Large-scale Phenotypic Screens Research Articles

A deep semantic network-based image segmentation of soybean rust pathogens.

Asian soybean rust is a highly aggressive leaf-based disease triggered by the obligate biotrophic fungus Phakopsora pachyrhizi which can cause up to 80% yield loss in soybean. The precise image segmentation of fungus can characterize fungal phenotype transitions during growth and help to discover new medicines and agricultural biocides using large-scale phenotypic screens. The improved Mask R-CNN method is proposed to accomplish the segmentation of densely distributed, overlapping and intersecting microimages. First, Res2net is utilized to layer the residual connections in a single residual block to replace the backbone of the original Mask R-CNN, which is then combined with FPG to enhance the feature extraction capability of the network model. Secondly, the loss function is optimized and the CIoU loss function is adopted as the loss function for boundary box regression prediction, which accelerates the convergence speed of the model and meets the accurate classification of high-density spore images. The experimental results show that the mAP for detection and segmentation, accuracy of the improved algorithm is improved by 6.4%, 12.3% and 2.2% respectively over the original Mask R-CNN algorithm. This method is more suitable for the segmentation of fungi images and provide an effective tool for large-scale phenotypic screens of plant fungal pathogens.

Large-Scale Evaluation of Indonesian Elite Maize Breeding Lines for Resistance Against Bacterial Stalk Rot Caused by Dickeya zeae

Bacterial stalk rot is one of the important diseases in maize caused by <em>Dickeya zeae</em>. Infection of this disease can lead to a considerable amount of loss in yield, with up to 98 percent loss of yield. The use of resistant materials is the most effective approach to managing bacterial stalk rot in maize. This study evaluates a large-scale phenotypic screening of 624 maize lines against bacterial stalk rot, divided into two groups based on a heterotic pool. These lines are used in a commercial breeding program in Indonesia. This study develops a stabbing method with a large gauge hypodermic needle, allowing for scalability in delivering inoculum while performing large-scale line evaluation. The result from ANOVA reveals a significant (P<0.05) effect of lines, day after infection, and interaction between day after infection and group. Group two displays fewer resistant lines compared to group one. This work presents a method for large-scale line evaluation for resistance against bacterial stalk rot, where the information obtained can be used in industrial breeding programs for routine material screening during the development of new lines or hybrids and for genomic studies of bacterial stalk rot resistance.

PHDs-seq: a large-scale phenotypic screening method for drug discovery through parallel multi-readout quantification

High-throughput phenotypic screening is a cornerstone of drug development and the main technical approach for stem cell research. However, simultaneous detection of activated core factors responsible for cell fate determination and accurate assessment of directional cell transition are difficult using conventional screening methods that focus on changes in only a few biomarkers. The PHDs-seq (Probe Hybridization based Drug screening by sequencing) platform was developed to evaluate compound function based on their transcriptional effects in a wide range of signature biomarkers. In this proof-of-concept demonstration, several sets of markers related to cell fate determination were profiled in adipocyte reprogramming from dermal fibroblasts. After validating the accuracy, sensitivity and reproducibility of PHDs-seq data in molecular and cellular assays, a panel of 128 signalling-related compounds was screened for the ability to induce reprogramming of keloid fibroblasts (KF) into adipocytes. Notably, the potent ATP-competitive VEGFR/PDGFR inhibitor compound, ABT869, was found to promote the transition from fibroblasts to adipocytes. This study highlights the power and accuracy of the PHDs-seq platform for high-throughput drug screening in stem cell research, and supports its use in basic explorations of the molecular mechanisms underlying disease development.

Abstract 1883: Large scale viability screening with PRISM underscores non-inhibitory properties of small molecules

Abstract Oncology drug discovery has been dependent on phenotypic screening that lead to discovery of many drugs that remain essential cancer treatments to this day. Historically, logistical and technological limitations have hampered the scale of phenotypic small molecule screening across genomic diverse models. Until recently, large scale phenotypic screening across many cell lines, reflecting the somatic diversity in cancer, has not been possible. The PRISM technology does allow for a high-throughput approach to phenotypic viability screening by pooling cell lines and deconvoluting their response to small molecules using unique integrated DNA-barcodes. Here, we report on the results of a large phenotypic screening effort to identify novel small molecules and targets that can expand our knowledge on cancer vulnerabilities and can spur development of phenotypic drug discovery. A library of 18.000 compounds was assembled based on structural diversity and ability to induce gene-expression changes. Of these 18.000 compounds, around 3.500 were found to reduce cell viability of at least 5 cell lines in the PRISM assay. Of the cytotoxic compounds, we did not identify many small molecules that correlated strongly with CRISPR KO dependency data from project Achilles, and instead found that gene-expression correlations hold much more predictive value. By using additional approaches, we were able to identify the target of select novel cytotoxic small molecules and their respective mechanism-of-action. To our surprise, we have identified few compounds that act as canonical inhibitors of cancer drivers, and instead found compounds that act as molecular glues, prodrugs or inhibitors of more general cellular processes. These findings make a strong case for large scale phenotypic viability screening across genomically diverse models as a complementary approach to target identification next to target based drug discovery. Citation Format: Lucian de Waal, Kevin Larpenteur, Li Wang, Cong Zhu, Jordan Bryan, Mustafa Kocak, Corrie Aghia, Joshua Bittker, Victor Jones, Anita Vrcic, Edward McBride, Samantha Bender, Kalea Gore, Frank Li, Zhenghao Chen, Aviad Tsherniak, Aravind Subramanian, Daphne Koller, Jeffery Settleman, Chris C. Mader, David Stokoe, Ari Firestone, Jennifer A. Roth, Todd R. Golub. Large scale viability screening with PRISM underscores non-inhibitory properties of small molecules [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 1883.

Controlling astrocyte-mediated synaptic pruning signals for schizophrenia drug repurposing with deep graph networks.

Schizophrenia is a debilitating psychiatric disorder, leading to both physical and social morbidity. Worldwide 1% of the population is struggling with the disease, with 100,000 new cases annually only in the United States. Despite its importance, the goal of finding effective treatments for schizophrenia remains a challenging task, and previous work conducted expensive large-scale phenotypic screens. This work investigates the benefits of Machine Learning for graphs to optimize drug phenotypic screens and predict compounds that mitigate abnormal brain reduction induced by excessive glial phagocytic activity in schizophrenia subjects. Given a compound and its concentration as input, we propose a method that predicts a score associated with three possible compound effects, i.e., reduce, increase, or not influence phagocytosis. We leverage a high-throughput screening to prove experimentally that our method achieves good generalization capabilities. The screening involves 2218 compounds at five different concentrations. Then, we analyze the usability of our approach in a practical setting, i.e., prioritizing the selection of compounds in the SWEETLEAD library. We provide a list of 64 compounds from the library that have the most potential clinical utility for glial phagocytosis mitigation. Lastly, we propose a novel approach to computationally validate their utility as possible therapies for schizophrenia.

Determining the Relationship Between Clofazimine Gastrointestinal Levels and anti‐CryptosporidiumEfficacy in Mice

IntroductionCryptosporidiosis, the disease caused by parasitic infections with Cryptosporidium spp. can cause long‐term adverse impacts and even death in malnourished children and immunocompromised patients. The only FDA‐approved drug for treating cryptosporidiosis, nitazoxanide, has limited efficacy in the populations that are impacted most by the disease. A large‐scale phenotypic screening campaign identified clofazimine as active in both in vitro and in vivo preclinical models of Cryptosporidium infection. However, clofazimine was not effective in controlling Cryptosporidium infection in a subsequent Phase 2a clinical trial in HIV‐positive adults with cryptosporidiosis. Our recent pharmacokinetic/pharmacodynamic (PK/PD) modeling work demonstrated that plasma clofazimine levels observed in the Phase 2a clinical trial fell well below projected efficacious plasma concentrations estimated with preclinical in vivo data. Therefore, suboptimal clofazimine exposure in the Phase 2a trial likely contributed to the observed lack of efficacy. Nevertheless, plasma clofazimine levels may not represent local concentrations at the gastrointestinal (GI) tract, the site of action for treatment of Cryptosporidium infection. Thus, the objective of our current study is to build a physiologically based pharmacokinetic (PBPK) model of clofazimine in mice to simulate clofazimine GI pharmacokinetics (PK), estimate efficacious clofazimine concentrations at the tissue level, and explore important factors that can impact clofazimine exposure at the site of infection.MethodsC57BL/6 mice were administered a single oral dose of clofazimine at dose levels ranging 0.03‐300 mg/kg. Blood samples were collected up to 240 hours post‐dosing and GI samples were collected at 240 hours post‐dosing. In tandem, an efficacy study treated C. parvum‐infected IFN‐γ‐/‐ mice with equivalent clofazimine dosing and oocysts in fecal samples were quantified over time. GastroPlus (Simulations Plus) was used for modeling and simulation.ResultsThe final PBPK model captured clofazimine plasma PK in mice well across most doses (within a 2‐fold margin). The simulated clofazimine GI concentrations increased with increasing dose in a non‐proportional manner, similar to what was observed with the plasma PK. As expected for a drug with poor solubility, the non‐linearity was attributed to diminished drug dissolution at higher dose levels. The simulated clofazimine tissue concentrations were used to construct PK/PD models to predict efficacious tissue concentrations.ConclusionA clofazimine PBPK model was generated in mice to simulate clofazimine levels in plasma and the GI tract. By incorporating clofazimine efficacy data from a mouse model of Cryptosporidium infection, we identified a relationship between tissue drug concentrations and reduction in infection burden. Establishing a PK/PD relationship informed by GI drug levels will assist future drug development efforts for cryptosporidiosis.

Machine Learning Models for Mycobacterium tuberculosisIn Vitro Activity: Prediction and Target Visualization.

Tuberculosis (TB) is a major global health challenge, with approximately 1.4 million deaths per year. There is still a need to develop novel treatments for patients infected with Mycobacterium tuberculosis (Mtb). There have been many large-scale phenotypic screens that have led to the identification of thousands of new compounds. Yet, there is very limited investment in TB drug discovery which points to the need for new methods to increase the efficiency of drug discovery against Mtb. We have used machine learning approaches to learn from the public Mtb data, resulting in many data sets and models with robust enrichment and hit rates leading to the discovery of new active compounds. Recently, we have curated predominantly small-molecule Mtb data and developed new machine learning classification models with 18 886 molecules at different activity cutoffs. We now describe the further validation of these Bayesian models using a library of over 1000 molecules synthesized as part of EU-funded New Medicines for TB and More Medicines for TB programs. We highlight molecular features which are enriched in these active compounds. In addition, we provide new regression and classification models that can be used for scoring compound libraries or used to design new molecules. We have also visualized these molecules in the context of known molecular targets and identified clusters in chemical property space, which may aid in future target identification efforts. Finally, we are also making these data sets publicly available, representing a significant increase to the available Mtb inhibition data in the public domain.

The apicoplast link to fever-survival and artemisinin-resistance in the malaria parasite

The emergence and spread of Plasmodium falciparum parasites resistant to front-line antimalarial artemisinin-combination therapies (ACT) threatens to erase the considerable gains against the disease of the last decade. Here, we develop a large-scale phenotypic screening pipeline and use it to carry out a large-scale forward-genetic phenotype screen in P. falciparum to identify genes allowing parasites to survive febrile temperatures. Screening identifies more than 200 P. falciparum mutants with differential responses to increased temperature. These mutants are more likely to be sensitive to artemisinin derivatives as well as to heightened oxidative stress. Major processes critical for P. falciparum tolerance to febrile temperatures and artemisinin include highly essential, conserved pathways associated with protein-folding, heat shock and proteasome-mediated degradation, and unexpectedly, isoprenoid biosynthesis, which originated from the ancestral genome of the parasite’s algal endosymbiont-derived plastid, the apicoplast. Apicoplast-targeted genes in general are upregulated in response to heat shock, as are other Plasmodium genes with orthologs in plant and algal genomes. Plasmodium falciparum parasites appear to exploit their innate febrile-response mechanisms to mediate resistance to artemisinin. Both responses depend on endosymbiont-derived genes in the parasite’s genome, suggesting a link to the evolutionary origins of Plasmodium parasites in free-living ancestors.

Identification and Validation of Molecular Markers of Tumor Cell Resistance to Natural Killer Cells through CRISPR-Based Screens and Large-Scale Phenotypic Screens of Pooled Tumor Cell Lines

Natural killer (NK) cellsexhibit potent activity in pre-clinical models of diverse hematologic malignancies and solid tumors. The in vivo anti-tumor effects of NK cells are not confounded by graft vs. host disease, providing a key advantage over strategies exploiting T cell reactivity. Indeed, infusion of high numbers of NK cells, either autologous or allogeneic, after their ex vivo expansion and activation, has been feasible and safe in clinical studies. Because of these considerations, NK cells had been viewed as a promising therapeutic approach for a disease like multiple myeloma (MM), which remains incurable, even after initial responses to novel immunotherapies, including anti-CD38 antibodies; combinations of thalidomide derivatives with immune checkpoint blockade; and even CAR-T cells. However, our prior studies and more recent unpublished work indicate that tumor cells can exhibit decreased response to NK due to the protective effect of nonmalignant mesenchymal stromal cells and depending on the genetic background of the tumor cells. To our knowledge, there have been limited, if any, reports of systematic screens to quantify NK cell cytotoxicity against large panels of molecularly characterized tumor cell lines. Moreover, while there have been earlier RNAi-based studies, to the best of our knowledge there have not been any genome-wide CRISPR-based screens for candidate NK cell resistance genes. To address this void, we sought to identify and validate mechanisms regulating tumor cell response to NK cells, by performing orthogonal screens in 2 systems: (i) Using similar approaches as in other screens conducted by our lab in other models, we performed a genome-wide CRISPR gene editing screen (GECKO library) in the NK cell-sensitive colorectal Ca line HCT15 and identified candidate genes for which knockout confers resistance to the NK cell line KHYG1. (ii) We studied the cytotoxicity of NK cells (primary NK cells or KHYG1; at different time points and effector-to-tumor [E2T] ratios) against a pool of DNA-barcoded 568 lineage diverse solid tumor lines from the Cancer Cell Line Encyclopedia (CCLE), using the PRISM multiplexed cell viability screening technology. This PRISM screen expanded on a previous screen across 100 cancer cell lines and yielded consistent results supporting the reproducibility of the method. We correlated NK cell cytotoxicity with the CCLE genomic feature data set (mRNA expression, copy number variation, and gene mutational statues) for each of the 568 cell lines. The screens provided candidate “hits” which were concordant in both studies and included recognizable regulators of NK cell biology (e.g. loss-of-function of death receptors such as FAS or gain-of-function of HLA-E correlating with NK cell resistance); as well as other plausible, but previously under-appreciated candidates, which will be considered for further studies in this project. We did not observe significant tumor type-specific differences in NK cell responses in this screen, suggesting that candidate markers identified in these studies may be relevant across lineages, including MM and blood cancers (even though these lines were not included so far in our PRISM or CRISPR screens). Additional ongoing “multi-omic” analyses seek to further examine the links between the phenotypic responses to NK cells with the molecular features of these cell lines (e.g. integrating mutations, DNA methylation, proteomic [RPPA] etc.). Our study highlights the feasibility of combining CRISPR/Cas9-based functional genomic screens (on genome-wide and sub-genome scale) on individual cell lines with large scale phenotypic screens on pooled barcoded cell lines to identify candidate biomarkers of tumor cell sensitivity vs. resistance to NK cells. Notably our PRISM screen represents, to our knowledge, the largest effort to-date to correlate the molecular annotation of tumor cells with their response to uniform treatment, pre-clinical or clinical, to a cell-based immunotherapy. DisclosuresMitsiades:Novartis: Research Funding; Janssen/Johnson & Johnson: Research Funding; Ono: Research Funding; TEVA: Research Funding; Abbvie: Research Funding; Takeda: Other: Employment of family member.

A High-Content Screening Assay for Small Molecules That Stabilize Mutant Triose Phosphate Isomerase (TPI) as Treatments for TPI Deficiency.

Triose phosphate isomerase deficiency (TPI Df) is an untreatable, childhood-onset glycolytic enzymopathy. Patients typically present with frequent infections, anemia, and muscle weakness that quickly progresses with severe neuromusclar dysfunction requiring aided mobility and often respiratory support. Life expectancy after diagnosis is typically ~5 years. There are several described pathogenic mutations that encode functional proteins; however, these proteins, which include the protein resulting from the "common" TPIE105D mutation, are unstable due to active degradation by protein quality control (PQC) pathways. Previous work has shown that elevating mutant TPI levels by genetic or pharmacological intervention can ameliorate symptoms of TPI Df in fruit flies. To identify compounds that increase levels of mutant TPI, we have developed a human embryonic kidney (HEK) stable knock-in model expressing the common TPI Df protein fused with green fluorescent protein (HEK TPIE105D-GFP). To directly address the need for lead TPI Df therapeutics, these cells were developed into an optical drug discovery platform that was implemented for high-throughput screening (HTS) and validated in 3-day variability tests, meeting HTS standards. We initially used this assay to screen the 446-member National Institutes of Health (NIH) Clinical Collection and validated two of the hits in dose-response, by limited structure-activity relationship studies with a small number of analogs, and in an orthogonal, non-optical assay in patient fibroblasts. The data form the basis for a large-scale phenotypic screening effort to discover compounds that stabilize TPI as treatments for this devastating childhood disease.

Current Strategies of Antiviral Drug Discovery for COVID-19.

SARS-CoV-2 belongs to the family of enveloped, single-strand RNA viruses known as Betacoronavirus in Coronaviridae, first reported late 2019 in China. It has since been circulating world-wide, causing the COVID-19 epidemic with high infectivity and fatality rates. As of the beginning of April 2021, pandemic SARS-CoV-2 has infected more than 130 million people and led to more than 2.84 million deaths. Given the severity of the epidemic, scientists from academia and industry are rushing to identify antiviral strategies to combat the disease. There are several strategies in antiviral drugs for coronaviruses including empirical testing of known antiviral drugs, large-scale phenotypic screening of compound libraries and target-based drug discovery. To date, an increasing number of drugs have been shown to have anti-coronavirus activities in vitro and in vivo, but only remdesivir and several neutralizing antibodies have been approved by the US FDA for treating COVID-19. However, remdesivir’s clinical effects are controversial and new antiviral drugs are still urgently needed. We will discuss the current status of the drug discovery efforts against COVID-19 and potential future directions. With the ever-increasing movability of human population and globalization of world economy, emerging and reemerging viral infectious diseases seriously threaten public health. Particularly the past and ongoing outbreaks of coronaviruses cause respiratory, enteric, hepatic and neurological diseases in infected animals and human (Woo et al., 2009). The human coronavirus (HCoV) strains (HCoV-229E, HCoV-OC43, HCoV-NL63, and HCoV-HKU1) usually cause common cold with mild, self-limiting upper respiratory tract infections. By contrast, the emergence of three deadly human betacoronaviruses, middle east respiratory syndrome coronavirus (MERS) (Zaki et al., 2012), severe acute respiratory syndrome coronavirus (SARS-CoV) (Lee et al., 2003), the SARS-CoV-2 (Jin et al., 2020a) highlight the need to identify new treatment strategies for viral infections. SARS-CoV-2 is the etiological agent of COVID-19 disease named by World Health Organization (WHO) (Zhu N. et al., 2020). This disease manifests as either an asymptomatic infection or a mild to severe pneumonia. This pandemic disease causes extent morbidity and mortality in the whole world, especially regions out of China. Similar to SARS and MERS, the SARS CoV-2 genome encodes four structural proteins, sixteen non-structural proteins (nsp) and accessory proteins. The structural proteins include spike (S), envelope (E), membrane (M), nucleoprotein (N). The spike glycoprotein directly recognizes and engages cellular receptors during viral entry. The four non-structural proteins including papain-like protease (PLpro), 3-chymotrypsin-like protease (3CLpro), helicase, and RNA-dependent RNA polymerase (RdRp) are key enzymes involved in viral transcription and replication. The spike and the four key enzymes were considered attractive targets to develop antiviral agents (Zumla et al., 2016). The catalytic sites of the four enzymes of SARS-CoV2 share high similarities with SARS CoV and MERS in genomic sequences (Morse et al., 2020). Besides, the structures of the key drug-binding pockets are highly conserved among the three coronaviruses (Morse et al., 2020). Therefore, it follows naturally that existing anti-SARS-CoV and anti-MERS drugs targeting these enzymes can be repurposed for SARS-CoV-2. Based on previous studies in SARS-CoV and MERS-CoV, it is anticipated a number of therapeutics can be used to control or prevent emerging infectious disease COVID-19 (Li and de Clercq, 2020; Wang et al., 2020c; Ita, 2021), these include small-molecule drugs, peptides, and monoclonal antibodies. Given the urgency of the SARS-CoV-2 outbreak, here we discuss the discovery and development of new therapeutics for SARS-CoV-2 infection based on the strategies from which the new drugs are derived.

The Small RNA sr8384 Is a Crucial Regulator of Cell Growth in Solventogenic Clostridia.

Small RNAs (sRNAs) are crucial regulatory molecules in organisms and are well-known not only for their roles in the control of diverse crucial biological processes but also for their value in regulation rewiring. However, to date, in Gram-positive anaerobic solventogenic clostridia (a group of important industrial bacteria with exceptional substrate and product diversity), sRNAs remain minimally explored, and thus there is a lack of detailed understanding regarding these important molecules and their use as targets for genetic improvement. Here, we performed large-scale phenotypic screens of a transposon-mediated mutant library of Clostridium acetobutylicum, a typical solventogenic clostridial species, and discovered a novel sRNA (sr8384) that functions as a crucial regulator of cell growth. Comparative transcriptomic data combined with genetic and biochemical analyses revealed that sr8384 acts as a pleiotropic regulator and controls multiple targets that are associated with crucial biological processes through direct or indirect interactions. Notably, the in vivo expression level of sr8384 determined the cell growth rate, thereby affecting the solvent titer and productivity. These findings indicate the importance of the sr8384-mediated regulatory network in C. acetobutylicum Furthermore, a homolog of sr8384 was discovered and proven to be functional in another important Clostridium species, C. beijerinckii, suggesting the potential broad role of this sRNA in clostridia. Our work showcases a previously unknown potent and complex role of sRNAs in clostridia, providing new opportunities for understanding and engineering these anaerobes.IMPORTANCE The uses of sRNAs as new resources for functional studies and strain modifications are promising strategies in microorganisms. However, these crucial regulatory molecules have hardly been explored in industrially important solventogenic clostridia. Here, we identified sr8384 as a novel determinant sRNA controlling the cell growth of solventogenic Clostridium acetobutylicum Based on a detailed functional analysis, we further reveal the pleiotropic function of sr8384 and its multiple direct and indirect crucial targets, which represents a valuable source for understanding and optimizing this anaerobe. Of note, manipulation of this sRNA achieves improved cell growth and solvent synthesis. Our findings provide a new perspective for future studies on regulatory sRNAs in clostridia.

Abstract C022: Large scale phenotypic screening using PRISM, integrated with functional genomic screening and transcriptional profiling accelerates target identification of cytotoxic small molecules

Abstract A comprehensive analysis of FDA approved drugs has shown that drug discovery through phenotypic screening yields a higher percentage of first-in-class compounds, compared to canonical target-based drug discovery. This has led to a renewed interest in phenotypic screening to identify novel mechanisms of action. The PRISM technology of adding a unique DNA barcode to cancer cell lines has been able to reduce the cost and time associated with phenotypic viability screening. Using the PRISM method, we profiled the cytotoxic activity of over 4,100 bioactive compounds in 578 barcoded cancer cell lines at one dose, followed by 8-point dose of 1100 active compounds. The use of genomically characterized cell lines from the Cancer Cell Line Encyclopedia (CCLE) allows us to identify baseline features associated with sensitivity to a particular compound. One of the compounds from the screen that we selected for follow up was BRD-K01353379, which killed 28 cancer cell lines of diverse lineages and showed a correlation between high expression of Ethanolamine Kinase 1 (ETNK1) and reduction of cell viability. Interestingly, BRD-3379 showed a similar PRISM cytotoxic profile to 2,3-DCPE, that was screened in our Repurposing effort. 2,3-DCPE is structurally similar to BRD-3379 and has been reported to induce cytotoxicity through upregulation of p21 and downregulation of Bcl-XL. To elucidate the mechanism of action of this compound we performed a genome-wide CRISPR-Cas9 resistance screen to identify molecular targets that are required for sensitivity to BRD-3379. Here, we found that loss of expression of the PRISM biomarker ETNK1 conferred resistance in a BRD-3379 sensitive cell line, JHH-7. Subsequently, we profiled metabolites in a cell line sensitive to BRD-3379 and found that the compound was phosphorylated. We confirmed in a biochemical assay that BRD-3379 is in fact a substrate of recombinant ETNK1, which phosphorylates the ethanolamine moiety of BRD-3379. Next, we transcriptionally profiled the effect of BRD-3379 by bulk RNAseq in a sensitive and non-sensitive cell line, as well as in a pool of 100 cell lines using single cell RNAseq [McFarland et al., manuscript in preparation]. Both bulk RNAseq and single cell transcriptional analysis showed that upregulation of both p21 and p57 is strongly associated with sensitivity to BRD-3379. This mechanism of action is similar to that of 2,3-DCPE. Here, we have demonstrated that by combining large-scale cell line viability screening, genomic characterization of these cell lines, genome wide CRISPR-Cas9 resistance screens, transcriptional profiling, and biochemical assays, we can facilitate rapid identification of molecular targets of small molecules and their respective mechanism of action by which they induce cytotoxicity. Citation Format: Kevin Larpenteur, Lucian de Waal, Li Wang, Mustafa Kocak, Jordan Bryan, Samantha Bender, Sulyman Barkho, Matt Kukurugya, Amy Goodale, Brenton Paolella, Min Cho, James McFarland, Michael Rothberg, Cong Zhu, Colin Trepiccio, Aviad Tsherniak, David Root, Bryson Bennett, Federica Piccioni, Josh Bittker, James Rong, Christopher Mader, David Stokoe, Ari Firestone, Todd R Golub, Jennifer Roth. Large scale phenotypic screening using PRISM, integrated with functional genomic screening and transcriptional profiling accelerates target identification of cytotoxic small molecules [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr C022. doi:10.1158/1535-7163.TARG-19-C022

A suite of phenotypic assays to ensure pipeline diversity when prioritizing drug-like Cryptosporidium growth inhibitors

Cryptosporidiosis is a leading cause of life-threatening diarrhea in children, and the only currently approved drug is ineffective in malnourished children and immunocompromised people. Large-scale phenotypic screens are ongoing to identify anticryptosporidial compounds, but optimal approaches to prioritize inhibitors and establish a mechanistically diverse drug development pipeline are unknown. Here, we present a panel of medium-throughput mode of action assays that enable testing of compounds in several stages of the Cryptosporidium life cycle. Phenotypic profiles are given for thirty-nine anticryptosporidials. Using a clustering algorithm, the compounds sort by phenotypic profile into distinct groups of inhibitors that are either chemical analogs (i.e. same molecular mechanism of action (MMOA)) or known to have similar MMOA. Furthermore, compounds belonging to multiple phenotypic clusters are efficacious in a chronic mouse model of cryptosporidiosis. This suite of phenotypic assays should ensure a drug development pipeline with diverse MMOA without the need to identify underlying mechanisms.

Abstract 3058: Data and resource sharing: Advancing our understanding of glioblastoma

Abstract Glioblastoma multiforme (GBM), characterized by an aggressive clinical course, therapeutic resistance, and striking molecular heterogeneity, remains incurable. GBM has a median survival of only 12-15 months post-surgery in adults even with standard of care chemotherapy and radiation. Through a collaborative pan-Canadian Stand up to Cancer (SU2C) Stem Cell Dream Team effort, our group is focusing on brain tumour stem cells (BTSCs), a subpopulation of cells within tumours, which are hypothesized to drive tumour growth, relapse, and resistance to conventional chemotherapies. We are completing large-scale phenotypic cell biology, tumourigenicity, drug screening, and multi-platform ‘omic studies (genomics, bulk and single cell transcriptomics, epigenomics, metabolomics, proteomics, ATACseq) on a large number of adult GBM patient-derived BTSC cultures. Using both in vitro and in vivo approaches and the above multi-platform approaches, we are investigating the unique growth and stem cell characteristics of the BTSCs. In addition, we are using BTSC cultures to investigate drug response and genetic targeting approaches to identify new therapeutic strategies for clinical translation and understand mechanisms of resistance and recurrence. The wealth of information derived from this project is being integrated using systems biology approaches and will be shared in the public domain via data visualization and big data sharing platforms. Contemporaneously, we are depositing a number of well-characterized BTSC cultures in the American Type Culture Collection repository for unrestricted access to the scientific community. Our cross-Canada multi-disciplinary Cancer Stem Cell Dream Team, comprised of scientists, clinicians and patient advocates supports the ideal of data and resource sharing to help improve patient outcome for this devastating disease. Our goal is to provide the broader community of scientists, clinicians and patient groups an enduring resource and data sharing model to advance our understanding of glioblastoma. Citation Format: Hema A. Luchman, Fiona Coutinho, Samuel Weiss, Stand Up to Cancer Stem Cell Dream Team Consortium, Peter Dirks. Data and resource sharing: Advancing our understanding of glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3058.

Comparing and Validating Machine Learning Models for Mycobacterium tuberculosis Drug Discovery.

Tuberculosis is a global health dilemma. In 2016, the WHO reported 10.4 million incidences and 1.7 million deaths. The need to develop new treatments for those infected with Mycobacterium tuberculosis ( Mtb) has led to many large-scale phenotypic screens and many thousands of new active compounds identified in vitro. However, with limited funding, efforts to discover new active molecules against Mtb needs to be more efficient. Several computational machine learning approaches have been shown to have good enrichment and hit rates. We have curated small molecule Mtb data and developed new models with a total of 18,886 molecules with activity cutoffs of 10 μM, 1 μM, and 100 nM. These data sets were used to evaluate different machine learning methods (including deep learning) and metrics and to generate predictions for additional molecules published in 2017. One Mtb model, a combined in vitro and in vivo data Bayesian model at a 100 nM activity yielded the following metrics for 5-fold cross validation: accuracy = 0.88, precision = 0.22, recall = 0.91, specificity = 0.88, kappa = 0.31, and MCC = 0.41. We have also curated an evaluation set ( n = 153 compounds) published in 2017, and when used to test our model, it showed the comparable statistics (accuracy = 0.83, precision = 0.27, recall = 1.00, specificity = 0.81, kappa = 0.36, and MCC = 0.47). We have also compared these models with additional machine learning algorithms showing Bayesian machine learning models constructed with literature Mtb data generated by different laboratories generally were equivalent to or outperformed deep neural networks with external test sets. Finally, we have also compared our training and test sets to show they were suitably diverse and different in order to represent useful evaluation sets. Such Mtb machine learning models could help prioritize compounds for testing in vitro and in vivo.

Development of Anti-Virulence Approaches for Candidiasis via a Novel Series of Small-Molecule Inhibitors of Candida albicans Filamentation.

ABSTRACTCandida albicans remains the main etiologic agent of candidiasis, the most common fungal infection and now the third most frequent infection in U.S. hospitals. The scarcity of antifungal agents and their limited efficacy contribute to the unacceptably high morbidity and mortality rates associated with these infections. The yeast-to-hypha transition represents the main virulence factor associated with the pathogenesis of C. albicans infections. In addition, filamentation is pivotal for robust biofilm development, which represents another major virulence factor for candidiasis and further complicates treatment. Targeting pathogenic mechanisms rather than growth represents an attractive yet clinically unexploited approach in the development of novel antifungal agents. Here, we performed large-scale phenotypic screening assays with 30,000 drug-like small-molecule compounds within ChemBridge’s DIVERSet chemical library in order to identify small-molecule inhibitors of C. albicans filamentation, and our efforts led to the identification of a novel series of bioactive compounds with a common biaryl amide core structure. The leading compound of this series, N-[3-(allyloxy)-phenyl]-4-methoxybenzamide, was able to prevent filamentation under all liquid and solid medium conditions tested, suggesting that it impacts a common core component of the cellular machinery that mediates hypha formation under different environmental conditions. In addition to filamentation, this compound also inhibited C. albicans biofilm formation. This leading compound also demonstrated in vivo activity in clinically relevant murine models of invasive and oral candidiasis. Overall, our results indicate that compounds within this series represent promising candidates for the development of novel anti-virulence approaches to combat C. albicans infections.

LITTLE FISH, BIG DATA: ZEBRAFISH AS A MODEL FOR CARDIOVASCULAR AND METABOLIC DISEASE.

The burden of cardiovascular and metabolic diseases worldwide is staggering. The emergence of systems approaches in biology promises new therapies, faster and cheaper diagnostics, and personalized medicine. However, a profound understanding of pathogenic mechanisms at the cellular and molecular levels remains a fundamental requirement for discovery and therapeutics. Animal models of human disease are cornerstones of drug discovery as they allow identification of novel pharmacological targets by linking gene function with pathogenesis. The zebrafish model has been used for decades to study development and pathophysiology. More than ever, the specific strengths of the zebrafish model make it a prime partner in an age of discovery transformed by big-data approaches to genomics and disease. Zebrafish share a largely conserved physiology and anatomy with mammals. They allow a wide range of genetic manipulations, including the latest genome engineering approaches. They can be bred and studied with remarkable speed, enabling a range of large-scale phenotypic screens. Finally, zebrafish demonstrate an impressive regenerative capacity scientists hope to unlock in humans. Here, we provide a comprehensive guide on applications of zebrafish to investigate cardiovascular and metabolic diseases. We delineate advantages and limitations of zebrafish models of human disease and summarize their most significant contributions to understanding disease progression to date.

Phenotypic Screens Identify Parasite Genetic Factors Associated with Malarial Fever Response in Plasmodium falciparum piggyBac Mutants.

Malaria remains one of the most devastating parasitic diseases worldwide, with 90% of the malaria deaths in Africa in 2013 attributable to Plasmodium falciparum. The clinical symptoms of malaria include cycles of fever, corresponding to parasite rupture from red blood cells every 48h. Parasite pathways involved in the parasite's ability to survive the host fever response, and indeed, the functions of ~40% of P.falciparum genes as a whole, are still largely unknown. Here, we evaluated the potential of scalable forward-genetic screening methods to identify genes involved in the host fever response. We performed a phenotypic screen for genes linked to the parasite response to febrile temperatures by utilizing a selection of single-disruption P.falciparum mutants generated via random piggyBac transposon mutagenesis in a previous study. We identified several mutants presenting significant phenotypes in febrile response screens compared to the wild type, indicating possible roles for the disrupted genes in this process. We present these initial studies as proof that forward genetics can be used to gain insight into critical factors associated with parasite biology. IMPORTANCE Though the P.falciparum genome sequence has been available for many years, ~40% of its genes do not have informative annotations, as they show no detectable homology to those of studied organisms. More still have not been evaluated via genetic methods. Scalable forward-genetic approaches that allow interrogation of gene function without any pre-existing knowledge are needed to hasten understanding of parasite biology, which will expedite the identification of drug targets and the development of future interventions in the face of spreading resistance to existing frontline drugs. In this work, we describe a new approach to pursue forward-genetic phenotypic screens for P.falciparum to identify factors associated with virulence. Future large-scale phenotypic screens developed to probe other such interesting phenomena, when considered in parallel, will prove a powerful tool for functional annotation of the P.falciparum genome, where so much remains undiscovered.

Open Source Drug Discovery: Highly Potent Antimalarial Compounds Derived from the Tres Cantos Arylpyrroles.

The development of new antimalarial compounds remains a pivotal part of the strategy for malaria elimination. Recent large-scale phenotypic screens have provided a wealth of potential starting points for hit-to-lead campaigns. One such public set is explored, employing an open source research mechanism in which all data and ideas were shared in real time, anyone was able to participate, and patents were not sought. One chemical subseries was found to exhibit oral activity but contained a labile ester that could not be replaced without loss of activity, and the original hit exhibited remarkable sensitivity to minor structural change. A second subseries displayed high potency, including activity within gametocyte and liver stage assays, but at the cost of low solubility. As an open source research project, unexplored avenues are clearly identified and may be explored further by the community; new findings may be cumulatively added to the present work.