FC014: Target Deconvolution from Phenotypic Screens—Identification of Ataxia Telangiectasia Mutated as a Potential Therapeutic Target for Podocyte Protection

Abstract

Abstract BACKGROUND AND AIMS As part of the renal filtration barrier, podocytes are critical to the function of kidney glomeruli. Their loss is one of the hallmarks of chronic kidney disease, including diabetic kidney disease (DKD). Indeed, many drug discovery efforts focus on preventing podocyte apoptosis triggered by diabetic stressors to delay kidney disease progression. In order to discover and characterize new therapeutic agents, phenotypic screens are a powerful tool to directly screen for the desired cellular outcome. These need to be complemented by target-centric assays to support the chemical optimization of compounds. Generally, the deconvolution and confirmation of molecular targets are considered challenging, but necessary in the follow-up of phenotype-based drug discovery programmes to allow progression towards the clinic. We have combined novel technologies like high-throughput next-generation sequencing (NGS) and Cell Painting [1] to develop an unbiased approach for deconvolution of targets from phenotypic screens. As an outcome, we identified a small molecule compound that can protect conditionally immortalized human podocytes (CIHP) from apoptosis in vitro. In a second step, we were able to identify the target responsible for this effect. METHOD Phenotypic screen: CIHP [2] were treated with high glucose and palmitate, a common DKD injury model, as the main driver for apoptosis as evaluated by caspase 3/7 activation [3]. Potentially protective agents were screened against apoptosis from a library of 50 000 small molecules. We selected a hit compound (CpdX) for subsequent target deconvolution to identify the target causing the protective effect. Target deconvolution and compound profiling: We compared the molecular signatures triggered by CpdX treatment to those of a reference panel of known, bio-annotated compounds (#L3500; Selleckchem). Dimensional reduction based on UMAP and PCA applied to transcriptomic data (NGS) as well as Cell Painting fingerprints were used to identify known compounds exhibiting a similar transcriptomic and morphological profile as the CpdX. We hypothesized that the molecular targets of these compounds also constitute likely targets for the new CpdX. RESULTS We developed a screening cascade (Figure 1A) to identify novel targets for kidney disease. A first-pass phenotypic podocyte apoptosis screen delivered a potent compound, able to rescue CIHP from a diabetic challenge in vitro (Figure 1B, C). Integrated analysis and alignment of cellular changes in transcriptomic profile and morphology following CpdX or reference compound treatment enabled us to identify the serine/threonine-protein kinase Ataxia Telangiectasia Mutated (ATM) as one of the putative targets of the orphan compound. Both KU-60019, a known inhibitor of ATM [4] (Figure 1D) and CpdX showed remarkably similar effects in NGS and Cell Painting (Figure 2A, B) and were also equally protecting the cells from apoptosis in vitro (Figure 1B). From this data, we inferred that ATM is a likely effector of CpdX treatment. CONCLUSION The initial lack of specific target candidates following primary phenotypic screens is a widespread problem in drug discovery. In the absence of a target deconvolution strategy, many potential candidates cannot be progressed easily to optimization. Here, we demonstrated that a combination of state-of-the-art technologies used to benchmark hit compounds from phenotypic screens versus known bio-annotated compound libraries represent a viable approach to perform target deconvolution of hits from phenotypic screening approaches. In this work, a phenotypic screen and subsequent deconvolution identified a novel target in combination with a novel compound, making a suitable starting point for progression into a potential kidney therapeutic agent.