ORIGINAL RESEARCH published: 18 July 2019 doi: 10.3389/fimmu.2019.01695 Regulation of TSHR Expression in the Thyroid and Thymus May Contribute to TSHR Tolerance Failure in Graves’ Disease Patients via Two Distinct Mechanisms Ana Marín-Sánchez 1,2,3 , Daniel Álvarez-Sierra 2,3 , Oscar González 4 , Ana Lucas-Martin 5 , Edited by: Alicia Sellés-Sánchez 2 , Francesc Rudilla 6 , Emma Enrich 6 , Roger Colobran 1,2,3* and Ludger Klein, Ricardo Pujol-Borrell 1,2,3* Ludwig Maximilian University of Munich, Germany 1 Immunology Division, FOCIS Center of Excellence, Hospital Universitari Vall d’Hebron, Barcelona, Spain, 2 Diagnostic Reviewed by: Immunology Group, Vall d’Hebron Research Institute, Barcelona, Spain, 3 Department of Cell Biology, Physiology and Sandra M. McLachlan, Immunology, Universitat Autònoma de Barcelona, Barcelona, Spain, 4 Surgery Department, Hospital Universitari Vall Cedars-Sinai Medical Center, d’Hebron, Barcelona, Spain, 5 Endocrinology Division, Hospital Universitari Germans Trias Pujol, Badalona, Spain, United States 6 Immunogenetics and Histocompatibility Laboratory, Blood and Tissue Bank, Transfusional Medicine Group, Vall d’Hebron Terry Francis Davies, Research Institute, Barcelona, Spain Icahn School of Medicine at Mount Sinai, United States Graves’ disease (GD) involves the presence of agonistic auto-antibodies against the *Correspondence: Roger Colobran thyrotropin receptor (TSHR), which are responsible for the clinical symptoms. While rcolobran@vhebron.net failure of TSHR tolerance is central to GD pathogenesis, the process leading to this Ricardo Pujol-Borrell failure remains poorly understood. Two mechanisms intimately linked to tolerance have rpujol@vhebron.net been proposed to explain the association of SNPs located in TSHR intron 1 to GD: Specialty section: (1) differential alternative splicing in the thyroid; and (2) modulation of expression in This article was submitted to the thymus. To elucidate the relative contribution to these two mechanisms to GD Immunological Tolerance and Regulation, pathogenesis, we analyzed the level of full-length and ST4 and ST5 isoform expression in a section of the journal the thyroid (n = 49) and thymus (n = 39) glands, and the influence of intron 1-associated Frontiers in Immunology SNPs on such expression. The results show that: (1) the level of flTSHR and ST4 Received: 06 May 2019 expression in the thymus was unexpectedly high (20% that of the thyroid); (2) while Accepted: 08 July 2019 Published: 18 July 2019 flTSHR is the predominant isoform, the levels are similar to ST4 (ratio flTSHR/ST4 = 1.34 Citation: in the thyroid and ratio flTSHR/ST4 in the thymus = 1.93); (3) next-generation sequencing Marín-Sánchez A, Álvarez-Sierra D, confirmed the effect of the TSHR intron 1 polymorphism on TSHR expression in the González O, Lucas-Martin A, Sellés-Sánchez A, Rudilla F, Enrich E, thymus with a bias of 1.5 ± 0.2 overexpression of the protective allele in the thymus Colobran R and Pujol-Borrell R (2019) compared to the thyroid; (4) GD-associated intron 1 SNPs did not influence TSHR Regulation of TSHR Expression in the alternative splicing of ST4 and ST5 in the thyroid and thymus; and (5) three-color confocal Thyroid and Thymus May Contribute to TSHR Tolerance Failure in Graves’ imaging showed that TSHR is associated with both thymocytes, macrophages, and Disease Patients via Two dendritic cells in the thymus. Our findings confirm the effect of intron 1 polymorphisms Distinct Mechanisms. Front. Immunol. 10:1695. on thymic TSHR expression and we present evidence against an effect on the relative doi: 10.3389/fimmu.2019.01695 expression of isoforms. The high level of ST4 expression in the thymus and its distribution Frontiers in Immunology | www.frontiersin.org 1 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance within the tissue suggest that this would most likely be the isoform that induces central tolerance to TSHR thus omitting most of the hinge and transmembrane portion. The lack of central tolerance to a large portion of TSHR may explain the relatively high frequency of autoimmunity related to TSHR and its clinical consequence, GD. Keywords: Graves’ disease, TSHR, tolerance, splicing isoforms, thymus, thyroid, next-generation sequencing INTRODUCTION with ophthalmic pathologies (17). As a reference, the classical housekeeping gene, GAPDH, has 813 transcripts per million in Graves’ disease (GD) is a highly prevalent autoimmune disease the thyroid. characterized by the presence of agonistic auto-antibodies against In addition to the full-length TSHR (flTSHR), five truncated the thyrotropin receptor (TSHR), which are responsible for TSHR transcripts have been reported in multiple studies, of hyperthyroidism and extrathyroidal manifestations (1, 2). Failure which ST4 (1.3 kb) and ST5 (1.7 kb) are the most highly expressed of tolerance to the TSHR is central to the pathogenesis of GD; (18–20). ST4 and ST5 share the first eight exons with the flTSHR however, our understanding of the process that leads to this but contain an additional ninth exon that is unique and different failure remains incomplete. in ST4 and ST5, both of which lack exon 10. These unique The strong contribution of genetic factors to GD is best exons are encoded in two different regions of intron 8, which demonstrated in twin concordance studies that suggest that are selected and retained in the mRNA of the corresponding as much as 79% of the risk of developing GD is hereditary alternatively spliced forms. If translated, both ST4 and ST5 (3). During the past 25 years, classical genetic approaches and would encode most of the leucine-rich repeats containing the more recently, genome-wide association studies (GWAS), have TSH-binding extracellular region, but neither the hinge nor the identified several gene loci whose polymorphisms may contribute transmembrane domain of the TSHR; therefore, is plausible that to GD susceptibility. Among the confirmed genes in the HLA they are secreted (Figure 1). region, CTLA4 and PTPN22 confer a higher risk; however, Different studies have demonstrated a significant association individual contributions, with the exception of HLA, are limited between SNPs in the 40 kb region in the 5′ side of the large (4–6). The products of these genes participate in the regulation intron 1 (106 kb) of TSHR with GD [reviewed in (5)] (Figure 1). of the immune response and have been implicated in other Two different but not mutually exclusive mechanisms have been autoimmune diseases but do not explain why the autoimmune postulated to explain this association. Brand et al. proposed response focuses on the thyroid gland. Common polymorphisms that these SNPs influence mRNA splicing, which results in have also been found in the genes specifically expressed in the increased levels of the ST4 and ST5 transcripts encoding the short thyroid [e.g., the thyrotropin receptor [TSHR] or thyroglobulin isoforms (21). The authors proposed that the short ST4 and ST5 [TG] (7, 8)]; however, only the association with TSHR has been variants would be translated into the putatively soluble TSHR repeatedly confirmed [reviewed in (4, 5, 9, 10)]. Since the loss of forms, released into the circulation where they become available tolerance to TSHR is the central mechanism for GD pathogenesis, to antigen presenting cells in the periphery, where they may there is a great interest in understanding how these TSHR contribute to inducing an autoimmune response to the TSHR. polymorphisms contribute to the failure of tolerance. However, it is unclear why these isoforms are immunogenic, TSHR consists of 10 exons encoding a 764 amino acid protein rather than tolerogenic or ignored, in this hypothesis. In one of ∼95 kDa that is converted to a 120 kDa protein following experimental model, the soluble TSHR A subunit is more substantial glycosylation (11). The original TSHR peptide chain immunogenic than the membrane-bound TSHR (22); however, undergoes proteolytic cleavage, generating two subunits: (1) ST4 and ST5 differ from Subunit A in one exon and is difficult to the A-subunit, encoded by exons 1–9, which constitutes the predict their immunogenicity. Therefore, Brand et al. postulate extracellular domain; and (2) the B-subunit, encoded by exon the failure in peripheral tolerance as the crucial checkpoint for 10, which consists of a transmembrane region plus a 5 kDa C the development of GD. connecting peptide (12). Following the excision of the C peptide, The second proposed mechanism is based on our finding the A- and B-subunits remain linked by disulfide bridges that that the GD-associated SNPs in the intron 1 alleles modulate can be subsequently reduced, and the A-subunit is partially TSHR expression in the thymus. By measuring the levels of shed, whereas B-subunit will remain anchored to the membrane allele-specific TSHR mRNA in the thymus, we demonstrated (13, 14). Controversy remains regarding the persistence of single that individuals carrying the protective genotype have higher chain (uncleaved) TSHR on the surface of thyrocytes and to what levels of thymic TSHR mRNA compared to those with the extent subunit A is physiologically shed (15, 16). disease-predisposing genotype. According to the well-established The TSHR gene is expressed at medium to low levels in the mechanisms of central tolerance, the expression of self-antigens thyroid (207 copies per million transcripts) and is extremely in the thymus is required to develop a tolerant T cell repertoire low in all other tissues (data from the GTEX portal, <0.3 (23, 24); a dose response relationship between the amount of transcripts per million in subcutaneous fat or 0.5 in the brain), a self-antigen in the thymus, and the frequency of T cells except for AIRE-expressing orbital fibrocytes from GD patients clones that can recognize it in the periphery, which has been Frontiers in Immunology | www.frontiersin.org 2 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance FIGURE 1 | TSHR gene, GD-associated intron 1 polymorphisms and the predicted proteins expressed as cell-anchored proteins and as soluble forms. This figure shows the unique structure of TSHR. Intron 1 contains the GD-associated SNPs scattered on a region of ∼30 kb at the 5′ end of intron 1. The numbers on top of the gene diagram correspond to the amino acids coded by each exon. The numbers on the bottom of the gene are the distances in kilobases (Kb). The primary associated SNPs are labeled in addition to the area of histone 3 lysine 4 monomethylated (H3K4me1), where the repressor factor, PLZF, was found to bind, reducing TSHR transcription. The exons used exclusively by the ST4 and ST5 isoforms are labeled in green and blue, respectively. The TSH holoreceptor (TSHR) and the two short isoforms (ST4 and ST5) are depicted, showing that the short isoforms contained less than half of the potential immunogenic regions of the receptor, including a large proportion of the extracellular domain, specially the hinge. C, C peptide; SP, signal peptide; LRR, Leucine Reach Repeats; TMD, transmembrane domain; CM, Cytoplasmic Motifs [This figure is based on Figure 1 and Figure 2 of manuscript (9). Reproduced with permission of Thieme editorial]. demonstrated for the insulin gene in type 1 diabetes (25, 26) and These two non-mutually exclusive mechanisms propose that is well-established in animal models (27). the SNPs located in intron 1 participate in the regulation of TSHR Interestingly, Tomer et al. has recently provided some expression through: (1) differential alternative splicing in the confirmatory evidence to our proposal, as they also found thyroid; and (2) modulation of expression levels in the thymus. that intrathymic TSHR expression was lower in individuals To date, the effect of TSHR intron 1 polymorphisms on the homozygous for the disease-associated allele. In addition, these differential expression of its isoforms (ST4 and ST5) has only authors have shown that the effect of SNPs in intron 1 on the been analyzed in the thyroid, in which the level of expression expression of TSHR may depend on the differential affinity of the is relatively high compared with the thymus. However, although allele for the PLZF transcription repressor factor. In turn, PLZF the level of TSHR expression and presumably its isoforms in the is regulated by IFN-α (28) and it is known that there is a strong thymus is lower, it can be critical since it is in the thymus where IFN signature in the GD thyroid glands (29, 30). Moreover, IFN-α the T lymphocyte repertoire is configured. therapy can trigger thyroid autoimmunity when administered for To better understand how the complex expression of TSHR in hepatitis C virus infection (31). These authors propose that IFN- both the thymus and thyroid can modulate central and peripheral α can contribute to triggering GD by reducing TSHR expression tolerance to TSHR, it is useful to precisely identify the levels in the thymus. of different isoform expression in these two tissues and analyze Frontiers in Immunology | www.frontiersin.org 3 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance the influence of intron 1-associated SNPs on their expression. Switzerland). The specific primers were designed for the TSHR The results presented here constitute a detailed analysis of TSHR gene intron 1 region containing the rs179247 SNP (233 bp isoform expression in each tissue and provide insight into how amplimer) (Supplementary Table 3). TSHR tolerance may fail. Library Preparation, DNA Sequencing, SNP Calling, MATERIALS AND METHODS and Genotyping Briefly, the target-specific primers were synthesized using Patients and Samples CS1/CS2 tags (Fluidigm, San Francisco, CA) followed by a In this retrospective study, the diagnosis was made based on two-step PCR reaction. First, the region of interest was amplified thyroid hormone levels, thyroid antibodies, including TSHR with target-specific tagged primers, and second, the sample- antibodies and ultrasound and scintiscan images, by experienced specific barcode and adapters were introduced to generate the endocrinologists. Thyroid tissue was obtained from 49 patients sequencing library (Fluidigm). A GeneAmp R PCR System 9700 (43 females, 7 males; age range: 15–71 years) recruited from Thermal Cycler (Applied Biosystems, Foster City, CA) was used the Endocrinology Clinics of Hospital Universitari Germans for the amplification. PCR products were pooled and purified Trias i Pujol (HUGTiP) and Hospital Universitari Vall d’Hebron using a MinElute Gel Extraction Kit (50) (Qiagen). The pool (HUVH), both of which are affiliated with the Universitat was run using the paired-end sequencing method on a MiSeq Autònoma de Barcelona (UAB) (Supplementary Table 1). The sequencing system (Illumina, San Diego, CA, USA) using the samples were processed as previously described (32). MiSeq reagent kit v2 (500 cycles) (Illumina). Finally, the FASTQ Thymic tissue was obtained from 39 patients undergoing files were analyzed for the SNP rs179247 by read mapping and corrective cardiac surgery (31 pediatric and 8 adult patients; variant calling using the CLC Genomic Workbench software 15 females and 24 males; age range: 4 days−72 years), version 9.5.1 (CLC Bio, Qiagen, Aarhus, Denmark). A total of 19 from the Departments of Heart Surgery of HUGTiP or thymus and 8 thyroid samples from heterozygous individuals for HUVH. All samples were processed within 4 h of resection the SNP rs179247 were tested (Supplementary Table 4). under sterile conditions, as previously described (33) (Supplementary Table 2). Quantification of the Relative Gene Informed consent was obtained from all participants and the Expression by qPCR studies have been approved by the local institutional ethics review The level of gene expression for the full-length human TSHR board of the participating institutions (ref PR AG-145/2011). mRNA (flTSHR) and two TSHR alternatively spliced transcripts (ST4 and ST5) were measured by qPCR using Taqman probes. DNA and RNA Isolation DNA-free RNA samples from the thyroid and thymus were Both genomic DNA (gDNA) and total RNA were isolated from obtained as described above. The cDNA was synthesized using the total thyroid and thymus tissue using standard methods Oligo(dT) primers (First Strand cDNA Synthesis Kit for RT- (QIAamp DNA Mini QIAcube Kit and RNeasy Mini Kit, PCR (AMV), Roche). To measure the level of flTSHR expression, QIAGEN, Hilden, Germany). An additional step of DNase I we used a pre-designed TaqMan R Gene Expression Assay that treatment was applied to the RNA samples (RNase-free DNase spans the TSHR exons 9 and 10 (assay ID: Hs01053841_m1). set kit, QIAGEN). To check for contaminating gDNA in the RNA To measure the flTSHR, ST4, and ST5 transcripts, we used samples, 100 ng of the total RNA were subjected to 45 cycles of the primers and probes as described by Brand et al. (21) PCR using specific primers for a 309 nt non-transcribed region (Supplementary Table 5). When checking the sequences of the of the CTLA4 promoter. Only samples free of DNA were used primers and probes, we realized that the ST5 probe sequence used for subsequent experiments. by Brand et al. did not coincide with the ST5 consensus sequence Genotyping TSHR Polymorphisms by one nucleotide, and we modified the ST5 probe accordingly DNA from the gland donors was genotyped for the two SNPs (Supplementary Table 5). Reactions were run on an Abi Prism R (rs179247 and rs12101255) known to be strongly associated with 7900 HT instrument (Applied Biosystems) in triplicate and the GD and for one non-associated SNP (rs2288495), by real-time CT average was used for further statistical analysis. The CV was PCR using TaqMan R SNP genotyping assays (TaqMan R SNP always < 15%. Normalization of the results was performed in Genotyping Assay, Applied Biosystems, UK). accordance with the relative quantification 2-11Ct method (34) with respect to the constitutive expression of the GAPDH gene. Allele-Specific Quantification Using To display the results of the ST4 and ST5 isoform expression, we Massive Parallel Sequencing also used the ratio to flTSHR (35). [Next-Generation Sequencing [NGS]] To more precisely assess the effect of the rs179247 allele on the TSHR Detection in Thymocytes and Thymic overall TSHR transcription, we used massive parallel sequencing APCs by Indirect Immunofluorescence (NGS). DNA-free RNA samples from the thyroid and thymus To investigate how TSHR is presented in the thymus, were obtained as described. To preserve the pre-mRNA in the we stained thymic 5 µm cryostat sections with monoclonal complementary DNA (cDNA) synthesis step, retro-transcription antibodies (mAbs) specific for TSHR (clone 49, Thermofisher, of the RNA samples was performed using random primers (First Waltham, MA, USA), CD68 as a macrophage marker (clone Strand cDNA Synthesis Kit for RT-PCR [AMV], Roche, Basel, Y1/82A Biologend, San Diego, CA, USA), and CD11c as a Frontiers in Immunology | www.frontiersin.org 4 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance dendritic cell marker (clone CBR-p150/4G1, Thermofisher). spliced transcripts (ST4 and ST5) from the cDNA of 39 To avoid cross-reactivity, the Ig isotype or species-specific and 49 samples collected from the thymus and thyroid secondary labeled anti-sera were used. For each protocol, tissue, respectively. The level of flTSHR expression in the the controls included sections incubated with the conjugated thymus was higher than expected, which was ∼20% of secondary anti-sera without the corresponding primary antibody the expression observed in the thyroid (median thymus: to assess the background. Sections in which each of the 651 ± 580 vs. median thyroid: 2860 ± 1792) (Figure 2A). primary antibodies in the protocol were omitted to detect Although these expression levels are similar to those mentioned any possible cross-reactions. Samples from five different glands by Kim van der Weerd (36), they are higher than the (1 male and four female donors; age range: 10 months expression levels reported in several databases [e.g., BioGps to 3 years and 7 months) were used. The sections were (37) or EMBL-EBI Expression Atlas [https://www.ebi.ac.uk/ first examined under a UV photomicroscope equipped with gxa]]. Recently, this relatively high level of ST4 expression in the adequate filters and selected sections were examined the thymus was also reported by Latif et al., who designated under a confocal microscope (FV1000, Olympus Corporation, it TSHR v1.3 (16). Tokyo, Japan). Age was found to have an effect on TSHR expression in the thymus. The average level of flTSHR expression in donors aged 0 Statistical Analysis to 11 years-old was 1,135 ± 1,110 and from donors ages 40 to 80 Descriptive data were presented as the mean ± standard years old was 436 ± 657 (p < 0.05) (Figure 2B). Since our thymus deviation (SD) or median and semi interquartile range (SIR). samples were from relatively young patients, this may explain the The data were analyzed using a non-parametric test, U-Mann discrepancy with levels recorded in the databases. Whitney test, or Kruskal-Wallis test with a multiple comparisons It is of interest that the relative levels of flTSHR and correction (Dunn’s test), except for the analysis of the NGS ST4 expression were not as different as expected in the data, which followed a normal distribution (t-test). The accepted thymus and thyroid and they were higher for flTSHR than p-value is <0.05. We used GraphPad Prism 5.0 program for ST4 (thyroid flTSHR 2,860 ± 1,792, ST4 2,135 ± 961, (GraphPad Software, La Jolla CA) for the statistical analysis and ratio flTSHR/ST4 = 1.34; thymus flTSHR 651 ± 580, ST4 figure generation. 337 ± 299, ratio flTSHR/ST4 = 1.93). In contrast, the level of isoform ST5 expression was substantially lower in RESULTS both tissues (thyroid ST5 303 ± 104 and thymus ST5 15 ± 8). Interestingly, the level of ST5 expression was Analysis of TSHR Transcripts in the proportionally much lower in the thymus compared with Thymus and Thyroid the thyroid (ST5 represents 6% compared to 1.5% of the We analyzed the relative expression of full-length TSHR total transcripts (flTSHR+ ST4 + ST5) in the thyroid and (flTSHR) transcripts by qPCR and its two major alternatively thymus, respectively) (Figure 2A). This difference is also FIGURE 2 | Gene expression of the TSHR isoforms in the thymus and thyroid. (A) mRNA expression of the full-length TSHR (flTSHR) and its isoforms (ST4 and ST5) in the thymus and thyroid. The relative level of mRNA expression was measured using quantitative PCR of the cDNA from 49 thyroid and 39 thymus samples. Each sample was tested in triplicate and the standard deviation was always <15% of the mean value. Values were normalized to GAPDH expression using the comparative CT method and expressed as copies per 100,000 copies of GAPDH. Each point represents the mean value of triplicate results from one sample. The median ± IQR of each group is shown. ***p < 0.0001, Kruskal Wallis test with a Dunn’s multiple comparison test. (B) The relative mRNA expression of flTSHR in the thymus samples was stratified into pediatric (0–11 years; n = 31) and adults (40–80 years; n = 8) samples. The mean ± SEM of each group is shown. *p < 0.05 using a t-test for independent samples. Frontiers in Immunology | www.frontiersin.org 5 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance FIGURE 3 | Proportionally lower expression of the TSHR ST5 isoform in the thymus compared with the thyroid. The expression ratios of ST5 with flTSHR (A) and ST4 (B) in the thymus and thyroid were calculated. Each point represents the mean value of triplicate results from one sample. The median ± IQR of each group is shown. ***p < 0.0001 using a t-test for independent samples. reflected when comparing the ratios of ST5 expression with TABLE 1 | Allele-specific transcript quantification by NGS. flTSHR or ST4 in both tissues (thyroid flTSHR/ST5 = 9.86 Sample Depth of Balanceb Counts Frequency vs. thymus flTSHR/ST5 = 39.01; p < 0.0001) (Figure 3A) type coveragea (G allele) (G allele) % (thyroid ST4/ST5 = 4.5 vs. thymus ST4/ST5 = 24.3; p < 0.0001) (Figure 3B). THYROID The above results indicate that substantial soluble TSHR likely 102 gDNA 29548x 0.49 15,054 50.95 exists within the thymus, which is primarily derived from the cDNA 9305x 0.49 4,926 52.94 transcription of ST4. Thus, this expression may have implications 218 gDNA 25800x 0.49 13,283 51.48 for establishing central TSHR tolerance. cDNA 5006x 0.49 2,746 54.85 NGS Confirms the Thymus-Specific 355 gDNA 15027x 0.49 7,771 51.71 Influence of TSHR Expression by cDNA 9200x 0.50 4,940 53.70 GD-Associated SNPs 381 gDNA 25044x 0.49 12,633 50.44 We previously demonstrated that the risk allele, rs179247 (one cDNA 23579x 0.49 12,361 52.42 of the main TSHR GD-associated SNPs), determined lower 409 gDNA 25049x 0.50 12,726 50.80 TSHR expression in the thymus but not in the thyroid (38). cDNA 14221x 0.49 7,211 50.71 However, since we used a semiquantitative technique (allele- specific transcript quantification by qPCR using FRET probes), 430 gDNA 29863x 0.49 15,038 50.36 we decided to apply massive parallel sequencing, which is a cDNA 1693x 0.49 764 45.13 more reliable quantitative technique, for the comparison of 452 gDNA 32564x 0.49 16,484 50.62 rs179247 TSHR allele expression in heterozygous individuals. It cDNA 1738x 0.50 877 50.46 should be emphasized that since rs179247 is located in intron 1, we took care in obtaining the cDNA from the pre-mRNA 92 gDNA 20911x 0.49 10,717 51.25 (immature RNA before the splicing process). The designed cDNA 13439x 0.49 6,916 51.46 primers do not discriminate between cDNA and gDNA and THYMUS consequently, the contaminating gDNA must be absent to obtain 101 gDNA 28664 0.49 14,453 50.42 reliable results. cDNA 6933 0.49 3,960 57.12 Allele-specific quantification was measured in the gDNA 117 gDNA 35762 0.49 18,282 51.12 and the cDNA from 19 thymus and 8 thyroid samples from cDNA 5912 0.49 3,764 63.67 heterozygous donors (previously genotyped for rs179247). We obtained good coverage, with a mean depth in the gDNA samples 119 gDNA 24245 0.49 12,165 50.18 of 27,000× (i.e., 27,000 sequences or “reads” interrogating the cDNA 4570 0.49 2,663 58.27 rs179247 position) and 9,000× in the cDNA samples, with no 126 gDNA 28552 0.50 14,280 50.01 significant differences between the thymus and thyroid samples cDNA 4510 0.49 2,481 55.01 (Table 1). The results showed that the ratio between the G (protecting) and the A (predisposing) alleles for gDNA was (Continued) Frontiers in Immunology | www.frontiersin.org 6 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance TABLE 1 | Continued Sample Depth of Balanceb Counts Frequency type coveragea (G allele) (G allele) % 145 gDNA 33290 0.50 16,932 50.86 cDNA 17575 0.49 9,996 56.88 159 gDNA 25178 0.50 12,664 50.30 cDNA 6616 0.49 4,299 64.98 16 gDNA 22447 0.49 11,221 49.99 cDNA 13429 0.49 8,649 64.41 171 gDNA 30800 0.49 15,640 50.78 cDNA 4138 0.49 2,438 58.92 188 gDNA 29066 0.49 14,720 50.64 cDNA 342 0.48 185 54.09 195 gDNA 31457 0.49 16,255 51.67 cDNA 3656 0.50 2,235 61.13 197 gDNA 31543 0.49 16,169 51.26 cDNA 8058 0.49 4,385 54.42 FIGURE 4 | Preferential transcription of the rs179247 GD protective allele in the thymus. We used massive parallel sequencing to determine the relative 225 gDNA 26017 0.49 13,152 50.55 abundance of TSHR alleles in 19 thymus and 8 thyroid cDNA samples from cDNA 18406 0.49 11,164 60.65 individuals heterozygous for the GD-associated SNP, rs179247. Specific primers were designed for the intron 1 region of the TSHR gene containing the 229 gDNA 31049 0.50 15,955 51.39 analyzed SNP. Allele-specific quantification was measured in the gDNA, as a cDNA 7373 0.50 4,392 59.57 control for equal biallelic representation, and the cDNA in all samples. Each point represents the ratio between the G (protecting) and the A (predisposing) 235 gDNA 31949 0.49 16,325 51.10 allele reads. ***p < 0.001 using a Kruskal Wallis test with Dunn’s multiple cDNA 14662 0.49 8,516 58.08 comparison test. 263 gDNA 33124 0.49 16,852 50.88 cDNA 16402 0.49 9,478 57.79 GD-Associated Intron 1 SNPs Do Not 28 gDNA 27355 0.49 13,931 50.93 cDNA 6143 0.49 3,593 58.49 Influence Alternative TSHR Splicing in the Thyroid and Thymus 38 gDNA 4047 0.49 2,110 52.14 To assess the effect of intron 1 GD-associated SNPs on the cDNA 6701 0.50 4,253 63.47 differential transcription of alternatively spliced TSHR isoforms, 4 gDNA 25273 0.49 12,879 50.96 gDNA samples from donors (n = 88, 49 thyroid and 39 thymus cDNA 16769 0.49 9,802 58.45 donors) were genotyped for the rs179247 and rs12101255 SNPs associated with GD, and for the control SNP, rs2288495 (located 72 gDNA 31112 0.49 15,957 51.29 in the 3’ UTR of TSHR, not in linkage disequilibrium with cDNA 7426 0.49 4,277 57.59 rs179247; r = 0.05) (21, 38). The relative level of flTSHR, ST4, Results of cDNA samples are indicated in bold. and ST5 expression was measured in the thymus and thyroid by a The number of unique reads (i.e., sequences) which interrogate the position of rs179247. b Balance between reads from the forward or reverse primer. A value of 0.5 indicates that qPCR and normalized to the level of GAPDH expression. The half of the reads come from each primer (i.e., there is no strand bias). ratio of each of the truncated ST4 and ST5 transcripts to flTSHR was calculated for each individual and the data were grouped by SNP genotype (Figure 5). In the thyroid samples, there was no overall significant always 1 in both the thyroid and thymus, reflecting the existence effect of the genotypes on ST4 or ST5 expression, either of the same number of maternal and paternal DNA copies in normalized to GAPDH or expressed as a ratio to flTSHR each cell (Figure 4; Table 1). For cDNA, the percentage of reads (Figures 5A–F). Specifically, there were no statistically corresponding to the G allele in the thyroid was ∼50% (range: significant differences between the protective (GG for rs179247 45.1–54.8%) with a G/A allele ratio of 1 ± 0.1 SD. In contrast, the and CC for rs12101255) and the predisposing genotypes G/A allele ratio in the thymus was 1.5 ± 0.2 SD, with a mean of (AA for rs179247 and TT for rs12101255). These results 59.1% (ranging: 54.1–65%). These findings indicate that there is differ from those reported by Brand et al. in their original unbalanced allele transcription only in the thymus (reflecting a publication (21). tissue-specific effect) and demonstrates that the protective TSHR In the analysis of the relative expression of the ST4 G allele is preferentially transcribed. and ST5 transcripts in the thymic samples, there is no Frontiers in Immunology | www.frontiersin.org 7 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance FIGURE 5 | Effect of GD-associated SNPs (rs179247 and rs12101255) on the expression of TSHR alternatively spliced isoforms in human thyroid and thymus samples. The relative level of flTSHR, ST4, and ST5 expression was measured by qPCR and normalized to the level of GAPDH expression. The ratio of each short ST4 and ST5 transcript to the flTSHR was calculated for each individual and the data were grouped by SNP genotype. (A–F) The relative ST4 and ST5 expression in the thyroid is presented. (G–L) Relative level of ST4 and ST5 expression in the thymus is presented. A non-GD-associated SNP (rs2288495) was also included as a control. Each point represents the mean value of triplicate results from one sample. The mean ± SEM of each group is shown. A Kruskal-Wallis with Dunn’s multiple comparison test was used. overall significant effect of the genotypes for ST4 or ST5 no statistical differences were found for either of these models expression relative to flTSHR (Figures 5G–L). There are no (Supplementary Figure 1). statistically significant differences between the protective and the predisposing genotypes. In both the thymus and thyroid, the TSHR Is Associated With Both control SNP was not associated with GD (rs2288495), but was Thymocytes, Macrophages, and Dendritic also not significantly different. Cells in the Thymus In addition to the codominant and the most likely recessive It would be extremely interesting to elucidate how TSHR models considered above, we have also considered the dominant and its isoforms are processed and presented in the thymus, model for the rs179247G and rs21201255C alleles; however, as this should determine the level of central tolerance and Frontiers in Immunology | www.frontiersin.org 8 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance FIGURE 6 | TSHR protein expression in the thymus detected by immunofluorescence. Abundant cells in the cortex that can be morphologically identified as thymocytes and other larger cells (left bottom panel) are positive for TSHR visualized in the green channel. The bottom-right pictures combine the macrophage marker, CD68, and the dendritic cell marker, CD11c, reveal some orange-yellow cells that correspond to macrophages that contain TSHR. There are also some violet cells (probably DCs), for which is more difficult to identify TSHR staining but some cytoplasmic light blue fluorescence (from the overlapping of dark blue and green), is observed. The arrow points to a macrophage visible in all micrographs except the CD11c staining, and the arrowhead points to a dendritic cell visible in all micrographs except in the CD68 staining. more specifically, which epitopes escape this central tolerance. thymus, resulting in similar levels of ST4 and ST5 transcripts Since there are no TSHR isoform-specific antibodies available, between samples from individuals homozygous for the risk or after testing several reagents, we finally selected a mAb that protective genotype. resulted in clear staining in the thyroid and low background This is the first study to simultaneously address the two in the lymphoid tissue (Supplementary Figure 2). As seen in mechanisms proposed to explain the functional role of TSHR Figure 6, the thymocytes located in the inner cortical areas of intron 1 GD-associated SNPs. The mechanism, which involves the thymus are clearly stained for TSHR. Interestingly, some defective central tolerance, was initially proposed by our group. positive dots were observed, indicative of small quantities of In our previous study, we showed that individuals with the TSHR associated with both CD68+ macrophages and CD11c+ rs179247 GD-protective genotype displayed higher levels of dendritic cells. thymic TSHR expression than those with the disease-associated genotype (38). This finding has been also later reported by DISCUSSION Yaron Tomer’s group (28). In both studies, the results were obtained measuring gene expression by qPCR in genotyped In this article, we report three significant findings: (1) relatively individuals; however, since this approach is vulnerable to high expression of the short ST4 and ST5 TSHR transcripts was individual confounding factors that can influence the results observed in the thymus, which may have implications for the (e.g., individual genetic background, age, and sex differences), establishment of central TSHR tolerance; (2) massive parallel we used an allele-specific quantification (ASQ) method to sequencing was used to definitively correlate the differential effect quantify the contribution of each allele (risk vs. protective) of GD-associated rs179247 SNP alleles on TSHR transcription in heterozygous individuals. This ASQ method was a qPCR- in the thymus but not in the thyroid; and (3) there was an based semiquantitative technique and confirmed higher levels absence of an effect of GD-associated SNPs (rs179247 and of protective allele expression in the thymus (38). In the rs12101255) in modulating mRNA splicing in the thyroid and present study, we aimed to definitely corroborate these results Frontiers in Immunology | www.frontiersin.org 9 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance using a more robust quantitative technique of massive parallel TSHR receptor is only minimally expressed by medullary thymic sequencing (NGS). The high on-target coverage that was epithelial cells (mTECs), it is expressed at relatively high levels obtained (>9,000 reads in cDNA samples) allowed us to precisely by double positive thymocytes themselves, for which it appears quantify the number of transcripts expressed from each allele in to play a role in maturation and differentiation (9, 36, 38, 41). heterozygous individuals. We confirmed the unbalanced tissue- Therefore, it should be assumed that there are two sources of specific transcription of the two alleles with the observation TSHR protein in the thymus: (1) the TSHR anchored in the that the GD-protective allele was expressed 1.5-fold more on thymocyte membrane and the soluble ST4 and ST5 isoforms average, compared to the GD-predisposing allele in the thymus, generated by alternative splicing. We and others consider that the but not in the thyroid. Although it remains unclear how these ST4 and ST5 isoforms are translated (ST4 and ST5 mRNA have moderate differences in the expression of a self-antigen in the all the features of mature transcripts), even if no experimental thymus have such an important effect on establishing the level evidence is available. According to the transcript levels, soluble of central tolerance, they may be related to the limited window short TSHR isoforms may be present at levels comparable to of opportunity that maturing thymocytes have to interact with that of membrane-anchored TSHR. It would be important to tissue-restricted antigens in the thymus (39). The magnitude of determine which of these two forms (soluble vs. membrane- this imbalance here reported by NGS is in the same order of anchored) play the main role in the establishment of central magnitude of that detected by ourselves using ASQ by qPCR tolerance to TSHR. It is difficult to answer to this question (38), and of that reported by Pugliese et al. and Vafiadis et with certainty given the present data and reagents. As previously al. for the insulin gene in the first publications describing this shown by van der Weerd (36) and ourselves (41), as well as by phenomenon (25, 40). our IFL experiments in this paper, TSHR is mainly expressed in The second mechanism used to explain the functional role of double positive thymocytes; however, low levels are also observed TSHR intron 1 GD-associated SNPs was proposed by Brand et in both macrophages and dendritic cells. There are currently no al. and highlighted the differential regulation of mRNA splicing reagents that can be used to identify the isoforms detected by IFLs by these SNPs. The authors showed that GD-risk alleles of in these APCs. Massive quantities of double positive thymocytes intron 1 SNPs (rs179247 and rs12101255) were associated with die by apoptosis in the thymus cortical compartment. It is known a relative increase in ST4 and ST5 expression, which would result that they are disposed by macrophages that ingest apoptotic cells. in a higher production of variants encoding putatively more Importantly, macrophages are much less efficient than mTEC or antigenic soluble TSHR isoforms (21). In our study, when we thymic dendritic cells at inducing tolerance to self-autoantigens, attempted to replicate this experiment, rs179247 and rs12101255 which should apply to thymocyte antigens (42, 43). In contrast, did not affect the transcription of the ST4 and ST5 isoforms soluble TSHR isoforms are likely engaged by thymic dendritic in the thyroid or thymus. Although these results differ with cells, which are highly efficient at inducing negative selection. those reported by Brand et al. for an unknown reason, this If central tolerance to TSHR is predominantly dependent on discrepancy may be the result of our study involving a much the two soluble ST4 and ST5 isoforms, the C terminal segment larger series of thyroid samples. We used 49 thyroid samples consisting of 533 aa out of the 764 aa of the full length TSHR compared with the 12 samples used by Brand et al. In their molecule would not be subjected to the tolerization process. In study, when the 12 samples were divided by genotype, the result GD, although the pathogenic stimulating antibodies are directed was small groups consisting of only three to five samples each at the N terminal LRR in the ectodomain, tolerance is ultimately (one of which had a high dispersion of the values), which the dependent on T cells. Thus, B cells that recognize the ectodomain authors stated required further confirmation (21). Additionally, may receive help from T cells that recognize epitopes in the long the ST5 probe sequence described in Brand et al. had an incorrect C terminal stretch of TSHR, to which tolerance is incomplete. nucleotide when compared with the current reference genome. This is an important issue because a demonstrated mechanism Furthermore, it is difficult to predict if this mismatch had an of central tolerance failure is the expression of different protein effect on ST5 quantification. isoforms in both the thymus and periphery. This was first We believe that our results are robust due to the use of highly demonstrated by Klein and Kyewski in experimental acute reliable and quantitative methods (NGS for ASQ and probe- encephalomyelitis induced by the myelin proteolipid protein in specific qPCR for gene expression). The age, clinical conditions, SJL/J mice (44). Since this original description, similar results and presence of thyroid autoantibodies in the donors had no have also been demonstrated for the islet antigen, I-A2 (45), effect on the results, and they were excluded as confounding and is also used to explain autoantibodies to post-translationally factors (data not shown). Therefore, we can conclude that modified proteins (e.g., citrullinated peptides in rheumatoid TSHR intron 1 GD-linked SNPs are associated with different arthritis) (46). In addition, the role of TSHR expression in the levels of thymic TSHR expression, but not with differential thymus for protection against the development of pathogenic mRNA splicing. anti-TSHR antibodies has recently been demonstrated in a mouse Perhaps the most striking finding in the present study is model (47). Therefore, we propose that a crucial mechanism the demonstration of high levels of short TSHR transcript for the failure of TSHR tolerance is the lack of presentation expression, especially ST4, in the thymus [recently confirmed of the full-length TSHR molecule by tolerogenic APCs in the by Latif et al. (16)]. This is relevant if we consider the peculiar thymic medulla. distribution of TSHR expression within the thymus. Different Collectively, the results of the present study support from most other restricted tissue antigens (RTAs), while the the functional role of TSHR intron 1 GD-associated SNPs Frontiers in Immunology | www.frontiersin.org 10 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance in modulating central tolerance through influencing the experiments. AL-M and OG reviewed the clinical and surgical intrathymic expression of TSHR. In addition, these findings data in patient records. RC and RP-B were responsible for provide a novel explanation as to why the loss of tolerance designing the study, checking the experimental protocols and to TSHR occurs with relatively high frequency (i.e., differential results and writing the manuscript, and approving the final draft. expression of isoforms in the thymus vs. thyroid). Furthermore, a deeper analysis of the differential isoform expression (i.e., FUNDING measurements of the corresponding proteins, the presence of TSHR peptides in the thymic ligandome) and epitope specificity This study was funded by Instituto de Salud Carlos III, grants of TSHR-specific autoreactive T cells should be assessed in PI14/00848, and PI17/00324, co-financed by the European future studies. Regional Development Fund (ERDF). DÁ-S is in recipient of a predoctoral fellowship from the Vall d’Hebron Research DATA AVAILABILITY Institute (VHIR). This manuscript contains previously unpublished data. The ACKNOWLEDGMENTS name of the repository and accession number(s) are not available. The authors are grateful to the patients and their parents ETHICS STATEMENT that agreed to donate part of the thymus and thyroid surgical specimen for research projects. We are also very grateful to the This study was carried out in accordance with the heart surgery teams at Pediatric Hospital Vall d’Hebron (Dr. recommendations of local institutional ethics review board Raul Abella) and Hospital Germans Trias i Pujol (Dr. X. Ruyra) of the participating institutions, with written informed consent and the endocrine surgery teams at Hospital Vall d’Hebron Dr. from all subjects. All subjects gave written informed consent in J. M. Balibrea, Dr. Enric Caubet, Dr. J. M. Fort and Dr. Aroa accordance with the Declaration of Helsinki. The protocol was Gómez and Dr. Irene Bello from the Surgery Department for the approved by Vall d’Hebron Hospital (HUVH) Research Ethics procurement of thyroid samples; Dr. Teresa Pont, the transplant Comittee (Approval reference PR AG-145/2011). coordinator for organizing the circuits for organ donor tissue collection. We thank ScienceDocs Inc. (https://www.sciencedocs. AUTHOR CONTRIBUTIONS com) for language editing. AM-S collected the samples, performed the qPCR experiments, SUPPLEMENTARY MATERIAL analyzed the data, and wrote part of the manuscript. AS-S contributed to the collection and processing of the samples, as The Supplementary Material for this article can be found well as the qPCR experiments. FR and EE performed the ASQ online at: https://www.frontiersin.org/articles/10.3389/fimmu. NGS experiments. DÁ-S performed the immunofluorescence 2019.01695/full#supplementary-material REFERENCES 8. Dai YD, Rao VP, Carayanniotis G. Enhanced iodination of thyroglobulin facilitates processing and presentation of a cryptic pathogenic peptide. J 1. Rapoport B, McLachlan SM. The thyrotropin receptor in Graves’ disease. Immunol. (2002) 168:5907–11. doi: 10.4049/jimmunol.168.11.5907 Thyroid. (2007) 17:911–22. doi: 10.1089/thy.2007.0170 9. Pujol-Borrell R, Álvarez-Sierra D, Jaraquemada D, Marín-Sánchez A, 2. Smith TJ, Hegedüs L. Graves’ disease. N Engl J Med. (2016) 375:1552–65. Colobran R. Central tolerance mechanisms to TSHR in Graves’ disease: doi: 10.1056/NEJMra1510030 contributions to understand the genetic association. Horm Metab Res. (2018) 3. Brix TH, Kyvik KO, Christensen K, Hegedüs L. Evidence for a 50:863–70. doi: 10.1055/a-0755-7927 major role of heredity in graves’ disease: a population-based study of 10. Stefan M, Faustino LC. Genetics of thyroid-stimulating hormone receptor— two danish twin cohorts. J Clin Endocrinol Metab. (2001) 86:930–4. relevance for autoimmune thyroid disease. Front Endocrinol. (2017) 8:1–6. doi: 10.1210/jcem.86.2.7242 doi: 10.3389/fendo.2017.00057 4. Brand OJJ, Gough SCLCL. Genetics of thyroid autoimmunity and 11. Kursawe R, Paschke R. Modulation of TSHR signaling by posttranslational the role of the TSHR. Mol Cell Endocrinol. (2010) 322:135–43. modifications. Trends Endocrinol Metab. (2007) 18:199–207. doi: 10.1016/j.mce.2010.01.013 doi: 10.1016/j.tem.2007.05.002 5. Pujol-Borrell R, Giménez-Barcons M, Marín-Sánchez A, Colobran R. 12. Rapoport B, McLachlan SM. TSH receptor cleavage into subunits and Genetics of Graves’ disease: special focus on the role of TSHR gene. Horm shedding of the A-subunit; a molecular and clinical perspective. Endocr Rev. Metab Res. (2015) 47:753–66. doi: 10.1055/s-0035-1559646 (2016) 37:114–34. doi: 10.1210/er.2015-1098 6. Lee HJ, Li CW, Hammerstad SS, Stefan M, Tomer Y. Immunogenetics of 13. Chazenbalk GD, Tanaka K, Nagayama Y, Kakinuma A, Jaume JC, autoimmune thyroid diseases: a cromprehensive review. J Autoimmun. (2014) Mclachlan SM, et al. Evidence that the thyrotropin receptor ectodomain 64:82–90. doi: 10.1016/j.jaut.2015.07.009 contains not one, but two, cleavage sites. Endocrinology. (1997) 138:2893–9. 7. Ban Y, Greenberg DA, Concepcion E, Skrabanek L, Villanueva R, Tomer doi: 10.1210/endo.138.7.5259 Y. Amino acid substitutions in the thyroglobulin gene are associated 14. Kleinau G, Worth CL, Kreuchwig A, Biebermann H, Marcinkowski P, with susceptibility to human and murine autoimmune thyroid disease. Scheerer P, et al. Structural–functional features of the thyrotropin receptor: Proc Natl Acad Sci USA. (2003) 100:15119–24. doi: 10.1073/pnas.24341 a class A G-protein-coupled receptor at work. Front Endocrinol. (2017) 8:86. 75100 doi: 10.3389/fendo.2017.00086 Frontiers in Immunology | www.frontiersin.org 11 July 2019 | Volume 10 | Article 1695 Marín-Sánchez et al. TSHR Isoforms and Tolerance 15. Chen CR, Chazenbalk GD, Wawrowsky KA, McLachlan SM, Rapoport 33. Sospedra M, Ferrer-Francesch X, Domínguez O, Juan M, Foz-Sala M, Pujol- B. Evidence that human thyroid cells express uncleaved, single-chain Borrell R. Transcription of a broad range of self-antigens in human thymus thyrotropin receptors on their surface. Endocrinology. (2006) 147:3107–13. suggests a role for central mechanisms in tolerance toward peripheral doi: 10.1210/en.2005-1514 antigens. J Immunol. (1998) 161:5918–29. 16. Latif R, Mezei M, Morshed SA, Ma R, Ehrlich R, Davies TF. A 34. Wong ML, Medrano JF. Real-time PCR for mRNA quantitation. modifying autoantigen in Graves’ disease. Endocrinology. (2019) 160:1008–20. Biotechniques. (2005) 39:75–85. doi: 10.2144/05391RV01 doi: 10.1210/en.2018-01048 35. Derveaux S, Vandesompele J, Hellemans J. How to do successful gene 17. Fernando R, Lu Y, Atkins SJ, Mester T, Branham K, Smith TJ. Expression expression analysis using real-time PCR. Methods. (2010) 50:227–30. of thyrotropin receptor, thyroglobulin, sodium-iodide symporter, and doi: 10.1016/j.ymeth.2009.11.001 thyroperoxidase by fibrocytes depends on AIRE. J Clin Endocrinol Metab. 36. Bogers AJJC, van Hagen PM, Schrijver B, Visser TJ, van der Weerd K, van der (2014) 99:1236–44. doi: 10.1210/jc.2013-4271 Lelij A-J, et al. Thyrotropin acts as a T-cell developmental factor in mice and 18. Hunt N, Willey KP, Abend N, Balvers M, Jähner D, Northemann W, humans. Thyroid. (2014) 24:1051–61. doi: 10.1089/thy.2013.0396 et al. Novel splicing variants of the human thyrotropin receptor encode 37. Wu C, Orozco C, Boyer J, Leglise M, Goodale J, Batalov S, et truncated polypeptides without a membrane-spanning domain. Endocrine. al. BioGPS: an extensible and customizable portal for querying and (1995) 3:233–40. doi: 10.1007/BF02994449 organizing gene annotation resources. Genome Biol. (2009) 10:R130. 19. Graves PN, Tomer Y, Davies TF. Cloning and sequencing of a doi: 10.1186/gb-2009-10-11-r130 1.3 KB variant of human thyrotropin receptor mRNA lacking the 38. Colobran R, Armengol Mdel P, Faner R, Gärtner M, Tykocinski LO, Lucas transmembrane domain. Biochem Biophys Res Commun. (1992) 187:1135–43. A, et al. Association of an SNP with intrathymic transcription of TSHR and doi: 10.1016/0006-291X(92)91315-H graves’ disease: a role for defective thymic tolerance. Hum Mol Genet. (2011) 20. Takeshita A, Nagayama Y, Fujiyama K, Yokoyama N, Namba H, Yamashita S, 20:3415–23. doi: 10.1093/hmg/ddr247 et al. Molecular cloning and sequencing of an alternatively spliced form of the 39. Brennecke P, Reyes A, Pinto S, Rattay K, Nguyen M, Küchler R, et al. Single- human thyrotropin receptor transcript. Biochem Biophys Res Commun. (1992) cell transcriptome analysis reveals coordinated ectopic gene-expression 188:1214–9. doi: 10.1016/0006-291X(92)91360-3 patterns in medullary thymic epithelial cells. Nat Immunol. (2015) 16:933–41. 21. Brand OJ, Barrett JC, Simmonds MJ, Newby PR, McCabe CJ, Bruce doi: 10.1038/ni.3246 CK, et al. Association of the thyroid stimulating hormone receptor 40. Pugliese A, Zeller M, Fernandez A, Zalcberg LJ, Bartlett RJ, Ricordi C, et al. gene (TSHR) with Graves’ disease. Hum Mol Genet. (2009) 18:1704–13. The insulin gene is transcribed in the human thymus and transcription levels doi: 10.1093/hmg/ddp087 correlated with allelic variation at the INS VNTR-IDDM2 susceptibility locus 22. Chen CR, Pichurin P, Nagayama Y, Latrofa F, Rapoport B, McLachlan SM. The for type 1 diabetes. Nat Genet. (1997) 15:293–7. doi: 10.1038/ng0397-293 thyrotropin receptor autoantigen in Graves disease is the culprit as well as the 41. Giménez-Barcons M, Colobran R, Gómez-Pau A, Marín-Sánchez A, Casteràs victim. J Clin Invest. (2003) 111:1897–904. doi: 10.1172/JCI17069 A, Obiols G, et al. Graves’ disease TSHR-Stimulating Antibodies (TSAbs) 23. Kappler JW, Roehm N, Marrack P. T cell tolerance by clonal elimination in induce the activation of immature thymocytes: a clue to the riddle of TSAbs the thymus. Cell. (1987) 49:273–80. doi: 10.1016/0092-8674(87)90568-X generation? J Immunol. (2015) 194:4199–206. doi: 10.4049/jimmunol.1500183 24. Cheng M, Anderson MS. Thymic tolerance as a key brake on autoimmunity. 42. Szondy Z, Garabuczi É, Tóth K, Kiss B, Köröskényi K. Thymocyte death Nat Immunol. (2018) 19:659–64. doi: 10.1038/s41590-018-0128-9 by neglect: Contribution of engulfing macrophages. Eur J Immunol. (2012) 25. Vafiadis P, Bennett ST, Todd JA, Nadeau J, Grabs R, Goodyer CG, et al. Insulin 42:1662–7. doi: 10.1002/eji.201142338 expression in human thymus is modulated by INS VNTR alleles at the IDDM2 43. Guerri L, Peguillet I, Geraldo Y, Nabti S, Premel V, Lantz O. Analysis locus. Nat Genet. (1997) 15:289–92. doi: 10.1038/ng0397-289 of APC types involved in CD4 tolerance and regulatory T cell generation 26. Sabater L, Ferrer-Francesch X, Sospedra M, Caro P, Juan M, Pujol-Borrell using reaggregated thymic organ cultures. J Immunol. (2013) 190:2102–10. R. Insulin alleles and autoimmune regulator (AIRE) gene expression both doi: 10.4049/jimmunol.1202883 influence insulin expression in the thymus. J Autoimmun. (2005) 25:312–8. 44. Klein L, Klugmann M, Nave K-AA, Tuohy VK, Kyewski B. Shaping of the doi: 10.1016/j.jaut.2005.08.006 autoreactive T-cell repertoire by a splice variant ofself protein expressed in 27. Malhotra D, Linehan JL, Dileepan T, Lee YJ, Purtha WE, Lu JV, et al. thymic epithelial cells. Nat Med. (2000) 6:56–61. doi: 10.1038/71540 Tolerance is established in polyclonal CD4 + T cells by distinct mechanisms, 45. Diez J, Park Y, Zeller M, Brown D, Garza D, Ricordi C, et al. Differential according to self-peptide expression patterns. Nat Immunol. (2016) 17:187–95. splicing of the IA-2 mRNA in pancreas and lymphoid organs as a permissive doi: 10.1038/ni.3327 genetic mechanism for autoimmunity against the IA-2 type 1 diabetes 28. Stefan M, Wei C, Lombardi A, Li CW, Concepcion ES, Inabnet WB, et al. autoantigen. Diabetes. (2001) 50:895–900. doi: 10.2337/diabetes.50.4.895 Genetic-epigenetic dysregulation of thymic TSH receptor gene expression 46. Raposo B, Merky P, Lundqvist C, Yamada H, Urbonaviciute V, triggers thyroid autoimmunity. Proc Natl Acad Sci USA. (2014) 111:12562–7. Niaudet C, et al. T cells specific for post-translational modifications doi: 10.1073/pnas.1408821111 escape intrathymic tolerance induction. Nat Commun. (2018) 9:353. 29. Ruiz-Riol M, Armengol Barnils Mdel P, Colobran Oriol R, Sánchez Pla A, doi: 10.1038/s41467-017-02763-y Borràs Serres FE, Lucas-Martin A, et al. Analysis of the cumulative changes 47. McLachlan SM, Aliesky HA, Banuelos B, Lesage S, Collin R, Rapoport in Graves’ disease thyroid glands points to IFN signature, plasmacytoid B. High-level intrathymic thyrotrophin receptor expression in thyroiditis- DCs and alternatively activated macrophages as chronicity determining prone mice protects against the spontaneous generation of pathogenic factors. J Autoimmun. (2011) 36:189–200. doi: 10.1016/j.jaut.2011. thyrotrophin receptor autoantibodies. Clin Exp Immunol. (2017) 188:243–53. 01.002 doi: 10.1111/cei.12928 30. Yin X, Sachidanandam R, Morshed S, Latif R, Shi R, Davies TF. MRNA- Seq reveals novel molecular mechanisms and a robust fingerprint Conflict of Interest Statement: The authors declare that the research was in Graves’ disease. J Clin Endocrinol Metab. (2014) 99:E2076–83. conducted in the absence of any commercial or financial relationships that could doi: 10.1210/jc.2014-1735 be construed as a potential conflict of interest. 31. Tomer Y, Blackard JT, Akeno N. Interferon alpha treatment and thyroid dysfunction. Endocrinol Metab Clin North Am. (2007) 36:1051–66; x–xi. Copyright © 2019 Marín-Sánchez, Álvarez-Sierra, González, Lucas-Martin, Sellés- doi: 10.1016/j.ecl.2007.07.001 Sánchez, Rudilla, Enrich, Colobran and Pujol-Borrell. This is an open-access article 32. Armengol MP, Juan M, Lucas-Martín A, Fernández-Figueras MT, distributed under the terms of the Creative Commons Attribution License (CC BY). Jaraquemada D, Gallart T, Pujol-Borrell R. Thyroid autoimmune The use, distribution or reproduction in other forums is permitted, provided the disease: demonstration of thyroid antigen-specific B cells and original author(s) and the copyright owner(s) are credited and that the original recombination-activating gene expression in chemokine-containing publication in this journal is cited, in accordance with accepted academic practice. active intrathyroidal germinal centers. Am J Pathol. (2001) 159:861–73. No use, distribution or reproduction is permitted which does not comply with these doi: 10.1016/S0002-9440(10)61762-2 terms. Frontiers in Immunology | www.frontiersin.org 12 July 2019 | Volume 10 | Article 1695