Induction of murine thyroiditis by a non dominant E^k-restricted peptide of human thyroglobulin

Abstract

We have previously shown that the human thyroglobulin (hTg) 20-mer peptide p2340 (aa 2340–2359) contains an epitope recognized by Tg-reactive B cells in patients with Graves' disease. The presence of several E^k-binding motifs within p2340 prompted us to examine whether this peptide can stimulate a T-cell response and elicit experimental autoimmune thyroiditis (EAT) in AKR/J (H-2^k) mice. The peptide was found to be immunogenic at the T-cell level since it induced specific proliferative responses as well as interleukin-2 and interferon-γ secretion in secondary cultures of peptide-primed lymph node cells (LNC). The p2340-specific proliferation was blocked almost completely by an E^k-specific monoclonal antibody (mAb) but was unaffected by a control A^k-specific mAb. Peptide-primed LNC did not respond to intact hTg and conversely, LNC primed in vivo with hTg did not respond to p2340 in culture, suggesting that p2340 contains non-dominant T-cell epitope(s). Direct subcutanaeous challenge of AKR/J mice (n = 9) with p2340 in adjuvant, elicited mild to moderate EAT (infiltration index of 1–2) and strong p2340-specific immunoglobulin G responses in all mice tested. These data delineate a new thyroiditogenic sequence within the carboxyl terminal region of hTg.

Introduction

Mice challenged with thyroglobulin (Tg) develop experimental autoimmune thyroiditis (EAT), a T-cell-mediated disease that resembles Hashimoto's thyroiditis in humans.¹ Tg is the most abundant protein of the thyroid² and the largest autoantigen known, consisting of two identical 330 000 MW subunits, each comprised of a 19 amino acid (aa) signal peptide plus 2749 aa residues.³ The large size of Tg has posed a unique challenge in attempts to map its pathogenic T-cell determinants by conventional biochemical methods. Despite this difficulty, work from various laboratories over the last 10 years has identified five immunopathogenic Tg peptides encompassing at least six distinct T-cell epitopes.⁴ Four of these peptides are clustered toward the C-terminal end of Tg, a region that shares significant homology with the catalytic subunit of acetylcholinesterase (AchE). ^5,6

Previous studies have shown that CD4⁺ T cells are both sufficient and necessary in EAT induction because the disease can be transferred into naïve recipients by Tg-specific CD4⁺ T-cell lines or clones^7,8 and it does not develop in nude mice.⁹ However, Tg epitopes that can activate autoreactive CD4⁺ cells still remain mostly uncharted given that, to date, only three Tg peptides are known to be MHC class II binders.⁴ With the view to continuing the search for MHC class II-restricted Tg determinants, we focused in this study our attention to Tg peptide p2340 (aa 2340–2359) which we have previously shown to be a target of Tg-reactive autoantibodies occurring in Tg/AChE-reacting sera from patients with Graves' disease.¹⁰ The presence of several E^k-binding motifs^11,12 within this site, and a previous finding that an E^k-binding Tg peptide causes EAT¹³, prompted us to examine whether p2340 could elicit T cells and thyroiditis in H-2^k mice that are known to be susceptible to Tg-induced EAT.¹⁴

Materials and methods

Animals and antigens

Female AKR/J mice were purchased from Jackson Laboratories (Bar Harbor, ME) and were used in experiments from 6 to 10 weeks of age. Human Tg (hTg) was purifed as described by Schardt et al.¹⁵ Briefly, frozen thyroids were homogenized in 0·15 m KCl and the supernatant was centrifuged at 78 000 g. After two successive precipitations of the supernatant with ammonium sulphate, Tg was isolated by gel filtration using a Sepharose 6B (Pharmacia) column. The Tg-containing fractions were pooled, concentrated to 5 mg/ml in phosphate-buffered saline (PBS), filter sterilized and stored in aliquots at −70°. The hTg peptide QVAALTWVQTHIRGFGGDPR (aa 2340–2359, p2340) was synthesized by Sigma Genosys (The Woodlands, TX) and had its N- and C-terminals blocked with an acetyl and amide group, respectively. The hTg peptide PYEFSRKVPTFATPWPDFVP (aa 2652–2671, p2652) was synthesized by Genosys Biotechnologies (Cambridge, UK). The amino acid coordinates of the peptides were previously¹⁰ given on the basis of the original hTg sequence published in 1987 by Malthiery and Lissitzky,¹⁶ but they were re-assigned in this study according to the revised hTg sequence data by van de Graaf et al.³ Both peptides were used in all experiments at >80% purity as determined by analytical high pressure liquid chromatography (HPLC) and mass spectral analyses.

Lymph node cell (LNC) proliferation assays

Mice were immunized subcutaneously (s.c.) at the base of the tail with hTg or p2340 emulsified in complete Freund's adjuvant (CFA) (Sigma Chemical Co., St. Louis, MO). Nine to 10 days later, the inguinal lymph nodes were collected aseptically and single cell suspensions were prepared in Dulbecco's modified Eagle's minimal essential medium (DMEM; Biochrom, Berlin, Germany) supplemented with 10% fetal bovine serum (FBS; Gibco, Paisley, UK), 20 mm HEPES buffer, 2 mm l-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin (all from Gibco) and 5 × 10⁻⁵ m 2-mercaptoethanol (2-ME; Sigma). The LNC were washed and cultured (4 × 10⁵/200 µl/well) with or without antigen in flat-bottomed 96-well plates (Costar, Corning Inc., NY) for 4 days at 37° in a 10% CO₂, 90% air-humidified incubator. Eighteen hr before harvesting, 1 µCi of [³H]TdR (6·7 Ci/mmol, Amersham Pharmacia, Uppsala, Sweden) was added to each well in 25 µl of medium. The cells were harvested using a semiautomatic cell harvester (Skatron, Inc., Tranby, Norway) and counted in a liquid scintillation counter (1450 microbetaTrilux, Wallac, Turku, Finland). Stimulation index is defined as: (c.p.m. in the presence of antigen/c.p.m. in the absence of antigen). Monoclonal antibodies (mAbs) were purified by affinity chromatography on protein G-Sepharose 4 Fast Flow columns (Pharmacia) from culture supernatants of the hybridomas 10–3.6.2 (immunoglobulin G2a, IgG2a) reactive with I-A^k17 and 14-4-4S (IgG2a) reactive with I-E^k18 purchased from the American Type Culture Collection (ATCC, Manassas, VA). Inhibition of proliferation by blocking mAbs is expressed as: percentage inhibition = [1 − (c.p.m. in the presence of mAb)/(c.p.m. in the absence of mAb)] × 100.

Detection of cytokines and peptide-specific IgG by enzyme-linked immunosorbent assay (ELISA)

Detection of cytokines in culture supernatants was performed by sandwich ELISA based on non-competing pairs of capture and detection (biotinylated) anti-cytokine mAbs as follows: interleukin (IL)-2, JES6-1A12 and JES6-5H4; interferon (IFN)-γ, R4-6A2 and XMG1.2; (PharMingen, San Diego, CA); IL-4, 11B11 (ATCC) and DVD-6-24G2 (PharMingen). Alkaline phosphatase-conjugated streptavidin was purchased from Sigma. Quantitation was extrapolated from standard curves generated for each cytokine using known amounts of murine rIL-2 and rIFN-γ (PharMingen) and rIL-4 (PeproTech, Rocky Hill NJ). Peptide-specific IgG in immune sera was determined by ELISA as previously described¹⁹ using an alkaline phosphatase-conjugated goat anti-mouse IgG (Sigma) as the second antibody. Light absorption of the p-nitrophenolate product at 405 nm was measured using a Dynatech MR5000 plate reader (Dynatech, Chantilly, VA).

EAT induction

Mice were challenged s.c. at the base of the tail with p2340 in CFA and three weeks later, they were boosted s.c. with the same peptide in incomplete Freund's adjuvant (IFA). Two weeks after the second challenge, mice were bled from the retrobulbar sinus to obtain sera for ELISA, and the thyroids were removed and fixed in buffered formalin. Thyroid sections were stained with hematoxylin and eosin and the mononuclear cell infiltration index (II) was scored as follows: 0 = no infiltration; 1 = interstitial accumulation of inflammatory cells distributed between two or more follicles; 2 = one to two foci of inflammatory cells at least the size of one follicle; 3 = extensive infiltration, 10–40% of total area; 4 = extensive infiltration, 40–80% of total area; 5 = extensive infiltration, >80% of total area.

Results

P2340 carries multiple E^k-binding motifs

The identification of the 20-mer hTg peptide p2340 as a site recognized by hTg-reactive IgG antibodies in Graves' disease invited a closer examination of its features. The fact that it differs only at four aa positions (80% identity) from its mouse Tg homologue (Fig. 1a), prompted us to examine whether it contained mouse major histocompatibility complex (MHC)-binding motifs that might render it immunogenic in H-2^k mice that are susceptible to Tg-induced EAT. It turned out that p2340 contains five overlapping E^k-binding motifs (Fig. 1a) according to the algorithm described by Altuvia et al.¹¹ which was developed following a compilation of an extended data base of helper T-cell sites and takes into account physical-chemical and structural properties of peptides that can be extrapolated by aa sequence data. Two of these motifs (C and E) are identically expressed in the mouse Tg (mTg) sequence, whereas one of the motifs D is shared between mTg and hTg with a conservative substitution T→S at aa position 2349 (Fig. 1a). In addition, p2340 contains three overlapping E^k-binding motifs (Fig. 1b) obeying the pattern described by Leighton et al.¹² in which hydrophobic, short chain, or aromatic residues are localized six to eight positions before a basic residue. One of these motifs (motif 3) is again identically shared with the mTg sequence with the exception of the conservative T2349S substitution (Fig. 1b). It is noteworthy that a similar number of motifs was not found in any of 10 randomly chosen 20-mer peptides in hTg (data not shown). The above features strongly suggested p2340 as an E^k-binder and because another E^k-binding Tg peptide (peptide (2496–2504)) had been previously shown to cause EAT¹³ these data encouraged us to test the EAT-inducing capacity of p2340 in H-2^k mice.

Figure 1.

Alignment of the 20-mer human Tg peptide p2340 with its mouse homologue (mTg) indicating the relative positions of E^k-binding motifs. (a) Boxed sequences delineate motifs identified by pattern searches according to Altuvia et al.¹¹ Motif B (aa 2344–2352): [AFILPVW]-X-[ACFILMPTVWY]-{DEAGS}-[KRHQY]-X-{NQ}-{NQFHWY}-[DEKNQR]; motif C (aa 2342–2348): [AILSTV]-{NQ}-[ACFILMPTVWY]-{DE}-[ACFILMPTVWY]-{NQ}-[DEKNQR]; motif D (aa 2345–2350) and (aa 2347–2352): [ACFILMPTVWY]-X-[ACFILMPTVWY]-{DE}-{DE}-{HKR}-{NQFHWY}-{DENQFHWY}-[KR]; motif E (aa.2342–2346): A-{HKR}-[ACFILMPTVWY]-{FHWY}-[KRWCHNQY]. [ ] denotes inclusion and { } denotes exclusion of the indicated aa at this position, X indicates any aa; (b) Overlapping E^k-binding motifs delineated by arrows according to the algorithm of Leighton et al.¹² in which two hydrophobic (A,V,I,L), short chain (S,T), or aromatic (Y,W,F) residues are localized six to eight positions before a basic residue (K,R,H). The aa substitutions between hTg and mTg sequences are underlined.

P2340 induces T helper 1 (Th1) cells but lacks dominant T-cell determinants

To determine the immunogenicity of p2340, AKR/J mice were s.c. challenged with 100 nmol of peptide, and 10 days later the antigen-specific proliferative responses of their inguinal LNC were examined. A strong recall response to p2340 was observed in the 0·1–10 µm range and it was specific because the same LNC population did not yield a detectable response to equimolar amounts of the control hTg peptide p2652. (Fig. 2a). In addition, p2340-primed LNC did not proliferate to 0·1–1 µm hTg in culture, suggesting that p2340 does not comprise of dominant T-cell epitope(s). This was further confirmed by the fact that hTg-primed LNC from AKR/J mice responded strongly to hTg in vitro but failed to respond under the same conditions to equimolar or much higher concentrations (0·1–10 µm) of free p2340 or p2652 in vitro (Fig. 2b). The strong recall response of p2340-primed LNC to the same peptide in vitro was strongly inhibited (78%) in the presence of an I-E^k-specific but not an I-A^k-specific mAb (Fig. 2c). The same response was characterised by release of significant amounts of IL-2 and IFN-γ, but not IL-4 in the culture supernatant (Fig. 2d). Collectively, the results support the view that p2340 is highly immunogenic in AKR hosts eliciting E^k-restricted Th1 cells and confirm the predictive value of the above algorithms in the identification of E^k -binding determinants.

Figure 2.

(a and b) Proliferative in vitro response of p2340-primed (a) or hTg-primed (b) LNC against the antigens shown. Inguinal LNC were pooled from two AKR/J mice s.c. primed 10 days earlier with 100 nmol of peptide (a), or 100 µg of hTg (b) in CFA. Background c.p.m. varied from 2400 to 2900 c.p.m. The results are representative of three independent experiments. (c) Inhibition of p2340-specific LNC proliferation in vitro in the presence of 2·5 µm p2340 and 40 µg/ml of I-A^k-specific or I-E^k-specific mAbs; the standard errors of the mean percentage inhibition values from triplicate wells were 4·2 and 5·7, respectively. Inguinal LNC were obtained from AKR/J mice primed as in (a). (d) Cytokine release in culture supernatants of p2340-primed LNC obtained as in (a) in the presence of 5 µm of the antigens shown. Cytokine concentrations were extrapolated from standard curves of sandwich ELISA as described in Materials and Methods. The data are representative of three independent experiments.

P2340 causes EAT in AKR/J mice

The pathogenicity of p2340 was assessed in two groups, (n = 4) and (n = 5), of AKR/J mice that were s.c. challenged with 100 nmol and 200 nmol p2340, respectively. Three weeks later, all mice were boosted with half the priming dose of p2340, and the thyroids were removed five weeks after the original challenge for histological examination. As shown in Table 1, all mice developed mild to moderate EAT (thyroid infiltration index of 1–2) with foci of mononuclear cell infiltrates concentrated mostly around capillaries or accumulated interstitially between follicles. (Fig. 3). Pathology appeared to be organ specific because mononuclear cell infiltrates were not detected in kidneys or livers of these animals (data not shown). All mice also elicited strong specific IgG responses to p2340 but these antibodies did not bind to intact hTg in ELISA (Fig. 4). These data support the view that p2340 encompasses epitopes that can elicit thyroid-infiltrating effector T cells in EAT.

Table 1.

EAT induction by the human thyroglobulin peptide p2340

		II†
Dose in vivo (nmol)*	Number of mice	0	1	2	3	Mice with EAT	Mean II ± SD
		100	4	0	2			2	0	4/4	1·5 ± 0·6
200	5	0	3	2	0	5/5	1·4 ± 0·5

^*AKR/J mice were s.c. primed with the indicated dose of peptide in CFA and 2 weeks later, they were boosted with half the immunizing dose of peptide in IFA. EAT was assessed 5 weeks after the initial challenge. Three control mice were immunized with 60 µg/mouse of ovalbumin (Sigma) in CFA and 2 weeks later, they were boosted with half of the immunizing dose of ovalbumin in IFA, Intrathyroidal mononuclear cell infiltration was not detected (II = 0) in any of the control mice.

^†Thyroid sections were stained with hematoxylin and eosin and scored for infiltration index as described in Materials and Methods.

Figure 3.

Histological appearance of thyroids of AKR/J mice: (a) normal gland, 100×; (b) and (c) interstitial or focal accumulation of mononuclear cell infiltrates after mouse challenge with p2340 in CFA. (b) II = 1, 200×; (c) II = 2, 200×.

Figure 4.

Representative IgG responses against the antigens shown in 5 week immune sera of AKR/J mice (n = 4) immunized with 100 nmol of p2340, as in Table 1. Reactivity was determined by an alkaline phosphatase-based ELISA as described in Materials and Methods.

Discussion

This study was initiated by the serendipitous finding of several E^k-binding motifs within the hTg peptide p2340 which was previously identified as a target site for hTg-reactive antibodies in Graves' disease.¹⁰ Of particular interest was the fact that some of these motifs are expressed either identically or with one conservative substitution (T2349S) in the mTg sequence, prompting an investigation into the pathogenic potential of p2340 in mouse EAT. Our results demonstrate that p2340 elicits E^k-restricted Th1 cells and thyroiditis in AKR/J mice confirming the predictive value of such algorithms in mapping pathogenic T-cell determinants in Tg. The choice of AKR/J mice in our study was arbitrary and based solely on the knowledge that this strain is susceptible to Tg-induced EAT,¹⁴ but preliminary experiments have shown that p2340 is also pathogenic in other H-2^k strains such as CBA/J (P. Lymberi, unpublished data).

Although the exact boundaries of the minimal T-cell epitope(s) within p2340 remain to be established, this peptide is being identified as only the second Tg site – after the 9-mer (2496–2504)^4,13 – that is known to encompass E^k-restricted pathogenic T-cell determinant (s). The existence of such epitopes is compatible with the fact that susceptibility to Tg-induced EAT maps to the I-A^k subregion²⁰ only if one assumes either that the immunodominant T-cell epitopes in Tg are A^k-restricted or that the majority of pathogenic Tg determinants bind to A^k, thus masking the contribution of E^k-restricted T cells to the pathology of EAT. The observation that p2340 is not immunodominant agrees with the former hypothesis but it is noteworthy that, so far, none of the mapped Tg T-cell epitopes has been identified as dominant. As mentioned previously^4,21 this may occur possibly because the large size of Tg poses constraints on the quantitative yield of any given 2000 MW epitope following its processing in vivo or in vitro. The arbitrary nature of the criteria for immunodominance using proliferative assays with intact antigen²² comes into focus when one considers that all pathogenic Tg peptides – dominant or cryptic – must be generated intrathyroidally in order to allow specific homing of effector lymphocytes. We envisage two mutually non-exclusive processes to explain this apparent conundrum. First, as previously postulated^4,19,23 intrathyroidal antigen presenting cells (APC) such as dendritic cells²⁴ may possess proteases that cleave intact Tg differently from those present in (APC in lymphoid organs, allowing generation of non-dominant epitopes. Second, during thyroxine secretion by thyrocytes, Tg may reach the blood in a partially digested form in which ‘cryptic’ fragments may become dominant, following its uptake by APC.²⁵ While these processes remain unclear, they necessitate the presence of tolerance mechanisms (anergy, suppression or ignorance) to prevent activation of naïve T cells specific for Tg epitopes, such as p2340, in situ or in the periphery.

Our data also raise the possibility that the hTg-derived p2340 may be presented by HLA-DR molecules which are homologous to the mouse I-E antigens. In this regard, it is quite interesting that HLA-DR4 has been weakly associated with Hashimoto's thyroiditis in some studies^26,27 and p2340 carries the DR4Dw4-binding motifs VAALTWVQT (aa 2341–2349) and VQTHIRGFG (aa 2347–2355).²⁸ Presentation of p2340 might ensue only under conditions that promote generation of cryptic determinants in Tg, e.g. after processing of Tg–antibody immune complexes²⁹ or highly iodinated Tg³⁰ in APC. In such cases, p2340 might constitute a target epitope during the spreading of the autoreactive T-cell response in the later stages of disease.³¹ The p2340-specific response could be further amplified by microbial peptides via molecular mimicry, as has been reported with other non-dominant Tg epitopes in EAT.³² Lastly, a role for p2340-specific IgG in pathogenesis, at least in EAT, remains doubtful in view of the fact that such antibodies do not bind to intact Tg. This is in apparent contrast with the capacity of Tg-reactive IgG from Graves' patients to bind to p2340.¹⁰ These observations suggest that p2340 is present on the Tg surface and accessible to antibodies but in free form it may not adopt an appropriate conformation to elicit Tg-binding antibodies.

Note added in proof: Five new thyroglobulin peptides with pathogenic potential in EAT were recently described by Verginis et al.³³

Abbreviations

EAT

experimental autoimmune thyroiditis

Tg

thyroglobulin

hTg

human thyroglobulin

LNC

lymph node cell(s)

mAb

monoclonal antibody

APC

antigen-presenting cell(s)

CFA

complete Freund's adjuvant

IFA

incomplete Freund's adjuvant

s.c.

subcutaneous

II

infiltration index