The transfer of membrane proteins from APC to T cells was initially described in the 1970s, and subsequent work has described two mechanisms of transfer: APC-derived exosomes and direct transfer of small packets, while cells remain conjugated. Using fibroblast APC expressing a GFP-tagged I-Ek molecule with covalently attached antigenic peptide, we observed a third mechanism in live cell imaging: T cells spontaneously dissociating from APC often capture MHC:peptide complexes directly from the immunological synapse. Using two I-Ek-restricted murine TCR transgenic T cells with different peptide specificity, we show in this study that the MHC transfer is peptide specific. Using blocking Abs, we found that MHC:peptide transfer in this system requires direct TCR-MHC:peptide interactions and is augmented by costimulation through CD28-CD80 interactions. Capture of the GFP-tagged MHC:peptide complexes correlates with an activated phenotype of the T cell, elevated CD69 with down-modulated TCR. The transferred MHC:peptide molecules transferred to the T cell are associated with molecules that imply continued TCR signaling; p56lck, phosphotyrosine, and polarization of the actin cytoskeleton.
Tlymphocyte recognition of cognate MHC:peptide complexes results in the accumulation of cytoplasmic and membrane-bound proteins to the contact region ( 1, 2). These molecules are spatially and temporally segregated into distinct supramolecular activation complexes (SMACs)4 ( 3). The central supramolecular activation complex (c-SMAC) contains engaged TCR/MHC:peptide, CD28/CD80, and signaling-associated molecules such as protein kinase C (PKC)-θ and tyrosine-phosphorylated proteins surrounded by a peripheral SMAC containing, among others, adhesion molecules such as ICAM-1/LFA-1 ( 3, 4). The role of this immunological synapse is still unclear. It is widely accepted that it functions in sustaining TCR-mediated intracellular signaling ( 4, 5, 6, 7), but data from others have suggested that its main function is the polarized secretion of effector molecules ( 8, 9, 10, 11, 12). Alternatively, Shaw and colleagues ( 13) have hypothesized that the synapse regulates TCR down-modulation and endocytosis.
Another potential role of the immunological synapse is the capture of membrane-bound molecules from APCs. The literature on acquisition of MHC molecules by T cells from APC dates back over 20 years ( 14, 15, 16). Capture has been characterized as exosome/vesicle mediated ( 17, 18, 19) or requiring direct cell-cell contact. Recently, several groups have shown that for TCR-mediated capture, the immunological synapse is the location of transfer, while the cells remain conjugated ( 20, 21, 22, 23). This transfer is Ag dependent ( 20, 22), but it is unknown whether transfer involves only specific MHC:peptide molecules or whether nonspecific MHC:peptide ligands are also transferred. The specificity of transfer may be critical in controlling immune responses in vivo, as T cells functioning as APCs in vivo can induce anergy ( 24, 25). The stripping of specific MHC:peptide complexes from an APC may also provide a mechanism for T cell competition for MHC:peptide ligands, thus increasing the affinity of a specific response ( 26). The potential role of the T cell immunological synapse in MHC transfer is strengthened by the fact that both B cells ( 27, 28) and NK cells (29, 30, 31, 32) form immunological synapses and capture membrane proteins directly from them.
The mechanism of direct cell-to-cell, contact-dependent transfer in T-APC conjugates is unknown. It has been characterized by the transfer of multiple small packets of APC membrane containing surface proteins such as MHC ( 20), CD80 (33), and OX40L ( 34), while the cells remain conjugated. It is possible that this form of transfer is the result of exosomal release from the APC and subsequent uptake in the region of the immunological synapse. This is supported by a recent report from Boes et al. ( 35) showing polarized MHC class II delivery to the T cell-dendritic cell interface. Additionally, electron microscopy data from Patel et al. ( 19) demonstrate the presence of APC-derived exosomes in the space between T cells and APCs at the T-APC interface. However, this is not the only method of transsynaptic capture. Stinchcombe et al. ( 11) showed by electron microscopy that during the process of spontaneous CTL release from target cells, small membrane bridges formed and that MHC class I molecules and yellow fluorescent protein-tagged APC membrane were subsequently transferred to the T cell upon dissociation. Such dissociation-associated transfer is consistent with the finding that during transendothelial migration, T cells acquire endothelial membrane and membrane proteins such as CD31, CD49d, CD54, CD61, and CD62E ( 36).
Using fibroblasts expressing a GFP-tagged covalent MHC:peptide construct and costimulatory molecules ( 37) as APC, in this study we have characterized the transfer of MHC:peptide complexes from APC to T cells directly from the immunological synapse as a result of spontaneous cellular dissociation. In this system, MHC transfer is mediated by TCR/MHC interactions and is augmented by CD28/CD80 costimulation. Using T cells from TCR transgenic mice specific for a different antigenic peptide presented by I-Ek, we show in this work for the first time that the transsynaptic transfer of MHC following Ag recognition is specific for MHC molecules loaded with cognate peptide. The extent of MHC:peptide transfer to the T cells is associated with an activated phenotype (CD69high, TCR down-modulated). Finally, we report that the transferred MHC:peptide molecules are found on the surface of the T cell associated with the TCR and molecules involved in TCR signaling. The data further suggest that the MHC:peptide molecules are in their native orientation on the surface of the T cell, which correlates well with previous reports showing that after MHC transfer from APC, T cells can present Ag to other T cells ( 20, 22, 24, 25), and has implications for the initiation and control of an immune response.
Materials and Methods
Heterozygous AD10 TCR transgenic mice (Vβ3+), specific for pigeon cytochrome c fragment 88–104 ( 38) and reactive against moth cytochrome c (MCC) fragment 88–103 on a B10.BR (H-2k) background, were kindly provided by S. Hedrick (University of California, San Diego, CA) by way of P. Marrack (National Jewish Medical Center, Denver, CO). Homozygous 3.L2 TCR transgenic mice (Vβ8.3+), specific for peptide 64–76 of murine hemoglobin d allele (Hb), were kindly provided by P. Allen (Washington University, St. Louis, MO) ( 39). The mice were bred and maintained in specific pathogen-free conditions in the Oregon Health & Science University animal care facility. AD10 TCR transgenic mice were identified by PCR and flow cytometry.
Abs and staining reagents
The following conjugated or unconjugated Abs were purchased from BD Pharmingen: anti-I-Ek (17-3-3 and 14-4-4s), anti-CD80 (16-10A1), anti-ICAM-1 (3E2), anti-CD69 (H1.2F3), anti-Vβ3 (KJ25), anti-Vβ8.3 (1B3.3), anti-TCRβ (H57), anti-CD3 (145-2C11), anti-CD4 (Gk1.5), anti-CD25 (3C7), and anti-CD71 (C2), in addition to streptavidin-allophycocyanin, and streptavidin-CyChrome. The following reagents were purchased from Molecular Probes (Eugene, OR): AlexaFluor 594-conjugated phalloidin, anti-GFP AlexaFluor 594, chicken anti-rabbit IgG AlexaFluor 594, and anti-goat IgG AlexaFluor 350. Anti-PKCθ (C18-G) was purchased from Santa Cruz Biotechnology. Anti-phosphotyrosine (4G10) was purchased from Upstate Biotechnology. Rabbit polyclonal anti-Lck (2752) was purchased from Cell Signaling Technology. Anti-γ-tubulin (GTU-88) was purchased from Sigma-Aldrich. Anti-mouse IgG PE was purchased from Southern Biotechnology Associates, and Texas Red-conjugated donkey anti-goat IgG and Cy5-conjugated anti-mouse IgG were purchased from Jackson ImmunoResearch Laboratories.
MCC:GFP fibroblasts expressing enhanced GFP-tagged I-Ek β-chain with covalent antigenic MCC were described previously ( 37). As a control in some experiments, a second transfected fibroblast line, MCC:FKBP (FK506-binding protein), was used. This cell line expresses similar levels of CD80 and ICAM-1 and surface MCC:I-Ek to MCC:GFP. It differs in that the MCC:I-Ek β-chain is fused to three repeats of the FKBP (Ariad) rather than GFP. The MCC:FKBP cells express no GFP molecules. Cells were maintained in DMEM (Invitrogen Life Technologies) containing 10% FBS (HyClone) and supplemented with 1 mM l-glutamine, sodium pyruvate (100 mg/ml), 50 μM 2-ME, essential and nonessential amino acids (Invitrogen Life Technologies), 100 U/ml penicillin G, 100 U/ml streptomycin, and 50 μg/ml gentamicin (complete DMEM).
In vitro T cell priming
Single cell suspensions of splenocytes from 6- to 12-wk-old AD10 or 3.L2 TCR transgenic mice were depleted of erythrocytes by hypotonic lysis and resuspended in RPMI 1640 (Invitrogen Life Technologies) containing supplements, as described for complete DMEM (complete RPMI 1640). Cells were primed in vitro with 2.5 μM peptide (MCC 88–103 for AD10; Hb 64–76 for 3.L2) for 4–6 days with addition of 10 U/ml exogenous IL-2 on day 2. Lymphocytes were isolated from primed cultures by density centrifugation using Lympholyte M (Cedarlane Laboratories). T cells were resuspended at 5 × 106/ml in complete phenol red-free RPMI 1640 for use.
Live cell microscopy
For live cell microscopy, 2.5 × 105 APCs were seeded into 0.17-mm Delta T culture dishes (Bioptechs) 1 day before the experiment in 1 ml of complete DMEM. Dishes were fitted into a Bioptechs TC3 heated stage adapter and maintained at 37°C for the duration of the imaging. After adding 2.5 × 105 AD10 T cells to the dish, alternating ×400 or ×600 green fluorescent (528 nm) and differential interference contrast (DIC) images were taken every 8–12 s for 45 min with the Applied Precision Instruments DeltaVision image restoration system. This includes the Applied Precision Instruments chassis with precision motorized XYZ stage, a Nikon TE200 inverted fluorescent microscope with standard filter sets, halogen illumination with Applied Precision Instruments light homogenizer, a CH350L camera (500 kHz, 12-bit, 2 Megapixel, liquid cooled), and DeltaVision software.
MHC transfer experiments
To assess MHC transfer, unless otherwise stated, 1 × 106 MCC:GFP cells were plated into individual wells of a six-well plate and incubated overnight at 37°C before addition of 2.5 × 106 in vitro primed T cells. In Ab blockade experiments, blocking reagents were added to the APC or T cells 1 h before addition of T cells to the wells. After a 90-min incubation at 37°C, T cells were recovered from the cultures by rinsing with PBS. No additional dissociating reagents were added (e.g., EDTA or trypsin) to aid in T cell recovery. After washing with PBS, cells were aliquoted for fixed cell microscopy or flow cytometry.
Fixed cell microscopy
T cells recovered from APC-containing wells, as above, were incubated for 10 min on poly-l-lysine-coated LabTek II eight-chambered 0.15-mm cover glasses in PBS at room temperature. Cells were fixed by addition of ice-cold fixative (4% paraformaldehyde, 0.5% glutaraldehyde in PBS) and incubated for 30 min at room temperature in the dark. For intracellular staining, cells were permeabilized for 5 min with 0.1% Triton X-100 in PBS after fixation. Cultures were then stained with primary Abs at 10 μg/ml in PBS, or phalloidin (1/500 dilution in PBS), for 2 h at room temperature in a humidified chamber. Following three PBS washes of 10 min each, cells were incubated with secondary Abs at 5 μg/ml for 2 h at room temperature. After three more PBS washes, SlowFade Light antifade reagent (Molecular Probes) was added to the wells. Cells were stored at 4°C, protected from light until imaged.
Cells to be imaged were chosen based upon morphology in DIC and presence of a GFP signal. A stack of 50–90 fluorescent images spaced 0.2 μm apart in the z-axis was obtained at ×600 on the DeltaVision system and deconvolved using an iterative, constrained algorithm. Deconvolution and three-dimensional reconstructions were performed on an SGI Octane workstation (Applied Biosystems) using the Applied Precision Instruments SoftWorx software package. Further image analysis and preparation of images for publication were performed using SlideBook 4 software (Intelligent Imaging Innovations, Denver, CO) on a Macintosh G4 (Apple Computer).
T cells were stained with the indicated reagents for 30 min at 4°C in FACS buffer (PBS plus 2% FBS plus 0.1% NaN3). After three washes, cells were stained for 20 min with secondary reagents in FACS buffer. After three additional washes, cells were analyzed on a four-color FACSCalibur (BD Biosciences) without fixation. Data were analyzed with CellQuest 3.3 software (BD Biosciences).
Intercellular transfer of GFP-tagged MHC:peptide complexes from APC to T cells via the immunological synapse upon spontaneous T-APC dissociation
Using fibroblasts expressing GFP-tagged MHC:peptide complexes as surrogate APC in live cell imaging experiments, we observed in ∼10% of the T-APC interactions leading to mature immunological synapse formation that the T cells spontaneously dissociated from the APC during the 1-h imaging period ( 37). This cellular dissociation was accompanied by the intercellular transfer of GFP-tagged MHC:peptide molecules from the fibroblast APC to the T cells. In this study, we have characterized the phenomenon of intercellular MHC:peptide transfer from the fibroblast APC to the T cell during spontaneous cellular dissociation.
The images in Fig. 1 are representative of >50 transfer events that we have observed in >20 live cell imaging experiments. The on-line supplement contains a movie of the entire image sequence (Movie 1).5 In this set of images, three separate T cells (indicated by the three different arrows) interact with a single MCC:GFP cell. Each T cell forms an immunological synapse, as shown by the significant increase in GFP intensity at the T-APC interface, and subsequently dissociates from the APC. At time 0 in Fig. 1, the T cell indicated by the yellow arrow has already formed a mature immunological synapse, and 2 min and 36 s later it has dissociated from the APC, carrying with it a GFP spot. This spot remains visible for the duration of imaging, although the intensity appears to decrease (which could be owing to movement of the T cell and associated spot out of the plane of focus). The second T cell, indicated by the gray arrow, makes contact with the APC at ∼40 s (data not shown), and by 2 min and 36 s, there is a clear accumulation of MHC:peptide at the T-APC interface. This nascent synapse continues to grow, and by 7 min and 7 s a very bright accumulation of GFP is present, signifying the formation of a mature immunological synapse. At 11 min and 13 s, the T cell has dissociated from the APC, carrying with it a GFP spot directly from the immunological synapse. Again, the GFP is visible as a coherent spot on the T cell surface over 6 min later. The third T cell in Fig. 1 (pink arrow) makes contact and induces accumulation of GFP at the interface rapidly (∼2 min), but just as rapidly it dissociates from the APC without forming a mature synapse, carrying a large GFP spot with it. Dissociation and transfer of MHC precede mature synapse formation in ∼20% of transfer events.
In each case, there is transfer of MHC:peptide from the APC to the T cell directly from the immunological synapse. After dissociation, the region of increased GFP intensity on the APC where the immunological synapse had formed diffuses back into the membrane and becomes indistinguishable from other regions of the APC membrane. This is best seen with the spot from the T cell marked with the pink arrow, where the spot on the APC disappears between 11 and 17 min, but the process can also be seen on the region of the APC interacting with the T cell marked with the gray arrow over the same time span. However, unlike the GFP regions on the APC, the spots on the T cell do not diffuse into the membrane, but remain as cohesive spots.
MHC:peptide complexes transferred to T cells associate with TCR, signaling molecules, and the cytoskeleton
The live cell data demonstrated that the MCC:I-Ek:GFP transferred from the APC to the T cell via the immunological synapse upon spontaneous T-APC dissociation remained a focused, distinct spot on the T cell surface. To examine the possibility that the transferred MHC:peptide complexes are associated with other molecules on or within the T cell, T cells were recovered from a coculture with APC after 90 min, and after a 10-min incubation on poly(l-lysine)-coated coverslips at 37°C, were fixed and stained. An early time point was chosen to correlate with the duration of our live cell imaging experiments, where we observed the direct APC to T cell MHC transfer (Fig. 1). Because we were interested only in the cells that had spontaneously detached from the APC, no dissociating agents, such as EDTA, were used in the T cell recovery process. T cells were recovered by simply rinsing the cultures with PBS, leaving behind the adherent MCC:GFP cells and any conjugated T cells. At least 20 cells were imaged for each staining combination in three separate experiments, and representative images are found in Fig. 2. In the majority of T cells imaged (147 of 165), the GFP on the T cell was found as a single spot or a few tightly clustered regions on the T cell surface, in complete agreement with the live cell imaging data (Fig. 1). In the remaining T cells, the GFP was found as multiple smaller regions of increased GFP intensity randomly distributed on the T cell. The intensity of these individual spots was generally lower compared with the large GFP spots, being <2-fold above background vs >3-fold for the large spots. When stained with Abs, these smaller, significantly less intense spots associated with several of the stains, but the numbers of individual cells for each staining regimen were too small to attach significance to the phenotype.
To confirm that the green spot on the T cell surface was comprised of GFP-tagged MHC:peptide molecules, we stained the recovered T cells with anti-GFP and anti-I-Ek without detergent permeabilization of the T cell. In these studies, we defined specific accumulation as being 3-fold or more above background compared with other regions of the same T cell. We observed that the green spot colocalized with anti-I-Ek, but not with anti-GFP in the nonpermeabilized T cells (Fig. 2,A). However, after permeabilization in 0.1% Triton X-100, there was very good colocalization of the green spot with anti-GFP (Fig. 2, B and C). Taken together, the anti-GFP and anti-I-Ek staining confirms that the GFP spot transferred to the T cell contains MCC:I-Ek:GFP molecules acquired from the APC. Furthermore, the inaccessibility of the GFP to Ab in intact cells indicates that the GFP is on the cytoplasmic side of the plasma membrane. Combined with the surface I-Ek staining on intact cells, this suggests that the transferred MHC:peptide molecules are present on the T cell surface as transmembrane molecules.
When permeabilized cells were stained with anti-GFP and anti-Vβ3 (specific for the TCR transgene), we observed that the TCR also colocalized with the MCC:I-Ek:GFP (Fig. 2,B) in 29 of 34 cells imaged (85%). The fact that the MHC:peptide captured by the T cell is still in close association with the TCR raises the possibility that these molecules may still be interacting on the T cell surface, continuing to generate intracellular signals. To examine this possibility, cells were stained with an Ab specific for phosphotyrosine, and it was observed that the transferred MCC:I-Ek:GFP molecules and phosphotyrosine colocalized in 29 of 33 cells (88%) (Fig. 2, C and D). In addition, when the cells were permeabilized and stained with anti-p56lck, we observed that the GFP spot on the T cell colocalizes with Lck as well as increased levels of phosphorylated tyrosine (Fig. 2,D). The Lck staining pattern in Fig. 2 D is representative of 24 of the 31 T cells imaged (77%) in which GFP and Lck colocalize over three separate experiments. In three other T cells, there is a faint Lck ring surrounding the GFP (data not shown) reminiscent of the pattern seen for phosphorylated Lck ( 40). We also observed a weak colocalization of PKC-θ with the GFP spot, but it was <2-fold above background (data not shown), and thus did not reach the 3-fold threshold we set for specific accumulation. The accumulation of GFP-tagged MHC:peptide, TCR, Lck, phosphotyrosine, and weakly accumulated PKC-θ strongly suggests that the GFP spot is an area of sustained T cell intracellular signaling.
Finally, we examined the T cell cytoskeleton to determine whether there was polarization toward the captured MHC:peptide complexes. When the T cells were stained with phalloidin, it was observed in >60% of T cells that F-actin preferentially accumulated under the GFP (Fig. 2,E, 90° merge). The en face view (Fig. 2 E, merge) showed that the F-actin surrounds the GFP region on the T cell. Unlike the actin cytoskeleton, there was no microtubule-organizing center polarization toward the GFP (data not shown).
Detection of MCC:I-Ek:GFP and CD80 transfer from APC to the T cells by flow cytometry
The imaging data in Figs. 1 and 2 demonstrate that when T cells dissociate from APC, they capture MHC:peptide molecules directly from the immunological synapse, and that the transferred material colocalizes with signaling-associated molecules. We turned to flow cytometry to evaluate the frequency of transfer, time and dose dependency, molecular requirements, and peptide specificity of transfer.
In this and subsequent flow cytometry experiments, T cells were cultured with APC for 90 min before recovery by PBS wash, 30 min longer than the live cell imaging experiments to increase the frequency of transfer events. With incubation times longer than 90 min, the level of transfer detected increased, but significant levels of APC death were observed after 6 h (data not shown). Cellular debris can efficiently be captured by the T cells, as shown by use of freeze/thawed cell extracts (Fig. 3 E). The 90-min incubation was chosen to optimize direct cell to cell transfer, but avoid the killing of the APC and concomitant release and potential uptake of apoptotic debris.
In Fig. 3, representative of seven separate experiments, the T cells became GFP positive (Fig. 3,A) after a 90-min incubation with APC, showing that the transfer of GFP and I-Ek from the APC to the T cells is detectable by flow cytometry. The mean fluorescence intensity increased on the T cells by 1.93-fold over the unstimulated controls (shaded histogram). This correlated very well with the 1.92-fold increase observed in surface I-Ek staining (Fig. 3,B). These data confirmed that T cells capture MHC:peptide complexes directly from APC. Of note is the profile of both the GFP and the surface I-Ek on the T cells. It might be expected that the transfer would result in the appearance of GFP-positive and GFP-negative populations. However, the profiles clearly show a heterogeneous monomodal population (Fig. 3, A and B), implying a continuum of the amounts of material transferred to T cells recovered after 90 min, when the majority of recovered T cells had interacted with APC, as shown by up-regulation of CD69 (see below, Fig. 4).
To determine whether other APC membrane proteins, particularly members of the c-SMAC, were transferred to the T cells along with the MHC:peptide complexes, T cells were also stained for the presence of CD80. It has previously been established that murine T cells do not express CD80 endogenously, but instead capture it from APC ( 33). We observed that like the MHC:peptide, CD80 is efficiently transferred to the T cells during dissociation, with a mean fluorescence intensity (MFI) increase of 2.93-fold vs unstimulated controls (Fig. 3 C). Thus, transfer is not limited to just MHC:peptide complexes, but includes other membrane proteins found in the c-SMAC.
To further characterize the MHC transfer event, we examined the dose dependency of transfer by the numbers of APC available to the T cells. When the number of APC is increased with a constant number of T cells (2.5 × 106), there is a dose-dependent increase in the amount of GFP transferred (Fig. 3,D). With 105 APC (an APC:T ratio of 1:25), the GFP MFI increased by only 28% above unstimulated controls. At 5 × 105 MCC:GFP (1:5 APC:T ratio), the increase was 54%, while at 106 APC (the 1:2.5 APC:T ratio used in all flow cytometry experiments), the MFI increased by 1.76, similar to the values in Fig. 3 A. Finally, at 5 × 106 APC (2:1 APC:T ratio), the GFP MFI was 2.8-fold above background.
To exclude the possibility that the GFP capture measured by flow cytometry is due to acquisition of fibroblast-derived exosomes and/or cellular debris, the T cells were incubated with a 24-h fibroblast-conditioned medium or with medium containing cellular debris after multiple rounds of freezing and thawing (Fig. 3 E). The data clearly show that the T cells become only faintly GFP positive when incubated with conditioned medium (17% increase in MFI vs unstimulated), suggesting that this passive acquisition of vesicles/exosomes is not responsible for the increase in GFP observed by flow cytometry.
Thus, the data from Fig. 3 show that the transfer of MHC:peptide from APC to T cells occurs alongside the transfer of another c-SMAC component CD80 (Fig. 3,C), and that transfer is dependent upon Ag dose (Fig. 3 D). The amount of transfer is also dependent on the time of incubation (data not shown) and is not due to the passive acquisition of exosomes released from the fibroblasts.
Transfer is mediated by TCR-MHC interactions
The finding that the capture of MHC:peptide from the APC was directly from the immunological synapse (Fig. 1) and that transfer was dose dependent (Fig. 3,D) suggests that MHC:peptide transfer is mediated by TCR-MHC interactions. To determine the molecular requirements for MHC:peptide transfer, APC or T cells were preincubated with blocking Abs specific for MHC class II, CD3, CD80, or CD71 (transferrin receptor) for 1 h before the initiation of the coculture. The blocking Abs were present for the duration of the coculture. The fibroblast APC used in this study are Fc receptor negative, and so do not display Abs in a multivalent fashion on their surface. At the end of the 90-min coculture, the T cells were recovered, gated on CD4+, Vβ3+ cells, and examined by flow cytometry for GFP levels. The data in Fig. 3, F–I, representative of three separate experiments, show the GFP levels on unstimulated cells (shaded) or on MCC:GFP cells without (black line) or with blocking Abs (gray line). Without any blocking reagents present, the GFP MFI of the T cells after incubation with the MCC:GFP cells increased from 8.6 to 12.5, a 1.45-fold increase. To control for potential steric effects of an Ab bound to the surface of either the T cell or the APC, we used anti-CD71 (Fig. 3 F). When this Ab was present, there was no effect on GFP capture.
To assess the importance of costimulation to capture, we blocked with anti-CD80 (Fig. 3,G). Blockade of CD80/CD28 interactions reduced the GFP MFI by a modest 17%. The costimulation results contrast sharply with the data involving blockade of Ag recognition. When TCR/MHC:peptide interactions are inhibited by addition of anti-I-Ek (Fig. 3,H), there is a significant reduction in the amount of MHC:peptide transferred to the T cells. Anti-CD3 completely inhibits MHC:peptide transfer to the T cells (Fig. 3 I).
Taken together, the results of the blocking experiments shown in Fig. 3 show that the transfer of MHC:peptide from APC to T cells in this system requires TCR engagement of the cognate MCC:I-Ek ligand. These results also show that the interaction of CD80 with CD28 (or CTLA-4), while augmenting TCR-mediated transfer, is not essential for MHC:peptide capture.
Expression of GFP on T cells correlates with T cell activation phenotype
We next turned to characterizing the activation phenotype of the T cells that had captured MHC:peptide. Knowing the activation state of the T cells in the culture could provide important information about the cells that captured MHC:peptide from the APC upon dissociation. It is possible that the T cells that dissociated and captured MHC:peptide were not fully activated and the transfer was a result of abortive activation. To assess their activation state, T cells were cocultured with APC for 90 min and recovered, as previously described. As a control in these experiments, T cells were stimulated with a fibroblast APC transfected with a construct encoding the same extracellular complex, but with a nonfluorescent cytoplasmic domain, MCC:FKBP. The extent of T cell proliferation and induction of activation markers such as CD69 and CD25 induced by MCC:FKBP cells is comparable to MCC:GFP cells. The T cells were stained with Abs to CD4, Vβ3 (specific for the transgenic TCR), and CD69. The flow cytometry results for CD4+ cells, representative of five separate experiments, are shown in Fig. 4.
The left column of Fig. 4, A–C, shows TCR vs GFP levels. The unstimulated cells (Fig. 4,A) are TCRhigh and GFP negative. When stimulated by the MCC:GFP cells for 90 min, a substantial proportion of the cells down-modulates their TCR (Fig. 4,B), as expected. The TCR down-modulated population has a clear shift in the GFP expression. The mean of the population increases, and there is a loss of the very dim cells. The GFP MFI difference between the TCRhigh and TCRlow populations in Fig. 4,B is 2.1-fold. The GFP signal in Fig. 4,B is not due to an increase in green autofluorescence due to T cell activation, because the T cells stimulated with the GFP-negative APC (MCC:FKBP) display a similar level of TCR down-modulation, but no increase in GFP (Fig. 4 C).
To further examine the T cell phenotype, we compared CD69 and TCR levels (right column of Fig. 4, A–C). As in the left panel of Fig. 4,A, the unstimulated T cells are TCRhigh and are CD69low, with a very small population being CD69intermediate. However, when the T cells were stimulated with either MCC:GFP or MCC:FKBP cells, two additional populations appeared, TCRhigh CD69high and TCRlow CD69high (right panels in Fig. 4, B and C). Each population was gated, and the GFP mean fluorescence intensity was determined. The GFP MFI value for each population is shown in Fig. 4 D.
The table in Fig. 4 D shows that there is a correlation between T cell activation phenotype and the expression of captured GFP-tagged MCC:I-Ek. For all CD4+ T cells, there is a 64% increase in GFP MFI for the MCC:GFP-stimulated cells vs unstimulated, 44% higher than the MCC:FKBP-stimulated T cells. When the results are broken down into the three TCR vs CD69 phenotypes, the results are more striking. There are only small differences in GFP levels for the nonactivated TCRhigh CD69low populations, but for the clearly activated TCRhigh CD69high population, the difference between unstimulated and MCC:GFP is fairly large (58% increase in MFI). The biggest difference is observed with the CD69high cells that have down-modulated TCR, greater than 2-fold. Thus, based upon TCR and CD69 levels, there is a clear positive correlation between GFP-tagged MHC:peptide complex capture and the T cell activation state.
The correlation between activation state and MHC:peptide capture is dependent upon both CD69 expression and TCR down-modulation. This would predict that T cells with the most highly down-modulated TCR levels would be the brightest for GFP; however, the relationship between TCR level and GFP capture is more complex. Careful examination of Fig. 4,B (left column) shows that the T cells that had down-modulated TCR ∼5-fold were the brightest for GFP. However, as the TCR levels continued to decrease, GFP expression decreases until cells with 15- to 20-fold less TCR were essentially GFP negative. This is clearer in the histogram in Fig. 4 E, in which the GFP level is plotted as a function of TCR level. As the cells down-modulate receptor, the GFP level increases sharply, reaching a maximum at ∼4-fold less TCR, then sharply declines. This implies that as TCR is internalized beyond a threshold of ∼4- to 5-fold, MHC:peptide disappears and may also be internalized.
Based upon these data, we conclude that the levels of GFP-tagged MHC:peptide captured from the APC found on T cells are a function of the TCR down-modulation and the activation state of the T cell. Cells that have down-modulated TCR to a moderate amount (5- to 10-fold) and are CD69high express the highest amount of GFP. Further down-modulation of the TCR is associated with a reduction of GFP expression. These data also suggest that capture is not the result of incomplete or abortive activation.
MHC:peptide transfer to T cells is peptide specific
The process of direct intercellular transfer of MHC:peptide from APC to T cells in this study is TCR dependent, raising the possibility that transfer is peptide specific. To test this hypothesis, we used the fact that in addition to expressing the GFP-tagged I-Ek:MCC complexes, the MCC:GFP cells also express unlabeled wild-type I-Ek molecules that can be exogenously loaded with murine hemoglobin (Hb64–76) peptide and used to stimulate T cell proliferation ( 37). To determine whether MHC transfer is peptide specific, Hb-specific 3.L2 T cells were cocultured with MCC:GFP cells with or without 20 μM Hb peptide prepulse. If MHC:peptide transfer is peptide specific, the Hb-specific 3.L2 T cells should not pick up the irrelevant MCC:I-Ek:GFP molecules from the APC. If, in constrast, transfer is not peptide specific, the 3.L2 T cells, responding to Hb-loaded unlabeled MHC:peptide complexes, would be expected to capture GFP-tagged I-Ek:MCC complexes and should become GFP+. The results, representative of five separate experiments, are shown in Fig. 5 A. There is no observable difference in green fluorescence on the 3.L2 T cells whether they were cultured on Hb-loaded MCC:GFP cells or Hb-negative MCC:GFP cells. Thus, the Hb-specific T cells do not pick up irrelevant GFP-tagged I-Ek:MCC molecules from the APC.
In Fig. 5, B and C, MCC-specific T cells were mixed with Hb-specific T cells at a 1:1 ratio and then added to MCC:GFP cells that had been preloaded with 20 μM Hb by overnight incubation. After a 90-min coculture, both MCC-specific and Hb-specific T cells were recovered, gated for their respective TCRs, and examined by flow cytometry. Fig. 5,B shows that the two T cells are easily separated based upon Vβ expression. In Fig. 5 C, Ag is present for both types of T cells, but only the MCC-specific T cells capture MCC:I-Ek:GFP from the APC. The Hb-specific T cells, while they are stimulated to express CD69 (data not shown) and proliferate by the Hb-loaded MCC:GFP cells ( 37), do not capture any of the MCC-loaded, GFP-tagged I-Ek. Based upon these findings, we conclude that MHC:peptide transfer in this system is peptide specific.
In this study, we have described the transfer of specific MHC:peptide complexes from transfected fibroblast APC to T cells directly from the immunological synapse upon spontaneous cellular dissociation. Imaging (Fig. 2) and flow cytometry examination of T cells recovered from a 90-min coculture with APC (Fig. 3) suggest that the transferred MHC:peptide molecules are incorporated into a membrane on the surface of the T cell. Specifically, an anti-I-Ek Ab stained intact cells (Figs. 2,A and 3 B), but anti-GFP stained only after detergent permeabilization of the T cell’s plasma membrane. These findings are in agreement with the previously reported ability of T cells to act as APC after capturing MHC ( 19, 20, 22, 25, 33, 41, 42) and may have important implications in controlling an ongoing immune response, as discussed below. The I-Ek molecules may be incorporated into the plasma membrane of the T cell, or could be retained on membrane vesicles physically associating with the surface of the T cells via engaged TCR.
The capture of plasma membrane and membrane-associated proteins from APC has previously been described for B cells ( 27, 28), NK cells ( 29, 30, 31, 43), γδ T cells (23), as well as CD4+ and CD8+ αβ T cells. For T and NK cells, the mechanism usually falls into two categories, vesicle-mediated transfer involving exosomes from APC ( 17, 18, 44, 45) or, as described in this study, direct T cell-APC contact ( 46). The contact-dependent capture that we describe differs from previous reports in that transfer occurs as the removal of a large patch of GFP-tagged I-Ek:MCC (along with CD80) from the immunological synapse during dissociation, rather than by transfer of multiple small packets of material when the T cells and APC are tightly conjugated ( 20). It is reminiscent of the transfer of yellow fluorescent protein-tagged membrane during CD8+ CTL-APC dissociation described by Stinchcombe et al. ( 11). Using transmission electron microscopy, they described the formation of small regions of fused T-APC membranes (“bridges”) only during the process of dissociation. The resolution of these T-APC fusion events leads to transfer of APC surface molecules onto T cells. Although we consider it highly unlikely, based upon images in Fig. 1, it is nevertheless possible that the transfer event observed in this study is due to directed secretion of exosomes that are subsequently captured by the T cells.
When examined by flow cytometry (Fig. 3), the level of GFP and anti-I-Ek staining is similar to that reported by Hudrisier et al. ( 22), but appears relatively low compared with other previously reported data for CD4+ T cells ( 15, 17, 19, 25, 45, 47). This is most likely due to differences in experimental details, as the MCC:I-Ek:GFP level on the surface of our MCC:GFP cells is relatively low. The GFP signal is ∼7-fold above background ( 37). Also, T cells were incubated with the APC for only 90 min before recovery from the culture without the use of any dissociating reagents (e.g., EDTA) to insure that we were examining only cells that had spontaneously dissociated from the APC. Although transfer efficiency increases with incubation duration ( 25), we found that there was significant death of APC after 6 h of coculture (data not shown), and Fig. 3 E shows that T cells can become GFP positive when cultured with cellular debris, calling into question the mechanism of transfer at later time points.
It is unknown whether the phenomenon described in this work with fibroblast APC reflects the situation with physiologic APC. However, 5 days after adoptive transfer into transgenic mice expressing a similar MHC:peptide construct on professional APC, TCR transgenic T cells with significant amounts of captured MHC:peptide can be detected by flow cytometry (S. Wetzel and C. Huddleston, unpublished observation).
Ab-blocking experiments showed that the transfer described in this study was TCR mediated. Physical blockade of TCR or MHC prevented appreciable transfer onto the T cells (Fig. 3, G and H). These findings are in line with previously published data for CD8+ T cells, which showed that Ag-dependent transfer required TCR/MHC interaction and subsequent TCR signaling ( 20, 22). In contrast to studies using CD28−/− T cells showing that MHC class I transfer to CD8+ CTL required CD28/CD80 interactions ( 21, 33), our Ab-blocking experiments showed that transfer was reduced, but not prevented when CD28/CD80 interactions were blocked (MCC:GFP cells do not express CD86 ( 37)). This suggests that CD28/CD80 interactions augment the transfer, but are not necessary for transfer to occur. These findings are consistent with our previous report showing that blockade of CD28/CD80 ligation significantly reduced synapse size and organization, inhibiting the formation of a mature immunological synapse, but not preventing MHC:peptide accumulation at the T-APC interface ( 37).
We have also shown that MHC transfer correlated with an activated phenotype of the T cells. T cells expressing the highest levels of CD69 were also the brightest with regard to transferred MCC:I-Ek:GFP (Fig. 4,B). Captured GFP levels are also related to another marker of T cell activation, TCR down-modulation (Fig. 4, B and E). However, the relationship between TCR down-modulation and GFP capture was more complex. As T cells down-modulated TCR, they had increasing levels of captured GFP, but beyond a critical TCR threshold, ∼5-fold lower than resting T cell blasts, the levels of GFP began to precipitously drop (Fig. 4,E). All of the cells with TCR down-modulation were CD69high, indicating that they were acutely activated. Internalization of the TCR may lead to internalization and degradation of the associated GFP-tagged MHC:peptide complexes. This is supported by the colocalization of TCR and GFP that we observed (Fig. 2 B) along with time course data that showed that GFP levels dropped by ∼50% during the first 2 h after removal from APC before stabilizing and slowly decaying over the next 46 h (data not shown). These results are consistent with those of Huang et al. ( 20), who showed that transferred MHC class I was internalized and delivered to lysosomes for degradation.
The activation phenotype is also relevant when considering the spontaneous dissociation of the T cells from the APC. In two previous in vitro imaging studies, CD4+ T cells were seen to repeatedly associate, then dissociate from macrophages ( 48) or dendritic cells in a three-dimensional collagen matrix ( 49). One interpretation was that the cells were interacting with multiple APC partners, summing the activation signals until full activation was achieved. An alternate explanation for the spontaneous dissociation was that it was due to an abortive activation event leaving the cells partially activated. In this study, the T cells formed mature immunological synapses, expressed high levels of CD69, and displayed significant TCR down-modulation, suggesting that the T cells that dissociated from the MCC:GFP cells were fully activated. Importantly, the rapid association-dissociation cycling seen previously in vitro ( 37, 48, 49) is very similar to the phenotype seen recently by Mempel et al. ( 50) and Miller et al. ( 51) using two-photon imaging of T-DC interactions during the initial phases of T cell priming in lymph nodes in vivo, suggesting that rather than being an artifact, the in vitro phenomenon reflects a process that is physiologically relevant.
Previous studies have shown that transfer is Ag dependent ( 20, 22), and that only the restricting MHC is transferred, but the peptide specificity of transfer has not been resolved ( 15, 20). Several studies have inferred peptide specificity using indirect measurements such as cross-reactivity of CTL to the T cells after capturing MHC ( 20) or absence of a decrease in detectable expression of irrelevant MHC class I on the surface of a dendritic cell in vivo ( 26), but the system used in this study allowed us to directly address this question. We used the fact that the transfected fibroblast APCs express GFP-tagged I-Ek:MCC covalent molecules along with unlabeled wild-type I-Ek that can be exogenously loaded with Hb peptide to present Ag to Hb-specific T cells. The Hb-specific T cells did not capture GFP-tagged irrelevant I-Ek:MCC from the APC during the culture period, but MCC-specific T cells did (Fig. 5, A and C). Thus, peptide-specific MHC transfer predominates during dissociation of the T cells from the APC. Although apparently only MHC loaded with the cognate peptide ligand is transferred, the transfer of another c-SMAC component CD80 suggests that the specific MHC:peptide molecules are not plucked from the APC surface. It must be noted that the immunological synapse formed between the Hb-specific T cells and Hb-pulsed MCC:GFP cells is enriched only for specific MHC:peptide ( 37), consistent with results from Wülfing et al. ( 52), showing that at high peptide concentrations, accumulation of MHC molecules at the immunological synapse is peptide specific.
Perhaps the most intriguing data in this study is the imaging done 10 min after recovery of the T cells from the 90-min coculture with APC (Fig. 2), which showed that the GFP-tagged I-Ek:MCC molecules remain associated with the TCR upon capture as a distinct, focused spot (Fig. 2). Furthermore, the GFP-tagged I-Ek:MCC colocalize with the src family kinase p56Lck and high levels of tyrosine-phosphorylated proteins. These findings suggest that the transferred MHC:peptide molecules may be interacting with the colocalized TCR and sustaining intracellular signaling. If MCC:I-Ek:GFP molecules are in their native orientation in the T cell plasma membrane, it implies that the transferred MHC:peptide molecules, along with the c-SMAC component CD80, are involved in autopresentation by the T cells. Although the data in Fig. 2 A, showing that the TCR and transferred MHC class II colocalize, are similar to imaging showing that CD8 and transferred MHC class I colocalize ( 11), this would appear to be contrary to recent evidence that on a CTL membrane TCR and potential MHC:peptide ligands are spatially segregated to prevent autolysis ( 53). However, the later study dealt with endogenously synthesized molecules, while in this study the MHC class II:peptide complexes are drawn onto the T cells as a result of interactions with the TCR. When the CD8+ T cells were treated with agents that disrupted cell surface charge, TCR, CD8, and MHC all colocalized, leading to cell death via suicide, consistent with autopresentation ( 53). It is plausible, therefore, that T cells might sustain intracellular signaling via the TCR after MHC:peptide capture from an APC.
The results in this study, along with several previous reports, suggest that MHC:peptide capture by T cells may be important in controlling an immune response. First, removal of specific MHC:peptide ligands from APC would limit their availability to other T cells. Such Ag stripping from DC has been observed in vivo by Kedl et al. ( 26). They suggested the stripping would limit access of lower affinity T cells to Ag, generating a higher affinity response. It would also have the added benefit of preventing activation of cross-reactive T cells as well as increasing the diversity of epitopes recognized by a population of T cells in an immune response.
Once transferred to the T cell, the MHC:peptide would continue to interact with the TCR and might sustain signaling. This would allow signaling of sufficient duration and strength necessary to fully activate effector functions ( 7). This continued signaling may help to resolve the apparent paradox between the signaling duration necessary to fully activate T cells and the recent findings that during the initial phases of T cell activation in vivo, contacts with lymph node DC are of short duration ( 35). If T cells capture MHC:peptide and sustain signaling via autopresentation, a short duration encounter with APC could allow for full T cell activation.
The ability of T cells to present Ag to each other leading to anergy ( 25, 41) or CTL-mediated killing of the presenting T cell (fratricide) after MHC:peptide transfer is well documented ( 22). Thus, the peptide-specific, MHC transfer from the immunological synapse during T-APC dissociation may also be a method to shut off an ongoing immune response.
Ultimately, the transferred MHC:peptide molecules disappear from the surface of the T cell over time due to TCR down-modulation ( 20, 42). This may explain the complex relationship between TCR levels and GFP levels in Fig. 4 E. Such internalization and degradation would extinguish autopresentation, and, therefore, the sustained TCR signaling that could lead to exhaustion of the T cells, rendering them nonfunctional. Thus, TCR-mediated capture of specific MHC:peptide complexes from an APC could lead to sustained signaling to generate effector cells, prevent activation of lower affinity, potentially cross-reactive T cells, and could serve to limit or terminate an ongoing response by inducing anergy or fratricide.
We thank Aurelie Snyder and the Oregon Health & Science University Molecular Microbiology and Immunology Microscopy Core Facility for expert technical assistance, Stephanie Lathrop for technical assistance with flow cytometry, Stan Barter for technical assistance with transfections, and Marcie Hackbarth for constructing the MCC:I-Ek:GFP and MCC:I-Ek:FKBP plasmids. We also thank Timothy Thauland and Cortny Huddleston for critical review of the manuscript, and other colleagues for providing essential materials.
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.
This work was supported by Grant AI29544 from the National Institutes of Health and a grant from the Oregon Health Sciences Foundation.
Abbreviations used in this paper: SMAC, supramolecular activation complex; c-SMAC, central SMAC; DIC, differential interference contrast microscopy; FKBP, FK506-binding protein; Hb, murine hemoglobin peptide; MCC, moth cytochrome c peptide; MFI, mean fluorescence intensity; PKC, protein kinase C.
The on-line version of this article contains supplemental material.