Reducing disparities is vital for equitable access to precision treatments in cancer. Socioenvironmental factors are a major driver of disparities, but differences in genetic variation likely also contribute. The impact of genetic ancestry on prioritization of cancer targets in drug discovery pipelines has not been systematically explored due to the absence of pre-clinical data at the appropriate scale. Here, we analyze data from 611 genome-scale CRISPR/Cas9 viability experiments in human cell line models to identify ancestry-associated genetic dependencies essential for cell survival. Surprisingly, we find that most putative associations between ancestry and dependency arise from artifacts related to germline variants. Our analysis suggests that for 1.2-2.5% of guides, germline variants in sgRNA targeting sequences reduce cutting by the CRISPR/Cas9 nuclease, disproportionately affecting cell models derived from individuals of recent African descent. We propose three approaches to mitigate this experimental bias, enabling the scientific community to address these disparities.

A hallmark of high-risk childhood medulloblastoma is the dysregulation of RNA translation. Currently, it is unknown whether medulloblastoma dysregulates the translation of putatively oncogenic non-canonical open reading frames (ORFs). To address this question, we performed ribosome profiling of 32 medulloblastoma tissues and cell lines and observed widespread non-canonical ORF translation. We then developed a stepwise approach using multiple CRISPR-Cas9 screens to elucidate non-canonical ORFs and putative microproteins implicated in medulloblastoma cell survival. We determined that multiple lncRNA-ORFs and upstream ORFs (uORFs) exhibited selective functionality independent of main coding sequences. A microprotein encoded by one of these ORFs, ASNSD1-uORF or ASDURF, was upregulated, associated with MYC-family oncogenes, and promoted medulloblastoma cell survival through engagement with the prefoldin-like chaperone complex. Our findings underscore the fundamental importance of non-canonical ORF translation in medulloblastoma and provide a rationale to include these ORFs in future studies seeking to define new cancer targets.

Aberrant expression of stem cell-associated genes is a common feature in acute myeloid leukemia (AML) and is linked to leukemic self-renewal and therapy resistance. Using AF10rearranged leukemia as a prototypical example of the recurrently activated "stemness" network in AML, we screened for chromatin regulators that sustain its expression. We deployed a CRISPR-Cas9 screen with a bespoke domain-focused library and identified several novel chromatin-modifying complexes as regulators of the TALE domain transcription factor MEIS1, a key leukemia stem cell (LSC)-associated gene. CRISPR droplet sequencing revealed that many of these MEIS1 regulators coordinately controlled the transcription of several AML oncogenes. In particular, we identified a novel role for the Tudor-domain-containing chromatin reader protein SGF29 in the transcription of AML oncogenes. Furthermore, SGF29 deletion impaired leukemogenesis in models representative of multiple AML subtypes in multiple AML subtype models. Our studies reveal a novel role for SGF29 as a nononcogenic dependency in AML and identify the SGF29 Tudor domain as an attractive target for drug discovery.

BACKGROUND:Genome-wide functional screening using the CRISPR-Cas9 system is a powerful tool to uncover tumor-specific and common genetic dependencies across cancer cell lines. Current CRISPR-Cas9 knockout libraries, however, primarily target protein-coding genes. This limits functional genomics-based investigations of miRNA function. METHODS:We designed a novel CRISPR-Cas9 knockout library (lentiG-miR) of 8107 distinct sgRNAs targeting a total of 1769 human miRNAs and benchmarked its single guide RNA (sgRNA) composition, predicted on- and off-target activity, and screening performance against previous libraries. Using a total of 45 human cancer cell lines, representing 16 different tumor entities, we performed negative selection screens to identify miRNA fitness genes. Fitness miRNAs in each cell line were scored using a combination of supervised and unsupervised essentiality classifiers. Common essential miRNAs across distinct cancer cell lines were determined using the 90th percentile method. For subsequent validation, we performed knockout experiments for selected common essential miRNAs in distinct cancer cell lines and gene expression profiling. RESULTS:We found significantly lower off-target activity for protein-coding genes and a higher miRNA gene coverage for lentiG-miR as compared to previously described miRNA-targeting libraries, while preserving high on-target activity. A minor fraction of miRNAs displayed robust depletion of targeting sgRNAs, and we observed a high level of consistency between redundant sgRNAs targeting the same miRNA gene. Across 45 human cancer cell lines, only 217 (12%) of all targeted human miRNAs scored as a fitness gene in at least one model, and fitness effects for most miRNAs were confined to small subsets of cell lines. In contrast, we identified 49 common essential miRNAs with a homogenous fitness profile across the vast majority of all cell lines. Transcriptional profiling verified highly consistent gene expression changes in response to knockout of individual common essential miRNAs across a diverse set of cancer cell lines. CONCLUSIONS:Our study presents a miRNA-targeting CRISPR-Cas9 knockout library with high gene coverage and optimized on- and off-target activities. Taking advantage of the lentiG-miR library, we define a catalogue of miRNA fitness genes in human cancer cell lines, providing the foundation for further investigation of miRNAs in human cancer.

ID 55303 Poster Board 246 A range of somatic alterations in both oncogenes and tumor suppressor genes promote cancer development. Synthetic lethal relationships or collateral lethalities have emerged in the context of these genetic driver mutations, revealing new targets for therapeutic development. Most variation within cancer genomes results from germline, as opposed to somatic, mutations, but the genetic dependencies engendered by these variants have not been systematically explored. We sought to define the landscape of these interactions by leveraging the genome-scale CRISPR/Cas9 screens that were performed as part of The Cancer Dependency Map (DepMap). In this study we systematically analyzed the associations between individual germline variants in 612 cancer cell lines and >16,000 genetic dependencies. Across all genes profiled in this analysis, we identified 121 genes whose dependency profiles are associated with the presence of specific germline variants. The top association in this analysis was between two functionally redundant Holliday junction resolvases. In this study we seek to define the nature of this synthetic lethal relationship, with a focus on how this variant impacts protein activity.

Stimulator of interferon genes (STING) is an intracellular sensor of cyclic di-nucleotides involved in the innate immune response against pathogen- or self-derived DNA. STING trafficking is tightly linked to its function, and its dysregulation can lead to disease. Here, we systematically characterize genes regulating STING trafficking and examine their impact on STING-mediated responses. Using proximity-ligation proteomics and genetic screens, we demonstrate that an endosomal sorting complex required for transport (ESCRT) complex containing HGS, VPS37A and UBAP1 promotes STING degradation, thereby terminating STING-mediated signaling. Mechanistically, STING oligomerization increases its ubiquitination by UBE2N, forming a platform for ESCRT recruitment at the endosome that terminates STING signaling via sorting in the lysosome. Finally, we show that expression of a UBAP1 mutant identified in patients with hereditary spastic paraplegia and associated with disrupted ESCRT function, increases steady-state STING-dependent type I IFN responses in healthy primary monocyte-derived dendritic cells and fibroblasts. Based on these findings, we propose that STING is subject to a tonic degradative flux and that the ESCRT complex acts as a homeostatic regulator of STING signaling.

To define cellular immunity to the intracellular pathogen Toxoplasma gondii , we performed a genome-wide CRISPR loss-of-function screen to identify genes important for (interferon gamma) IFN-γ-dependent growth restriction. We revealed a role for the tumor suppressor NF2/Merlin for maximum induction of Interferon Stimulated Genes (ISG), which are positively regulated by the transcription factor IRF-1. We then performed an ISG-targeted CRISPR screen that identified the host E3 ubiquitin ligase RNF213 as necessary for IFN-γ-mediated control of T. gondii in multiple human cell types. RNF213 was also important for control of bacterial ( Mycobacterium tuberculosis ) and viral (Vesicular Stomatitis Virus) pathogens in human cells. RNF213-mediated ubiquitination of the parasitophorous vacuole membrane (PVM) led to growth restriction of T. gondii in response to IFN-γ. Moreover, overexpression of RNF213 in naive cells also impaired growth of T. gondii . Surprisingly, growth inhibition did not require the autophagy protein ATG5, indicating that RNF213 initiates restriction independent of a previously described noncanonical autophagy pathway. Mutational analysis revealed that the ATPase domain of RNF213 was required for its recruitment to the PVM, while loss of a critical histidine in the RZ finger domain resulted in partial reduction of recruitment to the PVM and complete loss of ubiquitination. Both RNF213 mutants lost the ability to restrict growth of T. gondii , indicating that both recruitment and ubiquitination are required. Collectively, our findings establish RNF213 as a critical component of cell-autonomous immunity that is both necessary and sufficient for control of intracellular pathogens in human cells.

Mammary tumors in dogs hold great potential as naturally occurring breast cancer models in translational oncology, as they share the same environmental risk factors, key histological features, hormone receptor expression patterns, prognostic factors, and genetic characteristics as their human counterparts. We aimed to develop in vitro tools that allow functional analysis of canine mammary tumors (CMT), as we have a poor understanding of the underlying biology that drives the growth of these heterogeneous tumors. We established the long-term culture of 24 organoid lines from 16 dogs, including organoids derived from normal mammary epithelium or benign lesions. CMT organoids recapitulated key morphological and immunohistological features of the primary tissue from which they were derived, including hormone receptor status. Furthermore, genetic characteristics (driver gene mutations, DNA copy number variations, and single-nucleotide variants) were conserved within tumor-organoid pairs. We show how CMT organoids are a suitable model for in vitro drug assays and can be used to investigate whether specific mutations predict therapy outcomes. Specifically, certain CMT subtypes, such as PIK3CA mutated, estrogen receptor-positive simple carcinomas, can be valuable in setting up a preclinical model highly relevant to human breast cancer research. In addition, we could genetically modify the CMT organoids and use them to perform pooled CRISPR/Cas9 screening, where library representation was accurately maintained. In summary, we present a robust 3D in vitro preclinical model that can be used in translational research, where organoids from normal, benign as well as malignant mammary tissues can be propagated from the same animal to study tumorigenesis.

To define novel mechanisms for cellular immunity to the intracellular pathogen Toxoplasma gondii , we performed a genome-wide CRISPR loss-of-function screen to provide an unbiased assessment of genes important for IFN-γ-dependent growth restriction. We revealed a previously unknown role for the tumor suppressor NF2/Merlin for maximum induction of Interferon Stimulated Genes (ISG), which are positively regulated by the transcription factor IRF-1. We then performed an additional focused ISG-targeted CRISPR screen that identified the host E3 ubiquitin ligase RNF213 as essential for IFN-γ mediated control of T. gondii . RNF213 mediated ubiquitination of targets on the parasite-containing vacuole and growth restriction in response to IFN-γ in a variety of cell types, thus identifying a conserved factor that plays a prominent role in human cells. Surprisingly, growth inhibition did not require the autophagy protein ATG5, indicating that RNF213 initiates restriction independent of a non-canonical autophagy pathway that has previously been implicated in control of T. gondii . RNF213 was also important for control of unrelated intracellular pathogens in human cells treated with IFN, as shown here for Mycobacterium tuberculosis and Vesicular Stomatitis Virus. Collectively, our findings establish RNF213 as a critical component of cell-autonomous immunity to a broad spectrum of intracellular pathogens in human cells.

BACKGROUND: Genomic and transcriptomic studies have elucidated new insights into the landscape of diffuse intrinsic pontine glioma (DIPG). However, the role of long non-coding RNAs (lncRNAs) has not been explored at depth in these tumors, and there have not been studies focused on how lncRNAs interact with the K27M histone mutation. In a recent analysis of nearly 200 DIPGs and pediatric high-grade gliomas (pHGG), we previously detected a novel, recurring structural variant in the lncRNA CCDC26. This rearrangement occurs in nearly 10% of all DIPGs, and we have furthermore identified alterations in more than 100 lncRNAs in DIPG. METHODS: To identify lncRNAs required for proliferation of patient-derived DIPG cancer cells, we designed two custom genome-scale lncRNA libraries. We generated a genome-scale lncRNA CRISPR-Cas9 knockout pooled library, consisting of 45,766 single guide RNAs (sgRNAs). Additionally, we generated a genome-wide CRISPR interference pooled library consisting of 45,608 sgRNAs, targeting lncRNA transcription start sites (TSS). RESULTS: We utilized in vitro histone-mutant pHGG models as well as edited clones of these models with the K27M mutant corrected in order to compare lncRNA dependencies in these two contexts. We have successfully performed genome-scale CRISPR-Cas9 knockout and CRISPR interference screens targeting lncRNAs in these cell lines, revealing lncRNA dependencies. Candidate dependencies in our CRISPR-Cas9 knockout screen include LOC100507412, LOC105379524, and LINC02193. CONCLUSION: Genome-wide lncRNA CRISPR knock-out and CRISPR interference screens are a novel approach for the unbiased identification of lncRNAs that are required for pediatric high-grade glioma proliferation. Further validation of specific lncRNAs is required, and these lncRNA dependencies represent potential novel therapeutic targets.

ABSTRACT Aberrant expression of stem-cell-associated genes is a common feature in acute myeloid leukemia (AML) and is linked to leukemic self-renewal and therapy resistance. Using AF10-rearranged leukemia as a prototypical example displaying a recurrent “ stemness ” network activated in AML, we screened for chromatin regulators that sustain aberrant activation of these networks. We deployed a CRISPR-Cas9 screen with a bespoke domain-focused library and identified several novel chromatin-modifying complexes as regulators of the TALE domain transcription factor MEIS1, a key leukemia stem cell (LSC)-associated gene. CRISPR droplet sequencing revealed that many of these MEIS1 regulators coordinately controlled the transcription of several AML oncogenes. In particular, we identified a novel role for the Tudor-domain containing chromatin reader protein SGF29 in the transcription of key AML oncogenes. Furthermore, SGF29 deletion impaired leukemogenesis in models representative of multiple AML subtypes. Our studies reveal a novel role for SGF29 as a non-oncogenic dependency in AML and identify the SGF29 Tudor domain as an attractive target for drug discovery.

Reducing disparities is critical to promote equity of access to precision treatments for all patients with cancer. While socioenvironmental factors are a major driver behind such disparities, biological differences also are likely to contribute. The prioritization of cancer drug targets is foundational for drug discovery, yet whether ancestry-related signals in target discovery pipelines exist has not been systematically explored due to the absence of data at the appropriate scale. Here, we analyzed data from 611 genome-scale CRISPR/Cas9 viability experiments in human cell line models as part of the Cancer Dependency Map to identify ancestry-associated genetic dependencies. Surprisingly, we found that most putative associations between ancestry and dependency arose from artifacts related to germline variants that are present at different frequencies across ancestry groups. In 2-5% of genes profiled in each cellular model, germline variants in sgRNA targeting sequences likely reduced cutting by the CRISPR/Cas9 nuclease. Unfortunately, this bias disproportionately affected cell models derived from individuals of recent African descent because their genomes tended to diverge more from the consensus genome typically used for CRISPR/Cas9 guide design. To help the scientific community begin to resolve this source of bias, we report three complementary methods for ancestry-agnostic CRISPR experiments. This report adds to a growing body of literature describing ways in which ancestry bias impacts cancer research in underappreciated ways.

Several leukemia-associated oncoproteins activate transcriptional circuits resembling a stem-like state in acute myeloid leukemia (AML). This activation of “stemness” genes such as the clustered homeobox (HOX) genes, TALE domain proteins MEIS1, PBX1/3, and the Polycomb group gene BMI1 is achieved by enlisting the activity of specialized components of the epigenetic machinery. Leveraging a MEIS1 CRISPR knock-in fluorescent reporter and a pooled domain-focused high-density CRISPR screen, we comprehensively mapped epigenetic regulators critical for perpetuating these stemness networks in AML. Through this approach, we identified and independently validated members of eight distinct chromatin-modifying complexes as candidate regulators of the HOX/MEIS gene expression program. Single-cell and bulk RNA-seq studies revealed that some of the novel HOX regulators that we identified were required for the expression of a large proportion of the leukemia-associated transcriptome. Of particular interest to us was the TUDOR domain chromatin reader CCDC101. CRISPR-ko of CCDC101 significantly reduced the proliferation of cells with distinct HOX-activating mutations. ChIP-seq studies showed that CCDC101 occupied stem-cell associated gene loci in a TUDOR-domain dependent manner and CCDC101 deletion selectively attenuated their transcription. Further, CCDC101 knockout impaired blast colony formation and induced differentiation in the MLL-AF9 and CALM-AF10 mouse AML models but did not affect normal hematopoietic colony formation. CCDC101 deletion delayed disease latency in two distinct human AML cell line models in vivo and resulted in striking antiproliferative effects in AML patient cells. Our study revealed novel attractive nodes for therapeutic targeting of leukemia stem cells in AML and provides a framework to identify vulnerabilities against recalcitrant oncogenic networks.

STING is an intracellular sensor of cyclic di-nucleotides involved in response to pathogen- or self-derived DNA that induces protective immunity, or if dysregulated, autoimmunity. STING trafficking is tightly linked to its activity. We aimed to systematically characterize genes regulating STING trafficking and to define their impact on STING responses. Based on proximity-ligation proteomics and genetic screens, an ESCRT complex containing HGS, VPS37A and UBAP1 was found to be required for STING degradation and signaling shutdown. Analogous to phosphorylated STING creating a platform for IRF3 recruitment, oligomerization-driven STING ubiquitination by UBE2N formed a platform for ESCRT recruitment at the endosome, responsible for STING signaling shutdown. A UBAP1 mutant that underlies human spastic paraplegia and disrupts ESCRT function led to STING-dependent type I IFN responses at the steady-state, defining ESCRT as a homeostatic regulator of STING signaling.

Several leukemia-associated oncoproteins activate transcriptional circuits resembling a stem-like state in acute myeloid leukemia (AML). This activation of “stemness” genes is achieved by enlisting the activity of specialized components of the epigenetic machinery. We sought to comprehensively map epigenetic regulators critical for perpetuating these stemness networks in AML, as they may represent new therapeutic targets. To this end, we used a GFP-reporter knocked into the endogenous locus of the key leukemia oncogene and self-renewal-associated gene MEIS1 and conducted a pooled domain-focused CRISPR screen targeting >600 epigenetic modifiers. Through this screen, we identified and validated multiple members of eight distinct chromatin-modifying complexes that were required for sustaining MEIS1 expression in diverse AML subtypes. These included several novel MEIS1 regulators such as TAF6, LDB1, KAT2A, AFF2, JADE3, casein kinase 2 (CK2), ENY2 and SGF29, in addition to previously characterized regulators such as DOT1L, AF10, ENL and HBO1. A secondary pooled CRISPR screen, coupled with a single-cell transcriptome readout (CROP-seq) revealed that the deletion of several of these MEIS1 activators not only reversed MEIS1 activation but also reduced expression of stem-cell associated genes, including genes of the HOXA cluster, BMI1, SATB1 and Musashi 2, and concomitantly activated expression of differentiation-associated genes. Of particular interest to us was the TUDOR domain chromatin reader SGF29, a key component of the SAGA (Spt-Ada-Gcn5 acetyltransferase) complex. CRISPR-ko of SGF29 significantly reduced the proliferation of cells with distinct HOX-activating mutations including KMT2A rearranged and AF10-rearranged AML cells. ChIP-seq studies showed that SGF29 occupied the promoters and enhancers of key leukemia-oncogenes in a TUDOR-domain-dependent manner and SGF29 deletion selectively attenuated their transcription. Importantly, chromatin proteomics showed that SGF29 deletion led to the eviction of the SAGA complex subunit KAT2A from chromatin and into the cytoplasm. Further, SGF29 knockout impaired blast colony formation and induced differentiation in the KMT2A-MLLLT3, KMT2A-AF10, and CALM-AF10 mouse AML models but did not affect normal hematopoietic colony formation. Lastly, SGF29 deletion delayed disease latency in two distinct human AML cell line models as well as a patient-derived xenograft model in vivo and resulted in striking antiproliferative effects. Our study revealed several new attractive nodes for therapeutic targeting of leukemia stem cells in AML, including the chromatin reader SGF29, and provides a framework to identify vulnerabilities against recalcitrant oncogenic networks. Citation Format: Karina O. Barbosa Guerra, Anagha Deshpande, Ping Xiang, Anna Minkina, Fiorella Schischlik, Adam Brown, Neil A. Robertson, John Doench, Peter D. Adams, Keith Humphries, Eytan Ruppin, Jay Shendure, Prashant Mali, Aniruddha Deshpande. High-density CRISPR screens reveal mechanisms of chromatin regulation of stemness networks in acute myeloid leukemia. [abstract]. In: Proceedings of the AACR Special Conference: Cancer Epigenomics; 2022 Oct 6-8; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2022;82(23 Suppl_2):Abstract nr PR004.

In the 20 years since the completion of the Human Genome Project, cancer biology remains rooted in the assumption that the human genome encodes ~20,000 protein-coding genes. Yet, I and others have shown that thousands of “non-canonical” open reading frames (ncORFs) populate the human genome, potentially representing a dramatic expansion of the cancer proteome. Despite their abundance, little is known about the role of ncORFs as cancer driver genes. We developed functional genomics approaches to pursue this question across human cancers. To determine whether ncORFs represent biologically active proteins, we experimentally interrogated 553 candidates selected from ncORF datasets. Of these, 257 (46%) showed evidence of stable protein expression using multiple assays, and 401 (72%) induced gene expression changes when expressed in cancer cell lines. The bioactivity of ncORFs was dependent on their ability to translate a protein: mutation of the ORF start codon prevented induction of gene expression changes observed with the wild type ncORF in 48 of 51 (94%) cases. Using custom CRISPR/Cas9 knock-out screens targeting >2,000 ncORFs in 20 cancer cell lines, we found that genomic knock-out of approximately 10% of ncORFs induced viability defects in cancer cells. We focused on two candidates for functional studies. In breast cancer, we described G029442 - renamed glycine-rich extracellular protein-1 (GREP1) - as a secreted protein that is highly expressed and prognostic for poor patient outcomes. Knock-out of GREP1 in 263 cancer cell lines showed preferential essentiality in breast cancer-derived lines. The secretome of GREP1-expressing cells has an increased abundance of the oncogenic cytokine GDF15, and GDF15 supplementation mitigated the growth-inhibitory effect of GREP1 knockout. In medulloblastoma, we found that MYC-driven medulloblastoma cells are enriched for bioactive upstream ORFs (uORFs) that are encoded within the 5’ untranslated regions of mRNAs. We validated the ASNSD1 uORF as a top genetic vulnerability in multiple models of medulloblastoma, and its overexpression is sufficient to increase neural stem cell proliferation. Mechanistically, ASNSD1 uORF promotes a MYC-associated cellular program and interacts with the multiprotein prefoldin complex, which is required for tumors to maintain post-transcriptional regulation. Our work supports a generalizable principle that ncORFs commonly encode biologically-active proteins in diverse malignancies. Ongoing investigation of ncORFs therefore represents a new frontier in cancer research with the potential to define the next generation of therapeutic target genes. Citation Format: John R. Prensner, Ian Yannuzzi, Karl Clauser, Karsten Krug, Oana Enache, Adam Brown, Amy Goodale, David E. Root, Pratiti Bandopadhayay, Todd Golub. Non-canonical proteins are cancer cell vulnerabilities in diverse malignancies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3624.

Pooled variant expression libraries can test the phenotypes of thousands of variants of a gene in a single multiplexed experiment. In a library encoding all single-amino-acid substitutions of a protein, each variant differs from its reference only at a single codon-position located anywhere along the coding sequence. Consequently, accurately identifying these variants by sequencing is a major technical challenge. A popular but expensive brute-force approach is to divide the pool of variants into multiple smaller sub-libraries that each contains variants of a small region and that must each be constructed and screened individually, but that can then be PCR-amplified and fully sequenced with a single read to allow direct readout of variant abundance. Here we present an approach to screen very large variant libraries with mutations spanning a wide region in a single pool, including library design criteria and mutant-detection algorithms that permit reliable calling and counting of variants from large-scale sequencing data.### Competing Interest StatementW.C.H. is a consultant for ThermoFisher, Solasta, MPM Capital, iTeos, Jubilant Therapeutics, Tyra Therapeutics, RAPPTA Therapeutics, Frontier Medicines, KSQ Therapeutics and Paraxel. A.J.A has consulted for Oncorus, Inc., Arrakis Therapeutics, and Merck & Co., Inc, and has research funding from Mirati Therapeutics, Syros, Deerfield, Inc., and Novo Ventures that is unrelated to this work. A.O.G. is a share and option holder of 10X Genomics. D.E.R. receives research funding from members of the Functional Genomics Consortium (Abbvie, BMS, Jannsen, Merck, Vir), and is a director of Addgene, Inc.

Abstract The brain is the foremost non-gonadal tissue for expression of non-coding RNAs of unclear function. Yet, whether such transcripts are truly non-coding or rather the source of non-canonical protein translation is unknown. Here, we used functional genomic screens to establish the cellular bioactivity of non-canonical proteins located in putative non-coding RNAs or untranslated regions of protein-coding genes. We experimentally interrogated 553 open reading frames (ORFs) identified by ribosome profiling for three major phenotypes: 257 (46%) demonstrated protein translation when ectopically expressed in HEK293T cells, 401 (73%) induced gene expression changes following ectopic expression across 4 cancer cell types, and 57 (10%) induced a viability defect when the endogenous ORF was knocked out using CRISPR/Cas9 in 8 human cancer cell lines. CRISPR tiling and start codon mutagenesis indicated that the biological impact of these non-canonical ORFs required their translation as opposed to RNA-mediated effects. We functionally characterized one of these ORFs, G029442—renamed GREP1 (Glycine-Rich Extracellular Protein-1)—as a cancer-implicated gene with high expression in multiple cancer types, such as gliomas. GREP1 knockout in >200 cancer cell lines reduced cell viability in multiple cancer types, including glioblastoma, in a cell-autonomous manner and produced cell cycle arrest via single-cell RNA sequencing. Analysis of the secretome of GREP1-expressing cells showed increased abundance of the oncogenic cytokine GDF15, and GDF15 supplementation mitigated the growth inhibitory effect of GREP1 knock-out. Taken together, these experiments suggest that the non-canonical ORFeome is surprisingly rich in biologically active proteins and potential cancer therapeutic targets deserving of further study.

A key question in genome research is whether biologically active proteins are restricted to the ∼20,000 canonical, well-annotated genes, or rather extend to the many non-canonical open reading frames (ORFs) predicted by genomic analyses. To address this, we experimentally interrogated 553 ORFs nominated in ribosome profiling datasets. Of these 553 ORFs, 57 (10%) induced a viability defect when the endogenous ORF was knocked out using CRISPR/Cas9 in 8 human cancer cell lines, 257 (46%) showed evidence of protein translation when ectopically expressed in HEK293T cells, and 401 (73%) induced gene expression changes measured by transcriptional profiling following ectopic expression across 4 cell types. CRISPR tiling and start codon mutagenesis indicated that the biological effects of these non-canonical ORFs required their translation as opposed to RNA-mediated effects. We selected one of these ORFs,G029442--renamedGREP1(Glycine-Rich Extracellular Protein-1)--for further characterization. We found thatGREP1encodes a secreted protein highly expressed in breast cancer, and its knock-out in 263 cancer cell lines showed preferential essentiality in breast cancer derived lines. Analysis of the secretome of GREP1-expressing cells showed increased abundance of the oncogenic cytokine GDF15, and GDF15 supplementation mitigated the growth inhibitory effect ofGREP1knock-out. Taken together, these experiments suggest that the non-canonical ORFeome is surprisingly rich in biologically active proteins and potential cancer therapeutic targets deserving of further study.