An iPSC-Based Model Of MDS For Phenotype-Driven Gene and Drug Discovery

Despite past and ongoing research efforts, the pathogenetic mechanisms of MDS remain far from understood sufficiently to point to single genes or molecular pathways as putative targets for generation of better mouse models or drug development. In lack of clear targets, the manipulation of disease-associated phenotypes may at present be the best opportunity for disease study and therapeutic intervention. Furthermore, because cellular phenotypes are more proximal to molecular disease mechanisms than phenotypes seen at the level of the tissue or organism (i.e. the murine or human hematopoietic system), they may provide more sensitive and relevant readouts of the disease process.

With the advent of patient-specific induced pluripotent stem cells (iPSCs), disease models based on cellular phenotypes (“disease-in-a-dish”) can now be developed and their unique properties promise to revolutionize disease study and drug development. Unlimited cell numbers of biologically relevant cells can be obtained relatively easily and cost-effectively. Here we present a new MDS model based on patient-specific iPSCs and its use in two phenotype-driven screens: (a) a focused genetic screen and (b) a high-throughput chemical screen.

We have derived multiple MDS-iPSC lines with deletions of chromosomes 7q or 20q (characteristic chromosomal deletions that we are using as markers of iPSCs derived from the MDS clone) from bone marrow (BM) of 3 patients with MDS and 1 patient with sAML. We also derived isogenic karyotypically normal iPSC lines in parallel from the same BM samples. We identified two cellular phenotypes specific to the MDS-iPSCs: decreased proliferation rate and decreased potential for pan-hematopoietic differentiation and reduced clonogenic capacity of their hematopoietic progeny. These phenotypes are consistent across multiple iPSC lines from different patients and absent from their isogenic normal iPSC controls. They are reminiscent of the behavior of ex vivo cultured primary MDS BM cells and are therefore likely to be relevant to the disease process. Furthermore, they are rescued by spontaneous compensation for chr7q dosage through acquisition of an extra chr7 and recapitulated by the engineering of artificial chr7q deletions in normal iPSCs.

To identify critical MDS gene(s) on chr7q, we performed a screen of 62 candidate haploinsufficient genes (with reduced expression by at least 1.5-fold in our del(7q)-iPSCs compared to their isogenic normal iPSCs) for rescue of the proliferation and/or hematopoietic differentiation phenotype. We constructed a lentiviral library of all candidate ORFs (and 14 additional alternative transcripts) linked to eGFP through a P2A peptide and each tagged with a unique 4-nt barcode sequence to its 3’ UTR. The library was packaged as a pool and transduced into two different del(7q)-iPSC lines. The cells were passaged for 16 weeks and gDNA was isolated every 2 weeks. In parallel, the cells were differentiated along the hematopoietic lineage and gDNA was isolated from CD45⁺ cells FACS-sorted on day 15 of differentiation. High-throughput sequencing of the barcodes identified 5 ORFs that became enriched in undifferentiated iPSCs over time (rescue of proliferation) and 4 ORFs enriched in sorted CD45⁺hematopoietic progenitors (rescue of hematopoietic differentiation), 2 of which overlapped. All 9 hits are being further validated in more focused screens.

Second, we used the same platform for a high-throughput small molecule screen. We optimized the plating conditions and densities for a 384-well format using a luminescent cell viability assay and conducted a screen of 2000 compounds comprising known drugs, natural products, and other bioactives and chemicals, in an MDS-iPSC line (2.13) and its isogenic normal control (2.8). Primary hits were defined as compounds that enhanced the growth of the MDS-iPSC line, but not of the control normal iPSC line in a compound dose-dependent manner. 38 primary hits were retested in a dose-response survival assay in one additional MDS- (2.A1-3), one additional isogenic normal iPSC line (2.12) and one sAML iPSC line over 8 concentrations and 12 compounds were prioritized for further studies.

Our data highlight the potential of this new iPSC-based model of MDS for screens to identify genes or compounds that affect cellular phenotypes by acting on previously undefined targets.