CD8+ T Cell Cross-Priming via Transfer of Proteasome Substrates

See allHide authors and affiliations

Science  28 May 2004:
Vol. 304, Issue 5675, pp. 1318-1321
DOI: 10.1126/science.1096378


“Cross-priming” describes the activation of naïve CD8+ T cells by professional antigen-presenting cells that have acquired viral or tumor antigens from “donor” cells. Antigen transfer is believed to be mediated by donor cell–derived molecular chaperones bearing short peptide ligands generated by proteasome degradation of protein antigens. We show here that cross-priming is based on the transfer of proteasome substrates rather than peptides. These findings are potentially important for the rational design of vaccines that elicit CD8+ T cell responses.

CD8+ T cells play important roles in the elimination of pathogens, tumors, and transplanted tissues (1). Virus-specific CD8+ T cells recognize major histocompatibility complex (MHC) class I molecules bound to 8- to 10-residue peptides derived from viral proteins (2). Induction of primary CD8+ T cell responses requires that viral peptides are presented by class I molecules on “professional” antigen-presenting cells (pAPCs), particularly dendritic cells (3), because only these cells can provide costimulatory signals required to induce the proliferation of naïve T cells. Peptides can be generated from viral proteins synthesized by infected pAPCs themselves (“direct priming”) or from proteins originally synthesized by other infected “donor” cells (“cross-priming”) (4). Although direct priming is now relatively well understood, less is known about the mechanisms involved in cross-priming.

The principal hypothesis for cross-priming posits that molecular chaperones ferry peptides from donor cells to pAPCs (5). Although there is compelling evidence for the potent in vitro antigenicity of peptides bound to molecular chaperones (6), the participation of peptide-chaperone complexes in cross-priming in vivo has not been rigorously examined. One important prediction of the “chaperoned-peptide” hypothesis would be that the potency of donor cells in cross-priming should be directly proportional to the levels of peptides they express.

To study cross-priming, it is necessary to use donor cells that either do not express the same class I molecules as the responding mouse or cannot generate class I molecules owing to some defect in class I biosynthesis (7). In the first set of experiments, we used donor cells that cannot produce class I molecules due to a targeted deletion in the gene encoding the β2-microglobulin (β2m) subunit of class I molecules (8). Cells that were β2m-negative (β2mneg) were exposed to lactacystin, an irreversible proteasome inhibitor (9) (fig. S1), and then infected for 4 hours with a recombinant vaccinia virus (rVV) expressing chicken ovalbumin (Fig. 1). Infected cells were introduced into B6 (H-2b) mice that had previously received transgenic CD8+ T cells from OT-I mice specific for Kb class I molecules presenting a peptide consisting of ovalbumin residues 257 to 264 (Ova257-264) (10). Eighteen hours after immunization, OT-I activation was measured by increased cell surface expression of CD69. Rather than interfering with cross-priming, as predicted by the chaperoned-peptide hypothesis, lactacystin actually had an enhancing effect (Fig. 2A).

Fig. 1.

Schematic of expression vector–encoded gene products used to study cross-priming. The gene constructs utilized in this study are schematically diagrammed.

Fig. 2.

Cross-priming does not require proteasome activity in virus-infected cells. (A) Uninfected β2mneg cells (UN) or cells infected with VV-ovalbumin (OVA) were injected 4 hours after infection into mice that had received OT-I TCR CD8+ T cells 3 hours previously. “OVA/LC” cells were treated with the irreversible proteasome inhibitor lactacystin, which was started 1 hour before infection. Sixteen hours after immunization, activation of CD8+, Vα2+ OT-1 T cells was determined via measurement of increased expression of CD69 by flow cytometry. Data are expressed as the percentage of CD69 highly expressing cells. These data are representative of nine consecutive experiments in which immunogenicity was either enhanced (3 times) or remained unaffected by lactacystin treatment (6 times). (B) Mice were immunized intraperitoneally (i.p.) with IAV in the presence or absence of a neutralizing monoclonal antibody (mAb) against hemagglutinin (α-HA). Six days later, influenza virus–specific CD8+ T cells present in spleens were enumerated by flow cytometry using Db-NP366-374 molecules tetramerized by binding to fluorescent strepavidin. The percentage of CD8+ T cells that specifically bind tetramers is shown. (C) Uninfected β2mneg cells or β2mneg cells infected with IAV for 5 hours were injected i.p. in the presence of a neutralizing mAb against HA, and CD8+ T cells were enumerated as in (B). Where indicated, cells were treated with lactacystin (LC) from 1 hour before infection. The data shown represent the mean ±SEM of duplicate mice and are representative of four similar experiments.

The effect of proteasome inhibition on cross-priming was next examined by immunization of B6 mice with β2mneg cells infected with influenza A virus (IAV). We measured CD8+ T cell responses 6 days after immunization using tetrameric Db molecules complexed to an immunogenic peptide (residues 366 to 374) from IAV nucleoprotein (NP) (Fig. 1). To prevent infection of mice with IAV released from infected cells, we coinjected mice with a monoclonal antibody (“α-HA”) that neutralizes IAV infectivity and thereby abrogates induction of CD8+ T cells by infectious IAV (Fig. 2B). Once again, lactacystin failed to inhibit the immunogenicity of IAV-infected cells (Fig. 2C).

Because these experiments revealed that cross-priming in these systems could occur in the absence of proteasomal degradation of protein antigens, we tested whether preprocessed peptides could also cross-prime. Synthetic Ova257-264 peptide was loaded onto splenocytes derived from mice carrying the bm1 mutation of Kb (10). Although Ova257-264 binds to Kbm1 with relatively high affinity (Fig. 3A), these complexes fail to activate OT-I cells because of alterations in the region of Kbm1 that interacts with the T cell antigen receptor (10). We injected splenocytes pulsed with synthetic peptides into B6 mice and subsequently measured the proliferation of adoptively transferred OT-I cells. Whereas splenocytes from wild-type mice exposed to as little as 10–9 M peptide induced OT-I proliferation, Kbm1 splenocytes failed to elicit measurable OT-I responses, even when exposed to 10–5 M peptide (Fig. 3B).

Fig. 3.

Cross-priming is not mediated by exogenous or endogenous peptides. (A) We measured binding of Ova257-264 to splenocytes derived from B6 or Kbm1 mice by flow cytometry using a fluorescent version of the peptide that mimics the binding of the unmodified peptide. An additional experiment yielded similar results. MCF, mean channel flourescence. (B) Splenocytes from B6 or Kbm1 mice pulsed with synthetic Ova257-264 peptide at the indicated concentrations were injected into B6 mice into which fluorescent OT-I TCR transgenic CD8+ T cells were adoptively transferred 3 hours previously. The percentage of dividing OT-I cells was determined flow cytometrically by decreased cellular fluorescence. This experiment was repeated twice with similar results. (C and D). B6 mice were immunized with the rVV expressing the indicated gene product (C) or with P815 cells infected for 12 hours with rVVs expressing the gene product indicated (D). Six days later, numbers of responding Ova257-264-specific CD8+ T cells present in spleens was determined by their expression of IFN-γ. Data represent averages from three mice for each group; errors bars show ±SEM. The figure is representative of nine additional experiments performed with P815 or HeLa cells. (E) After 293 cells were transfected 24 hours earlier with plasmids encoding the indicted gene product, they were introduced into B6 mice, and the responding CD8+ T cells enumerated by expression of IFN-γ. Averages from four mice for each group with SEM are shown. An additional experiment yielded similar results.

Note that, under the conditions used for pulsing cells with peptide, synthetic Ova257-264 would have free access to the endoplasmic reticulum (ER) via a vesicular transport route (11), where it can potentially associate with gp96 and other molecular chaperones putatively involved in cross-priming. However, the inability of peptide-pulsed Kbm1 cells to activate CD8+ T cells indicates that neither class I molecules nor ER chaperones were capable of mediating cross-priming of exogenous peptide antigens in these experimental conditions.

We explored a more physiologically relevant system for studying peptide-based cross-priming by using donor cells in which peptides were generated by infection with rVVs expressing cytosolic (Ova257-264) or ER-targeted Ova257-264 (ER-Ova257-264) minigene products or a chimeric protein labeled with green fluorescent protein (NP-S-GFP) as a positive control (Fig. 1). After their introduction into B6 mice, each of these viruses induced measurable Ova257-264-specific CD8+ T cell responses as determined by interferon-γ (IFN-γ) expression (Fig. 3C). In marked contrast, immunization of B6 mice with MHC-mismatched P815 cells (H-2d) infected with either of the minigenes expressing rVVs failed to induce detectable Ova257-264-specific CD8+ T cells (Fig. 3D). Cross-priming was easily detectable, however, by means of P815 cells expressing the control chimeric protein (Fig. 3D).

In all experiments using rVV-infected cells, ultraviolet (UV) irradiation was used just before injection to prevent possible transfer of infectious virus. However, because UV irradiation inhibits protein synthesis (fig. S1) and minigene products are rapidly degraded within the cells (12), it remained possible that cells actively synthesizing minigene products could function in cross-priming. We eliminated this possibility by using a human cell line that produced cytosolic Ova257-264 from a transfected plasmid. These cells failed to induce a CD8+ T cell response, whereas control cells transfected with a plasmid expressing a metabolically stable chimeric protein (GFP-S) (Fig. 1) were immunogenic (Fig. 3E).

These experiments indicated that cells expressing stable proteins were capable of cross-priming, whereas those expressing minimal peptides were not. Minimal peptides (8 to 11 amino acids) are not, however, typically generated by proteasomes, which predominantly generate longer peptides (13, 14). To investigate the immunogenicity of cells producing natural proteasome products from a full-length protein, we infected P815 cells with a rVV chimeric protein designed to be rapidly degraded (R-NP-S-GFP) (Fig. 1). This virus is highly immunogenic when used to infect mice (Fig. 4A); however, cells expressing R-NP-S-GFP were unable to cross-prime unless they were first treated with lactacystin to prevent its degradation (Fig. 3B). This clearly demonstrates that natural proteasome products were not able to cross-prime in this system.

Fig. 4.

Rapidly degraded proteins are not substrates for cross-priming. Mice were immunized with rVV expressing the protein indicated (A) or P815 cells infected for 12 hours with rVVs expressing the gene product indicated (B). Six days after immunization, numbers of responding splenic Ova257-264-specific CD8+ T cells were determined by their expression of IFN-γ. Where indicated, 20 μM lactacystin was added to cells 1 hour after addition of virus and was maintained throughout the 12-hour infection. Averages from three mice for each group. This experiment was repeated three times with similar results.

Our findings, in conjunction with those of others (15, 16), demonstrate that cross-priming in vivo is based on the transfer of proteins, rather than peptides, as postulated by the chaperoned-peptide cross-priming hypothesis (5). This also extends a prior demonstration that cross-priming does not require transport of peptides into the ER by the transporter associated with antigen processing (17). Although these findings do not eliminate the possible participation of molecular chaperones in cross-priming, they do indicate that a putative chaperone ligand needs to achieve a sufficiently high steady-state level within cells. We further suggest that a role for chaperones in ferrying antigens to specific receptors on pAPCs might be superfluous in circumstances in which antigens are acquired through the process of phagocytosis of debris from dead or dying cells (18, 19). Consistent with this idea, a number of studies indicate that simply aggregating protein antigens (in the absence of molecular chaperones) greatly enhances their immunogenicity [reviewed in (20)].

It is important to recognize that our conclusions are strictly based on in vivo studies and do not address mechanisms of cross-presentation that occur in vitro, where solid evidence has been presented recently for the activities of proteasome-generated peptide products (21) and peptide-loaded chaperones (6). Further, our findings do not preclude therapeutic applications of molecular chaperones for inducing CD8+ T cells (5). Our study does, however, suggest a potentially important design principle for gene-based vaccines. Thus, vaccine antigens presented via direct priming should be engineered to maximally generate class I binding peptides (22), whereas vaccine antigens presented via cross-priming should be engineered for maximum metabolic stability.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article