Introduction
Medulloblastoma (MB) is one of the most frequent malignant brain tumors in children, accounting for around 10% of all childhood brain tumors [
1]. The prognosis of MB patients is very much dependent on the molecular subgroup of the tumor (WNT, SHH, Group 3 and Group 4) [
2,
3]. Group 3 and Group 4 tumors show a profound molecular heterogeneity and can be classified into 8 subtypes [
4]. Patients with Group 3 MB tumors with high expression of the oncogenic transcription factor MYC (subtype II) [
4] show poor survival rates despite highly aggressive treatment. New treatment approaches for this high-risk subset of patients are urgently needed [
5].
We and others have previously shown that group 3 MBs have an aberrantly high expression of class I histone deacetylase 2 (HDAC2) [
6,
7] and that
MYC-amplified group 3 MB cells are highly sensitive to class I HDAC inhibition in vitro and in vivo [
6,
8]. The cytotoxic effect of class I HDACi in
MYC-amplified MB is partially mediated by inhibition of HDAC2 in the MYC-HDAC2-protein complex leading to reduced MYC protein turnover and to an alteration of MYC chromatin binding patterns inducing an altered expression of MYC regulated genes [
9].
HDACis are established drugs for the treatment of hematological malignancies, currently four HDACi are FDA-approved for the treatment of T-cell lymphoma and multiple myeloma [
10]. However, in solid tumors single agent HDACi fail to achieve significant anti-tumor effects in clinical trials [
11]. It has become clear that a lack of knowledge on the molecular downstream effect of HDAC inhibition and lack of biomarkers for patient selection are currently impeding the translation of promising preclinical findings into clinical application [
12]. It may be that combination therapies will increase the clinical efficacy of HDACis in solid tumors [
13].
In this study, we investigate alterations of gene expression in response to class I HDAC inhibition in MYC-amplified MB cells to identify affected pathways. Clinically advanced compounds that target these pathways are evaluated for synergistic interaction with entinostat.
Discussion
Promising advances in the genomic, epigenomic, and proteomic characterization of MYC-driven MB have not yet resulted in biomarker-guided targeted treatment regimens. Patients with group 3 MYC-driven MBs face a very dismal prognosis and novel therapeutic approaches are urgently needed. Here, we analyzed the effect of entinostat treatment on the transcription profile of MYC-driven MB cells to better understand why these cells are highly susceptible to class I HDACi treatment and to deduce potential drug partners for synergistic or synthetic lethal combination therapies.
The majority of pathways that were affected by entinostat treatment in our data (DDR, cell cycle, RNA processing, differentiation) has previously been reported to be affected by HDACi also in other cancer entities [
33,
34], suggesting that some of these effects may be rather drug class (HDACi) than entity specific. In line with previous findings [
35] we also confirmed the enrichment of immunological pathways and upregulation of HLA class I heavy chain paralogues upon entinostat treatment. This supports the role for class I HDACis in immune cancer therapies, which is currently under clinical investigation in several clinical trials [
36] (including in pediatric oncology, see for example NCT03838042; nivolumab/entinostat).
Our synergy experiments identified synergism of entinostat and PARPis in
MYC-amplified MB cells lines at translationally relevant drug concentrations with regard to c
max reported in patients. Of note, we encountered previously described synergism model limitations in form of inconsistencies between applied synergism computations, leading us to integrate distinct models and emphasizing the importance of synergy validation experiments [
37].
We hypothesize that the functional explanation for the observed synergy of olaparib and entinostat is a joint effect on a cell’s DNA damage repair capacity, suggesting a drastically reduced DNA damage repair upon combination treatment. This is in line with our observation of increased levels of DSBs upon combination treatment, which could be dramatically increased by additional treatment with the DNA damaging agent doxorubicin. While PARP inhibition directly leads to reduced DNA damage detection signaling [
38], HDAC inhibition treatment has been described to impair NHEJ- or HR-repair by transcriptional repression of involved genes including
ATR,
ATM,
BRCA1,
BRCA2,
Ku70,
Ku80,
DNA-PK,
ligase IV,
XRCC4,
RAD51,
CHEK1 and
CHEK2 [
33,
39]. The expression of all of these genes was also affected by entinostat treatment in our data. Considering this highly consistent deregulation of DDR genes upon HDACi treatment, we assume that also other, clinically less advanced compounds interfering with DDR such as CHEK1/2 kinase inhibitors, DNA-PK inhibitors or ATR inhibitors may be promising combination partners for entinostat in this context.
Synergistic interaction of HDACi and PARPi has been previously described in vitro and in vivo including the combination of vorinostat/olaparib in GBM cell lines [
40], entinostat/olaparib in HR-proficient ovarian cancer cells [
39], as well vorinostat/veliparib, vorinostat/olaparib and belinostat/olaparib in triple negative breast cancer cells and xenograft models [
41,
42]. Three clinical trials are currently investigating this drug class combination (NCT03924245; NCT03742245; NCT04703920) in gynecological entities.
Notably, in our data
MYC-amplified MB models showed increased susceptibility to both single agent and combination treatment compared to non-
MYC amplified MB cells. Investigation of gene expression and protein expression data of primary MB confirmed high PARP1 expression and correlation with
MYC expression, particularly in subtype II MB samples. This is in line with previously published data showing elevated PARP1 expression in MB correlating with poor disease outcome [
43]. We hypothesize that MB cells with high
MYC expression are prone to acquire DNA damage caused by MYC-mediated replication stress, which may explain why
MYC-amplified MB cells are particularly dependent on a functioning DNA damage signaling and repair machinery. This is in line with previously published data showing increased sensitivity of cancer cells with high
MYC/MYCN-expression to PARPi including glioblastoma stem-like cells, neuroblastoma, multiple myeloma and medulloblastoma cells in vivo and in vitro [
44‐
47]. The data provided here on a possible co-localization of MYC and PARP1 has not been previously described and warrants further investigation since it may provide further insight into a mechanistic interaction of both proteins. The factor of potential underlying BRCAness, as reported previously in primary group 3 MB samples, might also play a role in the observed sensitivity to PARPi [
4,
48].
Our data suggest that additional induction of DNA damage (i.e. with doxorubicin) may be beneficial to fully exploit the reduced DDR capacity of cells treated with entinostat and olaparib. The combination of PARPi with a DNA-damaging chemotherapy is already being investigated in adult (NCT03161132; NCT00740805) and pediatric clinical trials (NCT02813135; NCT03749187; NCT02044120) [
32,
49]. Importantly in the field of neuro-oncology several trials investigate PARPi treatment in combination with radiotherapy and/or temozolomide in HGG (NCT03581292; NCT03212742; NCT03749187).
However, the combination of a PARPi plus a DNA damaging chemotherapeutic agent shows important and dose limiting myelotoxicity in the clinic [
50]. Accordingly, the identification of predictive biomarkers to select patients that are most likely to benefit from such a treatment, possibly at comparably low and thus better tolerable concentrations is crucial for a successful translation into the clinic. Our data suggests that entinostat might confer selectivity of DNA damage accumulation in
MYC-driven MB cells and reduce DNA damage tolerance selectively in transformed cells, thereby overcoming previously reported normal tissue toxicities [
50]. Further studies are needed to investigate the toxicity profile of this triple drug combination in in vivo models.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.