In vivo studies suggest that vascular endothelial cells (ECs) can acquire and cross-present exogenous Ag on MHC-I but the cellular mechanisms underlying this observation remain unknown. We tested whether primary female mouse aortic ECs could cross-present exogenous male Ag to the T cell hybridoma, MHH, specific for HYUty plus Db. MHC-I-deficient male spleen cells provided a source of male Ag that could not directly stimulate the MHH cells. Addition of male but not female MHC-I-deficient spleen cells to wild-type syngeneic female EC induced MHH stimulation, demonstrating EC cross-presentation. Lactacystin treatment of the donor male MHC-I-deficient spleen cells, to inhibit proteasome function, markedly enhanced EC cross-presentation showing that the process is most efficient for intact proteins rather than degraded peptide fragments. Additional experiments revealed that this EC Ag-processing pathway is both proteasome and TAP1 dependent. These studies demonstrate that cultured murine aortic ECs can process and present MHC-I-restricted Ag derived from a separate, live cell, and they offer insight into the molecular requirements involved in this EC Ag presentation process. Through this pathway, ECs expressing cross-presented peptides can participate in the effector phase of T cell-mediated inflammatory responses such as autoimmunity, anti-tumor immunity, and transplant rejection.
To initiate a T cell immune response, professional APCs (e.g., dendritic cells (DCs))3 process protein Ags into peptide determinants that are subsequently bound to MHC molecules for presentation to T lymphocytes. In general, proteins deriving from sources exogenous to the APC and/or from within the endolysosomal compartment of the APC are preferentially but not exclusively processed and expressed on the cell surface in the context of class II MHC (MHC-II) to activate CD4 T cells (1, 2). Proteins found primarily in the cytoplasm, e.g., self proteins and those derived from viruses, are generally degraded by proteasomes and transported in a TAP-dependent fashion to the endoplasmic reticulum where they are complexed to class I MHC (MHC-I) molecules and shuttled to the cell surface where they are poised to activate CD8 T cells (3). Exogenously derived Ags can be endocytosed or phagocytosed by DCs and directed through an increasingly well understood, alternative, cross-processing pathway to combine with MHC-I molecules and thereby stimulate CD8 T cells (2, 4, 5, 6). This latter process, commonly referred to as cross-presentation, may be essential for priming CD8 T cells specific for viruses that cannot infect professional APCs (4, 6).
Once activated, primed T cells exit the secondary lymphoid organs, circulate through the peripheral blood stream and peripheral tissues, and are attracted to sites of inflammation. If the primed T cell re-encounters its specific Ag in the periphery, it engages its effector machinery, leading to cytotoxicity and cytokine release. This requirement for a second encounter between a T lymphocyte and its peptide/MHC ligand to induce effector function underscores the importance of Ag processing and presentation in the peripheral tissues. How and where a given Ag is processed and presented by parenchymal or endothelial cells can influence the outcome of an immune response. As one illustration, we previously demonstrated that a primed TCR transgenic T cell could induce either acute graft destruction or chronic progressive graft vasculopathy, based on the cellular expression pattern of Ag within the target tissue (7). Acute graft destruction resulted when the Ag was expressed on the graft vascular endothelium and parenchymal cells whereas the same T cells caused vascular injury if the Ag was only found on graft-infiltrating mononuclear cells (but not parenchymal or endothelial cells).
Activated T cells initially encounter endothelial cells (ECs) lining the vasculature feeding sites of inflammation (8, 9, 10, 11, 12, 13). ECs express the protein machinery for Ag processing, including proteasome subunits, TAP proteins, and both MHC-I and -II, and can present endogenous Ag to activated T cells (11, 12, 13, 14, 15). Several in vivo studies, including those from our group, showed that ECs could cross-present exogenous Ag to activated CD8 T cells and thereby function as essential targets of the pathogenic immune responses (16, 17). Cross-presentation by ECs has not been directly demonstrated in vitro, and little is known about the cell biology of this process. Herein we definitively show that cultured ECs can cross-present exogenous, cell-derived Ag on MHC-I, and we provide new insight into the molecular requirements involved in this EC Ag presentation process.
Materials and Methods
Male and female C57BL/6 (B6, H-2b), C57BL/6 TAP1−/− (TAP1−/−, H-2b), C57BL/6 GFP transgenic (B6 GFP, H-2b), and C3H (H-2k) mice, age 6–10 wk, were purchased from The Jackson Laboratory. C56BL/6 DbKb−/− mice (originally produced by Hidde Pleogh (18)) were a kind gift of Alexander Chervonsky (The Jackson Laboratory). MataHari (CD8 TCR transgenic specific for Db + HYUty, backcrossed to RAG1−/−, C57BL/6 (16)) were a kind gift of Polly Matzinger (National Institutes of Health). All animals were maintained and bred in the pathogen-free animal facility at The Cleveland Clinic Foundation (CCF) and used under approved protocols.
Preparation and characterization of MHH hybridoma
MHH hybridoma cells were produced and cloned according to published methods (19, 20). To assess Ag specificity, hybridoma cells (1 × 105/well) were incubated with 5 × 104 B6 female spleen cells with or without peptide (5 μg/ml) in a total volume of 200 μl. After overnight incubation, supernatants (100 μl/well) were harvested and tested for IL-2 production by CTLL-2 proliferation using a colorimetric assay with Alamar Blue (21). Positive wells generally had an OD difference of >0.100 between Ag and medium control wells. Ag specific cell lines were expanded and retested to select the most sensitive T hybridomas. Peptides HYUty (WMHHNMDLI) and β-galactosidase96–103 (β-gal96–103) (DAPIYTNV) were synthesized by Research Genetics.
All Abs used for flow cytometry were obtained from BD Pharmingen. Cells were stained and analyzed on a FACScan (BD Biosciences), and 5,000–10,000 events were acquired for each sample.
EC isolation, passage, and characterization
Ag presentation assays
ECs were plated at 20,000 cells/well in 48-well plates in endothelial culture medium (RPMI 1640 + 20% FBS, 0.1% 2-ME (Eastman Kodak), 10,000 U/ml penicillin, 10,000 μg/ml streptomycin, and 0.05 mg/ml EC growth supplement (Fisher Scientific)). After an overnight incubation, the ECs were activated with 1 ng/ml TNF-α (R&D Systems), 1 ng/ml IL-1 (R&D Systems), and 100 U/ml IFN-γ (R&D Systems) plus peptides (1 μM) or “Ag donor” spleen cells (7 × 106 per well) overnight. The wells were washed three times, and 400,000 hybridoma cells were added to each well along with additional peptide or spleen cells. In selected experiments, the Ag donor spleen cells were pretreated with 20 μM lactacystin (Sigma-Aldrich) for 1 h followed by three washes in HBSS before addition to the cultures. In other experiments, the ECs were treated for 1 h with lactacystin followed by three washes before adding Ag. Culture supernatants were harvested at 96 h and tested for IL-2 production by CTLL proliferation using a colorimetric assay (20). Statistical comparisons were performed using Student’s t test.
In vitro T cell cytotoxicity assay
GFP+B6 aortic endothelial cells were plated at 20,000 per well in 48-well plates, activated with cytokines overnight as above and washed to remove residual cytokines. Splenic T cells isolated from female MataHari mice were activated with HYUty plus APCs in vitro for 72 h, washed and purified using commercially available T cell columns, and added to the activated ECs with peptide or DbKb−/− spleen cells as above. Digital images were obtained under a UV microscope, and the number of GFP+ ECs in each well (average of three different replicate wells) was assessed by computer-assisted image analysis at 72 and 96 h. Percentage of cytotoxicity was determined as follows: (1 − (number of ECs in test well/number of ECs in no Ag control well)) × 100%. Statistical comparisons were performed using Student’s t test.
Characterization of MHH hybridoma cells
We first developed a sensitive readout to detect expression of the immune dominant, MHC-I-restricted male antigenic peptide, HYUty, in the context of MHC-I Db on the surface of in vitro cultured murine aortic endothelial cells. Spleen cells from female MataHari TCR transgenic mice (RAG1−/− H-2b, Vβ8.3, specific for HYUty + Db) were fused with the TCR-deficient CD8+ BW5147 partner (19) to produce a CD8+ T cell hybrid (MHH). MHH cells expressed CD8, Vβ8.3, and MHC-I Db (Fig. 1, A and B). Ag specificity studies (Fig. 1 C) confirmed that MHH cells responded to HYUty peptide + MHC-Ib. MHH cells did not respond to H-2b APCs plus a control peptide, β-gal96–103, and did not respond to allogeneic H-2k APCs + HYUty peptide.
ECs cross-present exogenous intracellular Ag on class I
Vascular endothelial cells were isolated from murine aortas and maintained in short term culture (16, 22). To test whether the EC cells could stimulate MHH cells we bypassed any requirement for Ag processing and directly added HYUty peptide or a control peptide to cultures of female B6 ECs plus MHH cells (Fig. 2). As shown, HYUty peptide-loaded, but not β-gal96–103-loaded female B6 ECs stimulated MHH cells to produce IL-2, confirming specificity. As another control, female MHC-I-deficient DbKb−/− ECs were tested in this assay and did not stimulate MHH cells with or without HYUty peptide (Fig. 2). This latter result confirms that HYUty peptide is MHC-I restricted but more importantly shows that MHC-I+ MHH hybridoma cells in the culture system were not presenting the HYUty peptide to one another under our assay conditions.
With this highly sensitive tool capable of detecting HYUty + Db complexes on the surface of ECs, we designed an experimental system to test whether female B6 ECs (HY−, MHC-I+) could acquire the male Ag derived from live, exogenous male cells and present it to MHH cells (Fig. 3 A). Spleen cells obtained from male DbKb−/− mice were used as exogenous “donors” of the male Ag. In this manner the MHH cells could not recognize any Ag expressed on the DbKb−/− spleen cells (MHC-I−, HY+) and would only respond if the male Ag were cross-presented by female EC-expressed MHC-I.
Pilot studies showed that addition of DbKb−/− spleen cells inhibited detection of the IL-2 in culture supernatants (Fig. 3 B). Addition of 50 μg/ml anti-CD25 mAb (PC-61; Bio Express) to the wells specifically overcame this effect suggesting that IL-2Rs expressed on the large number of added DbKb−/− spleen cells bound to the secreted IL-2, precluding detection. Anti-CD25 mAb was therefore added to subsequent experiments to overcome this in vitro artifact.
Fig. 3,C reveals that female ECs mixed with male, but not female, DbKb−/− splenocytes stimulated a weak response by the MHH cells suggesting that ECs could process and cross-present this Ag in the context of MHC-I. Recently published studies showed that the stability of the donor Ag strongly influenced the ability of a DC to effectively cross-process and present the relevant determinant: inhibiting protein degradation within the cell supplying the exogenous Ag with the proteasome inhibitor lactacystin significantly enhanced the efficiency of DC cross-processing (23, 24). Consistent with these reports, EC cross-processing and presentation was markedly and significantly enhanced when the donor male DbKb−/− spleen cells were pretreated with lactacystin (Fig. 3,C, black bars). In contrast, the addition of lactacystin-treated female DbKb−/− spleen cells to female B6 ECs did not stimulate a response by the MHH cells, indicating that the effect of lactacystin was to augment Ag-specific cross-presentation. Additional control experiments showed that female DbKb−/− ECs mixed with lactacystin-treated male DbKb−/− spleen did not stimulate the MHH cells (Fig. 3 C, bottom panel). This latter result implies that 1) the ECs must express the relevant MHC-I restriction element (Db) to present the HYUty peptide derived from male cells and 2) the MHH cells cannot process and present the donor HY Ag to one another under the assay conditions used for these experiments.
We next sought to independently confirm that ECs could cross-process and present exogenous Ag to CTLs. Purified and in vitro-activated primary female MataHari TCR transgenic T cells were mixed with female GFP+B6 ECs and male (or control female) DbKb−/− spleen cells as a source of exogenous Ag. We used GFP+ ECs to be able to directly count the number of live ECs per test well and thereby calculate the percentage of CTL activity (see Materials and Methods). As shown in Fig. 4, the MataHari T cells efficiently killed the female ECs when male DbKb−/− spleen cells, but not female DbKb−/− spleen cells, were added to the cultures.
Endothelial cell cross-presentation is proteasome and TAP dependent
To further dissect the molecular basis of EC cross-processing/presentation, we tested whether EC-proteasome activity (rather than proteasome activity in the Ag donor cell as shown above) was required. As shown in Fig. 5, lactacystin treatment of female B6 ECs inhibited their ability to cross-present the male Ag to MHH. Importantly, the lactacystin had no effect on presentation of directly loaded, purified HYUty peptide to the MHH cells. This result confirmed that the lactacystin-treated ECs were capable of presenting Ag to the MHH cells if Ag processing was not required and ruled out the trivial possibility that the lactacystin killed the ECs.
The TAP facilitates proteosome-derived peptide entry into the ER for loading onto class I MHC (25). We tested whether TAP was involved in EC cross-processing and presentation by studying ECs derived from TAP1−/− (H-2b) mice. Endothelial cell cross-presentation was indeed TAP1-dependent because TAP1−/− ECs were unable to stimulate MHH cells mixed with male or female DbKb−/− spleen cells (Fig. 5, gray bars). The absence of TAP1 had no significant effect on presentation of directly loaded purified HYUty peptide to the MHH cells, again confirming that the ECs were alive and capable of presenting exogenous peptide, but not processed Ag.
This demonstration that primary aortic ECs can cross-present exogenous, cell-derived Ag on MHC-I provides new insight into the Ag-presenting capacity of ECs and extends their known role as targets of the cellular immune system. These data, in conjunction with previous reports (16, 17, 26, 27), show that proteins secreted by surrounding cells or intact proteins donated from live cells interacting with the endothelium in vivo can be acquired, processed, and presented on the surface of ECs in the context of MHC-I. Our data also strongly suggest that this cross-presentation pathway functions by acquiring intact proteins, rather than peptide fragments, because proteasome inhibition of the donor cell markedly enhanced the efficiency of the process (Fig. 3,C). The result is not surprising based on the recent descriptions of similar requirements for DC cross-presentation (23, 24), but our data extend previous reports by showing that ECs, “non-professional” APCs, function in an analogous manner. The fact that male DbKb−/− spleen cells did not function as efficient Ag donors unless treated with lactacystin (Fig. 3 C) further suggests that the HYUty gene product is normally short-lived (i.e., rapidly degraded) within the cell.
The cell biology of the Ag acquisition process is not yet fully elucidated but may involve standard endocytotic pathways. It has also been shown that intimate cellular interactions can result in transfer of membrane components from one cell to another (28, 29, 30), raising the possibility that proteins could be transferred from a transmigrating cell to ECs during diapedesis. Once the exogenous proteins are acquired by the ECs, our data reveal that the Ag presentation pathway is both proteasome and TAP1 dependent, similar to several known cross-presentation pathways characterized in professional APCs (4, 23, 24, 31, 32, 33). It is possible that the acquired proteins are directed to the cytosol where they join the cytoplasmic pool of endogenous Ags for processing. Alternatively, recent studies showed that early phagosomes can be derived from the endoplasmic reticulum and contain all of the machinery (including the proteasome, TAP proteins, and class I molecules) to function as self-sufficient Ag processing organelles (33).
Regardless of the specific intracellular pathways, the importance of EC cross-presentation is that exogenously derived ligands presented on EC surfaces could interact with, and activate primed effector or memory CD8 T cells. Such ligands may then function as targets of the CD8 T cell response, ultimately leading to destruction of the tissue fed by the relevant vasculature. Alternatively or in addition, the endothelium-expressed ligands may provide signals that are essential for primed CD8 T cells to enter the tissue and migrate toward the primary site of inflammation. Although cross-presentation by ECs has been shown in this regard to be relevant to models of skin graft rejection (16) and autoimmune diabetes (17), this process may more generally facilitate control of viruses that preferentially infect parenchymal cells.
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 National Institutes of Health Grants AI43578 (to P.S.H.), AI34343 and AI35726 (to C.V.H.), and AI36219 (to D.H.C.). A.V. is a recipient of a Scientist Development Award, and P.N.L. is a recipient of a Fellowship grant from the American Heart Association.
Abbreviations used in this paper: DC, dendritic cell; EC, endothelial cell; B6, C57BL/6; β-gal, β-galactosidase; MHC-I, MHC class I; MHC-II, MHC class II.