DNA-Methylation-Mediated lncRNA HOXB-AS4 Promotes Gastric Cancer Progression Through Regulating miR-130a-5p/PKP4

Background(cid:0)Gastric cancer (GC) is one of the most common cancer in the world, possessing the second leading cause of cancer-related mortality. Long noncoding RNAs (lncRNAs) have been shown to play important roles in tumorigenesis. However, the effect of lncRNA HOXB-AS4 in GC progression and the underlying mechanisms remain unknown. Methods(cid:0)Firstly, the expression of lncRNA HOXB-AS4 in gastric cancer tissues and cancer cells was investigated according to GEPIA database and Real time uorescence quantitative PCR(qRT-PCR). Then, MTT, clone formation, Transwell and Western blot were used to study the effects of overexpression or down-regulation of HOXB-AS4 on the proliferation, invasion and epithelial mesenchymal transformation of cancer cells. We further studied the molecular mechanism of HOXB-AS4 by uorescence in situ hybridization, bioinformatics analysis, luciferase reporting, methylation specic PCR (MSP) and chromatin immunoprecipitation (chip). Results:In the study, the GEPIA database and quantitative Real-Time PCR (qRT-PCR) assay showed that HOXB-AS4 was upregulated in GC tissues and cells. Then, MTT, clone formation, transwell, and western blot assays suggested that overexpression of HOXB-AS4 increased cell proliferation, migration, and invasion, and regulated epithelial-mesenchymal transition (EMT) markers expression, while knockdown of HOXB-AS4 showed the opposite effect. Fluorescence in situ hybridization (FISH) assay found that HOXB-AS4 localized in the cytoplasm of the GSE-1 and AGS cells. Further mechanism experiments, including bioinformatics, luciferase reporter, qRT-PCR, and western blot assays showed that HOXB-AS4 sponged to miR-130a-5p to regulate the PKP4 expression. Knockdown of miR-130a-5p obliterated the effect of HOXB-AS4, which was further abolished by knockdown of PKP4 in vitro and in vivo. Methylation-specic PCR (MSP) and chromatin immunoprecipitation (CHIP) assay showed that overexpression of HOXB-AS4 in GC was mediated by SP1-dependent DNA methylation. Abnormal upregulation of lncRNA HOXB-AS4 contributed to GC progression, which was mediated by DNA methylation. The study claried that DNA-methylation-mediated HOXB-AS4 played its role through miR-130a-5p/PKP4 axis. Conclusions: Our study provides new insights for the understanding of epigenetic regulation on lncRNA expression in GC, and indicates that HOXB-AS4 could be a biomarker of GC prognosis. Moreover, targeting HOXB-AS4 /miR-130a-5p/PKP4 axis might be a promising strategy to treat GC.

treatments [4]. The pathogenesis and molecular mechanism of GC are complicated, which resulted from the dysregulation of various signaling pathways [5]. So, it's of vital importance to clarify the molecular mechanism of GC progression and nd new markers and targets for GC diagnosis and treatment.
Long noncoding RNAs (LncRNAs) are a group of non-coding RNAs with the length longer than 200 nucleotides [6]. LncRNA has been proved to be involved in various biological processes, including cell proliferation, differentiation, and apoptosis [7,8]. LncRNA plays its role by regulating protein-coding genes, mRNA processing, and maintenance of genomic integrity [9]. LncRNA functioned as the competitive endogenous RNA (ceRNA) which compete with the mRNA to bind to the microRNA (miRNA), leading to the change of gene expression [10]. Although ceRNA mechanism has been reported to play roles in GC progression [11,12], the speci c molecular mechanism on the occurrence of ceRNA mechanism remains largely unknown.
Homeobox (HOX) genes are a highly conserved subgroup of the superfamily, which contain 39 transcription factors [13]. It is reported that HOX family is associated with the development and progression of numerous cancers, including GC [14]. For instance, the transcription factor HOXD9 was upregulated in GC, and accelerated the GC progression [15]. Moreover, lncRNAs generated in HOX family genes have been proved to be a vital biomarker when evaluating the prognosis of multiple cancers [16,17]. For instance, lncRNA HOXA11-AS served as an oncogene to participate in GC [18,19]. As is known, abnormal DNA methylation is related to many cancers, and this epigenetic disruption may lead to abnormal lncRNA expression in tumor tissues [20]. Evidence has shown that lncRNAs generated in HOX family genes, as HOTAIR, HOTTIP, HOXA11-AS, HOXB-AS4, and HOXC-AS3, were regulated by aberrant DNA methylation in GC [21]. However, the roles and mechanism of lncRNA HOXB-AS4 in GC remain unknown.
In the present study, we designed experiments to investigate the role of lncRNA HOXB-AS4 in the GC progression and the effect of epigenetic regulation on its abnormal expression (Additional e 1: Fig.S1).
Our results showed that HOXB-AS4 is epigenetically activated by DNA methylation at the CpG islands within its promoter region. HOXB-AS4 regulates the cell proliferation, migration, invasion, and EMT markers expression by competitively binding to miR-130a-5p, leading to the upregulation of plakophilin 4 (PKP4), which promotes the GC progression.

Materials And Methods
Patients and specimens A total of 16 pairs of GC tissues and the pair-matched non-tumoral gastric tissues were acquired from the rst people's hospital of suqian. The written consent agreement was signed and approved by the Ethics Committee of the A liated Suqian rst People's Hospital of Nanjing Medical University. All patients involved in the present study didn't receive chemotherapy or radiotherapy before the surgery. 16 tissues were stored at -80℃ immediately after the surgery.

Western blot
Western blot assay was used to assess epithelial-mesenchymal transition (EMT)-related makers Vimentin, N-Cadherin, and E-Cadherin and PKP4 protein expression. Total protein from tumor tissues or cells was extracted by using RIPA lysis buffer plus PMSF (Beyotime, China). BCA assay kit (Santa Cruz, USA) was used to detect total protein concentration. Prepared protein samples were separated by using SDS-PAGE electrophoresis, and transferred into PVDF membranes and incubated with prepared antibodies. Finally, enhanced chemiluminescence (ECL, ThermoFisher, USA) was used to visualize the membrane. The protein band analysis was conducted with ImageJ software. Antibodies against Vimentin, N-Cadherin, E-Cadherin, and GAPDH were purchased from CST (MA, USA). Antibody against PKP4 was purchased from Abcam (Cambridge, USA).

Cell viability assay
The viability of GSE-1 and AGS cells was assessed using MTT. Brie y, cells were plated in the 96-well microplates at a density of 1 × 10 4 cells per well. After transfection with indicated time, the cells were incubated with MTT (0.5 mg/ml) at 37 °C for 3 h. Then the medium was removed, and 100 mM DMSO solution was added to dissolve the formazan crystals. The absorbance at 570 nm wavelength was detected using a microplate reader (Molecular Devices, Sunnyvale, USA).

Clone formation assay
The GSE-1 or AGS cells were digested and plated in a 6 cm culture dish at a density of 5× 10 3 /well after indicated treatments. The cells were subjected to the normal culture condition at 37℃, 5% CO 2 for 14 days. Then the cells in the dish were washed with PBS, and xed with 3 mL methanol for 10 minutes. The cells were subjected to Giemsa dyeing for another 15 minutes. Then the cells were observed under a light microscope (Olympus, Japan), and the number of colonies was counted.

Cell migration and invasion assay
Cell migration and invasion were determined by transwell assay. GSE-1 and AGS cells were harvested and seeded to the upper transwell chamber at a density of 5 × 10 4 cells per well into an 8-mm pore transwell plate (Costar, USA) pre-treated with Matrigel (BD biosciences, USA) or not. Serum-free medium was added onto the upper chamber, and the culture medium containing 20% FBS was added into the lower chamber.
Following 24 h incubation, the unmigrated or uninvaded cells were cleaned using a cotton swab and then xed in 4% paraformaldehyde for 20 min and stained with hematoxylin. Images were taken, and the cell number was calculated.

Methylation-speci c PCR (MSP)
Genomic DNA was extracted from the GC tissues and adjacent normal tissues by using the DNA extraction Kit (Qiagen, Germany) according to the manufacturer's instruction. The puri ed DNA was then exposed to bisul te treatment by using the EpiTect Bisul te Kit (Qiagen, Germany). Then the MSP of bisul te-transformed DNA was performed with a nested, two-stage PCR method followed with the detection of the PCR products by using agarose gel electrophoresis.

Chromatin immunoprecipitation (ChIP)
ChIP assays were performed with an EZ-ChIP Kit (Millipore, USA) according to the manufacturer's instructions. Brie y, GES-1 and AGS cells were cross-linked with 1% formaldehyde for 10 min followed with glycine treatment. Cell lysates were then sonicated to generate chromatin fragments and then immunoprecipitated with SP1 antibody (CST, USA). IgG antibody (CST, USA) was used as the negative control. The DNA fragment was ampli ed by PCR. The HOXB-AS4 primer sequences are as follows: F: CTGAGTTTTCAGCCCTCCTG, R: AAGCTCCAATGAAGGGGTCT. The product was then analyzed by using agarose gel electrophoresis.

Fluorescence in situ hybridization (FISH)
To detect HOXB-AS4 expression, GSE-1 and AGS cells were xed in 4% formaldehyde for 15 min at room temperature and then permeabilized with 70% ethanol. After rehydrated for 5 min at room temperature, the cells were incubated by using biotin-labeled HOXB-AS4 probe (Genepharma, Shanghai, China) at 37°C for 8 h. Then the Alexa Fluor 647-conjugated secondary antibody (Abcam, USA) was used to detect the biotin-labeled HOXB-AS4.

Luciferase reporter assay
The wild-type or mutant 3'UTR of PKP4 and the full length of HOXB-AS4 were ampli ed and cloned into pGL3-basic vector (Promega, WI, USA) separately. The binding site of 3'UTR of PKP4 was mutated from UAU UAU GUU UUU UAA AAU GUG AG to TTT ATA GTA TTA AAT TTT GAG TG for miR-130a-5p, and the potential binding sites of HOXB-AS4 was mutated from AAAAGAGA to TTTTGTG for miR-130a-5p. Then, AGS cells were plated on a 24-well plate and co-transfected with wild-type or mutant luciferase plasmids and miR-23c mimic, miR-23c inhibitor or control miRNA. A Dual-Luciferase Reporter Assay System (Promega, WI, USA) was used to measure the luciferase activity. The relative luciferase activity of each sample was normalized to Renilla luciferase activity.
Tumor xenograft model 1 × 10 7 AGS cells were suspended in 200 μl PBS and subcutaneously injected into right ank of 4-6week-old BALB/c nu/nu male mice (Charles River Lab, Beijing, China). The mice were recorded the tumor volume (volume = (length × width 2 )/2) every 3 days. Knockdown of miR-130a-5p was performed by using miR-130a-5p antagomir (Ribo Bio, Guangzhou, China) which was administrated by tail injection at the dosage of 80 mg/kg. Knockdown of HOXB-AS4 or PKP4 was performed by using the related knockdown lentivirus (GenePharma, Shanghai, China) by intra tumoral injection of 50 μL virus (4×10 7 IU/mL) after the tumor cells injection. After 15 days, the mice were sacri ced. The tumor tissues were subjected to western blot analysis, hematoxylin-eosin (H&E) staining and immunohistochemical (IHC) analysis. The animal experiments were approved by the Animal Care and Use Committee of the rst people's hospital of suqian.

H&E and IHC analysis
For H&E staining, the prepared tumor tissue slices were dewaxed and hydrated. After washed by water, tumor tissue slices were stained in hematoxylin solution for 5 min. Next, after treated by 1% hydrochloric alcohol for 15s, the slices were washed with water. And then the slices were stained by eosin solution for 1 min. Finally, tissue slices were dehydrated, transparentized and sealed by neutral gum, observed under an optical microscope (Olympus, Tokyo, Japan). For IHC staining, the para n-embedded tumor sections were dewaxed and treated with 3% H 2 O 2 to deactivate endogenous peroxidase. After blocking nonspeci c antigen binding with 5% BSA at 37 °C for 1 h, the sections were incubated with a speci c primary antibody against PKP4 (1:100 dilution, Abcam, USA) at 4 °C overnight. After incubating with the corresponding secondary antibodies at 37 °C for 1 h, the sections were stained with diaminobenzidine and counterstained with hematoxylin. Representative images were taken using a light microscope (Olympus, Tokyo, Japan).

Kaplan-meier analysis
The overall survival data were derived from the online database Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/). Of the 110 GC samples, 108 samples showed high HOXB-AS4 expression, whereas 82 samples showed low HOXB-AS4 expression. The survival rates were analyzed by Kaplan-Meier curves. Then, the log-rank test was performed to evaluate the mean of the two groups.

Bioinformatics
The online database Starbase (http://starbase.sysu.edu.cn/) was used to predict possible miRNAs that may be sponged by HOXB-AS4. In addition, the online database miRCancer (http://mircancer.ecu.edu/) was used to predict the abnormal expression of miRNAs in gastric cancer. Then, the miRNAs that may be sponged by HOXB-AS4 in gastric cancer were screened through the intersection of the Starbase and miRCancer. The online databases miRDB (http://mirdb.org/) and Targetscan (http://www.targetscan.org/ were used to screen the potential targets of miR-130a-5p.

Statistical analysis
All data were shown as mean ± s.e.m. and experiments were carried out at least three times. GraphPad 6.0 version was adopted for statistical analyses. Student's t-test was performed to evaluate the mean of the two groups and one-way ANOVA followed Tukey's poc host was used to analyze signi cant differences among multiple groups. P < 0.05 was regarded as statistically signi cant.

Results
LncRNA HOXB-AS4 is up-regulated in GC tissues and GC cells To determine whether lncRNA HOXB-AS4 plays a role in the pathogenesis of GC, we rst investigated the expression of lncRNA HOXB-AS4 in the tumor tissues. According to the GEPIA database (http://gepia.cancer-pku.cn/), we found that lncRNA HOXB-AS4 was signi cantly up-regulated in various types of tumors, especially in GC (Fig. 1A). The HOXB-AS4 expression in the GC tissues was further veri ed by using qRT-PCR. The results showed that HOXB-AS4 was signi cantly up-regulated in the 16 pairs of GC tissues compared to the pair-matched non-tumoral gastric tissues (Fig. 1B). Moreover, Kaplan-meier analysis showed that HOXB-AS4 expression was associated with the prognosis of 110 of GC patients (Fig. 1C). 108 of GC patients with higher HOXB-AS4 expression showed a poorer prognosis than 82 of GC patients with lower HOXB-AS4 expression (Fig. 1C), indicating that HOXB-AS4 expression may contribute to the GC progression. We further investigated the expression level of HOXB-AS4 in various GC cells and in normal human gastric epithelial cell line GES-1. As shown in Fig. 1D, HOXB-AS4 exhibited higher expression level in the cancer cells compared to the normal gastric cells, among which AGS cells showed the highest expression.
As activation of EMT process provides tumor cells with invasive and migratory ability [22], we then investigated the effect of HOXB-AS4 on EMT markers expression. In GES-1 cells, overexpression of HOXB-AS4 reduced Vimentin expression and increased the N-Cadherin and E-Cadherin expression (Fig. 2E). In AGS cells, knockdown of HOXB-AS4 increased the Vimentin expression and inhibited the N-Cadherin and E-Cadherin expression (Fig. 2E).
HOXB-AS4 acts as a sponge for miR-130a-5p We then explored the mechanism of the regulatory effect of HOXB-AS4. We examined the intracellular localization of HOXB-AS4. Results of FISH showed that HOXB-AS4 localized in the cytoplasm of the GSE-1 and AGS cells (Fig. 3A). It is reported that cytoplasmic lncRNA exerts its regulatory function through several mechanisms, including mRNA turnover, translation, protein stability, sponging of cytosolic factors, and modulation of signaling pathways [23]. It's worth noting that ceRNA mechanism is one of the major functions of cytoplasmic lncRNA. Therefore, we then hypothesized that HOXB-AS4 may play its role through the ceRNA mechanism. By using the Starbase and miRCancer database, we identi ed two miRNAs that may be sponged by HOXB-AS4 (Fig. 3B). qRT-PCR showed that overexpression of HOXB-AS4 in AGS led to lower miR-130a-5p level in cells, and miR-130a-5p showed an increased level in HOXB-AS4knockdown cells (Fig. 3C). We then predicted the potential binding sites between HOXB-AS4 and miR-130a-5p (Fig. 3D). Luciferase reporter showed that miR-130a-5p signi cantly decreased the luciferase activity of the cells transfected with WT rather than the mutant HOXB-AS4, indicating that miR-130a-5p speci cally bound to the HOXB-AS4 (Fig. 3E). Moreover, miR-130a-5p showed signi cant downregulation in 16 pairs of GC tissues compared to the pair-matched non-tumoral gastric tissues (Fig. 3F), which negatively correlated with the HOXB-AS4 expression (Fig. 3G).
PKP4 is a direct target of miR-130a-5p To determine the targets of miR-130a-5p which might contribute to the progression of GC, we screened the potential targets of miR-130a-5p by using TargetScan and miRDB, which identi ed the PKP4 may be the potential targets of miR-130a-5p (Fig. 4A). We then predicted the potential binding sites (Fig. 4B), and performed luciferase reporter assays to verify whether PKP4 was a target of miR-130a-5p. miR-130a-5p signi cantly decreased the luciferase activity of the cells transfected with WT rather than the mutant PKP4, indicating that miR-130a-5p speci cally bound to the 3'UTR of PKP4 (Fig. 4C). Furthermore, western blot showed that miR-130a-5p overexpression markedly reduced the PKP4 protein expression, and miR-130a-5p knockdown increased the PKP4 expression (Fig. 4D). Then, 16 pairs of GC tissues and the pair-matched non-tumoral gastric tissues were collected for western blot assay. PKP4 showed an obvious upregulation in the GC tissues compared to the pair-matched non-tumoral gastric tissues (Fig. 4E). PKP4 showed signi cant positive correlation with HOXB-AS4 expression, and negative correlation with miR-130a-5p expression (Fig. 4F). All these data suggested that HOXB-AS4/miR-130a-5p/PKP4 may be involved in the GC progression.

Knockdown of HOXB-AS4 inhibits GC progression and modulates EMT markers in vivo
We further veri ed the effect and underlying mechanism of HOXB-AS4 in vivo. Mice were divided into four groups, with 10 mice in each group. Mice in HOXB-AS4 knockdown group showed an obvious smaller tumor size compared to the control group, while treatment of miR-130a-5p antagomir inhibited the tumorsuppressive effect of HOXB-AS4 knockdown (Fig. 6A, B). Knockdown of PKP4 further inhibited the tumor growth promoted by miR-130a-5p knockdown, showing smaller tumor size compared to the knockdown of HOXB-AS4 and miR-130-5p group (Fig. 6A, B). Western blot showed that knockdown of HOXB-AS4 increased E-Cadherin expression and decreased N-Cadherin and Vimentin expression (Fig. 6C). Knockdown of miR-130a-5p inhibited the effect of HOXB-AS4 knockdown on, and knockdown of PKP4 knockdown further repressed the effect of miR-130a-5p (Fig. 6C). IHC staining revealed that the PKP4 expression showed the consistency with the tumor progression (Fig. 6D). HOXB-AS4 knockdown induced lower PKP4 expression in the tumors, while knockdown of miR-130-5p recovered the PKP4 expression inhibited by HOXB-AS4 knockdown (Fig. 6D). All these data suggested that inhibition of HOXB-AS4/miR-130a-5p/PKP4 axis restrained the tumor progression in vivo.
DNA methylation and SP1 participate in the regulation of the expression of HOXB-AS4 We then explored the mechanism underlying the high expression of HOXB-AS4 in the GC tissues and cells. By using the UCSC Genome Bioinformatics Site(http://genome.ucsc.edu/), we identi ed CpG islands together with high enrichment and overlapping H3K427AC peaks within the promoter region of HOXB-AS4 (Fig. 7A), indicating a potential relationship between DNA methylation and HOXB-AS4 expression. We then performed MSP analysis to verify whether DNA methylation was involved in the upregulation of HOXB-AS4 in GC. As shown in Fig. 7B, DNA methylation levels were signi cantly decreased in the promoter region of HOXB-AS4 in the GC tissues. Accordingly, in the AGS cells, DNA methylation levels in the promoter region of HOXB-AS4 were lower than that in the GES-1 cells (Fig. 7C). We then treated the GES-1 cells or AGS cells with the DNA demethylation agent azacytidine for different time. We found that HOXB-AS4 expression increased with time in GSE-1 and AGS cells, indicating that the inhibition of DNA methylation promoted the HOXB-AS4 expression (Fig. 7D). We further used PROMO (http://alggen.lsi.upc.es/cgi-bin/promo_v3) to predict the potential transcription factors of HOXB-AS4, which showed that SP1 may be a transcription factor of HOXB-AS4. Results of CHIP revealed the enrichment of SP1 at the promoter of HOXB-AS4 (Fig. 7E). We then changed the SP1 expression in GES-1 and AGS cells, and the overexpression or knockdown e ciency was veri ed by western blot (Fig. 7G). Overexpression of SP1 enhanced HOXB-AS4 expression in GES-1 and AGS cells, and SP1 knockdown restrained HOXB-AS4 expression (Fig. 7F). All these data suggested that DNA methylation and transcription factor SP1 were involved in the regulation of SP1 expression.

Discussion
Increasing evidence showed that lncRNAs play important roles in the development of GC [24,25]. In the study, we revealed a novel function of HOXB-AS4 in GC progression. We reported that DNA-methylationmediated HOXB-AS4 promoted GC proliferation, migration, and invasion, and modulated EMT markers expression through regulating miR-130a-5p/PKP4 (Fig. 8).
HOXB-AS4 was HOXB cluster antisense RNA 4, whose effect has barely been reported. In the present study, by using the data of GEPIA database, we found that HOXB-AS4 was upregulated in various cancer, including GC. We further veri ed HOXB-AS4 overexpression in GC tissues and cells, and found its upregulation correlated with the poor prognosis of GC patients, indicating it may play potential roles in GC progression. Further in-vitro experiment showed that HOXB-AS4 overexpression promoted the AGS and GSE-1 cells proliferation, migration and invasion.
EMT encompasses complex and dynamic changes in phenotypes of cellular organization from epithelial characteristics to mesenchymal traits [26]. The epithelial and mesenchymal cell characteristics include EMT-inducing signals, EMT-Transcription factors and EMT markers [26]. One of the speci c changes is the decrease of epithelial marker proteins, such as E-cadherin, and the up-regulation of mesenchymal marker proteins, such as N-cadherin and Vimentin [26]. It is reported that activation of EMT process confers tumor cells with invasive and migratory ability, and supports tumor metastasis [27]. Various studies have reported that lncRNA regulates the EMT process in cancer development [28,29]. For example, lncRNA SNHG16 promotes colorectal cancer cell proliferation, migration, and modulates EMT markers expression through miR-124-3p/MCP-1 [30]. LncRNA LINC00525 regulates the proliferation and epithelial to mesenchymal transition of human glioma cells by sponging miR-338-3p [31]. In the present study, overexpression of HOXB-AS4 increased GSE-1 cell proliferation, migration and invasion, and modulated EMT markers expression. In AGS cells, knockdown of HOXB-AS4 inhibited the cancer cell proliferation, migration and invasion, accompanied with the modulation of EMT markers expression, indicating that HOXB-AS4 may function tumor-promoting effect through modulating EMT process.
LncRNA could act as miRNAs sponge to regulate gene expression, which were de ned as "ceRNA" mechanism [32,33]. CeRNA mechanism is the main regulatory model of cytoplasm lncRNA [34,35]. In the present study, we identi ed the HOXB-AS4 as a cytoplasm lncRNA by using FISH, indicating its regulatory role in GC may rely on the ceRNA mechanism. Screen of the Starbase and miRCancer database showed that miR-130a-5p and miR-140a-5p may interactwith HOXB-AS4. Luciferase reporter assay further showed that HOXB-AS4 was a miR-130a-5p sponge. Further experiments showed that miR-130a-5p targeted the 3'UTR of PKP4. MiR-130a-5p was reported to inhibit tumor invasion and metastatic potential in non-small cell lung cancer [36]. MiR-130a-5p enhanced the sensitivity of cisplatin-resistant gastric cancer cells to chemotherapy [37]. PKP4 was also called "p0071", which were reported to interact with Ecadherin to promote the invasion and metastasis of cancer cells [38]. In the present in-vitro and in-vivo experiments, we veri ed that knockdown of miR-130a-5p inhibited the anti-tumor effect of HOXB-AS4 knockdown, showing an increase in cell proliferation and modulation of EMT markers expression. Knockdown of PKP4 further abolished the effect of miR-130a-5p knockdown on recovering the cell proliferation and EMT markers expression. All these results indicated that HOXB-AS4/miR-130a-5p/PKP4 axis promoted GC development.
We further investigate the mechanism of HOXB-AS4 upregulation in GC. LncRNA expression is regulated by multiple factors, including chromatin state, the utilization of transcription factors, and some posttranscriptional regulation [39]. DNA methylation is a stable and heritable epigenetic mark which could in uence the chromatin state to change the transcriptional state of genes [40]. Many lncRNAs have been reported to be regulated by DNA methylation [41]. Although DNA methylation was studied much as a epigenetic modi cations in GC, the DNA methylation-mediated dysregulation of lncRNAs was barely reported [42][43][44]. In the present study, we identi ed abnormally low methylation levels of the promoter region of HOXB-AS4, which resulted in enhanced lncRNA expression. SP1 is a transcription factor which has been reported to mediate various lncRNA aberrant expression [45,46]. SP1-dependent DNA methylation was an important pattern for gene overexpression [47]. Consistently, we found that SP1 was involved in the DNA-methylation-mediated upregulation of HOXB-AS4, indicating that SP1-dependent DNA methylation was the molecular mechanism of HOXB-AS4 aberrant upregulation in GC. To data, there are some limitations in the study. For instance, the number of tissue samples seems inadequate. Only 16 pairs of GC tissues and the pair-matched non tumoral gastric tissues were acquired to detect HOXB-AS4, miR-130a-5p, and PKP4 expression. Furthermore, the involvement of EMT in any process cannot rely solely on a few salient molecular markers, such as E-cadherin and vimentin. As is known, cytoplasmic lncRNA exerts its regulatory function through several mechanisms, including mRNA turnover, translation, protein stability, sponging of cytosolic factors, and modulation of signaling pathways [23]. In the current study, we could not excluded the other mode of action, and just clari ed the role of HOXB-AS4 /miR-130a-5p/PKP4 axis on GC progression. We will further explore the mechanism of HOXB-AS4 in the future study.
In conclusion, we identi ed that abnormal upregulation of lncRNA HOXB-AS4 contributed to GC progression. HOXB-AS4 was found to be epigenetically activated by DNA methylation at the CpG islands within its promoter region. HOXB-AS4 regulated the cell proliferation, migration, invasion, and EMT markers expression by competitively binding to miR-130a-5p, leading to the upregulation of PKP4 in GC progression. Our study provides new insights for the understanding of epigenetic regulation on lncRNA expression in GC, and indicates that HOXB-AS4 could be a biomarker of GC prognosis.

Availability of data and materials
The data used to support the ndings of this study are available from the corresponding author upon request.

Ethics approval and consent to participate
The written consent agreement was signed and approved by the Ethics Committee of The A liated Suqian rst People's Hospital of Nanjing Medical University.

Consent for publication
The authors declared that there are no con icts of interest.

Consent for publication
All the listed authors have participated in the study, and have seen and approved the submitted manuscript.