**4. Discussion**

The epithelial–mesenchymal transition (EMT) plays a crucial role in promoting metastasis of carcinoma derived from epithelial cells. Tumor cells lose their epithelial characteristics such as cell polarity and gain mesenchymal features such as increased migratory and invasive potentials during EMT. Our data did not show an association of NME1 expression with tumor growth (Figure 4), consistent with a previous study showing that *NME1* silencing does not provide epithelial cancer cells with a selective growth advantage [17]. A number of groups have reported the relationship between NME1 expression and patient prognosis in NSCLC with different results. Some groups have reported no association between NME1 expression and overall survival [28,29]. In contrast, reduced NME1 expression has been found to be associated with bone metastasis and poor survival in patients with pulmonary adenocarcinoma [30]. Ohta et al. [31] have also reported that NME1 expression is inversely correlated with the microdissemination of tumor cells in stage I NSCLC. In addition, stage

I NSCLC patients with NME1-negative expression show a significantly poorer survival than those without [32,33]. The present study also showed the negative effect of reduced NME1 expression on RFS in early stage NSCLC, consistent with findings from previous groups [30–33].

The negative effect of reduced NME1 expression on RFS in this study was worse in patients with β-catenin overexpression than in those without. How does β-catenin overexpression influence the effect of reduced NME1 expression on RFS? Mechanisms underlying the metastasis suppression of NME1 have been addressed in multiple types of cancer cells. Early efforts have revealed that NME1 may mediate its inhibitory effects on cellular motility and invasion through interactions with signaling cascades [16,34–36]. For example, NME1 negatively regulates Rac1 (Rac family small GTPase 1) and Cdc42 (Cell division cycle 42) GTPase by interacting with Rac1-specific nucleotide exchange factors, TIAM Rac1 associated GEF1 (Tiam1) and TGF\_BETA\_2 domain-containing protein (Dbl-1), respectively [36]. A splicing variant of *NME1* inhibits the metastasis of lung cancer cells by interacting with Inhibitor of nuclear factor Kappa-B Kinase subunit beta (IKKβ) in an isotype-specific fashion and regulating tumor necrosis factor alpha (TNFα)-stimulated Nuclear Factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling negatively [16]. In addition, NME1 inhibits the liver metastasis of colon cancer cells by regulating the phosphorylation of myosin light chains in nude mice [35]. *NME1* silencing induces the nuclear translocation of β-catenin by disrupting adherence junction complexes mediated by E-cadherin and promotes extracellular matrix invasion by increasing invadopodia formation and pericellular matrix metalloproteinase (MMP) activity [17].

In addition to its effect on signaling pathways, NME1 is also known to regulate gene transcription by binding to single-stranded DNA. A promoter region between –922 to –846 of Kangai 1 (KAI1) is known to suppress metastasis through the inhibition of cell movement. It responds to NME1 in high-metastatic lung cancer cell line L9981 [37]. NME1 suppresses motile and invasive phenotypes of melanoma cells by inducing the transcription of integrin beta-3 (*ITG*β*3*) gene through direct physical interaction with the promoter [38]. NME1 also plays a role as a co-regulator of transcription by regulating expression of metastasis-related genes through direct or indirect interactions with transcription-regulatory elements [39–42]. However, the present study showed no relationship between NME1 expression and nuclear β-catenin expression (Figure 5), suggesting that the adverse effect of NME1 on RFS exacerbated by β-catenin overexpression might not be due to the nuclear translocation of β-catenin by reduced NME1. It is likely that NME1 may interact with β-catenin through other mechanisms such as upregulation of many genes related to cell cycle, apoptosis, and metastasis.

Platinum derivatives such as cisplatin are widely used chemotherapeutic agents for NSCLC. However, cisplatin resistance is a major challenge in the use of these drugs. The molecular mechanism of cisplatin resistance in lung cancer cells is not fully understood. Therefore, there are few efficient strategies to overcome such resistance. Cisplatin-based adjuvant chemotherapy in the present study did not affect RFS in univariate analysis. However, it worsened the RFS in patients with reduced NME1 expression (Supplementary Table S5). A functional link between NME1 expression and responsiveness to cisplatin-based adjuvant chemotherapy has been reported by several groups. Cisplatin increases interstrand DNA cross-links and inhibits pulmonary metastatic colonization in *NME1*-transfected breast cancer cells [43]. NME1 has 3 -5 exonuclease activity potentially involved in DNA proofreading [44]. Thus, reduced expression of NME1 may contribute to chemoresistance by allowing metastatic cells to escape from apoptosis. Knockdown of *NME1* by shRNA transfection in head and neck squamous carcinoma cells attenuates the chemosensitivity of cells to cisplatin by downregulating cyclins E and A and reducing cisplatin-induced S-phase accumulation [45]. These lines of evidence suggest that reduced NME1 expression might be involved in cisplatin resistance through various mechanisms. Therefore, restoring NME1 expression might be a therapeutic intervention strategy to surmount cisplatin resistance.

Previous studies have demonstrated Wnt/β-catenin-mediated resistance to cisplatin in various types of cancers [23,24]. Transient interference of cytoplasmic GSK-3β increases cisplatin resistance by activating Wnt/β-catenin signaling in cisplatin-resistant A549 cells [23]. Recently, Zhang and colleagues [24] have reported that the interference of β-catenin expression by siRNA can decrease mRNA and protein levels of anti-apoptotic gene *Bcl-xl* and increase cisplatin sensitivity in A549 wild-type cells. Despite these associations of β-catenin overexpression and cisplatin resistance in various types of cancer cells, β-catenin overexpression alone was not associated with cisplatin resistance in the present study. However, β-catenin overexpression aggravated RFS when patients with reduced NME1 received cisplatin-based adjuvant chemotherapy (Table 1). Further studies are needed to better understand the combined effect of β-catenin and NME1 on RFS of patients receiving platinum-based adjuvant chemotherapy in early stage NSCLC.

This study was limited by several factors. First, this was a retrospective study that was prone to selection and surveillance biases. Second, it is necessary to investigate combined effects of β-catenin and NME1 on apoptosis, migration, invasion, or metastasis in different cell types of lung cancer to clearly understand the molecular mechanisms underlying the effect of β-catenin and NME1 on poor RFS. Third, the lack of a negative effect of β-catenin in the univariate analysis (Supplementary Table S2) in this study might be due to the small sample size and short duration of follow-up. Fourth, *EcoR1* (rs34214448-G/T) polymorphism in *NME1* gene is associated with increased susceptibility to NSCLC [46] and could potentially affect the results of the current analysis, which is based only on expression levels. Fifth, the relationship between the Th1 (T helper cell type 1) and Th 2 (T helper cell type 2) ratio and β-catenin levels were not analyzed in this study. The balance between Th1 and Th2 in the tumor microenvironment is regulated by several factors, and β-catenin may affect the tumor microenvironment. Thus, for the understanding of their relationship and the analysis of β-catenin levels, it may be informative to know the Th1/Th2 ratio of patients. Sixth, the number of patients receiving adjuvant chemotherapy was too small. Accordingly, prospective large-scale studies are needed to validate the effect of β-catenin and cisplatin-based adjuvant chemotherapy on NME1-related RFS in early stage NSCLC.

In conclusion, the present study suggests that the adverse effect of reduced NME1 expression on RFS may be exacerbated by cisplatin-based adjuvant chemotherapy and β-catenin overexpression through other mechanisms rather than through the nuclear translocation of β-catenin in early stage NSCLC. Accordingly, it is recommended that cisplatin-based adjuvant chemotherapy in patients with completely resected stage I–IIIA NSCLC be carefully applied after examining the expression levels of β-catenin and NME1.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/9/10/3067/s1, Table S1: Relationship between NME1 expression and clinicopathological characteristics (*N* = 425). Table S2: Univariate analysis of RFS (*N* = 425). Table S3: Cox proportional hazards analysis of RFS according to NME1 in early stage NSCLC (*N* = 425), stratified by pathologic stages. Table S4: Cox proportional hazards analysis† for RFS in early stage NSCLCs (*N* = 425). Table S5: Cox proportional hazards analysis† for RFS according to NME1 expression in 425 early stage NSCLCs treated with and without cisplatin-based adjuvant chemotherapy. Figure S1: Kaplan–Meier plot of recurrence-free survival according to NME1 expression in histologic subtypes: (A) Total, (B) Adenocarcinoma, (C) Squamous cell carcinoma.

**Author Contributions:** Conceptualization D.K. (Dohun Kim) and D.-H.K.; data curation, Y.K. and W.-J.K.; formal analysis, Y.K., B.B.L., and D.K. (Dongho Kim); methodology, O.-J.L. and P.J.; resources, E.Y.C., J.H., and Y.M.S.; supervision, D.-H.K.; writing—original draft preparation, D.K. (Dohun Kim) and D-H.K.; writing—review and editing, D.K. (Dohun Kim), Y.K., B.B.L., D.K. (Dongho Kim), O.-J.L., P.J., W.-J.K., E.Y.C., J.H., Y.M.S., and D.-H.K.; funding acquisition, D.-H.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a grant from the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2019R1F1A1057654) and from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health and Welfare (HI18C1098), Republic of Korea.

**Acknowledgments:** The authors thank Eunkyung Kim and Jin-Hee Lee for data collection and management, and Hoon Suh for sample collection.

**Conflicts of Interest:** The authors declare no conflict of interest.
