Gli1 regulates stemness characteristics in gastric adenocarcinoma

Background Glioma-associated oncogene homolog 1 (Gli1), affects the progression and the stemness characteristics of malignant carcinoma. The aim of the present study was to identify the relation between Glioma-associated oncogene homolog 1 (Gli1) and stemness and determine its clinical significance in gastric adenocarcinoma (GA). We investigated Gli1 expression and its correlation with other stemness-associated proteins in 169 GA samples and 5 GA cell lines. Methods To elucidate the role of Gli1 in the clinicopathological significance and stemness of GA, tissues samples from 169 GA patients were collected for immunohistochemistry (IHC). Additionally, MKN74, MKN28, NCI-N87, SNU638, AGS cells were collected for western blotting, MKN28 cells were collected for spheroid formation assay. Results Results showed that Gli1 expression was closely related to tumor grade, primary tumor (pT) stage, distant metastasis, clinical stage, gross type, microvessel density, and shorter overall survival (OS). Cox regression analysis verified that Gli1 was an independent prognostic factor for OS. Furthermore, Gli1 expression correlated with the expression of stemness-related genes, CD44, LSD1, and Sox9. Gli1 inhibitor GANT61 significantly decreased the expression of CD44 and LSD1, and spheroid formation ability of the MKN28 cells. Conclusions In conclusion, Gli1 may be a poor prognostic indicator and a potential cancer stemness-related protein in GA.

homologs, PTCH1 and PTCH2, and both gene products can bind to Hh ligands. In the absence of the Hh ligand, PTCH inhibits Smo activity. The presence of Hh ligand relieves this inhibition, allowing smo to activate the Gli family of transcription factors. Gli1 has been implicated in several human cancers, including a role in the progression of pancreatic cancer [3] and an association with poor prognosis in glioblastoma [4], pancreatic cancer [5], and breast cancer [6]. Gli1 expression correlates with stemness in breast and lung cancers and is essential in the cellular proliferation and growth of these tumors [7,8]. Although Gli1 expression has been studied in many human cancers, its role as a prognostic indicator and its functional significance in determining the stemness of GA cells warrants further investigation.
In this study, we investigated the clinicopathological value of Gli1 and evaluated the correlation between Gli1 and stemness in GA.

Tissues
One hundred sixty-nine cases GA tissue in paraffin section were gained from the Affiliated Hospital of Yanbian University and comply with agreements approved by the institutional review committee. The tissue samples were collected from 1995 to 2000. Preoperative chemotherapy or radiotherapy was not implemented. Clinical and pathological reports were reviewed for age, sex, tumor size, tumor grade, tumor location, primary tumor (pT) stage, lymph node metastasis, distant metastasis, gross type and histological type.
The double immunostaining procedure was executed in the same section, the first step was to use anti-Gli1 antibody staining with 3,3′-diaminobenzidine, the second step was to use anti-CD105 antibody (Abcam) staining with AEC.
Two pathologists (WB Qi & YH Xuan) assessed the immunohistochemical results and the staining results were assessed according to previous study [9].

Western blotting
The tumor cells were lysed by using RIPA buffer with Phenyl methane sulfonyl fluoride (PMSF). The same amount of protein was separated with 10% SDS-PAGE gel and then was transferred onto PVDF membranes (Biotech). Subsequently, 5% skim milk (diluted in PBS) was used to blocked the PVDF membranes for 2 h at RT. And then the membranes were incubated with anti-Gli1 (Santa), anti-CD44 (Abcam), anti-LSD1 (ZSGB-BIO), anti-Sox9 (Abcam), anti-β-actin (Abcam). The next step is to incubate anti-rabbit/mouse for 2 h. Detection was performed by the ECL kit.

Immunofluorescence analysis
MKN28 cells were planted and were cultured to 60-70% density. 4% polyformaldehyde was used to fix cells for 20 min. And 0.5% Triton X-100 was used to permeabilize cells for 20 min. Next, 3% BSA was used to block cells for 1 h. Absorbent paper absorbs the sealing liquid and does not wash. Cells were incubated with anti-Gli1/LSD1, anti-Gli1/Sox9 for 2 h. The next day, cells were incubated with secondfluorescence antibodies (Invitrogen, A12380 and A11008) for 1 h. Finally, DAPI was used to stain the nuclear. Fluorescence detection was performed with the Axiovert200II (Carl-Zeiss).
Statistical analysis SPSS 25.0 statistical software (NO. 1975-01566-C), Pearson's chi-square test and mean ± standard deviation was used for the data analysis, and the results was evaluated by analysis of variance. The Kaplan-Meier method was used to identify the overall survival (OS) and were compared using the log-rank test. Univariate and multivariate analysis was used for the Cox proportional hazards model. The GraphPad Prism 7 software is used for statistics on the results of western blotting. P value less than 0.05 was considered to have statistical significance.

Association between the expression of Gli1 and clinical characteristics of GA
To understand if Gli1 is associated with GA progression, we investigated Gli1 expression in human GA by a Tissue Microarray (TMA) analysis. TMA analysis was performed for Gli1 expression by IHC staining in adjacent non-tumorous gastric epithelium and GA tissues. IHC staining revealed that Gli1 expression in GA ( Fig. 1b-c) was higher than non-tumorous gastric epithelium (Fig. 1a). Gli1 significantly correlated with tumor grade (P = 0.001), pT stage (P = 0.029), clinical stage (P = 0.005), distant metastasis (P = 0.007), and gross type (P = 0.021) ( Table 1), not with age, sex, tumor location, tumor size, lymph node metastasis, histological type. Interestingly, our results find a correlation between Gli1 expression and pT stage and distant metastasis, but no correlation with tumor size or lymph node metastasis. These results are accordance with the data in GEPIA (Gene Expression Profiling Interactive Analysis) and TCGA (The Cancer Genome Atlas) that Gli1 expression was higher in clinical stage (2/3/4) compared with clinical stage (1) (P < 0.001), and was not correlated with lymph node metastasis (Supplemental Figure).
The Kaplan-Meier survival analysis revealed that Gli1 expression in GA was associated with lower OS (P < 0.001; Fig. 1d). The univariate Cox regression analysis showed that tumor size, pT stage, lymph node metastasis, distant metastasis, and Gli1 expression (all P < 0.05) were independent prognostic factors for poor OS. The multivariate Cox regression analysis revealed that pT stage, lymph node metastasis, distant metastasis, and Gli1 expression (all P < 0.05) were independent prognostic predictors for OS ( Table 2). These results demonstrated that Gli1 is a potential prognostic biomarker of GA.
Furthermore, double-staining results proved that CD105 expression (blood vessels) was around Gli1 expression (cancer cells) (Fig. 1e). Microvessel density (MVD) was significantly higher in Gli1(+) group (55.51 ± 36.34) than in Gli1(−) group (36.86 ± 30.85) (P = 0.003; Fig. 1f). These results demonstrated that Gli1 may be likely to metastasize through the microangiogenesis and then promoting distant metastasis and finally promote tumor progression. This result further explained the potential reasons why there is an association between Gli1 and distant metastasis.

Correlation between Gli1 and stemness in GA
We have reported that Gli1 is associated with stemness in breast cancer and lung squamous cell carcinoma [9,10]. To identify the role of Gli1 in cancer stemness of GA, we studied Gli1 and stemness-related protein expression in GA. The result showed that Gli1 related with *Statistically significant findings cancer stemness proteins, CD44, LSD1, and Sox9 (all P < 0.05) (Fig. 2a) (Table 3). Gli1, LSD1, Sox9 were primarily expressed in the nucleus of cancer cells; CD44 primarily located in the membranes of cancer cells. To further confirm above results, we investigated Gli1, CD44, LSD1, Sox9 expression in GA cells by western blotting. Gli1 expression in GA cells (MKN74, MKN28, NCI-N87, SNU638, AGS) were similar with stemnessrelated proteins (Fig. 2b). Furthermore, an immunofluorescence assay indicated that the Gli1-positive cell population were strongly identical with the LSD1 and Sox9 positive cell population within the MKN28 cells (Fig. 2c).
To further understand the interaction between Gli1 and cancer stemness in GA cells, we blocked Gli1 expression using Gli1 inhibitor in MKN28 and MKN74 cells. Our studies showed that protein CD44, LSD1, Sox9 expression in MKN28 and MKN74 cells were significantly decreased after Gli1 was inhibited (P < 0.05, Fig. 3a-b). Subsequently, tumorsphere-forming experiment was performed to investigate the ability of clonogenic potential of MKN28 cells. Notably, MKN28 cells dealed with GANT61 reduced clonogenic potential compared with cells treated with DMSO (control group) (Fig. 3c). These results indicate a possibility that expression of Gli1 may enhances cancer cells to acquire stemness properties thereby promoting progression of GA.

Discussion
Gli1 promotes the progression of many types of cancers including, pancreatic and prostate cancer [10][11][12]. Our study showed that Gli1 is overexpressed in GA tissue samples, and its expression correlates with adverse clinicopathological parameters. This suggests a role for Gli1 in the initiation, progression, and metastasis of GA. In addition, Gli1 expression correlates with unfavorable prognosis of breast cancers [13][14][15]. Our survival analysis revealed similar poor prognosis correlated with Gli1 overexpression in GA. Angiogenesis is an important step in malignant tumor growth and progression [16]. Abnormally activated Hh/Gli1 pathway in gliomas promotes tumor microvessel formation [17], and Gli1 overexpression in esophageal tumors significantly correlates with increased microvessel density [18]. Similarly, Gli1 expression in our study was also associated with higher microvessel density, indicating that Gli1 may promote the progression of GA via angiogenesis. In human gliomas, the Hh/Gli1 pathway plays an important role in CSC self-renewal and tumorigenicity [19]. Gli1 expression also significantly correlates with stemness characteristics of esophageal carcinoma [18]. Moreover, some of stemness related proteins were used to identify the gastric CSC populations, such as CD44, CD133, LSD1, Sox9, LGR5 [20][21][22][23][24][25]. Evidence exists for the presence of cancer stem cells in colorectal cancer, with some phenotypes being CD44+/CD166+ enriched CSCs [26]. Our study demonstrated that Gli1 expression correlates with expression of cancer stemness-related proteins. Colocalization of Gli1 with cancer stemnessrelated proteins in GA tissue indicates that Gli1 may be an important stemness-related protein in GA. Furthermore, stemness-related protein expression in MKN28 and MKN74 cell lines and the ability of MKN28 cell lines to form spheroids were diminished after Gli1 inhibitor GANT61 was used. These factors prove that Gli1 regulates specific features of CSCs, such as self-renewal and proliferation. However, further investigation is necessary to elucidate the mechanism of Gli1 action in GA stem cells.
The SHh signaling pathway is a major regulator of tumorigenesis, tumor progression and therapeutic response. Downstream effectors of the SHh pathway include Smo and Gli family of zinc finger transcription factors. Both are regarded as important targets for cancer therapeutics. SMO inhibition prevents the downstream activation of Gli transcription factors, leading to suppression of those genes associated with cancer growth and progression. To date, SMO inhibitors include cyclopamine [27], LDE225 [28], and GDC-0449  [29] were investigated in GC. Gli1 is an extremely important part of the Hh signaling pathway and can activate most of the Hh pathway target genes. Developing Gli-targeted approach has its merit because of the fact that Gli proteins can be activated by both SHh liganddependent and -independent mechanisms. Gli1 and Gli2 inhibitor include GANT 61 [30] and Arsenic Trioxide [31] that have shown potent inhibition of Gli1 and Gli2 in many cancer cell lines, one of these is GC cells. Currently, many preclinical studies and clinical trials are being conducted to evaluate the efficacy of this exciting class of targeted therapy in a variety of cancers. We expect these inhibitors to be used clinically to help GC patients with targeted therapies.
Generally, targeting cancer cell stemness-associated genes may be an effective therapeutic strategy to overcome tumor relapse and chemoresistance. In this study, Gli1 proved to be a molecular marker for cancer stemness and a prognostic indicator of GA. We speculate that jointly targeting Gli1 and other cancer stemness biomarkers will provide a novel vision to treat GA.

Conclusions
Gli1 was upregulated in GA tissues and cancer cells and correlated with poor prognosis in GA patients. Knocking down Gli1 by specific inhibitor suppressed the expression of Gli1 protein levels. Reduced expression of Gli1 downregulated the protein levels of cancer stemness biomarkers while also decreasing cell clonogenic potential in GA cells. Thus, Gli1 may promote the progression of GA by maintaining GA cell stemness potential. Hh/Gli1 pathway may play an important role in CSC self-renewal and tumorigenicity. Taken the above results together, we speculated that Gli1 may play a potential role in cancer stemness and thus to accelerate the progression in GA.

Limitations of this current study
Our study has several innate limitations to note. First, Spheroid formation experiments were performed only on the MKN28 cell line. Immunofluorescence was performed only on Gli1, LSD1, Sox9. Other stemness related features have not been tested.
Additional file 1: Supplemental Figure. The association between Gli1 mRNA expression and clinical stage, lymph node metastasis in GEPIA and TCGA data in GA.