockdown of mouse NDE1 were as follows: 5’-CTTCACTGGCTACTAACTTAT-3’ (N1), 5’-AGTACCAGTGTGGGCGATAAA-3’ (N2), and 5’-ACCAACTGCAGAAATACATTA-3’ (N3). The non-target base scrambled negative control sequence was: 5’-CCTAAGGTTAAGTCGCCCTCG-3’ (C1). The target sequences for knockdown of mouse p150Glued were as follows: 5’-AGTGCAGTGTGGACGTGTATA-3’ (P1), 5’-CCATGCAAGAAGGCGAGTATG-3’ (P2), and 5’-CAGACGAGCGAATCGAGAAAG -3’ (P3). Primers were purchased from Sigma and cloned into pLKO.1-TRC cloning vector (Addgene, cat # 10878) as described by Moffat et al. (2006) (32). Insertion of intended sequences into the plasmid was confirmed by DNA sequencing. Plasmid constructs were transfected into HEK293 cells using Xfect transfection reagent (Clontech, cat # 631317). Viral supernatant was collected 48 hours post-transfection, filtered using a 0.45 μm syringe filter, mixed with 4 μg/ml of polybrene (Sigma-Aldrich, cat # H9268) and then stored in −80°C until used.
Naïve T cells were isolated from spleens of OT-I transgenic mice and were stimulated with syngeneic splenocytes pulsed with 1μM SIINFEKL. Splenocytes were pretreated with 50 μg/mL mitomycin C for 2 hours and thoroughly washed before peptide loading. CTLs were maintained in the same complete growth medium that was used for Jurkat cells but supplemented with 20 U/mL Il-2. After two days of stimulation, 0.5×106 CTLs were collected, washed and transduced with corresponding lentiviral supernatants (multiplicity of infection ~ 5-10) for 10 hours. Spinfection was performed at 2400 rpm for 30 min at 30°C. Cells were resuspended in fresh medium after 10 hours. Selection of transduced CTLs was started 48 hours post-transduction using puromycin (1 μg/mL initially for 1 day and at 2 μg/mL for an additional 3 days before analysis of expression on Western blots (WB).
For immunostaining, EL4 target cells were pulsed with 1 μM SIINFEKL peptide for 1 hour and labeled with CTB for 15 minutes. Cells were washed thoroughly before CTL-EL4 conjugates were prepared. Other conditions for IF, WB, or IP experiments using primary cells remained the same as described above.
A flow cytometry-based cytotoxicity assay was adapted from previously described protocols (33, 34). Briefly, NDE1-specific shRNA (N2) and p150Glued –specific shRNA (P1) treated CTLs, which showed highest level of corresponding protein knockdown, and scrambled control shRNA (C1) treated CTLs were mixed with EL4 target cells in a round bottom 96- well plate. To distinguish the target cells from the CTLs, EL4 cells were labeled with 250 nM carboxyfluorescein succinimidyl ester (CFSE) for 30 minutes in PBS and then washed twice. Labeled EL4 cells were then used directly for cytotoxicity assay or pulsed with 1 μM SIINFEKL peptide for 1 hour. 104 EL4 cells per well were mixed with corresponding CTLs at a ratio of 1:1, 4:1, and 10:1 (effector to target) to a final volume of 150 μL in each well. The plate was incubated at 37°C for 6 hours. Propidium iodide (PI) was added to a concentration of 100 μg/mL at the end of the assay and the cells were immediately analyzed for PI and CFSE positive cells using BD Accuri C6 Plus flow cytometer. FACS data were analyzed using FlowJo software and presented after background normalization.
Total mRNA pool was isolated from mouse CTLs and cDNA was synthesized following the protocol mentioned above. NCBI Primer BLAST program was used to design primers to detect unique regions of 4 commonly reported mouse NDE1 isoforms. The primers used were as follows: isoform a (NM_023317.2): Forward 5’-AAGAGCCAAACGAGCCACA-3’, Reverse 5’-AAGCGTTTTCCTGACCCTTTATC-3’; isoform b (NM_001114085.1): Forward 5’-GAACCGGGACCTCTTGTCAG-3’, Reverse 5’-GGAAGGGATCCTTTATCGCCC-3’; isoform c (NM_001285503.1): Forward 5’-GAATAACCGCCTTCGCATGG-3’, Reverse 5’-GATGAGACAGCAGTACCCCAG-3’; isoform d (NM_001285504.1): Forward 5’-AGTCTGTGAAGAGCCACAATCA-3’, Reverse 5’- CATCTCCGGCTTTACCACCC -3’.
To quantify CTLA4-mCherry clustering at the IS, we distinguished 4 main vesicle patterns in Jurkat cells CLTA4-mCherry. If vesicles were observed in a band at the Jurkat-Raji interface, or in a tight cluster elsewhere in the cell, we counted the vesicles as “clustered”. A second category was “unclustered” defined as cells having a large number of vesicles (at least 15) dispersed away a group of vesicles still clustered around the MTOC, to cells showing no clustering at all. A similar approach was also applied to quantify granzyme B-containing vesicles in CTLs.
The scoring of MTOC polarization has been described previously (35, 36). Briefly, a T cell was divided into four sections of equal width and assigned as region 1, 2, 3, or 4 (region 1 being the nearest to the IS). If the MTOC is located in region 1, the cell was counted as ‘polarized’.
Images were acquired using a Nikon Eclipse microscope coupled to an Andor Zyla scientific CMOS camera. Images were recorded using Micromanager and processed using ImageJ processing software. For determining protein accumulation at the synapse, a line (width: 70 pixels, length: 230 pixels) was drawn through conjugate pairs with the middle (pixel 115) corresponding to the IS. Background intensity obtained from a region outside the cell was subtracted and fluorescence was normalized by dividing all pixel values by the average intensity of row 1 (the rear of the Jurkat cell). Beginning with row 115, intensity values of 5 rows of pixels were grouped together to give 1 increment and used for statistics (mean ± S. E. of the mean). The computed mean together with error bars for each increment is plotted against normalized intensity. Student's t-test was performed for the first ten increments.
NDE1 was detected in Jurkat cells by RT-qPCR (data not shown) and on Western blots as a single band at 40 KD (Supplemental Fig. 1A). To examine the distribution of NDE1 before and after stimulation, Jurkat cells, alone or paired with SEE-coated Raji cells were immunostained for NDE1 and tubulin. Prior to stimulation, NDE1 was observed in punctate foci distributed throughout the cytoplasm (Fig. 1A). After Jurkat cells were activated with SEE-coated Raji cells, NDE1 became concentrated at the IS (Fig. 1B).
To determine if signal transduction through the TcR was required for NDE1 translocation, Jurkat cells were pretreated with either DMSO (vehicle) or the Src kinase inhibitor PP2, paired with SEE-coated Raji cells and then immunostained for NDE1. NDE1 accumulated at the IS in vehicle-treated cells (Fig. 1C), while treatment with PP2 prevented this accumulation (Fig. 1D). Quantification of the average cellular fluorescence through the T cell-target contact site for a population of Jurkat-Raji pairs confirmed that inhibition of Src family kinases blocked translocation of NDE1 to the IS (Fig. 1E).
To determine if dynein forms a complex with NDE1, the DIC was immunoprecipitated from Jurkat cell lysates and analyzed for dynein-binding proteins on Western blots. We observed that NDE1, Lis1 and p150Glued each co-immunoprecipitated with the DIC (Fig. 1F). However, when the reciprocal immunoprecipitation of NDE1 was carried out using a polyclonal anti-NDE1 antibody, the DIC and p150Glued were not detected in the immunoprecipitate (Fig. 1G). Failure of anti-NDE1 antibody to pull down the DIC suggested that the antibody might interfere with binding of NDE1 to dynein, an issue raised in a previous study (37). In a similar series of experiments, we also immunoprecipitated DISC1 and probed for the DIC and dynein-binding proteins. The results showed that the DIC, NDE1, Lis1 and p150Glued all coimmunoprecipitated with DISC1 (Supplemental Fig. 1B). Furthermore, in Jurkat cells paired with SEE-coated Raji cells, DISC1 was seen localized at the IS (Supplemental Fig. 1C-E).
Since anti-NDE1 antibody immunoprecipitates did not contain dynein, perhaps due to antibody and dynein binding to the same site on NDE1, we sought to determine if immunoprecipitation of NDE1-EGFP chimeras with anti-GFP would pull down the DIC. For these studies, a series of NDE1 EGFP chimeras were constructed and tested for their ability to bind dynein, localize at the IS and support MTOC translocation. The three most notable constructs were an N-terminal EGFP fusion to NDE1 (EGFP-NDE1) that potently blocked MTOC translocation, a C-terminal EGFP fusion to NDE1 (NDE1-EGFP) that slightly inhibited MTOC translocation, and a C-terminal monomeric EGFP fusion to NDE1 (NDE1-mEGFP) that showed no inhibition of MTOC translocation (Fig. 2 A-D). When these cells were activated by SEE-coated Raji B cells, EGFP-NDE1 remained in the cytosol, whereas both NDE1-EGFP and NDE1-mEGFP were localized at the IS (Fig. 2 A-C). Other constructs had intermediate effects and were not further explored (data not shown).
Although the NDE1-EGFP construct localized at the IS and did not block MTOC translocation (Fig. 2B and D) neither it nor the EGFP-NDE1 construct pulled down the DIC in immunoprecipitation studies (Fig. 2E and F). One possible explanation was that EGFP somehow interfered with NDE1 folding and reduced its interaction with the DIC. NDE1 is thought to fold into a hairpin that brings the dynein binding sites at the N- and C-terminus in close proximity so as to function as one contiguous binding site. Furthermore, it is thought that NDE1 can form dimers and tetramers (22). Given that EGFP can dimerize on its own, perhaps this interfered with the normal folding or dimerization of NDE1. To address this possibility, EGPF was mutated (A206K) to give monomeric EGPF (mEGFP-NDE1; NDE1-mEGFP) (38). When these constructs were immunoprecipitated with anti-GFP antibody and analyzed for the DIC we found that the NDE1-mEGFP construct co-immunoprecipitated with the DIC, endogenous NDE1 and Lis1 (Fig. 2G). The mEGFP-NDE1 construct pulled down Lis1 and endogenous NDE1 but not the DIC (data not shown). When these constructs were analyzed for their impact on MTOC translocation, the data show the NDE1-mEGFP gave the same levels of MTOC translocation as WT Jurkat cells (Fig. 2D). For subsequent studies, we continued to use the original EGFP-NDE1 as a potent inhibitory construct. The observation that NDE1-mEGFP localized at the IS, did not affect MTOC polarization and that it pulled down the DIC as well as endogenous NDE1 suggested that it was functioning normally.
Since immunoprecipitation of NDE1-EGFP using anti-GFP antibody did not pull down the DIC, we next sought to determine if immunoprecipitation of dynein would pull down NDE1-EGFP. To test that hypothesis, we immunoprecipitated the DIC from NDE1-EGFP expressing Jurkat cell lysates. Interestingly, we also did not detect any NDE1-EGFP in the DIC immunoprecipitates (Fig. 2H) although endogenous NDE1 was detected (Fig. 1F). These results indicate that NDE1-EGFP accumulates at the IS without binding to dynein.
Immunostaining for the DIC and Lis1 showed that both of these proteins colocalized with NDE1 at the IS in WT Jurkat-Raji pairs or in Jurkat cells expressing NDE1-EGFP or NDE1-mEGFP (Supplemental Fig. 2 A-F). On the other hand, neither the DIC nor Lis1 localized at the IS in Jurkat cells expressing the dominant negative EGFP-NDE1 (Supplemental Fig. 2G-H). Analysis of 81 Jurkat-Raji pairs from immunostained images showed that DIC was largely absent from the IS in Jurkat cells expressing EGFP-NDE1 proteins as compared to wild type Jurkat cells (Fig. 2I).
In previous studies, some perturbations of dynein seemed to affect cell signaling. For example, Martin-Cofreces et al. noted differences in phosphorylation in T cells overexpressing dynamitin (18). In other experiments, Yi et al. used 50 μM Ciliobrevin D to inhibit dynein (19). However, we found that 1 μM Ciliobrevin D severely reduced calcium signaling such that at 50 μM, calcium signaling was essentially abolished (Supplemental Fig. 2I). As a precaution, we thought it necessary to determine if the expressed NDE1 constructs affected calcium signaling through the TcR. To monitor calcium signaling, these cell lines were loaded with indo-1-AM and stimulated with anti-TcR antibody. The fluorescence ratio of calcium-bound / unbound indo-1 signal obtained after subtracting the fluorescence background suggested no abnormality in calcium signaling in any of the tested cell lines (Supplemental Fig. 2J).
Although inhibition of MTOC translocation by the dominant negative EGFP-NDE1 construct suggested that NDE1 might be important for MTOC translocation, there is always the possibility that EGFP fusion proteins could have non-specific effects. To further examine the role of NDE1 in MTOC translocation, NDE1 was depleted using NDE1-specific siRNA. By monitoring NDE1 expression at 24 hour intervals, we found that NDE1 expression was greatly reduced 24-72 hours after siRNA transfection, whereas the control siRNA had no effect (Fig. 3A). Using the 48 hour time point for subsequent assays, control and knockdown cells were paired with SEE-coated Raji cells and immunostained for tubulin and NDE1. The data show that MTOC translocation was greatly reduced in the NDE1 siRNA-treated cells (Fig. 3 B-D). As expected, visible NDE1 accumulation at the synapse in NDE1 knockdown cells was also greatly reduced compared to the control cells (Fig. 3E).
To determine if NDE1 depletion prevented the recruitment of dynein to the synapse, NDE1 siRNA and control siRNA- treated cells were paired with SEE-coated Raji cells and immunostained for the DIC. The data show that for the control siRNA-treated cells, NDE1 and DIC colocalized at the synapse as seen for untreated Jurkat cells (Fig. 3F). In cells treated with the NDE1-specific siRNA, little DIC was seen at the synapse (Fig. 3G). To quantify the difference in dynein accumulation, average pixel values were plotted as described previously. The results confirm that when NDE1 expression is reduced, little dynein accumulates at the synapse (Fig. 3H).
To determine if reduced expression of NDE1 impacted calcium signaling or the accumulation of actin at the IS, both were also monitored in control and NDE1 depleted cells. Depletion of NDE1 had little impact on the accumulation of actin at the synapse (Fig. 3 I and J) or on calcium signaling in response to TcR stimulation (Fig. 3K).
The previous data showed that NDE1 translocated from the cytoplasm to the synapse when Jurkat cells were stimulated by SEE-coated Raji cells. To determine if this translocation was microtubule-dependent the distribution of NDE1 was monitored in colchicine-treated Jurkat-Raji pairs. NDE1 accumulation at the synapse was profoundly reduced in colchicine-treated cells compared to DMSO treated cells (Fig. 4 A and B). Quantitative analysis of NDE1 pixel values confirmed that intact microtubules were required for NDE1 accumulation at the IS in Jurkat cells (Fig. 4C).
We next sought to determine the function of NDE1 in mouse primary CTLs. For these experiments, CTL blasts were prepared from splenocytes of OT-I mice by mixing them with mitomycin-treated C57BL/6 splenocytes in the presence of SIINFEKL peptide. Subsequently OT-I CTLs were mixed with EL-4 cells in the presence of absence of peptide and immunostained for NDE1. The results showed that in the absence of peptide NDE1 was scattered throughout the cytoplasm (Supplemental Fig. 3A-B). In the presence of peptide, NDE1 accumulated at the IS, similar to the results obtained using Jurkat cells (Supplemental Fig. 3C).
The dynein complexes in mouse CTLs were examined by immunprecipitating the DIC. The results showed that NDE1, Lis1 and p150Glued co-immunoprecipitated with the DIC from CTL cell lysates (data not shown). Similar to the results obtained from Jurkat cells, immunoprecipitation of NDE1 did not pull down the DIC or p150Glued (Supplemental Fig. 3D). The only obvious difference between the results obtained using Jurkat cells or mouse CTLs was that for the CTLs, NDE1 Ig appeared to detect multiple bands on the blot, perhaps due to the presence of multiple NDE1 isoforms. To test this idea, we designed specific primers targeting 4 common mouse NDE1 mRNA transcripts according to the NCBI database. PCR results indicated that NDE1 isoform b, c and d are present in OT-I mouse CTLs whose molecular weights corresponded to the 3 bands seen on the blot (Supplemental Fig. 3E). Immunostaining of mouse CTL-EL4 conjugates showed that NDE1 localization at the IS overlapped that of DIC and Lis1 (Supplemental Fig. 3F-G).
To deplete NDE1 in mouse CTLs, three NDE-1 specific (N1, N2, & N3) and one base-scrambled control shRNA sequence were packaged into Lentiviral particles and incubated with CTLs for 10 hours. Subsequently CTLs were selected with puromycin for 4 days. Of the three NDE1 shRNA target only N2 gave a strong reduction in NDE1 expression and was used for subsequent experiments (Fig. 5A).
OT-I CTLs pretreated with lentiviral NDE1-specific (N2) or scrambled control shRNA were prepared as described above and then incubated with peptide-pulsed EL4 cells. Subsequently cells were immunostained for tubulin (Fig. 5B-C) and scored for MTOC translocation (Fig. 5D). For the bulk population MTOC translocation was reduced from 80% to 48% for the NDE1-specific shRNA but was not affected by the scrambled control (Fig. 5B-C; 5D, ‘all pairs’). However, both the Western blot and immunostaining showed a partial knockdown of NDE1 such that NDE1 could sometimes still be detected at the IS (Supplemental Fig. 3H). When NDE1 was detected at the IS, the MTOC often translocated as well. When we examined CTL conjugates that showed no obvious NDE1 at the IS, the percentage of cells showing MTOC translocation was reduced to 20% (Fig. 5D, ‘selected pairs’).
Finally, we investigated how depletion of NDE1 affected the ability of mouse CTLs to lyse antigenic targets. CTLs transduced with either NDE1 or control shRNAs (as described above), were incubated for 6 hours with untreated EL4 or peptide-pulsed EL4 target cells at various effector to target ratios (1:1, 4:1, or 10:1). Subsequently, propidium iodide (PI) was added to label the dead cells. Flow cytometry was used to quantify CFSE-labeled target cells and PI-labeled dead cells as described in the Materials and Methods. Depletion of NDE1 significantly inhibited CTL-mediated target cell lysis, compared to control groups (Fig. 5E). At a 10:1 effector: target ratio, lysis was reduced from 76% for the control shRNA to 32% for CTLs receiving the N2 shRNA.
NDE1 together with Lis1 form one of the major complexes mediating dynein-dependent transport, the other being the dynactin complex. Using p150Glued as a marker for dynactin, we monitored the distribution of dynactin both by expression of EGFP- p150Glued and by immunofluorescence. In unactivated Jurkat cells, fluorescence was mainly localized to microtubules and the MTOC (data not shown). In Jurkat cells that expressed EGFP-p150Glued and that were activated by SEE-coated Raji cells, EGFP-p150Glued remained largely concentrated around the MTOC and along microtubules (Fig. 6A). A similar pattern of localization was seen when normal Jurkat cells were immunostained for p150Glued (Fig. 6B).
We next sought to determine if Jurkat cells had secretory lysosome-like vesicles similar to those of CTLs. Since we did not know if Jurkat cells had similar secretory vesicles or what the contents of those vesicles might be, we expressed CTLA4 fused to mCherry as a vesicle marker. CTLA-4 is a transmembrane protein resident in secretory lysosomes of CTLs (39, 40). The results show that CTLA-4 labeled vesicles accumulate near the center of the synapse of activated Jurkat-Raji pairs (Supplemental Fig. 4A).
To examine the role of dynactin in MTOC translocation and vesicle movements, Jurkat cells expressing CTLA4-mCherry were electroporated with either a p150Glued -specific siRNA or a scrambled control sequence and expression was monitored on Western blots. The results show that by 24 hours, expression of p150Glued was undetectable (Fig. 6C). Cells from the 48 hour time point were then conjugated with SEE-coated Raji cells, fixed, and immunostained for tubulin and p150Glued. The results show that for cells treated with the scrambled siRNA, secretory vesicles marked by CTLA4-mCherry concentrated at the IS. However, when cells were treated with p150Glued siRNA, the vesicles were often dispersed around the MTOC and did not concentrate at the IS (Fig. 6D-F). The p150Glued-depleted cells were also scored for MTOC translocation and the results were similar those obtained from untreated or scrambled siRNA control cells (Fig. 6G; supplemental Fig. 4B-D).
The finding that depletion of p150Glued blocked vesicle accumulation at the IS but not MTOC translocation suggested that there is little overlap in the functions of NDE1 and dynactin in these processes. In data presented previously, immunoprecipitation of NDE1-mEGFP pulled down the DIC and Lis1 but not p150Glued (Fig. 2G). The reciprocal immunoprecipitation of p150Glued showed that NDE1 was not detected on the blot (Fig. 6H). Thus our data indicate that NDE1 and p150Glued are not found in the same dynein complexes. Because CTLA4 will localize to the plasma membrane upon fusion of CTLA4-containing vesicles with the membrane in activated cells, we subsequently stained for granzyme B to confirm that p150Glued co-localized with secretory vesicles in mouse CTLs (Supplemental Fig. 4E).
We next investigated dynactin (p150Glued) distribution in mouse primary CTLs. In unactivated mouse CTLs, p150Glued remained largely concentrated around the MTOC and along microtubules (Supplemental Fig. 4E). Immunostaining data showed that granzyme B-containing vesicles were also clustered around the MTOC (Supplemental Fig. 4E).
We next sought to deplete p150Glued in mouse CTLs. For this experiment, 3 shRNA target sequences were tested as was done for NDE1 and one sequence (P1) gave substantial reduction in p150Glued expression (Fig. 7A). Next we prepared CTLs that were transduced with either p150Glued shRNA (P1) or a scrambled shRNA control, paired them with peptide pulsed EL4 cells and the immunostained them for tubulin (Fig. 7B-C). When these p150Glued – depleted conjugates were scored for MTOC translocation, the results showed no significant difference in MTOC translocation compared to those treated with the scrambled shRNA (Fig. 7D).
In control shRNA treated cells, granzyme B-immunostained vesicles formed a distinct band or cluster at the IS whereas p150Glued remained at the center of the contact site, presumably associated with the MTOC (Fig. 7E). When p150Glued was depleted, accumulation of granzyme B-vesicles at the IS was greatly reduced compared to those receiving a scrambled shRNA (Fig. 7F-G). Interestingly, in a few control knockdown cells, we observed overlapping p150Glued and granzyme B signals (Supplemental Fig. 4F-G).
Finally, we investigated how depletion of p150Glued affected the ability of mouse CTLs to lyse antigenic targets. For these experiments, CTLs transduced with either p150Glued or control shRNAs (as described above), were incubated for 6 hours with CFSE-labeled EL4 cells in the presence or absence of peptide at various effector to target ratios (1:1, 4:1, or 10:1). Subsequently, PI was added to label the dead cells. Flow cytometry was used to quantify CFSE-labeled target cells and PI-labeled dead cells as described above. Depletion of p150Glued significantly inhibited CTL-mediated target cell lysis, compared to control groups (Fig. 7H). At a 10:1 effector: target ratios, p150Glued depleted CTLs lysis was reduced to 50% as compared to the control shRNA value of 76%.
In this study we examined the roles of NDE1 and dynactin in what are generally regarded as two key aspects of focused secretion; movement of the MTOC and accumulation of secretory vesicles at the IS. The results based on both expression of an inhibitory EGFP-NDE1 construct or siRNA (or shRNA)-mediated depletion of NDE1, showed that NDE1 was essential for MTOC translocation. In contrast, dynactin was not required for MTOC translocation but was needed for accumulation of secretory vesicles at the IS.
Based on immunoprecipitation of the DIC, NDE1 forms a complex with dynein and Lis1. However, the reciprocal immunoprecipitation of NDE1 pulled down Lis1 but not dynein or p150Glued. This is consistent with other reports showing that immunoprecipitation of NDE1 did not pull down dynein (37). Efforts to circumvent this problem led us to construct a series of EGFP-NDE1 fusion proteins where anti-GFP antibody could be used to immunoprecipitate NDE1. This approach also avoided any confusion between NDE1 and its paralog, NDEL1 which behaved quite differently insofar as NDE1 accumulated at the IS whereas NDEL1 did not (Supplemental Fig. 1F).
Four EGFP-NDE1 fusion proteins were generated and tested. Of these, both C-terminal fusion constructs accumulated at the IS as did dynein and Lis1. A slight inhibition of MTOC translocation was seen with the NDE1-EGFP but not with NDE1-mEGFP. The key difference is that in immunoprecipitation experiments, pulldown of NDE1-mEGFP brings down dynein whereas pull-downs of NDE1-EGFP did not. Immunoprecipitation of the DIC also did not pull down NDE1-EGFP even though it did pull down native NDE1. These results lead to the conclusion that NDE1-EGFP does not bind to dynein.
The absence of any evidence that NDE1-EGFP binds to dynein was remarkable considering that it accumulated at the IS. Given that our data also show that NDE1 is required for accumulation of dynein at the IS we conclude that NDE1 must be serving directly or indirectly to anchor dynein at the IS, likely through connections to the actin cytoskeleton. NDE1 normally coimmunoprecipitates with LIS1 and DISC1 and this complex could link to the actin cytoskeleton in a number of ways. DISC1 is known to interact with components of the actin cytoskeleton (26, 28, 41). Lis1 is known form a complex with Cdc42 and CLIP-170 as well as the actin-binding protein IQGAP (42, 43). Finally, NDE1 reportedly binds to paxillin, a protein important in regulating focal adhesions and the actin cytoskeleton (44, 45).
In principle, NDE1 might accumulate at the IS by moving along microtubules but it seems that simple diffusion and binding is not adequate in itself. Data showing that NDE1 accumulation at the IS depends on intact microtubules might suggest that a motor is involved. In principle, either dynein or kinesin could carry NDE1 to the IS. However, since dynein does not seem to be required, a possible alternative is that a kinesin-like motor transports NDE1 to the IS. This seems plausible since dynein and NDE1 are known to form complexes with kinesin in other settings (7, 24, 46). Alternatively, NDE1 might be transported together with Lis1 at the tips of microtubules. Lis1 is reportedly a component of a microtubule tip-binding complex that includes CLIP-170 and IQGAP (43).
The association of NDE1 with the actin cytoskeleton would help resolve differences between dynein-dependent MTOC translocation and the model of Stinchcombe et al., who proposed that an expanding actin ring at the IS drives MTOC translocation (13). The actin-dependent mechanism is not mutually exclusive with a role for dynein, since NDE1 and ultimately dynein likely become anchored to structures involving actin. However, we question if actin is actually what drives MTOC translocation. For example, we showed that when a CTL is engaged with two target cells, the MTOC can oscillate between the two target contact sites (11). In order to generate these MTOC oscillations by the actin-based model, there would need to be coordinated cycles of microtubule release from actin anchors at one contact site and their reformation at the second site, combined with repeated cycles of actin spreading in opposite directions at the two contact sites. Yet actin seems to polymerize at the center of the IS and then spreads outward to the periphery where it remains. Such a mechanism is not impossible perhaps but is certainly harder to envision, whereas dynein-dependent oscillatory movements due to dynein are well known (47-50).
Using reciprocal immunoprecipitations and Western blots, we showed p150Glued was not present with NDE1 in the same dynein complexes. This result was expected given that NDE1 and p150Glued bind to the same site at the N-terminus of the DIC (51). This is also consistent with our data showing that NDE1 and p150Glued differ in their subcellular distributions. NDE1 localizes as a ring corresponding to the pSMAC of the IS whereas p150Glued is largely concentrated around the MTOC and along microtubules. However, p150Glued is sometimes clearly associated with granzyme-B containing vesicles. When these vesicles accumulate as a band at the IS, p150Glued can also be seen as a band that somewhat resembles the distribution of NDE1.
Consistent with their different distributions, NDE1 and dynactin play functionally different roles in the overall process of bringing vesicles to the IS. In an effort to study these vesicle movements, a number of methods were used to visualize acidic vesicles, but we found many more vesicles were labeled than accumulated at the IS. We decided to try labeling vesicles with CTLA4-mCherry because in CTLs, CTLA4 is known to concentrate in perforin and granzyme-B containing vesicles that move to the IS (39, 40, 52). Movement and secretion of these vesicles is ultimately what exposes CTLA4 on the surface where it can interact with B7 (53). When CTLA4-mcherry was expressed in Jurkat cells, a small subset of the total acidic vesicle pool was labeled and these vesicles accumulated at the IS when Jurkat cells engaged SEE-coated Raji cells.
In unactivated Jukat cells, many of the CTLA4-mCherry labeled vesicles are clustered around the MTOC and this is likely due to dynein-dependent transport (20). This is supported by our data showing that when p150Glued was depleted, CTLA4-labeled vesicles become more dispersed throughout the cytoplasm. However, when Jurkat cells engage SEE-coated Raji cells, even these MTOC-associated vesicles did not accumulate at the IS. Accumulation of granzyme B-containing vesicles was also reduced in p150Glued -depleted mouse CTLs. Both sets of data indicate that p150Glued plays an important role in the process of focused secretion.
There are two ways p150Glued could play a role in bringing vesicles to the IS. One is merely to cause vesicle accumulation around the MTOC prior to MTOC translocation. In this scenario, MTOC translocation is solely responsible for bringing secretory vesicles to the IS (12, 13). The other scenario is that the dynein dynactin complex independently and perhaps continuously moves vesicles towards MTOC and it is this movement that causes vesicle accumulation at the IS. This scenario would not eliminate the possibility that MTOC translocation brings vesicles to the IS but it also helps explain the data of Bertrand and colleagues (54) who showed that secretory vesicle accumulation and secretion can take place apart from MTOC translocation.
Bertrand et al. noted that microtubules appear at the IS about the same time as secretory vesicles. In a previous study, we showed evidence of microtubule interaction with the IS well before the MTOC had translocated (11). Assuming that accumulation of secretory vesicles at the IS is microtubule-dependent, we propose that the initial critical step in setting up focused secretion is association of microtubules with the IS. This may be a core function of the NDE1-dynein complex. In principle, dynactin-dynein-driven vesicle movements could be in progress before, during, and after T cell activation. Were there no association of microtubules with the IS, these vesicles might simply accumulate around the MTOC. When microtubules become associated with the IS, movement of vesicles along those anchored microtubules is now directed towards the IS, a process that could begin before the MTOC translocates.
Our model does not eliminate a role for MTOC translocation in bringing vesicles to the IS but it adds flexibility such as explaining the data of Bertrand et al. It also suggests a new way to think about how MTOC translocation could facilitate vesicle accumulation at the IS. In a previous study, we showed that vesicles can move along microtubules from the rear of the cell up to the IS (21). We have previously shown that microtubules form sharp turns and bend backwards where they contact the IS, possibly due to dynein-driven movements (11). As the MTOC translocates, dynein-driven sliding of microtubules would cause them to move further rearward into the cytoplasm where they might encounter new vesicles and create additional opportunities for vesicle movement towards the IS.
Yi et al. reported that dynein was required for MTOC translocation but in their studies, the process also depended on plus end microtubule depolymerization (19) rather than sliding past the IS towards the back of the cell as we have proposed. One of the supporting evidences for their model was that Taxol blocked MTOC translocation. Previous reports on the effects of taxol-related compounds are mixed with some studies showing that they had no effect on MTOC translocation (55-57). It is interesting to note that in the study of Bertrand and colleagues cited above, a fluorescent taxol analog was used in some experiments to label microtubules and follow MTOC translocation. Taxol does make microtubules considerably more rigid and might, at some concentrations, prevent their flexure at the IS as is required by our model (58). At present, we are not convinced that the taxol data proves that microtubule depolymerization is part of the process of MTOC translocation
Although the model of Yi et al. seems plausible, evidence is accumulating that microtubule stabilization rather than depolymerization is required for MTOC translocation. Microtubule acetylation is one of the mechanisms for increasing microtubule stability (59). Previous studies have shown that that there is a burst of microtubule acetylation upon TcR ligation and that overexpression of histone deacetylase 6, which deacetylates microtubules, blocked MTOC translocation (60). Recently it has also been found that formins acetylate microtubules and that formins are required for MTOC translocation (61, 62).
We have found that both NDE1 and dynactin complexes bind to DISC1. Our preliminary data show that DISC1 localizes to mitochondria and accumulates at IS but there is more than one isoform in Jurkat cells such that these differences might be isoform-specific. Mutations in DISC1 are associated with schizophrenia and other behavioral abnormalities. It has also long been noted in the literature that immune dysfunction, and in particular, T cell dysregulation is commonly associated with schizophrenia (63, 64). Indeed, epidemiological and genetic studies have linked inflammation in general and imbalanced cytokine secretion to the development of schizophrenia (65, 66). Given the association of DISC1 with neuronal secretion and the data presented in this study, perhaps DISC1 may provide a concrete link between schizophrenia and immune dysfunction.
This study shows that dynein complexed with NDE1 or p150Glued performs different functions that are both important in the delivery of secretory vesicles to the IS. Furthermore, depletion of either NDE1 or p150Glued had a large impact on CTL-mediated target cell lysis. While NDE1 depletion had a greater impact on lysis efficiency, this might be due to the incomplete knockdown of p150Glued. This study also provides hints as to how dynein might become associated with the actin cytoskeleton through the NDE1-DISC1 complex and through Lis1. Such an association could help resolve differences in various models that alternatively show that MTOC polarization depends on actin and actin-associated proteins or on dynein.
We thank Dr. Haley Tucker for critical reading of the manuscript and providing several reagents used in this study.
This work was supported in part by National Institute of Health grant R01AI104870 (to L.I.R.E).