Introduction
Pancreatic adenocarcinoma (PAAD) is a common malignant tumor in the digestive tract,1 which is known as the “king of cancer” in the field of cancer. As one of the worst prognosis malignancies, PAAD’s five-year survival was only 2–9% owing to its substantial invasion and migration abilities.2 Despite the low incidence of PAAD, it is the fourth leading cause of cancer-related death and is expected to be the second most common cancer-related cause of death by 2030, surpassing colorectal and breast cancer.3 Current new therapies for PAAD mainly include immunotherapy combinations, targeted DNA repair, and tumor metabolism. Although with new treatment options, the survival rate has not improved significantly in the past few decades.4
The Bloom’s syndrome helicase (BLM), a member of the human RecQ helicase family,5 has typical RecQ helicase family structural features and plays a vital role in the cellular metabolic processes, such as DNA replication and recombination, transcription, repair, and maintenance of the telomeres.6 Moreover, studies have shown that mutations in the BLM gene lead to the development of Bloom’s syndrome, patient genetic instability, and susceptibility to various cancers, such as breast, lung, prostate, and other malignancies.7 During the double-strand break (DSB) repair, BLM is rapidly recruited to the damaged DNA to participate in the homologous recombination repair (HRR),8 and is considered one of the 50 best candidate targets for cancer treatment in the DNA damage repair pathway.9 Regarding the relationship between BLM and PAAD, Kondo et al., by sequencing coding exons of HRR-related genes (including BLM), found that PAAD patients with HRR-related mutations had long median progression-free survival compared with those without mutations.10 However, the BLM expression and regulatory mechanism in PAAD and its effect on the prognosis still remain unclear.
In recent years, the study of the tumor immune microenvironment has received wide attention. In general, tumor cells can use immune checkpoint regulation to escape the immune response.11 Programmed death-ligand (PD-L1), the most representative immune checkpoint, has also been detected for its high expression in cancers, which is associated with a poor prognosis in breast cancer, colorectal cancer, gastric cancer, non-small cell lung cancer, and other cancers.12 Additionally, PAAD has been described as a type of tumor with rich immunosuppressive mechanisms, and its complex tumor microenvironment (TME) has been regarded as a barrier to systemic treatment. Simultaneously, the DNA repair pathway involved by BLM is also related to the development and maturation of immune cells.13 However, whether BLM involves regulating the immune microenvironment in PAAD has still not yet been determined.
Therefore, in this study, we focused on exploring the regulatory mechanism of BLM and researching the impact of BLM on the immune microenvironment and prognosis in PAAD. First, the pancancer and survival analysis of BLM in multiple cancer types was performed. Next, the genetic and epigenetic regulations of BLM were investigated, including the gene mutation, DNA methylation, transcription factors (TFs), and the noncoding RNA (ncRNA) in PAAD. Finally, the relationship of the expression of BLM with immune cell infiltration, biomarkers of immune cells, and immune checkpoints in PAAD were further explored. Taken together, our results suggested that the high expression of BLM regulated by abnormal methylation, the TFs, and the LINC01133-miR-30b-5p axis was associated with the poor prognosis and tumor immune infiltration in PAAD.
Materials and methods
Downloading and analysis of the cancer data
Raw data of 33 cancer types and normal samples were downloaded from the TCGA and GETx databases by UCSC Xena (https://xena.ucsc.edu/ ). The data were normalized using R software (the TPM values were extracted, log2(x+0.001) transformed, and combined).
Gene differential expression analysis
The R software, a tool with powerful data processing and mapping capabilities, has been widely used in genomic and transcriptomic studies.14 In this study, the differential expression analysis was performed by R software’s “limma” package.
Gene Expression Profiling Interactive Analysis (GEPIA) of the profiling of the gene expression (http://gepia.cancer-pku.cn/ ), an online database constructed by Peking University, China to analyze differentially expressed genes in cancer and normal tissues, has become an effective way to identify prognostic markers and molecular targets for cancer therapy.15 The BLM and related long noncoding RNAs (lncRNAs) expression and prognosis influence in cancer was validated in this database, in addition to assessing the correlation of the immune checkpoints and BLM gene in PAAD.
BLM protein differential expression analysis
The Human Protein Atlas (HPA) (https://www.proteinatlas.org/ ) is an antibody-based method for immunostaining on tissues and cell lines. It can also be used for differential expression analysis of normal tissue and tumor tissue proteins.16 The results of the immunohistochemistry of the BLM histochemical antibody (No: HPA005689) on the normal pancreas and PAAD tissues were quoted.
Gene mutation analysis
The cBioPortal website (https://www.cbioportal.org/ ) for cancer genomics provides a web resource for exploring, visualizing, and analyzing multidimensional cancer genomics data.17 We used this website to analyze the types and loci of the BLM mutations in cancer.
BLM methylation analysis
The Shiny Methylation Analysis Resource Tool (SMART; http://www.bioinfo-zs.com/smartapp/ ) is a user-friendly and easy-to-use web application for comprehensively analyzing DNA methylation based on TCGA data.18 Here we analyzed the differential expression of methylated BLM in different cancers and performed survival analyses.
Prediction of the potential upstream transcription factor targets of BLM
ARCHS4 (https://maayanlab.cloud/archs4/help.html ) was used to predict the upstream transcription factor targets (TFs), which predicted upstream TFs based on the ChIP-seq data from the ChEA and ENCODE gene set libraries.19 The JASPAR2022 database (https://jaspar.genereg.net/ ) was used to predict the binding sites for the transcription factors in the promoter regions of the genes.
Exploring the upstream ncRNAs of the BLM
Starbase (http://starbase.sysu.edu.cn/ ) is a database that can explore the upstream ncRNAs of the regulatory gene.20 Here, we used Starbase to select the upstream ncRNAs regulating BLM.
Survival analysis
The Kaplan-Meier plotter (http://kmplot.com/Analysis/ ), an open database, is capable of conducting the survival analysis of more than 20 types of cancers.21 We used this database to analyze the effect of the candidate microRNAs (miRNAs) on the prognostic OS in PAAD.
Analysis of the immune cell infiltration level and immune checkpoint expression
TIMER (https://cistrome.shinyapps.io/timer/ ) is an online database for comprehensively analyzing tumor-infiltrating immune cells.22 We used this database to analyze the relationship of the expression of BLM with the level of the immune cell infiltration or immune checkpoints in PAAD.
Other databases and software
UALCAN (http://ualcan.path.uab.edu/index.html ) was used to verify the gene and protein expression in PAAD, and Cytoscape software was used to map the relationship of the lncRNA-miRNA/BLM network.
Statistical analysis
For the results obtained from all the databases and analysis software, p-value < 0.05 was considered a statistically significant difference. A t-test was used to compare the mean between the two groups. Correlation analysis was performed using Pearson’s correlation based on bivariate normality. The survival curves were plotted with the Kaplan-Meier method and tested by the log-rank test.
Results
Pancancer analysis of the BLM expression
To explore the potential oncogenic role of BLM, the BLM expression level in 33 human cancers was analyzed by the TCGA and GTEx databases. The results displayed that BLM was highly expressed in ACC, BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KIRC, KIRP, LGG, LIHC, LUAD, LUSC, OV, PAAD, PCPG, PRAD, READ, SKCM, STAD, TGCT, THCA, UCEC, and UCS compared with the normal samples (p < 0.05), while there was no statistical difference in KIHC, LAML, MESO, SARC, and UVM (Fig. 1a). Subsequently, the expression of BLM was validated in the GEPIA database and found that it was also highly expressed in 14 cancers, including BLCA, BRAC, COAD, etc. (Fig. 1b). These results suggested that BLM could be a potential oncogene in human cancer carcinogenesis.
Prognosis analysis of BLM in cancer
The survival analysis, including two prognostic indices (OS and RFS), was performed on the above-mentioned cancer patients. The results indicated that the PAAD patients with a high BLM expression had a significantly poor prognosis for OS and RFS (Fig. 2). Thus, BLM could be a potential biomarker for PAAD prognosis.
The expression of the BLM protein in PAAD
The BLM protein level in the PAAD tissues was further researched from the HPA database. The high expression of the BLM protein only appeared in the endocrine cells of the PAAD tissue. However, the results did not indicate any significant differences compared to the normal tissue samples (Fig. 3a–b). In considering that the sample size of the normal tissue was too small to be compared with the HPA database, the CPTAC database was further used to verify the protein expression. Unsurprisingly, the results were consistent with our predictions, and the expression of the BLM protein was higher in the PAAD tissue (Fig. 3c).
Analysis of the genetic alteration of the BLM
For researching the genetic alteration of the BLM, we analyzed the different mutation types of BLM in cancer through the cBioPortal web. The results showed that BLM had mass mutations in multiple types of cancer, and the amplification mutations were the primary type in PAAD (Fig. 4a). The types and loci of the BLM mutation in PAAD were further explored. The missense and truncating were the dominating types, and the latter type occurred in the S1252F/Y on the HDRC domain (Fig. 4b). Thus, the mutations of BLM could be a cause of the high BLM expression in PAAD. However, mutations in the BLM gene did not significantly impact the prognosis of the PAAD patients (Fig. 4).
Analysis of the methylation level of the BLM promoter in PAAD
Methylation is an essential epigenetic regulatory mechanism for DNA. Therefore, a pancancer
analysis for the methylation level of the CpG site in the BLM promoter was performed (Fig. 5a). We found that the methylation level of the BLM promoter was significantly reduced in PAAD (Fig. 5b) and was positively correlated with OS (Fig. 5c–d). These findings indicated that the low methylation level of BLM could cause the high expression of BLM and affect the prognosis in PAAD patients.
Prediction of the transcription factor targets in the BLM upstream
TFs usually have competitive inhibitory relationships with DNA methylation and play an essential role in the gene transcription process. Therefore, we used the ARCHS4 resource to investigate the upstream TFs of the BLM further. Consequently, 48 upstream TFs of BLM were predicted in humans (Supplementary Table S1). Simultaneously, the JASPAR2022 database was used to verify whether these 48 TFs had binding sites in the BLM promoter, and 23 of them were revealed to have binding sites (Supplementary Table S2). Subsequently, the correlation analysis in PAAD showed that seven TFs (E2F4, E2F1, ETS1, MYCN, ELK1, IRF1, and FOXP3) were positively correlated with the BLM expression (Fig. 6a) and that five TFs (E2F1, ETS1, ELK1, IRF1, and FOXP3) were also highly expressed in PAAD (Fig. 6b). Therefore, these five TFs might be the most potential upstream TFs of BLM. Similarly, survival analysis was also performed, in which only E2F1 and ETS1, consistent with BLM, were correlated with a poor prognosis (OS or RFS) in PAAD (Fig. 6c).
Prediction and analysis of the ncRNAs upstream of the BLM
It is widely believed that ncRNAs, including miRNAs and lncRNAs, can regulate gene expression and are considered an epigenetic regulatory mechanism. Therefore, based on the Starbase database, we predicted the possible upstream miRNAs of BLM. Finally, we selected 11 miRNAs that were negatively correlated with the BLM expression (miRNA usually suppresses the target gene expression) (Fig. 7a). To intuitively visualize this situation, we drew the relationship network of these 11 miRNAs with BLM using Cytoscape software (Fig. 7b). Then, we determined these miRNAs’ expression and prognostic value individually in PAAD. Only miR-30b-5p was downregulated in PAAD, and its downregulation was positively correlated with the poor prognosis and disease course (Fig. 7c–e). These findings suggested that miR-30b-5p could be the most potential regulatory miRNA for the BLM in PAAD.
According to the competitive endogenous RNA (ceRNA) hypothesis, lncRNA could increase the mRNA expression by competitively binding to a shared miRNA.23 Consequently, we used the Starbase database to predict the upstream lncRNAs binding to miR-30b-5p and selected 12 lncRNAs (Fig. 8a–b). The correlation of these lncRNAs with BLM in PAAD was verified, and only LINC00657 (NORAD), LINC00707, LINC02535, and LINC01133 were positively correlated with BLM (Fig. 8c). To further verify whether they were also highly expressed in PAAD, we performed a differential expression analysis on the GEPIA database, and only LINC00657 and LINC01133 were confirmed (Fig. 8d). Finally, these two lncRNAs were defined as candidates for further prognostic analysis. LINC01133, consistent with the BLM, significantly correlated with the poor prognosis (OS) and progression in PAAD (Fig. 8e–f). For the preceding reasons, LINC01133 was selected as the most likely upstream lncRNA in the miR-30b-5p/BLM axis.
Differential expression analysis for the BLM related-genes in PAAD
The PAAD patients were divided into high and low expression groups according to the BLM expression. Differential expression analysis was performed to explore the BLM’s potential carcinogenic mechanism. A total of 106 differentially expressed genes (DEGs) with |logFC| ≥ 1 and p < 0.001 between the two groups were determined, as shown in the heatmap and the volcano plot (Fig. 9a–b). Subsequently, a GO and KEGG functional enrichment analysis for DEGs revealed these genes were involved in some biological processes and pathways closely related to cancer development, including maintaining chromatin stability, cell cycle, and p53 signaling pathway (Fig. 9c–d).
Relationship of immune cell infiltration with BLM in the PAAD cells
As mentioned in the introduction, there were some associations between BLM and the cancer’s immune microenvironment. Thus, by first comparing the expression of BLM in different immune cells from PAAD and the normal samples, we found that BLM was highly expressed in various immune cells from PAAD (Fig. 10a). To validate the regulatory role of BLM on the immune cells in PAAD, the level of immune cell infiltration at different BLM copy numbers in PAAD was examined using the TIMER database. There were significant differences in the immune infiltration level of the B cells and CD4+ cells under the different BLM copy number variations (Fig. 10b). Furthermore, a correlation analysis of BLM with immune cell infiltration level was performed. The immune infiltration of the B cells, neutrophil cells, and dendritic cells were positively correlated with BLM in PAAD (Fig. 10c–d). Therefore, BLM could facilitate the level of the PAAD immune infiltration.
Relationship of the immune cell biomarkers with BLM in PAAD
To further validate the role of BLM in PAAD immunity, we examined the correlation of BLM with the expression of the immune cell biomarkers in PAAD by using the GEPIA database. Unsurprisingly, the BLM was positively correlated with most immune cell biomarkers (Table 1). These results supported the relationship of BLM with immune cell infiltration in PAAD.
Table 1The correlation analysis of BLM with the biomarkers of the immune cells in PAAD determined by the GEPIA database
| Immune cell | Biomarker | R-value | p value |
|---|
| B cell | CD19 | 0.28 | 0.00013*** |
| CD79A | 0.29 | 0.00011*** |
| CD8+ T cell | CD8A | 0.14 | 0.063 |
| CD8B | 0.17 | 0.02* |
| CD4+ T cell | CD4 | 0.18 | 0.015* |
| M1 macrophage | NOS2 | 0.1 | 0.17 |
| IRF5 | 0.16 | 0.029* |
| PTGS2 | 0.11 | 0.16 |
| M2 macrophage | CD163 | 0.27 | 0.00022*** |
| VSIG4 | 0.23 | 0.0019*** |
| MS4A4A | 0.17 | 0.02* |
| Neutrophil | CEACAM8 | -0.051 | 0.5 |
| ITGAM | 0.067 | 0.37 |
| CCR7 | 0.25 | 0.00061*** |
| Dendritic cell | HLA-DPB1 | 0.15 | 0.052 |
| HLA-DQB1 | 0.073 | 0.33 |
| HLA-DRA | 0.15 | 0.043* |
| HLA-DPA1 | 0.13 | 0.072 |
| CD1C | 0.089 | 0.23 |
| NRP1 | −0.091 | 0.22 |
| ITGAX | 0.067 | 0.38 |
Relationship of BLM with the immune checkpoints in PAAD cells
To explore whether BLM may exert its oncogenic and prognostic role through immune mechanisms, we analyzed the critical immune checkpoints, PD1, PD-L1, CTLA-4, and CD27 responsible for tumor immune escape. From the TIMER database, we found that the BLM expression was positively correlated with the PD1 (PDCD1), PD-L1 (CD274), CTLA-4, and CD27 gene expressions in PAAD (Fig. 11a–c). Simultaneously, these results were confirmed in the GEPIA database (Fig. 11d–f). Consequently, the tumor immune escape could be involved in BLM-mediated PAAD carcinogenesis.
Discussion
There was the highest incidence of PAAD in all pancreatic tumors accounting for approximately 85% of all cases.24 Due to the lack of unique clinical symptoms in the early stage, the early detection and treatments for PAAD patients are more difficult, and the five-year survival rate of PAAD is the lowest among all pancreatic tumor types.25 Therefore, early prevention and effective judgment of prognosis are the keys. Although significant progress had been made in early diagnosis, the overall survival rate of PAAD has not improved in recent years.26,27 This requires us to exhaustively study its carcinogenic mechanisms, identify potential molecular markers, and find new therapeutic targets. In recent years, the role of BLM in DNA damage repair and the carcinogenesis mechanism has gradually received attention. However, the regulatory mechanism and oncogenic role in PAAD still remain unknown.
In this study, a pancancer analysis using the TCGA and GETx databases showed that BLM was highly expressed in most cancers, which meant that BLM had extensive oncogenic effects. We further performed a prognostic analysis in cancer patients through two survival indicators (OS and FRS), and found that the high expression of BLM had a significantly poor prognosis for the PAAD patients. Consequently, BLM could also be a prognostic marker of PAAD. To investigate the regulatory mechanisms of the high BLM expression in PAAD, we explored both the genetic and epigenetic regulations. As with most cancers and their mechanisms of origin, PAAD is also initiated by mutations in the proto-oncogene, tumor suppressor genes, and DNA repair genes. Both activating and inactivating mutations in these genes lead to cancer progression.28 Moreover, Bononi et al. found that mutations in DNA damage repair genes were present in various cancers.29 Thus, we analyzed the frequency and type of the BLM mutations and found many amplification mutations in BLM, and that structural mutation appeared on S1252F/Y of the HDRC domain, which could partly be responsible for the high BLM expression. At the same time, the mutations in BLM had no significant effect on the prognosis of the PAAD patients.
As for the epigenetic aspects, DNA methylation is a stable epigenetic modification that contributes to the spatiotemporal regulation of the gene expression. Traditionally, dense promoter DNA methylation has been associated with the transcriptional repression.30 Therefore, we verified whether the methylation level was also changed in the BLM promoter region. As expected, the methylation level of the BLM promoter was significantly reduced in PAAD compared with the normal tissues, thus suggesting that the high BLM expression could be regulated by hypomethylation. Next, we also explored the potential transcription factor targets of BLM, which could be bound to specific gene sequences to play a role in regulating their expression and form gene regulatory networks (GRNs).31 Combined with the target prediction and correlation analyses, we identified the five most likely potential upstream TFs of BLM. We found that TFs (E2F1 and ETS1), consistently with BLM, marked a poor prognosis in the PAAD patients.
In addition to DNA methylation and TFs, the effect of ncRNA on mRNA is also an essential epigenetic regulatory mechanism. The ceRNA hypothesis suggests that multiple RNAs are involved in the regulatory network of ceRNA, including lncRNA, miRNA, circular RNA, and mRNA. Among them, miRNA can bind to the untranslated region of mRNA to inhibit the translation of mRNA.32 In contrast, lncRNA competitively binds to miRNA through microRNA response elements (MREs) and inhibits the downregulation of mRNA mediated by miRNA.33 Many lncRNA-miRNA/mRNA axes have been confirmed to be involved in regulating tumor development.34 Therefore, we next explored the regulatory axis of the miRNA-lncRNA upstream of BLM by combining the consistency of the correlation and prognostic analysis, and finally determined that the LINC01133-miR-30b-5p axis could be involved in the regulation of BLM in PAAD. Moreover, the LINC01133-miR-30b-5p axis regulates Rab3D in renal cell carcinoma (RCC).35
It is well-known that DNA repair defects can lead to immunodeficiency, as DNA repair is essential for developing antigen receptors expressed in both the B and T lymphocytes. These antigen receptors are formed by the recombination of the variables (V), diversity (D), and the (J) genes at the antigen receptor locus.36 During the V, D, and J recombination, non-homologous end joining (NHEJ) and class-switch recombination (CSR) DNA repair pathway connections would be required.36 BLM is also a helicase involved in DNA repair, and immune cells are an essential factor affecting pancreatic cancer’s pathogenesis and disease severity.37 Therefore, we next further analyzed the relationship of BLM with the immune cell infiltration in PAAD. The results showed that BLM expression was positively correlated with the infiltration level of the multiple immune cells, including the B cells, neutrophil, and dendritic cells. Simultaneously, combined with the analysis of the immune markers, we could reasonably speculate that BLM promotes the infiltration of the immune cells in PAAD.
To further explore whether BLM exerts its oncogenic and influencing prognostic effects through the immune system, we considered the infiltration level of the immune cells and the ‘immune checkpoints’ mechanism. A higher level of immune cell infiltration usually suggests a good prognosis.38 In contrast, the immune checkpoint mechanism indicates that tumor cells can utilize checkpoint-regulated inhibitory effects to evade the immune response.39 PD-1, the most common immune checkpoint, is expressed in T lymphocytes and other immune cells, while PD-L1 is one of its ligands, and the binding of PD-1 and PD-L1 causes inhibitory signals in the T cells.40 Immune checkpoint CTLA-4 is a receptor with inhibitory effects on T lymphocyte activation.41 CD27, a co-stimulatory immune-checkpoint receptor, was expressed on a broad range of T-cells, NK-cells, and B-cells.42 Our results demonstrated that the high expression of BLM was correlated with the most immune cells infiltration level in PAAD, while the prognosis in PAAD with the high level of BLM was poor. Therefore, we validated the relationship of BLM with PD-1, PD-L1, CTLA-4, and CD27 in PAAD. Not unexpected, BLM showed a significant correlation with all three immune checkpoints. These results could explain why PAAD patients with a high level of BLM had a poor prognosis.
Taken together, the high expression of BLM regulated by gene mutation, hypomethylation, TFs, and the LINC01133-miR-30b-5p/axis was associated with a poor prognosis of PAAD. This may promote cancer development by regulating the immune checkpoints, consequently affecting its prognosis (Fig. 12). However, this deduction would need to be confirmed by biological experiments in the future.
Conclusions
Our study revealed the critical role of BLM as a potential biomarker in the PAAD prognosis and immune microenvironment by providing new targets for the treatment and prognosis research of PAAD.
Supporting information
Supplementary material for this article is available at https://doi.org/10.14218/CSP.2022.00015 .
Supplementary Table 1
Upstream TFs of BLM in humans based on ARCHS4 resource.
(XLSX)
Supplementary Table 2
The binding sides of 48 related TFs in BLM based on JASPAR2022 database.
(XLSX)
Abbreviations
- BLM:
Bloom’s syndrome helicase
- ceRNA:
competitive endogenous RNA
- DEGs:
differentially expressed genes
- HR:
homologous recombination repair
- ncRNA:
non-coding RNA
- OS:
overall survival
- PAAD:
pancreatic adenocarcinoma
- RFS:
relapse free survival
- TFs:
transcription factors
- TME:
tumor microenvironment
Declarations
Acknowledgement
We thank all of the participants for their contributions to the study.
Data sharing statement
Public data sets used for this study can be found in the Cancer Genome Atlas (https://portal.gdc.cancer.gov/).
Funding
This study was supported by the Scientific Research Project of Wenzhou Medical University for Talent (QTJ20010) and the National Natural Science Foundation of China (11635013).
Conflict of interest
The authors declare no competing interests.
Authors’ contributions
QL designed the study protocol and wrote the manuscript. A-HS and PZ were responsible for the literature review and the manuscript discussion. H-XH and Y-NZ contributed to the statistical analyses. J-FY gave valuable suggestions. B-RH designed, advised, and revised the manuscript. All authors contributed to the article and approved the submitted version.