Successful active immunization against cancer requires induction of immunity against self or mutated self Ags. However, immunization against self Ags is difficult. Xenogeneic immunization with orthologous Ags induces cancer immunity. The present study evaluated the basis for immunity induced by active immunization against a melanoma differentiation Ag, gp100. Tumor rejection of melanoma was assessed after immunization with human gp100 (hgp100) DNA compared with mouse gp100 (mgp100). C57BL/6 mice immunized with xenogeneic full-length hgp100 DNA were protected against syngeneic melanoma challenge. In contrast, mice immunized with hgp100 DNA and given i.p. tolerizing doses of the hgp100 Db-restricted peptide, hgp10025–33, were incapable of rejecting tumors. Furthermore, mice immunized with DNA constructs of hgp100 in which the hgp10025–27 epitope was substituted with the weaker Db-binding epitope from mgp100 (mgp10025–27) or a mutated epitope unable to bind Db did not reject B16 melanoma. Mice immunized with a minigene construct of hgp10025–33 rejected B16 melanoma, whereas mice immunized with the mgp10025–33 minigene did not develop protective tumor immunity. In this model of xenogeneic DNA immunization, the presence of an hgp100 heteroclitic epitope with a higher affinity for MHC created by three amino acid (25 to 27) substitutions at predicted minor anchor residues was necessary and sufficient to induce protective tumor immunity in H-2b mice with melanoma.
Induction of cancer immunity against self Ags can be difficult because of immune tolerance or ignorance. Immunity against a major class of self Ags, the differentiation Ags, can be induced by xenogeneic immunization, which is vaccination of a host from one species with DNA encoding an orthologous gene from another species (1, 2, 3, 4). Differentiation Ags are shared between cancer cells and their normal cell counterparts (5). The strategy of xenogeneic immunization has been successful, whereas immunization with the syngeneic differentiation Ag failed to induce immunity. Although it has been presumed that individual amino acid sequence differences in the xenogeneic protein create heteroclitic epitopes, which trigger immunity against the homologous self protein, this hypothesis has not been directly demonstrated.
Prototypical examples of differentiation Ags are glycoproteins shared by melanomas and melanocytes, categorized as the melanosomal differentiation Ags (6). Immunity against differentiation Ags is a form of autoimmunity that can mediate destruction of tumors and normal tissues (1, 2, 3, 4, 7, 8, 9, 10, 11). In fact, tumor-infiltrating lymphocytes from patients with melanoma most often recognize melanocyte differentiation Ags (12). Previous studies have shown that xenogeneic immunization against the melanosomal differentiation Ags tyrosinase-related protein-1 TRP-1/gp75 (3, 4), dopachrome tautomerase (1), and gp100/pmel 17 (2), either with DNA encoding the Ag or with recombinant protein, can result in the rejection of transplanted syngeneic melanomas in mice.
One mechanism that could mediate xenogeneic immunization is the creation of heteroclitic epitopes. A heteroclitic epitope describes an altered peptide that is a better agonist for inducing T cell responses than the native, unaltered peptide (13). This terminology has been applied to other altered peptides with a higher immunological potency than their unaltered counterparts (14, 15, 16). Heteroclitic peptides have increased potency either due to increased binding to MHC molecules (15, 17, 18, 19) or to increased agonist properties to stimulate TCRs (16).
We recently described xenogeneic DNA immunization of mice with the gene encoding human gp100 (hgp100)3 (2). The gp100 Ag is a melanosomal differentiation Ag expressed by melanomas and melanocytes (20). In this model, mice are protected against melanoma challenge, and this immunity correlates with T cell reactivity against the MHC class I Db-restricted epitope gp10025–33. Immunity after xenogeneic immunization against gp100 was achieved in mice deficient in MHC class II molecules, suggesting that immunity directed against MHC class I epitopes was sufficient without T cell help. In the present study, we assessed the role of hgp10025–33 after xenogeneic DNA immunization. We show that hgp10025–33 is a heteroclitic epitope that is both necessary and sufficient to induce autoimmunity and melanoma immunity in an H-2b mouse tumor model.
Materials and Methods
C57BL/6 female mice were acquired from the National Cancer Institute (Bethesda, MD) or The Jackson Laboratory (Bar Harbor, ME). Mice were maintained in a pathogen-free vivarium according to National Institutes of Health Animal Care guidelines under an institutional protocol reviewed and approved by the Animal Care Committee. All mice entered in the study were between 7 and 10 wk of age.
Cell lines and tissue culture
B16F10 (B16) is a pigmented mouse melanoma cell line of C57BL/6 origin provided by Dr. I. Fidler (M. D. Anderson Cancer Center, Houston, TX). This cell line has been maintained in our laboratory for tumor challenges by in vivo passage through mice, followed by in vitro expansion and storage at −70°C. The EL4 cell line is a C57BL/6 mouse lymphoma. Tumor cell lines were cultured as previously described (8).
hgp100-pWRG1644 was a gift of Dr. N. P. Restifo (National Cancer Institute). PCR-based site-specific mutagenesis was performed on hgp100 to create hgp100KVP→EGS-pWRG1644 and hgp10029(N→W)-pWRG1644 using the primers 5′-GCTGTGGGGGCTACAGAAGGATCCAGAAACCAGGACTG-3′ and 5′-CAGTCCTGGTTTCTGGATCCTTCTGTAGCCCCCACAGC-3′ (hgp100KVP→EGS) as well as 5′-GGGGGCTACAAAAGTACCCAFATGGCAGGACTGGCTTGG-3′ and 5′-CCAAGCCAGTCCTGCCATCTGGGTACTTTTGTAGCCCCC-3′ (hgp10029(N→W)). The cDNAs coding for hgp100, hgp100KVP→EGS, and hgp10029(N→W) were cleaved from the respective pWRG1644 plasmids using NotI, and each was cloned into the vector pWRG7077BEN (21). The inserts were then sequenced to assure that hgp100KVP→EGS and hgp10029(N→W) differed from hgp100 only at the intended sites. The oligonucleotides 5′-GATCCACCATGAGATACATGATCCTGGGCCTGCTGGCCCTGGCTGCA-3′ and 5′-GCCAGGGCCAGCAGGCCCAGGATCATGTATCTCATGGTG-3′ were annealed and ligated into pCR3 and digested with BamHI and PstI to make an expression vector coding for MRYMILGLLALAA. This construct incorporates most of the endoplasmic reticulum insertion sequence from the adenovirus E3/19kDa gene product (MRYMILGLLALAAVCSA), with a PstI site following the codon for the 13th amino acid, alanine. This vector was then expanded and sequenced. The remaining endoplasmic reticulum insertion sequence (VCSA) followed by minigenes coding for 1) hgp10025–33 (KVPRNQDWL) or mgp10025–33 (EGSRNQDWL), 2) the Db restriction epitope from the influenza matrix protein (Flu) (ASNENMETM), or 3) the Kb epitope from OVA (SIINFEKL) followed by a stop codon were ligated by annealing the respective primer pairs (5′-GTGTGCAGCGCCAAAGTACCCAGAAACCAGGACTGGCTTTGAC-3′ and 5′-TCGAGTCAAAGCCAGTCCTGGTTTCTGGGTACTTTGGCGCTGCACACTGCA-3′ (hgp10025–33), 5′-GTGTGCAGCGCCGCCTCCAACGAGAACATGGAGACAATGTGAC-3′ and 5′-TCGAGTCACATTGTCTCCATGTTCTCGTTGGAGGCGGCGCTGCACACTGCA-3′ (flu), and 5′-GTGTGCAGCGCCAGCATCATCAACTTCGAGAAGCTGTGAC-3′ and 5′-TCGAGTCACAGCTTCTCGAAGTTGATGATGCTGGCGCTGCACACTGCA-3′ (ova)) and cloning them into the minigene vector created above, after digestion with PstI and XhoI. Sequence fidelity was confirmed by DNA sequencing. All primers and oligonucleotides were obtained from Life Technologies (Gaithersburg, MD).
DNA and peptide immunization
DNA immunization by helium-driven gold particle bombardment via the gene gun has been previously reported (21). In brief, plasmid DNA was coated on 1.0-μm-diameter gold particles (Bio-Rad, Hercules, CA) and precipitated on bullets of Teflon tubing (McMaster-Carr, East Rutherford, NJ). Gold particles containing 1 μg of plasmid DNA were delivered to each abdominal quadrant of a mouse using a helium-driven gene gun (Accell; PowderJect Vaccines, Madison, WI) for a total of four injections. Each mouse was immunized three times at weekly intervals. Peptide immunization was performed by footpad injection with TiterMax emulsion (CytRx, Norcross, GA) containing 5 μg of peptide (10 μl total volume).
Injection of high dose peptide
Mice were injected i.p. with 100 μg of either the hgp10025–33 peptide (KVPRNQDWL) or an irrelevant Db-binding peptide SQPKNEEEI (R. Dyall, unpublished observations) every 2 days starting 30 days after the third full-length hgp100 DNA immunization, for a total of 7 injections. Mice were then boosted with full-length hgp100 DNA by gene gun 3 days after the final peptide injection. Peptides were provided by Research Genetics (Huntsville, AL).
CD4+ cell depletion
Mice were injected with the mAb GK1.5 from the American Type Culture Collection (Manassas, VA) produced by the Monoclonal Antibody Core Facility at Memorial Sloan-Kettering Cancer Center to deplete CD4+ cells. Mice were injected i.p. with 500 μg of GK1.5 on days 0 and 2, and then every 7 days until the end of the experiment. Depletion of >95% of CD4+ cells was confirmed by FACS of peripheral blood for all mice used in these studies. Mice began DNA immunization on day 4.
Mice were injected intradermally with 1 × 105 B16 melanoma cells on the right flank 5 days after the final DNA immunization. The mice were then followed for tumor onset by palpation every other day. Tumors were scored as present once they reached a 2-mm diameter and continued to grow. Mice were sacrificed once it was assured the tumors were progressing (usually at a size of ∼1 cm). Kaplan-Meier tumor-free survival curves were constructed and log rank analysis was performed to determine statistical significance.
Peptide-MHC binding assay
Peptide-MHC binding was quantified by determining relative expression of stable MHC molecules on the TAP-deficient cell line RMA-S in the presence of the indicated peptides (22). The Db restriction epitope of Flu (ASNENMETM) served as a positive control for Db. The A2.1-restricted HIV Nef peptide (VLEWRFDSRL) served as a negative control.
51Cr release assay
Inguinal lymph nodes and spleens obtained from five immunized mice 5 days after the last immunization were pooled, and 1 × 108 cells were cultured for 5 days with 1 μM concentrations of the appropriate peptide, as previously described (22). Lysis of EL4 tumor cells pulsed for 1 h at 37°C with 6 μM peptide was determined by a standard 4-h 51Cr release assay (23).
Depigmentation was objectively scored by producing digital images of the abdomens of immunized and control mice ∼6 wk after the final immunization (M. E. Engelhorn, unpublished method). The images were obtained by scanning mice on a flatbed document scanner (Hewlett-Packard, Palo Alto, CA). Mean and SD of the luminosity of the pixels for the unimmunized control mice were obtained by opening the image in Adobe PhotoShop 5.5 (Adobe, San Jose, CA), adjusting the contrast to +20, selecting the area for all the mouse abdomens, and constructing a histogram. The threshold for depigmentation was considered to be 2 SD greater than the mean. The mean luminosity of the pixels for the abdomen of each immunized mouse was then similarly obtained and scored as either depigmented or not.
ELISA for detection of Ab responses
Sera from immunized mice were collected ∼6 wk after the last immunization and tested against cell lysates from both a human (GMEL) and a mouse (B16) melanoma cell line in a modification of a previously described ELISA (24). Both cell lines express the gp100 protein.
Tumor immunity induced by xenogeneic DNA immunization with hgp100 is abolished by high dose hgp10025–33 peptide
We have previously shown that immunization of C57BL/6 mice with full-length hgp100 cDNA, but not syngeneic mouse mgp100, induces tumor protection against challenge with the syngeneic, poorly immunogenic B16 melanoma, which expresses mgp100 (2). We also showed that hgp100 DNA immunization induces T cell immunity against the Db-restricted peptide epitope hgp10025–33 and the corresponding syngeneic epitope mgp10025–33.
To test the hypothesis that T cells specific for hgp10025–33 are required for tumor protective immunity against B16, mice previously immunized with hgp100 were tolerized with high dose hgp10025–33 peptide administered without adjuvant. Injection of mice with a regimen of high dose peptide has been demonstrated to cause rapid expansion, followed by disappearance of peptide-specific T cells (25, 26). No cytotoxic T cell responses were detected in mice tolerized with peptide compared with nontolerized mice (data not shown). As shown in Fig. 1, mice immunized with full-length hgp100 DNA followed by high dose hgp10025–33 peptide are not protected against a subsequent challenge with B16 melanoma compared with nontolerized or naive mice (p = 0.91) (representative of two experiments). Mice immunized with hgp100 DNA followed by a similar treatment with a control peptide were provided the same degree of tumor protection as mice immunized with hgp100 DNA alone (p = 0.66 for control peptide vs no peptide after hgp100 immunization, p = 0.03 for hgp100 immunization plus control peptide vs no immunization). These results are consistent with a necessary role for the hgp10025–33 epitope in tumor immunity after xenogeneic immunization.
The peptide epitope hgp10025–33 has higher binding to Db than mgp10025–33, and binding of hgp10025–33 to Db is abrogated by an amino acid change (N→W) at anchor position 29
One mechanism for a heteroclitic response to a peptide epitope is through increased binding of the heteroclitic peptide to MHC molecules, leading to higher affinity (15, 17, 18, 19). The RMA-S stabilization assay was performed to determine relative binding of the Db epitopes used in this study (Fig. 2).
Table I shows the amino acid sequences of the relevant epitopes from hgp100 and mgp100, including flanking residues. Also shown is the corresponding amino acid sequence for the mutant hgp10029(N→W), in which the asparagine (N) residue at position 29 is changed to tryptophan (W). The N residue at this specific position is the major anchor for hgp100 binding to Db. As shown in Fig. 2, hgp10025–33 was a strong binder to Db, which corroborates a previous report (11), whereas mgp10025–33 was a weaker binder. The corresponding peptide (aa 25–33) from hgp10029(N→W) had no specific binding to Db.
Requirement for the high affinity Db-binding epitope of hgp10025–33 in tumor protection induced by immunization with full-length hgp100 DNA
To further test the hypothesis that the presence of the single higher affinity epitope hgp10025–33 is necessary for the tumor protection seen after xenogeneic immunization, site-specific mutants of hgp100 were constructed. The cDNA construct hgp100KVP→EGS was created from hgp100 by replacing the sequence coding for aa 25–27 (KVP) with the corresponding mgp100 sequence (EGS). These three amino acids include predicted minor anchors for binding to Db. Thus, hgp100KVP→EGS comprises the mgp10025–33 epitope embedded in the hgp100 framework, including hgp100-flanking residues (Table I). In contrast, hgp10029(N→W) contains hgp100 wild-type sequences with a single mutation at aa 29 (N→W), the Db anchor residue (Table I). As seen in Fig. 3, neither immunization with hgp100KVP→EGS nor immunization with hgp10029(N→W) full-length DNA significantly protected against tumor challenge with B16 melanoma, whereas immunization with hgp100 was protective (p < 0.01) (representative of three experiments). Combined results from all 3 experiments show that at 30 days after tumor challenge 19 of 30 mice immunized with full-length hgp100, 4 of 30 mice immunized with hgp100KVP→EGS, 3 of 23 immunized with hgp10029(N→W), and none of the 29 unimmunized mice were tumor free.
The hgp10025–33 epitope is not required for anti-gp100 Ab responses after hgp100 DNA immunization (Fig. 4). Immunization of mice with hgp100, hgp100KVP→EGS, and hgp10029(N→W) DNA produced similar Ab reactivities when measured against the human gp100+ human melanoma cell line GMEL (p < 0.01, compared with preimmune sera) (Fig. 4,A). Immunization with mgp100 DNA produced lower Ab reactivity (p = 0.02 to <0.01 compared with hgp100, hgp100KVP→EGS, and hgp10029(N→W)) but higher Ab reactivity compared with preimmune sera (p < 0.01). The same weaker Ab response against mgp100 (measured against the syngeneic melanoma B16 cell line) was observed in sera of mice immunized with hgp100, hgp100KVP→EGS, hgp10029(N→W), and mgp100 (p = 0.04 to <0.01 compared with preimmune sera; p < 0.15 compared with each other) (Fig. 4 B).
Tumor protection after hgp100 DNA immunization does not require CD4+ cells
We have previously shown that hgp100 DNA immunization can elicit tumor protection in MHC class II knockout mice (2). To confirm that CD4+ T cells are not required for tumor protection in mice with immune systems that developed under normal physiological conditions, CD4+ cells were depleted from adult mice immediately before hgp100 immunization and depletion was maintained throughout the tumor challenge (10 mice per group). As shown in Fig. 5, tumor protection was seen after xenogeneic hgp100 DNA immunization in CD4+ cell-depleted mice (p < 0.01). Although greater protection was observed in CD4+-competent mice, the difference did not reach statistical significance (p = 0.08). Further studies have shown that CD4+ T cells enhance immunity in this model, and we are presently searching for class II MHC-restricted epitopes within hgp100 (data not shown). In summary, these results confirm that the predominant immune mechanism in this model is through CD8+ T cells recognizing the class I MHC hgp10025–33 epitope and that T cell help by CD4+ T cells can further augment immunity.
Cytotoxic T cell responses against the self peptide mgp10025–33 can be induced by immunization with a minigene encoding hgp10025–33
Because immunity induced by full-length hgp100 DNA does not require CD4+ T cells or MHC class II, it is possible that immunity against mgp100 could be induced by the MHC class I peptide epitope hgp10025–33 alone. A minigene containing an endoplasmic reticulum insertion signal followed by hgp10025–33 was created and used to immunize mice. Lymphocytes were assessed in a standard 4-h 51Cr release assay. Mice immunized either with the hgp10025–33 minigene or with the full-length hgp100 cDNA had the same CTL activity against mgp10025–33, as measured by 51Cr release (Fig. 6).
Tumor immunity against B16 melanoma is induced by immunization with hgp10025–33 minigene
Mice immunized with the hgp10025–33 minigene and challenged with B16 melanoma demonstrated tumor protection that was comparable with protection observed in mice immunized with full-length hgp100 (Fig. 7) (representative of three experiments). In all experiments combined, at 30 days after tumor challenge, 10 of 40 mice immunized with full-length hgp100, 15 of 40 mice immunized with hgp10025–33 minigene, 1 of 18 mice immunized with mgp10025–33 minigene, and 1 of 40 mice left unimmunized were tumor free. Peptide immunization with the hgp10025–33 peptide with the TiterMax adjuvant did not elicit any tumor protection in 2 separate experiments with 10 mice per group (0 of 20 mice, data not shown). These results suggest that there are other unaltered epitopes within gp100 that have not yet been identified and that participate in immunity possibly through epitope spreading, although they are insufficient to induce tumor rejection.
Another measure of immunity against melanosomal differentiation Ags is autoimmunity, manifested as coat color depigmentation (1, 2, 3, 4, 7, 8, 9). Because melanosomal differentiation Ags are present in normal melanocytes, immunity can result in the destruction of pigment cells in hair follicles, and the degree of coat color hypopigmentation can be easily assessed in live animals. None of the 10 mice immunized with full-length hgp100, as well as none of the 10 mice immunized with a minigene construct based on the Db epitope of the influenza matrix-binding protein developed hypopigmentation. Two of nine mice immunized with the hgp10025–33 minigene were depigmented (J. S. Gold, unpublished observations). We believe that hypopigmentation is a real but infrequent consequence of minigene immunization. These results are consistent with a threshold for tumor immunity that is lower than that for autoimmunity after minigene vaccination.
The xenogeneic DNA immunization with hgp10025–33 epitope is necessary and sufficient to induce tumor immunity against melanoma in an H-2b mouse model, but this is accompanied by minimal autoimmunity.
Immunization with syngeneic melanosomal differentiation Ags in the form of recombinant proteins or naked DNA has generally been unsuccessful, whereas xenogeneic vaccination has proved to be a method of inducing immunity against these shared Ags (1, 2, 3, 4). It has been postulated that the mechanism underlying xenogeneic immunization for tumor immunity is the creation of heteroclitic peptide epitopes, but to our knowledge this has never been previously demonstrated for tumor immunity directly in vivo.
The gp100 Ag was originally defined by mouse mAb and subsequently shown to be recognized by autologous T cells of melanoma patients (20, 27). Previous studies by Overwijk et al. (11) showed that hgp100 contained an altered peptide ligand that could be used to activate T cells for adoptive therapy in a syngeneic mouse melanoma model. We have previously shown that active immunization with hgp100 was able to induce CD8+ T cell-mediated immunity independently of MHC class II and CD4+ T cells, providing a model to study the requirement of individual MHC class I epitopes (2). We have shown that the induction of tumor immunity depends on the presence of a single heteroclitic MHC class I epitope. Furthermore, this heteroclitic epitope is sufficient to induce tumor-protective immunity when delivered as a minigene construct, and this tumor immunity is associated with only minimal autoimmunity. These results provide support for a strategy to immunize against tumor Ags using minigenes containing defined heteroclitic epitopes that induce cross-reactivity to tumor Ags.
Delivery of this heteroclitic epitope as a minigene was effective in protecting against a subsequent tumor challenge, whereas vaccination with synthetic peptide in a strong adjuvant was not. The minigene construct presumably leads to more efficient processing and presentation through translation into the endoplasmic reticulum than uptake of peptide through an exogenous pathway, even in the presence of a potent immune adjuvant that presumably induces cytokines, heat shock proteins, and other components that could enhance immunity. Also, the prokaryotic unmethylated CpG motifs in plasmid DNA may enhance the potency of the minigene immunization (28).
Further elucidation of the mechanisms of tumor protection after xenogeneic immunization may allow the creation of more rational vaccines. The design of heteroclitic epitopes for differentiation Ags and their delivery to patients as minigenes may also be a valuable strategy for the immunotherapy of cancer.
We thank Francis Lim for his technical expertise and Dr. Rodica Stan for reviewing the manuscript.
This work was supported by National Cancer Institute Grants CA58621, CA33049, and CA59350; Swim Across America and the Experimental Therapeutics Center of Memorial Sloan-Kettering Cancer Center; the Louis and Anne Abrons Foundation; the Kennedy Family Fund; and Training Grant CA09501 (to J.S.G., W.G.H., and C.R.F.). J.S.G. was a Lawrence and Selma Rubin Fellow. M.E.E. is the recipient of a CRI fellowship award.
Abbreviations used in this paper: hgp100, human gp100; mgp100, mouse gp100.