Periostin-mediated activation of NF-κB signaling promotes tumor progression and chemoresistance in glioblastoma

Abstract

Glioblastoma (GBM) is the most aggressive form of diffuse glioma, characterized by high lethality. Temozolomide (TMZ)-based chemotherapy is a standard treatment for GBM, but development of chemoresistance poses a significant therapeutic challenge. Despite advances in understanding GBM biology, the mechanisms driving TMZ resistance remain unclear. Identifying vital molecular players involved in this resistance is crucial for developing new therapies. Our results indicated that periostin (POSTN) was significantly upregulated in GBM cell lines and patient samples, correlating with poorer clinical outcomes. POSTN overexpression enhanced GBM cell proliferation, migration, invasion, and chemoresistance, while lentiviral suppression of POSTN significantly reduced these behaviors. In vivo, bioluminescence imaging further confirmed the enhanced tumor growth associated with POSTN overexpression. Bioinformatics analysis was performed to explore the underlying molecular mechanism. The results revealed a strong correlation between POSTN and epithelial-mesenchymal transition (EMT) process and the tumor necrosis factor α (TNFα)-NF-κB signaling pathway. Moreover, exogenous POSTN silencing reduced IκB-kinase α (IKKα) phosphorylation, thereby decreasing NF-κB expression by limiting IκBα degradation. Collectively, our study demonstrated that POSTN-induced activation of NF-κB signaling and EMT processes promoted the malignancy and chemoresistance of GBM, suggesting that POSTN may serve as a reliable prognostic biomarker and potential therapeutic target for GBM.

Keywords: EMT-related phenotypes; Glioblastoma; NF-κB signaling; POSTN; TMZ resistance.

Fig. 1

Underlying chemoresistance genes could identify different prognostic subgroups for GBM. A and B, Differentially expressed genes (DEGs) in TMZ-resistant versus TMZ-sensitive cell lines, identified via limma analysis. Expression profiles were obtained from the GSE199689 (A) and GSE100736 (B) datasets. DEGs were defined as Log₂FC > 2.0 and adjusted P < 0.05. C, Venn diagram depicting intersection of upregulated genes from both datasets, revealing 220 shared candidate genes. D, E and G, Clustering analysis using The Cancer Genome Atlas (TCGA) database (E) and the Chinese Glioma Genome Atlas (CGGA) database (G), categorizing the candidate genes into two distinct clusters. Cluster optimization was achieved using the ConsensusClusterPlus package through distance analysis, resulting in two distinct clusters (D). F and H, Kaplan-Meier analysis and the distribution of resistance scores for the two clusters extracted from the TCGA (F) and CGGA (H) databases (TCGA: P = 0.00061, CGGA: P = 0.012, log-rank test; ****P < 0.001, independent sample t-test). I, Bubble plots illustrating the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of the 220 identified candidate genes, showing a positive correlation with extracellular matrix structural constituents^,.

Fig. 2

Periostin (POSTN) was markedly upregulated in GBM samples and cell lines, and increased expression of POSTN was strongly linked to unfavorable outcomes in GBM patients. A and B, Volcano plots depicting DEGs identified in patients with different prognoses using the datasets from TCGA (A) and CGGA (B). C, Venn diagram showing intersection analysis of upregulated genes from TCGA and CGGA datasets to identify common candidate genes. D and G, POSTN mRNA expression in gliomas categorized by WHO grade using the CGGA database (D) and TCGA database (G) (***P < 0.001, **P < 0.01, *P < 0.05, one-way ANOVA followed by Tukey’s Honest Significant Difference (HSD) test). E and H, Analysis of POSTN mRNA expression levels across different glioma subtypes (classical, mesenchymal, proneural) using data from the CGGA database (E) and TCGA database (H) (***P < 0.001, one-way ANOVA followed by Tukey’s HSD post-test). F and I, Kaplan-Meier survival analysis of POSTN expression in IDH-wt GBM samples across all subtypes from the CGGA (F) and TCGA datasets (I) (CGGA: P = 0.012, TCGA: P = 0.0002, log-rank test). J, IHC images showing POSTN distribution in gliomas of different WHO grades, including oligodendrogliomas (grade II), astrocytomas (grade III), and glioblastomas (grade IV), respectively. K, Western blot assays of POSTN expression in four glioma tissues compared to matched peritumoral infiltration tissues, using GAPDH as the reference control. qRT-PCR assays of POSTN mRNA levels in four gliomas tumor tissues versus matched peritumoral infiltration tissues (****P < 0.0001, **P < 0.01, Student’s t-test). Upper: Western blot analysis. Lower: qRT-PCR analysis. L, Western blot assays of POSTN expression in U87, LN229, and U251 GBM cells relative to normal human astrocytes (NHA), using GAPDH as the reference control. qRT-PCR assays for detecting POSTN mRNA levels in U87, LN229, and U251 GBM cell lines versus NHA (****P < 0.0001, Student’s t-test). Upper: Western blot analysis. Lower: qRT-PCR analysis.

Fig. 3

POSTN knockdown suppressed the malignant characteristics of glioblastoma cell lines. A, Immunofluorescence images illustrating U87 cell transduction efficiency with lentiviral shPOSTN#1, shPOSTN#2, and control (shNT). B, Western blot analysis of POSTN expression in U87 cells transfected with shPOSTN lentivirus or control, using GAPDH as the reference control. C, qRT-PCR assays of POSTN mRNA levels in U87 cells transfected with shPOSTN lentivirus or control (****P < 0.0001, one-way ANOVA followed by Dunnett’s post-test). D, EDU staining assay for measuring the proliferation of U87 cells transfected with shPOSTN lentivirus or control. Microscopic counts of EDU-positive cells demonstrated proliferation capacity (***P < 0.001, one-way ANOVA followed by Dunnett’s post-test). E, In vitro proliferation assay evaluating the proliferation of U87 cells transfected with shPOSTN lentivirus or control (***P < 0.001, one-way ANOVA followed by Dunnett’s post-test). F, Wound healing assays of U87 cells transfected with shPOSTN lentivirus or control at baseline (top) and day 3 (bottom). The cell migration distance was significantly reduced in the shPOSTN treatment group compared to the control group (*P < 0.05, one-way ANOVA followed by Dunnett’s post-test). G, Cell invasion assays of U87 cells transfected with shPOSTN lentivirus or control (*P < 0.05, one-way ANOVA followed by Dunnett’s post-test). H, Colony formation assays of U87 cells transfected with shPOSTN lentivirus or control (**P < 0.01, one-way ANOVA followed by Dunnett’s post-test).

Fig. 4

POSTN overexpression may contribute to enhanced malignancy and resistance to temozolomide, thereby corroborating its involvement in cancer processes. A, Immunofluorescence illustrating the efficiency of POSTN overexpression lentiviral transduction in the control (EV group) and POSTN group. B, Western blot analysis of POSTN protein expression in LN229 cells with or without POSTN overexpression lentivirus transduction, using GAPDH as the internal control. C, qRT-PCR analysis of POSTN mRNA levels in LN229 cells following POSTN lentiviral transduction (****P < 0.0001, Student’s t-test). D, EDU staining of LN229 cells with or without POSTN overexpression lentiviral transduction. Microscopic counts of EDU-positive cell clones represent proliferation ability (**P < 0.01, Student’s t-test). E, In vitro proliferation assay of LN229 cells with lentiviral POSTN or control transduction (**P < 0.01, one-way ANOVA followed by Dunnett’s post-test). F, Wound healing assays on LN229 cells transduced with POSTN overexpression lentivirus or control at day 0 (top) and day 3 (bottom). The migration distance was significantly increased in the POSTN-overexpressing group compared to the control group (*P < 0.05, Student’s t-test). G, Invasion assays of LN229 cells transfected with POSTN lentivirus or control (**P < 0.01, Student’s t-test). H, Colony formation assays of LN229 cells transfected with or without POSTN overexpression lentivirus (**P < 0.01, Student’s t-test). I, Cytotoxicity assays of LN229 cells transfected with POSTN lentivirus or control (**P < 0.01, one-way ANOVA followed by Dunnett’s post-test). J, In vivo bioluminescence imaging of POSTN-overexpressing LN229 cells at week 1 (top) and week 4 (bottom). The bar graph quantified the mean fluorescence intensity ratio at 4 weeks relative to 1 week (**P < 0.01, one-way ANOVA followed by Dunnett’s post-test).

Fig. 5

POSTN showed positive associations with EMT-like phenotype invasion and migration, as well as TNFα/ NF-κB signaling. A-D, Bubble plots were performed to elucidate pathways significantly enriched in the TCGA (A) and CGGA (C) databases, showing a positive association of POSTN with EMT and TNFα/NF-κB signaling. GSEA analysis indicated a substantial link between EMT and POSTN expression in the TCGA (B) and CGGA (D) databases using normalized enrichment scores (NES) and p-values. E, Western blot analysis was undertaken to determine EMT downstream target protein levels in control and POSTN knockdown samples, using GAPDH as the reference control. F, qRT-PCR was performed to analyze mRNA levels of EMT downstream targets in U87 cells transduced with lentiviral POSTN and controls. Relative mRNA levels were normalized to GAPDH (****P < 0.0001, one-way ANOVA followed by Dunnett’s post-test).

Fig. 6

POSTN regulated NF-κB signaling activation by modulating IkB-kinase α (IKKα) phosphorylation. A, Western blot analysis of NF-κB-regulated downstream proteins in U87 cells transfected with control (shNT) or shPOSTN#1 lentivirus and subsequently administered TNFα or maintained untreated, using GAPDH as the internal control. B, EDU staining assay on U87 cells transfected with shPOSTN#1 or control lentivirus and subsequently administered TNFα or maintained untreated. Proliferation levels were indicated by microscopic counts of EDU-positive cells (**P < 0.01, ***P < 0.001, Student’s t-test). C, Cytotoxicity assays on U87 cells transfected with shNT or shPOSTN#2 lentivirus and subsequently administered TNFα or maintained untreated (**P < 0.01, ***P < 0.001, one-way ANOVA followed by Dunnett’s post-test). D, Wound healing assays on U87 cells transfected with shNT or shPOSTN#1 lentivirus and subsequently administered TNFα or maintained untreated at day 0 (top) and day 3 (bottom) (*P < 0.05, **P < 0.01, one-way ANOVA followed by Dunnett’s post-test). E, Cell invasion assay on U87 cells transfected with shPOSTN#1 or control lentivirus and subsequently administered TNFα or maintained untreated (*P < 0.05, **P < 0.01, one-way ANOVA followed by Dunnett’s post-test). F, Colony formation assays on U87 cells transfected with shPOSTN#1 or control lentivirus and subsequently administered TNFα or maintained untreated (*P < 0.05, **P < 0.01, one-way ANOVA followed by Dunnett’s post-test).