Daily glucocorticoids promote glioblastoma growth and circadian synchrony to the host

Abstract

Glioblastoma (GBM) is the most common primary malignant brain tumor in adults with a poor prognosis despite aggressive therapy. Here, we hypothesized that daily host signaling regulates tumor growth and synchronizes circadian rhythms in GBM. We find daily glucocorticoids promote or suppress GBM growth through glucocorticoid receptor (GR) signaling depending on time of day and the clock genes, Bmal1 and Cry. Blocking circadian signals, like vasoactive intestinal peptide or glucocorticoids, dramatically slows GBM growth and disease progression. Analysis of human GBM samples from The Cancer Genome Atlas (TCGA) shows that high GR expression significantly increases hazard of mortality. Finally, mouse and human GBM models have intrinsic circadian rhythms in clock gene expression in vitro and in vivo that entrain to the host through glucocorticoid signaling, regardless of tumor type or host immune status. We conclude that GBM entrains to the circadian circuit of the brain, modulating its growth through clock-controlled cues, like glucocorticoids.

Keywords: Bmal1; TMZ; VIP; cancer neuroscience; circadian rhythms; clock genes; dexamethasone; glioblastoma; glucocorticoids; period gene; temozolomide; vasoactive intestinal peptide.

Figure 1:. Dexamethasone promotes GBM growth through glucocorticoid receptor signaling dependent on circadian time of treatment in vitro

(A) Schematic of cell transduction with reporters of clock gene (Period2, Per2, or Bmal1) expression. (B, C) Human LN229 (B) and murine GL261 (C) GBM cell lines transduced with a Per2- or Bmal1-driven luciferase reporter (Per2:luc (P2L) and Bmal1:luc (B1L), respectively) show circadian rhythms in clock gene expression in vitro (mean±SEM, n reported in figure, all recordings had cosine fits with correlation coefficients, CC > 0.9). (D, E) Acute treatment of LN229 (D) and GL261 (E) GBM WT cells with 100nM DEX in vitro suppressed cell proliferation about 3-fold when added at the peak of daily Per2 expression, and increased growth about 3-fold when administered at the trough (mean±SEM, n = 3 per group, two-way ANOVA with Šídák’s multiple comparisons test, *p < 0.05, **p < 0.01, ****p < 0.0001). (F, G) LN229 (F) and GL261 (G) cells lacking the glucocorticoid receptor (GR KD) did not proliferate in response to DEX treatment at either the daily peak or trough of Per2 expression in vitro (mean±SEM, n = 3 per group, two-way ANOVA with Šídák’s multiple comparisons test, ns p > 0.05). See also Figures S1 and S2.

Figure 2:. Dexamethasone promotes GBM growth through glucocorticoid receptor signaling when administered in the morning around the daily trough of Per2 expression in vivo

(A) Schematic of orthotopic xenograft into mouse basal ganglia and bioluminescence imaging. Mice were implanted with GBM cells expressing a Per2:luc (P2L) or Bmal1:luc (B1L) reporter to record tumor clock gene expression. (B-C) Representative bioluminescence images of LN229 (B) and GL261 (C) tumor xenografts in mice during the day (ZT0–12, yellow bar) and night (ZT12–24, grey bar) show high Bmal1 expression during the day, and high Per2 expression during the night (BLI counts are x10³, color bar depicts relative bioluminescence intensity). (D-E) GBM xenografts show reliable peak Per2 (yellow line) expression at night and Bmal1 (green line) during the day when implanted in nude (D) or C57BL/6NJ (E) male or female mice (Zeitgeber time (ZT), mean±SEM, n reported in figure, average traces scored circadian by JTK cycle p < 0.05). Light yellow background represents daytime and grey background represents nighttime. (F) Schematic of two DEX treatment paradigms after tumor implantation. Each dot represents a time of treatment; orange dots represent treatment with DEX, while grey dots represent treatment with vehicle. Yellow bars represent daytime and grey bars represent nighttime. (G-H) DEX in the morning promotes LN229 (G) and GL261 (H) tumor growth in vivo. Tumor size increased 2- to 5-fold when treating mice implanted with wild type (WT) LN229 or GL261 cells in the morning (ZT4, AM) compared to evening (ZT12, PM) or vehicle (Veh). DEX in the morning or evening did not significantly induce tumor growth in mice implanted with GR KD LN229 cells (mean±SEM, n = 8 in WT LN229-bearing mice receiving DEX AM or PM, n = 4 in all other groups, BLI counts on representative images are x10³, color bar depicts relative bioluminescence, one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). (I) Mice bearing LN229 WT tumors treated with DEX in the morning lost more body weight from start to end of the experiment compared to mice treated in the evening or with vehicle, or bearing GR KD cells (mean±SEM, n = 4 per group, left: two-way ANOVA with Tukey’s multiple comparisons test, right: one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01). (J) Mice bearing GL261 WT tumors treated with DEX in the morning lost more body weight from start to end of the experiment compared to mice treated in the evening or with vehicle (mean±SEM, n = 4 per group, left: two-way ANOVA with Tukey’s multiple comparisons test, right: one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01). See also Figures S3, S4, S5, and S6.

Figure 3:. Dexamethasone-induced growth (around the daily trough of Per2 or morning) and suppression (around the daily peak of Per2 or evening) depends on an intact circadian clock in GBM cells

(A) BMAL1 KD reduced the amplitude of daily rhythms in Per2 in human LN229-P2L cells in vitro (hereafter abbreviated as Bmal1 KD, shRNA construct 14, mean±SEM, n reported in figure, all recordings had cosine fits with correlation coefficients of CC < 0.7 Bmal1 KD). (B) LN229-P2L cells with Bmal1 KD showed no differences in growth in response to 100nM DEX or vehicle treatment around the peak or trough of Per2 expression in vitro (mean±SEM, n = 3 per group, two-way ANOVA with Šídák’s multiple comparisons test, ns p > 0.05). (C) DEX increased growth by about 3-fold when administered in the morning (ZT4, AM) compared to the evening (ZT12, PM) in vivo. No significant differences in tumor size were observed in mice bearing LN229 Bmal1 KD tumors, or treated with DEX in the morning or evening (mean±SEM, n = 4 per group, BLI counts on representative images are x10³, color bar depicts relative bioluminescence, one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05). (D) Mice bearing LN229 WT tumors treated with DEX in the morning (orange) lost more weight from start to end of the experiment compared to mice treated in the evening (blue), or bearing a LN229 Bmal1 KD tumor (mean±SEM, n = 4 per group, left: two-way ANOVA with Tukey’s multiple comparisons test, right: one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05). (E) Human LN229-P2L GBM cells show circadian rhythms in Per2 gene expression in vitro (mean±SEM, n reported in figure, all recordings had cosine fits with correlation coefficients of CC>0.9). (F) Treatment with a Cryptochrome agonist (KL001, 1μM) also blocked the time-of-day dependent growth response to DEX in vitro. In contrast, a Cryptochrome inhibitor (KS15, 1μM) increased cell growth to DEX at the trough, but not at the peak, of Per2 expression compared to vehicle (mean±SEM, n = 3 per group, two-way ANOVA with Šídák’s multiple comparisons test, *p < 0.05, **p < 0.01). See also Figure S7.

Figure 4:. Daily glucocorticoid signaling promotes GBM growth and accelerates disease progression

(A-B) Glucocorticoids promote cell growth in vitro in LN229 (A) and GL261 (B) WT, but not in GR KD cells (mean±SEM, n = 3 for GR KD GL261-P2L cells, n = 6 in all other groups. Two-way ANOVA with Bonferroni’s multiple comparisons test, ****p < 0.0001, ns p > 0.05). Cells treated with 100μM cortisol (LN229) or corticosterone (GL261) (Cort) grew an average of 4- to 2-fold compared to vehicle-treated or GR KD cultures. (C-D) Tumor size was higher in mice bearing LN229 (C) and GL261 (D) WT tumors (yellow) compared to GR KD (blue, mean±SEM, n reported in figure, two-way ANOVA with Šídák’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001. BLI counts on representative images are x10³, color bar depicts relative bioluminescence). (E-F) Probability of survival was higher in mice bearing LN229 (E) and GL261 (F) GR KD tumors compared to those implanted with WT tumors (n reported in figure, Log-rank Mantel-cox test, *p<0.05. Mice bearing LN229 (E) or GL261 (F) GR KD tumors lost less weight from start to the end of the experiment compared to mice implanted with WT tumors (mean±SEM, n reported in figure, Student’s t test, **p < 0.01, ****p < 0.0001). (G-H) Independent measurements of tumor size using a constitutive Ef1a-luc reporter in LN229 (G) and GL261 (H) tumors showed higher tumor bioluminescence in mice bearing WT tumors (yellow) compared to GR KD (blue, mean±SEM, n reported in figure, two-way ANOVA with Šídák’s multiple comparisons test, *p < 0.05, **p < 0.01). (I-J) Mice bearing LN299 (I) and GL261 (J)-Ef1a GR KD tumors lost less weight from start to the end of the experiment compared to mice implanted with WT tumors (mean±SEM, n reported in figure, Student’s t test, **p < 0.01). (K) NR3C1 (GR) mRNA expression was significantly higher in human IDH1 WT GBM samples compared to non-tumor histology (data obtained from The Cancer Genome Atlas Program (TCGA) database, n reported in figure, solid line indicates median, Wilcoxon rank sum test, ***p < 0.001). (L) Cox proportional hazard model on patients’ overall survival with NR3C1, with incorporation of known risk factors, showed a 60% increase in hazard of mortality for every one unit increase in NR3C1 expression in GBM (Wald test, abbreviations in table are defined as follows: hazard ratio (HR), overall survival (OS), confidence interval (CI), O6-methylguanine-DNA methyltransferase (MGMT). See also Figures S8 and S9.

Figure 5:. Disruption of circadian rhythms in the host slows GBM growth and disease progression

(A) Fecal Corticosterone (CORT) concentration was circadian in WT, but not VIP KO mice (mean±SEM, n = 3 per group, WT trace scored circadian by JTK cycle p < 0.05). (B) Tumor size was higher in WT mice bearing GL261 tumors compared to VIP KO mice (mean±SEM, n reported in figure, two-way ANOVA with Šídák’s multiple comparisons test, *p < 0.05, **p < 0.01. BLI counts on representative images are x10³, color bar depicts relative bioluminescence). (C) VIP KO mice bearing GL261 tumors lost less weight from start to the end of the experiment compared to WT mice (mean±SEM, n = 6 per group, Student’s t test, *p < 0.05). (D) GBM proliferation and tumor area, as measured by Ki67 expression, was higher in WT mice bearing GL261 tumors (images with yellow frame) compared to VIP KO mice (images with green frame). In violin plots, dotted lines indicate 25^th and 75^th percentile, solid line indicates 50^th percentile (median). (Scale bar = 1mm, n = 4 per group, Student’s t test, *p < 0.05, **p < 0.01). Composite images of DAPI to label cell nuclei (blue) and Ki67 (magenta) immunostaining of brain sections reveal tumor location and size.

Figure 6:. Peak timing of tumor Bmal1 and Per2 synchronizes to host rest-wake activity in different light cycles

(A) Schematic of light shifting paradigm after tumor implantation. Mice were housed in standard 12:12 light/dark conditions (LD), reversed 12:12 dark/light (DL), or constant darkness (DD). Light yellow background represents daytime and grey background represents nighttime. (B) Representative 36-hour in vivo bioluminescence imaging (top) and locomotor activity profile (bottom) of a mouse implanted with NF1^−/− DNp53-B1L cells, in a standard 12L:12D light schedule, two weeks post-implant, showed peak Bmal1 expression during the light phase (trace scored circadian by JTK cycle p < 0.05) and entrainment to the light cycle. Light yellow background represents daytime and grey background represents nighttime. (C) Average traces of 36-hour in vivo imaging of NF1^−/− DNp53-B1L (green line) and -P2L (yellow line) GBM tumors two weeks post-implant, in a standard 12L:12D light schedule, showed peak Bmal1 during the light phase and Per2 during the dark phase (mean±SEM, n reported in figure, average traces scored circadian by JTK cycle p < 0.05). Light yellow background represents daytime and grey background represents nighttime. (D) 36-hour in vivo imaging (top) and locomotor activity profile (bottom) of the same mouse after four weeks in a reversed 12L:12D light schedule, six weeks post-implant, showed Bmal1 synchronized to the new dark-light cycle, peaking during the light phase (trace scored circadian by JTK cycle p < 0.05). Light yellow background represents daytime and grey background represents nighttime. (E) Average traces of 36-hour in vivo imaging of NF1^−/− DNp53-B1L (green line) and -P2L (yellow line) GBM tumors after two weeks in a reversed 12L:12D light schedule (four weeks post-implant) showed tumor rhythms synchronized to the new light-dark schedule (mean±SEM, n reported in figure, average traces scored circadian by JTK cycle p < 0.05). Light yellow background represents daytime and grey background represents nighttime. (F) 36-hour in vivo imaging (top) and locomotor activity profile (bottom) of the same mouse after four weeks in constant darkness (eight weeks post-implant) showed Bmal1 peaking with the offset of daily locomotion (trace scored circadian by JTK cycle p < 0.05). Light grey background represents subjective daytime, and dark grey background represents subjective nighttime. (G) Average traces of 36-hour in vivo imaging of NF1^−/− DNp53-B1L (green line) and -P2L (yellow line) GBM tumors after two weeks in constant darkness (six weeks post-implant) showed peak expression of Per2 aligned with subjective dusk and Bmal1 with subjective dawn (mean±SEM, n reported in figure, average traces scored circadian by JTK cycle p < 0.05, Circadian Time 12, CT12, based on daily locomotor activity onset). Light grey background represents subjective daytime, and dark grey background represents subjective nighttime.

Figure 7:. Disruption of daily rhythms in the host desynchronizes tumor Per2 expression from the host rest-wake activity

(A) Representative 36-hour in vivo bioluminescence imaging (top) and locomotor activity profile (bottom) of a WT mouse implanted with GL261-P2L cells, two weeks post-implant and in constant darkness showed tumor Per2 expression peaking during the subjective dark phase (Circadian time (CT) 16, trace scored circadian by JTK cycle p < 0.05) and rhythmic locomotor activity starting at CT12. Light grey background represents subjective daytime, and dark grey background represents subjective nighttime. (B) Representative 36-hour in vivo bioluminescence imaging (top) and locomotor activity profile (bottom) of a VIP KO mouse implanted with GL261-P2L cells, two weeks post-implant and in constant darkness, showed tumor Per2 expression peaking during the subjective light phase (CT4, trace scored circadian by JTK cycle p < 0.05) and arrhythmic locomotor activity patterns. Light grey background represents subjective daytime, and dark grey background represents subjective nighttime. (C) Average trace of 36-hour in vivo imaging of WT mice (top) and average locomotor activity profiles (bottom) for all mice implanted with GL261-P2L tumors, two weeks post-implant and in constant darkness (subjective day = light grey and subjective night = dark grey background), showed reliable Per2 peak expression and rhythmic daily activity (mean±SEM, n reported in figure, average trace scored circadian by JTK cycle p < 0.05). (D) Average trace of 36-hour in vivo imaging of VIP KO mice (top) and average locomotor activity profiles (bottom) for all mice implanted with GL261-P2L tumors, two weeks post-implant and in constant darkness, showed desynchronized Per2 peak timing and arrhythmic daily activity (mean±SEM, n reported in figure, average trace not scored circadian by JTK cycle p > 0.05). Light grey background represents subjective daytime, and dark grey background represents subjective nighttime. See also Figure S10.

Figure 8:. Timing of daily Per2 expression in GBM xenografts depends on glucocorticoid receptor signaling

(A-B) Addition of 100nM DEX shifted circadian Per2 expression compared to vehicle in LN229 (A) and GL261 (B) GBM cells. Dashed line indicates time of DEX treatment, mean±SEM, n reported in figure, all recordings had cosine fits with correlation coefficients, CC > 0.9). (C-D) DEX did not phase shift Per2 expression in GR KD LN229 (C) and GL261 (D) GBM cells. Dashed line indicates time of DEX treatment (mean±SEM, n reported in figure, all recordings had cosine fits with correlation coefficients, CC > 0.9). (E-F) Representative images of in vivo tumor imaging during the light (ZT0–12, yellow bar) and dark (ZT12–24, grey bar) phases of LN229-P2L (E) and GL261-P2L (F), WT or GR KD cells (BLI counts on representative images are x10³, color bar depicts relative bioluminescence). (G-H) Daily Per2 profiles had high synchrony indices (i.e. peaked at similar times of day across mice) in LN229 (G) and GL261 (H) WT tumors. Synchrony was lower for tumors lacking GR (mean±SEM, n reports cohort number. Synchronization index was calculated as an average of all mice in one cohort. In LN229, n = 5 mice per one WT or GR KD cohort. In GL261, n = 12 mice in 3 WT cohorts, and n = 14 mice in 3 GR KD cohorts). See also Figures S11 and S12.