Downregulation of AAA-domain-containing protein 2 restrains cancer stem cell properties in esophageal squamous cell carcinoma via blockade of the Hedgehog signaling pathway

Downregulation of AAA-domain-containing protein 2 restrains cancer stem cell properties in esophageal squamous cell carcinoma via blockade of the Hedgehog signaling pathway. Am Physiol Physiol C93–C104, 2020. squamous cell carcinoma (ESCC) ranks among the ﬁve most common cancers in China and has a ﬁve-year survival rate of less than 15%. The transcription factor ATPase-family AAA-domain-containing protein 2 (ATAD2) has potential as a therapeutic target in various tumors, and microarray-based gene expression proﬁling reveals dysregulation of ATAD2 speciﬁcally in ESCC. Here we investigated whether ATAD2 could mediate a regulation of cancer stem cell (CSC) biological functions in ESCC. Immunohistochemical staining, reverse transcription quantitative polymerase chain reaction, and Western blot assays all revealed upregulation of ATAD2 in ESCC tissues and cell lines, which furthermore correlated with progression of ESCC. In loss-of-function experiments, silencing of ATAD2 inhibited activation of the Hedgehog signaling pathway, as indicated by reduced expression of glioma-associated oncogene family zinc ﬁnger 1 (Gli1), smoothened frizzled class receptor (SMO), and patched 1 (PTCH1). Investigations with 5-ethynyl-2 = -deoxyuridine (EdU), Transwell assay, scratch test, ﬂow cytometry, and colony formation assay showed that silencing of ATAD2 or inhibiting the Hedgehog signaling decreased the proliferation, invasion, and migration abilities along with colony formation, but elevated the apoptosis rate of CSCs. Furthermore, in vivo experiments validated the suppressive effect of siRNA-mediated ATAD2 silencing on tumor growth in nude mice. Thus, downregulation of ATAD2 can seemingly restrain the malignant phenotypes of ESCC cells through inhibition of the Hedgehog signaling pathway.


INTRODUCTION
Esophageal squamous cell carcinoma (ESCC) originates from cells on the surface of the middle and upper third of the esophagus (9). It is a highly prevalent and lethal cancer with poor prognosis, having a five-year survival rate of less than 15% (32). The occurrence of ESCC is generally associated with tobacco and alcohol consumption, intake of hot drinks and foods, and vitamin deficiency that is endemic in some regions of Asia (11). Surgical resection combined with chemotherapy is the predominant approach for ESCC treatment, although the outcome of existing treatments is unsatisfactory due to the high risk of metastasis (16). In recent years, neoadjuvant therapy has emerged as an effective treatment method for operable ESCC (2), despite provoking severe adverse reactions in~60 -70% of patients (25). Compelling new preclinical studies have demonstrated the potential effectiveness of gene therapy in tumorigenesis and progression of ESCC (20,25).
ATAD2 (also known as ANCCA) is a member of the AAA ϩ ATPase family, which is involved in various cancers via its participation in cell proliferation, apoptosis, invasion, and migration, and its overexpression in cancer tissue is an indicator of poor prognosis (5,37). High expression of ATAD2 has been identified in various types of tumors, such as breast cancer, hepatocellular carcinoma, and lung cancer (4,29). Additionally, ATAD2 protein is recognized as an oncogene, which may present a potentially tractable therapeutic target (14). Although the function of ATAD2 in ESCC has not been defined, gene expression profiling-based microarrays with GSE100942 and GSE20347 reveal a dysregulation of ATAD2 in ESCC, which suggests a promising functional significance of ATAD2 in ESCC progression. Interestingly, cell biology results suggest that ATAD2 could mediate the Hedgehog signaling pathway to affect progression in human hepatocellular carcinoma (29). The Hedgehog signaling pathway is mainly involved in embryogenesis, and has generally low activity in the mature organism (3). Furthermore, the Hedgehog signaling pathway does participate in cell differentiation and proliferation, as well as in establishing tissue polarity (34). As such, the Hedgehog signaling pathway is also implicated in the pathogenesis of a variety of tumors, including ESCC (17), and ectopic expression of proteins related to this signaling pathway shares an association with tumorigenesis (33). Moreover, the development of cancers in skin, brain, digestive tract, lung, and prostrate has been associated with an abnormal activation of the Hedgehog signaling pathway (1). Based on the aforementioned literature, we propose the hypothesis that ATAD2 affects the progression of ESCC via its interaction with the Hedgehog signaling pathway. Hence, we undertook a study aiming to elucidate the potential roles of ATAD2 in ESCC and the mechanisms involving the Hedgehog signaling pathway.

MATERIALS AND METHODS
Ethics statement. This study was performed with the approval of the Ethics Committee of the Fourth Affiliated Hospital of China Medical University, and all participants signed informed consent documentation. Experiments involved human beings were conducted in strict accordance with the Declaration of Helsinki and animal experiments were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol involving animals was approved by the Institutional Animal Care and Use Committee of the Fourth Affiliated Hospital of China Medical University.
Microarray-based gene expression profiling. The ESCC-related gene expression data sets were retrieved in the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), from which the data sets GSE100942 and GSE20347 were downloaded. The GSE100942 annotation platform was GPL570 (HG-U133 Plus 2) Affymetrix Human Genome U133 Plus 2.0 Array, and the GSE20347 annotation platform was GPL571 Affymetrix Human Genome U133A 2.0 Array. We included the gene expression in ESCC surgical specimens and adjacent normal tissues in our analysis. The data were standardized by the Limma package in R language (22). The differentially expressed genes (DEGs) related to ESCC were screened with a P Ͻ 0.05 and |logFoldChange| Ͼ2 as the threshold, and a heatmap of DEGs was plotted. The interacting DEGs of the two data sets were identified by a Venn diagram (http://bioinformatics.psb.ugent.be/ webtools/Venn/).
Clinical tissue collection. ESCC and adjacent normal tissues were collected from 52 ESCC patients (mean age 58.3 Ϯ 6.2 yr) composed of 34 men and 18 women hospitalized at the Fourth Affiliated Hospital of China Medical University from March 2018 to March 2019. All patients were diagnosed with malignant ESCC by pathologists and underwent resection of esophageal cancer before chemotherapy. After resection, tumor tissues and adjacent normal tissues were collected and stored at Ϫ80°C.
Immunohistochemistry. Paraffin-embedded sections were dewaxed by conventional methods using xylene ⌱ and II for 10 min each, followed by dehydration with gradient alcohol (100%, 95%, 80%, and 70%, 2 min each) and two washes in distilled water (5 min/time) in a shaker. Next, the sections were soaked in 3% H 2O2 for 10 min, and rewashed with distilled water. After antigen retrieval under high pressure for 90 s, the sections were cooled to room temperature and rinsed with phosphate-buffered saline (PBS). Then the sections were blocked with 5% bovine serum albumin (BSA) at 37°C for 30 min. Subsequently, the sections were incubated overnight with a primary antibody against ATAD2 (1:100, ab244431; Abcam Inc., Cambridge, UK) at 4°C followed by PBS rinses for 2 min. Next, the sections were incubated with biotinylated goat anti-rabbit immunoglobulin G (IgG; 1:100, SF8-0.3; Beijing Solarbio Science & Technology Co. Ltd., Beijing, China) at 37°C for 30 min. After reacting the sections with streptavidin-biotin and coloration by diaminobenzidine, the sections were counterstained with hematoxylin for 5 min. PBS instead of primary antibody was used as the negative control (NC). The positive cells presented with brown-yellow in the cytoplasm or membrane. The positive cells were visualized in five randomly selected high-power visual fields to calculate the percentage of ATAD2-positive cells in the total cells.
Cell culture and isolation of stem cells. Human ESCC cell line Eca-109 was purchased from iCell Bioscience Inc. (Shanghai, China). The normal human esophageal epithelial cell (HEEC) line was purchased from American Type Culture Collection (Manassas, VA). The cells were then cultured in complete medium supplemented with 10% fetal bovine serum (FBS), 90% Roswell Park Memorial Institute 1640 (RPMI-1640), and l% penicillin-streptomycin at 37°C with CO 2 in a humidified atmosphere. After 48 h, the culture medium was replaced with fresh medium. When the cell confluence reached 80%, the cells were passaged at the ratio of 1:2, and the ESCC stem-like cells were isolated by suspension culture. Eca-109 cells were detached, centrifuged, and counted. Here, the suspension medium consisted of equal volumes of serum-free Dulbecco's modified Eagle's medium and Nutrient F-12 Ham (DMEM-F12), supplemented with 2% B27, 20 ng/mL epidermal growth factor (EGF), and 10 ng/mL basic fibroblast growth factor. The cells were then resuspended in the medium, cultured in 25 cm 2 culture flasks at density of 1 ϫ 10 5 cells/mL, and supplemented with 2 mL suspension medium every other day. After 9 to 10 days, the morphology of Eca-109 microspheres was observed under an inverted microscope, and the microspheres were passaged. After two passages, the cell spheres were cultured with high-glucose DMEM containing 10% FBS. The cells were then cultured in medium with or without FBS alternatively for three times, and the adherent cells were collected for the subsequent experiments.
Plasmid construction and cell treatment. The sequence of ATAD2 was retrieved from National Center for Biotechnology Information. The overexpression and interference sequences as well as the corresponding NC sequences were constructed by Sangon Biotech Co., Ltd. (Shanghai, China) and synthesized by Shanghai Genechem Co., Ltd. (Shanghai, China) ( Table 1). The sequences were inserted into the p-AAV-CMV, respectively. The virus was packaged and purified. Next, the virus titer was determined. The efficiency of three small interfering RNAs (si-ATAD2-1, si-ATAD2-2, and si-ATAD2-3) against ATAD2 was assessed.
Reverse transcription quantitative polymerase chain reaction. Total RNA was extracted using the miRNeasy Mini Kit (217004, Qiagen, Hilden, Germany). The primers of ATAD2, glioma-associated oncogene family zinc finger 1 (Gli1), smoothened frizzled class receptor (SMO), patched 1 (PTCH1), B cell lymphoma-associated X (Bax), B cell CLL/lymphoma 2 (Bcl-2), matrix metalloproteinase 2 (MMP2), OCT4, and SOX2 were designed and synthesized by Takara , and HJURP) in top 100 differentially expressed genes from human GSE100942 and GSE20347 data sets. B: based on human GSE20347 data set, ATAD2 expressed at a relatively high level in ESCC samples relative to control samples, where the x-axis refers to sample type, y-axis refers to gene expression level, green box refers to normal controls, and red box refers to tumor samples. C: expression of ATAD2 in ESCC tissues was much higher than that in control tissues based on human GSE100942 data set, where the x-axis refers to sample type, y-axis refers to gene expression level, green box refers to normal controls, and red box refers to tumor samples. ATAD2, ATPase-family AAA-domain-containing protein 2; ESCC, esophageal squamous cell carcinoma. (Tokyo, Japan) ( Table 2). The extracted RNA was reverse transcribed into complementary DNA (cDNA) with the PrimeScript RT Kit (RR036A, Takara, Tokyo, Japan). The fluorescent quantitative PCR was carried out in accordance with the instructions of the SYBR Premix Ex Taq II kit (RR820A, Takara, Tokyo, Japan). Real-time quantitative PCR was conducted in the ABI 7500 Fast Real-Time PCR System (7500, ABI Company, Oyster Bay, NY). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal reference, and the relative expression of each gene was calculated with the 2 Ϫ⌬⌬Ct method. Western blot analysis. The total protein was extracted using 100 L lysis buffer containing 1 L proteinase inhibitor (Roche, Basel, Switzerland) on ice for 30 min. A total of 50 g protein was dissolved with 2 ϫ sodium dodecyl sulfate (SDS) loading buffer and boiled at 100°C for 5 min, then loaded for SDS-polyacrylamide gel electrophoresis. The separated proteins were transferred onto a polyvinylidene fluoride membrane and blocked with 5% skim milk powder at room temperature for 1 h. After rinsing with PBS for 2 min, the membranes were incubated overnight with the following diluted primary antibodies (purchased from Abcam Inc., Cambridge, UK): mouse antibodies against ATAD2 (1:5,000, ab176319), Gli1 (2.5 g/mL, ab49314), SMO (1:300, ab236465), PTCH1 (1:1,000, ab53715), Bax (1:5,000, ab32503), Bcl-2 (1:1,000, ab32124), MMP2 (2.0 g/ mL, ab37150), OCT4 (1:1,000, ab181557), and SOX2 (1 g/mL, ab97959) at 4°C. The following day, the membrane was washed three times with Tris-buffered saline Tween 20 (5 min/time) and then incubated with the secondary antibody, horseradish peroxidase-labeled goat-anti mouse antibody against IgG (HA1003, Yanhui Biotech Science & Technologies Co., Ltd, Shanghai, China) for 1 h. The immunocomplexes on the membrane were visualized using enhanced chemiluminescence solution (ECL808-25, Biomiga Inc., San Diego, CA) for 1 min. Next, the membrane was exposed to X-ray film (36209ES01, Shanghai QCbio Science & Technologies Co., Ltd., Shanghai, China). The ratio of the gray value of the target band to GAPDH was representative of the relative protein expression.

Control
Immunofluorescence for ATAD2 localization and flow cytometry for expression of stem cell markers CD44 ϩ and CD133 ϩ . Eca-109 microspheres were plated onto the cover glasses in a 24-well plate at a density of 5 ϫ 10 4 cells/well. After being cultured for 2 or 3 days, the microspheres on the cover glass were fixed with 4% paraformaldehyde, and then blocked with normal goat serum at 37°C for 30 min. After that, the microspheres were incubated with primary antibody to ATAD2 at 4°C overnight and with fluorescein isothiocyanate (FITC)labeled goat anti-rabbit IgG (Beijing Bioss Biotechnology Co., Ltd., Beijing, China) for 2 h, after which the cover glasses were mounted with glycerin-jelly. The microspheres were observed under a laser confocal microscope for determining the subcellular localization as well as the fluorescence intensity of ATAD2.
Eca-109 microspheres were detached with 0.25% trypsin, which was halted by the addition of culture medium. The microspheres were rinsed in 0.01% PBS and resuspended, followed by a centrifugation. The cells were reacted with 10 mL FITC-labeled antibody to CD44 (11-0441-82, eBioscience, San Diego, CA) and 10 mL PE-labeled antibody to CD133 (12-1339-42, eBioscience, San Diego, CA), 20 mL Fc receptor blocker and 80 mL buffer, and cooled down at 4°C for 10 min in subdued light. Finally, the cells were washed and resuspended with 500 L washing buffer. The CD44 ϩ CD133 ϩ cells were then analyzed using flow cytometry.
Sphere formation assay. Approximately 1 ϫ 10 4 cells were seeded into a 24-well ultra-low-attachment plate and cultured with serum-free DMEM-F12 medium containing 10 ng/mL EGF and 10 ng/mL fibroblast growth factor-␤. The culture medium was half-renewed every 2 days. After 10 days, the cells were observed, photographed, and counted under a microscope.
5-Ethynyl-2=-deoxyuridine staining. The 5-ethynyl-2=-deoxyuridine (EdU) solution was added to the cells and incubated for 2 h. The cells were fixed with 4% paraformaldehyde for 30 min and cultured in glycine solution for 8 min. Afterwards, cell permeation was performed with PBS containing 0.5% Triton X-100. Then the cells were   Flow cytometry for cell apoptosis. Cell apoptosis was detected using the Annexin V-FITC/propidium iodide (PI) double-staining method. In short, the cells were cultured in a 37°C incubator with 5% CO 2 for 48 h, followed by two PBS washes and a centrifugation. The cells were resuspended in 200 L binding buffer and reacted with 10 L Annexin V-FITC (ab14085, Abcam Inc., Cambridge, UK) and 5 L PI at room temperature for 15 min devoid of light. After that, the cells were mixed with 300 L binding buffer. Apoptosis was detected by a flow cytometer at an excitation wavelength of 488 nm.
Scratch test. The cells were seeded into six-well plates at a seeding density of 5 ϫ 10 5 cells/well. When the cell confluence reached 90%, a scratch line was drawn along the center of each well. The floating cells were removed by PBS washing, and the remaining cells were cultured with serum-free medium for 0.5-1 h. The cells were photographed after cell recovery, which represented the cell state at 0 h. Then, the medium was renewed with the complete medium, and the new cell state at 24 h was recorded. The cell migration distance was measured using the Image-Pro Plus Analysis Software (Media Cybernetics, Bethesda, MD).
Transwell assay. The apical chamber of the Transwell system was coated with Matrigel (50 L/well; 356234, Becton, Dickinson and Company, Franklin Lakes, NJ). The cells were resuspended with serum-free medium (1 ϫ 10 5 cells/mL) and plated in the Matrigelcoated apical chamber. Meanwhile, the medium containing 10% FBS was placed into the basolateral chamber. The Transwell chamber was placed in an incubator at 37°C for 24 h. After fixation in 5% glutaraldehyde at 4°C, the cells were stained with 0.1% crystal violet for 30 min, followed by PBS washing. The noninvaded cells were wiped off, and the number of cells passing through the Matrigel was counted.
Colony formation assay. The cells in the logarithmic growth phase were seeded into six-well plates at a density of 1 ϫ 10 3 cells/well and cultured at 37°C with 5% CO 2. When visible colonies appeared, the culture process was terminated. After being fixed with 4% polyoxymethylene for 20 min, the cells were stained with Giemsa's solution for 10 min, washed with tap water, and air-dried. The number of colonies per well was finally counted.
Xenograft tumor model in nude mice. A total of 36 severe combined immune deficient (SCID) nude mice (aged 4 -5 wk, half male and half female) were assigned into six groups: blank, NC, ATAD2, si-ATAD2, Cyc, and si-ATAD2 ϩ Cyc, with six mice per group. The ESCC stem cells (1 ϫ 10 6 ) were resuspended in 200 L normal saline and injected subcutaneously into the back of right hind legs of nude mice after ether anesthesia. The tumor dimensions were recorded once every 4 days under the same circumstances, and the tumor volume was calculated as length ϫ width 2 /2. The nude mice were euthanized on the 24th day, whereupon the tumors were dissected and collected.
Statistical analysis. Statistical analyses were conducted using SPSS 21.0 software (IBM Corp., Armonk, NY), Measurement data are expressed as means Ϯ SE. The Mann-Whitney U test was used for the comparison between two groups, while the Kruskal-Wallis test along with Dunn's multiple-comparison test were adopted for comparison among multiple groups. Enumeration data are expressed as percentage and were analyzed by 2 -test. P Ͻ 0.05 indicates that the difference is statistically significant.

RESULTS
The promising functional significance of ATAD2 in ESCC. The ESCC expression microarrays GSE100942 and GSE20347 were retrieved from the GEO database. There were 198 and 423 DEGs screened from the data sets of GSE100942 and GSE20347, respectively. The top 100 DEGs with larger fold expression change were selected to perform intersection analysis (Fig. 1A) and five DEGs (KIF4A, ATAD2, TRIP13, UBE2C, and HJURP) were identified. Among those DEGs, ATAD2 is recognized as an oncogene that closely associates with various human cancers such as ovarian carcinomas, hepatocellular carcinoma, cervical cancer, and gastric cancer (15,26,36,37). Since the role of ATAD2 in ESCC remains unclear, the study focused on the effects of ATAD2 on the progression of ESCC. As shown in the data from microarrays GSE20347 and GSE100942, expression of ATAD2 was up- regulated in ESCC (Fig. 1, B and C). Hence, we speculated the abnormal expression of ATAD2 in ESCC might affect cancer progression.

ATAD2 expresses highly in ESCC tissues and cell lines and its expression is correlated with clinicopathological progression.
To clarify the expression of ATAD2 in ESCC and its correlation with clinicopathological characteristics of patients, we first applied immunohistochemistry to determine the expression of ATAD2 protein in ESCC and adjacent normal tissues. The results showed that the expression of ATAD2 in ESCC tissues was increased compared with that in adjacent normal tissues (P Ͻ 0.05; Fig. 2A). Next, Western blot analysis was employed to measure the protein expression of ATAD2 in normal HEECs and human ESCC cell line Eca-109, which illustrated that ATAD2 protein was expressed at a higher level in Eca-109 cells than in HEECs (P Ͻ 0.05; Fig. 2B). Further, we analyzed the correlations between ATAD2 expression and the clinicopathological characteristics of ESCC patients, finding that ATAD2 expression was not linked to age or sex (P Ͼ 0.05), but correlated with tumor node metastasis (TNM) stage and histological grading (P Ͻ 0.05; Table 3). These results suggested that high ATAD2 expression correlates with ESCC progression.
si-ATAD2-1 exhibits best interference efficiency and is selected for subsequent experiments. To understand the effect of ATAD2 on ESCC cells, three siRNAs including si-ATAD2-1, si-ATAD2-2, and si-ATAD2-3 were designed to downregulate ATAD2 and silencing efficiency was determined by reverse transcription quantitative polymerase chain reaction (RT-qPCR) (Fig. 3A) and Western blot analysis (Fig. 3, B and C). We found that ATAD2 mRNA and protein levels were effectively downregulated in cells transfected with si-ATAD2-1, si-ATAD2-2, and si-ATAD2-3, where si-ATAD2-1 presented the highest efficiency. Therefore, si-ATAD2-1 was selected for subsequent experiments.

Downregulation of ATAD2 blocks the activation of the Hedgehog signaling pathway in ESCC stem cells.
Several studies have reported the link of ATAD2 to the Hedgehog signaling pathway (29,31). To clarify the effect of ATAD2 on the Hedgehog signaling pathway, we conducted RT-qPCR and Western blot analysis to determine the expression of the Hedgehog signaling pathway-related genes (Gli1, SMO, and PTCH1) in ESCC stem cells transfected with si-ATAD2 or oe-ATAD2, or treated with Cyc (a Hedgehog signaling pathway inhibitor). As shown in Fig. 4, A-C, mRNA and protein levels of ATAD2, Gli1, SMO, and PTCH1 were increased in the ESCC stem cells transfected with oe-ATAD2 (P Ͻ 0.05). On the contrary, ATAD2 expression was reduced by si-ATAD2 transfection as well as simultaneous ATAD2 silencing and Cyc treatment (P Ͻ 0.05), but it was not affected by Cyc treatment alone (P Ͼ 0.05). Meanwhile, mRNA and protein levels of Gli1, SMO, and PTCH1 were reduced by transfection with si-ATAD2, Cyc treatment, or combined treatment of si-ATAD2 and Cyc (P Ͻ 0.05). Compared with the cells treated with Cyc, mRNA and protein expression of Gli1, SMO, and PTCH1 was downregulated in the cells treated with si-ATAD2 and Cyc (P Ͻ 0.05). Collectively, these results show that ATAD2 silencing can disrupt the activation of the Hedgehog signaling pathway in ESCC stem cells.
ATAD2 silencing reduces the malignant phenotypes of ESCC stem cells. To identify the biological functions of ATAD2 in ESCC stem cells, ATAD2 loss-of-function experiments were carried out. First of all, immunofluorescence staining analysis revealed that ATAD2 was located in the nucleus (Fig. 5A). Flow cytometry was then performed to detect CD44 ϩ and CD133 ϩ cells, both of which are markers of cancer stem cells (CSCs). RT-qPCR and Western blot analysis were used to determine the expression of CSC-related genes OCT4 and SOX2. As displayed in Fig. 5, B-E, ATAD2 overexpression resulted in increased percentages of CD44 ϩ and CD133 ϩ cells along with upregulated mRNA and protein levels of OCT4 and SOX2 (P Ͻ 0.05) whereas ATAD2 silencing led to reduced percentages of CD44 ϩ and CD133 ϩ cells but reduced mRNA and protein levels of OCT4 and SOX2 (P Ͻ 0.05). Compared with the Cyc group, the si-ATAD2 ϩ Cyc group had relatively lower protein expression of CD44 ϩ and CD133 ϩ as well as reduced mRNA and protein expression of OCT4 and SOX2 (P Ͻ 0.05). Thus, silencing of ATAD2 could reduce the stemness of ESCC cells. ATAD2 silencing disrupts ESCC stem cell sphere formation. To clarify the effect of ATAD2 on the properties of ESCC stem cells to form colonies, we conducted sphere formation assay. As displayed in Fig. 6, A and B, sphere formation rate was enhanced by ATAD2 overexpression but inhibited by simultaneous ATAD2 silencing and Cyc treatment or either alone (P Ͻ 0.05). Compared with the Cyc group, the si-ATAD2 ϩ Cyc group had relatively lower rate of sphere formation (P Ͻ 0.05). These findings provide evidence that downregulation of ATAD2 could reduce the rate of sphere formation.
ATAD2 silencing reduces ESCC stem cell proliferation. The effects of ATAD2 and Hedgehog signaling pathway on the proliferation of ESCC stem cells were evaluated by EdU staining (Fig. 7, A and B). ATAD2 overexpression elevated cell proliferation rate, while simultaneous ATAD2 silencing and Cyc treatment or either alone relatively lowered cell proliferation rate (P Ͻ 0.05). Besides, the simultaneous ATAD2 silencing and Cyc treatment further suppressed cell proliferation rate than did Cyc treatment alone (P Ͻ 0.05). These findings confirmed that downregulation of ATAD2 could reduce ESCC stem cell proliferation via inhibition of Hedgehog signaling pathway.
ATAD2 silencing enhances ESCC stem cell apoptosis. The effects of ATAD2 and Hedgehog signaling pathway on the apoptosis of ESCC stem cells were evaluated by flow cytometry (Fig. 8A). Moreover, the expression of apoptosis-related genes was assessed by RT-qPCR and Western blot analysis (Fig. 8, B and C). ATAD2 overexpression led to a lower apoptosis rate, which was further reflected by increased expression of Bcl-2 and MMP2 yet decreased Bax expression (P Ͻ 0.05). On the contrary, simultaneous ATAD2 silencing and Cyc treatment or either alone resulted in a higher apoptosis rate, which was further reflected by decreased expression of Bcl-2 and MMP2 yet increased Bax expression (P Ͻ 0.05). In addition, compared with the Cyc treatment alone, simultaneous ATAD2 silencing and Cyc treatment caused a higher apoptosis rate in addition to diminished expression of Bcl-2 and MMP2 yet elevated Bax expression (P Ͻ 0.05). These results showed that ATAD2 silencing could increase apoptosis rate. ATAD2 silencing impedes ESCC stem cell invasion and migration. We applied scratch test and Transwell assay to elucidate the effects of ATAD2 on the migration (Fig. 9, A and  B) and invasion (Fig. 9, C and D) of ESCC stem cells, finding that overexpression of ATAD2 exhibited promotive effects on invasion and migration abilities of ESCC stem cells (P Ͻ 0.05). The invasion and migration abilities of ESCC stem cells were weakened by simultaneous ATAD2 silencing and Cyc treatment or either alone (P Ͻ 0.05). Compared with the Cyc treatment alone, simultaneous ATAD2 silencing and Cyc treatment contributed to further diminished abilities of invasion and migration (P Ͻ 0.05). These findings suggested that silencing of ATAD2 could suppress ESCC stem cell invasion and migration.

ATAD2 silencing causes decreased colony formation ability of ESCC stem cells.
Colony formation assays clarified that the number of colonies formed was increased by overexpression of ATAD2, but reduced by simultaneous ATAD2 silencing and Cyc treatment or either alone (P Ͻ 0.05). The simultaneous ATAD2 silencing and Cyc treatment resulted in a greater decline in the ability of colony formation than Cyc treatment alone (Fig. 10, A  and B). These findings verified that decrease of ATAD2 could suppress the colony formation of ESCC stem cells.
ATAD2 silencing reduce the tumorigenic ability of ESCC stem cells in vivo. The xenograft tumor assay was performed in nude mice to demonstrate the effect of ATAD2 and the Hedgehog signaling pathway on tumorigenic abilities of ESCC stem cells in vivo. The tumor sizes of mice did not differ remarkably among groups on the 4th and 8th day (P Ͼ 0.05). However, on the 12th day, compared with the blank and NC groups, the ATAD2 group exhibited increased tumor volume and weight, while the si-ATAD2, Cyc and si-ATAD2 ϩ Cyc groups presented with reductions in tumor volume and weight (P Ͻ 0.05). The tumor volume and weight in the si-ATAD2 ϩ Cyc group were further reduced versus the Cyc group (P Ͻ 0.05; Fig. 11, A-C). These results are evidence that ATAD2 silencing could suppress the tumorigenic ability of ESCC stem cells.

DISCUSSION
ESCC is a major histological subtype of esophageal cancer, which is lamentably common in China and has poor prognosis (24). Accumulating evidence has demonstrated the overexpression of ATAD2 in many aggressive tumors, such as esopha-geal, gastric, and breast cancers (28). ATAD2 also acts as a contributor to activation of the Hedgehog signaling pathway in human retinoblastoma (19). Furthermore, the Hedgehog signaling pathway is involved in the tumor-promotive role of oncogene MSI2 in ESCC (18). In this expanding research field, much remains unknown regarding the role of ATAD2 in ESCC. The aim of the present study was to investigate the potential roles of ATAD2 interacting with the Hedgehog signaling pathway in ESCC progression. Collectively, results show that ATAD2 silencing could potentially prevent the proliferation, invasion, and migration of ESCC stem cells through inhibition of the Hedgehog signaling pathway. We initially found that ATAD2 was highly expressed in ESCC resected tissues and cell lines. In addition, high expression of ATAD2 was correlated with advanced TNM stage and histological grading of ESCC patients. Highly expressed ATAD2 is frequently identified in various human tumors (7). For example, ATAD2 is observed to be upregulated in gastric cancer, and its high expression correlates with cancer progression, as shown by advanced clinical stage, and lymph node metastasis (LNM), as well as distant metastasis (36). Consistently, immunohistochemistry results revealed an upregulation of ATAD2 in colorectal cancer tissues, which is also associated with advanced TNM stage, LNM and recurrence (13).
Moreover, the downregulation of ATAD2 was found to induce the blockage of Hedgehog signaling pathway. A previous report showed that the Hedgehog signaling pathway was frequently activated in human cancers such as esophageal cancer (35). Identification of inactivation of Hedgehog signaling pathway is crucial to improve cancer diagnosis and treatment (33). In particular, activation of Hedgehog signaling pathway accelerates the progression of ESCC (38). PTCH1 and Gli1 are the essential components for activation of the Hedgehog signaling pathway (8). Wu et al. (29) have identified that ATAD2 could cooperate with c-Myc to control the expression of SMO and Gli1, whereby the Hedgehog signaling pathway is activated and also an active feedback of the Hedgehog signaling pathway is induced in human hepatocellular carcinoma. Similarly, downregulated ATAD2 expression induced by long non-coding RNA PCAT-14 is accompanied by decreased expression of PTCH1, SMO, and Gli2 in hepatocellular carcinoma cells (27). Moreover, we observed that the decline of ATAD2 expression was more pronounced when treated with Cyc and si-ATAD2 together relative to treatment with Cyc or si-ATAD2 alone. The possible mechanism may be explained that inhibited cell proliferation and promoted apoptosis by combined treatment with Cyc and si-ATAD2 subsequently slowed down cell metabolism, thus in turn contributing to reduced expression of ATAD2, a member of the ATP family.
In addition to these results, ATAD2 silencing was observed in this study to restrain CSC proliferation, migration, and invasion, while potentiating apoptosis in ESCC by disrupting the Hedgehog signaling pathway. A previous investigation by Morozumi et al. (21). has shown that ATAD2 is crucial in differentiating embryonic stem cells and also frequently shows abnormal activity in many cancers via controlling cells proliferation and differentiation. Likewise, ATAD2 has proven to promote hepatocellular carcinoma cell proliferation, migration, and invasion abilities (28). Silencing of ATAD2 inhibits migration and invasion of colorectal cancer cells (12). Furthermore, a previous study conducted by Caron et al. (4) has disclosed that ATAD2 is activated in upstream and basic cellular processes to enhance oncogenesis in various cell types. For instance, ATAD2 influences cell cycle and tumor growth in nude mice bearing hepatocellular carcinoma (29). siRNAmediated silencing of ATAD2 reduces retinoblastoma cell viability and invasive and migratory potentials, while also accelerating apoptosis via impairment of the Hedgehog signaling pathway (31). The blockade of Hedgehog signaling pathway has also been reported not only to impede hepatocellular carcinoma proliferation and invasiveness, but also to prevent cell growth (6). The Hedgehog signaling pathway is critically involved in cell proliferation and differentiation during embryogenesis, and its activation promotes CSC growth (23). Inhibition of the Hedgehog signaling pathway by SMO can effectively decrease cell proliferation and induce apoptosis in pancreatic cancer tissues (10). Our present findings were partially consistent with previous studies revealing the oncogene ATAD2 and the oncogenic Hedgehog signaling pathway. Finally, in vivo experiments validated that downregulation of ATAD2 could suppress carcinogenic ability of ESCC stem cells by suppressing the Hedgehog signaling pathway.
In conclusion, our study provides novel insights into the mechanisms underlying the regulation of biological functions mediated by ATAD2 in CSCs. We propose that ATAD2 silencing restrains CSC proliferation, invasion, and migration by blocking the activation of the Hedgehog signaling pathway. Moreover, the future work will be aimed at development of ATAD2 antibodies or inhibitors contributing to the treatment of ESCC.