21The appearance of genetic changes in human pluripotent stem cells (hPSCs) presents a 22 concern for their use in research and regenerative medicine. Variant hPSCs harbouring 23 recurrent culture-acquired aneuploidies display growth advantages over wild-type 24 diploid cells, but the mechanisms yielding a drift from predominantly wild-type to 25 variant cell populations remain poorly understood. Here we show that the dominance 26 of variant clones in mosaic cultures is enhanced through competitive interactions 27 resulting in elimination of wild-type cells. This elimination occurs through corralling 28 and mechanical compression by faster growing variants, causing a redistribution of F-29 actin and sequestration of YAP in the cytoplasm that induces apoptosis in wild-type 30 cells. Importantly, YAP overexpression in wild-type cells is sufficient to alleviate their 31 loser phenotype. Our results demonstrate that hPSC fate is coupled to mechanical cues 32 imposed by neighbouring cells and reveal that hijacking this mechanism allows variants 33 to achieve clonal dominance in cultures. 34 35 Keywords: human pluripotent stem cells, culture acquired variants, cell competition, 36 YAP 37 38 wild-type cells in mosaic cultures, akin to the super-competition-like behaviour 89 described in other cell types (de la Cova et al., 2004; Moreno and Basler, 2004). The 90 elimination of loser cells in hPSC cultures is exerted through mechanical cues, and is 91 mediated by YAP, downstream of the actomyosin cytoskeleton. Our findings illuminate 92 the reliance of hPSC fates on their mechanical environment and highlight the need for 93 consideration of culture space limitations in the scale up of hPSCs for research or 94 clinical use. 95 96 6 Results 97 Variant hPSCs selectively eliminate diploid wild-type counterparts from co-98 cultures 99 To uncover the reasons behind the rapid overtake of cultures by genetically variant 100 hPSCs (Olariu et al., 2010), we sought to examine how wild-type and genetically 101 variant hPSCs interact and whether they affect each other's growth. To this end, we 102 initially used two diploid H7 sublines (either non-modified or genetically engineered to 103 constitutively express red fluorescent protein (RFP), termed wild-type and wild-type-104 RFP, respectively), and their aneuploid variant harbouring a gain of chromosomes 1, 105 12, 17q and 20q CNV, and stably expressing green fluorescent protein (GFP) (termed 106 variant-GFP). Time-lapse microscopy of co-cultures containing wild-type-RFP and 107variant-GFP cells showed a selective elimination of wild-type-RFP cells during a three-108 day culture period (Video S1). To establish that the observed elimination is due to the 109 presence of variant cells in mixed cultures we compared the growth rates of wild-type-110 RFP or unlabelled wild-type cells in separate culture to how they grew in mixed cultures 111 with variant-GFP cells. Wild-type sublines were viable and created well-established, 112 large colonies in separate culture, but consistent with prev...