CRISPR Dropout Screens Identify DHODH,PIK3C3, and Crkl as Potential Therapeutic Targets in Acute Myeloid Leukemia

Relapsed disease is still a major challenge in AML despite recently approved targeted therapies. New therapeutic targets are needed. To identify vulnerabilities in AML, screening data from a lentiviral-based genome-wide pooled CRISPR-Cas9 library (DepMap) was analyzed by examining gene knockout (KO) dependency scores in 15 AML cell lines (HEL, MV411, OCIAML2, THP1, NOMO1, EOL1, KASUMI1, NB4, OCIAML3, MOLM13, TF1, U937, F36P, AML193, P31FUJ). 438 gene KOs were identified as essential to AML cell survival in 14/15 AML cell lines. 247/438 (56%) genes had no active compound or approved drug interacting with its gene-products (DGIdb, canSAR Black) but were tractable targets based on 3D modeling or ligand-based druggability scoring (canSAR Black). This refined list of targets made up a new screening platform termed disKO. For clinical relevance, a primary AML specimen was treated with a custom-made lentiviral-based pooled CRISPR-Cas9 library including disKO targets using guide RNA's constructed against a parsed version of the druggable genome. Following transduction, PCR and oligo-sequencing were performed on Days 0, 7, and 14. Model-based Analysis of Genome-wide CRISPR/Cas9 Knockout (MAGeCK) was used to identify gene KOs associating with AML dropout over time. Four gene KOs from disKO were identified as essential in primary AML: NUP93, DHODH, CRKL, and PIK3C3. NUP93 loss of function mutations have been associated with steroid-resistant nephrotic syndrome in patients, suggesting inhibition may be toxic. DHODH inhibitors are currently being tested in AML. PIK3C3 codes for lipid kinases involved in autophagocytosis, and PIK3C3 inhibitors are in preclinical stage of development. CRKL is a proto-oncogene encoding src homology domains (SH2, SH3), which activate RAS and JNK signaling pathways. Although CRKL is a substrate of BCR-ABL CML, its essentiality in our study was found in AML cell lines and primary AML lacking the BCR-ABL fusion. Differential gene expression analysis showed CRKL to have lower expression in AML when compared to normal hematopoietic cells. Its druggability score based on six 3D structures and 1.3 ligand z-score indicates promising pharmacologic inhibition. In conclusion, a new CRISPR-Cas9 screening method termed disKO led to the identification of three viable targets for AML: PIK3C3, DHODH, and CRKL. Current efforts are focused on target validation followed by compound screening to find inhibitors with therapeutic potential.