Investigations into the effects and related upstream/downstream regulatory mechanisms of long non-coding RNA WT1-AS in the maintenance and development of gastric cancer stem cells in vitro and in vivo

Cancer stem cells (CSCs) are proposed to be responsible for almost all malignant phenotypes (e.g. heterogeneity, uncontrolled growth, metastasis, recurrence, chemoresistance) of tumors. Long non-coding RNA WT1 antisense RNA (WT1-AS) has been found to be involved in the regulation of lung cancer cell stemness. However, the roles and molecular mechanisms of WT1-AS in the maintenance and development of gastric cancer stem cells (GSCs) have not been investigated. mRNA and protein expression was measured by RT-qPCR and western blot. CCK8 and Soft agar colony formation assays were performed to assess cell viability and colony clone formation ability. Cell cycle and apoptosis were determined by ow cytometry analysis. Cell transwell and wound healing analyses were carried out to assess cell migration ability. In vitro angiogenesis and 3D spheroid cultures assays were also performed. Moreover, in vitro experiments were carried out to explore the function of WT1-AS on tumor growth, metastasis and cell stemness. The upstream transcription factors or downstream genes of WT1-AS were screened through Bioinformatics, dual-luciferase assays and RNA-sequencing (RNA-seq) technology. cell EMT, increased CD44 + stem-like cell subpopulation, induced HUVEC angiogenesis, inhibited GSC cell apoptosis, and enhanced the resistance of GSCs to 5-FU in vitro. Moreover, 3D cell culture experiments demonstrated that WT1-AS loss or WT1 increase facilitated the formation of in-vitro 3D GSC aggregates. Moreover, our data presented that WT1-AS weakened the malignant behaviors and properties of GSCs by down-regulating WT in vitro. In addition, our in vivo experiments demonstrated that WT1-AS inhibited the growth and metastasis of GSCs-derived xenograft tumors and reduced CD44 + stem-like cell subpopulation in xenografts. These data suggested that WT1-AS functioned as a negative regulator in the maintenance and reprogramming of GSC stemness properties. MYB proto-oncogene E2F transcription factor 1 (E2F1), hypoxia inducible factor 1 alpha subunit (HIF1A), peroxisome proliferator activated receptor alpha (PPARA), upstream transcription factor 1 (USF1), ETS proto-oncogene 2 (ETS2), ETS transcription factor ELK1 (ELK1), androgen receptor (AR), estrogen receptor 1 (ESR1), X-box binding protein 1 (VDR), X-box binding protein 1 (XBP1), phosphoserine phosphatase (PSPH), phosphoglycerate dehydrogenase (PHGDH), glutathione S-transferase omega 2 (GSTO2), calcium channel α2δ1 subunit (CACNA2D1), monoamine oxidase B (MAOB), aminopeptidase-N (ANPEP), Integrin alpha 2 (ITGA2), cathepsin S (CTSS), protein kinase C alpha (PRKCA), 5'-Nucleotidase Domain Containing 2 (NT5DC2)


Introduction
Gastric cancer (GC), a heterogeneous disorder with multiple phenotypes, is a serious global healthcare problem with the fth highest incidence rate and the third highest mortality in all cancers worldwide [1,2].
It was estimated that more than 1,000,000 new GC cases and about 783,000 GC-related deaths occurred in 2018 globally, accounting for approximately 5.7% of all new cancer cases and 8.2% of all cancer deaths [1,2]. Despite the improvement in the diagnosis and screening strategies, many GC cases are still diagnosed at the advanced stages and patients with advanced GC have a poor prognosis [2][3][4]. It is widely accepted that 5-Fluorouracil (5-FU)-based combination chemotherapy is one of the standard therapeutic strategies for GC [5,6]. However, primary or acquired drug resistance strikingly limits the therapeutic e ciencies of chemotherapeutic drugs [7]. Cancer stem cells (CSCs) are a group of cancer cell subpopulation with the features of stem cells such as differentiation and unlimited self-renewal abilities [8]. Recently, CSC hypothesis, which proposes that CSCs are mainly responsible for the malignant phenotypes (e.g. heterogeneity, uncontrolled growth, metastasis, recurrence, chemoresistance) of tumors, has attracted much attention from researchers [8].
Long non-coding RNAs (lncRNAs), a class of long transcripts (over 200 nucleotides in size) with little or no protein-coding potential, have emerged as crucial regulators of CSC behaviors and properties such as proliferation, epithelial-mesenchymal transition (EMT), metastasis, differentiation [9,10]. LncRNA WT1 antisense RNA (WT1-AS), the large antisense transcript of Wilms' tumor 1 (WT1) gene, has been found to be speci cally expressed in different cancers and to be involved in the regulation of tumor growth, metastasis and invasion [11]. Moreover, Du et al. noted that WT1-AS expression was markedly reduced in GC tissues relative to adjacent normal tissues [12]. WT1-AS overexpression suppressed GC cell cycle progression, cell proliferation, migration and invasion in vitro and hindered GC xenograft tumor growth and metastasis in vivo [12]. Additionally, Jiang et al. demonstrated that the enforced expression of WT1-AS negatively regulated the stemness of non-small cell lung cancer cells [13]. However, the in uences of WT1-AS overexpression or loss on the properties and phenotypes of GSCs along with the molecular mechanisms of WT1-AS action have not been characterized.
WT1-AS has been reported to be a regulator of WT1 [14,15]. For instance, Lv et al. demonstrated that WT1-AS up-regulation triggered a notable reduction of WT1 expression level in HepG2 hepatocellular cancer cells [16]. WT1 also has been found to be involved in the regulation of multiple pathophysiologic processes such as cell EMT, growth, differentiation, and apoptosis [17,18]. Moreover, WT1 has been identi ed as an oncogene or a tumor suppressor in different cancers [18,19]. Additionally, previous studies showed that WT1 was highly expressed in primary GC tumor tissues [20], and WT1 silence led to the reduction of cell proliferative ability and increase of cell apoptotic percentage in AZ-521 GC cells [21,22]. Furthermore, Royer-Pokora et al. pointed out that Wilms tumor cells with WT1 mutation developed stem cell-like traits after long-term culture [23], suggesting the vital roles of WT1 in the development of cancer cell stemness.
In this text, we investigated whether WT1-AS could regulate the growth, metastasis and stemness of GSCs by WT1 by in vitro and in vivo experiments. Moreover, upstream transcriptional factors and downstream genes of WT1-AS were further explored in GSCs.

Soft agar colony formation assay
Cell proliferative potential was examined by soft agar colony formation assay. Firstly, the mixture solution (3 ml) containing equal volume of 1.2% agarose and 2× complete medium supplemented with 2×antibiotics and 20%FBS were added into the 6 cm culture dishes to prepare the base agar layer. Next, cells at the logarithmic phase (1000 cells/well) and the equal-volume mixtures (3 ml) of 0.7% agarose and 2× complete medium containing 2×antibiotics and 20%FBS were added into the upper layer. Next, the culture dishes were maintained for 14 days in a 5% CO 2 incubator at 37°C. Finally, the number of colonies was counted under a microscope.
Next, cell cycle distribution patterns were measured using a ow cytometry (BD Biosciences).

Cell apoptotic rate detection
Cell apoptotic percentage was determined using an Annexin V-PE/7-AAD Apoptosis Detection kit (Nan Jing KeyGen Biotech Co., Ltd., Nanjing, China) following the instructions of manufacturer. Brie y, cells were re-suspended in Binding Buffer and then stained with Annexin V-PE and 7-AAD solutions for 15 min at room temperature under a dark environment. Finally, cell apoptotic patterns were analyzed using a ow cytometry (BD Biosciences).
Transwell migration assay Cells (2×10 4 cells/well) suspended in serum-free medium were seeded into the upper chamber of Transwell plate (8µm pore size; Costar Corning, Corning, NY, USA) and complete medium containing 10% FBS was added into the low chamber of plate. The plates were cultured in a 5%CO 2 incubator for 24 h at 37 ˚C. Next, cells on the top surfaces of the Transwell lters were removed. Cells on the low surfaces were xed with 4% paraformaldehyde for 30 min and stained with 1% crystal violet solution (Sigma-Aldrich, lnc., St Louis, MO, USA) for 10 min. Finally, cells were imaged and counted using a microscope.
Wound healing assay Cells were seeded into 24-well plates and cultured overnight at 37°C. When cells grew to the full con uency, a straight scratch wound was created using a sterile 10μl pipette tip. After washed three times with PBS solution, cells were maintained in serum-free medium at 37°C in a 5% CO 2 incubator. The wound images were captured at 0 h and 24 h post scratching using a microscope (CKX31, Olympus, Tokyo, Japan). Finally, the migratory distance of cells under the same led were measured and statistically analyzed.
In vitro 3D culture Cells at the logarithmic growth phase were collected and re-suspended in the complete medium containing 2.5% Matrigel (volume ratio, Costar Corning) at a density of 10 4 cells/ml. Next, cells (200 µl) were added into 96-well plates pre-coated with agarose. After low-speed centrifugation (1000 ×g, 10 min, 4˚C), cells were cultured at 37°C in a 5% CO 2 incubator. Medium was replaced with fresh medium every other day. Next, these 3D multiple cell tumor spheres were imaged using a microscope after 7 days.

Promoter luciferase reporter assay
The luciferase reporter containing WT1-AS promoter sequences (-1K~+200) were constructed and corresponding WT1-AS reporter lentiviruses were generated by Novobio Biotechnology Co., ltd. GSC cells stably transduced with XBP1(+), XBP1(-), NC lentiviruses were also infected with the WT1-AS reporter lentiviruses, followed by the detection of luciferase activities. For subcutaneous xenograft experiments, GSCs (10 6 cells) stably transduced with WT1-AS(+), WT1-AS(-), or control (NC) lentiviruses were injected into the subcutaneous tissues of mice in corresponding groups, respectively. Tissue volume was measuredt 4 weeks after injection and calculated with the formula:
For peritoneal xenograft experiments, GSCs 10 6 cells) transduced with WT1-AS(+), WT1-AS(-), or control (NC) lentiviruses were intraperitoneally injected into the corresponding mice. The volumes of ascetic uid were measured. Tumors were resected and weighed at weeks upon injection.
For lung tumor metastasis models, GSCs10 6 cells) stably transduced with WT1-AS(+), WT1-AS(-), or control lentiviruses were administrated into corresponding mice via the tail vein. Lung tissues were obtained at Partial tumor tissues were xed with 10% formalin, embedded with para n and cut into 5 μm sections, followed by hematoxylin-eosin (HE) staining and Ki-67 immunohistochemistry (IHC), and cell subpopulation analysis. HE analysis was carried out using the Hematoxylin and Eosin Staining Kit (Beyotime) according to the manufacturer's instructions. Brie y, tumor sections were depara nized with xylene, rehydrated with different concentrations of ethanol, and stained with hematoxylin and eosin. After dehydrated, permeabilized and sealed, the sections were imaged under a microscope. For IHC analysis, tumor sections were depara nized and rehydrated. After the removal of endogenous peroxidase and treatment of antigen retrieval, sections were blocked with 10% normal goat serum for 30 min at room temperature and incubated overnight at 4˚C with primary antibody against Ki67 and then probed with horseradish peroxidase-conjugated secondary antibody for 30 min at room temperature. Next, the sections were incubated with 3,3'-diaminobenzidine (DAB) solution and counterstained with hematoxylin.
After the routine treatment of dehydration, clearing and mounting, the sections were imaged. Cell subpopulation analysis was performed with the experimental procedures similar with cell sorting except cell sorting procedures.

RNA-seq
RNA was isolated from GSCs stably transduced with lentiviruses using Trizol Reagent (Thermo Fisher Scienti c), followed by the measurement of RNA concentration, purity and integrality. After quality control, RNA was puri ed, fragmented, and transformed into cDNA library. Next, cDNA library was enriched, quanti ed by Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA) and sequenced using the Illumina HiSeq 2500 instrument (Illumina).
Raw data obtained after RNA-seq were processed into clean data by removing the sequences with 3' adapters or reads with the average quality score < Q20. Next, ltered clean data were aligned to the reference genome (Homo_sapiens.GRCh38.dna.primary_assembly.fa) using the HISAT2 software (http://ccb.jhu.edu/software/hisat2/index.shtml). Read counts were normalized using the FPKM (fragments per kilo bases per million fragments) method. Gene were regarded as differentially expressed when they satis ed the conditions: |log 2 FoldChange| > 1 and P value <0.05. Cluster analysis for differentially expressed genes were performed using the long distance hierarchical clustering method (Complete Linkage) and Euclidean distance metric (R language: Pheatmap software package) according the expression levels of the same genes in different samples and expression patterns of different genes in the same sample. Expression trends of genes in the same cluster were similar. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was performed using KAAS software. The top 5 KEGG pathways were picked out according to the P values (the 5 minimum P values).
Clinical samples GC tumors and adjacent normal tissues were collected from 3 cases of GC patientsnderwent surgery sections during May 2017 to July 2018. Our study was conducted with the approval of the Medical Ethics Committee of the A liated Hospital of Jining Medical University, and the written informed consents from all participants.

Statistics Analysis
Data (in vitro and in vivo experiments) analysis was performed using GraphPad Prism 7 software (GraphPad Software, Inc., San Diego, CA, USA) with the results presenting as mean ± standard deviation. Difference between groups was analyzed using paired or unpaired T test. Difference among groups was analyzed using two-way ANOVA along with Sidak post-hoc test or one-way ANOVA along with Tukey's post-hoc test. P < 0.05 was regarded a signi cant difference.

Results
Expression analysis of WT1-AS or WT1 in GSCs stably transduced with corresponding overexpression or knockdown lentiviruses.
WT1-AS overexpression inhibited cell proliferation, facilitated cell apoptosis and reduced 5-FU resistance by down-regulating WT1 in GSCs.
CCK-8 assay showed that the enforced expression of WT1-AS notably weakened cell viability in GSCs ( Fig. 2A). Soft agar colony formation assay also disclosed that WT1-AS1 overexpression led to the obvious decrease of cell colony number in GSCs (Fig. 2B). Moreover, WT1-AS hindered cell cycle progression in GSCs, as evidenced by the notable reduction of cell percentage in G2 and S phases in GSCs stably transduced with WT1-AS(+) lentiviruses compared with control group (Fig. 2C). These outcomes showed that WT1-AS overexpression weakened the proliferative activity of GSCs. Also, a noticeable increase in cell apoptotic rate was observed in WT1-AS-overexpressed GSCs compared to NC group (Fig. 2D). Moreover, WT1-AS overexpression hindered cell cycle progression (Fig. 2E) and induced cell apoptosis (Fig. 2F) in 5-FU-treated cells, suggesting that WT1-AS weakened the resistance of GSCs to 5-FU. Conversely, WT1-AS knockdown led to the increase of cell viability ( Fig. 2A), cell colony number (Fig.  2B) and cell percentage in G2 and S phases (Fig. 2C), and reduction of cell apoptotic rate (Fig. 2D) in GSCs. WT1-AS loss enhanced the resistance of GSCs to 5-FU ( Fig. 2E and 2F). Moreover, our data disclosed that WT1 overexpression facilitated cell proliferation, curbed cell apoptosis and potentiated 5-FU resistance in GSCs ( Fig. 2A-2F). Additionally, WT1 up-regulation weakened the effects of WT1-AS on cell proliferation, apoptosis and 5-FU resistance in GSCs ( Fig. 2A-2F).
WT1-AS hindered cell migration and EMT by down-regulating WT1 in GSCs.
Next, Transwell migration assay disclosed that enforced expression of WT1-AS triggered the notable reduction of cell migratory potential in GSCs (Fig. 3A). The wound healing assay also presented that WT1-AS overexpression hindered cell migration in GSCs (Fig. 3B). RT-qPCR and western blot assays demonstrated that WT1-AS up-regulation led to the notable reduction in mRNA and protein expression levels of vimentin, twist1 and ctnnb1 in GSCs (Fig. 3C and 3D). These outcomes suggested that WT1-AS overexpression inhibited cell migration and EMT in GSCs. Conversely, WT1-AS depletion facilitated cell migration and induced the expression of vimentin, twist1 and ctnnb1 in GSCs (Fig. 3A-3D). In addition, increased cell migratory capacity and elevated expression of vimentin, twist1 and ctnnb1 was observed in GSCs stably transduced with WT1(+) lentiviruses compared with control group (Fig. 3A-3D). Furthermore, WT1 overexpression abrogated the detrimental effects of WT1-AS on cell migration and EMT in GSCs ( Fig. 3A-3D).
WT1-AS overexpression inhibited angiogenesis, reduced GSC stemness, and hindered the formation of invitro 3D GSC aggregates by down-regulating WT1.
Next, our data further revealed that the addition of supernatants of GSCs transduced with WT1-AS(+) lentiviruses led to the reduction in the angiogenesis activity of HUVECs in vitro, suggesting that WT1-AS overexpression suppressed angiogenesis (Fig. 4A). Moreover, WT1-AS overexpression notably reduced the percentage of CD44+ GSCs relative to control group (Fig. 4B). Additionally, our outcomes revealed that WT1-AS loss or WT1 overexpression in GSCs promoted angiogenesis in vitro and increased CD44+ stem-like subpopulation ( Fig. 4A and 4B). Furthermore, WT1 increase markedly alleviated the detrimental effects of WT1-AS on angiogenesis and attenuated WT1-AS-mediated loss of CD44+ subpopulation ( Fig.  4A and 4B). Next, 3D cell culture experiments were performed to mimic the physiological microenvironment of GSCs in view of the strong dependence of stem cells on their environment. Results showed that WT1-AS overexpression triggered the notable reduction in the size of 3D spheroid cultures derived from GSCs (Fig. 4C). Conversely, WT1-AS loss promoted the formation of in-vitro 3D GSC aggregates (Fig. 4C). Also, a marked increase in the size of 3D spheroids was observed in WT1overexpressed GSCs compared to control group (Fig. 4C). Moreover, WT1 overexpression weakened the inhibitory effects of WT1-AS on the formation of 3D spheroid cultures (Fig. 4C), suggesting that WT1-AS hindered the development of GSCs by down-regulating WT1 in 3D cell culture model. WT1-AS inhibited tumor growth and reduced cell stemness in subcutaneous xenograft tumors derived from GSCs.
Next, the effects of WT1-AS overexpression or knockdown on the growth and stemness of GSCs-derived xenograft tumors were further investigated in vivo. Subcutaneous xenograft experiments demonstrated that enforced expression of WT1-AS led to the notable reduction in the volume of xenograft tumors (Fig.  5A). Also, Ki67-positive cell percentage was markedly decreased in xenograft tumors of WT1-AS(+) group compared to NC group (Fig. 5B). Moreover, HE staining analyses presented that WT1-AS overexpression led to the notable reduction in the blue staining area/intensity and abnormality in the morphology of cells and cell nucleus (Fig. 5B). These data suggested that WT1-AS1 inhibited DNA replication and hindered the development of subcutaneous xenograft tumors. Additionally, a noticeable reduction in the percentage of CD44 positive population was observed in subcutaneous xenograft tumors of WT1-AS(+) group versus NC group (Fig. 5C). Conversely, WT1-AS knockdown induced the tumor growth and reduced CD44+ subpopulation in GSCs-derived subcutaneous xenograft tumors (Fig. 5A-5C).
WT1-AS inhibited tumor growth and metastasis, and reduced tumor stemness in xenografts derived from GSCs.
Additionally, abdominal xenograft experiments demonstrated that enforced expression of WT1-AS led to the notable reduction of ascites volume (Fig. 6A) and abdominal xenograft tumor weight (Fig. 6B). Also, a conspicuous down-regulation in the cell percentage of Ki67+ (Fig. 6C) and CD44+ (Fig. 6D) was observed in abdominal xenograft tumors of WT1-AS(+) group than that in NC group. HE staining analyses also revealed that cell nuclear (blue) staining areas were reduced and cell injury was mitigated in WT1-ASoverexpressed abdominal xenograft tumors relative to NC group (Fig. 6C). These data demonstrated that WT1-AS hindered the development of GSCs-derived abdominal xenografts. Inversely, WT1-AS depletion promoted the development of GSCs-derived abdominal xenograft tumors. To further investigate the effects of WT1-AS overexpression or knockdown on lung metastasis of GSC-derived xenograft tumors, GSCs stably transduced with NC, WT1-AS(+) or WT1-AS(-) lentiviruses were injected into mice via tail vein. IHC and ow cytometry analyses respectively showed that the positive cell percentages of Ki67 and CD44 were notably reduced in the lung tissues of WT1-AS(+) group, but markedly increased in WT1-AS(-) group ( Fig. 6E and 6F). HE staining data also revealed that WT1-AS overexpression inhibited the lung metastasis of GSCs, as evidenced by the reduction in the blue staining area/intensity and alleviation of lung tissue lesions (Fig. 6E). Conversely, WT1-AS loss induced the lung metastasis of GSCs ( Fig. 6E and   6F). These results suggested that WT1-AS inhibited tumor growth and metastasis, and reduced tumor stemness in xenografts derived GSCs (i.p. and i.v.).

Identi cation of downstream genes of WT1-AS by RNA-seq technology
Next, RNA-seq technology was employed to search for potential downstream genes regulated by WT1-AS in GSCs stably transduced with NC, WT1-AS(+) (AS_OV), WT1-AS(-) (AS_KD), or WT1-AS(+) + WT1(+) lentiviruses (ASWT_OV). RNA-seq outcomes revealed that 35 genes were differentially expressed in ASWT_OV versus AS_OV group (Table 2). Also, 587 or 332 differentially expressed genes were identi ed in AS_OV versus NC group or AS_KD versus NC group ( Table 2). The heat map of all differentially expressed genes was presented in Fig. 8A and corresponding gene information was displayed in Table 3. To identify potential biological relationships among genes, cluster analysis for all differentially expressed genes (n=842) in AS_OV versus NC, AS_KD versus NC and ASWT_OV versus AS_OV groups was performed according to the co-expression patterns of genes. As presented in Fig. 8B and Table 4, these differentially expressed genes (n=842) were groups into 9 different clusters. Cluster analyses suggested that genes (n=516) belonging to cluster-2 and cluster-7 could be regulated by WT1-AS rather than WT1 in GSCs. Among the differentially expressed genes in the cluster-2 and cluster-7, 36 genes were found to be involved in the regulation of the top 5 KEGG enrichment pathways (Table 5). Among these 36 genes, 10 genes (phosphoserine phosphatase (PSPH), phosphoglycerate dehydrogenase (PHGDH), glutathione Stransferase omega 2 (GSTO2), calcium channel α2δ1 subunit (CACNA2D1), FYN, monoamine oxidase B (MAOB), aminopeptidase-N (ANPEP), Integrin alpha 2 (ITGA2), cathepsin S (CTSS), and protein kinase C alpha (PRKCA) with potential oncogenic effects were selected for further investigations. RT-qPCR outcomes showed that the expression levels of PSPH, GSTO2, FYN, and PHGDH were notably reduced in WT1-AS-overexpressed GSCs, but were markedly up-regulated in WT1-AS-depleted GSCs (Fig. 8C).

Discussion
In this text, we demonstrated that WT1-AS knockdown or WT1 overexpression in CD24+/CD44 + GSCs enhanced GSC cell proliferative and migratory capacities, promoted GSC cell EMT, increased CD44 + stem-like cell subpopulation, induced HUVEC angiogenesis, inhibited GSC cell apoptosis, and enhanced the resistance of GSCs to 5-FU in vitro. Moreover, 3D cell culture experiments demonstrated that WT1-AS loss or WT1 increase facilitated the formation of in-vitro 3D GSC aggregates. Moreover, our data presented that WT1-AS weakened the malignant behaviors and properties of GSCs by down-regulating WT in vitro. In addition, our in vivo experiments demonstrated that WT1-AS inhibited the growth and metastasis of GSCs-derived xenograft tumors and reduced CD44 + stem-like cell subpopulation in xenografts. These data suggested that WT1-AS functioned as a negative regulator in the maintenance and reprogramming of GSC stemness properties.
It is well known to us that transcription factors can regulate the expression of lncRNAs [29]. Thus, transcription factors that had the likelihood to bind with the promoter regions of WT1-AS were predicted.
Among these transcription factors, TP53 [26,30] and XBP1 [27,28,31] were found to be implicated in the formation and development of cancer stem cells. Our experiments demonstrated that TP53 increase did not in uence WT1-AS expression in GSCs. However, XBP1 could negatively regulated WT1-AS expression by binding with promoter region of WT1-AS in GSCs.
Recently, RNA-seq technology has been widely used to decipher the molecular mechanisms of lncRNAs in physiological and pathological conditions [32]. Also, a growing body of evidences shows that lncRNAs can exert their functions by regulating the expression of coding genes [33]. Hence, downstream genes (excluding WT1) that could be regulated by WT1-AS were investigated by RNA-sEq. Combined with the KEGG enrichment data and cluster analyzes, 10 genes (PSPH [34,35], PHGDH [36,37], GSTO2 [38], CACNA2D1 [39,40], FYN [41], MAOB [42,43], ANPEP [44,45], ITGA2 [46,47], CTSS [48] and PRKCA [49] were selected for further explorations in view of their close association with cancer progression and prognosis. Our RT-qPCR assay validated that PSPH, GSTO2, FYN, and PHGDH could be regulated by WT1-AS, but not WT1 in GSCs. In other words, PSPH, GSTO2, FYN, and PHGDH were the downstream targets of WT1-AS. A recent study pointed out that enforced expression of FYN promoted cell migration, invasion and EMT in vitro and induced GC lung metastasis in vivo [50]. Additionally, Guo et al. demonstrated that FYN expression was positively regulated by 5'-Nucleotidase Domain Containing 2 (NT5DC2), while NT5DC2 depletion markedly weakened GSC cell viability and tumorsphere formation potential in vitro and hampered the growth of GSCs-derived xenograft tumors in vivo [51], suggesting that FYN might function as a positive regulator in the development of GSCs. Our RNA-seq and RT-qPCR assay validated that FYN expression was notably reduced in WT1-AS-overexpressed GSCs, but markedly increased in WT1-ASdepleted GSCs. These data suggested that WT1-AS might hindered the development of GSCs by downregulating FYN.
PHGDH also has been found to be highly expressed in GC tissues and GC patients with high PHGDH expression has a poor prognosis [52]. Moreover, PHGDH depletion weakened 5-FU resistance in GCs [53]. Additionally, Sharif et al. demonstrated that PHGDH was required to maintain the self-renewal activity, stemness and pluripotency in embryonal cancer stem-like cells [54]. Our data showed that WT1-AS overexpression led to the loss of stem-like features and down-regulation of PHGDH level in GSCs.
Combined with the prior report [54], we supposed that WT1-AS might negatively regulated the selfrenewal, tumorigenesis and metastasis of GSCs by silencing PHGDH.

Conclusions
Taken together, our data demonstrated that WT1-AS weakened the stem-cell like behaviors and characteristics of GSCs in vitro and in vivo by down-regulating WT1, elucidating the vital roles of WT1-AS in the maintenance and development of GSCs and revealing a mechanism governing WT1-AS-induced loss of GSC stem-like phenotypes. Moreover, XBP1 as an upstream regulator of WT1-AS was identi ed in GSCs. Additionally, our data and previous studies suggested that FYN and PHGDH might be the potential downstream targets of WT1-AS. Furthermore, multitudinous downstream genes that could be regulated by WT1-AS or WT1 were identi ed by RNA-seq technology in GSCs. An in-depth understanding on the molecular mechanisms underlying the complex phenotypes of GSCs might contribute to the better management of intra-tumoral heterogeneity, drug resistance, tumor development and recurrence.  distribution patterns. *P < 0.05. **P < 0.01. ***P < 0.001. *: compared to NC group. ##P < 0.01.###P < 0.001. #: compared to WT1-AS(+) group.    alterations of lungs were examined by HE analysis. (F) The proportion of CD44+ cell population was measured using a ow cytometry. **P < 0.01. ***P < 0.001. *: compared to NC group.