Defective Proximal TCR Signaling Inhibits CD8+ Tumor-Infiltrating Lymphocyte Lytic Function
http://www.100md.com
免疫学杂志 2005年第3期
Abstract
CD8+ tumor-infiltrating lymphocytes (TIL) are severely deficient in cytolysis, a defect that may permit tumor escape from immune-mediated destruction. Because lytic function is dependent upon TCR signaling, we have tested the hypothesis that primary TIL have defective signaling by analysis of the localization and activation status of TIL proteins important in TCR-mediated signaling. Upon conjugate formation with cognate target cells in vitro, TIL do not recruit granzyme B+ granules, the microtubule-organizing center, F-actin, Wiskott-Aldrich syndrome protein, nor proline rich tyrosine kinase-2 to the target cell contact site. In addition, TIL do not flux calcium nor demonstrate proximal tyrosine kinase activity, deficiencies likely to underlie failure to fully activate the lytic machinery. Confocal microscopy and fluorescence resonance energy transfer analyses demonstrate that TIL are triggered by conjugate formation in that the TCR, p56lck, CD3, LFA-1, lipid rafts, ZAP70, and linker for activation of T cells localize at the TIL:tumor cell contact site, and CD43 and CD45 are excluded. However, proximal TCR signaling is blocked upon conjugate formation because the inhibitory motif of p56lck is rapidly phosphorylated (Y505) and COOH-terminal Src kinase is recruited to the contact site, while Src homology 2 domain-containing protein phosphatase 2 is cytoplasmic. Our data support a novel mechanism explaining how tumor-induced inactivation of proximal TCR signaling regulates lytic function of antitumor T cells.
Introduction
Lytic function of CD8+ effector T cells is initiated by recognition of target cells, which results in rapid activation (1). Activation requires oligomerization of the TCR dimer with cognate ligand/MHC class I molecules that induces either a conformational change in, or low-order aggregation of, the TCR (2, 3). Upon binding to target cells, the coreceptor CD8 is recruited to the TCR complex, which serves to stabilize binding to cognate ligand:MHC class I, therein enhancing sensitivity of Ag recognition (4). CD8 association with the TCR also recruits Src-related kinase p56lck into close proximity with its substrates, primarily CD3 (5, 6, 7, 8). Phosphorylated CD3 recruits ZAP70, which in turn phosphorylates linker for activation of T cells (LAT),3 followed by recruitment of additional proteins to phosphorylated LAT, culminating in the activation of phospholipase C-1 (PLC-1). Following this early activation phase (30 s), calcium flux is initiated, resulting in coordinated activation of additional downstream signaling pathways (9, 10, 11).
Assembly of the signaling complex results in the formation of dynamic membrane cholesterol- and sphingolipid-enriched domains, termed lipid rafts, into which the TCR:CD8 complex partitions (12, 13, 14, 15). The raft serves as a physical platform upon which various cell signaling molecules assemble after triggering, defined in biochemical terms as phosphorylation of CD3 and CD3 heterodimers (12, 15, 16). In quiescent cells, and immediately upon conjugate formation with APC or target cells, rafts are small (70 nm) (17), but reorganize after activation to form a large aggregate that colocalizes with the immunological synapse (IS) (18). Upon oligomerization of the TCR complex, proteins constitutively associated with raft lipids, by virtue of posttranslational S-acylation or addition of GPI moieties (e.g., Thy1, p56lck, CD8, or LAT), coalesce coincident with raft formation, rapidly increasing the local concentration of raft lipids and associated proteins.
Within 5–10 min of interaction between CD8+ T cells and MHC peptide on an APC, cell surface molecules segregate into different regions, forming an IS (19). A mature IS is characterized by proteins arranged in a bull’s eye pattern with an arrangement of supramolecular activation clusters (SMAC) in a central zone (c-SMAC) containing lipid rafts/TCR/CD3/costimulatory receptors, a peripheral zone (p-SMAC) containing adhesion receptors, primarily LFA-1, and a zone of additional proteins that are excluded from the core region. In addition to faster kinetics of formation compared with CD4+ T cells (20, 21), IS formation in CD8+ T cells has been shown to precede effector functions such as reorganization of the cytoskeleton, mobilization of the microtubule organizing center (MTOC) to the IS, and release of lytic granules (19, 22). Although there are several different theories on IS function (13, 18, 23), the exact function remains unknown.
CD8+ effector T cells that infiltrate tumors (TIL) provide an excellent natural model for study of signaling requirements for CTL function because they have severely defective lytic function (24, 25, 26). Freshly isolated TIL are nonlytic (CD44+CD62LlowCD69+) memory/effector cell, but recover lytic function following brief in vitro culture permitting direct comparison of lytic and nonlytic TIL purified from the same tumor (27, 28). The absence of manipulation during the in vitro culture makes this system ideal for study of the tumor-induced lytic defect. Because exocytosis of lytic granules may be dependent upon IS formation (19) and nonlytic TIL cannot exocytose granules (28), we considered that TIL IS formation or function is defective.
Using a combination of confocal microscopy and biochemical assays, we investigated the localization and activation status of proteins important in signal transduction, IS formation, and lytic function in conjugates formed between TIL and cognate tumor cells. Signal transduction in freshly isolated, nonlytic TIL is blocked at a proximal step because LAT is not phosphorylated and ZAP70, although recruited to the TIL:target contact site (CS), is only weakly phosphorylated. Significantly, the inhibitory motif in p56lck (Y505) becomes rapidly phosphorylated upon binding to cognate tumor cells. Consistent with inactivation of p56lck, COOH-terminal Src kinase (Csk) is localized at the TIL:tumor CS and Src homology 2 domain-containing protein phosphatase 2 (Shp-2) remains cytoplasmic. In addition, Shp-1 localizes at the CS, where it may mediate deactivation of various SH3-containing proteins (such as ZAP70), therein preventing propagation of the activation signal. Furthermore, we show that upon contact with cognate target cells, nonlytic TIL assemble many signaling components with kinetics typical of activated CD8+ T cells, showing that nonlytic TIL are triggered by conjugate formation. However, CD2, the CD3 complex, and CD8, which associate with the TCR before conjugate formation, rapidly dissociate and are excluded from the CS. Tumor-induced disruption of T cell activation at a point downstream of triggering, therein blocking proximal tyrosine kinase activity, calcium flux, and dependent lytic function, is a novel mechanism for inhibition of the CD8+ T cell effector phase.
Materials and Methods
Mice
C57BL/6 male mice were obtained from The Jackson Laboratory, were housed five per cage in a barrier facility, and were maintained on a 12-h light/dark cycle (7 a.m. to 7 p.m.) with ad libitum access to food and water. A sentinel program revealed that mice were mouse hepatitis virus negative and the MCA38 cell line is mouse hepatitis virus negative, as assessed by mouse assessment profile service testing. Experiments involving animals were conducted with the approval of the New York University School of Medicine Committee on Animal Research.
Tumor
MCA38 adenocarcinoma (a gift from Y. Liu, Ohio State University, Columbus, OH) was passaged from tissue culture plasticware by incubation in HBSS containing 3 mM EDTA, followed by washing three times in HBSS. Viability was determined by trypan blue dye exclusion, and 2–3 x 106 cells were injected i.p. in a volume of 0.2–0.3 ml of HBSS for tumor induction. Cells were passaged in vitro for 3–5 wk, following which new frozen stocks were thawed for usage.
Tissue culture
RPMI 1640 medium (BioWhittaker) was used for growth of MCA-38 cells and for culture of T cells, as described (28).
Isolation of TIL
Tumors were dissected (any necrotic portion was discarded), mechanically disrupted by passage through a tissue press, and enzymatically digested into single cell suspensions, and TIL were isolated by immunomagnetic separation using type LS+ columns and anti-CD8-conjugated magnetic beads (Miltenyi Biotec), as described previously (27). (Potentially inhibitory effects of the immunobead isolation protocol upon TIL lytic function is discounted because after recovery in vitro we have reisolated TIL with anti-CD8 magnetic beads (including exposure to tumor digest enzymes) and those cells are lytic. As further control, we have prepared CTL by primary MLR in vitro, followed by isolation with anti-CD8 immunobeads. These cells are fully lytic, again showing that the isolation protocol is without effect on lytic activity.) In each experiment, aliquots of isolated T cells were analyzed by flow cytometry and were routinely 95% CD8+. TIL were used immediately after isolation for experiments, except in some experiments in which TIL were plated in complete RPMI 1640 medium (2 x 106 cells/ml) for 6–18 h before usage.
Antibodies
Abs and reagents used for confocal microscopy and flow cytometry were: actin (phalloidin-FITC or -tetramethylrhodamine isothiocyanate; Sigma-Aldrich), CD2 (clone RM2-5; Caltag Laboratories), CD3 (clone 145-2C11; BD Pharmingen), CD3-Cy5 (clone 500A2; Caltag Laboratories), CD3 (goat IgG (W-12); Santa Cruz Biotechnology), CD3 (rabbit IgG; D. Wiest, Fox Chase Cancer Institute, Philadelphia, PA), CD5 (clone 4H8E6; Southern Biotechnology Associates), CD8 (clone 53-6.7; BD Pharmingen), CD8 (clone CT-CD8b; Caltag Laboratories), CD8-APC (clone 5H10; Caltag Laboratories), CD43 (clone S7; BD Pharmingen), CD45 (Alexa546-conjugated clone I3/2.3; M. Dustin, New York University School of Medicine, New York, NY), Csk (rabbit IgG (C-20); Santa Cruz Biotechnology), cholera toxin B (CT-B) FITC (Sigma-Aldrich), phospho-ERK (rabbit IgG; Cell Signaling Technology), granzyme B (goat IgG (N-19); Santa Cruz Biotechnology), LAT (rabbit IgG; L. Samelson, National Institutes of Health, Bethesda, MD), phospho-LAT (rabbit IgG; Upstate Biotechnology), LFA-1 (clone I21/7; BD Pharmingen), phospho-PLC-1 Y783 (rabbit IgG; Upstate Biotechnology), p56lck (clone 28; BD Pharmingen), phospho-p56lck (clone Y505; Cell Signaling Technology), phospho-Y (clone 4G10; Upstate Biotechnology), total proline-rich tyrosine kinase-2 (Pyk-2; rabbit IgG; Upstate Biotechnology), phospho-Pyk-2 Y881 (rabbit IgG; BioSource International), Shp-1 (rabbit IgG; Upstate Biotechnology), Shp-2 (rabbit IgG (sc-280); Santa Cruz Biotechnology), TCR (clone H57-597; BD Pharmingen), TCR-APC (clone H57-597; Caltag Laboratories), -tubulin (clone DM1A; Sigma-Aldrich), Wiskott-Aldrich syndrome protein (WASp; clone 7066; Upstate Biotechnology), phospho-ZAP70 Y319 (rabbit IgG; A. Chan, Genentech, South San Francisco, CA), and Y493 (rabbit IgG; Cell Signaling Technology). Secondary reagents for confocal microscopy were: goat anti-hamster Alexa488 (Molecular Probes), goat anti-hamster Alexa568 (Molecular Probes), goat anti-hamster rhodamine (Jackson ImmunoResearch Laboratories), goat anti-rabbit Alexa488 (Molecular Probes), goat anti-rabbit FITC (Caltag Laboratories), goat anti-mouse Alexa568 (Molecular Probes), donkey anti-goat Alexa488 (Molecular Probes), and donkey anti-rat rhodamine (Jackson ImmunoResearch Laboratories).
Flow cytometry
Cells were resuspended at 1 x 107/ml in complete RPMI 1640 medium, and Abs were added at empirically determined optimal concentrations. After incubation (4°C for 30 min), cells were washed once with 1 ml of FACS wash (PBS, 2% FBS, 0.1% sodium azide), and, if primary Ab was unlabeled, resuspended in complete RPMI 1640 medium at 1 x 107/ml, and appropriate secondary Abs were added. Conditions for reaction with secondary reagents were identical as for primary Ab. Cells were washed with 1 ml of FACS wash before fixation with 2% paraformaldehyde before analysis on a FACScan flow cytometer (BD Biosciences).
Conjugate formation and confocal microscopy
CD8+ TIL were mixed with tumor cells at a 3:2 ratio, centrifuged at 16,000 x g for 10 s to promote conjugate formation, resuspended in 0.05 ml of RPMI, transferred to poly-L-lysine-coated coverslips, and incubated for different intervals of time at 37°C before fixation in 4% paraformaldehyde (15 min at room temperature (RT)). (For analysis of intracellular molecules, cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min at RT (excepting phospho-ERK, in which case cells were permeabilized with 90% methanol). There was no staining of intracellular proteins (e.g., F-actin, granzyme B, etc.) when detergent was omitted, showing that paraformaldehyde fixation of cells does not introduce artifactual permeabilization (data not shown). For analysis of cell surface molecules, coverslips were not treated with detergent. After fixation, coverslips were washed three times with PBS, blocked in complete medium for 15 min before Abs were added, incubated for 45–60 min at RT, and washed in PBS. Coverslips were mounted with medium containing antifading agents (Biomedia) before analysis by confocal microscopy using a Zeiss confocal LSM510 microscope to obtain full-thickness Z-stacks of conjugates (15–20 per sample). Images were analyzed using Zeiss LSM510 software and are depicted as three-dimensional (3D) reconstructions of all Z-stacks, except where specifically noted. For quantitation of protein localization by confocal microscopy, conjugates were visualized and Ab staining that localized in a region of the TIL quartered into regions parallel to the plane of target contact within the most proximal 25% area closest to the conjugate CS was scored as positive. Representative examples of the localization phenotype for each molecule are shown.
All confocal and flow cytometric analyses were conducted repetitively using individual TIL preparations, and >50 conjugates were analyzed. The percentage of conjugates showing a specific phenotype is shown in the legend to corresponding micrograph for each confocal analysis.
Calcium flux
TIL were labeled with PE-conjugated anti-CD8 (4°C for 20 min, followed by washing with HBBS + 0.010 M HEPES). Cells were then resuspended in 1 ml of HBBS + 0.010 M HEPES containing 0.1 M Fluo-4 (Molecular Probes), incubated at RT for 30 min, washed twice with HBBS + 0.010 M HEPES, then incubated at RT for 30 min to allow de-esterification of the dye. MCA38 cells were labeled for 30 min with 0.002 mg/ml 4',6'-diamidino-2-phenylindole (DAPI; Molecular Probes), washed three times with cold RPMI, mixed together in 0.4 ml of RPMI + 0.020 M HEPES containing 0.002 M Probenecid (Sigma-Aldrich), and pulse centrifuged to induce conjugate formation. Cells were resuspended (RPMI, 0.020 M HEPES, 0.002 M Probenecid) and incubated at 37°C, and aliquots were removed for analysis using a FACS LSR. Concentrations of both Fluo-4 and DAPI used in these experiments were empirically determined to ensure that no leakage of either dye occurred during the course of the experiment (thereby potentially obfuscating analysis of authentic TIL:tumor conjugates). For ionomycin treatment, calculation of fold increase was determined by dividing the mean fluorescence intensity (MFI) of TIL alone treated with ionomycin by untreated TIL alone. Calcium flux analyses were conducted four times using individual TIL preparations. Fold induction of calcium flux in TIL was then calculated using this formula: ((MFI of TIL in conjugates) – (MFI of TIL not in conjugates))/(MFI of TIL not in conjugates). To control for possible fluctuations in baseline MFI associated with variability in dye loading, calculations of fold induction were internally controlled in each experiment using the MFI of TIL not in conjugates from each time point analyzed. The low magnitude of calcium flux signal is typical for murine T cells, which are considerably smaller compared with human cell lines often used in assay of calcium flux (29).
Intracellular phosphorylation flow cytometry assay
TIL were labeled with anti-CD8-TriColor (Caltag Laboratories; 4°C for 20 min), while MCA38 cells were labeled with the Red Cell Linker kit (Sigma-Aldrich). Cells were mixed at a 3:2-TIL:MCA38 ratio in RPMI at 4°C and induced to form conjugates for various times. Samples were fixed at the indicated time points with 3.5% paraformaldehyde for 15 min, washed with FACS wash, and permeabilized (PBS containing 0.1% Triton X-100 for 5 min at 4°C). After washing with complete medium, appropriate primary Abs were added, followed by appropriate FITC secondary reagents. Samples were analyzed for the level of fluorescence (FL)1 staining by gating on conjugates (FL2, FL3 double-staining population).
Fluorescence resonance energy transfer (FRET) analysis
FRET flow cytometry analysis was performed, as described (30), using PE-conjugated anti-TCR (H57-597) (donor) and APC-conjugated anti-CD8 (acceptor) or FITC-conjugated CT-B (donor) and appropriate fluorochrome-conjugated Abs to either CD3, TCR, or CD8 (acceptor), as indicated in the figure legend. Tumor cells were labeled with either PKH26 Red or PKH57 Green Fluorescent Cell Linker Kit (Sigma-Aldrich), and conjugates were formed for 10 min, as described above. Using single laser excitation at 488 nm, induced energy transfer was monitored by examining acceptor fluorescence due to energy transfer between PE/FITC and APC/Cy5 fluorophores. FRET was measured as enhancement of fluorescence intensity by donor:acceptor pair samples over donor alone samples. There was no FRET signal when any acceptor fluorochrome reagent was excited with the wavelength of the donor fluorochrome (data not shown) and the FRET signal when only acceptor reagent was bound was set to background. There is no down-regulation of TCR during the 10-min conjugation. FRET analyses were conducted more than five times using individual TIL preparations.
Chromium release assay
Cytolysis activity of TIL was assessed in standard 51Cr release assays performed in quadruplicate wells for each E:T ratio, as described (28).
Results
Freshly isolated TIL are: nonlytic, defective in F-actin localization, and unable to recruit molecules important for lytic granule release to the TIL:tumor cell interface
Like human TIL (24, 26), murine TIL lack ex vivo lytic function (28), but recover perforin-mediated, Ag-specific cytolysis following short-term culture in vitro (Fig. 1A). However, nonlytic TIL contain granzyme B- and perforin-positive granules consistent with previous data showing that de novo protein synthesis is not required for recovery of lytic function (28). Defective cytolysis in chromium release assay was substantiated by the observation that nonlytic TIL fail to mobilize both lytic granules (Fig. 1B) and the MTOC (Fig. 1C) to the CS. As was shown for lytic function, this phenotype is reversed upon short-term in vitro culture of TIL. Nonlytic and lytic TIL have equivalent levels of granzyme B and tubulin protein, as determined by flow cytometric and immunoblot analysis, respectively (data not shown).
FIGURE 1. Freshly isolated TIL are nonlytic, fail to localize lytic granules or the MTOC, and have defective cytoskeletal polarization. A, Freshly isolated TIL are unable to lyse cognate cell targets. Specific lysis of cognate target cells varied depending upon the experiment, but ranged from 5- to 7-fold higher than matched nonlytic T cells at a 20:1 E:T cell ratio. Error bars show the SD of quadruplicate wells. B, Lytic granules fail to mobilize to the target CS. The differential interference contrast images overlaid on the single channel are shown adjacent to the corresponding single channel confocal images as 3D reconstructions, wherein arrows indicate the TIL:target cell CS. C, MTOC and F-actin localization is deficient in conjugates formed between nonlytic TIL and cognate target cells. TIL were analyzed by confocal microscopy for localization of MTOC and F-actin in conjugates using anti--tubulin and tetramethylrhodamine isothiocyanate-conjugated phalloidin, respectively. Arrows indicate the MTOC (top panels) or the TIL:target cell CS (F-actin). The percentage of TIL with polymerized actin localized at the CS is: 8% (nonlytic) and 41% (lytic). Levels of F-actin in purified nonlytic and lytic TIL are shown in histogram analyses. This experiment shows that localization of the MTOC and F-actin in nonlytic TIL is deficient in comparison with lytic TIL. D, WASp and Pyk-2 fail to localize to the target CS. TIL were analyzed for localization of WASp or Pyk-2 by confocal microscopy of TIL conjugates. The arrows indicate the TIL:target cell CS. The percentage of TIL with WASp localized at the CS is: 8% (nonlytic) and 50% (lytic). The percentage of TIL with Pyk-2 localized at the CS is: 6% (nonlytic) and 49% (lytic). Failure to recruit WASp or Pyk-2 prevents the activation of these two intracellular molecules important in lytic function.
Because actin polymerization and localization at the target cell CS are hallmarks of T cell activation, we asked whether TIL could polymerize and redistribute F-actin upon conjugate formation (Fig. 1C). Although lytic and nonlytic TIL contain equivalent levels of polymerized actin, only lytic TIL localize F-actin to the CS; nonlytic TIL fail to accumulate F-actin at the CS. Moreover, densiometric pixel analysis of phalloidin labeling at the CS indicates that the few nonlytic TIL that do localize F-actin show significantly less pixel density than lytic TIL (data not shown).
Because TCR signaling is required for affinity maturation of LFA-1, which in turn is required for F-actin localization (31, 32), we assessed the activation of Pyk-2 and WASp, important mediators of integrin and actin signaling pathways, respectively (33, 34). In nonlytic TIL, neither protein is recruited and concentrated at the CS, showing that Pyk-2 and WASp cannot be activated (Fig. 1D). In lytic TIL, Pyk-2 and WASp are recruited to the CS, allowing them to become activated, again emphasizing the reversible nature of the lytic defect.
CD2, the CD3 complex, and CD8 do not colocalize with the TCR in TIL conjugates
The failure to localize F-actin to the CS implies a defect in TIL-proximal TCR-mediated signal transduction, which most likely prevents mobilization of the MTOC and exocytosis of lytic granules (35). Because actin polymerization and localization are also thought to be important in formation of a mature IS (36), using confocal microscopy of nonpermeabilized cells, we compared nonlytic and lytic TIL for distribution of molecules excluded from the SMAC (CD43 and CD45), localized to the p-SMAC (LFA-1), and localized to the c-SMAC (TCR, CD2, the CD3 complex, CD8, and lipid rafts). TIL were analyzed after 15 min of conjugation consistent with the kinetics of formation of mature IS in CD8+ T cells (19). (Confocal analyses were performed after fixation, obviating concern that Ab reagents may perturb the distribution of target molecules and, except where noted, are shown as 3D reconstructions of full-thickness Z-stack images, which reveal the distribution of label over the entire cell.) The TCR clusters at the CS to an equal extent in nonlytic and lytic TIL (Fig. 2A), and lytic and nonlytic TIL express equivalent levels of TCR, as determined by flow cytometry (data not shown).
FIGURE 2. CD2, CD3, CD3, and CD8 do not colocalize with TCR in TIL conjugates. TIL were analyzed by confocal microscopy, as described in Materials and Methods. Single Z-stack images of labeling taken at the top surface, the middle (at the CS), and the bottom surface of the TIL are included for experiments that display a puncta phenotype to emphasize the membrane distribution of labeling. Arrows indicate the TIL:target cell CS. The percentage of TIL with TCR localized at the CS is: 42% (nonlytic) and 46% (lytic). The percentage of TIL with CD43 and CD45 excluded from the CS is: 45% (nonlytic) and 49% (lytic), and 22% (nonlytic) and 30% (lytic), respectively. LFA-1 is localized at the CS in: 45% (nonlytic) and 53% (lytic) of conjugates, and lipid rafts localized at the CS in: 29% (nonlytic) and 41% (lytic). The percentage of TIL with CD2 in puncta is 76% (nonlytic) and 10% (lytic). Five percent of nonlytic TIL show localization of CD2 at the CS, and 51% of lytic TIL localize CD2 at the CS. The percentage of TIL with CD3 in puncta is: 71% (nonlytic) and 2% (lytic). Nine percent of nonlytic TIL show localization of CD3 at the CS, and 42% of lytic TIL localize CD3 at the CS. The percentage of TIL with CD3 in puncta is: 67% (nonlytic) and 10% (lytic). Fourteen percent of nonlytic TIL show localization of CD3 at the CS, and 40% of lytic TIL localize CD3 at the CS. The percentage of TIL with CD8 in puncta is: 79% (nonlytic) and 2% (lytic). Six percent of nonlytic TIL show localization of CD8 at the CS, and 49% of lytic TIL localize CD8 at the CS. Although TCR and lipid rafts localize to the CS (A and C) and CD43 and CD45 are excluded from the immunological synapse, LFA-1 localizes to the p-SMAC (B), and CD2, CD3, CD3, and CD8 do not localize to the CS in nonlytic TIL (D).
In both lytic and nonlytic TIL, conjugates CD43 and CD45 are excluded from the CS, and LFA-1 is properly localized to the p-SMAC (Fig. 2B). (Images of LFA-1 staining in the single channel were rotated to show en face labeling at the contact site.) Considered together with the observation that the TCR is localized to the CS (along with CD3 and p56lck; Fig. 5), these findings show that IS formation in nonlytic TIL is not grossly defective. However, CD2, CD3, CD3, and CD8 (and CD8; data not shown) are distributed in puncta over the cell surface of nonlytic TIL (Fig. 5D), whereas, in lytic TIL, these proteins localize exclusively to the CS. Appearance of proteins in puncta reflects distribution on the cell surface because single Z-stack images show label distributed at the plasma membrane, not in internal structures. For emphasis, we include images of labeling taken at the top surface, the middle of the cell at the CS (indicated by arrows), and the bottom surface. Therefore, in nonlytic TIL, the structure of the TCR signaling complex is aberrant as CD8, CD2, and the CD3 complex do not associate with the TCR at the tumor cell interface.
FIGURE 5. Localization and activation of tyrosine-phosphorylated proteins and IS components in nonlytic and lytic TIL. TIL:tumor conjugates were assessed for protein localization by confocal microscopy and activation-dependent phosphorylation events after conjugation using the intracellular flow cytometric-based phosphorylation assay, as described in Materials and Methods. p-Tyr localization (A) and conjugate-dependent activation levels (B, p-Tyr immunoblot of nonlytic TIL lysates before or after 5-min activation with anti-CD3 is shown in insert); LAT localization (C); levels of LAT p-Y191 (D); ZAP70 localization (E); activation of ZAP70 p-Y493 (F); localization of CD3 (G); and localization of p56lck (H). The percentage of TIL with p-Tyr localized at the CS is: 7% (nonlytic) and 59% (lytic). Seventy-two percent of nonlytic TIL and 0% of the lytic TIL show p-Tyr in puncta. The percentage of TIL with LAT localized at the CS is: 41% (nonlytic) and 23% (lytic). (Localization of LAT appears lower in lytic TIL because the anti-LAT Ab preferentially recognizes the nonphosphorylated form (J. Houtman, unpublished observation). Thus, because LAT is phosphorylated in lytic TIL, the Ab does not label as efficiently.) The percentage of TIL with localization at the CS is: ZAP70, 53% (nonlytic) and 60% (lytic); CD3, 40% (nonlytic) and 50% (lytic); and p56lck, 39% (nonlytic) and 47% (lytic). Arrows in single channel micrographs indicate the CS. Protein tyrosine phosphorylation in nonlytic TIL is deficient, reflected by the failure to increase levels upon conjugation and to localize p-Tyr to the immunological synapse. In nonlytic TIL, conjugates ZAP70, p56lck, and CD3 localize to the CS, but triggering is incomplete as ZAP70 is only partially activated and LAT is not phosphorylated at Y191.
Because CD8, the CD3 complex, and CD2 are associated with lipid rafts upon activation (12), we considered that the failure of those proteins in nonlytic TIL to localize at the CS with tumor targets may reflect the inability of nonlytic TIL to cluster rafts. However, upon analysis of raft localization (Fig. 2C), we found the majority of raft lipid coalesces at the CS in conjugates formed from both nonlytic and lytic TIL showing normal raft formation. Therefore, lack of mature IS formation in nonlytic TIL is not due to the inability to cluster rafts at the CS.
Conjugate formation dissociates CD2, CD3, and CD8 from the TCR in nonlytic TIL
In conjugates of nonlytic TIL, CD2, CD8, and the CD3 complex do not colocalize with the TCR; however, confocal microscopy does not permit analysis of these findings at a biochemical level. To definitively make this determination, we analyzed the interactions of selected cell surface proteins with each other and with lipid rafts using FRET flow cytometric analysis of TIL:tumor conjugates or TIL not in conjugates. (Numerical data shown in histogram panels are FRET units defined as: sample MFI minus the baseline MFI divided by the baseline MFI, where baseline is the MFI detected in the acceptor fluorochrome channel of samples labeled only with the donor fluorochrome-conjugated Ab.) In nonlytic and lytic TIL, the TCR and CD8 are closely associated with each other prior to formation of conjugates (Fig. 3A). However, after conjugation for 15 min, CD8 dissociates from the TCR in nonlytic TIL (Fig. 3A). Moreover, FRET analysis performed at earlier timepoints of conjugation indicates that in nonlytic TIL, TCR and CD8 are initially associated (at 1 min of conjugation), but dissociate by 5 min (Fig. 3C). This is in contrast to lytic TIL in which CD8 and TCR remain associated with each other after 15 min of conjugation (Fig. 3A). (Analysis of CD2 and TCR in lytic TIL does not show a FRET signal in keeping with the notion that in the mature IS CD2 localizes to a point interior to the LFA-1 ring, but clearly excluded from the c-SMAC that contains the TCR (37).)
FIGURE 3. FRET analysis. TIL conjugates were gated in scatter analysis, and FRET was recorded at 15 min of conjugate formation, as described in Materials and Methods (right panels, A and B). In control experiments, labeled TIL were analyzed in the absence of target cells (left panels, A and B). Control stains of donor alone are shown in the filled histogram, and the FRET signal is shown in the black tracings. (Numerical data shown are FRET units defined as: (sample MFI minus the baseline MFI) divided by the baseline MFI, where baseline is the MFI detected in the acceptor fluorochrome channel of samples labeled only with the donor fluorochrome-conjugated Ab.) Donor fluorochrome is the first molecule in the pair indicated in the x-axis label. A, CD8 and TCR associate in nonlytic TIL alone, but do not remain associated after conjugate formation. However, lytic TIL show CD8 and TCR association before and after conjugate formation. B, FRET analyses of lipid rafts (donor) associated with either CD3, TCR, or CD8 (acceptors), as indicated in the figure. Confocal microscopic analysis of nonlytic TIL in conjugates simultaneously stained for CT-B (green) and CD3 or CD8 (red) is shown as inserts in histograms. Arrows indicate the lipid raft at the CS. CD8 is found associated with raft lipid in both nonlytic and lytic TIL before conjugate formation, but only in lytic TIL do these two molecules remain associated after conjugation. C, FRET analysis of nonlytic TIL at 1, 5, and 15 min of conjugation.
We also analyzed interaction between lipid rafts and either CD3, CD8, or TCR after conjugate formation to further confirm association of these molecules with rafts in lytic TIL, and their exclusion in nonlytic TIL (Fig. 3B). Before conjugate formation, CD8 and CD3 associate with raft lipid, while TCR is excluded (Fig. 3B). However, upon conjugate formation, CD8 in nonlytic TIL dissociates from lipid rafts, while lytic TIL maintain CD8:raft association at 15 min. CD3 association with lipid rafts showed a FRET signal in conjugates of nonlytic TIL, albeit reduced in comparison with lytic TIL (Fig. 3B). This observation supports the confocal microscopy analysis (Fig. 3B), which showed appearance of raft lipids on the periphery of CD3+ puncta in nonlytic TIL conjugates, suggesting that some raft lipid is associated with CD3 even though the majority of lipid is found in a large aggregate that excludes CD3 (green label). In keeping with current notions of TCR inclusion in rafts (38), raft association of TCR is activation dependent because TIL alone (not in conjugates) showed no FRET signal. TCR rapidly associates with lipid rafts in nonlytic TIL conjugates (Fig. 3C), supporting the confocal microscopy, but the signal was low compared with lytic TIL, perhaps reflecting the relatively modest state of activation of nonlytic TIL.
When observed over a time course of conjugate formation (Fig. 3C), in nonlytic TIL lipid rafts associate with TCR, CD3, and CD8 at 1 min, but by 5 min of conjugation, CD8 dissociate from rafts with similar kinetics of TCR/CD8 dissociation.
Nonlytic TIL do not flux calcium
Because TCR signal transduction is required for lytic function, we compared lytic and nonlytic TIL for biochemical evidence of TCR-mediated signaling by analysis of calcium flux after different times of conjugate formation. For these experiments, TIL were labeled with PE-conjugated anti-CD8 and loaded with Fluo-4, and conjugates formed with DAPI-labeled MCA38 target cells for different times before fixation. Concentrations of dyes were used that permitted analysis of scatter patterns for gating of TIL in conjugates or nonconjugated TIL (Fig. 4). Nonlytic TIL display a 3- to 4-fold increase in calcium flux in response to ionomycin (green open histogram), as we noted previously using Indo-2 as calcium indicator (28), but do show an increase in calcium in response to conjugate formation (gate R2, purple filled histograms) remaining at the same level as unconjugated TIL (gate R3, red open histograms). However, elevation of calcium is observed in lytic TIL conjugates at the earliest time analyzed, 30 s. Lytic TIL maintain baseline calcium levels unless conjugates are formed (gate R3, red open histograms), showing that calcium flux is dependent upon Ag stimulation. Lytic TIL conjugates flux calcium to the same extent as that induced by ionophore, correlating recovery of lytic function with restoration of TCR-mediated signal transduction.
FIGURE 4. Nonlytic TIL do not flux calcium in T cell:target cell conjugates. A, By gating on appropriate populations of cells (shown in scatter plots), analyses were made on both TIL not in conjugates (gate R3) and on TIL conjugates (as described in Materials and Methods). Aliquots of TIL were separately activated with 0.01 M ionomycin for maximal levels of calcium flux (green trace) and are overlaid with TIL in conjugates (purple filled) and TIL not in conjugates (red trace). The average baseline MFI of calcium levels of TIL not in conjugates was 18 (nonlytic) and 15 (lytic). Conjugates were formed for 2.5 min. B, The table compares fold induction of calcium flux between nonlytic () and lytic () TIL at different times in multiple experiments (n = 4). This demonstrates that while lytic TIL flux calcium in conjugates, conjugate formation does not induce calcium flux in nonlytic TIL, although these cells are inducible by ionomycin treatment. C, Activation of PLC-1 was assessed after 2.5 min of conjugation using activation-specific Ab and showed that in contrast to lytic TIL, nonlytic TIL display no activated PLC-1 upon conjugate formation (arrow points to CS with target).
These data associate the lytic deficit of freshly isolated TIL with a block in TCR-mediated signaling. This observation furthermore suggests that the defect in signal transduction is both profound and proximal, affecting multiple downstream activation pathways and consequent lytic function. To further define our analysis of the proximal TCR signaling defect, we performed confocal microscopy of activated PLC-1 using phospho-specific Ab (Y783) (Fig. 4C). Low levels of active PLC-1 are detectable in lytic TIL at 1 min of conjugation (data not shown) and significant labeling localized to the CS is detected by 2.5 min. Even though both nonlytic and lytic TIL contain equivalent levels of nonactivated PLC-1 (data not shown), nonlytic TIL do not express detectable activated PLC-1 at any time after conjugate formation. The absence of phosphorylated PLC-1 in nonlytic TIL and the presence in lytic TIL (properly localized at the CS) corroborate the calcium flux data. These results suggest that the signaling defect in nonlytic TIL is located at a more proximal step of the activation pathway.
Defective activation and localization of tyrosine-phosphorylated proteins in nonlytic TIL
Because nonlytic TIL fail to increase calcium levels in response to conjugate formation, we asked whether recognition of cognate Ag induced tyrosine kinase activation as evidenced by an increase in protein tyrosine phosphorylation and CS localization of phosphotyrosinylated proteins, hallmarks of TCR activation. Nonlytic TIL in conjugates labeled brightly for phosphotyrosine (p-Tyr) (Fig. 5A); however, the subcellular distribution of label was dramatically distinct from lytic TIL in that p-Tyr in nonlytic TIL appeared in multiple small puncta, whereas p-Tyr in lytic TIL was localized primarily to the target cell CS. Flow cytometric analysis of conjugates showed lytic TIL increase p-Tyr levels with kinetics typical of activated T cells: increased levels appeared 2.5 min after initial conjugate formation and begin to decrease after 10 min. However, nonlytic TIL do not increase p-Tyr levels upon conjugate formation (Fig. 5B), although p-Tyr increases after stimulation of TIL with anti-CD3 (insert), showing that tyrosine kinase activity can be induced. Considered together with the deficit in calcium flux in nonlytic TIL, this observation suggests that upon conjugation with cognate target cells, activation of proximal and distal tyrosine kinases is defective in nonlytic TIL, and tyrosine-phosphorylated proteins are not reorganized.
p56lck, CD3, ZAP70, and LAT localize to the target cell CS in nonlytic TIL, but ZAP70 activation is incomplete
Site-specific phosphorylation and subcellular localization of proteins involved in signal transduction reflect the activation status of T cells; therefore, we determined the localization and activation status of additional essential components of the signaling complex: LAT, ZAP70, CD3, and p56lck. Confocal analyses of TIL in conjugates show LAT localization at the CS in both nonlytic and lytic TIL (Fig. 5C). However, flow analysis using an activation-specific anti-LAT Ab (p-Y191) showed that LAT is not phosphorylated in nonlytic TIL, but lytic TIL display normal LAT phosphorylation kinetics (Fig. 5D). This finding illustrates that defective signaling in nonlytic TIL is not due to a potential failure to recruit LAT to the signaling complex, but rather that LAT is not efficiently phosphorylated. Because LAT is the immediate downstream target of ZAP70, the localization of ZAP70 was analyzed by confocal microscopy (Fig. 5E) and its activation status by flow cytometry (Fig. 5F). In lytic TIL, ZAP70 localizes to the CS and is increasingly activated as a function of time in conjugates (average maximum increase of 60% in MFI). However, in nonlytic TIL, although ZAP70 localizes to the CS (implying that CD3 is phosphorylated), phosphorylation of ZAP70 Y493 is only modest (average maximum increase of 30% of MFI). This level of activated ZAP70 in nonlytic TIL is unable to propagate the activation signal, as p-Tyr is not increased and cells do not increase calcium levels. Nonlytic and lytic TIL contain similar levels of p56lck and CD3 proteins (data not shown), and both CD3 (Fig. 5G) and p56lck (Fig. 5H) are localized to the CS in nonlytic and lytic TIL; thus, weak ZAP70 activation in nonlytic TIL is not due to either deficient levels nor exclusion of CD3 or p56lck from the signaling complex.
Conjugate formation rapidly inactivates p56lck in nonlytic TIL
The activation status of p56lck was examined by flow cytometric analysis of phosphorylation of the inhibitory motif (containing Y505). Both lytic and nonlytic TIL have equivalent levels of phospho-Y505 immediately upon conjugation; however, Y505 is rapidly phosphorylated in nonlytic TIL, 1 min after target cell conjugation (Fig. 6A). The increase in Y505 phosphorylation averaged from multiple experimental repetitions reveals that p56lck in nonlytic TIL is significantly inactivated upon conjugation. Conversely, this motif is not significantly phosphorylated in lytic TIL until a much later time of conjugation, 30 min.
FIGURE 6. Localization and activation of proximal signaling components in nonlytic and lytic TIL. Cellular levels of phospho-p56lck Y505 were assessed using flow cytometry, as described in Materials and Methods. Baseline levels of p-Y505 in TIL in conjugates (A, histogram) and conjugation-induced levels (A, chart) are shown. Confocal images of TIL:tumor conjugates after 2.5 min show protein localization of Csk (B); Shp-2 (C); Shp-1 (D); and phospho-ERK (E). Single Z-stack images of top, the middle, and bottom slice are included to show inner membrane pERK distribution. The percentage of TIL with localization at the CS is: Csk, 27% (nonlytic) and 8% (lytic); Shp-1, 36% (nonlytic) and 13% (lytic); Shp-2, 10% (nonlytic) and 43% (lytic); and phospho-ERK, 7% (nonlytic) and 41% (lytic). Sixty-eight percent of nonlytic TIL and 0% of lytic TIL show pERK in puncta.
The p56lck Y505 is phosphorylated by tyrosine kinase Csk (itself phosphorylated by cAMP-controlled protein kinase A (39)), which is recruited into proximity with p56lck by Src homology 2 interaction with the raft-associated adapter Cbp (or phosphoprotein associated with glycosphingolipid-enriched microdomains (PAG)) (40). Therefore, we examined Csk localization (Fig. 6B) and found that the percentage of nonlytic TIL that localize Csk to the CS is 3- to 4-fold greater than lytic TIL at early times of conjugation, consistent with the observation that p56lck Y505 is phosphorylated. Localization of Csk to Cbp/PAG in the plasma membrane is regulated by the reciprocal activities of p56lck (and/or Fyn) and the phosphatase Shp-2 (40, 41). Thus, enhanced localization of Csk to the CS in nonlytic TIL could result from either increased phosphorylation or decreased dephosphorylation of Cbp/PAG. We examined Shp-2 localization in TIL conjugates and found that Shp-2 is localized to the CS in lytic TIL, whereas in nonlytic TIL Shp-2 does not localize and is mainly cytoplasmic (Fig. 6C). This finding supports the notion that phosphorylation of p56lck Y505 in nonlytic TIL is regulated by recruitment of Csk to the TIL:tumor interface.
Considered collectively, the observations that in nonlytic TIL p56lck and CD3 localize to the CS, but p56lck Y505 is phosphorylated upon target cell contact demonstrate that proximal TCR signaling becomes arrested at p56lck. To better understand the basis of the signaling block, the potential role of additional negative regulators of signaling was investigated. CD43 and CD45 localize equivalently in lytic and nonlytic TIL, both being excluded from the CS (Fig. 2), and CD5 localizes to the CS equivalently in both lytic and nonlytic TIL (data not shown). These observations diminish the likelihood of involvement of CD5, CD43, and CD45 in regulation of TIL signaling. However, localization of Shp-1 to the CS is 3-fold greater in nonlytic TIL compared with lytic TIL (Fig. 6D), supporting a potential role of this phosphatase in regulation of signaling in nonlytic TIL, as has recently been proposed (42, 43). In contrast, ERK can act as a positive regulator of TCR signaling by phosphorylating p56lck (at Ser59), preventing Shp-1 binding to p56lck. We examined the localization of activated ERK and found that pERK localized to the CS in lytic TIL, but in nonlytic TIL is sequestered to the inside of the cell membrane in a punctate pattern similar to that seen on the exterior of the cell for CD2, CD3, and CD8. This observation implies that ERK cannot provide positive feedback regulation of Shp-1 function in nonlytic TIL (44).
Discussion
The current study was predicated on the considerations that, because lytic function is dependent upon TCR-mediated signaling, and TIL are unable to exocytose lytic granules, residence of antitumor T cells in the tumor microenvironment may induce defective signal transduction. Supporting this notion, nonlytic TIL are deficient in protein tyrosine phosphorylation; thus, proximal tyrosine kinases are not activated upon recognition of cognate target cells. In addition, upon conjugate formation with cognate target cells in vitro, purified TIL are unable to flux calcium. This conclusion is supported by the observation that multiple substrates are not phosphorylated and ZAP70 is only modestly activated, deficiencies that undoubtedly underlie lytic dysfunction. Despite defective proximal signaling in nonlytic TIL, conjugation with cognate tumor cells induces IS formation, but the synapse is aberrant in that CD2, CD3 heterodimers, and CD8 do not colocalize with the TCR; instead, they localize in cell surface puncta, a unique phenotype not described in the literature. Confocal microscopy and FRET analysis show that the TCR complex is initially intact in nonlytic TIL, but is induced to dissociate upon conjugation with cognate target cells. The biochemical basis underlying the physical dissociation of selected molecules from the TIL TCR complex is currently unknown and may be multifaceted in nature.
Because TCR-mediated signaling is required for affinity up-regulation of LFA-1 (45, 46), which is in turn required for capacitation of F-actin-mediated signaling events (31, 47, 48), nonlytic TIL deficiencies in adhesion-dependent phenomena (decreased conjugation frequency, the failure to activate Pyk-2, and defective MTOC mobilization) most likely reflect the proximal nature of the TCR signaling defect. In support of this notion, CD4+ T cells lacking p56lck function have been shown by Burkhardt and colleagues (35) to have adhesion defects similar to those of CD8+ TIL. Similar to our findings, integrin-dependent signaling was also shown to be required for accumulation of F-actin filaments at the CS (formed between CD4+ T cells and peptide-pulsed APC), which facilitates sustained T cell activation (49). Therefore, the failure to affinity mature LFA-1 due to tumor-induced down-regulation of p56lck activity most likely prevents adhesion-dependent signaling possibly required to maintain association of the TCR with the CD3 complex, as well as accounting for TIL adhesion defects.
The physical dissociation of certain components of the TCR complex calls into question several notions concerning the initial stages of TCR-mediated signal transduction in CD8+ T cells. As was demonstrated by confocal microscopy and FRET analyses, CD8 and TCR dissociate in nonlytic TIL conjugates; nevertheless, TCR clusters at the CS. This observation suggests that CD8 interaction with the TCR is not required for stabilization of the Ag receptor at the CS. Our data are compatible with the notion that intact TCR complexes interact with Ag:MHC class I and initiate signaling; however, rapidly after cell triggering, coincident with both TCR clustering at the tumor cell:TIL interface and inactivation of p56lck function, CD8, CD2, and the CD3 complex dissociate from the TCR.
A related issue concerns the observation that, despite the failure of CD8 to remain associated with the TCR in nonlytic TIL conjugates, p56lck localizes to the CS. According to current notions about the structure of the TCR complex, CD8 is constitutively associated with lipid rafts due to palmitoylation of the -chain (5), where it associates with p56lck, whose residency in rafts is in turn mediated by activation-induced S-acylation (50). Therefore, upon recruitment to the TCR complex, by virtue of CD8 association with TCR, p56lck can phosphorylate CD3 and initiate T cell activation. The finding that p56lck and the TCR localize at the CS in nonlytic TIL conjugates coincident with TCR association with the raft suggests that p56lck does not require interaction with CD8 for stabilization at the IS (51, 52). The FRET analysis shows that before conjugation, CD8 is associated with raft lipid in nonlytic TIL and also interacts with the TCR. Upon conjugation, the TCR rapidly becomes raft associated (1 min), a time when CD8 is still associated with raft lipid, but immediately thereafter, CD8 dissociates from both TCR and the raft. This transient state wherein CD8 interacts with both TCR and the raft may be sufficient to recruit p56lck to the TCR upon activation, after which CD8 is no longer required to maintain p56lck association with TCR.
Recently, Valitutti and colleagues (53) have described experiments wherein a human CD8+ T cell clone was conjugated with peptide-pulsed B cells under conditions of low peptide concentration. In this model, the MTOC and lytic granules mobilize to the CS within 5–10 min of conjugation typical of CD8+ T cells, but localization of CD2 is weak, which was interpreted to mean that IS formation is not required for granule exocytosis. Curiously, although cells were not analyzed for localization of additional IS components and CD2 was not totally excluded, under these conditions of stimulation, CTL were not activated, as phosphotyrosinylated proteins did not strongly accumulate at the CS. In another recent paper by Davis and colleagues (54), the lack of mature IS formation in a murine T cell clone (stimulated by low concentration of peptide pulsed onto B cell APC) did not limit killing of target cells in vitro. However, in that model, lysis assays were conducted for 16 h, suggesting a profound experimental difference with our model, and lysis is Fas ligand mediated because blocking Fas receptor inhibited killing. (As we have shown previously, TIL lysis of targets is perforin and granzyme mediated.) Fas ligand-mediated cytotoxicity is thought to be important in regulation of the immune response, but less so as an effector mechanism in host defense (1), and may be independent of IS stability. Because the data of Valitutti and Davis were achieved under conditions of limiting Ag and used hemopoietic-derived APC, the relevancy of those observations in consideration of TIL lytic function is unclear.
Our data show that the block in TIL TCR signaling most likely results from rapid, conjugation-dependent inactivation of p56lck, depicted diagrammatically in Fig. 7. In normal T cells (or in lytic TIL), p56lck exists in a balance between the inactive and active forms due to opposing effects of Csk and CD45 on the phosphorylation of an inhibitory motif containing Y505: upon TCR ligation (when CD45 is colocalized with the TCR complex), Csk dissociates from the membrane, resulting in rapid Y505 dephosphorylation and coincident autoactivation, enabling phosphorylation of CD3, CD3 subunits, and ZAP70 (55, 56). Thus, tonic inhibition of TCR-mediated signaling permits fine control of modulating immune response (57). Our analysis of the status of p56lck using pY505-specific Ab showed that this inhibitory motif is equivalently phosphorylated in both lytic and nonlytic TIL immediately upon conjugation with target cells. However, Y505 in nonlytic TIL is substantially phosphorylated within 1 min of conjugation compared with lytic TIL, an event that rapidly inactivates p56lck kinase function (58, 59). In contrast, Y505 in lytic TIL does not become phosphorylated until late time of conjugation. Because phosphorylation of the inhibitory motif results in potent p56lck inactivation, small increases in phospho-Y505 are likely to be physiologically relevant (60). Supporting the finding of conjugation-dependent p56lck inactivation, at early times of conjugation Csk is localized at the CS, whereas in lytic TIL Csk is predominantly cytosolic. The finding that Shp-2 localizes at the CS of lytic TIL, but in nonlytic TIL is cytoplasmic, supports the notion that in nonlytic TIL Csk is recruited into proximity with its substrate p56lck (41).
FIGURE 7. TCR signaling in TIL. a, Lytic TIL. Because lytic TIL are no longer under the influence of the tumor, TCR stimulation induces the signaling cascade, which results in full activation of ZAP-70. This initiates subsequent downstream events such as: calcium flux, activation of the Ras/MAPK pathway, and MTOC and lytic granule mobilization. Because Shp-2 is recruited to the tumor cell:TIL contact site rapidly upon recognition of tumor cells, it can dephosphorylate Cbp, thereby blocking recruitment of Csk and proline-glutamic acid-, serine- and threonine-enriched protein tyrosine phosphatase (PEP) into proximity to their substrate p56lck. b, Nonlytic TIL. Upon TCR stimulation, nonlytic TIL initiate triggering: activation of p56lck, phosphorylation of CD3, and recruitment of ZAP-70. However, due to an unknown contribution by the tumor, transduction of the signal cannot progress past recruitment of ZAP-70. Instead, p56lck feedback inhibition pathways are activated involving recruitment to the CS of the negative regulator Csk, which phosphorylates p56lck at the inhibitory motif (Y505). (PEP recruitment is suggested by the recruitment of its interacting partner Csk, but has yet to be determined.) Csk is recruited to the CS by its binding partner Cbp. In addition, Shp-1 is recruited to the CS, where it may interact with its substrates. This causes early termination of the TCR signaling cascade, preventing calcium flux and MTOC and lytic granule recruitment.
In comparison with lytic TIL, nonlytic TIL in conjugates show CS localization of the tyrosine phosphatase Shp-1. Shp-1 can dephosphorylate the activation motif of p56lck (Y394), but may also dephosphorylate CD3, ZAP70, and LAT (43, 61, 62), therein uncoupling T cell triggering from activation, the phenotype of nonlytic TIL. Because ZAP70 localizes to the CS, indicating phosphorylation of CD3 (producing at least the p21 form (63)), our data suggest that Shp-1 may dephosphorylate ZAP70 in nonlytic TIL, therein blocking signal propagation. Supporting this notion is the finding that ZAP70 is only partially phosphorylated. An unresolved question is what is the mechanism of Shp-1 recruitment to the CS? The observation that pERK localizes to the CS in lytic TIL, but is excluded in nonlytic TIL supports the supposition that ERK may regulate the susceptibility of p56lck as a target of Shp-1, but does not identify the interacting partner with which Shp-1 binds. Our findings differ from those of Germain and colleagues (43) in that ERK is phosphorylated in nonlytic TIL, but is localized to puncta at the cytoplasmic face of the plasma membrane, not at the CS; therefore, it cannot prevent Shp-1 binding at the CS. Although purely speculative, it is possible that pERK may preferentially bind to a component of the puncta, therein sequestering it from p56lck, which localizes with the TCR at the TIL:tumor target interface.
Human cancers frequently contain CD8+ TIL, which are typically nonlytic, but Ag specific (24, 26). CD8+ T cells with this phenotype also characterize several chronic diseases resulting from infection by: Trypanosoma cruzi (64), SIV (65), HIV (66), human hepatitis C virus (67), herpes simplex virus (68), Friend virus (69), and lymphocytic choriomeningitis virus (70). The phenotype of defective CD8+ T cells in lymphocytic choriomeningitis virus infection closely resembles murine TIL that we have previously described: T cells are activated memory/effector cells, but are nonlytic, even when assessed by redirected assay (28). CD8+ T cells in the above-mentioned models of infection have been characterized as lytic defective on the basis of in vitro lysis assays without analysis of the biochemical basis for the nonlytic phenotype. (One exception was the study by Appay et al. (66), which showed that HIV-specific T cells were immature in that perforin protein was not present, a phenotype distinct from that which we have described.) Therefore, it is unclear whether the mechanism for induction of lytic dysfunction we describe is also used in chronic viral infection.
Collectively, our data suggest that proximal TCR-mediated signaling in nonlytic TIL is rapidly blocked after conjugation with cognate tumor targets. The signaling defect possibly involves two inhibitory mechanisms, Csk-mediated inactivation of p56lck function and enhanced Shp-1 activity, both of which may conspire to prevent proper activation of proximal tyrosine kinase activity, affinity up-regulation of LFA-1, and calcium flux, ultimately preventing exocytosis of lytic granules. Importantly, the TIL lytic defect is an acquired property restricted to T cells within the tumor microenvironment because in vivo systemic T cell function is not affected by tumor growth (71), whereas TIL are characterized by the inability to mobilize the MTOC to the CS and exocytose lytic granules. Acquired transient TIL lytic dysfunction in TIL has also been described in several transgenic TCR mouse tumor models (72, 73). Consideration of the observation that human TIL are Ag specific, but nonlytic, together with our description of defective lytic function of murine TIL, supports the notion that tumor-induced inhibition of TIL lytic function is a common characteristic that may contribute to tumor growth in the presence of antitumor immune response. Tumor-induced lytic dysfunction also may restrict T cell-based immunotherapy of cancer.
Acknowledgments
We thank Gabriele Campi and Mike Dustin for anti-CD45 Ab; Andrew Chan for anti-ZAP70 Y319 Ab; Lawrence Samelson for anti-LAT and anti-CD3 Abs; David Wiest for anti-CD3 Ab; Dan Rifkin and Dan Littman, respectively, for use of the FACScan and LSR equipment; Ed Ziff for use of the Zeiss microscope and Brian Fernholz for generous assistance in its usage; Abraham Kupfer for technical advice; and Andrew Bush, Andreas Dieffenbach, M. Dustin, and Stanislav Vukmanovic for editorial suggestions and/or helpful comments.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 M.K. and D.S. are cofirst authors of this manuscript and were supported by National Institutes of Health Training Grants T32 GM-07238 and T32 CA-09161, respectively.
2 Address correspondence and reprint requests to Dr. Alan B. Frey, Department of Cell Biology, New York University School of Medicine, 550 First Avenue, New York, NY 10016. E-mail address: freya01@med.nyu.edu
3 Abbreviations used in this paper: LAT, linker for activation of T cells; CS, tumor-infiltrating lymphocyte:target contact site; Csk, COOH-terminal Src kinase; c-SMAC, central zone supramolecular activation complex; CT-B, cholera toxin B; 3D, three-dimensional; DAPI, 4',6'-diamidino-2-phenylindole; FL, fluorescence; FRET, fluorescence resonance energy transfer; IS, immunological synapse; MFI, mean fluorescence index; MTOC, microtubule organizing center; PAG, phosphoprotein associated with glycosphingolipid-enriched microdomains; PLC, phospholipase C; p-SMAC, peripheral zone supramolecular activation complex; p-Tyr, phosphotyrosine; Pyk-2, proline-rich tyrosine kinase-2; RT, room temperature; Shp, Src homology 2 domain-containing protein phosphatase 1; SMAC, supramolecular activation complex; TIL, tumor-infiltrating lymphocyte; WASp, Wiskott-Aldrich syndrome protein.
Received for publication August 18, 2004. Accepted for publication November 11, 2004.
References
Kagi, D., B. Ledermann, K. Burki, R. M. Zinkernagel, H. Hengartner. 1996. Molecular mechanisms of lymphocyte-mediated cytotoxicity and their role in immunological protection and pathogenesis in vivo. Annu. Rev. Immunol. 14:207
Feito, M. J., A. Jimenez-Perianez, G. Ojeda, A. Sanchez, P. Portoles, J. M. Rojo. 2002. The TCR/CD3 complex: molecular interactions in a changing structure. Arch. Immunol. Ther. Exp. 50:263.
Fernandez-Miguel, G., B. Alarcon, A. Iglesias, H. Bluethmann, M. Alvarez-Mon, E. Sanz, A. de la Hera. 1999. Multivalent structure of an T cell receptor. Proc. Natl. Acad. Sci. USA 96:1547.
Zamoyska, R.. 1998. CD4 and CD8: modulators of T-cell receptor recognition of antigen and of immune responses?. Curr. Opin. Immunol. 10:82.
Arcaro, A., C. Gregoire, T. R. Bakker, L. Baldi, M. Jordan, L. Goffin, N. Boucheron, F. Wurm, P. A. van der Merwe, B. Malissen, I. F. Luescher. 2001. CD8 endows CD8 with efficient coreceptor function by coupling T cell receptor/CD3 to raft-associated CD8/p56lck complexes. J. Exp. Med. 194:1485.
Bachmann, M. F., A. Gallimore, S. Linkert, V. Cerundolo, A. Lanzavecchia, M. Kopf, A. Viola. 1999. Developmental regulation of Lck targeting to the CD8 coreceptor controls signaling in naive and memory T cells. J. Exp. Med. 189:1521.
Luescher, I. F., E. Vivier, A. Layer, J. Mahiou, F. Godeau, B. Malissen, P. Romero. 1995. CD8 modulation of T-cell antigen receptor-ligand interactions on living cytotoxic T lymphocytes. Nature 373:353
Renard, V., P. Romero, E. Vivier, B. Malissen, I. F. Luescher. 1996. CD8 increases CD8 coreceptor function and participation in TCR-ligand binding. J. Exp. Med. 184:2439
Fuller, C. L., K. S. Ravichandran, V. L. Braciale. 1999. Phosphatidylinositol 3-kinase-dependent and -independent cytolytic effector functions. J. Immunol. 162:6337.
Nesic, D., K. G. Jhaver, S. Vukmanovic. 1997. The role of protein kinase C in CD8+ T lymphocyte effector responses. J. Immunol. 159:582.(Mythili Koneru1, David Sc)
CD8+ tumor-infiltrating lymphocytes (TIL) are severely deficient in cytolysis, a defect that may permit tumor escape from immune-mediated destruction. Because lytic function is dependent upon TCR signaling, we have tested the hypothesis that primary TIL have defective signaling by analysis of the localization and activation status of TIL proteins important in TCR-mediated signaling. Upon conjugate formation with cognate target cells in vitro, TIL do not recruit granzyme B+ granules, the microtubule-organizing center, F-actin, Wiskott-Aldrich syndrome protein, nor proline rich tyrosine kinase-2 to the target cell contact site. In addition, TIL do not flux calcium nor demonstrate proximal tyrosine kinase activity, deficiencies likely to underlie failure to fully activate the lytic machinery. Confocal microscopy and fluorescence resonance energy transfer analyses demonstrate that TIL are triggered by conjugate formation in that the TCR, p56lck, CD3, LFA-1, lipid rafts, ZAP70, and linker for activation of T cells localize at the TIL:tumor cell contact site, and CD43 and CD45 are excluded. However, proximal TCR signaling is blocked upon conjugate formation because the inhibitory motif of p56lck is rapidly phosphorylated (Y505) and COOH-terminal Src kinase is recruited to the contact site, while Src homology 2 domain-containing protein phosphatase 2 is cytoplasmic. Our data support a novel mechanism explaining how tumor-induced inactivation of proximal TCR signaling regulates lytic function of antitumor T cells.
Introduction
Lytic function of CD8+ effector T cells is initiated by recognition of target cells, which results in rapid activation (1). Activation requires oligomerization of the TCR dimer with cognate ligand/MHC class I molecules that induces either a conformational change in, or low-order aggregation of, the TCR (2, 3). Upon binding to target cells, the coreceptor CD8 is recruited to the TCR complex, which serves to stabilize binding to cognate ligand:MHC class I, therein enhancing sensitivity of Ag recognition (4). CD8 association with the TCR also recruits Src-related kinase p56lck into close proximity with its substrates, primarily CD3 (5, 6, 7, 8). Phosphorylated CD3 recruits ZAP70, which in turn phosphorylates linker for activation of T cells (LAT),3 followed by recruitment of additional proteins to phosphorylated LAT, culminating in the activation of phospholipase C-1 (PLC-1). Following this early activation phase (30 s), calcium flux is initiated, resulting in coordinated activation of additional downstream signaling pathways (9, 10, 11).
Assembly of the signaling complex results in the formation of dynamic membrane cholesterol- and sphingolipid-enriched domains, termed lipid rafts, into which the TCR:CD8 complex partitions (12, 13, 14, 15). The raft serves as a physical platform upon which various cell signaling molecules assemble after triggering, defined in biochemical terms as phosphorylation of CD3 and CD3 heterodimers (12, 15, 16). In quiescent cells, and immediately upon conjugate formation with APC or target cells, rafts are small (70 nm) (17), but reorganize after activation to form a large aggregate that colocalizes with the immunological synapse (IS) (18). Upon oligomerization of the TCR complex, proteins constitutively associated with raft lipids, by virtue of posttranslational S-acylation or addition of GPI moieties (e.g., Thy1, p56lck, CD8, or LAT), coalesce coincident with raft formation, rapidly increasing the local concentration of raft lipids and associated proteins.
Within 5–10 min of interaction between CD8+ T cells and MHC peptide on an APC, cell surface molecules segregate into different regions, forming an IS (19). A mature IS is characterized by proteins arranged in a bull’s eye pattern with an arrangement of supramolecular activation clusters (SMAC) in a central zone (c-SMAC) containing lipid rafts/TCR/CD3/costimulatory receptors, a peripheral zone (p-SMAC) containing adhesion receptors, primarily LFA-1, and a zone of additional proteins that are excluded from the core region. In addition to faster kinetics of formation compared with CD4+ T cells (20, 21), IS formation in CD8+ T cells has been shown to precede effector functions such as reorganization of the cytoskeleton, mobilization of the microtubule organizing center (MTOC) to the IS, and release of lytic granules (19, 22). Although there are several different theories on IS function (13, 18, 23), the exact function remains unknown.
CD8+ effector T cells that infiltrate tumors (TIL) provide an excellent natural model for study of signaling requirements for CTL function because they have severely defective lytic function (24, 25, 26). Freshly isolated TIL are nonlytic (CD44+CD62LlowCD69+) memory/effector cell, but recover lytic function following brief in vitro culture permitting direct comparison of lytic and nonlytic TIL purified from the same tumor (27, 28). The absence of manipulation during the in vitro culture makes this system ideal for study of the tumor-induced lytic defect. Because exocytosis of lytic granules may be dependent upon IS formation (19) and nonlytic TIL cannot exocytose granules (28), we considered that TIL IS formation or function is defective.
Using a combination of confocal microscopy and biochemical assays, we investigated the localization and activation status of proteins important in signal transduction, IS formation, and lytic function in conjugates formed between TIL and cognate tumor cells. Signal transduction in freshly isolated, nonlytic TIL is blocked at a proximal step because LAT is not phosphorylated and ZAP70, although recruited to the TIL:target contact site (CS), is only weakly phosphorylated. Significantly, the inhibitory motif in p56lck (Y505) becomes rapidly phosphorylated upon binding to cognate tumor cells. Consistent with inactivation of p56lck, COOH-terminal Src kinase (Csk) is localized at the TIL:tumor CS and Src homology 2 domain-containing protein phosphatase 2 (Shp-2) remains cytoplasmic. In addition, Shp-1 localizes at the CS, where it may mediate deactivation of various SH3-containing proteins (such as ZAP70), therein preventing propagation of the activation signal. Furthermore, we show that upon contact with cognate target cells, nonlytic TIL assemble many signaling components with kinetics typical of activated CD8+ T cells, showing that nonlytic TIL are triggered by conjugate formation. However, CD2, the CD3 complex, and CD8, which associate with the TCR before conjugate formation, rapidly dissociate and are excluded from the CS. Tumor-induced disruption of T cell activation at a point downstream of triggering, therein blocking proximal tyrosine kinase activity, calcium flux, and dependent lytic function, is a novel mechanism for inhibition of the CD8+ T cell effector phase.
Materials and Methods
Mice
C57BL/6 male mice were obtained from The Jackson Laboratory, were housed five per cage in a barrier facility, and were maintained on a 12-h light/dark cycle (7 a.m. to 7 p.m.) with ad libitum access to food and water. A sentinel program revealed that mice were mouse hepatitis virus negative and the MCA38 cell line is mouse hepatitis virus negative, as assessed by mouse assessment profile service testing. Experiments involving animals were conducted with the approval of the New York University School of Medicine Committee on Animal Research.
Tumor
MCA38 adenocarcinoma (a gift from Y. Liu, Ohio State University, Columbus, OH) was passaged from tissue culture plasticware by incubation in HBSS containing 3 mM EDTA, followed by washing three times in HBSS. Viability was determined by trypan blue dye exclusion, and 2–3 x 106 cells were injected i.p. in a volume of 0.2–0.3 ml of HBSS for tumor induction. Cells were passaged in vitro for 3–5 wk, following which new frozen stocks were thawed for usage.
Tissue culture
RPMI 1640 medium (BioWhittaker) was used for growth of MCA-38 cells and for culture of T cells, as described (28).
Isolation of TIL
Tumors were dissected (any necrotic portion was discarded), mechanically disrupted by passage through a tissue press, and enzymatically digested into single cell suspensions, and TIL were isolated by immunomagnetic separation using type LS+ columns and anti-CD8-conjugated magnetic beads (Miltenyi Biotec), as described previously (27). (Potentially inhibitory effects of the immunobead isolation protocol upon TIL lytic function is discounted because after recovery in vitro we have reisolated TIL with anti-CD8 magnetic beads (including exposure to tumor digest enzymes) and those cells are lytic. As further control, we have prepared CTL by primary MLR in vitro, followed by isolation with anti-CD8 immunobeads. These cells are fully lytic, again showing that the isolation protocol is without effect on lytic activity.) In each experiment, aliquots of isolated T cells were analyzed by flow cytometry and were routinely 95% CD8+. TIL were used immediately after isolation for experiments, except in some experiments in which TIL were plated in complete RPMI 1640 medium (2 x 106 cells/ml) for 6–18 h before usage.
Antibodies
Abs and reagents used for confocal microscopy and flow cytometry were: actin (phalloidin-FITC or -tetramethylrhodamine isothiocyanate; Sigma-Aldrich), CD2 (clone RM2-5; Caltag Laboratories), CD3 (clone 145-2C11; BD Pharmingen), CD3-Cy5 (clone 500A2; Caltag Laboratories), CD3 (goat IgG (W-12); Santa Cruz Biotechnology), CD3 (rabbit IgG; D. Wiest, Fox Chase Cancer Institute, Philadelphia, PA), CD5 (clone 4H8E6; Southern Biotechnology Associates), CD8 (clone 53-6.7; BD Pharmingen), CD8 (clone CT-CD8b; Caltag Laboratories), CD8-APC (clone 5H10; Caltag Laboratories), CD43 (clone S7; BD Pharmingen), CD45 (Alexa546-conjugated clone I3/2.3; M. Dustin, New York University School of Medicine, New York, NY), Csk (rabbit IgG (C-20); Santa Cruz Biotechnology), cholera toxin B (CT-B) FITC (Sigma-Aldrich), phospho-ERK (rabbit IgG; Cell Signaling Technology), granzyme B (goat IgG (N-19); Santa Cruz Biotechnology), LAT (rabbit IgG; L. Samelson, National Institutes of Health, Bethesda, MD), phospho-LAT (rabbit IgG; Upstate Biotechnology), LFA-1 (clone I21/7; BD Pharmingen), phospho-PLC-1 Y783 (rabbit IgG; Upstate Biotechnology), p56lck (clone 28; BD Pharmingen), phospho-p56lck (clone Y505; Cell Signaling Technology), phospho-Y (clone 4G10; Upstate Biotechnology), total proline-rich tyrosine kinase-2 (Pyk-2; rabbit IgG; Upstate Biotechnology), phospho-Pyk-2 Y881 (rabbit IgG; BioSource International), Shp-1 (rabbit IgG; Upstate Biotechnology), Shp-2 (rabbit IgG (sc-280); Santa Cruz Biotechnology), TCR (clone H57-597; BD Pharmingen), TCR-APC (clone H57-597; Caltag Laboratories), -tubulin (clone DM1A; Sigma-Aldrich), Wiskott-Aldrich syndrome protein (WASp; clone 7066; Upstate Biotechnology), phospho-ZAP70 Y319 (rabbit IgG; A. Chan, Genentech, South San Francisco, CA), and Y493 (rabbit IgG; Cell Signaling Technology). Secondary reagents for confocal microscopy were: goat anti-hamster Alexa488 (Molecular Probes), goat anti-hamster Alexa568 (Molecular Probes), goat anti-hamster rhodamine (Jackson ImmunoResearch Laboratories), goat anti-rabbit Alexa488 (Molecular Probes), goat anti-rabbit FITC (Caltag Laboratories), goat anti-mouse Alexa568 (Molecular Probes), donkey anti-goat Alexa488 (Molecular Probes), and donkey anti-rat rhodamine (Jackson ImmunoResearch Laboratories).
Flow cytometry
Cells were resuspended at 1 x 107/ml in complete RPMI 1640 medium, and Abs were added at empirically determined optimal concentrations. After incubation (4°C for 30 min), cells were washed once with 1 ml of FACS wash (PBS, 2% FBS, 0.1% sodium azide), and, if primary Ab was unlabeled, resuspended in complete RPMI 1640 medium at 1 x 107/ml, and appropriate secondary Abs were added. Conditions for reaction with secondary reagents were identical as for primary Ab. Cells were washed with 1 ml of FACS wash before fixation with 2% paraformaldehyde before analysis on a FACScan flow cytometer (BD Biosciences).
Conjugate formation and confocal microscopy
CD8+ TIL were mixed with tumor cells at a 3:2 ratio, centrifuged at 16,000 x g for 10 s to promote conjugate formation, resuspended in 0.05 ml of RPMI, transferred to poly-L-lysine-coated coverslips, and incubated for different intervals of time at 37°C before fixation in 4% paraformaldehyde (15 min at room temperature (RT)). (For analysis of intracellular molecules, cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min at RT (excepting phospho-ERK, in which case cells were permeabilized with 90% methanol). There was no staining of intracellular proteins (e.g., F-actin, granzyme B, etc.) when detergent was omitted, showing that paraformaldehyde fixation of cells does not introduce artifactual permeabilization (data not shown). For analysis of cell surface molecules, coverslips were not treated with detergent. After fixation, coverslips were washed three times with PBS, blocked in complete medium for 15 min before Abs were added, incubated for 45–60 min at RT, and washed in PBS. Coverslips were mounted with medium containing antifading agents (Biomedia) before analysis by confocal microscopy using a Zeiss confocal LSM510 microscope to obtain full-thickness Z-stacks of conjugates (15–20 per sample). Images were analyzed using Zeiss LSM510 software and are depicted as three-dimensional (3D) reconstructions of all Z-stacks, except where specifically noted. For quantitation of protein localization by confocal microscopy, conjugates were visualized and Ab staining that localized in a region of the TIL quartered into regions parallel to the plane of target contact within the most proximal 25% area closest to the conjugate CS was scored as positive. Representative examples of the localization phenotype for each molecule are shown.
All confocal and flow cytometric analyses were conducted repetitively using individual TIL preparations, and >50 conjugates were analyzed. The percentage of conjugates showing a specific phenotype is shown in the legend to corresponding micrograph for each confocal analysis.
Calcium flux
TIL were labeled with PE-conjugated anti-CD8 (4°C for 20 min, followed by washing with HBBS + 0.010 M HEPES). Cells were then resuspended in 1 ml of HBBS + 0.010 M HEPES containing 0.1 M Fluo-4 (Molecular Probes), incubated at RT for 30 min, washed twice with HBBS + 0.010 M HEPES, then incubated at RT for 30 min to allow de-esterification of the dye. MCA38 cells were labeled for 30 min with 0.002 mg/ml 4',6'-diamidino-2-phenylindole (DAPI; Molecular Probes), washed three times with cold RPMI, mixed together in 0.4 ml of RPMI + 0.020 M HEPES containing 0.002 M Probenecid (Sigma-Aldrich), and pulse centrifuged to induce conjugate formation. Cells were resuspended (RPMI, 0.020 M HEPES, 0.002 M Probenecid) and incubated at 37°C, and aliquots were removed for analysis using a FACS LSR. Concentrations of both Fluo-4 and DAPI used in these experiments were empirically determined to ensure that no leakage of either dye occurred during the course of the experiment (thereby potentially obfuscating analysis of authentic TIL:tumor conjugates). For ionomycin treatment, calculation of fold increase was determined by dividing the mean fluorescence intensity (MFI) of TIL alone treated with ionomycin by untreated TIL alone. Calcium flux analyses were conducted four times using individual TIL preparations. Fold induction of calcium flux in TIL was then calculated using this formula: ((MFI of TIL in conjugates) – (MFI of TIL not in conjugates))/(MFI of TIL not in conjugates). To control for possible fluctuations in baseline MFI associated with variability in dye loading, calculations of fold induction were internally controlled in each experiment using the MFI of TIL not in conjugates from each time point analyzed. The low magnitude of calcium flux signal is typical for murine T cells, which are considerably smaller compared with human cell lines often used in assay of calcium flux (29).
Intracellular phosphorylation flow cytometry assay
TIL were labeled with anti-CD8-TriColor (Caltag Laboratories; 4°C for 20 min), while MCA38 cells were labeled with the Red Cell Linker kit (Sigma-Aldrich). Cells were mixed at a 3:2-TIL:MCA38 ratio in RPMI at 4°C and induced to form conjugates for various times. Samples were fixed at the indicated time points with 3.5% paraformaldehyde for 15 min, washed with FACS wash, and permeabilized (PBS containing 0.1% Triton X-100 for 5 min at 4°C). After washing with complete medium, appropriate primary Abs were added, followed by appropriate FITC secondary reagents. Samples were analyzed for the level of fluorescence (FL)1 staining by gating on conjugates (FL2, FL3 double-staining population).
Fluorescence resonance energy transfer (FRET) analysis
FRET flow cytometry analysis was performed, as described (30), using PE-conjugated anti-TCR (H57-597) (donor) and APC-conjugated anti-CD8 (acceptor) or FITC-conjugated CT-B (donor) and appropriate fluorochrome-conjugated Abs to either CD3, TCR, or CD8 (acceptor), as indicated in the figure legend. Tumor cells were labeled with either PKH26 Red or PKH57 Green Fluorescent Cell Linker Kit (Sigma-Aldrich), and conjugates were formed for 10 min, as described above. Using single laser excitation at 488 nm, induced energy transfer was monitored by examining acceptor fluorescence due to energy transfer between PE/FITC and APC/Cy5 fluorophores. FRET was measured as enhancement of fluorescence intensity by donor:acceptor pair samples over donor alone samples. There was no FRET signal when any acceptor fluorochrome reagent was excited with the wavelength of the donor fluorochrome (data not shown) and the FRET signal when only acceptor reagent was bound was set to background. There is no down-regulation of TCR during the 10-min conjugation. FRET analyses were conducted more than five times using individual TIL preparations.
Chromium release assay
Cytolysis activity of TIL was assessed in standard 51Cr release assays performed in quadruplicate wells for each E:T ratio, as described (28).
Results
Freshly isolated TIL are: nonlytic, defective in F-actin localization, and unable to recruit molecules important for lytic granule release to the TIL:tumor cell interface
Like human TIL (24, 26), murine TIL lack ex vivo lytic function (28), but recover perforin-mediated, Ag-specific cytolysis following short-term culture in vitro (Fig. 1A). However, nonlytic TIL contain granzyme B- and perforin-positive granules consistent with previous data showing that de novo protein synthesis is not required for recovery of lytic function (28). Defective cytolysis in chromium release assay was substantiated by the observation that nonlytic TIL fail to mobilize both lytic granules (Fig. 1B) and the MTOC (Fig. 1C) to the CS. As was shown for lytic function, this phenotype is reversed upon short-term in vitro culture of TIL. Nonlytic and lytic TIL have equivalent levels of granzyme B and tubulin protein, as determined by flow cytometric and immunoblot analysis, respectively (data not shown).
FIGURE 1. Freshly isolated TIL are nonlytic, fail to localize lytic granules or the MTOC, and have defective cytoskeletal polarization. A, Freshly isolated TIL are unable to lyse cognate cell targets. Specific lysis of cognate target cells varied depending upon the experiment, but ranged from 5- to 7-fold higher than matched nonlytic T cells at a 20:1 E:T cell ratio. Error bars show the SD of quadruplicate wells. B, Lytic granules fail to mobilize to the target CS. The differential interference contrast images overlaid on the single channel are shown adjacent to the corresponding single channel confocal images as 3D reconstructions, wherein arrows indicate the TIL:target cell CS. C, MTOC and F-actin localization is deficient in conjugates formed between nonlytic TIL and cognate target cells. TIL were analyzed by confocal microscopy for localization of MTOC and F-actin in conjugates using anti--tubulin and tetramethylrhodamine isothiocyanate-conjugated phalloidin, respectively. Arrows indicate the MTOC (top panels) or the TIL:target cell CS (F-actin). The percentage of TIL with polymerized actin localized at the CS is: 8% (nonlytic) and 41% (lytic). Levels of F-actin in purified nonlytic and lytic TIL are shown in histogram analyses. This experiment shows that localization of the MTOC and F-actin in nonlytic TIL is deficient in comparison with lytic TIL. D, WASp and Pyk-2 fail to localize to the target CS. TIL were analyzed for localization of WASp or Pyk-2 by confocal microscopy of TIL conjugates. The arrows indicate the TIL:target cell CS. The percentage of TIL with WASp localized at the CS is: 8% (nonlytic) and 50% (lytic). The percentage of TIL with Pyk-2 localized at the CS is: 6% (nonlytic) and 49% (lytic). Failure to recruit WASp or Pyk-2 prevents the activation of these two intracellular molecules important in lytic function.
Because actin polymerization and localization at the target cell CS are hallmarks of T cell activation, we asked whether TIL could polymerize and redistribute F-actin upon conjugate formation (Fig. 1C). Although lytic and nonlytic TIL contain equivalent levels of polymerized actin, only lytic TIL localize F-actin to the CS; nonlytic TIL fail to accumulate F-actin at the CS. Moreover, densiometric pixel analysis of phalloidin labeling at the CS indicates that the few nonlytic TIL that do localize F-actin show significantly less pixel density than lytic TIL (data not shown).
Because TCR signaling is required for affinity maturation of LFA-1, which in turn is required for F-actin localization (31, 32), we assessed the activation of Pyk-2 and WASp, important mediators of integrin and actin signaling pathways, respectively (33, 34). In nonlytic TIL, neither protein is recruited and concentrated at the CS, showing that Pyk-2 and WASp cannot be activated (Fig. 1D). In lytic TIL, Pyk-2 and WASp are recruited to the CS, allowing them to become activated, again emphasizing the reversible nature of the lytic defect.
CD2, the CD3 complex, and CD8 do not colocalize with the TCR in TIL conjugates
The failure to localize F-actin to the CS implies a defect in TIL-proximal TCR-mediated signal transduction, which most likely prevents mobilization of the MTOC and exocytosis of lytic granules (35). Because actin polymerization and localization are also thought to be important in formation of a mature IS (36), using confocal microscopy of nonpermeabilized cells, we compared nonlytic and lytic TIL for distribution of molecules excluded from the SMAC (CD43 and CD45), localized to the p-SMAC (LFA-1), and localized to the c-SMAC (TCR, CD2, the CD3 complex, CD8, and lipid rafts). TIL were analyzed after 15 min of conjugation consistent with the kinetics of formation of mature IS in CD8+ T cells (19). (Confocal analyses were performed after fixation, obviating concern that Ab reagents may perturb the distribution of target molecules and, except where noted, are shown as 3D reconstructions of full-thickness Z-stack images, which reveal the distribution of label over the entire cell.) The TCR clusters at the CS to an equal extent in nonlytic and lytic TIL (Fig. 2A), and lytic and nonlytic TIL express equivalent levels of TCR, as determined by flow cytometry (data not shown).
FIGURE 2. CD2, CD3, CD3, and CD8 do not colocalize with TCR in TIL conjugates. TIL were analyzed by confocal microscopy, as described in Materials and Methods. Single Z-stack images of labeling taken at the top surface, the middle (at the CS), and the bottom surface of the TIL are included for experiments that display a puncta phenotype to emphasize the membrane distribution of labeling. Arrows indicate the TIL:target cell CS. The percentage of TIL with TCR localized at the CS is: 42% (nonlytic) and 46% (lytic). The percentage of TIL with CD43 and CD45 excluded from the CS is: 45% (nonlytic) and 49% (lytic), and 22% (nonlytic) and 30% (lytic), respectively. LFA-1 is localized at the CS in: 45% (nonlytic) and 53% (lytic) of conjugates, and lipid rafts localized at the CS in: 29% (nonlytic) and 41% (lytic). The percentage of TIL with CD2 in puncta is 76% (nonlytic) and 10% (lytic). Five percent of nonlytic TIL show localization of CD2 at the CS, and 51% of lytic TIL localize CD2 at the CS. The percentage of TIL with CD3 in puncta is: 71% (nonlytic) and 2% (lytic). Nine percent of nonlytic TIL show localization of CD3 at the CS, and 42% of lytic TIL localize CD3 at the CS. The percentage of TIL with CD3 in puncta is: 67% (nonlytic) and 10% (lytic). Fourteen percent of nonlytic TIL show localization of CD3 at the CS, and 40% of lytic TIL localize CD3 at the CS. The percentage of TIL with CD8 in puncta is: 79% (nonlytic) and 2% (lytic). Six percent of nonlytic TIL show localization of CD8 at the CS, and 49% of lytic TIL localize CD8 at the CS. Although TCR and lipid rafts localize to the CS (A and C) and CD43 and CD45 are excluded from the immunological synapse, LFA-1 localizes to the p-SMAC (B), and CD2, CD3, CD3, and CD8 do not localize to the CS in nonlytic TIL (D).
In both lytic and nonlytic TIL, conjugates CD43 and CD45 are excluded from the CS, and LFA-1 is properly localized to the p-SMAC (Fig. 2B). (Images of LFA-1 staining in the single channel were rotated to show en face labeling at the contact site.) Considered together with the observation that the TCR is localized to the CS (along with CD3 and p56lck; Fig. 5), these findings show that IS formation in nonlytic TIL is not grossly defective. However, CD2, CD3, CD3, and CD8 (and CD8; data not shown) are distributed in puncta over the cell surface of nonlytic TIL (Fig. 5D), whereas, in lytic TIL, these proteins localize exclusively to the CS. Appearance of proteins in puncta reflects distribution on the cell surface because single Z-stack images show label distributed at the plasma membrane, not in internal structures. For emphasis, we include images of labeling taken at the top surface, the middle of the cell at the CS (indicated by arrows), and the bottom surface. Therefore, in nonlytic TIL, the structure of the TCR signaling complex is aberrant as CD8, CD2, and the CD3 complex do not associate with the TCR at the tumor cell interface.
FIGURE 5. Localization and activation of tyrosine-phosphorylated proteins and IS components in nonlytic and lytic TIL. TIL:tumor conjugates were assessed for protein localization by confocal microscopy and activation-dependent phosphorylation events after conjugation using the intracellular flow cytometric-based phosphorylation assay, as described in Materials and Methods. p-Tyr localization (A) and conjugate-dependent activation levels (B, p-Tyr immunoblot of nonlytic TIL lysates before or after 5-min activation with anti-CD3 is shown in insert); LAT localization (C); levels of LAT p-Y191 (D); ZAP70 localization (E); activation of ZAP70 p-Y493 (F); localization of CD3 (G); and localization of p56lck (H). The percentage of TIL with p-Tyr localized at the CS is: 7% (nonlytic) and 59% (lytic). Seventy-two percent of nonlytic TIL and 0% of the lytic TIL show p-Tyr in puncta. The percentage of TIL with LAT localized at the CS is: 41% (nonlytic) and 23% (lytic). (Localization of LAT appears lower in lytic TIL because the anti-LAT Ab preferentially recognizes the nonphosphorylated form (J. Houtman, unpublished observation). Thus, because LAT is phosphorylated in lytic TIL, the Ab does not label as efficiently.) The percentage of TIL with localization at the CS is: ZAP70, 53% (nonlytic) and 60% (lytic); CD3, 40% (nonlytic) and 50% (lytic); and p56lck, 39% (nonlytic) and 47% (lytic). Arrows in single channel micrographs indicate the CS. Protein tyrosine phosphorylation in nonlytic TIL is deficient, reflected by the failure to increase levels upon conjugation and to localize p-Tyr to the immunological synapse. In nonlytic TIL, conjugates ZAP70, p56lck, and CD3 localize to the CS, but triggering is incomplete as ZAP70 is only partially activated and LAT is not phosphorylated at Y191.
Because CD8, the CD3 complex, and CD2 are associated with lipid rafts upon activation (12), we considered that the failure of those proteins in nonlytic TIL to localize at the CS with tumor targets may reflect the inability of nonlytic TIL to cluster rafts. However, upon analysis of raft localization (Fig. 2C), we found the majority of raft lipid coalesces at the CS in conjugates formed from both nonlytic and lytic TIL showing normal raft formation. Therefore, lack of mature IS formation in nonlytic TIL is not due to the inability to cluster rafts at the CS.
Conjugate formation dissociates CD2, CD3, and CD8 from the TCR in nonlytic TIL
In conjugates of nonlytic TIL, CD2, CD8, and the CD3 complex do not colocalize with the TCR; however, confocal microscopy does not permit analysis of these findings at a biochemical level. To definitively make this determination, we analyzed the interactions of selected cell surface proteins with each other and with lipid rafts using FRET flow cytometric analysis of TIL:tumor conjugates or TIL not in conjugates. (Numerical data shown in histogram panels are FRET units defined as: sample MFI minus the baseline MFI divided by the baseline MFI, where baseline is the MFI detected in the acceptor fluorochrome channel of samples labeled only with the donor fluorochrome-conjugated Ab.) In nonlytic and lytic TIL, the TCR and CD8 are closely associated with each other prior to formation of conjugates (Fig. 3A). However, after conjugation for 15 min, CD8 dissociates from the TCR in nonlytic TIL (Fig. 3A). Moreover, FRET analysis performed at earlier timepoints of conjugation indicates that in nonlytic TIL, TCR and CD8 are initially associated (at 1 min of conjugation), but dissociate by 5 min (Fig. 3C). This is in contrast to lytic TIL in which CD8 and TCR remain associated with each other after 15 min of conjugation (Fig. 3A). (Analysis of CD2 and TCR in lytic TIL does not show a FRET signal in keeping with the notion that in the mature IS CD2 localizes to a point interior to the LFA-1 ring, but clearly excluded from the c-SMAC that contains the TCR (37).)
FIGURE 3. FRET analysis. TIL conjugates were gated in scatter analysis, and FRET was recorded at 15 min of conjugate formation, as described in Materials and Methods (right panels, A and B). In control experiments, labeled TIL were analyzed in the absence of target cells (left panels, A and B). Control stains of donor alone are shown in the filled histogram, and the FRET signal is shown in the black tracings. (Numerical data shown are FRET units defined as: (sample MFI minus the baseline MFI) divided by the baseline MFI, where baseline is the MFI detected in the acceptor fluorochrome channel of samples labeled only with the donor fluorochrome-conjugated Ab.) Donor fluorochrome is the first molecule in the pair indicated in the x-axis label. A, CD8 and TCR associate in nonlytic TIL alone, but do not remain associated after conjugate formation. However, lytic TIL show CD8 and TCR association before and after conjugate formation. B, FRET analyses of lipid rafts (donor) associated with either CD3, TCR, or CD8 (acceptors), as indicated in the figure. Confocal microscopic analysis of nonlytic TIL in conjugates simultaneously stained for CT-B (green) and CD3 or CD8 (red) is shown as inserts in histograms. Arrows indicate the lipid raft at the CS. CD8 is found associated with raft lipid in both nonlytic and lytic TIL before conjugate formation, but only in lytic TIL do these two molecules remain associated after conjugation. C, FRET analysis of nonlytic TIL at 1, 5, and 15 min of conjugation.
We also analyzed interaction between lipid rafts and either CD3, CD8, or TCR after conjugate formation to further confirm association of these molecules with rafts in lytic TIL, and their exclusion in nonlytic TIL (Fig. 3B). Before conjugate formation, CD8 and CD3 associate with raft lipid, while TCR is excluded (Fig. 3B). However, upon conjugate formation, CD8 in nonlytic TIL dissociates from lipid rafts, while lytic TIL maintain CD8:raft association at 15 min. CD3 association with lipid rafts showed a FRET signal in conjugates of nonlytic TIL, albeit reduced in comparison with lytic TIL (Fig. 3B). This observation supports the confocal microscopy analysis (Fig. 3B), which showed appearance of raft lipids on the periphery of CD3+ puncta in nonlytic TIL conjugates, suggesting that some raft lipid is associated with CD3 even though the majority of lipid is found in a large aggregate that excludes CD3 (green label). In keeping with current notions of TCR inclusion in rafts (38), raft association of TCR is activation dependent because TIL alone (not in conjugates) showed no FRET signal. TCR rapidly associates with lipid rafts in nonlytic TIL conjugates (Fig. 3C), supporting the confocal microscopy, but the signal was low compared with lytic TIL, perhaps reflecting the relatively modest state of activation of nonlytic TIL.
When observed over a time course of conjugate formation (Fig. 3C), in nonlytic TIL lipid rafts associate with TCR, CD3, and CD8 at 1 min, but by 5 min of conjugation, CD8 dissociate from rafts with similar kinetics of TCR/CD8 dissociation.
Nonlytic TIL do not flux calcium
Because TCR signal transduction is required for lytic function, we compared lytic and nonlytic TIL for biochemical evidence of TCR-mediated signaling by analysis of calcium flux after different times of conjugate formation. For these experiments, TIL were labeled with PE-conjugated anti-CD8 and loaded with Fluo-4, and conjugates formed with DAPI-labeled MCA38 target cells for different times before fixation. Concentrations of dyes were used that permitted analysis of scatter patterns for gating of TIL in conjugates or nonconjugated TIL (Fig. 4). Nonlytic TIL display a 3- to 4-fold increase in calcium flux in response to ionomycin (green open histogram), as we noted previously using Indo-2 as calcium indicator (28), but do show an increase in calcium in response to conjugate formation (gate R2, purple filled histograms) remaining at the same level as unconjugated TIL (gate R3, red open histograms). However, elevation of calcium is observed in lytic TIL conjugates at the earliest time analyzed, 30 s. Lytic TIL maintain baseline calcium levels unless conjugates are formed (gate R3, red open histograms), showing that calcium flux is dependent upon Ag stimulation. Lytic TIL conjugates flux calcium to the same extent as that induced by ionophore, correlating recovery of lytic function with restoration of TCR-mediated signal transduction.
FIGURE 4. Nonlytic TIL do not flux calcium in T cell:target cell conjugates. A, By gating on appropriate populations of cells (shown in scatter plots), analyses were made on both TIL not in conjugates (gate R3) and on TIL conjugates (as described in Materials and Methods). Aliquots of TIL were separately activated with 0.01 M ionomycin for maximal levels of calcium flux (green trace) and are overlaid with TIL in conjugates (purple filled) and TIL not in conjugates (red trace). The average baseline MFI of calcium levels of TIL not in conjugates was 18 (nonlytic) and 15 (lytic). Conjugates were formed for 2.5 min. B, The table compares fold induction of calcium flux between nonlytic () and lytic () TIL at different times in multiple experiments (n = 4). This demonstrates that while lytic TIL flux calcium in conjugates, conjugate formation does not induce calcium flux in nonlytic TIL, although these cells are inducible by ionomycin treatment. C, Activation of PLC-1 was assessed after 2.5 min of conjugation using activation-specific Ab and showed that in contrast to lytic TIL, nonlytic TIL display no activated PLC-1 upon conjugate formation (arrow points to CS with target).
These data associate the lytic deficit of freshly isolated TIL with a block in TCR-mediated signaling. This observation furthermore suggests that the defect in signal transduction is both profound and proximal, affecting multiple downstream activation pathways and consequent lytic function. To further define our analysis of the proximal TCR signaling defect, we performed confocal microscopy of activated PLC-1 using phospho-specific Ab (Y783) (Fig. 4C). Low levels of active PLC-1 are detectable in lytic TIL at 1 min of conjugation (data not shown) and significant labeling localized to the CS is detected by 2.5 min. Even though both nonlytic and lytic TIL contain equivalent levels of nonactivated PLC-1 (data not shown), nonlytic TIL do not express detectable activated PLC-1 at any time after conjugate formation. The absence of phosphorylated PLC-1 in nonlytic TIL and the presence in lytic TIL (properly localized at the CS) corroborate the calcium flux data. These results suggest that the signaling defect in nonlytic TIL is located at a more proximal step of the activation pathway.
Defective activation and localization of tyrosine-phosphorylated proteins in nonlytic TIL
Because nonlytic TIL fail to increase calcium levels in response to conjugate formation, we asked whether recognition of cognate Ag induced tyrosine kinase activation as evidenced by an increase in protein tyrosine phosphorylation and CS localization of phosphotyrosinylated proteins, hallmarks of TCR activation. Nonlytic TIL in conjugates labeled brightly for phosphotyrosine (p-Tyr) (Fig. 5A); however, the subcellular distribution of label was dramatically distinct from lytic TIL in that p-Tyr in nonlytic TIL appeared in multiple small puncta, whereas p-Tyr in lytic TIL was localized primarily to the target cell CS. Flow cytometric analysis of conjugates showed lytic TIL increase p-Tyr levels with kinetics typical of activated T cells: increased levels appeared 2.5 min after initial conjugate formation and begin to decrease after 10 min. However, nonlytic TIL do not increase p-Tyr levels upon conjugate formation (Fig. 5B), although p-Tyr increases after stimulation of TIL with anti-CD3 (insert), showing that tyrosine kinase activity can be induced. Considered together with the deficit in calcium flux in nonlytic TIL, this observation suggests that upon conjugation with cognate target cells, activation of proximal and distal tyrosine kinases is defective in nonlytic TIL, and tyrosine-phosphorylated proteins are not reorganized.
p56lck, CD3, ZAP70, and LAT localize to the target cell CS in nonlytic TIL, but ZAP70 activation is incomplete
Site-specific phosphorylation and subcellular localization of proteins involved in signal transduction reflect the activation status of T cells; therefore, we determined the localization and activation status of additional essential components of the signaling complex: LAT, ZAP70, CD3, and p56lck. Confocal analyses of TIL in conjugates show LAT localization at the CS in both nonlytic and lytic TIL (Fig. 5C). However, flow analysis using an activation-specific anti-LAT Ab (p-Y191) showed that LAT is not phosphorylated in nonlytic TIL, but lytic TIL display normal LAT phosphorylation kinetics (Fig. 5D). This finding illustrates that defective signaling in nonlytic TIL is not due to a potential failure to recruit LAT to the signaling complex, but rather that LAT is not efficiently phosphorylated. Because LAT is the immediate downstream target of ZAP70, the localization of ZAP70 was analyzed by confocal microscopy (Fig. 5E) and its activation status by flow cytometry (Fig. 5F). In lytic TIL, ZAP70 localizes to the CS and is increasingly activated as a function of time in conjugates (average maximum increase of 60% in MFI). However, in nonlytic TIL, although ZAP70 localizes to the CS (implying that CD3 is phosphorylated), phosphorylation of ZAP70 Y493 is only modest (average maximum increase of 30% of MFI). This level of activated ZAP70 in nonlytic TIL is unable to propagate the activation signal, as p-Tyr is not increased and cells do not increase calcium levels. Nonlytic and lytic TIL contain similar levels of p56lck and CD3 proteins (data not shown), and both CD3 (Fig. 5G) and p56lck (Fig. 5H) are localized to the CS in nonlytic and lytic TIL; thus, weak ZAP70 activation in nonlytic TIL is not due to either deficient levels nor exclusion of CD3 or p56lck from the signaling complex.
Conjugate formation rapidly inactivates p56lck in nonlytic TIL
The activation status of p56lck was examined by flow cytometric analysis of phosphorylation of the inhibitory motif (containing Y505). Both lytic and nonlytic TIL have equivalent levels of phospho-Y505 immediately upon conjugation; however, Y505 is rapidly phosphorylated in nonlytic TIL, 1 min after target cell conjugation (Fig. 6A). The increase in Y505 phosphorylation averaged from multiple experimental repetitions reveals that p56lck in nonlytic TIL is significantly inactivated upon conjugation. Conversely, this motif is not significantly phosphorylated in lytic TIL until a much later time of conjugation, 30 min.
FIGURE 6. Localization and activation of proximal signaling components in nonlytic and lytic TIL. Cellular levels of phospho-p56lck Y505 were assessed using flow cytometry, as described in Materials and Methods. Baseline levels of p-Y505 in TIL in conjugates (A, histogram) and conjugation-induced levels (A, chart) are shown. Confocal images of TIL:tumor conjugates after 2.5 min show protein localization of Csk (B); Shp-2 (C); Shp-1 (D); and phospho-ERK (E). Single Z-stack images of top, the middle, and bottom slice are included to show inner membrane pERK distribution. The percentage of TIL with localization at the CS is: Csk, 27% (nonlytic) and 8% (lytic); Shp-1, 36% (nonlytic) and 13% (lytic); Shp-2, 10% (nonlytic) and 43% (lytic); and phospho-ERK, 7% (nonlytic) and 41% (lytic). Sixty-eight percent of nonlytic TIL and 0% of lytic TIL show pERK in puncta.
The p56lck Y505 is phosphorylated by tyrosine kinase Csk (itself phosphorylated by cAMP-controlled protein kinase A (39)), which is recruited into proximity with p56lck by Src homology 2 interaction with the raft-associated adapter Cbp (or phosphoprotein associated with glycosphingolipid-enriched microdomains (PAG)) (40). Therefore, we examined Csk localization (Fig. 6B) and found that the percentage of nonlytic TIL that localize Csk to the CS is 3- to 4-fold greater than lytic TIL at early times of conjugation, consistent with the observation that p56lck Y505 is phosphorylated. Localization of Csk to Cbp/PAG in the plasma membrane is regulated by the reciprocal activities of p56lck (and/or Fyn) and the phosphatase Shp-2 (40, 41). Thus, enhanced localization of Csk to the CS in nonlytic TIL could result from either increased phosphorylation or decreased dephosphorylation of Cbp/PAG. We examined Shp-2 localization in TIL conjugates and found that Shp-2 is localized to the CS in lytic TIL, whereas in nonlytic TIL Shp-2 does not localize and is mainly cytoplasmic (Fig. 6C). This finding supports the notion that phosphorylation of p56lck Y505 in nonlytic TIL is regulated by recruitment of Csk to the TIL:tumor interface.
Considered collectively, the observations that in nonlytic TIL p56lck and CD3 localize to the CS, but p56lck Y505 is phosphorylated upon target cell contact demonstrate that proximal TCR signaling becomes arrested at p56lck. To better understand the basis of the signaling block, the potential role of additional negative regulators of signaling was investigated. CD43 and CD45 localize equivalently in lytic and nonlytic TIL, both being excluded from the CS (Fig. 2), and CD5 localizes to the CS equivalently in both lytic and nonlytic TIL (data not shown). These observations diminish the likelihood of involvement of CD5, CD43, and CD45 in regulation of TIL signaling. However, localization of Shp-1 to the CS is 3-fold greater in nonlytic TIL compared with lytic TIL (Fig. 6D), supporting a potential role of this phosphatase in regulation of signaling in nonlytic TIL, as has recently been proposed (42, 43). In contrast, ERK can act as a positive regulator of TCR signaling by phosphorylating p56lck (at Ser59), preventing Shp-1 binding to p56lck. We examined the localization of activated ERK and found that pERK localized to the CS in lytic TIL, but in nonlytic TIL is sequestered to the inside of the cell membrane in a punctate pattern similar to that seen on the exterior of the cell for CD2, CD3, and CD8. This observation implies that ERK cannot provide positive feedback regulation of Shp-1 function in nonlytic TIL (44).
Discussion
The current study was predicated on the considerations that, because lytic function is dependent upon TCR-mediated signaling, and TIL are unable to exocytose lytic granules, residence of antitumor T cells in the tumor microenvironment may induce defective signal transduction. Supporting this notion, nonlytic TIL are deficient in protein tyrosine phosphorylation; thus, proximal tyrosine kinases are not activated upon recognition of cognate target cells. In addition, upon conjugate formation with cognate target cells in vitro, purified TIL are unable to flux calcium. This conclusion is supported by the observation that multiple substrates are not phosphorylated and ZAP70 is only modestly activated, deficiencies that undoubtedly underlie lytic dysfunction. Despite defective proximal signaling in nonlytic TIL, conjugation with cognate tumor cells induces IS formation, but the synapse is aberrant in that CD2, CD3 heterodimers, and CD8 do not colocalize with the TCR; instead, they localize in cell surface puncta, a unique phenotype not described in the literature. Confocal microscopy and FRET analysis show that the TCR complex is initially intact in nonlytic TIL, but is induced to dissociate upon conjugation with cognate target cells. The biochemical basis underlying the physical dissociation of selected molecules from the TIL TCR complex is currently unknown and may be multifaceted in nature.
Because TCR-mediated signaling is required for affinity up-regulation of LFA-1 (45, 46), which is in turn required for capacitation of F-actin-mediated signaling events (31, 47, 48), nonlytic TIL deficiencies in adhesion-dependent phenomena (decreased conjugation frequency, the failure to activate Pyk-2, and defective MTOC mobilization) most likely reflect the proximal nature of the TCR signaling defect. In support of this notion, CD4+ T cells lacking p56lck function have been shown by Burkhardt and colleagues (35) to have adhesion defects similar to those of CD8+ TIL. Similar to our findings, integrin-dependent signaling was also shown to be required for accumulation of F-actin filaments at the CS (formed between CD4+ T cells and peptide-pulsed APC), which facilitates sustained T cell activation (49). Therefore, the failure to affinity mature LFA-1 due to tumor-induced down-regulation of p56lck activity most likely prevents adhesion-dependent signaling possibly required to maintain association of the TCR with the CD3 complex, as well as accounting for TIL adhesion defects.
The physical dissociation of certain components of the TCR complex calls into question several notions concerning the initial stages of TCR-mediated signal transduction in CD8+ T cells. As was demonstrated by confocal microscopy and FRET analyses, CD8 and TCR dissociate in nonlytic TIL conjugates; nevertheless, TCR clusters at the CS. This observation suggests that CD8 interaction with the TCR is not required for stabilization of the Ag receptor at the CS. Our data are compatible with the notion that intact TCR complexes interact with Ag:MHC class I and initiate signaling; however, rapidly after cell triggering, coincident with both TCR clustering at the tumor cell:TIL interface and inactivation of p56lck function, CD8, CD2, and the CD3 complex dissociate from the TCR.
A related issue concerns the observation that, despite the failure of CD8 to remain associated with the TCR in nonlytic TIL conjugates, p56lck localizes to the CS. According to current notions about the structure of the TCR complex, CD8 is constitutively associated with lipid rafts due to palmitoylation of the -chain (5), where it associates with p56lck, whose residency in rafts is in turn mediated by activation-induced S-acylation (50). Therefore, upon recruitment to the TCR complex, by virtue of CD8 association with TCR, p56lck can phosphorylate CD3 and initiate T cell activation. The finding that p56lck and the TCR localize at the CS in nonlytic TIL conjugates coincident with TCR association with the raft suggests that p56lck does not require interaction with CD8 for stabilization at the IS (51, 52). The FRET analysis shows that before conjugation, CD8 is associated with raft lipid in nonlytic TIL and also interacts with the TCR. Upon conjugation, the TCR rapidly becomes raft associated (1 min), a time when CD8 is still associated with raft lipid, but immediately thereafter, CD8 dissociates from both TCR and the raft. This transient state wherein CD8 interacts with both TCR and the raft may be sufficient to recruit p56lck to the TCR upon activation, after which CD8 is no longer required to maintain p56lck association with TCR.
Recently, Valitutti and colleagues (53) have described experiments wherein a human CD8+ T cell clone was conjugated with peptide-pulsed B cells under conditions of low peptide concentration. In this model, the MTOC and lytic granules mobilize to the CS within 5–10 min of conjugation typical of CD8+ T cells, but localization of CD2 is weak, which was interpreted to mean that IS formation is not required for granule exocytosis. Curiously, although cells were not analyzed for localization of additional IS components and CD2 was not totally excluded, under these conditions of stimulation, CTL were not activated, as phosphotyrosinylated proteins did not strongly accumulate at the CS. In another recent paper by Davis and colleagues (54), the lack of mature IS formation in a murine T cell clone (stimulated by low concentration of peptide pulsed onto B cell APC) did not limit killing of target cells in vitro. However, in that model, lysis assays were conducted for 16 h, suggesting a profound experimental difference with our model, and lysis is Fas ligand mediated because blocking Fas receptor inhibited killing. (As we have shown previously, TIL lysis of targets is perforin and granzyme mediated.) Fas ligand-mediated cytotoxicity is thought to be important in regulation of the immune response, but less so as an effector mechanism in host defense (1), and may be independent of IS stability. Because the data of Valitutti and Davis were achieved under conditions of limiting Ag and used hemopoietic-derived APC, the relevancy of those observations in consideration of TIL lytic function is unclear.
Our data show that the block in TIL TCR signaling most likely results from rapid, conjugation-dependent inactivation of p56lck, depicted diagrammatically in Fig. 7. In normal T cells (or in lytic TIL), p56lck exists in a balance between the inactive and active forms due to opposing effects of Csk and CD45 on the phosphorylation of an inhibitory motif containing Y505: upon TCR ligation (when CD45 is colocalized with the TCR complex), Csk dissociates from the membrane, resulting in rapid Y505 dephosphorylation and coincident autoactivation, enabling phosphorylation of CD3, CD3 subunits, and ZAP70 (55, 56). Thus, tonic inhibition of TCR-mediated signaling permits fine control of modulating immune response (57). Our analysis of the status of p56lck using pY505-specific Ab showed that this inhibitory motif is equivalently phosphorylated in both lytic and nonlytic TIL immediately upon conjugation with target cells. However, Y505 in nonlytic TIL is substantially phosphorylated within 1 min of conjugation compared with lytic TIL, an event that rapidly inactivates p56lck kinase function (58, 59). In contrast, Y505 in lytic TIL does not become phosphorylated until late time of conjugation. Because phosphorylation of the inhibitory motif results in potent p56lck inactivation, small increases in phospho-Y505 are likely to be physiologically relevant (60). Supporting the finding of conjugation-dependent p56lck inactivation, at early times of conjugation Csk is localized at the CS, whereas in lytic TIL Csk is predominantly cytosolic. The finding that Shp-2 localizes at the CS of lytic TIL, but in nonlytic TIL is cytoplasmic, supports the notion that in nonlytic TIL Csk is recruited into proximity with its substrate p56lck (41).
FIGURE 7. TCR signaling in TIL. a, Lytic TIL. Because lytic TIL are no longer under the influence of the tumor, TCR stimulation induces the signaling cascade, which results in full activation of ZAP-70. This initiates subsequent downstream events such as: calcium flux, activation of the Ras/MAPK pathway, and MTOC and lytic granule mobilization. Because Shp-2 is recruited to the tumor cell:TIL contact site rapidly upon recognition of tumor cells, it can dephosphorylate Cbp, thereby blocking recruitment of Csk and proline-glutamic acid-, serine- and threonine-enriched protein tyrosine phosphatase (PEP) into proximity to their substrate p56lck. b, Nonlytic TIL. Upon TCR stimulation, nonlytic TIL initiate triggering: activation of p56lck, phosphorylation of CD3, and recruitment of ZAP-70. However, due to an unknown contribution by the tumor, transduction of the signal cannot progress past recruitment of ZAP-70. Instead, p56lck feedback inhibition pathways are activated involving recruitment to the CS of the negative regulator Csk, which phosphorylates p56lck at the inhibitory motif (Y505). (PEP recruitment is suggested by the recruitment of its interacting partner Csk, but has yet to be determined.) Csk is recruited to the CS by its binding partner Cbp. In addition, Shp-1 is recruited to the CS, where it may interact with its substrates. This causes early termination of the TCR signaling cascade, preventing calcium flux and MTOC and lytic granule recruitment.
In comparison with lytic TIL, nonlytic TIL in conjugates show CS localization of the tyrosine phosphatase Shp-1. Shp-1 can dephosphorylate the activation motif of p56lck (Y394), but may also dephosphorylate CD3, ZAP70, and LAT (43, 61, 62), therein uncoupling T cell triggering from activation, the phenotype of nonlytic TIL. Because ZAP70 localizes to the CS, indicating phosphorylation of CD3 (producing at least the p21 form (63)), our data suggest that Shp-1 may dephosphorylate ZAP70 in nonlytic TIL, therein blocking signal propagation. Supporting this notion is the finding that ZAP70 is only partially phosphorylated. An unresolved question is what is the mechanism of Shp-1 recruitment to the CS? The observation that pERK localizes to the CS in lytic TIL, but is excluded in nonlytic TIL supports the supposition that ERK may regulate the susceptibility of p56lck as a target of Shp-1, but does not identify the interacting partner with which Shp-1 binds. Our findings differ from those of Germain and colleagues (43) in that ERK is phosphorylated in nonlytic TIL, but is localized to puncta at the cytoplasmic face of the plasma membrane, not at the CS; therefore, it cannot prevent Shp-1 binding at the CS. Although purely speculative, it is possible that pERK may preferentially bind to a component of the puncta, therein sequestering it from p56lck, which localizes with the TCR at the TIL:tumor target interface.
Human cancers frequently contain CD8+ TIL, which are typically nonlytic, but Ag specific (24, 26). CD8+ T cells with this phenotype also characterize several chronic diseases resulting from infection by: Trypanosoma cruzi (64), SIV (65), HIV (66), human hepatitis C virus (67), herpes simplex virus (68), Friend virus (69), and lymphocytic choriomeningitis virus (70). The phenotype of defective CD8+ T cells in lymphocytic choriomeningitis virus infection closely resembles murine TIL that we have previously described: T cells are activated memory/effector cells, but are nonlytic, even when assessed by redirected assay (28). CD8+ T cells in the above-mentioned models of infection have been characterized as lytic defective on the basis of in vitro lysis assays without analysis of the biochemical basis for the nonlytic phenotype. (One exception was the study by Appay et al. (66), which showed that HIV-specific T cells were immature in that perforin protein was not present, a phenotype distinct from that which we have described.) Therefore, it is unclear whether the mechanism for induction of lytic dysfunction we describe is also used in chronic viral infection.
Collectively, our data suggest that proximal TCR-mediated signaling in nonlytic TIL is rapidly blocked after conjugation with cognate tumor targets. The signaling defect possibly involves two inhibitory mechanisms, Csk-mediated inactivation of p56lck function and enhanced Shp-1 activity, both of which may conspire to prevent proper activation of proximal tyrosine kinase activity, affinity up-regulation of LFA-1, and calcium flux, ultimately preventing exocytosis of lytic granules. Importantly, the TIL lytic defect is an acquired property restricted to T cells within the tumor microenvironment because in vivo systemic T cell function is not affected by tumor growth (71), whereas TIL are characterized by the inability to mobilize the MTOC to the CS and exocytose lytic granules. Acquired transient TIL lytic dysfunction in TIL has also been described in several transgenic TCR mouse tumor models (72, 73). Consideration of the observation that human TIL are Ag specific, but nonlytic, together with our description of defective lytic function of murine TIL, supports the notion that tumor-induced inhibition of TIL lytic function is a common characteristic that may contribute to tumor growth in the presence of antitumor immune response. Tumor-induced lytic dysfunction also may restrict T cell-based immunotherapy of cancer.
Acknowledgments
We thank Gabriele Campi and Mike Dustin for anti-CD45 Ab; Andrew Chan for anti-ZAP70 Y319 Ab; Lawrence Samelson for anti-LAT and anti-CD3 Abs; David Wiest for anti-CD3 Ab; Dan Rifkin and Dan Littman, respectively, for use of the FACScan and LSR equipment; Ed Ziff for use of the Zeiss microscope and Brian Fernholz for generous assistance in its usage; Abraham Kupfer for technical advice; and Andrew Bush, Andreas Dieffenbach, M. Dustin, and Stanislav Vukmanovic for editorial suggestions and/or helpful comments.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 M.K. and D.S. are cofirst authors of this manuscript and were supported by National Institutes of Health Training Grants T32 GM-07238 and T32 CA-09161, respectively.
2 Address correspondence and reprint requests to Dr. Alan B. Frey, Department of Cell Biology, New York University School of Medicine, 550 First Avenue, New York, NY 10016. E-mail address: freya01@med.nyu.edu
3 Abbreviations used in this paper: LAT, linker for activation of T cells; CS, tumor-infiltrating lymphocyte:target contact site; Csk, COOH-terminal Src kinase; c-SMAC, central zone supramolecular activation complex; CT-B, cholera toxin B; 3D, three-dimensional; DAPI, 4',6'-diamidino-2-phenylindole; FL, fluorescence; FRET, fluorescence resonance energy transfer; IS, immunological synapse; MFI, mean fluorescence index; MTOC, microtubule organizing center; PAG, phosphoprotein associated with glycosphingolipid-enriched microdomains; PLC, phospholipase C; p-SMAC, peripheral zone supramolecular activation complex; p-Tyr, phosphotyrosine; Pyk-2, proline-rich tyrosine kinase-2; RT, room temperature; Shp, Src homology 2 domain-containing protein phosphatase 1; SMAC, supramolecular activation complex; TIL, tumor-infiltrating lymphocyte; WASp, Wiskott-Aldrich syndrome protein.
Received for publication August 18, 2004. Accepted for publication November 11, 2004.
References
Kagi, D., B. Ledermann, K. Burki, R. M. Zinkernagel, H. Hengartner. 1996. Molecular mechanisms of lymphocyte-mediated cytotoxicity and their role in immunological protection and pathogenesis in vivo. Annu. Rev. Immunol. 14:207
Feito, M. J., A. Jimenez-Perianez, G. Ojeda, A. Sanchez, P. Portoles, J. M. Rojo. 2002. The TCR/CD3 complex: molecular interactions in a changing structure. Arch. Immunol. Ther. Exp. 50:263.
Fernandez-Miguel, G., B. Alarcon, A. Iglesias, H. Bluethmann, M. Alvarez-Mon, E. Sanz, A. de la Hera. 1999. Multivalent structure of an T cell receptor. Proc. Natl. Acad. Sci. USA 96:1547.
Zamoyska, R.. 1998. CD4 and CD8: modulators of T-cell receptor recognition of antigen and of immune responses?. Curr. Opin. Immunol. 10:82.
Arcaro, A., C. Gregoire, T. R. Bakker, L. Baldi, M. Jordan, L. Goffin, N. Boucheron, F. Wurm, P. A. van der Merwe, B. Malissen, I. F. Luescher. 2001. CD8 endows CD8 with efficient coreceptor function by coupling T cell receptor/CD3 to raft-associated CD8/p56lck complexes. J. Exp. Med. 194:1485.
Bachmann, M. F., A. Gallimore, S. Linkert, V. Cerundolo, A. Lanzavecchia, M. Kopf, A. Viola. 1999. Developmental regulation of Lck targeting to the CD8 coreceptor controls signaling in naive and memory T cells. J. Exp. Med. 189:1521.
Luescher, I. F., E. Vivier, A. Layer, J. Mahiou, F. Godeau, B. Malissen, P. Romero. 1995. CD8 modulation of T-cell antigen receptor-ligand interactions on living cytotoxic T lymphocytes. Nature 373:353
Renard, V., P. Romero, E. Vivier, B. Malissen, I. F. Luescher. 1996. CD8 increases CD8 coreceptor function and participation in TCR-ligand binding. J. Exp. Med. 184:2439
Fuller, C. L., K. S. Ravichandran, V. L. Braciale. 1999. Phosphatidylinositol 3-kinase-dependent and -independent cytolytic effector functions. J. Immunol. 162:6337.
Nesic, D., K. G. Jhaver, S. Vukmanovic. 1997. The role of protein kinase C in CD8+ T lymphocyte effector responses. J. Immunol. 159:582.(Mythili Koneru1, David Sc)