Homeostatic modulation is pivotal in modern therapeutics. However, the discovery of bioactive materials to achieve this functionality is often random and unpredictive. Here, we enabled a systemic identification and functional classification of chemicals that elicit homeostatic modulation of signaling through Cdc42, a classical small GTPase of Ras superfamily. Rationally designed for high throughput screening, the capture of homeostatic modulators (HMs) along with molecular re-docking uncovered at least five functionally distinct classes of small molecules. This enabling led to partial agonists, hormetic agonists, bona fide activators and inhibitors, and ligand-enhanced agonists. Novel HMs exerted striking functionality in bradykinin-Cdc42 activation of actin remodelingand modified Alzheimer's disease-like behavior in mouse model. This concurrent computer-aided and experimentally empowered HM profiling highlights a model path for predicting HM landscape.
One sentence summary: With concurrent experimental biochemical profiling and in silico computer-aided drug discovery (CADD) analysis, this study enabled a systemic identification and holistic classification of Cdc42 homeostatic modulators (HMs) and demonstrated the power of CADD to predict HM classes that can mimic the pharmacological functionality of interests.
Introduction: Maintainingbody homeostasisis the ultimate keyto health. Thereare rich resources of bioactive materials for this functionality from both natural and synthetic chemical repertories including partial agonists (PAs) and various allosteric modulators. These homeostatic modulators (HMs) play a unique role in modern therapeutics for human diseases such as mental disorders and drug addiction. Buspirone, for example, acts as a PA for serotonin 5-HT 1A receptor but is an antagonist of the dopamine D 2 receptor. Such medical useto treat general anxietydisorders (GADs) has become one of the most-commonly prescribed medications. However, most HMs in current uses target membrane proteins and are often derived from random discoveries. HMs as therapeutics targeting cytoplasmic proteins are even more rare despite that they are in paramount needs (e. g. targeting Ras superfamily small GTPases).
Rationale: Cdc42, a classical member of small GTPases of Ras superfamily, regulates PI3K-AKT and Raf-MEK-ERK pathways and has been implicated in various neuropsychiatric and mental disorders as well as addictive diseases and cancer. We previously reported the high-throughput in-silico screening followed by biological characterization of novel small molecule modulators (SMMs) of Cdc42-intersectin (ITSN) protein-protein interactions (PPIs). Based on a serendipitously discovered SMM ZCL278 with PA profile as a model compound, we hypothesized that there are more varieties of such HMs of Cdc42 signaling, and the model HMs can be defined by their distinct Cdc42-ITSN binding mechanisms using computer-aided drug discovery (CADD) analysis. We further reasoned that molecular modeling coupled with experimental profiling can predict HM spectrum and thus open the door for the holistic identification and classification of multifunctional cytoplasmic target-dependent HMs as therapeutics.
Results: The originally discovered Cdc42 inhibitor ZCL278 displaying PA properties prompted the inquiry whether this finding represented a random encounter of PAs or whether biologically significant PAs can be widely present. The top ranked compounds were initially defined by structural fitness and binding scores to Cdc42. Because higher binding scores do not necessarily translate to higher functionality, we performed exhaustive experimentations with over 2,500 independent Cdc42-GEF (guanine nucleotide exchange factor) assays to profile the GTP loading activities on all 44 top ranked compounds derived from the SMM library. The N-MAR-GTP fluorophore-based Cdc42-GEF assay platform provided the first glimpse of the breadth of HMs. A spectrum of Cdc42 HMs was uncovered that can be categorized into five functionally distinct classes: Class I-partial competitive agonists, Class II-hormetic agonists, Class III- bona fide inhibitors (or inverse agonists), Class IV- bona fide activators or agonists, and Class V-ligand-enhanced agonists. Remarkably, model HMs such as ZCL278, ZCL279, and ZCL367 elicited striking biological functionality in bradykinin-Cdc42 activation of actin remodeling and modified Alzheimer's disease (AD)-like behavior in mouse model. Concurrently, we applied Schrödinger-enabled analyses to perform CADD predicted classification of Cdc42 HMs. We modified the classic molecular docking to instill a preferential binding pocket order (PBPO) of Cdc42-ITSN, which was based on the five binding pockets in interface of Cdc42-ITSN. We additionally applied a structure-based pharmacophore hypothesis generation for the model compounds. Then, using Schrödinger's Phase Shape, 3D ligand alignments assigned HMs to Class I, II, III, IV, and V compounds. In this HM library compounds, PBPO, matching pharmacophoric featuring, and shape alignment, all put ZCL993 in Class II compound category, which was confirmed in the Cdc42-GEF assay.
Conclusion: HMs can target diseased cells or tissues while minimizing impacts on tissues that are unaffected. Using Cdc42 HM model compounds as a steppingstone, GTPase activation-based screening of SMM library uncovered five functionally distinct Cdc42 HM classes among which novel efficacies towards alleviating dysregulated AD-like features in mice were identified. Furthermore, molecular re-docking of HM model compounds led to the concept of PBPO. The CADD analysis with PBPO revealed similar profile in a color-coded spectrum to these five distinct classes of Cdc42 HMs identified by biochemical functionality-based screening. The current study enabled a systemic identification and holistic classification of Cdc42 HMs and demonstrated the power of CADD to predict an HM category that can mimic the pharmacological functionality of interests. With artificial intelligence/machine learning (AI/ML) on the horizon to mirror experimental pharmacological discovery like AlphaFold for protein structure prediction, our study highlights a model path to actively capture and profile HMs in potentially any PPI landscape.
Identification and functional classification of cdc42 homeostatic modulators hms: Using Cdc42 HM model compounds as reference, GTPase activation-based screening of compound libraries uncovered five functionally distinct Cdc42 HM classes. HMs showed novel efficacies towards alleviating dysregulated Alzheimer's disease (AD)-like behavioral and molecular deficits. In parallel, molecular re-docking of HM model compounds established their preferential binding pocket orders (PBPO). PBPO-based profiling (Red reflects the most, whereas green reflects the least, preferable binding pocket) revealed trends of similar pattern to the five classes from the functionality-based classification.