Table 1.
Spearman’s rank correlation coefficient and associated p-values between the expression of hypoxia markers with respect to each other.
Table 2.
Spearman’s rank correlation coefficient and associated p-values between the expression of Notch receptors, ligands, target genes and the hypoxia markers.
Fig 1.
Heat maps showing gene cluster analysis in GBM tumor tertiles.
(A) Clustering of hypoxia genes in GBMs sorted in decreasing order of HIF-1α expression. PGK1, VEGF and OPN displayed greater upregulation in the high HIF-1α GBM tertile as compared to the low tertile. Clustering of Notch signaling genes in GBMs sorted in decreasing order of (B) HIF-1α expression (C) PGK1 expression and (D) OPN expression showed maximum number of Notch signaling genes (9/15) to be upregulated in the high HIF-1α GBM tertile, followed by high PGK1 and OPN tertiles which exhibited increased expression of 7/15 and 3/15 Notch genes, respectively. Gene expression found significantly different across the high and low tertiles (p ≤ 0.05) has been indicated by an asterisk (*).
Fig 2.
2D PCA plots showing clustering of high and low HIF-1α expressing GBM tertiles on the basis of differential expression of Notch pathway genes.
Each data point in a plot denotes a GBM and depicts the score created by weighted summation of Notch genes’ expression in that tumor. The colour gradient from darkest red to darkest green denotes the highest HIF-1α expression to the lowest HIF-1α expression in the data points representing tumors. The demarcation between high HIF-1α GBM tertile and low HIF-1α GBM tertile in each plot has been displayed by using either lines or curves (blue coloured) which have been drawn manually to reveal the closely spaced cluster. Cluster formation was assessed by counting the number of the darkest red and the darkest green data points on both sides of the line; or those enclosed within the curve. GBM scores were created by the combination of either (i) all 15 Notch genes or (ii) 9 Notch genes (Notch1, Dll1, Dll4, Hes1, Hes2, Hes5, Hes6, Hey1, Hey2) and (iii) 7 Notch genes (Notch1, Dll1, Hes1, Hes2, Hes6, Hey1, Hey2) selected on the basis of HIF-1α heat map. HIF-1α-ordering of the tumors produced well-separated clusters of high and low tertiles based on 15 Notch genes’ as well as 7 Notch genes’ expression.
Fig 3.
2D PCA plots showing clustering of high and low PGK1 expressing GBM tertiles on the basis of differential expression of Notch pathway genes.
Each data point in a plot denotes a GBM and depicts the score created by weighted summation of Notch genes’ expression in that tumor. The colour gradient from darkest red to darkest green denotes the highest PGK1 expression to the lowest PGK1 expression in the data points representing tumors. The demarcation between high PGK1 GBM tertile and low PGK1 GBM tertile in each plot has been displayed by using either lines or curves (blue coloured) which have been drawn manually to reveal the closely spaced cluster. Cluster formation was assessed by counting the number of the darkest red and the darkest green data points on both sides of the line; or those enclosed within the curve. GBM scores were created by the combination of either (i) all 15 Notch genes or (ii) 7 Notch genes (Notch1, Dll1, Dll4, Hes1, Hes6, Hey1, Hey2) and (iii) 6 Notch genes (Notch1, Dll1, Hes1, Hes6, Hey1, Hey2) selected on the basis of PGK1 heat map. Similar to the case of HIF-1α, PGK1-ordering of the tumors also generated well-separated clusters of high and low tertiles based on 15 Notch genes’ expression.
Fig 4.
3D PCA plots showing clustering of GBM tumors belonging to HIF-1α tertiles and PGK1 tertiles on the basis of differential expression of Notch pathway genes.
Colouring scheme in 3D plots: GBMs have been coloured as red data points for high tertile (12 tumors), yellow data points for intermediate tertile (11 tumors) and green data points for low tertile (12 tumors), respectively. Similar to 2D plots, the combinations of Notch genes used to create the principal components in each 3D plot were as follows: (i & ii) all 15 Notch genes; (iii) 9 genes (Notch1, Dll1, Dll4, Hes1, Hes2, Hes5, Hes6, Hey1, Hey2); (iv) 7 genes (Notch1, Dll1, Dll4, Hes1, Hes6, Hey1, Hey2); (v) 7 genes (Notch1, Dll1, Hes1, Hes2, Hes6, Hey1, Hey2); (vi) 6 genes (Notch1, Dll1, Hes1, Hes6, Hey1, Hey2). Both HIF-1α- and PGK1-ordering of the tumors produced well-separated clusters of high and low tertiles based on 15 or selected Notch genes’ expression.
Table 3.
Summary of sensitivity and specificity of HIF-1α, PGK1 and OPN expression in diagnosis of Notch genes’ overexpression in 35 GBM tumors.
Table 4.
Summary of results of linear regression analysis with each single hypoxia marker used as a regressor for its association with expression of individual Notch genes.
Table 5.
Summary of results of linear regression analysis using a combination of five hypoxia markers (HIF-1α, PGK1, VEGF, CA9 and OPN) as independent variables for their combined association with the expression level of individual Notch genes.
Table 6.
Summary of each computational/statistical analysis listing the significant associations of hypoxia markers with the expression of Notch signaling genes.
Fig 5.
Survival analysis of GBM patients in relation to expression of the components of hypoxia-Notch gene signature.
Kaplan-Meier plots showing comparison of survival of GBM patients (n = 21) segregated as high and low expressors of HIF-1α, OPN and Hes1, individually as well as in combination. The p-value indicates the significance of the difference in survival between the two groups of GBM samples. The fold expression thresholds that were used to draw the plots were: HIF-1α = 1.5; OPN = 5.0; Hes1 = 1.5 and HIF-1α/OPN/Hes1 = 8.0.
Table 7.
mRNA fold expression ratios of hypoxia markers and Notch signaling genes in monolayer cultures of GBM cell lines (U87MG, A172 and U373MG) exposed to 0.2% O2 for 24, 48 and 72 hours.
Table 8.
mRNA fold expression ratios of hypoxia markers and Notch signaling genes in U87MG monolayer and gliomasphere cultures exposed to 20%, 2% and 0.2% O2 at day 10.