70 cases of PTCs were collected by the Department of Pathology of Second Affiliated Hospital of Zhejiang University School of Medicine from 2019 to 2020. All PTC samples were surgically resected and the whole hematoxylin and eosin slides of all samples were reviewed by two senior pathologists. Formalin-fixed paraffin-embedded (FFPE) samples of PTCs were stored at room temperature (51 cases with BRAF^V600E mutation and 19 cases with BRAF^WT. 45 cases with lymph node metastasis and 25 cases with no metastasis).45 consecutive cases with lymph node metastasis were included in two years to compare the miR-222-3p in PTC tumors with and without lymph node metastasis.

This study was approved by the Human Research Ethics Committee of the Second Affiliated Hospital of Zhejiang University School of Medicine, Hangzhou, China. The IRB protocol number is 2019010.

B-CPAP (RRID: CVCL_0153) cell line was purchased from the Chinese Academy of Sciences (Shanghai, China). TPC-1 (RRID: CVCL_6298) cell line was purchased from Procell (Wuhan, China). K1 (RRID: CVCL_2537) cell line was purchased from Cellcook (Guangzhou, China). B-CPAP and TPC-1 were cultured in RPMI-1640 medium (Sigma), K1 was cultured in DMEM medium (Gibco). Both media were supplemented with 10% (v/v) fetal bovine serum. All cells were maintained at 5% CO₂ and 37°C in a humidified atmosphere.

Cells were seeded in a 12-well plate. After 24h, transfection was performed with Lipofectamine 3000 reagent (Thermo Fisher Scientific), following the manufacturer’s instructions. Transfection experiments were performed using 100nM miR-222-3p inhibitors or negative control (Genepharma). The inhibitor efficiency was greater than ninety-five percent. Sequences of miR-222-3p inhibitor and negative control were as following:

hsa-miR-222-3p inhibitor: 5’-ACCCAGUAGCCAGAUGUAGCU-3’; has-miR-222-3p negative control: 5’-CAGUACUUUUGUGUAGUACAA-3’.

PLX4720 (BRAF^V600E inhibitor) and PD0325901 (MEK inhibitor) were purchased from Selleck.

pcDNA3.1- plasmid was used as the vector plasmid, and Xba I (Takara, TCTAGA) and Hind III (Takara, AAGCTT) were determined as restriction sites. A BRAF^V600E point mutation (GTG→GAG) and the amplification primer sequence were shown below:

BRAF forward: 5’-GCTCTAGAATGGCGGCGCTGAGCGGTGG-3’, BRAF^V600E reverse: 5’-CTCCATCGAGATTTCTCTGTAGCTAGACCAA-3’; BRAF^V600E forward: 5’-TTGGTCTAGCTACAGAGAAATCTCGATGGAG-3’, BRAF reverse: 5’-CCCAAGCTTTCACTTGTCATCGTCGTCCTTGTAATCG TGGACAGGAAACGCACCATATC-3’.

Total RNA was isolated with RNA-Quick Purification Kit (YiShan Biotech) according to the manufacturer’s instructions. Reverse transcription of mRNA and miRNA (stem-loop primer) was performed following the manufacturer’s protocols of the PrimeScript™ RT reagent kit (Takara). Quantitative real-time PCR was performed following the manufacturer’s protocols of TB Green™ Premix Ex Taq™ II (Takara). Expression of miR-222-3p was normalized to internal control U6. Primer sequences as following: miR-222-3p forward: 5’-GGGAGCTACATCTGGCTA-3’ and reverse: 5’-GTGTCGTGGAGTCGGCAA-3’; U6 forward: 5’-CTCGCTTCGGCAGCACA-3’ and reverse: 5’-AACGCTTCACGAATTTGCGT-3’; miR-222-3p stem-loop primer: 5’- CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGGAGACC-3’.

Cells were lysed with Mammalian Protein Extract Reagent (Thermo Scientific) supplemented with protease inhibitor cocktail (1:100) (Thermo Scientific). Cell lysates were electrophoresed on 10% SDS-PAGE and then transferred to PVDF membranes. The membranes were blocked with 5% milk and then hybridized with a primary antibody at 4°C overnight. The membrane was washed and hybridized with horseradish peroxidase (HRP)-conjugated secondary antibody. Signals were visualized by SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Scientific). Primary antibodies were listed below: anti-E-cadherin (HUABIO, EM0502, 1:1000), anti-Snail (HUABIO, ER1706-22, 1:4000), anti-Zeb2 (HUABIO, ER64964, 1:1000), anti-NF-κB-P65 (HUABIO, ET1603-12, 1:2000), anti-p-P65(S536) (CST, EP2294Y, 1:1000), anti-p-P65(S468) (ABclonal, AP0446, 1:1000), anti-β-actin (HUABIO, M1210-2, 1:5000), anti-DYKDDDDK Tag (HUABIO, M1402-2, 1:2000).

For transwell migration assay, cell migration ability was measured by transwell permeable chamber with 8.0 μm pore size (Corning Costar, USA). Cell transfected with miR-222-3p inhibitor or negative control were collected after 72 h. 5 × 10⁴ cells were seeded on the upper chamber. 2.5% FBS was used as a chemokine in the lower chambers. After 4-5 h of incubation at cell incubator cells were fixed with 4% paraformaldehyde for 30 min and stained with crystal violet solution for 5 min. Cells were removed from the upper membrane with cotton swabs and migrated cells on the lower membrane were taken photos under a microscope (Random 5 fields, 40 ×).

In brief, FFPE tissue sections were deparaffinized in xylene and rehydrated in a series of decreasing alcohol concentrations. To expose miR-222-3p probes (MiR-222-3p probe sequence: 5’-ACCCAGTAGCCAGATGTAGC-3’), tissue sections were incubated with 15 μg/mL Proteinase K solution at 37°C for 10 min, after a titration experiment that established the optimal unmasking treatment while maintaining tissue morphology. 3% methanol-H₂O₂ were used to block endogenous peroxidase. Tissues were then pre-hybridized with ISH buffer for 30 min, and then hybridized with locked nucleic acid-based digoxigenin (DIG)-labeled miR binding oligonucleotides at 50°C. Tissues were then washed stringently in decreasing concentrations of saline-sodium citrate (SSC) buffer. Following the stringent wash, tissues were blocked in 2% sheep serum and 1% bovine serum albumin (BSA). Tissues were then incubated with anti-DIG-HRP at 37°C for 40 min. Freshly prepared DAB color developing solution was added and the color development time was controlled under the microscope. The sections were re-dyed by Harris hematoxylin for about 3 minutes and dehydrated. Positive is brownish yellow, and the nucleus is blue.

The scoring system (1-12) is done according to the depth of the color and the proportion of the positive cell. The images corresponding to scores 0, 3, 6, 9, and 12 are shown in Supplementary Figure 1 .

FFPE tissue sections were deparaffinized, rehydrated, antigen retrieved using Proteinase K, and blocked in sheep serum and BSA solution as described above in the ISH section. Snail antigen was retrieved by citrate antigen retrieval solution (Beyotime Biotechnology). Then, tissue sections were incubated with polyclonal rabbit anti-human primary antibody (Snail 1:150, ER1706-22, HUABIO) at RT for 1 h. A DAB Horseradish Peroxidase Color Development Kit (Beyotime Biotechnology) was used for the subsequent steps. Tissue endogenous peroxidase activity was blocked with peroxidase block for 5 min. Thereafter, sections were incubated with horseradish peroxidase (HRP)-conjugated secondary antibody for 30 min and visualized with diaminobenzidine (DAB) substrate at RT.

A two-tailed t-test was used for continuous variables (GraphPad Prism v9.0 and SPSS v25). Mann-Whitney t-test was used for ranked variables. Ranked correlation analysis was used to analyze the correlation of ISH score (hsa-miR-222-3p) and IHC score (Snail) in 60 PTC samples. Each sample was corresponding to an ISH score and an H-score. The bioinformatics data on the miRNA expression and mutation status of PTC were publicly available from UCSC Xena Browser under GDC TCGA Thyroid Cancer (THCA) datasets. PTC samples in TCGA were divided into two groups (BRAF^WT and BRAF^V600E mutation) according to BRAF gene status. A two-tailed t-test was used for comparing the miRNA expression. p<0.05 was considered as statistically significant.

Through differential analysis of miR-222-3p expression in PTC clinical samples from TCGA, we found miR-222-3p had a higher expression level in PTC patients harboring BRAF^V600E mutation (n=279) compared with that in PTC patients harboring BRAF^WT (n=212) (p<0.001) ( Figure 1A ). Subsequently, to further verify the clinical correlation of miR-222-3p, we evaluated miR-222-3p expression in PTC clinical samples from our research center through ISH analysis ( Table 1 ). ISH analysis of PTC patients harboring BRAF^V600E mutation (n=51) and BRAF^WT (n=19) demonstrated that PTC lesions with BRAF^V600E mutation showed a higher level of miR-222-3p than that with BRAF^WT (p<0.0001) ( Figures 1B, C ). In addition, we tested the expression of miR-222-3p in PTC cell lines, named TPC-1, B-CPAP, and K1 (TPC-1 cell line harbors BRAF^WT , while B-CPAP and K1 cell line harbor BRAF^V600E mutation). Through relative quantitative PCR (qPCR), we found the expression of miR-222-3p in B-CPAP (p=0.0154) and K1 (p=0.0035) cell lines was higher than that in TPC-1 ( Figure 1D ). We found that miR-222-3p was also overexpressed in the PTC cell line harboring BRAF^V600E mutation.

The above experiments showed that miR-222-3p expression in PTC was correlated with BRAF^V600E mutation. We speculated that the BRAF^V600E oncoprotein might promote miR-222-3p expression in PTC. To verify this hypothesis, we constructed a BRAF^V600E-Flag overexpression plasmid (OE) with pcDNA3.1- as a vector. 48 h after transfection, the differential expression of miR-222-3p could not be observed in TPC-1 transfected with OE plasmid (OE group) and empty vector (CTL group), neither was in B-CPAP ( Figures 2A–C ). However, 72 h after transfection, miR-222-3p expression in the OE group in both TPC-1 (p=0.0069) and B-CPAP (p=0.0385) were significantly increased compared with that in the CTL group ( Figures 2B, C ). These results confirmed that the BRAF^V600E oncoprotein up-regulated the expression of miR-222-3p in PTC cell lines.

To further clarify how the BRAF^V600E oncoprotein regulates miR-222-3p, RAF/ERK pathway inhibitors were used to treat PTC cell lines and the expression of miR-222-3p was measured. PLX4720 is a highly selective BRAF^V600E inhibitor, and its affinity to BRAF^V600E mutant protein is about ten times higher than that to BRAF wild-type protein. PD0325901 is a non-ATP-competitive MEK inhibitor that inhibits the phosphorylation of ERK1 and ERK2 ( Figure 2D ). We treated TPC-1, K1, and B-CPAP with 10μM PLX4720 and PD0325901, respectively, and measured the expression of miR-222-3p by qPCR after 24 h of treatment. These inhibitors did not affect the expression of miR-222-3p in TPC-1 ( Figures 2E, F ). For the K1 cell line, we found that 1μM PD0325901 significantly reduced the expression of miR-222-3p by 70% (p=0.0308). However, 1~10μM PLX4720 didn’t inhibit the level of miR-222-3p in K1 ( Supplementary Figures 2A, B ). After treatment of B-CPAP with PLX4720, the expression of miR-222-3p was reduced by about 50% (p=0.0433), and after treatment with PD0325901, the expression of miR-222-3p in B-CPAP was reduced by about 70% (p=0.0234) ( Figures 2E, F ). These results showed that the constitutive expression of miR-222-3p in TPC-1, K1, and B-CPAP could be distinct. The RAF/MEK/ERK pathway is responsible for the expression of miR-222-3p in B-CPAP; however, it seems that BRAF^V600E inhibitor PLX4720 plays no effect on the activation of the MAPK/ERK pathway and miR-222-3p expression in K1. As for TPC-1, there could be other mechanisms regulating miR-222-3p expression instead of MAPK/ERK pathway.

Through the above experimental data, the relationship between miR-222-3p expression and BRAF^V600E has been clarified. The ISH analysis also showed that PTCs with lymph node metastasis (n=45) had a higher level of miR-222-3p expression (p<0.0001) than those without lymph node metastasis (n=25) ( Figures 3A, B ). Besides, analysis of the TCGA database implied that the expression of miR-222-3p in PTCs with lymph node metastasis (n=226) was higher than that in PTCs without metastasis (n=231) (p < 0.0001) ( Figure 3C ). These results indicated that miR-222-3p overexpression might increase the risk of lymph node metastasis. Furthermore, miR-222-3p ISH score and BRAF mutation status were used to predict lymph node metastasis of PTCs. In the ROC curve analysis ( Figure 3D ), the area under the curve (AUC) was 0.8689, and the optimal cut-off value (0.5859) exhibited a sensitivity and specificity of 0.8667 and 0.76, respectively.

Next, we examined the role of miR-222-3p in PTC metastasis. The miR-222-3p inhibitor efficiency is greater than ninety-five percent ( Figure 4D and Supplementary Figure 2C ). Suppression of miR-222-3p in B-CPAP and TPC-1 cells by transfection of inhibitors suppressed cell migration (p=0.0009, p=0.0037) ( Figures 4A–C ). Nonetheless, the cell migration ability of K1 cell was not influenced by the miR-222-3p inhibitor ( Supplementary Figure 2D ). Meanwhile, EMT-related markers were measured including E-cadherin, N-cadherin, Twist1, Vimentin, Zeb1, Zeb2, Snail, and Slug through immunoblotting analysis. The decreased expression of Snail and Zeb2 and increased expression of E-cadherin were observed in the B-CPAP inhibitor group ( Figure 4E and Supplementary Figure 3A ). However, these changes cannot be observed in the TPC-1 cell line ( Figure 4F and Supplementary Figure 3B ). These results implicated that miR-222-3p influenced the cell migration ability and the underlying mechanism varied from types of cell lines or gene mutation.

In vitro experiment, we showed that miR-222-3p regulated the expression of Snail in B-CPAP. We further examined the expression of Snail in 60 clinical PTC samples by IHC. Through rank correlation analysis, we found that, Snail expression was correlated with miR-222-3p in the PTC clinical samples (r=0.33, p=0.0111) ( Figure 5D ).

Furthermore, we also analyzed the correlation between Snail expression and PTC lymph node metastasis. The Snail IHC score in the tumor tissue of PTC patients with lymph node metastasis was significantly higher than that without lymph node metastasis (p=0.0219) ( Figures 5A, B ). Furthermore, in the tumor tissues of PTC patients without lymph node metastasis, BRAF^V600E mutation is not significantly related to the Snail IHC score ( Figure 5C left); while in the tumor tissues of PTC patients with lymph node metastasis, Snail IHC score in the BRAF^V600E mutation group was significantly higher than that in the BRAF^WT group (p=0.0072) ( Figure 5C right). Consistent with the rank correlation analysis above, PTCs with both lymph node metastasis and BRAF^V600E mutation had the highest Snail IHC score and the highest miR-222-3p ISH score among the four groups. Taken together, these results showed that miR-222-3p could promote lymph node metastasis through upregulating Snail in PTCs harboring BRAF^V600E mutation.