Converging evidence for inhibition of transcriptional control in high-grade gliomas

See article by Katagi et al. pp. 1348–1359.

The identification of effective therapies for high-grade gliomas (HGGs) presents a stubbornly recalcitrant obstacle for the field of neuro-oncology. Great progress has been made in dissecting the heterogeneous biology and myriad oncogenic programs underlying the spectrum of both adult and pediatric HGG, but few of these insights have translated into meaningful new avenues of therapy. While ongoing work continues to define druggable subtype- or mutation-specific dependencies, increasing evidence is emerging for targeting the transcriptional machinery underlying these diverse oncogenic programs directly. Transcriptional addiction, or an acquired reliance on the continuous activity of an oncogenic transcriptional program, spans many cancer types and driver mutations. Importantly, it may represent a therapeutic vulnerability agnostic of much of the intra- or inter-tumoral heterogeneity observed in diseases like glioblastoma multiforme (GBM) or diffuse intrinsic pontine glioma (DIPG).

The centrality of this dependency was perhaps best exemplified in Miller et al,¹ in which the authors performed in vivo RNAi screening using xenograft models of glioblastoma. In contrast to the well-annotated cell cycle and proliferative programs enriched as hits in parallel in vitro screens, the in vivo tumors revealed striking dependencies on numerous genes involved in the regulation and mediation of transcription initiation and RNA polymerase II (Pol II) pause-release. These genes correlated with in vivo-derived, patient-validated expression signatures of GBM cell survival within the intact CNS microenvironment, and genetic and pharmacologic disruption of enhancer-mediated Pol II pause-release resulted in marked prolongation of survival in animal models of GBM.¹

The regulation of productive transcription by Pol II is a complex and multistep process involving several transcriptional cyclin-dependent kinases (CDKs). In a simplified model, the signal for initiation is provided by phosphorylation of the serine 5 residue of the Pol II C-terminal domain (CTD) by CDK7. At many genomic loci, this allows for transcription of only ~40-80 bp downstream before Pol II becomes paused. Phosphorylation of the serine 2 CTD residue by CDK9 is then required for Pol II pause-release and productive elongation into gene bodies. The CDK8-containing Mediator complex acts upstream of these events, linking enhancer elements to promoter-associated transcriptional machinery (Figure 1).

Fig. 1.

Transcriptional control as a therapeutic vulnerability in high-grade gliomas. Schematic of RNA polymerase II initiation and elongation control by transcriptional CDKs. Citations reflect studies identifying steps as therapeutic targets in high-grade gliomas. Citations in blue indicate studies in pediatric H3K27M-mutant diffuse intrinsic pontine glioma. Abbreviation: CDKs, cyclin-dependent kinases.

Notably, each of these catalytic steps has now been independently identified as druggable events in HGGs, including H3K27M-mutant DIPG. In preclinical models of adult GBM, the anti-tumor effects of CDK7 inhibition were first demonstrated in studies from Greenall et al² and Meng et al.³ Using the compound THZ1, the authors found that disruption of transcriptional addiction preferentially affected genes associated with super-enhancers (SEs) and led to cell death across the spectrum of GBM phenotypic and mutational states. The role of CDK8 in Mediator is complex and incompletely understood, with the highly modular nature of this complex allowing it to assume a variety of integrative roles in transcriptional regulation. Despite this, recent work from Fukasawa et al⁴ has identified CDK8 as a key mediator of stemness and tumorigenicity in adult GBM. The authors find that CDK8 expression in GBM patient samples is highly correlated with markers of stemness and poor prognosis. They then go on to show that depletion or inhibition of CDK8 attenuates GBM self-renewal through modulation of c-MYC signaling, again highlighting the utility of transcriptional inhibition as an indirect means of drugging a diverse range of oncogenic transcription factor circuitry. Finally, the precise control of productive transcriptional elongation by CDK9-mediated pause-release allows for either the fine-tuning of basal gene expression or for rapid, synchronous transcriptional induction in response to developmental or other exogenous signals. This was first described as relevant mechanism in adult gliomagenesis by Xie et al,⁵ in which the authors found that RBPJ, a central mediator of NOTCH signaling in glioma stem cells, exerted its tumorigenic effect via CDK9 recruitment to target promoters. Subsequent studies by Su et al⁶ and Le Rhun et al⁷ demonstrated that targeting CDK9 using the multi-kinase inhibitor TG02 showed anti-tumor effect in GBM and led to CDK9-dependent downregulation of oncoproteins such as MYC and MCL1. In a mouse model of GBM, the combination of TG02 and temozolomide was found to be synergistic in prolonging animal survival.⁶

Within preclinical models of pediatric H3K27M-mutant DIPG, CDK7-dependent transcriptional initiation was first described as a therapeutic target by Nagaraja et al.⁸ Again utilizing the CDK7-inhibitory compound THZ1, the authors demonstrated that transcriptional inhibition resulted in preferential disruption of SE-associated genes key to the DIPG malignant cell identity. They further showed that THZ1 exhibited a dose-dependent anti-tumor effect that both synergized with and overcame resistance to other promising preclinical agents, including both histone deacetylase and bromodomain inhibition.

Using an unbiased epigenome RNAi screen, Dahl et al⁹ recently identified the CDK9-containing super elongation complex (SEC) as a functional dependency in H3K27M-mutant DIPG. They demonstrated that the H3K27M mutation alters the regulation of SEC components by affecting changes to activating histone marks at SEC member promoters. By employing a combination of SEC RNAi depletion and CDK9 pharmacologic inhibition, the authors found that reduction of SEC activity resulted in an attenuation of DIPG clonogenic potential and self-renewal in favor of gene expression programs promoting differentiation. This was reflected in a ChIP-quantifiable elongation defect, with Pol II promoter occupancy preferentially shifted toward a paused state at genes regulating terminal morphogenesis. Treatment with the highly selective CDK9 inhibitors atuveciclib and AZD4573 resulted in prolonged survival with minimal toxicity in orthotopic xenograft models of DIPG.

In this issue, Katagi et al replicate these findings in H3K27M-mutant DIPG.¹⁰ They utilize a peptidomimetic compound, KL-1, to selectively disrupt the SEC subunit interaction between AFF4 and the CDK9-containing PTEFb. This approach enables the dissection of SEC-specific regulatory effects, in contrast to the many PTEFb-containing complexes which may be targeted by conventional CDK9 small molecule inhibitors. They find that SEC disruption similarly imparts a restricted elongation defect, with Pol II promoter occupancy primarily altered at genes involved in cell differentiation. This molecular analysis confirms prior data⁹ and highlights the relative genomic selectivity of SEC-directed intervention. Importantly, they demonstrate that KL-1 treatment in xenograft models of DIPG results in prolongation of animal survival, lending further data to the growing evidence for transcriptional inhibition as a viable therapeutic approach for DIPG and other HGGs.

Funding

This work was supported by the Cancer League of Colorado (R.V.) and Morgan Adams Foundation (N.A.D., R.V.).