Phospholipase C Beta 1: a Candidate Signature Gene for Proneural Subtype High-Grade Glioma

Abstract

Phospholipase C beta 1 (PLCβ1) expresses in gliomas and cultured glial cells, but its expression is barely detectable in normal glial cells. We analyzed data from Gene Expression Omnibus (GEO-GDSxxx), The Cancer Genome Atlas (TCGA), and the Repository for Molecular Brain Neoplasia Data (REMBRANDT) to explore the potential role of PLCβ1 as a biomarker in high-grade glioma (HGG). PLCβ1 expression is significantly higher in grade III gliomas than that in grade IV gliomas from GDS1815 (n = 24 vs. 76), GDS1962 (n = 19 vs. 81), and GDS1975 (n = 26 vs. 59). In GDS1815, PLCβ1 expression correlates with several known proneural (PN) signature genes; its expression from PN subtype (n = 15) is significantly higher than that from mesenchymal (Mes) subtype (n = 33) HGG. In GDS1962, PLCβ1 expression is the highest in nontumor brain tissue (n = 23) and is significantly higher than its expression in grade II gliomas [astrocytomas (n = 7) and oligodendrogliomas (n = 37)]. A Kaplan-Meier survival curve from a REMBRANDT cohort demonstrates that glioma patients with intermediate PLCβ1 expression (n = 103) survived significantly longer than PLCβ1 downregulated (2X) groups (n = 226). From TCGA data, PLCβ1 RNA-Seq signal inversely correlates with the pathological grades, and PLCβ1 expression in PN (n = 8) is of significantly higher levels than that in Mes (n = 8) subtypes of glioblastoma. The top 50 % of PLCβ1 expression subgroup (n = 294) of gliomas (grades II to IV merged) survived significantly longer than the low 50 percentile of the PLCβ1 expression subgroup (n = 293). p values are less than 0.05 for all these analyses. We conclude that PLCβ1 is a candidate signature gene for PN subtype HGG, and its expression inversely correlates with glioma pathological grade and is a potential prognostic factor.

Keywords: Biomarker; Glioblastoma; Glioma; PLCβ1; Proneural; REMBRANDT; Signature gene; TCGA.

Fig. 1

a Original microarray data showing PLCβ1 signal strength in GDS1815 dataset from the NIH website with minor modification. b Analysis of data from grade IV samples, primary and recurrent tumors combined. Data shows a significant lower level of PLCβ1 in Mes (n = 33) subtypes than in PN (n = 15; *p = 1.4E-4). c Analysis of combined samples of grades III and IV shows that the average PLCβ1 signal level in PN subtypes (2642 ± 207, n = 37) is significantly higher than its level in Mes (852 ± 120, n = 35; p = 2.78E-10) and proliferative (Prolif) subtypes (1242 ± 170, n = 28; p = 5.14E-6, *p < 0.05). The difference between the PLCβ1 signal level in the Mes and proliferative subtype HGG is not statistically significant (p = 0.126). NS nonsignificant. d In primary tumor cases, pooled data from grade III and grade IV gliomas shows that PLCβ1 microarray signal strengths are significantly higher in patients who survived over 2 years (mean = 240.3 weeks, range 106–477 weeks, n = 35) than those survived less than 2 years (mean = 55.4 weeks, range 3–102 weeks, n = 42; *p = 0.0035; 23 cases do not have survival data). e–g ERBB4, Olig2, and PLCβ1 expression, respectively. All of these signal strengths are significantly lower in grade IV glioma samples (n = 76) than those in grade III gliomas (n = 24; *p < 0.05)

Fig. 2

HGG in TCGA only has a few cases that are labeled as Mes and PN subtypes. a Normalized PLCβ1 expression (RNA-Seq data) is significantly higher in PN (n = 8) than in Mes subtypes (n = 8; *p = 4E-4). b In contrast, there is no statistical difference in GAPDH expression between PN and Mes subtypes (p = 0.41). c In the TCGA database, normalized PLCβ1 expression (RNA-Seq data) shows significant differences between grades II (10.87 ± 0.05; n = 212), III (10.23 ± 0.08; n = 231), and IV (9.61 ± 0.10; n = 152; *p < 0.0001). d GAPDH expression shows no significant difference between grades III (15.56 ± 0.03) and IV (15.54 ± 0.05; p = 0.68). NS nonsignificant. However, grade II gliomas (15.41 ± 0.03) contain less GAPDH than grade III (p = 0.017) and grade IV (p = 0.0018; *p < 0.05)

Fig. 3

ERBB4, Olig2, and PLCβ1 expressions presented as microarray data in low- and high-grade astrocytomas (n = 14). Original GDS2853 data from the NIH-maintained database are on top of each histogram. a, b ERBB4 and Olig2 microarray data used to do the same analysis; however, the ERBB4 signal was significantly reduced in high-grade samples (*p = 0.0037), but the Olig2 signal (p = 0.19) was not. c PLCβ1 signal strength among high-grade astrocytoma samples (n = 6) significantly decreased in comparison to its signal from low-grade astrocytoma (n = 8; *p = 0.0004)

Fig. 4

Analysis of PLCβ1 expression profile from GDS1962. a Original microarray data copied from the NIH website with minor modification. NA nonapplicable. b Collective PLCβ1 data analysis. Average PLCβ1 signal from nontumor controls (n = 23) is the highest, significantly higher than low-grade tumors including grade II astrocytomas (n = 7, p = 0.00097) and oligodendrogliomas (n = 37, p = 0.0075). The average PLCβ1 signal from grade II (n = 37) is significantly higher than that from grade III (n = 13) oligodendroglioma (p = 7.42E-05). c PLCβ1 signal levels from nontumor controls (n = 23) are significantly higher than those from astrocytomas cases (Astro, n = 107, p=9.63E-09), grades II to IV combined. d PLCβ1 signal levels from grade III (n = 19) astrocytomas are significantly higher than from grade IV astrocytomas (n = 81; *p = 0.044)

Fig. 5

Kaplan-Meier survival curve for samples with differential PLCβ1 and GAPDH gene expression, respectively; raw data extracted from the REMBRANDT cohort. a PLCβ1 downregulated glioma patients (n = 226) had significantly shorter survival times than intermediate level PLCβ1 glioma patients (n = 103) per the log-rank test (p = 3.0E-09). b PLCβ1 downregulated astrocytoma patients (n = 57) survived significantly shorter than the intermediate level PLCβ1 astrocytoma patients (n = 45) per the log-rank test (p = 2.0E-04). Both databases contain no cases stratified as upregulated PLCβ1 expression (≥2X). c Based on GAPDH expression level, only 1 glioma case is classified as upregulated (≥2X), not statistically significant in survival from the group of intermediate level GAPDH (n = 328; p = 0.23). There are no GAPDH downregulated glioma patients (n = 0). d Based on the GAPDH expression level, there are no astrocytoma cases being classified as up- (n = 0) or downregulated (n = 0) exceeding 2-fold. Intermediate level of GAPDH is the only group (n = 102). Note: Among the raw data we downloaded for analysis, 14 glioma cases (include 3 astrocytoma) lacked censored information. Thus, our analysis has 14 fewer glioma cases (include 3 astrocytoma) than the total cases reported in the REMBRANDT website (see Table 4)

Fig. 6

Kaplan-Meier survival curve for samples with differential PLCβ1 gene expression from the TCGA cohort. a Merged grade II–IV gliomas are stratified as high (n = 294) and low 50 % PLCβ1 expression (n = 293). There is statistical difference in survival between the two groups as indicated by the log-rank test (p = 4.8E-12). b GBM cases are stratified into the highest 5 % expression (n = 9) and the rest of 95 % (n = 142); there is statistical difference in survival between the two groups as indicated by the log-rank test (p = 0.039)

Fig. 7

IHC images obtained from the HPA website. Antibody CAB005334 is used for images a–c, HPA034743 antibody is used for d. a PLCβ1 staining from normal cerebral cortex tissue of a 52-year-old female subject (patient ID—3740). b Negative PLCβ1 staining of glioma from a 75-year-old male subject (patient ID—2851). Both c and d are from one 32-year-old female subject (patient ID—122). c Cytoplasmic/membranous stain of PLCβ1 in glioma cells. d Nuclear stain of PLCβ1 in glioma cells. Bar scale equals to 100 μm

Fig. 8

Graphic diagram displays a relationship between glioma PLCβ1 expression and pathological grades. Normal brain (NB) expresses the highest level of PLCβ1. Among the astrocytomas, the higher the pathological grade (II to IV), the lower is the PLCβ1 expression. This inverse relationship also applies to oligodendrogliomas (grades II and III)