P423: ADAPTING CRISPR CAS9 DROPOUT SCREENS TO IN VIVO PDX MODELS OF ACUTE LEUKEMIAS

Background: Acute leukemias require better treatment and targeted therapies represent interesting future therapeutic options. Such therapies precisely inhibit molecules with essential function, also called vulnerabilities or dependencies, so that leukemia cells die once the target is inhibited. Aims: Here, we aimed at identifying therapeutic targets on a patient individual level and in the surrounding of a living organism. Towards this aim, we established CRISPR Cas9 dropout screens in PDX models of acute leukemias in vivo. Methods: Primary patient acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) samples were transplanted into immunocompromised NSG mice and re-passaged to develop serially transplantable PDX models. To overcome limited transduction efficiency inherent to PDX leukemia cells, lentiviruses were used. Results: We first elaborated on the maximum library size to be used in PDX models in vivo. As only a limited number of PDX cells home into mice, library size is limited to ensure the coverage of the library. We used, screening sgRNAs as genetic barcodes in Cas9 negative samples to determine the maximum library size upon next generation sequencing (NGS); a regular distribution of all sgRNAs recovered from mice at the end of the experiments indicated a suitable library size. We found that PDX ALL models tolerated a larger library compared to PDX AML models that might reflect lower intra-sample heterogeneity and higher leukemia stem cell frequency in ALL compared to AML. To perform knockout screens, a split-construct for Cas9 was used to reduce plasmid size and optimize lentiviral transduction efficiency into PDX cells; concomitantly split-GFP was used to enrich cells expressing both Cas9 split plasmids by flow cytometry. A customized library consisting of 146 target genes combined with positive and negative control genes was designed at 5 sgRNAs per gene using the CLUE platform (www.crispr-clue.de) (Becker et al., Nucleic Acids Res. 2020) and cloned into a lentiviral vector backbone. The sgRNA library vectors include either a puromycin-resistance cassette or an H-2Kk surface marker, each combined with a BFP fluorescent marker. This allows to determine transduction efficiency by FACS-based detection of BFP and to enrich the transduced PDX cells either by puromycin selection or by H-2Kk-MACS selection. Input samples were collected after library transduction and enrichment to >90% CRISPR/Cas9 sgRNA library positive cells, while remaining cells were injected into immunocompromised NSG mice, grown in vivo and re-isolated at advanced disease stage. Analysis of input and endpoint samples using NGS enabled to determine the frequency of sgRNAs and bioinformatically compare the abundance of different sgRNAs between the control and the samples of interest using MAGeCK (Model-based Analysis of Genome-wide CRISPR-Cas9 Knockout). A significant sgRNA dropout was characterized by a p-value below 0.05 and a FDR below 0.1. The screening experiments led to the top 10 depleted dropout genes for ALL and AML PDX samples. Besides individual dropouts of the different samples, also shared dropouts were detected. Especially sample overlapping dropouts might represent interesting starting points for new therapeutic options. Summary/Conclusion: In summary, we have established a CRISPR Cas9 screening pipeline, which allows investigating therapeutic targets on a patient-individual level in ALL PDX models and, for the first time, in AML PDX models in vivo.