Metabolic imaging detects elevated glucose flux through the pentose phosphate pathway associated with TERT expression in low-grade gliomas

Abstract

Background

Telomerase reverse transcriptase (TERT) is essential for tumor proliferation, including in low-grade oligodendrogliomas (LGOGs). Since TERT is silenced in normal cells, it is also a therapeutic target. Therefore, noninvasive methods of imaging TERT are needed. Here, we examined the link between TERT expression and metabolism in LGOGs, with the goal of leveraging this information for noninvasive magnetic resonance spectroscopy (MRS)-based metabolic imaging of LGOGs.

Methods

Immortalized normal human astrocytes with doxycycline-inducible TERT silencing, patient-derived LGOG cells, orthotopic tumors, and LGOG patient biopsies were studied to determine the mechanistic link between TERT expression and glucose metabolism. The ability of hyperpolarized [U-¹³C, U-²H]-glucose to noninvasively assess TERT expression was tested in live cells and orthotopic tumors.

Results

TERT expression was associated with elevated glucose flux through the pentose phosphate pathway (PPP), elevated NADPH, which is a major product of the PPP, and elevated glutathione, which is maintained in a reduced state by NADPH. Importantly, hyperpolarized [U-¹³C, U-²H]-glucose metabolism via the PPP noninvasively reported on TERT expression and response to TERT inhibition in patient-derived LGOG cells and orthotopic tumors. Mechanistically, TERT acted via the sirtuin SIRT2 to upregulate the glucose transporter GLUT1 and the rate-limiting PPP enzyme glucose-6-phosphate dehydrogenase.

Conclusions

We have, for the first time, leveraged a mechanistic understanding of TERT-associated metabolic reprogramming for noninvasive imaging of LGOGs using hyperpolarized [U-¹³C, U-²H]-glucose. Our findings provide a novel way of imaging a hallmark of tumor immortality and have the potential to improve diagnosis and treatment response assessment for LGOG patients.

Key Points.

• TERT expression mechanistically reprograms glucose metabolism via the pentose phosphate pathway.

• TERT-linked metabolic reprogramming can be leveraged for noninvasive glioma imaging using hyperpolarized ¹³C-glucose.

Importance of the Study.

MRI is the mainstay of glioma imaging. However, MRI fails to distinguish true tumor from morphologically similar areas of gliosis, edema, and necrosis. MRI also does not distinguish tumor recurrence from treatment-induced effects such as pseudoprogression and pseudoresponse. Therefore, there is a need to identify noninvasive methods that track molecular hallmarks of tumor proliferation. Telomere maintenance is one such hallmark and most cancers, including LGOGs, maintain telomere length via TERT expression. Here, we show that TERT acts via the sirtuin SIRT2 to reprogram glucose metabolism through the PPP, resulting in elevated levels of the redox metabolites NADPH and glutathione in LGOGs. Importantly, we have leveraged this information for noninvasive hyperpolarized [U-¹³C, U-²H]-glucose-based imaging of TERT expression in patient-derived LGOG models in vivo. Our study enables noninvasive visualization of a hallmark of tumor proliferation and has the potential to improve longitudinal analysis of tumor burden and treatment response in LGOG patients.

Telomere maintenance is a fundamental hallmark of cancer.¹ Telomeres are specialized cap-like nucleoprotein structures that protect chromosomal ends from DNA damage.² Progressive telomere shortening during cell division leads to growth arrest and constitutes a barrier to uncontrolled proliferation.^1,2 Tumor cells, therefore, need a mechanism of maintaining telomere length. Most tumors maintain telomere length by reactivating the expression of telomerase reverse transcriptase (TERT), which is the catalytic component of the enzyme telomerase that synthesizes telomeric DNA.^2,3 TERT expression is silenced at birth in normal somatic cells, with the exception of stem cells, and is reactivated in tumor cells via TERT promoter mutations.^3–5 Importantly, TERT promoter mutations are a prerequisite for rapid glioma growth in vivo,⁶ further underscoring the crucial role of TERT in glioma proliferation.

The clinical relevance of TERT arises from its diagnostic, prognostic, and therapeutic potential. Among low-grade gliomas defined by mutations in isocitrate dehydrogenase (IDHmut),⁷ TERT promoter mutations occur frequently in low-grade oligodendrogliomas (LGOGs) but rarely in low-grade astrocytomas,^7,8 which use a TERT-independent telomere maintenance mechanism known as the alternative lengthening of telomeres (ALT) pathway.^9,10 As a result, assessment of TERT status is a valuable addition to glioma classification.¹¹ Research also suggests that clinically meaningful patient prognosis can be achieved on the basis of three molecular markers, ie, IDHmut, 1p19q co-deletion, and TERT promoter mutations.⁸ TERT is also an attractive therapeutic target since TERT expression is silenced in normal cells or expressed in a tightly regulated manner from the wild-type TERT promoter in stem cells.^2,3 Since TERT promoter mutations are specific to tumor cells,³ disrupting TERT expression from the mutant TERT promoter has therapeutic potential.⁵

Previous studies have linked TERT expression to metabolic reprogramming in cancer.¹² TERT inhibition in melanoma cells downregulates glucose transport and glycolysis.¹³ In primary glioblastoma cells, pharmacological or genetic inhibition of TERT abrogates glucose flux through the pentose phosphate pathway (PPP)¹⁴ and represses fatty acid synthesis.¹⁵ TERT has also been linked to altered cellular redox via modulation of reduced glutathione (GSH).^16,17

Magnetic resonance spectroscopy (MRS) is a safe, non-radioactive, noninvasive method of imaging metabolism.¹⁸ Thermally polarized ¹³C-MRS following administration of ¹³C-labeled precursors traces metabolic fluxes, but its inherently low sensitivity limits its translational value.¹⁹ However, hyperpolarization enhances the signal-to-noise ratio (SNR) of ¹³C-MRS by >10000-fold and provides a noninvasive method of imaging metabolic fluxes.¹⁸ Importantly, hyperpolarized ¹³C-MRS is a translational method that is in clinical trials,¹⁸ and its feasibility has been established in glioma patients.^19,20

Due to its inherent link to tumor proliferation, TERT has the potential to serve as a biomarker of tumor burden and response to therapy. The objective of this study was to examine the link between TERT expression and glucose metabolism in LGOGs with the ultimate goal of leveraging this information for noninvasive MRS-based metabolic imaging. Using genetically engineered and patient-derived LGOG models, we show that TERT expression is mechanistically associated with elevated PPP flux and higher antioxidant capacity. Importantly, we show that hyperpolarized [U-¹³C, U-²H]-glucose noninvasively reports on TERT expression in LGOG models in vivo.

Materials and Methods

Detailed experimental procedures are provided in the Supplementary material.

Cell Culture

Immortalized IDHmut-expressing p53/pRb-deficient normal human astrocytes (NHA_CONTROL) have been previously described.²¹ The doxycycline-inducible NHA_TERT+ model was generated by engineering NHA_CONTROL cells such that TERT expression was regulated from a tetracycline (TET)-off promoter and could be silenced upon doxycycline addition (0.4 µg/mL). RNA interference was carried out using two non-overlapping siRNA pools (Dharmacon). Human SIRT2 (Addgene) was overexpressed in NHA_TERT− cells by transient transfection with polyethyleneimine. NHAs that lack TERT and use the ALT pathway (NHA_ALT) have been previously described.²² The ALT status of NHA_ALT cells was confirmed by measuring c-circles using telomeric qPCR with and without amplification by φ29 polymerase as described.^9,23 BT54 neurospheres and SF10417 cells were derived from LGOG patients and are grown in serum-free media.^24–27 Cell lines were routinely tested for mycoplasma contamination, authenticated by short tandem repeat fingerprinting, and assayed within 6 months.

Patient Biopsies

Biopsies from LGOG or astrocytoma patients or gliosis biopsies from epileptic patients were obtained from the UCSF Brain Tumor Center Biorepository in compliance with the informed consent policy.

Gene Expression and Activity

Steady-state metabolite levels (NADPH, NADP+, GSH, oxidized glutathione [GSSG], 6-phosphogluconate, lactate, reactive oxygen species [ROS]) and enzyme activities (telomerase, glucose-6-phosphate dehydrogenase [G6PDH], phosphoglycerate kinase 1 [PGK1], hexokinase) were determined using commercial kits. γH2AX and GLUT1 protein levels were measured by ELISA. Gene expression was assessed by quantitative RT-PCR and normalized to β-actin (Supplementary Methods).

¹³C-MRS of Cell Extracts

Cells were cultured in a medium containing 5 mM [2-¹³C]-glucose for 48 hours and metabolites extracted by the methanol-chloroform method.^26,2713C-MR (30° flip angle, 3 second relaxation delay) spectra were obtained using a 11.7T spectrometer and peak integrals quantified to obtain metabolite concentrations. PPP flux was measured by quantifying the isotopomers of glutamate generated after PPP and glycolytic metabolism of [2-¹³C]-glucose.⁶ Glucose levels in media samples were quantified and uptake calculated as the difference in normalized fmol (Δfmol/cell) between 0 and 24 hours.

Hyperpolarized ¹³C-MRS of Live Cells

Hyperpolarized [U-¹³C, U-²H]-glucose prepared as described²⁸ was dissolved in isotonic buffer (50 mM Tris-HCl, 13.3 mM MgCl₂ in D₂O, pH 8) and added to live cells²³ at a concentration of 12 mM. ¹³C spectra were acquired on a 14T Bruker spectrometer (13° flip angle every 3 seconds for 300 seconds), and the area under the curve (AUC) for product was normalized to AUC of substrate and to cell number.

MRI

Animal studies were conducted in accordance with UCSF Institutional Animal Care and Use Committee guidelines. Doxycycline-inducible NHA_TERT+, NHA_ALT, or SF10417 cells (3 × 10⁵ cells/10 µL) were intracranially injected into athymic nude rats (male, rnu/rnu homozygous).^23,26,27 T2-weighted MRI was performed on a 3T Bruker scanner using a spin-echo TurboRARE sequence (TE/TR = 64/3484 ms, FOV = 43 × 43 mm,² 256 × 256, slice thickness = 1 mm, NA = 6).²³ Tumor volume was determined using in-house MATLAB codes (https://github.com/ViswanathLab/EPSI).²³ For doxycycline-mediated TERT silencing in vivo, doxycycline was administered in the feed ad libitum once tumors reached a volume of ~50 mm³ as previously described.²⁹

Hyperpolarized ¹³C-MRS in vivo

Studies were performed on a Bruker 3T spectrometer. Following tumor implantation, tumor volume was monitored longitudinally by MRI and ¹³C studies were performed once tumors reached a volume of ~50 mm.³ For slab studies, dynamic ¹³C spectra were acquired from a 12 mm axial slab through the brain every 3 seconds using a flyback spectral-spatial radiofrequency pulse³⁰ following the intravenous injection of 150 mM hyperpolarized glucose. For imaging, a 2D flyback spectral-spatial echo-planar spectroscopic imaging (EPSI) pulse³⁰ was used with a final hyperpolarized glucose concentration of 250 mM. ¹³C slab spectra were analyzed using Mnova. 2D EPSI data were analyzed using in-house MATLAB codes (https://github.com/ViswanathLab/EPSI).²³

Statistical Analysis

All experiments were performed on a minimum of 3 samples (n ≥ 3) and results presented as mean ± standard deviation. Statistical significance was assessed using an unpaired Student’s t test or ordinary 1-way ANOVA with P < .05 considered significant. *P < .05, **P < .01, ***P < .005, ****P < .0001, ns = non-significant.

Results

TERT Expression Is Associated With Higher Glucose Flux via the PPP in LGOGs

To begin with, we studied genetically engineered immortalized normal human astrocytes (NHAs) that express IDHmut, which is characteristic of low-grade gliomas⁷ and examined NHAs that lacked TERT (NHA_CONTROL) and those in which TERT expression was placed under the control of a doxycycline-inducible promoter. TERT expression and telomerase activity were observed in these cells in the absence of doxycycline (NHA_TERT+), significantly reduced upon doxycycline addition (NHA_TERT−) and restored following doxycycline removal (NHA_{TERT+ rescue}; Supplementary Figure S1A, B). As additional controls, we also examined NHAs that do not express TERT, but instead use the ALT pathway (NHA_ALT) for telomere maintenance.^22,23 The ALT phenotype of NHA_ALT cells has previously been verified via quantification of c-circles, which are extrachromosomal DNA circles characteristic of the ALT pathway^22,23 (Supplementary Figure S1C).

Thermally polarized ¹³C-MRS following administration of [2-¹³C]-glucose can inform on flux via the PPP, which produces [4-¹³C]-glutamate, and via glycolysis which yields [5-¹³C]-glutamate (Figure 1A).³¹ PPP flux was significantly higher in NHA_TERT+ cells relative to NHA_CONTROL and NHA_ALT (Figure 1B). Importantly, doxycycline-induced TERT silencing reduced PPP flux in NHA_TERT− cells to levels observed in NHA_CONTROL and NHA_ALT, an effect that was reversed in NHA_{TERT+ rescue} cells. In order to confirm the clinical relevance of these results, we examined PPP flux in the patient-derived BT54 and SF10417 models.^25–27 We confirmed elevated TERT expression and telomerase activity in BT54 and SF10417 cells relative to NHA_CONTROL (Supplementary Figure S1D, E). Silencing TERT in BT54 and SF10417 cells (Supplementary Figure S1F–I) significantly reduced PPP flux (Figure 1C and Supplementary Figure S1J), pointing to a causal link between TERT expression and PPP flux in LGOGs.

Fig. 1.

TERT expression is associated with elevated PPP flux in LGOG cells. (A) Representative ¹³C-MR spectrum from NHA_TERT+ cells incubated with [2-¹³C]-glucose. Peaks for glutamate, glutamine, and 2-hydroxyglutarate (2-HG; product of the IDHmut enzyme^26,27) are labeled. Insets show an expansion of the regions corresponding to [5-¹³C]-glutamate (left) and [4-¹³C]-glutamate (right). (B) PPP flux in NHA_CONTROL, NHA_ALT, NHA_TERT+, NHA_TERT−, and NHA_{TERT+ rescue} cells. (C) Effect of TERT silencing on PPP flux in BT54 neurospheres. (D) Schematic illustration of glucose metabolism via the PPP and glycolysis. NADPH (E) and GSH (F) in NHA_CONTROL, NHA_ALT, NHA_TERT+, NHA_TERT−, and NHA_{TERT+ rescue} cells. Effect of TERT silencing on NADPH (G), GSH (H), and ROS (I) in BT54 neurospheres. Effect of TERT silencing on ROS (J) in SF10417 cells. Abbreviations: GSH, glutathione; LGOG, low-grade oligodendrogliomas; PPP, pentose phosphate pathway; ROS, reactive oxygen species; TERT, telomerase reverse transcriptase.

TERT Expression Is Associated With Higher Antioxidant Capacity in LGOGs

Glucose flux through the PPP is the largest NADPH-producing pathway.³² NADPH, in turn, is essential for maintaining GSH in the reduced state³² (see schematic in Figure 1D). Since our results linked TERT to higher PPP flux, we examined whether there was a concomitant change in NADPH and GSH. In line with PPP flux, NADPH and GSH were elevated in NHA_TERT+ cells relative to NHA_CONTROL and NHA_ALT (Figure 1E, F). Doxycycline-induced TERT silencing reduced NADPH and GSH, an effect that was rescued by doxycycline removal (Figure 1E, F). Importantly, TERT silencing significantly reduced NADPH and GSH in the BT54 (Figure 1G, H) and SF10417 (Supplementary Figure S2A, B) models. There was no change in NADP+ or GSSG in our models (Supplementary Figure S2C–H).

NADPH and GSH mitigate oxidative stress via suppression of ROS (see Figure 1D). TERT silencing significantly increased ROS in BT54 and SF10417 cells (Figure 1I, J). Conversely, TERT expression was associated with reduced ROS in the NHA_TERT+ model (Supplementary Figure S2I). Taken together, these results link TERT expression to higher redox capacity and reduced oxidative stress in LGOGs.

Hyperpolarized [U-¹³C, U-²H]-Glucose Noninvasively Monitors TERT Expression in LGOG Cells

Hyperpolarized [U-¹³C, U-²H]-glucose has been used to probe glucose metabolism via glycolysis and the PPP in tumor cells other than gliomas.^28,33–36 Since our results linked TERT to higher PPP flux, we examined hyperpolarized [U-¹³C, U-²H]-glucose (T1 = 14 ± 2 seconds, 24% polarization, consistent with prior publications^28,33–36) metabolism in our models. As shown in Figure 2A, we observed buildup of the PPP metabolite [1-¹³C]-6-phosphogluconate (6-PG, 179.4 ppm; see schematic in Figure 1D) in NHA_TERT+ cells. We did not observe [1-¹³C]-lactate (183.4 ppm). Importantly, doxycycline-induced TERT silencing significantly reduced 6-PG production, and restoration of TERT expression via doxycycline removal restored 6-PG production (Figure 2B). The chemical shift of [1-¹³C]-6-PG (179.4 ppm) is indistinguishable from that of [1-¹³C]-3-phosphoglycerate (179.5 ppm) which can also be produced from glucose via glycolysis (see Figure 1D). Previous studies have disagreed on the assignment of this resonance and have variously assigned it to either 6-PG^28,33,35 or 3-phosphoglycerate.^34,36 In order to determine whether the resonance observed at 179.4 ppm in our experiments corresponded to 6-PG, we examined hyperpolarized [U-¹³C, U-²H]-glucose metabolism in NHA_TERT+ cells in which expression of G6PDH, the enzyme responsible for 6-PG synthesis (see Figure 1D), was silenced (Supplementary Figure S3A, B). As shown in Figure 2C, D, the resonance at 179.4 ppm was lost in NHA_TERT+ G6PDH− cells. In contrast, the 179.4 ppm peak was observed in NHA_TERT+ cells in which PGK1, the glycolytic enzyme catalyzing 3-phosphoglycerate synthesis (see Figure 1D), was silenced (Supplementary Figure S3C, D), suggesting that 179.4 ppm peak likely corresponds to 6-PG.

Fig. 2.

TERT-linked increase in PPP flux can be noninvasively monitored using hyperpolarized [U-¹³C, U-²H]-glucose in live LGOG cells. (A) Representative spectral array of hyperpolarized [U-¹³C, U-²H]-glucose metabolism to 6-PG in NHA_TERT+ cells. Inset shows an expansion of the region around 6-PG. (B) 6-PG/glucose ratio in the doxycycline-inducible NHA_TERT+ model. Representative summed ¹³C-MR spectra (C) and quantification of 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose (D) in NHA_TERT+ cells in which G6PDH or PGK1 has been silenced. (E) Representative spectral array of hyperpolarized [U-¹³C, U-²H]-glucose metabolism to lactate in NHA_ALT cells. Inset shows an expansion of the region around lactate. Representative summed ¹³C spectra showing the effect of TERT silencing on hyperpolarized [U-¹³C, U-²H]-glucose metabolism in the BT54 (F) or SF10417 (G) models. (H) Effect of 6-thio-2′-deoxyguanosine (6-thio-dG) on the 6-PG/glucose ratio in SF10417 cells. Abbreviations: LGOG, low-grade oligodendrogliomas; PPP, pentose phosphate pathway; TERT, telomerase reverse transcriptase.

In contrast to NHA_TERT+ cells, NHA_ALT cells produced the glycolytic metabolite lactate (183.4 ppm, Figure 2E), but not 6-PG, from hyperpolarized [U-¹³C, U-²H]-glucose. In order to further confirm these results, we quantified 6-PG and lactate by spectrophotometric assays in extracts from cell suspensions that were rapidly frozen at the end of our hyperpolarized ¹³C-MRS experiments. We detected 6-PG but not lactate in NHA_TERT+ extracts and lactate but not 6-PG in NHA_ALT extracts (Supplementary Figure S3E). NHA_TERT− extracts showed significantly reduced 6-PG relative to NHA_TERT+, but lactate could not be detected. Silencing G6PDH, but not PGK1, in NHA_TERT+ cells reduced 6-PG production (Supplementary Figure S3F). We did not observe lactate in extracts from NHA_TERT+ G6PDH− or PGK− cells (Supplementary Figure S3G). These results suggest that, within the temporal window of the hyperpolarized ¹³C-MRS experiment (~5 minutes), hyperpolarized [U-¹³C, U-²H]-glucose metabolism to 6-PG distinguishes TERT-expressing cells from cells lacking TERT.

We confirmed the clinical relevance of these results in the patient-derived BT54 and SF10417 models. TERT silencing abrogated 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose in both models (Figure 2F, G and Supplementary Figure S3H, I). To assess whether hyperpolarized [U-¹³C, U-²H]-glucose provides a readout of response to TERT inhibition, we examined SF10417 cells treated with the telomerase inhibitor 6-thio-2′-deoxyguanosine.³⁷ Induction of DNA damage (Supplementary Figure S3J) was associated with significantly reduced 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose in 6-thio-2′-deoxyguanosine-treated SF10417 cells (Figure 2H). Collectively, these results suggest that hyperpolarized [U-¹³C, U-²H]-glucose flux to 6-PG reports on TERT expression and response to telomerase inhibition in LGOG cells.

Hyperpolarized [U-¹³C, U-²H]-Glucose Metabolism to 6-PG Noninvasively Monitors TERT Expression in LGOGs In Vivo

Next, we examined the feasibility of imaging TERT in vivo using hyperpolarized [U-¹³C, U-²H]-glucose. Dynamic ¹³C spectra were acquired from a 12 mm slab of the brain following intravenous injection of hyperpolarized [U-¹³C, U-²H]-glucose into rats bearing orthotopic NHA_TERT+, SF10417, or NHA_ALT tumors (tumor volume ~50 mm³). Consistent with our cell studies, we observed buildup of [1-¹³C]-6-PG in NHA_TERT+ and SF10417 tumor-bearing animals (Figure 3A, B), while NHA_ALT tumor-bearing rats produced [1-¹³C]-lactate (Supplementary Figure S4A). Tumor-free controls produced neither 6-PG nor lactate (Supplementary Figure S4B). These differences are highlighted in the summed ¹³C spectra shown in Figure 3C. The 6-PG/glucose ratio differentiated NHA_TERT+ or SF10417 tumors from healthy brain (Figure 3D) while the lactate/glucose ratio differentiated NHA_ALT tumors from healthy brain (Supplementary Figure S4C). The temporal maximum (9 seconds post-injection; Figure 3E, F and Supplementary Figure S4D, E) and the SNR (Supplementary Figure S4F) of hyperpolarized [U-¹³C, U-²H]-glucose were similar in NHA_TERT+, SF10417, or NHA_ALT tumor-bearing animals and tumor-free controls, indicating that the observed differences in metabolism did not result from differences in glucose delivery. The maxima for 6-PG (15 seconds post-injection; Figure 3E, F) in TERT-expressing tumors and for lactate (21 seconds post-injection; Supplementary Figure S4D) in NHA_ALT tumors were delayed relative to glucose. Taken together with the lack of 6-PG or lactate in tumor-free rats, these results suggest that 6-PG or lactate production in tumor-bearing animals originates from metabolism within the tumor.

Fig. 3.

Hyperpolarized [U-¹³C, U-²H]-glucose metabolism can be noninvasively monitored in vivo. Representative ¹³C spectral array of hyperpolarized [U-¹³C, U-²H]-glucose metabolism in a rat bearing an orthotopic NHA_TERT+ (A) or SF10417 (B) tumor. (C) Representative summed ¹³C spectra of hyperpolarized [U-¹³C, U-²H]-glucose metabolism in rats bearing orthotopic NHA_TERT+ or NHA_ALT tumors or tumor-free rats. (D) 6-PG/glucose ratio from slab studies in tumor-free normal brain, NHA_TERT+ or SF10417 tumors. Dynamics of hyperpolarized [U-¹³C, U-²H]-glucose metabolism in rats bearing orthotopic NHA_TERT+ (E) or SF10417 (F) tumors. Representative summed ¹³C spectra (G) and quantification of 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose (H) in rats bearing orthotopic NHA_TERT+ tumors before (NHA_TERT+) or after doxycycline-mediated TERT silencing (NHA_TERT−). Abbreviation: TERT, telomerase reverse transcriptase.

To assess the ability of hyperpolarized [U-¹³C, U-²H]-glucose to report on response to TERT inhibition in vivo, we silenced TERT expression in a doxycycline-inducible manner in rats bearing orthotopic NHA_TERT+ tumors once tumors reached a volume of ~50 mm.³ TERT expression and telomerase activity were significantly reduced following doxycycline treatment without alterations in tumor volume (Supplementary Figure S5A–C), consistent with previous studies pointing to a lag period before cell death occurs due to TERT knockdown.³⁸ Importantly, doxycycline-induced TERT silencing abrogated 6-PG production in vivo (Figure 3G, H). There was no difference in the SNR of hyperpolarized glucose (Supplementary Figure S5D), indicating that loss of 6-PG production did not result from differences in glucose delivery.

Finally, to assess the spatial distribution of hyperpolarized [U-¹³C, U-²H]-glucose metabolism, we performed 2D EPSI on rats bearing orthotopic NHA_TERT+ or SF10417 tumors. Metabolic heatmaps generated by overlaying ¹³C data acquired at 15 seconds post-injection of hyperpolarized [U-¹³C, U-²H]-glucose (corresponding to the time-point of maximum 6-PG in slab studies; see Figure 3E, F) on the anatomical MRI indicated localization of 6-PG to the tumor while glucose was distributed throughout the brain (Figure 4A, B). Importantly, consistent with the slab studies, the 6-PG/glucose ratio differentiated tumor from contralateral normal brain while there was no difference in the SNR of hyperpolarized [U-¹³C, U-²H]-glucose (Figure 4C–F). Collectively, these results suggest that 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose is a metabolic imaging biomarker of TERT expression in vivo, including in clinically relevant patient-derived LGOG models.

Fig. 4.

Hyperpolarized [U-¹³C, U-²H]-glucose noninvasively monitors tumor burden in LGOGs in vivo. (A) Left panel: Representative T2-weighted MRI from an NHA_TERT+ tumor-bearing rat with the tumor contoured in white. Middle panel: Metabolic heatmap of hyperpolarized [U-¹³C, U-²H]-glucose distribution overlaid over the corresponding MRI; Right panel: Metabolic heatmap of [1-¹³C]-6-PG distribution. (B) Left panel: Representative T2-weighted MRI from a rat bearing an orthotopic SF10417 tumor. The tumor is outlined in white. Middle panel: Metabolic heatmap of hyperpolarized [U-¹³C, U-²H]-glucose distribution overlaid over the corresponding MRI; Right panel: Metabolic heatmap of hyperpolarized 6-PG production. 6-PG/glucose ratio (C) and the SNR of hyperpolarized [U-¹³C, U-²H]-glucose (D) in rats bearing orthotopic NHA_TERT+ tumors. 6-PG/glucose ratio (E) and the SNR of hyperpolarized [U-¹³C, U-²H]-glucose (F) in SF10417 tumor-bearing rats. Abbreviations: LGOG, low-grade oligodendrogliomas; SNR, signal-to-noise ratio.

TERT Acts via SIRT2 to Modulate GLUT1 and G6PDH Expression in LGOGs

To identify the molecular mechanisms by which TERT modulates PPP flux in LGOGs, we interrogated enzymes and transporters involved in glucose metabolism upstream of 6-PG (see schematic in Figure 1D). In line with PPP flux (see Figure 1B), G6PDH expression (Supplementary Figure S6A) and activity (Figure 5A) were downregulated by doxycycline-induced TERT silencing in the NHA_TERT+ model. Expression of the glucose transporter GLUT1 (Figure 5B and Supplementary Figure S6B) and [2-¹³C]-glucose uptake (Supplementary Figure S6C) were also reduced in NHA_TERT− cells. There was no alteration in hexokinase activity (Supplementary Figure S6D). TERT silencing also downregulated GLUT1 and G6PDH in the patient-derived SF10417 model (Supplementary Figure S6E, F). Importantly, GLUT1 and G6PDH were downregulated in ex vivo tumor tissues resected from rats bearing NHA_TERT− tumors relative to NHA_TERT+ (Supplementary Figure S6G, H).

Fig. 5.

TERT mechanistically modulates glucose metabolism in LGOGs. G6PDH activity (A) and GLUT1 protein (B) in NHA_TERT+, NHA_TERT−, and NHA_{TERT+ rescue} cells. Effect of GLUT1 (C) or G6PDH (D) silencing on 6-PG production from hyperpolarized glucose in BT54 neurospheres. (E) SIRT2 expression in NHA_TERT+, NHA_TERT−, and NHA_{TERT+ rescue} cells. (F) Effect of TERT silencing on SIRT2 expression in BT54 neurospheres. Effect of SIRT2 silencing on PPP flux (G), GLUT1 protein (H), and G6PDH activity (I) in NHA_TERT+ cells. Effect of SIRT2 overexpression on PPP flux (J), GLUT1 protein (K), and G6PDH activity (L) in NHA_TERT− cells (NHA_{TERT-SIRT2 OE}). Abbreviations: LGOG, low-grade oligodendrogliomas; PPP, pentose phosphate pathway; TERT, telomerase reverse transcriptase.

To assess the role of GLUT1 and G6PDH in TERT-linked upregulation of PPP flux, we examined the effect of silencing GLUT1 or G6PDH in BT54 neurospheres. GLUT1 silencing (Supplementary Figure S7A) significantly reduced [2-¹³C]-glucose uptake (Supplementary Figure S7B). Although PPP fractional flux was unaltered (both PPP-derived [4-¹³C]-glutamate and glycolysis-derived [5-¹³C]-glutamate were reduced; Supplementary Figure S7C–E), 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose was significantly reduced in BT54 GLUT1− neurospheres (Figure 5C). These results suggest that TERT-mediated increase in GLUT1 leads to higher glucose uptake, which, in turn, leads to higher availability of glucose for PPP flux. Silencing G6PDH in BT54 neurospheres (Supplementary Figure S7F) also reduced 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose (Figure 5D), in line with results from the NHA_TERT+ model (Figure 2D).

The sirtuin SIRT2 has been shown to activate G6PDH and enhance NADPH production in mammalian cells.^39,40 SIRT2 expression was significantly reduced by doxycycline-mediated TERT silencing in the NHA_TERT+ model (Figure 5E). TERT silencing also downregulated SIRT2 expression in BT54 neurospheres (Figure 5F). Importantly, silencing SIRT2 in NHA_TERT+ cells (Supplementary Figure S8A) downregulated PPP flux, GLUT1, and G6PDH (Figure 5G–I). Conversely, SIRT2 overexpression in NHA_TERT− cells (Supplementary Figure S8B) rescued the reduction in PPP flux, GLUT1, and G6PDH caused by TERT silencing (Figure 5J–L). Collectively, these results mechanistically link TERT to SIRT2, which, in turn, upregulates GLUT1 and G6PDH, resulting in higher glucose uptake and flux via the PPP.

TERT Expression Is Associated With Metabolic Reprogramming in LGOG Patient Biopsies

To evaluate the clinical relevance of our metabolic biomarkers, we examined biopsies from LGOG patients and compared them to non-neoplastic gliosis biopsies and biopsies from low-grade astrocytoma tumors that do not express TERT.^9,10 In line with the results from preclinical models, TERT expression and telomerase activity in LGOG biopsies (Figure 6A, B) were associated with elevated NADPH, GSH, 6-PG, SIRT2, GLUT1, and G6PDH relative to gliosis and astrocytoma (Figure 6C–J). These results link TERT to glucose metabolism in patient samples and point to the potential clinical utility of hyperpolarized glucose for assessing TERT status.

Fig. 6.

TERT expression is linked to glucose metabolism in LGOG biopsies. TERT mRNA (A), telomerase activity (B), NADPH (C), GSH (D), 6-PG (E), SIRT2 mRNA (F), GLUT1 mRNA (G), GLUT1 protein (H), G6PDH mRNA (I), and G6PDH activity (J) in LGOG, gliosis, and astrocytoma biopsies. Abbreviations: GSH, glutathione; LGOG, low-grade oligodendrogliomas; TERT, telomerase reverse transcriptase.

Discussion

TERT is indispensable for tumor proliferation, including in LGOGs.^1,2 Using NHAs that have been engineered to silence TERT in a doxycycline-inducible manner as well as patient-derived LGOG models, we show that TERT upregulates glucose flux via the PPP and elevates antioxidant capacity. Importantly, we show that hyperpolarized [U-¹³C, U-²H]-glucose flux to 6-PG via the PPP can be used for noninvasive imaging of TERT expression in preclinical LGOG models in vivo.

Although we cannot rule out the possibility that glucose metabolism is modulated by other molecular pathways, our results point to a mechanistic association between TERT and the PPP in LGOGs. TERT silencing downregulates PPP flux, NADPH, and GSH in the NHA_TERT+ and patient-derived LGOG models. Mechanistically, TERT acts by increasing GLUT1-mediated glucose uptake and upregulating G6PDH, the rate-limiting enzyme for 6-PG and NADPH synthesis. Importantly, the ability of SIRT2 overexpression to restore PPP flux, GLUT1, and G6PDH in NHA_TERT− cells and, conversely, the reduction in PPP flux, GLUT1, and G6PDH in NHA_TERT+ cells in which SIRT2 is silenced, identify SIRT2 as a novel mediator of TERT-induced alterations in glucose metabolism.

Studies have used positron emission tomography (PET) or optical imaging for TERT detection.^41,42 However, the radioactive nature of PET limits longitudinal imaging due to concerns of radiation exposure while the restricted depth penetration of optical imaging limits application to gliomas. Recent studies have also used MRI combined with radiomics or ¹H-MRS to assess TERT status in LGOGs.^43,44 Our study identifies a complementary hyperpolarized ¹³C-glucose-based method of imaging TERT status. 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose is consistently associated with TERT expression in LGOG cells and orthotopic tumors in vivo. In contrast, NHA_ALT cells and tumors that do not express TERT and, instead, use the ALT pathway for telomere maintenance,^22,23 produce lactate from hyperpolarized [U-¹³C, U-²H]-glucose, further confirming the utility of 6-PG production as a biomarker of TERT in LGOGs. Importantly, our results are significant in light of the limited utility of ¹⁸F-fluorodeoxyglucose-positron emission tomography (FDG-PET) for glioma imaging, which stems from high glucose uptake in the normal brain.⁴⁵ The ability of 6-PG production to demarcate tumor from contralateral brain in our EPSI studies indicates that monitoring hyperpolarized ¹³C-glucose metabolism to 6-PG, as opposed to monitoring glucose uptake by FDG-PET, can be useful for imaging glioma burden in vivo.

As an endogenous nutrient that readily crosses the blood-brain barrier⁴⁶ and is non-toxic at the doses in our study, hyperpolarized [U-¹³C, U-²H]-glucose has potential for clinical translation. The main limitation is the short T1 (~14 seconds at 3T). However, 6-PG has a relatively longer T1 (~32 seconds at 3T),⁴⁷ suggesting that clinically useful information can still be derived from the use of hyperpolarized [U-¹³C, U-²H]-glucose. A recent study⁴⁸ also achieved ~70% polarization for hyperpolarized [U-¹³C, U-²H]-glucose (as opposed to ~24% in our study), a method that could significantly improve the SNR. The use of singly labeled [2-¹³C, U-²H]-glucose³⁶ could also increase SNR since the resulting 6-PG peak would be a singlet.

Although some TERT inhibitors have encountered toxicity issues in clinical trials due to stem cell inhibition,³ next-generation TERT inhibitors such as 6-thio-2′-deoxyguanosine have been successful in preclinical studies and are now in clinical trials.⁴⁹ Our studies show that 6-thio-2′-deoxyguanosine inhibits 6-PG production from hyperpolarized [U-¹³C, U-²H]-glucose in the patient-derived SF10417 model. Importantly, doxycycline-induced TERT silencing in vivo abrogates 6-PG production in the NHA_TERT+ model. These results point to the potential utility of hyperpolarized [U-¹³C, U-²H]-glucose in monitoring response to TERT inhibitors and, thereby, enabling their clinical translation.

Glioma patient management is heavily dependent on MRI. However, distinguishing tumor from lesions such as gliosis and edema can be difficult using MRI. Importantly, MRI-based treatment response assessment is complicated by the occurrence of pseudoprogression, ie, treatment-related effects such as necrosis and inflammation that mimic tumor recurrence.⁵⁰ Our studies showing that TERT expression is associated with alterations in glucose metabolism in LGOG patient biopsies relative to gliosis point to the potential utility of metabolic imaging of TERT status for differentiating tumor from gliosis. Further studies in vivo are needed to fully determine the utility of hyperpolarized glucose in differentiating tumors from gliosis, pseudoprogression, or pseudoresponse.

In summary, our study mechanistically links TERT expression to elevated PPP flux and identifies hyperpolarized [U-¹³C, U-²H]-glucose as a potential metabolic imaging probe of TERT in LGOGs. By enabling noninvasive assessment of a hallmark of tumor immortality, our study has the potential to improve longitudinal tumor imaging and treatment response monitoring in LGOG patients.

Funding

This work was supported by: National Institutes of Health (R01CA239288 to P.V., S.M.R.; UCSF SPORE Career Enhancement Award P50CA97257 to P.V.; P01CA118816 to S.M.R., R01NS105087 to R.O.P., P41EB013598 Center grant), Department of Defense (W81XWH201055315 to P.V.), American Cancer Society (131715-RSG-18-005-01-CCE to P.E.Z.L.), NICO.

Conflict of interest statement. The authors have no conflicting interests to disclose.

Authorship statement. P.V. and S.M.R. conceptualized the research; P.V. designed the experiments; P.V., G.B., C.T., and V.A. performed experiments; A.M.G. assisted with cell and in vivo studies; P.E.Z.L. contributed to the design of imaging studies; H.A.L., J.F.C., and R.O.P. provided cell lines; P.V. wrote the manuscript; G.B. commented on the manuscript; P.V., S.M.R., and R.O.P. secured funding.