Uridine Synthesis Is the Metabolic Vulnerability in Relapse-Associated B-ALL Cells with Active Pre-BCR Signaling

Relapse stubbornly continues to limit survival for children with B-cell progenitor acute lymphoblastic leukemia (B-ALL). B-ALL cells accumulate at the developmental checkpoint regulating the transition from pro-B to pre-B cells and express the pre-B cell receptor (preBCR). B-ALL samples have previously been categorized as preBCR positive and preBCR negative subtypes. We previously found that pre-B cells with active preBCR signaling are present at diagnosis and highly predictive of relapse, termed relapse predictive cells (RPCs) (Good et al. Nature Med 2018). Whole transcriptome sequencing of pre-B RPCs from patient samples demonstrated enrichment of gene signature pathways is involved in oxidative phosphorylation (OXPHOS), pyrimidine metabolism, and glycolysis pathways compared to pre-B cells from patients in complete remission. Thus, we hypothesized that RPCs have unique metabolic demands driven by their active signaling.

Utilizing TARGET dataset, B-ALL patients were classified as preBCR high or low based on the sum of normalized counts of preBCR components and signaling molecules. Notably, preBCR^high patients (top 25%ile) had inferior overall survival compared to preBCR^low (bottom 75%ile; P=0.009). Further, preBCR^high patients demonstrated enrichment for the metabolic gene signatures identified in RPCs. CyTOF analysis in primary B-ALL samples demonstrated increased expression of glycolysis (PFKFB4, ENO1, LDHA) and pentose phosphate pathway (G6PD, PGD) proteins in preB cells from patients destined to relapse compared to those in continued remission. These data support the link between a distinct metabolic state in preB cells associated with relapse.

To investigate the metabolic state associated with the preBCR in B-ALL, we utilized cell line models (preBCR^high: Nalm6, 697, Kasumi2; preBCR^low: REH, Nalm16, RS4;11). We found that preBCR^high cells are more metabolically active than preBCR^low cells with significantly higher oxygen consumption rate and extracellular acidification rate. To identify their targetable dependencies, cells were starved of glucose or glutamine for 24 hours. Neither cell type was sensitive to glutamine starvation. However, preBCR^high cells were uniquely vulnerable to glucose deprivation (killing effect 64% vs 25%, P=0.027) indicating a distinct need for glucose in preBCR^high cells.

To unravel the unique glucose dependency of preBCR^high cells, we performed isotype tracing with U-¹³C-glucose and U-¹³C-glutamine. Both preBCR^high and preBCR^low cells use glutamine as the primary carbon source in the TCA cycle. However, glucose is primarily utilized for the pentose phosphate pathway and nucleotide synthesis in preBCR^high cells, as evidenced by higher fractional ¹³C labeling in m+5 UTP, UDP and ATP. Uridine, but not other pyrimidines or purines, rescued preBCR^high cells from glucose starvation, highlighting glucose-dependent uridine synthesis as the metabolic vulnerability of preBCR^high cells. To target this dependency, we treated the cells with the dihydroorotate dehydrogenase (DHODH) inhibitor (BAY-2402234), which disrupts a crucial step in uridine synthesis. Remarkably, DHODH inhibition resulted in preferential killing in preBCR^high but not preBCR^low cells. Evaluation of the in vivo efficacy of DHODH inhibition in xenografts will be presented.

Together, we found that preBCR^high B-ALL cells are associated with relapse and are highly energetic, relying uniquely on glucose for survival to fuel the PPP and uridine synthesis. Inhibition of uridine synthesis is a novel strategy to target these treatment resistant and relapse-associated cells.