Introduction

Diffuse midline glioma (DMG) is a malignant pediatric brain tumor that emerges in the midline region of the brain1. In the United States, the incidence rate of DMGs is ~0.8 in 100 000 and the median age of patients affected by it is between 6-7 years old2. Despite several efforts to combat the disease, DMGs remain incurable. Currently, median survival of DMG patients remains at <1 year post-diagnosis, with only 10% of patients exceeding 2 years of survival upon patient-specific therapies3.

Current standard-of-care for DMG patients includes radiation leading to prolonged survival3,4. In recent years, several therapeutic approaches have been developed to target the H3K27 lysine-to-methionine (H3K27M) alteration, a molecular characteristic defining 70–80% of DMGs5,6,7. H3K27M mutation leads to loss of H3K27-trimethyl (H3K27me3) and a global hypomethylation of DNA, thus affecting epigenetic regulation in DMGs8,9.

One such dysregulation is in the hypomethylation of O6-methyl-guanine-DNA methyltransferase (MGMT) promoter region10,11,12, rendering various alkylating agents ineffective on DMGs. Lack of MGMT promoter methylation and silencing, leads to its increased transcription, thus conferring resistance to drugs such as temozolomide (TMZ)12,13,14. TMZ is a DNA damaging alkylating agent that delivers methyl group to purine bases of DNA such as O6-guanine, N7-guanine and N3-adenine12. In the presence of MGMT, the methyl group on guanine bases are removed and transferred to cysteine residues, subsequently allowing tumor cell DNA to be repaired for continuous growth and proliferation15.

We set out to evaluate a functionally different chemotherapeutic agent, dianhydrogalactitol (DAG), also known as Val-083, on DMGs. Val-083 is a blood-brain-barrier (BBB) penetrant DNA targeting agent with a bi-functional unique structure16. It induces DNA N7-guanine interstrand crosslinks that is unrepairable by MGMT, subsequently leading to DNA double-strand breaks. DNA lesions triggered by interstrand crosslinks (ICL) are repaired by the Fanconia anaemia pathway that also intersects with other repair process such as homologous recombination and nucleotide excision repair17. In p53 mutant DMGs, we hypothesized that tumor cells with DNA damage will undergo cell cycle arrest at S/G2 phases for DNA repair to avoid genomic instability16. To circumvent DNA repair and propel cells with DNA damage into premature mitosis and apoptosis, we evaluated Val-083 treatment in combination with Wee1 inhibitor, AZD177518. AZD1775 inhibits Wee1 and prevents phosphorylation of CDK1 at the Y15 residue, thus hindering cell cycle arrest at the G2 phase18.

In previous studies, we demonstrated increased Wee1 expression in DMGs and assessed the efficacy of AZD1775 in combination with radiation on DMG preclinical models19. We subsequently tested this combination in Phase I clinical trials for children with newly diagnosed diffuse intrinsic pontine glioma (DIPG)20. While the combination was generally well tolerated by patients, it did not show a significant improvement in overall survival (OS) compared to historical cohort20. Based on these findings, we would like to investigate a novel therapeutic strategy involving Val-083 in combination with AZD1775 in our DMG preclinical models. In this study, we present compelling evidence for the efficacy and underlying mechanism behind the synergistic activity of Val-083 and AZD1775 combination on DMG preclinical models.

Results

MGMT and Wee1 gene expressions are upregulated in H3K27M DMGs

To validate the increased RNA expression of MGMT and Wee1 in H3K27M DMGs, we downloaded publicly available data from CCMA dataset and filtered for pediatric brain tumor cell lines and non-malignant cells21. Figure 1a shows that the MGMT expression is significantly higher in H3K27-DMGs compared to H3G34-diffuse hemispheric glioma (DHG) (p = 0.0114) and adult high-grade glioma (HGG) (p = 0.0026). This suggests that H3K27-DMGs are more prone to resistance towards alkylating chemotherapeutic agents such as temozolomide whose DNA damaging activity is easily repaired by MGMT. Hence, there is a higher need to test a functionally different chemotherapeutic agent such as Val-083 on H3K27-DMGs. In addition, to justify the use of Wee1 inhibitor, AZD1775, Wee1 gene expression was also determined in the same cohort of cell lines. Figure 1b shows that Wee1 expressions in H3K27-DMG along with adult HGG, ATRT, and H3WT-HGG are significantly higher (p < 0.0001) than in non-malignant cells, underscoring its potential as a promising target for combined therapy with Val-083.

Fig. 1: Efficacy and mechanism of action of Val-083 and AZD1775 on DMG cells.
figure 1

Gene expression analysis of CCMA datasets reveals significantly higher (a) MGMT expression in H3K27-DMG cell lines compared to adult HGG and H3G34-DHG cell lines, and (b) significantly higher WEE1 expression in adult HGG, ATRT, H3K27-DMG and H3WT-HGG cell lines compared to non-malignant cells. Kruskal-Wallis test with Dunn’s multiple comparisons test was conducted; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. c, d Dose response curve of Val-083 and AZD1775 on ten DMG/HGG cell lines. e IC50 range and median IC50 of Val-083 and AZD1775 on responsive cell lines based on their H3 status. f Val-083 induces DNA damage and DNA repair response through ATR-mediated signaling pathway. g AZD1775 inhibits Wee1 kinase to reduce inhibitory phosphorylation of CDK1 and CDK2 and induce cell cycle checkpoint impairment.

Full size image

Val-083 and AZD1775 reduces DMG cell viability in a dose-dependent manner

To evaluate the efficacy of Val-083 and AZD1775 as a single agent, we treated DMG and HGG patient-derived cell lines (Supp. Table 1) with H3.1K27M mutant (CNHDMG-1008, SF10693), H3.3K27M (HSJD-DIPG007, CNHDMG-762, SF8628, SF10423, CNHDMG-967) and H3 wildtype (WT) (SUDIPG-48, 7316-913, CNHDMG-760) statuses using varying concentrations of Val-083 and AZD1775 followed by cell viability assessment. Figure 1c shows the dose-response curve of Val-083 treated cell lines. The IC50 of Val-083 ranges between 0.84 and 258 µM with SF10423 showing the most sensitivity, and CNHDMG-762 showing the most resistance at the tested concentrations. We also evaluated the efficacy of AZD1775 as a single drug on the same panel of cell lines. Figure 1d shows the dose-response curve for the tested cell lines. The IC50 of AZD1775 ranges between 0.21 to 1.55 µM with SF10423 showing the most sensitivity and CNHDMG-760 showing the highest IC50 value

When comparing between cells with H3K27M mutation and H3WT statuses, (Fig. 1e), the IC50 of Val-083 ranges between 0.84 to 11.5 µM (median 3.52522 ± 4.06 µM) on H3.3 cells, 1.02 to 1.29 µM (median = 1.16 ± 0.13 µM) on H3.1 cells, and 0.45 to 258 µM (median = 0.49 ± 121 µM) on H3WT cells. As for AZD1775, the IC50 ranges between 0.21 to 1.03 µM (median = 0.47 ± 0.35 µM) on H3.3 cells, 0.30 to 0.67 µM (median = 0.49 ± 0.19 µM) on H3.1 cells, and 0.3 to 1.55 µM (median = 0.62 ± 0.53 µM) on H3WT cells. While the sensitivity of all the cells towards AZD1775 remains precisely similar, sensitivity towards Val-083 ranges widely in cells depending on their H3 status. One H3.3 mutant cell line, CNHDMG-762, did not respond to the highest concentration of Val-083 tested, while the IC50 of 7316-913 (H3WT) was higher than the others, at 258 µM.

Val-083 induces DNA double strand break and initiates DNA damage response through ATR/Chk1

We hypothesized that Val-083 induces DNA damage through interstrand crosslinks and will cause activation of ATM- and Rad3-Related (ATR)-mediated DNA damage response in DMGs. To identify the mechanism behind Val-083 activity, SF8628 cell line was used as a representative model as it portrayed the genetic architecture of a majority of DMGs harboring H3.3K27M mutation, p53 mutation, and unmethylated MGMT promoter12,19. Upon treatment with 2 µM, 4 µM and 6 µM doses of Val-083 for 24 h, cell lysates were probed for proteins, protein kinases and phospho-proteins mediating DNA damage checkpoints. As shown in Fig. 1f, phosphorylation of γH2AX was upregulated with increased concentration of Val-083. This further upregulated phospho-ATR and phospho-Chk1. Additionally, increased inhibitory phosphorylation of CDK1 at Thr-14 and Thr-15 was observed, indicating CDK1 inhibition. FANCD2 ubiquitination was also probed to assess FA pathway activation from ICL, but no difference between the treatment was observed (Supplementary Fig. 2b).

AZD1775 inhibits Wee1 kinase and activates CDK1 for G2/M checkpoint inhibition

AZD1775 is a Wee1 kinase inhibitor and potentially activates CDK2 or CDK1 to promote cell progression towards S and M phase of the cell cycle, respectively. To validate Wee1 kinase inhibition by AZD1775, SF8628 cells were treated with 0.1, 0.3 and 0.5 µM concentrations of AZD1775 for 24 h and cell lysates were immunoblotted. As shown in Fig. 1g, Wee1 expression was downregulated upon treatment with increased concentration of AZD1775. Additionally, activating phosphorylation (Thr160) of CDK2 decreased, and inhibitory phosphorylation of CDK1 (Thr14 and Thr15) decreased with treatment, suggesting inactivation of CDK2 and activation of CDK1. This suggests the gradual advancement of cells from the S phase of the cell cycle into the M phase after 24 h of treatment with AZD177522. As hypothesized, AZD1775 can progress cancer cells towards the mitotic phase upon treatment, making it crucial to be used in combination with Val-083.

Val-083 and AZD1775 combination is mostly additive on DMG cells

To evaluate the potential for synergistic activity between Val-083 and AZD1775 on p53 mutant DMGs, we tested this combination on three p53 mutant DMGs, CNHDMG-762, SF8628, and SF10693 (Fig. 2a–c) and three p53-WT cell lines, HSJD-DIPG007, CNHDMG-967, and 7316-913 (Fig. 2d−f). As the dose response chart and the ZIP synergy chart and synergy score (Fig. 2g) shows, AZD1775 and Val-083 combinations are synergistic on CNHDMG-762 (ZIP score: 21.43; most synergistic area score: 29.89) and additive on SF8628, (ZIP score: 4.64; most synergistic area score: 9.16) and SF10693 (ZIP score: 4.28; most synergistic area score: 6.43). While the combination was synergistic or additive on p53 mutant cells, it was also additive on the p53-WT cell lines, HSJD-DIPG007 (ZIP score: 4.38; most synergistic area score: 5.29), CNHDMG-967 (ZIP score: 0.26; most synergistic area score: 3.93), and 7316-913 (ZIP score: 8.91; most synergistic area score: 30.46), plausibly suggesting the efficacy of this drug combination on the cells regardless of p53 status. Figure 2g and Supplementary Fig. 1 also shows other mathematical models used to calculate the synergy score, with Bliss (drugs targeting different targets) and Loewe (drugs targeting the same target) both showing synergistic or additive scores on TP53 mutant and TP53-WT cells.

Fig. 2: Combination of Val-083 and AZD1775 is additive or synergistic on multiple DMG/HGG cells.
figure 2

Dose response curve and ZIP synergy scores reveal that Val-083 and AZD1775 combination is synergistic on (a) CNHDMG-762 and additive on (b) SF8628, (c) SF10693 (d) HSJD-DIPG-007, e CNHDMG-967 and (f) 7316-913. g ZIP, Bliss and Loewe synergy scores with the most synergistic area scores on the six DMG cell lines. h Western blot assay demonstrates that Val-083 and AZD1775 combination induces DNA damage (upregulated γ-H2AX), activates DNA damage response via ATR and Chk1, and reduces inhibitory phosphorylation of CDK1 and CDK2 initiated by Val-083.

Full size image

Val-083 and AZD1775 combination leads to DNA damage and perturbation in the cell cycle

To investigate the combined effects of Val-083 and AZD1775 on DMG cells, immunoblotting was performed to identify changes in DDR markers and cell cycle checkpoint markers. Figure 2h shows that the combination treatment induces γ-H2AX, ATR and Chk1 activation, as observed with Val-083 single treatment. A more prominent Chk1 activation is observed in the combo group compared to the Val-083 single treatment group. This is perhaps due to added effects of AZD1775, as Chk1 activation is also observed in its single treatment. As for Wee1 and phospho-CDK1, a reduced expression is observed upon combination treatment. Since Val-083 and AZD1775 have opposite effects on phospho-CDK1 expression, we noticed that in the combination group, phospho-CDK1 expression is similar to what was observed in AZD1775 single treatment group. As for phospho-CDK2 expression, there is a reduction in its expression, consistent with AZD1775 single treatment.

Next, flow cytometric analysis was performed to investigate the effects of 24- and 48 h treatment on cell cycle (Fig. 3a). Cell synchronization was confirmed upon serum starvation for 24 h (Supplementary Fig. 2a). After 24 h of treatment, no significant difference was observed between all groups at the SubG1 and G1 phases. In the S phase, a significantly higher number of cells were found at the S phase of Val-083 treatment group in comparison to the vehicle (p < 0.0001), AZD1775 (p < 0.0001) and combination (p = 0.0013) groups (Fig. 3c, Supp. Table 3). In addition, a significantly lower number of cells were found in the G2/M phase of cell cycle in the Val-083 (p < 0.0001), AZD1775 (p = 0.0004) and combination (p < 0.0001) treated group compared to the vehicle group (Supp. Table 3).

Fig. 3: Val-083 and AZD1775 combination leads to cell cycle perturbation and apoptosis.
figure 3

a 24 h and 48 h cell cycle assay reveal increased cell population at S and G2/M phases which is modified with the addition of AZD1775. b 24 h and 48 h apoptosis assay reveal increased cells in the early and late apoptotic phase upon Val-083 and AZD1775 single and combination treatment. c Percentage of cell population at SubG1, G1, S and G2/M phases after 24 h of treatment with vehicle, Val-083, AZD1775 and combination drugs. d Percentage of cell population at SubG1, G1, S and G2/M phases after 48 h of treatment with vehicle, Val-083, AZD1775 and combination drugs. e Percentage of total apoptotic cells is significantly increased upon AZD1775 and combination treatment at 24- and 48 h post-treatment, while Val-083 treatment is significant only after 48 h of treatments. f Western blot analysis of cleaved PARP shows that the combination treatment significantly increases cleaved PARP activation. All data are shown as mean ± standard deviation (n = 3). Statistically significant difference (Two-way ANOVA, Tukey’s post-hoc test) is shown as *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 or ns (not significant).

Full size image

At 48 h (Fig. 3d and Supp. Table 3), a comparison with the vehicle group showed that a significantly higher number of cells in SubG1 phase were found in the Val-083 (p = 0.0006) and combination (p = 0.0188) groups. Furthermore, significantly lower number of cells in G1 phase was found in the Val-083 (p < 0.0001), AZD1775 (p = 0.0003) and combination (p < 0.0001) groups compared to vehicle. AZD1775 also showed a significantly higher number of cells (p < 0.0001) compared to Val-083 and combination groups. At the S and G2/M phases, a significantly higher number of cells were found in the Val-083 (p < 0.0001) and combination (p < 0.0001) groups compared to vehicle. AZD1775 treatment led to significantly lower number of cells in S and G2/M phases (p < 0.0001) compared to Val-083 and combination groups.

Overall, combination of Val-083 and AZD1775 perturbs the 24 h and 48-h cell cycle by driving cells from G0/G1 phase to S and G2/M phases. These observations highlight the combined activity of Val-083 and AZD1775 in initiating S phase cycle arrest for DNA repair pathway, followed by CDK1 activation to disrupt the G2/M checkpoint for continuous cell progress towards the M phase. This overrides the completion of DNA repair and pushes cells with DNA damage to enter premature mitosis. Significant difference between Val-083 and combination groups was only seen at the S phase after 24 h of treatment, and no other distinct differences were observed in the G2/M phases of cell cycle. However, since cells in the G2 and M phases cannot be resolved to differentiate between cells undergoing DNA damage repair or apoptosis, we performed apoptosis assay to confirm premature mitosis and apoptotic activity induced by AZD1775 on Val-083 treated cells.

Val-083 and AZD1775 induces apoptosis

To investigate the effects of Val-083 and AZD1775 on apoptosis after 24 h and 48 h of treatment, Annexin V/7-AAD staining was performed on SF8628 cells and probed using flow cytometer (Fig. 3b, e). Figure 3e and Supp. Table 4 show that the number of cells undergoing apoptosis significantly increased in the AZD1775 (p = 0.0097) and combination (p = 0.0099) groups, in comparison to the vehicle group, after 24 h of treatment. 48 h treatment led to a significant increase (p < 0.0001) in the number of apoptotic cells in the Val-083, AZD1775 and combination groups. With combined treatment, the number of cells undergoing apoptosis significantly increased compared to Val-083 single treatment at 24 h (p = 0.0071) and 48 h (p < 0.0001), and AZD1775 single treatment at 48 h (p = 0.0006).

Western blot analysis assessing cleaved PARP and cleaved caspase 3 revealed no significant difference in cleaved caspase 3 expression in all samples at 24 and 48 h treatment (Supplementary Fig. 2c). However, cleaved PARP levels (Fig. 3f) were upregulated in the Val-083 treatment group after 48 h of treatment, and in the AZD1775 and combination group after 24- and 48 h treatment. Previously, Zhai and colleagues showed that Val-083-mediated cytotoxicity is due to replication-dependent DNA damage23. In our study we observed that Val-083 induced apoptosis after 48 h of treatment on SF8628 cells. These results suggest Val-083 induces DNA damage and apoptosis on cancer cells. As for AZD1775, it is an established inducer of apoptosis in cancers24,25.

Val-083 and AZD1775 reduce migration and invasion activity

Previous studies have investigated the role of Val-083 and AZD1775 as single drugs in inhibiting migration and invasion of glioblastoma and esophageal squamous cell carcinoma18,26. In this study, we explored the effects of these drugs as single agents and in combination, towards DMG cell migration and invasion. Figure 4a shows scratch assays conducted on SF8628 cells for 24 h and 48 h of treatment with Val-083 (4.5 µM) and AZD1775 (0.3 µM). Viability of cells were measured after treatment to ensure that the doses used were not significantly cytotoxic to the cells, and to ensure that the results will not be affected by cell death. A significantly lower percentage of wound recovery compared to vehicle was observed on the wells treated with AZD1775 (p = 0.0019) and combination (p = 0.0011) groups after 24 h and 48 h (p < 0.0001) of treatment (Fig. 4b, Supp. Table 5). When comparing the combination therapy with monotherapies, a significantly lower percentage of wound recovery was observed only in comparison to Val-083 (p < 0.0001) at 24- and 48 h post-treatment.

Fig. 4: Val-083 and AZD1775 combination inhibits migration and invasion of DMG cells.
figure 4

a 24 h and 48 h scratch assay image captured using Olympus EP50 microscope. Scale bar = 300 µm. b Plot shows that percentage of wound recovery in the combination group is significantly reduced in comparison to vehicle and Val-083 groups after 24 and 48 h of treatment. c Transwell inserts captured using EVOS M5000 after 24 h and 48 h invasion assay. Scale bar = 150 µm. d Combination treatment significantly inhibits invasion of cells when compared to vehicle, Val-083 and AZD1775 treated groups after 48 h of treatment. e Western blot analysis of MMP2 and VEGF shows that AZD1775 and combination treatment significantly reduces expression of these proteins. All plots are shown as mean ± standard deviation (n = 3). Statistically significant difference (One-way ANOVA or two-way ANOVA, Tukey’s post-hoc test) is shown as *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 or ns (not significant).

Full size image

Next, a Transwell invasion assay was conducted on Matrigel coated membranes (Fig. 4c) with 24 h and 48 h of treatment in serum free media. The number of invaded cells in the combination group were significantly lower than in the vehicle (p < 0.0001)), Val-083 (p = 0.0035) and AZD1775 groups (p = 0.0255) after 48 h of treatment (Fig. 4d, Supp. Table 6). No other significant differences were observed. The number of invaded cells in the AZD1775 group was also significantly lower than the vehicle group (p = 0.0154).

To investigate the mechanism behind the inhibitory effects of AZD1775 and the combination groups, immunoblotting was performed on conditioned media (concentrated 100 ×) of cells treated with Val-083 and AZD1775 as single drugs and in combination. Figure 4e and Supplementary Fig. 2c shows the expression of extracellular matrix metalloproteinase-2 (MMP-2), vascular endothelial growth factor (VEGF) and Ponceau staining blot to ensure equal loading of concentrated media. A downregulated MMP-2 expression is observed in the AZD1775 and combination group after 24 and 48 h of treatment. As for VEGF, a lower expression is observed in the combination group after 24 h and 48 h and in the AZD1775 group after 48 h of treatment. These results are consistent with the activity of AZD1775 in ESCC cells as previously published18. Although Val-083 has been shown to decrease migration, invasion and VEGF expression in glioblastoma cells26, this effect was not observed in the SF8628 cells.

Val-083 and AZD1775 reduce tumor growth and migration in zebrafish models

To test the efficacy of Val-083 and AZD1775 on zebrafish tumor models, luci-mCherry tagged- SF8628 xenograft zebrafish models were developed (Fig. 5a). Supplementary Figure 3 shows that the tumor successfully engrafted at midbrain or pericardial cavity by day 1 post-injection. These models were treated with solvent control (DMSO), 40 µM Val-083 (maximum tolerated concentration, MTC), 50 µM AZD1775 (MTC) and a combination of both drugs. Optimal doses of Val-083 and AZD1775 treatment were obtained from toxicity assessment on zebrafish models (Supplementary Fig. 4).

Fig. 5: Val-083 and AZD1775 reduces tumor growth and migration in zebrafish models.
figure 5

a Schematic figure of treatment plans for SF8628 zebrafish xenograft models. b SF8628 zebrafish xenograft (midbrain) models at 4 days post injection (DPI) treated with vehicle, Val-083, AZD1775 and combination (combo) treatments. Scale bar = 750 µm. c SF8628 pericardial cavity tumor models of zebrafish at 4 DPI, treated with vehicle, Val-083, AZD1775 and combo treatments. Scale bar = 750 µm. d Migration of SF8628 cells to the tail region of zebrafish models. Scale bar = 750 µm. e Val-083 and AZD1775 as a single drug or as a combination shows significantly reduced tumor growth in the midbrain region, f pericardial cavity and (g) reduced migration of tumor cells towards the tail region after 3 days of treatment. All plots are shown as mean ± standard deviation (n = 10 for midbrain; n = 10 for pericardial cavity). Statistically significant difference (One-way ANOVA, Tukey’s post-hoc test) is shown as **p < 0.01; ****p < 0.0001.

Full size image

Treatment was commenced 1 day post-injection (DPI), and the results show that at 4 DPI, the relative fluorescence intensity of tumors treated with Val-083, AZD1775 and combination drugs is significantly lower (p < 0.0001) than those treated with vehicle (Fig. 5b, e, and Supplementary Table 7). When comparing combination group with monotherapies, a significantly higher tumor intensity was observed in Val-083 and AZD1775 (p < 0.0001). A similar injection was conducted at the pericardial cavity of zebrafish (Fig. 5c, f and Supplementary Table 7), and the tumor intensity is significantly lower (p < 0.0001) in all treatment groups when compared to the vehicle treated group. In comparison to the combined therapy, the tumor intensity was higher in Val-083 (p = 0.0058) and AZD1775 (p = 0.0081) groups. Reduced migration of tumor cells due to regression in tumor growth, was also observed in the tail region at 4 DPI upon single and combination treatments (p < 0.0001) (Fig. 5d, g, and Supplementary Table 7). The number of migrated cells was also significantly lower in the combination group compared to Val-083 (p < 0.0001) and AZD1775 (0.0061) groups.

Val-083 and AZD1775 increase apoptotic activity in vivo and prolongs survival of nude mice

SF8628 mouse xenograft models were used to test the efficacy of Val-083 and AZD1775 on the tumor models (Fig. 6a). Figure 6b, c show that the median survival of animals in the Val-083 (p = 0.0004) and combination (p = 0.0001) groups was longer than the animals in the control groups. No significant difference is observed between AZD1775 and control. The median survival of Val-083 group is also significantly longer than the AZD1775 (p = 0.0101) group. When comparing single and combination treatment groups, the median survival of the combination group is significantly longer than AZD1775 (p = 0.0001) and Val-083 (p = 0.0401) groups.

Fig. 6: Val-083 and AZD1775 lead to apoptosis of tumor cells and prolonged survival of mouse xenograft models.
figure 6

a Schematic figure of SF8628 mice xenograft model treatment plans. b Survival curve of SF8628 mouse orthotopic tumor models treated with vehicle control, AZD1775, Val-083 and combination of both drugs. c Table indicates median survival of animals in different treatment groups. The combination treatment significantly extended survival of the animals in comparison to vehicle control and single treatments. d Images of H&E and IHC staining of tumor regions from harvested mouse brain of different treatment groups, captured using Olympus VS200. Scale bar = 100 µm. e IHC staining of cleaved caspase 3 shows that the combination treatment leads to significantly increased number of cells undergoing apoptosis in the Val-083, AZD1775 and combination groups. IHC staining of (f) Ki67 and (g) VEGF) shows no significant difference between treatment groups. All plots are shown as mean ± standard deviation (n = 10 for survival curve; n = 3 for IHC). Statistically significant difference (One-way ANOVA, Tukey’s post-hoc test) is shown as *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001.

Full size image

Furthermore, H&E staining was conducted to confirm tumor engraftment at the midbrain region (Fig. 6d) and the tumor sections were stained with cleaved caspase-3, Ki-67 and VEGF. A notable increase in cleaved caspase-3 expression was observed in the Val-083 (p = 0.04), AZD1775 (p = 0.0137) and combination (p = 0.0002) groups in comparison to the vehicle control group (Fig. 6e and Supp. Table 8). Cleaved caspase-3 expression was also significantly higher with combination treatment compared to Val-083 (p = 0.0084) and AZD1775 monotherapies (p = 0.0237). However, no significant differences were observed in the Ki-67 (Fig. 3f) and VEGF (Fig. 3g) expression, contrary to what was observed in vitro. This discrepancy could be attributed to staining being performed on tumors collected at the endpoint of the study, when the tumors were most aggressive, potentially diminishing the differences between treatment groups.

Discussion

Despite extensive research into chemotherapeutics and novel agents for treating DMGs, more than 250 clinical trials have failed to produce effective therapies for this aggressive disease27. These challenges are triggered by inherent and acquired chemoresistance driven by genetic and epigenetic alterations such as a mutation in the p53 gene and hypomethylation of DNA12,28. The p53 tumor suppressor gene mediates DNA repair, cell cycle arrest or apoptosis upon DNA damage induced by chemotherapeutic agents29,30. When p53 is mutated, cells with DNA damage continue to proliferate with mutations leading to high genomic instability31. Furthermore, global DNA hypomethylation, particularly at key promoters like MGMT, contribute to resistance to alkylating agents such as TMZ12. These factors highlight the need for novel approaches that can circumvent these resistance mechanisms.

In this study, we explored the therapeutic effects of a distinct DNA damaging agent, Val-083, which was designed to evade MGMT-mediated chemoresistance23. Val-083 has been previously shown to overcome TMZ resistance in glioblastoma models independent of their MGMT promoter methylation status32,33. It was granted orphan drug status by the US FDA, and was demonstrated to be safe when used in combination with radiotherapy on newly diagnosed GBM patients (NCT03050736)34. Our findings confirm that Val-083 is effective in both in vitro and in vivo DMG models. However, recent preliminary topline results of Val-083 Phase 2/3 clinical trials (NCT03970447) for glioblastoma revealed its lack of efficacy in outperforming standard-of-care therapy35, which echoes a common challenge in translating preclinical performance into clinical benefits. These results highlight the need to combine Val-083 with agents that target complementary pathways to achieve robust and sustained anti-tumor effects on aggressive tumors such as DMGs.

To address this, we investigated the combined effects of Val-083 with AZD1775, a Wee1 kinase inhibitor that overrides G2 checkpoint to drive cancer cells with DNA damage into premature mitosis and cell death18,24,25,36. We hypothesized that AZD1775 would work synergistically with drugs that induce DNA damage and cause replication stress, such as Val-083. Our results confirm this hypothesis, demonstrating that AZD1775 enhances the DNA damage caused by Val-083, leading to significant tumor growth inhibition in vitro and in vivo. Mechanistically, the combination induces S-phase cell cycle arrest via activation of the ATR-Chk1 pathway, followed by CDK1 activation, which drives cells into mitosis before DNA repair is complete.

Interestingly, AZD1775 monotherapy was superior to Val-083 in inducing apoptosis, reducing migration and invasion, and inhibiting tumor growth in zebrafish models. Prior studies have shown that AZD1775 monotherapy promotes DNA damage and elevates γH2AX expression37, further inducing apoptosis in many different cancer cell lines24. Furthermore, Wee1 inhibition on its own, has been found to suppress DNA damage repair and promote replication stress38,39, thus questioning the need to use Val-083 in combination. However, in our mice models, AZD1775 as a single agent did not show significant prolongation of survival. In line with this, AZD1775 in combination with cranial radiation therapy was well tolerated in children with newly diagnosed DIPG but did not improve overall survival compared to historical controls20. Additionally, AZD1775 monotherapy in Phase 2 clinical trials for SETD2-deficient advanced solid tumors (NCT03284385), failed to provide substantial clinical benefits, emphasizing the need for combination approaches with DNA damaging agents40.

Taken together, Val-083 and AZD1775 worked additively at specific concentrations to inhibit tumor growth in vitro regardless of p53 status. The combined treatment resulted in DNA damage and S phase cycle arrest to initiate HR repair pathway, followed by CDK1 activation to advance cancer cells through the G2 checkpoint. This overrode the completion of HR repair and pushes cells with DNA damage for entry into premature mitosis and apoptosis. Additionally, Val-083 and AZD1775 combination significantly reduced migration and invasion activity on our DMG cell lines, supporting its potential to limit tumor progression and metastasis. While we focused primarily on the SF8628 cell line, which harbors H3K27M and p53 mutations with an unmethylated MGMT promoter, future studies should explore the mechanism behind these drug combinations across a broader range of DMG subtypes, including those with wild-type p53 status.

Our results highlight the need for future research to further optimize this combination therapy. A more extensive analysis of DMG models, including those with varying genetic backgrounds, will be critical for understanding the broader applicability of this treatment. Additionally, expanding sample sizes in both preclinical models and controlled pharmacodynamic studies will provide stronger evidence for clinical translation. Lastly, long-term studies are needed to evaluate whether DMGs would develop resistance to this combination therapy, and to determine how it might be integrated with other standard-of-care treatments.

In conclusion, we have shown that the combination of Val-083 and AZD1775 offers a promising therapeutic strategy for treating DMGs. This combination targets multiple resistance pathways, induces significant tumor growth inhibition, and extends survival in preclinical models. While further studies are warranted, this combination represents a potential new avenue for treating this devastating pediatric cancer.

Methods

Cell culture and investigational agents

All primary human cell cultures (CNHDMG-762, CNHDMG-1008, SF8628, SF10693, SF10423, CNHDMG-967, CNHDMG-760) were generated from biopsy or autopsy samples collected in accordance with the Declaration of Helsinki, with informed consent and in compliance with the Institutional Review Board (IRB) of the Children’s National Hospital (CNMC) in Washington, DC (IRB protocols, #1339) and IRB of the University of California, San Francisco (UCSF) (CC#170817). Establishment of glioma cells from surgical specimens and post-mortem tissue was performed as previously described41,42. SF8628 with stable expression of firefly luciferase for intracranial xenografting was obtained from the Brain Tumor Research Center (BTRC) Tissue Bank at University of California, San Francisco (UCSF)19. SUDIPG-48 and HSJD-DIPG007 cells were kindly provided by Dr. Michelle Monje at the Stanford University, and Dr. Angel Montero Carcaboso at the Hospital Sant Joan de Deu, Barcelona. 7316-913, a high-grade glioma cell line, was obtained from Children’s Brain Tumor Network. All the cells were maintained as neurospheres in serum free media as previously reported41, while SF8628 and SF10693 were grown as adherent monolayers in Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine serum and 1% penicillin-streptomycin. Further details on the cell lines can be found from Supp. Table 1. Val-083 (HY-16513) for in vitro testing and AZD1775 (HY-10993) were purchased from MedChemExpress, NJ, USA. Val-083 for in vivo testing on mice models was obtained from Kintara Therapeutics.

Database

MGMT and Wee1 mRNA expression in pediatric CNS cell lines and normal cell lines (Childhood Cancer Model Atlas) were queried using the following database: https://ccma.shinyapps.io/ccma/. Data from cell lines included adult high-grade glioma (HGG) (n = 16), atypical teratoid rhabdoid tumor (ATRT) (n = 19), diffuse hemispheric glioma, H3G34-mutant (DHG) (n = 6), H3K27-DMG (n = 51), H3-Wild type (WT) HGG (n = 19), medulloblastoma (n = 7), and non-malignant (n = 37). All data are displayed as log2 counts per million [log2 (cpm + 1)]. Analyses were performed using GraphPad Prism 10 (LaJolla, CA).

Cell viability assay and synergy evaluation

2500 cells/well (for adherent cells) and 5000 cells/well (for neurospheres) were seeded in 96-well plates overnight in 100 µL media. After 24 h of incubation, cells were treated with different concentrations of Val-083 and/or AZD1775 for 72 h. Cell viability was determined using CellTiter-Glo 2.0 for adherent cells and CellTiter-Glo 3D for neurospheres (Promega, Fitchburg, WI, USA). Measurements were normalized to DMSO control. IC50 concentrations were analyzed using non-linear regression [log (inhibitor) vs. response—variable slope (four parameters)] in GraphPad Prism 10 (LaJolla, CA). For drug combination studies, SynergyFinder (version 3.0) was used to assess response due to drug interactions. Zero Interaction Potency (ZIP), Bliss, and Loewe scores were determined to evaluate drug-drug interactions with a score of <-10 indicating antagonism, score of -10 to 10 indicating additive effect, and score of >10 indicating synergism43.

Immunoblotting analysis

SF8628 cells were seeded overnight in T75 flasks before being treated with indicated doses of Val-083, AZD1775 and DMSO. After 24 h of treatment, cells were washed with ice-cold PBS and lysed using RIPA buffer (Millipore Sigma, MA, USA) supplemented with proteinase (Roche) and phosSTOP phosphatase inhibitor cocktail (Roche). Protein lysates were quantified using Pierce™ BCA Protein Assay Kits (ThermoFisher Scientific, MA, USA) and separated by SDS-PAGE. In the case of migratory and invasion markers in conditioned media, cells were first treated in serum-free media before the media was collected after 24- and 48 h treatment. Media was then concentrated 100 × using Vivaspin 500 and loaded at equal volumes with Laemmli buffer for SDS-Page. Ponceau stain was performed to ensure equal loading of samples before blocking. Immunoblotting was done using indicated primary antibodies overnight before being probed with horseradish peroxidase (HRP)-conjugated anti-mouse or anti-rabbit IgG antibodies. Blots were probed with Clarity Western ECL substrate and imaged using ChemiDocTM Touch Imaging system (BioRad, CA, USA). Antibodies used in this study are listed in Supp. Table 2.

Cell cycle analysis

60,000 SF8628 cells were seeded into 6-well plates overnight. The next day, media with serum was removed and replaced with serum-free media to initiate cell synchronization. After 24 h, cells were treated with 0.05% DMSO as vehicle control, and Val-083 (4.5 µM) and AZD1775 (0.3 µM) at their 72 h IC50 concentrations. Drugs were given as single treatments or in combination in media with serum. After 24 h of treatment, cells were harvested and fixed with ethanol. Fixed cells were treated with FxCycle PI/RNAse solution according to manufacturer’s protocol (ThermoFisher Scientific, MA, USA) and incubated at room temperature for 15−30 min. Stained cells were run through MACS Quant Analyzer (Militenyi Biotech, MD, USA) and the results were analyzed using FlowJo (Becton, Dickinson and Company, NJ, USA).

Apoptosis assay

SF8628 cells were treated as described above, harvested and stained using FITC Annexin V Apoptosis Detection Kit with 7-AAD (BioLegend, CA, USA) according to manufacturer’s protocol. Cells were washed twice with Stain Buffer (BD Pharmingen, NJ, USA) and resuspended in Annexin V Binding Buffer at a concentration of 2 × 106 cells/mL. One hundred µL of cell suspension was then incubated with five µL of FITC Annexin V and 7-AAD viability staining solution for 15 min at room temperature in the dark. Four hundred µL of Annexin V Binding Buffer was added to each tube before flow cytometry analysis. CytoFLEX Flow Cytometer was used to run samples and CytExpert software was used for analysis (Beckman Coulter, CA, USA). The proportion of cells undergoing apoptosis was determined.

Migration assay

Scratch assay was conducted to investigate the effects of the treatments on SF8628 cell migration. 200 000 cells were seeded in 12-well plates and incubated overnight at 37 °C in 5% CO2 for 24 h to allow monolayer formation with 80% confluence. The next day a sterile p-200 pipette tip was used to instill wounds of similar size to the monolayer. Cells were washed with 1× PBS twice to remove cell debris before treatment with 0.05% DMSO as vehicle control, and Val-083 (4.5 µM) and AZD1775 (0.3 µM). Drugs were given as single treatments or in combination in serum free media. They were incubated for 24 and 48 h at 37 °C. The images of the wound area before and after treatment were captured using Olympus EP50 microscope and the migrated area was measured using the ImageJ software (National Institutes of Health, MD, USA). Percentage area of wound recovery was determined.

Transwell invasion assay

Corning® BioCoat® control inserts with 8.0 µm PET membrane were coated with 100 µL of 300 µg/ml of Corning® Matrigel® hESC-Qualified matrix (Corning, NY, USA). The plate with inserts were left at 37 degrees Celsius for 2 h to allow solidification of Matrigel. Then, SF8628 cells were trypsinized, and ~1.5 × 105 cells/insert were seeded into each insert in 500 µL of serum-free media. The cells in each insert were then treated with Val-083 (4.5 µM) and AZD1775 (0.3 µM). Media with 10% (v/v) FBS were added to the receiving well as a chemo-attractant before being incubated at 37 °C. After 24 and 48 h, cotton swabs were used to gently remove the cells in the upper insert and the inserts were rinsed twice with 1X PBS. Then, the invading cells on the underside of the membrane were fixed in 70% ethanol for 10 min before staining with DAPI (1 µg/ml) for 10 min in the dark. The inserts were washed with PBS, air-dried, and imaged using EVOS M5000. The number of invaded cells in four random fields of each insert was counted using ImageJ (NIH, MD, USA).

Zebrafish tumor models

Zebrafish studies were conducted in accordance with NIH guidelines for the care and use of laboratory animals and were approved by the Georgetown University Institutional Animal Care and Use Committee (approved protocol no: 2017-0078). Zebrafish husbandry, injections, and mounting were performed at the Georgetown-Lombardi Animal Shared Resource. Wild type (WT) zebrafish (Tübingen) were obtained from Zebrafish International Resource Center, University of Oregon, and housed under standard conditions i.e., temperature, 28 °C; pH 7.2–7.4; 14:10-hr light-dark cycle. Embryos were collected from natural breeding, and healthy fertilized embryos were sorted under a stereomicroscope at the blastula stage (6-hr post-fertilization, hpf) according to the stages of embryonic development as previously described44, placed (50 embryos/dish) on separate 10 cm petri dishes (NuncTM, ThermoFisher, MA, USA), and incubated at 28.5 °C.

2 days post-fertilization (dpf) zebrafish embryos were manually dechorionated using Dumont #5 fine forceps (Fine Science Tools, CA, USA) and anesthetized with 0.6 mM tricaine. Anesthetized embryos were aligned properly inside 1% low melting point agarose on a petri dish. 1 million mCherry expressing SF8628 concentrated cells were then loaded into pulled glass capillaries (Narishige GD-1) using a Narishige PC-10 pipette puller. The injections were performed in the midbrain region of the embryos using a Narishige IM-31 microinjector (Narishige International, NY, USA). The injection volume was carefully calibrated to deliver ~80–100 cells per injection. After the injection, the xenografted zebrafish embryos were released from the agarose gel and kept in a 28.5 °C incubator for 1 h to recover. After 1 h the embryos were imaged with Leica M205 FA fluorescence stereomicroscope (Leica, Germany) and maintained in E3 media at 31.4 °C.

For drug testing, one day post injection, the xenografted embryos were exposed to different drug solutions for 72 h. After 4 days post injection, the embryos were fixed in 4% paraformaldehyde and imaged with Leica SP8 confocal microscope (Leica, Germany). Fluorescent intensity and the total migrated areas were quantified using ImageJ software (National Institutes of Health, MD, USA)45. Optimal doses of Val-083 and AZD1775 for treatments were obtained from toxicity assessment on zebrafish models (Supp. Method).

Mouse tumor model

All the animal studies were conducted with approval from the UCSF Institutional Animal Care and Use Committee (IACUC). Five and six-week-old female athymic nude mice (Nu/Nu) were purchased from Envigo, housed in cages (4–5 per cage) under aseptic conditions and allowed to acclimatize for 1 week prior to the start of the experiment Luciferase-modified SF8628 tumor cells (1.0 × 105 in 1 μL) were implanted into the pontine tegmentum of mice. Bioluminescence imaging (BLI) was performed with Xenogen IVIS Lumina System using Living Image software to monitor tumor growth in vivo. Once tumor engraftment was confirmed, the animals were randomized into four groups (n = 10) based on body weight measurements and BLI: vehicle control, Val-083, AZD1775 and combination of both drugs. Sample size was determined by performing a 1-way ANOVA pairwise comparison. 36 mice (plus 4) were implanted to cover casualties that might occur during the intracranial tumor transplantation by stereotactic surgeries, or unsuccessful tumor implants. This sample size has a power of 80% to detect a 0.39 difference in proportions between untreated and treated groups with a significance level of 0.05.

Exclusion of any mice after tumor injections were conducted if the animal showed symptoms due to neurological impairments such as constant hunched posture, tilted head, disorientation or twitching. These animals were euthanized using carbon dioxide administration in compliance with the IACUC protocol. Randomized animals were treated with 60 mg/kg AZD1775 daily by oral administration (Mon-Fri), 3 mg/kg Val-083 per day by intraperitoneal injection (IP) every other day (Mon-Wed-Fri) and continued until they reached endpoint. They were monitored daily for development of symptoms related to tumor growth or treatment, and euthanized once they exhibited symptoms indicative of significant neurological impairment. Brain samples were harvested and fixed for further analysis. Brain samples collected from nude mice were fixed in formalin, embedded with paraffin, sectioned on glass slides and then stained with hematoxylin and eosin by the Department of Pathology, Children’s National Hospital. The Kaplan-Meier estimator was used to generate survival curves and statistical differences were evaluated using log-rank test (GraphPad Software, CA, USA).

Immunohistochemistry

Mounted slides of FFPE brain sections were deparaffinized and rehydrated before antigen retrieval using eBioscience™ IHC Antigen Retrieval Solution (Invitrogen, Waltham, MA). Then, slides were permeabilized with 0.3% Triton X-100 (Sigma Aldrich, St. Louis, MO) for 20 min before being blocked with 5% normal goat serum for 1 h. The tissue sections were subsequently incubated with primary antibodies against cleaved-caspase 3, Ki67 and VEGF (Supp. Table 2). The next day, slides were washed, incubated with anti-mouse/anti-rabbit Biotin-conjugated secondary antibodies (1 h), before quenching endogenous peroxidase activity with 0.3% hydrogen peroxide solution in methanol (30 min). The slides were subsequently stained with VECTASTAIN Elite ABC Reagent (Vector Laboratories, Newark, CA) for 30 min. Diaminobenzidine (DAB) color development was then visualized using SignalStain® DAB Substrate Kit (Cell Signaling, Denvers, MA). After that, sections were counterstained with hematoxylin, rehydrated and mounted with cover slip. The percentage of positively stained cells for cleaved caspase 3, Ki67 and VEGF were determined using QuPath (version 0.3.2).

Statistical analyses

All the data are presented as mean ± s.d and were presented in at least three biological replicates. GraphPad PRISM software, version 10 (San Diego, CA, USA) was used for statistical analysis with significance being set at p < 0.05. Normality of distribution was first checked using Gaussian distribution and D’Agostino-Pearson, Anderson-Darling, Shapiro-Wilk, Kolmogorov-Smirnov methods. Once normality test was passed, one-way ANOVA or two-way ANOVA (grouped data) with Tukey’s post-hoc test was utilized. If the normality test failed, the Kruskal-Wallis model with Dunnett’s post-hoc test was performed.