Targeting inhibition of prognosis-related nicotinamide metabolism genes, including poly (ADP-ribose) polymerase 9 (PARP-9) attenuates glioma progression

doi:10.21203/rs.3.rs-4618172/v1

Download PDF

Article

Targeting inhibition of prognosis-related nicotinamide metabolism genes, including poly (ADP-ribose) polymerase 9 (PARP-9) attenuates glioma progression

https://doi.org/10.21203/rs.3.rs-4618172/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Nicotinamide (NAM) metabolism plays a significant role in glioma development. This study aimed to investigate the correlation between NAM metabolic genes and prognosis, immune microenvironment, and tumor progression in glioma.

Methods

We analyzed the expression and prognostic significance of NAM-metabolism-related genes in patients with glioma to develop a new NAM metabolism-related signature (NMRS) and nomograms using data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases and compared the differences in gene set enrichment analysis (GSEA), protein-protein interaction networks, competing endogenous RNA (ceRNA) regulatory network, prognosis, mutation load, and immune landscape between different groups. Additionally, we employed Western blotting, cell proliferation and apoptosis analysis, Semi-quantitative Reverse Transcription-Polymerase Chain Reaction(SqRT-PCR), and a xenograft model in nude mice to investigate the role of PARP9 in tumor progression.

Results

Our study identified eight genes, including NT5C1A, NNMT, CDKN1A, PTGS2, PNP, PARP10, PARP14, and PARP9, that exhibited a significant correlation with glioma prognosis and could act as an independent indicator. Risk stratification was conducted based on the NMRS, and the low-risk group exhibited more favorable clinical results. The GSEA revealed that the low-risk group exhibited a significant enrichment in immune-associated pathways, while the high-risk group showed significant enrichment in cancer-related pathways. The ESTIMATE and single-sample GSEA algorithms indicated that the low-risk group displayed higher antitumor immunocyte infiltration. TIDE analysis revealed that the low-risk group responded more favorably to immunotherapy. Furthermore, validation experiments revealed that PARP9 is a proto-oncogene associated with the PARP9-JAK2-STAT3 signaling pathway.

Conclusion

We developed a new NMRS for predicting prognosis and treatment efficacy in glioma. The identified gene, PARP9, is a potential therapeutic target for glioma.

Biological sciences/Cancer

Biological sciences/Computational biology and bioinformatics

Nicotinamide metabolism

PARP9

Tumor immune microenvironment

Immune infiltration

Bioinformatics

Glioma

Glioma is the most prevalent and lethal malignant neoplasm of the brain, accounting for approximately 100,000 newly diagnosed cases worldwide annually^[1]. Conventional therapeutic methods include surgical intervention, chemotherapy, and radiotherapy; however, these approaches have limitations. Given the severe invasive and angiogenic characteristics of glioblastoma, achieving a complete cure is difficult, and there is a considerable risk of recurrence. Although temozolomide is recommended in the treatment guidelines to improve survival rates, its impact on the 5-year survival rate remains restricted. Radiotherapy is limited because of the pronounced radioresistance ^[2]. Additional treatments, including the recombinant nonpathogenic polio-rhinovirus chimera and bevacizumab, have been prescribed for patients with recurrent glioblastoma. However, their effectiveness remains constrained ^{[3, 4]}. Despite multiple treatment approaches, the average overall survival in patients with glioblastoma is 14.6 months ^[5]. Consequently, investigating the molecular biological attributes of glioma, identifying biomarkers capable of predicting prognosis and immunotherapy response, and discovering new treatment approaches are crucial.

Glioma exhibits adaptive metabolic reprogramming through unique pathways, which alters energy metabolism in the tumor microenvironment, resulting in unregulated cancer cell proliferation. Nicotinamide (NAM) is a component of metabolic pathways ^[6] and a precursor for the coenzyme nicotinamide adenine dinucleotide (NAD⁺), which represents a crucial enzyme contributing to biological redox reactions, signal transmission, and cellular energy metabolism, including processes such as DNA damage repair, the tricarboxylic acid cycle, and epigenetic control ^{[7, 8]}. Previous studies reported that NAM metabolism participates in cancer immunosuppression, enhancing tumor blood flow and hypoxia microenvironment ^{[9, 10]}. Approximately 90% of solid tumors are sensitive to NAM depletion^[11]. Pharmacological inhibition of the NAD salvage biosynthesis pathway, specifically targeting the molecule NAM with the inhibitor KPT-9274, has been demonstrated to decrease NAD levels and the associated metabolites, hinder cell growth, and trigger cell death within in vitro cell lines and ex vivo human glioma tissue ^[6]. Consequently, we recognize the importance of NAM metabolism in tumor prognosis and immunity and its potential value as a prognostic indicator. Although Jiang C et al. and Jiang C et al. investigated the relationship between NAD⁺ metabolism and the progression and prognosis of patients with glioma, these studies only constructed a simple survival risk model^{[12, 13]}. They did not perform multi-level bioinformatics analysis, clarify the role of key genes in signaling pathways, or examine the immune environment. Additionally, there was no experimental validation of the reliability of the model .

Hence, this study constructed a prognostic model by fully mining key genes to predict the survival risk of patients with glioma and compared the differences between the two groups regarding gene set enrichment analysis (GSEA), protein-protein interaction (PPI) networks, ceRNA regulation networks, prognosis, clinical features, mutation load, immune cell infiltration, and tumor microenvironment (TME). Furthermore, this study validated key genes and performed cellular and animal-based experiments.

2.1 Study design

We obtained the required data from TCGA, GEO, and GeneCards databases. We identified hub genes using weighted gene co-expression network analysis (WGCNA) analysis. We used the Mann–Whitney U test and consensus clustering to analyze the expression differences of these hub genes. We constructed a prognostic model using least absolute shrinkage and selection operator (LASSO) regression, followed by gene ontology (GO)-based functional enrichment analysis, pathway enrichment Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, GSEA, and constructed PPI and ceRNA regulation networks. Additionally, we used the ESTIMATE, ssGSEA and TIDE algorithms to identify of the immune landscape. Subsequently, we constructed a nomogram based on univariate and multivariate Cox regression analyses, and evaluated the accuracy of the nomogram using calibration curve and decision curve analysis (DCA). Furthermore, we examined PARP9 expression in glioblastoma cell lines and tissue samples. By transfecting Small interfering RNA (siRNA) to knock down PARP9 in U251 cells, we observed changes in proliferation, apoptosis, and JAK2-STAT3 signaling. Subsequently, we established an orthotopic xenograft tumor model in nude mice to validate the changes in glioblastoma proliferation after downregulating PARP9 (specifically see Supplementary Materials: Materials and Methods section).

2.2 Statistical analysis

Data processing and analysis were performed using R software (version 4.2.2). Mann–Whitney U-test was used to analyze non-normally distributed continuous variables. The correlation coefficient between different molecules was calculated using Spearman correlation analysis. The Kaplan–Meier survival curve was used to display the survival difference, and the log-rank test was used to evaluate the significance of the difference in survival time between the two groups. Unless otherwise specified, all statistical P values are two-sided, and P < 0.05 is considered statistically significant.

3.1 Technology Roadmap

3.2 Data gathering

We standardized the two glioma data sets, and there are 284 samples in the GSE16011 data set, including 8 normal samples (control) and 276 glioma samples (GBMLGG) (Supplementary Figs. S1A-B). GSE50161 data set has 230 samples, including 13 cases of normal samples (control) and 117 cases of glioma samples (GBMLGG) (Supplementary Figs. S1C–D).

3.3 Nicotinamide metabolism score (NAMS) construction and clinical prognosis analysis

We initially intersected 50 NAM metabolism-related genes (NAMRGs) with genes in datasets TCGA-GBMLGG, GSE16011, and GSE50161 and obtained 39 NAMRGs for subsequent analysis (Fig. 2A). We obtained the NAMS of each sample in the three data sets through the single-sample GSEA (ssGSEA) algorithm, then we divided the glioma samples into high and low groups based on the median of NAMs (high score group: high, low score group: low). Stacked histograms (Figs. 2B–D) show the the differences between the two groups in patient age, overall survival, and gender (age, OS, gender). And OS Kaplan–Meier (KM) curve (Fig. 2E) show that low group exhibited a better prognosis than high group, with statistically significant (P < 0.001).Finally, we performed differential analysis on the counts data of the TCGA-GBMLGG dataset datasets GSE16011 and GSE50161, and plotted the volcano map (Figs. 2F–H) and the heat map of NAMRGs (Fig. 2I) to show the results.

3.4 WGCNA

In order to analyze the difference in gene expression between the high and low groups of NAMS, we performed WGCNA to screen for co-expression modules. We obtained the correlation between six modules (MEyellow, MEbrown, MEturquoise, MEblue, MEgreen, and MEgrey) and two groups (Fig. 3A-E). We intersected the 39 NAMRGs with the module genes contained in the MEbrown (Fig. 3F), MEgrey (Fig. 3G), and MEturquoise (Fig. 3H) modules and drew a Venn diagram to obtain the module NAMRGs (Figs. 3F–H). We obtained 11 module NAMRGs, PARP10, PARP14, PARP9, CDKN1A, NT5E, NAPRT, PTGIS, PTGS2, NNMT, NT5C1A, and PNP. These 11 genes were used as hub genes for subsequent analysis.

3.5 Mutation analysis and copy number variation of hub genes

The results of the genetic variation analysis of hub genes showed that the main mutation types were missense mutations, nonsense mutations, and frameshift mutations (Frame Shift Ins). Among them, missense mutations account for the main part. Additionally, the mutation types in the hub genes were mainly single nucleotide polymorphisms, and C > T was the most common single nucleotide variant, followed by C > A (Supplementary Fig. S2A). By analyzing the copy number variation of hub genes (Supplementary Fig. S2B), we found that most of the genes exhibited copy number changes (Altered group), among which the gene NT5C1A exhibited a higher frequency of copy number alterations. Furthermore, the copy number changes of half of the genes are concentrated in the deep deletion (Supplementary Fig. S2C). For example, the gene NT5C1A has a high frequency of copy number deep deletion (> 30%) and a low frequency (< 10%) of significant amplification.

3.6 Expression difference analysis of hub genes

The results of difference analysis showed that the ten hub genes in the TCGA-GBMLGG exhibited higher expression in the high group than in the low group, with statistically significant. (Supplementary Fig. S3A): CDKN1A, NNMT, NT5C1A, NT5E, PARP10, PARP14, PARP9, PNP, PTGIS, and PTGS2. And the same is true in GSE16011. (Supplementary Fig. S3B). In GSE50161, compared to low group, 8 hub genes were higher expressed in the high group, with statistical significance (Supplementary Fig. S3C): CDKN1A, NNMT, NT5E, PARP10, PARP14, PARP9, PNP, and PTGS2. According to the results of the group comparison chart, 8 hub genes were higher expressed in the high group in TCGA-GBMLGG, GSE16011, and GSE50161, with statistical significance: CDKN1A, NNMT, NT5E, PARP10, PARP14, PARP9, PNP, and PTGS2. And we plotted the receiver operating characteristic (ROC) curves displayed the results (Supplementary Figs. S3D–O). The ROC curve results are as follows: CDKN1A, NNMT, PARP10, PARP14, PARP9, PNP, PTGS2 (The AUC values were 0.705, 0.765, 0.701, 0.793, 0.827, 0.761 and 0.700, respectively) have certain diagnostic value in TCGA-GBMLGG, but NT5E is distinctive (AUC = 0.649) ; CDKN1A, NNMT, NT5E, PARP14, PARP9, PNP, and PTGS2 (The AUC values were 0.721, 0.810, 0.716, 0.758, 0.770, 0.779 and 0.759, respectively) have certain diagnostic value in GSE16011, but PARP10 is distinctive (AUC = 0.584); CDKN1A, NT5E, PARP10, PARP14, PARP9, PNP, and PTGS2 (The AUC values were 0.715, 0.787, 0.762, 0.825, 0.871, 0.715 and 0.761, respectively) have certain diagnostic value in GSE50161; but NNMT is distinctive (AUC = 0.665).

3.7 Construction of related subtypes of GBMLGG

To explore the heterogeneity of NAMRGs in GBMLGG, based on the expression levels of 11 hub genes, GBMLGG was divided into two subtypes (clusters 1 and 2) by consensus clustering (Supplementary Fig. S4A). Additionally, cluster 1 contained 249 samples, and cluster 2 contained 94 samples. We performed principal component analysis (PCA) on the dataset expression matrix of the two subtype, and the PCA clustering results showed that there were significant differences between the two subtype (Supplementary Fig. S4B). We showed the consistent clustering cumulative distribution function (CDF) graph and delta graph (Supplementary Fig. S4C-D) for different cluster numbers. The figure shows that clustering results for the TCGA-GBMLGG are best. We draw a heat map to show the expression difference of 11 hub genes in the two subtype. The results showed that in the TCGA-GBMLGG, the expression levels of 11 hub genes in the two subtype mostly showed a significant negative correlation (Supplementary Fig. S4E). The KM curve showed that two subtype have statistical heterogeneity between the outcomes of the patients (P < 0.0001, Supplementary Fig. S4F).

Subsequently, we analyzed the expression differences of 11 hub genes between the two subtypes. Supplementary Fig. S4G shows that the ten hub genes expression in the two subtypes has statistical significance (P < 0.05): PARP10, PARP14, PARP9, CDKN1A, NT5E, PTGIS, PTGS2, NNMT, NT5C1A, and PNP. NT5C1A has a higher expression level in cluster 1 than in cluster 2, while the opposite is true for the rest of the genes. (Supplementary Fig. S4G).

3.8 Construction of key genes prognostic model (NMRS)

To determine the prognostic value of the 11 hub genes, we perform a single-factor Cox regression analysis on them and screened out 8 genes significantly associated with OS time ( P < 0.001) that were used as the key genes (PARP10, PARP14, PARP9, CDKN1A, PTGS2, NNMT, NT5C1A, and PNP) for our follow-up research (Fig. 4A). Afterwards; we built a prognostic LASSO regression model based on the key genes (Fig. 4B). According to the minimum partial likelihood deviation, we kept all eight key genes. Additionally, we obtained the LASSO variable trajectory map by visualizing the results (Fig. 4C). To further verify the impact of the 8 key genes on the prognosis of glioma, a multivariate Cox model was constructed by performing multivariate Cox regression analysis on the multivariate Cox regression analysis. A risk factor map showed grouping in the multivariate Cox model ,and its included Risk score, SO, and Heat map (Fig. 4D). ( Risk score = PARP14 * -0.45538137 + PARP9 * 0.660476921 + CDKN1A * 0.034173338 + PTGS2 * 0.2545473 + NNMT * 0.150412959 + NT5C1A * − 0.66393584 + PNP * 0.143448889 )

Then, a time-dependent ROC curve (Fig. 4E) show the impact of the risk score on survival outcomes, and KM curve showed that risk score exhibited a statistical significance in predicting the survival outcome of glioma (P < 0.001, Fig. 4F). Finally, the boxplot showed the similarity between different key genes(Fig. 4G). To verify further the value of the multivariate Cox model, We drew the ROC curve based on the risk score and NAMs. The result showed that the diagnostic results of risk score in both the high and low groups are relatively accurate (AUC = 0.765, Fig. 4H).

3.9 Difference analysis of NMRS

The TCGA-GBMLGG patients had their individual risk scores determined, and they were assigned to different risk groups by the median. We perform differential analysis on the counts data of the TCGA-GBMLGG dataset, and drew the volcano map and the heat map of the eight key genes to show the results(Fig. 5A-B).The results of difference analysis of 8 key genes in the high and low risk groups showed that all key genes have higher expression in the high-risk group, with statistical significance (P < 0.001) (Fig. 5C).In addition, the heatmap shows the correlation between the eight key genes in the glioma sample group (Fig. 5D). Then We drawed correlation scatter plots of four most correlated gene immune cell pairs (Figs. 5E–H), and the results are as follows: PARP9 and NNMT ( R = 0.681, P < 0.001); PARP9 and PARP10 ( R = 0.602, P < 0.001); PARP14 and PARP10 ( R = 0.634, P < 0.001); PNP and PARP9 ( R = 0.569, P < 0.001) (Figs. 5E–H).

3.10 Functional enrichment analysis (GO) and pathway enrichment (KEGG) analysis of key genes

To investigate biological processes of eight key genes, We first performed GO analysis and KEGG enrichment analysis. GO analysis enriched key genes from two categories: biological processes (BPs), and molecular functions (MFs). BPs were primarily enriched in the pyridine-containing compound metabolic process, NAD biosynthesis via the NAM riboside salvage pathway, protein mono-ADP-ribosylation, NAM nucleotide biosynthetic process, and NAM nucleotide metabolic process. MFs were mainly enriched in pentosyl transferase activity, NAD+-protein ADP-ribosyl transferase activity, NAD + ADP-ribosyl transferase activity, glycosyl transferase activity, and transcription corepressor activity. The result of KEGG enrichment analysis showed that these genes were significantly enriched in nicotinate and NAM metabolism, nucleotide metabolism, small cell lung cancer, purine metabolism, and the oxytocin signaling pathway (Table S1, Supplementary Materials) (Figs. 6A–D). Additionally, we draw pathway diagrams to show the basic situation of the five main enriched pathways in the KEGG enrichment analysisc(Supplementary Figs. S5A–E). Finally, we performed the GO/KEGG enrichment analysis of the combined logFC of the eight key genes, the bubble diagram indicates that the eight key genes are mainly enriched in the BPs.

3.11 GSEA between high- and low-risk groups in NMRS

GSEA was conducted to complement and validate the functional annotation of GO analysis and KEGG enrichment analysis. The results show that the high- risk groups are significantly enriched in apoptosis, oxidative phosphorylation, TGF-beta receptor signaling, and JAK-STAT signaling pathway, and low-risk groups are mainly enriched in NO2IL12 pathway and interleukin-10 signaling pathway (Supplementary Figs. S6A–G; Table S2, Supplementary Materials).

3.12 Construction of PPI network, mRNA-miRNA, mRNA-binding protein (RBP), and mRNA-TF interaction network

We conducted a PPI analysis of the eight key genes and the PPI network composed of these eight key genes was visualized (Supplementary Fig. S7A).We predicted miRNAs interacting with eight key genes and drew the mRNA-miRNA interaction network (Supplementary Fig. S7B). The mRNA-miRNA interaction network consists of 6 key genes (CDKN1A, NNMT, PARP14, PARP9, PNP, and PTGS2) and 98 miRNA molecules, forming 196 pairs of mRNA-miRNA interaction relationships (specifically as reported in Table S3 (Supplementary Materials) ).We predicted the RBP interacting with eight key genes and draw the mRNA-RBP interaction network for visualization (Supplementary Fig. S7C). The mRNA-RBP interaction network consists of seven essential genes (CDKN1A, NNMT, PARP10, PARP14, PARP9, PNP, and PTGS2) and 92 RBP molecules, forming 204 pairs of mRNA-RBP interaction relationships (specifically as reported in Table S4 (Supplementary Materials) ). Finally, we searched for transcription factors (TF) that bind to eight key genes. The interaction relationship data of six key genes (PARP10, NT5C1A, PARP14, PARP9, PNP, and PTGS2) and 97 TF were obtained and visualized (Supplementary Fig. S7D). Table S5 (Supplementary Materials) shows the specific mRNA-TF interaction relationship.

3.13 The impact of NMRS on immune assessment

We calculated the immune score, stroma score, and ESTIMA score based on the expression matrix of each glioma using the ESTIMATE algorithm. The results disclosed that the stromal score, immune score, and ESTIMATE score in high-risk group significantly higher than those in low-risk group (P < 0.05) (Fig. 7A-C). This suggests that the high-risk group has stronger invasion and immune escape capabilities. Additionally, We also found that the stromal score, immune score, and ESTIMATE score were all significantly positively correlated with the NMRS. This implies that the higher the score, the more unfavorable the patient's prognosis may be.

3.14 Difference analysis of ssGSEA immune signature

We calculated and displayed the infiltration abundance of 28 types of immune cells in high- and low-risk groups through the ssGSEA algorithm and Mann–Whitney U. The findings indicated that high-risk groups exhibited greater infiltration abundance of immune cells with pro-tumor effects, when compared to the Low-risk group. This difference was statistically significant, with myeloid-derived suppressor cells (MDSC) being the most prominent (P < 0.05) (Fig. 8A) We further calculated the correlation between the abundance of 27 kinds of immune cell infiltration in high- and low-risk groups. The results show that The positive correlation between MDSCs in the high-risk group is stronger, suggesting a more potent mutual promotion effect and closer crosstalk communication.(Figs. 8B–C). Ultimately, we examined the correlation between the 27 kinds of immune cell infiltration abundance and eight key gene expression levels. The results showed that in low-risk groups, PARP14 and PARP9 have a significant positive correlation with immune cells, And in high-risk groups, NT5C1A negatively correlates with immune cells, and other key genes have a positive correlation.(Figs. 8D–E)

3.15 NMRS mutation analysis of high- and low-risk groups

We analyzed the differences in tumor mutation burden (TMB) and microsatellite instability (MSI) between two groups, and also evaluated the sensitivity of both groups of patients to immunotherapy using the tumor immune dysfunction and exclusion (TIDE) algorithm. The findings revealed that low-risk groups exhibit higher levels of MSI and lower TMB (P < 0.001), suggesting a better immune response. there was no statistical significance in TIDE between two groups ( Supplementary Fig. S8A-C). (Supplementary Figs. S8D–E). The results of the copy number variation (CNV) show that in the low-risk groups, PARP10 exhibited the most CNV amplification and the least CNV deletion; while NT5C1A exhibited the opposite pattern. In the high-risk groups (Supplementary Fig. S8E), PARP10 exhibited the most CNV amplification and the least CNV deletion; While PNP displayed the inverse pattern.

3.16 Nomogram establishment based on the multivariable Cox regression model

The clinical information of glioma patients in TCGA was statistically analyzed (Table S6, Supplementary Materials). We performed univariate Cox regression analysis on the expression levels of eight key genes and clinical variables (age, gender). Based on the outcomes of univariate analysis(Supplementary Fig. S9A), we incorporated these indicators into the multivariate Cox analysis in order to construct a multivariate Cox prognostic model (Table 1). Subsequently, based on the results of multivariate Cox analysis, we constructed nomogram and drew calibration curves. The nomogram showed better accuracy in predicting both short- and long-term survival, and was consistent with the results of the calibration curve.(Supplementary Figs.S9B–E) The decision curves demonstrated that the risk score was superior to other factors in forecasting the prognosis of glioma.(Supplementary Figs. S9F–H).

Table 1

COX regression analysis to key genes associated with OS in TCGA-GBMLGG.
Characteristics	Total(N)	Univariate analysis		Multivariate analysis
Characteristics	Total(N)	Hazard ratio (95% CI)	P value	Hazard ratio (95% CI)	P value
age	343		< 0.001
<=50	201	Reference		Reference
> 50	142	5.029 (3.422–7.391)	< 0.001	3.430 (2.211–5.320)	< 0.001
gender	343		0.214
MALE	210	Reference
FEMALE	133	0.796 (0.553–1.145)	0.219
PARP10	343	1.911 (1.574–2.320)	< 0.001	1.309 (0.966–1.774)	0.082
PARP14	343	1.859 (1.513–2.284)	< 0.001	0.830 (0.517–1.333)	0.441
PARP9	343	2.637 (2.156–3.225)	< 0.001	1.366 (0.816–2.285)	0.235
CDKN1A	343	1.397 (1.223–1.596)	< 0.001	0.976 (0.818–1.165)	0.789
PTGS2	343	1.612 (1.361–1.910)	< 0.001	1.200 (0.997–1.445)	0.054
NNMT	343	1.486 (1.388–1.591)	< 0.001	1.203 (1.063–1.362)	0.003
NT5C1A	343	0.229 (0.155–0.339)	< 0.001	0.528 (0.355–0.785)	0.002
PNP	343	2.428 (1.875–3.143)	< 0.001	1.196 (0.846–1.691)	0.311

OS, Overall Survival;TCGA, The cancer genome atlas༛GBMLGG, Glioma。

3.17 PARP9 promotes glioma cell proliferation in vivo and in vitro

PARP9 expression was detected in glioma tissue and normal brain tissue, which showed that PARP9 was significantly upregulated in glioma tissue. We detected PARP9 expression in HA, U87, U251, A172, SHY5Y, and SHG44 cells, and it showed that PARP9 was highly expressed in glioma cells, especially in U251 cells (Figs. 9A–E). The lentivirus was used to construct U251 cell lines with stably silenced PARP9-expression. The CCK-8 assay, Edu assay, and plate colony formation assay demonstrated that knocking down PARP9 expression level could inhibit the proliferation of U251 cells in vitro (Figs. 9F–I and L). In previous bioinformatic analysis, we found that PARP9 might be correlated with the JAK2-STAT3 signaling pathway. Subsequent cell experiments revealed that the phosphorylation levels of JAK2 (p-JAK2) and STAT3 (p-STAT3) in the shPARP9 group were significantly reduced compared to the scramble group. Therefore, PARP9 in U251 cells could activate JAK2–STAT3 signaling (Figs. 9J–K). Orthotopic homograft tumor models were constructed in nude mice to verify the change in the proliferation ability of glioma after the knockdown of PARP9 to determine the function of PARP9 in tumor propagation in vivo. The mice exhibited prolonged survival and decreased tumor volumes compared to those of the scramble group, suggesting the importance of PARP9 for the proliferation of U251 cells and tumorigenicity (Figs. 9M–O).

The prognosis of glioma has been disappointing due to the absence of early diagnosis and the constraints of conventional therapeutic approaches. Metabolic dysregulation is a hallmark of cancer and is vital in malignancy development, including glioma ^[14]. Glioma exhibits a distinct reliance on NAM metabolic pathways. NAM could be catalyzed by NAM phosphoribosyl transferase, known as the initiating factor in the immunosuppressive microenvironment ^[15], to generate NAM mononucleotide, converted into NAD⁺. The Warburg effect refers to the metabolic change of cancer cells caused by NAD⁺, which helps sustain a high glycolysis rate. Furthermore, NAD⁺ is a crucial cofactor and substrate involved in numerous biological processes, besides being important in essential metabolic processes contributing to the survival of cells ^[16–18]. These discoveries have led us to acknowledge the importance and ability to predict the diagnosis and therapy of glioma through NAM metabolism. Herein, we developed a new NMRS for predicting prognosis and investigated the relationship between these genes and the immune environment.

The NMRS comprised eight crucial genes, NT5C1A, NNMT, CDKN1A, PTGS2, PNP, PARP10, PARP14, and PARP9. The first key gene, NT5C1A, performs vital functions in maintaining energy balance, regulating metabolism, and facilitating cell reproduction ^[19]. A previous study reported that NT5C1A exhibits elevated expression in pancreatic cancer and makes cancer cells resistant to gemcitabine ^[20]. However, our research emphasizes the potential of NT5C1A as a potential biomarker in glioma research. The NNMT, the second key gene, exhibits a high expression level in the extracellular matrix surrounding metastatic cancer cells. Increased expression of NNMT enhances ovarian cancer progression and metastasis and hinders prognosis. ^[21]. The study has discovered that NNMT constitutes a key factor in glioma onset and progression ^[22], consistent with our research results. Previous studies and our study have demonstrated that CDKN1A, PTGS2, and PNP participate in the occurrence and development of gliomas^[23–25]

The three remaining key genes, PARP10, PARP14, and PARP9, belong to the PARP family and play a role in the genesis and progression of some cancers ^[26]. PARP10 represents the sole member possessing two ubiquitin-interaction motifs (UIMs). The UIMs of PARP10 specifically interact with K63-polyubiquitin, which serves as a scaffold in signaling pathways, including the nuclear factor-κB (NF-κB) signal transduction pathway. The NF-κB governs crucial cellular functions, including cell proliferation and survival, and significantly regulates inflammation and cancer ^[27]. Furthermore, PARP14 plays a complex role in the formation and advancement of various cancers, including multiple myeloma, prostate cancer, and hepatocellular carcinoma ^[28], and it has been poorly studied in glioma. However, our research helps fill this gap.

The study has provided evidence of dysregulated PARP9 expression in diverse human malignancies, and it promotes the recurrence and metastasis of prostate cancer and breast cancer ^{[29, 30]}. Herein, PARP9 was upregulated in glioma tissue and cell lines. The PARP9 downregulation inhibits U251 cells’ self-renewal ability and proliferation in vivo while reducing tumor growth in situ. Additionally, our study indicated that the role of PARP9 might be associated with the JAK2–STAT3 signaling pathway regulatory axis.

We conducted a mutation analysis of these genes to further understand the immunological properties of key genes. As previously mentioned, missense mutations were the most common. Our analysis of key gene mutations found that most genes exhibited copy number changes in patients with glioblastoma, particularly NT5C1A. Furthermore, NT5C1A mutations occur in glioma and have been associated with drug resistance and a poor prognosis in many cancers, including pancreatic cancer^[20]. In our study that investigated the immunological properties of NMRS, we compared the differences in immune assessment, immune infiltration, TMB, and MSI between groups. Our findings revealed that the results of NMRS’s immune assessment were consistent with those of the stromal, immune, and ESTIMATE scores. Furthermore, the NMRS low-risk groups exhibited higher levels of immune infiltration than the high-risk groups. These results are consistent with our survival analysis outcomes. Additionally, TMB has recently been evaluated as a potential biomarker for predicting the response to immune therapy and has been applied in prospective clinical trials for various tumor types, including glioma^[31]. Our study found that the low-risk groups exhibited a higher MSI than the high-risk groups, while the opposite was true for TMB. Previous studies demonstrated that MSI is prevalent in glioblastoma, and the high mutation burden caused by MSI renders tumors immunogenic^[32], potentially making them sensitive to immunotherapy. Similarly, tumor cells can evade immune detection by increasing TMB, while TMB itself is a predictor of the immune response.This suggests that these patients have lower levels of immune escape and better immune response.

There are limitations to this study. First, we investigated the in vitro and in vivo PARP9 expression and its underlying mechanism; however, further research is needed on other key genes of NMRS. Second, multicenter prospective trials are needed to validate the risk score model further and ensure its applicability in clinical practice.

We have developed a novel NMRS to predict glioma prognosis and immunotherapy response. We preliminarily validated the tumor-promoting role of the key gene PARP9 in gliomas. Our research may serve as a basis for future clinical tools to help doctors assess patient prognosis and provide support for individualized treatment plans.

Acknowledgments

We thank the Research Center of Clinical Medicine, Affiliated Hospital of Nantong University, for technical assistance and equipment support.

Funding

This work was supported by Jiangsu Province's Natural Science Foundation (BK20211115), the Science Research Funding Program of Yancheng (YCBE202317), China Postdoctoral Research Funding Program (2019M661906), and Major Project of Wuxi Municipal Health and Family Planning Commission (Z201705).

Conflict of Interest

The authors declare no conflict of interest.

Data availability statement

The datasets analysed during the current study are available in the The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), UCSC Xena and GeneCards repository, [PERSISTENT WEB LINK :https://portal.gdc.cancer.gGBMLGG/, http://genome.ucsc.edu, andhttps://www.genecards.org/]. The raw data in these databases are publicly available, and researchers can freely access and use them. Additional data are made available in supplementary materials of this manuscript. If there are further needs, the author will provide relevant data upon reasonable request.

Ostrom, Q.T.; Patil, N.; Cioffi, G.; Waite, K.; Kruchko, C.; Barnholtz-Sloan, J.S. Corrigendum to: CBTRUS Statistical Report: Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2013–2017. Neuro-oncology 2022, 24, 1214, doi:10.1093/neuonc/noaa269.
Mudassar, F.; Shen, H.; O'Neill, G.; Hau, E. Targeting tumor hypoxia and mitochondrial metabolism with anti-parasitic drugs to improve radiation response in high-grade gliomas. Journal of experimental & clinical cancer research: CR 2020, 39, 208, doi:10.1186/s13046-020-01724-6.
van den Bent, M.J.; Klein, M.; Smits, M.; Reijneveld, J.C.; French, P.J.; Clement, P.; de Vos, F.Y.F.; Wick, A.; Mulholland, P.J.; Taphoorn, M.J.B.; et al. Bevacizumab and temozolomide in patients with first recurrence of WHO grade II and III glioma, without 1p/19q co-deletion (TAVAREC): a randomised controlled phase 2 EORTC trial. The Lancet. Oncology 2018, 19, 1170–1179, doi:10.1016/s1470-2045(18)30362-0.
Desjardins, A.; Gromeier, M.; Herndon, J.E., 2nd ; Beaubier, N.; Bolognesi, D.P.; Friedman, A.H.; Friedman, H.S.; McSherry, F.; Muscat, A.M.; Nair, S.; et al. Recurrent Glioblastoma Treated with Recombinant Poliovirus. The New England journal of medicine 2018, 379, 150–161, doi:10.1056/NEJMoa1716435.
Zeng, W.; Jiang, H.; Wang, Y.; Wang, C.; Yu, B. TCF3 Induces DNMT1 Expression to Regulate Wnt Signaling Pathway in Glioma. Neurotoxicity research 2022, 40, 721–732, doi:10.1007/s12640-022-00510-w.
Sharma, P.; Xu, J.; Williams, K.; Easley, M.; Elder, J.B.; Lonser, R.; Lang, F.F.; Lapalombella, R.; Sampath, D.; Puduvalli, V.K. Inhibition of nicotinamide phosphoribosyltransferase (NAMPT), the rate-limiting enzyme of the nicotinamide adenine dinucleotide (NAD) salvage pathway, to target glioma heterogeneity through mitochondrial oxidative stress. Neuro-oncology 2022, 24, 229–244, doi:10.1093/neuonc/noab175.
Cairns, R.A.; Harris, I.S.; Mak, T.W. Regulation of cancer cell metabolism. Nature reviews. Cancer 2011, 11, 85–95, doi:10.1038/nrc2981.
Lv, H.; Lv, G.; Chen, C.; Zong, Q.; Jiang, G.; Ye, D.; Cui, X.; He, Y.; Xiang, W.; Han, Q.; et al. NAD(+) Metabolism Maintains Inducible PD-L1 Expression to Drive Tumor Immune Evasion. Cell metabolism 2021, 33, 110–127.e115, doi:10.1016/j.cmet.2020.10.021.
Wang, Y.; Wang, F.; Wang, L.; Qiu, S.; Yao, Y.; Yan, C.; Xiong, X.; Chen, X.; Ji, Q.; Cao, J.; et al. NAD(+) supplement potentiates tumor-killing function by rescuing defective TUB-mediated NAMPT transcription in tumor-infiltrated T cells. Cell reports 2021, 36, 109516, doi:10.1016/j.celrep.2021.109516.
Domínguez-Gómez, G.; Díaz-Chávez, J.; Chávez-Blanco, A.; Gonzalez-Fierro, A.; Jiménez-Salazar, J.E.; Damián-Matsumura, P.; Gómez-Quiroz, L.E.; Dueñas-González, A. Nicotinamide sensitizes human breast cancer cells to the cytotoxic effects of radiation and cisplatin. Oncology reports 2015, 33, 721–728, doi:10.3892/or.2014.3661.
Tateishi, K.; Wakimoto, H.; Iafrate, A.J.; Tanaka, S.; Loebel, F.; Lelic, N.; Wiederschain, D.; Bedel, O.; Deng, G.; Zhang, B.; et al. Extreme Vulnerability of IDH1 Mutant Cancers to NAD + Depletion. Cancer cell 2015, 28, 773–784, doi:10.1016/j.ccell.2015.11.006.
Jiang, C.; Zhou, Y.; Yan, L.; Zheng, J.; Wang, X.; Li, J.; Jiang, X. A prognostic NAD + metabolism-related gene signature for predicting response to immune checkpoint inhibitor in glioma. Frontiers in oncology 2023, 13, 1051641, doi:10.3389/fonc.2023.1051641..
Chen X.; Wu W.; Wang Y.; Zhang B.; Zhou H.; Xiang J.; Li X.; Yu H.; Bai X.; Xie W.; et al. Development of prognostic indicator based on NAD + metabolism related genes in glioma. Front Surg 2023, 10, 1071259. Doi: 10.3389/fsurg.2023.1071259
Pavlova, N.N.; Thompson, C.B. The Emerging Hallmarks of Cancer Metabolism. Cell metabolism 2016, 23, 27–47, doi:10.1016/j.cmet.2015.12.006.
Travelli, C.; Consonni, F.M.; Sangaletti, S.; Storto, M.; Morlacchi, S.; Grolla, A.A.; Galli, U.; Tron, G.C.; Portararo, P.; Rimassa, L.; et al. Nicotinamide Phosphoribosyltransferase Acts as a Metabolic Gate for Mobilization of Myeloid-Derived Suppressor Cells. Cancer research 2019, 79, 1938–1951, doi:10.1158/0008-5472.Can-18-1544.
Bi, J.; Chowdhry, S.; Wu, S.; Zhang, W.; Masui, K.; Mischel, P.S. Altered cellular metabolism in gliomas - an emerging landscape of actionable co-dependency targets. Nature reviews. Cancer 2020, 20, 57–70, doi:10.1038/s41568-019-0226-5.
Yan, H.; Parsons, D.W.; Jin, G.; McLendon, R.; Rasheed, B.A.; Yuan, W.; Kos, I.; Batinic-Haberle, I.; Jones, S.; Riggins, G.J.; et al. IDH1 and IDH2 mutations in gliomas. The New England journal of medicine 2009, 360, 765–773, doi:10.1056/NEJMoa0808710.
Hanahan, D.; Weinberg, R.A. Hallmarks of cancer: the next generation. Cell 2011, 144, 646–674, doi:10.1016/j.cell.2011.02.013.
Kulkarni, S.S.; Karlsson, H.K.; Szekeres, F.; Chibalin, A.V.; Krook, A.; Zierath, J.R. Suppression of 5'-nucleotidase enzymes promotes AMP-activated protein kinase (AMPK) phosphorylation and metabolism in human and mouse skeletal muscle. The Journal of biological chemistry 2011, 286, 34567–34574, doi:10.1074/jbc.M111.268292.
Patzak, M.S.; Kari, V.; Patil, S.; Hamdan, F.H.; Goetze, R.G.; Brunner, M.; Gaedcke, J.; Kitz, J.; Jodrell, D.I.; Richards, F.M.; et al. Cytosolic 5'-nucleotidase 1A is overexpressed in pancreatic cancer and mediates gemcitabine resistance by reducing intracellular gemcitabine metabolites. EbioMedicine 2019, 40, 394–405, doi:10.1016/j.ebiom.2019.01.037..
Wang, Y.; Zeng, J.; Wu, W.; Xie, S.; Yu, H.; Li, G.; Zhu, T.; Li, F.; Lu, J.; Wang, G.Y.; et al. Nicotinamide N-methyltransferase enhances chemoresistance in breast cancer through SIRT1 protein stabilization. Breast cancer research: BCR 2019, 21, 64, doi:10.1186/s13058-019-1150-z.
Sun, W.; Zou, Y.; Cai, Z.; Huang, J.; Hong, X.; Liang, Q.; Jin, W. Overexpression of NNMT in Glioma Aggravates Tumor Cell Progression: An Emerging Therapeutic Target. Cancers 2022, 14, doi:10.3390/cancers14143538..
Hu, K.; Li, J.; Wu, G.; Zhou, L.; Wang, X.; Yan, Y.; Xu, Z. The novel roles of virus infection-associated gene CDKN1A in chemoresistance and immune infiltration of glioblastoma. Aging 2021, 13, 6662–6680, doi:10.18632/aging.202519.
Tan, C.; Liu, L.; Liu, X.; Qi, L.; Wang, W.; Zhao, G.; Wang, L.; Dai, Y. Activation of PTGS2/NF-κB signaling pathway enhances radiation resistance of glioma. Cancer medicine 2019, 8, 1175–1185, doi:10.1002/cam4.1971.
Hong, J.S.; Waud, W.R.; Levasseur, D.N.; Townes, T.M.; Wen, H.; McPherson, S.A.; Moore, B.A.; Bebok, Z.; Allan, P.W.; Secrist, J.A., 3rd ; et al. Excellent in vivo bystander activity of fludarabine phosphate against human glioma xenografts that express the escherichia coli purine nucleoside phosphorylase gene. Cancer research 2004, 64, 6610–6615, doi:10.1158/0008-5472.Can-04-0012.
Wang, Z.; Grosskurth, S.E.; Cheung, T.; Petteruti, P.; Zhang, J.; Wang, X.; Wang, W.; Gharahdaghi, F.; Wu, J.; Su, N.; et al. Pharmacological Inhibition of PARP6 Triggers Multipolar Spindle Formation and Elicits Therapeutic Effects in Breast Cancer. Cancer research 2018, 78, 6691–6702, doi:10.1158/0008-5472.Can-18-1362.
Verheugd, P.; Forst, A.H.; Milke, L.; Herzog, N.; Feijs, K.L.; Kremmer, E.; Kleine, H.; Lüscher, B. Regulation of NF-κB signalling by the mono-ADP-ribosyltransferase ARTD10. Nature communications 2013, 4, 1683, doi:10.1038/ncomms2672.
Tauber, A.L.; Levonis, S.M.; Schweiker, S.S. Recent developments in PARP14 research. Future medicinal chemistry 2020, 12, 1657–1667, doi:10.4155/fmc-2020-0166.
Bachmann, S.B.; Frommel, S.C.; Camicia, R.; Winkler, H.C.; Santoro, R.; Hassa, P.O. DTX3L and ARTD9 inhibit IRF1 expression and mediate in cooperation with ARTD8 survival and proliferation of metastatic prostate cancer cells. Molecular cancer 2014, 13, 125, doi:10.1186/1476-4598-13-125.
Tang, X.; Zhang, H.; Long, Y.; Hua, H.; Jiang, Y.; Jing, J. PARP9 is overexpressed in human breast cancer and promotes cancer cell migration. Oncology letters 2018, 16, 4073–4077, doi:10.3892/ol.2018.9124.
Le Rhun E.; Preusser M.; Roth P.; Reardon DA.; van den Bent M.; Wen P.; Reifenberger G.; Weller M. Molecular targeted therapy of glioblastoma. Cancer Treat Rev 2019, 80,101896, doi: 10.1016/j.ctrv.2019.101896.
Touat M.; Li YY.; Boynton AN.; Spurr LF.; Iorgulescu JB.; Bohrson CL.; Cortes-Ciriano I.; Birzu C.; Geduldig JE.; Pelton K.; et al. Mechanisms and therapeutic implications of hypermutation in gliomas. Nature 2020, 580(7804),517–523, doi: 10.1038/s41586-020-2209-9.

No competing interests reported.

Download PDF

Editorial decision: Revision requested
01 Sep, 2024
Reviews received at journal
23 Aug, 2024
Reviews received at journal
21 Aug, 2024
Reviewers agreed at journal
20 Aug, 2024
Reviewers agreed at journal
20 Aug, 2024
Reviewers agreed at journal
11 Aug, 2024
Reviews received at journal
04 Aug, 2024
Reviewers agreed at journal
04 Aug, 2024
Reviewers invited by journal
21 Jul, 2024
Editor assigned by journal
21 Jul, 2024
Editor invited by journal
27 Jun, 2024
Submission checks completed at journal
27 Jun, 2024
First submitted to journal
21 Jun, 2024

You are reading this latest preprint version

Targeting inhibition of prognosis-related nicotinamide metabolism genes, including poly (ADP-ribose) polymerase 9 (PARP-9) attenuates glioma progression

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Materials and methods

2.1 Study design

2.2 Statistical analysis

Results

3.1 Technology Roadmap

3.2 Data gathering

3.3 Nicotinamide metabolism score (NAMS) construction and clinical prognosis analysis

3.4 WGCNA

3.5 Mutation analysis and copy number variation of hub genes

3.6 Expression difference analysis of hub genes

3.7 Construction of related subtypes of GBMLGG

3.8 Construction of key genes prognostic model (NMRS)

3.9 Difference analysis of NMRS

3.10 Functional enrichment analysis (GO) and pathway enrichment (KEGG) analysis of key genes

3.11 GSEA between high- and low-risk groups in NMRS

3.12 Construction of PPI network, mRNA-miRNA, mRNA-binding protein (RBP), and mRNA-TF interaction network

3.13 The impact of NMRS on immune assessment

3.14 Difference analysis of ssGSEA immune signature

3.15 NMRS mutation analysis of high- and low-risk groups

3.16 Nomogram establishment based on the multivariable Cox regression model

Discussion

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1