Identification and validation of obesity-related gene LEP methylation as a prognostic indicator in patients with acute myeloid leukemia

Background Obesity confers enhanced risk for multiple diseases including cancer. The DNA methylation alterations in obesity-related genes have been implicated in several human solid tumors. However, the underlying role and clinical implication of DNA methylation of obesity-related genes in acute myeloid leukemia (AML) has yet to be elucidated. Results In the discovery stage, we identified that DNA methylation-associated LEP expression was correlated with prognosis among obesity-related genes from the databases of The Cancer Genome Atlas. In the validation stage, we verified that LEP hypermethylation was a frequent event in AML by both targeted bisulfite sequencing and real-time quantitative methylation-specific PCR. Moreover, LEP hypermethylation, correlated with reduced LEP expression, was found to be associated with higher bone marrow blasts, lower platelets, and lower complete remission (CR) rate in AML. Importantly, survival analysis showed that LEP hypermethylation was significantly associated with shorter overall survival (OS) in AML. Moreover, multivariate analysis disclosed that LEP hypermethylation was an independent risk factor affecting CR and OS among non-M3 AML. By clinical and bioinformatics analysis, LEP may be also regulated by miR-517a/b expression in AML. Conclusions Our findings indicated that the obesity-related gene LEP methylation is associated with LEP inactivation, and acts as an independent prognostic predictor in AML.

the 2017 European LeukemiaNet (ELN) risk stratification by genetics is widely accepted, but there is a practical limitation to the definition of genetic risk, especially in patients falling in the intermediate group [2]. Therefore, additional prognostic factors are needed.
Obesity has been verified as an independent health risk and is significantly correlated with the development of metabolic disorders, including hyperlipidemia, type 2 diabetes mellitus, hypertension, stroke, and cardiovascular disease. Furthermore, strong evidences have proved the links between body mass index (BMI) and various cancers including the most forms of tumor-based cancer and hematological malignancies [3][4][5][6]. A number of studies have showed the increased risk of cancer including leukemia incidence in obese patients [7,8], and excess fat mass is associated with both enhanced incidence and lower survival for pediatric leukemia [9]. Moreover, obesity independently conferred poor prognosis in AML [10]. Yan et al. [11] revealed that Fatty acid-binding protein 4 mechanistically linked obesity with aggressive AML by enhancing aberrant DNA methylation. To date, various genes such as LEP, LEPR, NPY, ADIPOQ, FTO, MC4R, PCSK1, and POMC are implicated and have a direct role in obesity [12].
To the best of our knowledge, epigenetic alterations have been suggested as a molecular mechanism mediating gene expression, and also described as a potential early cancer-related biomarker with strategies for diagnostic, prognosis or cancer screening procedures being developed [13,14]. To date, epigenetic mechanisms including DNA methylation and aberrant microRNAs (miRNAs) expression involving in obesity-related genes have been reported especially in human cancers with prognostic significance [12,15]. However, the underlying role and clinical implication of DNA methylation of obesity-related genes in AML has yet to be elucidated.

Patients and samples
The first cohort of 200 AML patients from The Cancer Genome Atlas (TCGA) databases included in this study was used in the discovery stage for the identification of prognostic methylation-related genes [16]. Among the cohort, there are 173 cases with expression data whereas 194 patients with methylation data [17,18]. In addition, DiseaseMeth version 2.0 (http://bio-bigda ta.hrbmu .edu. cn/disea semet h/analy ze.html) was applied to compare the methylation difference between these AML patients with controls.
A second cohort of 25 healthy donors and 111 de novo AML patients treated at the Affiliated People's Hospital of Jiangsu University were also enrolled, and used in the validation stage for targeted bisulfite sequencing. In addition, expanded samples of the sequencing cohort, including 172 AML patients (161 de novo AML and 11 MDS-derived AML) and 46 healthy donors, were used in the validation stage for real-time quantitative methylation-specific PCR (qMSP). AML patients were diagnosed and classified according to the 2016 World Health Organization (WHO) criteria [19]. Treatment regimens for AML patients were as reported [20,21]. The detection of gene mutations in this study was described as our previous reports [20,21]. After informed consents were obtained from all participants, bone marrow (BM) was collected at diagnosed time, and were separated to obtain BM mononuclear cells (BMMNCs) by density-gradient centrifugation using Lymphocyte Separation Medium (Solarbio, Beijing, China) [20,21]. The current study was approved by the Ethics Committee of Affiliated People's Hospital of Jiangsu University.

Targeted bisulfite sequencing
The target gene methylation was detected by Targeted bisulfite sequencing-MethylTarget, which was performed in Genesky Biotechnologies Inc. (Shanghai, China) as our previous investigations [22,23]. The primers used for LEP were shown in Additional file 1: Table S1.

Reverse transcription and qPCR
Reverse transcription was carried out as our previous studies [20,21]. The detection of LEP mRNA expression was performed by real-time quantitative PCR (qPCR) using AceQ qPCR SYBR Green Master Mix (Vazyme, Piscataway, NJ). The reference gene ABL1 mRNA, examined by 2 × SYBR Green PCR Mix (Multisciences, Hangzhou, China), was detected to calculate the abundance of LEP mRNA expression. The qPCR primers used for LEP expression detection were listed in Additional file 1: Table S1. Relative LEP mRNA expression was calculated using 2 −∆∆CT method.

Bisulfite modification and qMSP
Genomic DNA was bisulfite converted as our previous reports [21,22]. The detection of LEP methylation level was evaluated by qMSP with primers shown in Additional file 1: Table S1. The reference gene ALU methylation level was also detected. Relative LEP methylation level was calculated using 2 −∆∆CT method.

Statistical analyses
SPSS 20.0 and GraphPad Prism 5.0 were conducted to perform statistical analyses. Mann-Whitney's U/ Kruskal-Wallis followed by Dunn's post hoc test and Pearson's χ 2 /Fisher's exact test were used for the comparison of continuous and categorical variables, respectively. Correlation analysis between LEP methylation and methylation/expression was performed by Spearman test. The Receiver operating characteristic (ROC) curve and area under the ROC curve (AUC) were used to evaluate LEP methylation level in distinguishing AML from controls. Complete remission (CR) was evaluated after 1-2 course of chemotherapy. Overall survival (OS) and leukemia free survival (LFS) were defined as previous report [20]. Survival analysis regarding the effect of LEP methylation on OS and LFS was analyzed by Kaplan-Meier analysis and Cox regression analysis (univariate and multivariate). A two-sided P less than 0.05 was seen as statistically significant.

Identification of prognostically obesity-related genes correlated with DNA methylation in AML
We first used TCGA data to identify the prognostically obesity-related genes including LEP, LEPR, NPY, ADI-POQ, FTO, MC4R, PCSK1, and POMC in AML. Prognostic value of these genes was analyzed in two groups divided by the median expression level of each gene respectively. In total AML and cytogenetically normal AML (CN-AML) patients, Kaplan-Meier analysis showed that only LEP expression was positively associated with OS (P = 0.013 and 0.007, Fig. 1a) and LFS (P = 0.025 and 0.062, Additional file 2: Figure S1), suggesting the prognostic effect of LEP expression in AML.
DNA methylation plays a crucial role in regulating gene expression. We next investigated the association between these obesity-related gene expression and methylation in AML. Among the eight genes, methylation data was available for LEP, LEPR, FTO, PCSK1, and POMC. Significantly negative association was shown in LEP 1b). These data suggested LEP, LEPR, FTO, PCSK1, and POMC genes methylation may play main roles in regulating gene expression during leukemogenesis, while LEP showed a very weak association. Moreover, we further identified that LEP promoter CpG island was hypermethylated in AML by using the DiseaseMeth version 2.0 (P < 0.001, Fig. 1c).

Abnormal LEP promoter methylation in AML by targeted bisulfite sequencing
To validate the methylation pattern of LEP in AML, we analyzed CpG island methylation located at the LEP promoter region (Fig. 2a) by targeted bisulfite sequencing in BMMNCs samples of 25 controls and 111 de novo AML patients. The sequencing mean bait coverage attached 1694 ×, and Q30 was 75.56% [22,23]. The targeted sequencing results exhibited that the level of LEP methylation in AML patients was markedly higher than that in controls (P < 0.001, Fig. 2b).

Further confirmation of LEP methylation in a larger cohort of AML by qMSP
In order to explore whether LEP methylation could be helpful utilized in patients diagnosis, prognosis and risk/ treatment assessment, we further expanded the patients samples including 46 controls, 161 primary AML and 11 secondary AML to explore clinical implication of LEP methylation by using a more rapid and convenient methodology-qMSP. The primers for qMSP were designed located inside the sequencing primer (Fig. 2c), and the results analyzed by qMSP results was positively associated with the results by targeted bisulfite sequencing (R = 0.404, P < 0.001, Fig. 2c). In addition, LEP promoter hypermethylation in primary and secondary AML was further confirmed by qMSP (both P < 0.001, Fig. 2d). However, LEP methylation showed no significant Fig. 1 Identification of prognostically obesity-related genes correlated with DNA methylation in AML. a The impact of obesity-related genes expression on overall survival among AML patients from TCGA databases. AML patients were divided into two groups by the median methylation level of each gene respectively. b Correlation between obesity-related genes expression and methylation among AML patients from TCGA databases. The correlation analysis was conducted by Spearman test. c LEP methylation level in AML patients and controls obtained by bioinformatics analysis. LEP promoter CpG island methylation level was obtained through the human disease methylation database DiseaseMeth version 2.0 (http://bio-bigda ta.hrbmu .edu.cn/disea semet h/analy ze.html). TCGA The Cancer Genome Atlas (See figure on next page.) Zhang et al. Clin Epigenet (2021) 13:16 difference between primary and secondary AML (P = 0.680, Fig. 2d). We next detected LEP expression in controls and AML patients with available RNA samples by qPCR. LEP expression was significantly decreased in AML (P < 0.001, Fig. 2e), and was inversely correlated with LEP methylation (R = − 0.338, P = 0.009, n = 59, Spearman test).

Clinical implication of LEP methylation in AML
ROC curve analysis exhibited that LEP promoter methylation may be severed as an underlying biological marker for distinguishing AML from controls with an AUC of 0.803 (95% CI 0.747-0.858, P < 0.001, Fig. 2f ). By the ROC analysis, LEP methylation at the value of 1.011 was set as cutoff point due to the sensitivity was 60.5% and the specificity was 100%. According to the set point, we divided AML patients into two groups to analyze the clinical significance of LEP methylation. No significant differences were found between two groups with regard to age, white blood cells, and hemoglobin (P > 0.05, Table 1). However, LEP hypermethylation tended to be associated with male patients and higher BM blasts (P = 0.057 and 0.064, respectively, Table 1), and significantly correlated with lower platelets (P = 0.046, Table 1). Moreover, there were significant differences between two groups in the distribution of French-American-British (FAB) classifications and karyotypes (P = 0.044 and 0.042, respectively, Table 1). LEP hypermethylation was less frequently occurred in M3/t(15;17) subtypes (P = 0.003 and 0.001, respectively). Moreover, there were no significant associations between LEP hypermethylation and gene mutations besides N/R-RAS mutations with a trend (P = 0.098, Table 1).

LEP methylation was associated with prognosis in AML
Firstly, we revealed the significant association of LEP methylation with CR rate in AML patients. Notably, CR rate in LEP hypermethylated patients was significantly lower than that in LEP non-hypermethylated patients among whole-cohort AML and non-M3 AML (P = 0.011 and 0.049, respectively, Table 1). In CN-AML, we did not observe the significant difference for CR between LEP hypermethylated and non-hypermethylated patients (P = 0.105, Table 1). Since the significant associations of LEP methylation with CR were observed among whole-cohort AML and non-M3 AML, Logistic regression analysis was performed to confirm the effect of LEP methylation on CR. After adjusting for the wellknown prognostic factors, LEP hypermethylation acted as an independent risk factor negatively affecting CR in both whole-cohort AML and non-M3 AML patients (P = 0.017 and 0.015, respectively, Tables 2 and 3). Secondly, we also analyzed the effect of LEP methylation on OS and LFS in AML patients. Kaplan-Meier analysis indicated that LEP hypermethylated patients exhibited shorter OS time than LEP non-hypermethylated patients among total AML, non-M3 AML and CN-AML patients (P = 0.010, 0.050, and 0.028, respectively, Fig. 3a, c, e). For LFS, significant difference was only observed in total AML between two groups (P = 0.030, 0.081, and 0.057, respectively, Fig. 3b, d, f ). Furthermore, by Cox regression analysis, LEP hypermethylation could severe as a prognostic biomarker independently affecting OS among total AML with a trend (P = 0.052, Table 4) and non-M3 AML patients (P = 0.041, Table 5).

MiRNA signatures correlated with LEP in AML
Due to a very weak correlation of LEP expression with LEP methylation in AML patients from both TCGA cohort and validation data, we thought that LEP expression in AML was not only regulated by LEP methylation, and other mechanism also involved such as miRNAs. To gain insights into the molecular signatures associated with LEP in AML, we first compared the transcriptomes of miRNAs expression signatures in lower and higher LEP expression groups (based on the median level of LEP expression) of AML patients from TCGA datasets. A total of 83 differentially expressed miRNAs (included 71 positively correlated and 12 negatively correlated) (FDR < 0.05, P < 0.05, |log2 FC|> 1; Fig. 4a; Additional file 3) were identified between two groups. The negatively correlated miRNAs such as miR-10a was identified to be significantly associated with AML with NPM1 mutation [27], whereas the other genes including miR-582, (See figure on next page.) Fig. 2 Validation and confirmation of LEP methylation in AML. a The genomic coordinates (GC) of LEP promoter region CpG island and primer locations. The panel plots the GC content as a percentage of the total. Each vertical bar in the bottom panel represents the presence of a CpG dinucleotide. Black horizontal lines indicate regions amplified by sequencing primer pairs and qMSP primer pairs. CpGplot (http://embos s.bioin forma tics.nl/cgi-bin/embos s/cpgpl ot) and Methyl Primer Express v1.0 software were used for creating the figure. TSS transcription start site, qMSP real-time quantitative methylation-specific PCR; b LEP methylation level in controls and AML patients detected by targeted bisulfite sequencing. c Correlation between targeted bisulfite sequencing and qMSP results for LEP methylation in AML patients. The correlation analysis was conducted by Spearman test. d LEP methylation level in controls and AML patients examined by qMSP. AML included de novo AML and sAML which indicated MDS-derived AML. e LEP expression level in controls and AML patients. LEP expression level was examined by qPCR. f ROC curve analysis of LEP methylation distinguishing AML from controls Zhang et al. Clin Epigenet (2021) 13:16 miR-517, miR-511, miR-508, miR-518c, miR-520g, and miR-187 were less investigated. Moreover, LEP was identified as a direct target of 69 miRNAs by bioinformatics prediction (Fig. 4b, Additional file 4). Of these miR-NAs, miR-517a/b was shared in both clinical data and bioinformatics prediction, suggesting that LEP may be also regulated by miR-517a/b expression in AML (Fig. 4c).

Discussion
Obesity confers enhanced risk for multiple diseases including cancer, and is increasingly recognized as a growing cause of preventable cancer risk [3][4][5][6]. The DNA methylation alterations in obesity-related genes have been implicated in several human solid tumors [12,15]. Previously, promoter methylation of obesity-related genes including LEP, NPY, and LEPR was involved in tumorigenesis of renal cell carcinoma, and LEPR methylation was associated with prognosis, and predicted renal cell carcinoma recurrence [12,28]. Herein, we for the first time evaluated prognostic value of obesity-related gene expression and methylation in AML. By the identification and validation stage, we finally revealed that LEP methylation, negatively associated with LEP expression, was independently associated clinical outcome in AML. The expression pattern and direct role of LEP remains controversial in AML. Functional studies have showed
In clinics, although no significant difference of serum leptin concentrations were found between de novo AML patients and controls in two previous reports [32,33], two independent investigations by Aref et al. and Bruserud et al. showed that serum leptin levels in AML patients were significantly lower than controls and had negative correlation with percentage of BM blasts and white blood cells [34,35]. In our study, we detected LEP mRNA level in BMMNCs but not in serum of AML patients, and were found to be significantly decreased.
The decreased expression of LEP may be caused by LEP promoter methylation in AML cells. In accordance with the previous study, we also observed LEP hypermethylation was associated with higher percentage of BM blasts and lower platelets. These results suggested DNA methylation-mediated leptin inactivation was a frequent event in AML cells. The reduction of autocrine of leptin in leukemia cells may negatively feedback regulates the increase of paracrine of leptin from adipose tissues into cancer microenvironment to promote leukemogenesis. Accordingly, further functional studies in vivo and in vitro are required to confirm our hypothesis. Besides the DNA methylation, miRNAs expression was also identified to be associated with LEP expression in AML. In this study, we identified that LEP expression may be also regulated by miR-517a/b expression. Although few investigations revealed the miR-517a/b expression pattern in AML, a number of studies have reported the oncogenic role of miR-517a/b in diverse human solid tumors [36][37][38]. These results suggested that multiple factors were involved in regulating LEP expression in AML biology. Obviously, additional studies are required to confirm the direct links of LEP with miR-517a/b by luciferase assay, and the direct role of miR-517a/b in AML needs further functional studies.

Conclusion
Our findings indicated that the obesity-related gene LEP methylation is associated with LEP inactivation, and acts as an independent prognostic predictor in AML.