The Role of Glutamic or Aspartic Acid in Position Four of the Epitope Binding Motif and Thyrotropin Receptor-Extracellular Domain Epitope Selection in Graves' Disease

Abstract

Context: Development of Graves' disease (GD) is related to HLA-DRB1*0301 (DR3),and more specifically to arginine at position 74 of the DRB1 molecule. The extracellular domain (ECD) of human TSH receptor (hTSH-R) contains the target antigen.

Objective and Design: We analyzed the relation between hTSH-R-ECD peptides and DR molecules to determine whether aspartic acid (D) or glutamic acid (E) at position four in the binding motif influenced selection of functional epitopes.

Results: Peptide epitopes from TSH-R-ECD with D or E in position four (D/E+) had higher affinity for binding to DR3 than peptides without D/E (D/E−) (IC₅₀ 29.3 vs. 61.4, P = 0.0024). HLA-DR7, negatively correlated with GD, and DRB1*0302 (HLA-DR18), not associated with GD, had different profiles of epitope binding. Toxic GD patients who are DR3+ had higher responses to D/E+ peptides than D/E− peptides (stimulation index 1.42 vs. 1.22, P = 0.028). All DR3+ GD patients (toxic + euthyroid) had higher responses, with borderline significance (Sl; 1.32 vs. 1.18, P = 0.051). Splenocytes of DR3 transgenic mice immunized to TSH-R-ECD responded to D/E+ peptides more than D/E− peptides (stimulation index 1.95 vs. 1.69, P = 0.036). Seven of nine hTSH-R-ECD peptide epitopes reported to be reactive with GD patients' peripheral blood mononuclear cells contain binding motifs with D/E at position four.

Conclusions: TSH-R-ECD epitopes with D/E in position four of the binding motif bind more strongly to DRB1*0301 than epitopes that are D/E− and are more stimulatory to GD patients' peripheral blood mononuclear cells and to splenocytes from mice immunized to hTSH-R. These epitopes appear important in immunogenicity to TSH-R due to their favored binding to HLA-DR3, thus increasing presentation to T cells.

Arginine at position 74 in HLA-DR molecules is linked to Graves' disease and causes TSH-R-ECD epitopes with glutamic or aspartic acid in binding motif position four to bind strongly to DRB1*0301, and stimulate Graves' disease patients' PBMCs.

Graves' disease is characterized by the presence of anti-TSH receptor (TSH-R) antibodies (1) that induce hyperfunction of thyroid cells. Specific genes, including HLA-DR, CTLA-4, TSH-R, and others contribute to the development of the disease (2).

Inheritance of HLA-DRB1*0301 (HLA-DR3) constitutes a significant risk factor (3,4). We postulated that this is because particular epitopes in TSH-R-extracellular domain (ECD) may bind effectively to HLA-DR3 and thus be presented to T cells, helping induce autoimmunity (1). In contrast, HLA-DRB1*0701 (HLA-DR7) was found to be a protective allele for Graves' disease (GD) (4). We also found that HLA-DRB1*0302 (HLA-DR18) did not increase the risk of GD among African-Americans (5), although this allele has a large structural similarity with HLA-DR3.

For presentation to T cells and induction of thyroid autoimmunity, epitopes derived from TSH-R-ECD must first bind to HLA-DR. The affinity of a given peptide epitope for a given human leukocyte antigen (HLA) molecule is related to the peptide's amino acid sequence and also its three-dimensional conformation. The nine-member central amino acid motif of the epitope must fit into the binding groove formed by the HLA molecule. To create a high-affinity bond, the anchoring side chains of the peptide (generally supposed to include binding positions 1, 4, 6, 7, and 9) must match, in terms of size, charge, and orientation, the pockets present in the floor of the binding groove. Variable-length tails, preceding and trailing the central binding motif, also make contact with the HLA molecule helping to stabilize the interaction. Bound peptides are transported to the surface membrane of antigen-presenting cells, such as dendritic cells (6,7,8,9), in which they may be recognized by passing T cells. The side chains present in binding positions 2, 3, 5, and 8 are assumed to be outward facing and available to the T cell receptor (TCR).

Recent studies suggest that the presence of arginine at position 74 in the β-chain of the HLA-DR molecule, with a strong positive charge, is more closely related to development of GD than is the HLA-DR3 molecule per se (10,11). Amino acids at positions 71 and 74 in the DR sequence are important in defining the shape and charge of pocket 4 on the surface of the DR molecule (12,13). Logically, epitopes with a negative charge in position four of the binding motif, due to aspartic acid (D) or glutamic acid (E) residues, might tend to bind well to HLA-DR3 and other alleles possessing the double-positive 71K/74R motif. Only a small number of known DRB1 alleles contain the double-positive 71K/74R motif. These include DRB1*0301 through DRB1*0308, DRB1*0422, and DRB1*1107. Amino acids that form four of five pocket profiles are virtually identical with the baseline allele DRB1*0301. These alleles will presumably present many of the peptides presented by DBR1*0301. Unfortunately, little to no information is available on the relationship between these DRB1*0301 homologs and GD. In this paper we restricted our analysis to the more common DRB1*0301, DRB1*0302 (DR 18), and DRB1*0701 alleles.

The amino acid sequences of alleles DRB1*0301 and DRB1*0302 are closely related differing in only four amino acids. A comparison of binding pockets suggests that only pocket 1 differs significantly. DRB1*0301, with a valine at position 86 in pocket 1, prefers epitopes with aliphatic amino acids (I, L, M, and V) in position one. DRB1*0302, with a glycine at position 86 in pocket 1 prefers aromatic amino acids (F, Y, and W). As mentioned above, only DRB1*0301 has been associated with GD. Possibly the combination of an aliphatic residue in position 1 and a negatively charged amino acid in position 4 as defined by the motif [L,I,M,V]XX[D,E]. is especially important.

We tested our hypothesis on the importance of D/E+ in position 4 using data developed in vitro (HLA binding assays), ex vivo in humans with GD (14), and in vivo in TSH-R immunized mice (15). We found significant association between D or E in position four of epitopes derived from TSH-R-ECD and several measures of binding to three HLA-DR molecules or T cell responses.

Materials and Methods

TSH-R-ECD peptides and competing peptides

Thirty-one 16-20 mer peptides covering the entire sequence of TSH-R-ECD with overlaps of five to six amino acids were synthesized as reported (16,17). In addition, 10 TSH-R-ECD peptides, listed as peptide 32-41, were newly synthesized, These peptides were predicted by EpiMatrix (18) to have high binding affinity for HLA-DR3 (peptides 32-36) or multiple DRs (peptides 37-41). Mycobacterium tuberculosis 65-kDa heat shock protein (HSP) peptide 3-13 (restricted by HLA-DR3) and influenza hemagglutinin (HA) peptide 307-319 (restricted by HLA-DR7 and DR18) were used as competitor peptides in DR binding assays (6,7,8).

TSH-R-ECD peptide binding assay to HLA-DR molecules

Binding assays quantifying each peptide's affinity for HLA-DR3, DR7, and DR18 were done as previously described (17). Homozygous Epstein-Barr virus-transformed B-lymphoblastoid cell lines QBL (HLA-DRB1*0301, DR3), PLH (HLA-DRB1*0701, DR7), and RSH (HLA-DRB1*0302, DR18) were cultured. HLA-DR molecules were purified from 10⁸ B-lymphoblastoid cell lines expressing each homozygous HLA-DR, using cyanogen bromide-activated-Sepharose 4B columns (Pharmacia, Uppsala, Sweden) conjugated with L243 (mouse anti HLA-DR monoclonal antibody).

Subsequently, serial dilutions of nonbiotinylated TSH-R-ECD peptides and HSP3-13 (or HA307-319) were incubated in a 96-well plate with a purified HLA-DR molecule, followed by the addition of biotinylated HSP3-13 or HA307-317 peptide. The plate was incubated at 37 C overnight. Each mixture was transferred to another plate precoated with L243 and incubated overnight at 4 C. After washing, europium-labeled streptavidin (Wallac, Gaithersburg, MD) was added, followed by enhancement buffer (Wallac) at room temperature. Fluorescence was measured with a Delfia 1232 fluorometer (Wallac). Each assay was done in triplicate, and at least three complete studies were done of binding of all peptides to each HLA-DR molecule.

In addition to the reported assays (17), TSH-R-ECD epitope binding to HLA-DR3, DR7 (peptides 32-41) and DR18 (peptides 1-36) were determined (Table 1). The affinities of TSH-R-ECD peptides for HLA-DR are presented as IC₅₀s and categorized into three groups: peptides with high binding affinities (IC₅₀ < 10), intermediate binding affinities (10 < IC₅₀ < 50), and low binding affinities (IC₅₀ > 50). EpiMatrix peptide binding predictions are presented as Z-scores (17,18). Generally, peptides with Z score greater than 1.64, approximately the top 5% of any given sample, are expected to have high binding affinity for the specific HLA-DR.

Table 1.

HLA-DR binding affinities of TSHR-ECD-derived peptides (IC₅₀ in micromoles) and binding predictions

No.	Position	Amino acid sequence	DR3	DR7	DR18	DR3z	P3	DR7z	P7	DR18z	P18
1	20–35	GGMGCSSPPCECHQEE	>100	65	>100	1.0	22–30	0	22–30	0.5	22–30
2	30–49	ECHQEEDFRVTCKDIQRIPS	>100	80	>100	1.3	33–41	1.6	36–44	1.3	37–45
3	44–62	IQRIPSLPPSTQTLKLIET	>100	2.1	46.5	0.8	44–52	1.5	49–57	0.6	45–53
4	57–75	LKLIETHLRTIPSHAFSNL	>100	0.3	42.5	1.7	57–65	2.2	64–72	1.4	57–65
5	70–88	HAFSNLPNISRIYVSIDVT	29	1.4	36.3	1.0	70–78	1.6	79–87	1.0	72–80
6	83–102	VSIDVTLQQLESHSFYNLSK	41.3	51	61.3	2.1	83–91	2.1	89–97	1.6	83–91
7	97–112	FYNLSKVTHIEIRNTR	42	0.9	42	2.0	104–112	2.5	98–106	2.2	104–112
8	109–124	RNTRNLTYIDPDALKE	5.5	43.3	>100	0.9	115–123	1.5	114–122	1.1	115–123
9	119–137	PDALKELPLLKFLGIFNTG	>100	18.3	>100	0.8	124–132	1.9	128–136	1.0	124–132
10	132–150	GIFNTGLKMFPDLTKVYST	0.64	13	43	2.3	140–148	0.9	133–141	1.8	140–148
11	145–163	TKVYSTDIFFILEITDNPY	19.3	10	15	1.3	154–162	1.9	147–155	1.6	148–156
12	158–176	ITDNPYMTSIPVNAFQGLC	50	16.7	28.3	1.2	163–171	1.9	163–171	1.8	163–171
13	172–186	FQGLCNETLTLKLYN	18.3	4.3	12.5	1.2	172–180	2.5	172–180	1.3	172–180
14	183–198	KLYNNGFTSVQGYAFN	>100	0.8	55.5	1.1	185–193	1.5	189–197	1.7	185–193
15	195–210	YAFNGTKLDAVYLNKN	46.6	55	79.8	1.6	196–204	1.8	199–207	1.4	196–204
16	207–222	LNKNKYLTVIDKDAFG	40	40	73	1.6	214–222	1.4	207–215	1.9	214–222
17	217–232	DKDAFGGVYSGPSLLD	90	28.3	>100	0.9	224–232	1.9	223–231	0.9	221–229
18	227–242	GPSLLDVSQTSVTALP	45	1.2	41.3	1.4	229–237	1.6	233–241	1.5	229–237
19	237–252	SVTALPSKGLEHLKEL	>100	20	>100	0.7	237–245	2.6	238–246	0.8	237–245
20	248–263	HLKELIARNTWTLKKL	40	31.6	50.5	1.0	250–258	2.3	252–260	1.2	250–258
21	258–277	WTLKKLPLSLSFLHLTRADL	16.7	4	38.8	2.0	269–277	1.5	261–269	2.2	269–277
22	272–291	LTRADLSYPSHCCAFKNQKK	14.2	30	71.3	1.8	273–281	1.2	278–286	2.1	273–281
23	286–305	FKNQKKIRGILESLMCNESS	60	46.7	21	2.6	286–294	1	292–300	2.8	286–294
24	301–320	CNESSMQSLRQRKSVNALNS	46.7	3	52.8	1.2	309–317	1.9	309–317	1.0	305–313
25	315–334	VNALNSPLHQEYEENLGDSI	33.8	>100	56.8	1.4	318–326	0.9	322–330	1.2	318–326
26	329–348	NLGDSIVGYKEKSKFQDTHN	80	70	82.5	1.6	329–337	0.9	336–344	1.8	329–337
27	343–362	FQDTHNNAHYYVFFEEQEDE	20	36.6	53	1.6	354–362	1	348–356	1.4	353–361
28	357–376	EEQEDEIIGFGQELKNPQEE	45	38.3	60	1.8	366–374	0.8	362–370	1.9	366–374
29	371–390	KNPQEETLQAFDSHYDYTIC	18.3	9.3	73.3	2.5	379–387	1.2	378–386	2.4	379–387
30	385–404	YDYTICGDSEDMVCTPKSDE	13	>100	>100	1.3	389–397	1.4	385–393	1.0	389–397
31	399–418	TPKSDEFNPCEDIMGYKFLR	>100	80	51.8	1.7	410–418	1.2	403–411	1.8	410–418
32 (like 8)	105–118	HIEIRNTRNLTYID	10	0.45	10	1.7	108–116	2.2	106–114	1.6	107–115
33 (like 10)	137–150	GLKMFPDLTKVYST	0.2	10.5	25	2.3	140–148	0.8	142–150	1.8	140–148
34 (21–22)	267–282	LSFLHLTRADLSYPSHC	27.6	0.6	>100	2.0	269–277	1.5	269–277	2.2	269–277
35 (26–27)	339–352	EKSKFQDTHNNAHY	37.5	8.5	>100	1.7	342–350	1	343–351	2.0	342–350
36 (like 29)	376–389	ETLQAFDSHYDYTI	25	15.7	>100	2.5	379–387	1.2	378–386	2.4	379–387
37 (5–6)	78–94	ISRIYVSIDVTLQQLES	0.3	42	ND	2.1	83–91	2.0	82–90	1.6	83–91
38 (7–8)	103–119	VTHIEIRNTRNLTYIDP	17.5	4.8	ND	2	104–112	2.2	106–114	2.2	104–112
39 (15–16)	201–217	KLDAVYLNKNKYLTVID	>100	>100	ND	2.2	204–212	2.0	205–213	2.0	204–212
40 (21–22)	266–284	SLSFLHLTRADLSYPSHCC	5	4	ND	2.0	269–277	1.5	269–277	2.2	269–277
41 (22–23)	283–297	CCAFKNQKKIRGILE	>100	>100	ND	2.6	286–294	0.7	288–296	2.8	286–294
ex1	HSP3–13	KITAYDEEARR	0.3	N/D	N/D	2.7	5–13
ex2	HA307–319	PKYVKQNTLKLAT	N/A	0.36	0.6			3.3	309–317	2.3	309–317

Column 1 is arbitrary sequence number; position is sequence position in TSH-R; amino acid sequence is one-letter amino acid designation of sequence, with the predicted amino acids present in the nine amino acid motif for DR3 underlined; DR3, DR7, and DR18 are IC₅₀s for each peptide binding to relevant DR protein. DR3Z + P3 is predicted binding affinity of peptides to DR3 and position (P3) of binding motif; DR7Z + P7 and DR18z + P18 are similar data for DR7 and DR18. EpiMatrix Z-scores 1.6 or greater are shown in bold. IC₅₀s for HLA-DR less than 50 are shown in bold. ex1 and ex2 are the two indicator peptides used in the binding assay. ND and N/D, Mean not determined. 

In vitro stimulation of lymphocytes of patients with GD

GD patients (n = 46) and controls (n = 23) were investigated under an institutional review board protocol and with informed consent of subjects. Thyroid status was diagnosed from clinical manifestations and laboratory examinations as reported (14). Peripheral blood mononuclear cells (PBMCs) were obtained from patients with GD and controls and separated using Ficoll-Hypaque (14). Then 1 × 10⁵ cells/well were incubated in the presence or absence of TSH-R-ECD peptides for 2 d at 37 C, 5% CO₂ in RPMI 1640, 5% human AB serum, and penicillin/streptomycin. Then 1 μCi of ³H was added to each well, and wells were incubated for 24 h. Incorporated ³H was determined in a liquid scintillation counter. Cells were stimulated using 1-31 TSH-R-ECD peptides. Results were shown as stimulation index (SI), calculated as (average counts per minute in sample of the subject − background)/(average counts per minute in sample of unstimulated cells − background).

DNA was obtained from blood in each subjects and typed for HLA-DR B1 at the University of Maryland (14).

Immunization of HLA transgenic mice with pcDNA3-TSH-R-ECD and AdCMVTSH-R-ECD (15)

On d 1, 7 μg cardiotoxin were injected to each mouse in one hind leg. On d 6 and 20, 100 μg of plasmid encoding hTSH-R-ECD 19-417 (pcDNA3-TSH-R-ECD) plasmid were injected. On d 34, 1 × 10¹¹ particles of adenovirus encoding hTSH-R-ECD 19-417 (AdCMVTSHR-ECD) were given to the mice. On d 48, the mice were killed. The 10⁵ splenocytes obtained from mice were assayed in the same way as with human PBMCs. All studies on HLA-DR3 transgenic mice were performed under a protocol approved by the Institutional Animal Care and Use Committee.

Statistical analyses

Pearson product moment correlation coefficients (r values) were calculated for IC₅₀ values vs. Z-scores and IC₅₀ values vs. IC₅₀ values in each HLA-DR. A t test was used for comparison between two groups. P < 0.05 was considered statistically significant.

Results

Binding affinity of TSH-R-ECD peptide epitopes for three HLA-DR molecules

The results of in vitro TSH-R-ECD peptide binding to HLA-DR3, DR7, and DR18 are shown in Table 1. Most of TSH-R-ECD peptides bound with high or intermediate affinities to HLA-DR3. Peptide 33 had highest affinity (IC₅₀ 0.2 μm) for HLA-DR3, and peptide 37 had second highest affinity (0.3 μm). Peptide 33 and peptide 10, which is also a high-affinity ligand for HLA-DR3, share a central 9-mer core. The superior binding affinity of the shorter peptide 33 demonstrates the value of using bioinformatics tools to identify the core 9-mer motif from within known HLA ligands. Peptide 37 is another case in point. The core 9-mer present in this peptide was present but truncated in peptide 5 and present at the extreme amino terminus in peptide 6. In both cases observed binding affinity was reduced relative to the optimized peptide 37. This ability to predict core 9-mers is central to the analysis presented in this report.

Alleles HLA-DR3 (HLA-DRB1*0301) and DR18 (HLA-DRB1*0302) are highly homologous. The amino acid sequences of HLA-DR3 and DR18 differ by only four amino acids (Table 2). Amino acids 28 and 47 are involved in forming pocket 7. Amino acid 86 defines the binding preferences for pocket (P) 1, an important anchoring position. The changes related to P7 are conservative, suggesting that the changes in P1 cause the differential binding of alleles HLA-DR3 and DR18. Allele HLA-DR3 contains a valine in position 86, whereas allele DR18 contains a glycine. Alleles containing valine at position 86 prefer aliphatic amino acids such as leucine, isoleucine, methionine, and valine in position 1, whereas alleles containing glycine at position 86 prefer aromatic amino acids such as phenylalanine, tryptophan, and tyrosine in the first position. As expected, binding results for HLA-DR3 and DR18 were positively correlated, although more peptides bound with higher affinity to HLA-DR3. For several of the peptides that bound with higher affinity to HLA-DR18, EpiMatrix predicted an aromatic epitope position one anchor (see peptides 12 and 23 as examples).

Table 2.

The amino acids differing between HLA-DR3 and DR18 and their importance in pocket structure are shown

Amino acid	HLA DR3	HLA DR18	Contributes to pocket
26	Y	F	N/A
28	D	E	7
47	F	Y	7
86	V	G	1

N/A, Not available. 

Many peptides exhibited high affinities for HLA-DR7, as previously reported (17). As expected, the predicted core 9-mer for allele HLA-DR7 rarely agreed with the predicted core 9-mer for allele HLA-DR3 (six cases in 41 total). None of the core 9-mers shared by allele DRB1*0301 and DRB1*0701 contain either D or E in position four. Five of six of these peptides bound with higher affinity to HLA-DRB1*0701. The sixth peptide bound with high affinity to both HLA-DRB1*0301 and HLA-DRB1*0701. The two best binding peptides, peptides 33 and 37, bound with higher affinity to HLA-DRB1*0301 than to HLA-DRB1*0701 (0.2 vs. 10.5 μm and 0.03 vs. 42 μm, respectively).

Pearson correlation (Table 3) was used to calculate the correlations, r value for in vitro TSH-R-ECD peptide binding affinities to different HLA-DRs. The r value between HLA-DR3 and HLA-DR7 was 0.076, P = 0.34, not significant. The r value between HLA-DR3 and DR18 was 0.32, P = 0.04, weakly significant, highlighting the importance of the shared binding pockets P4, P6, P7, and P9. Interestingly, the r value between HLA-DR7 and DR18 was 0.48, P = 0.003, and significant, highlighting the importance of the P1 pocket structure shared between alleles DR18 and DR7.

Table 3.

Pearson correlations (r value and P value) among HLA-DRs

	DR3	DR7	DR18
DR3	—	r = 0.076, P = 0.34	r = 0.32, P = 0.040
DR7	—	—	r = 0.48, P = 0.003
DR18	—	—	—

Data from peptides 1-31 (covering the TSH-R-ECD) were used to calculate r value for correlation of IC₅₀ of each peptide to each DR protein. Dash indicates redundant. 

Correlation of presence of aspartic or glutamic acid (D/E) in position four of predicted TSH-R-ECD binding motif and IC₅₀ of binding to HLA-DR3, DR7, or DR18

Peptides with D/E in position four of the binding motif (D/E+; 18 of 41 tested) bound with lower average IC₅₀ to HLA-DR3 than did 23 peptides without D/E (D/E−) (IC₅₀ 29.3 vs. 61.4, P = 0.0024). In other words, TSH-R-ECD peptides with D/E+ in position four bound on average more tightly to DR3. Only two of 41 peptides were found to have D/E in position four of the HLA-DR7 binding motif. Fifteen D/E+ peptides were predicted to have D/E in the HLA-DR18 motif. There was no significant correlation between IC₅₀s of D/E+ peptides and binding to HLA-DR7 (IC₅₀ 85.0 vs. 28.7, P = 0.14) or HLA-DR18 (IC₅₀ 62.7 vs. 61.1, P = 0.87).

Correlation between D/E in epitope binding motif position four and stimulation of GD patients' T cells

We compared data for stimulation of T cells from patients with GD and control subjects based on whether the stimulating epitope contained D or E in position four. Toxic GD patients who are DR3+, and all GD who are DR3+, had higher responses to D/E+ peptides than to D/E− peptides (SI 1.42 vs. 1.22, P = 0.028, and SI 1.34 vs. 1.20, P = 0.051, respectively) (Table 4). In contrast, toxic GD patients who are DR3− and all GD patients who are DR3− did not have significant differences in T cell responses between D/E+ and D/E− peptides (SI 1.16 vs. 1.20, P = 0.379, and SI 1.15 vs. 1.19, P = 0.439, respectively). In controls, T cell responses to D/E+ or D/E− peptides were similar (SI 1.16 and 1.16, respectively).

Table 4.

Average lymphocyte stimulation (SI) for DE+ or DE− peptides (relevant to DR binding motif), in different groups of GD patients and controls subjects

	Subjects		Age (yr)	Number		Average SI	t test
1	Control		51 ± 15.6	23	D/E+	1.16	0.908
					D/E−	1.16
2	Toxic GD	DR3+	43 ± 10.3	11	D/E+	1.42	0.028
					D/E−	1.22
3	Toxic GD	DR3−	40 ± 11.9	24	D/E+	1.16	0.379
					D/E−	1.20
4	Toxic GD	DR3+ and DR3−	41 ± 11.3	35	D/E+	1.24	0.518
					D/E−	1.21
5	All GD	DR3+	44 ± 11.6	14	D/E+	1.34	0.051
					D/E−	1.20
6	All GD	DR3−	42 ± 12.6	30	D/E+	1.15	0.439
					D/E−	1.19
7	All GD	DR3+ and DR3−	43 ± 12.2	44	D/E+	1.21	0.701
					D/E−	1.19

Data were obtained in assays using peptides 1-31. All DR3+ GD patients and toxic DR3+ GD patients (P < 0.028) have a strong response to DE+ epitopes. Toxic DR3+ GD patients had a higher significance than all DR3+ GD patients. Fourteen D/E+ and 17 D/E− TSH-R-ECD peptides were evaluated. 

Comparison between D/E in epitope motif position four and stimulation of splenocytes from HLA-DR3 transgenic mice immunized to recombinant human TSH-R-ECD

We did a similar comparison of responses to D/E+ or D/E− epitopes using data from T cell stimulation assays in mice immunized to TSH-R by adenovirus and plasmid. D/E+ TSH-R-ECD peptides were more stimulatory (average SI 1.95 vs. 1.69, P = 0.036) in assays on splenocytes from DR3+ mice immune to recombinant human TSH-R (Table 5).

Table 5.

Response (average SI) of T cells from TSH-R-ECD immunized HLA-DR3 transgenic mice to TSH-R peptides 1-31 that are D/E+ or D/E− in position four of the DR3 binding motif

No.	DR3 IC50	D/E in P4	D/E+	D/E−
1	100			1.66
2	100	D	2.41
3	100			1.59
4	100			1.85
5	29			1.20
6	41.3	D	1.32
7	42	E	1.54
8	5.5	D	2.14
9	100			1.21
10	0.64	D	2.00
11	19.3	E	1.88
12	50			1.42
13	18.3			1.74
14	100			1.76
15	46.6			1.84
16	40	D	1.74
17	90			2.15
18	45	D	2.24
19	100			1.28
20	40			1.71
21	16.7			2.06
22	14.2	D	1.92
23	60			2.13
24	46.7			1.99
25	33.8			1.49
26	80	D	1.56
27	20	E	1.86
28	45	E	1.83
29	18.3	D	2.46
30	13	D	2.38
31	100			1.59
		Avg. SI	1.95	1.69
		t test	0.036

Discussion

T cell epitopes of variable length are displayed to T cells by binding in the epitope binding groove in the HLA-DR protein on the antigen-presenting cell surface. The amino acids in the central epitope binding motif fit into a series of corresponding pockets in the surface of the major histocompatibility complex molecule (9,11,12,13,19). The pockets are formed by the structure of the DRα (largely invariant) and the highly variable first 90 amino acids of the DRβ chain. Epitope amino acid binding in P1, P4, P6, P7, and P9 interact with DR, and amino acids binding in P2, P3, P5, and P8 tend to make contact with TCR. P1 is known to play an important role in anchoring peptide epitopes to the floor of the HLA binding groove. The composition of amino acids determining P1 and P7 tend to be to be conserved, and composition of P6 is variable. P4 is narrow and significant in determining binding of an epitope (9). P4 probably plays an important role in determining epitope presentation and is in part responsible for the differences of affinities of epitopes for HLA-DR3 and DR7 (Table 1) (18,19). 71K (lysine) and 74R (arginine) in exon 3 of HLA-DR β1-chain in HLA-DR3 and DR18 were predicted to define selection of D/E for fitting into P4 (9,11,12,13,19).

HLA-DR3 provides a genetic risk for GD, and arginine at position 74 in the DR β-chain may be even more closely related to GD (10,11). Because arginine 74 is important in the structure of P4 of the epitope binding grove, we reasoned that TSH-R-ECD epitopes having a negative charge at position four in the binding motif might have higher affinity binding and induce strong T cell responses. We tested in several ways the hypothesis that D or E in binding motif position four could relate to more effective presentation of antigen to T cells in animals and man immune to TSH-R. We note that the affinity of binding of an epitope to DR molecules can be readily measured and that strong binding should correlate with good presentation of epitope to the TCR. Binding of epitope to DR is essential for function but does in itself prove immune function. Selection of epitopes in vivo depends on expression of the protein and its processing within antigen-presenting cells. Furthermore, binding of epitope to DR is one step away from binding of the epitope/DR complex to the TCR. The affinity of this latter step is presumably crucial in determining the T cell response but is difficult to measure except by using DR-tetramers.

We compared epitope peptides derived from TSH-R-ECD that have D/E predicted for binding motif position four, or do not have D/E, and the IC₅₀ of binding to HLA-DR3 (Table 1). We found that TSH-R-ECD peptides with D/E in position four had higher affinity (P = 0.0024). Peptides binding strongly to HLA-DR have an increased chance to be presented to immunogenic T cells. The same comparison for D/E+ epitope binding to HLA-DR7 or DR18 was not significant, presumably because of the different structure of epitopes favored by DR7 and DR18.

We next evaluated the ability of position four D/E+ epitopes to stimulate lymphocytes from patients with GD and controls. Toxic GD patients (Table 4) responded with significantly higher stimulation indexes to TSH-R-ECD peptides with D/E in position four than to non-D/E peptides. All DR3+ GD patients showed similar response as toxic DR3+ GD patients. This finding suggests that TSH-R-ECD peptides with D/E in position four can be functional T cell epitopes in DR3+ patients. Toxic DR3+ GD patients had higher correlation with D/E than did all DR3+ GD patients, including those previously treated and now euthyroid. This probably related to the generally higher SI responses among toxic GD patients, compared with euthyroid GD patients. Except for one patient, all DR3+ patients in the study were heterozygous for the DR3 allele (14). Possibly the correlation might be higher in DR3 homozygous subjects.

We also evaluated the responses of spleen cells from DR3+ transgenic mice immunized by injection of adenovirus and plasmid expressing human TSH-R-ECD. TSH-R-ECD peptides predicted to have D/E in position four produced higher SIs than D/E− epitopes. Although those mice demonstrated evidence for only low level thyroid stimulation, T cells were strongly activated. Peptide 37 (ISRIYVSIDVTLQQLES) strongly stimulated splenocytes of mice immunized with TSH-R-ECD protein (15). Both peptides 33 (GLKMFPDLTKVYST) and 37 (ISRIYVSIDVTLQQLES) have D/E in position four. Thus, TSH-R-ECD peptides with D/E in binding motif position four appear to be effective T cell epitopes.

Peptide 33 had the strongest affinity for HLA-DR3 among all TSH-R-ECD peptides, and peptide 37 had second strongest affinity to HLA-DR3, but these peptides did not have high affinity for HLA-DR7.

Previously we reported seven TSH-R-ECD peptide epitopes that induced strong immunogenic responses (peptides 10, 11, 12, 16, 20, 22, and 27) in PBMCs of patients with GD (17). Interestingly, five of these TSH-R-ECD peptides (peptide 10, 11, 16, 22, and 27) had D/E in position four, and the binding motif for peptide 12 and 20 may also possess D/E in position four in their sequence because an alternative binding motif is possible in these peptides. None of those immunogenic peptides had high binding affinities to HLA-DR7 (Table 1).

Together our data indicate that peptide epitopes derived from human TSH-R-ECD, having D or E at position 4 of the epitope binding motif and especially if an aliphatic residue is present at position 1, are important in the production of thyroid autoimmunity in DR3+ patients and animals. This is dependent on the presence of specific amino acids in the DRB1 chain that form P4 and P1. Within the structure of TSH-R-ECD, there are several epitopes with this composition that may be involved in pathogenesis of GD, including peptide 37 (ISRIYVSIDVTLQQLES), which may be especially important. These epitopes appear functional in immunogenicity against TSH-R, at least in patients who express the HLA DR3 molecule, due to their favored binding, thus increasing their chance for presentation to T cells.