Imaging biomarkers of TERT or GABPB1 silencing in TERT-positive glioblastoma

Abstract

Background: TERT promoter mutations are observed in 80% of wild-type IDH glioblastoma (GBM). Moreover, the upstream TERT transcription factor GABPB1 was recently identified as a cancer-specific therapeutic target for tumors harboring a TERT promoter mutation. In that context, noninvasive imaging biomarkers are needed for the detection of TERT modulation.

Methods: Multiple GBM models were investigated as cells and in vivo tumors and the impact of TERT silencing, either directly or by targeting GABPB1, was determined using 1H and hyperpolarized 13C magnetic resonance spectroscopy (MRS). Changes in associated metabolic enzymes were also investigated.

Results: 1H-MRS revealed that lactate and glutathione (GSH) were the most significantly altered metabolites when either TERT or GABPB1 was silenced, and lactate and GSH levels were correlated with cellular TERT expression. Consistent with the drop in lactate, 13C-MRS showed that hyperpolarized [1-13C]lactate production from [1-13C]pyruvate was also reduced when TERT was silenced. Mechanistically, the reduction in GSH was associated with a reduction in pentose phosphate pathway flux, reduced activity of glucose-6-phosphate dehydrogenase, and reduced NADPH. The drop in lactate and hyperpolarized lactate were associated with reductions in glycolytic flux, NADH, and expression/activity of GLUT1, monocarboxylate transporters, and lactate dehydrogenase A.

Conclusions: Our study indicates that MRS-detectable GSH, lactate, and lactate production could serve as metabolic biomarkers of response to emerging TERT-targeted therapies for GBM with activating TERT promoter mutations. Importantly these biomarkers are readily translatable to the clinic, and thus could ultimately improve GBM patient management.

Keywords: (MRS); glioblastoma; hyperpolarized 13C-MRS; imaging biomarkers; magnetic resonance spectroscopy; metabolism.

Fig. 1.

Lactate and GSH are the most significantly reduced metabolites in multiple GBM models following TERT silencing. (A) Representative spectra of NHARas/TERT treated with siCtrl, siTERT #1, and siTERT #2 respectively. (B) PCA comparing siCtrl (n = 6) and siTERT (n = 6, siTERT #1 n = 3, siTERT #2 n = 3). t1 on x axis corresponds to PC1 and t2 on y axis corresponds to PC2. (C) VIP predictive scores derived from OPLS-DA, illustrating metabolites with a score >1. (D) Univariate analysis of lactate and GSH levels comparing siCtrl (n = 6), siTERT #1 (n = 3), and siTERT #2 (n = 3). (E) Representative spectra of U251 cell lines stably expressing shRNAs. (F) PCA comparing U251shCtrl (n = 5) and U251shTERT (U251shTERT-2 n = 5, U251shTERT-3 n = 3). t1 on x axis corresponds to PC1 and t2 on y axis corresponds to PC2. (G) PCA analysis comparing U251shCtrl (n = 5) and U251shB1 (U251shB1-2 n = 5, U251shB1-1 n = 3). t1 on x axis corresponds to PC1and t2 on y axis corresponds to PC2. (H) Univariate analysis of lactate and GSH levels comparing U251shCtrl (n = 5), U251shTERT-2 (n = 5), U251shTERT-3 (n = 3), U251shB1-2 (n = 5), and U251shB1-1 (n = 3). (I) GSH and lactate levels in U251shCtrl, U251shTERT, and U251shB1 cells as a function of TERT expression (GSH; R² = 0.9007, P < .0001, lactate; R² = 0.9127, P < .0001).

Fig. 2.

TERT silencing is associated with a drop in glycolytic and PPP fluxes, as well as redox state. (A) Schema of metabolic pathways illustrating the fate of labeled carbons (black circles) from [2-¹³C]glucose into the glycolytic pathway and TCA cycle or pentose phosphate pathway and TCA cycle. (B) Levels of [5-¹³C]glutamate, [2-¹³C]lactate, [4-¹³C]glutamate, and [3-¹³C]lactate comparing siCtrl and siTERT-treated NHARas/TERT cells. (C) Levels of [5-¹³C]glutamate, [2-¹³C]lactate, [4-¹³C]glutamate, and [3-¹³C]lactate comparing U251shCtrl, U251shTERT, and U251shB1 cells. (D) NADPH levels in NHARas/TERT cells treated with siCtrl and siTERT (n = 5, respectively). (E) NADPH levels in U251shCtrl, U251shTERT, and U251shB1 cells (n = 5, respectively). (F) ROS levels in siCtrl and siTERT-treated NHARas/TERT cells (n = 6, respectively). (G) ROS levels in U251shCtrl, U251shTERT, and U251shB1 cells (n = 3, respectively).

Fig. 3.

TERT silencing is associated with alterations in enzymes of the glycolytic and PPP pathways. (A) Relative GLUT1 expression in NHARas/TERT cells treated with siCtrl and siTERT (n = 3, respectively). (B) Relative GLUT1 expression in U251shCtrl, U251shTERT, and U251shB1 cells (n = 6, respectively). (C) LDH activity in NHARas/TERT cells treated with siCtrl and siTERT (n = 4, respectively). (D) LDH activity in U251shCtrl, U251shTERT, and U251shB1 cells (n = 5, respectively). (E) Relative LDHA expression in NHARas/TERT cells treated with siCtrl and siTERT (n = 5, respectively). (F) Relative LDHA expression in U251shCtrl, U251shTERT, and U251shB1 cells (n = 3, respectively). (G) Relative MCT1 expression in NHARas/TERT cells treated with siCtrl and siTERT (n = 5, respectively). (H) Relative MCT1 expression in U251shCtrl, U251shTERT, and U251shB1 cells siTERT (n = 5, respectively). (I) Relative MCT4 expression in NHARas/TERT cells treated with siCtrl and siTERT (n = 5, respectively). (J) Relative MCT4 expression in U251shCtrl, U251shTERT, and U251shB1 cells (n = 4, respectively). (K) G6PD activity in siCtrl and siTERT in NHARas/TERT cells (n = 3, respectively). (L) G6PD activity in U251shCtrl, U251shTERT, and U251shB1 cells (n = 3, respectively). (M) Relative G6PD expression in NHARas/TERT cells treated with siCtrl and siTERT (n = 3, respectively). (N) Relative G6PD expression in U251shCtrl, U251shTERT, and U251shB1 cells (n = 3, respectively).

Fig. 4.

TERT silencing leads to a drop in hyperpolarized [1-¹³C]pyruvate conversion to lactate in live cells. (A) Spectral array of 13C spectra following [1-13C]pyruvate injection into NHARas/TERT cells treated with siCtrl. (B) Sum of the spectra comparing siCtrl and siTERT-treated NHARas/TERT cells. (C) Temporal evolution of [1-13C]lactate/[1-13C]pyruvate comparing siCtrl and siTERT-treated NHARas/TERT cells (n = 4, respectively). (D) The AUC of [1-13C]lactate/[1-13C]pyruvate comparing siCtrl and siTERT-treated NHARas/TERT cells (n = 4, respectively). (E) Temporal evolution of [1-13C]lactate/[1-13C]pyruvate comparing U251shCtrl, U251shTERT, and U251shB1 (n = 5, 3, 3, respectively). (F) The AUC of [1-¹³C]lactate/[1-¹³C]pyruvate comparing U251shCtrl, U251shTERT, and U251shB1 (n = 5, 3, 3, respectively).

Fig. 5.

In vivo ¹H MRS confirms that changes in lactate and GSH are associated with TERT silencing in glioblastoma. (A) Tumor volume change post-implantation comparing U251shCtrl, U251shTERT, and U251shB1 (Day28 ± 2 n = 3, 4, 3, Day35 ± 2 n = 3, 4, 3, Day42 ± 2 n = 0, 3, 3). The number of animals in U251Ctrl and U251shTERT groups dropped on Day42 ± 2 due to tumor-induced animal loss. (B) Representative ¹H-MRS spectrum and T₂-weighted image (insert) of U251shCtrl tumor. The tumor is contoured in white and the spectroscopic voxel position is illustrated in yellow in the anatomic image. (C) Univariate analysis of the composite lactate plus lipid peak comparing U251shCtrl, U251shTERT, and U251shB1 tumors, and normal brain (n = 3, respectively). (D) Univariate analysis of GSH levels comparing U251shCtrl, U251shTERT, and U251shB1 (n = 3, respectively).

Fig. 6.

In vivo ¹³C-MRS of hyperpolarized pyruvate metabolism confirms that a drop in hyperpolarized lactate production is associated with TERT silencing in glioblastoma. (A) Representative ¹³C spectra from each tumor model at 12 s after injection of hyperpolarized [1-¹³C]pyruvate. (B) Dynamic hyperpolarized ¹³C EPSI data of [1-¹³C]lactate acquired from one slice from each tumor model at 3 s temporal resolution. (C) Average dynamic lactate to pyruvate ratio of tumor voxel from each tumor model, contralateral brain and normal brain (n = 4, 3, 3, 9, 3, respectively). (D) AUC of [1-¹³C]lactate/[1-¹³C]pyruvate comparing U251shCtrl, U251shTERT, U251shB1 tumors, contralateral brain, and normal brain (n = 4, 3, 3, 9, 3, respectively).