A MicroRNA Signature for Evaluation of Risk and Severity of Autoimmune Thyroid Diseases

A MicroRNA Signature for Evaluation of Risk and Severity of Autoimmune Thyroid Diseases Abstract Context Circulating microRNAs (miRNAs) are emerging as an interesting research area because of their potential role as novel biomarkers and therapeutic targets. Their involvement in autoimmune thyroid diseases (AITDs) has not been fully explored. Objective To compare the expression profile of miRNAs in thyroid tissue from patients with AITD and controls, using next-generation sequencing, further validated our findings in thyroid and serum samples. Design Twenty fresh-frozen thyroid tissues (15 from patients with AITD and 5 from controls) were used for miRNA next-generation sequencing. Thirty-six thyroid samples were recruited for the qRT-PCR validation test and 58 serum samples for further validation in peripheral blood. Results Expression of several miRNAs that had been previously associated with relevant immunological functions was significantly dysregulated. Specifically, eight differentially expressed miRNAs (miR-21-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, and miR-766-3p) were confirmed using qRT-PCR in thyroid samples, and three had the same behavior in tissue and serum samples (miR-21-5p, miR-142-3p, and miR-146a-5p). Furthermore, when the expression of these miRNAs was assessed together with five additional ones previously related to AITD in peripheral blood, the expression of five (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) was significantly expressed in AITD and, in patients with Graves disease (GD), was correlated with a higher severity of disease, including active ophthalmopathy, goiter, higher antibody titers, and/or higher recurrence rates. Conclusions The present findings identify a serum five-signature miRNA that could be an independent risk factor for developing AITD and a predisposition of a worse clinical picture in patients with GD. Autoimmune thyroid disorders (AITDs) result from a dysregulation of the immune system directed against the thyroid and include a broad spectrum of diseases. The two main phenotypes of AITD, Hashimoto thyroiditis (HT) and Graves disease (GD), are both characterized by the presence of circulating thyroid antibodies and infiltration by autoreactive lymphocytes in the thyroid gland and sometimes the orbit. It has been traditionally thought that HT is mainly mediated by a cellular autoimmune response, whereas GD mainly has been considered to be mediated by a humoral response, mainly due to the presence of autoantibodies directed against the thyrotropin receptor [antithyrotropin receptor antibody (TSHR-Ab)], which lead to development of goiter and hyperthyroidism (1–3). Moreover, both immune responses, including TSHR-Ab, also have been reported in the pathogenesis of Graves ophthalmopathy (GO), one of the most common extrathyroid manifestations of AITD, which may occur clinically in up to 25% of patients with GD (4). The pathogenesis of AITD is probably related to a complex and multifactorial interplay of specific susceptibility genes and environmental exposures, leading to the breakdown of self-tolerance and subsequent development of autoimmune diseases. MicroRNAs (miRNAs) are small non-protein-coding RNA molecules of ∼22 nucleotides that interact with their targeted RNA in a sequence-dependent manner and therefore function as regulators of gene expression at a posttranscriptional level mainly by repressing the translation and also by decreasing target messenger RNA levels (5). Currently, nearly 3000 human miRNAs have been annotated in the miRBase (6). miRNAs have been recently involved in several biological processes, including immune functions, cellular apoptosis, cell differentiation and development, proliferation, and metabolism. The increasing importance of miRNAs highlights their potential role as biomarkers of disease and even their utility as therapeutic targets. In recent years, circulating miRNAs have been an interesting area of research due to their stability and reproducibility (7), and studies describing the role of miRNAs in the immune response and in autoimmune diseases have progressively developed. In this regard, evidence of differential miRNA expression has been reported in numerous immunological disorders such as rheumatoid arthritis, systemic lupus erythematosus, Sjögren disease, and psoriasis, among others (7–11). Several reports have been published over the past years regarding the specific scenario of miRNAs and AITD with discordant results (12–27). These discrepancies can be attributed to the differences between the techniques used [microarray and/or quantitative reverse transcription polymerase chain reaction (qRT-PCR) assays], the materials analyzed (peripheral blood mononuclear cells, serum, plasma, thyrocytes, orbital fibroblasts, and thyroid tissue), the different preservation methods (frozen vs paraffin embedded), and the specific type of individual AITD evaluated (all AITD, HT, GD, or GO). However, none of these previous studies have analyzed miRNAs in fresh thyroid tissue, in the whole spectrum of AITD, using next-generation sequencing (NGS). The interest in this latter technique lies in the fact that, in comparison with microarrays, NGS has a greater ability to capture the scale and complexity of whole transcriptomes. Furthermore, NGS allows the discovery of novel miRNAs and can identify miRNAs that are expressed at levels that fall below microarrays’ detectable threshold. In this study, therefore, we aimed to identify differentially expressed (DE) miRNAs in AITD thyroid tissue samples in comparison with controls by means of NGS and then corroborated their expression in serum samples. We then correlated their expression with patients’ clinical data to identify potential diagnostic biomarkers and therapeutic targets of the disease. Materials and Methods Patients Fresh-frozen thyroid tissue samples from 26 patients with AITD (9 with HT, 7 with GD without GO, and 10 with GD and GO) and 10 controls were collected. In addition, we collected formalin-fixed, paraffin-embedded (FFPE) samples from 19 patients with AITD (7 with HT, 6 with GD without GO, and 6 with GD and GO), of which 11 were the same as the frozen tissue, and 6 were healthy controls. Concomitantly, we also evaluated serum samples from 36 patients with AITD (14 with HT, 10 with GD without GO, and 12 with GD with GO) and 22 healthy controls. Clinical diagnoses were all reviewed by a single experienced endocrinologist; established on commonly accepted clinical, laboratory, and histological criteria (2); and based on information collected from clinical records (Tables 1 and 2). More details of clinical data are available in the Supplemental Material and Methods. Table 1. Clinical Features of Patients With AITD From Whom Thyroid Tissue Was Collected Characteristic  HT (n = 9)  GD (n = 17)  Controls (n = 10)  Sex, female/male  9/0  15/2  7/3  Age, y  63 (50.5–71.5)  39 (29.5–52.5)  43 (32.2–51.8)  TSH, μU/mL  2.3 (1–8.5)  1.8 (0.2–9.5)  0  FT4, ng/dL  1.1 (0.9–1.2)  1.1 (0.8–1.3)  0  Tg-Ab, UI/mL  311 (213–382.3)  456 (20–1103)  0  TPO-Ab, UI/mL  150 (20–306)  102 (43.2–635.5)  0  TSHR-Ab, U/L  0.5 (0.5–0.5)  5.6 (1.7–17.5)  0  Characteristic  HT (n = 9)  GD (n = 17)  Controls (n = 10)  Sex, female/male  9/0  15/2  7/3  Age, y  63 (50.5–71.5)  39 (29.5–52.5)  43 (32.2–51.8)  TSH, μU/mL  2.3 (1–8.5)  1.8 (0.2–9.5)  0  FT4, ng/dL  1.1 (0.9–1.2)  1.1 (0.8–1.3)  0  Tg-Ab, UI/mL  311 (213–382.3)  456 (20–1103)  0  TPO-Ab, UI/mL  150 (20–306)  102 (43.2–635.5)  0  TSHR-Ab, U/L  0.5 (0.5–0.5)  5.6 (1.7–17.5)  0  Values show number for categorical values and median (25th to 75th interquartile intervals) for continuous variables. FT4, free thyroxine (normal range, 0.93 to 1.7 ng/dL); Tg-Ab (negative <344 UI/mL); TPO-Ab (negative <100 UI/mL); TSH, thyrotropin (normal range, 0.27 to 4.20 mU/mL); TSHR-Ab (negative <0.7 U/L). View Large Table 2. Clinical Features of Patients With AITD From Whom Serum Was Collected Characteristic  HT (n = 14)  GD (n = 22)  Controls (n = 22)  Euthyroid (n = 8)  Hypothyroid (n = 6)  Hyperthyroid (n = 15)  Hypothyroid (n = 7)  Sex, female/male  6/2  4/2  11/4  7/0  9/13  Age, y  47 (31.2–62.7)  39 (28.5–54.2)  45 (30–54)  46 (41–54)  31 (25.5–40)  Ophthalmopathy  0  0  8  4  0  TSH, μU/mL  2.5 (1.5–2.9)  10.7 (4.9–23.9)  0  6.55 (3.21–23.96)  —  FT4, ng/dL  1.2 (0.9–1.3)  1.2 (0.9–1.4)  3 (1.9–4.5)  0.61 (0.36–1)  —  Tg-Ab, UI/mL  0 (0–986.8)  398 (0–1198)  0 (0–764)  0 (0–70.2)  0  TPO-Ab, UI/mL  424 (126.3–844.8)  252 (0–669.5)  156 (0–542.8)  98.6 (0–273)  0  TSHR-Ab, U/L  0.2 (0–1.1)  0 (0–0.8)  3.4 (1.9–5.6)  3.56 (1.66–6.3)  0.01 (0–0.07)  Characteristic  HT (n = 14)  GD (n = 22)  Controls (n = 22)  Euthyroid (n = 8)  Hypothyroid (n = 6)  Hyperthyroid (n = 15)  Hypothyroid (n = 7)  Sex, female/male  6/2  4/2  11/4  7/0  9/13  Age, y  47 (31.2–62.7)  39 (28.5–54.2)  45 (30–54)  46 (41–54)  31 (25.5–40)  Ophthalmopathy  0  0  8  4  0  TSH, μU/mL  2.5 (1.5–2.9)  10.7 (4.9–23.9)  0  6.55 (3.21–23.96)  —  FT4, ng/dL  1.2 (0.9–1.3)  1.2 (0.9–1.4)  3 (1.9–4.5)  0.61 (0.36–1)  —  Tg-Ab, UI/mL  0 (0–986.8)  398 (0–1198)  0 (0–764)  0 (0–70.2)  0  TPO-Ab, UI/mL  424 (126.3–844.8)  252 (0–669.5)  156 (0–542.8)  98.6 (0–273)  0  TSHR-Ab, U/L  0.2 (0–1.1)  0 (0–0.8)  3.4 (1.9–5.6)  3.56 (1.66–6.3)  0.01 (0–0.07)  Values show number for categorical values and median (25th to 75th interquartile intervals) for continuous variables. FT4, free thyroxine (normal range, 0.93 to 1.7 ng/dL); Tg-Ab (negative <344 UI/mL); TPO-Ab (negative <100 UI/mL); TSH, thyrotropin (normal range, 0.27 to 4.20 mU/mL); TSHR-Ab (negative <0.7 U/L). View Large The project was approved by the Internal Ethical Review Committee of the Hospital de la Princesa, and written informed consent was obtained from all patients prior to inclusion, in accordance with the Declaration of Helsinki. Tissue and serum samples All thyroid tissues were obtained from surgical thyroidectomies performed in our hospital and reviewed by an experienced pathologist. Samples were immediately snap-frozen in liquid nitrogen–cooled isopentane and transferred to a −80°C freezer for long-time preservation or processed by the Department of Pathology to create FFPE tissues. Serum samples were isolated by centrifugation (1500 × g) from 10 mL of total blood and stored at −80°C until use. Tissue miRNA isolation RNA was isolated from 36 fresh-frozen thyroid tissues: 20 samples were used for miRNA NGS and all 36 for the qRT-PCR validation test. Total RNA was isolated using miRNeasy Mini Kit (Qiagen, Germantown, MD), according to the manufacturer’s instructions. The quality and quantity of RNA and miRNA were assessed in a 2100 Bioanalyzer by using a RNA 6000 Nano kit and a Small RNA kit (Agilent Technologies, Palo Alto, CA). For the study, only those samples with an RNA integrity number >7 were included. To empower the statistical analysis, we isolated miRNA from 25 FFPE samples, 12 of which were repeated from frozen samples to verify their quality and reproducibility. FFPE samples were isolated using the miRneasy FFPE Mini kit (Qiagen). Serum miRNA isolation To test hemolysis in serum samples, the absorbance of free hemoglobin at 414 nm was measured. Those samples with a peak >0.2 were discarded. miRNA from serum samples were isolated using miRCURY RNA Isolation Kit Biofluids (Exiqon, Vedbaek, Denmark), according to the manufacturer’s instruction. A mixture containing 1.25 μg/mL MS2 bacteriophage RNA and RNA spike-ins (UniSp2, UniSp4, and UniSp5) was added to 200 μL serum. RNA was purified on a miRNA mini-spin column BF, eluted in 50 μL RNase-free water, and stored at −80°C. The robustness of the RNA isolation process and the quality of isolated miRNA were assessed using miRCURY microRNA QC Panels (Exiqon). miRNA NGS NGS of 20 thyroid samples (5 controls, 5 with HT, 5 with GD without GO, and 5 with GD with GO) was performed on the Illumina platform (HiSEquation 2000; Illumina, San Diego, CA) at Sistemas Genómicos (Valencia, Spain). The procedure included the following steps/phases: (1) Quality control of RNA: the quality and the quantity of RNA were determined in a Bioanalyzer 2100 Small RNA assay and Qubit 2.0 fluorometer. (2) Preparation of libraries: complementary DNA (cDNA) libraries were obtained following Illumina’s recommendations. Briefly, 3′ and 5′ adaptors were sequentially ligated to the RNA prior to reverse transcription and cDNA generation. cDNA was enriched with polymerase chain reaction (PCR) to create the indexed double-stranded cDNA library. Size selection was performed using 6% polyacrylamide gel. The quality of the libraries was analyzed using Bioanalyzer 2100, High Sensitivity assay, and the quantity of the libraries was determined by real-time PCR in Light Cycler 480 (Roche Farma, Madrid, Spain). (3) Equimolar pooling of libraries was performed before proceeding to the generation of clusters in cbot (Illumina). The pool of the cDNA libraries was sequenced by paired-end sequencing (100 × 2) in the Illumina HiSEquation 2000 sequencer. The bioinformatic analysis of NGS and raw data availability is described in Supplemental Material and Methods. Reverse transcription and qRT-PCR of tissue and serum samples First-strand cDNA was generated using a cDNA synthesis kit, and subsequent qRT-PCR was performed in triplicate using miRCURY LNA Universal RT microRNA PCR ExiLENT SYBR Green (both from Exiqon) with the CFX384 Touch Real-Time PCR Detection System (Bio-Rad, Alcobendas, Madrid). Expression of miRNAs was performed using miRNA LNA PCR primer sets (Exiqon). Expression stabilities of the reference genes were evaluated using geNorm (28), NormFinder (29), and the coefficient of variation score described by Marabita et al. (30). Data were normalized using the geometric mean Ct of the best gene combination generated by the algorithms (Supplemental Table 1 and Supplemental Fig. 1). Relative expression was determined using the log base 2 values of the difference Cts between miRNAs and the geometric mean of the selected housekeeping genes. See Supplemental Material and Methods for detailed technical descriptions of the normalization. Statistical analysis Descriptive results were expressed as mean ± standard deviation, mean ± error of the mean, or median and 25th to 75th percentiles, as appropriate. Spearman bivariate correlations were performed for all quantitative variables and differences between groups were compared using analysis of variance (Mann-Whitney U or Kruskal-Wallis analysis of variance, as appropriate). A logistic regression model was used to determine the differences in each normalized miRNA between controls and AITD groups, adjusted by age and sex, as recommended in previous reports (31). Receiver operating characteristic (ROC) curve analyses were performed to assess the classification power of each logistic regression model. Samples from all groups within an experiment were processed at the same time. The P values were two-sided, and statistical significance was considered when P < 0.05. Data are presented with the following specific P values: P < 0.05, P < 0.01, and P < 0.001. All statistical analyses were performed using GraphPad Prism 4 software (GraphPad Software, La Jolla, CA) and STATA (StataCorp LLC, College Station, TX). Results miRNA expression in patients with AITD Unsupervised hierarchical clustering and principal component analysis were used to investigate the potential identification of AITD subgroups, based on their molecular expression profile. We performed NGS using RNA obtained from thyroids from five patients with HT, five patients with GD with no GO, five patients with GD with GO, and five healthy participants. The analysis revealed a slightly but not substantial separation between AITD groups. Control samples clustered together and formed a separate subcluster, indicating that these samples have a very similar miRNA signature and that this signature was different from those of AITD samples. Regarding the principal component analysis, an outlier sample from the GD group was detected and excluded from the study (Fig. 1A and 1B). We, therefore, decided to assess the differential miRNA expression of all AITD samples together in the same group, in comparison with the control group. A total of 3431 sequences were detected. In particular, 19 miRNAs were upregulated and 1 was downregulated more than twofold in AITD vs normal thyroid. Of these, miR-146b-5p displayed the most significant upregulation, with a log twofold change of 6.303 and P < 0.00001. Figure 1. View largeDownload slide (A) Heatmap hierarchical clustering of DE miRNAs in AITD tissue samples on the basis of 19 DE miRNAs. The graph shows miRNA expression in thyroid tissue from AITD samples compared with controls, expressed in fold change. The key color bar indicates that miRNA expression levels increased from green to red, depending on the fold change. (B) Principal component analysis (PCA) projected as a two-dimensional scatterplot. The most correlated samples are circled in different colors (blue = controls, green = HT, brown = GD, and gray = GO). Circled in red is an outlier that was excluded from the analysis (see text for full explanation). (C) Cluster dendrogram of miRNAs that have been associated by immunological enrichment analysis (false discovery rate <0.05). miR-21-5p, miR-155-5p, and miR-146a-5p are the most associated with immunological pathways. (D) Top 10 dysregulated miRNAs in AITD samples selected for validation by qRT-PCR (P < 0.05) and fold change ± 2 calculated using the algorithm DESeq2; base mean indicates the mean of reads of the miRNAs adjusted by the mean of the total reads generated in the NGS. Figure 1. View largeDownload slide (A) Heatmap hierarchical clustering of DE miRNAs in AITD tissue samples on the basis of 19 DE miRNAs. The graph shows miRNA expression in thyroid tissue from AITD samples compared with controls, expressed in fold change. The key color bar indicates that miRNA expression levels increased from green to red, depending on the fold change. (B) Principal component analysis (PCA) projected as a two-dimensional scatterplot. The most correlated samples are circled in different colors (blue = controls, green = HT, brown = GD, and gray = GO). Circled in red is an outlier that was excluded from the analysis (see text for full explanation). (C) Cluster dendrogram of miRNAs that have been associated by immunological enrichment analysis (false discovery rate <0.05). miR-21-5p, miR-155-5p, and miR-146a-5p are the most associated with immunological pathways. (D) Top 10 dysregulated miRNAs in AITD samples selected for validation by qRT-PCR (P < 0.05) and fold change ± 2 calculated using the algorithm DESeq2; base mean indicates the mean of reads of the miRNAs adjusted by the mean of the total reads generated in the NGS. Gene ontology enrichment of differentially expressed miRNAs To evaluate the association of miRNAs with relevant immunological functions, target genes biologically annotated for the DE miRNAs were identified using miRTarBase and TargetScan. Gene ontology–selected immunological enriched processes were associated with DE miRNAS (Fig. 1C and Supplemental Table 2). The analysis of biological network exploration showed the correlation and cooperation of miRNAs related to the immunological system (Supplemental Fig. 2). In fact, we can observe the cooperation of pathway reactions relative to the immune and adaptive system of miR-96-5p, miR-142-3p, miR-146a-5p, and miR-146b-5p. At the same time, miR-21-5p and miR-6503-3p cooperate with miR-146a-5p in processes related to the Toll-like receptor and interferon signaling pathways, respectively. Considering this, 10 top-ranked selected miRNAs importantly associated with immune pathways were selected for validation by qRT-PCR: miR-21-5p, miR-96-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, miR-766-3p, and miR-6503-3p (Fig. 1D). Validation of significant differentially expressed miRNAs in thyroid tissue To validate the top-ranked DE selected miRNAs, qRT-PCR was performed in 36 fresh-frozen and 25 FFPE thyroid tissue from the same samples. There was a good correlation in 5 of 10 miRNAs in the same samples extracted from fresh-frozen and FFPE tissues (P < 0.05) (Supplemental Fig. 3). In this context, we decided to exclude FFPE samples from the study and continue only with the expressions related to frozen samples. Then, comparison of the expression levels performed between NGS data and qRT-PCR in frozen tissues demonstrated a similar expression pattern between both methods (Fig. 2A). Of the top 10 selected miRNAs, 8 were confirmed to be DE in AITD samples (miR-21-5p, P < 0.01; miR-142-3p, P < 0.01; miR-146a-5p, P < 0.001; miR-146b-5p, P < 0.001; miR-155-5p, P < 0.05; miR-338-5p, P < 0.05; miR-342-5p, P < 0.001; and miR-766-3p, P < 0.001). However, although there was a trend toward being increased in AITD, no significant differences were found for miR-96-5p and miR-6503-5p between the two groups (Fig. 2B). Figure 2. View largeDownload slide (A) Comparison of miRNA expression between NGS and qRT-PCR. miRNAs determined to be DE in patients with AITD by NGS were validated by qRT-PCR. The height of the columns in the chart represents the log-transformed average fold change in expression, comparing the expression levels of AITD samples with normal thyroid tissues for each of the validated miRNAs. A total of 10 miRNAs were selected for validation, and 8 were confirmed by qRT-PCR. Data are expressed as mean ± standard error of the mean. (B) miRNAs in the validation study in tissue and serum samples. Eight different miRNAs were DE in 26 patients with AITD compared with 10 controls measured in fresh-frozen thyroid tissue samples (miR-21-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, and miR-766-3p). Of the total circulating miRNAs evaluated in 56 samples (36 patients with AITD and 22 healthy controls), miR-21-5p, miR-96-5p, miR-142-3p, and miR-146a-5p were DE in patients with AITD vs controls. Furthermore, of the five miRNAs included in the study, miR-Let7d-5p, miR-142-5p, and miR-301a-3p were also DE in serum samples. (C) Normalized relative quantities (NRQs) of miRNAs in serum samples analyzed according to the different subgroups of AITD: HT and GD. miR-Let7d-5p was DE in HT, mir-142-3p in GD, and miR-21-5p, 96-5p, and 301-3p in both HT and GD. miR146a-5p and miR-6503-3p lost significance when we distinguished between HT and GD. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Figure 2. View largeDownload slide (A) Comparison of miRNA expression between NGS and qRT-PCR. miRNAs determined to be DE in patients with AITD by NGS were validated by qRT-PCR. The height of the columns in the chart represents the log-transformed average fold change in expression, comparing the expression levels of AITD samples with normal thyroid tissues for each of the validated miRNAs. A total of 10 miRNAs were selected for validation, and 8 were confirmed by qRT-PCR. Data are expressed as mean ± standard error of the mean. (B) miRNAs in the validation study in tissue and serum samples. Eight different miRNAs were DE in 26 patients with AITD compared with 10 controls measured in fresh-frozen thyroid tissue samples (miR-21-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, and miR-766-3p). Of the total circulating miRNAs evaluated in 56 samples (36 patients with AITD and 22 healthy controls), miR-21-5p, miR-96-5p, miR-142-3p, and miR-146a-5p were DE in patients with AITD vs controls. Furthermore, of the five miRNAs included in the study, miR-Let7d-5p, miR-142-5p, and miR-301a-3p were also DE in serum samples. (C) Normalized relative quantities (NRQs) of miRNAs in serum samples analyzed according to the different subgroups of AITD: HT and GD. miR-Let7d-5p was DE in HT, mir-142-3p in GD, and miR-21-5p, 96-5p, and 301-3p in both HT and GD. miR146a-5p and miR-6503-3p lost significance when we distinguished between HT and GD. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Expression of identified thyroid DE miRNAs in serum samples We then determined whether the selected DE miRNAs were detected in serum samples of 8 patients with active GO and 4 patients with inactive GO, 10 patients with GD with no GO, 14 patients with HT, and 22 healthy controls. In addition to the 10 DE miRNAs identified in thyroid tissue samples, 5 other miRNAs involved in immune functions or identified in previous studies were also included for validation in AITD serum samples (miR-Let7d-5p, miR-142-5p, miR-126-3p, miR-223-3p, and miR-301a-3p) (24, 32–34). Eight of this whole set of miRNAs analyzed were DE in AITD samples (miR-Let7d-5p, P < 0.001; miR-21-5p, P < 0.0001; miR-96-5p, P < 0.001; miR-142-3p, P < 0.01; miR142-5p, P < 0.001; miR146a-5p, P < 0.05; miR-301a-3p, P < 0.001; and miR-6503-3p, P < 0.05) (Fig. 2B). All were upregulated in AITD, except for miR-Let-7d, which was downregulated. To discriminate if these miRNAs were dysregulated according to the type of AITD (HT or GD), we analyzed the levels of DE miRNAs in each subgroup; miR-Let7d-5p was DE in HT, mir-142-3p in GD, and miR-21-5p, miR-96-5p, and miR-301-3p in both HT and GD. Although miR-146a-5p and miR-6503-3p lost significance when both groups were considered separately, we observed the same trend of dysregulation in all of them, denoting the distinction between controls and HT/GD (Fig. 2C and Fig. 3A). Figure 3. View largeDownload slide (A) Overall study design and workflow. NGS was performed in 20 thyroid tissues. Ten miRNAs were selected for validation by qRT-PCR, and 8 were confirmed. In five cases, miRNAs from NGS were confirmed in serum samples. Five additional miRNAs were selected bibliographically. Of the 15 miRNAs studied in serum samples (10 from NGS and 5 selected bibliographically), 5 were confirmed as a specific miRNA signature. (B) Roles of the five-signature miRNA (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) found in AITD samples. The lower levels of miR-Let7d-5p fall to the suppression of Th1 cells by Treg cells, leading to Th1 cell proliferation and interferon-γ (IFN-γ) secretion (32). The increase of miR-142-3p could impair the inhibitory effect of Treg cells on the proliferative response and cytokine production of CD4+CD25− T cells (35). miR-21-5p can regulate the Th1/Th2 balance and promote Th17 differentiation by inhibiting Smad-7 (36). miR-96-5p can increase cytokine production of Th17 cells (37). miR-301a-3p can downregulate PIAS3, promoting Th17 development (34). The altered expression of all these miRNAs can alter the normal function of immune cells, leading to a loss of immune tolerance in AITD. TNFα, tumor necrosis factor α. Figure 3. View largeDownload slide (A) Overall study design and workflow. NGS was performed in 20 thyroid tissues. Ten miRNAs were selected for validation by qRT-PCR, and 8 were confirmed. In five cases, miRNAs from NGS were confirmed in serum samples. Five additional miRNAs were selected bibliographically. Of the 15 miRNAs studied in serum samples (10 from NGS and 5 selected bibliographically), 5 were confirmed as a specific miRNA signature. (B) Roles of the five-signature miRNA (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) found in AITD samples. The lower levels of miR-Let7d-5p fall to the suppression of Th1 cells by Treg cells, leading to Th1 cell proliferation and interferon-γ (IFN-γ) secretion (32). The increase of miR-142-3p could impair the inhibitory effect of Treg cells on the proliferative response and cytokine production of CD4+CD25− T cells (35). miR-21-5p can regulate the Th1/Th2 balance and promote Th17 differentiation by inhibiting Smad-7 (36). miR-96-5p can increase cytokine production of Th17 cells (37). miR-301a-3p can downregulate PIAS3, promoting Th17 development (34). The altered expression of all these miRNAs can alter the normal function of immune cells, leading to a loss of immune tolerance in AITD. TNFα, tumor necrosis factor α. A five-signature miRNA to discriminate patients with AITD vs controls: correlation with clinical features To assess the existence of any possible associations between miRNA levels and clinical features and explore their clinical utility, we first investigated the correlations of all the analyzed miRNAs with patients’ corresponding serum levels of free thyroxine, thyrotropin, antithyroid peroxidase antibody (TPO-Ab), antithyroglobulin antibody (Tg-Ab), and TSHR-Ab (correlation heatmap in Fig. 4A). Spearman ρ analyses revealed that miR-Let7d-5p was negatively associated with levels of TPO-Ab (r = −0.38), whereas miR-21-5p and miR-96-5p were positively correlated with TPO-Ab (r = 0.39 and r = 0.48, respectively), Tg-Ab (r = 0.36 and r = 0.4, respectively), and TSHR-Ab (r = 0.45 and r = 0.4, respectively). miR-142-3p and miR-301a-3p were only positively associated with TSHR-Ab (r = 0.38 and r = 0.31), and miR-6503-3p correlated with TPO-Ab (r = 0.27). Interestingly, these miRNAs (except for miR-6503-3p) were DE in HT and/or GD. Also, all miRNAs were significantly correlated between themselves, with a remarkable similarity between miR-21-5p and miR-96-5, which exhibited an analogous behavior regarding their correlation with clinical features (Fig. 4B). Figure 4. View largeDownload slide (A) Correlation heatmap between the expression of miRNAs and clinical and laboratory parameters (Spearman ρ analysis). Significant negative correlations are shown in red and significant positive correlations in green (*P < 0.05; **P < 0.01). Bottom left triangle shows correlations between miRNAs and top square shows correlations between miRNAs and clinical data. miR-Let7d-5p is negatively associated with levels of TPO-Ab (r = −0.38). miR-21-5p and miR-96-5p are positively correlated with TPO-Ab (r = 0.39 and r = 0.48, respectively), Tg-Ab (r = 0.36 and r = 0.4 respectively), and TSHR-Ab (r = 0.45 and r = 0.4 respectively). miR-142-5p and miR-301a-3p are positively associated with TSHR-Ab (r = 0.38 and r = 0.31), and miR-6503-3p correlates with TPO-Ab (r = 0.27). (B) Correlation analysis between miR-21-5p and miR-96-5p (r = 0.58 and P < 0.001). (C) ROC curve analyses performed to assess the diagnostic value of circulating miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301-3p to discriminate between controls and patients with AITD. The table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. Figure 4. View largeDownload slide (A) Correlation heatmap between the expression of miRNAs and clinical and laboratory parameters (Spearman ρ analysis). Significant negative correlations are shown in red and significant positive correlations in green (*P < 0.05; **P < 0.01). Bottom left triangle shows correlations between miRNAs and top square shows correlations between miRNAs and clinical data. miR-Let7d-5p is negatively associated with levels of TPO-Ab (r = −0.38). miR-21-5p and miR-96-5p are positively correlated with TPO-Ab (r = 0.39 and r = 0.48, respectively), Tg-Ab (r = 0.36 and r = 0.4 respectively), and TSHR-Ab (r = 0.45 and r = 0.4 respectively). miR-142-5p and miR-301a-3p are positively associated with TSHR-Ab (r = 0.38 and r = 0.31), and miR-6503-3p correlates with TPO-Ab (r = 0.27). (B) Correlation analysis between miR-21-5p and miR-96-5p (r = 0.58 and P < 0.001). (C) ROC curve analyses performed to assess the diagnostic value of circulating miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301-3p to discriminate between controls and patients with AITD. The table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. We assessed the discriminating potency between AITD and controls of these circulating miRNAs using ROC curve analyses (Fig. 4C), and we confirmed that five miRNAs (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) were good discriminators [area under the curve (AUC) ≥0.848, with >76% correctly classified instances]. Then, we assessed the prognostic role of miRNAs in GD by correlating DE miRNAs of the signature with clinical parameters, including persistent and/or severe forms of hyperthyroidism. Relative expression of miR-Let7d-5p, miR-96-5p, and miR-301a-3p was significantly different in patients with GO than in controls (P = 0.0073, P = 0.0055, and P = 0.0374, respectively), although no differences were found between controls and patients with GD without GO (P > 0.99, P = 0.06, and P = 0.42, respectively) (Fig. 5A). However, miR-Let7d-5p was significantly lower in patients with GO compared with patients with GD without GO (P = 0.016) and more remarkably in those with active GO (P = 0.0016) (Fig. 5B), which also had a negative correlation with clinical activity score (r = −0.6504, P = 0.0014) (Fig. 5C). Patients with more severe disease had lower miR-Let7d-5p levels (P = 0.034) (Fig. 5D). For miR-21-5p, a higher expression was significantly correlated with a worse prognosis, with more patients presenting recurrent disease (P = 0.028) (Fig. 5E and Supplemental Table 3). In fact, after adjusting for age, sex, and TSHR-Ab, miR-21-5p remained significant (P = 0.042). Subsequent analysis using the method of all possible equations showed that combining miR-21-5p and TSHR-Ab in the regression model achieved the best predictive value for disease persistence. In this regard, ROC curves for miR-21-5p, TSHR-Ab, and a composite model of both markers (Fig. 5F) found that baseline miR-21-5p alone had an AUC of 0.85 and a percentage of correctly classified instances of 85%, whereas baseline TSHR-Ab alone had an AUC of 0.64 and a prediction of 65% at the cut point of 3. The composite marker had an AUC of 0.92 and a percentage of prediction of 90%. Figure 5. View largeDownload slide Correlation of levels of miRNAs with clinical data of patients with GD. (A) Classification of patients with GD according to the presence or absence of GO. Relative expression of miR-Let7d-5p, miR-96-5p, and miR-301-3p was DE in the GO group compared with controls. In addition, miR-Let7d-5p was DE between patients with GD with and without GO. (B) The expression levels of miR-Let7d-5p were DE in patients with GD with active GO [clinical activity score (CAS) >3]. (C) There was a negative correlation between the expression of miR-Let7d-5p and CAS. (D) Levels of miR-Let7d-5p according to disease severity; patients with severe disease had lower levels of miR-Let7d-5p. (E) miR-21-5p was higher in patients with persistent disease. (F) ROC curves were performed to analyze the predictive values of regression models for disease persistence based on miR-21-5p, TSHR-Ab, and the composite of both parameters. Table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Figure 5. View largeDownload slide Correlation of levels of miRNAs with clinical data of patients with GD. (A) Classification of patients with GD according to the presence or absence of GO. Relative expression of miR-Let7d-5p, miR-96-5p, and miR-301-3p was DE in the GO group compared with controls. In addition, miR-Let7d-5p was DE between patients with GD with and without GO. (B) The expression levels of miR-Let7d-5p were DE in patients with GD with active GO [clinical activity score (CAS) >3]. (C) There was a negative correlation between the expression of miR-Let7d-5p and CAS. (D) Levels of miR-Let7d-5p according to disease severity; patients with severe disease had lower levels of miR-Let7d-5p. (E) miR-21-5p was higher in patients with persistent disease. (F) ROC curves were performed to analyze the predictive values of regression models for disease persistence based on miR-21-5p, TSHR-Ab, and the composite of both parameters. Table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Discussion Accumulating evidence suggests that miRNAs can be used as biomarkers of various diseases, including various autoimmune disorders (38). In addition to tissue-derived miRNAs, circulating miRNAs, which are protected from endogenous RNase activity, are specific and stable and can be used as extracellular biomarkers. The advantage of using circulating miRNAs is that the diagnostic approach could be minimized to a single blood sample (7). Although studies have investigated miRNAs in patients with AITD, few have explored their clinical and diagnostic utility and their role in risk stratification. In this study, we describe the global patterns of miRNA expression found in AITD thyroid samples. We found a good correlation between the expression profiles using NGS and qRT-PCR in fresh-frozen thyroid tissues, suggesting that NGS could be a reliable method for miRNA profiling in AITD. In fact, using NGS, we selected 10 miRNAs in accordance to immune pathways (cytokine signaling, Toll-like receptor cascade, interferon signaling, B-cell receptor signaling) and confirmed that 8 miRNAs were DE in the thyroid and 3 (miR-21-5p, miR-142-3p, and miR-146a-5p) in serum samples of patients with AITD. These results suggest that some of these circulating miRNAs are synthetized in thyroid tissue and are then released to human body fluids, including the blood, as a form of intercellular communication, so they could be used to monitor the altered tissue. Some of the miRNAs found in the thyroid (miR-21-5p, miR-146a-5p, and miR-155-5p) have been widely reported to participate in lymphocyte differentiation and activation (36, 39–43). For instance, miR-146 has been associated with regulation of the immune response (39, 41, 43) and miR-155 to T-regulatory cell development and Th17 differentiation (40, 42). Regarding AITD, studies on these miRNAs have been discordant, with reports of decreased or increased expression of miR-21-5p, miR-146a-5p, and/or miR-155-5p, depending on the specific type of disease (AITD vs HT vs GD), the tissue studied (thyroid tissue, needle aspiration samples, peripheral blood mononuclear cells, microvesicles, and/or serum), or the conservation method used (fresh samples vs FFPE) (12, 13, 20–22, 24, 44), hence the importance of studying homogeneous groups of patients, samples, and preservation methods. In our study, we found some differences in DE miRNAs when we analyzed AITD as a whole or as subgroups or in thyroid vs serum samples. Moreover, there was a poor correlation between the relative expression of miRNA in formalin-fixed and frozen samples, which forced us to exclude FFPE samples from our analysis to provide more robustness to our results (45–48). Taken together, all these issues could potentially explain the discrepancies observed between previous publications and our current findings. To further evaluate if these miRNAs could behave as biomarkers of risk for developing AITD, we studied the top 10 miRNAs found in NGS and 5 additional miRNAs (miR-Let7d-5p, miR-126-3p, miR-142-5p, miR-223-3p, and miR-301a-3p), which have been previously reported to be involved in different autoimmune disorders, including AITD (24, 32–34). We found a good correlation between some of these miRNAs and all three thyroid autoantibody levels. Because these autoantibodies are frequently used to establish the risk of AITD, we assessed whether miRNAs could be also used in the same way. To do so, we performed a composite marker by a logistic regression model and found that a specific five-serum miRNA signature (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301-3p) was able to assign a risk for developing AITD independently of autoantibody titers. We also evaluated if this signature could be the used in the clinical evaluation of patients with GD. We found that three of the signature’s miRNAs (miR-Let7d-5p, miR-96-5p, and miR-301-3p) were DE in patients with GO, whereas miR-Let7d-5p was inversely correlated with a higher GD severity and GO clinical activity. In this regard, as miR-Let7 has a specific Treg cell-mediated function, which can prevent Th1 cell-mediated inflammation and interferon-γ secretion (49), its lower levels found in patients with AITD with more severe disease could be associated with a defective Treg function previously reported in patients with AITD (50). miRNA-21-5p, itself, was associated with a higher risk of developing GD and a worse clinical outcome, and it was correlated to persistent disease in the long-term follow-up, with a better predictive value than the currently used TRAb levels (51). Recent reports have highly implicated miR-21 in the regulation of immune functions and development of several autoimmune disorders (52–55), suggesting a shared mechanism of action of this miRNA in different immune cells. The fact that expression of miR-21 affects T-cell activation, including the Th1/Th2 balance and Th17 differentiation (20, 21, 36, 56, 57), may explain these findings. In fact, we have recently reported increased levels of pathogenic Th17 cells in the peripheral blood and thyroid tissue from patients with AITD (58, 59). Our results also point to a similar behavior between miR-21-5p and miR-96-5p in AITD serum samples. Recent findings have demonstrated that miR-96 is highly expressed in pathogenic Th17 cells, and its overexpression in CD4+ T cells significantly increases cytokine production of Th17 cells (37), denoting a role in Th17 cells (60). For miR-142-3p and miR-301a-3p, there have been reports of their implication in Treg and Th17 regulation, respectively, in some autoimmune diseases (34, 35, 61, 62). However, to our knowledge, data are still lacking regarding these miRNAs in AITD, and our results add new evidence to their role in Treg dysregulation in this particular disease (50) (Fig. 3B). In summary, our data reveal for the first time using NGS, to our knowledge, the validation of dysregulated miRNAs in thyroid tissue and serum from patients with AITD. In addition, we provide a five-signature miRNA, which may be used as a potential biomarker to assign a risk for developing AITD and also for the evaluation of the severity of GD. This signature could have a substantial value in selecting specific treatment options. These outcomes represent a meaningful advance in the field of miRNAs as biomarkers in AITD and could contribute to a better understanding of the powerful translational role of miRNAs. Abbreviations: AITD autoimmune thyroid disease AUC area under the curve cDNA complementary DNA DE differentially expressed FFPE formalin fixed, paraffin embedded GD Graves disease GO Graves ophthalmopathy HT Hashimoto thyroiditis miRNA microRNA NGS next-generation sequencing PCR polymerase chain reaction qRT-PCR quantitative reverse transcription polymerase chain reaction ROC receiver operating characteristic Tg-Ab antithyroglobulin antibody TPO-Ab antithyroid peroxidase antibody TSHR-Ab antithyrotropin receptor antibody. Acknowledgments We thank the Institut d'Investigació en Ciències de la Salut Germans Trias i Pujol (IGTP-HUGTIP) Biobank and the Department of Surgery of Hospital Universitario la Princesa (HUP) (José Luis Muñoz) for providing tissue samples. We also thank our colleagues from the Department of Pathological Anatomy of HUP, especially Magdalena Adrados, for her help classifying thyroid specimens. Financial Support: This work was supported by grants from the Ministerio de Economia y Competitividad and from the Instituto de Salud Carlos III [Proyectos de Investigación en Salud (FIS): PI13-01414, PI16-02091, and PIE13-0004-BIOIMID project] to M.M., and by grants from CIBER de Enfermedades Cardiovasculares (CIBERCV) and SAF2017-82886-R to F.S.-M. and cofinanced by FEDER funds. Disclosure Summary: The authors have nothing to disclose. References 1. Ramos-Leví AM, Marazuela M. Pathogenesis of thyroid autoimmune disease: the role of cellular mechanisms. Endocrinol Nutr . 2016; 63( 8): 421– 429. Google Scholar CrossRef Search ADS PubMed  2. Weetman AP. Autoimmune thyroid disease. Autoimmunity . 2009; 37( 4): 337– 340. Google Scholar CrossRef Search ADS   3. Weetman AP. Cellular immune responses in autoimmune thyroid disease. Clin Endocrinol (Oxf) . 2004; 61( 4): 405– 413. Google Scholar CrossRef Search ADS PubMed  4. Bartalena L, Fatourechi V. Extrathyroidal manifestations of Graves’ disease: a 2014 update. J Endocrinol Invest . 2014; 37( 8): 691– 700. Google Scholar CrossRef Search ADS PubMed  5. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell . 2004; 116( 2): 281– 297. Google Scholar CrossRef Search ADS PubMed  6. Kozomara A, Griffiths-Jones S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res . 2013; 42( D1): D68– D73. Google Scholar CrossRef Search ADS PubMed  7. Heegaard NHH, Carlsen AL, Skovgaard K, Heegaard PMH. Circulating extracellular microRNA in systemic autoimmunity. EXS . 2015; 106: 171– 195. Google Scholar PubMed  8. Nakamachi Y, Kawano S, Takenokuchi M, Nishimura K, Sakai Y, Chin T, Saura R, Kurosaka M, Kumagai S. MicroRNA-124a is a key regulator of proliferation and monocyte chemoattractant protein 1 secretion in fibroblast-like synoviocytes from patients with rheumatoid arthritis. Arthritis Rheum . 2009; 60( 5): 1294– 1304. Google Scholar CrossRef Search ADS PubMed  9. Wang H, Peng W, Ouyang X, Li W, Dai Y. Circulating microRNAs as candidate biomarkers in patients with systemic lupus erythematosus. Transl Res . 2012; 160( 3): 198– 206. Google Scholar CrossRef Search ADS PubMed  10. Tang Y, Luo X, Cui H, Ni X, Yuan M, Guo Y, Huang X, Zhou H, de Vries N, Tak PP, Chen S, Shen N. MicroRNA-146A contributes to abnormal activation of the type I interferon pathway in human lupus by targeting the key signaling proteins. Arthritis Rheum . 2009; 60( 4): 1065– 1075. Google Scholar CrossRef Search ADS PubMed  11. Xia P, Fang X, Zhang ZH, Huang Q, Yan KX, Kang KF, Han L, Zheng ZZ. Dysregulation of miRNA146a versus IRAK1 induces IL-17 persistence in the psoriatic skin lesions. Immunol Lett . 2012; 148( 2): 151– 162. Google Scholar CrossRef Search ADS PubMed  12. Bernecker C, Halim F, Lenz L, Haase M, Nguyen T, Ehlers M, Vordenbaeumen S, Schott M. microRNA expressions in CD4+ and CD8+ T-cell subsets in autoimmune thyroid diseases. Exp Clin Endocrinol Diabetes . 2014; 122( 2): 107– 112. Google Scholar CrossRef Search ADS PubMed  13. Bernecker C, Lenz L, Ostapczuk MS, Schinner S, Willenberg H, Ehlers M, Vordenbäumen S, Feldkamp J, Schott M. MicroRNAs miR-146a1, miR-155_2, and miR-200a1 are regulated in autoimmune thyroid diseases. Thyroid . 2012; 22( 12): 1294– 1295. Google Scholar CrossRef Search ADS PubMed  14. Dorris ER, Smyth P, O’Leary JJ, Sheils O. MIR141 expression differentiates hashimoto thyroiditis from PTC and benign thyrocytes in Irish archival thyroid tissues. Front Endocrinol (Lausanne) . 2012; 3: 102. Google Scholar PubMed  15. Kagawa T, Watanabe M, Inoue N, Otsu H, Saeki M, Katsumata Y, Takuse Y, Iwatani Y. Increases of microRNA let-7e in peripheral blood mononuclear cells in Hashimoto’s disease. Endocr J . 2016; 63( 4): 375– 380. Google Scholar CrossRef Search ADS PubMed  16. Liu R, Ma X, Xu L, Wang D, Jiang X, Zhu W, Cui B, Ning G, Lin D, Wang S. Differential microRNA expression in peripheral blood mononuclear cells from Graves’ disease patients. J Clin Endocrinol Metab . 2012; 97( 6): E968– E972. Google Scholar CrossRef Search ADS PubMed  17. Peng H, Liu Y, Tian J, Ma J, Tang X, Yang J, Rui K, Zhang Y, Mao C, Lu L, Xu H, Wang S. Decreased expression of microRNA-125a-3p upregulates interleukin-23 receptor in patients with Hashimoto’s thyroiditis. Immunol Res . 2015; 62( 2): 129– 136. Google Scholar CrossRef Search ADS PubMed  18. Qin Q, Wang X, Yan N, Song RH, Cai TT, Zhang W, Guan LJ, Muhali FS, Zhang JA. Aberrant expression of miRNA and mRNAs in lesioned tissues of Graves’ disease. Cell Physiol Biochem . 2015; 35( 5): 1934– 1942. Google Scholar CrossRef Search ADS PubMed  19. Shen L, Huang F, Ye L, Zhu W, Zhang X, Wang S, Wang W, Ning G. Circulating microRNA predicts insensitivity to glucocorticoid therapy in Graves’ ophthalmopathy. Endocrine . 2015; 49( 2): 445– 456. Google Scholar CrossRef Search ADS PubMed  20. Tong BD, Xiao MY, Zeng JX, Xiong W. miRNA-21 promotes fibrosis in orbital fibroblasts from thyroid-associated ophthalmopathy. Mol Vis . 2015; 21: 324– 334. Google Scholar PubMed  21. Wang Z, Fan X, Zhang R, Lin Z, Lu T, Bai X, Li W, Zhao J, Zhang Q. Integrative analysis of mRNA and miRNA array data reveals the suppression of retinoic acid pathway in regulatory T cells of Graves’ disease. J Clin Endocrinol Metab . 2014; 99( 12): E2620– E2627. Google Scholar CrossRef Search ADS PubMed  22. Wei H, Guan M, Qin Y, Xie C, Fu X, Gao F, Xue Y. Circulating levels of miR-146a and IL-17 are significantly correlated with the clinical activity of Graves’ ophthalmopathy. Endocr J . 2014; 61( 11): 1087– 1092. Google Scholar CrossRef Search ADS PubMed  23. Yamada H, Itoh M, Hiratsuka I, Hashimoto S. Circulating microRNAs in autoimmune thyroid diseases. Clin Endocrinol (Oxf) . 2014; 81( 2): 276– 281. Google Scholar CrossRef Search ADS PubMed  24. Zhu J, Zhang Y, Zhang W, Zhang W, Fan L, Wang L, Liu Y, Liu S, Guo Y, Wang Y, Yi J, Yan Q, Wang Z, Huang G. MicroRNA-142-5p contributes to Hashimoto’s thyroiditis by targeting CLDN1. J Transl Med . 2016; 14( 1): 166. Google Scholar CrossRef Search ADS PubMed  25. Lee JY, Yun M, Paik JS, Lee SB, Yang SW. PDGF-BB enhances the proliferation of cells in human orbital fibroblasts by suppressing PDCD4 expression via up-regulation of microRNA-21. Invest Ophthalmol Vis Sci . 2016; 57( 3): 908– 913. Google Scholar CrossRef Search ADS PubMed  26. Hiratsuka I, Yamada H, Munetsuna E, Hashimoto S, Itoh M. Circulating microRNAs in Graves’ disease in relation to clinical activity. Thyroid . 2016; 26( 10): 1431– 1440. Google Scholar CrossRef Search ADS PubMed  27. Pohl M, Grabellus F, Worm K, Arnold G, Walz M, Schmid KW, Sheu-Grabellus SY. Intermediate microRNA expression profile in Graves’ disease falls between that of normal thyroid tissue and papillary thyroid carcinoma. J Clin Pathol . 2016; 70( 1): 33– 39. Google Scholar CrossRef Search ADS PubMed  28. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol . 2002; 3: RESEARCH0034. 29. Andersen CL, Jensen JL, Ørntoft TF. Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res . 2004; 64( 15): 5245– 5250. Google Scholar CrossRef Search ADS PubMed  30. Marabita F, de Candia P, Torri A, Tegnér J, Abrignani S, Rossi RL. Normalization of circulating microRNA expression data obtained by quantitative real-time RT-PCR. Brief Bioinform . 2015; 17( 2): 204– 212. Google Scholar CrossRef Search ADS PubMed  31. Ameling S, Kacprowski T, Chilukoti RK, Malsch C, Liebscher V, Suhre K, Pietzner M, Friedrich N, Homuth G, Hammer E, Völker U. Associations of circulating plasma microRNAs with age, body mass index and sex in a population-based study. BMC Med Genomics . 2015; 8( 1): 61. Google Scholar CrossRef Search ADS PubMed  32. Okoye IS, Coomes SM, Pelly VS, Czieso S, Papayannopoulos V, Tolmachova T, Seabra MC, Wilson MS. MicroRNA-containing T-regulatory-cell-derived exosomes suppress pathogenic T helper 1 cells. Immunity . 2014; 41( 1): 89– 103. Google Scholar CrossRef Search ADS PubMed  33. O’Connell RM, Rao DS, Baltimore D. MicroRNA regulation of inflammatory responses. Annu Rev Immunol . 2012; 30( 1): 295– 312. Google Scholar CrossRef Search ADS PubMed  34. Mycko MP, Cichalewska M, Machlanska A, Cwiklinska H, Mariasiewicz M, Selmaj KW. MicroRNA-301a regulation of a T-helper 17 immune response controls autoimmune demyelination. Proc Natl Acad Sci USA . 2012; 109( 20): E1248– E1257. Google Scholar CrossRef Search ADS PubMed  35. Huang B, Zhao J, Lei Z, Shen S, Li D, Shen GX, Zhang GM, Feng ZH. miR-142-3p restricts cAMP production in CD4+CD25- T cells and CD4+CD25+ TREG cells by targeting AC9 mRNA. EMBO Rep . 2008; 10( 2): 180– 185. Google Scholar CrossRef Search ADS PubMed  36. Garo LP, Murugaiyan G. Contribution of microRNAs to autoimmune diseases. Cell Mol Life Sci . 2016; 73( 10): 2041– 2051. Google Scholar CrossRef Search ADS PubMed  37. Ichiyama K, Gonzalez-Martin A, Kim BS, Jin HY, Jin W, Xu W, Sabouri-Ghomi M, Xu S, Zheng P, Xiao C, Dong C. The microRNA-183-96-182 cluster promotes T helper 17 cell pathogenicity by negatively regulating transcription factor Foxo1 expression. Immunity . 2016; 44( 6): 1284– 1298. Google Scholar CrossRef Search ADS PubMed  38. Wang J, Chen J, Sen S. MicroRNA as biomarkers and diagnostics. J Cell Physiol . 2015; 231( 1): 25– 30. Google Scholar CrossRef Search ADS   39. Abou-Zeid A, Saad M, Soliman E. MicroRNA 146a expression in rheumatoid arthritis: association with tumor necrosis factor-alpha and disease activity. Genet Test Mol Biomarkers . 2011; 15( 11): 807– 812. Google Scholar CrossRef Search ADS PubMed  40. Kohlhaas S, Garden OA, Scudamore C, Turner M, Okkenhaug K, Vigorito E. Cutting edge: the Foxp3 target miR-155 contributes to the development of regulatory T cells. J Immunol . 2009; 182( 5): 2578– 2582. Google Scholar CrossRef Search ADS PubMed  41. Lu LF, Boldin MP, Chaudhry A, Lin LL, Taganov KD, Hanada T, Yoshimura A, Baltimore D, Rudensky AY. Function of miR-146a in controlling Treg cell-mediated regulation of Th1 responses. Cell . 2010; 142( 6): 914– 929. Google Scholar CrossRef Search ADS PubMed  42. Yao R, Ma YL, Liang W, Li HH, Ma ZJ, Yu X, Liao YH. MicroRNA-155 modulates Treg and Th17 cells differentiation and Th17 cell function by targeting SOCS1. PLoS One . 2012; 7( 10): e46082. Google Scholar CrossRef Search ADS PubMed  43. Taganov KD, Boldin MP, Chang KJ, Baltimore D. NF-kappaB-dependent induction of microRNA miR-146, an inhibitor targeted to signaling proteins of innate immune responses. Proc Natl Acad Sci USA . 2006; 103( 33): 12481– 12486. Google Scholar CrossRef Search ADS PubMed  44. Rodríguez-Muñoz A, Martínez-Hernández R, Ramos-Leví AM, Serrano-Somavilla A, González-Amaro R, Sánchez-Madrid F, de la Fuente H, Marazuela M. Circulating microvesicles regulate Treg and Th17 differentiation in human autoimmune thyroid disorders. J Clin Endocrinol Metab . 2015; 100( 12): E1531– E1539. Google Scholar CrossRef Search ADS PubMed  45. Glud M, Klausen M, Gniadecki R, Rossing M, Hastrup N, Nielsen FC, Drzewiecki KT. MicroRNA expression in melanocytic nevi: the usefulness of formalin-fixed, paraffin-embedded material for miRNA microarray profiling. J Invest Dermatol . 2009; 129( 5): 1219– 1224. Google Scholar CrossRef Search ADS PubMed  46. Klopfleisch R, Weiss AT, Gruber AD. Excavation of a buried treasure—DNA, mRNA, miRNA and protein analysis in formalin fixed, paraffin embedded tissues. Histol Histopathol . 2011; 26( 6): 797– 810. Google Scholar PubMed  47. Li J, Smyth P, Flavin R, Cahill S, Denning K, Aherne S, Guenther SM, O’Leary JJ, Sheils O. Comparison of miRNA expression patterns using total RNA extracted from matched samples of formalin-fixed paraffin-embedded (FFPE) cells and snap frozen cells. BMC Biotechnol . 2007; 7( 1): 36. Google Scholar CrossRef Search ADS PubMed  48. Xi Y, Nakajima G, Gavin E, Morris CG, Kudo K, Hayashi K, Ju J. Systematic analysis of microRNA expression of RNA extracted from fresh frozen and formalin-fixed paraffin-embedded samples. RNA . 2007; 13( 10): 1668– 1674. Google Scholar CrossRef Search ADS PubMed  49. Gandhi R, Healy B, Gholipour T, Egorova S, Musallam A, Hussain MS, Nejad P, Patel B, Hei H, Khoury S, Quintana F, Kivisakk P, Chitnis T, Weiner HL. Circulating microRNAs as biomarkers for disease staging in multiple sclerosis. Ann Neurol . 2013; 73( 6): 729– 740. Google Scholar CrossRef Search ADS PubMed  50. Marazuela M, García-López MA, Figueroa-Vega N, de la Fuente H, Alvarado-Sánchez B, Monsiváis-Urenda A, Sánchez-Madrid F, González-Amaro R. Regulatory T cells in human autoimmune thyroid disease. J Clin Endocrinol Metab . 2006; 91( 9): 3639– 3646. Google Scholar CrossRef Search ADS PubMed  51. Weetman AP. Graves’ disease. N Engl J Med . 2000; 343( 17): 1236– 1248. Google Scholar CrossRef Search ADS PubMed  52. Fenoglio C, Cantoni C, De Riz M, Ridolfi E, Cortini F, Serpente M, Villa C, Comi C, Monaco F, Mellesi L, Valzelli S, Bresolin N, Galimberti D, Scarpini E. Expression and genetic analysis of miRNAs involved in CD4+ cell activation in patients with multiple sclerosis. Neurosci Lett . 2011; 504( 1): 9– 12. Google Scholar CrossRef Search ADS PubMed  53. Garchow BG, Bartulos Encinas O, Leung YT, Tsao PY, Eisenberg RA, Caricchio R, Obad S, Petri A, Kauppinen S, Kiriakidou M. Silencing of microRNA-21 in vivo ameliorates autoimmune splenomegaly in lupus mice. EMBO Mol Med . 2011; 3( 10): 605– 615. Google Scholar CrossRef Search ADS PubMed  54. Ruan Q, Wang T, Kameswaran V, Wei Q, Johnson DS, Matschinsky F, Shi W, Chen YH. The microRNA-21-PDCD4 axis prevents type 1 diabetes by blocking pancreatic beta cell death. Proc Natl Acad Sci USA . 2011; 108( 29): 12030– 12035. Google Scholar CrossRef Search ADS PubMed  55. Punga AR, Andersson M, Alimohammadi M, Punga T. Disease specific signature of circulating miR-150-5p and miR-21-5p in myasthenia gravis patients. J Neurol Sci . 2015; 356( 1–2): 90– 96. Google Scholar CrossRef Search ADS PubMed  56. Murugaiyan G, da Cunha AP, Ajay AK, Joller N, Garo LP, Kumaradevan S, Yosef N, Vaidya VS, Weiner HL. MicroRNA-21 promotes Th17 differentiation and mediates experimental autoimmune encephalomyelitis. J Clin Invest . 2015; 125( 3): 1069– 1080. Google Scholar CrossRef Search ADS PubMed  57. Wang S, Wan X, Ruan Q. The microRNA-21 in autoimmune diseases. Int J Mol Sci . 2016; 17( 6): 864. Google Scholar CrossRef Search ADS   58. Vitales-Noyola M, Ramos-Levi AM, Martínez-Hernández R, Serrano-Somavilla A, Sampedro-Nuñez M, González-Amaro R, Marazuela M. Pathogenic Th17 and Th22 cells are increased in patients with autoimmune thyroid disorders. Endocrine . 2017; 57( 3): 409– 417. Google Scholar CrossRef Search ADS PubMed  59. Figueroa-Vega N, Alfonso-Pérez M, Benedicto I, Sánchez-Madrid F, González-Amaro R, Marazuela M. Increased circulating pro-inflammatory cytokines and Th17 lymphocytes in Hashimoto’s thyroiditis. J Clin Endocrinol Metab . 2010; 95( 2): 953– 962. Google Scholar CrossRef Search ADS PubMed  60. González-Amaro R, Marazuela M. T regulatory (Treg) and T helper 17 (Th17) lymphocytes in thyroid autoimmunity. Endocrine . 2015; 52( 1): 30– 38. Google Scholar CrossRef Search ADS PubMed  61. Mandolesi G, De Vito F, Musella A, Gentile A, Bullitta S, Fresegna D, Sepman H, Di Sanza C, Haji N, Mori F, Buttari F, Perlas E, Ciotti MT, Hornstein E, Bozzoni I, Presutti C, Centonze D. miR-142-3p is a key regulator of IL-1β-dependent synaptopathy in neuroinflammation. J Neurosci . 2016; 37( 3): 546– 561. Google Scholar CrossRef Search ADS   62. Tang X, Yin K, Zhu H, Tian J, Shen D, Yi L, Rui K, Ma J, Xu H, Wang S. Correlation between the expression of microRNA-301a-3p and the proportion of Th17 cells in patients with rheumatoid arthritis. Inflammation . 2016; 39( 2): 759– 767. Google Scholar CrossRef Search ADS PubMed  Copyright © 2018 Endocrine Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Clinical Endocrinology and Metabolism Oxford University Press

Loading next page...
 
/lp/ou_press/a-microrna-signature-for-evaluation-of-risk-and-severity-of-autoimmune-STfs9RbLhk
Publisher
Endocrine Society
Copyright
Copyright © 2018 Endocrine Society
ISSN
0021-972X
eISSN
1945-7197
D.O.I.
10.1210/jc.2017-02318
Publisher site
See Article on Publisher Site

Abstract

Abstract Context Circulating microRNAs (miRNAs) are emerging as an interesting research area because of their potential role as novel biomarkers and therapeutic targets. Their involvement in autoimmune thyroid diseases (AITDs) has not been fully explored. Objective To compare the expression profile of miRNAs in thyroid tissue from patients with AITD and controls, using next-generation sequencing, further validated our findings in thyroid and serum samples. Design Twenty fresh-frozen thyroid tissues (15 from patients with AITD and 5 from controls) were used for miRNA next-generation sequencing. Thirty-six thyroid samples were recruited for the qRT-PCR validation test and 58 serum samples for further validation in peripheral blood. Results Expression of several miRNAs that had been previously associated with relevant immunological functions was significantly dysregulated. Specifically, eight differentially expressed miRNAs (miR-21-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, and miR-766-3p) were confirmed using qRT-PCR in thyroid samples, and three had the same behavior in tissue and serum samples (miR-21-5p, miR-142-3p, and miR-146a-5p). Furthermore, when the expression of these miRNAs was assessed together with five additional ones previously related to AITD in peripheral blood, the expression of five (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) was significantly expressed in AITD and, in patients with Graves disease (GD), was correlated with a higher severity of disease, including active ophthalmopathy, goiter, higher antibody titers, and/or higher recurrence rates. Conclusions The present findings identify a serum five-signature miRNA that could be an independent risk factor for developing AITD and a predisposition of a worse clinical picture in patients with GD. Autoimmune thyroid disorders (AITDs) result from a dysregulation of the immune system directed against the thyroid and include a broad spectrum of diseases. The two main phenotypes of AITD, Hashimoto thyroiditis (HT) and Graves disease (GD), are both characterized by the presence of circulating thyroid antibodies and infiltration by autoreactive lymphocytes in the thyroid gland and sometimes the orbit. It has been traditionally thought that HT is mainly mediated by a cellular autoimmune response, whereas GD mainly has been considered to be mediated by a humoral response, mainly due to the presence of autoantibodies directed against the thyrotropin receptor [antithyrotropin receptor antibody (TSHR-Ab)], which lead to development of goiter and hyperthyroidism (1–3). Moreover, both immune responses, including TSHR-Ab, also have been reported in the pathogenesis of Graves ophthalmopathy (GO), one of the most common extrathyroid manifestations of AITD, which may occur clinically in up to 25% of patients with GD (4). The pathogenesis of AITD is probably related to a complex and multifactorial interplay of specific susceptibility genes and environmental exposures, leading to the breakdown of self-tolerance and subsequent development of autoimmune diseases. MicroRNAs (miRNAs) are small non-protein-coding RNA molecules of ∼22 nucleotides that interact with their targeted RNA in a sequence-dependent manner and therefore function as regulators of gene expression at a posttranscriptional level mainly by repressing the translation and also by decreasing target messenger RNA levels (5). Currently, nearly 3000 human miRNAs have been annotated in the miRBase (6). miRNAs have been recently involved in several biological processes, including immune functions, cellular apoptosis, cell differentiation and development, proliferation, and metabolism. The increasing importance of miRNAs highlights their potential role as biomarkers of disease and even their utility as therapeutic targets. In recent years, circulating miRNAs have been an interesting area of research due to their stability and reproducibility (7), and studies describing the role of miRNAs in the immune response and in autoimmune diseases have progressively developed. In this regard, evidence of differential miRNA expression has been reported in numerous immunological disorders such as rheumatoid arthritis, systemic lupus erythematosus, Sjögren disease, and psoriasis, among others (7–11). Several reports have been published over the past years regarding the specific scenario of miRNAs and AITD with discordant results (12–27). These discrepancies can be attributed to the differences between the techniques used [microarray and/or quantitative reverse transcription polymerase chain reaction (qRT-PCR) assays], the materials analyzed (peripheral blood mononuclear cells, serum, plasma, thyrocytes, orbital fibroblasts, and thyroid tissue), the different preservation methods (frozen vs paraffin embedded), and the specific type of individual AITD evaluated (all AITD, HT, GD, or GO). However, none of these previous studies have analyzed miRNAs in fresh thyroid tissue, in the whole spectrum of AITD, using next-generation sequencing (NGS). The interest in this latter technique lies in the fact that, in comparison with microarrays, NGS has a greater ability to capture the scale and complexity of whole transcriptomes. Furthermore, NGS allows the discovery of novel miRNAs and can identify miRNAs that are expressed at levels that fall below microarrays’ detectable threshold. In this study, therefore, we aimed to identify differentially expressed (DE) miRNAs in AITD thyroid tissue samples in comparison with controls by means of NGS and then corroborated their expression in serum samples. We then correlated their expression with patients’ clinical data to identify potential diagnostic biomarkers and therapeutic targets of the disease. Materials and Methods Patients Fresh-frozen thyroid tissue samples from 26 patients with AITD (9 with HT, 7 with GD without GO, and 10 with GD and GO) and 10 controls were collected. In addition, we collected formalin-fixed, paraffin-embedded (FFPE) samples from 19 patients with AITD (7 with HT, 6 with GD without GO, and 6 with GD and GO), of which 11 were the same as the frozen tissue, and 6 were healthy controls. Concomitantly, we also evaluated serum samples from 36 patients with AITD (14 with HT, 10 with GD without GO, and 12 with GD with GO) and 22 healthy controls. Clinical diagnoses were all reviewed by a single experienced endocrinologist; established on commonly accepted clinical, laboratory, and histological criteria (2); and based on information collected from clinical records (Tables 1 and 2). More details of clinical data are available in the Supplemental Material and Methods. Table 1. Clinical Features of Patients With AITD From Whom Thyroid Tissue Was Collected Characteristic  HT (n = 9)  GD (n = 17)  Controls (n = 10)  Sex, female/male  9/0  15/2  7/3  Age, y  63 (50.5–71.5)  39 (29.5–52.5)  43 (32.2–51.8)  TSH, μU/mL  2.3 (1–8.5)  1.8 (0.2–9.5)  0  FT4, ng/dL  1.1 (0.9–1.2)  1.1 (0.8–1.3)  0  Tg-Ab, UI/mL  311 (213–382.3)  456 (20–1103)  0  TPO-Ab, UI/mL  150 (20–306)  102 (43.2–635.5)  0  TSHR-Ab, U/L  0.5 (0.5–0.5)  5.6 (1.7–17.5)  0  Characteristic  HT (n = 9)  GD (n = 17)  Controls (n = 10)  Sex, female/male  9/0  15/2  7/3  Age, y  63 (50.5–71.5)  39 (29.5–52.5)  43 (32.2–51.8)  TSH, μU/mL  2.3 (1–8.5)  1.8 (0.2–9.5)  0  FT4, ng/dL  1.1 (0.9–1.2)  1.1 (0.8–1.3)  0  Tg-Ab, UI/mL  311 (213–382.3)  456 (20–1103)  0  TPO-Ab, UI/mL  150 (20–306)  102 (43.2–635.5)  0  TSHR-Ab, U/L  0.5 (0.5–0.5)  5.6 (1.7–17.5)  0  Values show number for categorical values and median (25th to 75th interquartile intervals) for continuous variables. FT4, free thyroxine (normal range, 0.93 to 1.7 ng/dL); Tg-Ab (negative <344 UI/mL); TPO-Ab (negative <100 UI/mL); TSH, thyrotropin (normal range, 0.27 to 4.20 mU/mL); TSHR-Ab (negative <0.7 U/L). View Large Table 2. Clinical Features of Patients With AITD From Whom Serum Was Collected Characteristic  HT (n = 14)  GD (n = 22)  Controls (n = 22)  Euthyroid (n = 8)  Hypothyroid (n = 6)  Hyperthyroid (n = 15)  Hypothyroid (n = 7)  Sex, female/male  6/2  4/2  11/4  7/0  9/13  Age, y  47 (31.2–62.7)  39 (28.5–54.2)  45 (30–54)  46 (41–54)  31 (25.5–40)  Ophthalmopathy  0  0  8  4  0  TSH, μU/mL  2.5 (1.5–2.9)  10.7 (4.9–23.9)  0  6.55 (3.21–23.96)  —  FT4, ng/dL  1.2 (0.9–1.3)  1.2 (0.9–1.4)  3 (1.9–4.5)  0.61 (0.36–1)  —  Tg-Ab, UI/mL  0 (0–986.8)  398 (0–1198)  0 (0–764)  0 (0–70.2)  0  TPO-Ab, UI/mL  424 (126.3–844.8)  252 (0–669.5)  156 (0–542.8)  98.6 (0–273)  0  TSHR-Ab, U/L  0.2 (0–1.1)  0 (0–0.8)  3.4 (1.9–5.6)  3.56 (1.66–6.3)  0.01 (0–0.07)  Characteristic  HT (n = 14)  GD (n = 22)  Controls (n = 22)  Euthyroid (n = 8)  Hypothyroid (n = 6)  Hyperthyroid (n = 15)  Hypothyroid (n = 7)  Sex, female/male  6/2  4/2  11/4  7/0  9/13  Age, y  47 (31.2–62.7)  39 (28.5–54.2)  45 (30–54)  46 (41–54)  31 (25.5–40)  Ophthalmopathy  0  0  8  4  0  TSH, μU/mL  2.5 (1.5–2.9)  10.7 (4.9–23.9)  0  6.55 (3.21–23.96)  —  FT4, ng/dL  1.2 (0.9–1.3)  1.2 (0.9–1.4)  3 (1.9–4.5)  0.61 (0.36–1)  —  Tg-Ab, UI/mL  0 (0–986.8)  398 (0–1198)  0 (0–764)  0 (0–70.2)  0  TPO-Ab, UI/mL  424 (126.3–844.8)  252 (0–669.5)  156 (0–542.8)  98.6 (0–273)  0  TSHR-Ab, U/L  0.2 (0–1.1)  0 (0–0.8)  3.4 (1.9–5.6)  3.56 (1.66–6.3)  0.01 (0–0.07)  Values show number for categorical values and median (25th to 75th interquartile intervals) for continuous variables. FT4, free thyroxine (normal range, 0.93 to 1.7 ng/dL); Tg-Ab (negative <344 UI/mL); TPO-Ab (negative <100 UI/mL); TSH, thyrotropin (normal range, 0.27 to 4.20 mU/mL); TSHR-Ab (negative <0.7 U/L). View Large The project was approved by the Internal Ethical Review Committee of the Hospital de la Princesa, and written informed consent was obtained from all patients prior to inclusion, in accordance with the Declaration of Helsinki. Tissue and serum samples All thyroid tissues were obtained from surgical thyroidectomies performed in our hospital and reviewed by an experienced pathologist. Samples were immediately snap-frozen in liquid nitrogen–cooled isopentane and transferred to a −80°C freezer for long-time preservation or processed by the Department of Pathology to create FFPE tissues. Serum samples were isolated by centrifugation (1500 × g) from 10 mL of total blood and stored at −80°C until use. Tissue miRNA isolation RNA was isolated from 36 fresh-frozen thyroid tissues: 20 samples were used for miRNA NGS and all 36 for the qRT-PCR validation test. Total RNA was isolated using miRNeasy Mini Kit (Qiagen, Germantown, MD), according to the manufacturer’s instructions. The quality and quantity of RNA and miRNA were assessed in a 2100 Bioanalyzer by using a RNA 6000 Nano kit and a Small RNA kit (Agilent Technologies, Palo Alto, CA). For the study, only those samples with an RNA integrity number >7 were included. To empower the statistical analysis, we isolated miRNA from 25 FFPE samples, 12 of which were repeated from frozen samples to verify their quality and reproducibility. FFPE samples were isolated using the miRneasy FFPE Mini kit (Qiagen). Serum miRNA isolation To test hemolysis in serum samples, the absorbance of free hemoglobin at 414 nm was measured. Those samples with a peak >0.2 were discarded. miRNA from serum samples were isolated using miRCURY RNA Isolation Kit Biofluids (Exiqon, Vedbaek, Denmark), according to the manufacturer’s instruction. A mixture containing 1.25 μg/mL MS2 bacteriophage RNA and RNA spike-ins (UniSp2, UniSp4, and UniSp5) was added to 200 μL serum. RNA was purified on a miRNA mini-spin column BF, eluted in 50 μL RNase-free water, and stored at −80°C. The robustness of the RNA isolation process and the quality of isolated miRNA were assessed using miRCURY microRNA QC Panels (Exiqon). miRNA NGS NGS of 20 thyroid samples (5 controls, 5 with HT, 5 with GD without GO, and 5 with GD with GO) was performed on the Illumina platform (HiSEquation 2000; Illumina, San Diego, CA) at Sistemas Genómicos (Valencia, Spain). The procedure included the following steps/phases: (1) Quality control of RNA: the quality and the quantity of RNA were determined in a Bioanalyzer 2100 Small RNA assay and Qubit 2.0 fluorometer. (2) Preparation of libraries: complementary DNA (cDNA) libraries were obtained following Illumina’s recommendations. Briefly, 3′ and 5′ adaptors were sequentially ligated to the RNA prior to reverse transcription and cDNA generation. cDNA was enriched with polymerase chain reaction (PCR) to create the indexed double-stranded cDNA library. Size selection was performed using 6% polyacrylamide gel. The quality of the libraries was analyzed using Bioanalyzer 2100, High Sensitivity assay, and the quantity of the libraries was determined by real-time PCR in Light Cycler 480 (Roche Farma, Madrid, Spain). (3) Equimolar pooling of libraries was performed before proceeding to the generation of clusters in cbot (Illumina). The pool of the cDNA libraries was sequenced by paired-end sequencing (100 × 2) in the Illumina HiSEquation 2000 sequencer. The bioinformatic analysis of NGS and raw data availability is described in Supplemental Material and Methods. Reverse transcription and qRT-PCR of tissue and serum samples First-strand cDNA was generated using a cDNA synthesis kit, and subsequent qRT-PCR was performed in triplicate using miRCURY LNA Universal RT microRNA PCR ExiLENT SYBR Green (both from Exiqon) with the CFX384 Touch Real-Time PCR Detection System (Bio-Rad, Alcobendas, Madrid). Expression of miRNAs was performed using miRNA LNA PCR primer sets (Exiqon). Expression stabilities of the reference genes were evaluated using geNorm (28), NormFinder (29), and the coefficient of variation score described by Marabita et al. (30). Data were normalized using the geometric mean Ct of the best gene combination generated by the algorithms (Supplemental Table 1 and Supplemental Fig. 1). Relative expression was determined using the log base 2 values of the difference Cts between miRNAs and the geometric mean of the selected housekeeping genes. See Supplemental Material and Methods for detailed technical descriptions of the normalization. Statistical analysis Descriptive results were expressed as mean ± standard deviation, mean ± error of the mean, or median and 25th to 75th percentiles, as appropriate. Spearman bivariate correlations were performed for all quantitative variables and differences between groups were compared using analysis of variance (Mann-Whitney U or Kruskal-Wallis analysis of variance, as appropriate). A logistic regression model was used to determine the differences in each normalized miRNA between controls and AITD groups, adjusted by age and sex, as recommended in previous reports (31). Receiver operating characteristic (ROC) curve analyses were performed to assess the classification power of each logistic regression model. Samples from all groups within an experiment were processed at the same time. The P values were two-sided, and statistical significance was considered when P < 0.05. Data are presented with the following specific P values: P < 0.05, P < 0.01, and P < 0.001. All statistical analyses were performed using GraphPad Prism 4 software (GraphPad Software, La Jolla, CA) and STATA (StataCorp LLC, College Station, TX). Results miRNA expression in patients with AITD Unsupervised hierarchical clustering and principal component analysis were used to investigate the potential identification of AITD subgroups, based on their molecular expression profile. We performed NGS using RNA obtained from thyroids from five patients with HT, five patients with GD with no GO, five patients with GD with GO, and five healthy participants. The analysis revealed a slightly but not substantial separation between AITD groups. Control samples clustered together and formed a separate subcluster, indicating that these samples have a very similar miRNA signature and that this signature was different from those of AITD samples. Regarding the principal component analysis, an outlier sample from the GD group was detected and excluded from the study (Fig. 1A and 1B). We, therefore, decided to assess the differential miRNA expression of all AITD samples together in the same group, in comparison with the control group. A total of 3431 sequences were detected. In particular, 19 miRNAs were upregulated and 1 was downregulated more than twofold in AITD vs normal thyroid. Of these, miR-146b-5p displayed the most significant upregulation, with a log twofold change of 6.303 and P < 0.00001. Figure 1. View largeDownload slide (A) Heatmap hierarchical clustering of DE miRNAs in AITD tissue samples on the basis of 19 DE miRNAs. The graph shows miRNA expression in thyroid tissue from AITD samples compared with controls, expressed in fold change. The key color bar indicates that miRNA expression levels increased from green to red, depending on the fold change. (B) Principal component analysis (PCA) projected as a two-dimensional scatterplot. The most correlated samples are circled in different colors (blue = controls, green = HT, brown = GD, and gray = GO). Circled in red is an outlier that was excluded from the analysis (see text for full explanation). (C) Cluster dendrogram of miRNAs that have been associated by immunological enrichment analysis (false discovery rate <0.05). miR-21-5p, miR-155-5p, and miR-146a-5p are the most associated with immunological pathways. (D) Top 10 dysregulated miRNAs in AITD samples selected for validation by qRT-PCR (P < 0.05) and fold change ± 2 calculated using the algorithm DESeq2; base mean indicates the mean of reads of the miRNAs adjusted by the mean of the total reads generated in the NGS. Figure 1. View largeDownload slide (A) Heatmap hierarchical clustering of DE miRNAs in AITD tissue samples on the basis of 19 DE miRNAs. The graph shows miRNA expression in thyroid tissue from AITD samples compared with controls, expressed in fold change. The key color bar indicates that miRNA expression levels increased from green to red, depending on the fold change. (B) Principal component analysis (PCA) projected as a two-dimensional scatterplot. The most correlated samples are circled in different colors (blue = controls, green = HT, brown = GD, and gray = GO). Circled in red is an outlier that was excluded from the analysis (see text for full explanation). (C) Cluster dendrogram of miRNAs that have been associated by immunological enrichment analysis (false discovery rate <0.05). miR-21-5p, miR-155-5p, and miR-146a-5p are the most associated with immunological pathways. (D) Top 10 dysregulated miRNAs in AITD samples selected for validation by qRT-PCR (P < 0.05) and fold change ± 2 calculated using the algorithm DESeq2; base mean indicates the mean of reads of the miRNAs adjusted by the mean of the total reads generated in the NGS. Gene ontology enrichment of differentially expressed miRNAs To evaluate the association of miRNAs with relevant immunological functions, target genes biologically annotated for the DE miRNAs were identified using miRTarBase and TargetScan. Gene ontology–selected immunological enriched processes were associated with DE miRNAS (Fig. 1C and Supplemental Table 2). The analysis of biological network exploration showed the correlation and cooperation of miRNAs related to the immunological system (Supplemental Fig. 2). In fact, we can observe the cooperation of pathway reactions relative to the immune and adaptive system of miR-96-5p, miR-142-3p, miR-146a-5p, and miR-146b-5p. At the same time, miR-21-5p and miR-6503-3p cooperate with miR-146a-5p in processes related to the Toll-like receptor and interferon signaling pathways, respectively. Considering this, 10 top-ranked selected miRNAs importantly associated with immune pathways were selected for validation by qRT-PCR: miR-21-5p, miR-96-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, miR-766-3p, and miR-6503-3p (Fig. 1D). Validation of significant differentially expressed miRNAs in thyroid tissue To validate the top-ranked DE selected miRNAs, qRT-PCR was performed in 36 fresh-frozen and 25 FFPE thyroid tissue from the same samples. There was a good correlation in 5 of 10 miRNAs in the same samples extracted from fresh-frozen and FFPE tissues (P < 0.05) (Supplemental Fig. 3). In this context, we decided to exclude FFPE samples from the study and continue only with the expressions related to frozen samples. Then, comparison of the expression levels performed between NGS data and qRT-PCR in frozen tissues demonstrated a similar expression pattern between both methods (Fig. 2A). Of the top 10 selected miRNAs, 8 were confirmed to be DE in AITD samples (miR-21-5p, P < 0.01; miR-142-3p, P < 0.01; miR-146a-5p, P < 0.001; miR-146b-5p, P < 0.001; miR-155-5p, P < 0.05; miR-338-5p, P < 0.05; miR-342-5p, P < 0.001; and miR-766-3p, P < 0.001). However, although there was a trend toward being increased in AITD, no significant differences were found for miR-96-5p and miR-6503-5p between the two groups (Fig. 2B). Figure 2. View largeDownload slide (A) Comparison of miRNA expression between NGS and qRT-PCR. miRNAs determined to be DE in patients with AITD by NGS were validated by qRT-PCR. The height of the columns in the chart represents the log-transformed average fold change in expression, comparing the expression levels of AITD samples with normal thyroid tissues for each of the validated miRNAs. A total of 10 miRNAs were selected for validation, and 8 were confirmed by qRT-PCR. Data are expressed as mean ± standard error of the mean. (B) miRNAs in the validation study in tissue and serum samples. Eight different miRNAs were DE in 26 patients with AITD compared with 10 controls measured in fresh-frozen thyroid tissue samples (miR-21-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, and miR-766-3p). Of the total circulating miRNAs evaluated in 56 samples (36 patients with AITD and 22 healthy controls), miR-21-5p, miR-96-5p, miR-142-3p, and miR-146a-5p were DE in patients with AITD vs controls. Furthermore, of the five miRNAs included in the study, miR-Let7d-5p, miR-142-5p, and miR-301a-3p were also DE in serum samples. (C) Normalized relative quantities (NRQs) of miRNAs in serum samples analyzed according to the different subgroups of AITD: HT and GD. miR-Let7d-5p was DE in HT, mir-142-3p in GD, and miR-21-5p, 96-5p, and 301-3p in both HT and GD. miR146a-5p and miR-6503-3p lost significance when we distinguished between HT and GD. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Figure 2. View largeDownload slide (A) Comparison of miRNA expression between NGS and qRT-PCR. miRNAs determined to be DE in patients with AITD by NGS were validated by qRT-PCR. The height of the columns in the chart represents the log-transformed average fold change in expression, comparing the expression levels of AITD samples with normal thyroid tissues for each of the validated miRNAs. A total of 10 miRNAs were selected for validation, and 8 were confirmed by qRT-PCR. Data are expressed as mean ± standard error of the mean. (B) miRNAs in the validation study in tissue and serum samples. Eight different miRNAs were DE in 26 patients with AITD compared with 10 controls measured in fresh-frozen thyroid tissue samples (miR-21-5p, miR-142-3p, miR-146a-5p, miR-146b-5p, miR-155-5p, miR-338-5p, miR-342-5p, and miR-766-3p). Of the total circulating miRNAs evaluated in 56 samples (36 patients with AITD and 22 healthy controls), miR-21-5p, miR-96-5p, miR-142-3p, and miR-146a-5p were DE in patients with AITD vs controls. Furthermore, of the five miRNAs included in the study, miR-Let7d-5p, miR-142-5p, and miR-301a-3p were also DE in serum samples. (C) Normalized relative quantities (NRQs) of miRNAs in serum samples analyzed according to the different subgroups of AITD: HT and GD. miR-Let7d-5p was DE in HT, mir-142-3p in GD, and miR-21-5p, 96-5p, and 301-3p in both HT and GD. miR146a-5p and miR-6503-3p lost significance when we distinguished between HT and GD. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Expression of identified thyroid DE miRNAs in serum samples We then determined whether the selected DE miRNAs were detected in serum samples of 8 patients with active GO and 4 patients with inactive GO, 10 patients with GD with no GO, 14 patients with HT, and 22 healthy controls. In addition to the 10 DE miRNAs identified in thyroid tissue samples, 5 other miRNAs involved in immune functions or identified in previous studies were also included for validation in AITD serum samples (miR-Let7d-5p, miR-142-5p, miR-126-3p, miR-223-3p, and miR-301a-3p) (24, 32–34). Eight of this whole set of miRNAs analyzed were DE in AITD samples (miR-Let7d-5p, P < 0.001; miR-21-5p, P < 0.0001; miR-96-5p, P < 0.001; miR-142-3p, P < 0.01; miR142-5p, P < 0.001; miR146a-5p, P < 0.05; miR-301a-3p, P < 0.001; and miR-6503-3p, P < 0.05) (Fig. 2B). All were upregulated in AITD, except for miR-Let-7d, which was downregulated. To discriminate if these miRNAs were dysregulated according to the type of AITD (HT or GD), we analyzed the levels of DE miRNAs in each subgroup; miR-Let7d-5p was DE in HT, mir-142-3p in GD, and miR-21-5p, miR-96-5p, and miR-301-3p in both HT and GD. Although miR-146a-5p and miR-6503-3p lost significance when both groups were considered separately, we observed the same trend of dysregulation in all of them, denoting the distinction between controls and HT/GD (Fig. 2C and Fig. 3A). Figure 3. View largeDownload slide (A) Overall study design and workflow. NGS was performed in 20 thyroid tissues. Ten miRNAs were selected for validation by qRT-PCR, and 8 were confirmed. In five cases, miRNAs from NGS were confirmed in serum samples. Five additional miRNAs were selected bibliographically. Of the 15 miRNAs studied in serum samples (10 from NGS and 5 selected bibliographically), 5 were confirmed as a specific miRNA signature. (B) Roles of the five-signature miRNA (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) found in AITD samples. The lower levels of miR-Let7d-5p fall to the suppression of Th1 cells by Treg cells, leading to Th1 cell proliferation and interferon-γ (IFN-γ) secretion (32). The increase of miR-142-3p could impair the inhibitory effect of Treg cells on the proliferative response and cytokine production of CD4+CD25− T cells (35). miR-21-5p can regulate the Th1/Th2 balance and promote Th17 differentiation by inhibiting Smad-7 (36). miR-96-5p can increase cytokine production of Th17 cells (37). miR-301a-3p can downregulate PIAS3, promoting Th17 development (34). The altered expression of all these miRNAs can alter the normal function of immune cells, leading to a loss of immune tolerance in AITD. TNFα, tumor necrosis factor α. Figure 3. View largeDownload slide (A) Overall study design and workflow. NGS was performed in 20 thyroid tissues. Ten miRNAs were selected for validation by qRT-PCR, and 8 were confirmed. In five cases, miRNAs from NGS were confirmed in serum samples. Five additional miRNAs were selected bibliographically. Of the 15 miRNAs studied in serum samples (10 from NGS and 5 selected bibliographically), 5 were confirmed as a specific miRNA signature. (B) Roles of the five-signature miRNA (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) found in AITD samples. The lower levels of miR-Let7d-5p fall to the suppression of Th1 cells by Treg cells, leading to Th1 cell proliferation and interferon-γ (IFN-γ) secretion (32). The increase of miR-142-3p could impair the inhibitory effect of Treg cells on the proliferative response and cytokine production of CD4+CD25− T cells (35). miR-21-5p can regulate the Th1/Th2 balance and promote Th17 differentiation by inhibiting Smad-7 (36). miR-96-5p can increase cytokine production of Th17 cells (37). miR-301a-3p can downregulate PIAS3, promoting Th17 development (34). The altered expression of all these miRNAs can alter the normal function of immune cells, leading to a loss of immune tolerance in AITD. TNFα, tumor necrosis factor α. A five-signature miRNA to discriminate patients with AITD vs controls: correlation with clinical features To assess the existence of any possible associations between miRNA levels and clinical features and explore their clinical utility, we first investigated the correlations of all the analyzed miRNAs with patients’ corresponding serum levels of free thyroxine, thyrotropin, antithyroid peroxidase antibody (TPO-Ab), antithyroglobulin antibody (Tg-Ab), and TSHR-Ab (correlation heatmap in Fig. 4A). Spearman ρ analyses revealed that miR-Let7d-5p was negatively associated with levels of TPO-Ab (r = −0.38), whereas miR-21-5p and miR-96-5p were positively correlated with TPO-Ab (r = 0.39 and r = 0.48, respectively), Tg-Ab (r = 0.36 and r = 0.4, respectively), and TSHR-Ab (r = 0.45 and r = 0.4, respectively). miR-142-3p and miR-301a-3p were only positively associated with TSHR-Ab (r = 0.38 and r = 0.31), and miR-6503-3p correlated with TPO-Ab (r = 0.27). Interestingly, these miRNAs (except for miR-6503-3p) were DE in HT and/or GD. Also, all miRNAs were significantly correlated between themselves, with a remarkable similarity between miR-21-5p and miR-96-5, which exhibited an analogous behavior regarding their correlation with clinical features (Fig. 4B). Figure 4. View largeDownload slide (A) Correlation heatmap between the expression of miRNAs and clinical and laboratory parameters (Spearman ρ analysis). Significant negative correlations are shown in red and significant positive correlations in green (*P < 0.05; **P < 0.01). Bottom left triangle shows correlations between miRNAs and top square shows correlations between miRNAs and clinical data. miR-Let7d-5p is negatively associated with levels of TPO-Ab (r = −0.38). miR-21-5p and miR-96-5p are positively correlated with TPO-Ab (r = 0.39 and r = 0.48, respectively), Tg-Ab (r = 0.36 and r = 0.4 respectively), and TSHR-Ab (r = 0.45 and r = 0.4 respectively). miR-142-5p and miR-301a-3p are positively associated with TSHR-Ab (r = 0.38 and r = 0.31), and miR-6503-3p correlates with TPO-Ab (r = 0.27). (B) Correlation analysis between miR-21-5p and miR-96-5p (r = 0.58 and P < 0.001). (C) ROC curve analyses performed to assess the diagnostic value of circulating miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301-3p to discriminate between controls and patients with AITD. The table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. Figure 4. View largeDownload slide (A) Correlation heatmap between the expression of miRNAs and clinical and laboratory parameters (Spearman ρ analysis). Significant negative correlations are shown in red and significant positive correlations in green (*P < 0.05; **P < 0.01). Bottom left triangle shows correlations between miRNAs and top square shows correlations between miRNAs and clinical data. miR-Let7d-5p is negatively associated with levels of TPO-Ab (r = −0.38). miR-21-5p and miR-96-5p are positively correlated with TPO-Ab (r = 0.39 and r = 0.48, respectively), Tg-Ab (r = 0.36 and r = 0.4 respectively), and TSHR-Ab (r = 0.45 and r = 0.4 respectively). miR-142-5p and miR-301a-3p are positively associated with TSHR-Ab (r = 0.38 and r = 0.31), and miR-6503-3p correlates with TPO-Ab (r = 0.27). (B) Correlation analysis between miR-21-5p and miR-96-5p (r = 0.58 and P < 0.001). (C) ROC curve analyses performed to assess the diagnostic value of circulating miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301-3p to discriminate between controls and patients with AITD. The table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. We assessed the discriminating potency between AITD and controls of these circulating miRNAs using ROC curve analyses (Fig. 4C), and we confirmed that five miRNAs (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301a-3p) were good discriminators [area under the curve (AUC) ≥0.848, with >76% correctly classified instances]. Then, we assessed the prognostic role of miRNAs in GD by correlating DE miRNAs of the signature with clinical parameters, including persistent and/or severe forms of hyperthyroidism. Relative expression of miR-Let7d-5p, miR-96-5p, and miR-301a-3p was significantly different in patients with GO than in controls (P = 0.0073, P = 0.0055, and P = 0.0374, respectively), although no differences were found between controls and patients with GD without GO (P > 0.99, P = 0.06, and P = 0.42, respectively) (Fig. 5A). However, miR-Let7d-5p was significantly lower in patients with GO compared with patients with GD without GO (P = 0.016) and more remarkably in those with active GO (P = 0.0016) (Fig. 5B), which also had a negative correlation with clinical activity score (r = −0.6504, P = 0.0014) (Fig. 5C). Patients with more severe disease had lower miR-Let7d-5p levels (P = 0.034) (Fig. 5D). For miR-21-5p, a higher expression was significantly correlated with a worse prognosis, with more patients presenting recurrent disease (P = 0.028) (Fig. 5E and Supplemental Table 3). In fact, after adjusting for age, sex, and TSHR-Ab, miR-21-5p remained significant (P = 0.042). Subsequent analysis using the method of all possible equations showed that combining miR-21-5p and TSHR-Ab in the regression model achieved the best predictive value for disease persistence. In this regard, ROC curves for miR-21-5p, TSHR-Ab, and a composite model of both markers (Fig. 5F) found that baseline miR-21-5p alone had an AUC of 0.85 and a percentage of correctly classified instances of 85%, whereas baseline TSHR-Ab alone had an AUC of 0.64 and a prediction of 65% at the cut point of 3. The composite marker had an AUC of 0.92 and a percentage of prediction of 90%. Figure 5. View largeDownload slide Correlation of levels of miRNAs with clinical data of patients with GD. (A) Classification of patients with GD according to the presence or absence of GO. Relative expression of miR-Let7d-5p, miR-96-5p, and miR-301-3p was DE in the GO group compared with controls. In addition, miR-Let7d-5p was DE between patients with GD with and without GO. (B) The expression levels of miR-Let7d-5p were DE in patients with GD with active GO [clinical activity score (CAS) >3]. (C) There was a negative correlation between the expression of miR-Let7d-5p and CAS. (D) Levels of miR-Let7d-5p according to disease severity; patients with severe disease had lower levels of miR-Let7d-5p. (E) miR-21-5p was higher in patients with persistent disease. (F) ROC curves were performed to analyze the predictive values of regression models for disease persistence based on miR-21-5p, TSHR-Ab, and the composite of both parameters. Table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Figure 5. View largeDownload slide Correlation of levels of miRNAs with clinical data of patients with GD. (A) Classification of patients with GD according to the presence or absence of GO. Relative expression of miR-Let7d-5p, miR-96-5p, and miR-301-3p was DE in the GO group compared with controls. In addition, miR-Let7d-5p was DE between patients with GD with and without GO. (B) The expression levels of miR-Let7d-5p were DE in patients with GD with active GO [clinical activity score (CAS) >3]. (C) There was a negative correlation between the expression of miR-Let7d-5p and CAS. (D) Levels of miR-Let7d-5p according to disease severity; patients with severe disease had lower levels of miR-Let7d-5p. (E) miR-21-5p was higher in patients with persistent disease. (F) ROC curves were performed to analyze the predictive values of regression models for disease persistence based on miR-21-5p, TSHR-Ab, and the composite of both parameters. Table shows the AUC, the percentage of correctly predicted instances, 95% confidence intervals (95% CIs), sensitivity, and specificity of the analyses. Data are presented as mean ± standard deviation. *P < 0.05. **P < 0.01. ***P < 0.001. ****P = 0.0001. Discussion Accumulating evidence suggests that miRNAs can be used as biomarkers of various diseases, including various autoimmune disorders (38). In addition to tissue-derived miRNAs, circulating miRNAs, which are protected from endogenous RNase activity, are specific and stable and can be used as extracellular biomarkers. The advantage of using circulating miRNAs is that the diagnostic approach could be minimized to a single blood sample (7). Although studies have investigated miRNAs in patients with AITD, few have explored their clinical and diagnostic utility and their role in risk stratification. In this study, we describe the global patterns of miRNA expression found in AITD thyroid samples. We found a good correlation between the expression profiles using NGS and qRT-PCR in fresh-frozen thyroid tissues, suggesting that NGS could be a reliable method for miRNA profiling in AITD. In fact, using NGS, we selected 10 miRNAs in accordance to immune pathways (cytokine signaling, Toll-like receptor cascade, interferon signaling, B-cell receptor signaling) and confirmed that 8 miRNAs were DE in the thyroid and 3 (miR-21-5p, miR-142-3p, and miR-146a-5p) in serum samples of patients with AITD. These results suggest that some of these circulating miRNAs are synthetized in thyroid tissue and are then released to human body fluids, including the blood, as a form of intercellular communication, so they could be used to monitor the altered tissue. Some of the miRNAs found in the thyroid (miR-21-5p, miR-146a-5p, and miR-155-5p) have been widely reported to participate in lymphocyte differentiation and activation (36, 39–43). For instance, miR-146 has been associated with regulation of the immune response (39, 41, 43) and miR-155 to T-regulatory cell development and Th17 differentiation (40, 42). Regarding AITD, studies on these miRNAs have been discordant, with reports of decreased or increased expression of miR-21-5p, miR-146a-5p, and/or miR-155-5p, depending on the specific type of disease (AITD vs HT vs GD), the tissue studied (thyroid tissue, needle aspiration samples, peripheral blood mononuclear cells, microvesicles, and/or serum), or the conservation method used (fresh samples vs FFPE) (12, 13, 20–22, 24, 44), hence the importance of studying homogeneous groups of patients, samples, and preservation methods. In our study, we found some differences in DE miRNAs when we analyzed AITD as a whole or as subgroups or in thyroid vs serum samples. Moreover, there was a poor correlation between the relative expression of miRNA in formalin-fixed and frozen samples, which forced us to exclude FFPE samples from our analysis to provide more robustness to our results (45–48). Taken together, all these issues could potentially explain the discrepancies observed between previous publications and our current findings. To further evaluate if these miRNAs could behave as biomarkers of risk for developing AITD, we studied the top 10 miRNAs found in NGS and 5 additional miRNAs (miR-Let7d-5p, miR-126-3p, miR-142-5p, miR-223-3p, and miR-301a-3p), which have been previously reported to be involved in different autoimmune disorders, including AITD (24, 32–34). We found a good correlation between some of these miRNAs and all three thyroid autoantibody levels. Because these autoantibodies are frequently used to establish the risk of AITD, we assessed whether miRNAs could be also used in the same way. To do so, we performed a composite marker by a logistic regression model and found that a specific five-serum miRNA signature (miR-Let7d-5p, miR-21-5p, miR-96-5p, miR-142-3p, and miR-301-3p) was able to assign a risk for developing AITD independently of autoantibody titers. We also evaluated if this signature could be the used in the clinical evaluation of patients with GD. We found that three of the signature’s miRNAs (miR-Let7d-5p, miR-96-5p, and miR-301-3p) were DE in patients with GO, whereas miR-Let7d-5p was inversely correlated with a higher GD severity and GO clinical activity. In this regard, as miR-Let7 has a specific Treg cell-mediated function, which can prevent Th1 cell-mediated inflammation and interferon-γ secretion (49), its lower levels found in patients with AITD with more severe disease could be associated with a defective Treg function previously reported in patients with AITD (50). miRNA-21-5p, itself, was associated with a higher risk of developing GD and a worse clinical outcome, and it was correlated to persistent disease in the long-term follow-up, with a better predictive value than the currently used TRAb levels (51). Recent reports have highly implicated miR-21 in the regulation of immune functions and development of several autoimmune disorders (52–55), suggesting a shared mechanism of action of this miRNA in different immune cells. The fact that expression of miR-21 affects T-cell activation, including the Th1/Th2 balance and Th17 differentiation (20, 21, 36, 56, 57), may explain these findings. In fact, we have recently reported increased levels of pathogenic Th17 cells in the peripheral blood and thyroid tissue from patients with AITD (58, 59). Our results also point to a similar behavior between miR-21-5p and miR-96-5p in AITD serum samples. Recent findings have demonstrated that miR-96 is highly expressed in pathogenic Th17 cells, and its overexpression in CD4+ T cells significantly increases cytokine production of Th17 cells (37), denoting a role in Th17 cells (60). For miR-142-3p and miR-301a-3p, there have been reports of their implication in Treg and Th17 regulation, respectively, in some autoimmune diseases (34, 35, 61, 62). However, to our knowledge, data are still lacking regarding these miRNAs in AITD, and our results add new evidence to their role in Treg dysregulation in this particular disease (50) (Fig. 3B). In summary, our data reveal for the first time using NGS, to our knowledge, the validation of dysregulated miRNAs in thyroid tissue and serum from patients with AITD. In addition, we provide a five-signature miRNA, which may be used as a potential biomarker to assign a risk for developing AITD and also for the evaluation of the severity of GD. This signature could have a substantial value in selecting specific treatment options. These outcomes represent a meaningful advance in the field of miRNAs as biomarkers in AITD and could contribute to a better understanding of the powerful translational role of miRNAs. Abbreviations: AITD autoimmune thyroid disease AUC area under the curve cDNA complementary DNA DE differentially expressed FFPE formalin fixed, paraffin embedded GD Graves disease GO Graves ophthalmopathy HT Hashimoto thyroiditis miRNA microRNA NGS next-generation sequencing PCR polymerase chain reaction qRT-PCR quantitative reverse transcription polymerase chain reaction ROC receiver operating characteristic Tg-Ab antithyroglobulin antibody TPO-Ab antithyroid peroxidase antibody TSHR-Ab antithyrotropin receptor antibody. Acknowledgments We thank the Institut d'Investigació en Ciències de la Salut Germans Trias i Pujol (IGTP-HUGTIP) Biobank and the Department of Surgery of Hospital Universitario la Princesa (HUP) (José Luis Muñoz) for providing tissue samples. We also thank our colleagues from the Department of Pathological Anatomy of HUP, especially Magdalena Adrados, for her help classifying thyroid specimens. Financial Support: This work was supported by grants from the Ministerio de Economia y Competitividad and from the Instituto de Salud Carlos III [Proyectos de Investigación en Salud (FIS): PI13-01414, PI16-02091, and PIE13-0004-BIOIMID project] to M.M., and by grants from CIBER de Enfermedades Cardiovasculares (CIBERCV) and SAF2017-82886-R to F.S.-M. and cofinanced by FEDER funds. Disclosure Summary: The authors have nothing to disclose. References 1. Ramos-Leví AM, Marazuela M. Pathogenesis of thyroid autoimmune disease: the role of cellular mechanisms. Endocrinol Nutr . 2016; 63( 8): 421– 429. Google Scholar CrossRef Search ADS PubMed  2. Weetman AP. Autoimmune thyroid disease. Autoimmunity . 2009; 37( 4): 337– 340. Google Scholar CrossRef Search ADS   3. Weetman AP. Cellular immune responses in autoimmune thyroid disease. Clin Endocrinol (Oxf) . 2004; 61( 4): 405– 413. Google Scholar CrossRef Search ADS PubMed  4. Bartalena L, Fatourechi V. Extrathyroidal manifestations of Graves’ disease: a 2014 update. J Endocrinol Invest . 2014; 37( 8): 691– 700. Google Scholar CrossRef Search ADS PubMed  5. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell . 2004; 116( 2): 281– 297. Google Scholar CrossRef Search ADS PubMed  6. Kozomara A, Griffiths-Jones S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res . 2013; 42( D1): D68– D73. Google Scholar CrossRef Search ADS PubMed  7. Heegaard NHH, Carlsen AL, Skovgaard K, Heegaard PMH. Circulating extracellular microRNA in systemic autoimmunity. EXS . 2015; 106: 171– 195. Google Scholar PubMed  8. Nakamachi Y, Kawano S, Takenokuchi M, Nishimura K, Sakai Y, Chin T, Saura R, Kurosaka M, Kumagai S. MicroRNA-124a is a key regulator of proliferation and monocyte chemoattractant protein 1 secretion in fibroblast-like synoviocytes from patients with rheumatoid arthritis. Arthritis Rheum . 2009; 60( 5): 1294– 1304. Google Scholar CrossRef Search ADS PubMed  9. Wang H, Peng W, Ouyang X, Li W, Dai Y. Circulating microRNAs as candidate biomarkers in patients with systemic lupus erythematosus. Transl Res . 2012; 160( 3): 198– 206. Google Scholar CrossRef Search ADS PubMed  10. Tang Y, Luo X, Cui H, Ni X, Yuan M, Guo Y, Huang X, Zhou H, de Vries N, Tak PP, Chen S, Shen N. MicroRNA-146A contributes to abnormal activation of the type I interferon pathway in human lupus by targeting the key signaling proteins. Arthritis Rheum . 2009; 60( 4): 1065– 1075. Google Scholar CrossRef Search ADS PubMed  11. Xia P, Fang X, Zhang ZH, Huang Q, Yan KX, Kang KF, Han L, Zheng ZZ. Dysregulation of miRNA146a versus IRAK1 induces IL-17 persistence in the psoriatic skin lesions. Immunol Lett . 2012; 148( 2): 151– 162. Google Scholar CrossRef Search ADS PubMed  12. Bernecker C, Halim F, Lenz L, Haase M, Nguyen T, Ehlers M, Vordenbaeumen S, Schott M. microRNA expressions in CD4+ and CD8+ T-cell subsets in autoimmune thyroid diseases. Exp Clin Endocrinol Diabetes . 2014; 122( 2): 107– 112. Google Scholar CrossRef Search ADS PubMed  13. Bernecker C, Lenz L, Ostapczuk MS, Schinner S, Willenberg H, Ehlers M, Vordenbäumen S, Feldkamp J, Schott M. MicroRNAs miR-146a1, miR-155_2, and miR-200a1 are regulated in autoimmune thyroid diseases. Thyroid . 2012; 22( 12): 1294– 1295. Google Scholar CrossRef Search ADS PubMed  14. Dorris ER, Smyth P, O’Leary JJ, Sheils O. MIR141 expression differentiates hashimoto thyroiditis from PTC and benign thyrocytes in Irish archival thyroid tissues. Front Endocrinol (Lausanne) . 2012; 3: 102. Google Scholar PubMed  15. Kagawa T, Watanabe M, Inoue N, Otsu H, Saeki M, Katsumata Y, Takuse Y, Iwatani Y. Increases of microRNA let-7e in peripheral blood mononuclear cells in Hashimoto’s disease. Endocr J . 2016; 63( 4): 375– 380. Google Scholar CrossRef Search ADS PubMed  16. Liu R, Ma X, Xu L, Wang D, Jiang X, Zhu W, Cui B, Ning G, Lin D, Wang S. Differential microRNA expression in peripheral blood mononuclear cells from Graves’ disease patients. J Clin Endocrinol Metab . 2012; 97( 6): E968– E972. Google Scholar CrossRef Search ADS PubMed  17. Peng H, Liu Y, Tian J, Ma J, Tang X, Yang J, Rui K, Zhang Y, Mao C, Lu L, Xu H, Wang S. Decreased expression of microRNA-125a-3p upregulates interleukin-23 receptor in patients with Hashimoto’s thyroiditis. Immunol Res . 2015; 62( 2): 129– 136. Google Scholar CrossRef Search ADS PubMed  18. Qin Q, Wang X, Yan N, Song RH, Cai TT, Zhang W, Guan LJ, Muhali FS, Zhang JA. Aberrant expression of miRNA and mRNAs in lesioned tissues of Graves’ disease. Cell Physiol Biochem . 2015; 35( 5): 1934– 1942. Google Scholar CrossRef Search ADS PubMed  19. Shen L, Huang F, Ye L, Zhu W, Zhang X, Wang S, Wang W, Ning G. Circulating microRNA predicts insensitivity to glucocorticoid therapy in Graves’ ophthalmopathy. Endocrine . 2015; 49( 2): 445– 456. Google Scholar CrossRef Search ADS PubMed  20. Tong BD, Xiao MY, Zeng JX, Xiong W. miRNA-21 promotes fibrosis in orbital fibroblasts from thyroid-associated ophthalmopathy. Mol Vis . 2015; 21: 324– 334. Google Scholar PubMed  21. Wang Z, Fan X, Zhang R, Lin Z, Lu T, Bai X, Li W, Zhao J, Zhang Q. Integrative analysis of mRNA and miRNA array data reveals the suppression of retinoic acid pathway in regulatory T cells of Graves’ disease. J Clin Endocrinol Metab . 2014; 99( 12): E2620– E2627. Google Scholar CrossRef Search ADS PubMed  22. Wei H, Guan M, Qin Y, Xie C, Fu X, Gao F, Xue Y. Circulating levels of miR-146a and IL-17 are significantly correlated with the clinical activity of Graves’ ophthalmopathy. Endocr J . 2014; 61( 11): 1087– 1092. Google Scholar CrossRef Search ADS PubMed  23. Yamada H, Itoh M, Hiratsuka I, Hashimoto S. Circulating microRNAs in autoimmune thyroid diseases. Clin Endocrinol (Oxf) . 2014; 81( 2): 276– 281. Google Scholar CrossRef Search ADS PubMed  24. Zhu J, Zhang Y, Zhang W, Zhang W, Fan L, Wang L, Liu Y, Liu S, Guo Y, Wang Y, Yi J, Yan Q, Wang Z, Huang G. MicroRNA-142-5p contributes to Hashimoto’s thyroiditis by targeting CLDN1. J Transl Med . 2016; 14( 1): 166. Google Scholar CrossRef Search ADS PubMed  25. Lee JY, Yun M, Paik JS, Lee SB, Yang SW. PDGF-BB enhances the proliferation of cells in human orbital fibroblasts by suppressing PDCD4 expression via up-regulation of microRNA-21. Invest Ophthalmol Vis Sci . 2016; 57( 3): 908– 913. Google Scholar CrossRef Search ADS PubMed  26. Hiratsuka I, Yamada H, Munetsuna E, Hashimoto S, Itoh M. Circulating microRNAs in Graves’ disease in relation to clinical activity. Thyroid . 2016; 26( 10): 1431– 1440. Google Scholar CrossRef Search ADS PubMed  27. Pohl M, Grabellus F, Worm K, Arnold G, Walz M, Schmid KW, Sheu-Grabellus SY. Intermediate microRNA expression profile in Graves’ disease falls between that of normal thyroid tissue and papillary thyroid carcinoma. J Clin Pathol . 2016; 70( 1): 33– 39. Google Scholar CrossRef Search ADS PubMed  28. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol . 2002; 3: RESEARCH0034. 29. Andersen CL, Jensen JL, Ørntoft TF. Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res . 2004; 64( 15): 5245– 5250. Google Scholar CrossRef Search ADS PubMed  30. Marabita F, de Candia P, Torri A, Tegnér J, Abrignani S, Rossi RL. Normalization of circulating microRNA expression data obtained by quantitative real-time RT-PCR. Brief Bioinform . 2015; 17( 2): 204– 212. Google Scholar CrossRef Search ADS PubMed  31. Ameling S, Kacprowski T, Chilukoti RK, Malsch C, Liebscher V, Suhre K, Pietzner M, Friedrich N, Homuth G, Hammer E, Völker U. Associations of circulating plasma microRNAs with age, body mass index and sex in a population-based study. BMC Med Genomics . 2015; 8( 1): 61. Google Scholar CrossRef Search ADS PubMed  32. Okoye IS, Coomes SM, Pelly VS, Czieso S, Papayannopoulos V, Tolmachova T, Seabra MC, Wilson MS. MicroRNA-containing T-regulatory-cell-derived exosomes suppress pathogenic T helper 1 cells. Immunity . 2014; 41( 1): 89– 103. Google Scholar CrossRef Search ADS PubMed  33. O’Connell RM, Rao DS, Baltimore D. MicroRNA regulation of inflammatory responses. Annu Rev Immunol . 2012; 30( 1): 295– 312. Google Scholar CrossRef Search ADS PubMed  34. Mycko MP, Cichalewska M, Machlanska A, Cwiklinska H, Mariasiewicz M, Selmaj KW. MicroRNA-301a regulation of a T-helper 17 immune response controls autoimmune demyelination. Proc Natl Acad Sci USA . 2012; 109( 20): E1248– E1257. Google Scholar CrossRef Search ADS PubMed  35. Huang B, Zhao J, Lei Z, Shen S, Li D, Shen GX, Zhang GM, Feng ZH. miR-142-3p restricts cAMP production in CD4+CD25- T cells and CD4+CD25+ TREG cells by targeting AC9 mRNA. EMBO Rep . 2008; 10( 2): 180– 185. Google Scholar CrossRef Search ADS PubMed  36. Garo LP, Murugaiyan G. Contribution of microRNAs to autoimmune diseases. Cell Mol Life Sci . 2016; 73( 10): 2041– 2051. Google Scholar CrossRef Search ADS PubMed  37. Ichiyama K, Gonzalez-Martin A, Kim BS, Jin HY, Jin W, Xu W, Sabouri-Ghomi M, Xu S, Zheng P, Xiao C, Dong C. The microRNA-183-96-182 cluster promotes T helper 17 cell pathogenicity by negatively regulating transcription factor Foxo1 expression. Immunity . 2016; 44( 6): 1284– 1298. Google Scholar CrossRef Search ADS PubMed  38. Wang J, Chen J, Sen S. MicroRNA as biomarkers and diagnostics. J Cell Physiol . 2015; 231( 1): 25– 30. Google Scholar CrossRef Search ADS   39. Abou-Zeid A, Saad M, Soliman E. MicroRNA 146a expression in rheumatoid arthritis: association with tumor necrosis factor-alpha and disease activity. Genet Test Mol Biomarkers . 2011; 15( 11): 807– 812. Google Scholar CrossRef Search ADS PubMed  40. Kohlhaas S, Garden OA, Scudamore C, Turner M, Okkenhaug K, Vigorito E. Cutting edge: the Foxp3 target miR-155 contributes to the development of regulatory T cells. J Immunol . 2009; 182( 5): 2578– 2582. Google Scholar CrossRef Search ADS PubMed  41. Lu LF, Boldin MP, Chaudhry A, Lin LL, Taganov KD, Hanada T, Yoshimura A, Baltimore D, Rudensky AY. Function of miR-146a in controlling Treg cell-mediated regulation of Th1 responses. Cell . 2010; 142( 6): 914– 929. Google Scholar CrossRef Search ADS PubMed  42. Yao R, Ma YL, Liang W, Li HH, Ma ZJ, Yu X, Liao YH. MicroRNA-155 modulates Treg and Th17 cells differentiation and Th17 cell function by targeting SOCS1. PLoS One . 2012; 7( 10): e46082. Google Scholar CrossRef Search ADS PubMed  43. Taganov KD, Boldin MP, Chang KJ, Baltimore D. NF-kappaB-dependent induction of microRNA miR-146, an inhibitor targeted to signaling proteins of innate immune responses. Proc Natl Acad Sci USA . 2006; 103( 33): 12481– 12486. Google Scholar CrossRef Search ADS PubMed  44. Rodríguez-Muñoz A, Martínez-Hernández R, Ramos-Leví AM, Serrano-Somavilla A, González-Amaro R, Sánchez-Madrid F, de la Fuente H, Marazuela M. Circulating microvesicles regulate Treg and Th17 differentiation in human autoimmune thyroid disorders. J Clin Endocrinol Metab . 2015; 100( 12): E1531– E1539. Google Scholar CrossRef Search ADS PubMed  45. Glud M, Klausen M, Gniadecki R, Rossing M, Hastrup N, Nielsen FC, Drzewiecki KT. MicroRNA expression in melanocytic nevi: the usefulness of formalin-fixed, paraffin-embedded material for miRNA microarray profiling. J Invest Dermatol . 2009; 129( 5): 1219– 1224. Google Scholar CrossRef Search ADS PubMed  46. Klopfleisch R, Weiss AT, Gruber AD. Excavation of a buried treasure—DNA, mRNA, miRNA and protein analysis in formalin fixed, paraffin embedded tissues. Histol Histopathol . 2011; 26( 6): 797– 810. Google Scholar PubMed  47. Li J, Smyth P, Flavin R, Cahill S, Denning K, Aherne S, Guenther SM, O’Leary JJ, Sheils O. Comparison of miRNA expression patterns using total RNA extracted from matched samples of formalin-fixed paraffin-embedded (FFPE) cells and snap frozen cells. BMC Biotechnol . 2007; 7( 1): 36. Google Scholar CrossRef Search ADS PubMed  48. Xi Y, Nakajima G, Gavin E, Morris CG, Kudo K, Hayashi K, Ju J. Systematic analysis of microRNA expression of RNA extracted from fresh frozen and formalin-fixed paraffin-embedded samples. RNA . 2007; 13( 10): 1668– 1674. Google Scholar CrossRef Search ADS PubMed  49. Gandhi R, Healy B, Gholipour T, Egorova S, Musallam A, Hussain MS, Nejad P, Patel B, Hei H, Khoury S, Quintana F, Kivisakk P, Chitnis T, Weiner HL. Circulating microRNAs as biomarkers for disease staging in multiple sclerosis. Ann Neurol . 2013; 73( 6): 729– 740. Google Scholar CrossRef Search ADS PubMed  50. Marazuela M, García-López MA, Figueroa-Vega N, de la Fuente H, Alvarado-Sánchez B, Monsiváis-Urenda A, Sánchez-Madrid F, González-Amaro R. Regulatory T cells in human autoimmune thyroid disease. J Clin Endocrinol Metab . 2006; 91( 9): 3639– 3646. Google Scholar CrossRef Search ADS PubMed  51. Weetman AP. Graves’ disease. N Engl J Med . 2000; 343( 17): 1236– 1248. Google Scholar CrossRef Search ADS PubMed  52. Fenoglio C, Cantoni C, De Riz M, Ridolfi E, Cortini F, Serpente M, Villa C, Comi C, Monaco F, Mellesi L, Valzelli S, Bresolin N, Galimberti D, Scarpini E. Expression and genetic analysis of miRNAs involved in CD4+ cell activation in patients with multiple sclerosis. Neurosci Lett . 2011; 504( 1): 9– 12. Google Scholar CrossRef Search ADS PubMed  53. Garchow BG, Bartulos Encinas O, Leung YT, Tsao PY, Eisenberg RA, Caricchio R, Obad S, Petri A, Kauppinen S, Kiriakidou M. Silencing of microRNA-21 in vivo ameliorates autoimmune splenomegaly in lupus mice. EMBO Mol Med . 2011; 3( 10): 605– 615. Google Scholar CrossRef Search ADS PubMed  54. Ruan Q, Wang T, Kameswaran V, Wei Q, Johnson DS, Matschinsky F, Shi W, Chen YH. The microRNA-21-PDCD4 axis prevents type 1 diabetes by blocking pancreatic beta cell death. Proc Natl Acad Sci USA . 2011; 108( 29): 12030– 12035. Google Scholar CrossRef Search ADS PubMed  55. Punga AR, Andersson M, Alimohammadi M, Punga T. Disease specific signature of circulating miR-150-5p and miR-21-5p in myasthenia gravis patients. J Neurol Sci . 2015; 356( 1–2): 90– 96. Google Scholar CrossRef Search ADS PubMed  56. Murugaiyan G, da Cunha AP, Ajay AK, Joller N, Garo LP, Kumaradevan S, Yosef N, Vaidya VS, Weiner HL. MicroRNA-21 promotes Th17 differentiation and mediates experimental autoimmune encephalomyelitis. J Clin Invest . 2015; 125( 3): 1069– 1080. Google Scholar CrossRef Search ADS PubMed  57. Wang S, Wan X, Ruan Q. The microRNA-21 in autoimmune diseases. Int J Mol Sci . 2016; 17( 6): 864. Google Scholar CrossRef Search ADS   58. Vitales-Noyola M, Ramos-Levi AM, Martínez-Hernández R, Serrano-Somavilla A, Sampedro-Nuñez M, González-Amaro R, Marazuela M. Pathogenic Th17 and Th22 cells are increased in patients with autoimmune thyroid disorders. Endocrine . 2017; 57( 3): 409– 417. Google Scholar CrossRef Search ADS PubMed  59. Figueroa-Vega N, Alfonso-Pérez M, Benedicto I, Sánchez-Madrid F, González-Amaro R, Marazuela M. Increased circulating pro-inflammatory cytokines and Th17 lymphocytes in Hashimoto’s thyroiditis. J Clin Endocrinol Metab . 2010; 95( 2): 953– 962. Google Scholar CrossRef Search ADS PubMed  60. González-Amaro R, Marazuela M. T regulatory (Treg) and T helper 17 (Th17) lymphocytes in thyroid autoimmunity. Endocrine . 2015; 52( 1): 30– 38. Google Scholar CrossRef Search ADS PubMed  61. Mandolesi G, De Vito F, Musella A, Gentile A, Bullitta S, Fresegna D, Sepman H, Di Sanza C, Haji N, Mori F, Buttari F, Perlas E, Ciotti MT, Hornstein E, Bozzoni I, Presutti C, Centonze D. miR-142-3p is a key regulator of IL-1β-dependent synaptopathy in neuroinflammation. J Neurosci . 2016; 37( 3): 546– 561. Google Scholar CrossRef Search ADS   62. Tang X, Yin K, Zhu H, Tian J, Shen D, Yi L, Rui K, Ma J, Xu H, Wang S. Correlation between the expression of microRNA-301a-3p and the proportion of Th17 cells in patients with rheumatoid arthritis. Inflammation . 2016; 39( 2): 759– 767. Google Scholar CrossRef Search ADS PubMed  Copyright © 2018 Endocrine Society

Journal

Journal of Clinical Endocrinology and MetabolismOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve Freelancer

DeepDyve Pro

Price
FREE
$49/month

$360/year
Save searches from Google Scholar, PubMed
Create lists to organize your research
Export lists, citations
Access to DeepDyve database
Abstract access only
Unlimited access to over
18 million full-text articles
Print
20 pages/month
PDF Discount
20% off