Epigenetic Control of Fatty-Acid Metabolism Sustains Glioma Stem Cells

Glioblastoma, the most common primary brain tumor, is among the most lethal forms of cancer. Despite aggressive treatment strategies, the 5-year survival rate remains at only 5% (1). The presence of glioma stem cells (GSC), functionally defined by their ability to self-renew, to initiate tumors in vivo, and to recapitulate tumor cell heterogeneity, is thought to contribute to glioblastoma pathogenesis and therapeutic resistance (1). Thus, there is impetus to develop novel therapies that effectively target GSCs to treat glioblastoma. In this issue of Cancer Discovery, Gimple and colleagues developed a novel therapeutic approach targeting GSCs from interrogation of superenhancer profiles, which revealed epigenetic regulation of fatty-acid metabolism (2).

The authors hypothesized that novel vulnerabilities in GSCs could be identified by studying their epigenomic features. They focused on superenhancers, large clusters of enhancers at which transcription factors and coactivators are heavily recruited, because of their important roles in regulating the expression of cell identity and disease genes (3). Gimple and colleagues proposed that identifying GSC-specific superenhancers could point toward genes or pathways required for the maintenance of GSCs that could then be therapeutically exploited.

To approach this question, the authors developed a bioinformatics pipeline to identify superenhancers that were unique to glioblastoma and GSCs. Histone H3 lysine 27 acetylation (H3K27ac), a mark of active enhancers, was analyzed by chromatin immunoprecipitation sequencing (ChIP-seq) in 10 surgically resected glioblastoma samples and 15 normal brain tissues, as well as in patient-derived GSCs and matched differentiated glioma cells (DGC). In this analysis, these samples were found to cluster by disease status (glioblastoma versus normal) and differentiation (GSC versus DGC). From 148 candidate superenhancer-associated genes initially identified, genes of interest were further prioritized by generating scores based on mRNA expression, the presence of H3K27ac peaks in GSCs compared with DGCs, and the ability to predict poor survival of patients with glioblastoma. The top-ranked genes that emerged were WSCD1, ELOVL2, and KLHDC8A.

Gimple and colleagues next sought to validate that the identified enhancers were responsible for regulating expression of these genes. Initial insight into whether each gene is superenhancer regulated was gained through the use of the BET bromodomain inhibitor JQ1. JQ1 selectively suppresses the expression of superenhancer-regulated genes because superenhancers are highly enriched for and dependent on the BET protein BRD4 (3). JQ1 treatment decreased expression of both ELOVL2 and KLHDC8A in GSCs. ELOVL2 was selected for further study because of the strong relationship between its expression and glioblastoma prognosis, its responsiveness to JQ1, and the potential to target its enzymatic function. To specifically determine whether the identified superenhancer regulates ELOVL2, the authors leveraged ChIP-seq data to show that the stem-cell transcription factors SOX2 and OLIG2 bound within a 500-bp region in the superenhancer (Fig. 1). SOX2-mediated silencing suppressed ELOVL2 gene expression. Moreover, targeting of the locus at which SOX2 binds, using the KRAB transcriptional repressor tethered to endonuclease dead Cas9 (dCas9), reduced ELOVL2 expression and proliferation in GSCs, functionally confirming the relevance of the enhancer.

ELOVL2 is an endoplasmic reticulum transmembrane protein that functions in long-chain polyunsaturated fatty acid (LC-PUFA) metabolism by elongating arachidonic (C20:4n-6) and eicosapentaenoic (C20:5n-3) acids (Fig. 1). LC-PUFAs exert signaling functions and are needed for producing membrane phospholipids where they affect membrane organization (4). Elevated levels of docosapentaenoic acid (C22:5n-6) and docosahexanoic acid (DHA; C22:6n-3), LC-PUFAs that are generated downstream of ELOVL2, were observed in GSCs as compared with DGCs, consistent with ELOVL2 expression. Functionally, ELOVL2 depletion decreased cell proliferation, promoted apoptosis, and reduced features of stemness in patient-derived GSCs, while having minimal effects on DGCs. Importantly, the effects of ELOVL2 knockdown were at least partially restored by supplementing its LC-PUFA products. ELOVL2 knockdown also resulted in changes in phospholipid profiles and altered membrane structural features, as assessed by fluorescence recovery after photobleaching assays. Altogether, these results reveal the importance of ELOVL2 in maintaining cell membrane composition and structure in GSCs.

Hypothesizing that altered membrane dynamics might impact cell signaling, Gimple and colleagues next queried reverse-phase protein array datasets to determine which signaling pathways correlated with ELOVL2 expression, finding a positive relationship with EGFR phosphorylation. Suppression of ELOVL2 either by RNAi or by targeting its superenhancer via dCas9-KRAB resulted in impaired EGFR membrane localization and signaling. Supplementation of the ELOVL2 LC-PUFA products produced a partial rescue of EGFR signaling. In addition, the proliferation defect mediated by ELOVL2 silencing could be blocked by EGFR overexpression. Thus, ELOVL2-dependent LC-PUFA synthesis was shown to sustain EGFR signaling and proliferation in GSCs (Fig. 1).

After elucidating the role and epigenetic regulation of ELOVL2 in GSC maintenance and proliferation, the authors sought to develop a therapeutic approach leveraging their new insights into the role of this LC-PUFA pathway in GSCs. Because there are no inhibitors that target ELOVL2, they chose to evaluate SC-26196, a small-molecule inhibitor of FADS2, the enzyme that desaturates the ELOVL2 products (Fig. 1). FADS2 is also upregulated in glioblastoma tissue and GSCs over normal brain and DGCs; thus, the authors hypothesized that the ELOVL2–FADS2 pathway could be targeted in glioblastoma. FADS2 loss by knockdown or treatment with SC-26196 inhibited cell proliferation and decreased EGFR signaling, phenocopying ELOVL2 deficiency. Having shown the importance of ELOVL2-dependent generation of LC-PUFAs for EGFR signaling, the authors next investigated the potential to combine FADS2 inhibition with the EGFR inhibitor lapatinib (Fig. 1). This combination was found to be synergistic in targeting GSCs in vitro and to improve survival of mice implanted orthotopically with GSCs that had been treated with this combination. Administration of the drug combination into mice after the implantation of GSCs resulted in a trend toward improved survival, but also substantial toxicity. The data suggest that development of inhibitors specifically targeting the ELOVL2–FADS2 pathway has potential to improve glioblastoma treatment.

This elegant study links epigenetic alterations to control of metabolic enzyme gene expression and thereby to the maintenance of growth factor receptor signaling in GSCs. Although much effort has gone into elucidating the roles of oncogenes and tumor suppressors in the regulation of cellular metabolism in cancer (5), substantially less is known about how epigenetic changes acquired in cancer cells contribute to the remodeling of cellular metabolism that sustains tumor growth. The findings of this study suggest that epigenetic alterations may play unique roles in driving tumor-specific metabolic features, allowing identification of metabolic nodes that can be exploited to selectively target cancer cells. Reciprocally, metabolic alterations are known to affect the epigenome. In glioma, IDH1 and IDH2 mutations result in production of the oncometabolite (R)-2-hydroxyglutarate, which inhibits α-ketoglutarate–dependent enzymes, including TET methylcytosine oxidases and Jumanji-C domain histone demethylases, resulting in hypermethylation profiles (5). AKT signaling also promotes the production of acetyl-CoA, the acetyl donor for histone acetylation, and pAKT-S473 and histone H4 acetylation levels positively correlate in human gliomas (6). It remains poorly understood, however, how changes in acetyl-CoA availability downstream of oncogenic signaling or nutrient availability affect histone acetylation at enhancers. Further elucidation of the reciprocal interplay between cellular metabolism and the epigenome promises to continue to be a fruitful area for further research.

This work also builds on prior studies that have identified roles for dysregulated fatty-acid metabolism and altered phospholipid composition in glioblastoma (7). Fatty-acid synthase (FASN) is upregulated in glioma and is associated with poor prognosis (8). EGFR signaling triggers activation of the transcription factor SREBP-1, which drives expression of de novo lipogenesis genes (9). Although many prior studies have emphasized de novo fatty-acid synthesis in cancer cells, this study highlights the importance of metabolism of essential fatty acids. Mammalian cells lack the ability to introduce double bonds beyond carbons 9 and 10 in fatty acids; thus, LC-PUFAs must be either ingested from the diet or synthesized from essential shorter chain PUFAs (18:3n-3 and 18:2n-6). Intake of LC-PUFAs can vary substantially with diet, with vegan diets for example lacking DHA (C22:6n-3; ref. 10), raising the question of whether modulating dietary intake of LC-PUFAs could offer another means of modulating membrane fluidity in cancer cells that are particularly reliant on these fatty acids.

In summary, this study by Gimple and colleagues elucidated a GSC-specific enhancer in control of LC-PUFA synthesis, a metabolic point of leverage in glioblastoma due to its importance in the maintenance of membrane dynamics and growth factor receptor signaling. With this knowledge, Gimple and colleagues explored an innovative combination therapy that targets GSCs for the treatment of glioblastoma. Their work demonstrates the power of probing the epigenetic landscape of tumors to gain deep mechanistic insights into the molecular processes on which cancer cells rely, with the ultimate goal of identifying improved strategies to treat cancer and improve the outlook for patients faced with the diagnosis of a lethal cancer such as glioblastoma.

No potential conflicts of interest were disclosed.

H.C. Affronti is supported by an NIH Pre-doctoral to Postdoctoral Transition Award 4K00CA212455-03. The lab of K.E. Wellen is supported by NIH grants R01DK116005, R01-CA174761, and R01-CA228339.