Iodination of Tyrosyls in Thyroglobulin Generates Neoantigenic Determinants That Cause Thyroiditis¹

Among the known autoantigens, thyroglobulin (Tg)³ is unique in its ability to incorporate and store available iodine in the form of iodotyrosyl residues (1). This process facilitates thyroid hormone, i.e., thyroxine (T4) and triiodothyronine, formation through intramolecular coupling of specific iodotyrosyls, but it also has immunological consequences: enhanced iodination of Tg has been known to increase its immunogenicity at the T and B cell level, as well as its pathogenicity in experimental animals (2, 3, 4, 5). The mechanisms underlying these observations remain mostly unknown, but progress with T cell epitope mapping in Tg has recently shed light on some of the processes involved.

So far, 13 Tg peptides encompassing T cell epitopes that cause experimental autoimmune thyroiditis (EAT) have been identified, and none of them has been classified as dominant (6). Eleven of these peptides elicit EAT of considerable severity but do not contain iodine, clearly highlighting that iodine atoms per se are not necessary for thyroiditogenicity. However, processing of highly iodinated Tg (I-Tg), containing 55–70 atoms per monomeric subunit, has been shown to facilitate generation of one of these pathogenic peptides (aa 2494–2510) (3), in agreement with earlier studies showing that the proteolytic degradation of Tg is affected by its iodine content (7, 8, 9). In contrast, experiments with Tg peptides containing hormonogenic sites have shown that iodine atoms can be an integral part of ligands recognized by thyroid-infiltrating T cells. The 12-mer Tg peptide (2549–2560), containing T4 at aa position 2553 (T4(2553)) has been reported to elicit both proliferative and cytotoxic T cell responses (10, 11, 12) and to cause lymphocytic as well as granulomatous EAT (13). Elegant studies by Wan et al. (12) using the thyronine (T0)-containing analog T0(2553) (T0 lacks the four iodine atoms of T4) found that lymph node cells (LNC) from mice challenged with the T4(2553) peptide could not be cross-stimulated in vitro with T0(2553), and this regimen failed to generate effector cells that could transfer EAT. These data suggested, for the first time, that iodine atoms sufficiently modify the peptide-MHC complex to elicit a distinct subset of thyroid-infiltrating T cells that recognize only the iodinated determinant. The presence of the bulky two-phenyl-ring side chain is not sufficient to impart immunogenicity, because other T4-containing peptides were either mildly pathogenic or were devoid of immunogenicity (12).

Prompted by the above observations, and by the fact that most of the iodine in Tg is found in iodotyrosyls outside the hormonogenic sites (1), we have hypothesized that Tg may harbor several pathogenic T cell epitopes that contain iodotyrosyls. In this study, we undertook a systematic search of Tg to localize such posttranslationally modified neoantigenic determinants that would have escaped detection in earlier mapping studies. Further goals were to examine whether iodine-modified epitopes were present in normal or I-Tg and whether they comprised immunodominant sites.

Female CBA/J (H-2^k) mice, 4–6 wk of age, were purchased from The Jackson Laboratory. All experimental procedures were reviewed and approved by the Animal Care Committee at Memorial University of Newfoundland (Newfoundland, Canada). All Tg peptides were synthesized by Dalton Chemical Laboratories and were used in experiments at >80% purity. F-moc-3-iodo-tyrosine, containing one iodine atom at position 3 of the phenolic ring, was used for the synthesis of iodinated peptide analogs. All peptides were blocked with an acetyl group at the N terminus and with an amide group at the C terminus; thiol group of internal Cys residues was blocked by acetamide. Tg was purified from frozen thyroid glands of outbred ICR mice (Harlan Bioproducts for Science) by passing thyroid homogenates through a Sepharose CL-4B column as described previously (14).

Mice were challenged s.c. with 100 nmol of Tg peptides in 100 μl of CFA emulsion (with Mycobacterium butyricum; Difco Laboratories). Nine days later, inguinal, axillary, and branchial lymph nodes were collected aseptically, and single cell suspensions were prepared in complete DMEM supplemented with 10% FBS (Cansera), 20 mM HEPES buffer, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin (all obtained from Invitrogen Life Technologies), and 5 × 10^-5 M 2-ME (Sigma Chemical). After centrifugation and washing, 4 × 10⁵ cells/well per 200 μl were cultured in 96-well plates for 4 days in the presence of titrated amounts of the appropriate Ags. Cell labeling and harvesting procedures were performed as described previously (15). Stimulation index is defined as follows: (cpm in the presence of Ag)/(cpm in the absence of Ag).

The IL-2-secreting hybridomas 4A6, 10C1, and 1H7 specific for I-p117, I-p304, and I-p1931, respectively, were produced by fusion of peptide-primed LNC with the BW5147 α⁻β⁻ variant (16), a gift from P. Marrack (Howard Hughes Medical Institute, Denver, CO), following a method described earlier (17). The A^k-restricted 4A12 T cell hybridoma specific for the Tg peptide p2498 (17) was used in inhibition assays. The IL-2-dependent CTLL-2 line (18) and the LK35.2 APC line (19) were purchased from American Type Culture Collection. Dendritic cells (DCs) were prepared from bone marrow precursor cells cultured in the presence of GM-CSF, as described previously (20). Detection of Tg- or Tg peptide-reactive IgG Abs in pooled mouse sera was determined by ELISA as described in an earlier study (15).

Mice were challenged s.c. with 100 nmol of peptide in CFA emulsion and were boosted 3 wk later with 50 nmol of the same peptide in IFA. EAT was assessed 35 days after the initial challenge. Induction of thyroiditis by adoptive transfer was performed as described previously (15). Briefly, LNCs from peptide-primed donor mice were cultured for 72 h in the presence of 20 μM of the respective peptide. After harvesting and washing, 2 × 10⁷ cells in 200 μl of PBS were injected i.p. into syngeneic recipients (six mice per group). EAT was assessed 14 days posttransfer. Fixation, embedding, and sectioning of thyroids were performed as described previously (15). Histological sections of thyroid lobes were stained with H&E, and the mononuclear cell infiltration index (II) was scored as follows: 0, no infiltration; 1, small interstitial accumulation distributed between two or more follicles; 2, one or two foci of inflammatory cells more than the size of one follicle; 3, diffuse infiltration of 10–40% of total area; 4, extensive infiltration, 40–80% of total area; and 5, extensive infiltration over 80% of total area.

Because I-A^k genes are known to control susceptibility to Tg-induced EAT (21, 22, 23), we scanned the complete murine Tg sequence (24) for the presence of I-A^k-binding motifs flanked by Tyr residue(s), using the algorithm described by Altuvia et al. (25). Twenty sites meeting these criteria were identified and from these, a total of 13 peptides, ranging in size from 11 to 17 aa residues, were synthesized in their noniodinated or iodinated, i.e., iodotyrosyl-containing, form (Table I). CBA/J mice were s.c. challenged with 200 nmol of each peptide analog in adjuvant, and 9 days later, cells from draining lymph nodes (LNC) were collected and cultured in the presence of varying concentrations of the respective peptides. Seven of 13 peptides elicited no responses regardless of their iodination status. Three Tg peptides were immunogenic in their noniodinated form, and iodotyrosyl formation had variable effects, because it increased (p179), decreased (p2540), or did not alter (p2529) their immunogenic profile (our unpublished observations). However, three peptides (p117, p304, and p1931) (Fig. 1,A) were nonimmunogenic in their noniodinated form, but their iodotyrosyl-containing analogs induced significant LNC responses (Fig. 1, B–D) accompanied by IL-2 as well as IFN-γ secretion in vitro (data not shown). Furthermore, each iodinated analog activated LNC that did not cross-react with the noniodinated form of the peptide (data not shown), indicating that iodine heavily influenced the immunogenicity as well as the antigenicity of these determinants. This was confirmed at the clonal T cell level using the hybridomas 4A6, 10C1, and 1H7, which secreted IL-2 upon culture with DC-presenting I-p117, I-p304, and I-p1931, respectively, but did not recognize equimolar amounts of the noniodinated analogs (Fig. 1, E–G). All three hybridomas were A^k-restricted (data not shown). The same hybridoma clones were not activated by DC processing intact Tg, suggesting that the iodinated peptides are not immunodominant (Fig. 1, E–G).

To investigate whether iodine enhanced the immunogenicity of these peptides by promoting peptide binding to MHC, we performed a competition assay using the 4A12 T cell hybridoma clone, which was previously (17) shown to be A^k-restricted, and reactive against the p2498 (aa 2498–2506) epitope of Tg (Fig. 2,A). The 4A12 cells were activated with 1 μM p2498 in the presence of increasing concentrations of inhibitor peptides using the LK35.2 (A^k-expressing) APC line. It was observed that p117 and p304 could not inhibit 4A12 activation at the 10–100 μM range (Fig. 2, B and C), whereas their iodinated analogs, at equimolar concentrations, significantly diminished 4A12 activation. These results suggested that p117 and p304 are not A^k-binders and that iodotyrosyl formation within these sequences promoted their binding to the A^k molecule, leading to enhancement of their immunogenicity in CBA/J mice. In contrast, both the iodinated and noniodinated analogs of p1931 significantly inhibited the activation of 4A12 (Fig. 2 D), suggesting that, in this case, the iodotyrosyl side chain must face away from the MHC cleft and make direct contact with the TCR. The enhanced immunogenicity of I-p1931 would result from recruitment of T cells able to recognize only the iodine-modified peptide-MHC complex.

The pathogenicity of I-p117, I-p304, and I-p1931 was subsequently tested by direct challenge of CBA/J mice (six mice per group) with the corresponding peptide in adjuvant or by adoptive transfer of peptide-primed LNC into naive CBA/J hosts. I-p117 and I-p1931 elicited thyroiditis by either experimental protocol, although the severity of EAT was higher by the adoptive transfer method (mean infiltration index of 0.67 vs 2.33 for I-p117, and 0.83 vs 1.67 for I-p1931) (Table II and Fig. 3). I-p304 was thyroiditogenic only by the adoptive transfer method (mean infiltration index of 0.83). At the B cell level, only I-p117 was strongly immunogenic eliciting serum IgG responses to itself and native Tg (Table II). There was no cross-reactivity with the noniodinated analog, suggesting that I-p117 is localized at the surface of the intact Tg molecule. Interestingly, previous studies have shown that the Tyr residue at position 130 is an early iodination site in Tg (26).

The iodine content of Tg varies widely (0.1–1.0% of weight) with iodine availability, but on average, 14 of the 26 atoms of iodine within each 19S dimer are stored in the form of mono- and di-iodotyrosines (1). The present findings demonstrate that iodotyrosyls at residues 130, 306, and 1942 contribute to the formation of neoantigenic T cell epitopes, which elicit EAT in CBA/J mice. These data and the lack of antigenicity of the noniodinated analogs p117, p304, and p1931 further suggest that Tg harbors iodotyrosyls at these residues under normal, steady-state conditions. Despite their nondominant nature, these peptides must be generated intrathyroidally and be recognized by the adoptively transferred effector cells mediating EAT. Thus, the current and earlier epitope mapping studies (6) delineate at least three distinct categories of T cell determinants with pathogenic potential in Tg: 1) T4-containing peptides in which iodine atoms are an integral part of the TCR ligand (10, 11, 27, 28); 2) iodotyrosyl-containing peptides, as described in this study; and 3) noniodinated epitopes (6). It is yet unknown whether di-iodotyrosyls contribute in the formation of determinants that elicit EAT.

Formation of pathogenic Tg epitopes by iodine presents another example of posttranslational modifications associated with autoimmune responses at both the T or B cell levels (29, 30). The bulky iodine atoms (atomic radius of ∼133 pm) in the iodotyrosyl side chain can be critical in forming a TCR-contact residue, as suggested by the results with the I-p1931 peptide in this study, and by earlier studies showing that iodine atoms on the longer thyroxyl side chain in T4(2553) influence its recognition by clonal or polyclonal T cells (11, 27). In addition, the findings with the I-p117 and I-p304, highlight, for the first time, that iodotyrosyl formation facilitates peptide binding to MHC. Specific recognition of iodine-modified Tg epitopes by B cells has been suggested by earlier studies (4) and is demonstrated by the observation that I-p117-specific IgG does not bind to p117. The Y130 residue within I-p117 was reported as an early iodination site in Tg (26), but this was not confirmed by another study (31). Recognition of iodine-modified Tg peptides by Abs has been also noted even following formation of peptide-MHC complexes. For example, a mAb recognizing the 5′ iodine atom of the outer phenolic ring of T4 has been shown to inhibit T cell recognition of the T4(2553)-MHC complex (27).

Enhanced iodotyrosyl formation, catalyzed by thyroid peroxidase (1), may account, in part, for the increased immunogenicity of I-Tg (2, 3, 4, 5). At the level of individual determinants, however, iodotyrosyls not only contribute to the formation of neoantigenic epitopes, but may have neutral or immunomodulatory effects (our unpublished observations with the p179, p2529, and p2540 peptides). Iodine atoms per se do not necessarily impart immunogenicity, as exemplified by I-p681 and I-p757, which encompass early iodination sites (31) but are not immunogenic. In addition, the noniodinated Tg peptide (306–320) was previously identified as pathogenic (15), whereas the overlapping noniodinated (304–318) fragment was not immunogenic in this study, possibly reflecting effects of flanking residues on recognition by TCR. These effects will be better understood once the relative position of iodotyrosyl vis-a-vis the minimal T cell epitopes within those sequences is delineated. As shown from the peptides studied, however, the relative location of the tyrosyl ring does not seem to influence peptide immunogenicity. For example, both 16-mer p117 and p179 contain Tyr at amino acid position 14, but only the latter peptide is immunogenic. Furthermore, both iodinated analogs of p304 and p610 contain iodotyrosine at amino acid position 3, but only the former peptide is immunogenic.

The iodine content of Tg is known to affect its structure (32) as well as its proteolytic degradation (8, 9), and altered processing of I-Tg in APC has been shown to facilitate generation of cryptic noniodinated pathogenic determinants, such as the p2494 peptide (3). The I-p117, I-p304, and I-p1931 peptides are not generated by the processing of I-Tg in DC (our unpublished observations), but they could be generated following processing of Tg-Ab immune complexes, as has been shown for other noniodinated determinants (14). Lastly, enhanced iodination of Tg may exert immunomodulatory influences because it has been reported that removal of iodine from tryptic human Tg fragments converts them from immunogenic to tolerogenic (33). Our results demonstrate the existence of a new group of pathogenic Tg determinants that cannot be detected by conventional mapping methods. Although their potential role in the development of clinical disease remains to be elucidated, they provide a new insight as to how an environmental trigger (iodine supply) may influence the development of thyroid disease.

The authors have no financial conflict of interest.