Drug Screening Research Articles

Unveiling the functional roles of patient-derived tumour organoids in assessing the tumour microenvironment and immunotherapy.

Recent studies have established the pivotal roles of patient-derived tumour organoids (PDTOs), innovative three-dimensional (3D) culture systems, in various biological and medical applications. PDTOs, as promising tools, have been established and extensively used for drug screening, prediction of immune response and assessment of immunotherapeutic effectiveness in various cancer types, including glioma, ovarian cancer and so on. The overarching goal is to facilitate the translation of new therapeutic modalities to guide personalised immunotherapy. Notably, there has been a recent surge of interest in the co-culture of PDTOs with immune cells to investigate the dynamic interactions between tumour cells and immune microenvironment. A comprehensive and in-depth investigation is necessary to enhance our understanding of PDTOs as promising testing platforms for cancer immunotherapy. This review mainly focuses on the latest updates on the applications and challenges of PDTO-based methods in anti-cancer immune responses. We strive to provide a comprehensive understanding of the potential and prospects of PDTO-based technologies as next-generation strategies for advancing immunotherapy approaches.

Patient-derived organoids and mini-PDX for predicting METN375S-mutated lung cancer patient clinical therapeutic response

Lung cancer as a molecularly and histologically high heterogonous disease, there is an urgent need to predict lung cancer patients' responses to anti-cancer treatment, and patient-derived organoids (PDOs) have been recognized as a valuable platform for preclinical drug screening. In this study, we successfully established 26 PDO lines from various subtypes of lung cancers including benign tumor, adenocarcinoma, squamous cell carcinoma, adenosquamous carcinoma, large-cell carcinoma, and small-cell carcinoma. These PDOs were shown to retain the major genomic and histological characteristics of primary tumors and remain stable during long-term culture. With the help of targeted genomic sequencing, we found that lung cancer that harbors METN375S mutation is selectively sensitive to afatinib, and a combination of afatinib and gemcitabine induced synthetic lethality in PDO and mini-PDX models. In summary, our findings demonstrate the potential of PDO in predicting lung cancer drug response, and reveal a promising strategy for METN375S mutant lung cancer treatment.

Recent advances and applications of artificial intelligence in 3D bioprinting.

3D bioprinting techniques enable the precise deposition of living cells, biomaterials, and biomolecules, emerging as a promising approach for engineering functional tissues and organs. Meanwhile, recent advances in 3D bioprinting enable researchers to build in vitro models with finely controlled and complex micro-architecture for drug screening and disease modeling. Recently, artificial intelligence (AI) has been applied to different stages of 3D bioprinting, including medical image reconstruction, bioink selection, and printing process, with both classical AI and machine learning approaches. The ability of AI to handle complex datasets, make complex computations, learn from past experiences, and optimize processes dynamically makes it an invaluable tool in advancing 3D bioprinting. The review highlights the current integration of AI in 3D bioprinting and discusses future approaches to harness the synergistic capabilities of 3D bioprinting and AI for developing personalized tissues and organs.

A human organoid drug screen identifies α2-adrenergic receptor signaling as a therapeutic target for cartilage regeneration.

Directed differentiation of stem cells toward chondrogenesis invitro and in situ to regenerate cartilage suffers from off-target differentiation and hypertrophic tendency. Here, we generated a cartilaginous organoid system from human expanded pluripotent stem cells (hEPSCs) carrying a COL2A1mCherry and COL10A1eGFP double reporter, enabling real-time monitoring of chondrogenesis and hypertrophy. After screening 2,040 FDA-approved drugs, we found that α-adrenergic receptor (α-AR) antagonists, especially phentolamine, stimulated chondrogenesis but repressed hypertrophy, while α2-AR agonists reduced chondrogenesis and induced hypertrophy. Phentolamine prevented cartilage degeneration in hEPSC cartilaginous organoid and human cartilage explant models and stimulated microfracture-activated endogenous skeletal stem cells toward hyaline-like cartilage regeneration without fibrotic degeneration in situ. Mechanistically, α2-AR signaling induced hypertrophic degeneration via cyclic guanosine monophosphate (cGMP)-dependent secretory leukocyte protease inhibitor (SLPI) production. SLPI-deleted cartilaginous organoid was degeneration resistant, facilitating large cartilage defect healing. Ultimately, targeting α2-AR/SLPI was a promising and clinically feasible strategy to regenerate cartilage via promoting chondrogenesis and repressing hypertrophy.

Paclitaxel improves thrombopoiesis in the absence of thrombopoietin receptor (Mpl)

BackgroundPlatelets are critical for thrombosis and hemostasis. The THPO-MPL pathway is the primary pathway for generating thrombocytes. Dysregulation of thrombopoiesis results in platelet formation and/or function-related disorders, such as thrombocytopenia. Paclitaxel is an extensively utilized chemotherapeutic agent and its activity may be related to platelets, but the effect of paclitaxel on thrombocytopoiesis warrants comprehensive exploration. ObjectivesWe focused on identifying factors that regulate thrombocyte production and elucidating paclitaxel’s regulatory mechanisms on thrombocytopoiesis, with a particular emphasis on discovering mechanisms that bypass THPO-MPL pathways. MethodsWe performed drug screenings using the Tg(mpl:eGFP) zebrafish model in vivo to identify Food and Drug Administration–approved compounds capable of boosting thrombocyte production. An injury experiment was used to evaluate thrombocyte function. Bromodeoxyuridine assays, terminal deoxynucleotidyl transferase dUTP nick-end labeling, and RNA sequencing analyses were performed to explore cytological and molecular mechanisms. Routine blood testing and flow cytometry were used to analyze mouse phenotypes. ResultsWe found that paclitaxel expands thrombocytes by accelerating the proliferation of thrombocytic lineage cells in zebrafish and elevates platelet levels in mice. This effect occurs by bypassing the thrombopoietin receptor (Mpl). We found that paclitaxel promotes thrombopoiesis, potentially involving the JAK2-ERK1/2 MAPK signaling cascade, a pathway integral to MPL and other regulators. Our results further demonstrate that ERK1/2 is at least partially downstream of JAK2 in paclitaxel-induced thrombopoiesis. ConclusionPaclitaxel could promote thrombopoiesis by bypassing Mpl but presumably via the JAK2-ERK1/2 MAPK pathways. It will aid in understanding the relationship between paclitaxel and platelets clinically, and paclitaxel may have potential value for safeguarding platelets and improving thrombocytosis in related diseases.

Virtual screening of drug materials for pharmaceutical tablet manufacturability with reference to sticking

The manufacturing of pharmaceutical solid dosage forms, such as tablets involves a large number of successive processing operations including crystallisation of the drug substance, granulation, drying, milling, mixing of the formulation, and compaction. Each step is fraught with manufacturing problems. Undesired adhesion of powders to the surface of the compaction tooling, known as sticking, is a frequent and highly disruptive problem that occurs at the very end of the process chain when the tablet is formed. As an alternative to the mechanistic approaches to address sticking, we introduce two different machine learning strategies to predict sticking directly from the chemical formula of the drug substance, represented by molecular descriptors. An empirical database for sticking behaviour was developed and used to train the machine learning (ML) algorithms to predict sticking properties from molecular descriptors. The ML model has successfully classified sticking/non-sticking behaviour of powders with 100% separation. Predictions were made for materials in the handbook of Pharmaceutical Excipients and a subset of molecules included in the ChemBL database, demonstrating the potential use of machine learning approaches to screen for sticking propensity early at drug discovery and development stages. This is the first-time molecular descriptors and machine learning were used to predict and screen for sticking behaviour. The method has potential to transform the development of medicines by providing manufacturability information at drug screening stage and is potentially applicable to other manufacturing problems controlled by the chemistry of the drug substance.

Expression and purification of SARS-CoV-2 receptor binding domain in Escherichia coli for diagnostic and therapeutic purposes

Background and purpose: SARS-CoV-2 causes a severe respiratory disease known as COVID-19 and is responsible for a global viral pandemic. The SARS-CoV-2 receptor binding domain (RBD) is located on the spike protein, which identifies and binds to the angiotensin-converting enzyme 2 (ACE2) receptor. The RBD is an important target for developing virus-neutralizing antibodies, vaccines, and inhibitors. Experimental approach: In this study, recombinant SARS-CoV-2 RBD was expressed in E. coli BL21 (DE3) and purified and its binding activity was determined. Purification was conducted using the Ni-NTA column. ELISA. flow cytometry assays were set to evaluate the binding ability of recombinant RBD to different anti-RBD antibodies and native ACE2 receptors on HEK293A cells, respectively. Findings/Results: The SDS-PAGE analysis revealed the corresponding band at 27 kDa in the culture after induction with 0.7 mM IPTG, while the corresponding band was not observed in the culture without IPTG induction. ELISA results showed that antibodies produced in the human sera could bind to the recombinant RBD protein and the commercial anti-RBD antibody. Also, flow cytometry analysis revealed that the recombinant RBD could bind to human ACE2 on the surface of HEK293A cells. Conclusion and implication: Our outcomes displayed that the recombinant RBD expressed in the E. coli strain has biological activity and can be used as an antigen for the development of diagnosis kits and vaccines as well as a tool for screening drugs against SASR-CoV-2.

Advances in 3D Bioprinting for Neuroregeneration: A Literature Review of Methods, Bioinks, and Applications

Recent advancements in 3D-bioprinting technology have sparked a growing interest in its application for brain repair, encompassing tissue regeneration, drug delivery, and disease modeling. This literature review examines studies conducted over the past five years to assess the current state of research in this field. Common bioprinting methods and key parameters influencing their selection are explored, alongside an analysis of the diverse types of bioink utilized and their associated parameters. The extrusion-based 3D-bioprinting method emerged as the most widely studied and popular topic, followed by inkjet-based and laser-based bioprinting and stereolithography. Regarding bioinks, fibrin-based and collagen-based bioinks are predominantly utilized. Furthermore, this review elucidates how 3D bioprinting holds promise for neural tissue repair, regeneration, and drug screening, detailing the steps involved and various approaches employed. Neurovascular 3D printing and bioscaffold 3D printing stand out as the top two preferred methods for brain repair. The recent studies’ shortcomings and potential solutions to address them are also examined and discussed. Overall, by synthesizing recent findings, this review provides valuable insights into the potential of 3D bioprinting for advancing brain repairment strategies.

Eupafolin hinders cross-talk between gastric cancer cells and cancer-associated fibroblasts by abrogating the IL18/IL18RAP signaling axis

BackgroundCancer-associated fibroblasts (CAFs) are involved in the progression of gastric cancer (GC) as a critical component of the tumor microenvironment (TME), yet specific interventions remain limited. Natural products hold a promising application prospect in the field of anti-tumor in view of their high activity and ease of binding with biological macromolecules. However, the role of natural products in modulating the cross-talk between CAFs and GC cells has not been fully investigated. PurposeThe aim of this study was to identify a potential therapeutic target in CAFs and then screen for natural small molecule drugs with anti-tumor activity against this target. MethodsIntegrating bioinformatics analysis of public databases and experimental validation of human samples and cell lines to identify a candidate target in CAFs. Molecular docking and biolayer interferometry technique were utilized for screening potential natural small molecule drugs. The efficacy and underlying mechanisms of the candidates were explored in vitro and in vivo through techniques such as lentiviral infection, cell spheroids culture, immunoprecipitation and cells-derived xenografts. ResultsIL18 receptor accessory protein (IL18RAP) was found to be overexpressed in CAFs derived from GC tissues and facilitated the protumor function of CAFs on GC. Based on virtual screening and experimental validation, we identified a natural product, eupafolin, that interfered with IL18 signaling. Phenotyping studies confirmed that the proliferation, spheroids formation and tumorigenesis of GC cells facilitated by CAFs were greatly attenuated by eupafolin both in vitro and in vivo. Mechanistically, eupafolin impeded the formation of IL18 receptor (IL18R) complex by directly binding to IL18RAP, thus blocking IL18-mediated nuclear factor kappa B (NF-κB) activation and reduced the synthesis and secretion of IL6 in CAFs. As a consequence, it inactivated signal transducer and activator of transcription 3 (STAT3) in GC cells. ConclusionThis study provides new evidence that IL18 signaling regulates the cross-talk between GC cells and CAFs. And it highlights a novel pharmacological role of eupafolin in inhibiting IL18 signaling, thereby curbing the development of GC via modulating CAFs.

Establishment and characterization of two novel patient-derived cell lines from myxofibrosarcoma: NCC-MFS7-C1 and NCC-MFS8-C1.

Myxofibrosarcoma (MFS), an aggressive soft tissue sarcoma, presents a significant challenge because of its high recurrence rate, distal metastasis, and complex genetic background. Although surgical resection is the standard treatment for MFS, the outcomes are unsatisfactory and effective non-surgical treatment strategies, including drug therapy, are urgently warranted. MFS is a rare tumor that requires comprehensive preclinical research to develop promising drug therapies; however, only two MFS cell lines are publicly available worldwide. The present study reports two novel patient-derived MFS cell lines, NCC-MFS7-C1 and NCC-MFS8-C1. These cell lines have been extensively characterized for their genetic profile, proliferation, spheroid-forming capacity, and invasive behavior, confirming that they retain MFS hallmarks. Furthermore, we conducted comprehensive drug screening against these cell lines and six others previously established in our laboratory to identify potential therapeutic candidates for MFS. Among the screened agents, actinomycin D, bortezomib, and romidepsin demonstrated considerable antiproliferative effects that were superior to those of doxorubicin, a standard drug, highlighting their potential as novel drugs. In conclusion, NCC-MFS7-C1 and NCC-MFS8-C1 are valuable research resources that contribute to the understanding of the pathogenesis and development of novel therapies for MFS.

In Silico Drug Screening for Hepatitis C Virus Using QSAR-ML and Molecular Docking with Rho-Associated Protein Kinase 1 (ROCK1) Inhibitors

The enzyme ROCK1 plays a pivotal role in the disruption of the tight junction protein CLDN1, a downstream effector influencing various cellular functions such as cell migration, adhesion, and polarity. Elevated levels of ROCK1 pose challenges in HCV, where CLDN1 serves as a crucial entry factor for viral infections. This study integrates a drug screening protocol, employing a combination of quantitative structure–activity relationship machine learning (QSAR-ML) techniques; absorption, distribution, metabolism, and excretion (ADME) predictions; and molecular docking. This integrated approach allows for the effective screening of specific compounds, using their calculated features and properties as guidelines for selecting drug-like candidates targeting ROCK1 inhibition in HCV treatment. The QSAR-ML model, validated with scores of 0.54 (R2), 0.15 (RMSE), and 0.71 (CCC), demonstrates its predictive capabilities. The ADME-Docking study’s final results highlight notable compounds from ZINC15, specifically ZINC000071318464, ZINC000073170040, ZINC000058568630, ZINC000058591055, and ZINC000058574949. These compounds exhibit the best ranking Vina scores for protein–ligand binding with the crystal structure of ROCK1 at the C2 pocket site. The generated features and calculated pIC50 bioactivity of these compounds provide valuable insights, facilitating the identification of structurally similar candidates in the ongoing exploration of drugs for ROCK1 inhibition.

Advanced strategies in 3D bioprinting for vascular tissue engineering and disease modelling using smart bioinks

ABSTRACT Advanced three-dimensional (3D) bioprinting technology enables the precise production of complex vascular structures and biomimetic models, driving advancements in tissue engineering and disease mechanism research. At the core of this technology is smart bioink, which is suitable for fabricating biomimetic models that can be vascularised to meet the property requirements of various tissues. Examples of smart bioinks include decellularized extracellular matrix (dECM), photocrosslinkable, reversible, and microgel-based biphasic (MB) bioinks, whose mechanical properties can be tuned through external stimuli. This tuning helps generate high-resolution and complex-shaped vascular networks essential for cell survival and functional maturation. This review explores advanced 3D bioprinting strategies using smart bioinks for spatially controlled and perfusable vascular in vitro models, emphasising the reconstruction of functional vascular structures within 3D bioprinted models. It also discusses challenges and future prospects, suggesting that 3D bioprinted models could serve as alternatives to traditional animal models for disease modelling and drug screening.

Drug evaluation platform based on non-destructive and real-time in situ organoid fate state monitoring by graphene field-effect transistor

3D organoid models have been widely used in drug evaluation because they effectively mimic physiological conditions and cell–cell interactions. However, current organoid-based drug evaluation heavily relies on invasive methods like qRT-PCR and immunofluorescence quantification, which disrupts the organoid’s structure and physiological function during the evaluation process. Therefore, it is urgent to develop a non-destructive drug evaluation platform to continuously monitor organoid fate. Here, a non-destructive and real-time in situ platform based on field effect transistor sensor (NDRS-FET) was developed to monitor organoid fate and obtained accurate drug evaluation by conducting the dynamic models. Using liver organoid as the model, the transition kinetics of liver organoid fate in response to candidate drug compounds were recorded by the NDRS-FET platform. OSM, FH1 and BG45 were successfully identified as effective regulators of liver organoid fate. Compared to traditional methods, the NDRS-FET platform could not only accurately capture the drug effect on organoid in real time, but also reduce the intergroup deviation among samples, providing more accurate and reliable tool for drug development and screening.

Psychometric Synthesis of the Drug Abuse Screening Test (DAST) Versions

Problem Statement Among individuals aged 12 years or older, 14.3% (40.0 million) reporting the use of an illicit drug in the previous year. Given the prevalence of drug abuse, it is increasingly important to determine effective screening practices, treatment procedures, and best practices among various subpopulations to identify drug use-related consequences. The DAST is one of the most commonly used and accurate drug screening tests. Method This psychometric synthesis of four versions of the Drug Abuse Screening Test (DAST-10, DAST-20, DAST-28, DAST-A) provided aggregated evidence from 346 articles over 40 years of published literature for score reliability, structure, diagnostic, and convergent validity, and descriptive statistics, all with the goal of informing counseling and medical practice and research. Results Results indicated adequate internal consistency (α = 0.81–0.84 across all four versions) and mostly medium to large effect size convergent correlations with comparison measures. Aggregated diagnostic validity data indicate optimal cutoff scores of 7 for DAST-10, 8 for DAST-20, 10 for DAST-28, and 6 for DAST-A. Discussion The DAST-10 appears the best choice for practical and psychometric reasons. Additional studies of the various DAST versions are needed to expand use across participant demographics. Public Statement of Relevance Drug use continues to be a societal problem and mental health practitioners need effective screening practices, treatment procedures, and best practices among various subpopulations to identify drug use-related consequences. This study synthesized 40 years of research on the four versions of the Drug Abuse Screening Test (DAST-10, DAST-20, DAST-28, DAST-A) and found acceptable levels of score reliability and validity for screening purposes.

Compressible Hollow Microlasers in Organoids for High-Throughput and Real-Time Mechanical Screening.

Mechanical stress within organoids is a pivotal indicator in disease modeling and pharmacokinetics, yet current tools lack the ability to rapidly and dynamically screen these mechanics. Here, we introduce biocompatible and compressible hollow microlasers that realize all-optical assessment of cellular stress within organoids. The laser spectroscopy yields identification of cellular deformation at the nanometer scale, corresponding to tens of pascals stress sensitivity. The compressibility enables the investigation of the isotropic component, which is the fundamental mechanics of multicellular models. By integrating with a microwell array, we demonstrate the high-throughput screening of mechanical cues in tumoroids, establishing a platform for mechano-responsive drug screening. Furthermore, we showcase the monitoring and mapping of dynamic contractile stress within human embryonic stem cell-derived cardiac organoids, revealing the internal mechanical inhomogeneity within a single organoid. This method eliminates time-consuming scanning and sample damage, providing insights into organoid mechanobiology.

An Array of Carbon Nanofiber Bundle_Based 3D In Vitro Intestinal Microvilli for Mimicking Functional and Physical Activities of the Small Intestine.

Researchers have developed in vitro small intestine models of biomimicking microvilli, such as gut-on-a-chip devices. However, fabrication methods developed to date for 2D and 3D in vitro gut still have unsolved limitations. In this study, an innovative fabrication method of a 3D in vitro gut model is introduced for effective drug screening. The villus is formed on a patterned carbon nanofiber (CNF) bundle as a flexible and biocompatible scaffold. Mechanical properties of the fabricated villi structure are investigates. A microfluidic system is applied to induce the movement of CNFs villi. F-actin and Occludin staining of Caco-2 cells on a 2D flat-chip as a control and a 3D gut-chip with or without fluidic stress is observed. A permeability test of FD20 is performed. The proposed 3D gut-chip with fluidic stress achieve the highest value of Papp. Mechano-active stimuli caused by distinct structural and movement effects of CNFs villi as well as stiffness of the suggested CNFs villi not only can help accelerate cell differentiation but also can improve permeability. The proposed 3D gut-chip system further strengthens the potential of the platform to increase the accuracy of various drug tests.

Harnessing the Biomimetic Effect of Macromolecular Crowding in the Cell-Derived Model of Clubfoot Fibrosis.

Fibrotic changes in pediatric clubfoot provide an opportunity to improve corrective therapy and prevent relapses with targeted drugs. This study defines the parameters of clubfoot fibrosis and presents a unique analysis of a simple pseudo-3D in vitro model for disease-specific high-throughput drug screening experiments. The model combines clubfoot-derived fibroblasts with a biomimetic cultivation environment induced by the water-soluble polymers Ficoll and Polyvinylpyrrolidone, utilizing the principle of macromolecular crowding. We achieved higher conversion of soluble collagen into insoluble collagen, accelerated formation of the extracellular matrix layer and upregulated fibrosis-related genes in the mixed Ficoll environment. To test the model, we evaluated the effect of a potential antifibrotic drug, minoxidil, emphasizing collagen content and cross-linking. While the model amplified overall collagen deposition, minoxidil effectively blocked the expression of lysyl hydroxylases, which are responsible for the increased occurrence of specific collagen cross-linking in various fibrotic tissues. This limited the formation of collagen cross-link in both the model and control environments. Our findings provide a tool for expanding preclinical research for clubfoot and similar fibroproliferative conditions.

Genetic and Environmental Factors Co-Contributing to Behavioral Abnormalities in adnp/adnp2 Mutant Zebrafish.

Human mutations of ADNP and ADNP2 are known to be associated with neural developmental disorders (NDDs), including autism spectrum disorders (ASDs) and schizophrenia (SZ). However, the underlying mechanisms remain elusive. In this study, using CRISPR/Cas9 gene editing technology, we generated adnp and adnp2 mutant zebrafish models, which exhibited developmental delays, brain deficits, and core behavioral features of NDDs. RNA sequencing analysis of adnpa-/-; adnpb-/- and adnp2a-/-; adnp2b-/- larval brains revealed altered gene expression profiles affecting synaptic transmission, autophagy, apoptosis, microtubule dynamics, hormone signaling, and circadian rhythm regulation. Validation using whole-mount in situ hybridization (WISH) and real-time quantitative PCR (qRT-PCR) corroborated these findings, supporting the RNA-seq results. Additionally, loss of adnp and adnp2 resulted in significant downregulation of pan-neuronal HuC and neuronal fiber network α-Tubulin signals. Importantly, prolonged low-dose exposure to environmental endocrine disruptors (EEDs) aggravated behavioral abnormalities in adnp and adnp2 mutants. This comprehensive approach enhances our understanding of the complex interplay between genetic mutations and environmental factors in NDDs. Our findings provide novel insights and experimental foundations into the roles of adnp and adnp2 in neurodevelopment and behavioral regulation, offering a framework for future preclinical drug screening aimed at elucidating the pathogenesis of NDDs and related conditions.

Signature Search Polestar: a comprehensive drug repurposing method evaluation assistant for customized oncogenic signature.

The burgeoning high-throughput technologies have led to a significant surge in the scale of pharmacotranscriptomic datasets, especially for oncology. Signature search methods (SSMs), utilizing oncogenic signatures formed by differentially expressed genes through sequencing, have been instrumental in anti-cancer drug screening and identifying mechanisms of action without relying on prior knowledge. However, various studies have found that different SSMs exhibit varying performance across pharmacotranscriptomic datasets. In addition, the size of the oncogenic signature can also significantly impact the result of drug repurposing. Therefore, finding the optimal SSMs and customized oncogenic signature for a specific disease remains a challenge. To address this, we introduce Signature Search Polestar (SSP), a webserver integrating the largest pharmacotranscriptomic datasets of anti-cancer drugs from LINCS L1000 with five state-of-the-art SSMs (XSum, CMap, GSEA, ZhangScore, XCos). SSP provides three main modules: Benchmark, Robustness, and Application. Benchmark uses two indices, Area Under the Curve and Enrichment Score, based on drug annotations to evaluate SSMs at different oncogenic signature sizes. Robustness, applicable when drug annotations are insufficient, uses a performance score based on drug self-retrieval for evaluation. Application provides three screening strategies, single method, SS_all, and SS_cross, allowing users to freely utilize optimal SSMs with tailored oncogenic signature for drug repurposing. SSP is free at https://web.biotcm.net/SSP/. The current version of SSP is archived in https://doi.org/10.6084/m9.figshare.26524741.v1, allowing users to directly use or customize their own SSP webserver.

From Static to Dynamic Structures: Improving Binding Affinity Prediction with Graph-Based Deep Learning.

Accurate prediction of protein-ligand binding affinities is an essential challenge in structure-based drug design. Despite recent advances in data-driven methods for affinity prediction, their accuracy is still limited, partially because they only take advantage of static crystal structures while the actual binding affinities are generally determined by the thermodynamic ensembles between proteins and ligands. One effective way to approximate such a thermodynamic ensemble is to use molecular dynamics (MD) simulation. Here, an MD dataset containing 3,218 different protein-ligand complexes is curated, and Dynaformer, a graph-based deep learning model is further developed to predict the binding affinities by learning the geometric characteristics of the protein-ligand interactions from the MD trajectories. In silico experiments demonstrated that the model exhibits state-of-the-art scoring and ranking power on the CASF-2016 benchmark dataset, outperforming the methods hitherto reported. Moreover, in a virtual screening on heat shock protein 90 (HSP90) using Dynaformer, 20 candidates are identified and their binding affinities are further experimentally validated. Dynaformer displayed promising results in virtual drug screening, revealing 12 hit compounds (two are in the submicromolar range), including several novel scaffolds. Overall, these results demonstrated that the approach offer a promising avenue for accelerating the early drug discoveryprocess.