Introduction

Esophageal carcinoma (EC) is a considerable medical and public health challenge in different regions worldwide, especially in Asia. Globally, it is ranked as the sixth cause of cancer-related deaths [1]. The overall age-adjusted incidence rate (ASR) of EC for men and women in the highly developed areas of the world is 6.5 and 1.2 per 100,000 persons, respectively. In contrast, the related rates in the developing areas are 11.8 and 5.7 [2]. Based on histological features, EC has two major types: squamous cell carcinoma and adenocarcinoma. Esophageal squamous cell carcinoma (ESCC) is the most common type of EC in Asian countries, in a region defined as the “esophageal cancer belt” which extends from north Iran eastward to China [3]. Risk factors in ESCC etiology include lack of dietary fruits and vegetables, tobacco and opium consumption, and the drinking of hot beverages [4]. Diagnosis of ESCC in an early stage is strongly associated with improved outcome. However, most patients are diagnosed in advanced stages, and the 5-year survival rate after surgery is only about 35 % [5]. Different genetic as well as epigenetic processes contribute to the development and progression of tumors [6]. Increasing evidence suggests that tumors are maintained by cancer stem cells (CSCs). CSCs are found in ESCC, and several well-known CSC genes including CD133, NANOG, OCT4, SALL4, and SOX2 have been proposed as ESCC CSC and disease progression biomarkers (for a recent overview, see [7]). Therefore, exploring CSC marker expression in ESCC could pave the road for better therapy.

The myeloid ecotropic viral integration site 1 (MEIS1) transcription factor (TF) gene was originally identified as a common viral integration site involved in myeloid leukemia [8]. A member of the three amino acid loop extension (TALE) family of homeodomain proteins, it is an important developmental TF, both in its own right and as a protein cofactor to other (TALE or HOX) homeobox proteins. Also, another TALE family members, including the other MEIS genes (MEIS2 and 3) and the pre-B cell leukemia homeobox (PBX) genes (PBX1-4), are important in normal development [9]. Conversely, upon deregulated expression, these genes can cause severe developmental disorders and cancer. MEIS, PBX, and HOX genes regulate the expression of their complex target gene network as protein-DNA complexes, in which the MEIS, PBX, and HOX proteins display specific DNA binding properties [10]. MEIS1 has a distinct role in self-renewal and maintenance of stemness state of different stem cell types, including neural and hematopoietic [11–13]. In addition, it has been shown that high MEIS1 expression has a role in the self-renewing of neural stem cells in developing olfactory epithelium [14] and can regulate the transcription of the critical self-renewal gene, OCT4, in neural stem cells [15].

Since normal stem cells share different properties such as self-renewal with CSCs, significant roles can also be construed for developmental TFs in cancer progression and maintenance. Among these TFs, SOX2 is involved in normal development of different organs as well as maintenance of self-renewal capacity of embryonic stem cells (ESCs) [16, 17]. It has been shown that both SOX2 gene amplification and mRNA overexpression are correlated to poor prognosis in several malignancies. High SOX2 expression is associated with lymph node metastasis, depth of tumor invasion, and poor differentiation in ESCC and lung cancer [18–20], also with metastasis in brain, breast, colorectal, and prostate malignancies [21–23]. Furthermore, it is involved in tumor initiation and apoptosis resistance in ovarian cancer [24]. Although SOX2 expression and function have been shown in a variety of cancers, its upstream regulatory mechanisms are almost completely unknown. The clarification about all of these mechanisms will enrich our knowledge about CSC self-renewal.

In this study, we aimed to evaluate the regulatory role of MEIS1 expression in ESCC and elucidate a possible interaction between MEIS1 and SOX2, which may be involved in maintaining the stemness state and self-renewal of ESCC cells.

Materials and methods

Clinical samples

Primary tumor tissue samples, with adjoining nontumoral tissue margins, from 50 ESCC patients were freshly collected during therapeutic surgery at the Omid Oncology Hospital of the Mashhad University of Medical Sciences, Mashhad, Iran. The clinical characteristics of the patients are summarized in Table 1. All specimens were obtained before any chemotherapy and radiotherapy treatments to prevent influence of treatment on clinically relevant features of the tumor samples. All tumor and adjacent nontumoral tissues were histopathologically confirmed, and the clinical features of the tumor samples were defined based on the 7th edition of Union International Cancer TNM classification guidelines [25]. The size of tumor samples ranged from 5 to 12 cm (mean ± SD, 4.12 ± 1.96). The mean age ± SD of patients was 61.5 ± 11.7. The study was approved beforehand by the local ethics committee, and the declared consent of all patients for enrollment in this study is on record.

Table 1 Clinical and immunohistochemical characteristics of the patients used in this study
Full size table

Cell lines and culture conditions

Human KYSE-30 ESCC cells [26] and human HEK293T embryonal kidney cells were cultured in RPMI-1640 medium (PAA, Pasching, Austria) and DMEM medium (Gibco, Grand Island, NY), respectively. Media were supplemented with 10 % heat-inactivated fetal bovine serum (FBS, Invitrogen, Carlsbad, CA), 10 mM glutamine, 100 U/mL, and 100 mg/mL penicillin-streptomycin (PAA) at 37 °C in a humidified atmosphere containing 5 % CO2. All cell lines were purchased from the Pasteur Institute Cell Bank of Iran (http://ncbi.pasteur.ac.ir/) and used at low passage numbers only. The KYSE-30 cell line was last successfully authenticated by short tandem repeat profiling at the Pasteur Institute Cell Bank of Iran on April 26, 2015, after completion of all experiments in this manuscript.

RNA extraction and qRT-PCR

RNA was extracted from ESCC cell lines and from tumor and adjacent nontumoral tissues of ESCC patients using TRIpure reagent (Roche, Nutley, NJ). Following cDNA synthesis, quantitative real-time PCR (qRT-PCR) to evaluate quantitative changes of MEIS1, EZH2, and SOX2 mRNA expression in ESCC samples using gene-specific primer sets (Table 2) with GAPDH as the reference mRNA was performed as described before [27]. Briefly, PCR was performed on 200 ng reverse-transcribed RNA in a total volume of 20 μL in 1× SYBR Green Real Time PCR Master Mix (Parstous, Mashhad, Iran) containing 0.5 μM of each primer. PCR consisted of an initial denaturation for 10 min at 94 °C, followed by 40 cycles of 15 s 94 °C, 30 s 60 °C, 30 s 72 °C, and was performed in an Mx-3000P real-time thermocycler (Stratagene, La Jolla, CA).

Table 2 Primer sequences used for qRT-PCR and MSP-PCR
Full size table

Immunohistochemistry

For MEIS1 protein tissue detection, we used the Novolink Polymer Detection kit (RE7200-CE, Leica Biosystems, Newcastle, UK) according to the manufacturer’s protocols. Briefly, immunohistochemistry was performed on formalin-fixed, paraffin-embedded esophageal tumor and adjacent nontumoral sections after deparaffinization and rehydration by xylene and ethanol dilutions, respectively. The sections were treated with antigen retrieval buffer for 30 min at 100 °C. Prediluted anti-MEIS1 antibody solution (ChIP Grade, ab19867, Abcam, Cambridge, UK) was applied to tissue sections for 30 min at room temperature. After washing with Tris-buffered saline, specific antigen-antibody binding was detected with Novolink polymer solution + DAB (Leica Biosystems, Buffalo Grove, IL). Tissue sections were immersed in hematoxylin-eosin for counterstaining and dehydrated using ethanol. Immunohistochemical staining was analyzed by light microscopy and scored according to Sincrope et al. based on either the percentage of cells with positive nuclear and cytoplasmic staining or on the overall cellular expression intensity, with scores <6 or ≥6 defined as low or high, respectively [28].

Public ESCC dataset analysis

For MEIS1 mRNA expression differences between ESCC and adjacent nontumoral tissue, all three ESCC mRNA profiling datasets available in the public domain were analyzed: Hu-34 (GSE20347) [29], Kimchi-24 (GSE1420) [30], and Su-106 (GSE23400) [31]. The transcript view genomic analysis and visualization tool (http://r2.amc.nl) were used to test whether the probe set selected uniquely mapped to an antisense position in an exon of the gene. The probe sets selected for MEIS1 in the Affymetrix Human Genome U133A arrays (204069_at) met all these criteria and showed the highest expression for MEIS1 in the datasets analyzed. The datasets were obtained from the Oncomine Web site (www.oncomine.org) and analyzed at standard settings. Other cohort details are available through GSE (www.ncbi.nlm.nih.gov/gds/) and PubMed (www.ncbi.nlm.nih.gov/pubmed/) links.

Bisulfite sequencing

Bisulfite conversion of genomic DNA for methylation detection was performed using the CpGenome DNA Modification kit (Chemicon International, Temecula, CA) according to the manufacturer’s protocol with the following exceptions: 2 μg DNA was resuspended in 0.3 M NaOH and heated at 50 °C for 10 min. The DNA was then incubated with reagent-I at 50 °C for 7 h with tube inverting every 30 min. Methylation-specific (MSP) PCR amplification of two CpG islands on the MEIS1 promoter region was performed using MSP primers (Table 2) on bisulfite-converted DNA from ESCC tissues showing low MEIS1 mRNA expression. Briefly, PCR was performed on 40 ng bisulfite-treated DNA in a total volume of 20 μL in 1× Hot Start Taq polymerase buffer (Finzymes, Espoo, Finland) containing 0.2 mM dNTP, 0.5 μM of each primer, and 0.5 U Taq Hot Start Polymerase (Takara, Shiga, Japan). PCR consisted of an initial denaturation for 12 min at 95 °C, followed by 35 cycles of 30 s 95 °C, 30 s 56 °C, 30 s 72 °C), and a final incubation for 15 min at 72 °C. The purified products were cloned into the pTZ57R/T vector using the T/A cloning kit (Fermentas, Vilnius, Lithuania) and sequenced using the M13 primer set (Macrogen, Seoul, Korea). The results were analyzed using the BiQAnalyser online tools (http://biq-analyzer.bioinf.mpi-inf.mpg.de/).

MEIS1 and EZH2 gene expression knockdown

The pLKO.1 lentiviral shRNA expression vector [32] encoding a validated shRNA sequence targeting human MEIS1 (TRCN0000015969) was obtained from Sigma-Aldrich (St. Louis, MO). The shc003 plasmid encoding GFP in the pLKO.1 backbone (Sigma) was used as a control. Lentiviral second-generation packaging plasmids psPAX2 and pMD2.G were purchased from Addgene (plasmids 12260 and 12259, respectively, Cambridge, MA). To produce lentiviral particles, the pLKO.1-MEIS1 plasmid was cotransfected into HEK293T cells along with the packaging plasmids according to the calcium phosphate-based Trono lab protocol [33]. Retrovirus vectors encoding EZH2-specific shRNA (RNAi-Ready pSIREN-RetroQ Vector, kindly provided by Yutaka Kondo, Nagoya, Japan) were used to target EZH2 expression. The shRNA vector was cotransfected with VSV-G and GP vectors into HEK293T cells as described above. After 48 h of transfection, the supernatant containing viral particles was harvested by ultracentrifugation (40-mL culture medium per 50-mL Beckman tube, ultracentrifugation for 120 min at 70,000×g, at 4 °C) and used to transduct KYSE-30 cells; the infected cells were selected by puromycin (Invitrogen Corporation, Carlsbad, CA) 48 h after infection. Quantification of MEIS1 and EZH2 mRNA knockdown was performed by qRT-PCR as described above.

Western blotting

Western blotting was performed as in reference [34] using BioRad equipment (Munich, Germany). The same quantity of protein from each sample was separated from 10 % SDS-PAGE gel, and then, protein was transferred to nitrocellulose membrane (N7892, Sigma-Aldrich). β-Actin was used as a loading control. The primary antibodies used were as follows: MEIS1 (ab19867, Abcam); SOX2 (NB110-37235, Novus Biologicals, Littleton, CO); β-actin (ab25894, Abcam) diluted at 1:1000, 1:400, and 1:2000, respectively. The secondary antibody used for MEIS1 and β-actin was anti-rabbit IgG Peroxidase (A0545, Sigma-Aldrich) diluted at 1:20,000; for SOX2, it was anti-mouse IgG1 (NBP1-51688, Novus Biologicals) diluted at 1:500. All antibodies were diluted in 2.5 % skim milk. Protein was incubated with specific primary antibodies at 4 °C overnight. After incubation with secondary antibody for 1 h at room temperature, the protein was detected by enhanced chemiluminescence (Clarity™ Western ECL Substrate kit #170-5060, BioRad).

Statistical analysis

Statistical analysis was performed using the SPSS 19.9 statistical package (SPSS, Chicago, IL). The correlations between gene expression and various histopathological features were assessed using both the χ2 and Fisher exact tests, and the correlation between MEIS1 and SOX2 expression was assessed using Pearson’s correlation (Table 1). To correlate gene expression levels (mRNA and protein), two-sided t tests were performed (Figs. 1, 4, and 5, and Supplemental Fig. 1). P < 0.05 was considered statistically significant.

Fig. 1
figure 1

MEIS1 mRNA analysis in 50 ESCC and matched adjacent nontumoral tissue samples and mRNA profiling datasets in the public domain. a MEIS1 mRNA expression is significantly lower in tumor than in adjacent nontumoral tissue. MEIS1 mRNA expression was determined with qRT-PCR, with GAPDH as the reference gene. Shown are the 2log values, median centered. A t test was used to assess the statistical difference between ESCC and adjacent nontumoral tissue. b MEIS1 expression is significantly lower in ESCC than in adjacent nontumoral tissue samples in three different ESCC mRNA profiling datasets in the public domain. a Hu-34: 17 ESCC and matched adjacent nontumoral tissue samples. b Kimchi-24: 8 ESCC and 8 adjacent nontumoral tissue samples. c Su-106: 53 ESCC and matched adjacent nontumoral tissue samples. Shown are the 2log median centered mean values, the error bars represent the SD. A two-sided t test was performed that was used to assess the statistical difference in mRNA expression. See “Materials and methods” for further details on the datasets and analysis

Full size image

Results

MEIS1 decreased expression at mRNA and protein level in ESCC tumor samples

To determine a possible role for MEIS1 expression in ESCC pathogenesis, we evaluated MEIS1 mRNA expression levels in 50 ESCC samples and compared these with the expression levels in the matched adjacent nontumoral margin tissues of esophageal epithelium, by qRT-PCR. As shown in Fig. 1a, MEIS1 mRNA expression in ESCC tissues was significantly lower than in the matched adjacent nontumoral tissue (P < 0.05). In 38 % of samples (19 of 50), ESCC MEIS1 mRNA expression was more than 2-fold lower than in the adjacent nontumoral esophageal tissue. To verify that our ESCC patient cohort is representative, and the results could be repeated in other ESCC series, we also analyzed ESCC mRNA expression profiling datasets in the public domain. Publicly available ESCC datasets are few in size and number. We found three datasets with MEIS1 expression data in ESCC and adjacent nontumoral samples. Two of these sets, Hu-34 and Kimchi-24, are quite small, and their analysis should be considered with some care. However, in all three sets, including the much larger Su-106 set, the MEIS1 expression was significantly lower in ESCC than in (matched) adjacent nontumoral esophageal tissue (Fig. 1b). We therefore tentatively concluded that our patient cohort was representative, and that the observation of lower MEIS1 ESCC expression is robust.

To extend this observation, we examined MEIS1 protein expression in 27 ESCC tissues and their adjacent nontumoral margin by immunohistochemical staining. Although both tissue types demonstrated immunoreactivity, in line with the mRNA expression pattern, the ESCC tissues showed significantly lower MEIS1 immunoreactivity than their adjacent nontumoral margin (P < 0.05). Only 1 out of 27 (3.7 %) tumor samples showed high MEIS1 protein expression, while 8 of 27 (29.6 %) adjacent nontumoral samples had similar high MEIS1 expression. Figure 2 shows representative images of tumor tissue (panel a) and matched adjacent nontumoral (panel b) with low and high MEIS1 immunostaining, respectively.

Fig. 2
figure 2

MEIS1 protein detection in ESCC and adjacent nontumoral tissue by immunohistochemistry. Representative pictures are shown for ESCC (a) compared to matched adjacent nontumoral tissue (b). c, d H&E staining of slides representative for a and b, respectively

Full size image

Downregulation of MEIS1 expression is correlated to metastasis, lymph node involvement, and tumor staging 

To assess potential clinical consequences of MEIS1 downregulation in ESCC tumor samples, we examined the correlation between MEIS1 mRNA/protein expression and different clinically relevant tumor parameters in our ESCC cohort. Low MEIS1 mRNA expression was significantly correlated with different indices of poor prognosis: tumor metastasis (P = 0.027) and lymph node involvement (P = 0.004). Interestingly, 84.2 % (16 out of 19) of samples with low MEIS1 expression showed invasion of tumor cells to adventitia (stage T3). In patients with low stage (stages I/II) ESCC, low MEIS1 mRNA expression was significantly correlated with poor tumor differentiation (P = 0.002). At MEIS1 protein level, we observed a significant correlation between MEIS1 protein expression in tumor samples and the number of involved lymph node (P = 0.019). Furthermore, MEIS1 protein expression was significantly correlated to lymph node involvement (P = 0.048) and high tumor stage (stages III/IV, P = 0.030). There was no other significant association between the level of MEIS1 mRNA or protein expression and clinical data (Table 1).

MEIS1 underexpression is not due to promoter methylation

To assess the role of promoter hypermethylation in the decreased MEIS1 expression in ESCC tumor samples, we amplified and analyzed MEIS1 gene promoter sequences from ESCC samples with low MEIS1 expression (6 samples selected arbitrarily) for probable methylated CpG islands using the UCSC Genome Browser (http://genome.ucsc.edu). A total of 58 CpG sites exist within the 1320-bp region upstream of the MEIS1 transcription start site. Two distinct segments of the promoter containing 23 CpG sites were selected for methylation analysis. Genomic DNA was isolated, subjected to methylation-specific (MSP) PCR, and cloned (Fig. 3). For every tumor sample, 10 separate clones were selected and sequenced. HL-60 cell line DNA was used as a positive control for methylation of these CpG’s, as based on previous work [35]. The results indicated that the promoter was not significantly more often methylated in ESCC samples with low MEIS1 expression than in matched adjacent nontumoral tissue samples.

Fig. 3
figure 3

MEIS1 promoter methylation status in ESCC. Bisulfite sequencing results of clonal MEIS1 promoter MSP-PCR products. Each horizontal line represents a different tumor sample. The CpG di-nucleotides are represented by circles, with open and closed circle for unmethylatedand methylated CpG, respectively. Shown are two representative tumor samples compared with their matched adjacent nontumoral tissue. HL60 cell line DNA was used as a positive control for methylation of these CpG’s as based on previous work [35]

Full size image

Knocking down epigenetic factor EZH2 to assess its effect on MEIS1 expression

The absence of MEIS1 promoter CpG island methylation in cells with low MEIS1 expression inspired us to modulate EZH2 expression and assess its effect on MEIS1 regulation in KYSE 30 cells. A retroviral EZH2 shRNA construct efficiently silenced EZH2 expression as shown by qRT-PCR. Additional qRT-PCR analysis showed that MEIS1 expression increased after EZH2 silencing in KYSE 30, to more than 2-fold (Fig. 4).

Fig. 4
figure 4

EZH2 silencing mediated by EZH2-specific retrovirus shRNA in KYSE 30 cells. MEIS1 is overexpressed (b) as result of EZH2 underexpression (a) confirmed by qRT-PCR, with GAPDH as the reference gene. The experiment was performed in triplicate. Shown are the mean values, the error bars represent the SD. A two-sided t test was used to analyze the statistical difference in mRNA expression. The difference was significant, P = 0.04

Full size image

Expression correlation of MEIS1 and SOX2 in ESCC

To analyze MEIS1 role as a stemness factor in ESCC, we performed SOX2 expression analysis in our ESCC cohort (Table 1), we found that SOX2 showed higher expression in ESCC than in matched adjacent nontumoral tissue (Supplemental Fig. 1). This led to a significant negative correlation between MEIS1 and SOX2 mRNA expression (P = 0.011, R = −0.790, Pearson test). To prove an actual, dynamic relationship between MEIS1 and SOX2 expression in ESCC cells, we performed lentiviral MEIS1 knockdown in KYSE-30 cells. The MEIS1 knockdown was confirmed on mRNA and protein levels by qRT-PCR and Western blot analysis (Fig. 5, panels a and b, respectively). Interestingly, we found that MEIS1 knockdown resulted in significant overexpression of SOX2 both at both mRNA (Fig. 5a, with over 3-fold SOX2 overexpression) and levels (Fig. 5b). Together, we take these results as a strong indication that MEIS1 is involved in ESCC cell differentiation, possibly in part through regulation of SOX2.

Fig. 5
figure 5

MEIS1 silencing mediated by MEIS1-specific lentivirus shRNA in KYSE-30 cells. MEIS1 knockdown was confirmed by a qRT-PCR and b Western blot. a Lower MEIS1 expression resulted in SOX2 mRNA overexpression, as established by qRT-PCR, with GAPDH as the reference gene. In a, the results of two separate qRT-PCR experiments are shown, with a, c showing MEIS1 and b, d SOX2 expression for the first and second experiment, respectively. The experiment was performed in triplicate, and repeated three times. Shown are the mean values, the error bars represent the SD. A two-sided t test was used to assess the statistical difference in mRNA expression. The difference was significant, P = 0.037. b MEIS1 knockdown and concomitant SOX2 overexpression was also confirmed on protein level using Western blot analysis. To confirm equal loading of the gel, β-actin was used as a reference

Full size image

Discussion

In this study, we found that MEIS1 expression is inversely correlated to metastasis, lymph node involvement, and tumor staging in ESCC (Figs. 1 and 2). In addition, we provided evidence that MEIS1 downregulation during ESCC development is caused by EZH2, and not by methylation of CpG islands (Fig. 3). Last, we determined an inverse correlation between MEIS1 and SOX2 in ESCC tumor samples and showed that MEIS1 knockdown led to SOX2 overexpression in an ESCC cell line (Figs. 4 and 5).

MEIS1 is a developmentally conserved member of TALE family and HOX gene clusters. Although many studies have focused on the function of MEIS1 as cofactor of different transcription machineries, its exclusive role as an independent transcription factor remains to be determined. Recent evidence has demonstrated the role of MEIS1 in cancer stem cells (CSCs), self-renewal in myeloid/lymphoid or mixed lineage leukemia (MLL), and potential rate limiting determinant in leukemia stem cell (LSC) [12, 36]. K. Okumura et al. have shown two roles for MEIS1 in epidermis: regulator of stem cells in normal tissues and as proto-oncogenic in skin tumorigenesis [37]. The oncogenic role of stem cell factor MEIS1 has been extensively determined in hematopoietic disorders. In other malignancies, including neuroblastoma [38], ovarian cancer [39], and Wilms tumor [11], high expression of MEIS1 has potential oncogenic properties with direct or indirect effects on the tumor cell growth and resistance to chemotherapy. In spite of these published MEIS1 oncogenic roles, we found downregulation of MEIS1 in ESCC compared to adjacent nontumoral tissue both on mRNA and protein level. In line with our finding in ESCC, Crist et al. showed downregulation of a specific isoform of MEIS1 in colon cancer [40]. Furthermore, it has been shown that low level of MEIS1 expression is correlated to poor prognosis in prostate cancer [41].

Upstream regulators of MEIS1 and its ensuing promoter methylation state were first studied in leukemia. Xiang et al. revealed that ELF1 is an important positive regulator of MEIS1 expression in K562 erythroleukemia cells [35]; moreover, Lasa et al. showed that MEIS1 expression is downregulated through promoter hypermethylation in AML1-ETO acute myeloid leukemias [42]. Our data showed that MEIS1 gene silencing might be caused by DNA methylation-independent mechanism (Fig. 3). Kondo et al. have described that downregulation of genes involved in prostate cancer progression can occur through H3K27me3 by EZH2 [43]. MEIS1 was recently identified as a target of Polycomb genes in bladder cancer [44].

The crosstalk between MEIS1 and core pluripotency circuit genes, including OCT4, SOX2, NANOG, and KLF4/5, has been identified in developmental and neurogenesis studies. Yamada T et al. showed a reciprocal relationship between MEIS1 and OCT4 expression in neural differentiation and observed induction of SOX2 by ectopic expression of MEIS1 [15]. Tucker and others demonstrated that olfactory epithelium precursors have slowly dividing lateral precursors that are regulated by antagonistic expression of SOX2 and MEIS1 [14].

Regarding context-dependent mechanisms for regulation of gene expression (Fig. 6), an inverse correlation between MEIS1 and SOX2 both in vivo in ESCC tumor samples and in vitro in an ESCC cell line would propose a crosstalk between MEIS1 and SOX2, where MEIS1 may suppress SOX2 gene expression, leading to tumor cell differentiation in ESCC. Altogether, our finding, regarding this correlation between MEIS1 and SOX2, established a novel important point for designing a model network between these genes in different ESCC cell lines in the near future.

Fig. 6
figure 6

Schematic drawing for MEIS1 and SOX2 expression in different cancers, in relation to differentiation status. Poorly differentiated cells with low MEIS1 expression (e.g., ESCC MEIS1-knockdown cells or prostate cancer [41] along with SOX2 overexpression may help CSCs to maintain a self-renewal and stemness state). In contrast, in well-differentiated cancer cells such as hematopoietic malignancies [36], high MEIS1 and low SOX2 expression could cause CSC maintenance and cancer progression. Similarly in neuroblastoma, high MEIS1 expression causes cancer

Full size image

In conclusion, the inverse correlation of MEIS1 with metastasis, tumor staging, and the role of EZH2 in methylation, together with its correlation with stemness factor SOX2 expression, led us to predict cancer stemness properties for MEIS1 in ESCC. These concepts will require deeper analysis.