Skip to main content
NCBI home page
Clinical and Experimental Immunology logoLink to Clinical and Experimental Immunology
. 2015 Apr 14;180(2):207–217. doi: 10.1111/cei.12572

Characterization of the autoimmune response against the nerve tissue S100β in patients with type 1 diabetes

I Gómez-Touriño *,1, R Simón-Vázquez †, J Alonso-Lorenzo ‡,2, S Arif §, C Calviño-Sampedro *, Á González-Fernández †, E Pena-González , J Rodríguez **, J Viñuela-Roldán ††, J Verdaguer ‡‡, O J Cordero *, M Peakman §, R Varela-Calvino *
PMCID: PMC4408155  PMID: 25516468

Abstract

Type 1 diabetes results from destruction of insulin-producing beta cells in pancreatic islets and is characterized by islet cell autoimmunity. Autoreactivity against non-beta cell-specific antigens has also been reported, including targeting of the calcium-binding protein S100β. In preclinical models, reactivity of this type is a key component of the early development of insulitis. To examine the nature of this response in type 1 diabetes, we identified naturally processed and presented peptide epitopes derived from S100β, determined their affinity for the human leucocyte antigen (HLA)-DRB1*04:01 molecule and studied T cell responses in patients, together with healthy donors. We found that S100β reactivity, characterized by interferon (IFN)-γ secretion, is a characteristic of type 1 diabetes of varying duration. Our results confirm S100β as a target of the cellular autoimmune response in type 1 diabetes with the identification of new peptide epitopes targeted during the development of the disease, and support the preclinical findings that autoreactivity against non-beta cell-specific autoantigens may have a role in type 1 diabetes pathogenesis.

Keywords: autoimmunity, peptide epitopes, S100β, Th1, type 1 diabetes

Introduction

Type 1 diabetes is an autoimmune disease resulting from the destruction of the insulin-producing β cells of the pancreatic islets. It is well established in preclinical models such as the non-obese diabetic (NOD) mouse that T lymphocytes, both CD4 and CD8, participate in a highly orchestrated process of infiltration (insulitis) of the pancreatic islet tissue, which is associated with β cell loss. In the same models, numerous studies have demonstrated the importance of autoantigen specificity among the infiltrating T cells. Up to 18 different autoantigens including preproinsulin (pre-PI), glutamic acid decarboxylase (GAD65 and GAD67) or insulinoma-associated protein 2β (IA-2β) are targets of the T cell response 1. Similar types of reactivity are observed in human type 1 diabetes, both for T and B cells 2,3 in the blood, and more recently for CD8 T cells in the insulitis 4.

However, it remains unclear whether immune cells with these target β cell specificities are sufficient alone to mediate an effective insulitic process. Studies in prediabetic NOD mice have highlighted the potential importance of selective immune cell engagement with, and destruction of, the peri-islet Schwann cells (pSC) located on the outside of the islet mass 5. This is associated with the presence of lymphocytes specific for autoantigens expressed by pSCs, such as glial fibrillary acidic protein (GFAP), as well as self-targets shared with β cells such as the calcium-binding neurotrophic factor S100β 5. Following these findings, it has become clear that similar autoreactivity exists in patients with type 1 diabetes. Autoantibodies against these autoantigens have been detected 5–7, and T cell reactivity against both GFAP and S100β, when tested as whole antigens, is seen in up to 60% of patients compared to only 11% of non-diabetic subjects 8. In NOD, GFAP-derived CD8 and CD4 T cells restricted by the class I and II major histocompatibility complex (MHC) molecules Kd- and I-Ag7 have been identified, and it has been shown that these epitopes are targeted spontaneously at an early age, making it likely that they can participate in β cell destruction in this model 9. In humans, GFAP epitopes restricted by the MHC class I molecule human leucocyte antigen (HLA)-A*02:01 have been identified, epitopes targeted by both antibody-positive relatives and recently diagnosed type 1 diabetes patients 10 but, to our knowledge, T cell responses against S100β remain poorly characterized in type 1 diabetes.

S100β is a member of a large family of small, acidic proteins of 10–12 kDa containing a calcium-binding EF-hand motif 11. The members of this protein family are involved in the regulation of diverse cellular processes such as cell growth and differentiation, cell cycle progression, transcription and secretion. Altered levels of these proteins have been associated with several diseases of the central nervous system, heart, inflammatory conditions and cancer 11. In autoimmunity, adoptive transfer of S100β-specific lymphocytes into Lewis rats induces intense inflammation throughout the central nervous system and in the uvea and retina of the eye 12, and S100β-specific CD4 T cell lines with a T helper type 1 (Th1)-like phenotype have been established from patients with multiple sclerosis (MS) 13. Although proliferative T cell responses to S100β as a whole antigen are seen at high frequencies in type 1 diabetes 8, to our knowledge a comprehensive analysis of the immunogenic determinants of S100β in type 1 diabetes and the nature of this response has not yet been carried out.

In this work we have identified several peptide epitopes derived naturally from S100β and presented to T lymphocytes by the MHC class II molecule DRB1*04:01. We tested the hypothesis that T cell responses against these epitopes might be present in type 1 diabetes and that the nature of these responses, through detection of secretion of interferon (IFN)-γ and interleukin (IL)-10, would be different when compared to those of matched healthy controls. Our study shows support for these hypotheses, as IFN-γ responses are more frequent in type 1 diabetes patients. Of interest is the finding that a dual response (secretion of IFN-γ and IL-10) may also be observed, suggesting active regulation of the response in some patients. These results confirm that S100β is an autoantigen in human type 1 diabetes and suggest that detection of immune responses against it could be useful for diagnostic/prognostic and potentially immunotherapeutic strategies.

Materials and methods

Subjects

Fresh heparinized blood samples were obtained from 34 type 1 diabetes patients (21 newly diagnosed type 1 diabetes patients – ND, mean time since diagnosis 3·0 months, and 13 type 1 diabetes patients with long-standing disease – LS, mean time since diagnosis 300·9 months) (Table 2) and 33 healthy donors (HC) with no family history of type 1 diabetes. HLA-DRB1*04:01 positivity was determined by polymerase chain reaction (PCR) using previously described methods 14. These studies were carried out with the approval of the local Research Ethics Committee and informed consent was obtained from all participants.

Table 2.

Demographic data [interferon (IFN)-γ/interleukin (IL)-10 cohort]

DRB1*04:01 subjects† Non-DRB1*04:01 subjects
HC (n = 8) ND (n = 11) LS (n = 13) HC (n = 25) ND (n = 10)
Age (years; mean ± s.d.) 30 ± 8·4 22·4 ± 7·4 49·2 ± 10·9 22·4 ± 3·5 25·5 ± 5·8
Time since diagnosis (months; mean ± s.d.) n.a. 2·4 ± 4·5 300·9 ± 130·4 n.a. 2·1 ± 1·9
Time since diagnosis (months; range) n.a. 1·0–16·0 96·0–504·0 n.a. 1·0–7·0
Autoantibody-positive (%) n.a. 62·5% n.a. n.a. 88·9%
PNS complications (%) n.a. n.a. 53·8% n.a. n.a.
Open in a new tab
†

Age: P < 0·001 [analysis of variance (anova)+Tukey's b; differences due to long-standing (LS)]. No correlation between age and any other variable analysed was observed (P > 0·05; Pearson's χ2 test). ND = newly diagnosed; HC = healthy controls; PNS = peripheral nervous system; n.a. = not applicable; s.d. = standard deviation. Autoantibody positivity was determined as positivity for glutamic acid decarboxylase (GAD) and/or insulinoma-associated protein 2 (IA-2).

S100β cloning and purification and identification of naturally processed and presented epitopes presented by HLA-DRB1*04:01

The open reading frame (ORF) corresponding to the sequence of human S100β in the pOTB7 vector (GenBank Accession no. BC001766·1) was obtained from Geneservice Ltd (Cambridge, UK). The nucleotide sequence coding for the BirA peptide tag GGGLNDIFEAQKIEWHE was added to the 5′ end of the S100β ORF by PCR and this construct was subcloned into pCR T7/NT-TOPO (Invitrogen, Barcelona, Spain), and the recombinant protein expressed in the BL21 (DE3) pLysS bacterial strain. S100β was purified by affinity chromatography in native conditions using a nickel resin (Ni-NTA agarose) (Qiagen, Madrid, Spain) and biotinylated in vitro using the BirA enzyme purified in our laboratory (clone and protocol kindly provided by Dr Ton Schumacher, the Netherlands Cancer Institute, Amsterdam). After the biotinylation reaction, the protein was dialyzed against phosphate-buffered saline (PBS) and purified by affinity chromatography using an avidin resin (SoftLinkTM Soft Release Resin; Promega Biotech Ibérica, Madrid, Spain). S100β was aliquoted and stored at −80°C at 150 μg/ml in 0·1 M phosphate buffer containing 5 mM d-biotin.

Identification of S100β naturally processed and presented epitopes (NPPEs) by mass spectrometry

The identification of naturally processed and presented epitopes has been described in detail elsewhere 15,16. Briefly, 1010 Priess cells (homozygous for DRB1*04:01) 16 were incubated sequentially with optimized concentrations of biotinylated pokeweed mitogen lectin, avidin and biotinylated S100β. As a negative control we incubated 1010 Priess with the biotinylated lectin and avidin only. HLA-DR4 molecules were purified by affinity chromatography using an HLA-DR-specific monoclonal antibody purified from the hybridoma HB-55 (clone L243; American Type Culture Collection, Manassas, VA, USA). HLA-DR4-bound peptides were eluted by 15 min incubation in 10% acetic acid at 70°C and filtered on a 10 kDa-cut off Amicon Ultra (Millipore Ibérica, Madrid, Spain). The peptide mix was fractionated by high-performance liquid chromatography (HPLC) in a 15-mm long reverse-phase BioBasic-18 column using a 0–60% acetonitrile gradient (containing 0·05% trifluoroacetic acid) at a flow speed of 0·150 ml/min. Each fraction was vacuum-dried and stored at −80°C.

Mass spectrometry analysis was carried out in the Spectrometry Service of the Instituto de Investigaciones Sanitarias (University of Santiago de Compostela). Each dried peptide fraction was dissolved in 4 μl of 0·5% formic acid and equal volumes of peptide and matrix [0·3% a-cyano-4-hydroxycinnamic acid in 0·1% trifluoroacetic acid (TFA)-supplemented 60% acetonitrile] were deposited onto an Opti-time-of flight (TOF) matrix-assisted laser desorption/ionization (MALDI) plate and analysed on a MALDI-TOF–TOF analyser (Applied Biosystems, Waltham, MA, USA). The mass spectrometry (MS) spectra were acquired in reflectron-positive ion mode with a Nd:YAG, 355 nm wavelength laser; 0·5 μl of Peptide Calibration Standard II (Bruker, Billerica, MA, USA) were loaded as external calibrators. All the MS/MS spectra were generated with a relative resolution of 300 [full-width half-maximum (FWHM)] and metastable suppression. The analysis was carried out with the 4000 Series Explorer software version 3·5 (Applied Biosystems) and Mascot version 2·1 (Matrix Science, Boston, MA, USA) to search against a non-redundant protein database (NCBInr) or in a S100β-specific database, with 300 parts per million (ppm) of precursor tolerance, 0·35 Da MS/MS fragment tolerance, carbamidomethyl cystein as fixed modification, oxidized methionine as variable modification and allowing one missed cleavage.

HLA-binding assays

The binding affinity of S100β peptides to the HLA-DRB1*04:01 molecule was performed by Proimmune Ltd using the ProImmune REVEAL™ Immunogenicity System (Proimmune Ltd, Oxford, UK). The consensus peptides, a library of one amino acid off-set overlapping 15-mer peptides for each consensus peptide and selected NPPEs were tested. Briefly, the peptide binding to HLA-DRB1*04:01 is evaluated by the presence or absence of the native conformation of the HLA–peptide complex using a labelled native conformation-dependent antibody. The score of the test peptide is reported quantitatively as a percentage of the signal generated by the positive control peptide. A stability index is calculated by comparing the binding at 0 versus 24 h. A peptide binding with intermediate affinity was included as an additional control. Peptides with a binding score above 15% are considered HLA-DRB1*04:01 binders.

Detection of S100β-specific responses

Enzyme-linked immunospot (ELISPOT) assays for the detection of IFN-γ and IL-10 (U-Cytech Biosciences, Utrecht, the Netherlands) were used to detect S100β-specific responses, as described previously 16,17;; this assay has proved to have significant discriminative ability for type 1 diabetes in blinded proficiency testing 18, and is more sensitive in the detection of autoreactive T cells when compared to other assays such as intracellular cytokine staining.

All peptides were synthesized by ProImmune Ltd to > 90% purity. Peripheral blood mononuclear cells (PBMCs) were isolated on density gradients, resuspended in RPMI-1640 containing 10% human AB+ serum (Sigma Aldrich, Madrid, Spain) and were tested in indirect ELISPOTs performed as described previously 16,17. Briefly, PBMCs were cultured in 48-well plates with the relevant peptides to a final concentration of 10 μM and incubated at 37°C for 48 h, before harvesting the non-adherent cells and culturing them in triplicate for a further 16 h in an anti-cytokine antibody-coated plate. Control wells contained an equivalent concentration of the peptide diluent [dimethylsulphoxide (DMSO)], a 15-mer peptide library mix derived from the human actin as a negative control, a 15-mer peptide library mix derived from the Epstein–Barr virus (EBV) pp65 as a positive control or the Penta™ vaccine Pediacel® (Sanofi-Pasteur, Maidenhead, UK) as an additional positive control 17.

Data were expressed as the total number of spots per triplicate divided by the total spot number in the presence of the peptide diluent alone [stimulation index (SI)]; the SI was preferred to the subtraction of background counts (delta spot number), as the latter masks the extent of the background response (see 16 for details). A response was considered positive when the SI was above the threshold determined using receiver-operating characteristic (ROC) curves, as described elsewhere 16.

Statistical analysis

Analyses were performed using Prism (GraphPad Software Inc., San Diego, CA, USA) and spss statistics version 19 (IBM, Armonk, NY, USA). When comparing two independent groups, a two-tailed Student's t-test analysis (for normal distributions) or the Mann–Whitney U-test (non-normal distributions) was carried out. For comparison of multiple independent groups, analysis of variance (ANOVA) (normal distributions) or Kruskal–Wallis (non-normal distributions) tests were used. In the case of qualitative variables, Fisher's exact test was used. For correlations, the Pearson's χ2 test was used. P-values were considered significant when less than 0·05.

Results

Identification of S100β peptides naturally processed and presented by HLA-DRB1*04:01

To identify T cell epitopes derived from S100β we used an efficient lectin-based antigen delivery system (ADS) followed by purification of HLA-DRB1*04:01 molecules and identification of bound peptides by high-accuracy mass matching, a method that has been validated previously with other autoantigens implicated in the development of type 1 diabetes 15,19.

Fifteen masses were identified as being unique to the S100β-pulsed Priess cells, corresponding to 18 S100β sequences (Table 1). The peptides circumscribed three potential nested sets, a characteristic feature of class II MHC processing, and when in-silico predictions were performed it became apparent that each of the three nested sets contains at least one potential DRB1*04:01 binding motif (Table 1 and Supporting information, Table S1). Consensus peptides were designed so that they included these putative HLA-DRB1*04:01 binding motifs, and were extended towards the amino- and carboxy-termini to provide an adequate number of peptide-flanking residues to favour binding to both the MHC molecule and the T cell receptor 20. These consensus peptides are shown in Table 1.

Table 1.

Experimental and expected m/z masses from S100β-derived peptides eluted from human leucocyte antigen (HLA)-DRB1*04:01 with the matching sequence and the consensus peptide used in the enzyme-linked immunospot (ELISPOT) assays

Experimental m/z Expected m/z Residues S100β sequence* Consensus peptide
2·136 048 2·136 054 6–24 KAMVALIDVFHQYSGREGD S100 6–25
2·136 054 7–25 AMVALIDVFHQYSGREGDK KAMVALIDVFHQYSGREGDK
2·137 851 2·137 063 2–20 SELEKAMVALIDVFHQYSG
2·610 379 2·610 269 1–23 MSELEKAMVALIDVFHQYSGREG
1·325 661 1·325 765 21–31 REGDKHKLKKS S100 21–36
1·352 681 1·352 826 26–33 HKLKKSELKEL REGDKHKLKKSELKEL
2·285 184 2·285 202 31–49 SELKELINNELSHFLEEIK
2·285 202 30–48 KSELKELINNELSHFLEEI
2·336 108 2·336 304 20–39 GREGDKHKLKKSELKELINN
2·365 166 2·365 308 22–41 EGDKHKLKKSELKELINNEL
2·423 172 2·423 336 19–39 SGREGDKHKLKKSELKELINN S100 25–46
2·677 497 2·677 503 25–46 KHKLKKSELKELINNELSHFLE KHKLKKSELKELINNELSHFLE
1·420 840 1·420 631 73–85 EFMAFVAMVTTAC S100 68–92
1·420 664 71–82 FQEFMAFVAMVT ECDFQEFMAFVAMVTTACHEFFEHE
1·592 802 1·592 749 71–84 FQEFMAFVAMVTTA
2·500 297 2·500 066 70–90 DFQEFMAFVAMVTTACHEFFE
2·650 556 2·651 140 71–92 FQEFMAFVAMVTTACHEFFEHE
2·967 461 2·967 194 60–86 TLDNDGDGECDFQEFMAFVAMVTTACH
Open in a new tab
*

In bold type and underlined sequences containing one or two overlapping potential DRB1*0401 binding motifs as predicted by SYFPEITHI (http://www.syfpeithi.de/).

HLA-DRB1*04:01 peptide-binding assays of S100β peptide epitopes

Consensus synthetic peptides, a library of 15-mer peptides offset by one amino acid based on each consensus sequence and selected NPPEs identified were analysed in an in-vitro MHC–peptide binding assay to determine the affinity of those sequences for HLA-DRB1*04:01. As shown in Fig. 1, the consensus peptide S10068–92 (Fig. 1d) and to a lesser extent S1006–25 (Fig. 1a), but not S10021–36 or S10025–46 (Fig. 1b,c, respectively), show binding affinity for HLA-DRB1*04:01 in vitro.

Fig 1.

Fig 1

Open in a new tab

In-vitro binding of S100β peptides to human leucocyte antigen (HLA)-DRB1*04:01. For each nested set shown in Table 1 the binding to DRB1*04:01 was measured for the designed consensus peptide, a library of 15-mer peptides off-set by one amino acid based on the consensus sequence and some of the naturally processed and presented epitopes (NPPEs), using the ProImmune REVEAL™ major histocompatibility complex (MHC)-peptide binding assay. (a) S1006–25 consensus peptide (1), 15-mer library (2–7) and NPPEs (8–10). (b) S10021–36 consensus peptide (1), 15-mer library (2–3) and NPPEs (4–7). (c) S10025–46 consensus peptide (1), 15-mer library (2–9) and NPPEs (10–13). (d) S10068–92 consensus peptide (1), 15-mer library (2–12) and NPPEs (13–16). For each peptide, the solid bar indicates the relative percentage of peptide bound at 0 h and the striped one the percentage of peptide still bound after 24 h. Hi = control peptide with high affinity for HLA-DRB1*04:01. Int = control peptide with intermediate affinity for HLA-DRB1*04:01. Peptides with a binding score above 15% are considered positives. The potential DRB1*04:01 binding regions predicted in silico with the highest scores are underlined in the consensus peptide.

The consensus peptide S10068–92 shows the highest binding affinity for HLA-DRB1*04:01 when compared to the other consensus peptides (Fig. 1d). Its binding stability is equivalent to that of the high-affinity control peptide, as shown by the percentage of peptide still bound after 24 h (Fig. 1d, striped bars). Moreover, the 15-mer peptide library shows that the phenylalanine in position 10 of the consensus peptide could correspond to the amino acid in position 1 of the peptide-binding motif, as its deletion considerably reduces the binding affinity of the corresponding 15-mer (Fig. 1d; compare peptides 11 and 12). Therefore, the amino acid sequence F77VAMVTTAC85 could constitute a high-affinity binding motif to HLA-DRB1*04:01, as predicted by MHC-binding algorithms (Table 1 and Supporting information, Table S1) and suggested by our data (Fig. 1).

The consensus peptide S1006–25 shows a lower binding affinity compared to S10068–92 but with a similar stability to an intermediate-affinity control peptide (Fig. 1a, striped bars). Due to this lower binding affinity and to the limited 15-mer peptide library it is difficult to confirm the peptide binding motif. However, MHC-binding predictions made by several algorithms suggest that the peptide binding motif in this peptide may lie in the amino acid sequence V9ALIDVFHQYS19 (Table 1 and Supporting information, Table S1).

In contrast, for the consensus peptides S10021–36 and S10025–46 no significant HLA-binding was detected (Fig. 1b,c, respectively). Almost all peptides show a very low affinity for DRB1*04:01, with the exception of the short NPPE HKLKKSELKEL that binds with a similar affinity to that of the intermediate-affinity control peptide (Fig. 1b, peptide 5 and Fig. 1c, peptide 11). However, it is well known that autoantigenic peptides may show no or very low affinity for specified HLA molecules 21, and for this reason these consensus peptides were also selected for functional assays (ELISPOTs) alongside S1006–25 and S10068–92.

Type 1 diabetes patients show an IFN-γ response against S100β peptides

PBMCs from 32 DRB1*04:01 subjects (11 ND, 13 LS type 1 diabetes patients and eight healthy donors (HC) (Table 2) were analysed by ELISPOT as described previously 16,17 to detect the presence of IFN-γ- and IL-10-secreting cells specific for the S100β consensus peptides. No differences among subject groups were found for responses against the vaccine control, either for IFN-γ (mean SIs: HC = 79·5; ND = 72·4; LS = 42·1) or for IL-10 (mean SIs: HC = 15·5; ND = 12·5; LS = 13·2) (P > 0·05, Kruskal–Wallis test).

For IFN-γ, the ND patients show higher responses against three of the consensus peptides, S1006–25, S10021–36 and S10025–46, when compared to HC, and this difference is statistically significant (Fig. 2a). In the case of LS patients, a statistically significant difference is seen for S10021–36 and S10068–92 (Fig. 2a); in the latter case this difference is still significant even when the outliers (SI = 6·0 and SI = 16·0) are removed from the analysis (P = 0·01 versus P = 0·02 without the outliers). This suggests that these S100β peptides are indeed the target of an IFN-γ response in patients with type 1 diabetes, which may be sustained for some epitopes into LS disease.

Fig 2.

Fig 2

Open in a new tab

Detection of S100β-specific responses. Peripheral blood mononuclear cells (PBMCs) from DRB1*04:01 subjects [healthy control (HC), newly diagnosed (ND) and long-standing (LS) patients] were stimulated with the consensus S100β peptides and both interferon (IFN)-γ and interleukin (IL)-10 secretion was determined by indirect enzyme-linked immunospot (ELISPOT). (a) IFN-γ; (b) IL-10; SI = stimulation index (see Material and methods for description). Horizontal lines represent the median (*P < 0·05; **P < 0·01; Mann–Whitney U-test). (c) Correlation between the IFN-γ and IL-10 responses for S10068–92 suggests a possible dual response against this peptide in LS patients but not HC or ND patients. Open squares = HC; triangles = ND; circles = LS. (d) The percentage of positive subjects for IFN-γ in the HC, ND and LS populations was calculated taking into account the threshold suggested by the receiver operating characteristic (ROC) curves. Higher responses against S1006–25 and S10068–92 were seen in ND and LS, respectively, compared to HC (*P < 0·05; Fisher's exact test).

For IL-10 (Fig. 2b) the ND patients show higher responses against S10021–36, when compared to HC, while in the case of LS patients an IL-10 response is seen against S10068–92; in the latter case, differences are not due to a single outlier (SI = 11·0), as the differences are still statistically significant when this sample is not included in the analysis (P = 0·02 versus P = 0·04). Interestingly, in the case of S10068–92 there is a significant positive correlation between the IFN-γ and IL-10 responses described above for LS patients (Fig. 2c; R = 0·67, P < 0·05; Pearson's χ2 test), but not for HC or ND patients; in fact, 100% of the IFN-γ positive LS patients are also positive for IL-10 (SI ≥ 1·5, as determined by ROC curve analysis, see next section), suggesting that the LS patients show a dual IFN-γ/IL-10 response against S10068–92. In the case of ND patients and S10021–36 there is no correlation, suggesting that the ND patients tested are either IFN-γ responders or IL-10 responders, but not dual responders. The IFN-γ/IL-10 responses are not correlated for any other peptide/type of subject (data not shown).

Interestingly, when these ELISPOT analysis are performed with 35 non-DRB1*04:01 subjects (10 ND patients and 25 HC; Table 2), no differences between ND patients and HC were found for any of the four consensus S100β peptides (data not shown), demonstrating that the differential response shown in Fig. 2 is specific to DRB1*04:01.

S100β responses were not related to autoantibody status in the case of ND [glutamic acid decarboxylase antibodies (GADA) and/or IA-2A] or to peripheral nerve system (PNS) complications in the case of LS (P > 0·05, Mann–Whitney U-test), and no correlations were found between the IFN-γ or IL-10 cytokine responses and age, time since diagnosis (P > 0·05, Pearson's χ2 test) or gender (P > 0·05, Mann–Whitney U-test). However, in the case of LS patients, all IFN-γ responses correlate significantly among the different epitope specificities (Fig. 3), whereas in the case of HC and ND patients no interepitope correlations were found (data not shown). As no correlations among these responses and age were found for any of the three groups, age alone cannot be the determinant of these correlations. These data suggest a highly inter-related response to S100β peptides in LS patients, but not in HC or ND patients. These interepitope correlations were not found for IL-10 (data not shown).

Fig 3.

Fig 3

Open in a new tab

Interferon (IFN)-γ response to S100β peptides is highly interrelated in long-standing (LS) patients. The correlations among the simulation indices (SIs) for IFN-γ and all four S100β peptides were calculated, showing that for all the six possible combinations there is a significant positive correlation. Pearson's correlation coefficients (R) and P-values for each correlation are shown. (a) S1006–25 versus S10021–36; (b) S1006–25 versus S10025–46; (c) S10021–36 versus S10025–46; (d) S1006–25 versus S10068–92; (e) S10021–36 versus S10068–92; (f) S10025–46 versus S10068–92.

Interestingly, in the case of LS patients the SIs for both IFN-γ and IL10 are higher for the consensus peptide with higher affinity for HLA-DRB1*0401 (S10068–92), when compared to those peptides with lower binding affinities (S10021–36 and S10025–46) (Supporting information, Fig. S1).

In summary, these results indicate the presence of specific IFN-γ and IL-10 responses against the S100β consensus peptides in type 1 diabetes patients which is restricted to patients with HLA-DRB1*04:01 and differs in ND and LS patients.

Prevalence and breadth of IFN-γ responses against S100β peptides

Analysis of the ROC curves of IFN-γ responses against any S100β consensus peptide shows an area under the curve (AUC) of > 0·5 (P < 0·001), indicating that responses can be disease discriminating. Applying the optimal threshold for positivity calculated from the ROC analysis (SI = 1·5), we observed a significantly higher prevalence of S1006–25-positive responses among ND type 1 diabetes patients when compared with healthy controls (Fig. 2d). Interestingly, for S10068–92 LS patients showed the highest prevalence of positive individuals (Fig. 2d). In addition, ND type 1 diabetes patients respond against a significantly higher number of S100β peptides when compared to HC (mean of 1·0 versus 1·8; P < 0·05, Mann-Whitney U-test). The ROC curve of IL-10 responses against any S100β consensus peptide showed an AUC not significantly > 0·5. In conclusion, the measurement of IFN-γ responses against four S100β peptides representing consensus epitope regions is disease discriminating, in terms of the level, prevalence and breadth of the response observed in newly diagnosed patients.

Discussion

The Ca2+-binding neurotrophic factor S100β has been shown to be a target of the immune response in the NOD mice as early as 4 weeks into the pathogenic process 5; this suggests that by halting the destruction of the S100β-expressing pSC cells pancreatic islets could remain intact, thereby preserving insulin secretion and preventing diabetes 5. Although in human type 1 diabetes previous research has shown autoimmune responses directed against the whole antigen, thereby indicating a potential role as a target for therapeutic intervention 8,18, no identification of the epitopes targeted or characterization of this autoimmune response has been performed.

We have identified several naturally processed and presented T cell epitopes derived from S100β, which trigger both IFN-γ and IL-10 responses in type 1 diabetes patients and healthy donors. IFN-γ responses were seen more frequently in type 1 diabetes patients compared to healthy donors to several of the S100β peptides. Similar T cell responses have been identified for other peptide–epitope autoantigens in human type 1 diabetes such as IA-2, PI and GAD65, for which the response is also dominated by the secretion of IFN-γ 16,17. These responses are also seen, although in small frequencies, in non-diabetic subjects against autoantigens such as GAD65 22, islet-specific glucose-6-phosphatase catalytic subunit related protein (IGRP) 23, the zinc transporter 8 (ZnT8) 24, whole S100β 8 and in the present work, pointing to the fact that some sort of ‘basal’ reactivity against S100β naturally exists in healthy donors. These results are consistent with autoimmunity to S100β being a component of the pathogenic processes associated with the development of type 1 diabetes. Interestingly, the specificity of these responses seem to be different depending on the duration of the disease, as ND patients show elevated responses to the consensus peptides located in the amino-terminus of the protein while LS patients only do so against the carboxy-terminal peptide S10068–92. These differential responses in LS patients do not depend upon factors such as PNS complications, so further studies need to be carried out to address the factors attributable to these different responses between ND and LS patients.

Interestingly, we also found that some type 1 diabetes patients also show IL-10 responses against some of the S100β-derived peptides. Dual responses, such as those shown mainly by the LS group and for the carboxy-terminus S10068–92 peptide, have been observed previously for peptides derived from other autoantigens such as IA-2 and pre-PI in ND patients 16 or against whole GAD65 in LS patients, where dual IFN-γ and IL-13 responses have been found 25; this probably indicates an attempt of the immune system to control the autoimmune response against these antigens. Moreover, this dual response against S100β has also been observed in MS, where DRB1*15:01-restricted T cell lines against whole S100β have been generated from both MS patients and healthy subjects, showing a mixed cytokine secretion pattern with production of IFN-γ and TNF-α while also secreting IL-10 and IL-4 13. Further studies, including the quantification of multiple cytokine in culture supernatant to determine the secretion of other proinflammatory and regulatory cytokines, together with assays such as two-colour fluorescent ELISPOTs to determine whether this ‘dual’ phenotype is due to the same cells producing both IFN-γ and IL-10 or due to cytokine production by different subsets of cells, should be carried out to fully determine the nature of these S100β-specific responses in type 1 diabetes.

We focused our analysis on one of the HLA molecules conferring susceptibility to the development of type 1 diabetes, the DRB1*04:01 molecule. The peptide epitopes described show a diverse range of binding affinity to this HLA molecule from the high affinity shown by the carboxy-terminal S10068–92 to the low binding shown by S10021–36 or S10025–46. Data shown in the present work suggest a DRB1*04:01-restricted response, as non-DRB1*04:01 type 1 diabetes patients and healthy donors do not show specific responses to the same epitopes. It remains a possibility that the S100β epitopes could bind to another HLA molecule such as DR3, or even to HLA-DQ molecules, and future studies should focus upon analysing this possibility.

The presence of autoreactive memory T cells in LS type 1 diabetes directed against antigens such as pre-PI and GAD65 is well described 26,27 28,29, and our findings in relation to S100β peptides also extend this observation. S100β is an antigen with a wide cellular expression, including cells of the nervous system including astrocytes and Schwann cells 30, and other non-nervous tissues, such as hepatocytes, adipocytes and smooth muscle 31,32. Elevated S100β serum levels are associated with neurological and neuropsychiatric disorders, and blood levels seem to correlate inversely with insulin levels 33. It is tempting to speculate that sustained low levels of insulin could influence maintenance of high levels of S100β in blood and therefore positively influence development of autoimmunity against this protein.

It is interesting to note that highly correlated IFN-γ responses against the four peptide epitopes described here are observed in LS but not ND patients. As S100β is an antigen expressed not only by the pancreatic β-cells but other cell types 27–29, it is tempting to speculate that this cellular response is maintained in these subjects by protein present in those cellular sources generating higher numbers of memory T cells directed against these epitopes. Moreover, as LS patients seem to respond more frequently to the carboxy-terminal S10068–92 region of S100β compared to the ND patients it is possible that after disease diagnosis an intramolecular epitope-spreading could be taking place, an observation made previously for the Th1 response against GAD65 in NOD mice, where the response seems to be limited initially to a confined region of the protein but later spreads intramolecularly to other epitopes during disease progression 34.

In conclusion, we have identified naturally processed and presented epitopes derived from the autoantigen S100β, which are targeted by both IFN-γ and IL-10 responses in type 1 diabetes patients. We also show that S100β autoreactivity changes with duration of disease and future research should focus upon larger longitudinal cohorts in order to understand this finding more clearly and to explain the role played by this anti-S100β cellular response in the development of clinical symptoms or future neurological complications.

Acknowledgments

We thank Victoria Illanes for recruiting some of the newly diabetic patients for this work. This research was supported by the Instituto de Salud Carlos III (grant PI10/00207 to R. V.-C.). A. G.-F. thank the European Commission BIOCAPS (316265, FP7/REGPOT) and Xunta de Galicia (Agrupamento INBIOMED and Grupo con potencial crecimiento). During this work I. G.-T. was supported sequentially by a Maria Barbeito predoctoral fellowship (Xunta de Galicia, Spain), a Barrié de la Maza's Foundation postdoctoral fellowship and currently a Marie Curie Intra-European fellowship.

Disclosure

The authors declare no financial or commercial conflicts of interest.

Author contributions

All authors edited and approved the written paper.

Supporting Information

Additional Supporting information may be found in the online version of this article at the publisher's web-site:

Table S1. In-silico analysis of the DR4-binding regions of S100β.

cei0180-0207-sd1.pdf (208.3KB, pdf)

Figure S1. Cytokine responses against S100β peptides in long-standing (LS) patients according to the human leucocyte antigen (HLA) affinities. Interferon (IFN)-γ (a) and interleukin (IL)-10 (b) responses against the S100β peptide epitopes in the long-standing patients are plotted according to the peptide affinities for HLA-DRB1*04:01. Statistically significant higher responses are detected against the S100β peptide with the highest affinity for HLA-DRB1*04:01 (S10068-92) compared to those peptides with the lower binding affinities (S10021-36 and S10025-46).

cei0180-0207-sd2.tif (879.9KB, tif)

References

  1. Babad J, Geliebter A, DiLorenzo TP. T-cell autoantigens in the non-obese diabetic mouse model of autoimmune diabetes. Immunology. 2010;131:459–465. doi: 10.1111/j.1365-2567.2010.03362.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Roep BO, Peakman M. Antigen targets of type 1 diabetes autoimmunity. Cold Spring Harb Perspect Med. 2012;2:a007781. doi: 10.1101/cshperspect.a007781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Chaparro RJ, Dilorenzo TP. An update on the use of NOD mice to study autoimmune (Type 1) diabetes. Expert Rev Clin Immunol. 2010;6:939–955. doi: 10.1586/eci.10.68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Coppieters KT, Dotta F, Amirian N, et al. Demonstration of islet-autoreactive CD8 T cells in insulitic lesions from recent onset and long-term type 1 diabetes patients. J Exp Med. 2012;209:51–60. doi: 10.1084/jem.20111187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Winer S, Tsui H, Lau A, et al. Autoimmune islet destruction in spontaneous type 1 diabetes is not beta-cell exclusive. Nat Med. 2003;9:198–205. doi: 10.1038/nm818. [DOI] [PubMed] [Google Scholar]
  6. Gomez-Tourino I, Camina-Darriba F, Otero-Romero I, et al. Autoantibodies to glial fibrillary acid protein and S100beta in diabetic patients. Diabet Med. 2010;27:246–248. doi: 10.1111/j.1464-5491.2009.02911.x. [DOI] [PubMed] [Google Scholar]
  7. Burbelo PD, Lebovitz EE, Bren KE, et al. Extrapancreatic autoantibody profiles in type I diabetes. PLOS ONE. 2012;7:e45216. doi: 10.1371/journal.pone.0045216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Banwell B, Bar-Or A, Cheung R, et al. Abnormal T-cell reactivities in childhood inflammatory demyelinating disease and type 1 diabetes. Ann Neurol. 2008;63:98–111. doi: 10.1002/ana.21244. [DOI] [PubMed] [Google Scholar]
  9. Tsui H, Chan Y, Tang L, et al. Targeting of pancreatic glia in type 1 diabetes. Diabetes. 2008;57:918–928. doi: 10.2337/db07-0226. [DOI] [PubMed] [Google Scholar]
  10. Standifer NE, Ouyang Q, Panagiotopoulos C, et al. Identification of novel HLA-A*0201-restricted epitopes in recent-onset type 1 diabetic subjects and antibody-positive relatives. Diabetes. 2006;55:3061–3067. doi: 10.2337/db06-0066. [DOI] [PubMed] [Google Scholar]
  11. Marenholz I, Heizmann CW, Fritz G. S100 proteins in mouse and man: from evolution to function and pathology (including an update of the nomenclature) Biochem Biophys Res Commun. 2004;322:1111–1122. doi: 10.1016/j.bbrc.2004.07.096. [DOI] [PubMed] [Google Scholar]
  12. Kojima K, Berger T, Lassmann H, et al. Experimental autoimmune panencephalitis and uveoretinitis transferred to the Lewis rat by T lymphocytes specific for the S100 beta molecule, a calcium binding protein of astroglia. J Exp Med. 1994;180:817–829. doi: 10.1084/jem.180.3.817. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Schmidt S, Linington C, Zipp F, et al. Multiple sclerosis: comparison of the human T-cell response to S100 beta and myelin basic protein reveals parallels to rat experimental autoimmune panencephalitis. Brain. 1997;120:1437–1445. doi: 10.1093/brain/120.8.1437. [DOI] [PubMed] [Google Scholar]
  14. Bunce M. PCR-sequence-specific primer typing of HLA class I and class II alleles. Methods Mol Biol. 2003;210:143–171. doi: 10.1385/1-59259-291-0:143. [DOI] [PubMed] [Google Scholar]
  15. Peakman M, Stevens EJ, Lohmann T, et al. Naturally processed and presented epitopes of the islet cell autoantigen IA-2 eluted from HLA-DR4. J Clin Invest. 1999;104:1449–1457. doi: 10.1172/JCI7936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Arif S, Tree TI, Astill TP, et al. Autoreactive T cell responses show proinflammatory polarization in diabetes but a regulatory phenotype in health. J Clin Invest. 2004;113:451–463. doi: 10.1172/JCI19585. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Arif S, Moore F, Marks K, et al. Peripheral and islet interleukin-17 pathway activation characterizes human autoimmune diabetes and promotes cytokine-mediated {beta}-cell death. Diabetes. 2011;60:2112–2119. doi: 10.2337/db10-1643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Herold KC, Brooks-Worrell B, Palmer J, et al. Validity and reproducibility of measurement of islet autoreactivity by T-cell assays in subjects with early type 1 diabetes. Diabetes. 2009;58:2588–2595. doi: 10.2337/db09-0249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Stevens EJ, Peakman M. Enhanced T cell proliferation and increased responder frequency following delivery of antigen to the antigen-presenting cell; B cell dependency and use in detection of autoreactive T cells. J Immunol Methods. 1998;215:59–70. doi: 10.1016/s0022-1759(98)00059-3. [DOI] [PubMed] [Google Scholar]
  20. O'Brien C, Flower DR, Feighery C. Peptide length significantly influences in vitro affinity for MHC class II molecules. Immunome Res. 2008;4:6. doi: 10.1186/1745-7580-4-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Gebe JA, Falk BA, Rock KA, et al. Low-avidity recognition by CD4+ T cells directed to self-antigens. Eur J Immunol. 2003;33:1409–1417. doi: 10.1002/eji.200323871. [DOI] [PubMed] [Google Scholar]
  22. Danke NA, Yang J, Greenbaum C, Kwok WW. Comparative study of GAD65-specific CD4+ T cells in healthy and type 1 diabetic subjects. J Autoimmun. 2005;25:303–311. doi: 10.1016/j.jaut.2005.08.007. [DOI] [PubMed] [Google Scholar]
  23. Yang J, Danke NA, Berger D, et al. Islet-specific glucose-6-phosphatase catalytic subunit-related protein-reactive CD4+ T cells in human subjects. J Immunol. 2006;176:2781–2789. doi: 10.4049/jimmunol.176.5.2781. [DOI] [PubMed] [Google Scholar]
  24. Dang M, Rockell J, Wagner R, et al. Human type 1 diabetes is associated with T cell autoimmunity to zinc transporter 8. J Immunol. 2011;186:6056–6063. doi: 10.4049/jimmunol.1003815. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Viglietta V, Kent SC, Orban T, Hafler DA. GAD65-reactive T cells are activated in patients with autoimmune type 1a diabetes. J Clin Invest. 2002;109:895–903. doi: 10.1172/JCI14114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Toma A, Haddouk S, Briand JP, et al. Recognition of a subregion of human proinsulin by class I-restricted T cells in type 1 diabetic patients. Proc Natl Acad Sci USA. 2005;102:10581–10586. doi: 10.1073/pnas.0504230102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Toma A, Laika T, Haddouk S, et al. Recognition of human proinsulin leader sequence by class I-restricted T-cells in HLA-A*0201 transgenic mice and in human type 1 diabetes. Diabetes. 2009;58:394–402. doi: 10.2337/db08-0599. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Laughlin E, Burke G, Pugliese A, Falk B, Nepom G. Recurrence of autoreactive antigen-specific CD4+ T cells in autoimmune diabetes after pancreas transplantation. Clin Immunol. 2008;128:23–30. doi: 10.1016/j.clim.2008.03.459. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Vendrame F, Pileggi A, Laughlin E, et al. Recurrence of type 1 diabetes after simultaneous pancreas–kidney transplantation, despite immunosuppression, is associated with autoantibodies and pathogenic autoreactive CD4 T-cells. Diabetes. 2010;59:947–957. doi: 10.2337/db09-0498. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Steiner J, Bernstein HG, Bielau H, et al. Evidence for a wide extra-astrocytic distribution of S100B in human brain. BMC Neurosci. 2007;8:2. doi: 10.1186/1471-2202-8-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Yamashita M, Qian ZR, Sano T, Horvath E, Kovacs K. Immunohistochemical study on so-called follicular cells and folliculostellate cells in the human adenohypophysis. Pathol Int. 2005;55:244–247. doi: 10.1111/j.1440-1827.2005.01820.x. [DOI] [PubMed] [Google Scholar]
  32. Donato R, Sorci G, Riuzzi F, et al. S100B's double life: intracellular regulator and extracellular signal. Biochim Biophys Acta. 2009;1793:1008–1022. doi: 10.1016/j.bbamcr.2008.11.009. [DOI] [PubMed] [Google Scholar]
  33. Steiner J, Bernstein HG, Schiltz K, et al. Decrease of serum S100B during an oral glucose tolerance test correlates inversely with the insulin response. Psychoneuroendocrinology. 2014;39:33–38. doi: 10.1016/j.psyneuen.2013.10.001. [DOI] [PubMed] [Google Scholar]
  34. Kaufman DL, Clare-Salzler M, Tian J, et al. Spontaneous loss of T-cell tolerance to glutamic acid decarboxylase in murine insulin-dependent diabetes. Nature. 1993;366:69–72. doi: 10.1038/366069a0. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1. In-silico analysis of the DR4-binding regions of S100β.

cei0180-0207-sd1.pdf (208.3KB, pdf)

Figure S1. Cytokine responses against S100β peptides in long-standing (LS) patients according to the human leucocyte antigen (HLA) affinities. Interferon (IFN)-γ (a) and interleukin (IL)-10 (b) responses against the S100β peptide epitopes in the long-standing patients are plotted according to the peptide affinities for HLA-DRB1*04:01. Statistically significant higher responses are detected against the S100β peptide with the highest affinity for HLA-DRB1*04:01 (S10068-92) compared to those peptides with the lower binding affinities (S10021-36 and S10025-46).

cei0180-0207-sd2.tif (879.9KB, tif)

Articles from Clinical and Experimental Immunology are provided here courtesy of British Society for Immunology

RESOURCES