Independent researchers have identified mutations in the histone H3.3-coding gene H3F3A in a large proportion of pediatric cases of gliobastoma multiforme (GBM) and diffuse intrinsic pontine glioma (DIPG), an aggressive brainstem astrocytic tumor that occurs almost exclusively in children. An international consortium led by researchers at McGill University sequenced the exomes of 48 pediatric GBM samples and found somatic mutations in the H3.3-ATRX-DAXX chromatin remodeling pathway in 44% of tumors and recurrent H3F3A mutations in 31% of tumors.
Researchers for the St. Jude Children’s Research Hospital–Washington University Pediatric Cancer Genome Project simultaneously reported data from exon sequencing studies demonstrating that 78% of DIPGs and 22% of nonbrainstem pediatric glioblastomas (non-BS-PGs) harbor somatic mutations in H3F3A or in the related gene HIST1H3B, which encodes H3.1.
Both teams suggest their respective works are the first to identify somatic mutations in histone H3. Reporting in Nature the McGill investigators suggest their findings “provide a rationale for targeting the chromatin remodeling machinery” in pediatric GBM. Their published paper is titled “Driver mutations in histone H3.3 and chromatin remodeling genes in pediatric glioblastoma.” The St. Jude-Washington University team reports its findings in Nature Genetics in a paper titled “Somatic histone H3 alterations in pediatric diffuse intrinsic pontine gliomas and nonbrainstem glioblastomas.”
GBM occurs far less often in children than in adults, Although primary GBM tumors in pediatric cases appear morphologically identical to those in adult patients, children with the disease have “dismal outcomes” in comparison with their adult counterparts even after aggressive therapy, the McGill team notes. Moreover, while previous work on adult GBM has identified transcriptome-based subtypes of the disease and led to molecular subclassification, pediatric GBM remains more of an enigma. This is despite experimental data demonstrating distinct molecular subsets of childhood GBM and indicating that the disease harbors different genetic alterations compared with adult disease.
McGill’s Nado Jabado, M.D., and colleagues carried out whole-exome sequencing (WES) on 48 pediatric GBMs including six for which there was matched nontumor (germline) DNA. Four of these latter six samples harbored one of two heterozygous single-nucleotide variants in H3F3A. One changed lysine 27 to methionine (K27M), and the other changed glycine 34 to arginine (G34R). Both mutations are sited at or very near the amino-terminal tail of the protein, which undergoes post-translational modifications associated with either transcriptional repression (K27) or activation (K36), the team states. All four of the samples also harbored mutations in ATRX, which encodes part of the transcription/chromatin remodeling complex involved in binding H3.3 to DNA.
The team then extended their whole-exome sequencing study to an additional 42 tumor samples. They found that 15 of these harbored heterozygous H3.3 mutations (either K27M, G34R or G34V) and 14 had a mutation in ATRX that was associated with a lack of detectable ATRX protein. Two of the samples displayed heterozygous mutations in DAXX, which codes for a protein that heterodimerizes with ATRX and is involved in H3.3 recruitment to DNA. Overall, 21 of 48 samples (44%) had a mutation in at least one of these three genes.
Because the three genes weren’t sequenced as part of the 600-gene The Cancer Genome Atlas (TCGA) Glioblastoma project, and mutations in H3F3A, ATRX, or DAXX weren’t identified in another adult GBM sequencing study, the researchers wanted to see whether H3F3A mutations were indeed specific to GBM and/or pediatric cases. To this end they sequenced the H3F3A gene in 784 glioma samples of different grades and histological diagnoses and from patients of all ages.
They found that H3.3 mutations were highly specific to GBM and were much more prevalent in pediatric cases (36%) than in young adults (3%). K27M-H3.3 mutations occurred mainly in younger patients (median age 11 years, range 5–29) and thalamic GBM, whereas G34R- or G34V-H3.3 mutations occurred in older patients (median age 20 years, range 9–42). A comparison of the dataset with adult GBM databases suggested there was only a limited overlap between the frequently mutated genes in pediatric GBM and any of the four described adult GBM subtypes.
Interestingly, immunostaining assays on samples from 124 pediatric GBM patients couldn’t find ATRX in 35% of cases or DAXX in 6% of cases. “Overall, 37% of samples had lost nuclear expression of either factor,” the investigators write. Moreover, unsupervised hierarchical clustering of gene expression from 27 of the whole-exome sequencing cohort samples revealed what they state is a clear separation in the expression of K27M versus G34R/V mutant samples.
“Our data indicate a central role of H3.3/ATRX-DAXX perturbation in pediatric GBM,” the team concludes. “Mutant H3.3 recruitment would occur across the genome and induce abnormal patterns of chromatin remodeling to yield distinct gene-expression profiles for the K27 and G34 mutations. Additional loss of ATRX may act to reduce H3.3 incorporation at a subset of genes important in oncogenesis, preventing mutant H3.3 from altering their transcription. ATRX loss will also impair H3.3 loading at telomeres and disrupt their heterochromatic state, facilitating alternative lengthening of telomeres (ALT).”
Richard K Wilson, M.D., James R Downing, M.D., Jinghui Zhang, M.D., and Suzanne J Baker, M.D., led the St. Jude-Washington University study. Focusing particularly on identifying potential molecular mechanisms underpinning DIPG, the team began with whole-genome sequencing DIPG tumor tissue and matched normal tissue from seven affected children. Samples from four patients were shown to carry a somatic adenine-to-thymine transversion (c.83A>T) mutation in H3.3, which results in the same lysine 27 to methionine substitution (which they designate p.Lys27Met) identified by the McGill team in the pediatric GBM patients. A fifth DIPG patient harbored an adenine to thymine transversion (c.83A>T) in the HIST1H3B gene, which results in a Lys27Met alteration in H3.1 isoform.
To determine the frequency of mutations in the H3 gene family, the researchers sequenced the Lys27 encoding exons of 16 histone H3-coding genes in 43 DIPG cases and another 36 non-BS-PG samples. These data combined with those from the original seven cases indicated that 78% of DIPG samples and 22% of non-BS-PGs harbored recurrent somatic adenine-to-thymine transversions encoding p.Lys27Met alterations in H3F3A or HIST1H3B. A guanine-to-adenine transition resulting in a p.Gly34Arg alteration in H3F3A was also identified in 14% of non-BS-PG samples but none of the GBM tumors analyzed. The three different mutations were also mutually exclusive.
Significantly, the H3 histone mutations appear to occur excusively in high-grade pediatric gliomas, the researchers continue. While a p.Llys27Met-causing H3F3A mutation was found in a patient with grade 3 pediatric anaplastic astrocytoma, Sanger sequencing and whole-exome sequencing studies didn’t find mutations in any of the 16 H3 genes in other pediatric brain tumors including brainstem gliomas, medullobastomas, and ependymomas. Nor were H3 mutations detected in noncentral nervous system pediatric tumors. Moreover, a search of large-scale public exome and SNP databases found no incidences of germline polymorphisms affecting Lys27 or Gly34 in the histone H3 gene.
How the Lys27 and Gly34 mutations affect histone H3 function isn’t yet clear, the authors admit. However, they remark, both the residues are sited within the highly conserved N-terminal tail of the histone H3 protein, which influences the dynamic regulation of chromatin structure and accessibility. “Alterations at these invariant residues, which are close to the site where the tail exits the globular histone core of the nucleosome, may affect nucleosome structure and function by affecting histone-DNA interactions, chromatin compaction, or interactions with other effectors that bind to histones ... these mutations could potentially affect epigenetic regulation of gene expression, selective regulation of developmental genes, or telomere length or stability.”