High ECT2 expression is an independent prognostic factor for poor overall survival and recurrence-free survival in non-small cell lung adenocarcinoma

Different subtypes of non-small cell lung cancer (NSCLC) have distinct sites of origin, histologies, genetic and epigenetic changes. In this study, we explored the mechanisms of ECT2 dysregulation and compared its prognostic value in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC). In addition, we also investigated the enrichment of ECT2 co-expressed genes in KEGG pathways in LUAD and LUSC. Bioinformatic analysis was performed based on data from the Cancer Genome Atlas (TCGA)-LUAD and TCGA-LUSC. Results showed that ECT2 expression was significantly upregulated in both LUAD and LUSC compared with normal lung tissues. ECT2 expression was considerably higher in LUSC than in LUAD. The level of ECT2 DNA methylation was significantly lower in LUSC than in LUAD. ECT2 mutation was observed in 5% of LUAD and in 51% of LUSC cases. Amplification was the predominant alteration. LUAD patients with ECT2 amplification had significantly worse disease-free survival (p = 0.022). High ECT2 expression was associated with unfavorable overall survival (OS) (p<0.0001) and recurrence-free survival (RFS) (p = 0.001) in LUAD patients. Nevertheless, these associations were not observed in patients with LUSC. The following univariate and multivariate analysis showed that the high ECT2 expression was an independent prognostic factor for poor OS (HR: 2.039, 95%CI: 1.457–2.852, p<0.001) and RFS (HR: 1.715, 95%CI: 1.210–2.432, p = 0.002) in LUAD patients, but not in LUSC patients. Among 518 genes co-expressed with ECT2 in LUAD and 386 genes co-expressed with ECT2 in LUSC, there were only 98 genes in the overlapping cluster. Some of the genes related KEGG pathways in LUAD were not observed in LUSC. These differences might help to explain the different prognostic value of ECT2 in LUAD and LUSC, which are also worthy of further studies.

Introduction Non-small cell lung cancer (NSCLC) is one of the leading causes of cancer death in the world [1]. NSCLC accounts for about 80% of all lung cancer cases and can be divided into three subtypes, including lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC) and large cell carcinoma (LCLC) [2]. Different subtypes have distinct sites of origin, histologies, genetic and epigenetic changes [3,4]. These differences are closely related to their unique responses to therapy [5,6]. Therefore, it is meaningful to investigate the difference in their molecular mechanisms.
Epithelial cell transforming sequence 2 (ECT2) is a guanine nucleotide exchange factor encoded by ECT2 gene in human [7]. In non-transformed cells, ECT2 is involved in the regulation of cytokinesis via catalyzing guanine nucleotide exchange on the small GTPases, RhoA, Rac1, and Cdc42 [7]. ECT2 is frequently upregulated in human cancers and acts as an oncogene [8,9]. In the transformed growth of ovarian and lung cancer cells, ECT2 has distinct regulative effects from its role in cytokinesis [9][10][11]. Nuclear ECT2 can activate Rac1 in the cancer cells and recruit Rac effectors to the nucleus, which is required for tumor initiation and transformation [9,11]. Knockdown of ECT2 can inhibit Rac1 activity and block transformed growth, invasion and tumorigenicity of LUAD cells [9,12]. One recent study found that ECT2 upregulation was associated with worse disease-free survival and overall survival (OS) of patients with LUAD [13].
ECT2 maps to 3q26.31 in the human genome. In fact, broad 3q chromosome amplification is the most common chromosomal aberration found in LUSC [14,15]. In this study, via bioinformatic analysis, we explored the mechanisms of ECT2 dysregulation in NSCLC and compared its prognostic value in LUAD and LUSC. In addition, we also investigated the enrichment of ECT2 co-expressed genes in KEGG pathways in LUAD and LUSC.

Materials and methods
Bioinformatic analysis using FireBrowse ECT2 expression in some solid tumors and in corresponding normal tissues was analyzed by using data from the Cancer Genome Atlas (TCGA). Data analysis was performed by using FireBrowse (http://firebrowse.org/), which provides access to analyze data generated by TCGA.

Bioinformatic analysis using UCSC Xena browser
The level 3 data of patients with primary NSCLC in TCGA-NSCLC were obtained by using the UCSC Xena browser (https://xenabrowser.net/) [16]. ECT2 mRNA expression, exon expression and DNA methylation in patients with primary LUAD or LUSC were also examined using data in TCGA-LUAD and TCGA-LUSC, by UCSC Xena browser. Kaplan-Meier curves of OS and recurrence-free survival (RFS) after initial therapy were generated by GraphPad Prism v6.0. loaded into ClueGo in Cytoscape [19] for analysis of KEGG pathways. Only pathways with pvalue 0.05 were included.

Statistical analysis
Statistical analysis was performed by using SPSS 19.0 (SPSS Inc., Chicago, IL, USA). The association between ECT2 RNA expression and the clinicopathological features was evaluated using χ 2 tests. Receiver operating characteristic (ROC) curves for death and recurrence detection were constructed and the optimal cut-off value of ECT2 expression was determined based on Youden index. Log-rank test was performed to assess the difference between the survival curves. Prognostic values were analyzed by univariate and multivariate Cox regression models. Welch's t-test was conducted to compare ECT2 RNA expression between LUAD and LUSC groups. p < 0.05 was considered to be statistically significant.

Results
ECT2 was significantly upregulated in both LUAD and LUSC compared with normal lung tissues By data mining using FireBrowse, we characterized ECT2 mRNA expression in several types of solid tumors, including LUAD and LUSC. Results indicated that ECT2 expression was approximately 4-fold higher in LUAD tissues than in normal lung tissues, while was about 9-fold higher in LUSC tissues than in normal lung tissues (Fig 1). To further compare ECT2 expression in LUAD and LUSC, ECT2 mRNA RNAseq and exon RNAseq data in TCGA-LUAD and TCGA-LUSC were extracted for analysis. Heatmap and following comparison showed that ECT2 expression was significantly higher in LUSC than in LUAD tissues (Fig 2A and 2B).
LUSC had a lower level of ECT2 DNA methylation and a higher level of ECT2 DNA amplification than LUAD Then, we tried to investigate the underlying mechanisms of dysregulated ECT2 expression in LUSC and LUAD. By comparing ECT2 expression and its DNA methylation, we observed that the level of ECT2 DNA methylation was significantly lower in LUSC cases than in LUAD cases ( Fig 3A). Some CpG loci were hypermethylated in LUAD, but not in LUSC ( Fig 3A). Then, we examined copy number alterations (CNA) in TCGA-LUAD and TCGA-LUSC. ECT2 mutation was observed in 5% of LUAD and in 51% of LUSC cases ( Fig 3B). Amplification was the predominant type of alteration and was associated with increased ECT2 mRNA expression in both LUAD and LUSC (Fig 3C and 3D).
ECT2 DNA mutation was associated with worse disease-free survival in LUAD, but not in LUSC patients Then, we studied the association between ECT2 DNA mutation and survival in LUAD and LUSC patients. Survival curves indicated that LUAD patients with ECT2 amplification had significantly worse disease-free survival (p = 0.023, Fig 4A). In comparison, although ECT2 amplification was common in LUSC patients, there was no significant association between ECT2 amplification and disease-free survival (Fig 4B).
High ECT2 expression was an independent prognostic factor for poor OS and RFS in LUAD, but not in LUSC patients The associations between ECT2 expression and the demographic and clinicopathological parameters in patients with primary LUAD and LUSC were summarized in Tables 1 and 2. In patients with LUAD, the high ECT2 expression group had significantly lower proportions of female (156/318, 49.1% vs. 115/184, 62.5%, p = 0.0036) and lifelong non-smoker (37/311,   11.9% vs. 35/177, 19.8%, p = 0.018) than the low ECT2 expression group (Table 1). Besides, the high ECT2 expression group also had a significantly higher ratio of death (136/318, 42.8% vs. 47/184, 25.5%, p<0.0001) compared with the low ECT2 expression group (Table 1). In contrast, these associations were not observed in LUSC patients ( Table 2). In LUAD, ECT2 expression gradually increased with the increase of pathological stages (Fig 5A). In comparison, this   trend was not observed in LUSC (Fig 5B). High ECT2 expression was associated with significantly worse OS (p<0.0001) and RFS (p = 0.001) in patients with LUAD (Fig 5C and 5D). Nevertheless, no significant association was observed in patients with LUSC (Fig 5E and 5F). By performing univariate analysis, we found that advanced stage (III/IV) and high ECT2 expression were associated with significantly shorter OS and RFS in LUAD patients (Table 3). Following multivariate analysis confirmed that the high ECT2 expression was an independent prognostic factor for poor OS (HR: 2.039, 95%CI: 1.457-2.852, p<0.001) and RFS (HR: 1.715, 95%CI: 1.210-2.432, p = 0.002) in LUAD patients (Table 3). In comparison, ECT2 had no prognostic value in LUSC patients (Table 4).

ECT2 was involved in different signaling pathways in LUAD and LUSC
By data mining using cBioPortal for Cancer Genomics, we identified the genes co-expressed with ECT2 in LUAD and LUSC (|Pearson's r| ! 0.4 and |Spearman's r| ! 0.4) (S1 Table).
Results indicated that 518 genes were co-expressed with ECT2 in LUAD and 386 genes were co-expressed with ECT2 in LUSC (Fig 6A and S1 Table). However, only 98 genes were in the overlapping cluster (Fig 6A). To further investigate the possible signaling pathways in which ECT2 might be involved in, ECT2 co-expressed genes in LUAD and LUSC were subjected to KEGG pathway analysis respectively. In LUAD, the genes were enriched in Pyrimidine metabolism, Ribosome biogenesis in eukaryotes, p53 signaling pathway, HTLV-I infection, RNA transport, Base excision repair, Homologous recombination, Fanconi anemia pathway, Cell cycle, Oocyte meiosis, Progesterone-mediated oocyte maturation, DNA replication, Nucleotide excision repair and Mismatch repair (Fig 6B and S2 Table). In comparison, ECT2 coexpressed genes in LUSC were enriched in Glycerophospholipid metabolism, Cell cycle, p53 signaling pathway, DNA replication, Mismatch repair, Homologous recombination and Fanconi anemia pathway (Fig 6C and S3 Table).

Discussion
Aberrant ECT2 expression was observed in both LUAD and LUSC [13,20,21]. In the current study, via characterizing ECT2 expression based on data in TCGA-LUAD and TCGA-LUSC, we also confirmed significantly upregulated ECT2 expression in LUAD and LUSC compared with normal lung tissues. In addition, we found that ECT2 expression was considerably higher in LUSC than in LUAD tissues. Previous studies indicated that copy number gains (CNGs) are one of the most common mechanisms of dysregulated genes at chromosome 3q26 [15,21]. In this study, we found that amplification is common in LUSC, but not in LUAD. Approximately 50% of LUSC cases had ECT2 amplification, but this rate was only around 5% in LUAD. These findings are consistent with the prevalence of chromosome 3q26 CNGs in LUSC and the relatively rare occurrence of 3q26 CNGs in LUAD [22]. In addition, we also observed that some CpG loci of ECT2 gene had higher levels of methylation in LUAD than in LUSC, suggesting that epigenetic alteration is also an important mechanism of dysregulated ECT2 in NSCLC. These results help to explain why ECT2 expression is significantly higher in LUSC than in LUAD.
As an oncogene, ECT2 upregulation also has prognostic values in some cancers. In patients with colorectal cancer, high expression level of ECT2 was significantly associated with tumor size, serum CEA levels and TNM stage [23]. Kaplan-Meier survival analysis indicated that patients with high ECT2 expression had a remarkably shorter OS [23]. High level of ECT2 expression was also associated with poor prognosis in patients with esophageal squamous cell carcinomas [20]. One study based on patients with LUAD indicated that high ECT2 expression was associated with unfavorable disease-free survival and overall survival [13]. However, the number of patients included in this study is relatively small (N = 88) [13]. In this study, based on large datasets in TCGA, we found that although ECT2 amplification is common in LUSC, its mutation had no influence on survival outcomes. Nevertheless, although ECT2 amplification was less frequent in LUAD, its mutation was associated with significantly worse disease-free survival. By generating Kaplan-Meier curves, we further demonstrated that in patients with LUAD, high ECT2 expression was related to unfavorable OS and RFS. But no significant association was observed in patients with LUSC. In addition, our univariate and multivariate analysis showed that high ECT2 expression was an independent prognostic factor for poor OS (HR: 2.039, 95%CI: 1.457-2.852, p<0.001) and RFS (HR: 1.715, 95%CI: 1.210-2.432, p = 0.002) in LUAD patients, but not in LUSC patients. Therefore, we hypothesized that ECT2 might play different roles in LUAD and LUSC.
In LUAD, one recent study indicated that ECT2 could activate rRNA synthesis by binding the nucleolar transcription factor upstream binding factor 1 (UBF1) on rDNA promoters and recruiting Rac1 and its downstream effector nucleophosmin (NPM) to rDNA [11]. However, whether other mechanisms are involved in the oncogenic properties of ECT2 in LUAD and whether ECT2 participates in different molecular pathways in LUAD and LUSC have not been fully revealed. By comparing ECT2 co-expressed genes in LUAD and LUSC, we found a considerable variation. Among 518 genes co-expressed with ECT2 in LUAD and 386 genes coexpressed with ECT2 in LUSC, there were only 98 genes in the overlapping cluster. The following KEGG pathway analysis of the enrichment of ECT2 co-expressed genes showed that Cell cycle, p53 signaling pathway, DNA replication, Mismatch repair, Homologous recombination and Fanconi anemia pathway are common in LUSC and LUAD. In LUAD, ECT2 co-expressed genes were additionally enriched in some cancer-related pathways, such as Pyrimidine metabolism, Ribosome biogenesis in eukaryotes, RNA transport and Base excision repair. Therefore, it is meaningful to further investigate the involvement of ECT2 in these pathways in LUAD in the future.

Conclusion
Both genetic and epigenetic alterations contributed to dysregulated ECT2 in NSCLC. High ECT2 expression was an independent prognostic factor for poor OS and RFS in LUAD patients, but not in LUSC patients. Some of the genes related KEGG pathways in LUAD were not observed in LUSC. These differences might help to explain the different prognostic value of ECT2 in LUAD and LUSC, which are also worthy of further studies.