Cytolytic CD8⁺ T Cells Directed against a Cryptic Epitope Derived from a Retroviral Alternative Reading Frame Confer Disease Protection¹

Cytotoxic T lymphocytes play a critical role in virus elimination and disease protection in several retrovirus infections, including murine retroviruses and HIV (1, 2), by responding to processed viral Ag presented as 8- to 11-residue peptide epitopes in the context of the MHC class I molecule on APCs, and subsequently infected target cells. Pathogens continually challenge the immune system by developing immune evasion mechanisms, including sequence variation within and around epitopes (3, 4, 5). CD8 T cells must be able to circumvent this problem, presumably in part by better detection. In recent years, studies have documented the presence of cryptic CTL epitopes, so-called for their nontraditional methods of generation. Their existence suggests a broader functional TCR repertoire via an expanded pool of potential antigenic epitopes. Although cryptic epitopes may be present in small abundance relative to traditional epitopes, the extreme sensitivity of CTL may compensate to allow efficient TCR detection and triggering of effector function (6).

Atypical CTL epitopes may potentially derive from all levels of gene and protein expression, including alternative reading frames (ARFs),³ which are of particular interest to our studies (7, 8, 9, 10, 11, 12, 13, 14, 15, 16). Our laboratory has previously demonstrated that normally resistant BALB/c mice depleted of CD8 T cells by anti-CD8 mAb treatment in vivo were rendered susceptible to LP-BM5 retrovirus infection, thus showing CD8⁺ T lymphocytes to be necessary for disease resistance (17). The retrovirus complex, comprised of a mixture of replication-incompetent defective/pathogenic (BM5def) and replication-competent ecotropic helper (BM5eco) viruses, induces a disease termed murine AIDS (MAIDS), featuring early nonspecific B and T cell activation and proliferation, and a subsequent progressive immunodeficiency (18, 19, 20). Characterization of CD8⁺ CTL raised against the viral gag gene showed their specificity for an immunodominant K^d-restricted antigenic epitope (SYNTGRFPPL, nt 653–682), derived from the +1 nt alternative translational open reading frame (ORF) 2 of the LP-BM5 genome (21).

Because of several similar disease manifestations in HIV-induced AIDS and this MAIDS model (22, 23, 24), the recent finding that cryptic MHC class I-restricted epitopes have also been identified in HIV-1 ARFs (25) is especially relevant to our present study. Yet, despite evidence that nontraditional CTL epitopes are generated from frame shifting of HIV translational sequences, a possible in vivo protective role against AIDS for CTL directed against ARF-derived epitopes could not be directly addressed. In fact, a recent report that functionally scanned the influenza genome for ARF-encoded epitopes has questioned their biological significance. (26).

To study CTL-mediated in vivo protection from LP-BM5-induced disease, we used susceptible CD8-deficient BALB/c mice. We show here that CD8 knockout (KO) mice reconstituted with CTL specific for the viral ORF2-derived SYNTGRFPPL epitope resolved an infection by the LP-BM5 retrovirus mixture. Ultimately, the mice receiving such adoptive transfer were protected from disease. These results thus demonstrate the physiological and functional significance of an ARF-derived epitope in a retrovirus-induced disease model.

BALB/c mice (5–6 wk old) were purchased from the National Cancer Institute (Bethesda, MD). KO breeders were obtained as follows: CD8 KO from P. Stuart (Washington University, St. Louis, MO) and T. Mak (Ontario Cancer Institute, Toronto, Canada), and IFN-γ KO from J. Gorham (Dartmouth Medical School). Mice were maintained under specific pathogen-free conditions in the Dartmouth Medical School animal research facility and all experimental procedures were conducted in accordance with the Institutional Animal Care and Use Committee (institutional assurance no. A3259-01).

The mouse macrophage cell line J774A.1 was grown in DMEM supplemented with 5% FCS, 2 mM l-glutamine, 30 μg/ml penicillin, 20 μg/ml streptomycin, 33 μg/ml gentamicin, and 10 mM HEPES. P815B mastocytoma cells were grown in RPMI 1640, supplemented as above, and including 50 μM 2-ME. The M2 (from E. Usherwood, Dartmouth Medical School) and 1229A (21) CTL lines were maintained with 15 U/ml IL-2, and restimulated every 2 wk with 10⁶/ml irradiated syngeneic splenocytes (3000 rad) and 10 μg/ml M2, or SYNTGRFPPL, peptide, respectively. Cells were cultured in a humidified incubator at 37°C with 10% v/v CO₂. The “empty” vaccinia vector, Vac-65, and recombinant virus, Vac-DG, in which the gag gene of the LP-BM5 replication-defective virus was inserted into Vac-65, were propagated as described (27). The LP-BM5 retroviral isolate was generated from chronically infected G6 fibroblasts provided by H. C. Morse and J. W. Hartley (National Institutes of Health/National Institute of Allergy and Infectious Diseases, Bethesda, MD) (28).

The LP-BM5 ORF2 peptide (SYNTGRFPPL) was purchased from Invitrogen Life Technologies at >95% purity. The control gammaherpesvirus latent peptide M2_91–99 (GFNKLRSTL) was provided by E. Usherwood (Dartmouth Medical School).

For priming for polyclonal CTL generation, 10⁷ PFU of Vac-65 or Vac-DG were administered i.p. to wild-type (WT) BALB/c mice (27). Ag-sensitized splenic lymphocytes were isolated at 3–5 wk postpriming. For depletion of CD8⁺ T cells, lymphocytes were subjected to two rounds of negative immunomagnetic cell sorting using the autoMACS system (Miltenyi Biotec). Flow cytometry confirmed >99% depletion of CD8⁺ T cells. Polyclonal effector cell populations, both total and CD8-depleted populations, were generated as described above, and ORF2/SYNTGRFPPL-specific cytolytic activity of unfractionated or CD8-depleted anti-ORF2, or control anti-Vac-65, effector populations, was determined before transfer into CD8 KO recipients. Briefly, effectors were restimulated as previously described, and cytolytic activity was measured in a standard ⁵¹Cr release assay using P815B target cells incubated with SYNTGRFPPL or infected with Vac-DG or Vac-65 (27). For short-term maintenance of CTL, cultures were carried with 5 U/ml rIL-2 for an additional 3 days before reconstitution of CD8 KO mice. For transfer of the 1229A and control M2 CTL lines, both populations were assayed for CD8 and K^d-SYNTGRFPPL tetramer specificity, and for ORF2/SYNTGRFPPL-dependent IFN-γ production and cytotoxicity before transfer into recipients.

Control WT (MAIDS-resistant) or experimental CD8 KO (susceptible) mice were infected i.p. with 5 × 10⁴ PFU LP-BM5 virus. Three, 6, 9, and 12 days postinfection (p.i.), 6–8 × 10⁶ polyclonal effector CTL (∼3 × 10⁷ total) were transferred i.v. into CD8 KO recipients. For 1229A and control M2 CTL lines, 10⁶ monospecific effector CTL (4 × 10⁶ total) were transferred.

Lymphocytes secreting IFN-γ were quantitated following peptide restimulation in a standard ELISPOT assay (29). Briefly, 96-well Multiscreen HTS nitrocellulose plates (Millipore) coated with capture anti-mouse IFN-γ were plated with 5 × 10⁵ spleen cells, 10⁶ irradiated syngeneic BALB/c spleen cells (3000 rad), and 1 μg/ml stimulating SYNTGRFPPL, or M2, peptide. Spot enumeration was performed by Zellnet Consulting Services.

An allophycocyanin-labeled tetramer consisting of K^d folded with SYNTGRFPPL peptide was provided by the NIH Tetramer Core Facility (Atlanta, GA). Effector cells were incubated with an anti-CD16/CD32 Fc block (BD Pharmingen) for 10 min on ice, tetramer for 1 h at room temperature, then FITC-anti-CD8α (clone 53-6.7; BD Pharmingen) for 20 min on ice. Stained cells were analyzed on a FACSCalibur flow cytometer using CellQuest software (BD Biosciences).

For disease analysis by activational parameters, spleen weight was determined, and sera were assayed for total IgG2a and IgM in a standard ELISA (30). Immunodeficiency was measured with Con A and LPS mitogen-induced proliferation assays using isolated splenocytes (30). For viral load, BM5 defective and ecotropic gag genes were amplified from purified splenic RNA by real-time RT-PCR according to published protocol (31).

The expansion of ORF2/SYNTGRFPPL-specific CD8 T cells in BALB/c mice following 3–5 wk of in vivo rVac-DG priming was measured in splenic lymphocytes after peptide restimulation in vitro. CD8 T cell responses were primed against the gag gene of the LP-BM5 replication defective virus, which was previously inserted into an “empty” vaccinia vector, Vac-65, to generate the rVac-DG virus (27). As controls for detection of Ag-specific polyclonal CTL, use of a K^d-folded SYNTGRFPPL tetramer showed specific staining of the K^d-restricted viral ORF2-specific CTL line 1229A, but not the irrelevant K^b-restricted CTL clone 0.3.1 (32) directed against LP-BM5-induced tumor cells (Fig. 1,A). To enumerate the Ag-specific CTL in polyclonal effector cell populations, it was necessary to boost the response by secondary stimulation with SYNTGRFPPL peptide in vitro, because tetramer-positive cells were below the level of detection after in vivo priming only (data not shown). In a representative experiment, 23.5% of the total cell population following specific in vitro restimulation was stained by both anti-CD8 and the ORF2/SYNTGRFPPL-specific tetramer (Fig. 2 A); 41.7% when the tetramer-positive population was represented as a percentage of only CD8 T cells present within the total population. Control spleen cells primed and restimulated with Vac-65 as the vector control showed much lower levels of tetramer staining–5.7% of all cells, and 7.9% of all CD8 T cells. Thus, despite the possibility of a decreased frequency of CTL directed to this atypical ARF-derived epitope compared with standard ORF1-encoded epitopes in general, as discussed above, the expansion of epitope-specific CD8 T cells was substantial and indicated a strong response to this ORF2-encoded determinant. Additionally, a K^d control tetramer folded with the irrelevant gammaherpesvirus M2 peptide did not stain Vac-DG/SYNTGRFPPL-stimulated CD8 T cells (data not shown).

Effector CD8 T cells with specificity for SYNTGRFPPL, generated from Vac-DG-primed cells restimulated for either 24 h (ELISPOT) or 6 days (lysis) with peptide in the absence of IL-2, were also functional in vitro (Fig. 2, B and C). In response to SYNTGRFPPL stimulation, primed cells produced a robust IFN-γ response, in comparison to the minimal response detected when the irrelevant M2 peptide was instead used (Fig. 2,B). As expected, IFN-γ production was not observed when splenic lymphocytes from either CD8⁻ or IFN-γ-deficient Vac-DG-primed mice were stimulated in parallel with SYNTGRFPPL peptide. Additionally, because SYNTGRFPPL is a class I K^d-restricted epitope, any reactivity to peptide stimulation by CD4 T cells seemed unlikely. Indeed, the lack of an ELISPOT response in CD8^−/− preparations argued against a CD4 T cell response (Fig. 2,B). Vac-DG/SYNTGRFPPL-stimulated polyclonal CTL also demonstrated substantial specific lytic activity against target cells pulsed with peptide (Fig. 2 C) or infected with Vac-DG (data not shown). Cytolytic function was indeed attributed to CD8 T cells, as effector preparations depleted of CD8⁺ cells before in vitro restimulation were nonresponsive. Furthermore, the response was highly specific: following priming and restimulation with Vac-65, control polyclonal CTL with irrelevant specificity for vaccinia were unable to lyse peptide-loaded targets, despite their efficient lysis of Vac-65-infected targets (90% specific lysis at 4:1 E:T ratio; data not shown).

Disease-susceptible CD8 KO recipient mice were infected with LP-BM5 virus and reconstituted at 3 days p.i. by adoptive transfer of highly lytic anti-ORF2/SYNTGRFPPL polyclonal CTL populations. ORF2/SYNTGRFPPL-specific cytotoxicity was negligible with CD8-depleted effectors, and with control Vac-65-primed polyclonal CTL (Fig. 2). At 9 wk p.i., mice were assayed for a standard panel of disease parameters: splenomegaly and hypergammaglobulinemia represent early nonspecific virus-induced activation, and mitogen nonresponsiveness indicates subsequently established immunodeficiency (Fig. 3). Activational and immunodeficiency readouts were highly correlative, and clearly demonstrated the disease resistance of intact BALB/c mice, whereas CD8 deficiency converted the host to a susceptible phenotype (Fig. 3). Resistance was restored in CD8 KO mice, upon their reconstitution with ORF2/SYNTGRFPPL-specific polyclonal CTL. Complete disease protection was observed with all parameters; infected CD8 KO mice that received transferred specific CTL demonstrated no signs of nonspecific polyclonal activation, and furthermore, remained immunocompetent. Importantly, resistance was evident despite the infection preceding the adoptive transfer by 3 days, thereby emphasizing the efficiency of protection by anti-ORF2/SYNTGRFPPL CTL.

In addition, protection was dependent specifically on CD8 T cells, presumably those directed against the ARF-derived SYNTGRFPPL epitope. Following depletion of CD8 T cells from Vac-DG-primed splenocytes before peptide restimulation, the remaining cell preparation was unable to transfer resistance; rather, the CD8 KO host remained disease susceptible (Fig. 3). Disease protection was also highly specific. Thus, a highly lytic polyclonal population but with irrelevant Ag specificity, i.e., the Vac-65-primed anti-vaccinia vector response, did not render the CD8 KO mice resistant (Fig. 3).

To confirm that the CD8 T cells required for protection were only those directed against the SYNTGRFPPL ARF-derived epitope, we conducted a parallel adoptive transfer experiment with a CTL line homogeneous for K^d-SYNTGRFPPL-tetramer and CD8 positivity (Fig. 1,A). The 1229A CTL produced a robust in vitro IFN-γ response, compared with the irrelevant M2 line, after stimulation with the SYNTGRFPPL peptide (Fig. 1,B). Cytolytic activity was correlative with IFN-γ induction; 1229A CTL, but not M2 CTL, were highly lytic against SYNTGRFPPL-pulsed target cells (Fig. 1,C). Upon restimulation with the gammaherpesvirus M2 peptide, however, M2 CTL produced strong IFN-γ (Fig. 1,B) and lytic (Fig. 1 C) responses.

The in vivo functional capabilities of the 1229A, vs M2, CTL lines were examined in disease-susceptible CD8 KO recipient mice, again reconstituted at 3 days p.i. by adoptive transfer of the effector cell lines. All activational and immunodeficiency parameters demonstrated complete disease protection in susceptible mice reconstituted with the ORF2/SYNTGRFPPL-specific 1229A CTL line, but not with the irrelevant M2 line (Table I).

Virus expression assessed by quantitative RT-PCR was directly and highly correlative with disease parameters in mice exhibiting disease, and those protected from viral pathogenesis by adoptive transfer of ORF2/SYNTGRFPPL-specific CTL. Indeed, viral load was very low in adoptively transferred mice, suggesting that CTL-mediated protection is based on clearance of virus-infected cells and/or lack of subsequent rounds of infection (Fig. 4). As determined by separate quantitative RT-PCR analyses, expression of pathogenic BM5def and BM5eco helper viruses was nearly undetectable in hosts reconstituted with Ag-specific effector cells. In contrast, expression of the BM5def and BM5eco viruses was present at levels ∼114- or 33-fold higher, respectively, in disease- susceptible mice (both CD8 KO mice with no transfer, and CD8 KO mice receiving either CD8-depleted anti-ORF2/SYNTGRFPPL or control Vac-65-specific effectors). Furthermore, only after transfer of the 1229A, but not the M2, CTL line, did gag expression of both viruses become nearly undetectable, demonstrating directly that ORF2/SYNTGRFPPL-specific CTL were sufficient for viral clearance and subsequent disease resistance (data not shown).

Our findings substantiate the in vivo function of ARF-encoded epitopes in disease protection, and dispute doubts concerning their biological significance. Previously, the potential for protection by CTL specific for an ARF-derived tumor epitope was examined in a melanoma patient, who demonstrated partial regression of lung metastases following transfers of IL-2-expanded tumor-infiltrating lymphocytes (33, 34, 35). However, doses of cyclophosphamide and IL-2 were simultaneously administered; additionally, the patient developed cerebral metastases at 3 mo posttherapy (33). To our knowledge, the report herein describes the only physiological system in which polyclonal CTL directed against an ARF-derived epitope conferred protection against disease, and the first study showing that anti-ARF CTL, in general, can protect against retroviral disease. Despite the delay in adoptive transfer of effectors until 3 days p.i., protection was associated with complete or near-complete viral clearance, indicating that ARF-specific CTL can efficiently control viral infection. Furthermore, long-lived, potentially memory, CD8 T cells were detectable and able to recall an ORF2/SYNTGRFPPL-specific IFN-γ response well after resolution of the initial infection (data not shown).

The ability of anti-ARF CTL to protect against disease is even more noteworthy, considering that anti-ARF CTL can mediate this protection independently of other immune mechanisms. Indeed, transfer of the 1229A CD8⁺ CTL line, homogeneous for ORF2/SYNTGRFPPL specificity, into susceptible LP-BM5-infected CD8 KO mice was sufficient to restore disease resistance. Protection was also evident in CD8 KO recipients upon reconstitution with CD8⁺ CTL purified from Vac-DG/SYNTGRFPPL-stimulated polyclonal effectors (preliminary data, not shown). A role for CD4 T cells in disease protection was explored in parallel, also using polyclonal effectors, but either depleted of, or purified for, CD4⁺ cells, following in vitro restimulation. In either case, CD8 KO recipients remained disease susceptible (preliminary data, not shown), suggesting that CD4 T cells do not have an effector role in protection, in keeping with the sufficiency of anti-ARF CD8⁺ CTL in disease resistance.

Beyond the physiological relevance demonstrated here for an ARF-derived epitope, there are significant implications for cryptic epitopes as possible additional sources of antigenic peptides that can direct T cell immunity. By engineering a bicistronic transgene capable of encoding in parallel two model CTL epitopes, one cryptic and one traditional, Schwab et al. (36) showed that both CTL responses as well as self-tolerance could be induced by the cryptic peptide, despite its reduced expression relative to the conventional peptide translated simultaneously. These findings support the concept that T cell immune surveillance extends beyond traditional sources of antigenic peptides. Indeed, this report illustrates the importance of understanding the mechanisms by which cryptic peptides arise, and how they function in the immune system (37).

Although our findings suggest the ability to generate protective effector T cells is comparable between the ARF-derived SYNTGRFPPL and traditional epitopes, notably, there is a void in information on the level of expression of the cryptic ORF2 epitope relative to other conventional epitopes in this mouse retrovirus system. Currently, no other H-2^d-restricted antigenic peptides have been identified in the LP-BM5 retrovirus complex, let alone in the retroviral gag gene, in particular a peptide which is derived conventionally from ORF1. Thus, it is not possible to directly compare the expression levels and efficacy of traditionally vs nontraditionally generated CTL epitopes. Rather, further reagents and studies will be required to determine how peptide/MHC expression levels may impact CTL induction, and ultimately, protection against disease. The availability of the protective ORF2-specific CTL population, however, allows us to begin addressing questions regarding possible quantitative and qualitative differences of the induced anti-ARF CD8 T cell response. Of interest are the effector mechanisms underlying the observed CD8 T cell-mediated protection. Any reasons for differences in the effector molecules used by anti-ARF CD8 T cells are not apparent; however, the cryptic nature of ARF-encoded epitopes may allow for a diminished density of cell surface-presented ARF epitope-MHC class I complexes on APCs and/or target cells. As a result, some effector mechanisms, but not others, may be triggered. To what extent these mechanisms function in concert and/or in compensatory fashion, or perhaps even in a hierarchical order, remains to be determined.

As a correlate to our preliminary observation that IFN-γ-producing, transferred ORF2/SYNTGRFPPL-specific CD8 T cells were present many weeks after virus infection, these likely memory T cells were also detected in LP-BM5-infected, resistant WT BALB/c mice (data not shown). Highly specific responses from priming against the ORF2 epitope have also been demonstrated with lymphocytes isolated from WT mice in the early stages of infection (data not shown, Refs.17 and 21). These observations indicate that the adoptive transfer of Ag-sensitized effectors into susceptible CD8 KO recipients, thereby restoring resistance, is an appropriate system to study the kinetics and mechanisms of disease protection operative in the normally resistant intact host. Such studies will provide insight into the degree of viral clearance, and therefore, the potential for residual viral Ag as a determinant in the persistence or induction of anti-ARF effector or memory T cells, respectively. Indeed, evidence of sustained protective ability may support the use of ARF-derived epitopes in vaccines, the feasibility of which has been shown by the induction of effective antitumor protection following immunization with ARF-encoded polypeptides, and priming of hepatitis B virus-specific CD8 T cell responses with ARFs engineered into DNA vaccines (38, 39, 40). Our current report provides the physiological evidence for ARF epitope-induced polyclonal CTL in disease protection, and equally significant, the platform to understand the role of cryptic epitopes in establishing long-lasting protective immunity.

We thank Drs. J. D. Gorham, P. M. Stuart, and T. W. Mak for kindly donating BALB/c KO mice; Drs. H. C. Morse III and J. W. Hartley for providing G6 cells; and K. A. Green for generation of LP-BM5 virus. We also thank Dr. E. J. Usherwood for helpful discussions, and along with J. J. Obar, for providing the gammaherpesvirus-specific CTL clone and M2 peptide. We thank Dr. E. Ravkov and A. Stout of the NIH Tetramer Facility for providing tetramer reagents.

The authors have no financial conflict of interest.