In Vivo Screening for Tumor Suppressors In Hematological Malignancies Using Barcode RNAi

Abstract 998

Chromosomal deletions are frequent cytogenetic defects in hematological malignancies. Common deleted regions (CDRs) defined by the minimal physical overlap of all deletion events are thought to harbor tumor suppressors relevant for pathogenesis. Haploinsufficiency is likely to be a common consequence of chromosomal deletions affecting the function of tumor suppressors within the deleted locus. The use of RNA interference (RNAi) offers a unique opportunity to mimick haploinsufficiency by partial knock-down of the gene transcripts. Since deletions are usually large containing a high number of candidate genes, we aimed to develop a screening method targeting several candidates at once and assaying for tumor suppressor features in a pool of knock-downs. As a model for our approach we selected a CDR on chromosome 20q frequently found clonal in myeloid diseases. We established a screen for knock-downs causing cytokine hypersensitivity. This CDR (physical position chr20:38.7- 42.2) spans 3.5Mb and contains 16 genes (MAFB-JPH2). We cloned 3 short hairpin RNAs (shRNAs) for each mouse homologue of the 16 target genes into the pLKO.2 lentiviral vector. The 48 shRNA constructs were “bar-coded” with different, unique 24bp DNA bar-codes, that can be identified and quantified using the microbead-based xMAP technology (Luminex). We lentivirally delivered the constructs independently into the erythropoietin (Epo) dependent murine cell line Baf3/EpoR, pooled the individual knock-downs in equal amounts and assayed for cytokine hypersensitivity based proliferation advantage under stringent Epo concentrations. Applying a scoring system based on the relative increase/decrease of the individual bar-codes in the pool over time, we could identify the knock-down of topoisomerase 1 (Top1) to induce Epo-concentration dependent outgrowth. Two different Top1 shRNA constructs succeeded in consistently mediating proliferative advantage in three biological replicates. We set up a validation experiment pooling Top1 knock-down cells with control cells (bar-coded, no shRNA) in a 1:1 ratio and could observe significant dominance of the Top1 knock-down cells establishing after 10–15 days in culture and increasing over time (p<0.0001). Knock-down efficiency of both successful constructs measured by qPCR and was consistently between 30–60% of control cell mRNA expression. Based on the convincing cell line data we set up an in vivo validation in a competitive repopulation mouse model. We established a protocol for lentiviral transduction of murine lineage depleted (lin-) bone marrow progenitor cells. Transplanting in a 1:1 ratio of Top1 knock-down and control progenitors into three lethally irradiated mice, we surprisingly observed an outcome opposite to the cell line results. Top1 knock-down progenitors showed a clear disadvantage in repopulation capacity, being underrepresented in the peripheral blood at week 3 post-transplantation and furthermore fully outcompeted by the control cells around week 15 post-transplantation. Based on the unreliability of cell line models in our setup we repeated the screen with the full library (48 shRNAs) in vivo. Two different constructs targeting phospholipase C gamma 1 (Plcg1) had the top score in 3 out of 5 transplanted mice, resulting in an impressive cumulative score. Showing equal representation with the other knock-downs (approximately 2%) in the donor pool, Plcg1 knock-down cells represented up to 13% of the peripheral blood cells at week 3 post- transplant, increasing to up to 29% at week 7 post-transplant. Our results suggest that cell lines often might not be the proper model for studying growth regulation. Further, our in vivo screen revealed Plcg1 as a promising candidate with a tumor suppressor function in hematopoiesis.