Loss of heterozygosity of essential genes represents a widespread class of potential cancer vulnerabilities

Alterations in non-driver genes represent an emerging class of potential therapeutic targets in cancer. Hundreds to thousands of non-driver genes undergo loss of heterozygosity (LOH) events per tumor, generating discrete differences between tumor and normal cells. Here we interrogate LOH of polymorphisms in essential genes as a novel class of therapeutic targets. We hypothesized that monoallelic inactivation of the allele retained in tumors can selectively kill cancer cells but not somatic cells, which retain both alleles. We identified 5664 variants in 1278 essential genes that undergo LOH in cancer and evaluated the potential for each to be targeted using allele-specific gene-editing, RNAi, or small-molecule approaches. We further show that allele-specific inactivation of either of two essential genes (PRIM1 and EXOSC8) reduces growth of cells harboring that allele, while cells harboring the non-targeted allele remain intact. We conclude that LOH of essential genes represents a rich class of non-driver cancer vulnerabilities.

a. Number of GEMINI variants (vertical axis) plotted against the fraction of individuals that are heterozygous for each variant in the ExAC cohort (horizontal axis). Bin width = 0.5%. b. Violin plot of the percent of genome affected by LOH across 33 TCGA tumor types. Tumor types are indicated by TCGA abbreviations (see https://gdc.cancer.gov/resources-tcga-users/tcga-code-tables/tcga-studyabbreviations). Plot width represents relative sample density and dots indicate median values. c. Violin plot demonstrating the rate of LOH for essential (blue) and non-essential genes (grey). Essential genes do not have a significantly higher rate of LOH (One-tailed Student's t-test, p=1 [n.s. = not significant]). Intersecting lines indicate median values.     Off-target reads and species representing less than 0.05% of total read pairs removed for clarity.    a. Growth of cells hemizygous for the major PRIM1 rs2277339 allele (SNU-175 PRIM1 T/-) expressing shGFP, positive control shPRIM1, or indicated putative allele-specific shRNAs targeting the T allele of PRIM1 rs2277339 , as measured by CellTiter-Glo luminescence, relative to day of assay plating. n = 3 biological replicates. b. Growth of cells hemizygous for the minor PRIM1 rs2277339 allele (SNU-175 PRIM1 G/-) expressing shGFP, positive control shPRIM1, or indicated putative allele-specific shRNAs targeting the G allele of PRIM1 rs2277339 as measured by CellTiter-Glo luminescence, relative to day of assay plating. n = 3 biological replicates. One-tailed Student's t-test, *p = 0.03. c. Growth of heterozygous cells (SNU-175 PRIM1 T/G ) and cells hemizygous for the major PRIM1 rs2277339 allele (SNU-175 PRIM1 T/-) expressing shGFP, positive control shPRIM1, or indicated putative allele-specific shRNAs targeting the T allele of PRIM1 rs2277339 as measured by CellTiter-Glo luminescence, relative to day of assay plating. n = 4 biological replicates. One-tailed Student's t-test, n.s. = not significant. d. Growth of heterozygous cells (SNU-175 PRIM1 T/G ) and cells hemizygous for the minor PRIM1 rs2277339 allele (SNU-175 PRIM1 G/-) expressing shGFP, positive control shPRIM1, or indicated putative allele-specific shRNAs targeting the G allele of PRIM1 rs2277339 as measured by CellTiter-Glo luminescence, relative to day of assay plating. n = 3 biological replicates. One-tailed Student's t-test, n.s. = not significant. Data for panels a-d are presented as mean values +/-s.d. Source data provided as a Source Data file. n.s.

Supplementary Note 1
We verified that allele-specific inactivation of essential genes is possible in a heterozygous context. Gene-disrupting indels introduced by the error-prone nonhomologous end joining (NHEJ) repair pathway make distinguishing the original genotype (S or R) of an edited allele challenging. Therefore, we compared the number of unaltered reads of each allele in heterozygous cells expressing either NA or AS sgRNA. If the sgRNA disrupts PRIM1 in a non-allele specific fashion, we would expect a ratio of 1 between unaltered reads of each allele. We infected a Cas9-stable PRIM1 R/S line with NA or AS sgRNA as described above and sequenced the target loci 4 and 18 days post-infection. As expected, we saw no substantial difference in the number of unaltered reads between the two alleles in PRIM1 R/S cells expressing NA sgRNA (Supplementary Figure 2l). In contrast, PRIM1 R/S cells expressing AS sgRNA showed significantly more unaltered reads from the resistant allele compared to the sensitive allele, with a ratio that increased over time (p<0.0001, Chi-square with Yates correction). We conclude that the AS sgRNA disrupts PRIM1 in a SNP-specific manner even in a heterozygous context.

Supplementary Note 2
We confirmed that AS CRISPR disrupts EXOSC8 in a SNP-dependent manner This result confirms that EXOSC8 AS sgRNA targets EXOSC8 in a SNP-specific manner.

Supplementary Note 3
We verified that the specificity of the AS sgRNA for EXOSC8S cell lines was not due to a lack of Cas9 activity or EXOSC8 essentiality in the EXOSC8R cell lines. We

Supplementary Discussion
As an enzyme, PRIM1 represents a potential target for small-molecule drug development. However, while several potential inhibitors of human DNA primase have been proposed, none have yet advanced to clinical stages of development 1 . Efforts to discover additional putative primase inhibitors may benefit from approaches used to target bacterial and viral primases. For instance, a combined fragment-based/virtual screening approach has been used to identify novel inhibitors of the T7 bacteriophage primase. Such an approach may aid PRIM1-targeting lead-compound discovery by eliminating the challenging task of initial primase functional screening for large numbers of compounds 2 .
The residue coded for by EXOSC8 rs117135638 lies on the interface of the EXOSC8 gene product Rrp43 and exosome complex member Mtr3, raising the possibility of developing an allele-specific inhibitor of exosome complex formation. Pharmacologic inhibition of protein-protein binding has generally proven challenging due to the large, often flat, surfaces involved in protein complex formation 3 . However, disruption of several clinically relevant protein-protein interactions has been achieved previously, as illustrated by small-molecule or peptidomimetic inhibitors of p53-Mdm2 binding 4-6 , Notch complex assembly 7 , and herpesvirus DNA polymerase complex assembly 8 . As with PRIM1, any such approach to targeting LOH of EXOSC8 must also overcome the substantial hurdle of achieving allele specificity.