Procollagen-lysine 2-oxoglutarate 5-dioxygenase 2 promotes hypoxia-induced glioma migration and invasion

Abstract

Poor prognosis of glioblastoma multiforme is strongly associated with the ability of tumor cells to invade the brain parenchyma, which is believed to be the major factor responsible for glioblastoma recurrence. Therefore, identifying the molecular mechanisms driving invasion may lead to the development of improved therapies for glioblastoma patients. Here, we investigated the role of procollagen-lysine 2-oxoglutarate 5-dioxygenase 2 (PLOD2), an enzyme catalyzing collagen cross-linking, in the biology of glioblastoma invasion. PLOD2 mRNA was significantly overexpressed in glioblastoma compared to low-grade tumors based on the Oncomine datasets and REMBRANDT database for human gliomas. Kaplan-Meier estimates based on the TCGA dataset demonstrated that high PLOD2 expression was associated with poor prognosis. In vitro, hypoxia upregulated PLOD2 protein in U87 and U251 human glioma cell lines. siRNA knockdown of endogenous HIF-1α or treatment of cells with the HIF-1α inhibitor PX-478 largely abolished the hypoxia-mediated PLOD2 upregulation. Knockdown of PLOD2 in glioma cell lines led to decreases in migration and invasion under normoxia and hypoxia. In addition, levels of phosphorylated FAK (Tyr 397), an important kinase mediating cell adhesion, were reduced in U87-shPLOD2 and U251-shPLOD2 cells, particularly under hypoxic conditions. Finally, orthotopic U251-shPLOD2 xenografts were circumscribed rather than locally invasive. In conclusion, the results indicated that PLOD2 was a gene of clinical relevance with implications in glioblastoma invasion and treatment strategies.

Keywords: FAK; glioma; hypoxia; invasion; procollagen-lysine 2-oxoglutarate 5-dioxygenase 2.

Figure 1. PLOD2 is overexpressed in glioblastoma and correlates with poor prognosis

(A) Forest plot of PLOD2 expression levels in glioblastoma (n = 861) versus non-neoplastic brain tissue samples from the publicly available Oncomine datasets. The x-axis is the standardized mean difference between glioblastoma and normal PLOD2 expression based on a log2 scale. (B) Expression of PLOD2 mRNA in human gliomas of all grades (n = 178) from cases in the publicly available REMBRANDT database. (C) Representative images of immunohistochemical staining for PLOD2 in sections (4 μm) from primary human tumor samples (Scale bars, 100 μm). (D) Comparison of PLOD2 mRNA levels among human gliomas based on tumor grade. Mean ± SD, **P < 0.01; ***P < 0.001; ns, not significant, versus normal samples. (E) ROC curve showing sensitivity of PLOD2 as a marker to distinguish non-glioblastoma and glioblastoma patients. (F and G) Kaplan-Meier analyses for overall survival (OS, n = 522) and progression-free survival (PFS, n = 386) for glioblastoma patients in TCGA cohorts based on PLOD2 expression levels (OS: P < 0.05; PFS: P < 0.0005, by the log-rank test).

Figure 2. Hypoxia induces PLOD2 expression through HIF-1α in U87 and U251 glioma cells

(A) Western blot for detection of PLOD2 levels in lysates prepared from U87 and U251 cells cultured under hypoxic conditions for 48 h. (B) Western blot for PLOD2 levels in U87 and U251 cultured 0, 6, 12, 24, 48 h under hypoxic conditions. (C) Western blot for PLOD2 and HIF-1α levels in U87 and U251 cells transfected with negative control siRNA (siNC) and HIF-1α siRNA (siHIF-1α#1-2; 100 nmol/L) and cultured under hypoxic conditions for 48 h. (D) Western blot for PLOD2 and HIF-2α levels in U87 and U251 cells transfected with negative control siRNA (siNC) and HIF-2α siRNA (siHIF-2α#1-2; 100 nmol/L) and cultured under hypoxic conditions for 48 h. (E) Western blot for PLOD2 and HIF-1α levels in U87 and U251 cells pre-treated with PX-478 (10 μM) and cultured under normoxic and hypoxic conditions for 48 h. β-actin was used as a loading control. (F) Correlation between PLOD2 and HIF-1α mRNA levels as determined from the publicly available TCGA data (n = 528).

Figure 3. Decreased PLOD2 expression in U87 and U251 cells inhibits migration and invasion

(A) Representative images from Transwell migration assays for U87-shNC and U87-shPLOD2 or U251-shNC and U251-shPLOD2 cells under normoxic or hypoxic conditions, as indicated, after 12 h (Scale bars, 50 μm). (B) Bar graphs representing quantification of the number of migrated cells per field for cell types indicated. (C) Representative images of 3D spheroid invasion assays for U87-shNC and U87-shPLOD2 or U251-shNC and U251-shPLOD2 under normoxic or hypoxic conditions for 24 h (Scale bars, 50 μm). (D) Bar graphs representing the longest invasion distance (fold change) for the cell type indicated. Data are presented as the mean ± SD (3 individual experiments). *P < 0.05; **P < 0.01; ***P < 0.001; versus shNC groups under the same condition; Student's t test.

Figure 4. PLOD2 mediates actin cytoskeleton remodeling through FAK

(A) Representative confocal images of U87-shNC and U87-shPLOD2 cells under normoxic and hypoxic conditions (Scale bars, 10 μm). Blue: DAPI (nuclei); red: phalloidin (polymerized actin/stress fiber); green: paxillin (focal adhesion). (B) Western blot for the detection of FAK-p³⁹⁷ levels in U87-shNC control and U87-shPLOD2 cells under normoxic and hypoxic conditions. (C) Representative confocal images of U251-shNC and U251-shPLOD2 cells under normoxic and hypoxic conditions (Scale bars, 10 μm). (D) Western blot for the detection of FAK-p³⁹⁷ levels in U251-shNC control and U251-shPLOD2 cells under normoxic and hypoxic conditions.

Figure 5. Treatment with FAK inhibitor attenuates U87 and U251 cell migration and invasion

(A) Western blot for the detection of FAK-p³⁹⁷ levels in total protein lysates (20 μg) from U87 and U251 cells treated with FAK inhibitor TAE226 (5 μM) for 12 h under normoxic and hypoxic conditions. (B) Representative images and quantification (cell number/field) of transwell migration assays for parental U87 and U251 cells treated with TAE226 (5 μM) under normoxic and hypoxic conditions (Scale bars, 50 μm). (C) Representative images and quantification of 3D invasion assays (longest invasive distance) for U87 and U251 cells treated with TAE226 (5 μM) under normoxic and hypoxic conditions (Scale bars, 50 μm). Results are presented as the mean ± SD (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001, versus U87-shNC or U251-shNC controls under the same conditions; Student's t test.

Figure 6. Knockdown of PLOD2 inhibits glioma cell invasion in vivo

(A) Nude mice bearing subcutaneous xenografts of injected U251-shPLOD2 or U251-shNC cells (U251-shPLOD2, n = 5; U251-shNC, n = 5). Subcutaneous xenograft tumor volume (mm³) was measured on day 43 with a caliper. (B) Concentration of hydroxyproline (μg/mL) in U251-shNC or U251-shPLOD2 subcutaneous xenografts as determined by ELISA. (C) Stiffness of U251-shNC or U251-shPLOD2 subcutaneous xenografts measured with the compression test. Measurements were made on 3 randomly chosen sites in each tumor. ns, not significant, Student's t test. (D) PLOD2 immunohistochemical staining of tissue sections from U251-shNC or U251-shPLOD2 orthotopic xenografts (Scale bars, 100 μm). (E) Hematoxylin and eosin staining of sections at tumor margins in intracranial U251-shNC or U251-shPLOD2 xenografts (Scale bars, 100 μm). (F) Picrosirius red staining of tumor sections from U251-shNC or U251-shPLOD2 intracranial xenografts captured under bright and polarized light (Scale bars, 100 μm).