Effects of mRNA expression of five Notch ligands on prognosis of gastric carcinoma

Notch ligands are expression changes in a great many malignancies including gastric cancer (GC) frequently. The prognostic value of each Notch ligands in GC patients remains lack of large sample data results. In present research, we researched the prognostic value of Notch ligands in GC patients in order to fill the shortage areas. We used an online database (http://kmplot.com/analysis/index.php?p=service&cancer=gastric) to identify the relationship between mRNA expression of each Notch ligand and overall survival (OS) in GC. We analyze the relevance of overall survival and clinical data which includes gender, Lauren's classification, differentiation, clinical stage and treatment. The study found that high DLL1, DLL3, DLL4 and JAG2 mRNA expression were tied to worse OS in all GC patients followed up for 10 years. There is no significant relevance to the expression of JAG1 mRNA and OS in patients with GC. We also did a survey of each Notch ligands in different clinical and pathological features present different prognosis. The information will help to better understand the biology of gastric cancer heterogeneity, provide more accurate prognostic evaluation tools and provide new targets for targeted drug development besides.

www.nature.com/scientificreports/ was set up using ten-year follow-up data and gene expression data of 876 carcinoma of stomach downloaded from GEO. The specific datasets are GSE14210, GSE15459, GSE22377, GSE29272, GSE51105 and GSE62254. In another word, five Notch ligands were entered into the database ((http:// kmplot. com/ analy sis/ index. php?p= servi ce& cancer= gastr ic) to acquire Kaplan-Meier survival plots. In Kaplan-Meier Plotter, three Probe set options are provided. This study we chose only JetSet best probe set. Jetset selects the optimal probe set for each gene using a scoring method established to assess each probe set for specificity, coverage and degradation resistance. When auto select best cutoff is selected, all possible cutoff values between the lower and upper quartiles are computed, and the best performing threshold is used as a cutoff. In general, p value < 0.05 were considered statistically significant.
Statement. This research was performed in accordance with relevant guidelines/regulations, and informed consent was obtained from all participants and/or their legal guardians. This Research have been performed in accordance with the Declaration of Helsinki.

Results
Kaplan-meier survival data of GC patients with differential expressions of Notch ligand can be discovered at www. kmplot. com. We were the first to analyze the prognostic value of DLL1 expression. The Affymetrix ID is: 224215_s_at. Figure 1A shows the survival curve of GC patients based on DLL1 mRNA expression level (n = 631  1D). For DLL4, the Affymetrix ID is: 223525_at. High expression of DLL4 was significantly tied to worsening OS in GC patients, HR 1.35 (1.08-1.68), p = 0.0091 ( Fig. 2A). For JAG2, the Affymetrix ID is: 209784_s_at. The OS of GC patients with high expression level of JAG2 mRNA was worse, HR 2.08 (1.69-2.55), p = 1.6e−12 (Fig. 2B). The same as Affymetrix ID is: 32137_at. HR 1.89 (1.57-2.29), p = 1.6e−11 (Fig. 2C).
We filled the gaps in the association between individual Notch ligands and clinicopathological features ( Table 1). Just as, DLL3 high expression was so clearly tied to worsening OS in male, HR 1.71 (1.38-2.14), p = 1.2e−6. DLL4 mRNA expression was so clearly tied to worsening OS in male, HR 1.     www.nature.com/scientificreports/

Discussion
The four receptors and five ligands constituted the Notch signaling pathway, a kind of intercellular communication system which is vitally important for embryogenesis and tissue homeostasis 13 . Current research has shown that DLL1 obviously has a crucial impact on tumorigenesis and the development of many cancers. The most widely studied and comparatively mature is breast cancer, DLL1 overexpression leads to poor prognosis through cell proliferation, maintenance of tumor stem cells and angiogenesis [14][15][16] . While, clustered DLL1 therapy results in significant enhancement of tumor antigen-specific T cell immune responses and memory, remarkably increased tumor infiltration by T cells, attenuated tumor vascularization, and suppression of tumor growth and produced remarkably enhanced progression-free survival 17 . These articles have been confirmed that DLL1 plays a crucial role in the formation of blood vessels. During the formation of new blood vessels, arterial endothelial cells induce a large amount of DLL1 expression. Angiogenesis in GC is formed under pathological conditions, and it is still regulated by molecular signals similar to the formation of blood vessels under physiological conditions. In this report, by analyzing 10-year follow-up data of GC, we were pleasantly surprised that patients with high expression of DLL1 had a worsening prognosis. Therefore, it also indirectly proved that DLL1 has great potential in tumor therapy of the formation of blood vessels. High expression of DLL1 leads to poor prognosis in patients with diffuse GC. This is consistent with the research conclusion of Giulia's 18 , Notch1 activity in gastric cancer is controlled by the epigenetic silencing of the ligand DLL1, and that Notch1 inhibition is associated with the diffuse type of gastric cancer. DLL3 is part of the Notch five transmembrane ligands family and presumably a crucial predictive biomarker and therapeutic target 19 . It has been reported that it has a pivotal role in gastrointestinal neuroendocrine carcinoma cells. There is a significant upregulation of DLL3 in gastrointestinal neuroendocrine carcinoma cell lines. The gene silencing of DLL3 caused significant growth inhibition through the induction of intrinsic apoptosis 20 . High-grade neuroendocrine carcinomas were negative for DLL3, and five-year OS and PFS were significantly improved in patients who received adjuvant chemotherapy. However, the patients with positive for DLL3 were no difference 21 . In vitro experiment have been confirmed DLL3 overexpression can promote the proliferation of gastric cancer cells in vitro, and down-regulation of DLL3 inhibits the proliferation of gastric cancer cells 22 . However, the literature we found only a few data have analyzed the relationship between DLL3 and prognosis in gastrointestinal tumors, and the expression of DLL3 in GC and its effect on prognosis have not been reported. In the study, we detect the expression of DLL3 was linked to worsening OS for GC patients followed for ten years, as well as in male patients, in intestinal, surgery, combined with 5-FU chemotherapy and stage II and stage III. These results could herald DLL3 as a negative regulator and an independent biomarker of negative prognosis in advanced GC.
The expression of DLL4 ligand in the Notch signaling pathway is most important in tumor endothelial cells. In the tumor microenvironment, tumor secreted angiogenic factors such as VEGF can induce the expression of DLL4 in endothelial cells 23 . According to the analysis of 383 patients with GC, the overexpression of DLL4 activates the Notch-1 signaling pathway, effectively promotes proliferation, lymph node metastasis and distant  25 . In summary, it can be said that DLL4 is a potential novel prognostic indicator and therapeutic target for GC to guide clinical practice.
There is now compelling evidence that overexpression of JAG1 can promote metastasis of colorectal carcinoma by inducing epithelial-mesenchymal transition. JAG1 can be used as a target for anti-tumor therapy 26,27 . In contrast, JAG1 has been deemed a tumor suppressor gene in acute myeloid leukemia, which cancer cell growth can be reduced and patients with JAG1 overexpression also have a better prognosis 28 . Xiao's reported that when miR-124 overexpression in GC the expression of the JAG1 and EZH2 was downregulated, and silencing of JAG1 or EZH2 by RNA interference also suppressed GC cell growth and metastasis. It has been certified that the JAG1 mRNA overexpressions were found in low differentiation adenocarcinoma, samples with lymph node metastasis, and samples at stage II, II and IV in GC tissue 29 . In our data, we found that high JAG1 expression was associated with worsening OS for 10 years of follow-up in GC patients. JAG1 overexpression was tied to better OS in stage I, intestinal, poorly differentiated and well differentiated cancer presents. However, patients treated alone with surgery had a worse prognosis. Our findings complement Xiao's results, but there are some notable differences. The latest research reported one of Notch ligand, JAG2 correlated to several carcinogenesis. In GC, Kang's reported that JAG2 was significantly higher in GC tissues than in pericarcinomatous tissue. High relative expression of JAG2 was tied to better OS in univariate analysis 30 . Our study shows that for ten years, JAG2 mRNA expression was found to be significantly related to worsening OS for GC patients. JAG2 mRNA overexpression was tied to worsen OS in all clinical stages except stage II. It was inconsistent with Kang's findings. The reason may be due to the difference in sample size. Although this study is the largest sample size reported to date. With the continuous updating and improvement of the database, we believe that there will be a larger sample size to verify our views in the future. Most studies currently focus on the biological effects of a single notch ligand. Whether the expression of the five Notch ligands is related to each other and affects each other is still unknown, especially in tumor cells, which needs to be explored in the future.
In conclusion, by exploiting the KM plotter database, we clear and definite the prognostic roles of five Notch ligands in GC, and update gene expression data and survival information from the largest sample library to date of 876 GC patients. In all patients with GC, there is an insignificant correlation between the expression of only JAG1 mRNA and OS. Our study preliminarily explored the prognostic role of Notch ligand in various clinicopathological characteristics. The results of the research will be useful for understanding the biology of GC heterogeneity, providing more accurate prognostic evaluation tools, and providing new potential targets for targeted drug development.

Data availability
The datasets generated during and/or analysed during the current study are available in the [Kaplan-Meier plotter] repository, [(http:// kmplot. com/ analy sis/ index. php?p= servi ce& cancer= gastr ic].