Highlights from the Literature

Targeting the Enzyme AMPK Impairs Energy Production in Glioblastoma Cells and Reduces Tumor Growth

Tumor cells have elevated energy demands to satisfy the requirements of their aberrant growth and proliferation rates. AMP-activated protein kinase (AMPK), a serine/threonine kinase, acts as a central regulator of cell energy by sensing the metabolic stress caused by hypoxia or glucose depletion.¹ Activation of AMPK in these stress conditions inhibits biosynthetic pathways and increases the production of ATP, restoring the energy pool of the cell.

Because AMPK inhibits anabolic processes required for normal cell growth, its upregulation has usually been associated with a tumor-suppressor role.² However, Chhipa and colleagues,³ have recently reported a notable upregulation in expression and activity of this enzyme in glioblastoma (GBM) tumors compared with both normal brain and lower-grade gliomas. The elevated expression of AMPK was retained in GBM stem-like cells (GSCs) and was necessary for cell viability. Accordingly, knockdown or deletion of the β1 subunit of this enzyme reduced GSC viability and self-renewal as well as tumor formation in vivo. The forced suppression of AMPK increased the production of superoxide radicals and reduced autophagy, both indicators of energy drainage in the cell. The investigators observed, however, that none of these effects was sufficient to explain the death of GSCs, which was more likely due to a failure in ATP production following AMPK suppression. Additional mechanistic results suggested that AMPK becomes upregulated in GSCs not as a consequence of external stressors but as a response to the sustained intracellular state of stress in cancer cells. This “hijacked” AMPK activity promotes continuous energy production through genetic programs regulated by the transcription factors HIF1α (which regulates glycolysis) and GAPBA (mitochondrial biogenesis and activity). Interestingly, AMPK silencing had no effect on conventional GBM cell lines, suggesting that either the process of differentiation or the long evolution of those cells in vitro have resulted in compensatory, AMPK-independent mechanisms to control energy generation. Although specific small-molecule inhibitors of AMPK have remained elusive, genetic suppression of this enzyme had no adverse effects in mice. This suggests that AMPK targeting could be a novel therapeutic avenue to treat GBM with limited effects on normal tissues.

References

Functional Diversity and Cooperativity Between Subclonal Populations of Pediatric Glioblastoma and Diffuse Intrinsic Pontine Glioma Cells

Tumor heterogeneity is now a well-established feature of both adult and pediatric brain tumors. The relevance of rare subclones in driving tumorigenesis and the functional cooperation amongst these subclones within these tumors remain understudied.

An elegant paper by Vinci et al explores the mechanisms of subclonal diversity and how this influences the tumor phenotype using pediatric glioblastoma (pGBM) and diffuse intrinsic pontine glioma (DIPG) as a model system.¹ Using published datasets from 142 specimens, the authors observed multiple co-existing subclones in a majority of the cases; gene mutations were identified that were predominantly or consistently clonal but also some that were associated with subclonal populations. Subclonal number associated with worse prognosis in DIPG, and comparison of pre- and post-treatments showed that although individual subpopulations changed in response to therapy, many subclones remained unchanged, highlighting that these subclones may cooperate to resist therapy. Using autopsied DIPG samples, the authors not only showed evidence of genetically distinct subclones that are separated spatially but also genetic divergence of cells that migrated away from the pons. To examine the functional relevance of these findings, patient-derived H3.3 K27M mutant tumors were used wherein both 2D and 3D cultures were employed to examine growth characteristics of these cells. Clones that expanded not only had shared but also private mutations in genes involved in cell shape and motility that was reflected in their ability to migrate and invade in vitro. In a specific case of a rare clone with a private mutation in the histone H4 methyltransferase KMT5B, although it did not differ from the wild-type subclone in its morphology or immune content, it exhibited considerable sensitivity to PARP inhibitors, suggestive of differential pathway activation between the subclones. Co-culture of the subclones showed that poorly migratory/invasive clones could be influenced by conditioned media from highly invasive subclones, suggesting that factors secreted by subclones can influence behavior. This was seen in vivo, where, although cells had similar proliferative capacity or immune infiltration, they showed differences in survival owing to invasive differences when subclones were mixed.

Overall, this study demonstrates that despite low mutation burden, DIPG and pGBM can exhibit aggressive behavior due to cooperative interactions amongst genetically distinct subclones. Disrupting these interactions will remain a challenge, yet they will need to be overcome as they present barriers to current therapies.

Reference

Cell of Origin of Human Glioblastoma

A number of studies have investigated the cell of origin of glioblastoma (GBM) and have suggested neural stem cells (NSC), oligodendrocyte precursor cells, and other central nervous system cells as likely candidates. However, most of these studies were conducted in mice, and direct genetic evidence form human GBM patients was lacking.

A recent study published in Nature provides new evidence from human patients that GBM arises from NSCs in the subventricular zone (SVZ) that harbor low-level driver mutations.¹ The authors first performed deep sequencing of triple-matched human GBM tumors, normal SVZ tissue distant from the tumor mass, and normal cortex or blood from 30 patients. They found that tumors, SVZ, and normal brain/blood harbored an average of 80.7, 23.0, and 4.3 somatic mutations, respectively. A majority of mutations in the SVZ were also found in the matching tumors and consisted mostly of mutations in cancer-driving genes including TERT, EGFR, PTEN, and TP53. On the other hand, the majority of mutations in the tumors were private (not shared with the SVZ). To determine if the presence of shared driver mutations in SVZ and tumors is due to clonal evolution and migration of SVZ cells into tumors or to micro-invasion of tumor cells into the SVZ, the authors performed single-cell cloning and sequencing as well as studies in mice. They found that all single-cell tumor clones that contained shared mutations also contained private mutations but that single-cell SVZ clones that contained shared driver mutations did not contain tumor private mutations. These data suggested that SVZ cells containing driver mutations evolve into GBM tumors. Using immunostaining and laser capture microdissection, the SVZ cells that contained the driver mutations were identified as astrocyte-like stem cells from the astrocytic ribbon layer. The authors then generated mouse models in which the shared driver mutations detected in human SVZ and tumors (p53, Pten, egfr) were introduced into either mouse NSCs of the SVZ or mouse cortex. When the driver mutations were introduced into SVZ NSCs, the cells harboring the mutations migrated from the SVZ into the cortex and brain tumors developed. Conversely, when the mutations were introduced into the cortex, no tumor formation or spread of the cells into the SVZ were observed.

Based on the above and additional data described in the manuscript, the authors proposed that NSCs carrying low-level driver mutations are the cell of origin of GBM and that they migrate from the SVZ to generate tumors in distant brain regions. This well-performed study provides powerful first evidence from human patients that NSCs from the SVZ are the (or one) cell of origin of GBM and identifies putative initiating driver mutations.

Reference

IDH Mutations in Glioma Target Host T Cells in the Tumor Microenvironment

Isocitrate dehydrogenase (IDH) mutations are thought to drive malignant transformation in glioma while also decreasing cellular fitness on a metabolic level. The epigenetic consequences of IDH mutation are well established, and their impact on differentiation state is hypothesized to promote neoplastic expansion. However, other non-tumor cell-autonomous mechanisms, potentially involving the immune system, may also be mobilized by IDH mutation and its associated oncometabolite (R)-2-hydroxyglutarate (2HG).

In a recent Nature Medicine article, Bunse et al. explored one such mechanism involving T lymphocytes and their responsiveness to tumor antigen.¹ The authors found that 2HG was imported by T cells in a paracrine fashion where it reduced proliferation and other antigen-mediated responses (e.g. cytokine production), including those generated by the IDH1 R132H mutation itself acting as a neoantigen. They corroborated these findings with an analysis of TCGA (the Cancer Genome Atlas) data demonstrating reduced markers of memory T cells and increased markers of naïve CD4+ T cells in IDH-mutant gliomas. Subsequent analyses in 2HG-treated T cells and TILs (tumor infiltrating lymphocytes) derived from IDH-mutant gliomas demonstrated that altered calcium and phosphatidylinositol signaling in these contexts led to downregulation and suppressed nuclear translocation of NFAT (nuclear factor of activated T cells) and NF-κB target genes. These changes were also associated with reduced levels of PD-1 and effector cytokines. The authors also implicated AMPK (AMP-activated protein kinase) and ODC (ornithine decarboxylase) as potential therapeutic targets in these disrupted signaling axes. Finally, the authors used an IDH-mutant sarcoma model to recapitulate the TIL phenotypes described above and demonstrated that disabling 2HG production in this context increased the efficacy of IDH1 R132H-targeted vaccine therapy. Similar results were obtained in orthotopic glioma xenograft studies, where vaccine reduced tumor growth only when coupled with inhibition of mutant IDH1. 2HG suppression also appeared to synergize with PD-1 inhibition in this latter model system.

Taken together, these data indicate that 2HG generated by mutant IDH suppresses T cell signaling and activation in the glioma microenvironment, incapacitating antitumor T cell immunity induced by vaccination, adoptive T cell transfer, and checkpoint blockade. These findings are highly significant in that they characterize targetable, non-tumor cell-autonomous mechanisms by which IDH mutations promote gliomagenesis in vivo.

Reference

Pharmacogenomic Landscape of Patient-Derived Tumor Cells and Precision Therapy

The success of molecularly driven, individualized therapy has been constrained for many tumor types by the complexity of underlying molecular alterations that confer resistance to targeted agents. A recent study aimed to overcome this limitation by integrating molecular profiles (expression, mutation, gene fusions, copy number aberrations) obtained for patient-derived tumor cells (PDCs) from 462 patients across 14 tumor types, including high-grade glioma and brain metastasis, with respective pharmacologic response data to 60 targeted agents in vitro.¹ To preserve the biology and inter-individual differences, short-term cultures were used for the large PDC library, e.g. neurosphere-like cultures for high-grade glioma. High similarity of the molecular makeup to the parental tumor was confirmed. The drug responses revealed lineage-specific drug sensitivities. In order to shed light on the association of genomic alterations with drug response, PDC genome-drug mapping was performed. Just to mention a few findings, drug sensitivity of somatic EGFR alterations, mostly found in GBM, were revealed for inhibitors of the EGFR (dacomitinib), VEGFR (vandetanib), and to the authors’ surprise inhibitors of Bruton’s tyrosine kinase (BTK; ibrutinib). On the other hand, multidrug resistance was found to be associated with an activated WNT signaling pathway across several tumor types, in line with previous reports.

A drug-centered analysis suggested sensitivity of GBM with EGFR alterations to the BTK inhibitor ibrutinib. The drug sensitivity was not associated with expression levels of BTK downstream targets, but with EGFR transcript levels. In accordance, longer survival was observed in an EGFRamp/vIII PDX model treated with ibrutinib in vivo. Evaluation of hits for resistance to EGFR inhibitors in EGFR-amplified GBM identified neuregulin 1 (NRG) expression as a candidate, which was supported by knock-down experiments. As a mechanism, induction of hetero-dimerization of the EGFR with HER3 was suggested based on previous studies.

In this comprehensive study, a good number of clinically interesting associations of specific genomic alterations and sensitivities across different drug classes and tumor types were discovered and discussed for potential clinical use. The next obvious question regarding the validation of the findings in patients was addressed in a retrospective study of a cohort of 31 patients across several tumor types treated with targeted drugs (against EGFR, VEGFR, PI3K/mTOR, MET, or PDGFR). The authors report good concordance of the clinical response rate and in vitro drug sensitivity of the corresponding PDCs. Based on these encouraging results, the authors are confident about the clinical feasibility of PDC-based drug sensitivity screening and advocate the combination of pharmacologic response screening and genomic profiling. A prospective exploratory trial for metastatic gastric cancer patients is ongoing that includes three targeted agents (NCT03170180).

This study provides a wealth of data and insights outlining new perspectives and opportunities. It remains to be seen whether the concept will be feasible for future standard care for difficult-to-treat cancers or whether it will be used as a powerful tool for identification of predictive biomarkers and refined design of clinical trials.

The pharmacogenomics dataset of the reported study is publically accessible as interactive web resource (cancerdrugexplorer.org).

Reference