March 11, 2019
Imagine how much more researchers could learn by examining the data of an underperforming deep learning method, or employing a tailored workflow to characterize endothelial cells or adopting a novel method that allows discovery of new, previously unknown phenotypes based only on negative controls. Screening technology and methods that ordinarily get buried deep in other published research surge to the surface in a new SLAS Discovery special issue.
Marc Bickle, Ph.D., guest editor of the SLAS Discovery Special Issue on Academic Screening Centers and Translational Drug Discovery Centers Showcase, wants to kick start partnership ideas for academia and industry. “We wanted to show the possibilities that exist,” says Bickle. “I don’t want pharmaceutical companies to invest in building yet another screening facility – I want them to seek possibilities and capacity in academia and exploit those resources.”
He is familiar with how potential gets overlooked. “In my own line of work, I hear people – even around our campus – mention that they have been looking for the type of work we do for years and only found out about us recently,” says Bickle, who is head of the Technology Development Studio (TDS) of the Max Planck Institute of Molecular Cell Biology and Genetics (MPI-CBG; Dresden, Germany). TDS, one of the institutions the special issue highlights, is an open-access screening facility that performs high-content screening (HCS) of RNAi and chemical libraries similar to the work conducted by the other featured organizations. “We need to raise the awareness of these services. There is a lot of interesting research going on and it’s all there for the taking.”
The lack of awareness stems from a simple truth: “Screens are difficult to publish,” says Bickle. “If your screen doesn’t have a specific biological story behind it, or if you don’t do some interesting medicinal chemistry on it, the story will not be published. Consequently, much work is done after the screen to have something that is publishable and the portion of the paper representing the actual screen gets smaller and smaller until it’s a footnote in the materials and methods. This special issue of SLAS Discovery allows screening centers to have visibility and a platform to shine instead of being buried deep down in the middle of the authors or not there at all.”
Bickle comments that the collection in the special issue is not exhaustive. "It’s just a glimpse of innovative life sciences research going on in the academic and not-for-profit drug screening world,” says Bickle, who profiles each of the featured institutions in his editorial for the special issue. “It illustrates an abundant landscape that offers a great diversity and a range of expertise.”
Relying solely on academic screening centers and translational drug discovery labs for submissions could have led to “a lot of repetition, but the spectrum here is broad, from neglected diseases technology to data mining to a label-free kinetic profiling method,” says Bickle, who serves as an associate editor for SLAS Discovery, and holds positions with the SLAS Europe Scientific Program Committee and the SLAS Knowledge Content and Delivery Council. “If you look beyond what we gathered for the special issue, you will find more labs doing even more exotic research.”
Of the 15 articles in the SLAS Discovery special issue, six are review articles and nine are original research articles, three of which describe analytical methods for phenotypic profiling of cellular responses. An example of this profiling, which includes some surprising results, is “Evaluation of Machine Learning Classifiers to Predict Compound Mechanism of Action When Transferred across Distinct Cell Lines,” from the Edinburgh Cancer Discovery Unit (ECDU), an academic research group located within the Cancer Research UK Edinburgh Centre, MRC Institute of Genetics and Molecular Medicine, University of Edinburgh (Scotland, United Kingdom (UK)).
“Deep learning networks seem to be beating every previous record of machine learning performance – until you read this paper,” says Bickle. “They found that the deep learning model they used was not better, and slightly worse than the random forest model.”
The article explores application of a convolutional neural networks (CNNs) classifier to deliver equivalent accuracy compared with an ensemble-based tree classifier at compound mechanism of action prediction within cell lines. However, the team's CNN analysis performs worse than an ensemble-based tree classifier when trained on multiple cell lines at predicting compound mechanism of action on an unseen cell line.
“The results are a bit of a damper on deep learning. However, it gets a bit boring when deep learning is always winning everything,” says Bickle. “It will be interesting to understand why deep learning did not perform as well as the random forest. The data is publicly available so that machine learning experts can access and examine it.”
A further exploration of analytical methods for phenotypic profiling of cellular responses unfolds in “Unbiased Phenotype Detection Using Negative Controls," from Bickle’s TDS at MPI-CBG (Dresden, Germany), in which the team presents a novel method to discover previously unknown phenotypes based only on negative controls. The method is compared with L1-norm regularization, a standard method using positive controls to obtain a sparse matrix. The analytical pipeline is implemented in the open-source software KNIME, allowing the implementation of the method in many laboratories, even ones without advanced computing knowledge.
Another article focused on analytical methods is "xCELLanalyzer: A Framework for the Analysis of Cellular Impedance Measurements for Mode of Action Discovery" from the Helmholtz Centre for Infection Research (HZI; Braunschweig, Germany), which describes a label-free kinetic profiling method based on an impedance readout to monitor the time-dependent cellular response profiles for the interaction of bioactive natural products and other small molecules with mammalian cells.
Six of the original research articles in the special issue describe novel model systems, assays and technologies that allow the screening of large collections of molecules, including “High-Density Cell Arrays for Genome-Scale Phenotypic Screening” from CellNetworks Advanced Biological Screening Core Facility in the BioQuant Center for Quantitative Analysis of Molecular and Cellular Biosystems at Heidelberg University (Heidelberg, Germany). The article highlights a high-density cell array (HD-CA) platform, which miniaturizes cell-based screening in the form of preprinted and ready-to-run screening arrays. With the HD-CA technology, up to 24,576 samples can be tested in a single experiment, thereby reducing costs and saving time for microscopy-based screening by 75 percent, which Bickle says is big news from the BioQuant team.
“The team works directly on high-density spotted arrays for virus applications,” explains Bickle. “That’s interesting because it demonstrates how academia can be very novel. At the same time, these arrays allow a certain democratization of the screening process because you don’t need big liquid handling robots or big automated microscopes. The screens use normal microscopy slides allowing the use of standard laboratory equipment. The applications to CRISPR, antibody uptake and so on make it an interesting technology.” Such research, he says, shows academia as a rich source for novel assays as basic research is transformed into screens, leading to unexplored therapeutic avenues.
An example of academia’s rich resources is found in the article, "Patient-Derived Phenotypic High-Throughput Assay to Identify Small Molecules Restoring Lysosomal Function in Tay–Sachs Disease." Here, the Moulder Center for Drug Discovery Research, located at the Temple University School of Pharmacy (Philadelphia, PA, USA), explores small molecules identified from high-throughput screening (HTS) that could provide leads suitable for chemical optimization to target the central nervous system in patients with Tay-Sachs disease, an inherited lysosomal storage condition resulting from mutations in a critical lysosomal enzyme. The team showcases a new high-throughput phenotypic assay utilizing infantile Tay–Sachs patient cells based on disrupted lysosomal calcium signaling as a monitor of diseased phenotype. The assay enables the screening of larger chemical compound collections that could lead to identification of new molecular targets previously unknown to impact the disease and accelerate the discovery of new treatments.
Similarly, a research team from the Drug Discovery Initiative (DDI) at the University of Tokyo (Tokyo, Japan) reports an innovative, simple assay protocol that requires only the mixing of test solutions with ADP detection solution and reading the fluorescence intensity of resorufin produced by the coupling reaction. Using an in-house collection that includes 63,000 samples deposited by industry since 2006, the team demonstrates an assay platform that is not only simple and cost-effective, but also sufficiently robust, showing good reproducibility and giving similar results to those obtained with the widely used ADP-Glo bioluminescent assay in their article, "Inexpensive High-Throughput Screening of Kinase Inhibitors Using One-Step Enzyme-Coupled Fluorescence Assay for ADP Detection."
In the article, "A Phenotypic Screening Assay Identifies Modulators of Diamond Blackfan Anemia," the Chemical Biology and Therapeutics group (CBT), located at Lund University (Lund, Sweden), and the Chemical Biology Consortium (CBCS), at the Karolinska Institutet (Stockholm, Sweden), establish a robust scalable assay for screening molecules that rescue erythropoiesis in Diamond-Blackfan anemia (DBA), a bone marrow failure syndrome caused by mutations in ribosomal protein genes. The team describes a screening assay developed using c-kit fetal liver erythroid progenitors from a doxycycline-inducible DBA mouse model.
Research from the Stem Cell Hotel, an innovative, collaborative phenotyping unit located within the Centre for Stem Cells and Regenerative Medicine (CSCRM), King's College (London, UK), reveals a tailored workflow to characterize endothelial cells (ECs) in "Integrated Multiparametric High-Content Profiling of Endothelial Cells." The study presents a parallel quantitative high-content multiparametric profiling of EC models, highlights a simple strategy to benchmark ECs in different conditions and develops new approaches for biological research and translational applications for regenerative medicine.
The article, "Implementation of the NCI-60 Human Tumor Cell Line Panel to Screen 2260 Cancer Drug Combinations to Generate >3 Million Data Points Used to Populate a Large Matrix of Anti-Neoplastic Agent Combinations (ALMANAC) Database," from the Pharmacy Chemical Biology Center (PCBC) in the University of Pittsburgh School of Pharmacy (Pittsburgh, PA, USA), predicts the efficacy of cancer drug combinations through systematic HTS of 100 approved drugs in the National Cancer Institute’s (NCI) panel of cancer cell lines. The team shows how miniaturizing growth inhibition assays into 384-well format, increasing the fetal bovine serum amount to 10 percent, lengthening compound exposure to 72 hours and using a homogeneous detection reagent determines the growth inhibition 50-percent values of individual drugs across 60 cell lines.
The special issue also offers six review articles that present overviews of innovative life sciences research going on in academic screening and translational drug discovery centers. “EU-OPENSCREEN: A Novel Collaborative Approach to Facilitate Chemical Biology” reveals recent projects from the 20 partner institutions of the European research infrastructure, known as EU-OPENSCREEN, which is poised to celebrate its first anniversary in April 2019. Highlights in this article show the benefits of small-molecule screening, the abundance of assay designs and the connection between screening and medicinal chemistry within EU-OPENSCREEN.
“The EU developed a big network to give screening access to many academic laboratories. Funders to EU-OPENSCREEN drive the goals to maximize research money, centralize research to ensure quality and avoid too much duplication,” says Bickle, adding that an important aspect of the partnership is the requirement for data to become publicly available after a certain amount of time. “This step further increases the value of that funding because more people can work on the data, reanalyze and draw conclusions.”
Another review article that captures a vast collaborative effort, according to Bickle, is "Accelerating Drug Discovery Efforts for Trypanosomatidic Infections Using an Integrated Transnational Academic Drug Discovery Platform," from the Fraunhofer-Institute IME (Hamburg, Germany). “The whole preclinical spectrum is represented in the IME’s work, and it demonstrates that academia can power such projects,” remarks Bickle. The article provides a look at the institute’s efforts to address neglected tropical disease treatment by using synthetic and natural product libraries, screening and a preclinical absorption, distribution, metabolism and excretion–toxicity (ADME-Tox) profiling platform to identify compounds that can enter the trypanosomatidic drug discovery value chain.
Other review articles in the collection include overviews from the Amsterdam Institute for Molecules, Medicines and Systems (AIMMS), a division of Bio-Analytical Chemistry at the Vrije Universiteit Amsterdam (Amsterdam, the Netherlands); the Centre for Integrative Biology High-Throughput Screening and Validation (CIBIO-HTS), a facility within the University of Trento, (Trento, Italy); LifeArc, formerly known as MRC Technology, (Stevenage, UK); and the Scripps Research Molecular Screening Center (SRMSC; Jupiter, FL).
AIMMS reviews drug discovery from natural products and in explaining this process focuses on ion-channel drug discovery in its article, "Drug Discovery on Natural Products: From Ion Channels to nAChRs, from Nature to Libraries, from Analytics to Assays." In particular, the article discusses identification of bioactives from natural products targeting nicotinic acetylcholine receptors (nAChRs) and serotonin type 3 receptors (5-HT3Rs).
CIBIO-HTS reviews the strategies applied for targeting RNA-binding proteins (RBPs) with emphasis on the technological advancements to study protein–RNA interactions and on the requirements of appropriate validation strategies to parallel HTS efforts in "Screening Approaches for Targeting Ribonucleoprotein Complexes: A New Dimension for Drug Discovery."
LifeArc describes the selection criteria used to compile the LifeArc diversity set from commercial suppliers and the process the team uses to generate its representative LifeArc index set in the article, "Design of the LifeArc Index Set and Retrospective Review of Its Performance: A Collection for Sharing.”
"The Scripps Molecular Screening Center and Translational Research Institute," focuses on work from SRMSC, founded in 2004 as part of the Translational Research Institute (TRI). SRMSC and TRI are recognized for discovering multiple leads, including Ozanimod, a selective sphingosine 1-phosphate (S1PR1) and 5 (S1PR5) receptor modulator now in advanced Phase 3 testing as a potential treatment for relapsing-remitting multiple sclerosis.
Bickle comments that the labs featured in the special issue all have operational models, allowing them to be accessed from the outside. "This is important for biotech companies,” he notes. “Screening centers offer access to expensive technologies that may give biotechs an edge in discovery but might not be practical for a start-up to own. It’s important to have those core facilities in academia to help develop others’ technology.”
Bickle encourages readers of the SLAS Discovery Special Issue on Academic Screening Centers and Translational Drug Discovery Centers Showcase to join the SLAS Americas Academic Drug Discovery and/or the SLAS European Academic Drug Discovery Special Interest Groups (SIGs) online to learn more about this exciting community. Members of these special interest groups also host meetings during the annual SLAS International Conference and Exhibition and the annual SLAS Europe Conference to discuss topics of interest and network with other professionals in this field.
From SLAS Discovery: In Vitro Tissue Microarrays for Quick and Efficient Spheroid Characterization
The Rise of Pharma R&D Outsourcing
