Expression and prognosis analysis of TET family in acute myeloid leukemia

TET family members (TETs) encode proteins that represent crucial factors in the active DNA demethylation pathway. Evidence has proved that TET2 mutation is associated with leukemogenesis, drug response, and prognosis in acute myeloid leukemia (AML). However, few studies revealed the TETs expression and its clinical significance in AML. We conducted a detailed expression and prognosis analysis of TETs expression in human AML cell lines and patients by using public databases. We observed that TETs expression especially TET2 and TET3 was closely associated with AML among various human cancers. TET1 expression was significantly reduced in AML patients, whereas TET2 and TET3 expression was significantly increased. Kaplan-Meier analysis showed that only TET3 expression was associated with overall survival (OS) and disease-free survival (DFS) among both total AML as well as non-M3 AML, and was confirmed by another independent cohort. Moreover, Cox regression analysis revealed that TET3 expression may act as an independent prognostic factor for OS and DFS in total AML. Interestingly, patients that received hematopoietic stem cell transplantation (HSCT) did not show significantly longer OS and DFS than those who did not receive HSCT in TET3 high-expressed groups; whereas, in TET3 low-expressed groups, patients that accepted HSCT showed significantly longer OS and DFS than those who did not accept HSCT. By bioinformatics analysis, TET3 expression was found positively correlated with tumor suppressor gene including CDKN2B, ZIC2, miR-196a, and negatively correlated with oncogenes such as PAX2 and IL2RA. Our study demonstrated that TETs showed significant expression differences in AML, and TET3 expression acted as a potential prognostic biomarker in AML, which may guide treatment choice between chemotherapy and HSCT.


INTRODUCTION
DNA methylation has contributed to the understanding of the complexities of genomic instability and gene regulation without altering the DNA sequence [1]. Aberrations in DNA methylation status are closely associated with tumor progression and prognosis of patients especially in blood cancers including acute myeloid leukemia (AML) [1,2]. During malignant transformation, CpG islands in the promoter region of numerous genes become hypermethylated, silencing the expression of suppressor genes, and leading to a loss in the control of cell apoptosis, proliferation, and differentiation [1]. Conversely, hypomethylation of oncogenes enhances the tumorigenic potential of normal cells [1]. The process of DNA methylation controlled by several molecules such as DNA methyltransferases (DNMTs) has been well characterized [3,4], but the underlying mechanism of demethylation remains to be elucidated. In recent years, Ten-eleven translocation (TET) proteins have been identified and expand the understanding about mechanisms of DNA demethylation [5].
The TET protein family includes TET1, TET2 and TET3, which can modify 5-methylcytosine (5-mC) by oxidation to 5-hydroxymethylcytosine (5-hmC) and further 5-formylcytosine (5-fC) and 5carboxycytosine (5-caC) [6][7][8]. TET family members (TETs) were dysregulated in multiple malignances, and loss-of-function mutations or decreased expression of TETs inhibited the DNA demethylation pathway, which prevents the removal of 5mC from genomic DNA [5]. Functional studies have revealed the direct role of TET2 in blood cancers especially in AML. Cimmino et al reported that restoration of TET2 reversed aberrant hematopoietic stem and progenitor cell self-renewal in vitro and in vivo, and suppressed human leukemic colony formation and leukemia progression of primary human leukemia patientderived xenografts [9]. Rasmussen et al indicated that loss of TET2 in hematopoietic cells lead to DNA hypermethylation of active enhancers and induction of leukemogenesis [10]. TET2 mutations frequently occur in AML, myelodysplastic syndromes (MDS) and chronic myelomonocytic leukemia (CMML), whereas TET1 and TET3 mutations rarely happen [11,12]. Moreover, TET2 mutations were important prognostic factors in AML and also predicted response to hypomethylating agents in MDS patients [13]. However, few studies investigated TETs expression and its clinical significance in AML [14,15]. Herein, we determined the clinical significance of TETs expression in AML among The Cancer Genome Atlas (TCGA) databases.

TETs expression associated with AML among human cancer cell lines
By assembling the Cancer Cell Line Encyclopedia (CCLE), we found that TETs expression especially TET2 and TET3 was highly expressed in AML cell lines among 40 types of human cancer cell lines ( Figure 1A-1C). Moreover, The Human Protein Atlas (HPA) also presented that TET2 and TET3 expression was also highly associated with myeloid cell lines ( Figure 1D-1F). The detailed comparison of TETs expression in AML cell lines was assessed by using the European Bioinformatics Institute (EMBL-EBI) website ( Figure 1G-1I). In addition, TET1/2/3 mutations in human cancer cell lines were given in Supplementary Table 1.

TETs expression associated with AML patients among human cancers
We further evaluated TETs expression in AML patients by using the Gene Expression Profiling Interactive Analysis (GEPIA) dataset including TCGA and the Genotype-Tissue Expression (GTEx) projects. Aberrant expression of all TETs members was only observed in AML patients among 33 types of human cancers (Figure 2A-2C). TET1 expression was significantly reduced in AML patients, whereas TET2 and TET3 expression was significantly increased in AML patients ( Figure 2D-2F). Moreover, TET1 expression did not show a significant correlation with TET2/TET3 expression in AML patients, whereas TET2 expression was positively correlated with TET3 expression in AML patients ( Figure 2G-2I). In addition, TET1 and TET3 mutations were identified in none of these AML patients, whereas TET2 mutation was identified in 8.5% (17/200) of these AML patients.

Prognostic value of TETs expression in AML
In order to evaluate the prognostic value of TETs expression in AML, we further divided these patients into two groups based on median level of TET1/2/3 transcript respectively (TET1 low vs. TET1 high ; TET2 low vs. TET2 high ; TET3 low vs. TET3 high ). Based on Kaplan-Meier analysis, we did not observe the significant associations of TET1 and TET2 expression with overall survival (OS) and disease-free survival (DFS) among both total AML and non-M3 AML ( Figure 3). However, TET3 high patients showed markedly longer OS and DFS than TET3 low patients among total AML (Figure 3, P=0.018 and 0.019, respectively). Moreover, if French-American-British (FAB)-M3 patients were excluded, patients with high expression of TET3 also had significantly longer OS and DFS than those with low expression of TET3 (Figure 3, P=0.006 and 0.007, respectively). We next determined the prognostic effect of TET3 expression in AML by using Cox regression analysis. Both univariate and multivariate analysis showed that TET3 expression may act as an independent prognostic factor for OS and DFS in total AML ( Table 1, P=0.011 and 0.026, respectively) and non-M3 AML ( Table 2, P=0.038 and 0.026, respectively).
In addition, the positive impact of high TET3 expression on OS in cytogenetically normal AML (CN-AML) patients was also validated by Gene Expression Omnibus (GEO) data (GSE12417) via online web tool Genomicscape ( Figure 4A-4D).

Association between TET3 expression and clinical/molecular characteristics
Due to the significant association of TET3 expression with AML prognosis, we next analyzed the clinical relevance of TET3 expression with clinical/molecular characteristics in AML. As presented in Table 3. There were no significant differences between TET3 high and TET3 low groups in sex, age, white blood cells (WBC), bone marrow (BM)/peripheral blood (PB) blasts, and the distributions of cytogenetics (P>0.05). Significant difference was observed between two groups in the distribution of FAB subtypes (P=0.009). TET3 high patients was frequently occurred in FAB-M1/M4 (P=0.083 and 0.022, respectively), and less frequently occurred in FAB-M0 (P=0.016). Among common gene mutations, high expression of TET3 was associated with FLT3 wildtype and NRAS mutation (P=0.018 and 0.018, respectively). No significant differences were found between TET3 expression with other gene mutations (P>0.05). Since TET2 mutation is frequent molecular event in AML, we further analyzed the relationship between TET2 mutation and TET1/2/3 expression in AML patients. As presented in Supplementary Figure 1, no significant differences were found between TET2 mutation (TET2 mu ) and TET2 wild-type (TET2 WT ) regarding TET1/2/3 expression (P>0.05).

TET3 expression may guide treatment choice between chemotherapy and HSCT
Because low expression of TET3 predicted poor clinical outcome in AML, we intended to investigate whether patients with low expression of TET3 could benefit from hematopoietic stem cell transplantation (HSCT). We compared OS and DFS between patients with and without HSCT among both TET3 high and TET3 low groups. In TET3 high groups, although patients who received HSCT presented longer OS and DFS compared with patients who did not receive HSCT among both total AML ( Figure 5A and 5B, P=0.052  and 0.221, respectively) and non-M3-AML ( Figure  5C and 5D, P=0.021 and 0.128, respectively), the P did not attach statistical significance especially for DFS. However, in TET3 low groups, patients who accepted HSCT showed significantly longer OS and DFS than patients who did not accept HSCT among both total AML ( Figure 5E and 5F, P=0.003 and 0.005, respectively) and non-M3-AML ( Figure 5G and 5H, P<0.001 and 0.001, respectively).

Correlations between TET3 expression and molecular signature
To gain insights into the biological function of TET3 in AML, we first compared the transcriptomes of TET3 high and TET3 low groups. A total of 464 differentially expressed genes were identified (FDR<0.05, |log2 FC|>1.5; Figure 6A and 6B; Supplementary Table 2), in which 300 genes were positively correlated with TET3 expression, and 164 were negatively correlated. Positively correlated genes such as CDKN2B and ZIC2 were reported to have anti-leukemia effects [16,17]. Among the negatively associated genes, several genes including PAX2, IL2RA, SOX11, and PAK7 played as oncogenes in leukemia [18][19][20][21]. Furthermore, the Gene Ontology analysis was also showed in Figure 6C.

DISCUSSION
Aberrant promoter methylation, an important hallmark of cancer cells, is considered as a major mechanism underlying the activation/inactivation of tumor-related genes [1]. In addition to DNMTs, TET gene family encodes proteins that represent crucial factors in the active DNA demethylation pathway [3][4][5]. A loss-offunction mutation in the TET2 gene is associated with leukemogenesis, drug response, and treatment outcome [11]. However, few studies investigated TETs expression and its clinical significance in AML [14,15]. Herein, we systemically explored the TETs expression and its clinical significance in AML, and we hope that our findings could provide new insight into AML biology, improve treatment designs, and enhance the accuracy of prognosis for patients with AML. In this study, we showed that TETs expression showed differentially expressed in AML, which indicated different role of TETs during AML pathogenesis. In solid tumors, a number of studies showed the direct role of TETs in cancer biology. For example, two studies have showed that TET1 was a tumor suppressor gene that inhibited colon cancer growth by derepressing inhibitors of the WNT pathway [23,24]. Xu et al AGING  disclosed that tumor suppressive role of TET2 promoted cancer immunity and immunotherapy efficacy [25]. Moreover, TET2 controlled chemoresistant slow-cycling cancer cell survival and tumor recurrence [26]. Cui et al demonstrated that TET3 as a potential tumor suppressor induced by the nuclear receptor TLX to regulate the growth and self-renewal in glioblastoma stem cells [27]. Moreover, several tumor suppressors, including BTG2, TUSC1, BAK1, LATS2, FZD6 and PPP2R1B, were regarded as common targets of TET3 [27]. Additionally, TET3 expression was decreased in ovarian cancer tissues, acted as a suppressor of ovarian cancer by demethylating miR-30d precursor gene promoter to block TGF-β1-induced epithelial-mesenchymal transition [28]. In our study, we showed that TET1 expression was significantly decreased in AML, whereas TET2 and TET3 expression was significantly increased in AML. Notably, we did not observe the direct association of TET3 with these factors, and found that several tumor suppressor genes (CDKN2B, ZIC2, and miR-196a) and oncogenes (PAX2, IL2RA, SOX11, and PAK7) were associated with TET3 in AML biology [16][17][18][19][20][21][22]. Moreover, these genes were important factors as cellular component or involving in many crucial biological processes contributing to cancer development.
Lastly, TET3 was differently expressed among the distributions of FAB subtypes in AML. These results suggested that the biological network of TETs in cancer was dependent on cancer type and stage specific.
Although previous studies showed the significant associations of TET1 and TET2 expression with AML prognosis [14,15], herein, we only observed that TET3 expression acted as an independent prognostic factor in AML, and could be overcame by HSCT. It was very interesting that TET3 expression was increased in AML, and its high expression showed a positive effect in AML. Possible reason was that TET3 expression may play a different role between cancer occurrence and development, and further functional studies are needed to explore the underlying mechanism in AML development. The expression pattern and clinical significance of TET3 have been determined in several human cancers. Several studies revealed that high expression of TET3 was revealed in renal cell carcinoma as well as endometrial cancers, and high mRNA levels of TET3 were independent predictors of poor outcome in renal cell carcinoma patients [29,30]; whereas, several other investigations reported that TET3 was low-expressed in diverse human cancers. For instance, Bronowicka-Kłys et al showed that TET3 transcript levels were lower in stage III samples of cervical cancer [31]. Moreover, TET3 mRNA was decreased in chronic lymphocytic leukemia cells compared with healthy B cells [32]. In colorectal cancer, reduced transcript level of TET3 was observed in cancerous tissue compared with their histopathologically unchanged counterparts [33]. In addition, Misawa et al reported that TET3 methylation was highly associated with poor survival in T1 and T2 tumor stages of oropharyngeal cancer and oral cancer patients [34]. All these results further indicated that the role of TET3 in diverse human cancers was specific among different cancer types. In summary, our study demonstrated that TETs showed significant expression differences in AML, and TET3 expression acted as a potential prognostic biomarker in AML, which may guide treatment choice between chemotherapy and HSCT.

CCLE, HPA, and EMBL-EBI dataset
Firstly, TETs expression in human cancer cell lines is assessed by the CCLE dataset (https://www. broadinstitute.org/ccle), which provides public access to genomic data, analysis, and visualization for about 1000 cell lines [35]. Secondly, we also used The HPA dataset (https://www.proteinatlas.org/) to verify TETs expression in human cancer cell lines [36]. Lastly, TETs expression in AML cell lines is verified by the EMBL-EBI dataset (https://www.ebi.ac.uk), which has provided free and open access to a range of bioinformatics applications for sequence analysis since 1998 [37].

Patients from TCGA and GEO
A total of 173 AML patients with available TETs expression data from TCGA (https://cancergenome .nih.gov/ and http://www.cbioportal.org/) were identified and included in this study [39]. Clinical and molecular characteristics were obtained, including, age, sex, WBC counts, PB blasts, BM blasts, FAB subtypes, and the frequencies of genetic mutations as presented in Table 3. After induction chemotherapy, consolidation treatment included chemotherapy (100 patients received) and HSCT (73 patients accepted).
In addition, two cohorts of 162 and 78 CN-AML patients from GEO data (GSE12417) were also included. The online web tool Genomicscape (http://genomicscape.com/microarray/survival.php) was applied to validate the prognostic value of TETs expression among CN-AML patients.

Bioinformatics analysis
The details for the identification of microRNAs targeting TET3 were reported as our previous study [40].

Statistical analysis
Statistical analysis and figures creation were performed on SPSS 22.0 software. Mann-Whitney's U test was used for the comparison of continuous variables, whereas Pearson Chi-square analysis or Fisher exact test was applied for the comparison of categorical variables. The prognostic effect of TETs expression on DFS and OS was evaluated analyzed though Kaplan-Meier analysis and Cox regression analysis. The twotailed P value < 0.05 in all statistical analysis was defined as statistically significant.

Ethical approval
All procedures performed in studies involving human participants were approved by the Ethics Committee of the Affiliated People's Hospital of Jiangsu University and the Washington University Human Studies Committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all patients included in this study.

AUTHOR CONTRIBUTIONS
Jingdong Zhou conceived and designed the study; Tingjuan Zhang, Yangli Zhao and Yangjing Zhao analyzed the data; Jingdong Zhou wrote the paper. All authors read and approved the final manuscript.

CONFLICTS OF INTEREST
The authors declare that they have no conflict of interest.

FUNDING
The work was supported by National Natural Science foundation of China (81900166), Zhenjiang Clinical Research Center of Hematology (SS2018009).