We examined the ability of OT-I CTL to induce target cell death, produce cytokine, form conjugates, and form immunological synapses in response to TCR signals of varying avidity. OT-I splenocytes were stimulated in vitro with 10 nM OVA_257-264 for 5 days to produce activated CTLs that were used for all subsequent experiments. Cytotoxicity was determined by lactate dehydrogenase (LDH) release from targets that had been peptide pulsed with either OVA_257-264 or G4 peptide at the concentrations shown. Decreasing concentrations of OVA_257-264 resulted in decreasing amounts of target cell lysis, whereas unpulsed targets were not killed (Figure 1A). Although EL4 targets pulsed with OVA_257-264 were killed very efficiently, G4-loaded targets were killed poorly (Figure 1B). Cytokine production is strongly affected by the strength of TCR signal and requires long, stable CTL-target interactions and synapse formation (Valitutti et al., 1996). No IFN-γ was detected after stimulating OT-I in the presence of 1 μM G4 peptide for 4 hr (Figure 1C), consistent with previous findings (Jameson et al., 1993).

We chose to study conjugates formed at 20 min throughout the study, in order to keep a standard set of conditions for comparing the avidity. Fixed conjugates were examined by confocal microscopy after labeling with antibodies against CD8 and lymphocyte protein tyrosine kinase, Lck, to mark the cSMAC (Figure 1D). Few conjugates formed in the absence of peptide (Figure 1E). Although there were fewer conjugates with a decrease in avidity between 1 μM OVA_257-264 and 1 μM G4, this difference was not statistically significant when compared via analysis of variance (p = 0.14). Our data indicate that the number of conjugates formed at 20 min was unaffected at lower-avidity interactions, consistent with previous studies that show a correlation between avidity and cytotoxicity (al-Ramadi et al., 1995; Rosette et al., 2001; Yachi et al., 2006).

Next we examined cSMAC formation by investigating the clustering of Lck at the immunological synapse. Lck is a tyrosine kinase, which associates with CD8 and clusters at the cSMAC upon TCR activation (Monks et al., 1998). It provides a marker of TCR signaling and early synapse formation. In the absence of any peptide, clustering of Lck occurred in only 7.4% of conjugates, which formed very rarely (Figures 1E and 1F). However, under all conditions in which OT-I cells were stimulated with either G4 or OVA_257-264 pulsed targets, >40% of conjugates displayed clustering of Lck and formation of a cSMAC (Figure 1F).

These results reveal that although the killing and cytokine production varied dramatically with avidity, conjugate formation and cSMAC formation occurred equally well over the range of avidities used in these experiments. This provided a system to examine the affect of avidity on downstream cytotoxic effector mechanisms.

Because OVA_257-264 and G4 provide different avidity interactions, we examined T cell signaling at the cSMAC for each. CTLs were conjugated to OVA_257-264 or G4-pulsed EL4 targets before labeling with antibodies against CD8, Lck, and phospho-Src-family kinases. The pSrc-family antibody (pY416) detects phosphorylated tyrosine 416 in Src family kinases, including the activated form of Lck (Ilani et al., 2009). Of the Src-family protein kinases, only Lck, Fyn, and Yes are expressed in T cells (http://symatlas.gnf.org; Samelson et al., 1990) and only Lck and Fyn have been shown to be able to cluster at the cSMAC (Monks et al., 1998). Conjugates were examined by confocal microscopy in the xy plane and xz plane to view the conjugates from the side and en face, respectively (Figures 2A and 2B). OVA_257-264-pulsed conjugates showed colocalization of Lck and pY416 staining, whereas no pY416 staining was detected in G4 conjugates. A projection through the z plane of the synapse of OVA_257-264 conjugates revealed pY416 staining at the cSMAC, colocalizing with Lck staining (Figure 2A). In contrast, no pY416 staining was detected in the majority of G4 conjugates (Figure 2F). The diameter of the cSMAC varied between 1.5 and 3 μM at the widest point in conjugates formed under any of the conditions tested.

G4 rapidly dissociates from the OT-I TCR, compared to OVA_257-264 (Rosette et al., 2001), consistent with the idea that it occupies the TCR more transiently. To establish how great this difference in the off-rate is in conjugates formed after 20 min, we examined the dissociation of the K^bG4-PE and K^b OVA_257-264-PE tetramers from OT-I, in the presence of an antibody to H-2D^b and H-2K^b that prevents rebinding of dissociated tetramers. CTLs were detected by flow cytometry and the percentage of tetramer-bound CD8⁺ cells were plotted over time to show the rate of dissociation (Figure 2C). OT-I CTL rapidly lost G4-PE tetramer compared to OVA_257-264-PE tetramer. At 20 min, only half the number of OT-I bound G4 tetramer compared to OVA_257-264. This suggested that any signaling in response to G4 might be inhibited by phosphatases, such as SHP-1 (Stefanova et al., 2003) in G4 conjugates.

To investigate this possibility, we generated OT-I conjugates in the presence of sodium orthovanadate (Na₃VO₄), which prevents dephosphorylation of tyrosine kinases, by inhibiting phosphatases (Figures 2D and 2E). Under these conditions, pY416 staining was seen in the cSMAC in more than 75% of OVA_257-264 and G4 conjugates (Figure 2F), colocalizing with Lck in both cases. Orthovanadate treatment did not affect amounts of pY416 staining in OT-I cells in the absence of peptide (Figure S1 available online). These results reveal that activation occurs at the cSMAC in response to both OVA_257-264 and G4; however, activation in response to G4 is inhibited by phosphatase activity.

Previous studies have shown that Lck activation leads to recruitment of the phosphatase, SHP-1, which downregulates Lck activity. This negative-feedback loop is balanced by a positive-feedback loop controlled by phospho-ERK (pERK), which prevents SHP-1 recruitment to Lck (Stefanova et al., 2003). Furthermore, pERK has been shown to accumulate at the synapse between CTL and the antigen-presenting cell more strongly in response to OVA_257-264 than G4 via RMAS antigen-presenting cells (APCs) (Yachi et al., 2006). In addition, pERK signaling has been implicated in granule polarization and actin remodeling in both T cells and NK cells (Chen et al., 2006; Robertson et al., 2005). For these reasons, we investigated whether pERK accumulated at the immunological synapses of both OVA_257-264 and G4 conjugates with EL4 APCs and pERK activation with these stimuli. Confocal microscopy of conjugates revealed that pERK accumulated at the synapse of both OVA_257-264 and G4 conjugates (Figures 3A and 3B), and the pattern of staining was distinctive when viewed in the xz plane (Figures 3A and 3B). pERK staining formed a ring, which overlapped with actin, in the dSMAC. Occasionally, a small patch of pERK was observed within this ring. The specificity of antibody staining was confirmed by preincubation with the mitogen-activated protein kinase kinase (MEK) inhibitor, U0126, which abolished staining (Figure S2). Although pERK staining accumulated in the dSMAC in both OVA_257-264 and G4 conjugates, the intensity of staining appeared weaker in G4 compared to OVA_257-264 conjugates (Figures 3A and 3B). pERK activation was assessed both by quantitation of conjugates and by immunoblotting of cell lysates. Only 25% of conjugates showed pERK accumulation at the synapse after G4 stimulation compared to 65% with OVA_257-264 (Figure 3C). Furthermore, immunoblotting on cell lysates stimulated with OVA_257-264 showed strong ERK activation, almost comparable to stimulation with the phorbol ester, phorbol 12-myristate 13-acetate (PMA), and well above the basal amount observed in OT-I in the absence of peptide (Figure 3D). Stimulation with G4 showed weak ERK activation, which was nevertheless above the basal level observed in control cells in the absence of peptide. Addition of the MEK inhibitor U0126 largely ablated the pERK signals, although the total amounts of ERK in the samples were comparable. These results implied a close association between actin and pERK and demonstrated that even the low-avidity peptide, G4, gave rise to low ERK activation that accumulated in the dSMAC of the synapse. Moreover, these results showed that actin clearance from its initial accumulation across the synapse to form the ring of the dSMAC occurred in both OVA_257-264- and G4-stimulated synapses.

Our previous studies have shown that, upon TCR activation, actin clears to form the dSMAC as the centrosome moves forward to dock at the plasma membrane, suggesting that the two events may be linked (Stinchcombe et al., 2006). Our findings that actin clearance could occur in both OVA_257-264 and G4 conjugates suggested that centrosome polarization might also be occurring in both sets of conjugates. Antibody staining for both actin and γ-tubulin revealed that actin cleared into a characteristic ring with the centrosome (labeled with γ-tubulin) contacting the plasma membrane in the center of this ring in both OVA_257-264 and G4 conjugates (Figure 4). This is particularly clear in xz reconstruction over 1 μM across the synapse.

We have previously shown that the centrosome not only polarized to the synapse but also moved right up to the plasma membrane (Stinchcombe et al., 2006). To confirm whether or not avidity affected the tightness of polarization, we examined the synapses formed between OT-I CTLs and targets pulsed with either 1 μM OVA_257-264 or G4 at high resolution via electron microscopy (EM) (Figure 5). Low-power images of OVA_257-264-stimulated OT-I (Figure 5, i) showed the centrosome, seen in cross-section through one centriole (indicated by an arrowhead), polarized to the plasma membrane, distant from the nucleus. These thin sections are consistent with 3D images taken of conjugates with the nucleus stained with Hoechst (compare to Figure 6), and with previous studies in other types of CTL (Stinchcombe et al., 2006). High-power images of transverse sections through the barrel of one centriole (Figure 5, ii–iv) revealed that both the centrosome and Golgi stacks (G) were clearly visible polarizing right up to the plasma membrane. Consistent with previous studies, the tightness of polarization of both the centrosome and Golgi stacks relative to the membrane varied between conjugates, with some showing the centriole at the plasma membrane and others, showing the centriole slightly back from it (compare to Stinchcombe et al., 2006). These differences are likely to reflect images of cells captured at different stages of interaction (i.e., polarization, killing, retraction) with their targets at the time of fixation. Electron-dense lytic granules were seen in the OVA_257-264-pulsed OT-I cell close to the centrosome (Figure 5; i, ii, and iv) as well as near the nucleus (Figure 5, i). Similar instances of tight centriole and Golgi polarization were also found in OT-I-G4 conjugates, indicating that lower avidity did not affect the tightness of either centriole or Golgi polarization (Figure 5; vi–viii). Transverse sections through the barrel of one centriole revealed how closely associated the centrosome is with the plasma membrane and suggests that it contacts the plasma membrane directly, consistent with immunofluorescent images (Figure 5; Stinchcombe et al., 2006).

In order to quantitate centrosome polarization in response to avidity, we examined the position of the centrosome (γ-tubulin labeling) relative to the cSMAC (Lck labeling) in conjugates formed with different concentrations of OVA_257-264 or G4 (Figure 6). The centrosome is dynamic and usually associated with the nucleus (perinuclear). In migrating T cells, the centrosome is localized on the trailing edge of the nucleus (Burkhardt et al., 2008), but on synapse formation reorients toward the contact site (Kupfer et al., 1985). In a fully polarized CTL, the centrosome migrates right up to and docks at the plasma membrane next to the cSMAC (Stinchcombe et al., 2006). By using immunofluoresence microscopy, we examined OT-I CTLs conjugated to EL4 target cells loaded with either OVA_257-264 or G4. Conjugates were classified into four groups (as illustrated in Figure 6A) based on the position of the centrosome within the cell: (i) centrosome tightly polarized at the plasma membrane, next to the cSMAC, (ii) partially polarized between the nucleus and the cSMAC, (iii) perinuclear and proximal to the cSMAC, or (iv) perinuclear and distal from the cSMAC (Figure 6A). In conjugates formed in the absence of peptide, the centrosome was always perinuclear and never reached the cSMAC (Figure 6B). However, in conjugates formed in the presence of OVA_257-264 or G4, the centrosome was tightly polarized to the cSMAC in the majority of cases. The mean numbers varied, with 80% of OVA_257-264 conjugates polarizing the centrosome compared with 53% for G4. A Student's t test comparing the two groups gave a value of p = 0.01, suggesting that this difference is statistically significant. This suggests that decreasing avidity may reduce the proportion of conjugates in which the centrosome polarizes and may contribute to the reduction in killing seen in response to G4. However, in spite of the fact that the centrosome polarized in the majority of G4 conjugates, there is virtually no killing of target cells observed under these conditions (Figure 1). We therefore examined granule polarization in response to G4 versus OVA_257-264 stimulation.

Lytic granule polarization to the synapse was compared in conjugates formed with OVA_257-264- and G4-pulsed targets. Immunofluorescence microscopy was used to identify lytic granule localization relative to the cSMAC by using antibodies against Lck (cSMAC) and the lysosomal membrane protein, LAMP-1 (CD107a) (Figure 7), a transmembrane protein that localizes to lytic granules (Peters et al., 1991). Because LAMP-1 staining detected lysosomes in both targets and CTLs, CD8 staining was used to identify CTLs, and in addition the lytic granules of OT-I CTLs were easily distinguished by their larger size (Figure 7A). Although the majority (54%) of OVA_257-264 conjugates showed granules clustered tightly by the cSMAC, the majority of G4 conjugates (90%) showed the granules dispersed and distant from the cSMAC (n ≥ 1115 for each group) (Figure 7B). Decreasing concentrations of OVA_257-264 led to a decrease in the proportion of conjugates displaying fully polarized granules, with a mean of 29% of conjugates with fully polarized granules with 1 pM OVA_257-264 compared with 54% at 1 μM OVA_257-264. These results suggest that the strength of the TCR-target interaction controls the efficiency of granule polarization. The granule polarization results correlated with the marked decline in cytotoxic effector function shown in Figure 1 and suggested that the loss of cytotoxicity associated with G4 interactions was likely to reflect decreased CTL degranulation. In order to address this, we measured the ability of the OT-I to release lytic granules in response to OVA_257-264 and G4 (Figure 7C). CTLs were stimulated with either G4 or OVA_257-264 peptide and the surface mobilization of LAMP-1 from lytic granule membranes was detected by flow cytometry. In these assays, α-LAMP1-PE antibodies were present in the media throughout the duration of the assay, and any LAMP-1 that appears on the surface becomes labeled, even when subsequently endocytosed. This provides a measure of degranulation. LAMP-1 labeling increased only in OVA_257-264-stimulated CTLs, but not in G4-stimulated cells where LAMP-1 labeling remained at the background count of cells without peptide. Assaying LAMP-1 labeling at 30, 60, and 120 min showed an increase in the percentage of CTLs labeled with LAMP-1-PE, indicating that an increasing number of OVA_257-264-stimulated CTLs degranulated with time. However, G4-stimulated cells showed no increase compared to unstimulated controls at any of the three time points assayed. These results show that TCR avidity affects the proportion of cells that can mobilize granules to the synapse.

Female C57BL/6 (B6)-OT-I naive spleens were stimulated with 10 nM OVA_257-264 peptide for 3 days in RPMI medium 1640 supplemented with 10% FCS, 50 μM β-mercaptoethanol, 10 U/mL human recombinant IL-2, L-glutamine, sodium pyruvate, and 50 U/mL penicillin and streptomycin (GIBCO). After 2 days, cells were washed and seeded into fresh media daily. Target H-2^b EL4 cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% FCS and L-glutamine.

For the immunofluorescence studies, reagents were obtained from the following: mouse anti-actin (AC-40), rabbit anti-actin, and rabbit anti-γ-tubulin (Sigma-Aldrich); mouse anti-mouse Lck (3A5) (Millipore); rat anti-mouse CD8 (YTS192) (H. Waldmann, Oxford University); rat anti-mouse CD107a (LAMP-1, 1BD4) (Developmental Studies Hybridoma Bank, IA); mouse-anti-Phospho-ERK (pT202/pY204) (BD Biosciences); rabbit polyclonal anti-mouse-phospho-Src-family (pY416) (Cell Signaling); and Hoechst 33342 (Invitrogen). All secondary Alexa Fluor antibodies (excited at 405, 488, 546, and 633 nm) were from Invitrogen. For the degranulation assays, rat anti-mouse CD107a-PE (1D4B) and rat anti-mouse CD8α-PerCP-Cy5.5 (53-6.7) were from PharMingen and BD Biosciences. For tetramer dissociation experiments: OVA_257-264/K^b-SA-PE (streptavidin-phycoerythrin) and G4/K^b-SA-PE tetramers (Beckman Coulter); and mouse anti-mouse-H-2D^b/K^b (28-8-6) was obtained from PharMingen, BD Biosciences. OVA_257-264 [SIINFEKL] and G4 [SIIGFEKL] peptides were obtained from AnaSpec, Cambridge Biosciences.

EL4 cells were pulsed with various concentrations of either SIINFEKL (OVA_257-264) or SIIGFEKL (G4) peptide at 37°C for 1 hr before washing three times in RPMI. In vitro activated OT-I CTL cells (day 6 after stimulation, unless otherwise indicated) were also washed in RPMI before CTL and target cell pellets were resuspended to a final concentration of 4 × 10⁶ cells/ml. CTLs and targets were mixed 1:1 and incubated in suspension for 5 min, before diluting to 10⁶/ml and aliquoting onto glass multiwell slides, and incubated for a further 15 min at 37°C to adhere to the glass. In some experiments (Figure 2; Figure S1), cells were conjugated in presence of 1 mM sodium orthovanadate (Na₃VO₄). Samples were fixed with ice-cold methanol for 5 min and washed six times in PBS before incubating for 1 hr at room temperature in blocking buffer (PBS, 1% BSA [Sigma-Aldrich]). Samples were incubated with primary antibodies in PBS, 0.2% BSA for 1 hr at room temperature, or overnight at 4°C and washed extensively in PBS, 0.2% BSA before adding secondary antibodies for 40 min at room temperature. Nuclei were stained with Hoechst (1:10,000) in PBS for 5 min before mounting with number 1.5 coverglass and Mowiol. Samples were examined with a Zeiss Confocal LSM510 microscope, with lasers exciting at 405, 488, 543, and 633 nm, with the 100× objective. Final images are displayed as projections of 3D stacks unless otherwise indicated. xz reconstructions were processed with Volocity software (Perkin-Elmer).

Cytotoxicity was examined with the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega). Target EL4 cells were pulsed with peptide at 37°C for 1 hr, washed three times, and resuspended in phenol red-free RPMI, 2% FBS (killing assay media), at 10⁵ cells/mL in a round-bottom 96-well plate. Activated CTL were added at titrated effector:target (E:T) ratio and plates were incubated at 37°C for 4 hr. The absorbance of the supernatants at 490 nm determined the release of lactate dehydrogenase (LDH) and percent cytotoxicity. For degranulation assays, effector CTL and targets were mixed at a ratio of 1:1 and incubated in 96-well plates at 37°C, in the presence of CD107a-PE (LAMP-1). Cells were harvested into cold PBS at each time point and incubated on ice. At the end of the time course, cells were resuspended in PBS/0.2% BSA/0.02% NaN₃ (FACS buffer) and labeled with antibodies against CD8α-PerCPCy5.5 and anti-CD107a-FITC (LAMP-1).

Activated OT-I CTL were cultured for 5 hr at 37°C in 96-well round-bottom plates at approx. 0.5–2 × 10⁶ cells/well in complete RPMI medium containing 10% FCS, 10 U/ml recombinant human IL-2, and 5 μg/ml GolgiPlug (Becton Dickinson), with or without 1 μM of the OVA_257-264 (SIINFEKL) or G4 (SIIGFEKL) peptides. The cells were then washed with PBS (containing 0.1% BSA and 0.02% sodium azide), labeled with anti-mouse CD8α-PerCPCy5.5 for 30 min on ice, fixed and permeabilized with the BD Cytofix/Cytoperm kit (Becton Dickinson), and labeled for intracellular cytokine production with anti-mouse IFNγ-FITC (clone XMG1.2, PharMingen). Cells were washed and analyzed on a FACSCalibur with CellQuestPro software (Becton Dickinson). In each assay, any cytokine-positive cells isolated from wells with no peptide were subtracted from the percent cytokine-positive cells incubated with peptide to yield the final value.

CTL were stained with OVA_257-264/K^b-PE, or G4/K^b-PE tetramers for 1 hr at room temperature and then washed with FACS buffer (PBS, 0.2% BSA, 0.02% NaN₃). Cells were then incubated at 37°C in medium containing 50 μg/ml 28-8-6, anti-H-2D^b/K^b to prevent tetramer rebinding. Cell aliquots were removed at different times into cold FACS buffer, washed, and stained with anti-CD8α-CD8α-PerCpCy5.5 for 30 min on ice. After further washing, the cells were analyzed for residual tetramer staining with a FACSCalibur (BD Biosciences).

OT-I CTL were left untreated or stimulated with PMA (50 nM), OVA_257-264 (1 μM), or G4 (1 μM) peptide in the presence or absence of 100 μM sodium orthovanadate for 15 min. Cell pellets were lysed for 45 min at 4°C, at 2 × 10⁷ cells/ml in lysis buffer (PBS, 2% Triton X-100, 150 mM NaCl, 50 mM Tris-Cl [pH 8.0], 1 mM MgCl₂, complete protease inhibitor cocktail [Roche]). Cell debris and nuclei were removed by centrifugation at 13,000 rpm. 15 μl of each lysate were separated on a 10% acrylamide gel with 15 μl SDS loading buffer (Novex Tris-Glycine SDS Sample Buffer [2×], Invitrogen) after denaturation for 10 min at 95°C, before transfer onto nitrocellulose. Membranes were blocked with PBS, 0.2% Tween-20, 5% BSA, probed overnight with rabbit anti-ERK or pERK (Cell Signaling) at 4°C, washed with PBS-T, and incubated in secondary goat anti-mouse-horseradish peroxidase (HRP) for 30 min at room temperature, washed, and developed with ECL developing solution (Amersham).

OT-I CTL were incubated with 1 mg/ml HRP (Boehringer) to load the endocytic pathway, washed, and resuspended at ∼3 × 10⁷/ml in serum-free medium. OT-1 were mixed with equal numbers of EL4 pulsed with 1 μM OVA_257-264 or G4 at 37°C for 4 min, diluted to ∼7 × 10⁶/ml in 24-well plates, incubated for a further 20–30 min, fixed and processed for DAB cytochemistry (Stinchcombe et al., 2001), postfixed in 1% osmium in H₂O for 1 hr at RT, washed extensively in H₂O and incubated overnight in 0.5% uranyl acetate in H₂O, washed, and EPON embedded. Sections (50–150 nm) were post-stained with lead citrate and viewed with a Phillips CM100 Electron Microscope and images captured on Kodak photographic film. Negatives were scanned and recorded digitally.

Quantitative data are displayed as the mean of three or more independent experiments ± standard deviation. A one-way ANOVA was used to determine whether there were statistically significant differences within each experiment. If significance was found, each individual set of conditions (avidity group) was compared to the reference group (OVA_257-264 1 μM) via a two-tailed Student's t test. Significance was reached when p ≤ 0.05.