Activation of CD4+ T cells by APCs occurs by multiple Ag recognition events including the exchange of costimulatory signals and cytokines. Additionally, the T cells acquire APC-derived surface molecules. Herein, we describe for the first time the transfer of human and murine T cell surface receptors to APCs after Ag-specific interaction. This transfer occurs in two qualitatively different phases. The first group of molecules (e.g., CD2) derived from the T cell surface was transferred rapidly after 2 h of interaction, was strongly bound on the DC surface (acid wash-resistant), was strictly dependent on dendritic cell-T cell contact, and transferred independently of T cell activation. The second group, including the CD3/TCR complex, CD27, and OX40, was of intracellular origin, transferred later after 10–16 h in a cell-cell contact-independent fashion, was noncovalently bound, and was strictly dependent on Ag-specific T cell activation. Functionally, murine dendritic cells that received TCR molecules from OVA-specific CD4+ T cells after Ag-specific interaction were less efficient in priming naive CD4+ T cells of the same specificity without losing their ability for CD8+ T cell stimulation, indicating that the transferred TCR molecules mask the Ag-bearing MHC II molecules, thereby reducing their accessibility to following Ag-specific CD4+ T cells. While the first group of transferred T cell surface molecules might facilitate the detachment of the CD4+ T cell from the dendritic cell during the early scanning phases, the second group could play an important immunomodulatory role in intraclonal competition of T cells for APC access, making the physical presence of CD4+ T cells unnecessary.
The adaptive immune response is initiated in secondary lymphoid organs by cognate physical interaction of APCs and Ag-specific CD4+ T cells. This Ag-specific activation of CD4+ T cells by APCs is a dynamic process including distinct phases of interactions (1) and it leads to conditioning (“licensing”) of APCs. During this cognate interaction the cells not only deliver signals via receptor-ligand interactions and cytokine production, they also exchange cell surface molecules (2, 3).
Surface molecule transfer has been observed for various cell types but has been predominantly reported for the transfer of dendritic cell (DC)3 surface molecules to T cells in a unidirectional way, mostly in the form of exosomes that contain MHC I/II-peptide molecules (4, 5, 6, 7, 8) and costimulatory molecules such as CD80/86 (8, 9, 10, 11, 12), ICAM-1 (9, 13), and OX40L (14). The documented molecule transfer after cell-cell interaction is dependent on cognate interaction and physical contact as well as the presence of the ligand of the observed transferred molecule, for example, the appropriate TCR for transferred MHC-peptide molecules (5, 9, 11).
The transfer kinetics and the fate of the exchanged molecules are variable and dependent on the cell type and activation status of the cells. APCs transfer MHC, ICAM-1, and CD80/86 to T cells in less than 1 h (5). However, in case of the MHC I-peptide transfer, the complex is lost on the cell surface to 50% after 4 h and is undetectable after 16 h (5). Presumably TCR internalization accounts for this rapid loss of the acquired molecules. In the case of CD80 transfer, the molecules disappear even faster from the T cell surface and are shuttled to endocytic compartments (9). APCs are known to direct absorbed ligands to endosomes to allow Ag processing and subsequent loading of MHC molecules. Using this internalization process they have been shown to acquire Ag from other cells by direct contact (15, 16) or in the form of tumor- or APC-derived exosomes (17, 18, 19). On the other hand, follicular DCs do not internalize exosomes but carry them on their surface for long time periods (20). To date, only one report describes the transfer of rat T cell molecules to B cells and DCs by exosomes (21). The transferred MHC II-peptide complexes from activated CD4+ T cells were recognized by allogenic responder T lymphocytes and caused them to proliferate. However, mouse CD4+ T cells do not up-regulate MHC II upon activation, and the influence of other T cell molecules on DC function involving their ability to prime following CD4+ T cells or CD8+ T cells has not been investigated. We now show that, after Ag-specific interaction, human and murine DCs can acquire T cell surface molecules, including the CD3/TCR complex. Furthermore, we demonstrate that the transfer from DC surface molecules to naive T cells during the priming process is conducted in two phases. The first group of molecules originates from the T cell surface and is transferred rapidly in a cell-cell contact-dependent manner, whereas the second group is transferred through the release of intracellularly formed surface molecules during the later priming phase. Finally, we investigated the accessibility of these licensed, T cell surface molecule-carrying DCs to naive CD4+ T cells. Our observations showed a reduced capacity of DCs that had received the TCR complex by cognate CD4+ T cell-DC interaction to stimulate naive CD4+ T cells without losing their ability for CD8+ T cell stimulation, indicating a role of the transferred CD4+ T cell molecules in the specific feedback regulation of the adaptive CD4+ T cell immune response.
Materials and Methods
Mice and reagents
Mice were maintained under specific pathogen-free conditions at the animal facilities of the House of Experimental Therapy and the Institute for Molecular Medicine in Bonn. Experiments were performed according to animal experimental ethics committee guidelines. C57BL/6 and BALB/c mice were purchased from Charles River Laboratories. OVA323–339 TCR transgenic OT-II (22) and DO11.10 (23) mice and OT-I OVA257–264 TCR transgenic mice (24) have been described previously and were bred in our animal facility.
Staphylococcal enterotoxin B superantigen (SEB) and OVA (grade II) were purchased from Sigma-Aldrich. The following human Abs were used: CD2-FITC, CD3-FITC, CD27-FITC, and CD28-FITC (ImmunoTools); OX40-FITC and CD69-FITC (eBioscience); HLA-DR-Alexa Fluor 647, CD94-FITC, CD62L-FITC, and CD158b-FITC (BioLegend); HLA-DR blocking (clone L243, kindly provided by Gerhard Moldenhauer, Heidelberg, Germany); CD2 blocking (clone MEM-63, AbD Serotec); annexin V (apoptosis detection kit, Roche); and annexin I and CD3ζ (BD Biosciences). Murine cells were stained with CD2-FITC (Serotec); CD11c-Alexa Fluor 647 (eBioscience); and CD3-PE, CD62L-PE, TCRβ5-PE, and TCRKJ1–26-PE (BD Biosciences). For secondary staining of biotinytated proteins, streptavidin-Alexa Fluor 647 (Invitrogen) was used.
Generation of human monocyte-derived DCs, human CD4+ T cells, and NK cells
Human PBMC were isolated from buffy coats obtained from healthy donors using LSM 1077 lymphocyte separation medium (PAA Laboratories). Monocytes were purified from PBMC using a human CD14 EasySep selection kit (StemCell Technologies), and DCs were generated by culturing monocytes in serum-free CellGro DC medium (CellGenix) containing antibiotics, GM-CSF (800 U/ml; ImmunoTools), and IL-4 (1000 U/ml; ImmunoTools) for 6 days. CD4+ T cells and NK cells were purified from buffy coats by positive selection using CD4 or CD56-biotin plus streptavidin-coupled magnetic beads (Miltenyi Biotec), respectively.
Purification of murine CD11c+ DCs and CD4+ T cells from spleen
For murine DC preparation, collagenase-digested spleen cells (collagenase A, Roche) were pressed through a sterile fine wire mesh, suspended in MACS buffer (PBS, 1% FCS, 2 mM EDTA), and separated using CD11c magnetic beads (Miltenyi Biotec). The resulting DCs were routinely ≥85–90% positive for CD11c as determined by flow cytometry. CD4+ T cells were purified by positive selection using CD4 magnetic beads (Miltenyi Biotec).
For transfer experiments, DCs and CD4+ T cells were incubated in RPMI 1640 medium (Invitrogen) containing GM-CSF (mouse) or GM-CSF plus IL-4 (human). T cell activation was achieved by adding SEB (1 μg/ml), OVA protein (10 μM), or OVA323–339 peptide (1 μM). For cell-cell contact-dependent transfer, 3.3 × 105 DCs were cocultured with 2.3 × 106 T cells (ratio of 1:7) in the upper compartment of 24-well transwell plates (membrane with 0.4-μm pore size; Nunc); for cell-cell contact-independent transfer, 5 × 105 DCs were cultured in the lower compartment. To obtain sufficient material for biochemical analysis of the transferred molecules by Western blotting, bulk cultures of 1 × 107 DCs were cocultured with 7 × 107 T cell blasts (ratio of 1:7) supplemented with 1 μg/ml SEB plus 20 ng/ml IL-12. T cell blasts were generated from freshly buffy coat-isolated CD4+ T cells that were stimulated with 5 μg/ml PHA (Sigma-Aldrich) for 48 h. Cells were then restimulated with 200 U/ml IL-2 (ImmunoTools) every 3 d for 14 d. Supernatants from DC/SEB-activated T cell blasts were collected after 24 h, purified from remaining cells and debris by centrifugation at 1500 rpm for 10 min, followed by another centrifugation step at 4000 rpm for 20 min, and subjected to ultracentrifugation followed by biochemical analysis via Western blotting.
NK cell-DC transfer experiments were conducted as described above, and NK cell activation was achieved by addition of 20 ng/ml IL-12 (ImmunoTools).
Generation of apoptotic and necrotic T cells
To induce apoptosis, CD4+ T cells were UV irradiated at 100,000 μJ/cm2 (254 nm UV) using a Ultraviolet Products crosslinker. After overnight culture, the cells were 52% apoptotic and 20% dead as verified by annexin V/Hoechst 33258 staining. For induction of necrosis, T cells were subjected to five freeze-thaw cycles. Thereafter, all cells were dead as verified by trypan blue staining.
Biotinylation of cell surface proteins
Cells were washed thoroughly with PBS and incubated with 200 μM N-hydroxysuccinimido-biotin (NHS-x-Biotin, Interchim) in PBS for 15 min. The reaction was stopped by addition of an equal volume of FCS for 1 min and the cells were washed thoroughly thereafter.
Cells were washed twice in PBS and resuspended for 4 min at 20°C in citrate buffer (0.133 M citric acid and 0.066 M Na2HPO4 (pH 3.3)) at a density of 5 × 106 cells/ml. The treatment was stopped by the addition of an excess of 5% FCS in PBS. After being washed, the cells were stained for flow cytometry.
For better identification, cells were labeled with CellTracker orange CMTMR ((5-(and-6)-(((4-chloromethyl)benzoyl)amino)tetramethyl-rhodamine)) or green CMFDA (5-chloromethylfluorescein diacetate) (Invitrogen) as suggested by the manufacturer. Cells were washed with MACS buffer after coculture, stained with designated Abs, and fixed with 4% (w/v) paraformaldehyde in PBS for 20 min. After a final wash, cells were mounted onto slides with Gel-Mount mounting medium (BioMeda) containing 50 mg/ml DABCO (1,4 diazobicyclo-(2.2.2) octane; Sigma-Aldrich) and samples were examined by laser scanning confocal microscopy. Confocal data were acquired and processed with an Olympus FluoView1000 confocal microscope performing two-dimensional reconstructions of Z-stacks (mean width 0.9 μm/slice) and scanning with 40 μs/pixel using Kalman mode. Images were taken with a PLAPON60×O/TIRFM-SP (numerical aperture 1.45) oil immersion lens at room temperature. Reconstructions were made with the Olympus FluoView software. All images were collected in a sequential mode to minimize the crossover between channels, and microscope parameters for the collection of images were kept constant between samples.
Single-cell suspensions were stained with designated Abs in MACS buffer for 10 min at room temperature, washed once with MACS buffer, and immediately processed for measurement. Appropriate isotype controls were always included. Dead cells were excluded from the analysis by Hoechst 33258 incorporation. DC-T cell aggregates were electronically gated out by large forward scatter width exclusion (digital acquisition) and strong streptavidin staining. Cells were measured on a LSR II fluorocytometer (BD Biosciences) and data were analyzed using the FlowJo software (Tree Star).
Before Western blotting, cell culture supernatants were fractionated by ultracentrifugation at 100,000 × g for 2.5 h. Thereafter, supernatants were TCA-precipitated, adding 10% volume of 100% TCA. Precipitated pellets were washed twice with ether, air dried, and redissolved in SDS sample buffer. Samples were separated by SDS-PAGE using 10% polyacrylamide gels and transferred to an Immobilon membrane using semidry blotting. After blocking for 30 min with 5% skim milk in TBS, the blots were developed with the respective primary Ab overnight followed by peroxidase-conjugated secondary Ab and ECL detection.
Ag-specific T cell activation
Murine DCs from C57BL/6 mice were incubated with CD4+ T cells from OT-II mice at a ratio of DC-to-T cells of 1:4 for 24 h with 10 μM OVA protein. The cells were washed thereafter, stained with CD11c Ab, and separated from the coculture by cell sorting using a FACSVantage (BD Biosciences). Dead cells were excluded from the analysis by Hoechst 33258 incorporation. DC-T cell aggregates were electronically gated out by large forward scatter width exclusion (digital acquisition). DC purity was consistently >95% as confirmed by flow cytometry. OT-II, DO11.10, and OT-I cells were purified by magnetic bead separation and labeled with CFSE. Sorted DCs and isolated T cells were cultured at a 1:1 cell ratio in 96 wells at 5 × 104 cells/ml for 24 h (cytokine secretion) or 72 h (proliferation). Quantitative ELISA for IL-2 were performed using commercially available Abs (BD Biosciences) according to the manufacturer’s recommendations. Proliferation was assessed by visualization of CFSE dye dilution upon cell division using flow cytometry.
Human CD4+ T cells transfer surface molecules to DCs during cognate interaction
The transfer of DC-derived surface molecules to T cells has been extensively investigated. Here we asked the question whether the same occurs in a reverse way, that is, from T cells to DCs. To show surface protein transfer and exclude endogenous up-regulation of surface molecules by the DCs, all T cell surface receptors were biotinylated. After coculture of these labeled CD4+ T cells with DCs, biotinylated proteins were detected by streptavidin staining. Twenty-four hours after coculture of biotinylated T cells and DCs in the presence of SEB, these biotinylated surface molecules could be detected on the DC cell membrane (Fig. 1,A). They colocalize with the T cell surface molecule CD2, a receptor generally not expressed by DCs (Fig. 1 B).
Transfer of various T cell surface receptors to DCs
To obtain more detailed information on which other T cell surface molecules besides CD2 were transferred, the DCs were stained for various T cell surface Ags after coculture with biotinylated T cells in the presence of SEB. Results showed a great number of T cell surface Ags on the DC surface, including CD2, CD3, CD27, OX40, CD69, CD5, and PD-1, whereas others (CD28, CD45, CD11a, CD29, CD7, CD4) were not transferred (selected markers shown in Fig. 2,A). The fact that only particular T cell surface proteins were transferred indicates the measured transfer is not due to culture artifacts (e.g., bound cell debris). Similar results were obtained with murine OVA-specific CD4+ T cells after unspecific stimulation by superantigen-induced TCR crosslinking to MHC II on DCs or Ag-specific stimulation with OVA peptide-pulsed DCs (Fig. 2 B). In addition to the surface molecules transferred from human T cells to DCs, the specific TCR of transgenic T cells was detected on the DC surface. The amount of transferred molecules achieved by superantigen stimulation was similar to that observed by Ag-specific activation. These findings strongly suggest that the transfer observed in the human system is not due to artifacts caused by SEB stimulation but are rather reflecting a process conducted by Ag-specific activation of CD4+ T cells. Similar to the human system OX40, CD5 and CD27 were transferred by murine CD4+ T cells whereas CD28 was absent from the DC surface (data not shown).
Transfer of CD4+ surface proteins to DCs by two different mechanisms
To investigate whether the transferred molecules were only cell-cell contact-dependently or also independently transferred, we employed a transwell system containing an upper well with biotinylated CD4+ T cells and DCs with SEB separated from DCs alone in the lower well by a cell-impermeable membrane (0.4 μm). T cell activation was mediated by SEB. Surprisingly, some of the analyzed molecules such as CD3, CD27, or OX40 did not require DC-T cell contact for transfer: They were transferred cell-cell contact-independently in the transwell system (Fig. 3 A). In contrast, CD2 is transferred only through cell-cell contact. Again, CD28 was neither transferred cell-cell contact-dependently or contact-independently. The cell-cell contact-dependently transferred CD4+ T cell surface molecules in the upper well originate from T cell molecules present on the cell surface since a great amount of biotin was transferred onto the DCs. Acquisition by trogocytosis or ectodomain shedding are possible transfer mechanisms. In contrast, the cell-cell contact-independently transferred CD4+ T cell surface molecules in the lower well are not of cell surface origin since biotin was absent on the surface of DCs, which indicates the secretion of intracellular assembled molecules rather than shedding or nibbling of biotinylated surface molecules. Since is has been reported that activated CD4+ T cells secrete exosomes bearing the CD3/TCR complex (25), we assumed that this transfer of T cell surface molecules might be mediated by T cell-secreted exosomes, that is, vesicles that are formed intracellularly and subsequently exocytosed. For the reverse transfer of DC surface molecules to other cells, exosomes are known to be important vehicles (17, 26).
Intracellularly derived but not cell surface-derived T cell molecules are dissociated by low pH
Noncovalent protein-protein interactions like association of β2-microglobulin with the MHC I H chain or Ab binding can be dissociated by acid wash treatment. Therefore, we tested the sensitivity of the transferred CD4+ T cell surface molecules to this low pH treatment to obtain information about their binding nature. Interestingly, surface proteins that were intracellularly synthesized after the initial biotinylation and transferred in the transwell system such as CD3, CD27, and OX40 could be removed by lowering the pH to 3.3, whereas the T cell surface-derived, strictly cell-cell contact-dependently transferred CD4+ T cell surface molecule CD2 remained bound to the DC surface (Fig. 3 B). Thus, transferred T cell surface proteins that were cell surface-derived are considerably more stably bound than were intracellularly derived molecules.
Transfer of T cell molecules is receptor-mediated
To investigate whether the cell surface-derived CD4+ T cell surface molecules such as CD2 and/or the intracellularly derived molecules such as CD3 or CD27 are transferred by specific receptor-ligand interaction or randomly by unspecific membrane transfer, we used blocking Abs to their receptors on the DC surface (Fig. 3,C). Interaction of CD2 with CD58 was blocked by an anti-CD58 Ab, and interaction of the CD3/TCR complex with its MHC II counterpart was inhibited by an anti-HLA-DR Ab. To prevent inhibition of T cell activation by anti-HLA-DR before intracellularly derived T cell molecule transfer, T cells were activated by DCs and SEB in the absence of anti-HLA-DR for 24 h and fresh DCs were incubated subsequently with the resulting cell-free supernatants in the presence/absence of anti-HLA-DR. Both Abs inhibited the T cell surface molecule transfer to DCs in a dose-dependent manner (Fig. 3 C), indicating that both surface-derived (i.e., CD2) and intracellularly derived (i.e., CD3) CD4+ T cell molecule transfer is receptor-mediated.
Transfer kinetics: rapid transfer of cell surface-derived molecules; transfer of intracellularly derived molecules only after later T cell activation
The two observed transfer mechanisms do not only differ in their origin of the transferred molecules and their binding stability but also in their transfer kinetics. Whereas cell surface-derived molecules like CD2 were already transferred within 2 h, intracellularly synthesized proteins like CD3, CD27, or OX40 were detected not until 10–16 h after T cell activation on the DC surface (Fig. 4 A). These findings further support our findings that T cell surface molecule transfer to DCs occurs via two entirely different mechanisms that might also differ in their biological function.
To assign a biological function to the molecules transferred onto DCs, it is important to know how long they stay on the DC surface even after the T cell has left. To this end, we cultured DCs in a transwell system containing DCs and CD4+ T cells in the upper well and DCs alone in the lower well, leading to cell surface-derived molecule transfer of T cell surface molecules to DCs in the upper well and intracellularly derived molecule transfer in the lower well during a 24-h culture time. Thereafter, the upper well containing the activated T cells was removed, and the DCs in the lower well were thoroughly washed and cultured for an additional 18 h. After this additional 18-h culture, the intracellularly derived molecules CD3, CD27, and OX40 could still be detected on the DC surface, indicating they are long-lived and are not internalized (Fig. 4 B).
Transfer of intracellularly derived surface proteins to DCs requires activation of the transferring cell
CD4+ T cells only transferred the intracellularly derived T cell surface molecules CD3 or CD27 onto DCs after superantigen-induced activation, whereas resting T cells did not transfer significant amounts (Fig. 5 A). However, transfer of the T cell surface-derived molecule CD2 was only marginally affected by the activation status of the CD4+ T cells.
Since DCs also interact with NK cells in the periphery, we investigated whether, as for CD4+ T cells, NK cells also transfer cell surface molecules to DCs. We employed the same transfer coculture system as described for CD4+ T cells containing NK cell-DC coculture in the upper well and DCs alone in the lower well. Similar to CD4+ T cells, which require TCR triggering to induce surface molecule transfer of the intracellularly derived T cell surface molecules, we observed surface molecule transfer of several NK cell surface receptors to DCs in the lower well only when NK cells were activated by IL-12 for 24 h (Fig. 5,B, lower panels). Resting NK cells did not transfer any of the examined surface molecules to DC (Fig. 5 B, upper panels). Transferred molecules included CD27, CD158b, and CD62L, whereas CD94 and CD2 where not transferred.
Thus, activation of the transferring cell seems to be crucial for the transfer of the intracellularly derived surface molecule, whereas surface-derived surface molecules are transferred independently of the activation status of the transferring cell.
To rigorously exclude the transfer of T cell molecules originating from apoptotic bodies or necrotic cell debris, DC-T cell coculture was established using apoptotic/necrotic T cells (Fig. 5 C). Neither apoptotic nor necrotic CD4+ T cells transferred T cell surface molecules onto DCs: cell surface-derived CD2 as well as intracellularly derived CD3 and CD27 molecules were undetectable on the DC surface after coculture, indicating that the T cells have to be intact to mediate molecule transfer.
CD3/TCR molecules are distributed over the entire DC surface
Because we observed a transfer of T cell-derived CD3/TCR complexes (Fig. 2) onto murine DCs, we expected the complexes to be bound Ag-specifically to their natural ligands (e.g., MHC II molecules). The strong fluorescence of CD3/TCR molecules on the DC surface suggests that most MHC II molecules are covered by these complexes. This observation was confirmed in the human, superantigen-induced transfer model. CD3 molecules were distributed over the entire DC surface to the same extent as were surface-localized HLA-DR molecules (Fig. 6 A).
To investigate whether the T cell-derived CD3 molecules were transferred to the DC surface as single molecules or in the context of intracellularly released vesicles/exosomes, we subjected the cell- and debris-depleted supernatants of DC/SEB-activated CD4+ T cells to a 100,000 × g ultracentrifugation step. Biochemical analysis of the pellets and supernatants after ultracentrifugation revealed large amounts of CD3 molecules in the pellet fraction, whereas CD3 was undetectable from pellet-depleted supernatants (Fig. 6 B). This indicates that the transferred CD3 molecules are released in context of larger molecular mass structures such as exosomes, as single molecules cannot be pelleted at 100,000 × g. Another protein present in the pelleted fraction was annexin 1, a prominent marker present on exosomes of various cellular sources (17, 27).
Although one cannot exactly colocalize single CD3/TCR molecules to surface HLA-DR molecules due to resolution limitations of light microscopy, the distribution pattern of both molecules is the same, suggesting that all MHC II molecules were covered by CD3/TCR complexes. If the majority, if not even all, of the MHC II-peptide molecules are covered by Ag-specific CD3/TCR complexes, this could then hamper the Ag-presenting capacity of the covered DCs. To test this hypothesis, the Ag-presenting function of TCR-covered murine DCs to OVA-specific CD4+ T cells was investigated in vitro.
Reduced activation of Ag-specific CD4+ T cells by TCR-covered DCs
CD4+ T cells of the same Ag specificity are reported to compete for access to the peptide-MHC II complex on the APC. It is thought that sterical hindrance or Ag extraction by previous T cells is the reason for this intraclonal competition (28, 29). Since CD3 molecules from activated T cells were distributed over the entire DC surface and appeared to be associated with HLA-DR (Fig. 6), we hypothesized that transferred surface molecules from CD4+ T cells might mask MHC II-peptide complexes and thereby affect the ability of the DCs to subsequently stimulate CD4+ T cells of the same specificity (intraclonal competition/feedback regulation). To test this hypothesis, we incubated DCs with OVA-specific OT-II T cells in the presence of OVA protein. After 24 h a great amount of transgenic TCR molecules from OT-II T cells could be detected on the DC surface (Fig. 7 B). Thereafter, the DCs were sorted to high purity (>98%) to exclude sterical hindrance by bound T cells and then incubated with fresh naive CD4+ T cells of the same specificity or OVA-specific CD8+ T cells.
Notably, DCs that had received CD3/TCR complexes through prior Ag-specific contact with OT-II T cells induced less proliferation (Fig. 7,C, left) and considerably less cytokine production (OT-II, 4-fold less; DO11.10, 2.5-fold less; Fig. 7,D) of naive OVA-specific OT-II/DO11.10 CD4+ T cells. In contrast, fresh and “used” TCR-covered DCs showed similar capacity to stimulate naive OVA-specific OT-I CD8+ T cells (Fig. 7 D). This indicates that after cognate interaction of DCs and CD4+ T cells, TCR molecules were only transferred to MHC II molecules, blocking CD4+ T cell access without affecting access of CD8+ T cells.
To provide further evidence that the lack of CD4+ T cell stimulation was due to transferred TCR molecules and not due to a functional defect of the TCR-covered DCs, OVA323–339 peptide was added to the sorted DCs in excess before coculture with naive OVA-specific CD4+ T cells. In this setup, all available MHC II molecules were loaded with the high-affinity OVA peptide (Fig. 7,A). TCR-covered DCs induced similar proliferation (Fig. 7,C, right) and cytokine production (Fig. 7 D, right) of naive OVA323–339-specific CD4+ T cells when they were re-pulsed with OVA323–339 peptide. The capacity of OVA peptide-loaded TCR-covered DCs to stimulate OVA-specific naive CD4+ T cells, as well as their ability to stimulate OVA-specific naive CD8+ T cells in a comparable amount to that induced by DCs that were Ag-loaded in the absence of CD4+ T cells, excludes a functional defect of these TCR-covered DCs.
We herein show that CD4+ T cells transfer particular T cell surface molecules including the CD3/TCR complex onto DCs after cognate interaction. This transfer had functional implications for the immune function of the receiving DCs. Specific binding of CD3/TCR complexes to peptide-loaded MHC II molecules prevented further stimulation of CD4+ T cells of the same specificity but allowed activation of OVA-specific CD8+ T cells. These findings suggest that transfer of T cell surface molecules regulates further DC-T cell interactions.
We observed the transfer of two groups of T cell surface molecules after the interaction of DCs with CD4+ T cells. The first group of molecules (i.e., CD2) was derived from the T cell surface, was only transferred upon cell-cell contact, was transferred rapidly after 2 h of interaction, was relatively stably bound on the DC surface (possibly membrane-anchored), and was transferred independently of the T cell activation state; the other group of molecules, including the CD3/TCR complex, CD27, and OX40, was of intracellular origin, was transferred during the later activation phase after 10–16 h, was easily removable by acidic treatment, and was strictly dependent on Ag-specific T cell activation. This “two-class system” has not been reported to date.
Surface molecule exchange between cells can be conducted by trogocytosis, exosomes, or receptor shedding. Trogocytosis is a process that describes the active uptake of membrane fragments by interacting cells (30, 31). Exosomes are small vesicles that are secreted by fusion of multivesicular endosomal bodies with the plasma membrane (17, 26). Receptor shedding or enzymatic cleavage is primarily mediated by membrane-bound metalloproteinases (32). Publications about molecule transfer between various cell types describe only one particular type out of these described transfer mechanisms. The given information (e.g., the time that was required for the transfer from one cell to the other) might reflect the type of transfer observed by the investigators. The rapid transfer of HLA-C from target cells to NK cells is described to be an active process of the NK cell (33, 34). Furthermore, phosphoantigen recognition-initiated transfer of γδ T cell surface molecules after short contact with Daudi target cells is dependent on γδ T cell activation (11). Thus, these observations describe a rapid, active process of the receiving cell that points to a mechanism described as “nibbling” or trogocytosis (30). Because of its rapid nature, our observed transfer of surface-derived T cell molecules might be conducted by trogocytosis. However, this step might be accompanied by receptor shedding from the T cell surface. Since only live T cells transfer CD2 molecules onto the DC surface, but not apoptotic or dead cells (Fig. 5,C), an active mechanism on the T cell side might be involved. The colocalization of biotinylated T cell surface proteins with CD2 (Fig. 1 B) implies a common mechanism for the transfer of these biotinylated molecules and CD2. Interestingly, both biotinylated and CD2 molecules display a dotted distribution on the DC surface. This may indicate that they are concentrated in lipid rafts in the DC membrane, as it has been shown for transferred MHC II molecules (35).
Using the transwell system, our observation that considerably more biotin was transferred by direct T cell-DC contact in the upper well in comparison to that from this coculture separated DCs in the lower well sheds light on the origin of the transferred molecules. Since biotinylated surface molecules were almost absent on DCs in the lower well, they must have originated from an intracellular T cell compartment. This might point to intracellularly formed molecules or vesicles by the activated T cells. Exosomes are originating from endosomes and contain intracellular proteins that would not be labeled by surface biotinylation. Consequently, there would be much less biotinylated protein transfer (as opposed to CD3, CD27, or OX40) to DCs by the cell-cell independent mechanism through the transwell membrane, which is supported by our observations. The fact that the intracellular ζ domain of CD3 was present in the 100,000 × g fraction of DC/SEB-activated T cell supernatants (Fig. 6 B) further suggests that CD3 was bound as a complete molecule in the context of a structure of larger molecular mass rather than as a single ectodomain molecule from vesicle surfaces or cellular membranes. Together with the published data of exosome secretion by activated CD4+ T cells bearing the TCR/CD3 complex (25), our data strongly suggest that transfer of intracellularly derived T cell surface molecules is mediated by exosomes. Notably, this transfer is only occurring if the T cells are activated; resting T cells do not release exosomes (25), thereby depending on the activation status of the transferring cell and not, as reported for trogocytosis, on the status of the receiving cell. In summary, we conclude that the rapidly surface-derived molecules might be transferred by trogocytosis, whereas the slower transfer of intracellularly derived molecules is most likely mediated by exosomes. However, identification of the exact cellular mechanisms underlying the transfer of these two sets of CD4+ T cell surface molecules is currently under investigation.
Before entering secondary lymphoid organs, DCs can interact in the periphery with NK cells. Activated NK cells secrete great amounts of IFN-γ and they are thought to play an activating role upon interaction with DCs (36, 37). Our observation that activated NK cells leave some of their surface molecule behind after interaction with DCs may indicate a general mechanism for cells to change the phenotype and possibly the function of the cells they are leaving. NK cells might simply just mark the DCs, as it has been speculated for transferred NK cell molecules to target cells (38), or they may provide further signaling.
For CD4+ T cells the two observed processes of surface molecule transfer might have essential functions during the Ag-specific activation of T cells: 1) the rapidly transferred, surface-derived molecules would be helpful for disengagement of DCs and T cells, and 2) during the later activation phase, the intracellularly derived molecules might be important for further immunomodulation. Our observations that transferred molecules like CD3 are bound to the DC surface for a long time (at least 18 h) strengthen the hypothesis that these transferred molecules indeed might have another function than just facilitating the dissociation of the two interacting cells. This seems to be different to APC-derived surface molecules, which are rapidly internalized by CD8+ T cells (2) or DCs that take up allogenic DC-derived exosomes and internalize them rapidly to present the alloantigen (18). NK cells are also reported to rapidly internalize target cell HLA-C molecules (33).
Our findings contribute an additional important aspect to the current understanding of the process of CD4+ T cell priming. CD4+ T cell activation in vivo is initiated by a scanning phase that leads to transient interactions, followed by formation of stable DC-T cell clusters (1). The detachment of the CD4+ T cell from the DC during the scanning phase could be facilitated by the transfer of surface-derived T cell surface molecules, thus facilitating maximal scanning events. Interestingly, after formation of stable DC-T cell clusters the T cells have been described to regain motility and “swarm” around DCs before they finally proliferate and exit the secondary lymphoid organ to the periphery (1). During this stage, T cells are visibly enlarged and migrate slowly among DCs in the local region with which they make dynamic serial interactions. During this stage they may deposit intracellularly formed T cell surface molecules on the DC surface.
We herein show that T cells modulate DC function in vitro after Ag-specific interaction by transferring T cell surface molecules such as costimulatory molecules and the TCR/CD3 complex to DCs. These T cell surface molecules might mask the corresponding MHC II-peptide molecules, leading to a diminished access of naive CD4+ T cells of the same specificity to these modulated DCs and leaving the cell still accessible to Ag-specific CD8+ T cells. This may play an important role in intraclonal competition and contraction of the Ag-specific immune response. CD4+ T cells have been described to compete with other T cells of the same specificity for access to DCs (28, 29). Since these deposited molecules comprise the CD3/TCR complex, they might mask peptide-MHC II molecules and would thereby play an important role in intraclonal competition. Our observations strengthen this hypothesis since DCs that had acquired TCR molecules by Ag-specific interaction with CD4+ T cells had reduced capacity to activate naive CD4+ T cells of the same specificity even when the T cells from the first interaction had been sorted out (and were therefore not responsible for physical sterical hindrance). Furthermore, a diminished amount of MHC molecules detected on the DC surface after T cell interaction (9, 39) may be explained by MHC masking through transferred T cell surface molecules. The DC being the limiting factor during the priming of CD4+ T cells of one specificity would be particularly meaningful in the dynamic model of Ag-specific T cell activation where the DC is thought to first interact with CD4+ T cells, which leads to DC conditioning and subsequently recruits CD8+ T cells (40, 41, 42). The CD4+ T cell would be predominantly important for the DC conditioning to put it into a state ready for CD8+ T cell activation. An exceeding activation of too many CD4+ T cells, however, would not make sense. This could be achieved by masking MHC II molecules with T cell-derived CD3/TCR molecules rather than, as presumed so far, the physical presence of CD4+ T cells (28).
We acknowledge the assistance of E. Endl, A. Dolf, and P. Wurst from the Flow Cytometry Core Facility at the Institute of Molecular Medicine and Experimental Immunology, University of Bonn.
The authors have no financial conflict of interest.
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 a Lise-Meitner Stipend to A.B. and SFB704 to A.L.
A.B. designed and performed the research, analyzed the data, and wrote the paper. A.L. coordinated and designed the research and revised the paper. T.Q. provided substantial help for confocal microscopy experiments and confocal data analysis. W.H. is the head of the Life and Medical Sciences (LIMES) Institute where all confocal microscopy experiments were performed, and he thereby provided essential equipment; he further coordinated the research and revised the paper. S.K. designed and performed substantial research for the characterization of the intracellularly mediated transfer. P.A. coordinated the research for the intracellularly mediated transfer and revised the paper. P.K. is the head of the Institute of Molecular Medicine and Experimental Immunology where all experiments except confocal microscopy experiments were performed, and he thereby provided essential equipment and facilities; he further coordinated the research and revised the paper.
Abbreviations used in this paper: DC, dendritic cell; SEB, staphylococcal enterotoxin B superantigen.