WNT and CDK4 inhibition might provide a potential therapeutic strategy for difficult-to-treat epithelial 35 tumors. 36 37 42 regression in animal models [7][8][9][10][11][12] . We recently showed that restoring endogenous regulation of the WNT 43 pathway by re-expressing normal levels of APC, is sufficient to induce disease regression in established 44 colorectal cancers 8,9 . 45 Efforts to target WNT signaling pharmacologically have explored multiple nodes of inhibition, 46 including blocking WNT ligand secretion and direct targeting the downstream transcriptional effector, 47 b-catenin 13,14 . We reasoned that modulating WNT signaling through the same mechanism by which 48 APC acts would provide the best opportunity for well-controlled WNT regulation. While it is not 49 feasible to genetically restore APC in human CRCs, it is possible to reinforce the activity of the 50 APC/AXIN/b-catenin destruction complex by increasing the levels of the scaffold proteins AXIN1/2. 51 Tankyrase enzymes (TNKS and TNKS2) are functionally-redundant members of the poly-ADP ribose 52 polymerase (PARP) family that mediate the PARsylation and degradation of AXIN1/2, and thus, TNKS 53 inhibition stabilizes AXIN and reinforces endogenous suppression of WNT signaling. Numerous small 54 molecules have been described that enable selective TNKS1/2 blockade and WNT pathway suppression 55 in CRC cell lines 15-18 . 56 Despite their ability to dampen hyperactive WNT and suppress tumor growth in model systems, TNKS 57 inhibitors do not always show anti-proliferative effects as single agents in human CRC cell lines 17,19,20 . 58 However, when mitogenic signaling is limited by the reduction of serum in the culture media, TNKS 59 inhibition can block growth and proliferation of otherwise resistant cells 16,21 . This suggests that 60 inhibition of parallel mitogenic signals may synergize with TNKS inhibitors to control tumor cell 61 growth. Indeed, blocking MEK, PI3K, or EGFR can enhance the anti-tumor effect of TNKS 62inhibitors 17,19,20 . Here we perform a domain-focused kinome CRISPR library screen 22 , aiming to identify 63 the most critical mitogenic signaling components that mediate resistance to TNKS inhibition, with the 93 We next introduced a domain-focused kinome library targeting 482 human kinases (~6 sgRNAs/gene) 22 94 into each LC9 line with an average representation of 2000x (Fig 1e). Analysis of gRNA abundance in 95 flow-sorted (GFP-positive) populations showed a high concordance between replicates with more than 96 99% of expected sgRNAs represented in each of the pools and good correlation across biological 97 replicates (Tables 1-4; Supplementary Fig 2c). Following expansion for 7 days to deplete sgRNAs 98 targeting essential genes, each population was treated with either DMSO or 1uM XAV for 14 population 99 doublings and relative abundance of individual sgRNAs was compared to post-sort (pre-treatment) 100 samples (Fig 1e). As expected, control, neutral sgRNAs showed no significant change in abundance 101 (Fig 1f and Supplement...