Following confirmation by pathological diagnosis, thyroid cancer tissues and corresponding normal thyroid tissues at least 1 cm away from the tumor were obtained from 196 patients. These patients were diagnosed and had received a thyroidectomy between 2003 and 2012 at the Department of Head and Neck Surgery, Fudan University Shanghai Cancer Center (Shanghai, China). All tissue samples were fixed in formalin, embedded in paraffin and sectioned into 5-μm-thick slices. The criteria of the American Joint Committee on Cancer (8th edition) for thyroid cancer was used for TNM staging classification (19). This study was approved by the Human Ethics Committee/Institutional Review Board of Fudan University Shanghai Cancer Center. All 196 patients signed the written informed consent. A highly sensitive streptavidin-biotin-peroxidase detection system was used to perform immunohistochemical staining with thyroid cancer tissue microarrays derived from the samples of the 196 patients. To investigate the association between the expression of NRG1 and that of NRF2, we extracted and analyzed the data of 490 patients with PTC from The Cancer Genome Atlas (TCGA) database. NRG1 and NRF2 expression and clinical data of the TCGA database are available from the link directly: http://www.cbioportal.org/index.do?cancer_study_id=thca_tcga&Z_SCORE_THRESHOLD=2.0&RPPA_SCORE_THRESHOLD=2.0&data_priority=0&case_set_id=thca_tcga_all&gene_list=NRG1%2520NFE2L2&geneset_list=+&tab_index=tab_visualize&Action=Submit&genetic_profile_ids_PROFILE_MUTATION_EXTENDED=thca_tcga_mutations&genetic_profile_ids_PROFILE_COPY_NUMBER_ALTERATION=thca_tcga_gistic. Patients who fit the following criteria were included: i) Patients who had PTC as the only malignancy or the first of multiple malignancies; and ii) patients who received surgical therapy. Patients who fit the following criteria were excluded: i) Patients with insufficient data or unknown clinicopathological profiles; ii) patients with an undetermined histology; and iii) patients with other types of thyroid cancer (follicular thyroid cancer, medullary thyroid cancer, anaplastic thyroid cancer, etc.), or secondary tumors.

Briefly, tissue samples obtained from the 196 patients were fixed with formalin (37% formaldehyde dissolved in water, containing 10% methanol in addition) at room temperature for 24 h. The fixed samples were then embedded in paraffin and sectioned into 5-μm-thick slices. The slices were mounted on slides and were then deparaffinized with xylol, hydrated with gradient ethanol in multiple steps. Citrate buffer (pH 6.0) was used for antigen retrieval and the slides were heated in this buffer at 121°C for 15 min. Following the blockage of endogenous peroxidase by 3% H₂O₂, the slides were then blocked by 10% normal goat serum and incubated with primary antibodies. After being washed with PBS, the slides were developed using Dako EnVision + Rabbit Polymer (cat. no. K4003) from Dako (Carpinteria, CA, USA), followed by another thorough washing. The slides were counterstained with hematoxylin and coverslipped. The immunohistochemically-stained samples were scored by two pathologists blinded to the clinical parameters, separately. The staining intensity was scored as follows: 0 (negative), 1 (weak), 2 (medium) or 3 (strong). The extent of staining was scored as follows: 0, <5%; 1, 5–25%; 2, 26–50%; 3, 51–75%; and 4, >75% according to the percentages of the positively stained areas in relation to the whole tumor area. The immunoreactivity score (IS) for each sample was generated by multiplying the score for staining intensity with the score for staining extent. Final staining scores for samples of <4, 4, 6 and ≥8 were considered to be -, +, ++ and +++, respectively.

Two human papillary thyroid cancer cell lines, K1 and TPC1, were purchased from the University of Colorado 125 Cancer Center Cell Bank and were cultured in RMPI-1640 medium containing 10% FBS (Invitrogen, Carlsbad, CA, USA). 293T cells were purchased from the Type Culture Collection Cell Bank, Chinese Academy of Sciences and were cultured in Dulbecco’s modified Eagle’s medium containing 10% FBS (Invitrogen). All cells were maintained at 37°C with 5% CO₂ in proper humidity. It should be noted that the K1 cells are considered to be a mixed thyroid gland papillary carcinoma type, as it has been reported that the K1 cells are contaminated by GLAG-66, which is also derived from thyroid gland papillary carcinoma (20,21).

NRG1 antibody (10527-1-AP), NRF2 antibody (16396-1-AP), GAPDH (60004-1-Ig) and ERK1/2 antibody (16443-1-AP) were purchased from Proteintech Group, Inc. (Chicago, IL, USA) β-actin antibody (sc-47778) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The secondary antibodies, anti-mouse (SA00001-1) antibody conjugated with horseradish peroxidase (HRP) and anti-rabbit (SA00001-2) antibody conjugated with HRP, were purchased from Proteintech Group, Inc. All antibodies were diluted as per the manufacturer’s instructions when they were applied for specific experiments. NRG1 antibody and NRF2 antibody were diluted at a ratio of 1:50 for IHC and were diluted 1:1,000 for western blot analysis. GAPDH antibody and ERK1/2 antibody were diluted at a ratio of 1:1,000 for western blot analysis. β-actin antibody was diluted 1:500 for western blot analysis.

The pLKO.1-TRC cloning vector (Addgene plasmid 10878) was used to generate shRNA constructs against NRG1. The 21-bp target sequences against NRG1 were CGTGGAATCAAACGAGATCAT and CCACAGAAGGAGCAAATACTT respectively. These two constructs were afterwards used to produce lentiviruses and establish stable the cell lines named as ‘shNRG1-1’ and ‘shNRG1-2’, respectively. For lentivirus production, the packaging plasmid psPAX2 and the envelope plasmid pMD2.G were kind gifts from Dr Yi Qin, Pancreatic Cancer Institute, Fudan University.

To produce lentiviral particles, the shRNA constructs were co-transfected with the psPAX2 and pMD2.G plasmids at a ratio of 4:3:1 into the 293T cells using Lipofectamine™ 2000 Transfection Reagent (11668027; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Lentiviral particles were harvested after 48 h of transfection. After being infected by the lentiviral particles, the K1 and TPC1 cells were selected by puromycin, respectively to obtain cell lines that stably expressed shRNA.

An oxidant-sensitive fluorescent probe (DCFH-DA) was used to detect intracellular ROS levels (Sigma-Aldrich, St. Louis, MO, USA). Briefly, the cells were washed twice by phosphate-buffered saline (PBS), and were then stained with DCFH-DA (10 μmol/l) at 37°C for 20 min as per the manufacturer’s instructions. DCFH-DA was deacetylated by intracellular non-specific esterase, and was then oxidized by ROS to generate the fluorescent compound, 2,7-dichlorofluorescein (DCF). The DCF fluorescence intensity was detected using a FACScan flow cytometer (BD Biosciences, San Jose, CA, USA). Intracellular GSH activity that reflects the oxidative status of cells was determined using the GSH/GSSG Ratio Detection Assay kit (ab138881; Abcam, Cambridge, MA, USA).

TRIzol reagent (10296010; Thermo Fisher Scientific, Inc.) was used to extract total RNA from the cells as per the manufacturer’s instructions. cDNA reverse transcription was achieved using the PrimeScript™ RT Master Mix (RR036A; Takara Bio, Inc., Beijing, China). Relative quantitative (real-time) PCR was performed using SYBR^® Premix Ex Taq™ II (Tli RNaseH Plus) (RR820Q; Takara Bio, Inc.) to determine the mRNA expression levels of candidate genes normalized to β-actin, using the ABI 7900HT Real-Time PCR system (Applied Biosystems, Waltham, MA, USA) by the standard protocol. The thermocycling conditions are as follows: step 1, 50°C for 2 min; step 2, 95°C for 5 min (step 1 and step 2 are the hold stage); step 3, 95°C for 10 sec; step 4, 60°C for 1 min (recycle step 3 and step 4 to a total of 40 cycles, making the PCR stage); step 5, 95°C for 15 sec; step 6, 60°C for 1 min; step 7, 95°C for 15 sec (step 5 to step 7 are the melt curve stage). All reactions were conducted in triplicate. Relative quantification (ΔΔCq) was carried out as previously described (22). The primers used for PCR were as follows: NRG1 forward, 5′-CTAACATAGGAGAGTTAGGTGGC-3′ and reverse, 5′CTGTGGGCCAGTTAAACCTCTT-3′; ME1 forward, 5′-CCTCACTACTGCTGAGGTTATAGC-3′ and reverse, 5′-CGGTTCAGGATAAACTGTGGCTG-3′; TXNRD1 forward, 5′-GCAATCCAGGCAGGAAGATTGCT-3′ and reverse, 5′-CTCTTGACGGAATCGTCCATTCC-3′; GCLC forward, 5′-GTGGTACTGCTCACCAGAGTG-3′ and reverse, 5′-AGCTCCGTGCTGTTCTGGGCCTT-3′; GCLM forward, 5′-ATCTTGCCTCCTGCTGTGTGATGC-3′ and reverse, 5′-CAATGACCGAATACCGCAGTAGCC-3′; HMOX1 forward, 5′-GCTCTGGAAGGAGCAAAATCACACC-3′ and reverse, 5′-TATGACCCTTGGGAAACAAAG TCTGG-3′; NQO1 forward, 5′-AAGCCCAGACCAACTTCT-3′ and reverse, 5′-GCGTTTCTTCCATCCTTC-3′; GAPDH forward, 5′-GAACGGGAAGCTCACTGG-3′ and reverse, 5′-GCCTGCTTCACCACCTTCT-3′; and β-actin forward, 5′-CACCATTGGCAATGAGCGGTTC-3′ and reverse, 5′-AGGTCTTTGCGGATGTCCACGT-3′.

Cell lysates were obtained from 1×10⁶ cultured cells with a mixture of RIPA protein extraction reagent, protease inhibitor and phosphatase inhibitor (11836153001 and 4906845001; Roche, Shanghai, China). The protein concentration was determined by bicinchoninic acid assay (BCA). Equal amounts (50 μg) of total protein lysate were separated by 5 to 10% SDS-PAGE and then transferred onto PVDF membranes. The membranes were then blocked in 5% non-fat milk at room temperature for 1 h. Following this treatment, the membranes were probed with primary antibodies against NRG1, ERK1/2, NRF2, β-actin and GAPDH at 4°C overnight. Following incubation in a solution of goat anti-rabbit or anti-mouse IgG at room temperature for 1 h, the membranes were washed using TBST and then treated with enhanced chemiluminescence reagents (Thermo Fisher Scientific, Inc.).

The Cell Counting kit-8 (CCK-8; CK04; Dojindo Laboratories, Shanghai, China) assay and plate clone formation assay were used to evaluate cell viability and the cell proliferative capability. Briefly, 1,000 cells were seeded in 1 well of a 96-well plate for detection for 5 days. At the same time each day, the cell culture medium was changed and CCK-8 reagents were added at a ratio to medium of 1:10. Following 2 h of incubation, the optical density at 450 nm of each well was measured using a Synergy™ H4 Hybrid Multi-Mode Microplate Reader (BioTek, Winooski, VT, USA) and transformed into cell numbers accordingly. For clone formation assay, 500 cells were plated into a well of a 6-well plate and cultured in a cell incubator for 14 days with regular medium changes. The cells were then washed with PBS and fixed with 4% paraformaldehyde (P1110; Solarbio Biotechnology, Shanghai, China) at room temperature for 30 min. Then cells were stained with crystal violet staining solution (0.5%) (60506ES60; Yeasen Biotechnology, Shanghai, China) at room temperature for 30 min and then washed with PBS 3 times. The plate was dried at room temperature and images were aqcuired using a NEM-AL10 device developed by Honor, Huawei (Guangdong, China). Clone numbers were counted by Image J software and analyzed.

All statistical analyses were conducted using the SPSS software program (version 22.0; IBM Corp., Armonk, NY, USA). Pearson’s χ² test was conducted to assess the association between NRG1 expression and patient outcomes (Table I) and between NRG1 expression and NRF2 expression in thyroid cancer samples (Table II). A value of P<0.05 was considered to indicate a statistically significant difference. Pearson’s correlation coefficient test was used to assess the correlation between the expression of NRG1 and the expression of NRF2 (Fig. 5). Values of P<0.05 and R2>0.1 were considered to indicate statistically significant differences. One-way ANOVA with Dunnett’s Multiple Comparison test was used to determine the statistical significance and P-values in Figs. 2–4 and a value of P<0.05 was considered to indicate a statistically significant difference.

The clinicopathological characteristics of the patients have been provided in a previous study by our group (23). IHC staining and scoring was used to detect the protein levels of NRG1 in the PTC samples. The NRG1 level was higher in the PTC samples than that in the normal tissues (Fig. 1). In order to examine the role of NRG1 in the lymph node metastasis (LNM) of PTC, we examined its level in a PTC tissue microarray composed of 196 patient samples. Further analysis demonstrated that NRG1 was a positive indicator of PTC LNM at diagnosis (Table I).

To determine the biological role of NRG1 in thyroid cancer, we generated PTC cells K1 and TPC1) in which NRG1 was knocked down using shRNA (two shRNA constructs effectively decreased NRG1 expression. The knockdown efficiency was detected by RT-qPCR and western blot analysis after 48 h of transfection (Fig. 2A). To ascertain the contribution of NRG1 to the proliferation of PTC cells, a CCK-8 cell proliferation assay was conducted. Cell viability was significantly decreased when NRG1 was knocked down (Fig. 2B). Plate clone formation assay was also performed to detect the effect of NRG1 on thyroid cancer cell proliferation. The colony numbers of TPC1 and K1 cells were significantly decreased when NRG1 expression was silenced (Fig. 2C).

To explore the potential mechanisms through which NRG1 regulates the viability and proliferation of thyroid cancer cells, we further examined the association between NRG1 and other thyroid cancer-related pathways. The results revealed that both the expression of ERK1/2 and NRF2 decreased when NRG1 was knocked down (Fig. 3A).

We further detected the downstream targets of the NRF2/ARE pathway to explore the biological effects of NGR1. The results of RT-qPCR revealed that NRG1 silencing decreased the expression of ARE-driven genes, including GCLC, GCLM, HMOX1, NQO1, ME1 and TXNRD (Fig. 3B and C). These results suggest that NRG1 is an upstream transcriptional regulator of antioxidant enzymes, whose regulatory function is dependent on the NRF2/ARE pathway.

Subsequently, we attempted to explore the contribution of NRG1 to redox homeostasis. We found that when NRG1 was knocked down, the GSH/GSSG ratios in the PTC cells were significantly decreased, suggesting that NRG1 maintains the reductive status of cells (Fig. 4A and B). The intracellular ROS levels were examined to examine the effect of NRG1 on the oxidative stress level of PTC cells directly. Consistent with the changes observed in the GSH/GSSG ratio levels, the knockdown of NRG1 significantly elevated the intracellular ROS levels of the PTC cells, indicating that PTC cells in which NRG1 expression is knocked down are subjected to more severe oxidative stress (Fig. 4C and D). These results demonstrate that NRG1 plays an important regulatory role in the redox balance of PTC cells.

The association between NRG1 and NRF2 was analyzed by IHC staining of PTC tissues from 196 patients treated at the Department of Head and Neck Surgery, Fudan University Shanghai Cancer Center and by the analysis of data from 490 patient data extracted from TCGA database. In the 196 PTC samples, the NRG2 level was high, in parallel with the high level of NRG1 in the PTC samples. Statistical analysis indicated a positive association between NRG1 and NRG2 in the PTC tissue microarray (Table II). Linear regression analysis of the expression of NRG1 and NRF2 of 490 patients with PTC derived from the TCGA database revealed a similar pattern (Fig. 5). These results suggest a positive correlation between the expression of NRG1 and NRF2, implying the role of NRG1 in redox homeostasis clinically.

According to the results mentioned above, we explored the mechanisms through which NRG1 regulates redox homeostasis in PTC cells. NRG1 positively regulates NRF2, which upregulates ARE-related antioxidant genes (GCLC, GCLM, HMOX1, NQO1, ME1 and TXNRD). Through these mechanisms, NRG1 reduces ROS-induced oxidative stress and maintains the moderate redox homeostasis that is essential for the proliferation of papillary thyroid cancer cells (Fig. 6).