A novel role for the extended MYC network in cancer cell survival

Abstract

Cancer cells exhibit altered metabolic requirements compared to their normal counterparts. This metabolic reprogramming is mediated by activation of oncogenes such as MYC. Here, we summarize our recent findings demonstrating a metabolic dependency of deregulated MYC on MLXIP-MLX, critical components of the nutrient-sensing arm of the extended MYC transcriptional network.

The proto-oncogene MYC is a mitogen-activated basic helix loop helix leucine zipper (bHLHZ) transcription factor that plays a major role in controlling and coordinating cellular metabolism with cell cycle progression and is deregulated in the majority of human cancers. MYC exists within a larger network made up of 3 Myc paralogs (MYC, MYCN, and MYCL1), their obligate heterodimerization partner MAX, 6 MAX-binding antagonists of MYC (MXD1, MXI1, MXD3, MXD4, MNT, and MGA), as well as the more distantly related Max-like protein X (MLX) and its glucose-responsive binding partners MLXIP (commonly known as MONDOA) and MLXIPL (commonly known as carbohydrate-response element binding protein [CHREBP]).¹ Although the role of MXD–MAX heterodimers as MYC antagonists is well characterized, much remains to be learned concerning how the individual transcription factors within the MLX arm of the network influence MYC function.

Our recent findings² that MYC overexpression enhances MONDOA–MLX expression and that loss of MONDOA or MLX is synthetic lethal (SL) in cells with either deregulated MYC or MYCN indicate that the MAX-binding and MLX-binding arms of the network are functionally interconnected and interdependent. To identify the molecular mechanism responsible for this SL phenotype we undertook a combined transcriptomic/metabolomic approach coupled with isotope-tracing and metabolic inhibitor experiments. Together, these data indicated that loss of MONDOA–MLX in neuroblastoma cells led to decreased expression of critical transporters and enzymes involved in glutaminolysis, gluconeogenesis, mitochondrial function, and de novo nucleotide and fatty acid synthesis. Notably, targeting gluconeogenesis or lipid metabolism, either by siRNA or chemical inhibitors, negatively affected the growth of MYC-driven cancer cells. Consistent with the hypothesis that the SL phenotype is due to altered metabolism, we were able to rescue both apoptosis induction and loss of viability with oleic acid, a monounsaturated fatty acid and the terminal metabolic product of the proteins encoded by the MONDOA-MLX target genes fatty acid synthase (FASN) and stearoyl-CoA desaturase (SCD).

We demonstrated co-occupancy of MYCN and MONDOA on the promoter of thioredoxin-interacting protein (TXNIP), a locus that contains 3 enhancer box or E-Box sequences capable of direct binding by MAX or MLX heterodimers. Although this supports direct transcriptional coordination between the 2 arms of the network, we did not determine global genome occupancy of MYC and MONDOA to directly evaluate the overlap between all of their transcriptional targets. Interestingly, genome-wide CHREBP target genes have been previously described and include critical regulators of glucose and lipid metabolism.³ Surprisingly, the number of genes identified is much lower than the number of MYC target genes, numbering in the hundreds rather than thousands. Since both MYC–MAX and MONDOA–MLX can recognize and bind to the same E-Box sequences in vitro, other factors such as the presence of non-canonical or multiple E-Boxes, epigenetic context, or absence/presence of other transcriptional modifiers may play an important role in determining which member of the network would bind to a specific promoter and the sum of these parts may dictate relative transcriptional output from a given promoter.

Importantly, other network members have been linked to modification of metabolism and cellular viability, including the MYC-antagonists MXD4 and MNT in T cells,⁴ MONDOA in B-cell acute lymphoblastic leukemia,⁵ and its homolog CHREBP in hepatoma and colon cancer cells.⁶ All of the aforementioned proteins have been reported to heterodimerize with MLX, suggesting that the MLX arm integrates distinct cues to modify the network's transcriptional output as required for something as dynamic as the regulation of metabolic flux. Therefore, analysis of the genome-wide binding profiles of these factors under different growth and nutrient conditions is highly warranted.

These experiments and the results obtained should also be interpreted in the light of MYC's propensity to invade the promoters and enhancers of already active genes when overexpressed.⁷ Is it possible that MONDOA–MLX and other member of the network may act as priming factors for invasion and subsequent transcriptional amplification when MYC is deregulated? Could MONDOA–MLX or MNT–MLX heterodimers respectively potentiate or attenuate the ability of deregulated MYC to amplify expression?

The tumor environment is drastically different from the normal cellular environment and is characterized by heterogeneity and variations in the level of growth factors, nutrients, and oxygen. Because MYC–MAX is mitogen-responsive, MONDOA–MLX is nutrient-sensing, and the MXD protein MNT is degraded under conditions of low oxygen or hypoxia,⁸ we hypothesize that the cellular response to these different conditions is transcriptionally influenced by the overall composition and relative activity of the extended MAX/MLX network in a cell type-specific manner (Fig. 1). Importantly, our recent review of the TCGA expression data indicated that multiple components of the network are simultaneously expressed in a wide range of both normal tissues and cancers, and that most tumor types show a specific expression pattern of MAX/MLX network members, arguing that each tissue has a distinct balance among members of the network.⁹ In support of this hypothesis, Myc elicits alternative metabolic responses when overexpressed in murine lung or liver tissues, which predominantly express either MondoA or Chrebp.¹⁰ Thus, evaluating the cell- or tissue-specific distribution and role of MAX/MLX network members, alone and in combination, will contribute to our understanding of both cancer cell metabolic flexibility and vulnerability.

Figure 1.

Transcriptional integration of extrinsic signals by the extended MYC network. The MAX/MLX network integrates various extrinsic cues, such as mitogen, oxygen, and nutrient levels. These factors alternatively activate or stabilize different MAX- and MLX-binding partners including mitogen-responsive MYC, hypoxia-degraded MNT, and the glucose-activated MLXIP (commonly known as MONDOA). These proteins compete for both heterodimer binding and occupancy of the common enhancer box (E-Box) DNA element. The final biological output of these signals is dependent upon the transcriptional activities of the network members, the expression of which varies in a cell-type and context-dependent manner (Top). Perturbation of the network composition, such as synthetic lethal loss of MONDOA in the presence of deregulated MYC, prevents MYC-driven metabolic reprogramming and results in apoptosis (Bottom).

Finally, our findings have 2 clear implications for the therapeutic treatment of MYC overexpressing malignancies. First, they suggest that agents that block lipid metabolism may be employed to selectively target cancer cells with deregulated MYC. Second, because, unlike Myc or Max knockouts, Mlxip, Mlxipl, and Mlx knockout mice are all viable, it is feasible to consider development of MONDOA-MLX inhibitors as a potential means of targeting malignant cells.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgements

PAC and DD would like to acknowledge R.N. Eisenman for discussion of and editing of this commentary.

Funding

The work from our laboratory cited in this review was funded by NIH NCI RO1 CA57138 to R.N. Eisenman.