Highlights from the Literature

Three’s a Crowd: Breast Cancer Cells in Brain Steal Glutamate in Three-cell Pseudo-synapses

Central nervous system metastasis is a frequent and dreaded complication of breast cancer, with obstacles such as the blood-brain barrier impeding the success of many systemic therapies in this setting. While there has been modest progress in recent years in treating breast cancer metastasis to the brain, particularly with stereotactic radiosurgery, challenges remain and mortality remains relatively high. Some recent reports have shed light on unique aspects of the biology of breast cancer brain metastases, and a new report from Zeng et al. provides a potentially key insight into how breast cancer may thrive in the milieu of the brain.¹

Building on prior studies, the authors initially assessed the expression of glutamate receptor components in human cancers. They noted high expression of components of the NMDA receptor in breast cancer, most notably in basal-like breast cancers. In particular, high expression of the NMDA receptor component GRIN2B was associated with triple-negative breast cancers and high risk of recurrence and was also linked to metastasis. Given the connection of a neurotransmitter receptor and metastasis, the authors then focused on breast cancer metastasis to brain. They noted high activation of NMDA receptors in breast cancer brain metastases from patients. In mouse models, they derived highly brain-metastatic versions of breast cancer lines and showed that these lines had higher NMDA glutamate receptor activity. As these lines did not produce more glutamate, the authors assessed whether they might be getting glutamate from brain synapses. They were indeed able to demonstrate that breast cancer cells were intimately involved at glutamatergic synapses, forming a pseudo-tripartite synapse with presynaptic and postsynaptic neurons—essentially replacing astrocytes. In addition, the authors also showed in mouse models that expression of GRIN2B was associated with the extent of brain metastasis.

These results provide an intriguing biologic insight on breast cancer cells in the brain. While a novel finding, the authors do note its complementarity to a prior study showing that breast cancer cells are supported in the brain by forming gap junctions with astrocytes.² The fact that this newly described phenomenon involves NMDA glutamate receptors may also provide near-term therapeutic leverage, as the NMDA receptor inhibitor memantine is already in the clinic. In fact, a course of memantine is already being used in some patients with numerous brain metastases from breast cancer to alleviate side effects from whole-brain radiation therapy. Additional follow-up is warranted on both the biologic and therapeutic fronts, and this will hopefully provide new traction against the scourge of breast cancer brain metastasis.

Targeting Glioblastoma Stem Cells Through Disruption of the Circadian Clock

Circadian clocks are observed in most normal cells and tissues as well as in cancer cells. Molecularly, the clock is controlled by a transcriptional and translational feedback circuitry involving a set of “clock genes” including, among others, Clock, Bmal1, Periods (Per1/2/3), and Cryptochromes (Cry1/2) that code for transcription factors and transcriptional regulators to generate rhythms of clock gene and clock-controlled gene expression. Through their rhythmic transcription, the actions of these circadian clock genes have been shown to regulate key aspects of cellular metabolism and respiration and are considered important regulators of cellular division. Although a robust link between circadian dysfunction and cancer is well established, mechanistic details are not well understood. Glioblastoma (GBM) displays a high level of cellular heterogeneity and contains self-renewing glioblastoma stem cells (GSCs). While it is known that GBM cells are intrinsically circadian, the extent and function of this circadian rhythmicity in GSCs and differentiated GBM cells remains mostly unaddressed. A recent study, published in the journal Cancer Discovery,¹ demonstrates that although GSCs, differentiated GBM cells, and normal brain cultures all displayed robust circadian rhythms, only GSCs are exquisitely dependent on the core clock transcription factors, BMAL1 and CLOCK, for optimal cell growth.

The study established that genetic ablation of Bmal1 or Clock expression induces a cell cycle arrest and an apoptotic response in GSCs but not in differentiated GBM cells. Disruption of Bmal1 and Clock also reduced the maintenance of stemness in GSCs, and BMAL1 was shown to bind to the promoters of the stemness factor SOX2, OLIG2, and MYC. These studies established that key members of the circadian machinery are essential for maintaining a stem cell transcriptional profile in GSCs along with the self-renewal properties of GSCs. Pharmacological inhibition of key regulators of the BMAL1:CLOCK machinery REV-ERBs and CRY led to disruption in the self-renewal properties of GSCs, increased apoptosis, and prolonged survival of GSC xenograft models in vivo.

These results led the authors to suggest that core clock proteins likely regulate gene expression programs in GSCs that are distinct from those in normal stem cells. Indeed, BMAL1 ChIP-seq experiments revealed a GSC-specific BMAL1 binding site profile that suggests a function beyond the generation of circadian oscillations but rather a repurposing of this key circadian regulator to gain novel functions in GSCs to support aberrant tumor metabolism. Metabolic analyses suggest that BMAL1:CLOCK heterodimers may directly regulate mitochondrial oxidative phosphorylation (OXPHOS) and glycolysis in GSCs by binding to the promoters of glycolysis and TCA cycle genes. One gene affected by Bmal1 and Clock disruption is succinate dehydrogenase (SDHA). Reduction of SDHA levels in GSCs resulted in the induction of apoptosis, reduction of GSC cell growth, and decreased expression of stem markers.

Taken together, this study reveals a novel role for the key circadian proteins BMAL1 and CLOCK in promoting and maintaining stemness in GSCs and offers new therapeutic approaches for the treatment of GBM.

Adjuvant Whole-brain Radiation Therapy Compared With Observation After Local Treatment of Melanoma Brain Metastases: A Multicenter, Randomized Phase III Trial

Several clinical trials have tried to tease out the role of whole-brain radiation therapy (WBRT) for metastatic brain tumors, resulting in an evolution in management overtime for these challenging patients. Most experts now favor a tailored approach to patients with <4 brain metastases (BM). Often WBRT is avoided because of the lack of an overall survival benefit and concern for neurotoxicity. Hong et al continue this effort to evaluate the role of WBRT in a randomized, phase III trial restricted to patients with 1–3 BM from melanoma.¹ Following local treatment (surgery or stereotactic radiosurgery, SRS), patients were randomized to WBRT (N = 107) or observation (N = 108). Stratification factors included the number of metastases, presence or absence of extracranial disease, gender, age, and planned dose of WBRT. Patients could also receive any systemic therapy. The primary endpoint was distant intracranial failure within 12 months.

Forty-two percent of the patients in the WBRT cohort and 50.5% in the observation group developed distant failure within 12 months (not statistically significant). The median time to development of distant intracranial failure was 26.4 months in the WBRT group and 11.5 months in the observation group, but this was also not statistically significant, likely because of the small sample size. There was no difference in the rate of neurologic death or median time to deterioration in performance status. In several subanalyses, the authors tried to identify subgroups who benefited. For example, in patients with a single metastasis, the intracranial failure rate was significantly lower in the WBRT group than in the observation group, but this was driven by those who received surgery; in patients who received SRS as local therapy, there was no difference. Not unexpectedly, WBRT was associated with more grade 1 to 2 acute toxicity.

Similar to prior trials evaluating RT in BM patients, this trial suffered from slow accrual—it took 8 years and 935 patients were screened to enroll 215 patients from 24 centers. During this prolonged accrual period, systemic therapy with checkpoint inhibitors and BRAF targeting therapies have shown responses in melanoma BM patients. Thus, how to incorporate these results into clinical practice is challenging, although the results do seem to confirm that WBRT is not clearly needed in patients with <4 BM from melanoma.

Mitogenic and Progenitor Gene Programs in Single Pilocytic Astrocytoma Cells

Pilocytic astrocytomas (PAs) are characterized by their simple genetic make-up with usually only a single known driver mutation, the BRAF rearrangement. This pediatric brain tumor, classified as a WHO grade I glioma, presents in general with an indolent clinical course and may be cured upon complete surgical resection. However, little is known about the intra-tumoral molecular heterogeneity and the hierarchic organization of these tumors.

Taking advantage of single-cell technology Reitman et al¹ characterized the cellular composition of 6 PAs using RNA sequencing. The cells were first labeled and sorted with the marker for glial progenitor cells A2B5+ that is known to be enriched in PA. The KIAA1549-BRAF fusion, detected in 5/6 PAs, was exclusively found in the A2B5+ cells. The 6^th PA had a non-canonical BRAF duplication. Clustering the single cells by their expression profiles yielded two A2B5+ tumor clusters expressing glial markers associated with PAs such as OLIG2, APOD, and PDGFRA, and three A2B5- clusters expressing markers of immune cells. These were subdivided into a cluster enriched for microglia, one for T cells, and one for macrophages. The most common non–tumor component was microglia with up to 40% of cells, in a tumor location-specific manner. Deconvolution of expression profiles of bulk tumors revealed more tumor cell-related gene expression in infratentorial tumors, whereas supratentorial tumors expressed more genes related to the microglia gene cluster. Markers relevant for current immune checkpoint inhibitor therapies, such as the ligands for PD1 CTL4, were rarely expressed on tumor cells, whereas the antigen-presenting capacity appeared intact.

The PA cancer cells exerted oligodendrocyte precursor cell (OPC)-related expression signatures. When comparing these signatures to those from higher-grade pediatric glioma, such as H3K27M mutated midline gliomas or IDH mutant tumors and normal OPCs, the authors concluded that the profiles of PAs seemed more committed than those of the higher-grade gliomas, in line with their more indolent clinical course.

Unsupervised analyses revealed that PA cells express a wide spectrum of developmental signatures. A majority of cells express a mixture of more mature astrocyte-like (AC-like) signatures and less MAPK pathway signaling, with a wide range of proportions of one or the other signature. A minority of cells with a high MAPK expression program was associated with higher expression of BRAF transcripts. Cells displaying a MAPK expression program comprised a higher proportion of cycling cells and expressed higher levels of progenitor-associated transcription factors such as SOX2. The authors hypothesized that these may reflect an OPC-like cell population that gives rise to an AC-like population, similar to normal development. Activated BRAF signaling associated with a high MAPK gene expression program seemed to give rise to two cell populations with mutually exclusive expression programs: one exerting an oncogene-induced senescence expression phenotype and the other a proliferative phenotype. The authors speculate that the dosage of MAPK signaling may dictate cell fate, potentially with other cellular factors that together may provide an opportunity for therapeutic intervention.