The H3.3K27M Oncohistone Affects Replication Stress Outcome And ...

Discussion

Whereas the changes in epigenetic patterning in histone mutant gliomas have been extensively investigated [1–6,9], the exact mechanisms driving oncogenesis remain elusive to date. Rather than being solely dependent on transcriptional changes, we hypothesize that pediatric high-grade glioma results from an interplay of multiple pro-tumorigenic processes that are related to the different functions of the multifaceted (onco)histone H3.3. These include H3.3 as an epigenome/transcriptome regulator, but also as an important actor in different DNA damage repair mechanisms and in maintenance of genome integrity, which can explain the tendency for H3.3 mutant gliomas to be highly aneuploid [4,16,24,69–71]. The latter idea was explored in this study, in which we discovered that histone mutant H3.3K27M cells are more susceptible to acquire genome instability resulting from an intrinsic sensitivity to stress during DNA replication.

While studying the mechanism underlying the strong genomic instability found in H3.3K27M HGG, we made a number of observations that point at H3.3K27M deregulating several processes that may be related. For instance, the more rapid resolution of the ɣ-H2AX signal following UVC exposure could indicate that DNA damage signaling ceases prematurely in the presence of the H3.3K27M mutant histone. This could allow damaged cells to pass the G2 cell cycle checkpoint and progress into mitosis. Interestingly, in our mitotic pulldown experiments, we uncovered an enrichment for DNA replication components, comprising PCNA, Mini-Chromosome-Maintenance proteins (MCMs) and replicative histone chaperones, in H3.3K27M cells. This was unexpected, as although MCM proteins have been identified as direct histone H3 interactors previously, these factors are typically unloaded prior to mitosis [49,50,72,73]. Under physiological conditions, the MCM complex assembles in G1 as a pre-replisome complex on DNA in a process called origin licensing [74]. In late G1 and S-phase, interaction with the replicative helicase converts the structure into the replicative-complex that initiates replication, or origin firing [75]. Of note, MCM complexes are widely deposited throughout the genome, but most remain dormant and only become activated during replication stress to ensure complete genome duplication [76–78]. When this process fails, PCNA and MCMs remain stably locked with un-replicated chromatin throughout the cell cycle until the next S-phase, and this might lead to MCM binding to mitotic DNA [50,51,79,80]. Thus, the increased binding of MCM proteins to H3.3K27M may result from stalled replication forks that were not resolved prior to mitosis. The reason for this remains speculative, but it is possible that replication stress-induced damage remained undetected until mitosis due to improper DNA damage signaling. Alternative explanations are H3.3K27M-induced changes in the epigenetic landscape impairing origin licensing or firing; or altering the distribution of H3.3 proteins to alternative genomic localizations, the latter potentially leading to interactions with novel binding partners. Yet it is also possible that the Lysine 27 substitution directly interferes with MCM binding [49,50,72,73,81,82].

Additional evidence for unresolved DNA replication problems in mitotic H3.3K27M cells is provided by the increased numbers of DNA UFBs. Normally, mitotic DNA synthesis (MiDAS) of under-replicated DNA provides the lattermost opportunity to complete DNA replication and therefore tackle extensive genome instability [51]. Hence, it is possible that MiDAS is also impaired in H3.3K27M cells, especially since H3.3K27M inhibits the EZH2 histone methyltransferase that places the H3K27me3 mark that serves as docking platform for key MiDAS factor MUS81 [10,15,51,83–87]. Furthermore, it is quite remarkable that relatively small changes in H3.3K27M expression can have such profound effects on UFB formation. In our RPE1 cell model, the fraction of mutant H3.3 was around thirty percent of endogenous H3.3. This is largely in line with what has been reported for tumors in vivo, in which mutant H3.3 constitutes up to twenty percent of all H3.3—yet with a dramatic effect on outcome [10]. Importantly, when we deleted the mutant H3F3A allele from primary H3.3K27M HGG cells, leaving the three wild type alleles intact (i.e, one remaining copy of H3F3A and two copies of H3F3B), the number of UFBs decreased significantly. This suggests that the H3.3K27M oncohistone, despite relatively low expression, remains an important driver of genomic instability even in a malignant background.

Finally, we found that the increase in UFBs did not trigger an increased recruitment of 53BP1 nuclear bodies in H3.3K27M cells, suggesting that not all UFB breaks are buffered by this salvage mechanism. This observation was further accentuated by the H3.3 mutant tumors, which exhibited relatively low numbers of 53BP1 nuclear bodies compared to histone wild type tumors. Indeed, H3.3 gliomas tend to co-mutate DNA repair enzymes that also have a role in 53BP1 nuclear body formation, which could suggest that bypass of the 53BP1 response is favorable for H3.3K27M (tumor) cells as disruption of such a repair mechanism might prevent resolving the DNA damage resulting from UFBs and thus enhance genome plasticity and heterogeneity of the tumor, and ultimately help the cancer cells to thrive [4].

In conclusion, efficient targeting of H3.3 mutant pediatric high-grade glioma remains highly challenging and it is likely that the multi-faceted functions of (mutant) H3.3 play an important part in this. The findings presented in this study highlight a few of these functions and provide a possible explanation for the GIN/CIN phenotype observed in H3.3K27M high-grade glioma. They also suggest that these tumors might be more susceptible to compounds that exacerbate replication stress, however this should be investigated in follow up studies.