The long non-coding RNA HOTAIR increases tumour growth and invasion in cervical cancer by targeting the Notch pathway

Evidence suggests that the long non-coding RNA (lncRNA), HOTAIR, is involved in cervical cancer pathogenesis. We examined serum HOTAIR expression levels in cervical cancer patients and determined the relationships between HOTAIR expression and several clinicopathological factors, including survival. We also examined the functional consequences of HOTAIR overexpression both in vitro and in vivo. Compared with control patients, HOTAIR expression was significantly greater in the serum of cervical cancer patients (P < 0.001). The results indicated that this increase was significantly associated with tumour size (P = 0.030), lymphovascular space invasion (P = 0.037), and lymph node metastasis (P = 0.043). Univariate analysis revealed that disease-free survival and overall survival times were significantly shorter in cervical cancer patients with high HOTAIR expression (hazard ratio [HR] = 4.27, 4.68 and P = 0.039, 0.031, respectively). Cell proliferation and invasion in vitro increased as a result of lentiviral-mediated HOTAIR overexpression in cervical cancer cell lines. HOTAIR knockdown inhibited these properties and increased apoptosis. In vivo xenograft experiments using the HOTAIR-overexpressing SiHa cell line revealed that HOTAIR was a strong inducer of tumour growth and modulated the expression of epithelial-mesenchymal transition and Notch-Wnt signalling pathway-related genes. This result suggested that HOTAIR overexpression promoted cell proliferation and invasion. In conclusion, increased HOTAIR expression was associated with decreased patient survival times. HOTAIR may be a useful target for treatment of cervical cancer patients.


INTRODUCTION
In 2012, cervical cancer was ranked fourth in incidence and mortality among cancers affecting females worldwide. Globally, there were approximately 528,000 new cervical cancer cases, and 266,000 deaths, in 2012 [1]. Recently, decreased incidence and mortality rates in developed countries have been attributed to the effectiveness of cervical cancer screening tests. However, the incidence rates remain high in developing countries, where 85% of all cervical cancer cases occur [2]. The cure rate for cervical cancer is up to 80-90% in the early stages (Stages I-II) and 60% in Stage III. However, the prognosis is still poor following cancer progression to an advanced stage or recurrence.
Circulating, non-coding RNAs are a novel class of diagnostic and prognostic biomarkers for malignant tumours [3,4]. Long non-coding RNAs (lncRNAs) are present in the body fluids of humans [5,6] and may be useful diagnostic biomarkers [7]. However, it is unknown

Research Paper
whether circulating lncRNAs are useful and valid biomarkers for cervical cancer diagnosis and treatment.
The well-characterised lncRNA, HOX transcript antisense RNA (HOTAIR), is expressed by the human HOXC locus on chromosome 12q13. 13. Epigenetic silencing metastasis suppressor gene results when the PRC2 complex is recruited by this lncRNA to specific HOXD locus target genes [8]. Increased HOTAIR levels are present during advanced-stage cervical cancer [9,10]. HOTAIR is a risk factor for increased metastasis and cervical cancer patient death [10,11]. However, previous studies have mostly emphasised study of microRNAs (miRNAs) expressed in tumour cells and tissues. A variety of cancer types (e.g., cervical cancer) can be diagnosed using tissue miRNA. However, collection of tissue samples can consist of invasive procedures, and depends on biopsy samples after the initial categorisation of clinical status.
Serum and plasma miRNAs represent new bloodbased markers useful for cancer detection, and detection of other diseases [12][13][14][15]. We previously found that HOTAIR is upregulated during cervical cancer. However, whether circulating HOTAIR can be a useful and valid cervical cancer biomarker remains to be determined. The function of HOTAIR in cervical cancer development, and its fundamental molecular mechanisms, are also unclear. This study revealed that expression of serum HOTAIR increased in cervical cancer patients. Higher serum HOTAIR expression was associated with changes in overall survival times of cervical cancer patients. HOTAIR overexpression (in SiHa cells) resulted in tumour growth rate increase via the Notch signalling pathway. These results suggested that HOTAIR has a crucial role during cervical cancer cell growth.

HOTAIR is elevated in serum from patients and overexpression is associated with a poor prognosis
We previously observed that HOTAIR transcription is upregulated by more than 30-fold in cervical cancer tissues by qRT-PCR [10]. We also evaluated HOTAIR expression in serum. The HOTAIR serum level was also greater in cancer patients (4.20794 ± 0.89) compared with control patients (0.76813 ± 0.24) (Figure 1).
To reveal whether HOTAIR circulating in serum is linked to clinicopathological features of cervical cancer, we measured the relative HOTAIR concentrations in serum of 153 patients at different stages of cervical cancer. The mean HOTAIR concentrations at stages I and II were 4.35 and 3.82, respectively; the differences were not statistically significant. However, relatively greater HOTAIR expression (≥ 1.5) was significantly correlated with features also associated with tumour recurrence (e.g., tumour size (P = 0.030), lymphovascular space invasion (LVSI) (P = 0.037), and lymph node (LN) metastasis (P = 0.043)) ( Table 1).
Next, we examined the relationships between HOTAIR expression and outcomes. Clinicopathological and outcome information was available for 153 cervical cancer patients. The lengths of the follow-up periods were Figure 1: Elevated expression of HOTAIR in human cervical cancer serum. HOTAIR expression was significantly higher in cervical cancer serum (n = 153) than in normal serum (n = 80). Relative HOTAIR expression was determined using qRT-PCR with U6 as an internal control. Data are expressed as the mean ± SD. **P < 0.001 vs. normal control.

HOTAIR overexpression promotes cell growth, invasion, and migration
Next, lentiviral-mediated overexpression of HOTAIR was performed to determine its functional role. SiHa and CasKi cells were used for the overexpression of HOTAIR, because expression is lower in these cells compared with HeLa cells [10]. Following lentiviralmediated overexpression, HOTAIR was upregulated in SiHa, and CasKi cells ( Figure 3A), which significantly enhanced colony formation and cell invasion ( Figure 3B). Overexpression in these cell types also promoted cell migration ( Figure 3C).

HOTAIR overexpression in SiHa cells increases xenograft tumour growth in mice
To explore whether HOTAIR can affect tumour growth in vivo, we inoculated SiHa cells as xenografts into nude mice ( Figure 5A). Tumour volume and weight were measured. Mean tumour volumes and weights (387.4 ± 23.5 mm 3 and 531 ± 27 mg, respectively) at day 50 in mice receiving HOTAIR-overexpressing SiHa cells were significantly larger than those (309.3 ± 32.7 mm 3 and 380 ± 93 mg, respectively) in mice receiving empty vectorexpressing cells (P < 0.001) ( Figure 5B). HOTAIR expression in tumour tissue was significantly greater in HOTAIRoverexpressing cells compared with control cells ( Figure  5C, 5D). Tumour weight correlated with tumour volume, as determined by calipers (P < 0.001; r 2 = 0.935). Histological examination revealed that more cells with large nucleoli and irregular nuclear membranes were present in HOTAIRoverexpressing xenografts than in control xenografts ( Figure 5E). We further evaluated tumour size and activity using magnetic resonance imaging (MRI) and positron emission tomography (PET) ( Figure 5F, 5G). Tumour growth was strongly induced by HOTAIR overexpression; tumour size and fluorodeoxyglucose (FDG) accumulation were significantly larger and greater, respectively, in mice inoculated with HOTAIR-overexpressing cells than in mice inoculated with empty vector-expressing cells. These findings suggested that HOTAIR promoted tumour growth in vivo and further supported our hypothesis that HOTAIR is involved in the pathogenesis of the malignant transformation of cervical cancer cells.

HOTAIR promotes a malignant phenotype via the epithelial-mesenchymal transition (EMT) and Notch signalling pathways
To determine the mechanism by which HOTAIR promotes a malignant phenotype in cervical cells, we assessed the status of important signalling cascades controlled by Notch in HOTAIR-overexpressing cells. HOTAIR overexpression in SiHa cells resulted in increased NOTCH1, HES1, and p300 expression in RNA ( Figure  6A) and protein ( Figure 6B, 6C).
Next, we determined the expression of EMTrelated genes and NOTCH1 levels in xenografts derived from HOTAIR-overexpressing SiHa cells. The β-catenin, N-cadherin, Vimentin, Snail, and Twist levels of expression were greater in HOTAIR-overexpressing tumours compared with the control tumours ( Figure 7).
Using qRT-PCR, we also examined the expression of NOTCH1, HES1, and p300 in xenografts derived from HOTAIR-overexpressing SiHa cells ( Figure 8A, 8B). NOTCH1, HES1, and p300 expression was significantly higher in HOTAIR-overexpressing xenograft tumours. These findings suggested that HOTAIR promoted tumour growth in vivo via the EMT and Notch signalling pathways.

DISCUSSION
In addition to identifying novel diagnostic markers for cervical cancer, a further understanding of molecular mechanisms fundamental to the progression and metastasis of cervical cancer is essential for development of more targeted, and effective, therapeutic treatments. Among various genes and proteins previously identified to be overexpressed or modified in cervical cancer, the HOTAIR lncRNA has attracted research interest due to its marked and consistent overexpression in all cell lines and its reported role in cell growth and survival in other cancers [16]. We investigated the functional role and clinical significance of circulating HOTAIR using serum from cervical cancer patients, cervical cancer cell lines, and mouse xenograft models. HOTAIR expression was greater in tumours with nodal metastasis compared with non-nodal metastatic tumours. High HOTAIR status and lymphatic invasion were prognostic factors for the lengths of disease-free and overall survival times. Taken together, these results suggested that HOTAIR may have a significant role in the pathogenesis of cervical cancer. However, our relatively small clinical sample size was a limitation of this study and may have reduced the power of our clinical analysis. Another limitation was that only subcutaneous xenograft models were used to investigate in vivo cancer cell behaviour. Subcutaneously implanted cancer cells allow for rapid and quantitative tumour formation; they are more appropriate for use in studies that include continuous measurement of the tumour. In contrast, it is inherently difficult to quantitatively monitor tumour growth using intraperitoneal and orthotopic xenograft models, but these models can represent a more relevant tumour microenvironment.
The serum lncRNA cervical cancer signature may be useful for monitoring cancer progression in the www.impactjournals.com/oncotarget clinical setting. No statistically significant differences were revealed by the analyses stratified by patient age, histological tumour type, or SCC Ag levels. However, the results indicated that there was a distinct correlation between serum HOTAIR expression and tumour size, LVSI, and LN metastasis. Greater serum HOTAIR expression was associated with an advanced disease status.
HOTAIR targets HOX genes and binds to PRC2. It targets the repressive PRC2 complex towards different PRC genomic sites [8,9,17,18]. The initiation, and progression, of many types of cancer involve this process [19][20][21][22]. In cervical cancer patients, HOTAIR expression is associated with tumour stage and EMT and is associated with a poorer prognosis [10,11]. Increasing evidence suggests that HOTAIR has an important role in cell growth, invasion, cancer metastasis, EMT, and stemness [23,24]. We therefore investigated the potential roles of HOTAIR. This investigation was based on the results of our prior analysis of clinical cervical cancer specimens. HOTAIR overexpression resulted in significant increases in cell proliferation and clonogenicity. Conversely, HOTAIR knockdown led to the suppression of these phenotypes and increased apoptosis. HOTAIR overexpression was positively correlated with cell migration and invasion in cervical cancer cell in vitro cultures, and with in vivo tumour growth in xenograft-bearing mice. Our findings suggested that HOTAIR is one of the critical lncRNAs contributing to cervical cancer carcinogenesis and progression.
Notch1 signalling contributes to tumour progression characteristics (e.g., invasion, EMT, metastasis, and angiogenesis) [25,26]. In cervical carcinoma cells, induction of anoikis resistance, inhibition of p53 activity, and upregulation of Myc are affected by Notch 1; during cell growth, it also impedes TGF-β's growth inhibition effects [27][28][29]. We previously reported that EMT correlates with HOTAIR expression in cervical cancer patients [10]. Results support Notch1's putative tumour progression function, but there is little evidence of associated downstream factors that contribute to the regulation of these processes associated with HOTAIR.
We also investigated whether the EMT and Notch-Wnt signalling pathways were compromised upon HOTAIR overexpression or knockdown. There was an obvious relationship between increased SiHa cell levels of EMT-related genes, NOTCH1, HES1, p300, and overexpression of HOTAIR. Conversely, HOTAIR knockdown decreased EMT-related gene expression in HeLa cells. Thus, HOTAIR may contribute to malignant cervical cancer cell phenotypes via the activation of Wnt-Notch signalling and EMT.
In summary, we found that cervical cancer patients had elevated serum HOTAIR levels. Enhanced serum HOTAIR expression was positively correlated with clinicopathological parameters in in vitro and in vivo cervical cancer cells. This result supports the use of Figure 6: HOTAIR regulates the Notch signalling pathway. A. NOTCH1, p300, and HES1 levels were determined by qRT-PCR in HOTAIR-overexpressing SiHa cells. Each assay was performed in triplicate. Data are mean ± SD. *P < 0.05 vs. control. B. Protein lysates were obtained from HOTAIR-overexpressing SiHa cells. NOTCH1, p300, and HES1 expression levels were analysed by western blotting. C. Band intensities were quantified and normalised to that of β-actin. www.impactjournals.com/oncotarget HOTAIR to determine clinicopathological stage and/or prognosis in cervical cancer patients. HOTAIR may also be a potential therapeutic target, because the results indicated that HOTAIR has a mechanistic role in the promotion of cell growth and invasion via modulation of the Notch-Wnt signalling pathway and EMT.

Ethics statement
Investigation has been conducted in accordance with the ethical standards and according to the Declaration of Helsinki and according to national and international guidelines and has been approved by the authors' institutional review board.

Patient and sample collection
Samples of blood serum were taken from female cervical cancer patients (n = 153) who underwent surgery between 2007 and 2012 (Yonsei Severance Hospital, Yonsei University, Seoul, Korea). These patients had newly-diagnosed, untreated, invasive, cervical cancer. Eighty control samples were obtained from patients who had a simple hysterectomy to treat uterine leiomyoma.
No samples from patients with concurrent gynaecological cancer were included in the analysis.
The treatment and control serum samples were obtained following the same protocol. A total of 4.0 ml peripheral blood were collected in BD Vacutainer tubes. Each sample was then incubated (30-60 min), and centrifuged (15 min, 2000 rpm), at room temperature. The supernatant was then transferred to an Eppendorf tube and was stored (-80°C) until RNA extraction was performed.
The results for the clinical information are presented in Table 1.

Cell culture
The SiHa (squamous cell cervical carcinoma) and Caski (epidermoid cell cervical carcinoma cells, originally derived from a small-bowel mesentery metastasis) human cervical cancer cell lines (American Type Culture Collection, Rockville, MD, USA) were maintained in medium (Dulbecco's modified Eagle's medium, Gibco-BRL, Gaithersburg, MD, USA) that included 10% (vol/ vol) foetal bovine serum and penicillin/streptomycin. The cell line maintenance conditions were 37°C, 95% air, and 5% CO 2 . Only cells that had been passaged < 20 times were used in the experiments.

Plasmid constructs, generation of stable cell lines
PCR was used to amplify the human HOTAIR transcript variant 3 cDNA. It was then inserted into a pLenti6/V5-D-TOPO vector (ViraPower™ Lentiviral Expression Systems, Invitrogen, Carlsbad, CA, USA, protocol). The plasmid was then transfected into the 293FT cell line for packaging. The resultant lentivirus was used to infect the cell lines. Medium containing blasticidin (Invitrogen) was used for selection of HOTAIR stably transfected cells.

Small interfering RNA (siRNA) transfection
The siRNAs (HOTAIR and negative control (siNC)) were purchased from Bioneer (Daejeon, Korea). A concentration of 5×10 4 cells/well was added to 6-well plates. The G-Fectin kit (Genolution Pharmaceuticals Inc., Seoul, Korea) was used according to the manufacturer's instructions to transfect these cells with 10 nM siRNA in phosphate-buffered saline (PBS). After a 48-hour post-transfection period, the siRNAtransfected cells were used for the in vitro assays. The target HOTAIR siRNA sequences were siRNA, 5′-AAUUCUUAAAUUGGGCUGG-3′.

Quantitative real-time PCR analysis (qPCR)
RNA was extracted from the samples and cell lines (TRIzol® reagent, Invitrogen). An Invitrogen reverse transcription reagent kit was used according to the manufacturer's instructions for reverse-transcription of 2 μg total RNA into first-strand cDNA. The qRT-PCR (ABI StepOnePlus Real-Time PCR system, Applied Biosystems, Foster City, CA, USA) included use of the SYBR® Green real-time PCR kit (Toyobo, Co., Ltd., Osaka, Japan) and a 20-μl reaction volume (10 μl SYBR-Green master PCR mix, 5 pmole, each, forward and reverse primers, 1 μl diluted cDNA template, and sterile distilled water). The Gene amplification conditions consisted of an initial denaturation at 95°C for 3 min, then 40 cycles of denaturation at 95°C for 15 sec, annealing at 60°C for 60 sec, and elongation at 72°C for 60 sec. The final elongation was performed at 72°C for 5 min. U6 was the internal standard used for all quantifications. The primer sequences were: HOTAIR,

Xenografts in mice
BALB/c mice (n = 16, 5-6 weeks of age, Orient Bio, Seongnam, Korea) were kept in aseptic, constant temperature and humidity, conditions (Yonsei Medical University protocol). Each mouse received a subcutaneous injection of a 150-μL suspension of SiHa cells that was placed into the dorsal scapula area. Calipers were used to measure tumour size two times per week. Tumour volume was calculated using a simplified equation to estimate a rotational ellipsoid (length × width 2 × 0.5). Each tumour was harvested at 50 days post-treatment.

Magnetic resonance (MR) imaging in mice
A Bruker Biospec 94/24 USR (9.4T) small animal scanner (35-mm diameter birdcage coil, Bruker BioSpin MRI, Ettlingen, Germany) was used to obtain the MR images. A custom-built cradle was used to immobilise each mouse during the MR experiment. T 2 -weighted images were obtained using the rapid acquisition setting and were acquired at the beginning of each imaging session. These images were used confirm that the animal was in the correct position inside the magnet bore. The T 2 -weighted images were acquired using the rapid acquisition setting. A 1.5% isoflurane and a O 2 / N 2 O (1:1) mixture, at a 0.7 L/min flow rate, was used for anaesthesia during the MR experiment. An air pillow was used to monitor respiration. Circulating warm water was used to maintain mouse body temperature within acceptable limits.

Micro positron emission tomography (PET) imaging in mice
A [ 18 F]-fluorodeoxy-glucose (FDG) reference image was used to evaluate the agent as a diagnostic and therapy follow-up tracer, for each mouse. The mouse was then injected with the imaging and photodynamic therapy bifunctional agent, 124 I-labeled pyropheophorbide-a derivative. The long half-life of 124 I (4.2d) allowed for multiple scans of the same mouse and the same agent over time (i.e., a longitudinal study). Tumour uptake relative to the rest of the body increased over time. This result indicated that the agent has therapeutic and tumour-monitoring potential. After FDG incubation, a microPET scanner (Inveon™ Dedicated PET, Siemens) was used to obtain an image of each cervix tumour. Iterative reconstruction and segmented attenuation were used to reconstruct the acquisition data. Standard software (Inveon™ Acquisition workplace) was used for the PET image analyses. Fusion and co-registration of the PET and CT data were performed using a software module. However, the data were manually corrected, if necessary. Volumes of interest were placed around the calcified areas for analysis of fused images. www.impactjournals.com/oncotarget

Matrigel invasion assay
A BD Biocoat Matrigel Invasion Chamber (pore size: 8 mm, 24-well; BD Biosciences, Bedford, MA, USA, following the manufacturer's protocol) was used for the Matrigel invasion assay. In brief, the cells (0.5×10 5 viable cells/well) were seeded into the Matrigelcoated upper chamber (BD Transduction Lab, San Jose, CA, USA). Serum-free medium (0.1-10% FBS) was added to the lower chamber, and the cells were incubated for 24 h. A cotton swab was then used to remove nonmigrating cells from the upper chamber. Differential Quik Stain Kit (Triangle Biomedical Sciences, Inc., Durham, NC, USA) was used to stain the cells on the lower surface of the insert. The invading cell number was quantified by counting at least six random fields (200× magnification). The experiment was repeated three times.

Statistical analysis
The results of the statistical analyses (SPSS application, standard version 20.0, IBM, Chicago, IL) were expressed as mean ± standard deviation (SD) values. Pearson's χ2 test, the Student's t-test, and Fisher's exact test were used to evaluate the associations between HOTAIR expression and the clinicopathological characteristics. The median value (1.5) was set as the cut-off value. The groups were categorised into HOTAIR expression status at above (high HOTAIR), and below (low HOTAIR), the cut-off value (1.5). The Kaplan-Meier method was used to analyse overall survival times. The log-rank test was used to estimate betweengroup differences. Stepwise Cox regression model analysis was used for the multivariate survival analysis of the parameters that were significant in the univariate analysis. The statistical tests were two-sided; a P value < 0.05 was considered to indicate a statistically significant result.