Loss of CD4 T-cell–dependent tolerance to proteins with modified amino acids

Abstract

The site-specific incorporation of the unnatural amino acid p-nitrophenylalanine (pNO₂Phe) into autologous proteins overcomes self-tolerance and induces a long-lasting polyclonal IgG antibody response. To determine the molecular mechanism by which such simple modifications to amino acids are able to induce autoantibodies, we incorporated pNO₂Phe, sulfotyrosine (SO₃Tyr), and 3-nitrotyrosine (3NO₂Tyr) at specific sites in murine TNF-α and EGF. A subset of TNF-α and EGF mutants with these nitrated or sulfated residues is highly immunogenic and induces antibodies against the unaltered native protein. Analysis of the immune response to the TNF-α mutants in different strains of mice that are congenic for the H-2 locus indicates that CD4 T-cell recognition is necessary for autoantibody production. IFN-γ ELISPOT analysis of CD4 T cells isolated from vaccinated mice demonstrates that peptides with mutated residues, but not the wild-type residues, are recognized. Immunization of these peptides revealed that a CD4 repertoire exists for the mutated peptides but is lacking for the wild-type peptides and that the mutated residues are processed, loaded, and presented on the I-A^b molecule. Overall, our results illustrate that, although autoantibodies are generated against the endogenous protein, CD4 cells are activated through a neo-epitope recognition mechanism. Therefore, tolerance is maintained at a CD4 level but is broken at the level of antibody production. Finally, these results suggest that naturally occurring posttranslational modifications such as nitration may play a role in antibody-mediated autoimmune disorders.

Previously we showed that the site-specific incorporation of the unnatural amino acid p-nitrophenylalanine (pNO₂Phe) into a protein can induce autoantibodies against the endogenous, unmodified protein (1, 2). The murine proteins TNF-α and retinol-binding protein (RBP4) were modified genetically to incorporate pNO₂Phe at several surface-exposed positions and were used as immunogens. Vaccination of mice with these altered proteins resulted in an IgG polyclonal antibody response against the endogenous proteins that lasted for at least 40 wk. Experiments involving hybridomas generated from the TNF-α–immunized mice revealed that the IgG antibody response was directed against different regions of the protein and not exclusively against the region containing the pNO₂Phe residue. The autoantibodies generated by the mutated proteins were tested in an endotoxemia model for their ability to neutralize wild-type TNF-α. Protection from endotoxemia indicated that vaccination with pNO₂Phe-modified self-protein is able to elicit physiologically relevant autoantibody responses. Although these initial studies implicated a role for CD4 T cells in autoantibody production, the exact mechanism by which the site-specific incorporation of pNO₂Phe breaks immunological self-tolerance remained to be elucidated.

Recent studies have indicated that posttranslational modifications (PTMs) can affect T-cell–mediated immune responses and have suggested that such modifications may play a role in the initiation and/or progression of autoimmune disorders (3–6). PTMs, including citrullination, glycosylation, isomerization, sulfation, and nitrosylation, transform the structure of proteins and result in generation of neo-epitopes (7–14). In particular, the nitration of Tyr forms 3-nitrotyrosine (3NO₂Tyr), a PTM induced by stress, viral infection, and mitochondrial dysfunction (10, 11, 15), which shares remarkable structural similarity with the nitroaryl moiety of pNO₂Phe. In mouse models, immunization with class I and class II peptides modified with 3NO₂Tyr residues elicits stronger immune responses than do the unmodified peptides (16, 17). Moreover, the presence of 3NO₂Tyr-modified proteins has been associated with several autoimmune diseases (10, 11, 15, 18–24). Thus, we decided to investigate further the mechanism by which nitration of phenylalanine and Tyr residues can lead to a cross-reactive antibody response to native proteins. Specifically, we demonstrate that autoantibody responses are elicited when mice are immunized with TNF-α mutants containing pNO₂Phe and 3NO₂Tyr, as well as with sulfotyrosine (SO₃Tyr) residues, a PTM involved in the regulation of chemokine signaling (25, 26). Mechanistic studies indicate that, although CD4 T cells remain tolerant to endogenous epitopes, neo-epitopes activate CD4 T cells that render help for the generation of autoantibody responses. In addition, when unnatural amino acids that correspond to the naturally occurring PTMs 3NO₂Tyr and SO₃Tyr are substituted for Tyr²⁹ in EGF, self-tolerance is broken, and antibodies to native EGF are generated. These experiments suggest that PTMs may naturally induce an autoimmune response and also provide a useful strategy for enhancing the immunogenicity of vaccine antigens.

Results

Loss of Immunological Tolerance Is Position and Strain Dependent.

Our previous experiments showed that when surface-exposed residues in murine TNF-α and RBP4 proteins are mutated to pNO₂Phe, the resulting proteins become immunogenic (1, 2). In particular, substitution of pNO₂Phe for Lys at position 11 in TNF-α was shown to break immunological tolerance against the self-protein. In the current study, we extended these earlier studies by incorporating two PTMs (3NO₂Tyr and SO₃Tyr) and a somatic mutation (Tyr) at the same position in TNF-α. Furthermore, to determine if the site of mutation affects the production of autoantibodies, pNO₂Phe also was substituted for Gln²¹ in TNF-α. Amino acid substitutions were generated by mutating the codon for the amino acid of interest to the amber stop codon TAG using site-directed mutagenesis. Orthogonal, amber suppressor tRNA/ aminoacyl-tRNA synthetase pairs specific for pNO₂Phe, 3NO₂Tyr, or SO₃Tyr were used to incorporate the unnatural amino acid selectively into the protein in response to the TAG codon (27, 28). All antigens were confirmed by protein liquid chromatography - MS, SDS-PAGE, gel filtration, and endotoxin level estimation (Figs. S1–S5 and Tables S1–S4). C57BL6 (B6) mice then were immunized individually once a week for 4 wk with wild-type TNF-α or with TNF-α mutants containing Tyr, pNO₂- Phe, 3NO₂Tyr, or SO₃Tyr residues at positions 11 and 21 in alum. ELISA analysis revealed that autoantibodies were generated for some, but not all, mutants. In particular, all the site-specific modifications at Lys¹¹ (pNO₂Phe, Tyr, 3NO₂Tyr, and SO₃Tyr) resulted in autoantibody production (Fig. 1 and Fig. S6); consequently these mutants were used for subsequent studies.

Fig. 1.

Immunogenicity of pNO₂Phe is position and strain dependent. (A and E) B6 (H-2^b), (B and F) FVB/N (H-2^q), (C and G) BALB/c (H-2^d), and (D and H) C3H (H-2^k) mice were immunized with 5 μg wild-type TNF-α or pNO₂Phe¹¹ (A–D) or with 3NO₂Tyr¹¹ or SO₃Tyr¹¹ TNF-α (E–H) mutants adjuvanted in alum, once a week for 4 wk. In all experiments, sera were collected 1 wk after each immunization, and serial dilutions were tested for the presence of antibodies against the wild-type TNF-α protein. The data indicate autoantibody responses at a 1:1,000 titer measured by ELISA. Data shown are representative of two independent experiments (n = 3–5 mice per group).

We next determined whether the mouse strain affects the immune response. Mice with different genetic backgrounds [B6, FVB/NJ (FVB/N), BALB/c, and C3H/HeNHsd (C3H)], were vaccinated with pNO₂Phe¹¹, 3NO₂Tyr¹¹, or SO₃Tyr¹¹ TNF-α mutants, and the production of autoantibodies against the wild-type TNF-α was determined by ELISA (Fig. 1). All modifications at position 11 are immunogenic in B6 mice, but the same mutant proteins are not immunogenic in FVB/N or C3H mice. In BALB/c mice, a low antibody titer is observed only after immunization with the pNO₂Phe¹¹ mutant. To confirm that immunogenicity is position and strain dependent, B6 and FVB/N mice were immunized with the pNO₂Phe²¹ mutant (wild type is Gln²¹) (Fig. S7). ELISA analysis indicated that pNO₂Phe¹¹, but not pNO₂Phe²¹, is immunogenic in B6 mice, whereas the opposite result was observed in FVB/N mice, which developed autoantibodies when immunized with pNO₂Phe²¹ TNF-α but not with pNO₂Phe¹¹ TNF-α.

A loss of tolerance to self-antigens can be achieved through vaccination with strong adjuvants (29). To assess whether pNO₂Phe¹¹, 3NO₂Tyr¹¹, and SO₃Tyr¹¹ are immunogenic because of the mutations, and not because of the aggressive vaccination schedule, we immunized B6 mice with the mutated proteins in PBS. The results indicate that even in the absence of adjuvant, the pNO₂Phe¹¹, 3NO₂Tyr¹¹, and SO₃Tyr¹¹ mutants are able to elicit a humoral immune response against the unaltered TNF-α protein (Fig. S8). Overall, these results show that both the site of the modification and the genetic background of the immunized population significantly affect autoantibody production.

Loss of Immunological Tolerance Depends on MHC Class II.

The difference in immune response to the TNF-α mutants in one genetic background relative to another could result from overall genetic differences between the different strains or, more probably, from differences exclusively at the MHC locus. To address this question, we generated F2 mice from F1 intercrosses between the responder background (B6) and the nonresponder background (FVB/N) and phenotyped the expression of I-A for each mouse (Fig. 2 A and D). Three groups of F2 (FVB × B6) mice expressing I-A^b, I-A^q, or I-A^bq were immunized with the pNO₂Phe¹¹, 3NO₂Tyr¹¹, or SO₃Tyr¹¹ TNF-α mutants, and autoantibodies against wild-type TNF-α were measured by ELISA. Autoantibody production strictly correlated with mice harboring the I-A^b or I-A^bq molecules, indicating that the autoimmune phenotype is not dependent on the genetic diversity of the different backgrounds but rather correlates with the MHC locus. To confirm the requirement for MHC restriction, mice with identical genetic backgrounds and congenic for the MHC locus were studied. C57BL/10SnJ (B10) mice and B10.D1-H2^q/SgJ (B10.Q) mice are genetically identical except for the MHC molecules (H-2^b for B10 and H-2^q for B10.Q mice); therefore, any differences in autoantibody production depend on MHC H-2 expression. Presumably, if immune responses are dependent on the H-2^b locus, only the B10 mice should produce autoantibodies, and mice with the H-2^q locus should not. Indeed, B10 mice immunized with the pNO₂Phe¹¹, 3NO₂Tyr¹¹, or SO₃Tyr¹¹ TNF-α mutant proteins are able to generate autoantibody responses, whereas B10.Q mice do not produce detectable autoantibody (Fig. 2 B and E). To demonstrate further that these responses are restricted not only to the H-2^b locus but specifically to the MHC class II molecule, congenic B6 and B6(C)-H2-Ab1^bm12/KhEgJ (Bm12) mice (genetically identical except for three amino acids in the class II molecule) (30) were vaccinated with the pNO₂Phe¹¹, 3NO₂Tyr¹¹, or SO₃Tyr¹¹ mutant proteins (Fig. 2 C and F). Autoantibody production was observed only for the B6 mice, confirming that loss of tolerance to wild-type TNF-α by vaccination with the Lys¹¹ TNF-α mutants is mediated through I-A^b–restricted responses.

Fig. 2.

Immunogenicity of pNO₂Phe¹¹ is I-A^b dependent. Different strains of congenic mice were immunized once a week for 4 wk with 5 μg pNO₂Phe¹¹ (■), 3NO₂Tyr¹¹(▲), and SO₃Tyr¹¹ (●) TNF-α mutants adjuvanted in alum. (A and D) F2 mice derived from a cross of F1 (FVBxB6) mice were haplotyped by FACS analysis for the expression of I-A^b and I-A^q. Mice were assigned to three groups depending on the zygosity of these alleles. (B and E) B10 (H-2^b) and B10.Q (H-2^q) mice are genetically identical except for the MHC locus. (C and F) B6 (I-A^b) and Bm12 (I-A^bm12) mice are genetically identical except for three amino acids in the β chain of the class II molecule. The data represent serial dilutions of sera collected 26 d after mice were vaccinated and the autoantibody responses at 1:1,000 titer measured by ELISA. The data shown are representative of two independent experiments (n = 3–5 mice per group).

pNO₂Phe¹¹, 3NO₂Tyr¹¹, and SO₃Tyr¹¹ Mutants of TNF-α Generate CD4 T-Cell Neo-Epitopes.

Modifications of self-proteins may affect their recognition by the immune system in a number of ways. For example, the immunogenic nitrophenyl moiety may enhance binding to natural antibodies or increase uptake, processing, and presentation of antigen (31–34). Alternatively, a single modified residue may induce changes in endosomal/lysosomal processing and allow epitopes that usually are not presented to be exposed on the surface to the MHC molecules (35–38). Additionally, the modified residues could alter the MHC–T-cell receptor (TCR) interaction. For example, if the modified residue enhances the binding affinity of the peptide to the MHC molecule, the lifetime of the MHC–TCR complex could be increased, resulting in T-cell activation (39, 40). A modified residue also could create an epitope that interacts directly with the TCR and triggers T-cell activation (16, 17). Because our data indicate that immunization with Lys¹¹ TNF-α mutants depends on MHC class II restriction, we next determined whether CD4 T cells are able to recognize the mutated neo-epitopes. On the basis of T-cell epitope prediction software (41–43), we synthesized a panel of wild-type and pNO₂Phe¹¹ peptides spanning a small portion of the Lys¹¹ region in TNF-α (Table S5). Mice with the B6 background (H-2^b) then were immunized with wild type TNF-α or with pNO₂Phe¹¹ TNF-α mutants, and CD4 T cells were purified from the spleens 1 wk after the last immunization. A panel of the pNO₂Phe¹¹ and wild-type peptides was added to an antigen-presenting cell (APC)-CD4 coculture, and IFN-γ production was determined by ELISPOT analysis. The CD4 T cells derived from mice immunized with wild-type TNF-α do not recognize either the wild-type or pNO₂Phe¹¹ peptides. Conversely, CD4 T cells isolated from mice immunized with the pNO₂Phe¹¹ TNF-α are able to recognize, become activated, and produce IFN-γ only in response to the pNO₂Phe¹¹ peptides but not in response to the native, nonaltered peptides (Fig. 3A). Immunization of B6 mice with a Tyr¹¹ mutant TNF-α also resulted in responses against the Tyr¹¹ epitope but not against the wild-type Lys¹¹ epitope (Fig. S9). Immunization of B10, B10.Q, Bm12, and F2 (FVB × B6) mice with pNO₂Phe¹¹ TNF-α further confirmed that CD4 T cells isolated from mice vaccinated with the mutant TNF-α are able to recognize only the peptides containing pNO₂Phe¹¹ (Fig. 3 B–D).

Fig. 3.

CD4 T cells isolated from vaccinated mice recognize the mutated epitopes exclusively. (A) B6 mice were immunized once a week for 4 wk with 5 μg wild-type TNF-α or pNO₂Phe¹¹ TNF-α mutant adjuvanted in alum. (B) B10 and B10.Q mice were immunized once a week for 4 wk with 5 μg pNO₂Phe¹¹ TNF-α mutant adjuvanted in alum. For A and B, CD4 T cells were isolated 1 wk after the last immunization and were incubated with APC pulsed with 4 μg of a panel of wild-type peptides (WT-pep) or pNO₂Phe-containing peptides (pNO₂Phe-pep) for 48 h at 37 °C (see Table S5 for sequences). (C and D) B6, Bm12, and F2 mice derived from a cross of F1 (FVB × B6) mice were immunized once a week for 4 wk with 5 μg pNO₂Phe¹¹ TNF-α mutant adjuvanted in alum. (E and F) B6, Bm12, B10, B10.Q, and F2 mice were immunized once a week for 4 wk with 5 μg 3NO₂Tyr¹¹ TNF-α mutant adjuvanted in alum. (G and H) B6, Bm12, B10, B10.Q, and F2 mice were immunized once a week for 4 wk with 5 μg SO₃Tyr¹¹ TNF-α mutant adjuvanted in alum. In all experiments, CD4 T cells were incubated with APC pulsed with 4 μg of WT-pep5, peptide Ova₃₃₉ known to bind I-A^b, or a peptide containing the modification with which the mice were immunized. The results indicate the number of IFN-γ–producing CD4 T cells by ELISPOT analysis. The data are representative of two independent experiments (n = 3–5 mice per group).

We also determined whether immunization with the 3NO₂Tyr¹¹ and SO₃Tyr¹¹ TNF-α mutants induces a mutant-specific CD4 T-cell response. B6, Bm12, B10, B10.Q, and F2 (FVB × B6) mice were immunized with the 3NO₂Tyr¹¹ or SO₃Tyr¹¹ TNF-α mutants, and the specificity of CD4 T-cell recognition was tested by IFN-γ ELISPOT analysis. Again, immunization with “posttranslationally modified” TNF-α induces specific CD4 activation only against the mutated residues but not against the endogenous Lys¹¹ epitope (Fig. 3 E–H). Overall, these results indicate that site-specific modifications of TNF-α at position Lys¹¹ result in CD4 recognition of the neo-epitopes and no recognition of the wild-type epitope.

CD4 T-Cell Repertoire Against Wild-Type Lys¹¹TNF-α Epitope Is Absent.

The observation that no CD4 T-cell responses were detected against the wild-type epitope could be caused by differences in the processing of the proteins, by a weak immunization strategy not capable of activating T cells (the epitope recognized is a weak binder), or by the complete absence of a CD4 repertoire able to recognize the endogenous epitope. Peptide vaccination in the presence of a strong adjuvant (TiterMax) bypasses antigen processing, directly loads peptides onto the MHC molecules, and induces responses even to peptides with low affinity for the MHC molecules (44). To determine whether CD4 T-cell responses exist against a peptide presented directly onto I-A^b, mice were immunized with wild-type Lys¹¹ peptide (WT-pep5) or with pNO₂Phe¹¹ peptide (pNO₂Phe-pep5) as a positive control in TiterMax (Fig. 4). Draining lymph nodes were isolated 1 wk after vaccination, and CD4 T cells were purified and placed in a coculture with APC. To test for specific responses, APC were left untreated (no peptide) or were pulsed with either the same peptide contained in the immunization (cognate peptide) or with a peptide derived from Mycobacterium tuberculosis known to bind I-A^b (irrelevant peptide). The results shown in Fig. 4 demonstrate that CD4 T cells isolated from mice immunized with pNO₂Phe-pep5 produce IFN-γ and indicate, not surprisingly, the presence of a T-cell repertoire against this mutant. Conversely, CD4 T cells isolated from mice immunized with the WT-pep5 did not produce IFN-γ in the presence of the WT-pep5 peptide, demonstrating that no CD4 T cells in the peripheral repertoire of B6 mice recognize this endogenous epitope. These results are consistent with the ontogeny of the immune system and the development of CD4 T cells, with autoreactive T cells absent from peripheral circulation.

Fig. 4.

Absence of CD4 T-cell responses against wild-type TNF-α. B6 mice were vaccinated once with 250 μg of either the WT-pep5 or pNO₂Phe-pep5 peptide in TiterMax in the footpad. CD4 T cells isolated 1 wk later from the draining lymph nodes were incubated with 4 μg of the same peptide used for vaccination (cognate) or with a peptide derived from Mycobacterium tuberculosis Ag85 known to bind I-A^b. Results indicate number of IFN-γ–producing CD4 T cells by ELISPOT analysis. The data are representative of two independent experiments (n = 3 mice per group).

Posttranslational Modifications of Tyr Break Immunological Tolerance in EGF.

The immunization experiments described above showed that unnatural amino acids corresponding to the PTMs 3NO₂Tyr and SO₃Tyr can break tolerance in a fashion similar to pNO₂Phe. However, the native amino acid that was mutated in TNF-α is Lys, and not a Tyr. Under physiological conditions, a native Tyr residue is nitrated or sulfated posttranslationally to form 3NO₂Tyr or SO₃Tyr, respectively. To determine whether nitration or sulfation of a native Tyr residue can break immunological tolerance (and, at the same time, to test this approach in the context of a second protein), Tyr²⁹ in EGF was mutated to Phe²⁹, pNO₂Phe²⁹, 3NO₂Tyr²⁹, or SO₃Tyr²⁹. These proteins were administered in alum to B6 mice once a week for 4 wk, with a weekly dose of 50 μg. Sera were collected 1 wk after each immunization and were tested for antibodies against wild-type EGF protein. Autoantibody production was observed for all the site-specific modifications, including the PTMs (3NO₂Tyr and SO₃Tyr), pNO₂Phe, and Phe (Fig. 5 and Fig. S10). Therefore, both natural PTMs, generated by nitration or sulfation of a Tyr residue, and somatic mutations can break tolerance and induce antibody production against corresponding wild-type protein when introduced in the protein at the correct site and in the context of a specific MHC.

Fig. 5.

Immunogenicity of pNO₂Phe, 3NO₂Tyr, and SO₃Tyr EGF mutants. B6 mice were immunized once a week for 4 wk with 50 μg wild-type EGF, pNO₂Phe²⁹, 3NO₂Tyr²⁹, or SO₃Tyr²⁹ EGF mutants adjuvanted in alum. In all experiments, sera were collected 1 wk after each immunization, and serial dilutions were tested for the presence of antibodies against the wild-type EGF protein. The data indicate autoantibody responses at a 1:1,000 titer measured by ELISA. The data are representative of two independent experiments (n = 3–5 mice per group).

Discussion

In this study, we show that the incorporation of somatic mutations (Tyr in TNF-α, and Phe in EGF), unnatural amino acids (pNO₂Phe), and PTMs (3NO₂Tyr and SO₃Tyr) at specific sites in a self-protein are able to induce a strong CD4-dependent (IgG isotype) autoantibody response against the native protein. Mechanistic studies with TNF-α indicate that the immunogenicity of each mutant depends on the ability of a specific class II molecule to present peptides with the altered residue. In particular, the incorporation of pNO₂Phe, Tyr, 3NO₂Tyr, and SO₃Tyr at position 11 of the TNF-α protein is immunogenic in hosts bearing the H-2^b haplotype. Furthermore, pNO₂Phe¹¹, 3NO₂Tyr¹¹, and SO₃Tyr¹¹ immunogenicity is MHC class II restricted, and the mutants require I-A^b hosts to generate autoantibodies, as confirmed by experiments in mice that are genetically identical except for the MHC locus. In addition, immunogenicity is dependent upon the recognition of the modified pNO₂Phe¹¹, 3NO₂Tyr¹¹, and SO₃Tyr¹¹ epitopes by CD4 T cells, and this recognition does not cross-react with the endogenous Lys¹¹ epitope in TNF-α. Lack of CD4 T-cell responses upon TiterMax-peptide immunization demonstrates that the peripheral CD4 T-cell repertoire could not be elicited against the endogenous Lys¹¹ epitope, consistent with the notion of a “hole in the T-cell repertoire,” probably caused by central tolerance. Modeling (45) and in silico analysis (41–43) of the Lys¹¹– and Tyr¹¹–TNF-α epitopes in different pockets of the I-A^b binding groove suggest that the wild-type peptide-binding core, Lys¹¹-Pro-Val-Ala-His-Val-Val-Ala-Asn (KPVAHVVAN), is a low-affinity binder, whereas the mutant peptide core, Tyr¹¹-Pro-Val-Ala-His-Val-Val-Ala-Asn (YPVAHVVAN), is a stronger binder and therefore has a better probability of being presented in vivo (Tables S6 and S7). Overall, we demonstrate that site-specific modifications of specific residues can induce the generation of autoreactive antibodies through the activation of CD4 T cells, which recognize these modifications restricted by a particular class II molecule. Because the activated CD4 T cells recognize only the mutant, but not the wild-type epitopes, self-tolerance is maintained at a T-cell level. Presumably, autoreactive antibodies are generated with assistance from CD4 T cells that recognize the mutated epitope presented on class II molecules of B cells, allowing class-switching and somatic hypermutation to proceed and generate long-lived polyclonal IgG antibodies.

The mechanism by which pNO₂Phe, 3NO₂Tyr, and SO₃Tyr break tolerance suggests that naturally occurring PTMs also can lead to a loss in self-tolerance. Indeed, it is postulated that PTMs may play an important but not clearly understood role in the initiation/progression of inflammatory diseases and autoimmune disorders. PTMs, including citrullination, glycosylation, phosphorylation, isoaspartylation, and nitrosylation, have been implicated in several autoimmune disorders such as multiple sclerosis, systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), celiac disease, and type I diabetes (3–6). Moreover, PTMs result from various biological processes, such as cellular stress, inflammation, and trauma, or from exposure to infectious agents such as viruses—processes that have been associated with autoimmune diseases. PTMs alter the chemical structure of native antigens and generate neo-epitopes that may not have been subject to thymic selection. These new epitopes then are potential targets for APCs and may result in an autoimmune response involving autoantibody production by B cells. For instance, citrullinated fibrinogen has been observed in RA patients, and autoantibodies against citrullinated fibrinogen generally are used as a diagnostic tool to detect RA (9, 13, 14). Moreover, the induction of RA by vaccination of citrullinated fibrinogen with Freund's incomplete adjuvant in DR4-IE transgenic mice has been reported (9). Another PTM, isoaspartyl, has been observed in small nuclear ribonucleoproteins in patients suffering from SLE (46), and immunohistochemical evidence of 3NO₂Tyr has been reported in patients with acute multiple sclerosis lesions (47), experimental autoimmune encephalomyelitis (48, 49), and type I diabetes (24). Posttranslational modifications also result from the nonenzymatic glycosylation of Arg and Lys residues in diabetes and age-related diseases (50–53). This modification results in the formation of advanced glycation end products (AGEs), which also may generate neo-epitopes that result in self-reactive antibodies and contribute to vascular disease and neuropathy.

Considerable literature suggests a role for PTMs in autoimmune disorders; however, most experimental models investigating PTMs involve vaccination with modified peptides in the presence of adjuvants. The relevance of our approach lies in the use of chemically defined, site-specific modifications that can be studied in the context of full-length proteins that subsequently are processed and presented. For example, we explored the effects of protein nitration or sulfation by replacing Tyr²⁹ in EGF with the corresponding modified residues. Our results indicate that these “synthetic” posttranslational modifications lead to the generation of antibodies that are cross-reactive with native, unaltered EGF. Moreover, we show that nitration of a distinct residue at a particular site in a full-length self-antigen breaks tolerance in the context of a specific MHC, consistent with the MHC association of antibody-mediated autoimmune diseases. We currently are attempting to extend this proof of principle experiment to (i) known antigens in antibody-dependent autoimmune disorders, including the noncollagenous 1 domain of α3 and α5 chains of type IV collagen (54, 55) in Goodpasture disease, which is mediated by anti-glomerular basement membrane antibodies, and (ii) AGE-modified proteins, which have been characterized in diabetes.

PTMs also have been associated with malignant transformation and are believed to contribute to oncogenic signaling (18, 56, 57). Most cancer antigens described to date are nonmutated self-antigens against which tolerance has been established (58); however, the immune system still is able to recognize tumors, suggesting that the presence or absence of PTMs may play a role in their recognition. For instance, the mucin-like glycoprotein MUC1 is overexpressed in epithelial tumor cells, but it bears incomplete glycosylation (59) compared with the healthy tissues. Studies have determined that the presence of hypoglycosylated motifs is required for effective immune responses (60), and the presence of a tumor-associated Tn antigen (GalNAc-O-S/T) aids in strong T-cell and antibody responses. Phosphorylation is one of the most common and ubiquitous PTMs, and tumor-associated phosphoproteins give rise to phosphoepitopes complexed to the MHC molecules (61–64) that interact with TCRs (65), as evidenced by studies on phosphopeptides. Thus, enhancements in binding affinity or alterations in the antigenic identity of presented epitopes can increase T-cell recognition. These studies suggest that other PTMs, such as 3NO₂Tyr and SO₃Tyr, may be recognized by the immune system in cancers. Indeed, increased nitration of Tyr has been observed in different cancer tissues; specifically, higher levels of 3NO₂Tyr modified serum protein have been observed in lung cancer patients (66, 67). Furthermore, 3NO₂Tyr immunostaining experiments on lung tumors and normal lung tissues have shown that structural protein, metabolic enzymes, and other proteins in tumor tissues undergo a higher degree of nitration than seen in normal tissue (68). Therefore, it is foreseeable that a vaccine incorporating a site-specific 3NO₂Tyr-modified antigen might activate T cells that would be directed differentially to tumors relative to nontumor tissues.

Materials and Methods

Mice and Immunization Strategies.

All experiments were performed in accordance with the National Institutes of Health Animal Protection Guidelines and were approved by the Genomics Institute of the Novartis Research Foundation Animal Care and Use Committee. C57BL/6 (B6), BALB/c, C3H, and FVB/N mice were bred internally. C57BL/10SnJ (B10), B10.D1-H2^q/SgJ (B10.Q), and B6(C)-H2-Ab1^bm12/KhEgJ (Bm12) mice were purchased from the Jackson Laboratory. CD-1 mice were purchased from Charles River. All mice used in this study were 6- to 10-wk-old females. Mice were vaccinated s.c. once a week for 4 wk with 5 μg (TNF-α) or 50 μg (EGF) of protein at a 1:1 ratio with Alu-Gel-S (Serva Elecrophoresis GmbH) in a final volume of 200 μL. Serum was obtained through retro-orbital eye bleeding at 2, 3, and 4 wk after the initial vaccination. In some experiments, mice were killed 1 wk after the last immunization, and purified CD4 T cells were analyzed by an ELISPOT assay. For peptide immunization, 25 μL of 10 mg/mL peptide mixed with 25 μL of Titermax Gold adjuvant (Sigma) was vortexed for 15 min and injected into the left footpad. Cells were isolated from the left popliteal lymph node 1 wk after immunization and were used in an ELISPOT assay.