Cross-priming of CD8+ T cells by viral and tumor antigens is a robust phenomenon



"Cross-priming" refers to the activation of naive CD8+ T cells by antigen-presenting cells that have acquired nominal antigens from another cell. The biological relevance of cross-priming of CD8+ T cells has recently been challenged (Zinkernagel, R. M., Eur. J. Immunol. 2002. 32: 2385–2392), on the basis that responses are weak or poorly quantitated, and the determinants recognized are undefined. Here we show that cross-priming is a robust process that elicits vigorous primary responses to multiple peptides in two well-defined systems. Our findings support the relevance of cross-priming in CD8+ T cell responses to viruses and tumor cells, and demonstrate that cross-priming elicits CD8+ T cells to determinants generated by the endogenous processing pathway.


Influenza A virus


Intracellular cytokine staining


Professional APC


Peritoneal exudate cells


Complete RPMI medium


T antigen


CD8+ T cells

1 Introduction

Shortly after the discovery of the MHC-restricted nature of CTL responses by Zinkernagel and Doherty 1, Bevan reported an interesting exception 2. Injection of mice with MHC-mismatched cells expressing minor H antigens primed self-MHC-restricted CTL responses specific for the immunizing minor H antigens. Bevan termed this phenomenon "cross-priming". Soon after, Trinchieri et al. reported MHC-independent priming of MHC-restricted CTL specific for simian virus 40 (SV40)-transformed cells 3.

We now recognize that the CTL of yore represent CD8+ T cells (TCD8+), and that these cells recognize short peptides bound to MHC class I molecules. Current immunology dogma dictates that activation of naive TCD8+ requires costimulatory signals that are uniquely provided in vivo by professional APC (pAPC), principally dendritic cells (DC). DC have the capacity to process and present cell-associated exogenous antigens to TCD8+4, providing a potential mechanism for cross-priming. Cross-priming can be rationalized as a means for the immune system to prevent viruses from avoiding TCD8+ responses by simply not expressing their genes in pAPC.

The contribution of cross-priming to the generation of anti-viral TCD8+ responses remains an open and important question that constitutes a thorny experimental problem. Ideally, it would be possible to selectively and completely block direct priming of viral antigens and qualitatively and quantitatively characterize anti-viral TCD8+ responses. (Even this would not unequivocally establish the relative contribution of cross-priming due to the possibility that TCD8+ activation by cross-priming is enhanced in the absence of direct priming.) Unfortunately, there is no surefire way of specifically blocking direct priming that can be generally applied to virus systems.

Weighing the contribution of cross-priming then must be based on results from less direct approaches. One important indirect approach is to measure the ability of virus-infected cells lacking restricting MHC molecules to cross-prime for anti-viral TCD8+ responses. At the very least, cross-priming should elicit a vigorous TCD8+ response to many of the determinants recognized by TCD8+ induced following in vivo infection. Based on the absence of such evidence in the literature, Zinkernagel has questioned the relevance of cross-priming in anti-viral responses, and even to tumor cells 5. He takes particular aim at the questionable inactivation of virus when infected cells are used for cross priming, and the absence of direct quantitation of TCD8+ responses ex vivo to defined determinants. In this communication we directly address these concerns by quantitating primary TCD8+ responses following immunization of mice with viral antigen-expressing cells that fail to generate self-MHC peptide complexes due to antigen processing defects or the absence of the restricting MHC allomorph.

2 Results and discussion

2.1 Cross-priming by influenza A virus-infected cells

We studied cross-priming by immunizing animals with influenza A virus (IAV)-infected cells via i.p. injection and quantitating primary responses ex vivo via intracellular IFN-γ staining. Responding TCD8+ were activated using APC sensitized with synthetic peptides corresponding to the three most dominant determinants in H-2b mice; PA224–233, NP366–374, and PB1-F262–706, 7 (Table 1). To prevent infection of host cells by input or progeny IAV released from infected cells, we mixed infected cells with a hemagglutinin-specific mAb with potent neutralizing activity. (Note that even in the absence of neutralizing antibody, progeny virus is very unlikely to be a problem, since IAV replicates poorly in non-epithelial cell lines.) As seen in Fig. 1A, the mAb was used at a concentration sufficient to block induction of TCD8+ by a large dose of infectious IAV. Under the same conditions, IAV-infected syngeneic cells elicited vigorous peritoneal TCD8+ responses against the two immunodominant determinants, demonstrating that the mAb treatment does not nonspecifically suppress TCD8+ responses to IAV. Antibody treatment was used in subsequent experiments to ensure that TCD8+ induction reflects bona fide cross-priming.

Since the MC57G cells used in this experiment are not pAPC (they are fibroblasts), it is likely that their immunogenicity is due to cross-priming. This particularly applies to PA224–233, which, as shown below, is not presented by MC57G cells. To further reduce the possibility for direct priming, we immunized mice with IAV-infected fibroblasts (1E12 cells 8) generated from a TAP1-knockout mouse. These cells, like TAP2-deficient RMA/S cells used in the same experiment, fail to detectably present any of the three determinants to TCD8+ (Fig. 1C). By contrast, MC57G cells activated nearly all of the NP366–374- and PB1-F262–70-specific TCD8+. As previously described 6, PA224–233 is not presented by MC57G cells, probably due to a requirement for immunoproteasomes 9. Notably, each of the three TCD8+ populations recognizes RMA/ S cells pulsed with pM concentrations of the respective peptides (Fig. 1D). Despite their inability to present IAV peptides to these highly sensitive TCD8+ (it can be calculated that about ten complexes are generated at the lowest peptide concentrations capable of sensitizing target cells), IAV-infected TAP–/– cells still induced a robust peritoneal response to all three determinants (Fig. 1B).

To remove all doubts about the class I-independent nature of cross-priming with IAV-infected cells, we immunized mice with IAV-infected cells derived from H-2d mice. As seen in Fig. 2, despite the coexisting strong alloreactive response (evinced by the response against the non-infected cells used for priming), immunization with either IAV-infected LTA-4 cells, and to a lesser extent A20 cells, resulted in extremely robust local and splenic TCD8+ responses that compare favorably to the response elicited by MC57G cells, and are of similar magnitude to responses elicited by IAV itself 10. We are at a loss to explain the differential abilities of LTA-4 and A20 cells in cross-priming and differences in relative cross-priming of different determinants in different populations (e.g. poor response to PB1-F262–70 in spleen). We note, however, that this provides a potential handle for future investigations into the mechanism of cross-priming.

Figure 1.

Cross-priming by IAV-infected TAP–/– cell line. (A, B) B6 mice were primed with IAV (A), or IAV-infected MC57G (A) or TAP–/– (B) cells (Table 2), all incubated with anti-HA antibody prior to injection to neutralize infectious virus. Seven days later, the numbers of responding TCD8+ specific for the indicated peptide present in peritoneal washout cells were determined by ICS. Each data point represents the average for three individuals. (C, D) The IAV-infected cells indicated (Table 2) were tested for recognition by TCD8+ lines specific for the peptides indicated (C). Note that PA224–233 is not detectably presented by MC57G cells. The sensitivities of the TCD8+ lines were determined by titrating the concentration of synthetic peptides needed for TCD8+ activation using low-temperature-induced RMA/S cells as APC.

Table 1. Peptides used in this study
Tag206 – 215206SAINNYAQKLDb
Tag223 – 231224CKGVNKEYLDb
Tag404 – 411404VVYDFLKCKb
Tag489 – 497489QGINNLDNLDb
Figure 2.

 Cross-priming by IAV-infected allogeneic cell lines. B6 mice were primed i.p. with anti-HA antibodies mixed with IAV-infected syngeneic (MC57G) or allogeneic cell lines (Table 2). Seven days later, the numbers of responding TCD8+ specific for the indicated experimental, control (OVA 55), or no peptide ("nil") present in spleen or peritoneal washout cells were determined by ICS. TCD8+ populations were also tested against their own priming line to determine the magnitude of allogeneic responses and against MC57G as another specificity control. Each data point represents the average from three individuals.

2.2 Cross-priming of SV40 T antigen

We extended these results to SV40 T antigen (Tag)-expressing cells. These cells are transformed with sub-genomic fragments of SV40 and are incapable of producing SV40 virions due to the absence of information encoding viral structural proteins. This eliminates any possibility that cross-priming results from SV40 infection of host APC. Thanks to the efforts of Tevethia and colleagues, four determinants have been defined in Tag 11 (Table 1). Following infection with SV40, TCD8+ responses to these determinants fall into the typical Tag immunodominance hierarchy: anti-Tag404–411> anti-anti-Tag223–231 ≅ anti-Tag206–215 as determined by intracellular cytokine staining (ICS) 7 days following immunization using either splenic or peritoneal TCD8+ (Fig. 3A, B). We next immunized mice with MHC-mismatched SV40-transformed cells derived from either H-2d, H-2s, or H-2k×d mice and again measured local and splenic TCD8+ responses on day 7. This elicited extremely robust responses in the spleen and peritoneum (Fig. 3C, D), recapitulating the immunodominance hierarchy observed following infection with SV40 (Fig. 3A, B). As typically seen for responses to Tag following immunization with either SV40 or recombinant vaccinia virus encoding Tag 12, responses to Tag489–497 were not detectable above background levels.

We next examined the effector function of TCD8+ induced by cross-priming. Splenocytes and peritoneal exudate cells (PEC) from mice immunized 9 days previously with SV40-transformed syngeneic or allogeneic cells were assessed immediately ex vivo for their ability to lyse syngeneic SV40-transformed cells or EL4 cells sensitized with Tag peptides in a 51Cr-release assay. As seen in Fig. 4, both PEC and splenic cells exhibited ex vivo SV40 Tag-specific lytic activity following immunization with syngeneic or allogeneic transformed cells. As with ICS, responses were dominated by TCD8+ specific for Tag404–411. Responses to allogeneic cells ranged from 50% less lysis at a given effector-to-target ratio to slightly more lysis, but were generally comparable.

We extended these results to the classical cross-priming P→F1 scenario by injecting SV40 H-2d cells or H-2b cells into H-2d×b mice and measuring peptide-specific H-2b-restricted responses as described above (Fig. 5). Although responses were somewhat lower in this experiment, they were still easily detected ex vivo. Importantly, the magnitude of response under these conditions is similar to that obtained in the same experiment following immunization with SV40-transformed syngeneic cells.

Figure 3.

Cross-priming by Tag-expressing allogeneic cell lines. (A, B) B6 mice were primed by i.p. injection with SV40 (∼107 plaque-forming units). After 7 days, the numbers of responding TCD8+ specific for the indicated peptide present in spleen (A) or peritoneal washout cells (B) were determined by ICS (data averaged from two animals). (C, D) B6 mice were primed i.p. with 5×06 γ-irradiated (100 Gy) cells of the allogeneic cell lines indicated. Seven days later, spleen cells (C) and peritoneal washout cells (D) were assessed by ICS for their recognition of the specific peptides indicated or syngeneic SV40-transformed cells (C57.SV), syngeneic control cells (MC57G) that do not express Tag, or the allogeneic cells used for priming. Data points were averaged from three animals; bars represent standard deviations.

Figure 4.

Killing by cross-priming induced TCD8+. B6 mice were immunized i.p. with 2×107 syngeneic SV40-transformed (C57-SV) or allogeneic cells (KD2SV). Nine days later, spleen cells or PEC were tested for their abilities to lyse EL4 cells incubated with the synthetic peptide indicated or C57-SV cells. Spleen cells were used at an effector-to-target ratio of 300:1; similar patterns of lysis were generated using splenocytes at 100:1 and 30:1. PEC were used at an effector-to-target ratio of 50:1.

Figure 5.

P→F1 Tag cross-priming. H-2b×d F1 mice were immunized i.p. with 5×106 syngeneic SV40-transformed (C57-SV) or allogeneic cells (KD2SV). Seven days later, the numbers of responding TCD8+ specific for the indicated peptide present in spleen was determined by ICS.

2.3 Conclusions

These data demonstrate that cross-priming of TCD8+ responses to viral antigens is a robust process that occurs in multiple antigens provided by multiple cell lines. Priming can be detected directly ex vivo in normal (i.e. non-TCR transgenic) mice either by ICS or by lytic activity. Up to 8% of responding TCD8+ in PEC induced by i.p. injection of allogeneic cells recognize defined protein or peptide antigens. Cross-priming does not require immunization with great quantities of cells: we show robust cross-priming following transfer of 5×106 cells, which is far lower than would occur during physiological anti-tumor responses, and also far lower than the number of infected cells following viral infections. Cross-priming by SV40-transformed cells demonstrates that the phenomenon is not due to transfer of virus, since these cells do not express SV40 virion structural proteins and are incapable of producing virus. These findings directly refute the criticisms directed by Zinkernagel at the physiological significance of cross-priming (summarized in Table 1 in 5).

Importantly, in conjunction with other findings we recently reported 13, these data demonstrate that TCD8+ induced by cross-priming not only recognize the determinants generated by the endogenous class I antigen processing pathway, but also recapitulate the immunodominance hierarchy. The cell biological basis for this similarity remains to be determined, but the functional significance of this overlap is unmistakable: it enables the immune system to generate by either direct priming or cross-priming, effector TCD8+ that recognize endogenously processed determinants.

Three factors could contribute to the evolutionary fitness of cross-priming: first, thwarting the ability of viruses to evade TCD8+ surveillance by simply failing to synthesize their gene products in pAPC; second, enabling surveillance of tumors, a concept recently resurrected based on findings with knockout mice with severely compromised immunity 14; third, serving as a mechanism for tolerizing TCD8+ to self antigens not synthesized by pAPC (cross-tolerance) 15. It is high time to weigh the biological significance of these factors.

3 Materials and methods

3.1 Cells

The various cell lines used are shown in Table 2. All cells were cultured in RPMI 1640 containing 10% fetal calf serum, 50 μM β-mercaptoethanol, antibiotics, and 2 mM glutamine (complete RPMI medium, RP-10). RMA/S cells were cultured at 26°C overnight to increase cell surface peptide-receptive class I molecules for the assay shown in Fig. 1. For culturing TCD8+ lines, briefly, 3×107 splenocytes were cultured with 1.5×105 irradiated (100 Gy) peptide-pulsed APC, in six-well plates with RP-10 containing 10 U/ml of recombinant human IL-2. Media were changed every 3–4 days. Activated viable T cells were collected through Ficoll-Hypaque gradients.

Table 2. Cell lines used in this study
Cell lineDescriptionH-2 type of donor cells
kxd SVC3H × BALB/c SV 40 transformantH-2k × H-2d
SS SVB10.S mouse SV 40 transformantH-2s
KD2 SVB10.D2 SV 40 transformantH-2d
C57 SVB6 SV 40 transformantH-2b
1E12TAP1 mutant T cell lymphomaH-2b
A20BALB/c B cell lymphoma cell lineH-2d
MC57GB6 metyhlcholanthrene transformantH-2b
LTA-5CBA fibroblastsH-2k

3.2 mAb, virus

Fluorescein-labeled anti-IFN-γ, Cy-Chrome-labeled anti-CD8α, and purified anti-B220 were purchased from Becton Dickinson (Sunnyville, CA). All mAb were used at 1:100 dilution in PBS. Purified B220 antibody was used to coat M450 Dynal beads (Dynal, The Netherlands) for depleting B220+ cells as described 16. IAV (Puerto Rico/8/34) was grown in 10-day embryonic chicken eggs and used as infectious allantoic fluid. Concentrated SV40 virus was the generous gift of Satvir Tevethia (Penn State Medical School, Hershey, PA).

3.3 TCD8+ priming

For in vivo priming, 8–10-week-old female C57BL/6J mice were either injected i.p. with ∼600 hemaggutinating units IAV, or with 5×106 of infected or uninfected cell lines. In the case of priming with IAV-infected cells, the cells were infected for 4–5 h before irradiation (100 Gy). Cells were washed to remove free virus and resuspended in 0.5 ml of PBS containing 100 μl of anti-hemagglutinin ascites (H28-E23 17) to neutralize residual virus. The cells were then injected i.p. Seven days later, primed splenic cells and peritoneal cells were collected and directly assessed by ICS as described. Priming with irradiated SV40-transformed cells for ICS analysis was performed similarly.

3.4 Intracellular cytokine staining

Peptides were procured and characterized by the Biologic Resource Branch, NIAID (Rockville, MD). In each case, substances with the predicted mass constituted more than 95% of the material analyzed. Peptides were dissolved in DMSO at 1 mM and stored at –20°C. For ICS, splenic cells and peritoneal cells from primed animals were suspended in 200 μl RP-10 at 1.5×107–2×107/ml in round-bottom 96-well plates. Peptides were added to wells to a final concentration of 0.5 μM. "No peptide" addition wells were used as a control for background activation. TCD8+ were incubated with peptides for 2 h at 37°C and then for 4 h with brefeldin A (Sigma-Aldrich, St. Louis, MO) at 5 μg/ml. When TCD8+ lines were used, the 2-h pre-incubation step was omitted. Cells were then stained with Cy-Chrome anti-CD8α mAb on ice for 30 min, washed and fixed with 1% paraformaldehyde in PBS at room temperature for 20 min, then further stained with fluorescein-anti-IFN-γ in PBS containing 0.1% saponin (Sigma). Stained cells were analyzed on a FACSCalibur (Becton Dickinson) with live gate on the CD8+ cells. Normally, 100,000 cells were acquired and analyzed.

3.5 Cytotoxicity assay

Mice were injected i.p. with 2×107 SV40-transformed cells. Animals were sacrificed and spleen cells and PEC were prepared on day 9 post-injection. Spleen cells or PEC were pooledand used at indicated effector-to-target ratio(s) in a conventional cytotoxicity assay using 51Cr-labeled EL4 thymoma target cells (H-2b) sensitized with 100 nM Tag-derived peptides and seeded at 104 cells/well in 96-well U-bottom microtiter plates. The plates were centrifuged at 400×g for 5 min at the end of a 6- or 12-h incubation at 37°C. A 100-μl aliquot of supernatant was then harvested from each well and the 51Cr content of the samples was determined by γ-counting. Specific lysis of the target cells was determined using following formula: % specific lysis = [(ER–SR)/(TR–SR)] ×100, where ER (experimental release) is obtained from wells containing both effector and target cells, while SR (spontaneous release) and TR (total release) are determined from wells receiving only target cells and medium or a 1:7 dilution of cetrimide 3.5% (w/v), respectively. Each data point represents the average of quadruplicate samples; data pointsexhibited less than 10% standard deviation.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5