Abstract
Acute myeloid leukemia is characterized by its high biological and clinical heterogeneity, which represents an important barrier for a precise disease classification and accurate therapy. While epigenetic aberrations play a pivotal role in acute myeloid leukemia pathophysiology, molecular signatures such as change in the DNA methylation patterns and genetic mutations in enzymes needed to the methylation process can also be helpful for classifying acute myeloid leukemia. Our study aims to unveil the relevance of DNMT3A and TET2 genes in global and specific methylation patterns in acute myeloid leukemia. Peripheral blood samples from 110 untreated patients with acute myeloid leukemia and 15 healthy control individuals were collected. Global 5-methylcytosine and 5-hydroxymethylcytosine in genomic DNA from peripheral blood leukocytes were measured by using the MethylFlashTM Quantification kits. DNMT3A and TET2 expression levels were evaluated by real-time quantitative polymerase chain reaction. The R882A hotspot of DNMT3A and exons 6–10 of TET2 were amplified by polymerase chain reaction and sequenced using the Sanger method. Methylation patterns of 16 gene promoters were evaluated by pyrosequencing after treating DNA with sodium bisulfite, and their transcriptional products were measured by real-time quantitative polymerase chain reaction.Here, we demonstrate altered levels of 5-methylcytosine and 5-hydroxymethylcytosine and highly variable transcript levels of DNMT3A and TET2 in peripheral blood leukocytes from acute myeloid leukemia patients. We found a mutation prevalence of 2.7% for DNMT3A and 11.8% for TET2 in the Mexican population with this disease. The average overall survival of acute myeloid leukemia patients with DNMT3A mutations was only 4 months. In addition, we showed that mutations in DNMT3A and TET2 may cause irregular DNA methylation patterns and transcriptional expression levels in 16 genes known to be involved in acute myeloid leukemia pathogenesis. Our findings suggest that alterations in DNMT3A and TET2 may be associated with acute myeloid leukemia prognosis. Furthermore, alterations in these enzymes affect normal methylation patterns in acute myeloid leukemia– specific genes, which in turn, may influence patient survival.
Introduction
Acute myeloid leukemia (AML) is a hematologic neoplasia characterized for its high biological and clinical heterogeneity, which is still an important barrier for the development of a precise classification and accurate therapy. Clinical observations suggest that cytogenetic alterations such as chromosomal translocations, mutations, and more recently specific epigenetic alterations play an important role in determining the origin of a particular AML and assess disease prognosis.
The importance of epigenetic changes in the process of leukemia cell transformation is well known.1,2 DNA methylation is one of the best known and deeply studied epigenetic mechanisms, which can alter gene expression by creating or eliminating protein binding sites in the DNA. 3 For instance, some of these genes could be tumor suppressors, and their silencing could lead to oncogene activation.4,5 Through this mechanism, DNA methylation affects not only leukemia development but also its progression and remission. 6
Methylation is a dynamic and complex process that involves numerous proteins and molecules. The concentrations of 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC) are considered as indicators of global methylation and demethylation status, respectively. 7 Alternatively, the status of enzymes responsible for modifying cytosine residues, such as DNMT3A for 5mC and TET2 for 5hmC, has also been associated with global methylation patterns. Interestingly, in recent years, changes in global 5mC and 5hmC patterns, as well as gene mutations or altered messenger RNA (mRNA) levels of DNMT3A and TET2, have been reported in different types of cancer.8,9
In the this work, in patients with AML, we evaluated the concentrations of 5mC and 5hmC, mutations present in DNMT3A and TET2 genes, and the activity of promoters whose transcription is mediated by DNMT3A and TET2. We report a possible correlation between AML patient survival and the activities of DNMT3A and TET2, via alterations in methylation patterns that have a clear effect on the activity of gene-specific promoters.
Material and methods
Human tissue samples
Peripheral blood (PB) samples were collected from 110 patients with AML during routine clinical evaluation to diagnose the disease while receiving services at the Hospital General Centro Médico Nacional, “La Raza” (Hematology Service) or at the Hospital General de México, “Dr Eduardo Liceaga” (Hematology Service). Only individuals diagnosed de novo were included in this population, excluding individuals with secondary AML and myelodysplastic syndrome (MDS). PB from 15 healthy control individuals was obtained from the Hospital General de México “Dr Eduardo Liceaga” (Blood Bank). Ethylenediaminetetraacetic acid (EDTA)-treated PB was centrifuged through a Ficoll–Histopaque (Sigma-Aldrich) gradient following the manufacturer’s instructions to collect leukocyte-enriched plasma (peripheral blood leukocytes (PBL)).
Global DNA methylation and hydroxy-methylation assays
Genomic DNA from leukocytes was isolated using the standard phenol–chloroform extraction method. The concentrations of 5mC and 5hmC were determined using the MethylFlash Methylated DNA 5-mC Quantification Kit (Colorimetric) (EpiGentek) and MethylFlash Hydroxymethylated DNA 5-hmC Quantification Kit (Colorimetric) (EpiGentek), according to the manufacturer’s instructions.
Total RNA isolation and real-time quantitative polymerase chain reaction
RNA was isolated from leukocytes using TRIzol reagent (Invitrogen), following the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized through reverse transcription using 2 µg of total RNA, oligo (dT) 15 primer (Promega), and M-MLV Reverse Transcriptase (Promega). Quantitative polymerase chain reaction (qPCR) was performed using gene-specific primer pairs for each gene, including β-actin as an internal control (Table 1) and the Maxima SYBR Green qPCR kit (Thermo Fisher Scientific), following the manufacturer’s instructions. The amplification reaction was carried out in a StepOne thermocycler (Applied Biosystems) using the following thermal profile: 94°C for 15 min followed by 40 cycles of 94°C for 45 s and 60°C for 60 s.
Specific primers for qPCR evaluation.
qPCR: quantitative polymerase chain reaction.
TET2 and DNMT3 PCR amplification and sequencing
Genomic DNA from PBL was isolated using the standard phenol–chloroform extraction method. We used gene-specific primer pairs to amplify genomic regions of TET2 and DNMT3A by PCR (Table 2). Sequencing of the PCR products was performed using the Sanger method.
Specific primers for sequencing.
Gene-specific DNA methylation
Bisulfite DNA conversion was done using the EZ DNA Methylation™ Kit (Zymo Research), following the manufacturer’s instructions. Methylation of specific cytosines in CpG dinucleotides was determined by pyrosequencing using the Pyromark Q24 instrument (Qiagen). Oligonucleotides used for PCR and sequencing were obtained from pre-designed assays (Table 3). For the PCR reactions, the PyroMark PCR Kit (Qiagen) was used with HotStarTaq DNA Polymerase (Qiagen) and 20 ng of previously bisulfite-converted DNA. PCR products were verified by capillary electrophoresis using QIAxcel (Qiagen). Results from the pyrosequencing experiments were analyzed using the PyroMark Q24 Software v2.0.8 (Qiagen).
Specific primers for Pyrosequencing.
Bioinformatic analysis of differential DNA methylation
The HeatMap analyses were conducted using the Complex HeatMap software from Bioconductor server and the “R project” by Free Software Foundation Inc.
Statistical analysis
Data were analyzed using GraphPad Prism 6 (GraphPad Software). Analysis of variance (ANOVA) test was used for statistical analysis between multiple groups and confirmed by post hoc test. Statistical significance was defined with a p value <0.05.
Results
5mC and 5hmC levels are altered in AML
To evaluate whether methylation and demethylation dynamics are altered in patients with AML, we measured 5mC and 5hmC in PBL from a total of 15 healthy individuals and compared levels against 110 patients with AML. In healthy individuals, we found little variation in the 5mC concentration ranging just between 2.2 and 3.4 ng of 5mC for every 100 ng of total DNA (ng 5mC/100 ng DNA). In contrast, in individuals with AML, a broad dispersion in 5mC concentration was observed, ranging between 0.5 and 13.2 ng, 5mC/100 ng DNA (Figure 1(a)). Similarly, 5hmC concentrations in healthy individuals range between 2 and 7 ng for every 100 ng of total DNA (ng 5hmC/100 ng DNA; Figure 1(b)), while individuals with AML showed a greater range of variation from 0.2 to 9.5 ng, 5hmC/100 ng DNA.
Total 5mC and 5hmC content and mRNA levels of DNMT3A and TET2 in the individuals included in the study. (a and b) 5mC and 5hmC concentrations in healthy individuals and AML patients. In AML patients, 5mC concentrations range from 1 to 10 ng/100 ng DNA, while for 5hmC, it ranges from 1 to 8 ng/100 ng DNA. (c and d) Expression of DNMT3A and TET2 was highly variable in AML patients, showing from lower levels to those of healthy individuals up to 30 times higher levels.
The transcriptional activity of DNMT3A and TET2 genes is altered in patients with AML
Because DNMT3A and TET2 are enzymes responsible for the methylation status of cytosines, we quantified the mRNA levels of both genes through real-time qPCR (RT-qPCR). In AML patients, DNMT3A expression range from 0.02 to 17-fold, and TET2 from 0.1 to 35-fold the average expression observed in healthy individuals (Figure 1(c) and (d)). Despite this dispersion, in most patients, we found a correlation between the mRNA levels of these genes and the enzymatic activity of their protein products, measured indirectly by 5mC and 5hmC levels. However, several patients with a very high level of DNMT3A mRNA showed a low 5mC levels and some individuals with very low 5hmC levels showed higher or normal levels of TET2 mRNA, thus suggesting a failure in the enzymatic activity of their protein products.
Patients with AML harbor mutations in DNMT3A and TET2 genes
Considering that changes in the DNA sequence of these genes could have an impact on its function in all the AML patients included in the study, we analyzed DNMT3A for R882A hotspot mutations and TET2 for mutations in exons 6–10. We identified 3 individuals with mutations in the hotspot R882A of DNMT3A and 13 individuals with mutations in any of the evaluated TET2 exons.
Mutations in DNMT3A or TET2 genes could explain the variation in 5mC and 5hmC concentrations among AML patients
In order to assess whether DNMT3A and/or TET2 mutations are associated with variation in 5mC and 5hmC concentrations, we grouped AML patients based on mutation status into patients harboring DNMT3A (mDNMT3A) mutations, patients harboring TET2 mutations (mTET2), and patients harboring no mutations (AML noMUT). After organizing 5mC and 5hmC results by concentration, we observed that mDNMT3A patients matched individuals with the lowest 5mC concentration (average 0.7 ng 5mC/100 ng DNA), showing a significant decrease (75%) compared with the average observed in healthy individuals and in AML noMUT patients (2.7 ng; p = 0.001). In contrast, mTET2 patients showed a 5mC average concentration of 8.2 ng 5mC/100 ng DNA, which represents a significant increase of 200% compared with healthy individuals (p = 0.001; Figure 2(a)). In general, patients with AML had a decrease in 5hmC concentration; however, patients with mutations in either DNMT3A or TET2 showed lower levels of 5hmC that reached concentrations below 0.3 ng 5hmC/100 ng DNA in mTET2 patients (p = 0.001) and below 0.5 ng in mDNMT3A patients (p = 0.008; Figure 2(b)).
Total 5mC and 5hmC content and mRNA levels of DNMT3A and TET2 in AML patients considering the mutational status of these genes. (a and b) TET2 mutations lead to an increase while DNMT3A mutation leads to a decrease in 5mC concentrations, as compared with healthy individuals. Concentration of 5hmC decreases in both groups of mutants. (c and d) DNMT3A expression increases when mutations are present in either gene. Meanwhile, TET2 expression increases in individuals with DNMT3A mutation and decreases in individuals where TET2 is mutated, as compared with healthy individuals. All the statistical tests were considered significant (****p = 0.001; **p = 0.005; *p = 0.05).
AML patients with mutations in DNMT3A or TET2 genes present altered levels of the DNMT3A and TET2 mRNA
Data from the transcriptional activity of DNMT3A and TET2 genes showed that DNMT3A mRNA levels were significantly higher in AML patients from both mDNMT3A and mTET2 groups, with an average 6-fold increase (p = 0.001) in mDNMT3A patients and an average 11-fold increase (p = 0.001) in mTET2 patients (Figure 2(c)). Moreover, TET2 mRNA levels were eight times higher in mTET2 patients (p = 0.001) than in mDNMT3A patients which also showed an 80% average decrease in TET2 levels compared with healthy individuals (p = 0.001; Figure 2(d)). Interestingly, in AML noMUT patients, the transcription levels of both DNMT3A and TET2 genes were increased; however, this variation was not statistically significant.
AML patients with DNMT3A and/or TET2 gene mutations show changes in DNA methylation patterns and transcriptional activity of specific gene promoters
In order to determine whether these DNMT3A and TET2 mutations affected the methylation status of specific genes, we selected 16 genes previously reported to have promoters with altered methylation patterns in patients with leukemia and analyzed them in the mDNMT3A (3 patients), mTET2 (12 patients) groups, and 8 healthy individuals, using specific probes and pyrosequencing after treating DNA with bisulfite. A HeatMap was done with the percentage results of the CpG-site-by-CpG-site methylation obtained from these genes in every individual, using “R project” by Free Software Foundation Inc. (http://cran.r-project.org/) as well as the Complex HeatMap pack. The dendrogram generated with these data showed the presence of specific methylation patterns that matched the groups included in this work (Figure 3, top panel); the first cluster generated grouped all 8 healthy individuals included in the analysis; the second cluster grouped 6/8 AML noMUT patients; and the third cluster included the all 3 mDNMT3A patients and 11/12 mTET2 patients.
Methylation and expression patterns in specific genes associated with DNMT3A and TET2 mutation. The hierarchy grouping and HeatMap based either on the methylation status (top panel) or the expression levels (bottom panel) of 19 genes showed in both cases the formation of four clusters on the x-axis, corresponding to healthy individuals, non-mutant AML patients, TET2 mutants, and DNM3A mutants. The blue scale corresponds to methylation percentage, from highest to lowest. The red scale represents expression fold, from higher to lowest.
The relative abundance of the mRNAs of these 16 AML-associated genes was quantified in all the groups through RT-qPCR. The same bioinformatic tools used previously to generate the methylation HeatMap were also used to generate a 16-gene transcription HeatMap level using the RT-qPCR (Figure 3, bottom panel). The generated dendrogram showed three main clusters that also match with the groups included in this work. One cluster included all of the healthy individuals included in the analysis; the second cluster included all three mDNMT3A patients and one mTET2 patient; and the third cluster divided into two smaller groups that included mTET2 patients or AML noMUT patients.
Some of the evaluated genes show a clear correlation between methylation and gene expression
Based on the dendrogram results, eight key genes (EFNA1, OLIG2, PTCH, LAG3, PGRMC1, P15, HOXB3, and LHX9) were used to analyze the relationship between methylation and gene expression.
Although HOXB3 (Figure 4(g)) and EFNA1 (Figure 4(a)) genes showed a clear inverse methylation-expression correlation, differences between groups were observed only in EFNA1, where the lowest methylation values (3% on average) were observed in mDNMT3A patients and matched the highest expression values (20 times higher than control), while the highest methylation values belonging to healthy individuals (66% on average) matched the lowest expression values (1.3 times).
Correlation between changes in gene promoter methylation and its transcriptional activity. Of the genes that were shown to be important for the hierarchic grouping and HeatMap, (a and g) two had positive correlation, (b–f) five had negative correlation, and (h) one had no correlation between methylation and expression.
Five of the remaining genes showed a direct methylation-expression correlation (Figures 4(b)–(f)), being OLIG2 the one with the strongest correlation (Figures 4(b)), with the highest expression values (five times than control) and the highest methylation values (36% on average) in mDNMT3A patients. In contrast, the lowest OLIG2 expression values (40% decrease compared with the control) and the lowest methylation values (15% on average) were observed in AML noMUT patients.
LHX9 showed no methylation-expression correlation (Figure 4(h)), with a constant methylation status (20% on average), but high expression in all AML patients, being more noticeable in mTET2 and mDNMT3A patients (Figure 4(h)).
The presence of mutations in TET2 and DNMT3A could have an impact on the survival rate of the AML patients
The analysis of patient groups using the survival outcome data showed a clear and significant correlation between the presence of mutations and overall patient survival. This is more prominent in mDNMT3A patients, with an average overall survival of 4 months compared to AML noMUT patients whose survival was 40 months (p = 0.001; Figure 5).
The survival rate of patients with mDNMT3A was considerably lower than that observed in AML noMUT individuals (p = 0.0001); in the case of patients in the mTET2 group, the mean survival rate was higher than that observed in the AML noMUT group.
Discussion
In this study, we described a relationship between AML and the two main enzymes governing DNA methylation dynamic (DNMT3A and TET2) in patients with untreated disease. We showed that alterations in these enzymes correlated with the methylation status of 16 AML-associated gene promoters and had a possible role in patient survival. These findings suggest that methylation patterns could be used as early biomarkers for AML prognosis.
Altered levels of 5mC and 5hmC in DNA have been observed in a variety of cancers, including solid and liquid neoplasias.10,11 Global hypomethylation as well as significant low levels of these modified cytosines12,13 is often reported both in cancer-derived cell lines 14 and primary tumors. 15 In this work, we found a wide variation in 5mC and 5hmC levels in cells from AML patients (Figure 1(a) and (b)), which is consistent with that reported by Kroeze et al. (5hmC concentration) and Lehman et al. (5mC concentration), who suggest not only a possible relationship of these variations with the disease prognosis but also with the survival rate. 16
Variations in 5mC and 5hmC concentrations could be caused by changes in the expression and function of genes encoding the enzymes responsible for DNA methylation and demethylation.8,17 In agreement with this, we detected variable mRNA levels of the genes DNMT3A and TET2 in AML patients. Even though we found a correlation between levels of modified cytosines and levels of DNMT3A or TET2 mRNA in most patients, this did not occur in nearly 20% of patients, suggesting that in the latter cases a different mechanism, such as the presence of mutations in these genes may be responsible.
A hotspot mutations affecting the residue R882 in the DNMT3A gene has been demonstrated to decrease its enzymatic activity by at least 50%. Meanwhile, mutations along the whole coding region of TET2 gene have been reported to affect its enzymatic activity, but a clear hotspot has not been established.9,18 Therefore, we searched for possible mutations in the hotspot of DNMT3A and for regions with higher ratio of mutation/number of nucleotides, which include important domains for the enzymatic function of TET2. 19
The mutation rate we found for both DNMT3A and TET2 genes was similar to what was previously reported.20,21 However, this work represents the first report of such mutations detected in Mexican AML patients, with an incidence of 11.8% for the hotspot of DNMT3A and 2.7% for any significant mutation in TET2.
Unlike a previous study conducted by Nguyen and collaborators where both DNMT3A and TET2 mutations were detected in T lymphoma samples, 22 our observations did not identify AML patients harboring mutation in both genes. This suggests that these epigenetic protagonists play a different role depending on which disease these enzymes are promoting.
As expected, in mDNM3A patients, 5mC concentration was lower than in those where no mutation was detected (Figure 2(a)), suggesting a decrease in the activity of the enzyme responsible for 5mC synthesis. 20 In contrast, the 5mC concentration was clearly increased among mTET2 patients, possibly as a result of the low demethylating activity of the enzyme. 17 The association between 5hmC levels and mutations in DNMT3A and TET2 mutations was less clear because individuals with mutations in either gene showed a clear decrease in 5hmC levels (Figure 2(b)). Although previous reports show a similar trend in mTET2 patients, 17 this has not been shown in mDNMT3A patients. This suggests a complex mechanism in AML for regulating 5hmC levels, with DNMT3A possibly playing a key role. Further studies are needed to clarify the involvement of DNMT3A.
Although several reports indicate that mutations in DNMT3A or TET2 either decrease their mRNA levels or have no effect at all,23,24 our data indicate that mutations in DNMT3A or TET2 lead to an increase in the expression of the mutated gene (Figure 2(c) and (d)). Interestingly, mDNMT3A patients showed a clear decrease in TET2 mRNA level (Figure 2(d)) that could explain the low 5hmC concentration observed in them, because a decrease in TET2 mRNA levels with a drastic decrease in 5hmC concentration has been reported. These results confirm that mutant TET2 or low transcription has a similar effect on 5hmC concentration. 25
Even though mTET2 patients had high levels of TET2 mRNA, the low concentration of 5hmC observed in these patients could be caused by mutations in the gene that hinder TET2’s enzymatic activity. 18
To our knowledge, the decrease in TET2 mRNA levels in individuals harboring DNMT3A mutations has not been reported previously and represents a new and open field for future research. Besides the presence of undetected mutations in TET2 (i.e. in the promoter region), several epigenetic mechanisms activated only when DNMT3A is mutated could be also involved, such as the expression of particular miRNAs that target TET2. 8 Even though this kind of interactions has been predicted in experimental models by several research groups, 26 our work conducted in AML patients represents one of the first reports of simultaneous alterations in two of the most important members in the DNA methylation and demethylation processes, in which the presence of DNMT3A mutations correlate with a decrease in TET2 transcript levels.
Previous works suggest that the global methylation status (measured as total 5mC concentration) could have a particular impact on the methylation pattern of specific genes, sometimes associated with particular clinical parameters. 27 For this reason, we experimentally evaluated the methylation status of the CpG islands in the promoters of 16 genes included in different studies where their expression level or methylation status has been evaluated in AML.28–30
The HeatMap dissimilarity index was generated with the methylation status data of these 16 genes, sorted the individuals included in our study into two big groups with clearly distinct methylation patterns (Figure 3, top panel); one group corresponding to healthy individuals and the other to AML patients. Similar results have been reported in different diseases, including various types of cancer, where a specific methylation pattern differs from that observed in healthy individuals. 31
Interestingly, the methylation group that includes AML patients comprised three clusters. Two of them match completely with the groups of mutants and non-mutants, indicating that the evaluated DNMT3A and TET2 mutations are not only associated with changes in global methylation patterns but also cause changes in methylation pattern of particular gene promoters.
The association between TET2 or DNMT3A mutations and specific methylation patterns has been reported in certain types of cancer such as glioblastoma; a subgroup of these patients exhibit specific clinical characteristics that appear to be associated with particular methylation patterns. 32 Importantly, Figueroa and collaborators, suggest that abnormal TET2 activity could be related to the start or progression of leukemogenesis through the appearance of specific methylation patterns associated with mutations in TET2-affecting genes. 33
Different studies have reported a direct correlation between transcriptional activity of specific genes and methylation pattern changes in their promoter sequences.10,34–36 The HeatMap generated using the expression data of the evaluated genes also allowed us to identify two clearly distinctive patterns corresponding to healthy individuals and AML patients. Similarly, to the methylation data, expression profiles also sorted the AML patients in three clusters that corresponded to non-mutant individuals, DNMT3A mutants, and TET2 mutants (Figure 3, bottom panel). This confirms that the presence of mutations in either of these genes is not only associated with global and promoter-specific methylation patterns but also with transcriptional expression. Therefore, methylation and transcription profiles vary depending on whether patients with AML harbor mutations in DNMT3A or TET2.
Although typically gene expression is inversely correlated to promoter methylation, recent evidence describes genes where expression and methylation are not correlated or show a different involvement, thus suggesting a high complexity in this relationship. 10 From the group of genes evaluated in our study, only EFNA1 and HOXB3 showed the typical inverse correlation between methylation and expression (Figure 4(a) and (g)). However, a direct correlation was observed in most of the remaining genes (Figure 4(b)–(f)). Moreover, in fewer genes such as LHX9, these events did not appear to be associated. In spite of its promoter methylation being similar among all individuals, expression was clearly increased in some groups and not others (Figure 4(h)).
Although associations between observed gene methylation patterns and particular clinical characteristics are difficult to detect in patients, the survival data indicate important differences between AML patients.
Diverse evidence suggests that alteration in the expression of particular genes induced by DNMT3A and TET2 activity could play a role in both, initiation and progression of different types of cancer.
For instance, it has been reported that EFNA1, one of the genes included in this study that has an altered expression in AML patients, is overexpressed in urinary bladder carcinoma, breast cancer, gastric cancer, glioma, mesothelioma, and also recently in colorectal cancer. 37 In addition, authors observed a lower survival rate in patients with high expression of this gene. In agreement with this observation, we detected high EFNA1 expression in all AML patients compared to healthy individuals. Interestingly, in patients harboring DNMT3A mutations, EFNA1 expression was considerably higher (Figure 4(a)), and the survival rate of these patients was the lowest observed in our study (Figure 5).
However, the expression of tissue-specific genes, such as OLIG2 that is normally restricted to neural tissue, has been observed in breast cancer–derived and lung cancer–derived cell lines as well as in leukemia patients. 38 According to Aplan et al., 38 OLIG2 overexpression in mice, induces a pre-T leukemia phenotype with an efficiency of 60%, suggesting that this gene could be important at the beginning of the leukemogenesis. In our study, we observed OLIG2 overexpression in both, mTET2 and mDNMT3A groups, strengthening the idea that this event might constitute an important step in the very early stages of AML development.
In summary, the transcriptional modifications that result from the alterations in the methylation pattern associated with TET2 and DNMT3A mutations could play an important role both at the beginning of the disease (OLIG2) and in the patient survival (EFNA1).
The survival data indicate important differences between AML patients depending on the mutational status of the DNMT3A and TET2 genes. However, the impact of mutations in the DNMT3A hotspot is outstanding, because the average overall survival in these patients was only one tenth that of patients with no detectable mutations in this gene (p = 0.001). However, patients harboring TET2 mutations had longer average overall survival, but the number of these AML patients were too limited to establish a trustworthy association.
All these data demonstrate that in AML, mutations in TET2 and DNMT3A are associated with DNA methylation changes at a global level; in the case of DNMT3A however, there was a drastic shortening in the survival rate of these patients. These findings suggest that these genes could be useful as prognostic biomarkers in AML, along with mutations in JAK2, cKIT, and FLT3 genes, recently proposed for their high frequency in AML (30%) and their association with a decrease in survival rate. 39 In fact, it has been suggested that large groups generated by the presence of these alterations could be divided into small subgroups, depending on the presence of less common mutations, 39 such as the ones reported in our work. These data highlight the importance of finding new biomarkers that allow a better diagnostic, prognostic, and therapy in AML.
Conclusion
In summary, we demonstrated that DNMT3A and TET2 mutations in patients with AML are associated with varying mRNA levels of these genes and changes in 5mC and 5hmC levels, thus indicating an important effect on the global DNA methylation process. Besides, mutations in these genes are associated with methylation and expression patterns that differ from those present in healthy individuals and from patients without mutations but more importantly with changes in the survival rate. Finally, our data showed for the first time a possible interaction between the two main enzymes that govern the DNA methylation dynamics in patients with AML, reporting the association of DNMT3A mutations and a clear decrease in TET2 mRNA level.
Footnotes
Acknowledgements
The authors are grateful to the patients who consented to participate in the study. The authors thank Qiagen for kindly permitting access to their facilities to use the PyroMark Q24 equipment, to Dr. Jacobo Zuñiga for technical help in the pyrosequencing process, and to Dr Daniel Piñero for the support provided. Ponciano-Gómez acknowledges CONACyT for the scholarship 234020. The authors also acknowledge Dr Guillermo Rivera, Dr Edgar Farrioni, and Dr Christian D. Cuevas for the critical review and edition process of the manuscript (MYR521).
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical approval
The present protocol was approved by the Ethics Committee and Research from the Hospital General de México, “Dr. Eduardo Liceaga,” with the reference number: DI/16/103/03/035.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the National Council of Science and Technology (CONACyT; Grant No. 256514).