Pan-cancer analysis of DNMT3A as a robust prognostic biomarker to predict the immunotherapy response

doi:10.21203/rs.3.rs-3149495/v1

Download PDF

Article

Pan-cancer analysis of DNMT3A as a robust prognostic biomarker to predict the immunotherapy response

https://doi.org/10.21203/rs.3.rs-3149495/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

DNA methyltransferase 3A (DNMT3A) is essential for de-novo methylation and cell development. Recent studies have shown that dysregulation of methylation regulated by DNMT3A is highly implicated in cancer progression. However, the regulatory roles of DNMT3A in various cancers are not completely clear and need further investigation.

Methods

The RNA-seq data in The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression databases (GTEx) are the source of this study. Western blot assays were performed to exhibit the relative expression level of DNMT3A in clinical glioma samples. CBioportal was utilized to explore the genomic alternation of DNMT3A. The images of immunofluorescence downloaded from the Human Protein Atlas (HPA) help to show the subcellular distribution of DNMT3A proteins. ComPPI is a powerful tool for studying protein interactions. Single-cell sequencing cohorts from TISCH were used to reveal the DNMT3A expression levels in different cell types. Two types of survival algorithms were conducted to assess the prognostic value of DNMT3A in pan-cancer. Gene Set Enrichment Analysis (GSEA) was applied to explore various cellular pathways and hallmarks. Immune cells infiltration in pan-cancer was summarized using data available on TIMER 2.0 website.

Results

The expression level of DNMT3A is significantly up-regulated in tumor tissue compared with that in normal tissue in most cancers. DNMT3A is discovered to have great accordance with the immune-related hallmarks like immune response signaling. In addition, the infiltration of DNMT3A in various subtypes of immune cells showed obvious aggregation of Treg, MDSC, B cell, Neutrophil, and Monocyte. At last, the robust prognostic ability of DNMT3A was further enhanced in several independent immunotherapy cohorts.

DNMT3A

pan-cancer analysis

tumor prognosis

immune infiltration

immune response

Cancer is the biggest threat to the safety of human life and property[1]. As a global public health issue, cancer deprives millions of people of their lives every single year, and the death toll continues to rise. Currently, it is widely believed that cancer is the result of a combination of genetic and environmental factors. Environmental factors include exposure to various harmful chemicals and microbes[2], while genetic factors include changes in the expression level of related genes, resulting in dysfunction of DNA methylation, histone modification, genomic stability, and so on. Although the research on cancer treatment has made great progress, the mortality rate is still high. Therefore, the research and development of novel cancer treatment methods is the most urgent task at present and will make a huge contribution to the tumor patients.

In recent years, immunosuppressive therapy targeting immune checkpoints has gradually occupied a pivotal place in the field of tumor treatment and has been developing extremely rapidly[3]. The drugs targeting immune checkpoints such as Programmed Cell Death Protein 1 (PD1)[4], Lymphocyte Activation Gene 3 (LAG3)[5], and Cytotoxic T-Lymphocyte Associated Antigen 4 (CTLA4) can effectively inhibit the growth and invasion of tumor cells and promote the lethality of the immune system on tumors[6, 7]. The effectiveness of anti-immune checkpoint drugs has been verified in clinical practice. However, these immune checkpoints cannot provide effective and lasting curative effects for all patients[8], so it is urgent to develop new immunotherapy regimens continuously to achieve long-term and stable responses with the anti-tumor drug in immunotherapy.

According to the difference in the methylated substrate, methylation can occur at DNA[9], RNA[10], and protein[11] levels. DNA methylation is a common epigenetic modification mechanism[12] in which methyl groups are directed to the C5 position of cytosine to form 5-methylcytosine under the catalysis of methyltransferase. The binding of 5-methylcytosine recruitment protein to the binding of transcription factors on the binding effect of space, promoter activity, and gene expression regulation. Incorrect cytosine methylation pattern is a risk factor for many cancers[13]. As essential regulators of DNA methylation, DNMT3A and DNMT3B play essential roles in de novo DNA methylation[14]. The methylation pattern of non-CpG sites established by DNMT3A is of great importance for the regulation of development, but surprisingly, DNMT3A does not participate in the maintenance of DNA methylation, while DNMT1 is the main gene responsible for the maintenance of methylation[15]. However, with the gradual deepening of research, more and more evidences reveal the important roles of DNMT3A and DNMT3B in methylation maintenance, and DNMT1 is also involved in de novo methylation in specific loci, therefore, there is an intersection between the functions of the two types of genes[16, 17]. DNA methylation is the most common type of methylation, which has great clinical transformation potential and is widely used in the early diagnosis of diseases[18]. Its role in tumor treatment and prognosis evaluation has attracted more and more attention[19]. In addition to affecting epigenetics, DNMT3A is also involved in the expression regulation of DNA repair-related genes[20, 21]. For example, in DNMT3A-deficient cells, the expression of PARP1 gene is significantly down-regulated, and the alternate non-homologous end-joining (Alt-NHEJ) mediated by PARP1 is inhibited while favored the homologous recombination (HR)[22].

In view of the importance of methylation regulated by DNMT3A for development[23], an increasing number of studies have shown that DNMT3A functions in promoting the growth and development of cancer [24, 25]. Prognostic analysis of pan-cancer suggests that DNMT3A may be an effective prognostic factor in a variety of tumors, and its specific association is that high expression of DNMT3A protein predicts a worse prognosis.

Acquisition and processing of raw data

The mRNA-seq data and its matched clinical information were downloaded from TCGA and GTEx datasets in UCSC Xena (https://xenabrowser.net/datapages/)[26]. Toil strategy[27] was applied to transform the Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) into Transcripts Per Kilobase Million (TPM). The genomic alteration data of DNMT3A in 33 kinds of cancers was downloaded and analyzed on the cBioprotal website (https://www.cbioportal.org/). cBioportal is an open-source website for comprehensive analysis of the genomic data in cancers, which examines the mutation type and mutation frequency in all cancer types[28]. The images of immunofluorescence obtained from HPA (https://www.proteinatlas.org/) were used for observing the distribution and abundance of DNMT3A protein. The source data of protein-protein interaction (PPI) network including the name of the interacting protein, the subcellular site of action, and the corresponding comprehensive score was gained from the ComPPI website (http://comppi.linkgroup.hu)[29]. Correlation analysis between immune cell infiltration abundance and DNMT3A expression in various tumor cells was obtained from Timer2.0 (http://timer.cistrome.org/). The Single-cell sequencing data and related studies were conducted on the TISCH website (http://tisch.comp-genomics.org/home/).

Clinical GBM samples and Western blot analysis

We obtained GBM surgically resected tumor samples and adjacent peritumoral tissue samples from the Neurosurgery Department of the Second Affiliated Hospital of Nanchang University. All patients who provided clinical samples have signed informed consent and the study has been reviewed and approved by the Ethics Committee of the Second Affiliated Hospital of Nanchang University. All clinical samples were in accordance with the Declaration of Helsinki. All samples were stored in liquid nitrogen (-196℃) immediately once they were surgically removed to prevent protein molecules from being degraded until tissue protein extraction experiments. After the samples were minced with tissue shears, protein extraction was performed using Radio Immunoprecipitation Assay (RIPA) lysis buffer mixed with 1% phenylmethanesulfonylfluoride (PMSF). We extracted proteins from seven tumor samples and their adjacent tissues respectively. 20 samples were loaded in 20 lanes and separated under the condition of 100 V for half an hour and then 120 V for 1 hour. The separated protein was then immobilized to the polyvinylidene difluoride (PVDF) membrane by membrane transfer process at an atmosphere of 400mA for 1 hour. 5% skim milk was used to block the nonspecific antigen on the PVDF membrane surface. After 3 times of rinsing with Tris-BufferedSalineTween-20 (TBST) solution, the membrane strips were incubated with DNMT3A and Tubulin antibody at 4℃ overnight. Then we rinsed the strips with TBST another 3 times, and the horseradish peroxidase (HRP)-conjugated secondary antibody was applied to incubate the strips at room temperature for 4 hours. At last, we utilized the hypersensitive enhanced chemiluminescence (ECL) kit to incubate the strips and the images were visualized through GV6000M. ImageJ software was applied to the semi-quantitative analysis of protein content.

Immunohistochemistry and immunofluorescence

Glioma samples were first fixed with 4% paraformaldehyde. After dehydration, they were embedded in paraffin and cut into slices of appropriate thickness with a cryostat. Dewaxing treatment made the sections fixed on glass slides to make glioma tissue sections. Samples treated with 3% hydrogen peroxide for 10 minutes at room temperature were used to inactivate endogenous peroxidase activity in the sections. We used 5% bovine serum albumin 5 for 30 min incubation for non-specific antigen blocking. Next, anti-DNMT3A antibody and anti-GFAP antibody were incubated overnight at 4°C as primary antibodies. After washed withphosphate buffer solution (PBS) for three times, 50µL of secondary antibody was used for 15 min incubation at room temperature, and then washed with PBS for another three times. Add 50 µL of streptavidin-peroxidase solution to each slide and incubate at room temperature for 10 minutes. Add 100 µL of DAB solution to the slides for 5 minutes and rinse with water. The final sections were counterstained with hematoxylin and differentiated with 1% hydrochloric acid alcohol for 3 seconds, and rinsed with tap water for three minutes. After dehydration, the microscopic imaging system can be used to take pictures. Glioma tissue sections were also used for immunofluorescence histochemical experiments. We used 3% Bovine albumin (BSA) solution to incubate for 1h for antigen blocking. Similarly, anti-GFAP (1:100, proteintech, Cat No. 16825-1-AP) and anti-DNMT3A (1:100, proteintech, Cat No. 20954-1-AP) antibodies were applied to incubate the samples overnight, and PBS solution was used to rinse the sections for three times to remove residual antibodies. Cy3 (1:300) and AF488 (1:400) were used as fluorescent labeled secondary antibodies to incubate the sections with for 1h. After three times of rinse, DAPI was used to dye the nucleus. The results were observed and recorded with fluorescence microscope.

Analysis of DNMT3A expression level in Single-cell seq cohort.

We conducted the single-cell analysis to reveal the abundance of DNMT3A in various types of immune cells through the TISCH web tool[30]. TISCH (http://tisch.comp-genomics.org/documentation/) is a powerful website tool that can quickly search the expression level of target genes in different immune-related cells of 79 independent single-cell cohorts recorded in library, and the results were summarized in the heatmap. The expression of DNMT3A in diverse single-cell cohorts has been normalized for further analysis. In addition, we can further explore the expression level of target gene in each type of cell in a separate cohort.

Multimodal survival analysis of DNMT3A in pan-cancer

To comprehensively analyze the linkage between the survival time and the prognosis status of the patients, we collected four kinds of survival data of each sample in the TCGA pan-cancer database, including overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI), combined with the mRNA expression level of DNMT3A and survival status, and carried out two tumor prognosis methods mentioned above to comprehensively evaluate the prognostic function of DNMT3A[31]. The gene expression and survival time in the univariate Cox regression were analyzed as continuous variables. Kaplan-Meier survival analysis used “survival” and “surviminor” R packages to obtain the optimal cut-off value and drew the survival curve[32]. We integrated the data of the DNMT3A expression and four types of survival time to perform the survival analysis with two methods and finally drew a heatmap to show the prognostic results of DNMT3A in pan-cancer clearly. In addition, the log-rank p-value and hazard ratios (HR) with 95% confidence intervals (CI) of DNMT3A in each tumor were calculated and displayed in the forest plot with the assistance of the “froestplot” R package[33].

Gene Set Enrichment Analysis of DEGs

The whole patient sample library is sorted according to the expression level of DNMT3A, and we regard the top 30% as DNMT3A high-expression group and the bottom 30% as DNMT3A low-expression group respectively that to identify the DNMT3A-associated differentially expressed genes (DEGs). We used the “limma” package to obtain the consequence of DEGs analysis, and the genes with p-value < 0.05 are seen as DEGs. The log₂FC and adjusted p-value of each DEGs was calculated and will be applied to conduct GSEA through the R software. The hallmark gene set file was downloaded from the Molecular Signatures Database (MSigDB, https://www.gsea-msigdb.org/gsea/index.jsp)[34]. We can obtain the normalized enrichment score (NES) and false discovery rate (FDR) for each kind of biological process collected in the hallmark file of the DEGs between DNMT3A high- and low- expression groups. At last, we visually analyze the above results through the “clusterProfiler” package [35], and the final results are presented in histogram and bubble chart. In view that too many kinds of tumors and hallmarks, we conclude the results and show them in the heatmap.

Immune cell infiltration in tumors

Infiltration analysis of immune-related cells in tumors can help us understand the relationship between DNMT3A and immune cell infiltration[36]. TIMER (http://timer.cistrome.org/) is a powerful tool for analyzing different types of immune cell infiltration in various tumors[37, 38], and the correlation prediction between the relative expression of DNMT3A and the infiltration of various immune cells in diverse tumors was calculated by the deconvolution statistical method. Relevant immune-related cell types include B cells, CD4 + T cells, CD8 + T cells, Tregs, Neutrophil, Monocyte, Macrophage, DC, NK cells, Mast cells, Cancer-associated fibroblast, Common lymphoid progenitor, Common myeloid progenitor, Endothelial cell, Eosinophil, Granulocyte-monocyte progenitor, Hematopoietic stem cell, T cell follicular helper, γδT cell, NK T cell and Myeloid-derived suppressor cells, and these cells constitute the immune microenvironment of the tumor together. Spearman correlation analysis was performed to analyze the correlations between the DNMT3A expression and 21 kinds of immune cells mentioned above.

Correlation analysis of biomarkers and DNMT3A expression

In order to study the relationship between the abundance of immune checkpoints (ICPs) and the expression of DNMT3A, we extracted the mRNA sequencing data of 47 immune regulators in the TCGA database and combined them with the expression of DNMT3A for Spearman’s correlation analysis. In addition, we calculated the tumor mutation burden (TMB) and microsatellite instability (MSI) with the previously reported algorithm in various tumors[39]. TMB and MSI are classical tumor biomarkers for evaluating tumor immune microenvironment, which is of great significance for patient's condition and tumor immunological analysis.

Prognostic validation in immunotherapy cohorts

Previous studies have demonstrated that DNMT3A has a significant prognostic role in the TCGA cohort. We randomly selected two immunotherapy cohorts (GSE91061 and vanallen2015) to study the relationship between DNMT3A expression and the prognosis of patients receiving immunotherapy. In the cohort, the responses of the high and low expression groups to immunotherapy were also statistically analyzed to indicate the potential link between DNMT3A expression and immunotherapeutic response. There are 51 melanoma patients receiving nivolumab (anti-PD1 drug) treatment in GSE91061 cohort[40], and 42 melanoma patients receiving ipilimumab (anti-CTLA4 drug) treatment in VanAllen2015 cohort[41]. Through the analysis of DNMT3A and the response effect of immunotherapy in the above two immunotherapy cohorts, we can explore the application prospect of DNMT3A in patients receiving immunotherapy effect.

Statistical processing method

The expression difference among various types of tissue was evaluated through the Kruskal–Wallis test. The Wilcoxon rank-sum test method was used for the comparison of gene expression and the correlation test between the two cohorts. For the difference analysis of DNMT3A expression in tumor tissue and adjacent tissue in the same sample, we used paired t-test to calculate its statistical significance. For the survival correlation analysis of each cohort, we used two analysis methods Kaplan-Meier method and Univariate Cox regression analysis to conduct the DNMT3A-related prognosis analysis of various tumors. Spearman’s correlation test was applied to evaluate the correlation between DNMT3A expression and relative scores like TMB, MSI, immune cell infiltration score and immune regulators. Finally, Chi-square analysis was applied to compare the immune responses of the patients in the DNMT3A-high and DNMT3A-low expression groups in the immunotherapy cohort. The result is considered statistically different when the log rank p-value < 0.05.

Overview of DNMT3A related information

In order to make a preliminary evaluation of the mRNA level of DNMT3A in various types of tumors, we integrated the expression data in TCGA and GTEx database and performed t-test. In most types of tumors, the mRNA level in tumor tissues is lower than that in normal tissues. This difference may be due to the relative specificity between tumors (Fig. 1A). To observe the difference in the expression of DNMT3A in different tumors in more detail and given the fact that we are mainly engaged in GBM research, we selected the GBM samples to analyze the expression difference and found that the mRNA expression of DNMT3A in GBM tissue was remarkably upregulated (Fig. 1B). To further confirm the conclusion, we designed the western blot assay and the result shows that tumor tissue has higher expression compared to adjacent tissue (Fig. 1C). We quantified the expression level by detecting the gray value with the assistance of ImageJ software and visualized the difference in R software(Fig. 1D). Exploration of DNMT3A genomic alternation showed that the alternation rate of DNMT3A was not high in most tumors, indicating that DNMT3A is a relatively conserved gene in tumor cells. Mutations account for the vast majority of DNMT3A genomic alternation, especially in AML. However, in UCS patients, amplification accounted for the largest proportion of the five gene alterations, and interestingly, deep deletions accounted for nearly all of the genetic mutations in DLBC patients (Fig. 1E). Immunofluorescence images showed that the distribution of DNMT3A protein was co-located with the nucleus in A431 and U251 cell lines, indicating that DNMT3A was mainly distributed in the nucleus and played a role (Fig. 1F). The PPI analysis showed that 60 kinds of proteins interacted with DNMT3A in different parts of the cell, which further proved the extensiveness of DNMT3A involved in the regulation of cellular functions (Fig. 1G).

Single-cell sequencing analysis of DNMT3A

To determine the expression of DNMT3A in 33 classical cell types, we performed DNMT3A expression analysis in 79 independent single-cell sequencing cohorts. From the heatmap, we can find that the expression level of DNMT3A in most types of immune cells is not high, but the expression level in Neutrophils and Epithelial is significantly higher than that in other types of cells. And the overall expression levels of DNMT3A were different in different single-cell cohorts (Fig. 2). In the figure, the data is processed for better visualization, and the larger the value, the higher the DNMT3A expression. Judging from the result above, we found DNMT3A exhibit a higher expression in the glioma cohorts which attract us to know whether the expression of DNMT3A has a correlation with glial cells. Immunohistochemistry was performed to determine the difference of DNMT3A expression between GBM tissues and normal brain tissues (NBT). The results showed that the expression of DNMT3A in GBM tissues and NBT showed great differences (Fig. 3A). ImageJ software was used for quantitative analysis, and the staining gray value was converted into optical density (OD) value to draw a bar chart (p < 0.0001) (Fig. 3B). Then we selected GFAP which is specially expressed in glial cells to perform the immunohistochemical fluorescence along with DNMT3A. The result shows that DNMT3A is lowly expressed in the NBTs, while DNMT3A is highly expressed in the glioma samples and have an obvious colocalization with GFAP protein (Fig. 3C). The results above proved that cancer cells have a higher DNMT3A expression in glioma samples.

Higher DNMT3A expression predicts a worse prognosis in pan-cancer

We performed the survival analysis with the overall survival (OS) time, and the result shows that DNMT3A plays a risky role in various tumors, like ACC, GBM, LIHC, MESO, and SARC and UVM. Beyond that, in order to reduce the occurrence of accidental results, we performed the analogous analysis with DSS, DFI, and PFI. All the results are summarized and characterized on the heatmap. The result shows that DNMT3A acts as a hazardous biomarker in a wide range of cancers. But situations were not the same in all tumors such as in the patients of HNSC, OV, READ, THYM, and USC, higher DNMT3A expression exhibit a better prognosis on the contrary. The situations in four kinds of survival time and two kinds of algorithms were basically the same which can be regarded as strong support for the conclusion above (Fig. 4A). To further check the prognostic role of DNMT3A, we performed a univariate COX regression analysis, calculated the hazard rate (HR) and p-value in 32 types of tumors, and displayed them in a forest plot (Fig. 4B). To observe the prognostic value in more detail, we selected tumors with positive results (ACC, GBM, LIHC, MESO) in the forest map for Kaplan-Meier curve analysis to further confirm the prognostic effect of DNMT3A and the relationship between DNMT3A high expression and poor prognosis. The results showed that high expression of DNMT3A predicted a poor prognosis, indicating DNMT3A is an effective biomarker (Fig. 4C-F). In addition, we also integrated three GEO databases (GSE16011, GSE61374, and Rembrandt) and conducted differential expression analysis and survival analysis, respectively. The results showed that the expression of DNMT3A was upregulated with the increase of glioma grade, and DNMT3A is a poor prognostic factor (Supplementary Fig. 1A-B).

GSEA analysis of DNMT3A in pan-cancer

After the DEGs of the DNMT3A high- and low- expression subgroups were screened out by “limma” package, GSEA analysis was performed to preliminarily identify their functions in tumor cells and summarize cell functions with cancer hallmarks (Fig. 5). Many immune and inflammation-related pathways were involved in the DNMT3A-related DEGs, such as the interferon-α and γ response, inflammatory response, IL6-JAK-STAT3, and IL2-STAT5 signaling pathways in most kinds of cancers. These results demonstrate that the expression level of DNMT3A is highly correlated with inflammatory and immune-related pathways[42]. The initial formation of tumor cells results from the immune escape of tumor cells caused by abnormal immune system[43]. There are various inflammatory cells in the infiltration environment of the tumor, and the activation of these inflammatory cells is inhibited by some molecules so that the killing effect towards tumor cells is weakened, and the tumor cells are retained in the body. Furthermore, the NES in the G2M checkpoint and E2F targets is relatively high, and as we know G2M and E2F are members of transcription factors which proves that DNMT3A may be involved in regulating the expression of other genes by affecting the expression of some transcription factor-related genes. In conclusion, DNMT3A is of great significance for the immune process and the regulation of inflammation, and can influence the inflammatory state of the internal environment by regulating the expression of DNMT3A, which can provide early inhibition and even elimination of tumor cells.

Infiltration analysis of immune-related cells

To reveal the close relationship between DNMT3A expression and immune cell infiltration in tumors, we conducted an in-depth mining analysis of 21 common immune-related cell types. Spearman correlation analysis was applied to reveal the relationship between the expression of DNMT3A and the abundance of immune cells in pan-cancer (Fig. 6). Overall, the expression of DNMT3A showed a clear positive correlation with various types of immune cells, especially in Tregs, MDSC, B cells, Neutrophil, Monocyte, CAF, Endo, and Tfh. However, the positive correlation between other types of immune cells and DNMT3A expression is not so obvious or even show a significant negative correlation, such as NKT and CD8 + T cells. Recent studies have shown that the treatment of immune cells including B cells, CD8 + T cells, NK cells, CAF, Neutrophil, and Macrophage have been proved to be of great significance for tumor suppression and clearance[44–47]. Most of the aforementioned cells were positively enriched in our results, demonstrating a strong association between DNMT3A and immune-related cells in the tumor microenvironment. Our results reveal the potential of DNMT3A to influence tumorigenesis, progression, and prognosis by modulating immune cells.

Analysis of DNMT3A-related immune characteristics

Many immunomodulatory factors have been proved to be highly correlated with tumor immunoregulation. We have extracted the mRNA level of 47 immune regulators and performed a correlation analysis combined with the expression levels of DNMT3A. The result was exhibited in Fig. 7A. From the diagram, we can conclude that in DLBC, KICH, and UVM tumors, DNMT3A showed a robust positive correlation with most immune regulators, while in TGCT, DNMT3A expression had a significant negative correlation with most immunomodulators. In addition, we found that ADORA2A and CD276 molecules were significantly positively associated with DNMT3A expression in more than half of tumor types. To further help predict patients' response to immune checkpoint inhibitor therapy, we calculated TMB and MSI indices in various tumors and preliminarily evaluated their correlation with DNMT3A expression. As shown in the figure, DNMT3A expression showed a positive trend with TMB in most tumors, such as in BLCA, CESC, LUAD, LUSC, and TGCT tumors, while showed a significant negative correlation in DLBC and STAD tumors (Fig. 7B). Additionally, DNMT3A expression and MSI index also showed a strong correlation in many tumors. A significant positive correlation was found in BLCA, BRCA, LAML, LGG, LUAD, PRAD, and other tumors, and a negative correlation was found in COAD, DLBC, ESCA, and other tumors (Fig. 7C). The above results show that DNMT3A is an effective biomarker to predict the prognosis after immunotherapy in the corresponding tumors.

The predictive ability of DNMT3A in immunotherapy cohort

As the mainstream tumor treatment method in recent years, immunotherapy is widely used in the clinical treatment of cancers. In order to further understand the prognostic function of DNMT3A in the immunotherapy cohorts, we randomly selected several independent immunotherapy cohorts for survival analysis using DNMT3A expression levels and patients' overall survival time. In the GSE91061 immunotherapy cohort (an anti-PD-L1 immunotherapy cohort), DNMT3A expression was considered to be a risk factor for tumor treatment, as high-DNMT3A expression patients had a worse prognosis, showing that high-DNMT3A expression patients had a shorter survival time and lower survival rate. The effective response rate in the low DNMT3A expression group was higher than that in the high DNMT3A expression group obviously, proving that the cellular functions involved in DNMT3A may hinder the effect of related immunotherapy drug (Fig. 7D). Similarly, we explored the prognostic role of DNMT3A expression in the VanAllen2015 immunotherapy cohort (Fig. 7E), and the results were consistent with those in the GSE91061 cohort. It further proved that DNMT3A is a negative prognostic factor for tumor patients. In both tumor immunotherapy cohorts, the potential role of DNMT3A on tumor prognosis was confirmed, and the potential of DNMT3A as a tumor prognostic marker was further strengthened.

Nowadays, tumor immunotherapy has become the mainstream treatment. Chimeric antigen receptor (CAR) -T therapy[48–50] and immune checkpoint inhibitors therapy[51] are of great help to the treatment and prognosis of tumor patients. Given the individual differences between each patient, the response effect of immunotherapy for the same tumor varies greatly[52]. Immunotherapy that has been proved to be effective in some individuals may not be effective in other patients. Therefore, we try to evaluate the abundance of biomarkers in patients as a whole, to accurately predict the response effect of the patient to specific immunotherapy. In this study, we preliminarily determined the potential of DNMT3A as a tumor prognostic marker to predict the prognosis of tumor patients. Then we tried to apply it to the immunotherapy response effect of tumor patients and achieved gratifying results. Our conclusions can provide some reference value for the further application of DNMT3A in immunotherapy.

In our study, we first comprehensively evaluated the expression of DNMT3A. From the mRNA level analysis, the mRNA expression of DNMT3A showed an upward trend in most tumors. For example, in BLCA, BRCA, CHOL, GBM, LICH, SKCM, TGCT, and UVM, the expression of DNMT3A showed a significant overexpression trend. At the protein level, we extracted the tissue proteins of GBM patients and carried out WB experiments. The results further confirmed the above conclusion. The expression of DNMT3A protein in tumor tissues was significantly higher than that in precancerous tissues (p < 0.001). This suggests that DNMT3A may play a role as a cancer-promoting factor in most tumors. In other words, DNMT3A may be a potential oncogene and can be regarded as a promising biomarker in specific cancers. From the perspective of single-cell cohort and immunohistochemical fluorescence, the high expression of DNMT3A is mainly concentrated in cancer cells, and the infiltrating immune cells also exhibit relatively high DNMT3A expression, indicating that the high expression of DNMT3A in tumor tissue is caused by the high expression of DNMT3A in both cancer cells and infiltrating immune cells.

Another aspect of our research is the overall evaluation of the survival related to DNMT3A. Through a comprehensive and in-depth analysis of four survival data and two algorithms, we conclude that the high expression of DNMT3A indicates a poor prognosis, such as ACC, BRCA, CESC, COAD, GBM, LGG, LICH, LUAD, LUSC, MESO, PRAD, TGCT, THCA, and UVM. In a few special tumors, such as HNSC, OV, THYM, UCS, and so on, DNMT3A plays a protective factor. Patients with higher expression of DNMT3A have a better prognosis. Survival plots drawn through the Kaplan-Meier method in several tumors show in detail the adverse prognostic role of DNMT3A in tumor patients. The above analysis shows that DNMT3A is competent as a prognostic biomarker in specific tumors.

Through the correlation analysis of differentially expressed genes and GSEA, we found that DNMT3A related differential genes were enriched in immune and inflammatory processes, including γ-Interferon, α-Interferon response, inflammatory response, and interleukin related pathways suggesting that DNMT3A may be indirectly or even directly involved in tumor immune and inflammatory regulation[53, 54]. Another related pathway is the transcription factor-related regulatory pathway. E2F is a famous family of transcription factors, which is involved in the regulation of multiple gene expression, and studies have shown that DNMT3A can participate in the biological phenotype regulated by E2F[55].

More than that, one more interesting fact is that the DNMT3A expression is strongly correlated with various immune-related cells, such as Treg, MDSC, B cell, Neutrophil, Monocyte, CAF, Endo, etc. The degree of infiltration of these immune-related cells is correlated with the expression of DNMT3A strongly, suggesting that DNMT3A may affect the disease progression and prognosis of tumor patients by participating in the regulation of the tumor immune microenvironment. The co-expression correlation analysis between the abundance of DNMT3A and that of multiple immune regulatory factors indicates that in DLBC, KICH, and UVM, the expression of DNMT3A has a positive correlation with almost all immune-related factors[56, 57]. In addition, DNMT3A has an obvious correlation with CD276 and ADORA2A in all types of tumors, and the research of CD276 and ADORA2A as novel immunotherapeutic targets has also continued and achieved preliminary success. Given the co-expression of DNMT3A and CD276, it will be an interesting study of whether DNMT3A may be a potential therapeutic target in combination with CD276 for immunotherapy.

As two commonly used tumor-related indexes, TMB and MSI have a certain ability to predict the clinical response of tumor immunotherapy[58]. More and more evidences show that TMB and MSI can be regarded as potential efficacy prediction markers. TMB mainly reflects the mutation in tumor tissue. The higher TMB index indicates more mutant genes in tumor cells. The amount of abnormal protein produced by patients in the high TMB group is also significantly increased, so it is easier to be recognized by the immune system, which helps to improve the sensitivity of the immune system[59]. The anti-cancer response of the immune system is also easier to be activated. Therefore, the frontal killing effect of immune drugs on tumor cells is enhanced, so it can harvest a better curative effect. Similarly, the MSI index represents gene instability in the process of DNA replication. The higher the MSI index, the more unstable the genome is. The possibility of base mismatch and the gene mutation is increasing greatly, which can generate abnormal protein and activates the anti-cancer immune response of immune-related cells.

In the tumor immunotherapy cohort, there is a significant correlation between the expression of DNMT3A and the survival time of patients receiving anti-PD-1 treatment. The overall survival time of patients with high expression of DNMT3A is significantly shortened, which proves that DNMT3A can stably predict the prognosis of melanoma patients. The response effect of the high expression group to immunotherapy was significantly lower than that of the low expression group. The above results show that DNMT3A can not only increase the pathogenicity of the tumor to the body and make patients achieve worse survival, but also may be involved in hindering the role of immunosuppressive drugs. It is a biomarker with a strong prediction ability for tumor prognosis.

It is undeniable that most of the data used in the above research come from queues in public databases, which makes our research have certain limitations. And our conclusions are based on the theoretical level. We reasonably speculate about the cell function of DNMT3A and its impact on tumor patients on the basis of the existing theoretical basis. We also need to comprehensively verify the function of DNMT3A in tumors through later design corresponding experiments. In conclusion, we conducted a comprehensive pan-cancer analysis of DNMT3A from all aspects and repeatedly verified its robust prognostic role in a variety of tumors. The results make us confident that DNMT3A can be used as a tumor prognostic marker.

Date Availability Statement

The original data of this study have been published in TCGA database and can be downloaded through the website: https://xenabrowser.net/datapages.

Ethics statement

All tumor and normal samples from human body in this study have been certified by the Ethics Committee of The Second Affiliated Hospital of Nanchang University. All the patients have written the informed consent.

Author Contributions

FRL, JYL, and QKJ designed this subject and revised the manuscript. XYS and JZL performed the data collection, bioinformatic and statistical analysis, figure visualization, and manuscript writing. ZWT did some necessary revision of this subject. All authors have agreed to publish the manuscript.

Funding

This research was funded by the National Natural Science Foundation (grant nos. 81960521), and Jiangxi Province Department of Education Science and technology research project, China (grant no. GJJ210177).

Acknowledgments

We sincerely acknowledge the contributions from the TCGA database and Sanger box to this study.

Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A: Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018, 68(6):394-424.
Tiffon C: The Impact of Nutrition and Environmental Epigenetics on Human Health and Disease. Int J Mol Sci 2018, 19(11).
Singh S, Hassan D, Aldawsari HM, Molugulu N, Shukla R, Kesharwani P: Immune checkpoint inhibitors: a promising anticancer therapy. Drug Discov Today 2020, 25(1):223-229.
Kumagai S, Togashi Y, Kamada T, Sugiyama E, Nishinakamura H, Takeuchi Y, Vitaly K, Itahashi K, Maeda Y, Matsui S et al: The PD-1 expression balance between effector and regulatory T cells predicts the clinical efficacy of PD-1 blockade therapies. Nat Immunol 2020, 21(11):1346-1358.
Maruhashi T, Sugiura D, Okazaki IM, Okazaki T: LAG-3: from molecular functions to clinical applications. J Immunother Cancer 2020, 8(2).
Havel JJ, Chowell D, Chan TA: The evolving landscape of biomarkers for checkpoint inhibitor immunotherapy. Nat Rev Cancer 2019, 19(3):133-150.
Zhang H, Dai ZY, Wu WT, Wang ZY, Zhang N, Zhang LY, Zeng WJ, Liu ZX, Cheng Q: Regulatory mechanisms of immune checkpoints PD-L1 and CTLA-4 in cancer. J Exp Clin Canc Res 2021, 40(1).
Ribas A, Wolchok JD: Cancer immunotherapy using checkpoint blockade. Science 2018, 359(6382):1350-1355.
Moore LD, Le T, Fan G: DNA methylation and its basic function. Neuropsychopharmacology 2013, 38(1):23-38.
Yang B, Wang JQ, Tan Y, Yuan R, Chen ZS, Zou C: RNA methylation and cancer treatment. Pharmacol Res 2021, 174:105937.
Chung WM, Chen L, Chang WC, Su SY, Hung YC, Ma WL: Androgen/Androgen Receptor Signaling in Ovarian Cancer: Molecular Regulation and Therapeutic Potentials. International journal of molecular sciences 2021, 22(14).
He XJ, Chen T, Zhu JK: Regulation and function of DNA methylation in plants and animals. Cell Res 2011, 21(3):442-465.
Traube FR, Carell T: The chemistries and consequences of DNA and RNA methylation and demethylation. RNA Biol 2017, 14(9):1099-1107.
Zhang ZM, Lu R, Wang P, Yu Y, Chen D, Gao L, Liu S, Ji D, Rothbart SB, Wang Y et al: Structural basis for DNMT3A-mediated de novo DNA methylation. Nature 2018, 554(7692):387-391.
Uysal F, Cinar O, Can A: Knockdown of Dnmt1 and Dnmt3a gene expression disrupts preimplantation embryo development through global DNA methylation. J Assist Reprod Genet 2021, 38(12):3135-3144.
Ren W, Gao L, Song J: Structural Basis of DNMT1 and DNMT3A-Mediated DNA Methylation. Genes (Basel) 2018, 9(12).
Gujar H, Weisenberger DJ, Liang G: The Roles of Human DNA Methyltransferases and Their Isoforms in Shaping the Epigenome. Genes (Basel) 2019, 10(2).
Delpu Y, Cordelier P, Cho WC, Torrisani J: DNA methylation and cancer diagnosis. Int J Mol Sci 2013, 14(7):15029-15058.
Dai X, Ren T, Zhang Y, Nan N: Methylation multiplicity and its clinical values in cancer. Expert Rev Mol Med 2021, 23:e2.
Cali CP, Park DS, Lee EB: Targeted DNA methylation of neurodegenerative disease genes via homology directed repair. Nucleic Acids Res 2019, 47(22):11609-11622.
Walsh CP, Xu GL: Cytosine methylation and DNA repair. Curr Top Microbiol Immunol 2006, 301:283-315.
Maifrede S, Le BV, Nieborowska-Skorska M, Golovine K, Sullivan-Reed K, Dunuwille WMB, Nacson J, Hulse M, Keith K, Madzo J et al: TET2 and DNMT3A Mutations Exert Divergent Effects on DNA Repair and Sensitivity of Leukemia Cells to PARP Inhibitors. Cancer Res 2021, 81(19):5089-5101.
Heyn P, Logan CV, Fluteau A, Challis RC, Auchynnikava T, Martin CA, Marsh JA, Taglini F, Kilanowski F, Parry DA et al: Gain-of-function DNMT3A mutations cause microcephalic dwarfism and hypermethylation of Polycomb-regulated regions. Nat Genet 2019, 51(1):96-105.
Zhou Y, Yang Z, Zhang H, Li H, Zhang M, Wang H, Zhang M, Qiu P, Zhang R, Liu J: DNMT3A facilitates colorectal cancer progression via regulating DAB2IP mediated MEK/ERK activation. Biochimica et biophysica acta Molecular basis of disease 2022, 1868(4):166353.
Zhang Y, Gao Y, Zhang G, Huang S, Dong Z, Kong C, Su D, Du J, Zhu S, Liang Q et al: DNMT3a plays a role in switches between doxorubicin-induced senescence and apoptosis of colorectal cancer cells. International journal of cancer 2011, 128(3):551-561.
Consortium GT: The Genotype-Tissue Expression (GTEx) project. Nat Genet 2013, 45(6):580-585.
Vivian J, Rao AA, Nothaft FA, Ketchum C, Armstrong J, Novak A, Pfeil J, Narkizian J, Deran AD, Musselman-Brown A et al: Toil enables reproducible, open source, big biomedical data analyses. Nat Biotechnol 2017, 35(4):314-316.
Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E et al: Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 2013, 6(269):pl1.
Veres DV, Gyurko DM, Thaler B, Szalay KZ, Fazekas D, Korcsmaros T, Csermely P: ComPPI: a cellular compartment-specific database for protein-protein interaction network analysis. Nucleic Acids Res 2015, 43(Database issue):D485-493.
Sun D, Wang J, Han Y, Dong X, Ge J, Zheng R, Shi X, Wang B, Li Z, Ren P et al: TISCH: a comprehensive web resource enabling interactive single-cell transcriptome visualization of tumor microenvironment. Nucleic Acids Res 2021, 49(D1):D1420-D1430.
George B, Seals S, Aban I: Survival analysis and regression models. J Nucl Cardiol 2014, 21(4):686-694.
Rich JT, Neely JG, Paniello RC, Voelker CC, Nussenbaum B, Wang EW: A practical guide to understanding Kaplan-Meier curves. Otolaryngol Head Neck Surg 2010, 143(3):331-336.
Kim J, Kaufman JS, Bang H: Graphing Ratio Measures on Forest Plot. J Am Coll Cardiol 2018, 71(5):585-586.
Liberzon A, Subramanian A, Pinchback R, Thorvaldsdottir H, Tamayo P, Mesirov JP: Molecular signatures database (MSigDB) 3.0. Bioinformatics 2011, 27(12):1739-1740.
Yu G, Wang LG, Han Y, He QY: clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 2012, 16(5):284-287.
Zhang H, Wang Z, Dai Z, Wu W, Cao H, Li S, Zhang N, Cheng Q: Novel Immune Infiltrating Cell Signature Based on Cell Pair Algorithm Is a Prognostic Marker in Cancer. Front Immunol 2021, 12:694490.
Li T, Fu J, Zeng Z, Cohen D, Li J, Chen Q, Li B, Liu XS: TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res 2020, 48(W1):W509-W514.
Li T, Fan J, Wang B, Traugh N, Chen Q, Liu JS, Li B, Liu XS: TIMER: A Web Server for Comprehensive Analysis of Tumor-Infiltrating Immune Cells. Cancer Res 2017, 77(21):e108-e110.
Chalmers ZR, Connelly CF, Fabrizio D, Gay L, Ali SM, Ennis R, Schrock A, Campbell B, Shlien A, Chmielecki J et al: Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden. Genome medicine 2017, 9(1):34.
Riaz N, Havel JJ, Makarov V, Desrichard A, Urba WJ, Sims JS, Hodi FS, Martin-Algarra S, Mandal R, Sharfman WH et al: Tumor and Microenvironment Evolution during Immunotherapy with Nivolumab. Cell 2017, 171(4):934-949 e916.
Van Allen EM, Miao D, Schilling B, Shukla SA, Blank C, Zimmer L, Sucker A, Hillen U, Foppen MHG, Goldinger SM et al: Genomic correlates of response to CTLA-4 blockade in metastatic melanoma. Science 2015, 350(6257):207-211.
Owen KL, Brockwell NK, Parker BS: JAK-STAT Signaling: A Double-Edged Sword of Immune Regulation and Cancer Progression. Cancers (Basel) 2019, 11(12).
Wu SY, Fu T, Jiang YZ, Shao ZM: Natural killer cells in cancer biology and therapy. Molecular cancer 2020, 19(1):120.
Tokunaga R, Naseem M, Lo JH, Battaglin F, Soni S, Puccini A, Berger MD, Zhang W, Baba H, Lenz HJ: B cell and B cell-related pathways for novel cancer treatments. Cancer Treat Rev 2019, 73:10-19.
Kishton RJ, Sukumar M, Restifo NP: Metabolic Regulation of T Cell Longevity and Function in Tumor Immunotherapy. Cell Metab 2017, 26(1):94-109.
Shimasaki N, Jain A, Campana D: NK cells for cancer immunotherapy. Nat Rev Drug Discov 2020, 19(3):200-218.
Chen Y, McAndrews KM, Kalluri R: Clinical and therapeutic relevance of cancer-associated fibroblasts. Nat Rev Clin Oncol 2021, 18(12):792-804.
Sterner RC, Sterner RM: CAR-T cell therapy: current limitations and potential strategies. Blood Cancer J 2021, 11(4):69.
Wang H, Kaur G, Sankin AI, Chen F, Guan F, Zang X: Immune checkpoint blockade and CAR-T cell therapy in hematologic malignancies. J Hematol Oncol 2019, 12(1):59.
Prinzing B, Zebley CC, Petersen CT, Fan Y, Anido AA, Yi Z, Nguyen P, Houke H, Bell M, Haydar D et al: Deleting DNMT3A in CAR T cells prevents exhaustion and enhances antitumor activity. Science translational medicine 2021, 13(620):eabh0272.
Wilky BA: Immune checkpoint inhibitors: The linchpins of modern immunotherapy. Immunol Rev 2019, 290(1):6-23.
Bagchi S, Yuan R, Engleman EG: Immune Checkpoint Inhibitors for the Treatment of Cancer: Clinical Impact and Mechanisms of Response and Resistance. Annu Rev Pathol 2021, 16:223-249.
Castro F, Cardoso AP, Goncalves RM, Serre K, Oliveira MJ: Interferon-Gamma at the Crossroads of Tumor Immune Surveillance or Evasion. Front Immunol 2018, 9:847.
Borden EC: Interferons alpha and beta in cancer: therapeutic opportunities from new insights. Nat Rev Drug Discov 2019, 18(3):219-234.
Su K, Lin N, Xie S, Han Y, Yang Z, Zhang H, He H, Zhou SA, Ma W, Zhang T et al: DNMT3A inhibits E2F1-induced arterial marker expression and impairs angiogenesis in human umbilical artery endothelial cells. Acta Biochim Biophys Sin (Shanghai) 2020, 52(11):1236-1246.
Zhou WT, Jin WL: B7-H3/CD276: An Emerging Cancer Immunotherapy. Front Immunol 2021, 12:701006.
Picarda E, Ohaegbulam KC, Zang X: Molecular Pathways: Targeting B7-H3 (CD276) for Human Cancer Immunotherapy. Clin Cancer Res 2016, 22(14):3425-3431.
Chan TA, Yarchoan M, Jaffee E, Swanton C, Quezada SA, Stenzinger A, Peters S: Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann Oncol 2019, 30(1):44-56.
Jardim DL, Goodman A, de Melo Gagliato D, Kurzrock R: The Challenges of Tumor Mutational Burden as an Immunotherapy Biomarker. Cancer cell 2021, 39(2):154-173.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Pan-cancer analysis of DNMT3A as a robust prognostic biomarker to predict the immunotherapy response

Status:

Version 1

Abstract

Background

Methods

Results

Figures

Introduction

Methods And Materials

Clinical GBM samples and Western blot analysis

Immunohistochemistry and immunofluorescence

Multimodal survival analysis of DNMT3A in pan-cancer

Gene Set Enrichment Analysis of DEGs

Immune cell infiltration in tumors

Correlation analysis of biomarkers and DNMT3A expression

Prognostic validation in immunotherapy cohorts

Statistical processing method

Result

Single-cell sequencing analysis of DNMT3A

Higher DNMT3A expression predicts a worse prognosis in pan-cancer

GSEA analysis of DNMT3A in pan-cancer

Infiltration analysis of immune-related cells

Analysis of DNMT3A-related immune characteristics

The predictive ability of DNMT3A in immunotherapy cohort

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1