Integrated High-Throughput Screening and Large-Scale Isobolographic Analysis to Accelerate the Discovery of Radiosensitizers With Greater Selectivity for Cancer Cells.

PURPOSE: High-throughput screening (HTS) platforms have been widely used to identify candidate anticancer drugs and drug-drug combinations; however, HTS-based identification of new drug-ionizing radiation (IR) combinations has rarely been reported. Herein, we developed an integrated approach including cell-based HTS and computational large-scale isobolographic analysis to accelerate the identification of radiosensitizing compounds acting strongly and more specifically on cancer cells. METHODS AND MATERIALS: In a 384-well plate format, 160 compounds likely to interfere with the cell response to radiation were screened on human glioblastoma (U251-MG) and cervix carcinoma (ME-180) cell lines, as well as on normal fibroblasts (CCD-19Lu). After drug exposure, cells were irradiated or not and short-term cell survival was assessed by high-throughput cell microscopy. Computational large-scale dose-response and isobolographic approach were used to identify promising synergistic drugs radiosensitizing cancer cells rather than normal cells. Synergy of a promising compound was confirmed on ME-180 cells by an independent 96-well assay protocol, and finally, by the gold-standard colony forming assay. RESULTS: We retained 4 compounds synergistic at 2 isoeffects in U251-MG and ME-180 cell lines and 11 compounds synergistically effective in only one cancer cell line. Among these 15 promising radiosensitizers, 5 compounds showed limited toxicity combined or not with IR on normal fibroblasts. CONCLUSIONS: Overall, this study demonstrated that HTS chemoradiation screening together with large-scale computational analysis is an efficient tool to identify synergistic drug-IR combinations, with concomitant assessment of unwanted toxicity on normal fibroblasts. It sparks expectations to accelerate the discovery of highly desired agents improving the therapeutic index of radiation therapy.

Transfer learning for versatile and training free high content screening analyses.

High content screening (HCS) is a technology that automates cell biology experiments at large scale. A High Content Screen produces a high amount of microscopy images of cells under many conditions and requires that a dedicated image and data analysis workflow be designed for each assay to select hits. This heavy data analytic step remains challenging and has been recognized as one of the burdens hindering the adoption of HCS. In this work we propose a solution to hit selection by using transfer learning without additional training. A pretrained residual network is employed to encode each image of a screen into a discriminant representation. The deep features obtained are then corrected to account for well plate bias and misalignment. We then propose two training-free pipelines dedicated to the two main categories of HCS for compound selection: with or without positive control. When a positive control is available, it is used alongside the negative control to compute a linear discriminant axis, thus building a classifier without training. Once all samples are projected onto this axis, the conditions that best reproduce the positive control can be selected. When no positive control is available, the Mahalanobis distance is computed from each sample to the negative control distribution. The latter provides a metric to identify the conditions that alter the negative control's cell phenotype. This metric is subsequently used to categorize hits through a clustering step. Given the lack of available ground truth in HCS, we provide a qualitative comparison of the results obtained using this approach with results obtained with handcrafted image analysis features for compounds and siRNA screens with or without control. Our results suggests that the fully automated and generic pipeline we propose offers a good alternative to handcrafted dedicated image analysis approaches. Furthermore, we demonstrate that this solution select conditions of interest that had not been identified using the primary dedicated analysis. Altogether, this approach provides a fully automated, reproducible, versatile and comprehensive alternative analysis solution for HCS encompassing compound-based or downregulation screens, with or without positive controls, without the need for training or cell detection, or the development of a dedicated image analysis workflow.

Cell painting transfer increases screening hit rate.

Drug discovery uses high throughput screening to identify compounds that interact with a molecular target or that alter a phenotype favorably. The cautious selection of molecules used for such a screening is instrumental and is tightly related to the hit rate. In this work, we wondered if cell painting, a general-purpose image-based assay, could be used as an efficient proxy for compound selection, thus increasing the success rate of a specific assay. To this end, we considered cell painting images with 30,000 molecules treatments, and selected compounds that produced a visual effect close to the positive control of an assay, by using the Frechet Inception Distance. We then compared the hit rates of such a preselection with what was actually obtained in real screening campaigns. As a result, cell painting would have permitted a significant increase in the success rate and, even for one of the assays, would have allowed to reach 80% of the hits with 10 times fewer compounds to test. We conclude that images of a cell painting assay can be directly used for compound selection prior to screening, and we provide a simple quantitative approach in order to do so.

Cytidine analogs are synthetic lethal with base excision repair default due to MBD4 deficiency.

Inactivating mutations of MBD4 have been reported in subsets of various tumors. A deficiency of this DNA glycosylase, recognizing specifically T:G mismatch resulting from the deamination of methyl-cytosine, results in a hypermutated phenotype due to the accumulation of CpG>TpG transitions. Here, we hypothesize that the difference in DNA metabolism consecutive to MBD4 deficiency may result in specific cytotoxicities in MBD4-deficient tumor cells in a synthetic lethality fashion. After a large-scale drug repurposing screen, we show in two isogenic MBD4 knock-out cell models that the inactivation of MBD4 sensitizes cancer cells to cytidine analogs. We further confirm the exquisite activity of gemcitabine in an MBD4-deficient co-clinical model as (i) it completely prevented the development of an MBD4-deficient uveal melanoma patient-derived xenograft and (ii) treatment in the corresponding patient resulted in an exceptional tumor response. These data suggest that patients harboring MBD4-deficient tumors may be treated efficiently by cytidine analogs.

Homosalate boosts the release of tumour-derived extracellular vesicles with protection against anchorage-loss property.

Eukaryotic cells, including cancer cells, secrete highly heterogeneous populations of extracellular vesicles (EVs). EVs could have different subcellular origin, composition and functional properties, but tools to distinguish between EV subtypes are scarce. Here, we tagged CD63- or CD9-positive EVs secreted by triple negative breast cancer cells with Nanoluciferase enzyme, to set-up a miniaturized method to quantify secretion of these two EV subtypes directly in the supernatant of cells. We performed a cell-based high-content screening to identify clinically-approved drugs able to affect EV secretion. One of the identified hits is Homosalate, an anti-inflammatory drug found in sunscreens which robustly increased EVs' release. Comparing EVs induced by Homosalate with those induced by Bafilomycin A1, we demonstrate that: (1) the two drugs act on EVs generated in distinct subcellular compartments, and (2) EVs released by Homosalate-, but not by Bafilomycin A1-treated cells enhance resistance to anchorage loss in another recipient epithelial tumour cell line. In conclusion, we identified a new drug modifying EV release and demonstrated that under influence of different drugs, triple negative breast cancer cells release EV subpopulations from different subcellular origins harbouring distinct functional properties.

Upregulation of the Mevalonate Pathway through EWSR1-FLI1/EGR2 Regulatory Axis Confers Ewing Cells Exquisite Sensitivity to Statins.

Ewing sarcoma (EwS) is an aggressive primary bone cancer in children and young adults characterized by oncogenic fusions between genes encoding FET-RNA-binding proteins and ETS transcription factors, the most frequent fusion being EWSR1-FLI1. We show that EGR2, an Ewing-susceptibility gene and an essential direct target of EWSR1-FLI1, directly regulates the transcription of genes encoding key enzymes of the mevalonate (MVA) pathway. Consequently, Ewing sarcoma is one of the tumors that expresses the highest levels of mevalonate pathway genes. Moreover, genome-wide screens indicate that MVA pathway genes constitute major dependencies of Ewing cells. Accordingly, the statin inhibitors of HMG-CoA-reductase, a rate-limiting enzyme of the MVA pathway, demonstrate cytotoxicity in EwS. Statins induce increased ROS and lipid peroxidation levels, as well as decreased membrane localization of prenylated proteins, such as small GTP proteins. These metabolic effects lead to an alteration in the dynamics of S-phase progression and to apoptosis. Statin-induced effects can be rescued by downstream products of the MVA pathway. Finally, we further show that statins impair tumor growth in different Ewing PDX models. Altogether, the data show that statins, which are off-patent, well-tolerated, and inexpensive compounds, should be strongly considered in the therapeutic arsenal against this deadly childhood disease.