Introduction
Thyroid cancer is the most common endocrine malignancy and presents a rapid growing annual incidence in recent decades. The increased incidence is probably due to early detection using high-resolution ultrasonography and commonly diagnosed at a younger age [
1]. Histologically, there are five major types of thyroid cancer including papillary, follicular, poorly differentiated, anaplastic, and medullary thyroid cancer. Among them, papillary thyroid cancer (PTC) accounts for the majority (around 80%) of thyroid cancer [
2]. The first-line treatment of PTC is total thyroidectomy, then followed with radioactive iodine ablation of thyroid remnant and thyroid hormone suppression of thyroid-stimulating hormone (TSH). Generally, the overall survival is more than 90% [
3]. However, recent research disclosed there are still around 20% of PTC patients encountered disease recurrence with lymph nodes metastasis, causing the consequence of increased mortality [
4]. Indeed, heterogeneity within the category of PTC has been reported and only could be roughly characterized by some histological features and molecular markers, such as
BRAFV600E mutation and
RET/PTC rearrangements [
5]. Besides, using transcriptional and mutational landscape, well-differentiated thyroid cancer could be classified as three molecular subtypes including
BRAF-like,
RAS-like, and Non-
BRAF-Non-
RAS [
6]. Notably, increased prevalence of metastasis, gross extra-thyroidal extension, and multifocality were found in the
BRAF-like subtype [
7]. Therefore, further investigation on the genetic and epigenetic alternations of
BRAF-like PTC to identify novel prognosis associated markers is of vital importance.
Long non-coding RNAs (lncRNAs) were defined as RNA transcripts with more than 200 nucleotides and not translated into proteins. Recently, many lncRNAs have been recognized to present dynamic expression during the mammalian organ development [
8] and their dysregulations were observed among numerous cancers [
9], indicating lncRNAs could function as important regulators during organogenesis and involve in carcinogenesis.
HOTAIR (HOX Transcript Antisense Intergenic RNA), as one of the well-known lncRNAs, has recently been noted to possess potential roles in PTC.
HOTAIR could be detected in serum and the levels might serve as a diagnostic value to distinguish thyroid benign nodule and PTC [
10]. Also, PTC patients with lymph node metastasis have significantly higher serum
HOTAIR levels than them without metastasis [
11]. Moreover, PTC tissues were shown to express increased
HOTAIR levels, which further positively correlated with advanced pathological stages and poor prognosis of patients [
12]. These results point out
HOTAIR likely plays functional roles during PTC carcinogenesis and metastasis and can serve as a potential marker for evaluating disease status of PTC patients.
To explore
HOTAIR mediated molecular mechanisms in thyroid cancer, several groups have examined the
HOTAIR-miRNA-mRNA competitive endogenous RNA networks in multiple thyroid cancer cell lines. For examples, using papillary (TPC-1) and follicular (FTC-133) thyroid cancer cell lines,
HOTAIR was showed to sponge miR-1, activate
CCND2 expression, and promote thyroid cancer progression [
13]. Also,
HOTAIR can down-regulate miR-488-5p with upregulation of NUP205 and Bcl-2 to enhance growth, migration, and invasion of the papillary (BCPAP) thyroid cancer cell [
14]. Similarly,
HOTAIR promotes cell viability, migration, and invasion in papillary (TPC-1) and follicular (FTC-133) thyroid cancer cells via counter-regulating miR-17-5p [
15]. Using papillary (BCPAP) and anaplastic (HTh-7 and CAL-62) thyroid cancer cells,
HOTAIR/miR-761 sponge can regulate PPME1 to promote cell proliferation and inhibit cell apoptosis [
16]. Therefore, it is undoubtedly that
HOTAIR can sponge various miRNAs to enhance cell malignant behaviours in thyroid cancer cells.
However,
HOTAIR mediated epigenetic mechanisms are not limited to miRNAs sponge.
HOTAIR can also silence genes
in trans via interacting with polycomb repressive complex 2 (PRC2) at the 5’ end and lysine-specific histone demethylase 1 (LSD1) at the 3’ end to modulate H3K27 trimethylation and H3K4 demethylation, respectively [
17]. To our knowledge, this mechanism in papillary thyroid cancer has not been investigated before. Additionally, previous research mainly used the traditional PTC cell lines, such as BCPAP and TPC-1. Therefore, in this study, we utilized two newly characterized PTC cell lines (MDA-T32 and MDA-T41) [
18] to evaluate the
HOTAIR phenotypic effects on cellular biology after gain or loss of its expression. The Cancer Genome Atlas (TCGA) database were used to select genes that potentially repressed by
HOTAIR in PTC tissues. The cellular models, chromatin immunoprecipitation assay and immunohistochemical (IHC) staining of PTC tissues were further applied to validate the putative
HOTAIR-suppressed gene and evaluate its association with the clinical staging of PTC. We aim to strengthen current understandings of
HOTAIR-mediated carcinogenesis and epigenetic alternations in PTC with identification of novel prognosis associated markers for clinical application.
Discussion
Although patients with PTC generally have good overall survival rate under standard therapy, but there is still a concern for the high disease recurrence, which could reach up to 20% during the long-term follow-up [
4]. Also, there is still a difficulty to determine the optimal initial operational choice and postoperative therapeutic strategies (such as if radioiodine ablation is required post total thyroidectomy) for PTC patients. For examples, patients with 1–4 cm intrathyroidal PTC without evidence of contralateral lobe or cervical lymph nodes metastasis can receive either solely thyroid lobectomy or total thyroidectomy with/without radioiodine therapy [
24]. One important issue is that the heterogenicity within the histological features and genetic/epigenetic signatures of PTC have not been thoroughly understood. Therefore, there is a vital importance to recognize or discover novel prognosis-associated markers of PTC patients for clinical evaluation.
Recently, lncRNA
HOTAIR has been recognized as a practicable biomarker in patients with PTC. The
HOTAIR expression in serum or in thyroid tissues collected via fine-needle aspiration biopsies can serve as one of useful markers to differentiate benign nodules and thyroid cancers [
10,
25].
HOTAIR expression was shown to be significantly increased in PTC tissues, positively correlated with the advanced PTC stages, and negatively associated with overall survival probability of PTC patients [
12]. In vitro experiments using PTC cell lines further demonstrate
HOTAIR could promote cell proliferation, migration, and invasion via competitively sponging numerous microRNAs including miR-1 [
13], miR-488-5p [
14], miR-17-5p [
15] and miR-761 [
16]. In the present study, we observed similar findings that
HOTAIR was highly expressed in PTC tissues and its higher expression associated with lower overall survival. Notably, we further found there are differential
HOTAIR expression among four PTC cell lines (BCPAP, K1, MDA-T41 and MDA-T32), supporting there is indeed heterogenicity within the category of PTC. We further modulated
HOTAIR expression in two newly developed and well-characterized PTC cell lines. We performed
HOTAIR knockdown in high
HOTAIR-expressed MDA-T32 and conducted
HOTAIR overexpression in MDA-T41, which presented significantly lower
HOTAIR levels compared to MDA-T32. In line with previous reports, we demonstrated
HOTAIR can foster cell proliferation, colony formation and migration, redistribute cell cycle phases toward S phase (DNA replication) and G2/M phase (cell mitosis) and correspondingly change the signaling of PTEN/p-AKT pathway. Overall,
HOTAIR is thought as an important epigenetic regulator that vigorously participates in carcinogenesis, tumor growth, progression, and metastasis of PTC.
Currently, there is still limited understanding regarding the upstream regulatory mechanisms for
HOTAIR transcription.
HOTAIR can be upregulated by estradiol in vitro via enriching the bindings of estrogen receptors (ERs) to its promoter [
26] and genetic variants such as single nucleotide polymorphisms (SNPs) can also potentially influence the
HOTAIR expression with tissue-specific pattern. Indeed,
Zhu et al
. revealed
HOTAIR rs920778 T allele as a functional genetic variant associated with higher
HOTAIR levels in both normal thyroid and PTC tissues [
27].
HOTAIR expression could also be induced by the tumor microenvironment such as hypoxia [
28]. Therefore, the intrinsic factor (genetic variation) and extrinsic factors (estradiol and hypoxia) might coordinate together to modulate
HOTAIR expression and contribute PTC tumorigenesis and will require future investigation to clarify in depth.
To extend HOTAIR mediated epigenetic alterations in PTC beyond microRNAs sponge, we examine several genes presented negative correlation with HOTAIR expression in PTC tissues. Then, our attention was centred on the DLX1 gene since it retained negative corresponding changes after gain or loss of HOTAIR in our PTC cellular system. We further identified DLX1 gene was epigenetically repressed by HOTAIR via recruitments of PRC2 (EZH2) and H3K27me3 on its promoter region. Moreover, using the TCGA dataset and IHC staining of PTC tissues, we disclosed that DLX1 expression was significantly reduced in PTC tissues and decrease of DLX1 was associated with advanced PTC stages and lowered progression free survival. These results indicate DLX1 as a novel marker with a promising application for predicting the disease prognosis of PTC patients.
DLX1 belongs to the
DLX homeobox family genes that encode transcription factors possessing critical roles in the morphogenesis of craniofacial structures, pharyngeal arches, forebrain, and sensory organs [
29]. Particularly, pharyngeal arches are closely linked to the development of thyroid gland. The thyroid originates from a diverticular outgrowth of the primitive pharynx between the first and second pharyngeal arches [
30] and the developmental genes such as
DLX and
HOX genes are both important for patterning the anterior/posterior and dorsal/ventral axes of the pharyngeal arches [
31].
HOX genes are the most well-known subset of homeobox genes containing four chromosome clusters (
HOXA,
HOXB,
HOXC and
HOXD), which involved in the formation of various body structures during embryonic development [
32].
HOTAIR is transcribed from the antisense strand of
HOXC cluster in chromosome 12 and has been demonstrated to repress the expression of
HOXD genes in chromosome 2 [
17]. In this study, we further identified
DLX1 gene was epigenetically supressed by
HOTAIR. Intriguingly,
DLX1 gene just locates next to the
HOXD gene cluster, indicating the
HOTAIR mediated histone modifications can inhibit several genes in the near position of the same chromosome. Based on RNA sequencing data of GTEx (Genotype-Tissue Expression) Analysis Release V8 (dbGaP Accession phs000424.v8.p2), thyroid tissue is listed as the 2nd rank tissue with high
DLX1 expression (only below to the brain tissues), whereas
HOTAIR was lowly expressed in thyroid tissue (only above to the whole blood). Therefore, maintenance of low
HOTAIR expression and high
DLX1 levels in the thyroid gland is essential for its normal development and functionality. In this paper, we further demonstrated their dysregulation can promote the carcinogenesis and disease progression of PTC. However, future investigation is still warranted to ascertain the functional roles of DLX1 in PTC.
In summary, HOTAIR is an important epigenetic regulator for steering the tumorigenesis and disease progression of PTC through modulating the cellular malignancy. We validated HOTAIR can modulate histone modifications to inhibit DLX1 expression and loss of DLX1 was associated with advanced PTC stages and decreased progression free survival. Our results broaden HOTAIR mediated epigenetic regulatory networks in PTC to include histone modifications. Moreover, DLX1 was recognized as a novel prognosis associated biomarker in PTC likely to be practicable for clinical utilization.
Materials and methods
Analyze gene expression in PTC tissues and its association with clinical prognosis
The Cancer Genome Atlas (TCGA) thyroid cancer data were obtained from UCSC Xena (
http://xena.ucsc.edu). A total of 59 PTC tissues and matched adjacent normal thyroid tissues were applied to compare the
HOTAIR and
DLX1 expression. In survival analysis,
HOTAIR and
DLX1 expression were obtained from 513 and 515 PTC tissues, respectively, and stratified as high or low based on the expression level with significant differences in the survival outcomes and the lowest log-rank
P-value among subgroups using
Cutoff Finder [
33].
HOTAIR expression in 308 TCGA PTC tissues with different genetic alternations including
BRAFV600E,
NRAS,
HRAS mutations and
RET fusion were further compared. We also utilized the TCGA data to select the putative
HOTAIR suppressed genes that present a significantly negative correlation with
HOTAIR expression. The correlation between
HOTAIR and
DLX1 expression (n = 513) was further plotted and analyzed by Pearson correlation.
Cell lines and culture conditions
The human papillary thyroid cancer cell lines including BCPAP, K1, MDA-T41 and MDA-T32 were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin-amphotericin B at 37 ℃ in 5% CO2.
HOTAIR knockdown using small interfering (si)RNA transfection
The HOTAIR siRNAs (Cat#N-187951-02, GE Healthcare Dharmacon) and scramble siRNA (Cat#D-001210-01, GE Healthcare Dharmacon) were used to generate the HOTAIR knockdown cells (MDA-T32 si-HOTAIR) and scramble control cells (MDA-T32 Scramble). 10 nM siRNAs with transfection reagent were added to the cells following the protocols supplied by the manufacturer. Further assays or experiments were performed at 48 h post-transfection.
Generation of constitutive HOTAIR overexpression
The sequence of HOTAIR (NR_003716) was cloned from the LZRS-HOTAIR plasmid (Addgene plasmid #26110) and inserted into the pEGFP-Lv105 vector (Capital Biosciences) as the working HOTAIR plasmid. pEGFP-Lv105 empty vector was used as a control. Then, we applied lentiviral packaging kit (Cat#3D5F03, Origene) in HEK293T cells to produce lentiviral particles. The MDA-T41 cells were transduced with HOTAIR plasmid- or empty vector-lentiviral particles in 8 µg/ml hexadimethrine bromide (Sigma)-containing growth medium to generate MDA-T41 HOTAIR-OE cells or Control cells, respectively. Then, cells were revived in normal growth medium for an additional 24–28 h with subsequent selection in puromycin-containing growth medium.
Cell proliferation assay
The Cell Counting Kit-8 (CCK-8) assay purchased from Dojindo Laboratories (Cat#TJ557, Kumamoto) was used to evaluate cell proliferation rate. All the experimental protocols were performed as described in the manufacturer’s instructions. Briefly, 1 × 103 cells per well were seeded into a 96‑well plate and cultured at 37 °C. At the indicated time points, CCK‑8 solution was added to each well with 1:9 ratio (CCK-8: media) and cells were incubated at 37 °C for a further 3 h. The absorbance at 450 nm was measured with a microplate reader (The Synergy™ HT, Bio-Tek, Taiwan).
1 × 103 cells per well were seeded in a 6-well plate for 7–14 days at 37 ℃ in 5% CO2. The cells were fixed with 4% formalin for 30 min, washed by PBS and stained with 0.1% crystal violet (Cat#MKCH6258, Sigma). The colony forming units were photographed and counted using Image J.
Wound-healing assay
5 × 103 cells per well were seeded in a 12-well plate to reach 80–90% confluence. Then, a small wound was created in each well with a gentle wash to remove cell fragments. Cells were cultured at 37 ℃ in 5% CO2 and images were taken under a microscope at 0 and 16 h.
Flow cytometric analysis of the cell cycle
Briefly, cells with the culture medium were harvested, washed with PBS following 1300 rpm spinning for 5 min at 4 ℃ × 3 times, fixed with ice cold ethanol (75%), washed with 1% FBS contained PBS following 1300 rpm spinning for 5 min at 4 ℃ twice, then stained with propidium iodide (PI) solution containing 50 μg/ml PI (Sigma) and 10 μg/ml RNase A (Thermo Fisher Science). Images of the cell cycle were obtained by the FACSCalibur system (BD Biosciences).
Gene expression analysis
Total RNA was extracted from cells using TRIzol reagent and cDNA synthesis was performed using a high-capacity cDNA reverse transcription kit (Cat#00984365, Thermo Fisher Science). SensiFAST SYBR No-ROX mix (Cat#98005, Bioline) was used for running quantitative real-time PCR in a LightCycler 480 (Roche). The ΔCT values of target genes were normalized to the ΔCt of stably expressed reference transcripts (
GAPDH). The primer sequences used are listed in Additional file
1: Table S2.
Protein extraction and western blot
Total cells are washed with ice-cold PBS and whole cell extracts were prepared using protein lysis buffer. The Pierce
™ bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Science) was used to quantify protein amount of the samples. All immunoblots are standardized to the same amounts of proteins per well. Then, proteins were separated by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes. Western blotting analysis was subsequently performed via incubating the membranes with primary antibodies against to proteins of interest under appropriate dilution. Then, horseradish peroxidase-conjugated secondary antibodies and the enhanced chemiluminescence assay were applied to visualize the signal. Band intensities are determined via using an UVP GelStudio
™ PLUS Imager (Analytik Jena, Germany). The primary antibodies against CDK1 (1:1000, Cat#9116), CDK2 (1:1000, Cat#2546), CDK4 (1:1000, Cat#12790), CDK6 (1:1000, Cat#13331), Cyclin A2 (1:1000, Cat#91500), Cyclin B1 (1:1000, Cat#4135), Cyclin D1 (1:1000, Cat#55688), Cyclin E1 (1:1000, Cat#4129), p-AKT (Ser473) (1:1000, Cat#9271S), AKT (1:3000, Cat#9272S) and PTEN (1:1000, Cat#9552) were purchased from Cell Signaling Technology. Other primary antibodies used here including DLX1 (1:1000, Cat#PA5-28899, Thermo Fisher Science) and β-actin (1:10000, Cat#NB600-501, Novus). The complete photos of western blotting on these proteins were presented in the Additional file
1: Figure S4.
Chromatin Immunoprecipitation (ChIP)
We used the EZ-Magna ChIP
™ A/G Chromatin Immunoprecipitation kit (Cat#17–10086, Millipore) to perform ChIP experiments as the standard protocols. Briefly, in vivo cross linking was conducted via adding 1% formaldehyde to the cells (1 × 10
6 per ChIP) for 10 min. Sonication (2*10 bursts of 30 s ON/OFF at high-level output in the sonicator (XL2015, Misonix)) of the cross-linked chromatin was performed to generate < 500 bp DNA fragments with confirmation via running electrophoresis on the agarose gel stained with DNA VIEW (Cat#TT-DNA01, TOOLS). 1% of chromatin supernatant was collected as the input. Then, antibodies with ChIPAb + H3K27me3 (Cat#17–622, Millipore), ChIPAb + EZH2 (Cat#17–662, Millipore) or mouse IgG (Cat#12-371B, Millipore) were used to perform immunoprecipitation. For binding the antibody/antigen/DNA complex, magnetic protein A/G Beads (Cat# CS204457, Millipore) were applied. Then, ChIP samples were going through washing, reversal of cross-linking, and ChIP DNA isolation. The eluted ChIP DNA was used as the template for real-time qPCR using SYBR Green. To assess the occupancy of H3K27me3 and EZH2 on the
DLX1 gene, the primers target on
DLX1 promoter region were designed as listed in Additional file
1: Table S2.
Immunohistochemistry staining of normal thyroid and papillary thyroid cancer tissues
The paraffin embedded sections of normal thyroid tissues (n = 22) and papillary thyroid cancer tissues with different stages (n = 52 with stage I, n = 20 with stage II, n = 15 with stage III and n = 3 with stage IVA) were purchased from US Biomax, Inc (Cat#TH8010a, Cat#TH208, Cat#TH961). Immunohistochemistry (IHC) staining of DLX1 was performed as the standard protocol including deparaffinized, rehydrating, antigen retrieval, immunohistochemical staining using the antibody against DLX1 (Cat#PA5-28899, Thermo Fisher Science), dehydrating and stabilizing with mounting medium and viewing the staining under the microscope. The staining intensity from 0 to 3 was scored, with 3 referred as the section with maximum intensity, while 0 indicated negative. The percentage of staining was estimated, and 100% staining was scored as 1. The IHC scores were calculated by staining intensity*staining percentage and assessed by an independent pathologist.
Statistical analysis
Graphpad Prism 6.0 and SPSS version 22 were used for data analysis. The details of statistical methods were described in the figure legends.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.