T Cell-Dendritic Cell Immunological Synapses Contain TCR-dependent CD28-CD80 Clusters That Recruit Protein Kinase Cθ¹

Primary helper T cell activation requires the presentation of MHC class II-antigenic peptide complexes (MHCp)⁴ and costimulatory ligands by dendritic cells (DC). These interactions induce T cell deceleration and formation of stable T cell-DC synapses (1, 2, 3, 4, 5). Most of what we know about the molecular details of the stable immunological synapse (IS) comes from the study of simplified model systems that replace the DC with a transformed non-DC cell type or supported planar bilayer (6, 7, 8, 9, 10, 11, 12).

Model ISs are characterized by formation of radially symmetric multimicrometer scale supramolecular activation clusters (SMACs). When MHCp density is high, the bulk of the TCR can be located in the central SMAC (cSMAC), which is also enriched in protein kinase Cθ (PKCθ) (7, 9). The amount of TCR accumulated in the cSMAC is proportional to the density of MHCp presented (13). The cSMAC is surrounded by a symmetrical peripheral SMAC (pSMAC) enriched in LFA-1 and talin (9) and a distal SMAC enriched in F-actin and CD45 (14, 15). Although this organization was observed with mature DC in one study, others have found that T cell-DC interfaces have multiple large TCR clusters rather than a single cSMAC (16, 17).

Before SMAC formation, TCR are engaged in the periphery of the nascent IS (7) in microclusters (10). After SMACs are established, the ongoing formation of TCR microclusters is required for sustained signaling when T cells are activated by planar bilayers (13, 18, 19), but this remains to be clearly demonstrated with cells presenting MHCp. Microclusters are dynamic and have a lifetime of ∼2 min, the time required for their disappearance after addition of anti-MHCp Abs that block new TCR-MHCp interactions (13). The cSMAC is relatively more stable than microclusters in that it persists for >20 min after anti-MHCp Abs (13). In summary, TCR are engaged in two structures, short-lived microclusters linked to sustain TCR signaling and a relatively long-lived cSMAC that focuses on PKCθ.

The key costimulatory receptor CD28 has also been studied primarily in model systems and not previously with DC. Presentation of CD80 along with MHCp on B cell tumors enhanced IS formation (20). CD28-CD80 interactions accumulate in the cSMAC in a TCR-dependent manner (21, 22). Before cSMAC formation, CD28 accumulates in TCR microclusters (23). This suggests that the CD28 focused in the cSMAC is recruited via the TCR microclusters and this may account for the TCR dependence. The recruitment of PKCθ to the cSMAC is dependent upon CD28-CD80 interactions and specific sequences in the cytoplasmic domain of CD28 (11). We have previously observed using a model system that the cytoplasmic domain of CD80 controls the multimicrometer scale segregation of TCR clusters from CD28-CD80 clusters (12). This model also revealed that when CD28-CD80 clusters are segregated from TCR clusters, the PKCθ is focused in the CD28-CD80 clusters (12). In summary, CD28-CD80 clusters can be segregated from TCR clusters in model systems, but the dynamics of these structures are not known in any system and costimulation has not been studied at all in critically important T cell-DC IS.

In this study, we examined the dynamics of CD80, TCR, and LFA-1 accumulation in the T cell-DC IS and tested the ongoing requirement for new TCR-MHCp interactions in sustaining CD80, TCR, and LFA-1 clusters. Using DC from CD80-enhanced cyan fluorescence protein (eCFP) bacterial artificial chromosome (BAC)-transgenic (Tg) mice and immunofluorescence microscopy, we were able to monitor CD80 clustering longitudinally with live conjugates without modification of CD28 on the T cells. This approach allowed us to analyze the dependence of CD80 clustering in the IS with respect to dynamic TCR interactions that were controlled temporally using anti-MHC Abs. In T cell-DC IS, we observed multiple clusters containing CD80, TCR, or LFA-1, but not in combination. CD80 clusters uniquely focused PKCθ. Inhibition of new TCR-MHCp interactions with anti-I-A^b, MHC-blocking Ab, rapidly reduced Ca²⁺ signaling to baseline and abrogated CD80 clustering, reduced TCR clustering, but did not alter LFA-1 clustering in 4–5 min. This suggests that CD80 clusters share properties of microclusters (short lifetime after TCR blocking) and the cSMAC (focusing of PKCθ in stable IS).

The vector containing CD80-eCFP used to generate BAC Tg mice were constructed by inserting eCFP (BD Clontech) to the end of the CD80 cytoplasmic tail of BAC clone RP23-69GS (National Center for Biotechnology Information). The linker sequence used and the BAC Tg construct generation method has been described elsewhere (12, 24). In brief, PCR-positive RP23-69GS clone was grown to log phase, pelleted, and washed three times with 10% glycerol while keeping everything at 4°C. H5.3 shuttle vector (1 μg/μl) containing 1 kb of CD80 sequences flanking both sides of eCFP was added to 50 μl of the resuspended cells and electroporated at 1800 Ω and 2.5 W. After shaking for 30 min in 1 ml of Super Optimal Broth medium, 900 μl of the cell pellet was added to plates containing 12.5 μg/ml chloramphenicol and 30 μg/ml ampicillin. Colonies positive for integration were identified by PCR and streaked on sucrose plates to grow for 3 days at 37°C. Positive colonies for resolution were first identified by PCR and by pulse-field gel electrophoresis and confirmed by Southern blot. Purified BAC DNA was injected into F2 B6 blastocytes by the staff of the University of Michigan transgenic core facility. Positive founders were screened by PCR with 5′-CACTATGCTATGAGA-3′ and 5′-GGACACGCTGAACTTGTG-3′. Expression of CD80-eCFP was confirmed by immunofluorescence microscopy.

The F2 C57BL/6 founders were crossed to CD80/CD86^−/− on the C57BL/6 background purchased from The Jackson Laboratory. OT-II C57BL/6 (H-2^b) CD4⁺ TCR Tg mice purchased from The Jackson Laboratory were crossed to C57BL/6 Rag2^−/− mice. B10.A 5C.C7 TCR (Vα11Vβ3) Tg mice were purchased from Taconic Farms. All mice were maintained under specific pathogen-free conditions at the Skirball Institute animal care facility at New York University in accordance with the Institutional Animal Care and Use Committee guidelines. MCC_88–103 (ANERADLIAYLKQTK) and OVA_323–339 (ISQAVHAAHAEINEAGR) were synthesized by the Dana-Farber Cancer Institute peptide synthesis facility (Boston, MA).

LPS/anti-CD40- treated and untreated CD11c⁺ splenocytes were lysed using an immunoprecipitation kit with protein A (Roche Applied Science) according to the manufacture’s protocol. The target samples were immunoprecipitated overnight at 4°C with rabbit polyclonal anti-GFP Ab (Sigma-Aldrich). The immunoprecipitated protein samples were reduced with 2-ME and separated by 10% SDS-PAGE (Bio-Rad) with Kaleidioscope Prestained Standards (Bio-Rad), than transferred to polyvinylidene difluoride membranes (Bio-Rad) at 4°C for 4 h. The sample was immunoblotted with rabbit anti-GFP Ab followed by HRP-conjugated anti-rabbit IgG from a Rabbit IgG True Blot Kit (eBioscience). The sample was developed using ELC (Roche Applied Science).

To isolate DC, spleens were placed in a petri dish with 1 mg/ml collagenase D and 100 μg/ml DNase I (Roche). The spleens were infused with collagenase D using a 1-ml syringe with a 25-gauge needle, cut into small pieces, and filtered through nylon mesh. Larger pieces of tissue that did not go through the nylon mesh were incubated in 5 ml of collagenase D/DNase I for 45 min at 37°C and filtered through nylon mesh again. The cells were spun at 200 × g for 10 min at 4°C and washed twice. Cells were resuspended in PBS (Life Technologies) plus 0.5% BSA (Sigma-Aldrich) and 2 mM EDTA (Life Technologies), Fc receptors were blocked with 2 mg/ml mouse Ig (Sigma-Aldrich) for 15 min at 4°C, and CD11c beads (Miltenyi Biotec) were added for an additional 15 min at 4°C. Cells were washed once and purified using an LS column (Miltenyi Biotec). Purified CD11c⁺ DC were plated in petri dishes in the presence of 12.5 μg/ml LPS (Sigma-Aldrich) and 1 μg/ml anti-CD40 (BD Biosciences). After 16–18 h, cells in suspension were collected, separated on a Ficoll-Hypaque gradients (>1.033 g/ml), and washed twice with medium.

OT-II × Rag^−/− TCR Tg splenocytes were activated in RPMI 1640 (Invitrogen) medium supplemented with 10% FBS (HyClone), nonessential amino acids, l-glutamine, sodium pyruvate (Mediatech) plus 2 μM of 2-ME (Sigma-Aldrich) in the presence of 5 μM OVA peptide. T cells were used on days 5–8 after activation. CD80-eYFP I-E^k Chinese hamster ovary (CHO) cells and 5C.C7 TCR Tg T cell activation has been previously described (12).

Lymph node suspensions were first stained with CD4-FITC and CD25-allophycocyanin. Stained cells were then fixed and permeabilized in BD Cytofix/Cytoperm (BD Biosciences) for 30 min at 24°C with 1% paraformaldehyde, 0.5% Tween 20, and stained with FoxP3-PE for 1 h at 24°C. All Abs were purchased from BD Biosciences.

For anti-Golgi staining, Ab clone 53FC3 against mannosidase II (Covance Research Products) was used at a 1/10,000 dilution with a goat anti-mouse secondary Alexa Fluor 488 conjugated (Invitrogen). For microtubule staining, cells were fixed with 2% paraformaldehyde in PBS at 37°C for 30 min and washed three times with PBS/human serum albumin. Cells were permeabilized with 0.1% Triton X-100 for 5 min at 24°C, washed three times, blocked with 5% goat serum in PBS for 1 h, and stained with 20 μg/ml anti-β-tubulin, clone TUB 2.1(Sigma-Aldrich), for 1 h at 24°C. Ab was diluted in staining buffer, 10% FBS, and 0.1% Tween 20 in PBS. Cells were washed three times and goat anti-mouse secondary Ab (Molecular Probes) was added at 1/250 for 30 min at 24°C and washed three times.

For Immunofluorescent staining and flow cytometry, allophycocyanin-conjugated anti-CD80 (1610A), anti-CD3ε;, and anti-CD86 were purchased from eBioscience. Anti-CD28 (14-4-4s) and anti-CTLA-4 were purchased from BD Biosciences and anti-PKCθ polyclonal Ab was purchased from Santa Cruz Biotechnology. Alexa Fluor 546- conjugated secondary goat anti-Syrian hamster and goat anti-rabbit were purchased from Molecular Probes and Cy5- conjugated goat anti-Armenian hamster and anti-mouse were purchased from Jackson ImmunoResearch Laboratories. For acute blocking assay, anti-CD11c and anti-I-A^b Abs were purchased from BD Biosciences.

For T cell-DC interaction, glass-bottom chambers were coated with 50 μl of 200 μg/ml human fibronectin (Fisher Scientific) for 1 h and washed twice with sterile H₂O. CD11c⁺ purified and overnight-activated DC in serum-free medium were plated onto the glass chamber for 1 h at 37°C with 5% CO₂, and medium with OVA peptide was added for an additional 2 h. Days 5–8 after activated or naive OT-II × Rag^−/− T cells were then added to the glass-bottom chambers for 30 min and fixed as described above. For live cell images, H57-F(ab′)₂-Alexa Fluor 568 or H155-F(ab′)₂-Alexa 568 (H57 and H155 are nonblocking Abs to TCR and LFA-1, respectively) were used.

For fixed cell images, samples were prepared as follows: after T cells-CHO cells interacted for 10–15 min at 37°C and 5% CO₂ in a Lab-Tek II Chamber 1.5 German Cover Glass System, 8-well glass-bottom slides (Nalge Nunc International), they were briefly washed with PBS to remove excess T cells that failed to conjugate with the CHO cells. Sample were then fixed at 24°C with 4% methanol-free formaldehyde/PBS for 10 min (for permeabilized membrane) or with 2% methanol-free formaldehyde/PBS for 2 min (for nonpermeabilized membrane). The cells were washed three times with HBSS and incubated with primary Abs for 1 h at room temperature, washed, blocked with goat serum overnight at 4°C, and followed by secondary Abs for 1 h at 4°C.

For calcium imaging, OT-II T cells were labeled with 3 μM Fluo-LOJO (K_Ca²⁺ = 440 nM) for 30 min at 24°C in 1 ml of serum-free medium, washed twice, resuspended in medium with 10% serum. and incubated for 30 min at 37°C. Images of different fields were taken 5 min after adding T cells to DC, Alexa Fluor 647-conjugated H155 Fab and Alexa Fluor 568- conjugated H57 Fab. The 5-min time point was used as the standard observation point for T cell-DC interactions. Subsequently, anti-I-A^b (30 μg/ml) was added and images were collected every 4–5 min. Fluo-LOJO-loaded T cell anti-MHC isotype control (30 μg/ml) were imaged as control for photobleaching. LFA-1, CD80, and TCR cluster fluorescent intensity were measured at the interface using Image J (National Institutes of Health) before and after anti-MHCp treatment. Data analysis was done in Metamorph 5.0 (Universal Imaging) and Image J (National Institutes of Health).

Images were taken with an LSM 510 laser scanning confocal microscope (Zeiss) and ×100 PLAN Apochromat objective having a numerical aperture of 1.4. Image stacks consisted of 8–15 planes spaced by 0.48 μm. The images were analyzed using LSM (Zeiss) and Volocity (Improvision). For three-dimensional (3D) visualization of intercellular contacts, only those complexes were taken into consideration whose contact areas were oriented properly to be contained in a rectangular volume for an en face projection. Quick-time virtual reality (qtvr) allows multiple angle of 3D viewing of images.

Supported planar bilayers were formed using unilamellar liposomes containing Cy5-ICAM-1-GPI at 200 molecules/μm² and Ni²⁺ chelating 1, 2-dioleoyl-sn-glycero-3-succinyl-nitriloacetic acid (DOGS-NTA; Avanti Polar Lipids) to allow attachment of 6His-tagged soluble I-A^b. I-A^b-6His-OVA_323–339 was expressed in S2 insect cells and purified by immunoaffinity chromatography using a mAb to I-A^b (M5/114) coupled at agarose. The OVA_323–339 peptide (ISQAVHAAHAEINEAGR) was covalently linked to the β-chain of I-A^b. Varying concentrations of I-A^b were applied to planar bilayers containing 10% NTA lipids to generate different site densities. The concentration necessary to generate each site density was quantified by FACS using FITC-labeled M5/114 with calibration using FITC standard beads (Bangs Laboratories). All components were >80% mobile in the final bilayer preparations.

Previously activated day 6 OT-II TCR Tg T cells were loaded onto bilayers and fixed after 15 min with 2% paraformaldehyde. They were then permeabilized with 0.05% saponin. TCR was visualized by Alexa Fluor 568-H57 anti-TCR Fab and PKCθ was visualized with a rabbit polyclonal anti-PKCθ Ab from Santa Cruz Biotechnology and an Alexa Fluor 488-goat anti-rabbit F(ab′)₂.

Bilayer imaging was performed using wide-field fluorescence or total internal reflection illumination on an Olympus microscope with a PLAPO ×60 with a 1.45 numerical aperture objective (Olympus America). Images were acquired using an Orca-ER (Hamamatsu) camera with a pixel size of 0.2 μm. Images were acquired using IPLab software, background was subtracted using Image J, and analyzed using Metamorph.

Pearson’s correlation coefficient (r) was used to determine the degree of colocalization, segregation, or no correlation of the CD80, CD28, and PKCθ clusters in the IS (25). Pearson’s correlation coefficient (R_r) is one of the standard techniques applied in pattern recognition for matching one image to another to describe the degree of overlap between the two patterns. In Pearson’s correlation, the average pixel intensity values are subtracted from the original intensity values. As a result, the value of this coefficient ranges from −1 to 1, with a value of −1 representing a total lack of overlap between pixels from the images, and a value of 1 indicating perfect image registration. Pearson’s correlation coefficient accounts only for the similarity of shapes between the two images and does not depend upon image pixel intensity values.

A single sample t test was performed to establish whether the correlation coefficient mean value is significantly different from zero. The Fisher exact test was used to determine whether the pattern accumulation is associated with wild-type (WT) DC and CD80-eCFP DC. The multinomial test for equal probabilities was used to determine whether there is a difference among CD80 accumulation patterns. The exact binomial test was used to determine whether the proportion of CD80-eCFP response to anti-MHC Ab treatment by the lost of clustering is the same as the group of no effect. Fisher’s exact test was used to determine whether the difference in CD80 cluster loss by anti-MHC and anti-CD11c treatment is significant. The paired t test (after natural log transform on the data) and the Wilcoxon-signed rank test were used to determine whether there exists a difference in fluorescent intensity before and after anti-MHC Ab treatment of Ca²⁺, LFA-1, CD80, and TCR fluorescence.

To study the dynamics of CD80 clustering in the IS between primary T cells and DC, we generated BAC Tg mice that expressed a CD80-eCFP chimera under control of CD80’s endogenous promoter. Escherichia coli containing BAC clone RP23-69GS were purchased from the National Center for Biotechnology Information. This clone spans a 185-kb chromosome insert containing the complete CD80 gene near its center. The CD80 gene was cloned into a shuttle vector and eCFP appended to the end of the cytoplasmic tail by overlapping PCR. Homologous recombination between the shuttle vector and the purified BAC DNA RP23-69GS was performed in E. coli (Fig. 1,A). Positive clones containing CD80-eCFP in RP23-69GS were confirmed by PCR and Southern blot, in which digestion with HindIII and XbaI produced a fragment size of either 2 kb from the WT BAC or 2.8 kb from the recombinant BAC encoding CD80-eCFP (Fig. 1 B). The CD80-CFP band appeared stronger than the endogenous CD80 band due to the different amount of DNA loaded. CD80-eCFP BAC DNA was purified from E. coli and the size and integrity were confirmed by pulse-field gel electrophoresis before injection into C57BL/6 oocytes to generate Tg mice (data not shown).

Tg founders were crossed to CD80/CD86^−/− mice on the C57BL/6 background (26) and screened for total CD80 expression on mature DC relative to WT mature DC. Tg founders with similar total CD80 (with one WT CD80 allele and one CD80-eCFP allele) to WT DC would have physiological expression from the CD80-eCFP BAC Tg. We obtained one founder in which CD11c⁺ splenic DC matured in vitro had a similar mean fluorescent intensity after anti-CD80 staining, 987 for CD80-eCFP BAC Tg × CD80/CD86^−/+ vs 826 for WT (Fig. 1,C). We then confirmed that eCFP was indeed fused to CD80 to generate a 75-kDa protein in mature CD11c⁺ spleen cells via anti-GFP immunoprecipitation, SDS-PAGE, and detection by immunoblotting with anti-GFP Ab (Fig. 1,D). There was no evidence of proteolytic generation of eCFP, which would have a relative molecular mass of 25 kDa. We were not able to detect CD80-eCFP on unstimulated CD11c⁺ spleen cells consistent with expected induction of CD80 on DC maturation (Fig. 1 D).

We determined whether CD80-eCFP expression and accumulation pattern during a T cell-DC interaction followed that of the endogenous CD80. Splenic CD11c⁺ DC were matured overnight with LPS/anti-CD40 and then pulsed with 5 μM OVA peptide. OT-II TCR Tg T cells labeled with Alexa Fluor 633-conjugated anti-LFA-1 (H155 clone, nonblocking) Fab were added to these DC and fixed at 30 min. The DC were obtained from CD80-eCFP BAC Tg mice that were heterozygous for endogenous CD80 and CD86. Anti-CD80 staining colocalized with the eCFP fluorescence in the IS (Fig. 2,A). We noticed that there were intracellular accumulations of CD80 in the DC and hypothesized that this was associated with the Golgi apparatus. To test this. we performed colocalization experiments with mannosidase II, a marker for the cis-face of the Golgi apparatus. Anti-CD80 mAb staining colocalized with anti-mannosidase II Ab staining in both the WT DC (Fig. 2,B), and CD80-eCFP BAC Tg × CD80/CD86^−/− DC (Fig. 2,C). We further characterized the intracellular pool of CD80 accumulation in relation to the microtubule-organizing center (MTOC) by staining with anti-β-tubulin. We found that intracellular CD80-eCFP was closely apposed to the MTOC (Fig. 2 D) in mature CD11c⁺ DC, consistent with Golgi localization. In subsequent experiments, we use the Golgi-associated CD80 fluorescence for orientation in the DC and excluded the Golgi-associated CD80 fluorescence from analysis of CD80 clusters in the IS.

To demonstrate functional reconstitution of costimulation by the CD80-eCFP BAC transgene, we exploited the observation that mice deficient in both CD80 and CD86 molecules have a greatly reduced regulatory T cell (Treg) population (27, 28). T cells from the lymph nodes of CD80-eCFP BAC Tg × CD80/CD86^−/− mice were examined for Tregs by flow cytometry based on CD25 and FoxP3 expression. As seen in Fig. 3, ∼10% of the CD4⁺ T cells were double positive for CD25 and FoxP3 in both WT (Fig. 3,A) and CD80-eCFP BAC Tg × CD80/86^−/− (Fig. 3,C), but only ∼1% were double positive in CD80/86^−/− (Fig. 3 B). This demonstrates that one allele of the CD80-eCFP BAC Tg is functional in maintaining Tregs in vivo. Thus, CD80-eCFP functions similarly to WT CD80 and validates its use for IS studies.

We have previously shown that CD80-eCFP or CD80 WT colocalized with CD28 and CTLA4 in model IS formed between T cells and MHCp expressing CHO cells, but were mainly segregated from TCR clusters (12). Since the location of CD80 clusters and its relation to TCR clusters in the T cell-DC IS has not been determined, we first performed fixed immunofluorescence analysis to compare the pattern formed by CD80 WT to that formed by CD80-eCFP BAC Tg. We evaluated the localization of CD80 in relation to the TCR in the conjugates formed between OT-II T cells and WT DC or CD80-eCFP BAC Tg × CD80^+/−/86^−/− (CD80-eCFP) DC. Immunofluorescence staining for CD80 and TCR was performed with conjugates fixed after coincubation for 30 min at 37°C. Three accumulation patterns were scored for >40 fixed cell contacts for each DC type in two independent experiments. CD80 clusters segregated from the TCR clusters were scored as segregated and CD80 clusters colocalized with the TCR clusters partially or entirely were scored as colocalized. When there were no CD80 clusters in the interface, these were scored as none, which may reflect no CD80 redistribution or redistribution at a level that cannot be distinguished from the basal CD80 at the plasma membrane.

Both WT DC (Fig. 4^,A; see supplemental qtvr 1 for 3D views)⁵ and CD80-eCFP DC (Fig. 4,B; see supplemental qtvr 2 for 3D views) had CD80 clusters (solid yellow arrow) that were segregated from the TCR clusters (solid white arrow) in 63% (SD = 0.7%) and 59% (SD = 2.8%) of IS, respectively. In contrast, only 20% (SD = 1.4%) and 25% (SD = 3.5%) of the time did CD80 colocalize with TCR, respectively (Fig. 4 C). CD80 and TCR clusters were considered segregated when >50% of the CD80 clusters had no overlap with TCR clusters. We previously reported similar finding in the T cell-CHO cell model system (12). Unlike the IS formed with the CHO cell bilayer and B lymphoma presenting MHCp, a central TCR cluster was rarely observed in the T cell-DC IS.

Statistical analysis was performed to ask whether the IS patterns formed by WT DC vs CD80-eCFP DC are different. Fisher’s exact test from two independent experiments gave a p value of 0.521 and 0.787, respectively, and a combined p value of 0.850, suggesting no difference between accumulation patterns in IS of T cell-WT DC and T cell-CD80-eCFP DC. Additional statistical testing was performed to ask whether the CD80 clusters were segregated from the TCR clusters significantly more than CD80 clusters colocalized with the TCR cluster. Using the multinomial test for equal probabilities, in the T cell-WT DC IS, individual experiments have p values of 0.011 and 0.036, respectively, and a combined p value of 0.0005. In the T cell-CD80-eCFP DC IS, individual experiments have p values of 0.044 and 0.0002, respectively, and a combined p value < 0.0001. Thus, the results indicate that CD80-eCFP expressing DC form similar molecular patterns in the IS as WT DC and that the molecular segregation of CD80 clusters from TCR clusters in the IS is statistically significant. These results suggest that the multiple close contacts observed in recent electron microscopy studies may represent distinct TCR clusters and CD80 clusters (17).

To determine whether the failure to observe a well-defined pSMAC and cSMAC in the OT-II T cell-DC IS is a property of the OT-II TCR system or a function of DC, we used the supported planar bilayer system. The supported planar bilayer system has previously been shown to fully reconstitute cSMAC formation similarly to B cells. The I-A^b-OVA_323–339 was attached to supported planar bilayers at 2, 20, or 100 molecules/μm². The OVA_323–339 peptide binds to I-A^b in three different registers such that <10% of the complexes are ligands for the OT-II TCR and thus we used 10-fold higher pMHC densities than we previously used for I-E^k-MCC_91–103, which binds in a single register. We included ICAM-1 and CD80 in the supported planar bilayers to enable pSMAC and cSMAC formation, respectively. Fig. 5 shows representative IS formed by previously activated OT-II T cell-formed IS with pSMAC and cSMACs in an I-A^b-OVA_323–339 dose-dependent manner (p < 0.0001, χ² test). I-A^b-OVA_323–339 at 0.2 molecules/μm² did not induce any TCR clustering or ICAM-1 rings. As previously described, the lowest density of MHCp that induced activation did not produce a significant TCR accumulation in the cSMAC, but did generate an ICAM-1 ring and PKCθ clustering in a single central structure. When 10–50 times higher densities of I-A^b-OVA_323–339 were used, a single central TCR accumulation was observed. It is notable that even in the supported planar bilayer system the TCR cluster in the cSMAC was segregated from the PKCθ cluster, which often formed a well-defined ring between the central TCR cluster and ICAM-1 ring. Based on precedents defined by Kupfer and colleagues (9), we would refer to the central TCR clusters and tightly apposed PKCθ ring as together comprising the cSMAC. In our analysis of T cell-DC IS, we used MHCp doses that are >10-fold over the threshold for T cell activation. Under these conditions, the OT-II T cells were capable of forming a single, well-defined cSMAC with a planer bilayer, but we have shown that they instead formed distinctly multifocal IS with activated splenic DC.

We next wanted to take advantage of the CD80-eCFP DC to perform real-time imaging of T cell-DC interfaces to determine whether these were stable IS. OT-II T cells interacting with CD80-eCFP DCs were imaged by 3D confocal microscopy in which each 3D data set required ∼10 s to acquire, and 3D data sets were acquired of the same conjugates at intervals of 30–120 s. The time series started 5 min after adding T cells to OVA peptide-pulsed mature CD11c⁺ splenic DC. The TCRs were visualized on live T cells by staining with Alexa Fluor 547-labeled anti-TCRβ (H57 clone, nonblocking) Fab, which was maintained at a low concentration of 1 μg/ml in the coculture during imaging to maintain staining. The series of 3D projections revealed that the T cells remained with the same OVA peptide-pulsed DC (Fig. 6,A) for the entire imaging period up to 10 min, even though the pattern of CD80 and TCR clusters changed over this time (Fig. 6 B; see supplemental qtvr 3 and 4 for 3D views and supplemental movie 1).

Although the pattern of the CD80 and TCR clusters in the IS changed overtime, CD80 and TCR remained segregated. The degree of CD80 and TCR cluster segregation over time in Fig. 6,B were quantified by Pearson’s correlation coefficient (r). The values of this coefficient range from −1 to 1. A value of −1 represents perfect segregation, a value of zero represents random localization, and a value of 1 represents perfect colocalization. For the interface of Fig. 6,B time series, we obtained r = −0.568 with SE = 0.131 (Fig. 6 C). Pearson’s correlation coefficient accounts only for the similarity of shapes between the two images after application of a threshold and does not depend upon image pixel intensity values. To test the significance of the r value obtained, we asked whether CD80 and TCR clusters are negatively correlated. One sample t test gave a p = 0.0076, suggesting that the r value is significantly different from zero and that CD80 and TCR clusters are negatively correlated. These data are consistent with segregation of dynamic CD80 clusters and TCR clusters with overlap due to resolution limits of the imaging system.

The question remains whether these CD80 clusters in the IS were bound by CD28. We have previously shown in the effector T cell-CHO cell system that CD28 and CD80 colocalized at 30 min (12). We used this system as a positive control here to demonstrate that CD80-eYFP and CD28 are extensively colocalized in the T cell-CHO cell conjugates at 15 min (Fig. 7, A and B). Occasionally, we did observe CD80 clusters that did not have coassociated CD28 staining, but this was <10% of CD80-eYFP clusters. The degree of CD80-eYFP and CD28 cluster colocalization was quantified by the mean Pearson correlation coefficient, which was determined to be r = 0.411 (n = 13) with SE = 0.045 (Fig. 7 F, ▴). We asked whether CD80-eYFP and CD28 clusters are positively correlated. One sample t test gave a p < 0.0001, suggesting that the r value is significantly different from zero and that CD80-eYFP and CD28 clusters are colocalized.

In the same system, there was no evidence for significant CTLA-4 clustering with CD80-eYFP at 15 min (Fig. 7 C). The conjugates were fixed with a nonpermeabilized condition to ensure that only surface CTLA-4 was stained. Therefore, in the absence of CD86, CD28 preferentially bound to CD80 and these clusters participated in T cell activation as evidenced by their colocalization with PKCθ (data not shown). This is supported by our previous result where CD28, CD80, and PKCθ were colocalized at 30 min in effector T cell-CHO cell conjugates (12).

We then determined whether CD80 and CD28 are colocalized in the T cell-DC IS. The DCs used in these studies were CD80-eCFP BAC Tg × CD80^+/−/CD86^−/−. At the 15-min time point, CD80-eCFP and CD28 were colocalized in the IS (Fig. 7,D). There was a relatively small area of CD28 accumulation that did not contain CD80-eCFP accumulation and the converse. This observation is consistent with extensive engagement of CD28 and CD80-eCFP where some component of CD28 and CD80-eCFP clustering is not strictly dependent upon a 1:1 receptor-ligand interaction. Nonetheless, CD80-eCFP clusters extensively colocalized with PKCθ staining in the T cell-DC IS (Fig. 7,E), which is consistent with CD28-mediated signaling activity associated with each CD80-eCFP cluster even when equivalent CD28 clustering was not evident. The degree of CD80-eCFP and CD28 cluster colocalization (Fig. 6,F, ○) and CD80-eCFP and PKCθ (Fig. 7 F, ▵e) were quantified by Pearson’s correlation coefficient, r = 0.217 (n = 16) with SE = 0.063 and r = 0.315 (n = 13) with SE = 0.048, respectively. We asked whether CD80-eCFP and CD28 clusters and CD80-eCFP and PKCθ are positively correlated, and one sample t test gave a p = 0.0035 and p < 0.0001, respectively. This suggests that the r values are significantly different from zero and that CD80-eCFP is colocalized with CD28 and PKCθ clusters.

To study the relationship between TCR signaling and CD80 interactions in the IS, we set out to disrupt new TCR-MHCp interactions at defined times with a MHC-blocking Ab (29). We first tested whether anti-I-A^b can block TCR signal by labeling OT-II TCR Tg T cells with the Ca²⁺ sensitive dye Fluo-LOJO. OT-II T cells showed elevated Fluo-LOJO fluorescence within 5 min of introduction to OVA-pulsed DC (Fig. 8, A and C). This elevated fluorescence was reduced to basal levels within 4 min of anti-I-A^b (anti-MHC) Ab treatment (Fig. 8, B and D), as previously described in cellular systems (29). This is also similar to results on supported planar bilayers in which Ca²⁺ signaling is reduced to baseline within 2 min as the last TCR microcluster reach the cSMAC (13). We quantified Fluo-LOJO-labeled T cell fluorescence before and after anti-MHC treatment (Fig. 8,E) and asked whether the reduction in fluorescent intensity is statistically significant. Both the Wilcoxon-signed rank test and the paired t test (after log transform) gave a p < 0.0001, demonstrating that the difference is statistically significant. This reduction in fluorescence was not due to photobleaching since the same number of imaging iterations produced no reduction in fluorescence when the cells were treated with a control Ab (data not shown) and the addition of ionomycin still increased Fluo-LOJO fluorescence after anti-MHC treatment (data not shown). No changes in fluorescence were observed in the absence of Fluo-LOJO dye. We also quantified LFA-1 fluorescence intensity before and after anti-MHC treatment in T cells that showed a Ca²⁺ reduction (Fig. 8 F) and asked whether the change in LFA-1 fluorescent intensity is statistically significant. Statistical analysis using the Wilcoxon-signed rank test and the paired t test gave a p = 0.6122 and 0.3236, respectively. Thus, the change in LFA-1 fluorescent intensity after anti-MHC treatment is not significant. LFA-1 remained polarized toward the IS for at least 5 min, and the IS remained intact as previously described in the T cell-B cell lymphoma model system (30). The LFA-1 staining was due to LFA-1 expression on T cells since LFA-1 expression on DC is not detectable with our image parameters.

In contrast, the treatment of T cell-DC IS with the anti-MHC Ab dramatically reduced CD80 clusters within 4 min (Fig. 9, A, B, D, and E). We quantified this result in two ways. First, we measured the CD80 fluorescent intensity at the interface before and after anti-MHC treatment using areas outside the IS for reference. A total of 12 interfaces were quantified and we found that the fluorescent intensity relative to the baseline was reduced 70–90% after anti-MHC treatment (Fig. 9, C, F, and G). We asked whether the difference in CD80 cluster fluorescent intensity before and after Ab treatment is statistically significant. Statistical analysis using the Wilcoxon-signed rank test and paired t test resulted in p = 0.0005 and p < 0.0001, respectively. Thus, the change in CD80 cluster fluorescent intensity after anti-MHC treatment is significant. We also scored 36 ISs for reduction in the number of CD80 clusters vs no affect on CD80 clusters after anti-MHC treatment (Fig. 9 I). Our result suggests that 89% of the IS had CD80 cluster loss 4 min after anti-MHC treatment. We asked whether the loss of CD80 clusters vs no effect after anti-MHC treatment was statistically significant. Statistical analysis using the exact binomial test gave a p < 0.0001. In the irrelevant Ab control group, anti-CD11c treatment resulted in loss of CD80 clusters in 26% of the ISs, which we attributed to fluctuations in CD80 cluster number due to normal dynamics. We asked whether the difference in CD80 cluster loss between anti-MHC and anti-CD11c Ab treatment was significant. Statistical analysis using the Fisher exact test gave a p < 0.0001, demonstrating that significantly more clusters are lost after anti-MHC than with anti-CD11c treatment.

The anti-MHC treatment had a less dramatic effect on TCR accumulation. The fluorescent intensity of the TCR clusters in the IS was not dramatically reduced as compared with the CD80 clusters (Fig. 9, C and F–H). The TCR clusters continued to be dynamic with either no changes in fluorescence intensity (Fig. 9, C and H) or some decrease in fluorescence intensity of ∼20–40% relative to the baseline (Fig. 8, F and H). Although there was a small decrease in TCR cluster fluorescence intensity compared with CD80 clusters, the difference in TCR clusters fluorescence intensity before and after anti-MHC treatment was statistically significant with p = 0.016 and 0.0203,respectively, using the Wilcoxon-signed rank test and paired t test. Our result shows that the loss of CD80-eCFP clusters was not due simply to loss of IS formation or IS stability, but the loss of continuous TCR signaling. Thus, the maintenance of CD80-CD28 interaction in the IS requires continuous TCR signaling.

Our primary goal in this study was to advance knowledge of how TCR and costimulatory signaling are coordinated in the T cell-DC IS, an area that has not been studied. Our work builds on a substantial body of knowledge from model systems that have identified short-lived TCR microclusters and longer-lived cSMAC structures, both of which can associate with CD28-CD80 interactions that mediate an important component of costimulation. We used primary splenic DC purified from functionally validated BAC Tg mice in which CD80 is the only CD28 ligand and is labeled with eCFP. Our results provide insight into the organization of the T cell-DC IS and the dynamics of the CD80 clusters allows us to relate these to previously defined structures. We have found that the organization of CD80 clusters in the T cell-DC interface is most similar to that in T cell-transformed fibroblast (CHO) model systems (12, 31), more so than in T cell-B cell tumor systems, in that CD28-CD80 clusters are largely segregated from visible TCR clusters. Second, we found that CD80 clusters focus PKCθ in T cell-DC IS even though they are segregated from TCR clusters. We also found that the CD80 clusters are highly dynamic and require new TCR-MHCp interactions for maintenance, whereas the visible TCR clusters and LFA-1 clusters had greater “memory” for signals generated by new TCR-MHCp interactions. Thus, CD80 clusters have characteristics of microclusters in that they require recently established (within 4–5 min) TCR-MHCp interactions, but also share a characteristic with the cSMAC in that they are sites of PKCθ focusing in the IS.

We also verified that the multifocal TCR clusters observed between OT-II TCR Tg T cell and DC IS was not likely due to the inability of the T cells to form a cSMAC consisting of TCR clusters in the center of the IS with a suitable substrate. We found that OT-II T cells form a well-defined IS with supported planar bilayers that includes a cSMAC and pSMAC. As recently described for a different TCR system restricted by I-E^k, the cSMAC formed by the previously activated OT-II cells on bilayers with I-A^b-Ova_323–339 complexes, ICAM-1, and CD80 contains a readily resolved TCR-rich core surrounded by a ring of CD28 and PKCθ (32). These results suggest that T cells actively segregate TCR and PKCθ, but the failure to organize into a single cSMAC may be dependent on the DC.

One of the reasons that we chose to make BAC CD80-eCFP is because it has been shown that the expression of large DNA transgenes can accurately reflect the pattern of the endogenous chromosomal gene transcription (33). A major advantage of BAC Tg mice is that it allows the recombinant gene of interest to be under the control of its endogenous promoter, thus protecting its expression from position effect variegation. As a consequence of the size of BACs, a low copy number of concatamers is generally observed.

We found that CD80-eCFP is localized at the plasma membrane, but is also colocalized at the light microscopy level with the Golgi apparatus. The area around the centrioles is crowded with many vesicles; therefore, it is possible that the intracellular CD80 is associated with a distinct compartment from the Golgi. All type I transmembrane proteins that are expressed at the plasma membrane must traverse the Golgi apparatus (34). We did not perform an extensive comparison to other type I transmembrane proteins to determine whether this signal represents normal trafficking, retention in the Golgi or Golgi proximal compartments, or an intermediate in regulated secretion (33). Endogenous CD80 and CD80-eCFP displayed this localization to similar extents, demonstrating that this is not an effect of appending eCFP. We were surprised that we were unable to find any prior reference to CD80 localization near the Golgi apparatus, but most cell biology studies have focused on CD86 (35). Practically, this Golgi proximal signal is strong enough to require care in scoring of T cell-DC conjugates to make sure that the Golgi proximal staining is not misinterpreted as an IS- associated cluster.

Previous work has demonstrated an absolute requirement for either CD80 or CD86 for Treg maintenance in the periphery (27, 36). In contrast, CD80 and CD86 are not absolutely required for activation of conventional αβ T cells by most Ags (37). Thus, the single most robust test for in vivo CD80 function was the ability to maintain peripheral Tregs. We showed that the CD80-eCFP transgene crossed to CD80/CD86^−/− mice maintained peripheral Treg in normal numbers. This demonstrates that the expression pattern and cell biology of CD80-eCFP is similar to that of endogenous CD80, although we did not determine whether CD80-eCFP fully replaces function of both CD80 and CD86 in immune responses. Nonetheless, the ability of CD80-eCFP to maintain Tregs demonstrates that it is functional in vivo.

The OT-II T cell-DC IS contained many large TCR clusters that appeared to be scattered throughout the IS, consistent with recent reports (17). It is not clear whether these are structures involved in active signaling, postsignaling complexes, or a mixture of these. Our findings using OT-II Tg T cells and DC are distinct from earlier data that B lymphomas presenting suboptimal levels of MHCp induced multiple TCR clusters without a cSMAC (38), because we did not observe a cSMAC in T cell-DC IS at MHCp densities leading to maximal T cell stimulation with LPS/anti-CD40 matured DC. It is not clear whether the multifocal T cell-DC IS are due to distinct signals from DC surface molecules that direct the T cell cytoskeleton to general multiple TCR foci, due to resistance of the DC cytoskeleton to convergent movement by the T cell cytoskeleton, or due to specific characteristics of the OT-II TCR. The idea that the DC cytoskeleton plays an important role is consistent with observations that DC cytoskeleton contributes to T cell activation (39, 40, 41). Cytoskeletal restriction of MHCp and CD80 mobility on the DC may be important for DC to simultaneously maintain multiple ISs with different T cells without allowing the first IS formed to sequester resources from other IS (5, 42). However, treatments of DC with cytochalasin D and nocodazole before adding T cells did not disrupt the multiple TCR clusters observed in the T cell-DC interface (17). Further molecular analysis of different TCR Tg T cell systems and DC with higher spatial and temporal resolution is needed to answer these questions.

We previously observed that CD80 clusters segregated from TCR clusters and were more peripheral compared with the TCR in IS formed between naive T cells and CHO cells engineered to express I-E^k and CD80 (12). This peripheral segregation correlates with enhanced costimulatory signals through CD28. Such peripheral segregation was not previously observed in T cell-B lymphoma or T cell-planar bilayer IS (21, 22), but it is possible that this is due to physical overlap of structures that are segregated on the molecular scale. Parallel studies with microcontact printing on solid substrates in which anti-CD3 and anti-CD28 Abs were presented in a unifocal colocalized pattern (cSMAC like), a multifocal segregated pattern, or a multifocal colocalized pattern demonstrate that multifocal CD28 engagement, not segregation of CD28 and TCR engagement, is the critical parameter for enhanced IL-2 production (47).

Our primary motivation for generating CD80-eCFP BAC Tg mice was to be able to study the dynamics relationship of CD80 interactions in relation to the TCR in a physiological IS. We first performed dynamic 3D imaging and then used blocking anti-MHC Abs to perturb the system. Dynamic 3D imaging demonstrated that CD80 clusters and TCR clusters are mobile in the IS. Krummel and colleagues (23) noted that TCR and CD28 tend to colocalize in the early stages of TCR microcluster formation. These observations led us to hypothesize that CD80 clusters are formed around early TCR clusters that are too faint to detect by conventional confocal imaging. We speculate that with time these clusters gain TCR, turn off TCR signaling, after which CD80 disperses, leaving a TCR-only cluster. A reciprocal relationship between TCR and CD80 has been observed in the cSMAC in T cell-planar bilayer systems (43). In the dynamic T cell-DC IS, this process rapidly leaves larger visible TCR clusters without associated CD80 interactions. This is one possible explanation for why visible TCR clusters are segregated from CD80 clusters. The CD80-eCFP BAC Tg mice provide an excellent reagent to pursue this hypothesis using higher speed 3D-imaging approaches.

Acute blockade of new TCR interactions with MHC- blocking Ab stops signaling within 4 min and eliminated CD80 clusters in the same time frame. Interestingly, TCR clusters decreased less and LFA-1 clusters were stable in this time frame. This suggests that these TCR clusters visible by confocal microscopy in the T cell-DC interface are more cSMAC-like and supports the idea that these are postsignaling complexes.

The persistence of the LFA-1 clusters for longer after cessation of new TCR-MHCp interactions than CD80 clusters was surprising. LFA-1 activity on resting T cells is regulated in minutes by TCR signals (44). Perhaps LFA-1 clusters generate their own positive feedback signals once engaged and thus can persist even when TCR signals are extinguished (45).

The rapid dispersal of the CD80 clusters following blockage of new TCR engagement is consistent with a model in which the CD80 clusters form in response to TCR-nucleated CD28 clustering. In CHO cells expressing CD80-eYFP, the Ag- dependent clustering of CD80 corresponded exactly to sites of CD28 and PKCθ clustering (12). The segregation of CD80 from TCR clusters was observed in most, but not all T cell-DC IS. The mechanism for segregation may be based on loss of signaling in larger TCR clusters leading to loss of associated CD28 clustering activity or the physical segregation of MHCp clusters from CD80 clusters by the DC. Addressing these questions will require new imaging technologies that combine super-resolution and speed (46).

We thank Dr. G. Eberl for advice on BAC Tg construction and Dr. J. Lafaille for help with the in vivo characterization of Treg cells. We thank Drs. T. Cameron and R. Varma for critical reading of this manuscript. We acknowledge the New York University Cancer Institute Biostatistics shared core facility.

The authors have no financial conflict of interest.