Cancer Cell Line Encyclopedia Research Articles

GJB3: a comprehensive biomarker in pan-cancer prognosis and immunotherapy prediction.

A wide range of connexins are situated between normal-normal cells, cancer-cancer cells, and cancer-normal cells. Abnormalities in connexin expression are typically accompanied by cancer development; however, no systematic studies have examined the role of Gap Junction Protein Beta 3 (GJB3) in the context of tumor progression and immunity, especially when considering a broad range of cancer types. In this study, data on GJB3 expression were gathered from Genotype-Tissue Expression, Cancer Cell Line Encyclopedia, and The Cancer Genome Atlas databases. Then, we analyzed the relationship between GJB3 expression and tumor characteristics. In vitro experiments using colony formation, EdU, CCK8, transwell migration assays, immunohistochemistry and western blot were performed to investigate the function of GJB3 in tumor progression of various cell lines. A drug sensitivity analysis of GJB3 was performed using the Genomics of Drug Sensitivity in Cancer database. Our findings demonstrate that GJB3 is widely expressed in various cancers and correlates significantly with disease stages, patient survival, immunotherapy response, and pharmaceutical guidance. Additionally, GJB3 plays a role in different cancer pathways, as well as in different immune and molecular subtypes of cancer. Co-expression of GJB3 with immune checkpoint genes was observed. Further experiments showed that knockdown of GJB3 inhibited the PI3K/AKT pathway and resulted in reduced proliferation, migration, and viability of different cancer cells. Overall, GJB3 shows potential as a molecular biomarker and therapeutic target for various cancers, particularly lung adenocarcinomas, mesothelioma, pancreatic adenocarcinoma. Thus, GJB3 may represent a new therapeutic target for a wide range of cancers.

HANSynergy: Heterogeneous Graph Attention Network for Drug Synergy Prediction.

Drug synergy therapy is a promising strategy for cancer treatment. However, the extensive variety of available drugs and the time-intensive process of determining effective drug combinations through clinical trials pose significant challenges. It requires a reliable method for the rapid and precise selection of drug synergies. In response, various computational strategies have been developed for predicting drug synergies, yet the exploitation of heterogeneous biological network features remains underexplored. In this study, we construct a heterogeneous graph that encompasses diverse biological entities and interactions, utilizing rich data sets from sources, such as DrugCombDB, PubChem, UniProt, and cancer cell line encyclopedia (CCLE). We initialize node feature representations and introduce a novel virtual node to enhance drug representation. Our proposed method, the heterogeneous graph attention network for drug-drug synergy prediction (HANSynergy), has been experimentally validated to demonstrate that the heterogeneous graph attention network can extract key node features, efficiently harness the diversity of information, and further enhance network functionality through the incorporation of a multihead attention mechanism. In the comparative experiment, the highest accuracy (Acc) and area under the curve (AUC) are 0.877 and 0.947, respectively, in DrugCombDB_early data set, demonstrating the superiority of HANSynergy over the competing methods. Moreover, protein-protein interactions are important in understanding the mechanism of action of drugs. The heterogeneous attention mechanism facilitates protein-protein interaction analysis. By analyzing the changes of attention weight before and after heterogeneous network training, we investigated proteins that may be associated with drug combinations. Additionally, case studies align our findings with existing research, underscoring the potential of HANSynergy in drug synergy prediction. This advancement not only contributes to the burgeoning field of drug synergy prediction but also holds the potential to provide valuable insights and uncover new drug synergies for combating cancer.

Glioma hexokinase 3 positively correlates with malignancy and macrophage infiltration.

Glioma is the main subtype of primary central nervous system (CNS) tumor with high malignancy and poor prognosis under current therapeutic approaches. Glycolysis and suppressive tumor microenvironment (TME) are key markers of glioma with great importance for aggressive features of glioma and inferior clinical outcomes. Hexokinase 3 (HK3) is an important rate-limiting enzyme in glycolysis, but its function in glioma remains unknown. This study comprehensively assessed the expression distribution and immunological effect of HK3 via pan-cancer analysis based on datasets from Genotype Tissue Expression (GTEx), Cancer Cell Line Encyclopedia (CCLE), and The Cancer Genome Atlas (TCGA). Furthermore, it explored the malignant phenotype and genomic landscape between low-HK3 and high-HK3 expression groups in gliomas from Chinese Glioma Genome Atlas (CGGA) and TCGA. Moreover, data from the TIMER website predicted the relationship between macrophage infiltration and HK3 expression. Also, single-cell sequencing data were used to validate the relationship. For pan-cancer patients, HK3 was expressed in various cancers. The results showed that HK3 was highly expressed in gliomas and positively correlated with tumor-infiltrating immune cells (TIICs), immune checkpoints, immunomodulators, and chemokines. Meanwhile, HK3 expression was highest in normal immune cells and tissues. In gliomas, the expression of HK3 was found to be closely correlated with the malignant clinical characteristics and the infiltration of macrophages. Also, HK3 was proven to be positively associated with macrophage through single-cell sequencing data and immunohistochemistry techniques. Finally, it is predicted that samples with high HK3 expression are often malignant entities and also significant genomic aberrations of driver oncogenes. This is the first comprehensive research to figure out the relationship between HK3 and TME characteristics in gliomas. HK3 is positively associated with macrophage infiltration and can induce the immunosuppressive TME and malignant phenotype of gliomas.

Identification of genes related to ribosomal proteins in colorectal cancer: exploring their potential as biomarkers, prognostic indicators, and therapeutic targets.

Colorectal cancer (CRC) ranks as the third most commonly diagnosed cancer in both females and males, underscoring the need for the identification of effective biomarkers. We assessed the expression levels of ribosomal proteins (RPs) at both mRNA and protein levels. Subsequently, leveraging the STRING database, we constructed a protein-protein interaction network and identified hub genes. The co-expression network of differentially expressed genes associated with CRC and their target hub RPs was constructed using the weighted gene co-expression network analysis algorithm. Gene ontology and molecular signatures database were conducted to gain insights into the biological roles of genes associated with the identified module. To confirm the results, the expression level of the candidate genes in the CRC samples compared to the adjacent healthy was evaluated by the RT-qPCR method. Our findings indicated that the genes related to RPs were predominantly enriched in biological processes associated with Myc Targets, Oxidative Phosphorylation, and cell proliferation. Also, results demonstrated that elevated levels of GRWD1, MCM5, IMP4, and RABEPK that related to RPs were associated with poor prognostic outcomes for CRC patients. Notably, IMP4 and RABEPK exhibited higher diagnostic value. Moreover, the expression of IMP4 and RABEPK showed a significant association with drug resistance using cancer cell line encyclopedia and genomics of drug sensitivity in cancer databases. Also, the results showed that the expression level of IMP4 and RABEPK in cancerous samples was significantly higher compared to the adjacent healthy ones. The general results of this study have shown that many genes related to RPs are increased in cancer and could be associated with the death rate of patients. We also highlighted the therapeutic and prognostic potentials of RPs genes in CRC.

The prognostic value of MEK pathway-associated estrogen receptor signaling activity for female cancers.

Other than for breast cancer, endocrine therapy has not been highly effective for gynecologic cancers. Endocrine therapy resistance in estrogen receptor positive gynecologic cancers is still poorly understood. In this retrospective study, we examined the estrogen receptor (ER) signaling pathway activities of breast, ovarian, endometrial, and cervical cancers to identify those that may predict endocrine therapy responsiveness. Clinical and genomic data of women with breast and gynecological cancers were downloaded from cBioPortal for Cancer Genomics. Estrogen receptor alpha (ESR1) expression level and sample-level pathway enrichment scores (EERES) were calculated to classify patients into four groups (low/high ESR1 and low/high EERES). Correlation between ESR1/EERES score and survival was further validated with RNAseq data from low-grade serous ovarian cancer. Pathway analyses were performed among different ESR1/EERES groups to identify genes that correlate with endocrine resistance, which are validated using Cancer Cell Line Encyclopedia gene expression and Genomics of Drug Sensitivity in Cancer data. We identified a novel combined prognostic value of ESR1 expression and the corresponding estrogen response signaling (EERES score) for breast cancer. The combined prognostic value (ESR1/EERES) may be applicable to other gynecologic cancers. More importantly, we discovered that ER signaling can cross-regulate MEK pathway activation. We identified downstream genes in the MEK pathway (EPHA2, INAVA, MALL, MPZL2, PCDH1, and TNFRSF21) that are potential endocrine therapy response biomarkers. This study demonstrated that targeting both the ER and the ER signaling activity related MEK pathway may aid the development of endocrine therapy strategies for personalized medicine.

Abstract LB247: TCPAplus: An LLM-empowered chatbot for analyzing a large protein expression atlas of human cancers

Abstract We present a comprehensive functional proteomics resource comprising nearly 9,000 samples from The Cancer Genome Atlas and Cancer Cell Line Encyclopedia, using ~500 high-quality antibodies through reverse phase protein arrays. To enhance its utility, we have developed an analytical chatbot harnessing advanced large language models. This intuitive platform empowers researchers to effortlessly query protein-centered cancer omics data, conduct related analyses, and visualize results using natural language, significantly easing the translation from intricate proteogenomic data to valuable biological insights. Citation Format: Wei Liu, Jun Liu, Yitao Tang, Chaozhong Liu, Meiyi Song, Zhenlin Ju, Shweth V. Kumar, Yiling Lu, Rehan Akbani, Gordon Mills, Han Liang. TCPAplus: An LLM-empowered chatbot for analyzing a large protein expression atlas of human cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB247.

Abstract NG04: Targeting DHX9 to trigger viral mimicry and immunotherapy responsiveness in small cell lung cancer

Abstract Immune checkpoint blockade (ICB) therapy has permanently altered cancer medicine and can result in impressive clinical responses of multiple cancer types. However, therapy resistance is common and only a small subset of patients mounts lasting responses to ICB monotherapies, highlighting the urgent need to identify mechanisms of resistance as well as new strategies to improve responsiveness. In large part, such resistance is related with the fact that many tumors are immunologically silent, or ‘cold’. Emerging data suggest that activating innate immunity in tumor cells is a potent means by which cold tumors can be turned ‘hot’, invigorating responsiveness to ICB modalities. One promising strategy to trigger an innate immune response within tumors, and boost cancer immunotherapy is by inducing the intracellular accumulation of endogenous ‘virus-mimetic’ nucleic acids in cancer cells and triggering an active antiviral response in the tumor, phenomenon known as ‘viral mimicry’. These nucleic acids are sensed by the host innate immune system as evidence of viral replication, and provoke an immediate and robust Interferon (IFN)-driven antiviral response. Such an antiviral response is potently immunogenic, and ‘viral mimicry’ inducing therapies have shown very promising results in pre-clinical models and in clinical trials. Reawakening Endogenous Retroviral Elements (EREs), including Endogenous Retroviruses (ERVs) has emerged as an effective strategy to induce ‘viral mimicry’ and boost ICB responsiveness in cold tumors. In fact, we previously reported a novel subclass of ERVs that engages innate immune signaling in mesenchymal subpopulations of Small Cell Lung Cancer (SCLC) (Canadas I, et al. Nat Med 2018). The unique feature of these ERVs is that they are located in the 3’ untranslated regions (3’UTRs) of IFN-induced genes, and oriented in an antisense fashion, which results in double-stranded RNA (dsRNA) production and IFN response. More recently, and to discover novel dsRNA regulators, we carried out a curated CRISPR screen of RNA helicases and identified DHX9 as a potent inducer of dsRNA. DHX9 is an abundant RNA/DNA helicase capable of unwinding both RNA and DNA duplexes, as well as more complex nucleic acid structures. Its functions include regulation of DNA replication and transcription, RNA processing and transport, and maintenance of genomic stability. We initially focused on SCLC, since it is the tumor type with higher DHX9 gene expression levels in the Cancer Cell Line Encyclopedia (CCLE) database. SCLC is the most aggressive and lethal form of lung cancer, with a 5-year overall survival of less than 5%. Importantly, while having one of the highest mutational burdens because of its strong association with tobacco smoking, SCLC is often characterized by a reduced antigen presentation and an immunologically cold Tumor Microenvironment (TME). Despite promising advances in the use of immunotherapy, SCLC remains a devastating disease and only a very small fraction of SCLC patients respond to these therapies. We now present a novel and effective way to restore the immunogenicity of the TME through ‘viral mimicry’, and also to selectively kill SCLC cells while doing so. Our study demonstrates that targeting DHX9 is a promising and unexplored strategy by which ‘viral mimicry’ can be triggered in tumors. Ablating DHX9 induces the formation of dsRNA derived from EREs, provoking a strong antitumor immune response in ‘cold’ tumors. Intriguingly, eliminating DHX9 also induced the accumulation of RNA-DNA hybrids (R-loops), which resulted in DNA replication stress and DNA damage in SCLC cells. In vivo, eliminating DHX9 causes a significant decrease in tumor growth while inducing a more immunogenic TME in mouse models of SCLC, dramatically enhancing responsiveness to ICB therapy. Notably, DHX9 expression is negatively correlated with immune signatures and associated with poor clinical outcomes across cancer patient datasets. These discoveries have potentially profound implications for inducing ‘viral mimicry’ in ‘cold’ tumors, because if we can induce a more immunogenic TME while selectively killing cancer cells, we could more efficiently convert these ‘cold’ tumors to ‘hot’, and therefore targeting DHX9 could be an attractive therapeutic strategy as an adjuvant for ICB-based cancer immunotherapy. To our knowledge, this is the first description of the role of DHX9 on tumor immunity and genomic instability and the identification of a completely novel viral mimicry-inducing strategy to enhance antitumor immunity and boost immunotherapy in immunologically ‘cold’ tumors, such as SCLC, one of the least responsive and deadliest tumors. All this work, which I am the senior corresponding author of, has been recently published in Cancer Discovery (Murayama T, et al. Cancer Discov. 2024). In summary, our study demonstrates that targeting DHX9 (1) induces immunogenic dsRNAs, causing tumor cell-intrinsic antiviral signaling; that (2) alters transcription and DNA replication in cancer cells to induce formation of R-loops, compromising genome stability and cell viability; and that (3) triggering these tumor-intrinsic events greatly improves ICB treatment outcomes. Thus, our work outlines an entirely new viral mimicry-inducing strategy with widespread application in cancer treatment. Citation Format: Israel Cañadas. Targeting DHX9 to trigger viral mimicry and immunotherapy responsiveness in small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr NG04.

DBDNMF: A Dual Branch Deep Neural Matrix Factorization method for drug response prediction.

Anti-cancer response of cell lines to drugs is in urgent need for individualized precision medical decision-making in the era of precision medicine. Measurements with wet-experiments is time-consuming and expensive and it is almost impossible for wide ranges of application. The design of computational models that can precisely predict the responses between drugs and cell lines could provide a credible reference for further research. Existing methods of response prediction based on matrix factorization or neural networks have revealed that both linear or nonlinear latent characteristics are applicable and effective for the precise prediction of drug responses. However, the majority of them consider only linear or nonlinear relationships for drug response prediction. Herein, we propose a Dual Branch Deep Neural Matrix Factorization (DBDNMF) method to address the above-mentioned issues. DBDNMF learns the latent representation of drugs and cell lines through flexible inputs and reconstructs the partially observed matrix through a series of hidden neural network layers. Experimental results on the datasets of Cancer Cell Line Encyclopedia (CCLE) and Genomics of Drug Sensitivity in Cancer (GDSC) show that the accuracy of drug prediction exceeds state-of-the-art drug response prediction algorithms, demonstrating its reliability and stability. The hierarchical clustering results show that drugs with similar response levels tend to target similar signaling pathway, and cell lines coming from the same tissue subtype tend to share the same pattern of response, which are consistent with previously published studies.

Abstract 6210: The cancer omics and drug experimental response dataset (CODERData): A harmonized benchmark dataset for machine learning models of drug response prediction

Abstract Background: Determining a patient’s response to a specific therapy is a vital step in developing personalized cancer treatment. Personalized treatment relies on two key technological advancements: algorithms tailored to each patient’s molecular profiles, and comprehensive datasets detailing the effects of patient diversity on drug response. Despite recent advancements in machine learning-based models, each new algorithm requires curation of numerous datasets, which poses a significant burden to researchers who need to evaluate and compare various algorithms. Currently, datasets include extensive genomics, transcriptomics, proteomics and metabolomics measurements along with drug response data (i.e., cell viability) for each cell line, collected through the Cancer Cell Line Encyclopedia (CCLE) and other initiatives. Additional omics data is available for tumor organoids and patient samples generated by consortia such as Human Cancer Model Initiative (HCMI) and the Clinical Proteomic Tumor Analysis Consortium (CPTAC). However, to date there has been little public effort to annotate and harmonize these data across repositories. Methods: To address this need, we developed the Python package coderdata that collates the most up-to-date cancer omics and drug response data. We curated four public cancer datasets, establishing reproducible pipelines that standardize data retrieval, formatting, and integration so the datasets can be built on a cluster if needed. The datasets undergo bi-monthly synchronization with FigShare, ensuring the data is up to date and available. Scientists can use coderdata to retrieve and format data on the command line interface as well as within Python scripts. Results: Coderdata is a straightforward Python package that will enable the development and benchmarking of diverse machine learning applications in cancer research. At its core, it represents reproducible data harmonization that enables heterogeneous datasets to be analyzed in bulk. This Python package downloads and standardizes data from multiple consortia-related resources including HCMI, CPTAC, CCLE, and BeatAML, together representing 4931 multi-omics samples from across 250+ cancer types. These datasets include cancer data metrics such as copy number, mutation, transcriptomics, proteomics, miRNA, methylation, and drug response data. Conclusions: The applications of machine learning in cancer biology are hampered by the distributed nature of existing datasets. As such, the collection and standardization of data by coderdata substantially reduces the time investment required for data curation in cancer research. Direct access to benchmark multi-omics and drug response datasets enables scientists to focus on algorithm development for tumor drug response prediction, potentially accelerating the discovery of therapeutic strategies. Citation Format: Jeremy Jacobson, Sydney Schwartz, M. Ryan Weil, Neeraj Kumar, Sara Gosline. The cancer omics and drug experimental response dataset (CODERData): A harmonized benchmark dataset for machine learning models of drug response prediction [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6210.

Abstract 5829: Elucidating the significance of amphiregulin and its implications for EGFR inhibition strategies in LKB1-mutant NSCLC

Abstract LKB1-mutant non-small cell lung cancer (NSCLC) is aggressive and treatment-refractory, underscoring the need for new therapeutic strategies. Our comparative study involving mouse lung tumors driven by oncogenic KrasG12D and LKB1 loss as well as tumors driven solely by KrasG12D, revealed that LKB1 loss upregulated amphiregulin (Areg), a pro-growth EGFR ligand. Therefore, this study investigates how the inactivation of LKB1 enhances AREG expression, the significance of elevated AREG in sustaining lung cancer growth, and the responses of AREG-high lung cancer to EGFR inhibitors. We first evaluated the potential clinical significance of AREG using The Cancer Genome Atlas (TCGA) lung adenocarcinoma dataset and the Cancer Cell Line Encyclopedia (CCLE) database. We observed that LKB1-null tumors with high AREG levels were associated with shorter patient survival and exhibited transcriptional signatures similar to EGFR-mutant tumors. High AREG expression also correlated positively in NSCLC cell lines with LKB1 mutations and Western blot analysis confirmed that elevated AREG levels in lung cancer cells were associated with constitutive EGFR signaling. We next explored the impact of LKB1 loss on AREG expression in lung cancer. Our prior RNA-seq and ChIP-seq investigations demonstrated CRTC2 and CREB enrichment on the AREG promoter in LKB1-null lung cancer cells. Upon LKB1 reintroduction, the enrichment diminished, indicating that CRTC-CREB transcriptionally induces AREG in LKB1-null lung cancer. To corroborate, reintroducing LKB1 or expressing a dominant-negative construct (dnCRTC), disrupting CREB/CRTC interaction, decreased AREG levels, as confirmed by RT-qPCR in LKB1-null lung cancer cells. This supports that AREG is a direct target of CRTC-CREB activation induced by LKB1 loss. To assess the impact of elevated AREG in LKB1-mutant lung cancer growth, we used RNA interference and antibodies to inhibit AREG. Both interventions decreased cell growth and colony formation, underscoring AREG is critical for LKB1-mutant lung cancer growth. Moreover, we investigated the potential correlation between AREG and the response to the EGFR tyrosine-kinase inhibitor Erlotinib in NSCLC cell lines, utilizing AREG expression data from CCLE and Erlotinib response data from the Genomics of Drug Sensitivity in Cancer (CDSC) databases. Stratifying NSCLC cell lines into sensitive, intermediate, and resistant categories revealed a potential positive association between AREG expression and Erlotinib sensitivity. Notably, LKB1-null lung cell lines with high AREG expression demonstrated heightened sensitivity compared to those with low AREG expression. Our study revealed that LKB1 loss increases AREG, triggering aberrant EGFR signaling critical for lung cancer growth, and AREG expression might impact the Erlotinib response in LKB1-inactive cells. Citation Format: Xzaviar Kaymar Solone, Xin Zhou, Mu Yu, Jennifer W. Li, Finn Rimay, Frederic Kaye, Lizi Wu. Elucidating the significance of amphiregulin and its implications for EGFR inhibition strategies in LKB1-mutant NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5829.

Abstract 862: Multi-omic analysis of allelic expression patterns of imprinted and non-imprinted genes in cancer and their association with drug response

Abstract Malignant cells undergo a broad array of epigenetic changes, including changes in imprinting patterns. In normal cells, imprinted genes have parent of origin-specific monoallelic expression, which is controlled via epigenetic mechanisms. Imprinted gene loci are often complex and frequently include multiple isoforms, which have different individual patterns of parent-of-origin monoallelic or biallelic expression in the normal tissues. Many imprinted genes promote cell proliferation and body growth, and their expression is important during embryonic and postnatal development. Dysregulation of imprinting in tumors has been suggested to contribute to growth and proliferation of cancer cells. Previously we reported that chemotherapy responses of cancer cell lines from a variety of tumors and of patient-derived acute myeloid leukemia samples were associated with copy number, expression, or DNA methylation of selected imprinted genes. In the current project, we examined allele-specific patterns of expression (monoallelic vs. biallelic) of imprinted genes and isoforms in tumors and studied their potential effect on response of cancer cells to drug treatment. We analyzed expression patterns at the whole gene, isoform, and exon levels of 59,283 autosomal protein-coding genes and ncRNA genes, with a special focus on 94 imprinted genes and their isoforms. We developed a computational pipeline to evaluate monoallelic and biallelic expression patterns of each gene in cancer cell lines and in tumor samples based on available matched RNA-seq expression data, whole exome sequencing information, and copy number data. Our initial analysis included 108 cell lines from the Cancer Cell Line Encyclopedia (CCLE), which had been derived from 9 pediatric and adult tumor histologies. We observed variation of allelic expression patterns of imprinted genes and their isoforms in tumor cell lines, including differences across and within cancer histologies. We also observed predominantly monoallelic expression of multiple additional genes. Some of them (PRIM2, BCLAF1, MAP2K3, and SEC22B) had been previously reported as potentially imprinted. Monoallelic expression of other genes identified in our analysis (e.g., GRK2, CBX4, and additional genes) was likely caused by molecular mechanisms other than imprinting. Allelic expression patterns and overall expression levels of isoforms of multiple imprinted and non-imprinted monoallelically expressed genes (e.g., CPA4, MAP2K3, and other genes) were associated with in vitro drug response. We also report our ongoing validation of allelic expression patterns of imprinted and non-imprinted genes, isoforms, and exons in primary bladder and breast tumors and in matched solid normal tissue samples using patient data from the Cancer Genome Atlas (TCGA). These results provide new knowledge about epigenetic dysregulation in cancer. Citation Format: Julia Krushkal, Travis L. Jensen, George Wright, Yingdong Zhao. Multi-omic analysis of allelic expression patterns of imprinted and non-imprinted genes in cancer and their association with drug response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 862.

Abstract 1753: SHOC2 is a genetic dependency in NRAS-mutant melanoma

Abstract Melanoma is the most lethal form of skin cancer, with over 300,000 new cases worldwide every year. Specifically, melanoma with NRAS mutations are of particular clinical concern due to their association with a poor prognosis and lack of specific treatment options. Therefore, there is a pressing need for novel approaches to address the treatment of NRAS-mutant melanoma. A current approach to identify novel drug targets is based on the genetic concept of induced essentiality, where functional interactions that occur in response to oncogene addiction create a dependency on another gene. To identify genetic dependencies in NRAS-mutant melanoma, whole-genome CRISPR-Cas9 knockout screens were conducted in 6 NRAS-mutant and 7 NRAS-wild-type (wt) early-passage New Zealand melanoma (NZM) cell lines that were established and maintained at 5% oxygen to mimic physiological oxygen conditions. The NZM cell lines were stably transduced with the whole-genome Brunello lentiviral single guide (sg) RNA library and screened for up to 35 days. BAGEL (Bayesian Analysis of Gene Essentiality) analyses of the NZM whole-genome knockout screens, alongside CRISPR-Cas9 screening data using the Avana sgRNA library from an additional 28 melanoma cell lines, available on the Cancer Cell Line Encyclopaedia (CCLE) database, revealed prospective gene candidates that exhibit greater detrimental effects on the fitness of NRAS-mutant cell lines compared to the NRAS-wt lines. These genes are being further validated as essential genes for NRAS-mutant melanoma cells through custom sgRNA library knockout screens and in vitro individual gene knockout studies. In particular, we demonstrate that knockout of SHOC2, a gene that encodes a scaffold protein essential for activation of the MAPK signalling pathway, results in the prevention of ERK phosphorylation and a more substantial reduction in cell growth in NRAS-mutant NZM cell lines when compared to NRAS-wt lines. These results support previous studies that have identified SHOC2 as a potential therapeutic target for RAS-driven cancers and suggest that targeting of SHOC2 may have utility in NRAS-mutant melanoma, where greater treatment options are urgently needed. Citation Format: Andrea Y. Gu, Tet-Woo Lee, Aziza Khan, Francis W. Hunter, Dean C. Singleton, Stephen M. F. Jamieson. SHOC2 is a genetic dependency in NRAS-mutant melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1753.

Abstract 3042: Unraveling the mechanism of YAP toxicity in neuroendocrine solid tumors: Insights from Merkel cell carcinoma

Abstract The Hippo pathway, a pivotal regulator orchestrating organ size and tissue growth, has emerged as a central player in cancer biology. The downstream transcription factor YAP within this pathway has been implicated in diverse oncogenic processes. YAP activation contributes to drug resistance, metastasis, and poor clinical outcomes in many solid tumors. Conversely, YAP exhibits growth-suppressive effects in neuroendocrine (NE) solid cancers, a paradox that remains poorly understood. Analysis of RNAseq data from the Cancer Cell Line Encyclopedia (CCLE) revealed a distinct classification of tumors into YAP1-expressing (YAPon) and non-expressing (YAPoff) subgroups, with hematopoietic malignancies and NE solid tumors featured prominently in the latter. Merkel cell carcinoma (MCC) is a rare, high-grade, NE carcinoma of the skin with two distinct etiologies. Virus-positive MCC contains integrated Merkel cell polyomavirus that expresses the viral T antigens and virus-negative MCC caused by sunlight exposure and extensive UV damage. Bulk RNAseq performed with more than 60 MCC tumors revealed that YAP expression displayed an inverse correlation with an NE gene expression signature. To gain insight into why YAP expression is so low in MCC, we generated MCC cell lines with inducible expression of YAP. Expression of YAP led to growth arrest and cell death. We performed a genome-wide CRISPR-Cas9 screen to identify genes that when lost enable MCC cells to tolerate YAP expression. Functional enrichment analysis of the positively selected genes uncovered a variety of signaling pathways. Intriguingly, the CRISPR screen revealed a positive enrichment of non-canonical BAF complex members, including BRD9 and SMARCD1. To validate this finding, we treated MKL-1 cells with a BRD9 inhibitor, observing a substantial reduction in YAP toxicity. These findings open avenues for the development of targeted therapeutic strategies for high grade NE cancers, addressing a critical gap in our understanding of YAP biology in the context of MCC and beyond. Citation Format: Furkan Bahar, Thomas C. Frost, Varsha Ananthapadmanabhan, Julia L. Schnabel, Manisha Thakuria, James A. DeCaprio. Unraveling the mechanism of YAP toxicity in neuroendocrine solid tumors: Insights from Merkel cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3042.

Abstract 3502: Multi-omics approach to cholangiocarcinoma-related biology and development of precision medicine

Abstract Introduction: Cholangiocarcinoma is a diverse tumor entity characterized by distinct tumor biology and clinical phenotypes, which are influenced by the unique tumor microenvironment (TME) and anatomical location—whether intrahepatic, peri-hilar, or distal extrahepatic. Consequently, there is a clear need for a precision strategy that relies on cancer-specific, multi-omic features, moving away from traditional molecular subtypes derived from bulk RNA sequencing. Methods: We conducted thorough multi-omic analyses utilizing data from the Cancer Dependency Map (DepMap) project. This encompassed cancer-specific molecular characterization through various omics data, genome-wide loss-of-function screening employing the CRISPR-Cas9 system, and drug sensitivity assessments via compound screening. These efforts aimed to reveal cancer-specific molecular subtypes with clinical relevance. The identified cancer-specific molecular signatures were subsequently validated in independent translational cohorts. Results: Through integrative profiling of the transcriptome in bile duct cancer cell lines (n=39) obtained from the Cancer Cell Line Encyclopedia (CCLE), we unveiled cancer-specific molecular subtypes exhibiting distinct tumor biology across all omics layers, coupled with clinical relevance and prognostic significance. The major molecular features of each subtype demonstrated reproducibility in validation cohorts. Additionally, subtype-specific molecular biomarkers, encompassing mutational signatures and metabolites, were identified. Finally, we analyzed target drugs with subtype-specific genetic dependencies to formulate a precision strategy tailored to the distinct molecular characterizations of each subtype. Conclusions: A thorough analysis utilizing a multi-omics dataset uncovered precision strategies founded on cancer-specific molecular subtypes of cholangiocarcinoma, both in terms of tumor classification and discriminative therapeutic opportunities. It is imperative to conduct prospective translational studies in conjunction with clinical trials, all centered around the cancer-specific molecular subtypes, to establish a precision strategy for effectively managing cholangiocarcinoma. Citation Format: Woo Young Kwon, Incheon Kang, Sung Hwan Lee. Multi-omics approach to cholangiocarcinoma-related biology and development of precision medicine [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3502.

Abstract 3527: Application of a deep learning based drug sensitivity prediction model on a novel anticancer drug

Abstract Discovery of a candidate drug molecule through the preclinical stages is an iterative multiparametric process until a candidate is selected for investigational new drug submission enabling studies. This iterative approach takes up significant time and effort to design, synthesize, and test new compounds. A specific testing step is evaluation of half maximal inhibitory concentration (IC50) of compounds on cell line (CL) models. To augment CL testing, computational approaches predicting the IC50 of a drug can be used to guide choice of optimization as the molecule goes through hit to lead optimization onto candidate selection. DeepDSC [1] is a deep learning (DL) model developed to predict the IC50 in a specific CL of a given drug by integrating public data from large-scale drug screens and high-throughput RNAseq profiling. The Cancer Cell Line Encyclopedia (CCLE) [2] and Genomics of Drug Sensitivity (GDSC) [3] projects compiled pharmacological profiles of 23 and 139 drugs across 491 and 655 cancer CLs, respectively. A 10-fold cross validation (CV) approach was employed to train DeepDSC on the datasets independently and reported the average root mean squared error (RMSE) and the coefficient of determination (R2) on the held-out sets across the folds. However, the model’s generalizability to novel compounds not in the dataset, specifically compounds that are structurally dissimilar, is yet to be tested. In this regard, the DeepDSC model was retrained, with modifications, on a subset of the GDSC dataset - GDSC2 - to access the generalizability of DL models on BRG399, a proprietary microtubule targeting agent (MTA) developed by BPGbio Inc. GDSC2 was used since data generated for BRG399 and GDSC2 share the same potency assay (CellTiter-Glo [6]), The GDSC2 set was split into training and held-out sets using a 80-20 random split. Further, we chose to optimize the learning rate for both the autoencoder and feed forward networks (FFN), and the L2 regularization factor for the FFN only. Hyperparameters were optimized using 5-fold CV on the training set. Finally, we evaluated the predictive performance of the model on BRG399, which was not present in the training set. We encoded each compound in GDSC2 (n=234) using Circular Fingerprints [4] and observed an average Jaccard (or Tanimoto) distance (JD) [5] of 0.892 across 27261 drug pairs. BRG399 showed an average JD of 0.873 across 234 drug pairs, with the most similar compound JD of 0.785. Our retrained DeepDSC model had RMSE 1.15 and R2 0.82 on the training set and RMSE 1.19 and R2 0.86 on the held-out set. However, on BRG399 tested CLs (n=101) model had RMSE 1.84 and R2 < 0, which indicated that predictions were on average worse compared to a mean-predictor model for BRG399. These results suggest that DL models that can generalize predictions to a held-out set for compounds that were present in the training set may not accurately predict IC50 of novel compounds not in the training set. Citation Format: Nischal M. Chand, Vivek Vishnudas, Michael A. Kiebish, Anjan Thakurta, Niven R. Narain, Stephane Gesta, Gregory M. Miller. Application of a deep learning based drug sensitivity prediction model on a novel anticancer drug [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3527.

Abstract 7348: CytoCellDB: A gold standard database for classification and analysis of extrachromosomal DNA in cancer

Abstract Extrachromosomal DNA (ecDNA), or double minute chromosomes, are established cytogenetic markers for malignancy and genome instability. More recently, the cancer community has gained a heightened awareness of the roles of ecDNA in cancer proliferation, drug resistance and epigenetic remodeling. A current hindrance to understanding the biological roles of ecDNA is the lack of available cell line model systems with experimental cytogenetic data that confirm ecDNA status. Although several recent landmark studies have identified numerous cell lines and tumor models with ecDNA, the current sample size limits our ability to detect ecDNA-driven molecular differences due to limitations in power. Increasing the number of model systems that are known to express ecDNA would provide new avenues for understanding the fundamental underpinnings of ecDNA biology and would unlock a wealth of potential targeting strategies for ecDNA-driven cancers. To bridge this gap, we created CytoCellDB, a resource that provides karyotype annotations, including ecDNA status, and leverages publicly available global cell line data from the Cancer Dependency Map (DepMap) and the Cancer Cell Line Encyclopedia (CCLE). We expanded the total number of cancer cell lines with ecDNA annotations to 632, or 34% of the cell lines in CCLE/DepMap. Of these, 154 cell lines express ecDNA, which is a 240% increase from the current sample size. We demonstrate that a strength of our framework is the ability to interrogate ecDNA, and the compendium of other chromosomal aberrations, in the context of cancer-specific vulnerabilities, drug sensitivities, and other multi-omic features (genomics, transcriptomics, methylation, proteomics). We anticipate that this resource will lower the barrier to systematic, population-scale discoveries related to ecDNA and will provide insights into strategies and best practices for determining novel therapeutics that overcome ecDNA-mediated drug resistance. Citation Format: Liz Brunk. CytoCellDB: A gold standard database for classification and analysis of extrachromosomal DNA in cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7348.

Abstract 4351: Differential dependency mapping of chimeric RNAs across cancer reveals a new landscape of functional fusion transcripts

Abstract Chimeric RNAs are RNA transcripts containing sequences originating from multiple distinct genetic loci and can form through a variety of mechanisms such as DNA rearrangement, cis-splicing of adjacent genes, or RNA trans-splicing. Chimeras have long been known to be a hallmark of cancer and due to their unique properties are promising targets for precision medicine. Appropriately, fusion proteins transcribed from chimeras such as BCR-ABL1 and TPM3-NTRK1 have been shown to be effective targets. Due to these past successes, there exists an opportunity to identify new chimeras as therapeutic targets. While modern chimera prediction software allows for the fast and accurate identification of chimeric RNAs from RNA sequencing data, investigations separating therapeutically relevant transcripts from “transcriptional noise” remain lacking. In this study we performed an in-silico functional screen of chimeras across cell lines representing a wide variety of cancers. We used state-of-the-art chimeric RNA prediction software to create a database of chimeric RNAs across 1017 cell lines from The Cancer Cell Line Encyclopedia. For each chimera we assessed factors such as frequency, recurrence, breakpoint coordinates, frame, and coding potential. To identify specific functional transcripts, we integrated publicly available shRNA knockdown data with our predictions and developed an in-silico functional analysis pipeline comparing differential knockdown effects between chimera and corresponding parental transcripts. For each transcript we assessed the average fold change of chimera-mapping probes, a function score defined as the difference between the fold change of chimera-mapping and parent-specific probes, and a p-value assessing the confidence of our functional score. From our initial screen of 127,819 transcripts, we identified 1088 high-confidence functional chimeras. We successfully identified nearly all known functional chimeras screened including PAX3-FOXO1, EWSR1-FLI1, and TCF3-PBX1. We also identified previously unknown chimeras that that we predict have a function in cell growth and proliferation. Overall, the results of our study reveal a new landscape of functionally relevant chimeric transcripts in cancer. Follow-up studies can further investigate these transcripts to determine their potential as therapeutic targets. Citation Format: Samir Lalani, Sehajroop Gadh, Justin Elfman, Sandeep Singh, Hui Li. Differential dependency mapping of chimeric RNAs across cancer reveals a new landscape of functional fusion transcripts [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4351.

Abstract 3590: Tumor intrinsic production of glucocorticoids in ovarian carcinoma

Abstract Background: Cortisol is a glucocorticoid (GC) known to be produced by the hypothalamic-pituitary-adrenal (HPA) axis. Dysregulation of the HPA axis is prevalent in patients with ovarian cancer. These patients have elevated nighttime levels of cortisol and increased tumoral expression of the glucocorticoid receptor (GR); both of which correlate with shortened overall survival (OS). We hypothesize that outside the HPA axis, ovarian cancer may have tumor-intrinsic production of GC. In this study, we examined whether 11 beta-hydroxysteroid dehydrogenase 1 (11β-HSD1; enzyme which activates cortisone to cortisol) has steroidogenic activity in the ovarian cancer tumor microenvironment (TME). Methods: We used the Genotype-Tissue Expression (GTex) Portal and the Cancer Cell Line Encyclopedia (CCLE) to determine expression of steroidogenic enzymes in the ovary and ovarian cancer cell lines. We further performed UMAP analysis of a single-cell RNA-seq dataset of tumor samples from patients with ovarian carcinoma to cluster for cell type. ELISAs were performed to quantify cortisol production and secretion by cells stimulated with metabolite, cortisone. Subcellular fractionation and Western blotting were performed to measure protein expression of 11β-HSD1, GR, and phospho-Ser 211 GR. Results: HSD11B1 mRNA (which encodes 11β-HSD1) is detectable in bulk ovarian tissue at a median expression of 78.19 transcripts per million (TPM) as measured in the GTex data. In vitro, primary human ovarian fibroblasts express functional 11β-HSD1 and produce and secrete cortisol upon cortisone stimulation in a time-dependent manner, with a mean total secretion of 17.91 ± 0.21 ng/ml at 12 hours. Analysis of scRNA-seq data identifies that cancer-associated fibroblasts (CAFs) may be a source of HSD11B1 in the ovarian TME, with roughly 9% of CAFs expressing HSD11B1. In HeyA8 ovarian cancer cells (positive for 11β-HSD1), cortisone stimulation in vitro results in cortisol production with a mean total secretion of 22.04 ± 5.46 ng/ml at 12 hours. Moreover, this cortisol results in GR activation in an 11β-HSD1-dependent manner. Discussion: We identify 11β-HSD1 as a potential mechanism of GC production within ovarian cancer, with CAFs being a putative source. We further show that 11β-HSD1 activity can lead to activation of the GR within cancer cells. Citation Format: Pahul Hanjra, Elaine Stur, Steven Cole, Susan Lutgendorf, Anil Sood. Tumor intrinsic production of glucocorticoids in ovarian carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3590.

Abstract 5591: ARID domain containing BAF subunits exhibit neomorphism: Have we identified the new Achilles heel in cancer

Abstract The SWI/SNF chromatin remodeler utilizes energy of ATP hydrolysis to slide or evict nucleosomes or histones, respectively, thus enabling nuclear processes by driving an ‘open’ chromatin architecture. The mammalian SWI/SNF, also termed the BrG1/Brm associated factor (BAF) complex, is the major chromatin remodeler in ontogeny and adult life. Three major BAF complexes namely canonical (cBAF), polybromo (pBAF), and non-canonical (ncBAF) have been studied. BAF components, classified as a tumor suppressors, exhibit frequent mutations in cancers. Though, studies have evaluated functional roles of BAF subunits, very few attempted to study BAF mutations identified across cancer types. In our previous studies, we identified tumor specific mutations in the gene encoding ‘A-T rich interaction domain containing 1B’ (ARID1B), within its nuclear localization signal (NLS). NLS-inactivated ARID1B exhibited cytoplasmic localization and promoted tumorigenesis by activating the ERK and Wnt/β-catenin pathways. More recently, we detected inactivating mutations in the gene encoding ARID2, a component of the pBAF complex and an ARID1B paralogue. We now present detailed characterization of tumor-specific ARID2 truncations that were presumed to be inactivating in nature. One such truncation mutation viz. p.Ser989fsArg21*, identified in our previous study, is expected to yield an ARID2 protein truncating at amino acid position 1010 (as against the full length of 1835 amino acids). Scrutiny of The Cancer Genome Atlas cBioPortal and Cancer Cell Line Encyclopedia mutation data bases revealed several ARID2 mutations expected to truncate the protein around the 1010 position. Surprisingly, ectopic expression of truncated ARID2 forms promoted oncogenic function in both ARID2 proficient and deficient backgrounds, thus validating a neomorphic gain of function for the truncated protein. Fluorescence based intracellular localization assays revealed a surprising cytoplasmic localization of truncated ARID2. Further, we mapped the ARID2 NLS by in silico analysis followed by evaluation of several deletion constructs, to be located at 1488-1518 position; thus, explaining the cytoplasmic localization of ARID2 forms truncating around 1010 position. Further, truncated ARID2 was unable to bind other pBAF components and did not sequester other pBAF components to the cytoplasm, in both ARID2 proficient and deficient cells. Tandem Affinity Purification Mass-spectrometry revealed putative interaction partners for truncated ARID2; analysis is currently underway to reveal the mechanism of tumorigenic induction exhibited by truncated ARID2. Immunohistochemistry on colorectal and breast cancer tissue microarrays revealed ARID2 cytoplasmic localization in a significant fraction of samples. We have discovered a novel YIN-YANG function of BAF components expected to yield efficient therapeutic options in cancer. Citation Format: Sanjana Sarkar, Murali Dharan Bashyam. ARID domain containing BAF subunits exhibit neomorphism: Have we identified the new Achilles heel in cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5591.

Comprehensive Pan-cancer Analysis of CMPK2 as Biomarker and Prognostic Indicator for Immunotherapy.

UMP-CMP kinase 2 (CMPK2) is involved in mitochondrial DNA synthesis which can be oxidized and released into the cytoplasm in innate immunity. It initiates the assembly of NLRP3 inflammasomes and mediates various pathological processes such as human immunodeficiency virus infection and systemic lupus erythematosus. However the role of CMPK2 in tumor progression and tumor immunity remains unclear. In this study we conducted a systematical analysis of CMPK2 across 33 different cancers based on datasets such as Genotype Tissue-Expression (GTEx) The Cancer Genome Atlas (TCGA) the Cancer Cell Line Encyclopedia (CCLE) and Tumor Immune Syngeneic Mouse (TISMO). Our focus encompassed the characterization of CMPK2 expression patternsclinical significance potential regulatory mechanisms and its relationship with the tumor immune profile including responsiveness to immune checkpoint inhibitor treatment. CMPK2 expression was elevated in 23 cancers and decreased in two cancers. Receiver operating characteristic curve analysis indicated that CMPK2 expression had a high diagnostic value for 16 cancers. Kaplan-Meier survival analysis showed that high CMPK2 expression was associated with Lower Overall Survival (OS)Disease- Specific Survival (DSS) and Progression-Free Interval (PFI) in Kidney Cutaneous Chromophobe (KICH) Uterine Corpus Endometrial Carcinoma (UCEC) and Uveal Melanoma (UVM) and the opposite was true in Skin Cutaneous Melanoma (SKCM). Immune microenvironment-related analysis revealed strong associations between CMPK2 expression and immune cell infiltration as well as immune checkpoint expression across various tumors. Notably in four mouse immunotherapy cohorts CMPK2 expression in treated mouse tumors was post-treatment. In five clinical immunotherapy cohorts patients with high CMPK2 expression show better responses to immunotherapy. Furthermore the methylation level of the CMPK2 gene was closely correlated to its expression and tumor prognosis. Among these cancers the clinical and immunological indications of SKCM are particularly closely related to CMPK2 expression. Our analysis preliminarily describes the complex function of CMPK2 in cancer progression and immune microenvironment highlighting its potential as a diagnostic and therapeutic target for immunotherapy.