Despite advances in chemotherapy and radiation over the past 40 years, the outcome for children with diffuse intrinsic pontine gliomas (DIPGs) remains almost uniformly fatal, with survival of less than 12 months despite numerous trials of chemotherapy, targeted agents and radiation therapy1. These diffuse tumors are admixed with the normal neural structures of the pons, a portion of the brainstem that comprises several life-sustaining nuclei, precluding surgical resection, and they undergo only transient responses to radiation. Development of new therapies has been hampered by a paucity of surgical samples and model systems.

Recently, large genome-sequencing studies of pediatric high-grade gliomas have been carried out and have consistently identified a lysine to methionine (K27M) substitution in histones H3.1 and H3.3 in over 80% of midline high-grade gliomas and DIPGs2. Hemispheric high-grade gliomas of childhood commonly harbor a H3.3 Gly34 to valine mutation3. These mutations in H3 in DIPG lead to inactivation of the polycomb repressive complex 2 (PRC2) through an interaction between EZH2, the H3K27 methyltransferase of PRC2, and the mutant histone4,5,6. This results in global hypomethylation at H3K27 with consequent transcriptional derepression at these loci4. Restoration of physiological Lys27 methylation may therefore represent a new therapeutic strategy to treat H3K27M–mutated pediatric high-grade gliomas. In this issue of Nature Medicine, Hashizume et al.7 demonstrate that small-molecule inhibition of the H3K27 demethylase in patient-derived intrapontine xenografts of pediatric DIPGs harboring the H3K27M mutation results in a reversal of H3K27 demethylation and significantly prolongs survival in these models (Fig. 1).

Figure 1: The H3.1 K27M altered chromatin state is reversible by JMJD3 inhibition.
figure 1

Katie Vicari/Nature Publishing Group

Top, in DIPG, there is global hypomethylation of Lys27 of H3, which promotes a more accessible chromatin state characterized by H3K27 acetylation and aberrant gene expression. In DIPG, histone H3.1 or H3.3 harbors a K27M aberration. The mutant K27 histone inhibits PRC2, which is the major H3K27 methylase. Bottom, Hashizume et al.7 show that treatment of DIPG with GSKJ4, an inhibitor of the histone demethylase JMJD3, restores methylation at H3K27 toward the physiological state. Treatment with GSKJ4 promotes increased survival of animals with intrapontine DIPG xenografts with the H3K27M mutation.

Full size image

The authors studied a panel of patient-derived cell lines: two lines derived from patients with DIPG that harbor a heterozygous H3.3K27M mutation, two patient-derived pediatric glioblastoma lines wild type for H3 and a fifth pediatric glioblastoma line that carries the H3.3 G34V mutation7. Additional isogenic cell lines derived from normal human astrocytes (NHAs) with and without transgene expression of the H3.3 K27M mutation were also studied. Both patient-derived cell lines had global hypomethylation of H3K27 compared with the wild-type models. Treatment of K27M mutant cells in vitro with GSKJ4, a small-molecule inhibitor of the H3K27 demethylase JMJD3, resulted in increased K27me2 and K27me3 and a dose-dependent reduction in cell viability and proliferation. Notably, the therapy in this case was not targeting the mutant protein (H3.3 K27M) directly but rather inhibiting a wild-type enzyme in the methylation pathway, hence promoting the methylation of the nonmutant protein allele.

Treatment of two intrapontine H3K27M murine xenografts with GSKJ4 increased animal survival and reduced tumor growth but did not achieve the same effect in animals with xenografts containing wild-type H3.3 or the mutant H3 G34V (Fig. 1). Although the evidence that K27M mutations drive the initiation of DIPG is strong, it was uncertain whether or not continued expression of the mutant protein was necessary for tumor maintenance and, therefore, whether the mutant protein is a valid target for development of novel therapies. The efficacy of GSKJ4 in vivo suggests that H3K27M is indeed an attractive target for therapy.

Furthermore, the development of many potential brain tumor therapies has been halted due to their inability to penetrate the blood-brain barrier. Hashizume et al.7 confirmed the delivery of GSKJ4 into the pons using high-performance liquid chromatography analysis of non–tumor-bearing mice, thereby further supporting the use of GSKJ4 as a potentially clinically effective targeted therapy.

It may have been expected that other H3K27 demethylases such as UTX would compensate for the antitumor effects of GSKJ4 on JMJD3 through restoration of H3K27 demethylation8. Loss-of-function mutations in the histone lysine methylase EZH2 drive T cell acute lymphoblastic leukemia (T-ALL) development, suggesting a common mechanism of tumorigenesis with DIPG9,10. Treatment of T-ALL cells with GSKJ4 also resulted in reduced cell proliferation and increased K27M methylation8. In a recent study, inhibition of UTX in T-ALL cell lines and in K27M DIPG cell lines did not result in impaired proliferation, and genes downregulated with GSKJ4 overlapped significantly with genes upregulated with UTX knockdown8. This suggests that UTX and JMJD3 demethylate different targets and are not redundant H3K27 demethylases. Collectively, these reports highlight a previously unknown functional specificity of H3K27 demethylases and dictate that the correct demethylase must be selected for targeted therapy.

Given the canonical mutation (K27M) was discovered only 2 years ago, it is striking that Hashizume et al.7 have already identified GSKJ4 as a targeted therapeutic that penetrates the pons, have identified the molecular mechanism by which it acts and had success in preclinical models. There are no known effective chemotherapeutic agents for DIPG, so the bar for novel therapies is necessarily low. As such, this study strongly supports transition into clinical trials.

Almost all children with DIPG are currently treated on a palliative basis with upfront radiation therapy, which prolongs survival and improves symptoms. Children with DIPG are often enrolled in phase 1 clinical trials of novel agents combined with radiation. Regarding a phase 1 study of JMJD3 inhibition, we believe the ideal clinical trial would commence therapy at the time of diagnosis before tumor biology is altered by the radiotherapy.

It is also necessary to identify potential combinatorial agents to proactively counteract the development of resistance to single-agent small-molecule therapy. Evidence from Drosophila melanogaster constitutively expressing H3K27M indicates that H3K27 acetylation levels and bromodomain-containing protein 1 (BRD1) and bromodomain-containing protein 4 (BRD4) are increased in H3.3 K27M–containing nucleosomes, suggesting bromodomain inhibition could be a potential combinatorial strategy to pursue in preclinical models11. Although inhibition of PRC2 and loss of H3K27 methylation appear to drive the formation of DIPG and T-ALL, activation of PRC2 and hypermethylation of H3K27 may be driving the initiation of medulloblastoma, ependymoma and lymphoma, suggesting that agents driving hypermethylation of H3K27 should be used with caution12,13,14. JMJD3 inhibition has not yet been therapeutically tested in humans, so translation into clinical trials should proceed cautiously within the strict confines of a phase 1 safety and dose-escalation trial.

The past 2 years, comprising just two generations of patients with DIPG, have yielded a wealth of data that could allow our understanding of DIPG to progress from a black box to target identification and finally to clinical trials.