Acute Myeloid Leukemia Cells Rely on the Glycolytic Enzyme PGK1 to Support Energy Production and Amino Acid Metabolism

Acute myeloid leukemia (AML) cells frequently display an increased dependence on glucose compared to healthy hematopoietic stem and progenitors (HSPCs). Furthermore, increased glucose metabolism is associated with worse outcomes in AML. However, the upstream regulators that drive glucose metabolism as well as how glucose is utilized in AML remains largely unknown. Here, we have discovered that the oncogenic transcription factor (TF), JUN supports AML, in part, by regulating glycolytic enzymes that utilize glucose to support both cellular energetics and amino acid metabolism.

Having previously reported that certain aggressive AML sub-types depend on JUN, we next set out to identify JUN-regulated transcriptional programs that support AML. To do so, we engineered the human AML cell line, THP-1 to express either control (shNT) or JUN-targeting shRNAs and then performed RNA-seq on cells from each condition. Next, we carried out a functional enrichment analyses and discovered that genes encoding numerous enzymatic regulators of glycolysis were significantly down-regulated by JUN-inhibition. Using qPCR validation, we confirmed that JUN inhibition results in reduced mRNA expression of Hexokinase 1 & 2 (HK1 & HK2), Glucose-Phosphate Isomerase (GPI), Phosphofructose Kinase (PFKP), Aldolase A (ALDOA), Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH), Phosphoglycerate Kinase (PGK1), Enolase 1 (ENO1) and Pyruvate Kinase Muscle (PKM). ChIP assays confirmed that JUN directly localizes to the promoters of each of these genes except for ALDOA and PKM. To assess whether the JUN-dependent changes in glycolytic gene expression translated to defects in the ability of AML cells to catabolize glucose, we performed a glycolysis stress test on shNT- and shJUN-expressing human AML cell lines (e.g. THP-1, NOMO-1 and MOLM-14). From this assessment, we found that JUN inhibition resulted in a complete loss of glucose catabolism in all cell lines tested.

We postulated that JUN activates glycolytic gene expression to support AML and thus, inhibition of JUN-regulated glycolytic genes will impede leukemia cell function. To test this hypothesis, we generated shRNAs targeting each of the JUN-regulated glycolytic enzymes. We then engineered mouse leukemia cells expressing the human leukemogenic allele, MLL-AF9 - found in both childhood and adult AML - to express these shRNAs and then assessed their colony forming capacity (CFC) in cytokine-enriched semisolid culture. In parallel, we also assessed how knockdown of each glycolytic enzyme impacted the CFC of healthy HSPCs. Notably, we observed that inhibition of each enzyme significantly impaired the CFC of MLL-AF9 leukemia cells but that inhibition of Pfkp, Aldoa or Pgk1 did not impact the CFC of healthy HSPC, suggesting that these enzymes were selectively important for AML. We next evaluated how knockdown of Pfkp, Aldoa or Pgk1 impacted leukemia propagation in vivo. Interestingly, inhibition of Pfkp or Pgk1, but not Aldoa, significantly delayed the onset of MLL-AF9 driven leukemia in vivo. Moreover, we found that a chemical inhibitor of PGK1, CBR-470-1, reduces the CFC of patient-derived AML cells in vitro.

Since Pgk1 had not previously been studied in AML, we focused our attention on the its role in leukemia cell metabolism. We thus performed untargeted metabolomics of shNT- and shPgk1-expressing mouse MLL-AF9 cells using hydrophilic liquid chromatography mass spectrometry (HILIC-MS). PGK1 utilizes ADP to convert 1,3-bisphosphoglycerate into 3-phosphoglycerate (3-PG) and ATP. 3-PG can subsequently be used to generate phosphoenolpyruvate (PEP) or in the de novo synthesis of serine (Ser). As expected, we found that Pgk1 inhibition resulted in a significant decrease in 3-PG, PEP and Ser. We also observed that Pgk1 inhibition resulted in the decrease of several tricarboxylic acid cycle (TCA) intermediates that support oxidative phosphorylation energy production. Surprisingly, we also found that Pgk1 inhibition resulted in a marked decrease in several amino acids, including Alanine, Arginine and Histidine. We are currently investigating whether the decrease in these amino acids mediated by Pgk1 inhibition is due to an impediment of their biosynthesis or their consumption in gluconeogenic pathways.

Collectively, our studies identify JUN as a previously unrecognized regulator of AML cell metabolism and shed new into how glucose is catabolized in AML cells.

No relevant conflicts of interest to declare.