Stemming the growth of pediatric gliomas through histone modification

See the article by Vitanza et al. in this issue pp. 376–386.

A surprising theme that has emerged from sequencing the cancer genome is that mutations in genes encoding histone proteins are very common. Histones bind DNA to form bead-like structures (nucleosomes) along the DNA strand. At each nucleosome, the DNA double helix wraps around a complex of four pairs of core histone proteins (H2A, H2B, H3, and H4), spanned by a single linker protein (H1). Histones are no longer considered to be simply DNA-packaging proteins but instead are understood to be dynamic regulators of gene expression.¹ Covalent modifications of histones, such as addition and removal of methyl and acetyl groups on lysine residues, induce conformational changes in the DNA structure that cause the DNA to unwind or condense. Unwinding DNA creates permissive chromatin, where transcription factor complexes can move in and transcribe nearby genes. Conversely, tightening the DNA structure creates sites of repressive chromatin, where gene transcription is suppressed. The idea that tumor behavior is determined by such epigenetic modifications, in addition to mutations in the primary DNA sequence of cancer-causing genes, has increased the frustration of cancer biologists in their quest to understand the molecular origins of cancer. Nevertheless, adding histone modification to the seemingly endless complexity of cancer biology has revealed new molecular targets for therapeutic intervention.

One can scarcely imagine a group of patients in greater need of a treatment breakthrough than children with brainstem gliomas. Surgical resection is out of the question, chemotherapy has been ineffective, and radiotherapy offers a median overall survival of only 11 months.² The long-standing practice of avoiding diagnostic biopsy in these patients has limited our ability to understand the molecular biology of the disease through laboratory analysis of tissue specimens; however, genome sequencing of pediatric gliomas collected through international consortia has revealed a frequent mutation (K27M) that replaces lysine with methionine residues in histone protein H3.^3,4 In fact, this mutation is so prevalent that neuropathologists now consider gliomas arising in the brainstem and other midline brain regions, especially in children, as a single diagnostic entity, which they call diffuse midline glioma, H3 K27-mutant (DMG).

How the H3 K27M mutation promotes the growth of DMGs and other tumor types remain the subject of fervent scientific research. Removal of K27 from histone H3 and replacement with a methionine residue, which is resistant to methylation and acetylation, would likely alter the transcription of many genes that govern cell growth. Hope that histone modification could be exploited to treat DMGs came from a project to screen a library of compounds for growth suppression of DMG cells in culture.⁵ The most promising candidate was panobinostat, a member of a class of drugs that inhibit histone deacetylase (HDAC) enzymes, which catalyze the removal of lysine-bound acetyl groups from histone proteins. Natural products that inhibit HDACs were discovered in the 1970s, and their first clinical application was the use of trichostatin to treat fungal infection in 1990.⁶ Clinical trials of panobinostat in pediatric DMG patients are currently underway, but efficacy has not yet been determined.⁷

Recognizing the potential of HDAC inhibitors and anticipating shortcomings of ongoing panobinostat trials, Vitanza et al. tested various HDAC inhibitors for growth suppression in DMG cell lines, which they prepared from tumors explanted from previously untreated patients.⁸ They discovered that two compounds, quisinostat and romidepsin, potently induced apoptotic cell death in culture and suppressed the growth of xenografts implanted subcutaneously in mice. The investigators compared transcription profiles of DMG cells before and after drug treatment and observed that histone modification can induce widespread changes in gene expression.

Designing accurate methods for preclinical drug testing, as Vitanza and colleagues have done, is especially important for HDAC inhibitors. The human genome encodes 18 different HDACs, which are variably expressed in different tissues.⁶ The fact that HDACs can also deacetylate nonhistone proteins broadens the range of unpredictable side effects. Furthermore, acetylation of a lysine residue can change the methylation state of a different lysine on the same histone protein, further altering gene transcription.⁹

Future advancements in HDAC inhibitor therapy will require tackling some thorny questions. Would selective HDAC inhibitors be more effective than broad-spectrum drugs, like panobinostat, quisinostat, and romidepsin, that inhibit all HDACs? In general, increasing drug specificity reduces toxic side effects. Developing selective HDAC inhibitors is challenging because the structure of the catalytic active site is very similar across HDAC enzymes.⁶ How do amino acid substitutions at histone regulatory sites alter the response of tumor cells to HDAC inhibitors? The common H3 K27M mutation in DMG occurs at a methylation site and therefore seems unlikely to affect acetylation or the sensitivity of tumors to HDAC inhibitors.

Finally, the therapeutic efficacy of HDAC inhibitors is limited by poor blood-brain barrier penetration. Vitanza and coworkers report that after systemic administration, quisinostat and romidepsin did not enter DMG tumors implanted into the brain in an orthotopic mouse model.⁸ They speculated that conjugating HDAC inhibitors to molecules, like chlorotoxin, that shuttle compounds across the blood-brain barrier could circumvent the drug delivery problem. Despite these obstacles, histone modification gives us hope for finally making inroads into one of the most vexing problems in neuro-oncology.

Conflict of interest statement. None declared.

Authorship statement. The text is the sole product of the author and no third party had input or gave support to its writing.