Endocrine epithelial cells, targets of the autoimmune response in thyroid and other organ-specific autoimmune diseases, express HLA class II (HLA-II) molecules that are presumably involved in the maintenance and regulation of the in situ autoimmune response. HLA-II molecules thus expressed by thyroid cells have the “compact” conformation and are therefore expected to stably bind autologous peptides. Using a new approach to study in situ T cell responses without the characterization of self-reactive T cells and their specificity, we have identified natural HLA-DR-associated peptides in autoimmune organs that will allow finding peptide-specific T cells in situ. This study reports a first analysis of HLA-DR natural ligands from ex vivo Graves’ disease-affected thyroid tissue. Using mass spectrometry, we identified 162 autologous peptides from HLA-DR-expressing cells, including thyroid follicular cells, with some corresponding to predominant molecules of the thyroid colloid. Most interestingly, eight of the peptides were derived from a major autoantigen, thyroglobulin. In vitro binding identified HLA-DR3 as the allele to which one of these peptides likely associates in vivo. Computer modeling and bioinformatics analysis suggested other HLA-DR alleles for binding of other thyroglobulin peptides. Our data demonstrate that although the HLA-DR-associated peptide pool in autoimmune tissue mostly belongs to abundant ubiquitous proteins, peptides from autoantigens are also associated to HLA-DR in vivo and therefore may well be involved in the maintenance and the regulation of the autoimmune response.

Class II MHC3 molecules are transmembrane glycoproteins that bind mostly exogenous peptides to be presented to CD4+ T lymphocytes. Constitutive expression of HLA-II is restricted to professional APCs such as dendritic cells, macrophages, and B cells. HLA-II expression can be induced in other cell types in inflammatory conditions as shown in vivo in epithelial cells from various target organs in autoimmune diseases such as autoimmune thyroiditis (1, 2, 3, 4).

Human autoimmune thyroid diseases (AITD), such as Hashimoto’s thyroiditis and Graves’ disease (GD), are characterized by T and B cell infiltration of the thyroid gland and an intense immune response to well-characterized thyroid autoantigens such as thyroglobulin (Tg), thyroid peroxidase, and the thyrotropin receptor. Thyroid follicular cells (TFC) are polarized epithelial cells that have a very important endocytic activity in their apical end, from where they endocytose thyroglobulin-rich colloid that is processed to generate thyroid hormones that are released by exocytosis to the capillaries in the basal pole. The expression of HLA-II by TFC in these diseases is well documented (5). In situ dendritic cells, macrophages, and B cells infiltrating the organ also express HLA-II and can process and present self Ags. The importance of nonprofessional APCs such as TFCs in this context is not known, but it is clear that the autoimmune response is modulated by the interaction between cells of the immune system and cells of the target organ, with a strong influence of the local microenvironment, which may determine the formation of tertiary lymphoid tissue (6).

TFCs in Graves’ disease overexpress class I molecules and express ectopic HLA-II, invariant chain, and HLA-DM. The HLA-DM expression level by TFCs is low but sufficient to allow peptide loading into the HLA-II molecules and express stable HLA-II-peptide complexes at the cell surface (7). Thus, HLA-II-expressing epithelial cells in autoimmune diseases could present autoantigens to infiltrating CD4+ cells and play a role in the maintenance and/or the regulation of the autoimmune response. Consequently, the antigenic repertoire displayed at the surface of HLA-II-expressing epithelial cells must be relevant to understand self-reactive T cell responses in situ. However, other cells in autoimmune thyroid glands also express HLA-II, namely resident and infiltrating dendritic cells, endothelial cells, activated T cells, B cells, and macrophages (5).

Studies characterizing HLA-DR-associated peptides were mostly done using lymphoblastoid cell lines as the source of DR-peptide complexes. They led to the identification of the allele-specific structural motifs common to peptides that associate to most DR and some DQ alleles (8, 9, 10). Studying natural DR ligands in other cell types is more difficult, mostly due to low HLA-II expression and limited access to cells. In previous work, we analyzed the peptide repertoire associated to HLA-DR4 molecules expressed by a transfected insulinoma cell line. We detected a heterogeneous pool of self peptides derived from cell surface proteins and proteins from internal cell compartments, including the cytosol and the secretory pathway, some of which corresponded to tissue-specific proteins (11). Although these data were very informative and showed a different pattern from that of DR4-associated peptides from lymphoblastoid cells, this material gave only a very indirect assessment of what the repertoire of peptide ligands occupying the cavity of HLA-II molecules in a real-life autoimmune tissue would be.

Therefore, studies from ex vivo tissue samples are still necessary to gain insight into the real repertoire of DR-associated natural peptides from autoimmune tissue. Little has been done with human tissue, where sample heterozygosity adds to the other difficulties. In the first published work on HLA-DR-peptide purification from human tissue in 1995, 17 peptides were identified from 50 g of the spleen of a patient with rheumatoid arthritis (12). Later, 55 peptides were sequenced from 9 human intestine samples from patients with inflammatory bowel disease (13). Recently, a large number of HLA class II-associated peptides have been identified from primary renal carcinoma samples (14).

In the present report, we describe the characterization of natural peptides bound to HLA-II molecules from thyroid glands of GD patients, using mass spectroscopy (MS) techniques. No cell separation methods were applied and whole tissue from several glands was processed to analyze the HLA-DR-associated peptides. A total of 162 DR ligands from 7 different donors were identified. All peptides complied with the common features of HLA-DR natural ligands: they were variable in length, formed “nested” sets with a common core and varying flanking sequences, and were derived from proteins present in different compartments, including the extracellular milieu, the plasma membrane, vesicular compartments, and the cytosol. Most extracellular peptides were derived from complement factors, extracellular matrix proteins, or colloid components such as albumin. Most interestingly, we were able to identify eight different peptides corresponding to the well-defined thyroid autoantigen thyroglobulin, which is also one of the major components of the colloid. Although we cannot be certain that all of these peptides are associated with class II molecules expressed by the thyroid follicular cells and not by other class II-expressing cells, the common presence of colloid components together with the important endocytic activity at the apical end of the TFCs make us think that some of these peptides can indeed be associated to DR molecules expressed by TFCs.

Thyroid surgery samples from patients with Graves’ disease were used after the patients’ informed consent, using a protocol approved by the ethical committees of University Hospitals Germans Trias i Pujol and Vall d’Hebron. Donors were women, 26–41 years of age, and their HLA class II typing was obtained by exon 2 sequencing (Table I). HLA-II expression by TFC was assessed by immunofluorescence staining on cryostat sections and by flow cytometry, as described (5).

Table I.

Thyroid glands from GD patients used for the assays

SampleSexAge (years)HLA-II ExpressionaSample Processed (g)HLA-II
TB449 35 ++ 25.5 DRB1*0301,1501, DQB1*0201,0602 
TB471 28 ++ 15 DRB1*0301,081, DQB1*0201,0402 
TB448 26 ++ DRB1*0407,1501, DQB1*0301,0602 
TB 270 32 +++ 3.31 DRB1*0701, DQB1*0202,0303 
TB269 30 ++ DRB1*1101,1104, DQB1*0301 
TB 237 41 1.1 DRB1*1501, DQB1*0602 
TB190 35 ++ DRB1*1101,1104, DQB1*0301 
SampleSexAge (years)HLA-II ExpressionaSample Processed (g)HLA-II
TB449 35 ++ 25.5 DRB1*0301,1501, DQB1*0201,0602 
TB471 28 ++ 15 DRB1*0301,081, DQB1*0201,0402 
TB448 26 ++ DRB1*0407,1501, DQB1*0301,0602 
TB 270 32 +++ 3.31 DRB1*0701, DQB1*0202,0303 
TB269 30 ++ DRB1*1101,1104, DQB1*0301 
TB 237 41 1.1 DRB1*1501, DQB1*0602 
TB190 35 ++ DRB1*1101,1104, DQB1*0301 
a

HLA-II expression level in thyroid follicular cells as assessed by flow cytometry and immunofluorescence (see Materials and Methods). Expression scores are based on immunofluorescence staining on cryostat sections as follows: +, positive staining in <10% follicles (focal); ++, 10–50% positive follicles (multifocal); +++, >50% positive follicles (generalized) (14 ).

Anti-HLA-DRαβ mouse IgG2a mAb L243 (American Type Culture Collection) and anti-HLA-DRβ mouse IgG2b mAb B8.11.2 (kindly provided by Dr. F. Koning, Immunohaematology Department, University of Leiden, Leiden, The Netherlands) were used for affinity chromatography.

A modification of the previously described method (15) was applied. Briefly, frozen thyroid samples were mechanically homogenized with a tissue homogenizer in H buffer (50 mM Tris-HCl (pH 7.6), 150 mM NaCl, 5 mM EDTA, 1 mM PMSF, 5 μg/ml aprotinin, 5 μg/ml leupeptin, 5 μg/ml pepstatin, and 5 mM iodoacetamide) and resuspended in lysis buffer (H buffer plus 0.5% Nonidet P-40 (Surfact-Amps Nonidet P-40; Cultek)). HLA-II-peptide complexes from the lysate were purified by affinity chromatography using 1 ml Sepharose bead columns with 5 mg mAb L243. The unretained fraction of sample TB449 was repurified by eluting through a column with mAb B8.11.2, due to the low affinity of L243 for the DR3 molecule. TB471 peptides were purified using the B8.11.2 column directly.

HLA-II-peptide complexes were eluted with trifluoroacetic acid 0.1%, following standard procedures (11). Eluted class II-peptide mixtures were spun through 10-kDa cut-off ultrafiltration filters (Centriprep-10; Millipore) and collected from the flow-through. The low molecular mass fraction containing the peptide pools was separated by rpHPLC using an acetonitrile gradient column (Vydac protein and peptide C18, 2.1 mm × 25 cm) (15). Chromatographic analysis was monitored at 214-nm UV wavelength. One-minute fractions were collected from minute 20 to 100 and then analyzed by MS.

HPLC fractions were evaporated and redissolved in 5 μl MeOH/H2O 1/1, 1% AcOH. The molecular mass of each fraction’s components was measured in a MALDI-TOF mass spectrometer (Voyager-DE PRO, Applied Biosystems) in positive ion reflector mode. Peptide sequences were obtained by electrospray tandem MS using an LCQ ion trap mass spectrometer (Thermo Fisher Scientific) with a nanospray electrospray ionization (nESI) source (nESI-ion trap MS (ITMS)/MS from Protana A/S, Odense, Denmark), by a MALDI-TOF/TOF in a mass spectrometer (AB 4700 mass spectrometer, Applied Biosystems), and by vMALDI-MS/MS in an LTQ instrument (Thermo Fisher Scientific).

For nESI-ITMS/MS analysis, 2 μl of each HPLC fraction was loaded in the nanospray needle. Product ion spectra were obtained by collision-induced dissociation, mainly of the double- and triple-protonated peptide molecules. Precursor masses were selected from the MALDI-TOF data spectra and from the signals observed in the nanospray full MS scan. Isolation windows of 3- and 4-unit widths were used for the MS/MS and MS3 experiments, respectively, and a 25–35% relative collision energy, depending on the charge state of the precursor ion. For MALDI-TOF/TOF analysis, 0.4 μl of each sample was automatically analyzed: a full scan was done within the mass-to-charge (m/z) range 900-6000, and the four most abundant signals detected over the m/z range 1100–3500 Da were fragmented. For vMALDI-MS/MS analysis, 0.5 μl of each sample was shot automatically in the Crystal Positioning System mode. Instrument setup was optimized in a data-dependent scan mode as follows: Automatic Spectral Filter parameters were restrictive in the m/z 1200–2500 range, wide band activation was set up for the MS/MS experiments, dynamic exclusion was 1, and 5 MS/MS were acquired for each full MS.

Sequences from the MALDI-TOF/TOF fragmentation spectra were identified using the MASCOT software program (Matrix Science) without protein size or enzyme cleavage site restriction. Using the same parameters, peptide identification from the nESI spectra was done with the MS-Tag (ProteinProspector; Mass Spectrometry Facility, University of California–San Francisco) and Mascot software programs, by comparing the MS/MS spectrum of each peptide with the virtual spectra derived from protein sequences in public databases (Swiss-Prot, European Bioinformatics Institute, Heidelberg, Germany; and GenPept, National Center for Biotechnology Information). The Sequest software (Thermo Fisher Scientific) was used for the vMALDI-MS/MS spectra identification using the parameters described above.

Thyroglobulin and other peptides were synthesized at the Peptide Facility, Universitat Pompeu Fabra, Barcelona.

HLA-DR3 and HLA-DR15 binding experiments were performed by ProImmune, using their cell-free class II REVEAL binding assay. A total of 11 peptides (plus 2 overlapping peptides) were tested, including 7 thyroglobulin peptides and others that we could assign to one HLA-DR allele (see Fig. 3). The tests consisted of the detection by a conformational Ab of HLA-DR-peptide complex formation after mixing soluble DR molecules with each peptide. Positive control for DR15 was hCMV pp65 109–123 peptide MSIYVYALPLKMLNI and for DR3, M. tuberculosis Ab 85B 14–27, PSPSMGRDIKVQFQ. Negative control peptides did not show any binding (data not shown), but the reference analysis was done comparing the binding of each peptide to each allele with a known borderline affinity control peptide (score of 100), named pass/fail control binder, and to each positive control peptide. Each experiment was conducted twice at two different dilutions, generating four data points for assembly reaction (peptide/allele). Thus, values between 0 and 100 were considered weak binding and values >100 were considered positive binding.

Modeling the complex between peptide Tg726–743 and the HLA-DR51 or HLA-DR15 molecules was performed using a novel simulation protocol (X. Daura, unpublished). The essential aspects are summarized below.

Structures for DR51 and DR15 were obtained from the Protein Data Bank (PDB entries 1H15 and 1BX2). The 1h15.pdb and 1bx2.pdb files contain crystallographic coordinates for the complexes between DR51 and an EBV DNA polymerase (EBVP) peptide and between DR15 and a human myelin basic protein (HMBP) peptide. Initial system configurations included the following atoms: residues Glu28–Thr105 of DRA1*0101 and Pro34–Arg122 of DRB5*0101 and DRB1*1501, and the backbone atoms of residues Val630–His639 of EBVP and Val113–Thr122 of HMBP (residue numbers according to Swiss-Prot). The protocol developed generates the missing peptide atoms via a molecular dynamics slow-growth technique (16) under nonequilibrium conditions, thus transforming a generic 10-residue backbone (EBVP630–639 or HMBP113–122) into the Tg726–743 peptide within the binding grooves of DR51 and DR15. Additionally, the method allows the assignment of a relative binding score. The following steps were followed:

1) Analysis of the Tg726–743 sequence and the known binding motifs to DR alleles (17) suggested Leu732 as the most probable candidate for binding to P1 in both DR51 and DR15. This alignment was therefore fixed for subsequent modeling of DR/Tg726–743 interactions. The following steps were performed in parallel for the DR51 and DR15 alleles.

2) The positions of (missing) Tg726–743 backbone and side-chain atoms were assigned by successive addition of a small vector (10−4 nm in the x direction) to the coordinates of the closest existing backbone atom.

3) The atoms added in step 2 were defined as dummies in the biomolecular force field of the GROMOS96 simulation package (18). A dummy atom is one that has no interactions with the rest of the system. All other atoms (which we shall call crystallographic) were assigned parameters from the vacuum (45B3) force field (19), which includes shielding of charge-charge interactions in a very crude way to mimic a basic solvent effect. The protonation state of ionizable residues was chosen to mimic a neutral pH.

4) Following equilibration of the system, a nonequilibrium slow-growth molecular dynamics simulation (200 ps) in vacuum at a very low temperature (10 K) was performed. In the course of this simulation the dummy atoms were progressively transformed into regular atoms (i.e., atoms with standard force field parameters) using a soft-core potential to avoid atomic clashes, while the positions of the crystallographic atoms were restrained with a harmonic potential. During the process, the growing atoms (i.e., atoms that were initially dummies) accommodate to the environment under the sole influence of the force field. The simulation delivers the work involved in the transformation (16). To accumulate statistics on the nonequilibrium process, the simulation was repeated 1000 times with different initial velocity distributions.

5) The free-energy change associated to the transformation was estimated by using the following equation (20):

\[{\Delta}G\ {=}\ {-}\mathrm{k}_{\mathrm{B}}T\ \mathrm{ln}{[}\mathrm{exp}({-}W/\mathrm{k}_{\mathrm{B}}T){]}\]

where kB is Boltzmann’s constant, T is the absolute temperature, W is the work, and the brackets denote an average over the 1000 work values. This free-energy difference estimate was taken as a score for binding affinity. Because the transformation affects exactly the same atoms in the two complexes, the scores for DR51/Tg726–743 and DR15/Tg726–743 can be directly compared.

6) The 1000 model structures obtained for the complex were structurally clustered, using an atom-position root-mean-square difference of 0.1 nm as similarity cutoff for the buried side chains (Leu732, Glu735, Ala737, Phe738, and Arg740). The structural variability within the ensemble of 1000 model structures could be thus evaluated. The above equation can then be used to estimate the free energy change for each structural cluster or family (ΔGclus). Here, the brackets denote an average over the work values obtained for the specific cluster.

Note that the estimated free energies do not have a direct connection to the experimental binding free energies. The calculated free energies are associated, strictly, to the creation of residues Cys726–Cys730 and Thr741–Gln743 and the side chains of residues Gln731–Arg740 in the binding grooves of DR51 (ΔGDR51/Tg726–743) and DR15 (ΔGDR15/Tg726–743) at 10 K and in the absence of solvent effects. Additionally, 1000 simulations may not suffice to guarantee convergence of ΔG. Nevertheless, we make the plausible hypothesis that the sign of the difference ΔG51–15 = ΔGDR51/Tg726–743 − ΔGDR15/Tg726–743 is equal to the sign of the difference between the experimental binding free energies. That is, we assume that the free energy change involved in the creation of this set of atoms of Tg726–743 in the binding groove of DR51 and DR15, under the specified conditions, is correlated with the binding affinities of Tg726–743 for the two molecules.

A total of 150 peptide sequences were obtained from DR molecules isolated from three GD thyroid samples: TB448 (8 g, 19 sequences), TB449 (25.5 g, 61 sequences), and TB471 (15 g, 70 sequences). Of the 19 TB448 peptides, 11 were sequenced by nESI, 6 by MALDI-TOF/TOF, and 2 were obtained using both techniques. Similarly, 47 of 61 TB449 peptides were sequenced by nESI, 5 by MALDI-TOF/TOF, and 9 by both. Of the 70 peptides from TB471, 49 were sequenced by nESI, 7 by MALDI-TOF/TOF, and 10 by vMALDI. Some peptides were detected by all three methods. A summary of the sequences obtained and their characteristics is shown in Table II.

Table II.

Peptide sequences obtained from three thyroid samples from GD patients

Mr (Da)aSizeb (aa)PPPPPPPPPc123456789Allele Assignedd% Affinity IndexePromiscuityfSource ProteingId No.hLi
Sample: TB448 (DRB1*1501, DRB5*0101/DRB1*0407, DRB4*0103)         
 1770.0 15 DIPELVNMGQWKIRA DRB1*1501 41.84 27/51 Complement C3 (194–208) P01024 
 1594.0 15 AAGLLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α1 chain (XV) (1243–1257) P39059 EM 
 1522.8 14 AGLLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α1 chain (XV) (1244–1257) P39059 EM 
 1687.0 13 LKKYLYEIARRHP DRB1*1501 33.67 6/51 Serum albumin (159–171) P02768 Co 
 2014.2 16 YPKSLHMYANRLLDHR DRB1*1501 55.10 19/51 Apolipoprotein B (1200–1215) P04114 
 1629.8 15 DVGVYRAVTOQGRPD DRB5*0101 68.37 7/51 DQB1*0602 β-chain (75–89) P03992 
 1676.8 15 DVGEYRAVTELGRPD DRB5*0101 44.90 3/51 HLA II β-chain DRB1*4 (72–86) P13760 
 1633.0 15 GGELRFTVTQRSQPG DRB5*0101 29.59 39/51 Heparan sulfate proteoglycan (1006–1020) P98160 EM 
 1795.0 16 YGGELRFTVTQRSQPG DRB5*0101 29.59 39/51 Heparan sulfate proteoglycan (1005–1020) P98160 EM 
 1660.8 14 YDHNFVKAINAIQK DRB5*0101 54.08 19/51 Cathepsin C (170–183) P53634 Cv 
 1661.0 16 RLPIIDVAPLDVGAPD DRB5*0101 13.27 (LB) 7/51 Collagen α1 chain (I) C-term propeptide (1436–1451) P02452 EM 
 2030.0 18 CPTPCQLQAEQAFLRTVQ DRB5*0101 29.59 2/51 Thyroglobulin (726–743) P01266 Co 
 1678.9 12 REIFNMARQRSR DRB5*0101 46.94 30/51 Apolipoprotein B (427–439) P04114 
 1720.0 16 IRASYAQQPAESRVSG DRB1*0407 ND 0/51 Heparan sulfate proteoglycan (1089–1104) P98160 EM 
 1827.0 16 NIFSFYLSRDPDAQPG DRB1*0407 ND 3/51 Cathepsin D (227–241) P07339 Cv 
 1863.0 16 RPKDYEVDATLKSLNN DRB1*0407 ND 15/51 Collagen α2 chain (I) C-term propeptide (1129–1144) P08123 EM 
 1613.8 15 KPGQFIRSVDPDSPA DRB1*0407 ND 0/51 NHERF (174–188) O14745 
 1449.8 12 EDFRDGLKLMLL DRB1*0407 ND 3/51 α-Actinin 1 (57–68) or α-Actinin 4 (76–87) P12814 
 1895.1 16 DVPKWISIMTERSVPH DRB1*0407 ND 19/51 Annexin II (208–223) P07335 EM 
Sample: TB449 (DRB1*1501, DRB5*0101/DRB1*0301, DRB3*0101)         
 1362.7 12 VVDLMAHMASKE DRB1*1501 19.39 (LB) 5/51 GAPDH (324–335) NP002037 
 1763.8 14 LISWYDNEFGYSNR DRB1*1501 16.33 (LB) 0/51 GAPDH (310–323) NP002037 
 1650.2 13 ISWYDNEFGYSNR DRB1*1501 16.33 (LB) 0/51 GAPDH (311–323) NP002037 
 1779.2 14 LISwYDNEFGYSNR DRB1*1501 16.33 (LB) 0/51 GAPDH (310–323) NP002037 
 1274.6 10 LLVVYPWTQR DRB1*1501 39.80 21/51 Hemoglobin δ-chain (31–40) P02042 
 1308.2 10 LVVYPWTQRF DRB1*1501 39.80 21/51 Hemoglobin δ-chain (32–41) P02042 
 1421.8 11 LLVVYPWTQRF DRB1*1501 39.80 21/51 Hemoglobin δ-chain (31–41) P02042 
 2301.3 22 LVTLAAHLTAEFTPAVHASLDK DRB1*1501 24.49 (LB) 7/51 Hemoglobin α-chain (106–127) P69905 
 1687.0 13 LKKYLYEIARRHP DRB1*1501 33.67 6/51 Serum albumin (159–171) P02768 Co 
 1567.0 15 TNVPPEVTVLTNSPV DRB1*1501 23.47 (LB) 3/51 HLA-II, DR α-chain (108–122) P01903 
 1690.9 15 GNRIAQWQSFQLEGG DRB1*1501 29.60 1/51 α-2-macroglobulin (172–186) P01023 
 1633.9 14 GNRIAQWQSFQLEG DRB1*1501 29.60 1/51 α-2-macroglobulin (172–185) P01023 
 1522.8 14 AGLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α-1 (XV) chain (1244–1257) P39059 EM 
 1593.6 15 AAGLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α-1 (XV) chain (1243–1257) P39059 EM 
 1521.0 13 DGVIKVFNDMKVR DRB1*1501 57.10 3/51 Cofilin, non-muscle isoform (Cofilin-1) (8–20) P23528 
 1310.8 10 WRLEEFGRFA DRB5*0101 20.41 (LB) 0/51 HLA-II, DR α-chain (68–76) P01903 
 1409.8 11 VWRLEEFGRFA DRB5*0101 22.45 (LB) 1/51 HLA-II, DR α-chain (67–76) P01903 
 1687.1 16 RVMTIAPGLFGTPLLT DRB5*0101 34.69 4/51 3-Hydroxyacyl-CoA-dehydrogenase type II (192–207) Q99714 
 1815.0 16 DKFLASVSTVLPSKYR DRB5*0101 16.33 (LB) 4/51 Hemoglobin α-chain (126–141) P69905 
 1571.6 14 FLASVSTVLPSKYR DRB5*0101 16.33 (LB) 4/51 Hemoglobin α-chain (128–141) P69905 
 1600.9 14 RGNSIIMLEALERV DRB5*0101 53.06 29/51 Small nuclear ribonucleoprotein G (63–76) Q15357 
 1857.0 16 IGSSYFPEHGYFRAPE DRB5*0101 32.7 2/51 Multimerin 2 (851–866) Q9H8L6 EM 
 1728.1 15 QLQPDNSTLTWVKPT DRB1*0301 32.63 7/51 Phospholipase C (883–897) Q9HBX6 
 1914.4 18 IVNTNVPRASVPDGFLSE DRB1*0301 36.84 23/51 Macrophage migration inhibitory factor (5–22) P14174 
 2245.3 21 LAQATGKPPQYIAVHVVPDQL DRB1*0301 23.16 (LB) 11/51 Macrophage migration inhibitory factor (27–47) P14174 
 2305.6 21 PmFIVNTNVPRASVPDGFLSE DRB1*0301 36.84 32/51 Macrophage migration inhibitory factor (2–22) P14174 
 2089.6 19 PmFIVNTNVPRASVPDGFL DRB1*0301 36.84 32/51 Macrophage migration inhibitory factor (2–20) P14174 
 2384.6 22 IVNTNVPRASVPDGFLSELTQQ DRB1*0301 36.84 23/51 Macrophage migration inhibitory factor (5–26) P14174 
 1362.4 12 mLLIPTSFSPLK DRB1*0301 47.05 30/51 NipSnap4 protein (236–247) Q9UFN0 
 1744.0 16 WLRNGKPVTTGVSETV DRB1*0301 35.79 12/51 HLA-II, DR α-chain (146–161) P01903 
 1845.1 17 TWLRNGKPVTTGVSETV DRB1*0301 35.79 12/51 HLA-II, DR α-chain (145–161) P01903 
 1489.0 14 EDIVADHVASYGVN DRB1*0301 62.11 12/51 HLA-II, DQ(2,6,3) α-chain (24–37) P04226 
 1937.1 15 EQFYVDLDKKETVWH DRB1*0301 35.76 10/51 MHC-II (61–75) Q95HB9 
 1730.9 15 FDPIIEDRHGGYKPS DRB1*0301 26.30 0/51 Creatine kinase, B chain (B-CK) (89–103) P12277 
 1643.8 14 FDPIIEDRHGGYKP DRB1*0301 26.30 0/51 Creatine kinase, B chain (B-CK) (89–102) P12277 
 1919.3 18 RLLLPGELAKHAVSEGTK DRB1*0301 17.58 (LB) 0/51 Histone H2B K (99–116) P62807 
 1416.9 13 RLLLPGELAKHAV DRB1*0301 17.58 (LB) 0/51 Histone H2B (99–111) P62807 
 2031.2 19 AFTPEGERLIGDAAKNQLT DRB1*0301 26.32 5/51 BIP (67–85) P11021 ER 
 1670.0 15 LSSVVVDPSIRHFDV DRB1*0301 49.47 16/51 Thyroglobulin (2098–2112) P01266 Co 
 2043.2 23 TGGASGLGLATAERLVGQGASAV DRB1*0301 28.11 4/51 3-Hydroxyacyl-CoA-dehydrogenase type II (16–38) Q99714 
 1530.4 14 GTLIIRDVKESDQG DRB1*0301 48.42 34/51 Heparan sulfate proteoglycan (462–475) P98160 EM 
 1626.9 14 KEYFAIDNSGRIIT DRB1*0301 35.79 10/51 Cadherin-5 (197–210) P33151 
 1684.0 15 GKEYFAIDNSGRIIT DRB1*0301 35.79 10/51 Cadherin-5 (196–210) P33151 
 1171.6 11 VLSPADKTNV DRB1*0301 15.79 (LB) 0/51 Hemoglobin α-chain (1–11) P69905 
 1504.9 13 RYSVDIPLDKTVV DRB1*0301 21.05 (LB) 6/51 60S Ribosomal protein L27 P61353 
 1175.7 10 FLENVIRDAV NA  0/51 Histone H4 (61–70) P62805 
(Table continues        
Mr (Da)aSizeb (aa)PPPPPPPPPc123456789Allele Assignedd% Affinity IndexePromiscuityfSource ProteingId No.hLi
Sample: TB448 (DRB1*1501, DRB5*0101/DRB1*0407, DRB4*0103)         
 1770.0 15 DIPELVNMGQWKIRA DRB1*1501 41.84 27/51 Complement C3 (194–208) P01024 
 1594.0 15 AAGLLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α1 chain (XV) (1243–1257) P39059 EM 
 1522.8 14 AGLLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α1 chain (XV) (1244–1257) P39059 EM 
 1687.0 13 LKKYLYEIARRHP DRB1*1501 33.67 6/51 Serum albumin (159–171) P02768 Co 
 2014.2 16 YPKSLHMYANRLLDHR DRB1*1501 55.10 19/51 Apolipoprotein B (1200–1215) P04114 
 1629.8 15 DVGVYRAVTOQGRPD DRB5*0101 68.37 7/51 DQB1*0602 β-chain (75–89) P03992 
 1676.8 15 DVGEYRAVTELGRPD DRB5*0101 44.90 3/51 HLA II β-chain DRB1*4 (72–86) P13760 
 1633.0 15 GGELRFTVTQRSQPG DRB5*0101 29.59 39/51 Heparan sulfate proteoglycan (1006–1020) P98160 EM 
 1795.0 16 YGGELRFTVTQRSQPG DRB5*0101 29.59 39/51 Heparan sulfate proteoglycan (1005–1020) P98160 EM 
 1660.8 14 YDHNFVKAINAIQK DRB5*0101 54.08 19/51 Cathepsin C (170–183) P53634 Cv 
 1661.0 16 RLPIIDVAPLDVGAPD DRB5*0101 13.27 (LB) 7/51 Collagen α1 chain (I) C-term propeptide (1436–1451) P02452 EM 
 2030.0 18 CPTPCQLQAEQAFLRTVQ DRB5*0101 29.59 2/51 Thyroglobulin (726–743) P01266 Co 
 1678.9 12 REIFNMARQRSR DRB5*0101 46.94 30/51 Apolipoprotein B (427–439) P04114 
 1720.0 16 IRASYAQQPAESRVSG DRB1*0407 ND 0/51 Heparan sulfate proteoglycan (1089–1104) P98160 EM 
 1827.0 16 NIFSFYLSRDPDAQPG DRB1*0407 ND 3/51 Cathepsin D (227–241) P07339 Cv 
 1863.0 16 RPKDYEVDATLKSLNN DRB1*0407 ND 15/51 Collagen α2 chain (I) C-term propeptide (1129–1144) P08123 EM 
 1613.8 15 KPGQFIRSVDPDSPA DRB1*0407 ND 0/51 NHERF (174–188) O14745 
 1449.8 12 EDFRDGLKLMLL DRB1*0407 ND 3/51 α-Actinin 1 (57–68) or α-Actinin 4 (76–87) P12814 
 1895.1 16 DVPKWISIMTERSVPH DRB1*0407 ND 19/51 Annexin II (208–223) P07335 EM 
Sample: TB449 (DRB1*1501, DRB5*0101/DRB1*0301, DRB3*0101)         
 1362.7 12 VVDLMAHMASKE DRB1*1501 19.39 (LB) 5/51 GAPDH (324–335) NP002037 
 1763.8 14 LISWYDNEFGYSNR DRB1*1501 16.33 (LB) 0/51 GAPDH (310–323) NP002037 
 1650.2 13 ISWYDNEFGYSNR DRB1*1501 16.33 (LB) 0/51 GAPDH (311–323) NP002037 
 1779.2 14 LISwYDNEFGYSNR DRB1*1501 16.33 (LB) 0/51 GAPDH (310–323) NP002037 
 1274.6 10 LLVVYPWTQR DRB1*1501 39.80 21/51 Hemoglobin δ-chain (31–40) P02042 
 1308.2 10 LVVYPWTQRF DRB1*1501 39.80 21/51 Hemoglobin δ-chain (32–41) P02042 
 1421.8 11 LLVVYPWTQRF DRB1*1501 39.80 21/51 Hemoglobin δ-chain (31–41) P02042 
 2301.3 22 LVTLAAHLTAEFTPAVHASLDK DRB1*1501 24.49 (LB) 7/51 Hemoglobin α-chain (106–127) P69905 
 1687.0 13 LKKYLYEIARRHP DRB1*1501 33.67 6/51 Serum albumin (159–171) P02768 Co 
 1567.0 15 TNVPPEVTVLTNSPV DRB1*1501 23.47 (LB) 3/51 HLA-II, DR α-chain (108–122) P01903 
 1690.9 15 GNRIAQWQSFQLEGG DRB1*1501 29.60 1/51 α-2-macroglobulin (172–186) P01023 
 1633.9 14 GNRIAQWQSFQLEG DRB1*1501 29.60 1/51 α-2-macroglobulin (172–185) P01023 
 1522.8 14 AGLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α-1 (XV) chain (1244–1257) P39059 EM 
 1593.6 15 AAGLSTYRAFLSSH DRB1*1501 33.67 2/51 Collagen α-1 (XV) chain (1243–1257) P39059 EM 
 1521.0 13 DGVIKVFNDMKVR DRB1*1501 57.10 3/51 Cofilin, non-muscle isoform (Cofilin-1) (8–20) P23528 
 1310.8 10 WRLEEFGRFA DRB5*0101 20.41 (LB) 0/51 HLA-II, DR α-chain (68–76) P01903 
 1409.8 11 VWRLEEFGRFA DRB5*0101 22.45 (LB) 1/51 HLA-II, DR α-chain (67–76) P01903 
 1687.1 16 RVMTIAPGLFGTPLLT DRB5*0101 34.69 4/51 3-Hydroxyacyl-CoA-dehydrogenase type II (192–207) Q99714 
 1815.0 16 DKFLASVSTVLPSKYR DRB5*0101 16.33 (LB) 4/51 Hemoglobin α-chain (126–141) P69905 
 1571.6 14 FLASVSTVLPSKYR DRB5*0101 16.33 (LB) 4/51 Hemoglobin α-chain (128–141) P69905 
 1600.9 14 RGNSIIMLEALERV DRB5*0101 53.06 29/51 Small nuclear ribonucleoprotein G (63–76) Q15357 
 1857.0 16 IGSSYFPEHGYFRAPE DRB5*0101 32.7 2/51 Multimerin 2 (851–866) Q9H8L6 EM 
 1728.1 15 QLQPDNSTLTWVKPT DRB1*0301 32.63 7/51 Phospholipase C (883–897) Q9HBX6 
 1914.4 18 IVNTNVPRASVPDGFLSE DRB1*0301 36.84 23/51 Macrophage migration inhibitory factor (5–22) P14174 
 2245.3 21 LAQATGKPPQYIAVHVVPDQL DRB1*0301 23.16 (LB) 11/51 Macrophage migration inhibitory factor (27–47) P14174 
 2305.6 21 PmFIVNTNVPRASVPDGFLSE DRB1*0301 36.84 32/51 Macrophage migration inhibitory factor (2–22) P14174 
 2089.6 19 PmFIVNTNVPRASVPDGFL DRB1*0301 36.84 32/51 Macrophage migration inhibitory factor (2–20) P14174 
 2384.6 22 IVNTNVPRASVPDGFLSELTQQ DRB1*0301 36.84 23/51 Macrophage migration inhibitory factor (5–26) P14174 
 1362.4 12 mLLIPTSFSPLK DRB1*0301 47.05 30/51 NipSnap4 protein (236–247) Q9UFN0 
 1744.0 16 WLRNGKPVTTGVSETV DRB1*0301 35.79 12/51 HLA-II, DR α-chain (146–161) P01903 
 1845.1 17 TWLRNGKPVTTGVSETV DRB1*0301 35.79 12/51 HLA-II, DR α-chain (145–161) P01903 
 1489.0 14 EDIVADHVASYGVN DRB1*0301 62.11 12/51 HLA-II, DQ(2,6,3) α-chain (24–37) P04226 
 1937.1 15 EQFYVDLDKKETVWH DRB1*0301 35.76 10/51 MHC-II (61–75) Q95HB9 
 1730.9 15 FDPIIEDRHGGYKPS DRB1*0301 26.30 0/51 Creatine kinase, B chain (B-CK) (89–103) P12277 
 1643.8 14 FDPIIEDRHGGYKP DRB1*0301 26.30 0/51 Creatine kinase, B chain (B-CK) (89–102) P12277 
 1919.3 18 RLLLPGELAKHAVSEGTK DRB1*0301 17.58 (LB) 0/51 Histone H2B K (99–116) P62807 
 1416.9 13 RLLLPGELAKHAV DRB1*0301 17.58 (LB) 0/51 Histone H2B (99–111) P62807 
 2031.2 19 AFTPEGERLIGDAAKNQLT DRB1*0301 26.32 5/51 BIP (67–85) P11021 ER 
 1670.0 15 LSSVVVDPSIRHFDV DRB1*0301 49.47 16/51 Thyroglobulin (2098–2112) P01266 Co 
 2043.2 23 TGGASGLGLATAERLVGQGASAV DRB1*0301 28.11 4/51 3-Hydroxyacyl-CoA-dehydrogenase type II (16–38) Q99714 
 1530.4 14 GTLIIRDVKESDQG DRB1*0301 48.42 34/51 Heparan sulfate proteoglycan (462–475) P98160 EM 
 1626.9 14 KEYFAIDNSGRIIT DRB1*0301 35.79 10/51 Cadherin-5 (197–210) P33151 
 1684.0 15 GKEYFAIDNSGRIIT DRB1*0301 35.79 10/51 Cadherin-5 (196–210) P33151 
 1171.6 11 VLSPADKTNV DRB1*0301 15.79 (LB) 0/51 Hemoglobin α-chain (1–11) P69905 
 1504.9 13 RYSVDIPLDKTVV DRB1*0301 21.05 (LB) 6/51 60S Ribosomal protein L27 P61353 
 1175.7 10 FLENVIRDAV NA  0/51 Histone H4 (61–70) P62805 
(Table continues        
Table IIA.

(Continued)

Mr (Da)aSizeb (aa)PPPPPPPPPc123456789Allele Assignedd% Affinity IndexePromiscuityfSource ProteingId No.hLi
 1442.8 14 SRAGLELGAEPETI NA  0/51 Nidogen-2 (NID-2) (1167–1180) Q14112 EM 
 1733.9 15 GELEKEEAQPIVTKY NA  0/51 Hypothetical protein (490–504) Q8N3B3  
 1397.7 12 RFDSDVGEFRAV NA  0/51 HLA-II, DR, DP DQ β-chain (68–79) P01911 
 1257.6 11 FDSDVGEYRAV NA  0/51 HLA-II, DRB3-1 β-chain (69–79) P79483 
 1807.0 18 IYTRNTKGGDAPAAGEDA NA  6/51 Ribosomal protein S25 (108–125) P62851 
 1714.4 16 AVPSPPPASPRSQYNF NA  0/51 Serine protease (134–149) O43464 
 2013.4 19 AVPSPPPASPRSQYNFIAD NA  0/51 Serine protease (134–152) O43464 
 1311.6 12 ASVSTVLPSKYR NA  4/51 Hemoglobin α-chain (130–141) P69905 
 1735.8 14 SWYDNEFGYSNRVV NA  0/51 GAPDH (312–325) P04406 
 1631.9 16 SWNSGALTSGVHTFPA NA  0/51 Ig γ-1 chain C (40–55) P08157 
 1637.6 18 AKVAVLGASGGIGQPLSL NA  0/51 Malate dehydrogenase (25–42) P40926 mit 
 2167.3 19 LLTEAPLNPKANREKMTQI NA  3/51 Actin (106–124) P63261 
 1296.7 10 DRVYIHPFHL NA  0/51 Angiotensin α (34–43) ANHU 
 1543.8 13 DTNADKQLSFEEF NA  0/51 Calgranulin B (67–79) P06702 
 1004.6 IKEESKTAV NA  0/51 CENP-F kinetochore protein (2346–2354) P49454 
Sample: TB471 (DRB1*0301, DRB3*0101/DRB1*0801)         
 3274.66 31 VHLTPEEKSAVTALWGKVNVDEVGGEALGRL DRB1*0301 48.42 3/51 Hemoglobin β-chain (1–31) P02023 
 3387.83 32 VHLTPEEKSAVTALWGKVNVDEVGGEALGRLL DRB1*0301 48.42 3/51 Hemoglobin β-chain (1–32) P02023 
 1308.63 10 LVVYPWTQRF DRB1*0301 36.84 12/51 Hemoglobin β-chain (32–41) P02042 
 1956.92 18 QVTQPTVGMNFKTPRGPV DRB1*0301 40 18/51 40S ribosomal protein S17 (107–134) P08708 
 1470.73 12 YYKVDENGKISR DRB1*0301 31.58 8/51 40S ribosomal protein 27 (29–40) P62979 
 1905.95 16 AHLLILRDTKTYMLAF DRB1*0301 66.32 26/51 α-1 acid glycoprotein (117–132) P02763 
 1758.92 15 AHLLILRDTKTYMLA DRB1*0301 66.32 26/51 α-1 acid glycoprotein (117–131) P02763 
 1374.6 13 KAVLTIDEKGTEA DRB1*0301 51.58 5/51 α-1-antitrypsin (359–371) P01009 
 1501.8 15 KAVLTIDEKGTEAAG DRB1*0301 51.58 5/51 α-1-antitrypsin (359–373) P01009 
 1737.6 17 VHKAVLTIDEKGTEAAG DRB1*0301 51.58 5/51 α-1-antitrypsin (357–373) P01009 
 1774.09 15 KVKKIYLDEKRLLAG DRB1*0301 48 12/51 SPARC-like protein 1 (556–570) Q14515 
 1683.87 15 GKEYFAIDNSGRIIT DRB1*0301 35.79 11/51 VE-cadherin (196–210) P33151 
 1915.05 18 DPSPSPVLGYKIVYKPVG DRB1*0301 42.11 23/51 Collagen α-1 type XII (2137–2154) Q5VYK1 EM 
 1765.8 15 EGLTFQMKKNAEELK DRB1*0801 65.1 17/51 Apolipoprotein A-IV (239–253) P06727 
 1636.78 14 GLTFQMKKNAEELK DRB1*0801 65.1 17/51 Apolipoprotein A-IV (240–253) P06727 
 1395.6 12 GLTFQMKKNAEE DRB1*0801 65.1 17/51 Apolipoprotein A-IV (240–251) P06727 
 1893.97 17 AYVRLAPDYDALDVANK DRB1*0801 30.23 8/51 60S ribosomal protein L23a (136–152) P62750 
 1609.8 14 SVPRYLPRPANPDE DRB1*0801 38.37 7/51 Epithelial-cadherin (793–806) P12830 
 1970.94 18 QVDQFLGVPYAAPPLAER DRB1*0301 15,79 (LB) 0/51 Thyroglobulin (2224–2241) P01266 Co 
 1059.4 10 LSLQEPGSKT DRB1*0301 11,58 (LB) 2/51 Thyroglobulin (2756–2765) P01266 Co 
 1448.71 14 VVAGVANALAHKYH DRB1*0301 26,00 (LB) 9/51 Hemoglobin β-chain (133–146) 68871/P0204 
 1683.91 16 VLSPADKTNVKAAWGK DRB1*0301 15,79 (LB) 0/51 Hemoglobin α-chain (1–16) P01922 
 3196.9 31 VLSPADKTNVKAAWGKVGAHAGEYGAEALER DRB1*0301 15,79 (LB) 2/51 Hemoglobin α-chain (1–31) P01922 
 3327.44 32 VLSPADKTNVKAAWGKVGAHAGEYGAEALERM DRB1*0301 15,79 (LB)  Hemoglobin α-chain (1–32) P01922 
 3344.8 32 VLSPADKTNVKAAWGKVGAHAGEYGAEALERm DRB1*0301 15,79 (LB) 2/51 Hemoglobin α-chain (1–32) P01922 
 3473.61 33 VLSPADKTNVKAAWGKVGAHAGEYGAEALERMF DRB1*0301 15,79 (LB) 2/51 Hemoglobin α-chain (1–33) P01922 
 2582.24 24 VGAHAGEYGAEALERMFLSFPTTK DRB1*0301 28,42 (LB) 6/51 Hemoglobin α-chain (17–40) P01922 
 1324.49 12 LITPAVVSERLK DRB1*0301 21,79 (LB) 0/51 40S ribosomal protein S25 (67–78) P62851 
 1925.95 18 GRMHAPGKGLSQSALPYR DRB1*0301 24,21 (LB) 0/51 40S ribosomal protein S13 (1–18) P62277 
 1709.88 16 TGAAPIIDVVRSGYYK DRB1*0301 12,32 (LB) 0/51 60S ribosomal protein L27α (94–109) P46776 
 1688.6 16 ASGNYATVISHNPETK DRB1*0301 14,74 (LB) 0/51 60S ribosomal protein L8 (128–143) P62917 
 1457.6 15 AGNLGGGVVTIERSK DRB1*0301 21,05 (LB) 0/51 60S ribosomal protein L22 (52–66) P35268 
 2094.02 18 SYELPDGQVITIGNERFR DRB1*0301 17,89 (LB) 2/51 Actin (α, β, γ) (239–256) P60709 
 1161.4 11 EITALAPSTMK DRB1*0301 17,58 (LB) 0/51 Actin (α, β, γ) (316–326) P60709 
 1886.93 17 GTQGKIVDLVKELDRDT DRB1*0301 13,68 (LB) 0/51 α-1 antitrypsin (188–204) P01009 
 1829.9 16 TQGKIVDLVKELDRDT DRB1*0301 13,68 (LB) 0/51 α-1 antitrypsin (189–204) P01009 
 1613.86 14 TQGKIVDLVKELDR DRB1*0301 13,68 (LB) 0/51 α-1 antitrypsin (189–202) P01009 
 2953.23 33 SGNFGGSRNMGGPYGGGNYGPGGSGGSGGYGGR DRB1*0301 14,74 (LB) 4/51 hnRNP A2/B1 (318–350) P22626 
 1739.83 18 DGRGALQNIIPASTGAAK DRB1*0301 24,21 (LB) 2/51 GAPDH (197–214) P04406 
 3095.54 29 AAWGKVGAHAGEYGAEALERMFLSFPTTK DRB1*0801 28,42 (LB) 7/51 Hemoglobin α-chain (12–40) P01922 
 2690.51 23 AVTELGRPDAEYWNSQKDLLEQK NA  0/51 MHC-II (78–100) P13760 
 2755.38 24 FLSFPTTKTYFPHFDLSHGSAQVK NA  6/51 Hemoglobin α-chain (33–56) P01922 
 3262.68 29 FLSFPTTKTYFPHFDLSHGSAQVKGHGKK NA  6/51 Hemoglobin α-chain (33–61) P01922 
 3134.58 28 FLSFPTTKTYFPHFDLSHGSAQVKGHGK NA  6/51 Hemoglobin α-chain (33–60) P01922 
 1833.69 16 TYFPHFDLSHGSAQVK NA  0/51 Hemoglobin α-chain (41–56) P01922 
 2341.18 21 TYFPHFDLSHGSAQVKGHGKK NA  0/51 Hemoglobin α-chain (41–61) P01922 
 2996.49 29 VADALTNAVAHVDNMPNALSALSDLHAHK NA  3/51 Hemoglobin α-chain (62–90) P01922 
 3427.86 32 AAHLPAEFTPAVHASLDKFLASVSTVLTSKYR NA  3/51 Hemoglobin α-chain (110–141) P01922 
 2792.50 27 AAHLPAEFTPAVHASLDKFLASVSTVL NA  3/51 Hemoglobin α-chain (110–136) P01922 
 1699.97 15 KFLASVSTVLTSKYR NA  4/51 Hemoglobin α-chain (127–141) P01922 
 2846.44 24 AVTELGRPDAEYWNSQKDLLEQKR NA  0/51 HLA-II, DRB1–4 β-chain (78–101) P13760 
 2744.39 26 KTNVKAAWGKVGAHAGEYGAEALERM NA  2/51 Hemoglobin α-chain (7–32) P01922 
 2174.04 21 AAWGKVGAHAGEYGAEALERM NA  2/51 Hemoglobin α-chain (12–32) P01922 
 1998.98 20 PAAPAAAPPAEKAPVKKKAR NA  0/51 Histone H1.4 (histone H1b) (5–24) P10412 
 2044 21 SETAPAAPAAAPPAEKAPVKK NA  0/51 Histone H1.2 (1–21) P16403 
 2171.8 22 SETAPAAPAAAPPAEKAPVKKK NA  0/51 Histone H1.2 (1–22) P16403 
 1331.73 15 GAPAAATAPAPTAHK NA  0/51 Histone H1 (129–143) Q92522 
 1511.66 12 AKFEELNMDLFR NA  0/51 BIP protein (325–336) P11021 ER 
 1274.4 12 IGSSQDDGLINR NA  0/51 Thyroglobulin (2508–2519) P01266 Co 
 1220.52 11 SRTSGLLSSWK NA  0/51 Thyroglobulin (1080–1090) P01266 Co 
 1880.78 17 AKAVKQFEESQGRTSSK NA  0/51 Thyroglobulin (2520–2536) P01266 Co 
 1477.69 13 AKAVKQFEESQGR NA  0/51 Thyroglobulin (2520–2532) P01266 Co 
 952.4 VHLTPEEK NA  0/51 Hemoglobin β-chain (1–8) P02023 
 2139.11 20 FESFGDLSTPDAVMGNPKVK NA  1/51 Hemoglobin β-chain (42–61) P68871 
 1910.88 18 FESFGDLSTPDAVMGNPK NA  1/51 Hemoglobin β-chain (42–59) P68871 
 1528.2 15 VGAHAGEYGAEALER NA  0/51 Hemoglobin α-chain (17–31) P01922 
 1214.6 11 TVDGPSGKLWR NA  0/51 GAPDH (186–196) P04406 
 1087.5 LRVDPVNFK NA  4/51 Hemoglobin α-chain (91–99) P01922 
(Table continues        
Mr (Da)aSizeb (aa)PPPPPPPPPc123456789Allele Assignedd% Affinity IndexePromiscuityfSource ProteingId No.hLi
 1442.8 14 SRAGLELGAEPETI NA  0/51 Nidogen-2 (NID-2) (1167–1180) Q14112 EM 
 1733.9 15 GELEKEEAQPIVTKY NA  0/51 Hypothetical protein (490–504) Q8N3B3  
 1397.7 12 RFDSDVGEFRAV NA  0/51 HLA-II, DR, DP DQ β-chain (68–79) P01911 
 1257.6 11 FDSDVGEYRAV NA  0/51 HLA-II, DRB3-1 β-chain (69–79) P79483 
 1807.0 18 IYTRNTKGGDAPAAGEDA NA  6/51 Ribosomal protein S25 (108–125) P62851 
 1714.4 16 AVPSPPPASPRSQYNF NA  0/51 Serine protease (134–149) O43464 
 2013.4 19 AVPSPPPASPRSQYNFIAD NA  0/51 Serine protease (134–152) O43464 
 1311.6 12 ASVSTVLPSKYR NA  4/51 Hemoglobin α-chain (130–141) P69905 
 1735.8 14 SWYDNEFGYSNRVV NA  0/51 GAPDH (312–325) P04406 
 1631.9 16 SWNSGALTSGVHTFPA NA  0/51 Ig γ-1 chain C (40–55) P08157 
 1637.6 18 AKVAVLGASGGIGQPLSL NA  0/51 Malate dehydrogenase (25–42) P40926 mit 
 2167.3 19 LLTEAPLNPKANREKMTQI NA  3/51 Actin (106–124) P63261 
 1296.7 10 DRVYIHPFHL NA  0/51 Angiotensin α (34–43) ANHU 
 1543.8 13 DTNADKQLSFEEF NA  0/51 Calgranulin B (67–79) P06702 
 1004.6 IKEESKTAV NA  0/51 CENP-F kinetochore protein (2346–2354) P49454 
Sample: TB471 (DRB1*0301, DRB3*0101/DRB1*0801)         
 3274.66 31 VHLTPEEKSAVTALWGKVNVDEVGGEALGRL DRB1*0301 48.42 3/51 Hemoglobin β-chain (1–31) P02023 
 3387.83 32 VHLTPEEKSAVTALWGKVNVDEVGGEALGRLL DRB1*0301 48.42 3/51 Hemoglobin β-chain (1–32) P02023 
 1308.63 10 LVVYPWTQRF DRB1*0301 36.84 12/51 Hemoglobin β-chain (32–41) P02042 
 1956.92 18 QVTQPTVGMNFKTPRGPV DRB1*0301 40 18/51 40S ribosomal protein S17 (107–134) P08708 
 1470.73 12 YYKVDENGKISR DRB1*0301 31.58 8/51 40S ribosomal protein 27 (29–40) P62979 
 1905.95 16 AHLLILRDTKTYMLAF DRB1*0301 66.32 26/51 α-1 acid glycoprotein (117–132) P02763 
 1758.92 15 AHLLILRDTKTYMLA DRB1*0301 66.32 26/51 α-1 acid glycoprotein (117–131) P02763 
 1374.6 13 KAVLTIDEKGTEA DRB1*0301 51.58 5/51 α-1-antitrypsin (359–371) P01009 
 1501.8 15 KAVLTIDEKGTEAAG DRB1*0301 51.58 5/51 α-1-antitrypsin (359–373) P01009 
 1737.6 17 VHKAVLTIDEKGTEAAG DRB1*0301 51.58 5/51 α-1-antitrypsin (357–373) P01009 
 1774.09 15 KVKKIYLDEKRLLAG DRB1*0301 48 12/51 SPARC-like protein 1 (556–570) Q14515 
 1683.87 15 GKEYFAIDNSGRIIT DRB1*0301 35.79 11/51 VE-cadherin (196–210) P33151 
 1915.05 18 DPSPSPVLGYKIVYKPVG DRB1*0301 42.11 23/51 Collagen α-1 type XII (2137–2154) Q5VYK1 EM 
 1765.8 15 EGLTFQMKKNAEELK DRB1*0801 65.1 17/51 Apolipoprotein A-IV (239–253) P06727 
 1636.78 14 GLTFQMKKNAEELK DRB1*0801 65.1 17/51 Apolipoprotein A-IV (240–253) P06727 
 1395.6 12 GLTFQMKKNAEE DRB1*0801 65.1 17/51 Apolipoprotein A-IV (240–251) P06727 
 1893.97 17 AYVRLAPDYDALDVANK DRB1*0801 30.23 8/51 60S ribosomal protein L23a (136–152) P62750 
 1609.8 14 SVPRYLPRPANPDE DRB1*0801 38.37 7/51 Epithelial-cadherin (793–806) P12830 
 1970.94 18 QVDQFLGVPYAAPPLAER DRB1*0301 15,79 (LB) 0/51 Thyroglobulin (2224–2241) P01266 Co 
 1059.4 10 LSLQEPGSKT DRB1*0301 11,58 (LB) 2/51 Thyroglobulin (2756–2765) P01266 Co 
 1448.71 14 VVAGVANALAHKYH DRB1*0301 26,00 (LB) 9/51 Hemoglobin β-chain (133–146) 68871/P0204 
 1683.91 16 VLSPADKTNVKAAWGK DRB1*0301 15,79 (LB) 0/51 Hemoglobin α-chain (1–16) P01922 
 3196.9 31 VLSPADKTNVKAAWGKVGAHAGEYGAEALER DRB1*0301 15,79 (LB) 2/51 Hemoglobin α-chain (1–31) P01922 
 3327.44 32 VLSPADKTNVKAAWGKVGAHAGEYGAEALERM DRB1*0301 15,79 (LB)  Hemoglobin α-chain (1–32) P01922 
 3344.8 32 VLSPADKTNVKAAWGKVGAHAGEYGAEALERm DRB1*0301 15,79 (LB) 2/51 Hemoglobin α-chain (1–32) P01922 
 3473.61 33 VLSPADKTNVKAAWGKVGAHAGEYGAEALERMF DRB1*0301 15,79 (LB) 2/51 Hemoglobin α-chain (1–33) P01922 
 2582.24 24 VGAHAGEYGAEALERMFLSFPTTK DRB1*0301 28,42 (LB) 6/51 Hemoglobin α-chain (17–40) P01922 
 1324.49 12 LITPAVVSERLK DRB1*0301 21,79 (LB) 0/51 40S ribosomal protein S25 (67–78) P62851 
 1925.95 18 GRMHAPGKGLSQSALPYR DRB1*0301 24,21 (LB) 0/51 40S ribosomal protein S13 (1–18) P62277 
 1709.88 16 TGAAPIIDVVRSGYYK DRB1*0301 12,32 (LB) 0/51 60S ribosomal protein L27α (94–109) P46776 
 1688.6 16 ASGNYATVISHNPETK DRB1*0301 14,74 (LB) 0/51 60S ribosomal protein L8 (128–143) P62917 
 1457.6 15 AGNLGGGVVTIERSK DRB1*0301 21,05 (LB) 0/51 60S ribosomal protein L22 (52–66) P35268 
 2094.02 18 SYELPDGQVITIGNERFR DRB1*0301 17,89 (LB) 2/51 Actin (α, β, γ) (239–256) P60709 
 1161.4 11 EITALAPSTMK DRB1*0301 17,58 (LB) 0/51 Actin (α, β, γ) (316–326) P60709 
 1886.93 17 GTQGKIVDLVKELDRDT DRB1*0301 13,68 (LB) 0/51 α-1 antitrypsin (188–204) P01009 
 1829.9 16 TQGKIVDLVKELDRDT DRB1*0301 13,68 (LB) 0/51 α-1 antitrypsin (189–204) P01009 
 1613.86 14 TQGKIVDLVKELDR DRB1*0301 13,68 (LB) 0/51 α-1 antitrypsin (189–202) P01009 
 2953.23 33 SGNFGGSRNMGGPYGGGNYGPGGSGGSGGYGGR DRB1*0301 14,74 (LB) 4/51 hnRNP A2/B1 (318–350) P22626 
 1739.83 18 DGRGALQNIIPASTGAAK DRB1*0301 24,21 (LB) 2/51 GAPDH (197–214) P04406 
 3095.54 29 AAWGKVGAHAGEYGAEALERMFLSFPTTK DRB1*0801 28,42 (LB) 7/51 Hemoglobin α-chain (12–40) P01922 
 2690.51 23 AVTELGRPDAEYWNSQKDLLEQK NA  0/51 MHC-II (78–100) P13760 
 2755.38 24 FLSFPTTKTYFPHFDLSHGSAQVK NA  6/51 Hemoglobin α-chain (33–56) P01922 
 3262.68 29 FLSFPTTKTYFPHFDLSHGSAQVKGHGKK NA  6/51 Hemoglobin α-chain (33–61) P01922 
 3134.58 28 FLSFPTTKTYFPHFDLSHGSAQVKGHGK NA  6/51 Hemoglobin α-chain (33–60) P01922 
 1833.69 16 TYFPHFDLSHGSAQVK NA  0/51 Hemoglobin α-chain (41–56) P01922 
 2341.18 21 TYFPHFDLSHGSAQVKGHGKK NA  0/51 Hemoglobin α-chain (41–61) P01922 
 2996.49 29 VADALTNAVAHVDNMPNALSALSDLHAHK NA  3/51 Hemoglobin α-chain (62–90) P01922 
 3427.86 32 AAHLPAEFTPAVHASLDKFLASVSTVLTSKYR NA  3/51 Hemoglobin α-chain (110–141) P01922 
 2792.50 27 AAHLPAEFTPAVHASLDKFLASVSTVL NA  3/51 Hemoglobin α-chain (110–136) P01922 
 1699.97 15 KFLASVSTVLTSKYR NA  4/51 Hemoglobin α-chain (127–141) P01922 
 2846.44 24 AVTELGRPDAEYWNSQKDLLEQKR NA  0/51 HLA-II, DRB1–4 β-chain (78–101) P13760 
 2744.39 26 KTNVKAAWGKVGAHAGEYGAEALERM NA  2/51 Hemoglobin α-chain (7–32) P01922 
 2174.04 21 AAWGKVGAHAGEYGAEALERM NA  2/51 Hemoglobin α-chain (12–32) P01922 
 1998.98 20 PAAPAAAPPAEKAPVKKKAR NA  0/51 Histone H1.4 (histone H1b) (5–24) P10412 
 2044 21 SETAPAAPAAAPPAEKAPVKK NA  0/51 Histone H1.2 (1–21) P16403 
 2171.8 22 SETAPAAPAAAPPAEKAPVKKK NA  0/51 Histone H1.2 (1–22) P16403 
 1331.73 15 GAPAAATAPAPTAHK NA  0/51 Histone H1 (129–143) Q92522 
 1511.66 12 AKFEELNMDLFR NA  0/51 BIP protein (325–336) P11021 ER 
 1274.4 12 IGSSQDDGLINR NA  0/51 Thyroglobulin (2508–2519) P01266 Co 
 1220.52 11 SRTSGLLSSWK NA  0/51 Thyroglobulin (1080–1090) P01266 Co 
 1880.78 17 AKAVKQFEESQGRTSSK NA  0/51 Thyroglobulin (2520–2536) P01266 Co 
 1477.69 13 AKAVKQFEESQGR NA  0/51 Thyroglobulin (2520–2532) P01266 Co 
 952.4 VHLTPEEK NA  0/51 Hemoglobin β-chain (1–8) P02023 
 2139.11 20 FESFGDLSTPDAVMGNPKVK NA  1/51 Hemoglobin β-chain (42–61) P68871 
 1910.88 18 FESFGDLSTPDAVMGNPK NA  1/51 Hemoglobin β-chain (42–59) P68871 
 1528.2 15 VGAHAGEYGAEALER NA  0/51 Hemoglobin α-chain (17–31) P01922 
 1214.6 11 TVDGPSGKLWR NA  0/51 GAPDH (186–196) P04406 
 1087.5 LRVDPVNFK NA  4/51 Hemoglobin α-chain (91–99) P01922 
(Table continues        
Table IIB.

(Continued)

Mr (Da)aSizeb (aa)PPPPPPPPPc123456789Allele Assignedd% Affinity IndexePromiscuityfSource ProteingId No.hLi
 1059.51 IPDWFLNR NA  0/51 40S ribosomal protein S18 (79–86) P62269 
 1699.85 13 TGKNKWFFQKLRF NA  0/51 60S ribosomal protein L27 (123–135) P61353 
Mr (Da)aSizeb (aa)PPPPPPPPPc123456789Allele Assignedd% Affinity IndexePromiscuityfSource ProteingId No.hLi
 1059.51 IPDWFLNR NA  0/51 40S ribosomal protein S18 (79–86) P62269 
 1699.85 13 TGKNKWFFQKLRF NA  0/51 60S ribosomal protein L27 (123–135) P61353 
a

Molecular weight.

b

Length in amino acid number.

c

DR motif core P1-P9, sequences aligned according to their anchor position P1 when known, anchor positions in bold.

d

HLA allele to which each sequence was assigned.

e

ProPred % affinity index: LB, low binder, peptides with low binding capacity or without a correct P1 residue. NA, not assigned. ND, not done.

f

Ratio between the number of alleles to which peptides are good binders and the total number of HLA molecules analyzed. We considered good binders peptides with correct P1 residue for the corresponding allele and with a theoretical affinity score over 25%.

g

The source protein with position of each peptide in parentheses.

h

Database sequence identification number.

i

Subcellular location of the source protein: EM, extracellular matrix; E, exogenous; C, cytoplasmic; Co, colloid; Cv, cytoplasmic vesicula; M, plasma membrane; N, nucleus; ER, endoplasmic reticulum; mit, mitochondria.

All peptides were autologous, as expected from the starting material, thus confirming the absence of contamination from other sources. Peptide size ranged between 9 and 33 aa length, most of them (76.6%) 12–22 aa, the standard class II size range (Fig. 1,a). The mean molecular mass was 1837.6 Da (952.4–3473.6 Da). Fig. 1,b shows the cellular distribution of the DR-associated peptides in two groups: peptides expected to be generated by degradation in the endocytic pathway (E) or in the cytosol (C). The size distribution (Fig. 1 a) of endocytic peptides was within the standard, although it included some small (9–12 aa) and very large (23–33 aa) peptides, whereas the size of “cytosolic” peptides was more heterogeneously distributed.

FIGURE 1.

Location and size analysis of peptides sequenced from thyroid samples TB448, TB449, and TB471. a, Peptide size distribution of TB448 and TB449 peptides: all peptides (hatched bars) and peptides putatively generated in the endocytic pathway (filled bars) or in the cytosolic pathway (open bars). b, Top, cellular distribution of the peptides’ source proteins: E proteins from the endocytic pathway (filled bars) and C, from the cytosolic pathway (open bars); bottom, each group on the top was analyzed according to their intracellular localization: filled bars, E peptides; open bars, C peptides; E, exogenous; Co, colloid; EM, extracellular matrix; M, cellular membrane; Cv, vesicles; C, cytosol; ER/mit, endoplasmic reticulum or mitochondria; N, nucleus.

FIGURE 1.

Location and size analysis of peptides sequenced from thyroid samples TB448, TB449, and TB471. a, Peptide size distribution of TB448 and TB449 peptides: all peptides (hatched bars) and peptides putatively generated in the endocytic pathway (filled bars) or in the cytosolic pathway (open bars). b, Top, cellular distribution of the peptides’ source proteins: E proteins from the endocytic pathway (filled bars) and C, from the cytosolic pathway (open bars); bottom, each group on the top was analyzed according to their intracellular localization: filled bars, E peptides; open bars, C peptides; E, exogenous; Co, colloid; EM, extracellular matrix; M, cellular membrane; Cv, vesicles; C, cytosol; ER/mit, endoplasmic reticulum or mitochondria; N, nucleus.

Close modal

Endocytic peptides were the most abundant, but 46 (30%) peptides were derived from proteins potentially degraded in the cytosol, including nuclear proteins. Of 150 DR ligands identified, 103 were “endocytic”, of which 10 were derived from colloid components. One sequence from TB448, one from TB449, and six different ones from TB471 belonged to the major component of the colloid, thyroglobulin, a thyroid-specific protein known as a target autoantigen in autoimmune thyroid diseases. As a confirmation of these sequences, all seven peptides were synthesized and their MS spectra compared with those obtained from the thyroid samples (see Fig. 2). Albumin peptides were also found: albumin is ubiquitous but is also the most abundant protein in the thyroid colloid after the thyroglobulin (21). Of the remaining endocytic peptides, 14 derived from extracellular matrix proteins, 18 from membrane proteins, and 46 from soluble serum proteins. Many of these peptides derived from hemoglobin, known also to be abundant in the colloid in hyperplasic thyroids. Additionally, there were two peptides from vesicular resident proteins, both derived from cathepsins, very abundant proteases in the TFC endocytic vesicles. Peptides putatively generated by degradation in the cytosol were: 10 nuclear, 2 from the endoplasmic reticulum, 1 from mitochondria, and 33 cytosolic peptides. Only one sequence was not identified in the database.

FIGURE 2.

Fragmentation spectra comparison of five natural and synthetic thyroglobulin peptides. Spectra corresponding to (a) natural peptides and (b) synthetic peptides. All spectra were obtained by mass spectrometry with an ion trap coupled to a nanoelectrospray source (nESI-ITMS/MS).

FIGURE 2.

Fragmentation spectra comparison of five natural and synthetic thyroglobulin peptides. Spectra corresponding to (a) natural peptides and (b) synthetic peptides. All spectra were obtained by mass spectrometry with an ion trap coupled to a nanoelectrospray source (nESI-ITMS/MS).

Close modal

Analyzing the peptides sequenced from TB448, TB449, and TB471 eluates, the patterns were very similar (Table II). Membrane peptides derived mostly from HLA-II proteins, 2 in TB448, 9 of 12 in TB449, and 2 of 4 in TB471. Endocytic peptides derived from cathepsins, extracellular matrix proteins like heparan sulfate proteoglycan, nidogen-2, multimerin, or collagen; from serum proteins C3 or apolipoprotein B; from albumin, angiotensin, or α-2-macroglobulin; or from the C chain of Igγ-1. A large number of hemoglobin peptides were identified from TB471. In all samples, colloid peptides from albumin and thyroglobulin were sequenced. A large number of cytosolic peptides belonging to a heterogeneous set of proteins were also identified. Peptides derived from GAPDH, NHERF (Na+/H+ exchanger regulatory factor), a serine protease, several ribosomal proteins, the enzyme creatine kinase, 3-hydroxyacyl-CoA dehydrogenase type II, calgranulin B, NipSnap4, actin, the actin-binding protein cofilin-1 and α-actinin, an RE resident chaperone, BIP (HSP79), a mitochondrial enzyme (malate dehydrogenase), and nuclear proteins such as histones, the small nuclear ribonucleoprotein G, or the kinetochore protein CENP-F.

Thyroglobulin peptides Tg726–743 and Tg2098–2112 were sequenced from TB448 and TB449 samples, and Tg1080–1090, Tg2224–2241, Tg2508–2519, Tg2520–2532, Tg2520–2536, and Tg2756–2765 from TB471. Sequencing of Tg726–743 peptide was done manually because it contained a disulfide bridge between the 2 N-terminal Cys, making the fragments containing the disulfide bridge 2 mass units lower than the ones found in the databases.

Despite all three donors being heterozygous for HLA-DR, many peptides could be theoretically assigned to one of the alleles expressed by the donor. Interestingly, although two of the donors shared the DR2 haplotype, the two Tg peptides from their repertoire were apparently associated to different DR alleles: Tg726–743 to DR51 of the DR2 haplotype, and Tg2098–2112 to DR17 of the DR3 haplotype. Additionally, only one of the six Tg peptides from TB471 did show any theoretical binding capacity: Tg2224–2241 to DR3. None of these peptides was identical with the DR3-associated Tg peptide from TB448. The HLA binding assignments were based on the ProPred software analysis (www.imtech.res.in/raghava/propred/) (10), which contains a database of the allele-binding matrixes for 51 HLA-DR alleles and calculates theoretical affinities for each peptide to each of the alleles. To allow comparison between the different alleles, the theoretical affinity score was calculated as the percentage of the affinity index for each peptide-allele combination, in relation to the maximum theoretical affinity index for each allele. On the basis of this analysis, we considered potential “good binders” those peptides that had the correct P1 residue for the corresponding allele and showed a theoretical affinity score of >25%. Most of the peptides that could be assigned to any alleles showed good binder affinity scores, although 16 of them did not reach the minimum affinity score (marked as low binders (LB) in Table II). When one allele was not included in the ProPred database (such as DRB1*0407, DRB3, and DRB4 alleles) but the binding motif was available (http://www.syfpeithi.de/), we simply assigned the peptide on the basis of the presence of the correct P1 residue and at least one of the other anchor residues (15). However, in those cases we could not calculate a theoretical binding score. As a whole, 104 (37%) sequences could be assigned to any one allele, with some giving a relatively high affinity score and some considered “low binders”. These assignations are only theoretical and need to be confirmed by binding data.

Another observation was the large number of potentially promiscuous peptides that were identified. Again using the ProPred software, we ran all the sequences against all the 51 alleles of the database. The promiscuity ratio (Table II) for each peptide indicates how many of the 51 alleles analyzed could potentially bind the peptide (i.e., with an affinity score of >25%). Some peptides showed that in addition to the canonical allele expressed by the donor tissue, these peptides could theoretically bind more than two other alleles (Table II). An interesting feature of TB471 was the number of long peptides (i.e., 23–33 aa). This can be related to the second DR allele expressed by this donor, HLA-DR8 (DRB1*0801). A number of long epitopes have been described as ligands or epitopes associated to HLA-DR8, but no binding motif has been defined so far (http://www.syfpeithi.de). This may account for the difficulty to assign many TB471 peptides to any allele.

Therefore, the DR molecules expressed in GD thyroid glands bind to autologous peptides mostly of the standard size that are recruited from proteins abundant in the colloid and other tissue compartments, including internal cell proteins. Within this repertoire, tissue-specific proteins such as thyroglobulin can also generate peptides capable of associating to HLA-DR.

Once we had established that the HLA-DR peptide repertoire could be analyzed from whole tissue samples, we tried to apply the method to smaller samples from frozen blocks of thyroid tissue from our tissue collection to see whether we could confirm the patterns obtained in the previous samples. We selected four thyroid samples of different weights: TB270 (3.3 g), TB269 (2 g), TB237 (1.1 g), and TB190 (1 g), all expressing high HLA-DR levels. Despite the small sample size, 12 peptides could be sequenced by nESI (Table III). Four exogenous peptides were obtained from TB270, three from serum albumin, and one from the extracellular matrix protein fibronectin. The three albumin peptides formed a nested set, varying in one residue at the N- or C-terminal ends. Two nested albumin peptide sets of identical core sequence were identified from the DR-identical thyroid samples TB269 and TB190. These sequences were different from those found in TB270, of different HLA type. One sequence from the β-chain of the DQB1*0602 molecule was identified from the DQB1*0602+ TB237 sample (DR15 homozygous). This same peptide was obtained from DR15-positive sample TB448 but not from TB449, also DR15, and it had been previously described as one of the major DR15- and DR51-associated peptides in studies of DR repertoires from EBV-transformed cell lines (22). Thus, we demonstrated that bona fide peptides could be isolated from very small samples of frozen tissue using our protocol. Nevertheless, the yield obtained from these small samples was far too low.

Table III.

Sequences identified from small GD thyroid samplesa

SamplePhenotypeMr (Da)Size (aa)P P P P P P P P P 1 2 3 4 5 6 7 8 9Allele Assigned% Affinity IndexPromiscuitySource ProteinId No.L.
TB270 (3.31g) DRB1*0701 1601.8 15 A P S N L R F L A T T P N S L DRB1*0701 57.76 31/51 Fibronectin (1905–1919) P02751 EM 
  1815.8 15 L G E Y K F Q N A L L V R Y T DRB1*0701 39.66 17/51 Serum albumin (422–436) P02768 Co 
  1925.8 16 Q L G E Y K F Q N A L L V R Y T DRB1*0701 39.66 17/51 Serum albumin (421–436) P02768 Co 
  2053.6 17 Q L G E Y K F Q N A L L V R Y T K DRB1*0701 39.66 17/51 Serum albumin (421–437) P02768 Co 
TB269 (2g) DRB1*1101 1569.9 15 T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (444–458) P02768 Co 
 DRB1*1104 1656.9 16 S T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (443–458) P02768 Co 
 DRB3*0202 1656.9 16 T P T L V E V S R N L G K V G S DRB3 ND 0/51 Serum albumin (444–459) P02768 Co 
TB237 (1,1g) DRB1*1501 1629.8 15 D V G V Y R A V T O Q G R P D DRB5*0101 68.37 7/51 DQB1*0602 β-chain (75–89) P03992 
TB190 (1g) DRB1*1101 1569.9 15 T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (444–458) P02768 Co 
 DRB1*1104 1656.9 16 S T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (443–458) P02768 Co 
 DRB3*0202 1656.9 16 T P T L V E V S R N L G K V G S DRB3 ND 0/51 Serum albumin (444–459) P02768 Co 
  1755.9 17 V S T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (442–458) P02768 Co 
SamplePhenotypeMr (Da)Size (aa)P P P P P P P P P 1 2 3 4 5 6 7 8 9Allele Assigned% Affinity IndexPromiscuitySource ProteinId No.L.
TB270 (3.31g) DRB1*0701 1601.8 15 A P S N L R F L A T T P N S L DRB1*0701 57.76 31/51 Fibronectin (1905–1919) P02751 EM 
  1815.8 15 L G E Y K F Q N A L L V R Y T DRB1*0701 39.66 17/51 Serum albumin (422–436) P02768 Co 
  1925.8 16 Q L G E Y K F Q N A L L V R Y T DRB1*0701 39.66 17/51 Serum albumin (421–436) P02768 Co 
  2053.6 17 Q L G E Y K F Q N A L L V R Y T K DRB1*0701 39.66 17/51 Serum albumin (421–437) P02768 Co 
TB269 (2g) DRB1*1101 1569.9 15 T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (444–458) P02768 Co 
 DRB1*1104 1656.9 16 S T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (443–458) P02768 Co 
 DRB3*0202 1656.9 16 T P T L V E V S R N L G K V G S DRB3 ND 0/51 Serum albumin (444–459) P02768 Co 
TB237 (1,1g) DRB1*1501 1629.8 15 D V G V Y R A V T O Q G R P D DRB5*0101 68.37 7/51 DQB1*0602 β-chain (75–89) P03992 
TB190 (1g) DRB1*1101 1569.9 15 T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (444–458) P02768 Co 
 DRB1*1104 1656.9 16 S T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (443–458) P02768 Co 
 DRB3*0202 1656.9 16 T P T L V E V S R N L G K V G S DRB3 ND 0/51 Serum albumin (444–459) P02768 Co 
  1755.9 17 V S T P T L V E V S R N L G K V G DRB3 ND 0/51 Serum albumin (442–458) P02768 Co 
a

Table heading as in Table II.

To verify the capacity of the thyroid DR-eluted thyroglobulin peptides to bind DR molecules, binding assays were performed by direct detection of HLA-DR-peptide complexes in a cell-free assay using soluble DR15 and DR3 molecules. DR8 and DR51 soluble molecules were not available for these assays. In addition to the thyroglobulin peptides, we also tested four more thyroid peptides that had been well assigned to one or other of the alleles involved, that is, VE-cadherin 196–210, hemoglobin δ-chain 31–41, DQB1*0602 β-chain 75–89, and collagen α1-chain (XV) 1243–1257, assigned to DR3, DR15, DR51, and DR15, respectively. The data confirmed the assignations to DR15 and DR3 of the above peptides and demonstrated the positive binding of Tg2098–2112 to DR3. Tg2224–2240, considered a low binder to DR3, showed no binding (Fig. 3).

FIGURE 3.

Peptide binding assay. List of peptides, source protein, and binding data against alleles DR15 (DRB1*1501) and DR3 (DRB1*0301). The data are also shown as a bar graph. All thyroglobulin peptides isolated were tested (peptides 1–9), as were other peptides that had been theoretically assigned to one DR allele (peptides 10–13). Binding data are expressed as the percentage binding compared with a low-binder peptide (pass/fail, see Materials and Methods). Most of the thyroglobulin peptides could not be assigned to any allele, except for Tg726–743 to DR51 and Tg2098–2112 to DR3. Only the latter could be tested and confirmed as a medium binder to DR3. The alleles assigned to peptides 10, 11, and 15 were also confirmed. Assignation to DR51 could not be confirmed because this allele is not available for binding.

FIGURE 3.

Peptide binding assay. List of peptides, source protein, and binding data against alleles DR15 (DRB1*1501) and DR3 (DRB1*0301). The data are also shown as a bar graph. All thyroglobulin peptides isolated were tested (peptides 1–9), as were other peptides that had been theoretically assigned to one DR allele (peptides 10–13). Binding data are expressed as the percentage binding compared with a low-binder peptide (pass/fail, see Materials and Methods). Most of the thyroglobulin peptides could not be assigned to any allele, except for Tg726–743 to DR51 and Tg2098–2112 to DR3. Only the latter could be tested and confirmed as a medium binder to DR3. The alleles assigned to peptides 10, 11, and 15 were also confirmed. Assignation to DR51 could not be confirmed because this allele is not available for binding.

Close modal

In the absence of binding data for Tg726–743 to DR51, modeling studies were performed for its binding to HLA-DR51 compared with HLA-DR15. The calculations predicted a relatively higher binding affinity (∼10% higher) for the pair DR51/Tg726–743 than for DR15/Tg726–743, with an average relative binding score ΔZ51–15 of −25 kJ/mol (Tables IV and V). Although the computational method used is not meant to be quantitative, this result is in remarkable agreement with the trend shown by the binding data (Fig. 3).

Table IV.

Summary of modeling results

DR51/Tg726–743DR15/Tg726–743Δ51-15h
min(W)a (kJ/mol) −300 −278 −22 
min(ΔGclus)b (kJ/mol) −297 −268 −29 
ΔGc (kJ/mol) −282 −261 −21 
Zc (kJ/mol) −293 −269 −24 
VDRA/Tg726–743e (kJ/mol) −392 −456 64 
VDRB/Tg726–743f (kJ/mol) −405 −410 
Structural familiesg 388 140 248 
DR51/Tg726–743DR15/Tg726–743Δ51-15h
min(W)a (kJ/mol) −300 −278 −22 
min(ΔGclus)b (kJ/mol) −297 −268 −29 
ΔGc (kJ/mol) −282 −261 −21 
Zc (kJ/mol) −293 −269 −24 
VDRA/Tg726–743e (kJ/mol) −392 −456 64 
VDRB/Tg726–743f (kJ/mol) −405 −410 
Structural familiesg 388 140 248 
a

Lowest work value obtained within the 1000 simulations (corresponds to model structure in Fig. 4).

b

Lowest free-energy value associated to a structural cluster. In both cases (DR51/Tg726–743 and DR15/Tg726–743), the model structure with the lowest W is a member of the cluster with the lowest ΔGclus.

c

Free energy calculated over the entire set of 1000 work values.

d

Average of the previous three quantities: average binding score.

e

Interaction potential energy between chain DRA and Tg726–743 in the model structure with the lowest W (from the ensemble of 1000 model structures generated for each of the two complexes; see Fig. 4).

f

Interaction potential energy between chain DRB and Tg726–743 in the model structure with the lowest W.

g

Number of structural families obtained by clustering of the 1000 model complexes based on the structure of the Tg726–743 buried side chains (Leu732, Glu735, Ala737, Phe738, and Arg740).

h

Difference between the first two columns.

Table V.

DR/Tg726–743 side-chain contactsa in model structures with lowest Wb

Tg726–743DR51-DRA1DR15-DRA1DR51-DRB5DR15-DRB1
Leu732 Phe49 Phe49  His110 
 Phe57 Phe57 Asn111 Asn111 
 Trp68 Trp68 Val114 Val114 
 Ala77 Ala77  Val115 
 Phe79 Phe79 Phe118 Phe118 
Glu735 Gln34 Gln34 Tyr42 Arg42 
   Phe55 Phe55 
    Gln99 
   Arg100  
   Ala103 Ala103 
   Tyr107 Tyr107 
Ala737 Glu36 Glu36 Asp40  
 Asn87 Asn87 Tyr42 Arg42 
 Val90 Val90  Tyr59 
 Asp91 Asp91   
  Asn94   
Phe738    Tyr59 
    Phe76 
   Trp90 Trp90 
   Phe96 Ile96 
    Gln99 
   Arg100  
Arg740  Asn94  Trp38 
  Ile97  Tyr59 
  Met98  Ser66 
 Arg101 Arg101  Val67 
   Pro85  
   Asp86 Asp86 
   Tyr89  
    Trp90 
Tg726–743DR51-DRA1DR15-DRA1DR51-DRB5DR15-DRB1
Leu732 Phe49 Phe49  His110 
 Phe57 Phe57 Asn111 Asn111 
 Trp68 Trp68 Val114 Val114 
 Ala77 Ala77  Val115 
 Phe79 Phe79 Phe118 Phe118 
Glu735 Gln34 Gln34 Tyr42 Arg42 
   Phe55 Phe55 
    Gln99 
   Arg100  
   Ala103 Ala103 
   Tyr107 Tyr107 
Ala737 Glu36 Glu36 Asp40  
 Asn87 Asn87 Tyr42 Arg42 
 Val90 Val90  Tyr59 
 Asp91 Asp91   
  Asn94   
Phe738    Tyr59 
    Phe76 
   Trp90 Trp90 
   Phe96 Ile96 
    Gln99 
   Arg100  
Arg740  Asn94  Trp38 
  Ile97  Tyr59 
  Met98  Ser66 
 Arg101 Arg101  Val67 
   Pro85  
   Asp86 Asp86 
   Tyr89  
    Trp90 
a

Two side chains are considered in contact when their distance is ≤6 Å.

b

See Table IV.

The potential energy of DR/Tg726–743 interactions in the model structure with the best binding score (VDR/Tg726–743, Table IV) is more favorable (more negative) for the DR15/Tg726–743 complex than for DR51/Tg726–743. The difference, which might appear surprising at first sight, is mainly associated with the interaction between Tg726–743 and the DR α-chain. A close examination of the environment of the five Tg726–743 residues expected to anchor the peptide to the P1–P5 DR pockets (Table V and Fig. 4) shows that Tg726–743 establishes more interactions in the binding groove of DR15 than in that of DR51. In particular, as a consequence of the reduced size of P5 in DR51, Arg740 does not enter into this pocket but stays on the surface. In fact, His639 of EBVP also stays out of P5 in the crystallographic structure 1H15 (PDB code). If the energetics of the interaction between Tg726–743 and the DR molecules favor complexation with DR15, why do the binding scores indicate the opposite binding preferences? The answer must be found in the entropy. Table IV shows that the structural variability of these five residues among the model complex structures generated, as reflected by the number of structural families obtained, is larger for DR51/Tg726–743 than for DR15/Tg726–743. Binding is the result of a fine balance between enthalpy and entropy. Optimal complementarity between protein and peptide may lead, despite a favorable enthalpy component, to poor binding due to a large (nonfavorable) entropy loss. A good binder is one that maximizes binding enthalpy (interactions) while minimizing the unavoidable entropy loss upon binding. These considerations seem especially relevant for DR molecules that, because of the nature of their function, must escape highly specific interactions.

FIGURE 4.

Model structure of Tg726–743 peptide binding to DR15 and DR51. Graphical representation of the model structure with best binding score (min(W) in Table I) for DR51/Tg725–743 (a) and for DR15/Tg725–743 (b). Peptide color code (stick representation): yellow (C), red (O), blue (N), green (S), white (H). Color code for DR51 and DR15 (space-filling representation): cyan (C), red (O), blue (N), white (H).

FIGURE 4.

Model structure of Tg726–743 peptide binding to DR15 and DR51. Graphical representation of the model structure with best binding score (min(W) in Table I) for DR51/Tg725–743 (a) and for DR15/Tg725–743 (b). Peptide color code (stick representation): yellow (C), red (O), blue (N), green (S), white (H). Color code for DR51 and DR15 (space-filling representation): cyan (C), red (O), blue (N), white (H).

Close modal

The results of this study provide a first glance to the repertoire of peptides associated with HLA-DR molecules in thyroid glands of patients with autoimmune thyroid disease. To our knowledge, this is the first such analysis from human autoimmune tissue. The identified sequences constitute only a fraction of the total pool of peptides associated with the different DR alleles. They are certainly the most abundant peptides within our system’s capability, and therefore peptides poorly represented in the tissue or those with low efficiency of ionization will not have been detected. Despite the limitations, we were able to identify seven different peptides belonging to one of the major autoantigens of thyroid autoimmune disease, thyroglobulin.

Most studies characterizing HLA-II-associated peptides have been conducted using immortalized lymphoid cell lines. When analyzing peptides from cell lines in culture, not only is the cell content different from the original tissue, but also the extracellular context is artificial and does not mimic the environment surrounding the cells in vivo. However, the use of human tissue is difficult due to low availability, low HLA-II expression, and heterozygosity of most samples. In the first study of HLA-II peptides from human tissue, 17 HLA-DR ligands where identified from a rheumatoid arthritis patient’s spleen sample (12). Oshitani et al. reported 55 HLA-DR peptides from human intestine samples of 4 controls and 18 patients affected by inflammatory bowel disease (13). More recently, 78 peptides have been reported from a pool of 16 bronchoalveolar lavage samples from patients with sarcoidosis, being the first such study in human autoimmunity, although the nature of the samples was very different (23). Using a different approach, a recent study has shown that most peptides bound to MHC-II from mouse insulinoma cells derive from proteins normally expressed in β cells from normal islets and that some of them are recognized by CD4+ T cells from NOD mice (24). In the present report, 162 peptides have been obtained from seven Graves’ disease-affected thyroid tissue samples. We were able to sequence peptides from as little as 1 g of human thyroid tissue, despite the low expression of HLA class II in the thyroid and the heterogeneity of this expression.

Our data do not allow discrimination of whether these peptides were associated to DR molecules expressed by TFC, by B cells, dendritic cells, or macrophages, all present in the thyroid gland. In fact, B cells are abundant in GD thyroid infiltrates, where they form autoantigen-specific germinal centers (6, 25). We have indeed identified one Ig peptide that could have been generated from an internalized BCR. However, there is not much tissue destruction in GD thyroids; on the contrary, TFCs are hyperfunctional by their interaction with agonistic anti-TSH-R Abs and therefore produce thyroid hormones, requiring continuous endocytosis of thyroglobulin and other colloid components by the TFCs. In this context, we can hypothesize that by being most accessible to thyroid follicular cells, the thyroglobulin peptides were associated to TFC-expressed DR molecules. Peptide identification from a single cell population in the tissue is at the moment very difficult because the amount of material that can be obtained using the best separation techniques is still very small, and thus it is hard to be sure which APCs would be presenting most peptides identified. Another limitation of human tissue, sample HLA heterozygosity, was mostly overcome because most sequences could be statistically assigned to one or the other DR allele expressed by each sample. The large available database of peptides specific to most HLA-DR alleles facilitated the search. The only limiting alleles were DR407 (DRB1*0407) and DR8, expressed by TB448 and TB471, respectively, for which there are still no well-defined peptide binding motives.

One common protein that generated peptides was albumin, an ubiquitous protein that, after the thyroglobulin, is the most abundant in the thyroid gland and, like the thyroglobulin, is present in the colloid. Another protein from which many peptides were derived was hemoglobin, also found in the colloid in hyperplasic thyroid tissue (21). Many exogenous peptides sequenced belonged to proteins of the extracellular matrix. Autoantibodies reacting with extracellular matrix proteins have been extensively studied in various autoimmune diseases affecting connective tissue. Both T cells and Abs specific for matrix proteins including collagen, laminin, and fibronectin have been detected in patients with Graves’-associated ophthalmopathy (26), diagnosed to patient TB449. Extracellular matrix proteins and proteinases involved in their degradation are related to thyroid function, and many are dependent of thyroid hormone production (27, 28, 29). Therefore, the chronic stimulation of thyroid hormone production characteristic of Graves’ disease may well be related to the presence of extracellular matrix peptides associated with HLA-DR in the thyroid. Equally, the activation of thyroid hormone production requires active endocytosis of the colloid content by thyroid follicular cells, leading to a higher availability of Tg and even albumin peptides to bind to DR molecules expressed by the TFCs.

The most interesting peptides isolated from the Graves’ thyroid glands were the Tg peptides, Tg726–743 from TB448 and Tg2098–2112 from TB449, that preferentially associated to DR51 and DR3, respectively, and five other from TB471, one of which (Tg2224–2240) appeared to be a DR3 low binder, but could not be confirmed by binding. Thyroglobulin, one of the main autoantigens in autoimmune thyroiditis, constitutes 75–80% of the total protein contents of the colloid (21). Although thyroglobulin is highly expressed in the thymus during maturation (30), and therefore a high degree of tolerance to this molecule is expected in the periphery, the identification of thyroglobulin peptides as natural ligands in the repertoire indicates that these DR-peptide complexes must be relatively abundant in the affected thyroids. This putative high ligand density in the inflamed tissue may be a reason for otherwise ignorant T cells to become stimulated. The analysis of T cell responses to these peptides both from in situ and peripheral T cells will be a next step to define which functional population of T cells may respond. Much evidence has demonstrated the role of Tg in the etiology of AITD: anti-Tg Abs are detected in almost all patients with AITD, and experimental autoimmune thyroiditis (EAT) can be induced in susceptible animals by both mouse and human Tg immunization (31). EAT can also be induced by direct immunization with thyroiditogenic peptides, including some human Tg peptides (32, 33). Although the pathological relevance of the Tg peptides that we have identified is not known, data from another group have demonstrated the pathogenicity of Tg2098–2112 in a DR3 transgenic model of Tg-induced EAT (34). A computer-based predicted set of 39 synthetic DR3-binding Tg peptides was screened against T cells from EAT+ DR3-transgenic mice, and only 4 could stimulate T cell responses. Interestingly, only one of those, peptide LSSVVVDPSIRHFDV, identical with the DR3-associated Tg2098–2112 peptide from TB449, could induce the disease both by direct peptide immunization and by adoptive transfer of peptide-activated T cells. Our data therefore demonstrate that a potentially pathogenic Tg peptide is indeed generated in vivo and presented by DR3 molecules in human Graves’ disease.

Thus, we can now confirm what the authors suggested, that is, that the peptide was likely a natural peptide generated in DR3 individuals (34), demonstrating that a computer algorithm can predict a peptide identical in size and sequence to a naturally processed peptide. It remains to be clarified whether DR51-Tg726–743 or any other DR-Tg complex expressed in human GD thyroids are also potentially pathogenic or related to the regulation of the autoimmune response. Thyroglobulin is a very large autoantigen of 2749 aa that is extensively modified by iodination and other posttranslational events. Iodinated Tg peptides are highly immunogenic and can trigger thyroid autoreactive T cells (35). Iodine has been proposed as capable of modifying the processing of Tg by APCs, resulting in the generation of pathogenic epitopes (32). Indeed, two of the TB471 peptides, Tg2508–2519 and Tg2520–2532, are contained in a sequence that has been shown to be a cryptic peptide in mouse experimental thyroiditis, being only generated when mice are immunized with highly iodinated thyroglobulin (36).

When we compared our sequences with DR3- or DR51-predicted Tg epitopes, we found both the DR3 peptide Tg2098–2112 and the DR51 peptide Tg726–743 were within the highest scored peptides by ProPred (www.imtech.res.in/raghava/propred/), RANKPEP (www.mifoundation.org/Tools/rankpep.html), or SYFPEITHI (www.syfpeithi.de/) (only DR3) (data not shown). Computer predictions gave us the assignation of these peptides to their respective alleles, and computer modeling supported these results.

Most other natural peptides identified also derived from highly expressed and ubiquitous proteins. These are by and large the proteins that one would expect to generate peptides capable of associating to class II molecules in any tissue. Some of these were analyzed for binding to their assigned alleles DR3 or DR15 and some demonstrated to be really high binders, such as VE-cadherin 196–210 to DR3 and hemoglobin δ-chain 31–41 or collagen α-chain 1243–1257 to DR15. One interesting feature of many peptides identified was their theoretical “promiscuity”, that is, their capability to bind to many different HLA-DR alleles, which may be related to autoreactive TCR degeneracy. In fact, DR association with thyroid diseases is not very high, despite the preference for DR3 (37, 38, 39, 40). The main associated alleles are DR3 (DRB1*0301) and DR5 (DRB1*1101). Tg726–743 was presumably associated with DR51, a DR molecule expressed within the DR2 (DR15) haplotype. Although this haplotype has not been associated with AITD, it is not negatively associated and we have found 16% DR15+ Graves’ patients in a larger study (41). DR51 is the dominant restriction element for myelin basic protein peptide recognition by multiple sclerosis patients’ T cells (42) and it has more stringent requirements for peptide binding than does DR15 (22, 43). These requirements include a negatively charged Arg or Lys residue at position P9 (±2), which was confirmed in all but two peptides assigned to DR51. The modeling studies suggest that Tg726–743 binds more strongly to DR51 than to DR15 as a consequence of the major flexibility of the DR51/Tg726–743 interface (entropic effect), despite the more favorable network of interactions between Tg726–743 and DR15. The low binding of this peptide to DR15 favors this interpretation.

Our data thus demonstrate that tissue-specific peptides are associated in vivo to HLA-DR in GD-affected tissue and therefore may well be involved in the autoimmune process by potentiating the autoimmune responses, but also that the in situ HLA-DR-associated peptide pool mostly derive from abundant and ubiquitous proteins common to other tissues. The relevance of all these peptides may not only relate to their pathogenicity but also to their capacity to regulate responses. Ex vivo studies can well be a first step toward the clarification of the role of HLA class II molecules expressed in autoimmune organs.

We thank the patients who donated thyroid tissue and Dr. A. Lucas-Martín (Endocrinology Department, Hospital Germans Trias i Pujol, Badalona) and Dr. G. Obiols (Endocrinology Department, Hospital Vall d’Hebron, Barcelona) for clinical data. Thanks also to Dr. E. Palou (Laboratory of Immunobiology for Research and Application to Diagnosis, Blood and Tissue Bank, Hospital Germans Trias i Pujol) for HLA typing and Dr. E. Borràs (Universitat Pompeu Fabra, Barcelona), for help with peptide synthesis. Special thanks to the Port d’Informació Científica and the Òliba- Universitat Autònoma de Barcelona project for providing data storage and computational resources.

The authors have no financial conflicts of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was funded by the Spanish Ministry of Education Grants SAF2003-08843-C02-01 and SAF00-0131-C02-01. X.D. was supported by the Spanish MEC/FEDER Grant BIO2003-02848. L.M. is a Formación de Personal Investigador fellow of the Spanish Ministry of Education. I.A. was partially funded by the Eurothymaid Integrated European Project LSHB-CT-2003-503410. C.P. was funded by the Multiple Sclerosis National Research Institute.

3

Abbreviations used in this paper: HLA-II, HLA class II; AITD, autoimmune thyroid diseases; EAT, experimental autoimmune thyroiditis; EBVP, EBV DNA polymerase; GD, Graves’ disease; HMBP, human myelin basic protein; ITMS, ion trap mass spectroscopy; MHC-II, MHC class II; MS, mass spectroscopy; m/z, mass-to-charge; nESI, nanoelectrospray; rpHPLC, reversed-phase HPLC; TFC, thyroid follicular cells; Tg, thyroglobulin; vMALDI, vacuum MALDI.

1
Hanafusa, T., R. Pujol-Borrell, L. Chiovato, R. C. Russell, D. Doniach, G. F. Bottazzo.
1983
. Aberrant expression of HLA-DR antigen on thyrocytes in Graves’ disease: relevance for autoimmunity.
Lancet
2
:
1111
-1115.
2
Guardiola, J., A. Maffei.
1993
. Control of MHC class II gene expression in autoimmune, infectious, and neoplastic diseases.
Crit. Rev. Immunol.
13
:
247
-268.
3
Bottazzo, G. F., R. Pujol-Borrell, T. Hanafusa, M. Feldmann.
1983
. Role of aberrant HLA-DR expression and antigen presentation in induction of endocrine autoimmunity.
Lancet
2
:
1115
-1119.
4
Weetman, A. P..
2003
. Autoimmune thyroid disease: propagation and progression.
Eur. J. Endocrinol.
148
:
1
-9.
5
Lucas-Martin, A., M. Foz-Sala, I. Todd, G. F. Bottazzo, R. Pujol-Borrell.
1988
. Occurrence of thyrocyte HLA class II expression in a wide variety of thyroid diseases: relationship with lymphocytic infiltration and thyroid autoantibodies.
J. Clin. Endocrinol. Metab.
66
:
367
-375.
6
Aloisi, F., R. Pujol-Borrell.
2006
. Lymphoid neogenesis in chronic inflammatory diseases.
Nat. Rev. Immunol.
6
:
205
-217.
7
Catalfamo, M., L. Serradell, C. Roura-Mir, E. Kolkowski, M. Sospedra, M. Vives-Pi, F. Vargas-Nieto, R. Pujol-Borrell, D. Jaraquemada.
1999
. HLA-DM and invariant chain are expressed by thyroid follicular cells, enabling the expression of compact DR molecules.
Int. Immunol.
11
:
269
-277.
8
Chicz, R. M., R. G. Urban, J. C. Gorga, D. A. Vignali, W. S. Lane, J. L. Strominger.
1993
. Specificity and promiscuity among naturally processed peptides bound to HLA-DR alleles.
J. Exp. Med.
178
:
27
-47.
9
Rammensee, H. G., T. Friede, S. Stevanoviic.
1995
. MHC ligands and peptide motifs: first listing.
Immunogenetics
41
:
178
-228.
10
Singh, H., G. P. Raghava.
2001
. ProPred: prediction of HLA-DR binding sites.
Bioinformatics
17
:
1236
-1237.
11
Muntasell, A., M. Carrascal, L. Serradell, P. van Veelen, F. Verreck, F. Koning, G. Raposo, J. Abian, D. Jaraquemada.
2002
. HLA-DR4 molecules in neuroendocrine epithelial cells associate to a heterogeneous repertoire of cytoplasmic and surface self peptides.
J. Immunol.
169
:
5052
-5060.
12
Gordon, R. D., J. A. Young, S. Rayner, R. W. Luke, M. L. Crowther, P. Wordsworth, J. Bell, G. Hassall, J. Evans, S. A. Hinchliffe, et al
1995
. Purification and characterization of endogenous peptides extracted from HLA-DR isolated from the spleen of a patient with rheumatoid arthritis.
Eur. J. Immunol.
25
:
1473
-1476.
13
Oshitani, N., F. Hato, S. Kitagawa, K. Maeda, K. Higuchi, T. Matsumoto, T. Arakawa.
2003
. Analysis of intestinal HLA-DR bound peptides and dysregulated immune responses to enteric flora in the pathogenesis of inflammatory bowel disease.
Int. J. Mol. Med.
11
:
99
-104.
14
Dengjel, J., M. D. Nastke, C. Gouttefangeas, G. Gitsioudis, O. Schoor, F. Altenberend, M. Muller, B. Kramer, A. Missiou, M. Sauter, et al
2006
. Unexpected abundance of HLA class II presented peptides in primary renal cell carcinomas.
Clin. Cancer Res.
12
:
4163
-4170.
15
Muntasell, A., M. Carrascal, I. Alvarez, L. Serradell, P. van Veelen, F. A. Verreck, F. Koning, J. Abian, D. Jaraquemada.
2004
. Dissection of the HLA-DR4 peptide repertoire in endocrine epithelial cells: strong influence of invariant chain and HLA-DM expression on the nature of ligands.
J. Immunol.
173
:
1085
-1093.
16
van Gunsteren, W. F., X. Daura, A. E. Mark.
2002
. Computation of free energy.
Helv. Chim. Acta
85
:
3113
-3129.
17
Nelson, C. A., D. H. Fremont.
1999
. Structural principles of MHC class II antigen presentation.
Rev. Immunogenet.
1
:
47
-59.
18
Scott, W. R. P., P. H. Hunenberger, I. G. Tironi, A. E. Mark, S. R. Billeter, J. Fennen, A. E. Torda, T. Huber, P. Kruger, W. F. van Gunsteren.
1999
. The GROMOS biomolecular simulation program package.
J. Phys. Chem. A
103
:
3596
-3607.
19
Schuler, L. D., X. Daura, W. F. Van Gunsteren.
2001
. An improved GROMOS96 force field for aliphatic hydrocarbons in the condensed phase.
J. Comput. Chem.
22
:
1205
-1218.
20
Jarzynski, C..
1997
. Nonequilibrium equality for free energy differences.
Phys. Rev. Lett.
78
:
2690
-2693.
21
Baggio, M. C., G. Medeiros-Neto, Y. Osawa, N. Y. Nguyen, P. Santisteban, M. Knobel, E. E. Groliman.
1996
. Amino acid composition of proteins extracted from endemic goiter glands.
Endocr. Pathol.
7
:
137
-143.
22
Vogt, A. B., H. Kropshofer, H. Kalbacher, M. Kalbus, H. G. Rammensee, J. E. Coligan, R. Martin.
1994
. Ligand motifs of HLA-DRB5*0101 and DRB1*1501 molecules delineated from self-peptides.
J. Immunol.
153
:
1665
-1673.
23
Wahlstrom, J., J. Dengjel, B. Persson, H. Duyar, H. G. Rammensee, S. Stevanovic, A. Eklund, R. Weissert, J. Grunewald.
2007
. Identification of HLA-DR-bound peptides presented by human bronchoalveolar lavage cells in sarcoidosis.
J. Clin. Invest.
117
:
3576
-3582.
24
Suri, A., J. J. Walters, H. W. Rohrs, M. L. Gross, E. R. Unanue.
2008
. First signature of islet β-cell-derived naturally processed peptides selected by diabetogenic class II MHC molecules.
J. Immunol.
180
:
3849
-3856.
25
Armengol, M. P., M. Juan, A. Lucas-Martin, M. T. Fernandez-Figueras, D. Jaraquemada, T. Gallart, R. Pujol-Borrell.
2001
. Thyroid autoimmune disease: demonstration of thyroid antigen-specific B cells and recombination-activating gene expression in chemokine-containing active intrathyroidal germinal centers.
Am. J. Pathol.
159
:
861
-873.
26
Bednarczuk, T., J. Kiljanski, T. Mrowiec, M. Slon, E. Ing, C. Stolarski, J. S. Kennerdell, A. Gorski, J. Nauman, J. R. Wall.
1998
. T cell interactions with extracellular matrix proteins in patients with thyroid-associated ophthalmopathy.
Autoimmunity
27
:
221
-230.
27
Ribeiro-Carvalho, M. M., D. A. Farias-de-Oliveira, D. M. Villa-Verde, W. Savino.
2002
. Triiodothyronine modulates extracellular matrix-mediated interactions between thymocytes and thymic microenvironmental cells.
Neuroimmunomodulation
10
:
142
-152.
28
Trentin, A. G., C. B. De Aguiar, R. C. Garcez, M. Alvarez-Silva.
2003
. Thyroid hormone modulates the extracellular matrix organization and expression in cerebellar astrocyte: effects on astrocyte adhesion.
Glia
42
:
359
-369.
29
Damjanovski, S., M. Puzianowska-Kuznicka, A. Ishuzuya-Oka, Y. B. Shi.
2000
. Differential regulation of three thyroid hormone-responsive matrix metalloproteinase genes implicates distinct functions during frog embryogenesis.
FASEB J.
14
:
503
-510.
30
Sospedra, M., X. Ferrer-Francesch, O. Dominguez, M. Juan, M. Foz-Sala, R. Pujol-Borrell.
1998
. Transcription of a broad range of self-antigens in human thymus suggests a role for central mechanisms in tolerance toward peripheral antigens.
J. Immunol.
161
:
5918
-5929.
31
Esquivel, P. S., N. R. Rose, Y. C. Kong.
1977
. Induction of autoimmunity in good and poor responder mice with mouse thyroglobulin and lipopolysaccharide.
J. Exp. Med.
145
:
1250
-1263.
32
Carayanniotis, G..
2003
. The cryptic self in thyroid autoimmunity: the paradigm of thyroglobulin.
Autoimmunity
36
:
423
-428.
33
Balasa, B., G. Carayanniotis.
1993
. Induction of experimental autoimmune thyroiditis in rats with the synthetic peptide (2495–2511) of thyroglobulin.
Cell. Immunol.
148
:
259
-268.
34
Flynn, J. C., D. J. McCormick, V. Brusic, Q. Wan, J. C. Panos, A. A. Giraldo, C. S. David, Y. C. Kong.
2004
. Pathogenic human thyroglobulin peptides in HLA-DR3 transgenic mouse model of autoimmune thyroiditis.
Cell. Immunol.
229
:
79
-85.
35
Champion, B. R., K. R. Page, N. Parish, D. C. Rayner, K. Dawe, G. Biswas-Hughes, A. Cooke, M. Geysen, I. M. Roitt.
1991
. Identification of a thyroxine-containing self-epitope of thyroglobulin which triggers thyroid autoreactive T cells.
J. Exp. Med.
174
:
363
-370.
36
Dai, Y. D., V. P. Rao, G. Carayanniotis.
2002
. Enhanced iodination of thyroglobulin facilitates processing and presentation of a cryptic pathogenic peptide.
J. Immunol.
168
:
5907
-5911.
37
Farid, N. R., L. Sampson, E. P. Noel, J. M. Barnard, R. Mandeville, B. Larsen, W. H. Marshall, N. D. Carter.
1979
. A study of human leukocyte D locus related antigens in Graves’ disease.
J. Clin. Invest.
63
:
108
-113.
38
Schleusener, H., G. Schernthaner, W. R. Mayr, P. Kotulla, U. Bogner, R. Finke, H. Meinhold, K. Koppenhagen, K. W. Wenzel.
1983
. HLA-DR3 and HLA-DR5 associated thyrotoxicosis: two different types of toxic diffuse goiter.
J. Clin. Endocrinol. Metab.
56
:
781
-785.
39
Tomer, Y., T. F. Davies.
2003
. Searching for the autoimmune thyroid disease susceptibility genes: from gene mapping to gene function.
Endocr. Rev.
24
:
694
-717.
40
Wan, Q., R. Shah, J. C. Panos, A. A. Giraldo, C. S. David, Y. M. Kong.
2002
. HLA-DR and HLA-DQ polymorphism in human thyroglobulin-induced autoimmune thyroiditis: DR3 and DQ8 transgenic mice are susceptible.
Hum. Immunol.
63
:
301
-310.
41
Sospedra, M., G. Obiols, L. F. Babi, E. Tolosa, F. Vargas, C. Roura-Mir, A. Lucas-Martin, G. Ercilla, R. Pujol-Borrell.
1995
. Hyperinducibility of HLA class II expression of thyroid follicular cells from Graves’ disease: a primary defect?.
J. Immunol.
154
:
4213
-4222.
42
Jaraquemada, D., R. Martin, S. Rosen-Bronson, M. Flerlage, H. F. McFarland, E. O. Long.
1990
. HLA-DR2a is the dominant restriction molecule for the cytotoxic T cell response to myelin basic protein in DR2Dw2 individuals.
J. Immunol.
145
:
2880
-2885.
43
Jones, E. Y., L. Fugger, J. L. Strominger, C. Siebold.
2006
. MHC class II proteins and disease: a structural perspective.
Nat. Rev. Immunol.
6
:
271
-282.