Leveraging Cell Painting Images to Expand the Applicability Domain and Actively Improve Deep Learning Quantitative Structure-Activity Relationship Models.

The search for chemical hit material is a lengthy and increasingly expensive drug discovery process. To improve it, ligand-based quantitative structure-activity relationship models have been broadly applied to optimize primary and secondary compound properties. Although these models can be deployed as early as the stage of molecule design, they have a limited applicability domain─if the structures of interest differ substantially from the chemical space on which the model was trained, a reliable prediction will not be possible. Image-informed ligand-based models partly solve this shortcoming by focusing on the phenotype of a cell caused by small molecules, rather than on their structure. While this enables chemical diversity expansion, it limits the application to compounds physically available and imaged. Here, we employ an active learning approach to capitalize on both of these methods' strengths and boost the model performance of a mitochondrial toxicity assay (Glu/Gal). Specifically, we used a phenotypic Cell Painting screen to build a chemistry-independent model and adopted the results as the main factor in selecting compounds for experimental testing. With the additional Glu/Gal annotation for selected compounds we were able to dramatically improve the chemistry-informed ligand-based model with respect to the increased recognition of compounds from a 10% broader chemical space.

Automated quality control tool for high-content imaging data by building 2D prediction intervals on reference biosignatures.

Recent advances in automated microscopy and image analysis enables quantitative profiling of cellular phenotypes (Cell Painting). It paves the way for studying the broad effects of chemical perturbations on biological systems at large scale during lead optimization. Comparison of perturbation biosignatures with biosignatures of annotated compounds can inform on both on- and off-target effects. When building databases with phenotypic profiles of thousands of compounds, it is vital to control the quality of Cell Painting assays over time. A tool for this to our knowledge does not yet exist within the imaging community. In this paper, we introduce an automated tool to assess the quality of Cell Painting assays by quantifying the reproducibility of biosignatures of annotated reference compounds. The tool learns the biosignature of those treatments from a historical dataset, and subsequently, it builds a two-dimensional probabilistic quality control (QC) limit. The limit will then be used to detect aberrations in new Cell Painting experiments. The tool is illustrated using simulated data and further demonstrated on Cell Painting data of the A549 cell line. In general, the tool provides a sensitive, detailed and easy-to-interpret mechanism to validate the quality of Cell Painting assays.

Development of a cellular high-content, immunofluorescent HBV core assay to identify novel capsid assembly modulators that induce the formation of aberrant HBV core structures.

Hepatitis B Virus (HBV) core protein has multiple functions in the viral life cycle and is an attractive target for new anti-viral therapies. Capsid assembly modulators (CAMs) target the core protein and induce the formation of either morphologically normal (CAM-N) or aberrant structures (CAM-A), both devoid of genomic material. To date a diverse family of CAM-N chemotypes has been identified, but in contrast, described CAM-As are based on the heteroaryldihydropyrimidine (HAP) scaffold. We used the HBV-inducible HepG2.117 cell line with immunofluorescent labeling of HBV core to develop and validate a cellular high-content image-based assay where aggregated core structures are identified using image analysis spot texture features. Treatment with HAPs led to a dose- and time-dependent formation of aggregated core appearing as dot-like structures in the cytoplasm and nucleus. By combining a biochemical and cellular screening approach, a compound was identified as a novel non-HAP scaffold able to induce dose-dependent formation of aberrant core structures, which was confirmed by electron microscopy and native gel electrophoresis. This compound displayed anti-HBV activity in HepG2.117 cells, providing proof-of-concept for our screening approach. We believe our combined biochemical and cellular high-content screening method will aid in expanding the range of CAM-A chemotypes.

Discovery and Pharmacological Characterization of JNJ-64619178, a Novel Small-Molecule Inhibitor of PRMT5 with Potent Antitumor Activity.

The protein arginine methyltransferase 5 (PRMT5) methylates a variety of proteins involved in splicing, multiple signal transduction pathways, epigenetic control of gene expression, and mechanisms leading to protein expression required for cellular proliferation. Dysregulation of PRMT5 is associated with clinical features of several cancers, including lymphomas, lung cancer, and breast cancer. Here, we describe the characterization of JNJ-64619178, a novel, selective, and potent PRMT5 inhibitor, currently in clinical trials for patients with advanced solid tumors, non-Hodgkin's lymphoma, and lower-risk myelodysplastic syndrome. JNJ-64619178 demonstrated a prolonged inhibition of PRMT5 and potent antiproliferative activity in subsets of cancer cell lines derived from various histologies, including lung, breast, pancreatic, and hematological malignancies. In primary acute myelogenous leukemia samples, the presence of splicing factor mutations correlated with a higher ex vivo sensitivity to JNJ-64619178. Furthermore, the potent and unique mechanism of inhibition of JNJ-64619178, combined with highly optimized pharmacological properties, led to efficient tumor growth inhibition and regression in several xenograft models in vivo, with once-daily or intermittent oral-dosing schedules. An increase in splicing burden was observed upon JNJ-64619178 treatment. Overall, these observations support the continued clinical evaluation of JNJ-64619178 in patients with aberrant PRMT5 activity-driven tumors.

Tales of 1,008 small molecules: phenomic profiling through live-cell imaging in a panel of reporter cell lines.

Phenomic profiles are high-dimensional sets of readouts that can comprehensively capture the biological impact of chemical and genetic perturbations in cellular assay systems. Phenomic profiling of compound libraries can be used for compound target identification or mechanism of action (MoA) prediction and other applications in drug discovery. To devise an economical set of phenomic profiling assays, we assembled a library of 1,008 approved drugs and well-characterized tool compounds manually annotated to 218 unique MoAs, and we profiled each compound at four concentrations in live-cell, high-content imaging screens against a panel of 15 reporter cell lines, which expressed a diverse set of fluorescent organelle and pathway markers in three distinct cell lineages. For 41 of 83 testable MoAs, phenomic profiles accurately ranked the reference compounds (AUC-ROC ≥ 0.9). MoAs could be better resolved by screening compounds at multiple concentrations than by including replicates at a single concentration. Screening additional cell lineages and fluorescent markers increased the number of distinguishable MoAs but this effect quickly plateaued. There remains a substantial number of MoAs that were hard to distinguish from others under the current study's conditions. We discuss ways to close this gap, which will inform the design of future phenomic profiling efforts.

Binding kinetics of darunavir to human immunodeficiency virus type 1 protease explain the potent antiviral activity and high genetic barrier.

The high incidence of cross-resistance between human immunodeficiency virus type 1 (HIV-1) protease inhibitors (PIs) limits their sequential use. This necessitates the development of PIs with a high genetic barrier and a broad spectrum of activity against PI-resistant HIV, such as tipranavir and darunavir (TMC114). We performed a surface plasmon resonance-based kinetic study to investigate the impact of PI resistance-associated mutations on the protease binding of five PIs used clinically: amprenavir, atazanavir, darunavir, lopinavir, and tipranavir. With wild-type protease, the binding affinity of darunavir was more than 100-fold higher than with the other PIs, due to a very slow dissociation rate. Consequently, the dissociative half-life of darunavir was much higher (>240 h) than that of the other PIs, including darunavir's structural analogue amprenavir. The influence of protease mutations on the binding kinetics was tested with five multidrug-resistant (MDR) proteases derived from clinical isolates harboring 10 to 14 PI resistance-associated mutations with a decreased susceptibility to various PIs. In general, all PIs bound to the MDR proteases with lower binding affinities, caused mainly by a faster dissociation rate. For amprenavir, atazanavir, lopinavir, and tipranavir, the decrease in affinity with MDR proteases resulted in reduced antiviral activity. For darunavir, however, a nearly 1,000-fold decrease in binding affinity did not translate into a weaker antiviral activity; a further decrease in affinity was required for the reduced antiviral effect. These observations provide a mechanistic explanation for darunavir's potent antiviral activity and high genetic barrier to the development of resistance.

Repurposing High-Throughput Image Assays Enables Biological Activity Prediction for Drug Discovery.

In both academia and the pharmaceutical industry, large-scale assays for drug discovery are expensive and often impractical, particularly for the increasingly important physiologically relevant model systems that require primary cells, organoids, whole organisms, or expensive or rare reagents. We hypothesized that data from a single high-throughput imaging assay can be repurposed to predict the biological activity of compounds in other assays, even those targeting alternate pathways or biological processes. Indeed, quantitative information extracted from a three-channel microscopy-based screen for glucocorticoid receptor translocation was able to predict assay-specific biological activity in two ongoing drug discovery projects. In these projects, repurposing increased hit rates by 50- to 250-fold over that of the initial project assays while increasing the chemical structure diversity of the hits. Our results suggest that data from high-content screens are a rich source of information that can be used to predict and replace customized biological assays.

A time-resolved fluorescence assay to identify small-molecule inhibitors of HIV-1 fusion.

Fusion of host cell and human immunodeficiency virus type 1 (HIV-1) membranes is mediated by the 2 "heptad-repeat" regions of the viral gp41 protein. The collapse of the C-terminal heptad-repeat regions into the hydrophobic grooves of a coiled-coil formed by the corresponding homotrimeric N-terminal heptad-repeat regions generates a stable 6-helix bundle. This brings viral and cell membranes together for membrane fusion, facilitating viral entry. The authors developed an assay based on soluble peptides derived from the gp41 N-terminal heptad-repeat region (IQN36) as well as from the C-terminal region (C34). Both peptides were labeled with fluorophores, IQN36 with allophycocyanin (APC) and C34 with the lanthanide europium (Eu3+). Formation of the 6-helix bundle brings both fluorophores in close proximity needed for Förster resonance energy transfer (FRET). Compounds that interfere with binding of C34-Eu with IQN36-APC suppress the FRET signal. The assay was validated with various peptides and small molecules, and quenching issues were addressed. Evaluation of a diversified compound collection in a high-throughput screening campaign enabled identification of small molecules with different chemical scaffolds that inhibit this crucial intermediate in the HIV-1 entry process. This study's observations substantiate the expediency of time-resolved FRET-based assays to identify small-molecule inhibitors of protein-protein interactions.

Ellipsoid Segmentation Model for Analyzing Light-Attenuated 3D Confocal Image Stacks of Fluorescent Multi-Cellular Spheroids.

In oncology, two-dimensional in-vitro culture models are the standard test beds for the discovery and development of cancer treatments, but in the last decades, evidence emerged that such models have low predictive value for clinical efficacy. Therefore they are increasingly complemented by more physiologically relevant 3D models, such as spheroid micro-tumor cultures. If suitable fluorescent labels are applied, confocal 3D image stacks can characterize the structure of such volumetric cultures and, for example, cell proliferation. However, several issues hamper accurate analysis. In particular, signal attenuation within the tissue of the spheroids prevents the acquisition of a complete image for spheroids over 100 micrometers in diameter. And quantitative analysis of large 3D image data sets is challenging, creating a need for methods which can be applied to large-scale experiments and account for impeding factors. We present a robust, computationally inexpensive 2.5D method for the segmentation of spheroid cultures and for counting proliferating cells within them. The spheroids are assumed to be approximately ellipsoid in shape. They are identified from information present in the Maximum Intensity Projection (MIP) and the corresponding height view, also known as Z-buffer. It alerts the user when potential bias-introducing factors cannot be compensated for and includes a compensation for signal attenuation.

A duplex real-time RT-PCR assay for profiling inhibitors of four dengue serotypes.

We have developed a duplex real-time RT-PCR assay for profiling antiviral inhibitors of four dengue virus (DENV) serotypes. In this assay, the primers and the probe for amplifying DENV were designed in the conserved regions of the genome after aligned more than 300 nucleotide sequences of four dengue serotypes deposited in the GeneBank. To discriminate the antiviral activity from the cytotoxicity of compounds, a housekeeping gene of the Vero cells, ß-actin, was used to design the primers and the probe for the second set of PCR as an internal control, which is used to normalize the RNA levels of dengue-specific PCR due to the cellular toxicity of test compounds. For compound profiling, the duplex PCR is performed using LightCycler(®) in a single tube to simultaneously amplify both the dengue target gene and the Vero cell housekeeping gene from the compound-treated Vero cell lysates. This assay was validated against a panel of reference compounds. The results show that the universal primers and probe in this duplex RT-PCR assay can efficiently amplify all four dengue serotypes and that the PCR efficiency for both the dengue target gene and the Vero cells ß-actin gene is 100%.

A homogeneous time-resolved fluorescence assay to identify inhibitors of HIV-1 fusion.

The human immunodeficiency virus type 1 (HIV-1) initiates infection through sequential interactions with CD4 and chemokine coreceptors unmasking the gp41 subunit of the viral envelope protein. Consequently, the N-terminal heptad repeats of gp41 form a trimeric coiled-coil groove in which the C-terminal heptad repeats collapse, generating a stable six-helix bundle. This brings the viral and cell membrane in close proximity enabling fusion and the release of viral genome in the cytosol of the host cell. In this chapter, we describe a homogeneous time-resolved fluorescence assay to identify inhibitors of HIV-1 fusion, based on the ability of soluble peptides, derived from the N- and C-terminal domains of gp41, to form a stable six-helix bundle in vitro. Labeling of the peptides with allophycocyanin and the lanthanide europium results in a Föster resonance energy transfer (FRET) signal upon formation of the six-helix bundle. Compounds interfering with the six-helix bundle formation inhibit the HIV-1 fusion process and suppress the FRET signal.

Translation of a tumor microenvironment mimicking 3D tumor growth co-culture assay platform to high-content screening.

For drug discovery, cell-based assays are becoming increasingly complex to mimic more realistically the nature of biological processes and their diversifications in diseases. Multicellular co-cultures embedded in a three-dimensional (3D) matrix have been explored in oncology to more closely approximate the physiology of the human tumor microenvironment. High-content analysis is the ideal technology to characterize these complex biological systems, although running such complex assays at higher throughput is a major endeavor. Here, we report on adapting a 3D tumor co-culture growth assay to automated microscopy, and we compare various imaging platforms (confocal vs. nonconfocal) with correlating automated image analysis solutions to identify optimal conditions and settings for future larger scaled screening campaigns. The optimized protocol has been validated in repeated runs where established anticancer drugs have been evaluated for performance in this innovative assay.

Phaedra, a protocol-driven system for analysis and validation of high-content imaging and flow cytometry.

High-content screening has brought new dimensions to cellular assays by generating rich data sets that characterize cell populations in great detail and detect subtle phenotypes. To derive relevant, reliable conclusions from these complex data, it is crucial to have informatics tools supporting quality control, data reduction, and data mining. These tools must reconcile the complexity of advanced analysis methods with the user-friendliness demanded by the user community. After review of existing applications, we realized the possibility of adding innovative new analysis options. Phaedra was developed to support workflows for drug screening and target discovery, interact with several laboratory information management systems, and process data generated by a range of techniques including high-content imaging, multicolor flow cytometry, and traditional high-throughput screening assays. The application is modular and flexible, with an interface that can be tuned to specific user roles. It offers user-friendly data visualization and reduction tools for HCS but also integrates Matlab for custom image analysis and the Konstanz Information Miner (KNIME) framework for data mining. Phaedra features efficient JPEG2000 compression and full drill-down functionality from dose-response curves down to individual cells, with exclusion and annotation options, cell classification, statistical quality controls, and reporting.