LETTERS https://doi.org/10.1038/s41588-018-0155-3 Genome-scale analysis identifies paralog lethality as a vulnerability of chromosome 1p loss in cancer Srinivas R. Viswanathan1,2,3, Marina F. Nogueira! !1,2,12, Colin G. Buss! !4,5,12, John M. Krill-Burger2, Mathias J. Wawer6, Edyta Malolepsza2,7, Ashton C. Berger1,2, Peter S. Choi1,2,3, Juliann Shih! !2, Alison M. Taylor1,2,3, Benjamin Tanenbaum! !2, Chandra Sekhar Pedamallu1, Andrew D. Cherniack! !2, Pablo Tamayo! !2,8, Craig A. Strathdee2, Kasper Lage2,7, Steven A. Carr2, Monica Schenone! !2, Sangeeta N. Bhatia2,3,4,5,9,10,11, Francisca Vazquez2, Aviad Tsherniak! !2, William C. Hahn! !1,2,3 and Matthew Meyerson! !1,2,3* Functional redundancy shared by paralog genes may afford on a gene5 and loss of function of its paralog across 10,287 paralog protection against genetic perturbations, but it can also result pairs (Supplementary Fig. 1; Supplementary Note). We identified in genetic vulnerabilities due to mutual interdependency1– 5. 167 genes for which dependency was significantly correlated with Here, we surveyed genome-scale short hairpin RNA and loss of a paralog (1.6% of paralog test pairs at q < 0.05), includ- CRISPR screening data on hundreds of cancer cell lines and ing many previously reported paralog dependencies (for example, identified MAGOH and MAGOHB, core members of the splic- ARID1B dependency with ARID1A inactivation10, SMARCA2 ing-dependent exon junction complex, as top-ranked paralog dependency with SMARCA4 inactivation11, UBC dependency with dependencies6– 8. MAGOHB is the top gene dependency in cells UBB inactivation5, and FERMT1 dependency with FERMT2 inac- with hemizygous MAGOH deletion, a pervasive genetic event tivation5). However, of these 167 paralog dependency pairs, only that frequently occurs due to chromosome 1p loss. Inhibition 7 were ‘symmetric’, in which dependency for each of the genes in of MAGOHB in a MAGOH-deleted context compromises via- the pair was significantly correlated with inactivation of its partner bility by globally perturbing alternative splicing and RNA sur- paralog (Fig. 1a,b; Supplementary Table 1). A similar analysis of data veillance. Dependency on IPO13, an importin-β receptor that from genome-scale CRISPR screening of 341 cell lines15 identified mediates nuclear import of the MAGOH/B-Y14 heterodimer9, 125 significant paralog dependencies (1.4% of paralog test pairs at is highly correlated with dependency on both MAGOH and q < 0.05), of which 7 pairs were symmetric (Supplementary Table 2; MAGOHB. Both MAGOHB and IPO13 represent dependencies Supplementary Note). Paralog genes arise via ancestral duplica- in murine xenografts with hemizygous MAGOH deletion. Our tion events and may functionally diverge over time1,16. Symmetric results identify MAGOH and MAGOHB as reciprocal paralog paralog pairs likely share complete functional redundancy, making dependencies across cancer types and suggest a rationale for them particularly attractive targets for ‘collateral lethality’ strate- targeting the MAGOHB-IPO13 axis in cancers with chromo- gies2. An enrichment for RNA-splicing related genes was noted some 1p deletion. among symmetric, but not asymmetric, paralog pairs in the shRNA The systematic integration of data from genomic characteriza- and CRISPR screening datasets (Supplementary Table 3), suggest- tion and genetic screening of cancer cell lines can identify gene ing that redundant essentiality may be exploited to target splicing- dependencies induced by specific somatic alterations and inform related pathways. the development of targeted therapeutics. For example, several One symmetric paralog pair was shared between the shRNA studies have shown that inactivation of specific driver or pas- and CRISPR datasets: MAGOH-MAGOHB; a second pair, FUBP1- senger genes may confer dependency on functionally redundant KHSRP, was highly significant for symmetry in the shRNA data 2,3,10–13 paralogs . Paralog dependencies have also emerged as impor- and borderline significant in the CRISPR dataset (q1 = 0.0547) tant targets in recent genome-scale functional genomic screens4,5, (Fig. 1a,b; Supplementary Fig. 1; Supplementary Tables 1 and 2)15. underscoring the importance of further characterizing this class of We focus here on validation of the former pair. MAGOH and cancer vulnerabilities. MAGOHB encode core members of the exon–junction complex To systematically identify paralog dependencies that may rep- (EJC), a multiprotein complex that is deposited on messenger resent attractive cancer targets, we analyzed data from pooled, RNAs at the time of splicing and that mediates diverse downstream genome-scale short hairpin RNA (shRNA) screening of 501 cancer processes including mRNA transport, stability, and nonsense- cell lines5,14. We determined the correlation between a dependency mediated decay (NMD)6,17. 1Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA. 2Broad Institute of MIT and Harvard, Cambridge, MA, USA. 3Harvard Medical School, Boston, MA, USA. 4Harvard-MIT Department of Health Sciences and Technology, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Boston, MA, USA. 5Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA, USA. 6Chemical Biology and Therapeutics Science Program, Broad Institute of Harvard and MIT, Cambridge, MA, USA. 7Department of Surgery, Massachusetts General Hospital, Boston, MA, USA. 8UCSD Moores Cancer Center and Department of Medicine, University of California, San Diego, La Jolla, CA, USA. 9Howard Hughes Medical Institute, Chevy Chase, MD, USA. 10Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, USA. 11Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA. 12These authors contributed equally: Marina F. Nogueira, Colin G. Buss. *e-mail: [email protected] NATURE GENETICS | VOL 50 | JULY 2018 | 937–943 | www.nature.com/naturegenetics 937 © 2018 Nature America Inc., part of Springer Nature. All rights reserved. LETTERS NATURE GENETICS Using both shRNA and CRISPR technologies, we individually of MAGOHB knockdown, with or without ectopic re-expression of validated MAGOHB dependency in the setting of MAGOH loss, MAGOH. We observed an increased expression of NMD biotype as well as MAGOH dependency in the setting of MAGOHB loss. transcripts on MAGOHB knockdown in ChagoK1 cells (Fig. 2a, Furthermore, in a cell line without hemizygous deletion of either left). In contrast, MAGOHB knockdown in MAGOH-reconstituted paralog, knockdown of either MAGOH or MAGOHB individu- ChagoK1 cells was well tolerated without a notable shift in NMD ally was tolerated, but the combination was lethal (Supplementary biotype transcript distribution (Fig. 2a, right). We next sought to Fig. 2). We noted that MAGOHB dependency in the setting of determine whether the upregulation of NMD isoforms on MAGOHB MAGOH inactivation was particularly pronounced based on (1) knockdown in ChagoK1 cells was occurring at the expense of other effect size (log-fold difference in MAGOHB dependency between transcript biotypes. Among genes that had significantly upregulated MAGOH-inactivated and non-MAGOH-inactivated cell lines) and NMD isoform(s) on MAGOHB knockdown, we observed a signifi- (2) MAGOHB scoring as a robust 6σ differential dependency (hav- cant proportional decrease in coding isoform expression in ChagoK1 ing a dependency score in some cell lines greater than six standard cells but not MAGOH-reconstituted ChagoK1 cells (Fig. 2b, deviations below its mean dependency score across all cell lines) in compare left and right). To investigate whether particular splice both the RNA interference (RNAi) and CRISPR screening data. We event classes were driving this redistribution of isoform types, we therefore sought to further characterize MAGOHB dependency in quantified the proportion of differentially spliced events of each the setting of MAGOH loss. class that were more common in either the absence (Fig. 2c, red) or MAGOHB was the top differential dependency in cells with presence (Fig. 2c, blue) of MAGOHB knockdown in either ChagoK1 hemizygous deletion of MAGOH (Fig. 1c; Supplementary Tables cells or MAGOH-reconstituted ChagoK1 cells. As compared with 4 and 5; Supplementary Note) and dependency on MAGOHB MAGOHB knockdown in MAGOH-reconstituted ChagoK1 cells, was predicted by low expression of MAGOH, consistent with the MAGOHB knockdown in ChagoK1 cells resulted in reduced cas- notion that hemizygous deletion of MAGOH leads to its decreased sette exon inclusion and increased intron retention (Fig. 2c). expression (Supplementary Fig. 3). shRNA-mediated knockdown Therefore, many global transcriptomic effects of MAGOH/B insuf- of MAGOHB led to a decrease in cell viability and colony-forming ficiency appear attributable to alterations in these two splice event capacity in three MAGOH-deleted cell lines, but not in control cell types, indicative of a defect in exon definition/recognition. lines euploid for MAGOH (Fig. 1d,e; Supplementary Fig. 4). Ectopic We identified 22 instances in which there was both a signifi- expression of MAGOH in an MAGOH-deleted cell line fully rescued cant absolute upregulation of an NMD isoform (beta > 1 in dif- MAGOHB dependency, indicating that MAGOHB dependency in ferential expression analysis using Kallisto19) and corresponding MAGOH-deleted cells is solely due to MAGOH loss, and consis- downregulation of at least one protein