Naturally Occurring TAP-Dependent Specific T-Cell Tolerance for a Variant of an Immunodominant Retroviral Cytotoxic T-Lymphocyte Epitope

Abstract

Upon immunization and restimulation with tumors induced by the endogenous AKR/Gross murine leukemia virus (MuLV), C57BL/6 mice generate vigorous H-2K^b-restricted cytotoxic T-lymphocyte (CTL) responses to a determinant (KSPWFTTL) derived from the p15E transmembrane portion of the viral envelope glycoprotein. By contrast, the highly homologous determinant RSPWFTTL, expressed by tumor cells induced by Friend/Moloney/Rauscher (FMR) MuLV, is not immunogenic, even when presented to the immune system as vaccinia virus-encoded cytosolic or endoplasmic reticulum (ER)-targeted minigene products. Such minigene products are usually highly immunogenic since they bypass the need for cells to liberate the peptide or transport the peptide into the ER by the transporter associated with antigen processing (TAP). Using KSPWFTTL-specific CTLs that cross-react with RSPWFTTL, we previously demonstrated that presentation of RSPWFTTL from its natural viral gene product is TAP dependent. Here, we show first that C57BL/6 mice express mRNA encoding RSPWFTTL but not KSPWFTTL and second that the ER-targeted RSPWFTTL minigene product is highly immunogenic in C57BL/6 mice with a targeted deletion in TAP1. These findings provide the initial demonstration of TAP-dependent tolerance induction to a specific self peptide and demonstrate that this contributes to the differential recognition of RSPWFTTL and KSPWFTTL by C57BL/6 mice.

Major histocompatibility complex (MHC) class I-restricted processing and presentation involve a series of coordinated events resulting in the cell surface display of endogenously synthesized viral or self peptides to cytotoxic T lymphocytes (CTLs) (7). Cytosolic peptides, largely generated by proteasomes, are shuttled into the lumen of the endoplasmic reticulum (ER) by the transporter associated with antigen processing (TAP), which is dedicated to this function. In the lumen of the ER, MHC class I heavy chains loosely assemble with β₂-microglobulin and acquire peptides transported from the cytosol. General-purpose and dedicated molecular chaperones play important roles in facilitating the folding of nascent class I molecules and in forming a complex between class I molecules and TAP (13, 22, 31). Class I molecules that acquire high-affinity peptides form a highly stable complex that is rapidly transported from the ER to the cell surface for perusal by CTLs.

Viruses have evolved numerous mechanisms for evading immune detection to persist in host cells. In the case of retroviruses, which exhibit extremely high mutation rates, sequence variation can enable the virus to evade CTL-mediated destruction (6). In theory, such mutations can affect any of the steps needed to generate immunogenic complexes: proteolytic liberation, TAP transport, class I binding, or TcR recognition.

C57BL/6 mice (H-2^b) have a low incidence of spontaneously occurring thymic leukemias (relative to AKR mice [H-2^k]) and characteristically generate vigorous H-2K^b-restricted CTL responses directed against tumor cells induced by AKR/Gross murine leukemia retroviruses (MuLV), the retroviruses that are carried in the genome of AKR mice or derived by recombination from these proviruses (8, 9, 12). Such CTLs (designated AKR/Gross MuLV-specific CTLs) predominantly recognize a determinant (KSPWFTTL) (10, 11, 27, 30) located within the p15E transmembrane region of the viral envelope protein. This determinant is present in nearly all endogenous ecotropic MuLV examined (4) and several mink cell focus-inducing (MCF) MuLV (3, 27).

AKR/Gross MuLV-specific CTLs are type-specific, efficiently recognizing Gross cell surface antigen-expressing tumor cells, but not Friend/Moloney/Rauscher (FMR) MuLV-induced tumor cell lines (8, 12). Interestingly, FMR (and MCF13) MuLV express the highly homologous determinant RSPWFTTL (which differs only by a highly conserved K-to-R substitution in position 1), yet FMR-induced tumors are very poorly recognized by AKR/Gross MuLV-specific CTLs (8, 27). Previous studies have demonstrated that the K-to-R substitution does not negatively affect peptide binding to H-2K^b or the turnover rates of peptide-stabilized cell surface H-2K^b complexes (17, 27). Nor does the substitution affect the TAP-mediated transport of the peptides into microsomes, although the P at position 3 greatly hinders the transport of both peptides (20, 21). There is evidence, however, that this substitution has a significant effect on cleavage of the peptide (and extended versions) by purified 20S proteasomes which cleave RSPWFTTL between R and S while sparing the K-to-S junction (21). Based on the fragments generated from a 26-mer synthetic peptide precursor and the negative impact of P at position 3 on TAP-mediated transport, it was proposed that KSPWFTTL was produced by proteasomes as an intermediate with two or three NH₂-terminal flanking residues that were trimmed in the ER, while the presence of R at position 1 prevents the generation of this precursor (21). In support of this model, Sijts et al. reported that C57BL/6 mice were capable of responding to synthetic RSPWFTTL peptide emulsified in incomplete Freund's adjuvant (27), suggesting that the poor immunogenicity of RSPWFTTL reflects a defect in antigen presentation and not in the CTL repertoire.

Attempts by our laboratory to elicit RSPWFTTL-specific CTLs, by immunizing C57BL/6 mice with synthetic RSPWFTTL peptide emulsified in incomplete Freund's adjuvant (17), have been unsuccessful. Interestingly, the only CTL line induced by the synthetic RSPWFTTL preparation proved to recognize a minor contaminant present in the preparation. The poor immunogenicity of RSPWFTTL was confirmed by expression as minigene cytosolic or ER-targeted gene products expressed by recombinant vaccinia virus (rVV) or Sindbis virus vectors that were used to immunize mice (4, 17).

Our findings suggested that limitations in the TcR repertoire contribute to the poor immunogenicity of RSPWFTTL. Given that C57BL/6 mice possess many proviruses in their genome, perhaps including RSPWFTTL-encoding MuLV, it is plausible that RSPWFTTL-specific CTLs are deleted or silenced by a tolerance mechanism. In the present study we provide strong support for this hypothesis.

The ability to elicit B6-derived secondary AKR/Gross MuLV (KSPWFTTL; “peptide 12”)-specific CTLs, following in vivo priming with synthetic peptide or via a variety of endogenous sources of KSPWFTTL, including rVV-driven minigene expression, has been well documented (10, 11, 27, 30). Our laboratory, however, has been unable to elicit RSPWFTTL (“peptide 13”)-specific CTLs from B6 mice by these approaches (4, 17) despite demonstrating the ability to improve target cell surface presentation of RSPWFTTL in vitro by targeting to the ER with signal-sequence (ss)-tagged minigene constructs (Vac 13ss) (17). Figure 1 demonstrates that B6 mice immunized with Vac 12ss (expressing KSPWFTTL) could easily generate secondary KSPWFTTL-specific CTLs when the responding splenocytes were stimulated in vitro with MC57 cells infected with recombinant Sindbis virus expressing KSPWFTTL (Sin 12; data not shown) (17) or KSPWFTTL coupled to an ER-targeting sequence (Sin 12ss). When the same approach was adopted to obtain RSPWFTTL-specific CTLs, however, by restimulation of splenocytes from B6 mice immunized with Vac 13ss, RSPWFTTL-specific CTLs could not be obtained. However, Vac 13ss targeting of the RSPWFTTL epitope into the ER did allow for its cell surface presentation on MC57 target cells in vitro, as demonstrated by the clear susceptibility of MC57 cells to the cross-reactive recognition of CTLs raised by secondary KSPWFTTL stimulation (Fig. 1), as we have previously reported (17). The extent of lysis of ER-targeted RSPWFTTL expressing target cells by KSPWFTTL-specific CTLs is variable, but frequently is quite high, as shown in Fig. 1.

FIG. 1.

C57BL/6 mice generate KSPWFTTL-specific, but not RSPWFTTL-specific, CTLs. C57BL/6 mice were immunized with 3 × 10⁷ PFU of the indicated rVV (intraperitoneally) and responding splenocytes stimulated in vitro with MC57 cells infected with various recombinant Sindbis viruses expressing either KSPWFTTL, RSPWFTTL, or the negative control, chloramphenicol transferase (CAT). Spontaneous release values for MC57 target cells ranged from 7.9 to 13.4%. This experiment was repeated two times with similar results.

To pursue the possibility that endogenously expressed RSPWFTTL serves to tolerize RSPWFTTL-specific CD8⁺ CTLs in B6 mice, experiments were first designed to determine whether the RSPWFTTL peptide is expressed in lymphoid tissues. While B6 mice are known to contain multiple endogenous proviruses in their genome, the expression of individual gene products from proviral genomes in given tissues is not well characterized. In Fig. 2, we examined lymphoid tissue, as self antigens are known to be more tolerogenic when expressed by lymphoid cells. To detect the expression of the RSPWFTTL- containing viral peptide precursor, p15E (27, 30), reverse transcriptase (RT)-PCR was performed on mRNA derived from spleen and thymus. The PCR was positively controlled by including extracted genomic DNA to detect the presence of integrated proviral p15E and also included primers used in separate reactions to amplify a ubiquitously expressed message, β-actin, to control for differences in RNA extraction and cDNA synthesis.

FIG. 2.

Confirmation of RSPWFTTL expression in lymphoid tissue by RT-PCR. Total mRNA was extracted from either spleen or thymic tissue from 6- to 8-week-old C57BL/6 or C57BL/6 TAP1-deficient mice for reverse transcription into cDNA and PCR amplification. The PCR products, generated by using cycling conditions programmed at 95°C for 1 min, 55°C for 1 min, and 72°C for 2 min for 25 cycles, were verified for the predicted size by gel electrophoresis, purified, and directly sequenced to confirm the presence of the correct TM134-141 sequences. RT-PCR utilizing the primers 623/7499 (GTACGGGATAGCATGGCCAAACTTAGAGAA) and 623/7694 (CTACCGAAATCCTGTCTTTGATAAACTG), spanning the region encoding KSPWFTTL, could not detect the 225-bp PCR product representative of KSPWFTTL mRNA expression (A). The primers p13B (AGATCCCCTTGGTTTACCACCTTG) and LTR3GEN (TACAGAAGCGAGAAGCGAGC) (19), however, were able to detect the 662-bp PCR product representative of RSPWFTTL mRNA expression (B). The RT-PCRs were positively controlled by including a reaction containing B6 genomic DNA isolated from the spleen. Lanes −RT, conditions of cDNA synthesis where RT was not included; lanes +RT, RT was present. This experiment was performed two times, with similar results.

As depicted in Fig. 2, DNA sequences predicted to encode either KSPWFTTL (Fig. 2A) or RSPWFTTL (Fig. 2B) were readily detected in genomic DNA isolated from the spleens of B6 mice. This is consistent with the presence of multiple endogenous proviruses in all inbred mouse strains. Although a KSPWFTTL DNA sequence was detected, we failed to detect mRNA sequences encoding KSPWFTTL in either the spleen or thymus (Fig. 2A). In contrast, RSPWFTTL-specific mRNA was easily detected in these tissues from B6 mice (Fig. 2B). The identity of this mRNA was confirmed by directly sequencing the RT-PCR product (data not shown). These data demonstrate that RSPWFTTL mRNA is expressed in B6 splenic and thymic tissues, whereas KSPWFTTL mRNA is undetectable. This is consistent with the contribution of tolerance to the discrepancy between the immunogenicities of RSPWFTTL and KSPWFTTL. To determine whether mutation of the tap1 gene had any effect on the expression of RSPWFTTL, mRNA extracted from the spleen and thymus of B6 TAP1-deficient mice (see experiments below) was also assessed. Again, mRNA encoding RSPWFTTL was readily detected (Fig. 2B).

The presentation of RSPWFTTL (as well as KSPWFTTL) is TAP dependent (17). Since TAP functions during thymic selection, we reasoned that TAP1^−/− mice would not present RSPWFTTL and could respond to RSPWFTTL if such CTLs were among those that are positively selected despite the absence of TAP1 (1, 25, 26). To enable the generation of RSPWFTTL-K^b complexes in TAP-deficient antigen-presenting cells (APCs), we primed mice with rVVs expressing ER-targeted versions of KSPWFTTL (Vac 12ss) or RSPWFTTL (Vac 13ss). Two weeks following immunization, memory CTL activity was determined by in vitro restimulation with synthetic peptides.

We initially examined (C57BL/6J × 129/Sv)F₂ TAP1-deficient mice. Immunization with VV 13ss resulted in the generation of memory CTLs specific for RSPWFTTL as demonstrated by their lysis of human T2.K^b transfectant target cells infected with VV 13ss but not a control VV (Vac 65) (Fig. 3A). In the same experiment, TAP1^+/+ (C57BL/6J × 129/Sv)F₁ mice failed to generate an RSPWFTTL-specific response, although they were perfectly capable of responding to KSPWFTTL (Fig. 3B and C). This pattern of results was confirmed in two additional experiments, including the demonstration that the control TAP1^+/+ F₁ mice were able to generate AKR/Gross MuLV-specific CTLs following secondary KSPWFTTL stimulation, and these CTLs showed substantial cross-reactive lysis of Vac 13ss-infected T2.K^b target cells, similar to the results shown for B6 antiviral CTLs in Fig. 1. In short, only TAP1-deficient mice immunized with Vac 13ss responded to RSPWFTTL (Fig. 3A).

FIG. 3.

RSPWFTTL-specifics CTLs can be induced in (C57BL/6J × 129/Sv)F₂ TAP1-deficient mice. (C57BL/6J × 129/Sv)F₂ TAP1-deficient (A) and (C57BL/6J × 129/Sv)F₁ (B) mice were immunized with rVV expressing RSPWFTTL coupled to an ER-targeting sequence (Vac 13ss). Secondary stimulation of responder splenocytes was conducted in vitro with synthetic RSPWFTTL. (C57BL/6J × 129/Sv)F₁ (C) mice were alternatively immunized with rVV expressing KSPWFTTL coupled to an ER-targeting sequence (Vac 12ss). Secondary stimulation of responder splenocytes was conducted in vitro with synthetic KSPWFTTL. Human T2.K^b transfectant target cells, which were used to decrease “background” CTL lysis of uninfected target cells (due to increased recognition of class I molecules presenting self peptides as a result of the altered thymic selection processes in TAP^−/− mice), were infected with either Vac 12ss, Vac 13ss, or control Vac 65 at a multiplicity of infection of 10:1 4 h before the addition of effector CTLs. Spontaneous release values for T2.K^b target cells ranged from 13.9 to 15.2%. This experiment was conducted three times, with similar results.

These observations were confirmed in fully backcrossed C57BL/6J TAP1-deficient mice. In this experiment, we additionally determined that memory in vitro RSPWFTTL peptide-restimulated CTLs from Vac 13ss-infected TAP1^−/− mice lysed both target cells that were infected with Vac 13ss, but not Vac ES-OVA_257-264 (Fig. 4A), and targets pulsed with synthetic RSPWFTTL, but not the ovalbumin-derived H-2K^b-restricted SIINFEKL immunodominant epitope (Fig. 4C). Thus, although these anti-RSPWFTTL/K^b CTLs were cross-reactive for T2.K^b cells infected with Vac 12ss or pulsed with the KSPWFTTL peptide (Fig. 4A and C), similar to the converse cross-reactivity of anti-KSPWFTTL/K^b CTLs of B6 (Fig. 1) or (C57BL/6 × 129/Sv)F₁ TAP^+/+ origin as discussed above, the CTLs raised against RSPWFTTL were not broadly reactive. As expected from our previous studies (4, 17) and the data of Fig. 1 and 3, synthetic RSPWFTTL peptide-restimulated splenocytes from normal C57BL/6 mice immunized with Vac 13ss failed to generate CTLs that recognized Vac 13ss-infected (Fig. 4B) or RSPWFTTL peptide-pulsed (Fig. 4D) targets.

FIG. 4.

RSPWFTTL-specific CTLs can be induced in C57BL/6 TAP1-deficient mice. C57BL/6J TAP1-deficient (A and C) and C57BL/6J (B and D) mice were immunized with Vac 13ss. Secondary stimulations of responding splenocytes were conducted in vitro with synthetic RSPWFTTL, and CTL activity was assessed by testing against T2.K^b targets infected with Vac 12ss, Vac 13ss, or Vac ES OVA_257-264, as indicated in the legend to Fig. 3, before the addition of effector CTLs (A and B). Alternatively, primed and restimulated splenocytes were tested against T2.K^b targets pulsed with either KSPWFTTL, RSPWFTTL, or SIINFEKL synthetic peptide (C and D). Spontaneous release values for T2.K^b target cells ranged from 0.9 to 4.9%. This experiment was repeated two times, with similar results.

We interpret these findings to mean that expression of type-specific MuLV p15E in C57BL/6 mice results in the TAP-dependent tolerance of CTLs specific for RSPWFTTL. We believe that this is the first demonstration of TAP-dependent tolerance to a defined self-determinant. Although we detected mRNA encoding RSPWFTTL in the thymus of B6 and B6 TAP-deficient mice, we cannot be certain that deletion occurs centrally and not in the periphery, as has apparently been demonstrated following infection of adult C57BL/6 mice with an exogenous MuLV expressing KSPWFTTL (5). It will be possible, however, to address the question of central versus peripheral tolerance in future experiments by thymic transplantation.

The specificity of tolerance induction for CTL responses to the RSPWFTTL variant of this retroviral epitope is somewhat curious, if not paradoxical, given the observed reciprocal cross-reactions observed for RSPWFTTL recognition by C57BL/6- derived anti-KSPWFTTL CTLs (Fig. 1) (17) and recognition of KSPWFTTL by anti-RSPWFTTL CTLs from C57BL/6 TAP1-deficient mice (Fig. 4). Our explanation is based on the evidence that in wild-type C57BL/6 mice, RSPWFTTL from endogenous sources is antigenic but not immunogenic, consistent with functional tolerance of high-affinity or -avidity RSPWFTTL-specific CTL clones. It is only in C57BL/6 TAP1-deficient mice where these high-affinity or -avidity CTL clones are not subject to tolerance induction and persist, allowing for CTL responses to be raised against (endogenous) RSPWFTTL stimulation. These RSPWFTTL-specific CTLs can also cross-reactively recognize KSPWFTTL (Fig. 4), analogous to the cross-reactive recognition of RSPWFTTL by anti-KSPWFTTL CTLs (Fig. 1).

Regarding the originating source of RSPWFTTL, the only reported ecotropic provirus in C57BL/6 mice is emv-2, which has not been molecularly cloned or sequenced (15, 29). RSPWFTTL may be present in the emv-2 envelope. Alternatively, emv-2 sequences may contribute to the formation, in C57BL/6 mice, of other ecotropic or polytropic infectious recombinant MuLV that express RSPWFTTL. For example, our laboratory has reported that the BM5 ecotropic helper MuLV (i.e., from the LP-BM5 retroviral complex that causes murine AIDS), which originates from C57BL/6 mice, expresses the RSPWFTTL epitope (4). Other endogenous MuLV may serve as a source of expressed RSPWFTTL. Endogenous xenotropic Bxv-1 MuLV, for example, is known to encode RSPWFTTL, and its DNA is detectable in C57BL/6 mice (28). Since Bxv-1 does not infect murine cells due to its xenotropic host range, Bxv-1 provides the donor sequences that encode RSPWFTTL, such as in the recombinant MCF-13 MuLV that was previously shown to be poorly recognized by KSPWFTTL-specific CTLs (3).

In summary, we have provided strong evidence supporting a TAP-dependent, specific tolerance mechanism as a basis for nonresponsiveness to the RSPWFTTL variant epitope in C57BL/6 mice. These findings do not exclude the possibility that inefficient liberation of RSPWFTTL from p15E (21) limits its antigenicity in RMA or other cell lines used as APCs in vitro. Thus, in vivo tolerance induction leading to the inability of B6 mice to generate high-affinity, RSPWFTTL-specific antiviral CTLs to endogenous sources of this peptide, as discussed herein, is compatible with additional mechanistic defects in the presentation of RSPWFTTL/K^b complexes for CTL-mediated recognition. Indeed, using proteasome in vitro reconstitution assays and/or following RSPWFTTL/K^b expression by use of cross-reactive CTLs raised to KSPWFTTL, evidence has been provided not only for poor processing (21) but also for relatively inefficient TAP-mediated transport (17) for RSPWFTTL compared to that for KSPWFTTL. Our results here do suggest, however, that generation of the RSPWFTTL peptide in vivo is not limiting for inducing tolerance in a large fraction of RSPWFTTL-specific CTLs.

ADDENDUM IN PROOF

Regarding the MuLV source of the tolerizing RSPWFTTL-encoding sequences, Li et al. (M. Li, X. Huang, Z. Zhu, and E. Gorelik, J. Virol. 73:9178–9186, 1999) have recently sequenced emv-2 and determined that the p15E envelope specifies the KSPWFTTL version of the CTL epitope, thus suggesting that RSPWFTTL expression originates from nonecotropic MuLV sequences.