Large-Scale In Vivo Zebrafish Drug Screen Uncovers Novel Insights into Leukemia Stem Cell Biology

Relapsed T-cell Acute Lymphoblastic Leukemia (T-ALL) patients have a dismal prognosis, with 5-year overall survival rates of 10% for adults and 36% for children. Relapse is attributed to chemotherapy’s inability to target Leukemia Stem Cells (LSCs), that have the potential to re-populate the cancer. LSCs exists with a unique behavior compared to the rest of the leukemia, finding targeted therapies toward LSCs is critical for disease eradication and elimination of relapse. However, investigating LSC is particularly challenging in T-ALL for two reasons 1) The lack of well-established surface markers that distinguish LSCs and 2) LSCs exist at very low frequencies in T-ALL murine models and patient samples.

Here, we used a transgenic model of rag2-myc derived T-ALL in zebrafish that offers solutions to these challenges. Our lab had previously generated leukemia lines from this model with LSC frequency of ~ 10%, compared to <0.01% in murine models, generating an ideal environment to interrogate T-ALL LSC biology and explore specific targets. We utilized this model to screen over 770 FDA-approved drugs in vivo using >2,500 syngeneic CG1 zebrafish for their effect on self-renewal. Further secondary screening in human T-ALL cells narrowed down the potential hits into 4 compounds that specifically interfere with self-renewal. Finally, the hit drug Amiloride was confirmed by an in vivo limiting dilation assay (LDA), resulting in a more than 4-fold decrease in the LSC frequency (p=0.0026). The top hit, Amiloride, which reduced the frequency of LSCs in zebrafish and human T-ALL cells, is an inhibitor of the Sodium Hydrogen Exachanger-1 (NHE1). Notably, NHE1 has not been previously linked to self-renewal in hematologic malignancies, presenting a novel therapeutic strategy to inhibit LSCs.

To validate the NHE1 as the target responsible for the effect on self-renewal by Amiloride, we used shRNA knock down (KD) of NHE1. We found more than 70% decrease (p< 0.0001) in self-renewal in vitro with the KD cells compared to the scrambled control. Additionally, there was a concurrent decrease in gene expression, assessed by RT-PCR, of Nanog (p< 0.0001), CD7 (p< 0.0001) and CD44 (p= 0.0079), among other self-renewal genes. Further, the inhibition of the NHE1 by KD studies and pharmacological treatment results in a significant G1 cell cycle arrest. Bulk mRNA sequencing revealed a down-regulation in the Myc Target genes, KRAS, Hedgehog and Notch signaling in KD cells compared to the control, all critical pathways in leukemia self-renewal. Interestingly, the EphA3 gene was among the top down-regulated genes (FDR=2.01*10^-25), EphA3 is expressed by LSCs and involved in maintaining their leukemogenicity. Collectively, these data suggest a central role for the NHE1 in T-ALL self-renewal.

Inhibiting the NHE1 impairs mitochondrial function in cancer cells, offering a possible mechanism for the effect on LSCs. In human T-ALL cells, we found that Amiloride significantly reduced mitochondrial basal respiration (p=0.0074) and ATP production (p=0.0008). When probing for different genes associated with mitochondrial processes, PRKN expression was constantly upregulated by more than 8-folds in Amiloride-treated cells, compared to DMSO control (p< 0.0001). Parkin is an E3 ubiquitin ligase and a well-established mediator of mitophagy, which led us to investigate the impact of Amiloride on mitochondrial morphology. We found that Amiloride treatment reduced mitochondrial footprint (p< 0.0001) and mitochondrial branching (p=0.0028) in T-ALL cells. The inhibition of the exchanger by KD studies revealed similar mitochondrial alterations including decreased ATP production, basal respiration, spare capacity, decreased mitochondrial mass, and decreased mitochondrial membrane potential. Further, Gene set enrichment analysis identified a downregulation in the OXPHOS pathway. Importantly, we found that high mitochondrial content was significantly associated with self-renewal in T-ALL cells, suggesting that LSCs may be highly sensitive to metabolic disruption.

Overall, in this work we capitalized on a transgenic model of zebrafish T-ALL with high frequency of LSC to identify the physiologically accurate hit, Amiloride. In addition, we are the first to describe its target, the NHE1 as a vulnerability of self-renewal in hematological malignancies, potentially through regulating the cellular energy homeostasis.