*3.5. No Correlation between Reduced RUNX1 Expression and Expression Levels of Phospho-Rb and Ki67 Proliferation Index*

RUNX proteins are implicated in diverse signaling pathways and cellular processes, including the cell cycle and stress response. In order to elucidate the effect of RUNX1 on the cell cycle and cell proliferation in NSCLC, we analyzed the phospho-pRb (Ser-807/811) levels and Ki-67 proliferation index according to the expression status of RUNX1. The average phospho-pRb (Ser-807/811) levels were 2.7% in tumor tissues with reduced RUNX1 expression and 2.1% in tumor tissues without reduced RUNX1 expression. The difference was not statistically significant (*p* = 0.18), irrespective of histology (Figure 5A). The Ki-67 proliferation index in tumor tissues with reduced RUNX1 expression was slightly higher than in tumor tissues without reduced RUNX1 expression, but the difference was also not statistically significant (28.8% vs. 22.3%, *p* = 0.18), irrespective of histology (Figure 5B).

**Figure 5.** The effect of RUNX1 expression on phospho-pRb (Ser-807/811) level and Ki-67 proliferation index. The expression levels of phosphorylated pRb stained using polyclonal anti-phospho-pRb (Ser-807/811) antibody (Cell Signaling, Danvers, MA, USA) (**A**) and Ki-67 proliferation index (**B**) were compared according to the expression status of RUNX1 using Student *t*-test. The phospho-pRb levels and Ki-67 proliferation index were not significantly different between tumor tissues with normal or reduced RUNX1 expression irrespective of histologic subgroup.

#### **4. Discussion**

RUNX1 causes a wide range of leukemias through translocation with genes such as eight-twenty-one (ETO) [7] and acts as an oncogene in various solid tumors such as ovarian cancer [15], and endometrial cancer [16], as well as in the mouse mammary tumor virus-polyoma middle tumor-antigen (MMTV-PyMT) transgenic mouse model of breast cancer [17], and in the transgenic adenocarcinoma of mouse prostate (TRAMP) model of prostate cancer [18]. RUNX1 is also known to function as a tumor suppressor in different types of cancer. For example, the ectopic expression of RUNX1 in esophageal adenocarcinoma cells reduced the anchorage-independent growth [19], and the knockdown of *RUNX1* by siRNAs enhanced androgen-independent proliferation of prostate cancer cells [20]. In addition, the inhibition of endogenous *RUNX1* using short-hairpin RNA targeting RUNX1 (shRunx1) in breast cancer cells resulted in loss of epithelial morphology and promotion of epithelial-mesenchymal transition [9], and the ectopic expression of RUNX1 reduced the population of breast cancer stem cells [21]. RUNX1 also inhibited the migration and stemness of mammary epithelial cells [22]. Ramsay et al. [10] reported that lentiviral-mediated RNAi knockdown of *RUNX1* increased the proliferation and migration of lung cancer cells. In the present study, *RUNX1* showed abnormal methylation in primary NSCLCs, and the reduced expression of RUNX1 was associated with poor overall survival, suggesting that RUNX1 may play a role as a tumor suppressor in normal bronchial epithelial cells.

Functional disruption of RUNX1 usually occurs by chromosomal translocation, point mutation, or deletion in leukemia and some solid tumors. *RUNX1* mutation has been reported rarely in lung cancer [23], although changes in its methylation have been reported by a couple of studies [24,25]. In this study, *RUNX1* was found to be abnormally methylated at the CpG island of *RUNX1* in NSCLC tumor tissues, and the methylation and mRNA levels of *RUNX1* showed a linear negative correlation. Unlike most genes whose transcription is regulated by a single promoter, *RUNX1* is regulated by two promoters in the upstream region of 5 UTR [26]. The three hypermethylated CpGs in this study might affect the transcription of *RUNX1*, which may also be affected by tissue-specific control factors. Further studies are needed to understand the mechanisms underlying transcriptional repression mediated by abnormal methylation of *RUNX1* in NSCLC.

A CpG (cg04228935) for the prediction of NSCLC was identified using tumor and matched normal tissues obtained from 42 NSCLC patients. Although the number of normal samples in the TCGA lung cancer data is small and the prevalence of lung cancer is not the exact same between Koreans and Americans, the present study suggests that RUNX1 hypermethylation may be a useful biomarker for the early detection of NSLC in other populations worldwide. Screening of lung cancer using low-dose computed tomography (LDCT) reduces mortality; however, approximately 20% of pulmonary nodules were found to be false positive [27,28]. A biopsy is needed for a more accurate diagnosis of lung cancer, but it is very difficult to obtain tissue in some patients. Methylation levels of a CpG (cg04228935) from bronchial biopsy were comparable to those from surgically resected lung tumor tissues. Accordingly, bronchial biopsy specimens may be used for the molecular analysis of *RUNX1*, and advances in technology such as electromagnetic navigation bronchoscopy (ENB) and endobronchial ultrasonography using a guided sheath (EBUS-GS) may provide more adequate specimens with fewer complications.

The association of *RUNX1* mutations or changes in expression with the prognosis of patients has been reported in various carcinomas, and the effect of RUNX1 on prognosis varies considerably depending on the type of cancer. *RUNX1* mutations are associated with poor overall survival in adult acute myelogenous leukemia (AML) as well as in pediatric AML [29,30]. The RUNX1 expression in prostate cancer tissues was negatively associated with poor prognosis [20]. The low RUNX1 expression in breast cancers is associated with metastasis to lymph nodes and poor survival [9,21]. In addition, the RUNX1-RUNX3 expression showed a significant effect on the survival of breast cancer patients with high YAP-signature expression levels [22]. Lung adenocarcinomas with low RUNX1 expression were associated with poor overall survival compared to tumors with high RUNX1 expression [10]. Our data also showed that reduced RUNX1 expression was associated with poor overall survival in adenocarcinomas. Based on these observations, it is likely that the reduced expression of RUNX1 may serve as an indicator of poor prognosis in patients with lung adenocarcinoma.

RUNX proteins are known to regulate a wide range of biological processes via various interacting proteins in human cancer and to be implicated in carcinogenesis mediated via TGF-β and Wnt signaling pathways, and in cell cycle or stress response. For example, RUNX1 promoter is regulated by EZH2 (enhancer of zeste homolog 2)-dependent histone H3 lysine 27 (K27) trimethylation in prostate cancer cells [20]. RUNX1 directly regulates E-cadherin, and rescues TGFβ-induced EMT phenotype in breast cancer cells [9]. RUNX1 suppresses breast cancer growth by repressing the activity of breast cancer stem cells and inhibiting ZEB1 expression directly [21]. RUNX1 acts as a negative regulator of oncogenic function of YAP that is involved in solid tumor progression [22]. The effect of RUNX1 on cell cycle in lung cancer differs between study groups. RUNX1 stimulated G1 to S progression in hematopoietic cells, partly via transcriptional induction of cyclin D2 promoter [31], whereas RUNX1 depletion resulted in an increased E2F1 mRNA levels in lung cancer cells [10]. In this study, tumor tissues with reduced RUNX1 expression did not show high levels of pRb phosphorylation (Ser-807/811) or the Ki67 proliferation index, suggesting that the reduced expression of RUNX1 may be involved in lung carcinogenesis through other mechanisms rather than cell-cycle regulation and growth control.

This study was limited by several factors. First, the effect of the two promoters on abnormal methylation of three CpGs in *RUNX1* gene and the tissue-specific factors affecting the expression of RUNX1 were not fully elucidated. Second, we failed to analyze *RUNX1* methylation in circulating cell-free DNA and to evaluate the prediction performance of the model due to assay failure. Third, the present study was a retrospective case-control study, which can result in a biased estimate of the population prevalence of NSCLC. In addition, sputum analysis was limited to the very few specimens from NSCLC patients only. Accordingly, the prediction performance of the model needs to be validated using several molecular techniques such as droplet digital polymerase chain reaction (ddPCR) in sputums and cell-free DNAs from a large cohort. Fourth, it is unclear the abnormal methylation of RUNX1 as a predictive biomarker can also be applied to tissue samples from metastatic lesions because the data from the present study and TCGA was from surgical specimens of early stage

tumors. Fifth, methylation levels from benign lung tumors such as localized organizing pneumonia and hamartoma were not analyzed due to lack of samples.

In conclusion, the present study suggests that abnormal methylation at the CpG island of the *RUNX1* gene may be a valuable biomarker for the detection of NSCLC regardless of races. Reduced expression of RUNX1 may be associated with poor overall survival in patients with lung adenicarcinoma.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/9/6/1694/s1: Table S1: Prediction performance of logistic regression model based on three CpGs in a test dataset (N = 269) of TCGA lung cancer. Table S2: Clinicopathological characteristics. Supplementary Figure S1: Relationship between differentiation and *RUNX1* methylation. Supplementary Figure S2: Association between RUNX1 expression and pathologic stage.

**Author Contributions:** Conceptualization, Y.K. and D.-H.K.; Data curation, Y.K., B.B.L., and D.K.; Formal analysis, B.B.L., E.Y.C., J.H., and D.K.; Methodology, Y.K. and D.-H.K.; Software, Y.K. and D.-H.K.; Supervision, D.-H.K.; Validation, Y.K., B.B.L., and D.K.; Resources, S.U., E.Y.C., J.H., and Y.M.S.; Visualization, Y.K.; Writing—original draft preparation, Y.K. and D.-H.K.; Writing—review and editing, Y.K., B.B.L., D.K., S.U., E.Y.C., J.H., Y.M.K., 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 grants from Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2019R1F1A1057654), 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.
