Endocrine DOI 10.1007/s12020-014-0487-4
ORIGINAL ARTICLE
Circulating microRNA predicts insensitivity to glucocorticoid therapy in Graves’ ophthalmopathy Liyun Shen • Fengjiao Huang • Lei Ye • Wei Zhu • Xiaofang Zhang • Shu Wang Weiqing Wang • Guang Ning
•
Received: 25 September 2014 / Accepted: 18 November 2014 Ó Springer Science+Business Media New York 2015
Abstract Glucocorticoid (GC) insensitivity occurs commonly in Graves’ ophthalmopathy (GO), and GC therapy is associated with major adverse effects. A reliable and easily accessible biomarker is required to predict the outcome of GC therapy. This study aimed to evaluate the performance of circulating microRNA (miRNA) to predict GC insensitivity in GO patients. A total of 35 consecutive patients were included in this study. A cumulative dose of 4.5 g of methylprednisolone (MP) was administered intravenously for 12 weeks. Pretreatment serum miRNAs from the best(N = 5) and worst- (N = 4) responding patients were profiled using miScript PCR arrays and validated by quantitative PCR in all patients. We calculated the Liyun Shen and Fengjiao Huang have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1007/s12020-014-0487-4) contains supplementary material, which is available to authorized users. L. Shen F. Huang L. Ye (&) W. Zhu X. Zhang S. Wang W. Wang (&) G. Ning Shanghai Key Laboratory for Endocrine Tumors, Shanghai Clinical Center for Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases and Shanghai E-institute for Endocrinology, Ruijin Hospital, School of Medicine, Shanghai Jiaotong University, 197 Ruijin 2nd Road, Shanghai 200025, People’s Republic of China e-mail: [email protected] W. Wang e-mail: [email protected] G. Ning Laboratory for Endocrine & Metabolic Diseases of Institute of Health Science, Shanghai Jiaotong University School of Medicine and Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 227 South Chong Qing Road, Shanghai 200025, People’s Republic of China
predictive value of pretreatment assays of serum miRNAs with regard to GC insensitivity. We further investigated the roles of target miRNAs in modulating NF-jB activity and restoring transrepression of an NF-jB reporter by dexamethasone. Nine miRNAs displayed significant differences between responsive and resistant patients by miScript PCR arrays. Validation of the top two miRNAs in all 35 patients confirmed a significantly lower serum level of miR-224-5p (p = 0.0048) in resistant patients. A multivariate logistic regression model identified a composite biomarker combining baseline serum miR-224-5p and TRAb was independently associated with GC response (OR: 2.565, 95 % CI 1.011–6.505, p = 0.047). Receiver operating characteristic (ROC) curves analysis revealed the composite marker combining miR-224-5p and TRAb led to a 91.67 % positive prediction value (PPV) and a 69.56 % negative prediction value (NPV) with regard to GC resistance. Overexpression of miR-224-5p restored transrepression of the NF-jB reporter by dexamethasone under induced resistance, which may be via targeting GSK-3b to increase GR protein level. Our study demonstrated baseline serum miR-224-5p was associated with GC sensitivity in GO and in vitro overexpression of miR-224-5p restored GC sensitivity in a resistant cell model. A parameter combined serum miR-224-5p and TRAb could effectively predict GC sensitivity in GO patients. Keywords Circulating miRNA GC sensitivity Graves’ ophthalmopathy MiR-224-5p Glucocorticoid receptor
Introduction Graves’ ophthalmopathy (GO) is the most prominent extrathyroidal manifestation of Graves’ disease, which
123
Endocrine
profoundly reduces the quality of life of afflicted patients. Intravenous glucocorticoid (GC) administration is currently the first-line therapeutic choice for active moderate-tosevere cases [1, 2]. Patient responses vary greatly between individuals, with approximately 75–80 % of patients displaying a satisfactory response, 6.5 % morbidity and 0.6 % mortality [3, 4]. The major side effects, such as severe cardiovascular and liver dysfunction, are mainly related to preexisting diseases or treatment. Thus, careful patient selection and personalized treatment based on GC outcome prediction is difficult. There are currently no known reliable biomarkers to predict the therapeutic outcome of GC treatment in GO patients. The effectiveness of GC therapy in GO patients has been associated with clinical parameters such as early response [5], short disease duration [6], and prolonged T2 relaxation time on magnetic resonance scan (MR) [7]. In addition, Ujhelyi et al. demonstrated that 99mTc-diethylenetriamine-pentaacetic-acid (DTPA) uptake on a SPECT scan could be used to predict GC therapy outcome in GO patients [8]. However, that study was based upon circular reasoning, as the DTPA approach was used for both the initial and follow-up test. In addition to clinical parameters, the pharmacogenomics of the glucocorticoid receptor (GR) have also been investigated in GO patients. Although polymorphism and mutations that alter GR transcriptional activity cause GC resistance in rheumatoid arthritis and acute lymphoblast leukemia [9, 10], a GR polymorphism study in GO patients failed to find associations with the ER22/23EK (increases GR activity) and N363S loci (decreases GR activity) [5]. In recent years, studies revealed that microRNAs (miRNA) could modulate drug response via targeting various pharmacogenomics relevant genes [11–15]. Studies have demonstrated that miRNA downregulates genes that are important for GC functions. The GC response in sepsis comprises miRNA-124-induced downregulation of glucocorticoid receptor a (GRa) in T cells [16]; miR-18 and miR124a not only decrease GR protein but also reduce GRmediated events in neuroscreen cells [17]. In GO, miRNAs might be also dysregulated, as hypothesized recently [18]. However, no study has investigated the roles of miRNAs in modulating GC response of GO patients. In the present study, we aimed to identify circulating miRNAs that could predict GC response in GO patients and investigate the causal roles of miRNAs in modulating GC response in vitro.
during July 2010 to March 2012 in Ruijin Hospital, Shanghai Jiaotong University, School of Medicine. After ruling out patients with obesity, liver disease, heart disease, or history of other autoimmune diseases, we recruited 35 participants into this study (Online resource 1 for inclusion and exclusion criteria). A cumulative dose of 4.5 g of methylprednisolone (MP) was administered intravenously for 12 weeks. For each patient, the same ophthalmologist calculated the Clinical Activity Score (CAS), measured parameters such as proptosis, intraocular pressure, diplopia (Gorman score), lid width, and visual acuity at baseline and follow-ups using the modified EUGOGO patient form [19]. The 7-point CAS (spontaneous retrobulbar pain, pain on attempted eye movements, conjunctival hyperemia, eyelid redness, chemosis, swelling of the caruncle and swelling of the eyelids) and NOSPECS score: (1) absent/mild (NOSPECS 0–1 including lid retraction or lid lag); (2) moderate (NOSPECS 2–3 including periorbital edema and proptosis); (3) severe (NOSPECS 4–6 including eye muscle involvement/corneal involvement or sightloss) was evaluated as previously reported [20, 21]. A Hertel exophthalmometer (Keeler Instruments Inc., Broomall, PA, USA) was used for the measurement of proptosis, and the upper limit of the normal Chinese patients in our study was 18.6 mm [22]. Thyroid function and antibody measurements were performed in a clinical laboratory at the Shanghai Institution of Endocrine and Metabolic Diseases certified by College of American Pathologists. Serum free triiodothyronine (FT3), free thyroxine (FT4), thyroidstimulating hormone (TSH), thyroid peroxidase antibodies (TPOAb) and thyroglobulin antibody (TgAb) were determined by chemiluminescent microparticle immunoassay (Abbott laboratories, Abbott Park, IL). Serum TSH receptor antibody (TRAb) was tested based on the M22 monoclonal antibody (Roche Diagnostics, Switzerland). The normal ranges were as followed: FT3: 2.62–6.49 pmol/l, FT4: 9.01–19.047 pmol/l, sTSH: 0.35–4.94 lIU/ml, TRAb: \1.75 U/l, and TPOAb: \5.61 IU/ml. Serum was collected before the first dose of MP and at each monthly follow-up (before i.v.MP). Serum samples were kept frozen at -80 °C until use. The board of medical ethics of Ruijin Hospital approved the study, and all patients gave their written informed consent. Grouping
Patients, materials, and methods Patients A total of 54 consecutive patients with untreated, active, and moderately severe GO were eligible for GC therapy
123
Patients were classified into three groups based upon changes of CAS and objective ophthalmological parameters at the end of therapy (12 weeks). Response was defined as CAS decreased by at least two points and CAS \ 3/7, together with at least two of the following parameters improved: (i) reduction of proptosis by at least
Endocrine
2 mm; (ii) reduction of lid width by at least 3 mm; (iii) degrade of Gorman Score (from constant to inconstant, inconstant to intermittent and intermittent to absent); (iv) reduction in any of the class two signs of NOSPECS by at least two grades; (v) decrease of intraocular pressure by at least 2 mmHg; (vi) improvement of visual acuity by one snellen line. Partial response was defined as CAS decreased by at least two points and CAS \ 3/7, together with less than two of the above parameters improved. Resistance was defined as CAS dropped less than two points or stayed active (CAS C 3/7). MiScript PCR array and real-time PCR MicroRNAs from 200 ll of sera exogenously spiked with syn-cel-mir-39 miScript miRNA mimic were extracted using a serum miRNeasy kit and reverse transcribed with miScript RT II, following the manufacturer’s instructions. All of these kits and reagents used in this procedure were purchased from Qiagen (Hilden, Germany). Human serum miScript PCR Arrays (Qiagen, Hilden, Germany) were applied to screen candidate miRNAs. Fluorescent PCR products were detected using a Light Cycler 480 (Roche Applied Science, Indianapolis, IN). Each array profiled 84 miRNAs, including replicated reverse transcription controls (miRTC), positive PCR controls (PPC) and a panel of control miRNA for QC as well as normalization. We chose C. elegans mir-39 as the normalization control. Finally, the data were analyzed using the miScript PCR Array online analysis tool and the DDCT method of relative quantification (http://pcrdataana lysis.sabiosciences.com/mirna). Real-time PCR was applied to validate the screened candidate miRNAs with a miScript SYBR Green PCR kit. C. elegans mir-39 served as the internal control for RNA input. Light Cycler 480 software (Roche Applied Science, Indianapolis, IN) was used to analyze data, and fold changes were determined by the 2-DDCT method. To detect hsa-miR-224-5p (sequence: CAAGUCACUAGU GGUUCCGUU) and hsa-miR-155-5p (sequence: UUAA UGCUAAUCGUGAUAGGGGU), commercially available primers purchased from Qiagen were used as followed: Hs_miR-224_1 miScript Primer Assay, Catalog Number:MS00003878 and Hs_miR-155_2 miScript Primer Assay, Catalog Number:MS00031486. Cell culture Retrobulbar adipose tissue was obtained from euthyroid GO patients who underwent surgical rehabilitation for severe exophthalmos (n = 3). Tissues were minced and dissociated with collagenase type A (1 mg/ml, Roche Applied
Science, Penzberg, Germany) at 37 °C with gentle agitation for 3.5 h in DMEM/F12 (1:1, Invitrogen, Grand Island, NY) containing 15 % FBS (Invitrogen, Australia) and 1 % penicillin–streptomycin (Invitrogen, Grand Island, NY). After passed through a disposable 100-lm strainer (BD Falcon, Palo Alto, CA, USA) and centrifugation, the cells were cultured in DMEM/F12 (1:1) containing 15 % FBS, 1 % penicillin–streptomycin and 2 mM L-glutamine (Invitrogen, Grand Island, NY) at 37 °C. The media were changed the next day and every 3 days until passage. Ocular fibroblasts of passages 2–5 were used in the following experiments. 293T cells were maintained in DMEM supplemented with 10 % FBS, 1 % penicillin–streptomycin and 2 mM Lglutamine at 37 %. The cells were passed every 3 days. DNA constructs The wild type GSK-3b forward and reverse oligonucleotides were annealed to form a segment containing the miR224-5p binding site on GSK-3b 30 UTR predicted by Targetscan program (http://www.targetscan.org), flanked by XbaI and NotI restriction enzyme sites. The mutant seed region was formed by mutant forward and reverse oligonucleotides. The segments were cloned downstream of the Renilla luciferase gene in pRL-null plasmid (gift from Dr. Mingwei Li), generating pRL-GSK-3b-30 UTR wild type and mutant constructs. The NF-jB reporter plasmid was a kind gift from Dr. Qinyun Ma. It was constructed with pGL3 reporter plasmid (Promega) and a consensus sequence (50 -GGGACTTT CC-30 ) for NF-jB binding. Luciferase assay For NF-jB repression study, 293T cells were co-transfected with 200 ng of NF-jB luciferase reporter and 10 ng of Renilla luciferase reporter (gift from Dr. Xuelian Xiong) together with miRNA mimics (Genepharma, China) or negative control mimics using Lipofectamine 2000 (Invitrogen). After 24 h, the cells were treated with 0, 1, 5, and 20 ng/ml TNFa (Peprotech, USA) ± 1 lM dexamethasone (Dex, Sigma-Aldrich, St Louis, MO) for another 24 h. For GSK-3b 30 UTR binding assay, 293T cells were co-transfected with 50 ng pGL3 reporter and 10 ng of pRL-GSK3b-30 UTR wild type or mutant plasmids together with miRNA mimics (Genepharma, China) or negative control mimics using Lipofectamine 2000 (Invitrogen). Cells were cultured for 48 h after transfection. Then, cells were lysed in passive cell lysis buffer, and the cells were subjected to a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Firefly luciferase activity was normalized to Renilla luciferase activity.
123
Endocrine
Western blot Whole cells were lysed in RIPA with 1 % proteinase inhibitor (Roche Applied Science). The protein concentration was determined with a Pierce BCA Protein Assay Kit (Thermo Scientific, Rockford, IL, USA) by following the manufacturer’s instructions. A total of 15 lg of protein lysate was loaded in each lane on 10 % PAGE gels and then transferred to PVDF membranes (Millipore, Bedford, MA, USA). GR antibody was purchased from Santa Cruz Biotechnologies. GSK3b antibody was purchased from Abcam. HRP-conjugated anti-mouse, anti-rabbit, and anti-GAPDH were purchased from Cell Signaling Technology. Images were captured using ImageQuant LAS4000 (Fujifilm, Tokyo, Japan). Statistical analysis Statistical analysis systems (SAS) software for Windows, Version 8.1 (SAS Institute Inc., Cary, NC) and Graph Pad (Prism 5 for Windows, version 5.2) were applied in the analysis. Continuous variables are reported as the mean ± SD or median and interquartile range (IQR). Categorical variables are reported as the frequency and proportions. The values of serum FT3, FT4, sTSH, TRAb, TPOAb, and TgAb were Log10 normal transformed to achieve a normal distribution before analysis. The mean values of the two eyes’ ophthalmological parameters, such as proptosis, lid width, ocular pressure, and visual acuity, are presented for each patient. Student’s t test or the Mann– Whitney U test was used to compare continuous variables between the two groups. A v2 or Fisher’s exact test was used to compare frequency and proportions between the three groups. A logistic regression model was used to determine the independent associations between miRNA levels and GC response adjusted for age, gender, current/passive smoking status, baseline CAS, TRAb, thyroid function status (euthyroid, hyper/hypothyroid), GO duration, and use of medication. Receiver operating characteristic (ROC) curves were used to evaluate the predictive value of miRNAs and GC response-related factors. The optimal cutoff point was chosen when the sum of sensitivity and specificity was maximal. In vitro data are presented as the mean ± SD from at least three experiments as analyzed using Student’s t test. All P values were two-sided, and a p value \ 0.05 was considered statistically significant.
Results Clinical features of GO patients A total of 35 participants were categorized into responsive group, partial responsive group, and resistant group based
123
upon clinical response at the end of GC therapy. As shown in Table 1, there were no significant differences between the three groups pre-therapy except for TRAb [2.95(0.9–9.0) vs. 4.74 (2.86–6.68) vs. 13.6 (4.8–31.7), p = 0.005] (Fig. 1a). Post-GC CAS was significantly lower in responders and partial responders than in resistant patients [1 (1–2) vs. 2 (1.5–2) vs. 4 (3–5), p \ 0.0001]. In addition, conjunctival hyperemia (p = 0.006) and diplopia (p = 0.02) were significantly improved in responsive and partial responsive patients post-therapy. A greater proportion of patients showed improved CAS (p = 0.002), proptosis (p = 0.05), and intraocular pressure (p = 0.02) in responsive and partial responsive groups. More improved items were found in responsive and partial responsive patients than in resistant patients [4 (3–5) vs. 2 (2) vs. 1 (1, 2), p \ 0.0001, Table 1]. Discovery and validation of baseline circulating miRNA associated with GC response A total of 84 circulating miRNAs were screened using baseline serum with a miScript PCR array comparing the five best-responding patients (responsive patients who improved in at least four items and showed early response in the first month of therapy without relapse at the end of the therapy) with the four worst-responding patients (resistant patients who had only one or no items improved, even deteriorated at the end of the therapy) (Online resource 2). We found nine miRNAs that displayed significant differences (p \ 0.05) (Online resource 3), eight of which were upregulated in the responders. MiR-155-5p and miR-224-5p displayed the greatest fold-change: 14.14 (p = 0.02) and 11.04 (p = 0.02), respectively. We further validated the serum level of miR-155-5p and miR-224-5p using real-time PCR in all 35 patients, including those in the discovery set. We found the DCt of miR-224-5p was significantly higher in the resistant group (responsive vs. resistant, p = 0.0048; partial responsive vs. resistant, p = 0.0098, Fig. 1b), whereas no significant difference was found for miR-155 among the three groups (Fig. 1c). To be noted, for serum miR-224-5p level, we found no significant difference among patients with different thyroid-related medications (Online resource 4). Risk factors associated with GC outcome Univariate analysis found that baseline miR-224-5p, TRAb, and current medication (with/without either antithyroid drug or levothyroxine or both) were associated with risks of GC resistance. After adjusting for age, gender, current/passive smoking status, baseline CAS, TRAb, thyroid function status (euthyroid, hyper/hypothyroid), GO duration, and current medication, we found no parameter
Endocrine Table 1 Clinical features of GO patients Responsive
Partial responsive
Resistant
p Value –
Number (n)
14
4
17
Age (year)
44.57 ± 11.65
42.67 ± 13.61
51.07 ± 9.56
0.21
Female (n, %)
11 (78.57 %)
1 (25 %)
9 (52.94 %)
0.11
Smoking history (n, %) Current smoker
0
1 (25 %)
3 (21.43 %)
0.20
Ex-smoker
0
1 (25 %)
3 (21.43 %)
0.20
Passive smoker
9 (69.23 %)
2 (50 %)
7 (50 %)
0.72
Never-smoker
4 (30.77 %)
0
1 (7.14 %)
0.14
Graves’ hyperthyroidism
13 (92.86 %)
4 (100 %)
Primary hypothyroidism Hashimotos’ thyroiditis
0 0
History of thyroid disease (n, %) 16 (94.12 %)
0.86
0 0
– – 0.06
Previous antithyroid treatments Anti-thyroid drugs
8 (57.14 %)
4 (100 %)
15 (88.24 %)
Radioiodine
4 (28.57 %)
0
1 (5.88 %)
0.14
Thyroidectomy
1 (7.14 %)
0
0
0.46
Euthyroid
10 (71.43 %)
4 (100 %)
15 (88.24 %)
Hyperthyroid
4 (28.57 %)
0
1 (5.88 %)
Hypothyroid
0
0
1 (5.88 %)
Current thyroid status (n, %)
0.30
Current thyroid treatments None
9 (64.29 %)
1 (25 %)
4 (23.53 %)
0.06
Levothyroxine only
2 (14.28 %)
1 (25 %)
1 (5.88 %)
0.51
Tapazole only
2 (14.28 %)
1 (25 %)
4 (23.53 %)
0.48
Propylthiouracil only
0
0
0
–
Levothyroxine and tapazole
1 (7.14 %)
1 (25 %)
6 (35.29 %)
0.18
0 7 (5–11.5)
0 17 (6.5–25)
2 (11.76 %) 5.5 (4–9)
0.32 0.32
Baseline
0.18 (0.005–0.90)
2.72 (0.42–5.03)
0.07 (0.005–0.67)
0.52
3 months
2.65 (1.56–3.67)
3.67 (1.84–5.20)
1.94 (0.62–2.89)
0.84
Levothyroxine and Propylthiouracil Duration of eye symptoms (months) sTSH(lIU/ml)
FT3 (pmol/l) Baseline
4.89 (4.55–6.08)
4.32 (4.14–4.66)
4.91 (4.45–5.51)
0.64
3 months
4.23 (3.75–4.73)
4.12 (3.9–4.32)
4.11 (3.69–4.45)
0.81
FT4 (pmol/l) Baseline
14.48 (12.8–19)
13.09 (11.40–14.58)
13.73 (12.34–15.81)
0.31
3 months
12.46 (12.07–14.19)
12.97 (12.23–13.71)
12.66 (10.1–13.5)
0.35
TRAb (U/l) Baseline
2.95 (0.9–9)
4.74 (2.86–6.68)
13.6 (4.8–31.7)
0.005
3 months
1.8 (0.83–2.82)
2.77 (1.5–11.1)
2.58 (1.5–5.18)
0.35
2.58 (1.07–12.49) 2.28 (1.0–6.52)
3.43 (2.0–313.86) 2.25 (1.40–76.61)
1.99 (1.07–172.95) 2.58 (1.07–12.49)
0.79 0.93
Baseline
40.78 (0.37–363.57)
210.26 (15.47–694.9)
56.63 (0.52–561.69)
0.47
3 months
9.55 (0.18–59.40)
32.8 (3.18–157.43)
7.65 (0.25–104.16)
0.64
4 (3–5)
4.5 (3.5–5)
5 (4–6)
0.22
TgAb (IU/ml) Baseline 3 months TPOAb (IU/ml)
a
CAS
Baseline
123
Endocrine Table 1 continued Responsive
Partial responsive
Resistant
p Value
3 months
1 (1–2)
Improved
14 (100 %)
2 (1.5–2)
4 (3–5)
ORIGINAL ARTICLE
Circulating microRNA predicts insensitivity to glucocorticoid therapy in Graves’ ophthalmopathy Liyun Shen • Fengjiao Huang • Lei Ye • Wei Zhu • Xiaofang Zhang • Shu Wang Weiqing Wang • Guang Ning
•
Received: 25 September 2014 / Accepted: 18 November 2014 Ó Springer Science+Business Media New York 2015
Abstract Glucocorticoid (GC) insensitivity occurs commonly in Graves’ ophthalmopathy (GO), and GC therapy is associated with major adverse effects. A reliable and easily accessible biomarker is required to predict the outcome of GC therapy. This study aimed to evaluate the performance of circulating microRNA (miRNA) to predict GC insensitivity in GO patients. A total of 35 consecutive patients were included in this study. A cumulative dose of 4.5 g of methylprednisolone (MP) was administered intravenously for 12 weeks. Pretreatment serum miRNAs from the best(N = 5) and worst- (N = 4) responding patients were profiled using miScript PCR arrays and validated by quantitative PCR in all patients. We calculated the Liyun Shen and Fengjiao Huang have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1007/s12020-014-0487-4) contains supplementary material, which is available to authorized users. L. Shen F. Huang L. Ye (&) W. Zhu X. Zhang S. Wang W. Wang (&) G. Ning Shanghai Key Laboratory for Endocrine Tumors, Shanghai Clinical Center for Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases and Shanghai E-institute for Endocrinology, Ruijin Hospital, School of Medicine, Shanghai Jiaotong University, 197 Ruijin 2nd Road, Shanghai 200025, People’s Republic of China e-mail: [email protected] W. Wang e-mail: [email protected] G. Ning Laboratory for Endocrine & Metabolic Diseases of Institute of Health Science, Shanghai Jiaotong University School of Medicine and Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 227 South Chong Qing Road, Shanghai 200025, People’s Republic of China
predictive value of pretreatment assays of serum miRNAs with regard to GC insensitivity. We further investigated the roles of target miRNAs in modulating NF-jB activity and restoring transrepression of an NF-jB reporter by dexamethasone. Nine miRNAs displayed significant differences between responsive and resistant patients by miScript PCR arrays. Validation of the top two miRNAs in all 35 patients confirmed a significantly lower serum level of miR-224-5p (p = 0.0048) in resistant patients. A multivariate logistic regression model identified a composite biomarker combining baseline serum miR-224-5p and TRAb was independently associated with GC response (OR: 2.565, 95 % CI 1.011–6.505, p = 0.047). Receiver operating characteristic (ROC) curves analysis revealed the composite marker combining miR-224-5p and TRAb led to a 91.67 % positive prediction value (PPV) and a 69.56 % negative prediction value (NPV) with regard to GC resistance. Overexpression of miR-224-5p restored transrepression of the NF-jB reporter by dexamethasone under induced resistance, which may be via targeting GSK-3b to increase GR protein level. Our study demonstrated baseline serum miR-224-5p was associated with GC sensitivity in GO and in vitro overexpression of miR-224-5p restored GC sensitivity in a resistant cell model. A parameter combined serum miR-224-5p and TRAb could effectively predict GC sensitivity in GO patients. Keywords Circulating miRNA GC sensitivity Graves’ ophthalmopathy MiR-224-5p Glucocorticoid receptor
Introduction Graves’ ophthalmopathy (GO) is the most prominent extrathyroidal manifestation of Graves’ disease, which
123
Endocrine
profoundly reduces the quality of life of afflicted patients. Intravenous glucocorticoid (GC) administration is currently the first-line therapeutic choice for active moderate-tosevere cases [1, 2]. Patient responses vary greatly between individuals, with approximately 75–80 % of patients displaying a satisfactory response, 6.5 % morbidity and 0.6 % mortality [3, 4]. The major side effects, such as severe cardiovascular and liver dysfunction, are mainly related to preexisting diseases or treatment. Thus, careful patient selection and personalized treatment based on GC outcome prediction is difficult. There are currently no known reliable biomarkers to predict the therapeutic outcome of GC treatment in GO patients. The effectiveness of GC therapy in GO patients has been associated with clinical parameters such as early response [5], short disease duration [6], and prolonged T2 relaxation time on magnetic resonance scan (MR) [7]. In addition, Ujhelyi et al. demonstrated that 99mTc-diethylenetriamine-pentaacetic-acid (DTPA) uptake on a SPECT scan could be used to predict GC therapy outcome in GO patients [8]. However, that study was based upon circular reasoning, as the DTPA approach was used for both the initial and follow-up test. In addition to clinical parameters, the pharmacogenomics of the glucocorticoid receptor (GR) have also been investigated in GO patients. Although polymorphism and mutations that alter GR transcriptional activity cause GC resistance in rheumatoid arthritis and acute lymphoblast leukemia [9, 10], a GR polymorphism study in GO patients failed to find associations with the ER22/23EK (increases GR activity) and N363S loci (decreases GR activity) [5]. In recent years, studies revealed that microRNAs (miRNA) could modulate drug response via targeting various pharmacogenomics relevant genes [11–15]. Studies have demonstrated that miRNA downregulates genes that are important for GC functions. The GC response in sepsis comprises miRNA-124-induced downregulation of glucocorticoid receptor a (GRa) in T cells [16]; miR-18 and miR124a not only decrease GR protein but also reduce GRmediated events in neuroscreen cells [17]. In GO, miRNAs might be also dysregulated, as hypothesized recently [18]. However, no study has investigated the roles of miRNAs in modulating GC response of GO patients. In the present study, we aimed to identify circulating miRNAs that could predict GC response in GO patients and investigate the causal roles of miRNAs in modulating GC response in vitro.
during July 2010 to March 2012 in Ruijin Hospital, Shanghai Jiaotong University, School of Medicine. After ruling out patients with obesity, liver disease, heart disease, or history of other autoimmune diseases, we recruited 35 participants into this study (Online resource 1 for inclusion and exclusion criteria). A cumulative dose of 4.5 g of methylprednisolone (MP) was administered intravenously for 12 weeks. For each patient, the same ophthalmologist calculated the Clinical Activity Score (CAS), measured parameters such as proptosis, intraocular pressure, diplopia (Gorman score), lid width, and visual acuity at baseline and follow-ups using the modified EUGOGO patient form [19]. The 7-point CAS (spontaneous retrobulbar pain, pain on attempted eye movements, conjunctival hyperemia, eyelid redness, chemosis, swelling of the caruncle and swelling of the eyelids) and NOSPECS score: (1) absent/mild (NOSPECS 0–1 including lid retraction or lid lag); (2) moderate (NOSPECS 2–3 including periorbital edema and proptosis); (3) severe (NOSPECS 4–6 including eye muscle involvement/corneal involvement or sightloss) was evaluated as previously reported [20, 21]. A Hertel exophthalmometer (Keeler Instruments Inc., Broomall, PA, USA) was used for the measurement of proptosis, and the upper limit of the normal Chinese patients in our study was 18.6 mm [22]. Thyroid function and antibody measurements were performed in a clinical laboratory at the Shanghai Institution of Endocrine and Metabolic Diseases certified by College of American Pathologists. Serum free triiodothyronine (FT3), free thyroxine (FT4), thyroidstimulating hormone (TSH), thyroid peroxidase antibodies (TPOAb) and thyroglobulin antibody (TgAb) were determined by chemiluminescent microparticle immunoassay (Abbott laboratories, Abbott Park, IL). Serum TSH receptor antibody (TRAb) was tested based on the M22 monoclonal antibody (Roche Diagnostics, Switzerland). The normal ranges were as followed: FT3: 2.62–6.49 pmol/l, FT4: 9.01–19.047 pmol/l, sTSH: 0.35–4.94 lIU/ml, TRAb: \1.75 U/l, and TPOAb: \5.61 IU/ml. Serum was collected before the first dose of MP and at each monthly follow-up (before i.v.MP). Serum samples were kept frozen at -80 °C until use. The board of medical ethics of Ruijin Hospital approved the study, and all patients gave their written informed consent. Grouping
Patients, materials, and methods Patients A total of 54 consecutive patients with untreated, active, and moderately severe GO were eligible for GC therapy
123
Patients were classified into three groups based upon changes of CAS and objective ophthalmological parameters at the end of therapy (12 weeks). Response was defined as CAS decreased by at least two points and CAS \ 3/7, together with at least two of the following parameters improved: (i) reduction of proptosis by at least
Endocrine
2 mm; (ii) reduction of lid width by at least 3 mm; (iii) degrade of Gorman Score (from constant to inconstant, inconstant to intermittent and intermittent to absent); (iv) reduction in any of the class two signs of NOSPECS by at least two grades; (v) decrease of intraocular pressure by at least 2 mmHg; (vi) improvement of visual acuity by one snellen line. Partial response was defined as CAS decreased by at least two points and CAS \ 3/7, together with less than two of the above parameters improved. Resistance was defined as CAS dropped less than two points or stayed active (CAS C 3/7). MiScript PCR array and real-time PCR MicroRNAs from 200 ll of sera exogenously spiked with syn-cel-mir-39 miScript miRNA mimic were extracted using a serum miRNeasy kit and reverse transcribed with miScript RT II, following the manufacturer’s instructions. All of these kits and reagents used in this procedure were purchased from Qiagen (Hilden, Germany). Human serum miScript PCR Arrays (Qiagen, Hilden, Germany) were applied to screen candidate miRNAs. Fluorescent PCR products were detected using a Light Cycler 480 (Roche Applied Science, Indianapolis, IN). Each array profiled 84 miRNAs, including replicated reverse transcription controls (miRTC), positive PCR controls (PPC) and a panel of control miRNA for QC as well as normalization. We chose C. elegans mir-39 as the normalization control. Finally, the data were analyzed using the miScript PCR Array online analysis tool and the DDCT method of relative quantification (http://pcrdataana lysis.sabiosciences.com/mirna). Real-time PCR was applied to validate the screened candidate miRNAs with a miScript SYBR Green PCR kit. C. elegans mir-39 served as the internal control for RNA input. Light Cycler 480 software (Roche Applied Science, Indianapolis, IN) was used to analyze data, and fold changes were determined by the 2-DDCT method. To detect hsa-miR-224-5p (sequence: CAAGUCACUAGU GGUUCCGUU) and hsa-miR-155-5p (sequence: UUAA UGCUAAUCGUGAUAGGGGU), commercially available primers purchased from Qiagen were used as followed: Hs_miR-224_1 miScript Primer Assay, Catalog Number:MS00003878 and Hs_miR-155_2 miScript Primer Assay, Catalog Number:MS00031486. Cell culture Retrobulbar adipose tissue was obtained from euthyroid GO patients who underwent surgical rehabilitation for severe exophthalmos (n = 3). Tissues were minced and dissociated with collagenase type A (1 mg/ml, Roche Applied
Science, Penzberg, Germany) at 37 °C with gentle agitation for 3.5 h in DMEM/F12 (1:1, Invitrogen, Grand Island, NY) containing 15 % FBS (Invitrogen, Australia) and 1 % penicillin–streptomycin (Invitrogen, Grand Island, NY). After passed through a disposable 100-lm strainer (BD Falcon, Palo Alto, CA, USA) and centrifugation, the cells were cultured in DMEM/F12 (1:1) containing 15 % FBS, 1 % penicillin–streptomycin and 2 mM L-glutamine (Invitrogen, Grand Island, NY) at 37 °C. The media were changed the next day and every 3 days until passage. Ocular fibroblasts of passages 2–5 were used in the following experiments. 293T cells were maintained in DMEM supplemented with 10 % FBS, 1 % penicillin–streptomycin and 2 mM Lglutamine at 37 %. The cells were passed every 3 days. DNA constructs The wild type GSK-3b forward and reverse oligonucleotides were annealed to form a segment containing the miR224-5p binding site on GSK-3b 30 UTR predicted by Targetscan program (http://www.targetscan.org), flanked by XbaI and NotI restriction enzyme sites. The mutant seed region was formed by mutant forward and reverse oligonucleotides. The segments were cloned downstream of the Renilla luciferase gene in pRL-null plasmid (gift from Dr. Mingwei Li), generating pRL-GSK-3b-30 UTR wild type and mutant constructs. The NF-jB reporter plasmid was a kind gift from Dr. Qinyun Ma. It was constructed with pGL3 reporter plasmid (Promega) and a consensus sequence (50 -GGGACTTT CC-30 ) for NF-jB binding. Luciferase assay For NF-jB repression study, 293T cells were co-transfected with 200 ng of NF-jB luciferase reporter and 10 ng of Renilla luciferase reporter (gift from Dr. Xuelian Xiong) together with miRNA mimics (Genepharma, China) or negative control mimics using Lipofectamine 2000 (Invitrogen). After 24 h, the cells were treated with 0, 1, 5, and 20 ng/ml TNFa (Peprotech, USA) ± 1 lM dexamethasone (Dex, Sigma-Aldrich, St Louis, MO) for another 24 h. For GSK-3b 30 UTR binding assay, 293T cells were co-transfected with 50 ng pGL3 reporter and 10 ng of pRL-GSK3b-30 UTR wild type or mutant plasmids together with miRNA mimics (Genepharma, China) or negative control mimics using Lipofectamine 2000 (Invitrogen). Cells were cultured for 48 h after transfection. Then, cells were lysed in passive cell lysis buffer, and the cells were subjected to a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Firefly luciferase activity was normalized to Renilla luciferase activity.
123
Endocrine
Western blot Whole cells were lysed in RIPA with 1 % proteinase inhibitor (Roche Applied Science). The protein concentration was determined with a Pierce BCA Protein Assay Kit (Thermo Scientific, Rockford, IL, USA) by following the manufacturer’s instructions. A total of 15 lg of protein lysate was loaded in each lane on 10 % PAGE gels and then transferred to PVDF membranes (Millipore, Bedford, MA, USA). GR antibody was purchased from Santa Cruz Biotechnologies. GSK3b antibody was purchased from Abcam. HRP-conjugated anti-mouse, anti-rabbit, and anti-GAPDH were purchased from Cell Signaling Technology. Images were captured using ImageQuant LAS4000 (Fujifilm, Tokyo, Japan). Statistical analysis Statistical analysis systems (SAS) software for Windows, Version 8.1 (SAS Institute Inc., Cary, NC) and Graph Pad (Prism 5 for Windows, version 5.2) were applied in the analysis. Continuous variables are reported as the mean ± SD or median and interquartile range (IQR). Categorical variables are reported as the frequency and proportions. The values of serum FT3, FT4, sTSH, TRAb, TPOAb, and TgAb were Log10 normal transformed to achieve a normal distribution before analysis. The mean values of the two eyes’ ophthalmological parameters, such as proptosis, lid width, ocular pressure, and visual acuity, are presented for each patient. Student’s t test or the Mann– Whitney U test was used to compare continuous variables between the two groups. A v2 or Fisher’s exact test was used to compare frequency and proportions between the three groups. A logistic regression model was used to determine the independent associations between miRNA levels and GC response adjusted for age, gender, current/passive smoking status, baseline CAS, TRAb, thyroid function status (euthyroid, hyper/hypothyroid), GO duration, and use of medication. Receiver operating characteristic (ROC) curves were used to evaluate the predictive value of miRNAs and GC response-related factors. The optimal cutoff point was chosen when the sum of sensitivity and specificity was maximal. In vitro data are presented as the mean ± SD from at least three experiments as analyzed using Student’s t test. All P values were two-sided, and a p value \ 0.05 was considered statistically significant.
Results Clinical features of GO patients A total of 35 participants were categorized into responsive group, partial responsive group, and resistant group based
123
upon clinical response at the end of GC therapy. As shown in Table 1, there were no significant differences between the three groups pre-therapy except for TRAb [2.95(0.9–9.0) vs. 4.74 (2.86–6.68) vs. 13.6 (4.8–31.7), p = 0.005] (Fig. 1a). Post-GC CAS was significantly lower in responders and partial responders than in resistant patients [1 (1–2) vs. 2 (1.5–2) vs. 4 (3–5), p \ 0.0001]. In addition, conjunctival hyperemia (p = 0.006) and diplopia (p = 0.02) were significantly improved in responsive and partial responsive patients post-therapy. A greater proportion of patients showed improved CAS (p = 0.002), proptosis (p = 0.05), and intraocular pressure (p = 0.02) in responsive and partial responsive groups. More improved items were found in responsive and partial responsive patients than in resistant patients [4 (3–5) vs. 2 (2) vs. 1 (1, 2), p \ 0.0001, Table 1]. Discovery and validation of baseline circulating miRNA associated with GC response A total of 84 circulating miRNAs were screened using baseline serum with a miScript PCR array comparing the five best-responding patients (responsive patients who improved in at least four items and showed early response in the first month of therapy without relapse at the end of the therapy) with the four worst-responding patients (resistant patients who had only one or no items improved, even deteriorated at the end of the therapy) (Online resource 2). We found nine miRNAs that displayed significant differences (p \ 0.05) (Online resource 3), eight of which were upregulated in the responders. MiR-155-5p and miR-224-5p displayed the greatest fold-change: 14.14 (p = 0.02) and 11.04 (p = 0.02), respectively. We further validated the serum level of miR-155-5p and miR-224-5p using real-time PCR in all 35 patients, including those in the discovery set. We found the DCt of miR-224-5p was significantly higher in the resistant group (responsive vs. resistant, p = 0.0048; partial responsive vs. resistant, p = 0.0098, Fig. 1b), whereas no significant difference was found for miR-155 among the three groups (Fig. 1c). To be noted, for serum miR-224-5p level, we found no significant difference among patients with different thyroid-related medications (Online resource 4). Risk factors associated with GC outcome Univariate analysis found that baseline miR-224-5p, TRAb, and current medication (with/without either antithyroid drug or levothyroxine or both) were associated with risks of GC resistance. After adjusting for age, gender, current/passive smoking status, baseline CAS, TRAb, thyroid function status (euthyroid, hyper/hypothyroid), GO duration, and current medication, we found no parameter
Endocrine Table 1 Clinical features of GO patients Responsive
Partial responsive
Resistant
p Value –
Number (n)
14
4
17
Age (year)
44.57 ± 11.65
42.67 ± 13.61
51.07 ± 9.56
0.21
Female (n, %)
11 (78.57 %)
1 (25 %)
9 (52.94 %)
0.11
Smoking history (n, %) Current smoker
0
1 (25 %)
3 (21.43 %)
0.20
Ex-smoker
0
1 (25 %)
3 (21.43 %)
0.20
Passive smoker
9 (69.23 %)
2 (50 %)
7 (50 %)
0.72
Never-smoker
4 (30.77 %)
0
1 (7.14 %)
0.14
Graves’ hyperthyroidism
13 (92.86 %)
4 (100 %)
Primary hypothyroidism Hashimotos’ thyroiditis
0 0
History of thyroid disease (n, %) 16 (94.12 %)
0.86
0 0
– – 0.06
Previous antithyroid treatments Anti-thyroid drugs
8 (57.14 %)
4 (100 %)
15 (88.24 %)
Radioiodine
4 (28.57 %)
0
1 (5.88 %)
0.14
Thyroidectomy
1 (7.14 %)
0
0
0.46
Euthyroid
10 (71.43 %)
4 (100 %)
15 (88.24 %)
Hyperthyroid
4 (28.57 %)
0
1 (5.88 %)
Hypothyroid
0
0
1 (5.88 %)
Current thyroid status (n, %)
0.30
Current thyroid treatments None
9 (64.29 %)
1 (25 %)
4 (23.53 %)
0.06
Levothyroxine only
2 (14.28 %)
1 (25 %)
1 (5.88 %)
0.51
Tapazole only
2 (14.28 %)
1 (25 %)
4 (23.53 %)
0.48
Propylthiouracil only
0
0
0
–
Levothyroxine and tapazole
1 (7.14 %)
1 (25 %)
6 (35.29 %)
0.18
0 7 (5–11.5)
0 17 (6.5–25)
2 (11.76 %) 5.5 (4–9)
0.32 0.32
Baseline
0.18 (0.005–0.90)
2.72 (0.42–5.03)
0.07 (0.005–0.67)
0.52
3 months
2.65 (1.56–3.67)
3.67 (1.84–5.20)
1.94 (0.62–2.89)
0.84
Levothyroxine and Propylthiouracil Duration of eye symptoms (months) sTSH(lIU/ml)
FT3 (pmol/l) Baseline
4.89 (4.55–6.08)
4.32 (4.14–4.66)
4.91 (4.45–5.51)
0.64
3 months
4.23 (3.75–4.73)
4.12 (3.9–4.32)
4.11 (3.69–4.45)
0.81
FT4 (pmol/l) Baseline
14.48 (12.8–19)
13.09 (11.40–14.58)
13.73 (12.34–15.81)
0.31
3 months
12.46 (12.07–14.19)
12.97 (12.23–13.71)
12.66 (10.1–13.5)
0.35
TRAb (U/l) Baseline
2.95 (0.9–9)
4.74 (2.86–6.68)
13.6 (4.8–31.7)
0.005
3 months
1.8 (0.83–2.82)
2.77 (1.5–11.1)
2.58 (1.5–5.18)
0.35
2.58 (1.07–12.49) 2.28 (1.0–6.52)
3.43 (2.0–313.86) 2.25 (1.40–76.61)
1.99 (1.07–172.95) 2.58 (1.07–12.49)
0.79 0.93
Baseline
40.78 (0.37–363.57)
210.26 (15.47–694.9)
56.63 (0.52–561.69)
0.47
3 months
9.55 (0.18–59.40)
32.8 (3.18–157.43)
7.65 (0.25–104.16)
0.64
4 (3–5)
4.5 (3.5–5)
5 (4–6)
0.22
TgAb (IU/ml) Baseline 3 months TPOAb (IU/ml)
a
CAS
Baseline
123
Endocrine Table 1 continued Responsive
Partial responsive
Resistant
p Value
3 months
1 (1–2)
Improved
14 (100 %)
2 (1.5–2)
4 (3–5)
