bioRxiv preprint doi: https://doi.org/10.1101/2020.07.23.218875; this version posted July 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 1 Cohesin mutations are synthetic lethal with stimulation of WNT signaling 2 3 4 Chue Vin Chin1,2, Jisha Antony1,2, Sarada Ketharnathan1, Gregory Gimenez1, Kate M. 5 Parsons3, Jinshu He3, Amee J. George3, Antony Braithwaite1,2, Parry Guilford4, Ross D. 6 Hannan3 and Julia A. Horsfield1,2,5 7 8 1 Department of Pathology, Dunedin School of Medicine, University of Otago, Dunedin, New 9 Zealand. 10 2 Maurice Wilkins Centre for Molecular Biodiscovery, Private Bag 92019, The University of 11 Auckland. 12 3 The John Curtin School of Medical Research, Australian National University, Canberra, 13 Australia. 14 4 Department of Biochemistry, University of Otago, Dunedin, New Zealand. 15 5 Correspondence: Julia Horsfield, Department of Pathology, Dunedin School of Medicine, 16 University of Otago, PO Box 913, Dunedin 9016, New Zealand. Tel.: 64 3-479-7436; E-mail: 17 [email protected]. 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.23.218875; this version posted July 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 18 Abstract 19 Mutations in genes encoding subunits of the cohesin complex are common in several cancers, 20 but may also expose druggable vulnerabilities. We generated isogenic MCF10A cell lines 21 with deletion mutations of genes encoding cohesin subunits SMC3, RAD21 and STAG2 and 22 screened for synthetic lethality with 3,009 FDA-approved compounds. The screen identified 23 several compounds that interfere with transcription, DNA damage repair and the cell cycle. 24 Unexpectedly, one of the top ‘hits’ was a GSK3 inhibitor, an agonist of Wnt signaling. We 25 show that sensitivity to GSK3 inhibition is likely due to stabilization of b-catenin in cohesin 26 mutant cells, and that Wnt-responsive gene expression is highly sensitized in STAG2-mutant 27 CMK leukemia cells. Moreover, Wnt activity is enhanced in zebrafish mutant for cohesin 28 subunit rad21. Our results suggest that cohesin mutations could progress oncogenesis by 29 enhancing Wnt signaling, and that targeting the Wnt pathway may represent a novel 30 therapeutic strategy for cohesin mutant cancers. 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.23.218875; this version posted July 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 31 Introduction 32 33 The cohesin complex is essential for sister chromatid cohesion, DNA replication, DNA 34 repair, and genome organization. Three subunits, SMC1A, SMC3 and RAD21, form the core 35 ring-shaped structure of human cohesin [1, 2]. A fourth subunit of either STAG1 or STAG2 36 binds to cohesin by contacting RAD21 and SMC subunits [3], and is required for the 37 association of cohesin with DNA [1-3]. The STAG subunits of cohesin are also capable of 38 binding RNA in the nucleus [4]. Cohesin associates with DNA by interaction with loading 39 factors NIPBL and MAU2 [5], its stability on DNA is regulated by the acetyltransferases 40 ESCO1 [6] and ESCO2 [7], and its removal is facilitated by PDS5 and WAPL [8, 9]. The 41 cohesin ring itself acts as a molecular motor to extrude DNA loops, and this activity is 42 thought to underlie its ability to organize the genome [10, 11]. Cohesin works together with 43 CCCTC-binding factor (CTCF) to mediate three-dimensional genome structures, including 44 enhancer-promoter loops that instruct gene accessibility and expression [12, 13]. The 45 consequences of cohesin mutation therefore manifest as chromosome segregation errors, 46 DNA damage, and deficiencies in genome organization leading to gene expression changes. 47 48 All cohesin subunits are essential to life: homozygous mutations in genes encoding complex 49 members are embryonic lethal [14]. However, haploinsufficient germline mutations in 50 NIPBL, ESCO2, and in cohesin genes, cause human developmental syndromes known as the 51 “cohesinopathies” [14]. Somatic mutations in cohesin genes are prevalent in several different 52 types of cancer, including bladder cancer (15-40%), endometrial cancer (19%), glioblastoma 53 (7%), Ewing’s sarcoma (16-22%) and myeloid leukemias (5-53%) [15-17]. The prevalence of 54 cohesin gene mutations in myeloid malignancies [18-22] reflects cohesin’s role in 55 determining lineage identity and differentiation of hematopoietic cells [23-26]. Of the 56 cohesin genes, STAG2 is the most frequently mutated, with about half of cohesin mutations in 57 cancer involving STAG2 [17]. 58 59 While cancer-associated mutations in genes encoding RAD21, SMC3 and STAG1 are always 60 heterozygous [21, 27, 28], mutations in the X chromosome-located genes SMC1A and STAG2 61 can result in complete loss of function due to hemizygosity (males), or silencing of the wild 62 type during X-inactivation (females). STAG2 and STAG1 have redundant roles in cell 63 division, therefore complete loss of STAG2 is tolerated due to partial compensation by 64 STAG1. Loss of both STAG2 and STAG1 leads to lethality [29, 30]. STAG1 inhibition in 3 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.23.218875; this version posted July 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 65 cancer cells with STAG2 mutation causes chromosome segregation defects and subsequent 66 lethality [31]. Therefore, although partial depletion of cohesin can confer a selective 67 advantage to cancer cells, a complete block of cohesin function will cause cell death. The 68 multiple roles of cohesin provides an opportunity to inhibit the growth of cohesin mutant 69 cancer cells via chemical interference with pathways that depend on normal cohesin function. 70 For example, poly ADP-ribose polymerase (PARP) inhibitors were previously shown to 71 exhibit synthetic lethality with cohesin mutations [17, 31-34]. PARP inhibitors prevent DNA 72 double-strand break repair [35], a process that also relies on cohesin function. 73 74 To date, only a limited number of compounds have been identified as inhibitors of cohesin- 75 mutant cells [17]. Here, we sought to identify additional compounds of interest by screening 76 libraries of FDA-approved molecules against isogenic MCF10A cells with deficiencies in 77 RAD21, SMC3 or STAG2. Unexpectedly, our screen identified a novel sensitivity of 78 cohesin-deficient cells to a GSK3 inhibitor that acts as an agonist of the Wnt signaling 79 pathway. We found that b-catenin stabilization upon cohesin deficiency likely contributes to 80 an acute sensitivity of Wnt target genes. The results raise the possibility that sensitization to 81 Wnt signaling in cohesin mutant cells may participate in oncogenesis, and suggest that Wnt 82 agonism could be therapeutically useful for cohesin mutant cancers. 83 84 Results 85 86 Cohesin gene deletion in MCF10A cells results in minor cell cycle defects 87 To avoid any complications with pre-existing oncogenic mutations, we chose the relatively 88 ‘normal’ MCF10A line for creation and screening of isogenic deletion clones of cohesin 89 genes SMC3, RAD21, and STAG2 (Supplementary Fig. 1a). MCF10A is a near-diploid, 90 immortalized, breast epithelial cell line that exhibits normal epithelial characteristics [36] and 91 has been successfully used for siRNA and small molecule screening [37]. We isolated several 92 RAD21 and SMC3 deletion clones, and selected single clones for further characterization that 93 grew normally and were essentially heterozygous. In the selected RAD21 deletion clone, one 94 of three RAD21 alleles (on chromosome 8, triploid in MCF10A) was confirmed deleted, with 95 one wild type and one undetermined allele (Supplementary Fig. 1c, e-g). In the selected 96 SMC3 deletion clone, one of two SMC3 alleles (on chromosome 10) was deleted 97 (Supplementary Fig. 1d, e-g). In the selected STAG2 deletion clone (Supplementary Fig. 1b), 4 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.23.218875; this version posted July 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 98 homozygous loss of STAG2 was determined by the absence of STAG2 mRNA and protein 99 (Supplementary Fig. 1e-g). For convenience here, we named the three cohesin mutant clones 100 RAD21+/-, SMC3+/- and STAG2-/-. a b Hours after double thymidine release 0 6 12 36 48 100 WT WT ** RAD21+/- ce (%) RAD21+/- en 50 SMC3+/- **** STAG2-/- Conflu SMC3+/- 0 0 24 48 72 96 120 STAG2-/- Time (h) 100 c Chromosome spreads d Normal Partial loss Complete loss Normal ds (%) ds ea Partial loss 50 Complete loss ase spr aph Met Normal 0 PCS 1+/- 2-/- WT 2 G D A A R SMC3+/-ST e f g WT RAD21+/- SMC3+/- STAG2-/- 300 **** 0.25 **** 0.20 M) 2 (μ 250 * 0.15 DAPI r area 0.10 ea 200 Nucl 0.05 Micronuclei per cell (%) 150 0.00 1+/- 2-/- WT 1+/- 2-/- WT 2 G 2 G D A D A A A SMC3+/- R SMC3+/-ST R ST Figure 1.