Coupling and Uncoupling of Tumor Immunity and Autoimmunity

Abstract

Self-antigens, in the form of differentiation antigens, are commonly recognized by the immune system on melanoma and other cancers. We have shown previously that active immunization of mice against the melanocyte differentiation antigen, a tyrosinase-related protein (TRP) gp75^TRP-1 (the brown locus protein) expressed by melanomas, could induce tumor immunity and autoimmunity manifested as depigmentation. In this system, tumor immunity and autoimmunity were mediated by autoantibodies. Here, we characterize immunity against another tyrosinase family glycoprotein TRP-2 (the slaty locus protein), using the same mouse model and method of immunization. As observed previously for gp75^TRP-1, immunity was induced by DNA immunization against a xenogeneic form of TRP-2, but not against the syngeneic gene, and depended on CD4⁺ cells. Immunization against TRP-2 induced autoantibodies and autoreactive cytotoxic T cells. In contrast to immunization against gp75^TRP-1, both tumor immunity and autoimmunity required CD8⁺ T cells, but not antibodies. Only autoimmunity required perforin, whereas tumor immunity proceeded in the absence of perforin. Thus, immunity induced against two closely related autoantigens that are highly conserved throughout vertebrate evolution involved qualitatively different mechanisms, i.e., antibody versus CD8⁺ T cell. However, both pathways led to tumor immunity and identical phenotypic manifestations of autoimmunity.

Self-antigens on human cancer are the most prevalent antigens recognized by the immune system 1. This reflects the fact that cancers arise from the hosts' own tissues, and are not truly foreign. Thus, in some respects, immune recognition of cancer appears to be different from immune recognition of viruses or bacteria, and typically more akin to autoimmunity. Recognition of self-antigens on cancer presents problems. First, immunity to cancer may not develop because of immune tolerance. Second, even when the immune system can recognize and respond to self-antigens, immunity may not be sufficient to reject cancers. Finally, if immunity to self-antigens develops, there are potential autoimmune sequelae.

Differentiation antigens form one prototype of self-antigens on cancer 2 3 4. A differentiation antigen distinguishes a cell lineage from other cell types, and is typically expressed at specific stages of differentiation 5. Immune recognition of human cancer has been intensively investigated in melanoma, a cancer arising from melanocytes 1. Studies of melanoma have surprisingly shown that the immune system commonly recognizes products of genes that are specifically expressed by melanocytes, particularly genes that are involved in synthesis of pigment 1 6. Examples include tyrosinase, the critical enzyme required for synthesis of the pigment melanin, and tyrosinase-related proteins (TRPs) that determine the type of melanin synthesized (TRP-2 and gp75^TRP-1) [3, 4, 7]). These melanoma/melanocyte differentiation antigens can be recognized by antibodies and by CD8⁺ and CD4⁺ T cells, and thus can be broadly seen by the immune system 3 4 7 8 9 10 11 12 13 14 15 16 17 18.

We have reported previously that immunization of C57BL/6 mice against the tyrosinase family antigen gp75^TRP-1 induces tumor immunity and autoimmunity that is mediated by autoantibodies 13 15 16 17. Using DNA immunization strategies, we have investigated immunity induced against another homologous tyrosinase family member, TRP-2. The mouse gp75^TRP-1 and TRP-2 proteins are closely related, having 52% identity and 67% homology, and these genes have been highly conserved throughout vertebrate evolution (e.g., the identity of mouse gp75^TRP-1 and chicken TRP-2 is also 52%).

Materials and Methods

Mice.

C57BL/6 mice (6–8-wk-old females) were acquired through the National Cancer Institute breeding program. Homozygotic mice genetically deficient for β2-microglobulin (β2m^−/−), MHC II (Abb^−/−), and perforin (B6.pfp ^−/−), all in a C57BL/6 background, were obtained from Taconic Farms, Inc. Immunoglobulin μ chain (μMT^−/−) and gld/gld (B6.gld) mice were acquired from The Jackson Laboratory. These mice were bred and kept in a pathogen-free Memorial Sloan-Kettering Cancer Center vivarium according to National Institutes of Health Animal Care guidelines. All mice entered the study between 7 and 10 weeks of age.

Cell Lines and Tissue Culture.

B16F10/LM3 15 is a pigmented mouse melanoma cell line of C57BL/6 origin, derived from the B16F10 line, provided by Dr. Isaiah Fidler (M.D. Anderson Cancer Center, Houston, TX). The EL-4 cell line was derived from a C57BL/6 mouse lymphoma, and SK-MEL-188 is a human melanoma cell line. Tumor cell lines were cultured as described 15.

Plasmid Constructs.

The human TRP-2 (hTRP-2) and mouse TRP-2 (mTRP-2) expression vectors (supplied by Drs. S.A. Rosenberg and J.C. Yang, National Cancer Institute, Bethesda, MD) were previously described 9. These genes were cloned into the PCR3 vector, which was used as a control vector without inserts. The mouse GM-CSF gene, provided by PowderJect Vaccines Inc., was cloned into the WRGBEN vector 13.

DNA Immunization.

The method of DNA immunization has been reported 13. In brief, plasmid DNA encoding hTRP-2, mTRP-2, or GM-CSF was coated onto 1.0 μm gold bullets. Animals were immunized by delivering gold–DNA complexes using a helium-driven gun (Accell^®; PowderJect Vaccines Inc.) into each abdominal quadrant (1 μg plasmid DNA/quadrant) for a total of four injections.

Depletion and Depigmentation Studies.

NK cell depletion was performed as described 15. Animals were injected intraperitoneally with mAb PK136 (anti–NK-1.1) 3 d before immunization, and every 4 d thereafter. Depigmentation experiments were performed as described 13 15. In brief, after the final immunization, mice were shaved and depilated over the posterior flank and observed for 8 wk. Scoring of depigmentation was performed by dividing the abdomen into four equal quadrants. Quadrants were recorded as positive when they had estimated >50% depigmented hairs. Depigmentation was scored 0–4+ according to the number of quadrants that were depigmented in each mouse (e.g., 3+ if three of four quadrants are depigmented >50%).

Antibody Responses to TRP-2.

Antibody responses to syngeneic mTRP-2 were measured by immunoprecipitation, followed by Western blot assay as described 13. B16F10/LM3 melanoma cells were lysed, followed by immunoprecipitation with anti–PEP-8, a rabbit polyclonal antisera raised against a carboxyl-terminal peptide of TRP-2 (from Vincent Hearing, National Cancer Institute).

CTL Assay.

Lymphocytes (2 × 10⁷) from draining inguinal lymph nodes were cocultured for 5 d with naive irradiated (3,000 rads) splenocytes (2 × 10⁷), mitomycin C (100 μg/ml, 120 min at 37°C)-treated human SK-MEL-188 cells (4 × 10⁵) as a source of antigen, and pristane-induced macrophages (2 × 10⁶) from C57BL/6 mice. H-2K^b–restricted lymphocyte lysis of EL-4 tumor cells pulsed for 1 h at 37°C with 4 μg mTRP-2_181–188 peptide (sequence VYDFFVWL) was determined by a 4-h standard ⁵¹Cr release assay 16.

Mouse Tumor Studies.

All mice were injected intravenously via the tail vein with 2 × 10⁵ B16F10/LM3 melanoma cells. Tumor challenge was performed 5 d after the final immunization, unless otherwise indicated. Mice were killed at 14–22 d after tumor challenge, all lobes of both lungs were dissected, and surface lung metastases were scored and counted for all lobes under a dissecting microscope. Statistical analysis of tumor growth was performed using the two-sided Student's t test, assuming unequal variances and Wilcoxon Scores (Rank Sums).

Results

Xenogeneic, but Not Syngeneic, TRP-2 DNA Induces Tumor Rejection.

hTRP-2 has 90% homology and 83% identity to the amino acid sequence of C57BL/6 mTRP-2. DNA immunization with xenogeneic hTRP-2 decreased B16F10/LM3 lung metastases by ≥90% (P < 0.0001) in tumor protection experiments (Fig. 1A–C). There was no significant evidence of tumor immunity after immunization with syngeneic mTRP-2 DNA compared with untreated mice or mice injected with control null vector (Fig. 1A and Fig. C).

Figure 1.

Protection and rejection of mouse melanoma B16F10/LM3 after immunization with human TRP-2 (hTRP-2), but not mouse TRP-2 (mTRP-2) DNA. C57BL/6 mice, 10–12 per group, were immunized cutaneously with hTRP-2 or mTRP-2 DNA by particle bombardment. Mice were challenged with B16 melanoma cells intravenously and scored for surface lung metastases after 14–17 d. (A and B) Mice were immunized three times at weekly intervals with hTRP-2, mTRP-2, or PCR3 control vector (vector), and compared to untreated mice. Mice were challenged with B16F10/LM3 melanoma 5 d after the last immunization. B is a repeat of the experiment in A, without the vector control and mTRP-2 groups. Significant tumor protection was observed in both experiments (P < 0.0001). (C) Mice were immunized five times at weekly intervals with hTRP-2, or mTRP-2, or were untreated. Significant protection was observed with hTRP-2 compared with no treatment (P = 0.001) or with mTRP-2 (P = 0.001). The difference between mice treated with mTRP-2 and untreated mice was not significant (P = 0.16). (D) Immunization with hTRP-2 DNA started 4 d after tumor challenge, or mice remained untreated. A significant therapeutic effect was observed (P < 0.001). (E) Immunization with hTRP-2 plus GM-CSF DNA started 10 d after tumor challenge. Significant therapeutic effect was observed compared with other treatment and control groups (P < 0.01). GM-CSF treatment alone yielded 692 ± 69 metastases, mTRP-2 gave 902 ± 65 metastases, hTRP-2 gave 783 ± 75 metastases, and control null vector gave 705 ± 61 metastases (not shown in figure). Results are shown as mean number of lung metastases ± SEM.

To assess the potency of DNA immunization using xenogeneic hTRP-2 DNA, mice were immunized 4 d after tumor challenge or 10 d after tumor challenge, when lung metastases were numerous and macroscopic. Immunization at 4 d decreased metastases by >80% (P < 0.001; Fig. 1 D). Therapeutic effects were observed 10 d after tumor challenge using immunization with hTRP-2 DNA plus recombinant mouse GM-CSF DNA as an immune adjuvant. Vaccination significantly decreased lung metastases by approximately half (P = 0.004; Fig. 1 E). No significant decrease in lung metastases was observed after treatment with hTRP-2 or mTRP-2 DNA, or GM-CSF DNA alone (Fig. 1 E, see legend), although there was a trend towards decreased metastases with GM-CSF alone that did not reach significance (P > 0.05). These results showed a requirement for xenogeneic antigen and the adjuvant effect of GM-CSF in the treatment of established tumors.

Xenogeneic hTRP-2 DNA Vaccination Induces Autoantibodies and Autoreactive CTLs.

We next determined whether immunization with mTRP-2 or xenogeneic hTRP-2 generated antibody and CD8⁺ T cell responses against syngeneic mTRP-2. 6 of 12 mice immunized with hTRP-2 had detectable IgG antibodies (IgG1 and IgG2b isotype) against mTRP-2 (data not shown). No autoantibodies against syngeneic mouse TRP-2 were generated after immunization with mTRP-2 (0/12). Generation of autoantibodies after immunization with hTRP-2 required both CD4⁺ and CD8⁺ T cells, because no autoantibodies were detected in mice deficient in MHC class I (0/11) or II molecules (0/10).

CTL responses against TRP-2 were detected after immunization with xenogeneic hTRP-2, but not syngeneic mTRP-2 DNA. Specifically, CD8⁺ CTL from draining lymph nodes (supraclavicular nodes), stimulated in vitro for 5 d, recognized an MHC class I H-2K^b–restricted peptide of mTRP-2 after immunization with hTRP-2 DNA (Fig. 2 [9]). Interestingly, the H-2K^b–restricted peptide of mTRP-2, mTRP-2_181–188, is identical between mouse and human TRP-2, including the immediate flanking amino acid residues. Thus, this self-peptide in the context of self–TRP-2 DNA does not induce CTL responses, but presentation of the same peptide in the context of xenogeneic hTRP-2 is immunogenic.

Figure 2.

CTL response to TRP-2_181–188 peptide in mice that were not immunized (A), or immunized with mTRP-2 (B) or hTRP-2 (C). C57BL/6 mice (2 or 3 per group) were immunized as described in the legend to Fig. 1. 7 d after the last immunization, draining lymph nodes were pooled and stimulated as described in Materials and Methods for 5 d and tested for cytotoxicity against EL-4 target cells, either pulsed with TRP-2 peptide or unpulsed. Results are representative of two experiments.

Tumor Rejection Requires CD4⁺ and CD8⁺ Cells, but Not B Cells or NK Cells.

These results suggested that either antibody or CTL responses, or both, mediated tumor rejection. Roles for critical cell types were investigated by immunizing β2m^−/− mice deficient in MHC class I and CD8⁺ T cells, MHC II^−/− mice deficient in MHC class II and CD4⁺ T cells, Igμ^2/− mice deficient in mature B cells, and mice depleted of NK1.1⁺ cells, including NK cells (Fig. 3). Both MHC class I and II molecules were required for tumor rejection, supporting a central role for both CD8⁺ and CD4⁺ T cells. Neither NK cells nor B cells were necessary for tumor immunity. Noticeably, mice deficient in B cells developed fewer baseline metastases compared with wild-type C57BL/6 mice, and were completely free of any detectable tumor after treatment with hTRP-2 (12 of 12 mice). This phenomenon of enhanced T cell–dependent tumor rejection associated with B cell deficiency has been reported previously 10. These results showed that T cell immunity, including both CD8⁺ and CD4⁺ T cells, was required for tumor rejection, but antibodies were not.

Figure 3.

Tumor rejection in C57BL/6 mice deficient in immune molecules and cells. Mice, 10–12 per group, were immunized cutaneously with hTRP-2 DNA as described in the legend to Fig. 1. Lung metastases were evaluated after 18 d. Groups included mice deficient in β2 microglobulin (β2^−/−), MHC II, immunoglobulins (μ^2/−), and depleted of NK1.1⁺ cells (NK deplete). All mice were treated in the same experiment, and differences in lung metastases in the no treatment groups reflect tumorigenicity in different mouse strains or under different conditions (e.g., NK deplete). Significant protection was observed in NK depleted (P < 0.0001) and μ^2/− (P = 0.03), but not β2^−/− or MHC II^−/− (P > 0.10) mice. This experiment is representative of two experiments. Results are shown as mean ± SEM.

Xenogeneic Immunization Induces Autoimmunity That Also Requires T Cells.

Signs of autoimmunity, manifested as depigmentation, were observed in mice immunized with hTRP-2, but not generally in mice immunized with syngeneic mTRP-2 (Fig. 4). Depigmentation appeared 4–5 wk after starting immunization over depilated and shaved areas of the mouse coat, spreading to unshaved areas in most mice. Autoimmunity also required T cells, but not antibodies or NK cells, showing that tumor immunity and autoimmunity were coupled by a requirement for class I and II MHC expression leading to a requirement for T cells.

Figure 4.

Depigmentation in mice treated with hTRP-2 or mTRP-2 DNA. Groups of mice (12–15 per group) were immunized with hTRP-2 or mTRP-2 DNA as described in the legend to Fig. 1, or remained untreated. Each dot represents a separate mouse. Groups include C57BL/6 (wt) mice treated with hTRP-2 (h) or mTRP-2 (m) DNA. In addition, immunoglobulin-deficient (μ^2/−), β2 microglobulin-deficient (β2^−/−), MHC II–deficient (MHC II^−/−), NK-depleted, and perforin-deficient (perf^−/−) mice were immunized with hTRP-2.

Requirement for Perforin in T Cell–dependent Autoimmunity, but Not Tumor Immunity.

CTLs have been proposed to be critical effector cells that mediate tumor rejection. Cytotoxicity of T cells can be mediated by exocytic granules involving perforin, or by cell membrane molecules that induce death of target cells. Tumor immunity proceeded in the complete absence of perforin (P = 0.0002) in pfp ^−/− mice (Fig. 5), whereas autoimmunity was mostly inhibited (Fig. 4). Fas ligand was not necessary for either autoimmunity (Fig. 4) or tumor immunity (Fig. 5). These results are consistent with perforin-mediated killing of normal melanocytes in hair follicles playing a central role in autoimmunity. However, tumor immunity could proceed in a perforin-independent manner.

Figure 5.

Tumor rejection in perforin-deficient (pfp ^−/−) mice and gld/gld mice (deficient in fas ligand) treated with hTRP-2, compared with mice not treated. Mice (9 or 10 per group) were immunized cutaneously with hTRP-2 DNA as described in the legend to Fig. 1. Significant tumor protection was observed in pfp ^−/− (P = 0.0002) and gld/gld (P < 0.0001) mice.

Discussion

TRP-2 is recognized by CTLs of patients with melanoma 7 11 12, and has also been defined as a potential tumor-rejection antigen in C57BL/6 mice 9. Thus, TRP-2 provides a model for a differentiation antigen with relevance to a human cancer. Xenogeneic DNA vaccination is one strategy to immunize against potentially weak self-antigens. The approach using a xenogeneic source of antigen is well known to produce autoimmunity, but has also been used to induce tumor immunity against gp75^TRP-1 and another melanocyte/melanoma differentiation antigen gp100 13 14 16 17 18. Other strategies have shown that syngeneic mouse gp75^TRP-1 expressed in insect cells 16 or by vaccinia virus 14 can also induce tumor immunity and trigger depigmentation. Expression of the syngeneic protein in the context of xenogeneic cells that may package the antigen in insoluble inclusion bodies (e.g., insect cells) or with strongly immunogenic viral proteins are alternative strategies, at least for inducing antibody-dependent immunity against tyrosinase family antigens.

As noted above, the CTL response directed against the H-2K^b-restricted peptide of mTRP-2, mTRP-2_181–188, is remarkable because this peptide is identical between mouse and human TRP-2, including the immediate flanking amino acid residues. It is possible the distant amino acid residues alter processing of this peptide in hTRP-2, allowing more efficient presentation. Alternatively, amino acid differences in other peptides of hTRP-2 may provide T cell help, which in turn is sufficient to trigger CTL responses to the mTRP-2_181–188 self-peptide. The observation that only hTRP-2 DNA (but not mTRP-2 DNA) induced CD8⁺ T cell responses suggests that T cell tolerance was broken by xenogeneic DNA immunization, although we recognize this could reflect differences in efficiency of processing.

Tumor Immunity and Autoimmunity: Two Means to the Same End.

Immunity against gp75^TRP-1 led to tumor protection and to depigmentation that was indistinguishable from autoimmunity induced by immunization against TRP-2 13. Similar results were observed by Naftzger et al. 16 and Overwijk et al. 14 after immunization with syngeneic gp75^TRP-1 expressed by baculovirus in insect cells or in vaccinia virus, respectively. In gp75^TRP-1 systems, tumor immunity and autoimmunity were mediated by autoantibodies, and tumor immunity depended on NK cells and functional Fc receptors; neither tumor immunity nor autoimmunity required CD8⁺ T cells 13 14 15 16 17. In contrast, tumor immunity and autoimmunity against TRP-2 required MHC class I molecules, and by implication CD8⁺ T cells, without a requirement for antibodies or NK cells. Thus, either autoantibodies or T cells can provide specificity for tumor immunity and autoimmunity.

Antibody Versus CD8⁺ T Cells: Both Need CD4⁺ Cells.

The final phenotype of tumor immunity and autoimmunity can be dependent either on antibodies or CD8⁺ T cells, but in both cases CD4⁺ cells are required. Vaccination against TRP-2 as late as 10 d after tumor challenge induced substantial decreases in tumor burden, whereas active immunization against gp75^TRP-1 at these later time points was ineffective 13. We propose that this reflects in part the kinetics of T cell versus antibody responses, where effective T cell immunity can be generated over days, whereas effective antibodies may require weeks. Passive transfer of antibodies against gp75^TRP-1 had significant effects 4 and 7 d after tumor challenge 15, consistent with this view.

Tumor Immunity and Autoimmunity.

The potential coupling of tumor immunity with autoimmunity has been suggested by the clinical observation that patients with metastatic melanoma who develop vitiligo have a better prognosis and are more likely to respond to therapy 19 20 21 22. The differences in mechanisms underlying tumor immunity and autoimmunity could be a consequence of fundamental differences in effector mechanisms used to kill tumor cells versus normal melanocytes. Alternatively, the differences could reflect the different tissue sites of melanocytic cells. In this scenario, the effector mechanisms in the skin require perforin, whereas other mechanisms are used in the lung. Mechanisms other than perforin or fas ligand may be involved in tumor rejection, but it is also possible that perforin and fas ligand provide redundant mechanisms for effector functions against tumors. Finally, the uncoupling of tumor immunity from autoimmunity in this model shows that one can inhibit autoimmunity and still be permissive for tumor immunity. This opens strategies for inducing immunity to treat cancer where autoimmunity can be inhibited while cancer immunity proceeds.