Biology subjects

Lee, A. A.

Publications and source records attributed to Lee, A. A..

13 recordsLinked to original sources

bioRxiv2026

High-throughput biochemical phenotyping of SHP2 variants reveals molecular basis of diseases and allosteric drug inhibition

Interpreting clinical and functional consequences of genetic variants remains challenging due to limited quantitative biochemical data at scale. We applied high-throughput microfluidic enzyme kinetics to profile 190 clinical variants of SHP2, a protein tyrosine phosphatase linked to developmental disorders and cancers. Through >300,000 reaction progress curves, we derived kinetic and thermodynamic parameters quantifying variant effects on catalysis, autoinhibition, stability, phosphopeptide binding, and drug responses. This multidimensional dataset reveals that dysregulated autoinhibition, rather than altered stability or catalysis, predominantly determines SHP2-associated pathogenesis. Thermodynamic modeling reveals that clinical-stage allosteric inhibitors preferentially stabilize a previously underappreciated, partially active conformation over the fully inactive state, leading to variant-dependent drug responses. Our high-throughput biochemical framework establishes a general approach to decipher the biochemical logic connecting protein variants to clinical outcomes.

biochemistry↗

bioRxiv2025

Structure-Activity Relationship Studies Towards Analogues of Pleconaril as Novel Enterovirus-D68 Capsid-Targeting Antivirals

Non-polio enteroviruses (NPEV) such as enterovirus D68 (EV-D68) that are highly infectious and associated with polio-like neurological complications have caused out-breaks, globally, in recent years. While some clinical and preclinical compounds have shown efficacy against NPEV in-vitro, liabilities that caused historical compounds such as pleconaril to fall short of FDA approval still remain. We present herein SAR and SPR studies of analogues of clinical compounds such as pleconaril and vapendavir against EV-D68 as a representative NPEV. Numerous structurally differentiated analogues with EV-D68 antiviral activity and useful ADME properties were discovered, which could serve as starting points for future EV drug discovery campaigns. Screening against a panel of enteroviruses revealed moderately broad-spectrum anti-EV activity of compound 26.

pharmacology and toxicology↗

bioRxiv2025

Discovery of orally bioavailable Zika and West Nile Virus antiviral compounds targeting the NS2B-NS3 protease

Flaviviruses are a class of pathogenic viruses with pandemic potential that are typically transmitted via infected arthropods. In addition, Zika virus is sexually transmissible and causes congenital malformations if infection occurs during pregnancy. Although over 1.5 million people were infected during the 2015-2016 Zika outbreak, to date, there are no clinical-stage vaccines or antivirals. Herein, we report the discovery of potent inhibitors of the Zika virus NS2B-NS3 protease that also show activity against the West Nile virus NS2B-NS3 protease. Starting from a crystallographic fragment screen, we employed a pharmacophore approach coupled with high-throughput library chemistry to elaborate fragments in the active site. Potent, metabolically stable, non-covalent, non-peptidomimetic inhibitors were identified with antiviral activity in vitro. The lead compound was progressed to a three-day Zika virus challenge study in AG129 mice, where it displayed significant viral RNA load reduction in both mouse plasma and spleen.

microbiology↗

bioRxiv2025

Open-science discovery of DNDI-6510, a compound that addresses genotoxic and metabolic liabilities of the COVID Moonshot SARS-CoV-2 Mpro lead inhibitor

The 2020 SARS-CoV-2 coronavirus pandemic highlighted the urgent need for novel small molecule antiviral drugs. (S)-x38 DNDI-6510 is a non-covalent SARS-CoV-2 main protease inhibitor developed by the open science collaboration COVID Moonshot. Here, we report on the metabolic and toxicologic optimization of the lead series previously disclosed by the COVID Moonshot Initiative, leading up to the selection of (S)-x38 DNDI-6510 as the preclinical candidate. We describe the thorough profiling of the series, identifying key risks such as formation of genotoxic metabolites and high clearance, which were successfully addressed during lead optimization. In addition, we disclose the in vitro and in vivo evaluation of (S)-x38 DNDI-6510 in pharmacokinetic and pharmacodynamic models, exploring multiple approaches to ameliorate rodent-specific metabolic clearance, and show that both co-dosing of (S)-x38 DNDI-6510 with an ABT inhibitor and utilizing a metabolically humanized mouse model (8HUM) achieve significant improvements in exposure. Through comparisons of ABT co-dosing and humanized mouse models in efficacy experiments, we demonstrate that continuous exposure over cellular EC90 is required for SARS-CoV-2 antiviral efficacy in vivo in an antiviral model using a mouse-adapted SARS-CoV-2 strain. Finally, (S)-x38 DNDI-6510 was assessed in maximum tolerated dose experiments in two species, demonstrating significant in vivo PXR-linked auto-induction of metabolism, leading to the discontinuation of this compound. In summary, we report the successful effort to overcome series-specific AMES liabilities in a lead development program. Downstream optimization of existing series will require in-depth optimization of rodent-specific liabilities and metabolic induction profile.

pharmacology and toxicology↗

bioRxiv2024

Discovery of potent SARS-CoV-2 nsp3 macrodomain inhibitors uncovers lack of translation to cellular antiviral response

A strategy for pandemic preparedness is the development of antivirals against a wide set of viral targets with complementary mechanisms of action. SARS-CoV-2 nsp3-mac1 is a viral macrodomain with ADP-ribosylhydrolase activity, which counteracts host immune response. Targeting the virus immunomodulatory functionality offers a differentiated strategy to inhibit SARS-CoV-2 compared to approved therapeutics, which target viral replication directly. Here we report a fragment-based lead generation campaign guided by computational approaches. We discover tool compounds which inhibit nsp3-mac1 activity at low nanomolar concentrations, and with responsive structure-activity relationships, high selectivity, and drug-like properties. Using our inhibitors, we show that inhibition of nsp3-mac1 increases ADP-ribosylation, but surprisingly does not translate to demonstrable antiviral activity in cell culture and iPSC-derived pneumocyte models. Further, no synergistic activity is observed in combination with interferon gamma, a main protease inhibitor, nor a papain-like protease inhibitor. Our results question the extent to which targeting modulation of innate immunity-driven ADP-ribosylation can influence SARS-CoV-2 replication. Moreover, these findings suggest that nsp3-mac1 might not be a suitable target for antiviral therapeutics development.

microbiology↗

bioRxiv2024

Discovery of orally active SARS-CoV-2 papain-like protease (PLpro) inhibitors

Vaccines and first-generation antiviral therapeutics have provided important protection against coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, there remains a need for additional therapeutic options that provide enhanced efficacy and protection against potential viral resistance. The SARS-CoV-2 papain-like protease (PLpro) is one of two essential cysteine proteases involved in viral replication. While inhibitors of the SARS-CoV-2 main protease (Mpro) have demonstrated clinical efficacy, known PLpro inhibitors have to date lacked the inhibitory potency and requisite pharmacokinetics to demonstrate that targeting PLpro translates to in vivo efficacy in a preclinical setting. Herein, we report the machine learning-driven discovery of potent, selective, and orally available SARS-CoV-2 PLpro inhibitors, with lead compound PF-07957472 (4) providing robust efficacy in a mouse-adapted model of COVID-19 infection.

microbiology↗

bioRxiv2023

Bimodality in Ras signaling originates from processivity of the Ras activator SOS without classic kinetic bistability

Ras is a small GTPase that is central to important functional decisions in diverse cell types. An important aspect of Ras signaling is its ability to exhibit bimodal, or switch-like activity. We describe the total reconstitution of a receptor-mediated Ras activation-deactivation reaction catalyzed by SOS and p120-RasGAP on supported lipid membrane microarrays. The results reveal a bimodal Ras activation response, which is not a result of classic kinetic bistability, but is rather driven by the distinct processivity of the Ras activator, SOS. Furthermore, the bimodal response is controlled by the condensation state of the scaffold protein, LAT, to which SOS is recruited. Processivity-driven bimodality leads to stochastic bursts of Ras activation even under strongly deactivating conditions. This behavior contrasts classic kinetic bistability and is distinctly more resistant to pharmacological inhibition.

biophysics↗

bioRxiv2022

Fragment-Based Hit Discovery via Unsupervised Learning of Fragment-Protein Complexes

The process of finding molecules that bind to a target protein is a challenging first step in drug discovery. Crystallographic fragment screening is a strategy based on elucidating binding modes of small polar compounds and then building potency by expanding or merging them. Recent advances in high-throughput crystallography enable screening of large fragment libraries, reading out dense ensembles of fragments spanning the binding site. However, fragments typically have low affinity thus the road to potency is often long and fraught with false starts. Here, we take advantage of high-throughput crystallography to reframe fragment-based hit discovery as a denoising problem - identifying significant pharmacophore distributions from a fragment ensemble amid noise due to weak binders - and employ an unsupervised machine learning method to tackle this problem. Our method screens potential molecules by evaluating whether they recapitulate those fragment-derived pharmacophore distributions. We retrospectively validated our approach on an open science campaign against SARS-CoV-2 main protease (Mpro), showing that our method can distinguish active compounds from inactive ones using only structural data of fragment-protein complexes, without any activity data. Further, we prospectively found novel hits for Mpro and the Mac1 domain of SARS-CoV-2 non-structural protein 3. More broadly, our results demonstrate how unsupervised machine learning helps interpret high throughput crystallography data to rapidly discover of potent chemical modulators of protein function.

bioinformatics↗

bioRxiv2021

Turning high-throughput structural biology into predictive inhibitor design

A common challenge in drug design pertains to finding chemical modifications to a ligand that increases its affinity to the target protein. An underutilised advance is the increase in structural biology throughput, which has progressed from an artisanal endeavour to a monthly throughput of up to 100 different ligands against a protein in modern synchrotrons. However, the missing piece is a framework that turns high throughput crystallography data into predictive models for ligand design. Here we designed a simple machine learning approach that predicts protein-ligand affinity from experimental structures of diverse ligands against a single protein paired with biochemical measurements. Our key insight is using physics-based energy descriptors to represent protein-ligand complexes, and a learning-to-rank approach that infers the relevant differences between binding modes. We ran a high throughput crystallography campaign against the SARS-CoV-2 Main Protease (MPro), obtaining parallel measurements of over 200 protein-ligand complexes and the binding activity. This allows us to design a one-step library syntheses which improved the potency of two distinct micromolar hits by over 10-fold, arriving at a non-covalent and non-peptidomimetic inhibitor with 120 nM antiviral efficacy. Crucially, our approach successfully extends ligands to unexplored regions of the binding pocket, executing large and fruitful moves in chemical space with simple chemistry.

biophysics↗

bioRxiv2021

Membrane-mediated dimerization potentiates PIP5K lipid kinase activity

The phosphatidylinositol 4-phosphate 5-kinase (PIP5K) family of lipid modifying enzymes generate the majority of phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) lipids found at the plasma membrane in eukaryotic cells. PI(4,5)P2 lipids serve a critical role in regulating receptor activation, ion channel gating, endocytosis, and actin nucleation. Here we describe how PIP5K activity is regulated by cooperative binding to PI(4,5)P2 lipids and membrane-mediated dimerization of the kinase domain. In contrast to constitutively dimeric phosphatidylinositol 5-phosphate 4-kinase (PIP4K, type II PIPK), solution PIP5K exists in a weak monomer-dimer equilibrium. PIP5K monomers can associate with PI(4,5)P2 containing membranes and dimerize in a protein density dependent manner. Although dispensable for PI(4,5)P2 binding and lipid kinase activity, dimerization enhances the catalytic efficiency of PIP5K through a mechanism consistent with allosteric regulation. Additionally, dimerization amplifies stochastic variation in the kinase reaction velocity and strengthens effects such as the recently described stochastic geometry sensing. Overall, the mechanism of PIP5K membrane binding creates a broad dynamic range of lipid kinase activities that are coupled to the density of PI(4,5)P2 and membrane bound kinase.

biochemistry↗

bioRxiv2020

Machine learning models for predicting protein condensate formation from sequence determinants and embeddings

Intracellular phase separation of proteins into biomolecular condensates is increasingly recognised as an important phenomenon for cellular compartmentalisation and regulation of biological function. Different hypotheses about the parameters that determine the tendency of proteins to form condensates have been proposed with some of them probed experimentally through the use of constructs generated by sequence alterations. To broaden the scope of these observations, here, we established an in silico strategy for understanding on a global level the associations between protein sequence and condensate formation, and used this information to construct machine learning classifiers for predicting liquid-liquid phase separation (LLPS) from protein sequence. Our analysis highlighted that LLPS-prone sequences are more disordered, hydrophobic and of lower Shannon entropy than sequences in the Protein Data Bank or the Swiss-Prot database, and have their disordered regions enriched in polar, aromatic and charged residues. Using these determining features together with neural network based word2vec sequence embeddings, we developed machine learning classifiers for predicting protein condensate formation. Our model, trained to distinguish LLPS-prone sequences from structured proteins, achieved high accuracy (93%; 25-fold cross-validation) and identified condensate forming sequences from external independent test data at 97% sensitivity. Moreover, in combination with a classifier that had developed a nuanced insight into the features governing protein phase behaviour by learning to distinguish between sequences of varying LLPS propensity, the sensitivity was supplemented with high specificity (approximated ROC-AUC of 0.85). These results provide a platform rooted in molecular principles for understanding protein phase behaviour. The predictor is accessible from https://deephase.ch.cam.ac.uk/. Significance StatementThe tendency of many cellular proteins to form protein-rich biomolecular condensates underlies the formation of subcellular compartments and has been linked to various physiological functions. Understanding the molecular basis of this fundamental process and predicting protein phase behaviour have therefore become important objectives. To develop a global understanding of how protein sequence determines its phase behaviour, here, we constructed bespoke datasets of proteins of varying phase separation propensity and identified explicit biophysical and sequence-specific features common to phase separating proteins. Moreover, by combining this insight with neural network based sequence embeddings, we trained machine learning classifiers that identified phase separating sequences with high accuracy, including from independent external test data. The predictor is available from https://deephase.ch.cam.ac.uk/.

biophysics↗

bioRxiv2019

Temporal Gut Microbial Changes Predict Recurrent Clostridium difficile in Patients with and without Ulcerative Colitis

BackgroundUlcerative colitis (UC) carries an increased risk of primary and recurrent Clostridium difficile infection (rCDI) and CDI is associated with UC flares. We hypothesized that specific fecal microbial changes associate with UC flare and rCDI.\n\nMethodsWe conducted a prospective observational cohort study of 57 patients with UC and CDI, CDI only, and UC flare only. Stool samples were collected at baseline, at the end of antibiotic therapy, and after reconstitution for 16S rRNA sequencing. The primary outcomes were recurrent UC flare and rCDI. Logistic regression and Lasso models were constructed for analysis.\n\nResultsThere were 21 (45.7%) patients with rCDI, while 11 (34.4%) developed UC flare. Patients with rCDI demonstrated significant inter-individual (P=.008) and intra-individual differences (P=.004 relative to baseline samples) in community structure by Jensen-Shannon distance (JSD) compared with non-rCDI. Two cross-validated models identified by Lasso regression predicted risk of rCDI: a baseline model with female gender, hospitalization for UC in the past year, increased Ruminococcaceae and Verrucomicrobia, and decreased Eubacteriaceae, Enterobacteriaceae, Lachnospiraceae, and Veillonellaceae (AuROC=0.94); and a model 14 days after completion of antibiotics with female gender, increased Shannon diversity, Ruminococcaceae and Enterobacteriaceae, and decreased community richness and Faecalibacterium (AuROC=0.9). Adding JSD between baseline and post-treatment samples to the latter model improved fit (AuROC=0.94). A baseline model including UC hospitalization in the past year and increased Bacteroidetes showed good fit characteristics for predicting increased risk of UC flare (AuROC=0.88).\n\nConclusionFecal microbial features at baseline and following therapy predict rCDI risk in patients with and without UC. These results may help risk stratify patients to guide management.

microbiology↗

bioRxiv2017

Intracellular production of hydrogels and synthetic RNA granules by multivalent enhancers

Non-membrane bound, hydrogel-like entities, such as RNA granules, nucleate essential cellular functions through their unique physico-chemical properties. However, these intracellular hydrogels have not been as extensively studied as their extracellular counterparts, primarily due to technical challenges in probing these materials in situ. Here, by taking advantage of a chemically inducible dimerization paradigm, we developed iPOLYMER, a strategy for rapid induction of protein-based hydrogels inside living cells. A series of biochemical and biophysical characterizations, in conjunction with computational modeling, revealed that the polymer network formed in the cytosol resembles a physiological hydrogel-like entity that behaves as a size-dependent molecular sieve. We studied several properties of the gel and functionalized it with RNA binding motifs that sequester polyadenine-containing nucleotides to synthetically mimic RNA granules. Therefore, we here demonstrate that iPOLYMER presents a unique and powerful approach to synthetically reconstitute hydrogel-like structures including RNA granules in intact cells.

synthetic biology↗