Differential activation of human T cells to allogeneic endothelial cells, epithelial cells and fibroblasts in vitro

Single donor endothelial cells (EC) were isolated from human umbilical cords as previously described [35], and were grown in M199 medium (Bio Whittaker) containing 20% FCS (GIBCO BRL, Grand Island NY) or 10% human serum, EC growth supplement, 1% penicillin/streptomycin, l-glutamine, and heparin. Single donor fibroblasts and renal tubular epithelial cells (RPTEC) were purchased from Clonetics and cultured in CC-4126 FGM 2 or CC-4127 REGM (Clonetics, Walkersville MD) according to the manufacturers protocol. EC, fibroblasts and RPTEC were treated with IFN (1000U/ml, R&D Systems, Minneapolis, MN) for 72h prior to use. EC were used in subculture 34. Apoptosis was induced in EC monolayers following treatment with TNF- (200 U/ml, Biogen, Cambridge MA) and cyclohexamide (2ng/ml, Sigma, St. Louis, MO) for 7h. Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Hypaque gradient centrifugation from healthy volunteers. CD4⁺ T cells were isolated from PBMC by positive selection using anti-CD4-coated magnetic beads (Invitrogen, Grand Island NY) according to the manufacturers protocol. CD4⁺CD45RA⁺ and CD4⁺CD45RO⁺ cells were isolated from CD4⁺ cells by negative selection using magnetic beads coated with mouse anti-human CD45RO and CD45RA antibodies (Invitrogen, Grand Island NY).

Transmigration experiments were conducted using EC, fibroblasts or RPTEC monolayers cultured on fibronectin (50 g/ml) coated 3m pore size transwell inserts (Costar, Cambridge, MA, USA) as previously described [35, 36]. A total of 310³ EC, fibroblasts or RPTEC were seeded onto inserts and were allowed to grow for 45days. Monolayers were treated with IFN (1000U/ml, R&D systems) for the last 72 culture. Confluence of the monolayers was confirmed by Coomasie staining using standard techniques [37]. Monolayers were labeled with DiOC-16 (5 g/ml, Molecular Probes, Eugene OR, USA), which incorporates into the membranes of cells and has an emission at 501nm. A total of 1x10⁶ PBMC were placed into the upper transwell and cells were collected from the bottom well after 1.5h. Following collection, DiOC uptake was analyzed on CD14⁺ CD4⁺ and CD8⁺ cells by flow cytometry.

Confluent monolayers were harvested in Trypsin/EDTA (Sigma, St. Louis, MO) and were analyzed by indirect immunofluorescence staining and flow cytometry as previously described [23]. Briefly, cells were incubated in optimal concentrations of the primary antibodies anti-HLA DR (LB3.1, ATCC, Manassas, VA), negative control mouse IgG (K16/16, a gift from Michael Gimbrone, Brigham and Womens Hospital) or positive control anti-ICAM-1 (RR1/1, a gift from TS Springer) for 30 mins on ice. Cells were washed and subsequently incubated in FITC-conjugated anti-mouse secondary antibody (Jackson, Immunoresearch, West Grove, PA) for an additional 30 mins on ice, and stained cells were washed and fixed in 1% paraformaldehyde prior to analysis. Leukocytes were stained using Phycoerythrin (PE) conjugated anti-human -CD14, -CD4 and -CD8 antibodies (PharMingen, San Diego CA) or isotype negative controls using standard techniques. All stained cells were analyzed using a FACSCalibur cell sorter (Becton Dickinson, Mountain View, CA) and CellQuest and FlowJo software.

EC, fibroblasts or RPTEC were cultured to confluence in fibronectin-coated 24-well flat-bottom plates (Costar) coated with fibronectin (50 g/ml) and treated with IFN (1000U/ml) for 72h. PBMC (1x10⁶/well) were added and were cultured for 7days in standard RPMI 1640 medium containing 10% autologous serum. Subsequently, PBMC were harvested by washing with HBSS and CD4⁺ T cells were isolated by positive selection (Invitrogen, Grand Island NY) and were rested in culture medium containing 2.5% of human T-stim without PHA (Becton Dickinson Labware, Bedford, MA) for 5days. The allogeneic HLA mismatch between PBMCs and stimulator EC, fibroblasts or RPTEC was unknown, but was presumed based on consistency of the response in multiple repeated experiments.

Cell membranes were prepared as sonicates from IFN- treated EC, fibroblasts and RPTEC. Exactly 8 10⁶ of each cell type was washed and suspended in 1ml of sterile PBS and was sonicated using a tip sonicator (Branson Sonifer 250) fitted with a 2mm probe. Disrupted cells were centrifuged for 10min at 500g to remove debris. A total of 20l of sonicate was used in each assay. As a quality control, the protein content per ml was measured in occasional sonicate supernatants by standard Bradford assay. Sonicates were frozen at 20C and thawed at 37C before use.

CD4⁺ T cells (1x10⁵/well) isolated from primary cultures (above) were used with autologous APCs in proliferation assays. Autologous APCs (1x10⁵/well) were isolated from T cell-depleted PBMC, were irradiated (1750rad) and used as stimulators. Assays were performed in round-bottom 96-well plates (Costar). Sonicates from the allogeneic cells, were added (as above) to each culture and, after 6days proliferation was assessed by standard [³H]thymidine incorporation assays (1Ci/well) for the last 16h of the coculture.

ELISpot was performed in 96-well plates (Cellular Technology Ltd., Cleveland, OH) coated overnight with capture cytokine antibodies diluted in sterile PBS. Antibodies were mouse anti-human IFN (clone 2G1, Endogen, Wolburn, MA, 4g/ml), mouse anti-human IL-2 (clone 5355, R&D Systems Inc., Minneapolis, MN, 5g/ml), mouse anti-human IL-4 (clone 8D4-8, BD PharMingen, San Diego, CA, 5g/ml) or rat anti-human IL-5 (clone TRFK5, PharMingen, San Diego, CA, 5 g/ml). After blocking for 1h with PBS/1%BSA, the plates were washed and CD4⁺ T cells (2x10⁵/well) were cultured with autologous APCs (2x10⁵/well) in the absence or presence of sonicate. The plates were incubated at 37C in 5% CO₂ for 24-36h. Detection antibodies were biotinylated mouse anti-human IFN- (clone B133.5, Endogen, 2g/ml), mouse anti-human IL-2 (clone 5334, R&D Systems Inc., 50ng/ml), rat anti-human IL-4 (clone MP4-25D2, BD PharMingen, 2g/ml) and rat anti-human IL-5 (clone JES1-5A10, PharMingen, 2g/ml). After overnight incubation at 4C, the plates were washed and cytokines were detected using streptavidin HRP (Daco, Carpenteria, CA) and 3-amino-9-ethyl carbazole (Pierce Chemical Co., Rockford, IL). Spots were counted using a computer-assisted ELISpot image analyzer (Cellular Technologies Limited).

Statistical analysis was performed using the student t test for two groups of data and by one-way ANOVA for three or more groups. P values <0.05 were considered statistically significant.

We initially evaluated whether APCs acquire membrane fragments from allogeneic cells during brief interactions in the course of transmigration. We used a standard transwell model in which PBMC were allowed to transmigrate through confluent IFN-treated EC, fibroblasts or RPTEC. Prior to the assay, cells were labeled with lipophylic DiOC-16, which is well established to stably incorporate into cell membranes. As illustrated in Figure 1, we found that 3565% of CD14⁺ monocytes acquired dye after interaction with both fibroblasts and RPTEC. However, surprisingly, the transfer of dye was very limited after interaction with EC. We also found that neither CD4⁺ T cells nor CD8⁺ T cells acquire dye from any allogeneic cell type indicating that the transfer was related to phagocytosis of membrane rather than through cell surface membrane transfer (as can occur in the semi-direct pathway of allorecognition [7, 38]). To further confirm that intact EC fail to transfer membrane to APCs, we also assessed transfer when PBMC transmigrated across EC undergoing apoptosis (TNF- and cyclohexamide- treated cells). As illustrated in Figure 1B, we find that APCs acquire DiOC-labeled membrane from apoptotic EC (15-25% cells) as compared to untreated or IFN-treated EC (3-10% cells). In contrast, the transmigration of PBMC across apoptotic fibroblasts or RPTEC did not alter DiOC-labeled membrane uptake from that described above (data not shown). Therefore, it is possible that acute injury or alloimmune targeting of EC may be a factor in the initiation of indirect processing of alloantigen by APCs. This process may result in crosstalk between both pathways of allorecognition, as described [8].

We next wished to compare the ability of EC, fibroblasts and RPTEC to induce direct and indirect pathway alloactivation of nave CD45RA⁺ and memory CD45RO⁺ CD4⁺ T cells. CD45RA⁺ and CD45RO⁺ cells were isolated by negative selection from pure populations of CD4⁺ T cells (>90% purity by FACS, data not shown) and were cocultured with IFN-treated EC, fibroblasts or RPTEC for 5days. As illustrated in Figure 2A, and consistent with other reports [23, 39, 40], we find that our IFN-treated EC, fibroblasts and RPTEC express high levels of MHC class II. However, CD45RO⁺CD4⁺ T cells showed significant proliferation only in response to culture with allogeneic EC, and had minimal proliferative responses when cultured with fibroblasts or RPTEC (Figure 2B). In contrast, there was a minimal proliferative response of CD45RA⁺CD4⁺ T cells to all allogeneic stimulator cell types (Figure 2B). This observation is consistent with previously published reports indicating that EC function in the activation of allogeneic T cells [22, 23, 41, 42]. They are also consistent with previous studies indicating that EC fail to provide sufficient costimulation for the initiation of nave T cell activation responses [23, 28].

In order to evaluate the effect of each cell type in indirect alloactivation, we next cocultured CD45RA⁺ or CD45RO⁺ CD4⁺ T cells with autologous APCs in the absence or presence of sonicate derived from allogeneic 72h IFN-treated EC, fibroblasts and RPTEC (as shown in Figure 2A). In these assays, T cells cultured in the presence of autologous APCs without sonicate served as a negative control. Illustrated in Figure 3A, we find that both nave CD45RA⁺ as well as memory CD45RO⁺ CD4⁺ T cells proliferate to sonicates prepared from all cell types. As a control, each T cell subset failed to proliferate to sonicate alone in the absence of autologous APCs (Figure 3B). These observations indicate that indirect allorecognition is a most potent mechanism for the allogeneic activation of both nave and memory T cells.

We next created an in vitro model to evaluate whether there is a difference in indirect responses among unsensitized and sensitized CD4⁺ T cells. For this purpose, we cultured PBMC with IFN-treated EC, fibroblasts or RPTEC. After 7days, the CD4⁺ T cells were isolated from the cultures and were rested in culture medium alone. After 5days of rest, these sensitized T cells (or freshly isolated unsensitized cells) were then used as responders in cocultures with autologous APCs in the absence or presence of sonicates derived from the allogeneic EC, fibroblast or RPTEC used in the primary culture. Proliferation was again assessed on day 6. Illustrated in Figure 4, we found no difference in the proliferative responses of unsensitized or sensitized T cells in these indirect allorecognition assays. By ELISpot (48h), we found variations in individual cytokine production between sensitized and unsensitized cells (Figure 4B). However, the overall production of cytokines was at high levels reflective of the activation status of the T cells and there were no significant differences between responses in unsensitized and sensitized cells (data not shown). ELISpot counts for IFN were generally higher when EC were used as the alloantigen, as compared to assays where fibroblasts and RPTEC were used (Figure 4B). Collectively, these observations further confirm that the indirect response is potent to initiate T cell activation, and indicate that T cells primed through the indirect pathway may be reactivated within allografts upon encounter with self-APCs.