Dendritic Cells Acquire Antigens from Live Cells for Cross-Presentation to CTL¹

Dendritic cells (DC)³ are sentinels of the immune system that function by surveying peripheral tissues for Ag (1). To facilitate this function, DC acquire Ag from a variety of sources. DC can uptake Ag released from cells undergoing apoptosis or necrosis for presentation to MHC class I-restricted CTL (2, 3, 4, 5, 6, 7, 8). Similarly, DC acquire soluble proteins by macropinocytosis for presentation in the MHC class I pathway (9, 10). It has recently been demonstrated that DC acquire Ag from other DC for presentation in MHC class II molecules, a process also thought to be mediated by apoptosis of donor cells (11). These Ag acquisition processes have in common an initiating event involving cell death, which is thought to provide danger signals for DC activation and the subsequent stimulation of CTL (6).

Several reports have indicated that physiologic interactions between hemopoietic cells in the absence of cell death can result in transfer of membrane components between cells. Passive transfer of MHC molecules between allogeneic murine T cell clones has been reported (12), and the internalization by T cells of MHC molecules derived from the surface of APC has been demonstrated (13, 14). DC are highly interactive cells with extensive membrane processes that facilitate cell clustering and interaction with other cells, including T cells and B cells. We reasoned that during such interactions DC might acquire Ag in the absence of apoptosis or necrosis of the donor cell. Ag transfer between live DC has been suggested after the observation that both Ag-naive and Ag-exposed DC are important in T cell proliferation, although transfer was not directly demonstrated (15). Here, we used flow cytometry and live cell imaging techniques to visualize interactions of DC with other cells in a well-characterized monkey model (16). We demonstrate that DC but not macrophages have a remarkable capacity to acquire plasma membrane and intracellular proteins from other live APC and lymphocytes, and that such transfer results in cross-presentation of Ag to MHC class I-restricted CTL.

Immature DC were generated by culturing normal monkey CD14⁺ monocytes in GM-CSF and IL-4 for 4 days, with an additional 48-h exposure to CD40L (generously provided by Elaine Thomas, Immunex, Seattle, WA) to induce maturation, as described previously (16). Human DC from HLA-typed donors were generated in the same manner. Macrophages were generated by culturing monkey CD14⁺ monocytes for 24 h in the absence of cytokines. T cells used in the transfer experiments were monocyte-depleted from monkey PBMC and cultured in the presence of 5 μg/ml Con A and 20 U/ml IL-2 for 3 days to generate activated T cells, or for an additional 5 days to generate resting T cells. The EBV-transformed human HLA-A2⁺ B cell line Croft was generously provided by Olivera Finn (University of Pittsburgh, Pittsburgh, PA). The oligoclonal T cell line TIL620, which recognizes two epitopes of the melanoma Ag gp100 (17), was derived in the laboratory of Dr. Steven Rosenberg (National Institutes of Health, Bethesda, MD).

Cells were incubated with the lipophilic probe 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodi-carbocyanine perchlorate (DiD; excitation/emission spectra = 644 nm/663 nm) as described (18) and the thiol-reactive chloromethyl probe 5-chloromethylfluorescein diacetate (CMFDA; 492 nm/516 nm; both from Molecular Probes, Eugene, OR) at a concentration of 0.5μg/ml in 5 mM EDTA for 30 min at 37°C. Any dead cells were removed by centrifugation through sodium diatrizoate and Ficoll (Sigma, St. Louis, MO).

Unlabeled DC, macrophages, B cells, or T cells with or without pretreatment with 10 μg/ml cytochalasin D were cocultured with labeled DC at various ratios at 4 or 37°C in 24- or 48-well plates at a concentration of 1 × 10⁵ cells/ml. For transwell studies, labeled DC were added to 0.4-μm pore transwell chambers (Millipore, Bedford, MA) inserted into wells containing unlabeled DC. Cells were harvested at various intervals and maintained on ice until analyzed on a FACSCaliber flow cytometer (Becton Dickinson, Mountain View, CA) by using CellQuest software (Becton Dickinson). All analyses were done by using log₁₀ fluorescence. In some experiments, cells were fixed in 2% paraformaldehyde before analysis by flow cytometry. We determined in preliminary studies that fixation did not lead to leakage of dye from DiD- and CMFDA-labeled cells (data not shown).

DC were labeled with DiD and incubated with unlabeled DC for various time periods. Apoptosis was measured by using the TUNEL assay (Boehringer Mannheim, Indianapolis, IN) and staining with Alexa Fluor 488 annexin V (Molecular Probes) as per manufacturer’s instructions. As a positive control for apoptosis, T cells were treated with a UVB light source calibrated at 0.6 mJ/cm²/sec for 8 min and analyzed 12–16 h later.

DC labeled with DiD were introduced into a Bioptechs FCS2 (Butler, PA) closed microscope chamber maintained at 37°C and allowed to adhere. CMFDA-labeled DC were then added in medium containing Hoescht 33342 to label nuclei. Cells were imaged with a Zeiss Axiovert 135 microscope (Oberkochen, Germany) equipped with an XYZ stage, dual excitation filter wheels (Ludl, Hawthorne, NY), shuttered xenon and halogen light sources, and an Orca 12-bit cooled charge-coupled device camera (Hamamatsu, Tokyo, Japan). Images from five random fields were collected every 4 min for each of the illumination conditions with Metamorph software (Universal Imaging, West Chester, PA). The time-lapse sequences were integrated into movies and observed. Cell-cell interactions were tagged and tracked by particle tracking methods within the Metamorph software.

Recombinant adenovirus encoding gp100 (Ad-gp100) was generated by cloning a SalI-NotI fragment containing full-length human gp100 (generously provided by Dr. Stephan Wagner, University of Essen, Essen, Germany) into the pAdlox shuttle vector, with subsequent cotransfection with Ψ5 helper virus into the CRE8 packaging cell line as described previously (19). DC or macrophages were transduced with Ad-gp100 at a multiplicity of infection of 100 by adding virus directly to cells in culture as described (16). Cells were washed extensively at 24 h after infection to remove free virus and cultured for an additional 24–36 h to allow for protein expression. We routinely achieve 90% transduction efficiency using this method (16).

TIL620 cells (3 × 10⁴) were cultured with the same number of Ad-gp100-transduced immature monkey DC or macrophages with and without immature human DC or macrophages on Multiscreen-HA plates (Millipore) coated with anti-human IFN-γ Ab (1-D1K; MABTECH, Stockholm, Sweden). After 18 h of incubation, IFN-γ production was detected by labeling with a second anti-human IFN-γ Ab (7-B6-1; MABTECH) as described previously (19). Spots were enumerated by using a dissecting microscope.

Our initial studies focused on interactions between immature DC, as these cells are actively phagocytic and function in Ag capture (16, 20). DC were dual-labeled with DiD and CMFDA to track movement of plasma membrane and intracellular proteins, respectively, and cocultured with unlabeled cells for various time periods before flow cytometric analysis. To establish the validity of the fluorescence transfer system, we did a number of control experiments. When labeled DC were cultured for 4 h with unlabeled DC pretreated with cytochalasin D, an inhibitor of membrane ruffling, transfer of fluorescence was substantially inhibited. Similarly, unlabeled and labeled DC populations remained separate when cultured for 4 h at 4°C (Fig. 1,a). In addition, coculture of labeled DC with resting T cells for 4 h at 37°C did not result in fluorescence transfer between cell populations (Fig. 1,a). These experiments indicate that fluorescent labels remain cell associated and do not passively transfer between cells. In contrast to the control experiments, when labeled and unlabeled DC were cultured together at 37°C in the absence of phagocytic inhibitors transfer of DiD-labeled plasma membrane occurred rapidly, as by 1 h of coculture, 52% of unlabeled cells were DID⁺, and by 4 h, up to 95% were DiD⁺ (Fig. 1,b). Transfer of CMFDA occurred in cells that were DiD^bright, indicating that movement of plasma membrane was accompanied by movement of cytoplasmic protein to unlabeled DC. Labeled DC were viable at all time points of the experiment as measured by lack of staining using both the TUNEL assay and annexin V binding (Fig. 1 b), indicating that transfer of membrane and cytoplasmic proteins between cells was not the result of uptake of apoptotic material.

To investigate the mechanisms of transfer between DC we used a transwell system. Separation of labeled and unlabeled DC by a 0.4-μm transwell membrane completely abrogated transfer of label when cocultured for 4 h at 37°C (Fig. 2,a). This indicates that direct cell-cell contact is required and reinforces the earlier finding that passive movement of dye does not contribute to transfer of fluorescence between cells. In addition, by using different ratios of labeled and unlabeled DC, we demonstrate that transfer of plasma membrane and intracellular proteins between DC is an extremely efficient process, as 43% of unlabeled DC acquired fluorescence within 4 h when mixed with labeled DC at a 4:1 ratio (Fig. 2 b). Together, these data indicate that DC can readily transfer plasma membrane and, to a lesser extent, cytoplasmic proteins to other live DC.

To determine whether the capacity to acquire labeled plasma membrane and intracellular proteins from live cells was unique to immature DC, we performed similar studies with different DC populations as well as macrophages, which are highly effective at acquiring cellular Ag via uptake of apoptotic material (4). We first compared immature DC with DC that had been treated with CD40L to induce maturation, which is known to down-regulate phagocytic function (16, 21). Immature DC were able to acquire plasma membrane and intracellular labels with similar efficiency irrespective of the maturation state of the labeled population. When DC were matured with CD40L, the capacity to acquire labeled membrane and intracellular proteins from either immature or mature DC was suppressed by ∼50%, indicating that acquisition was primarily a function of immature DC (Fig. 3,a). The process of transfer did not result in maturation of DC, as coculture of unlabeled immature DC with DiD- and CMFDA-labeled DC for 6 h did not result up-regulation of CD86 or expression of the maturation marker CD83 (data not shown). Expression of CD83 is induced on immature DC after exposure to CD40L for the same time period (data not shown). In contrast to DC, macrophages were extremely limited in the capacity to acquire labeled plasma membrane and intracellular proteins from either live macrophages or immature DC (Fig. 3,b). Conversely, immature DC readily acquired Ag from labeled macrophages (Fig. 3 b).

To determine whether cells other than APC could serve as donors, we cocultured immature DC with DiD- and CMFDA-labeled activated T cells and B cells. Although neither of these cell populations was able to acquire label from DC, immature DC readily acquired labeled plasma membrane and cytoplasmic proteins from both activated T cells and B cells (Fig. 3 c). Similar findings were generated when immature DC were cultured with labeled resting monkey T cells, the Jurkat T cell line, and the 174XCEM.T1 T-B cell hybrid (data not shown). These experiments indicate that DC are unique in the capacity to acquire cellular components from other APC and from lymphocytes.

To study the dynamics of transfer between individual DC, we used live cell fluorescence microscopy, which allows real-time imaging of cellular interactions in culture. DC labeled with DiD were seeded into the microscope chamber and an equal number of CMFDA-labeled DC then were added (Fig. 4, 0 min). Detection of double-labeled cells over time signaled transfer of cellular components between cells. DC were highly motile and frequently moved in and out of small clusters, while rapidly extending and retracting plasma membrane processes (see supplemental video at www.sbic.pitt.edu/harshyne/figure 4). During some of these interactions, large portions of DiD-labeled plasma membrane were physically pulled from one DC to another (Fig. 4, 48–168 min). Transferred membrane appeared to be contained in endocytic vesicles of up to 1 μm in diameter, and frequently multiple vesicles were present within single DC (Fig. 4, 192–216 min). DC that had acquired membrane could be seen in isolation as double-labeled cells less than 4 h after the initiation of the experiment (Fig. 4, 216 min). Importantly, DC that had donated membrane also moved away and interacted with other DC, indicating continued viability of these cells (Fig. 4, 192–216 min). To definitively rule out the contribution of apoptotic cell death in transfer of fluorescent label between DC, we used Hoescht 33342 to label nuclei during live cell imaging. Nuclei of cells remained intact and noncondensed during all stages of cell interaction and transfer, as highlighted by the images collected at 0, 96, and 216 min (Fig. 5). These data provide direct evidence that DC involved in plasma membrane transfer were viable.

To ascertain the immunological significance of the observed transfer between DC, we developed a xenogeneic T cell stimulation assay with TIL620, an HLA-A2-restricted CTL line specific for melanoma Ag gp100 (17). Ad-gp100 was used to express gp100 protein as an endogenous cytoplasmic Ag in DC. Immature monkey DC, confirmed to be HLA-A2⁻ by Ab staining (data not shown), were transduced with Ad-gp100 and cocultured in equal numbers with immature HLA-A2⁺ human DC. TIL620 cells were added to the culture and IFN-γ production was assayed 18 h later by using ELISPOT. As expected, culturing Ad-gp100 transduced HLA-A2⁺ DC with TIL620 resulted in vigorous IFN-γ release (Fig. 6,a). However, coculture of Ad-gp100 transduced monkey DC with HLA-A2⁺ human DC was also effective at inducing cytokine release from TIL620 (Fig. 6,a). Cytokine production was not attributable to direct stimulation of TIL620 by transduced monkey DC, as coculture with HLA-A2⁻ human DC did not elicit significant IFN-γ production. Similarly, coculture of monkey DC and HLA-A2⁺ human DC in the absence of gp100 resulted in negligible stimulation of TIL620 (Fig. 6,a). When HLA-A2⁺ macrophages were used as acceptor cells with Ad-gp100-transduced monkey DC minimal IFN-γ release by TIL620 was observed (Fig. 6,b), consistent with the flow cytometry data indicating that macrophages cannot acquire plasma membrane or intracellular protein from live cells. Directly transduced HLA-A2⁺ macrophages efficiently stimulated TIL620 (Fig. 6 b).

To determine whether Ad-gp100-transduced cells were dying in the ELISPOT assays, which could provide Ag to DC in the form of apoptotic bodies, we performed TUNEL staining at the conclusion of the 18-h coculture. Monkey DC were TUNEL^neg at 18 h, ruling out this possibility (Fig. 6,c). Apoptosis of human DC directly transfected with Ad-gp100 did occur after 18 h of coincubation with TIL620 (Fig. 6 c). In summary, the ELISPOT data are consistent with direct transfer of intracellular Ag between live DC resulting in processing and presentation in MHC class I.

The results of this study demonstrate for the first time that simple interactions between DC and other healthy cells lead to substantial acquisition of plasma membrane and cytoplasmic proteins by DC. Moreover, the data show that such transfer is immunologically relevant, resulting in cross-presentation of endogenous cytoplasmic Ag to MHC class I-restricted CTL. These data challenge the current view that DC only acquire Ag from dead and dying cells.

Several findings provide conclusive evidence that transfer occurred independently of apoptosis, the common mechanism cited for acquisition of cellular Ag by DC (2, 3). First, labeled DC used in the 4 h coculture experiments had intact DNA as measured by TUNEL assay. Importantly, DC did not undergo apoptosis during the assays as measured by annexin V staining, which detects the early apoptotic event of phosphatidylserine membrane translocation. Second, live cell imaging experiments clearly demonstrated that DC maintained intact nuclear structure, cell morphology, and motility before, during, and after transfer of fluorescent label to other DC. Third, gp100-transduced monkey DC were TUNEL^neg for the duration of the 18-h ELISPOT assay, indicating that apoptosis of these cells did not contribute to Ag acquisition by human DC with subsequent stimulation of T cells. TIL620 were clearly capable of killing Ag-bearing DC as evidenced by apoptosis in gp100-transduced HLA-A2⁺ DC. We hypothesize that apoptosis would become apparent in HLA-A2⁺ DC cocultured with gp100-transduced monkey DC had a longer time period elapsed to allow for sufficient Ag transfer and processing in recipient DC. Finally and most notably, macrophages did not acquire fluorescently labeled membrane and cytoplasm from DC and other macrophages, nor did they cross-present Ag when cocultured with live Ag-expressing DC, despite the exquisite capacity of these cells to phagocytose apoptotic material (4). This finding highlights an apparent distinction between DC and macrophages in that both cell types internalize particulate material, but only DC are able to acquire significant levels of Ag from other live cells.

The observed mechanism of Ag acquisition by DC also appears to be independent of exosomes, which are 50- to 90-nm diameter vesicles released from DC and other cells that have been shown to function as Ag-presenting moieties (22, 23). Studies using DC pulsed with FITC have suggested a role for secreted exosomes in uptake by other DC, although Ag transfer was more effective after direct cell contact (15). In our experiments, acquisition of fluorescently labeled membrane and cytoplasm was completely abrogated by a 0.4-μm transwell filter, a pore size that would not restrict flow of exosomes. The live cell microscopy images revealed that intimate DC-DC contact resulted in physical stripping of membrane from one cell by another, leading to the formation of relatively large vesicles up to 1 μm in diameter. DC appeared to interact via plasma membrane extensions, suggesting that these characteristic cell processes play an important role in Ag transfer between DC. Hence, DC can secrete small exosomes into the supernatant but have the capacity to readily produce substantially larger vesicles after direct interaction with other DC.

The functional studies indicated that endogenous tumor Ag expressed in one DC could be acquired by another DC for cross-presentation to class I-restricted CTL. Cross-presentation by DC of cell-associated Ag in MHC class I previously only had been demonstrated for Ag derived from dead or dying cells (2, 3, 4, 5, 6, 7, 8). In our studies, it is likely that cytoplasmic Ag was carried within vesicles that were observed being transferred between DC by live-cell imaging. This is consistent with the flow cytometry data showing that cells that had acquired substantial membrane label also acquire labeled intracellular proteins. Whether Ag is transferred intact and requires further processing in the recipient DC before loading on MHC class I molecules is not known. An alternative mechanism for cross-presentation could be trafficking of plasma membrane containing MHC molecules in a reverse direction, supported by recent findings that T cells can acquire MHC-peptide complexes from APC after TCR engagement (13, 14). However, transferred HLA-A2 would have to be internalized by the monkey DC, loaded with processed gp100 Ag, and trafficked intact to the cell surface for subsequent recognition by TIL620. We currently are investigating the precise mechanism of cross-presentation and whether reciprocal transfer of Ag and MHC molecules occurs in this system.

The marked degree of Ag transfer detected in this study suggests that Ag acquisition by DC from other viable cells may be an important process in vivo. Acquisition of Ag was significantly more efficient by immature than mature DC, indicating that Ag transfer in vivo may occur at the level of the skin and other peripheral tissues where immature DC reside. Hence, DC directly infected with a pathogen could provide a continuous source of Ag for resident Langerhan’s cells or immature DC recruited to the site through release of inflammatory cytokines (24), thereby rapidly and effectively amplifying the immune response. Transfer of Ag between DC in skin would enable subpopulations that are apparently nonmigratory to nevertheless stimulate robust CTL responses in vivo, as has been suggested for CD8α⁺ lymphoid DC in the mouse (25). This may be the case for mature DC administered as vaccines, which we have shown in the primate model have a tendency to remain at the site of injection (16). Alternatively, Ag transfer could occur after migration of Ag-exposed DC to lymph nodes as has been suggested by others (11), mediated by lymph node DC subpopulations that have Ag-acquiring capacity (26). Moreover, our data indicate that cells other than DC can serve as Ag donors, including macrophages, T cells, and B cells. Hence, macrophages containing phagocytosed material could serve as a source of Ag for DC in peripheral tissues. Similarly, lymphocytes or macrophages infected with pathogens could donate Ag to DC. This latter scenario suggests a mechanism whereby DC could acquire Ag derived from pathogens that preferentially infect macrophages and lymphocytes without necessarily causing cell death, such as HIV and EBV.

It is believed that maturation is required for DC to become efficient stimulators of Ag-specific T cells (6, 27). Studies indicate that uptake of apoptotic material by itself may not provide sufficient stimulus for maturation of DC and that additional stimuli provided by necrotic cells may be required for effective Ag presentation (7). In addition, uptake of apoptotic material by DC that have migrated from the gut has been reported, suggesting that DC exposed to Ag in this manner may serve to induce T cell tolerance to intestinal Ag in draining lymph nodes (28). However, in our studies, immature DC readily acquired Ag from other immature DC and were able to cross-present Ag to CTL, suggesting that maturation signals may not be essential for effective immunostimulatory function. It remains to be determined whether Ag transfer between live DC is involved in induction of tolerance to self.

In summary, our studies reveal that interactions between DC and other healthy cells result in acquisition of membrane and cytoplasmic components by the DC, and subsequent cross-presentation to T cells. We suggest that this previously unrecognized means of Ag transfer might be important in the development and regulation of immune responses in vivo.

We thank M. Zimmer for stimulating discussion, O. Finn and S. Rosenberg for cell lines, Immunex Corporation and Schering-Plough Research Institute for cytokines, M. Murphey-Corb and the University of Pittsburgh Primate Facility for Infectious Disease Research for monkey samples, C. Castillo for technical assistance, and J. Ahearn for access to the flow cytometer.