Machine learning (ML) is increasingly used in DNA-encoded library (DEL) screening for ligand discovery, but its success depends on access to suitable data sets, which are often proprietary and costly. To overcome this, we present the first fully open, automated DEL-ML framework using public DEL data sets and chemical fingerprints to enable reproducible, accessible drug discovery. Our workflow─from model training to virtual screening and compound selection─requires no human intervention. As a proof of concept, we identified binders for WDR91 by training ML models on the HitGen OpenDEL library (3B molecules) and screening the Enamine REAL Space library (37B molecules), yielding 50 candidates. Experimental testing confirmed seven novel binders with dissociation constants between 2.7-21 μM. Our open-source approach matches the performance of proprietary methods, demonstrating that public DEL data can support robust ML-driven ligand discovery and fostering transparency and broader community participation in drug development.

Molecular glue degraders enable targeted protein degradation by bridging interactions between target proteins and E3 ubiquitin ligases. Whereas some target-E3 interfaces exhibit the capacity to accommodate structurally diverse degraders, the extent of this adaptability across molecular glue targets remains unclear. We recently identified (S)-ACE-OH as a molecular glue degrader that recruits the E3 ubiquitin ligase TRIM21 to the nuclear pore complex by recognizing NUP98, thereby inducing the degradation of nuclear pore proteins. Here, we analyzed public compound toxicity data across a large collection of cell lines and identified two additional molecular glue degraders, PRLX 93936 and BMS-214662, which engage the TRIM21-NUP98 interface to induce selective degradation of nuclear pore proteins. Additionally, we confirmed that HGC652, another TRIM21-dependent molecular glue degrader, also binds at this interface. Together with our previously characterized degrader (S)-ACE-OH, these findings demonstrate that the TRIM21-NUP98 interface can accommodate structurally diverse molecular glue degraders.

Despite the exciting progress of bifunctional degrader molecules, also known as proteolysis-targeting chimeras (PROTACs), the rapidly expanding field is still significantly hampered by the lack of available E3 ligase ligands. Our research bridges this gap by uncovering a series of small-molecule ligands to the E3 ligase TRIM21 through DNA-Encoded Library (DEL) technology. We confirmed their interaction with TRIM21 using crystallography and demonstrated their antiproliferative effects across various cancer cell types. Furthermore, proteomic studies identified that the mRNA Export Factor GLE1 and the Nuclear Pore Complex Protein NUP155 were significantly downregulated on TRIM21 ligand treatment. This degradation required TRIM21 and was ubiquitin-proteasome-dependent. More specifically, NUP155 was the primary target for the TRIM21 ligands, while GLE1 was considered a passenger target on initial degradation of NUP155. Using immunofluorescence techniques, we further demonstrated that the degradation of GLE1 and NUP155 proteins impaired the integrity of the nuclear envelope, leading to cell death. Highlighted by this research, a novel mode of action has been discovered for the TRIM21 E3 ligase ligand, acting as a monovalent degrader that triggers de novo interaction with functional complex proteins and induces their degradation.

To better understand how pre-installed covalent warheads affect the ligand discovery in DNA encoded library (DEL), we have designed three individual covalent DELs incorporating 7, 32 and 64 cysteine-targeting covalent warheads respectively, and screened these DELs against JAK3 purified protein. The experiments resulted 6 novel series of covalent inhibitors with drug-like properties, where the most potent compounds achieved picomolor IC50 and good selectivity against a mini panel of kinases. The mass spec study confirmed their covalent MOAs by targeting JAK3 Cys909. More importantly, we confirmed the synergistic effect of the binding moiety and warhead by comparing the activities with their close analogs, suggesting that these compounds may not able to be designed by installation of covalent warheads to reversible binders. Further analysis revealed that 7 warheads were sufficient for identifying JAK3 covalent ligands. This work deepens our understanding of the design and screening of covalent DEL, and also demonstrate the power of DEL in the identification of diverse inhibitors.

TRIM21 is a highly efficient and versatile E3 ubiquitin ligase that plays a crucial role in targeted protein degradation through its specific binding capabilities. To investigate its role in chemically mediated degradation, we utilized two high-affinity binders identified from a DNA-encoded compound library (DEL) to design and synthesize several TrimTACs (TRIM21-based degraders) targeting BRD4. Degradation assays revealed that TrimTACs incorporating distinct TRIM21-binding moieties produced markedly different levels of BRD4 degradation. Further, to elucidate mechanisms underlying ternary complex formation and selective degradation, we employed precise ternary complex modeling combined with Gaussian accelerated molecular dynamics (GaMD) simulations. These computational analyses demonstrated that the diversity of TRIM21-binding moieties significantly affects the stability and conformation of the complex. Structurally stable complexes efficiently induce BRD4 degradation, and the binding orientation of BRD4 directly affects ubiquitin transfer and degradation efficiency. Moreover, key residues critical for complex formation were identified, shedding light on the cooperative interactions driving TRIM21-mediated degradation. Building on these findings, we conducted degradation experiments to validate novel TrimTAC designs derived from our ternary complex-based free energy perturbation (FEP) calculations. The results confirmed the reliability and accuracy of the ternary complex model, providing key insights for the rational design and optimization of TrimTACs, thereby advancing the development of targeted protein degraders.

NLRP3 is an intracellular sensor protein that detects a broad range of danger signals and environmental insults. Its activation results in a protective pro-inflammatory response designed to impair pathogens and repair tissue damage via the formation of the NLRP3 inflammasome. Assembly of the NLRP3 inflammasome leads to caspase 1-dependent secretory release of the pro-inflammatory cytokines IL-1 beta and IL-18 as well as to gasdermin d-mediated pyroptotic cell death. Herein, we describe the discovery of a novel indazole series of high affinity, reversible inhibitors of NLRP3 activation through screening of DNA-encoded libraries and the potent lead compound 3 (BAL-0028, IC50 = 25 nM) that was identified directly from the screen. SPR studies showed that compound 3 binds tightly (K range 104-123 nM) to the NACHT domain of NLRP3. A CADD analysis of the interaction of compound 3 with the NLRP3 NACHT domain proposes a binding site that is distinct from those of ADP and MCC950 and includes specific site interactions. We anticipate that compound 3 (BAL-0028) and other members of this novel indazole class of neutral inhibitors will demonstrate significantly different physical, biochemical, and biological properties compared to NLRP3 inhibitors previously identified.

Despite the exciting progress of the bifunctional degrader molecules, also known as proteolysis-targeting chimeras (PROTACs), the rapidly expanding field is still significantly hampered by the lacking of available E3 ligase ligands. Our research bridges this gap by uncovering a series of small-molecule ligands to the E3 ligase TRIM21 through DNA-Encoded Library (DEL) technology. We confirmed their interaction with TRIM21 using crystallography and demonstrated their anti-proliferative effects across various cancer cell types. Furthermore, proteomic studies identified that the mRNA Export Factor GLE1, and the Nuclear Pore Complex Protein NUP155, were significantly down-regulated upon TRIM21 ligand treatment. This degradation required TRIM21 and was ubiquitin-proteasome-dependent. More specifically, NUP155 was the primary target for the TRIM21 ligands, while GLE1 was considered a passenger target upon the initial degradation of NUP155. Using immunofluorescence techniques, we further demonstrated that the degradation of GLE1 and NUP155 proteins impaired the integrity of the nuclear envelope, leading to cell death. Highlighted by this research, a novel mode of action has been discovered for the TRIM21 E3 ligase ligand, acting as a monovalent degrader that triggers de novo interaction with functional complex proteins and induces their degradation. ### Competing Interest Statement The authors have declared no competing interest.

Recent advances in DNA-encoded library (DEL) screening have created bioactivity datasets containing billions of molecules, unlocking new opportunities for machine learning (ML) in drug discovery. However, most ultra-large DEL libraries are proprietary, limiting the advancement of ML tools for big chemical data analytics and hindering the democratization of DEL-ML technology. We address this gap by developing an open, end-to-end DEL-ML framework using public datasets, where enriched binders are represented by common chemical fingerprints, ensuring proprietary data protection. We demonstrate that ML models can be built and validated on fingerprinted DEL data and then applied to virtual screening (VS) of billion-sized, publicly accessible chemical libraries. As a proof-of-concept, we screened the human protein WDR91 using the HitGen OpenDEL library (3 billion molecules) and trained ML models, which were used to screen the Enamine REAL Space library (37 billion molecules). Fifty potential binders were identified, 48 of which were tested, and seven were confirmed as novel binders with dissociation constants (KD) from 2.7 to 21 μM that were successfully co-crystalized with WDR91. This fully automated, open-source workflow demonstrates the potential of DEL-ML models in discovering novel binders and promotes the use of open chemical bioactivity datasets and ML to accelerate drug discovery.

Peptidyl-prolyl cis-trans isomerase NIMA-interacting 1 (PIN1) has emerged as a promising therapeutic target for cancer treatment. However, the current PIN1 inhibitors have shown limited efficacy in animal models, leaving the question of whether PIN1 is a proper oncologic target still unanswered. By screening a 1 trillion DNA-encoded library (DEL), we identified novel nonacidic compounds. Among resynthesized DEL compounds, DEL1067-56-469 (A0) is the most potent one (KD = 430 nM, IC50 = 420 nM). Further optimization of A0 resulted in compound C10 with much improved potency (KD = 25 nM, IC50 = 150 nM). As an alternative approach, C10 was then converted into proteolysis targeting chimeras (PROTACs) in order to achieve deeper downregulation of the PIN1 protein in cancer cell lines. Unfortunately, neither PIN1 inhibitors nor PIN1 PROTACs demonstrated meaningful antiproliferation activity. In addition, siRNA knock-down experiments provided unfavorable evidence of PIN1 as an oncologic target. Our findings highlight the complexity of targeting PIN1 for cancer therapy.

The proteolysis targeting chimera (PROTAC) strategy results in the down-regulation of unwanted protein(s) for disease treatment. In the PROTAC process, a heterobifunctional degrader forms a ternary complex with a target protein of interest (POI) and an E3 ligase, which results in ubiquitination and proteasomal degradation of the POI. While ternary complex formation is a key attribute of PROTAC degraders, modification of the PROTAC molecule to optimize ternary complex formation and protein degradation can be a labor-intensive and tedious process. In this study, we take advantage of DNA-encoded library (DEL) technology to efficiently synthesize a vast number of possible PROTAC molecules and describe a parallel screening approach that utilizes DNA barcodes as reporters of ternary complex formation and cooperative binding. We use a designed PROTAC DEL against BRD4 and CRBN to describe a dual protein affinity selection method and the direct discovery of novel, potent BRD4 PROTACs that importantly demonstrate clear SAR. Such an approach evaluates all the potential PROTACs simultaneously, avoids the interference of PROTAC solubility and permeability, and uses POI and E3 ligase proteins in an efficient manner.

DNA-encoded library (DEL) technology is a powerful tool for small molecule identification in drug discovery, yet the reported DEL selection strategies were applied primarily on protein targets in either purified form or in cellular context. To expand the application of this technology, we employed DEL selection on an RNA target HIV-1 TAR (trans-acting responsive region), but found that the majority of signals were resulted from false positive DNA-RNA binding. We thus developed an optimized selection strategy utilizing RNA patches and competitive elution to minimize unwanted DNA binding, followed by k-mer analysis and motif search to differentiate false positive signal. This optimized strategy resulted in a very clean background in a DEL selection against Escherichia coli FMN Riboswitch, and the enriched compounds were determined with double digit nanomolar binding affinity, as well as similar potency in functional FMN competition assay. These results demonstrated the feasibility of small molecule identification against RNA targets using DEL selection. The developed experimental and computational strategy provided a promising opportunity for RNA ligand screening and expanded the application of DEL selection to a much wider context in drug discovery.

We have developed a graph-based Variational Autoencoder with Gaussian Mixture hidden space (GraphGMVAE), a deep learning approach for controllable magnitude of scaffold hopping in generative chemistry. It can effectively and accurately generate molecules from a given reference compound, with excellent scaffold novelty against known molecules in the literature or patents (97.9% are novel scaffolds). Moreover, a pipeline for prioritizing the generated compounds was also proposed to narrow down our validation focus. In this work, GraphGMVAE was validated by rapidly hopping the scaffold from FDA-approved upadacitinib, which is an inhibitor of human Janus kinase 1 (JAK1), to generate more potent molecules with novel chemical scaffolds. Seven compounds were synthesized and tested to be active in biochemical assays. The most potent molecule has 5.0 nM activity against JAK1 kinase, which shows that the GraphGMVAE model can design molecules like how a human expert does but with high efficiency and accuracy.

Two novel compounds were identified as Naa50 binders/inhibitors using DNA-encoded technology screening. Biophysical and biochemical data as well as cocrystal structures were obtained for both compounds (3a and 4a) to understand their mechanism of action. These data were also used to rationalize the binding affinity differences observed between the two compounds and a MLGP peptide-containing substrate. Cellular target engagement experiments further confirm the Naa50 binding of 4a and demonstrate its selectivity toward related enzymes (Naa10 and Naa60). Additional analogs of inhibitor 4a were also evaluated to study the binding mode observed in the cocrystal structures.

DNA-encoded library (DEL) technology has been used as an ultra-high-throughput screening approach for hit identification of drug targets. This process is an affinity-based selection and requires incubation of DEL molecules with the target. Currently, in most reported cases, the input (i.e., the copy number) of individual DEL molecules varies from 105 to 107. With the ever-increasing DEL size and screening cost, lowering the input of DEL molecules while maintaining an appropriate signal-to-noise ratio in a selection is of paramount importance. In this article, we varied the input of DEL ranging from 103 to 105 in selections with two different protein targets to explore the lower limit of DEL molecule input. The results could facilitate the optimization of the DEL selection process and reduce costs related to library consumption.

Biofilm formation is a critical determinant in the pathopoiesis of Pseudomonas aeruginosa It could significantly increase bacterial resistance to drugs and host defense. Thus, inhibition of biofilm matrix production could be regarded as a promising attempt to prevent colonization of P. aeruginosa and the subsequent infection. PpgL, a periplasmic gluconolactonase, has been reported to be involved in P. aeruginosa quorum-sensing (QS) system regulation. However, the detailed function and catalysis mechanism remain elusive. Here, the crystal structure of PpgL is described in the current study, along with biochemical analysis, revealing that PpgL is a typical β-propeller enzyme with unique metal-independent lactone hydrolysis activity. Consequently, comparative analysis of seven-bladed propeller lactone-catalyzing enzymes and mutagenesis studies identify the critical sites which contribute to the diverse catalytic and substrate recognition functions. In addition, the reduced biofilm formation and attenuated invasion phenotype resulting from deletion of ppgL confirm the importance of PpgL in P. aeruginosa pathogenesis. These results suggest that PpgL is a potential target for developing new agents against the diseases caused by P. aeruginosa.

The invention discloses a method for solid-phase synthesis of a DNA-coded compound database. The method comprises the following steps: a, reacting a solid-phase carrier G-1 with a junctional molecule L-1 so as to prepare L-G-1; b, reacting DNA with a junctional molecule L-0 so as to prepare L-2; c, reacting L-G-1 with L-2 to prepare L-G-2; d, removing the protective group of L-G-2 so as to obtain L-G-2-1; e, reacting L-G-2-1 with a synthesized building block and carrying out DNA coding; and f, cutting off a solid-phase carrier so as to obtain the DNA-coded compound database. Compared with the prior art, the method has the advantage that aftertreatment purification of a reaction can be completed through simple filtering and washing operations a plurality of times; operation is simple; a production period for synthesis of the DNA-coded compound database is reduced by 50% or above, production efficiency and uniqueness of the DNA-coded compound database can be substantially improved, and the purity of the final products is greatly enhanced; and the method has high economic value and is suitable for industrial application.

This chapter provides a survey of advanced technologies in lead generation, which demonstrate capabilities for challenging targets. DNA-encoded library (DEL) technology enables billions of compounds to be screened within a highly miniaturized format, facilitated by high-capacity DNA sequencing and provides opportunities in identifying allosteric modulators. The effectiveness of a DEL eventually depends on library design, synthesis quality and how robust the screening process and the sequencing power are. The chapter focuses on the recent peptide drug improvement, especially for the stapled peptide approach. After careful design and modification, hydrocarbon-stapled peptides are able to achieve enhanced in vitroand in vivobiological activity coupled with oral absorption. Phenotypic screening is very helpful for screening compounds that would upon binding modulate a variety of signal transduction events, especially for targets with unknown function. It is generally believed that active compounds identified through physiologically relevant phenotypic assays are more likely to translate into clinical success.

The present invention provides a precursor compound is administered for a permeable membrane, and a pharmaceutical carrier based on the precursor formulation. Administration of the compound and a pharmaceutical carrier precursor formulation prepared according to the invention, can be difficult to improve membrane permeability of the compound-permeable membrane, provide a new choice for clinical medication.

Provided is a method for cell membrane permeation for a compound, comprising the following steps: (1) acquisition of raw materials: the compound and a DNA or an RNA; (2) connection: the compound is connected with either the DNA or the RNA to acquire a molecular conjugate; and, (3) transfer: a gene transfer method is used to transfer the molecular conjugate acquired in step (2) into a cell. Also provided are a molecular conjugate for use in transmembrane transfer and a method for synthesizing the molecular conjugate.

We reported previously that insect acetylcholinesterases (AChEs) could be selectively and irreversibly inhibited by methanethiosulfonates presumably through conjugation to an insect-specific cysteine in these enzymes. However, no direct proof for the conjugation has been published to date, and doubts remain about whether such cysteine-targeting inhibitors have desirable kinetic properties for insecticide use. Here we report mass spectrometric proof of the conjugation and new chemicals that irreversibly inhibited African malaria mosquito AChE with bimolecular inhibition rate constants (k(inact)/K(I)) of 3,604-458,597 M(-1)sec(-1) but spared human AChE. In comparison, the insecticide paraoxon irreversibly inhibited mosquito and human AChEs with k(inact)/K(I) values of 1,915 and 1,507 M(-1)sec(-1), respectively, under the same assay conditions. These results further support our hypothesis that the insect-specific AChE cysteine is a unique and unexplored target to develop new insecticides with reduced insecticide resistance and low toxicity to mammals, fish, and birds for the control of mosquito-borne diseases.