Biology subjects

Cho, K.

Publications and source records attributed to Cho, K..

At least 37 records · Page 2Linked to original sources

bioRxiv2025

Induction of TFEB promotes Kupffer cell survival and reduces lipid accumulation and inflammation in MASLD

Kupffer cells (KCs) are the tissue-resident macrophage of the liver where they serve a critical role in maintaining liver tissue homeostasis and as a filter for circulation. The composition of liver macrophages changes during metabolic dysfunction-associated liver disease (MASLD), with the loss of resident KCs being a hallmark of disease progression. The mechanism(s) and consequences of KC death in metabolic liver disease have yet to be defined. Transcription factor EB (TFEB) is a master regulator of lysosome function and lipid metabolism which has been shown to protect macrophages from lipid stress in atherosclerosis. We hypothesized that TFEB would improve KC fitness in MASLD. To investigate this possibility, we created a transgenic mice in which TFEB was induced specifically in KCs. We found that TFEB induction protected KCs from cell death in two mouse models of MASLD. KC preservation through TFEB induction reduced liver steatosis via a mechanism that was dependent on macrophage lysosomal lipolysis and mitochondrial fatty acid oxidation. The protection from cell death in TFEB KCs was the result of reduced oxidative stress and ferroptosis through a mechanism that involved enhanced NADPH levels. Together, we provide evidence that TFEB promotes KC fitness during MASLD and orchestrates beneficial effects on liver pathology, thus providing potential targets to develop cell-specific therapeutics.

immunology↗

bioRxiv2025

U2AF1 mutations rescue deleterious exon skipping induced by KRAS mutations

The mechanisms by which somatic mutations of splicing factors, such as U2AF1S34F in lung adenocarcinoma, contribute to cancer pathogenesis are not well understood. Here, we used prime editing to modify the endogenous U2AF1 gene in lung adenocarcinoma cells and assessed the resulting impact on alternative splicing. These analyses identified KRAS as a key target modulated by U2AF1S34F. One specific KRAS mutation, G12S, generates a cryptic U2AF1 binding site that leads to skipping of KRAS exon 2 and generation of a non-functional KRAS transcript. Expression of the U2AF1S34F mutant reverts this exon skipping and restores KRAS function. Analysis of cancer genomes reveals that U2AF1S34F mutations are enriched in KRASG12S-mutant lung adenocarcinomas. A comprehensive analysis of splicing factor/oncogene mutation co-occurrence in cancer genomes also revealed significant co-enrichment of KRASQ61R and U2AF1I24T mutations. Experimentally, KRASQ61R mutation leads to KRAS exon 3 skipping, which in turn can be rescued by the expression of U2AF1I24T. Our findings provide evidence that splicing factor mutations can rescue splicing defects caused by oncogenic mutations. More broadly, they demonstrate a dynamic process of cascading selection where mutational events are positively selected in cancer genomes as a consequence of earlier mutations.

cancer biology↗

bioRxiv2025

Precise, Specific, and Sensitive De Novo Antibody Design Across Multiple Cases

The precision design of antibodies, which naturally recognize diverse molecules through six variable loops, remains a critical challenge in therapeutic molecule discovery. In this study, we demonstrate that precise, sensitive, and specific antibody design can be achieved without prior antibody information across eight distinct target proteins. For each target, binders were identified from a yeast display scFv library of approximately 106 sequences, constructed by combining 102 designed light chain sequences with 104 designed heavy chain sequences. Binders with varying binding strengths were identified for all eight targets, including a case where no experimentally resolved target protein structure was available, demonstrating the highest level of precision compared to previous de novo antibody design reports. To further validate the designed antibodies, they were characterized in the IgG format for five distinct targets. The antibodies exhibited favorable developability properties, positioning them as promising hit or lead candidates. Notably, for one target in particular, the IgG-formatted antibodies exhibited affinity, activity, and developability, comparable to a commercial antibody, highlighting the sensitivity of the design method. Furthermore, binders capable of distinguishing closely related protein subtypes or mutants were identified, demonstrating that the method can achieve high molecular specificity. Cryo-EM analysis experimentally validated the designs accuracy by confirming that the key binding interactions were precisely as designed. These findings underscore the effectiveness of precision molecular design based on atomic-accuracy structure prediction. This study establishes computational antibody design as a viable approach for generating therapeutic molecules with tailored properties, with promising potential for achieving the efficacy and safety required for successful therapeutics.

bioengineering↗

bioRxiv2025

Pyrimethamine and a potent analogue WCDD115 inhibit NRF2 by suppressing DHFR and one-carbon metabolism

Nuclear factor erythroid 2-related factor 2 (NFE2L2/NRF2) is a critical mediator of the cellular oxidative stress response. Aberrant activation of NRF2 is common in lung and upper aerodigestive cancers, where it promotes tumor initiation and progression and confers resistance to chemotherapy, radiation therapy, and immune checkpoint inhibitors. As such, NRF2 therapeutic inhibitors are actively being sought. We previously reported that the antiparasitic drug Pyrimethamine (PYR) inhibits NRF2 in cell lines and in a NRF2-inducible genetically engineered mouse model. Here we design, synthesize, and define structure-activity relationships across a series of 25 PYR-based derivatives to reveal WCDD115 as a 22-fold more potent inhibitor of NRF2 (57nM versus 1.2{micro}M). PYR is known to inhibit plasmodial and human dihydrofolate reductase (DHFR). We found that WCDD115 inhibits hDHFR with 31-fold greater potency than PYR (144nM versus 4.49{micro}M). Metabolomics showed strong similarities between PYR, WCDD115 and methotrexate. Genetic, pharmacological and metabolic epistasis studies reveal that DHFR inactivation is required for NRF2 suppression by WCDD115 and PYR. Global and targeted proteomics revealed overlapping profiles for WCDD115, PYR and methotrexate, including suppression of NRF2 oxidative stress response and activation of TP53 and the DNA damage response. Therefore, PYR and a novel potent derivative WCDD115 are effective, indirect inhibitors of NRF2 and its antioxidant functions. These data underscore the importance of one- carbon metabolism for the NRF2 signaling pathway and support a new therapeutic strategy to suppress NRF2-driven cancer biology.

cancer biology↗

bioRxiv2025

DyAb: sequence-based antibody design and property prediction in a low-data regime

Protein therapeutic design and property prediction are frequently hampered by data scarcity. Here we propose a new model, DyAb, that addresses these issues by leveraging a pair-wise representation to predict differences in protein properties, rather than absolute values. DyAb is built on top of a pre-trained protein language model and achieves a Spearman rank correlation of up to 0.85 on binding affinity prediction across molecules targeting three different antigens (EGFR, IL-6, and an internal target), given as few as 100 training data. We employ DyAb in two design contexts: as a ranking model to score combinations of known mutations, and combined with a genetic algorithm to generate new sequences. Our method consistently generates novel antibody candidates with high binding rates, including designs that improve on the binding affinity of the lead molecule by more than ten-fold. DyAb represents a powerful tool for engineering therapeutic protein properties in low data regimes common in early-stage drug development.

bioengineering↗

bioRxiv2024

Generating All-Atom Protein Structure from Sequence-Only Training Data

AbstractO_ST_ABSPurposeC_ST_ABSDesigning proteins with atomic-level functional control remains a central challenge in de novo design, exacerbated by limitations in structural data availability. MethodsWe introduce PLAID (Protein Latent Induced Diffusion), a generative model that efficiently co-generates discrete sequence and all-atom structure by sampling directly in the shared sequence-structure latent of a pretrained sequence-to-structure predictor. Unlike existing de novo generative models, PLAID trains its diffusion model exclusively on sequences, expanding effective training corpora by 2-4 orders of magnitude relative to structural databases. Using classifier-free guidance, PLAID supports controllable generation based on function (Gene Ontology) and organism keywords. ResultsIn silico, PLAID can unconditionally generate all-atom structures without explicit structural supervision during diffusion model training. Function conditioned proteins can recapitulates catalytic side-chain positions for residues at non-adjacent positions, and transmembrane proteins with expected hydrophobicity patterns and predicted topologies. We further experimentally validate that PLAID can be prompted to generate heme binding proteins with high sequence novelty. ConclusionOverall, PLAID unifies sequence-scale training with atomic-level generation, enabling more precise functional control in protein design. Code and weights: github.com/amyxlu/plaid.

bioinformatics↗

bioRxiv2024

Characterization of tumor heterogeneity through segmentation-free representation learning

The interaction between tumors and their microenvironment is complex and heterogeneous. Recent developments in high-dimensional multiplexed imaging have revealed the spatial organization of tumor tissues at the molecular level. However, the discovery and thorough characterization of the tumor microenvironment (TME) remains challenging due to the scale and complexity of the images. Here, we propose a self-supervised representation learning framework, CANVAS, that enables discovery of novel types of TMEs. CANVAS is a vision transformer that directly takes high-dimensional multiplexed images and is trained using self-supervised masked image modeling. In contrast to traditional spatial analysis approaches which rely on cell segmentations, CANVAS is segmentation-free, utilizes pixel-level information, and retains local morphology and biomarker distribution information. This approach allows the model to distinguish subtle morphological differences, leading to precise separation and characterization of distinct TME signatures. We applied CANVAS to a lung tumor dataset and identified and validated a monocytic signature that is associated with poor prognosis.

cancer biology↗

bioRxiv2024

Ketone Body Metabolism is Not Required for Improvement of Heart Failure by Ketogenic Diet in Mice

Failing hearts increasingly metabolize ketone bodies, and enhancing ketosis improves heart failure (HF) remodeling. Circulating ketones are elevated by fasting/starvation, which is mimicked with a high-fat, low-carbohydrate "ketogenic diet" (KD). While speculated that KD improves HF through increased ketone oxidation, some evidence suggests KD paradoxically downregulates cardiac ketone oxidation despite increased ketone delivery. We sought to clarify the significance of cardiac ketone metabolism during KD in HF. Mice were subjected to transverse aortic constriction with apical myocardial infarction (TAC-MI) and fed either low-fat (LF) control or KD. Cardiac-specific mitochondrial pyruvate carrier 2 (csMPC2-/-) mice were used as a second model of heart failure. In both mice, feeding a KD improved HF, determined by echocardiography, heart weights, and gene expression analyses. Although KD increases plasma ketone bodies, gene expression for ketone metabolic genes is decreased in the hearts of KD-fed mice. Cardiac-specific {beta}-hydroxybutyrate dehydrogenase 1 (csBDH1-/-), the first enzyme in ketone catabolism, mice were also studied and crossed with the csMPC2-/-mice to create double knockout (DKO) mice. These mice were aged to 16 weeks and switched to LF or KD, and KD was able to completely normalize the hearts of both csMPC2-/- and DKO mice, suggesting that ketone metabolism is unnecessary for improving heart failure with ketogenic diet. These studies were then repeated, and mice injected with U-13C-{beta}-hydroxybutyrate to evaluate ketone metabolism. KD feeding significantly decreased the enrichment of the TCA cycle from ketone body carbons, as did the BDH1-deletion in DKO mice. Gene expression and respirometry suggests that KD instead increases cardiac fat oxidation. In conclusion, these results suggest that ketogenic diet decreases cardiac ketone metabolism and does not require ketone metabolism to improve heart failure.

physiology↗

bioRxiv2024

Tokenized and Continuous Embedding Compressions of Protein Sequence and Structure

Existing protein machine learning representations typically model either the sequence or structure distribution, with the other modality implicit. The latent space of sequence-to-structure prediction models such as ESMFold represents the joint distribution of sequence and structure; however, we find these embeddings to exhibit massive activations, whereby some channels have values 3000x higher than others, regardless of the input. Further, on continuous compression schemes, ESMFold embeddings can be reduced by a factor of 128x along the channel and 8x along the length, while retaining structure information at <2[A] scale accuracy, and performing competitively on protein function and localization benchmarks. On discrete compression schemes, we construct a tokenized all-atom structure vocabulary that retains high reconstruction accuracy, thus introducing a tokenized representation of all-atom structure that can be obtained from sequence alone. We term this series of embeddings as CHEAP (Compressed Hourglass Embedding Adaptations of Proteins) embeddings, obtained via the HPCT (Hourglass Protein Compression Transformer) architecture. CHEAP is a compact representation of both protein structure and sequence, sheds light on information content asymmetries between sequence and structure, democratizes representations captured by large models, and is designed to have flexible downstream applications such as generation, search, and prediction.

bioinformatics↗

bioRxiv2024

Phosphoglycerate mutase regulates Treg differentiation through control of serine synthesis and one-carbon metabolism

The differentiation and suppressive functions of regulatory CD4 T cells (Tregs) are supported by a broad array of metabolic changes, providing potential therapeutic targets for immune modulation. In this study, we focused on the regulatory role of glycolytic enzymes in Tregs and identified phosphoglycerate mutase (PGAM) as being differentially overexpressed in Tregs and associated with a highly suppressive phenotype. Pharmacologic or genetic inhibition of PGAM reduced Treg differentiation and suppressive function while reciprocally inducing markers of a pro-inflammatory, T helper 17 (Th17)-like state. The regulatory role of PGAM was dependent on the contribution of 3-phosphoglycerate (3PG), the PGAM substrate, to de novo serine synthesis. Blocking de novo serine synthesis from 3PG reversed the effect of PGAM inhibition on Treg polarization, while exogenous serine directly inhibited Treg polarization. Additionally, altering serine levels in vivo with a serine/glycine-free diet increased peripheral Tregs and attenuated autoimmunity in a murine model of multiple sclerosis. Mechanistically, we found that serine limits Treg polarization by contributing to one-carbon metabolism and methylation of Treg-associated genes. Inhibiting one-carbon metabolism increased Treg polarization and suppressive function both in vitro and in vivo in a murine model of autoimmune colitis. Our study identifies a novel physiologic role for PGAM and highlights the metabolic interconnectivity between glycolysis, serine synthesis, one-carbon metabolism, and epigenetic regulation of Treg differentiation and suppressive function.

immunology↗

bioRxiv2024

Loss of mitochondrial pyruvate transport initiates cardiac glycogen accumulation and heart failure

Heart failure involves metabolic alterations including increased glycolysis despite unchanged or decreased glucose oxidation. The mitochondrial pyruvate carrier (MPC) regulates pyruvate entry into the mitochondrial matrix, and cardiac deletion of the MPC in mice causes heart failure. How MPC deletion results in heart failure is unknown. We performed targeted metabolomics and isotope tracing in wildtype (fl/fl) and cardiac-specific Mpc2-/- (CS-Mpc2-/-) hearts after in vivo injection of U-13C-glucose. Failing CS-Mpc2-/- hearts contained normal levels of ATP and phosphocreatine, yet these hearts displayed increased enrichment from U-13C-glucose and increased glycolytic metabolite pool sizes. 13C enrichment and pool size was also increased for the glycogen intermediate UDP-glucose, as well as increased enrichment of the glycogen pool. Glycogen levels were increased [~]6-fold in the failing CS-Mpc2-/- hearts, and glycogen granules were easily detected by electron microscopy. In young, non-failing CS-Mpc2-/- hearts, increased glycolytic 13C enrichment occurred, but glycogen levels remained low and unchanged compared to fl/fl hearts. Inhibiting glycogen synthase with MZ-101 reduced cardiac glycogen levels and improved heart failure. Feeding a ketogenic diet to CS-Mpc2-/- mice reversed the heart failure and normalized the cardiac glycogen and glycolytic metabolite accumulation. Cardiac glycogen levels were also elevated in mice infused with angiotensin-II, and both the cardiac hypertrophy and glycogen levels were improved by ketogenic diet. Thus, loss of MPC in the heart causes glycogen accumulation and heart failure, while inhibition of glycogen synthesis or a ketogenic diet can reverse both the glycogen accumulation and heart failure.

physiology↗

bioRxiv2024

Novel miRNA-inducing drugs enable differentiation of retinoic acid-resistant neuroblastoma cells

Tumor cell heterogeneity in neuroblastoma, a pediatric cancer arising from neural crest-derived progenitor cells, poses a significant clinical challenge. In particular, unlike adrenergic (ADRN) neuroblastoma cells, mesenchymal (MES) cells are resistant to chemotherapy and retinoid therapy and thereby significantly contribute to relapses and treatment failures. Previous research suggested that overexpression or activation of miR-124, a neurogenic microRNA with tumor suppressor activity, can induce the differentiation of retinoic acid-resistant neuroblastoma cells. Leveraging our established screen for miRNA-modulatory small molecules, we validated PP121, a dual inhibitor of tyrosine and phosphoinositide kinases, as a robust inducer of miR-124. A combination of PP121 and BDNF-activating bufalin synergistically arrests proliferation, induces differentiation, and maintains the differentiated state of MES SK-N-AS cells for 8 weeks. RNA-seq and deconvolution analyses revealed a collapse of the ADRN core regulatory circuitry (CRC) and the emergence of novel CRCs associated with chromaffin cells and Schwann cell precursors. Using a similar protocol, we differentiated and maintained MES neuroblastoma GI-ME-N and SH-EP cell lines, as well as glioblastoma LN-229 and U-251 cell lines, for over 16 weeks. In conclusion, our novel protocol suggests a promising treatment for therapy-resistant cancers of the nervous system. Moreover, these long-lived, differentiated cells provide valuable models for studying mechanisms underlying differentiation, maturation, and senescence.

cancer biology↗

bioRxiv2024

Accelerating Genome- and Phenome-Wide Association Studies using GPUs - A case study using data from the Million Veteran Program

The expansion of biobanks has significantly propelled genomic discoveries yet the sheer scale of data within these repositories poses formidable computational hurdles, particularly in handling extensive matrix operations required by prevailing statistical frameworks. In this work, we introduce computational optimizations to the SAIGE (Scalable and Accurate Implementation of Generalized Mixed Model) algorithm, notably employing a GPU-based distributed computing approach to tackle these challenges. We applied these optimizations to conduct a large-scale genome-wide association study (GWAS) across 2,068 phenotypes derived from electronic health records of 635,969 diverse participants from the Veterans Affairs (VA) Million Veteran Program (MVP). Our strategies enabled scaling up the analysis to over 6,000 nodes on the Department of Energy (DOE) Oak Ridge Leadership Computing Facility (OLCF) Summit High-Performance Computer (HPC), resulting in a 20-fold acceleration compared to the baseline model. We also provide a Docker container with our optimizations that was successfully used on multiple cloud infrastructures on UK Biobank and All of Us datasets where we showed significant time and cost benefits over the baseline SAIGE model.

genetics↗

bioRxiv2024

Cramming Protein Language Model Training in 24 GPU Hours

Protein language models (pLMs) are ubiquitous across biological machine learning research, but state-of-the-art models like ESM2 take hundreds of thousands of GPU hours to pre-train on the vast protein universe. Resource requirements for scaling up pLMs prevent fundamental investigations into how optimal modeling choices might differ from those used in natural language. Here, we define a "cramming" challenge for pLMs and train performant models in 24 hours on a single GPU. By re-examining many aspects of pLM training, we are able to train a 67 million parameter model in a single day that achieves comparable performance on downstream protein fitness landscape inference tasks to ESM-3B, a model trained for over 15, 000x more GPU hours than ours. We open source our library1 for training and inference, LBSTER: Language models for Biological Sequence Transformation and Evolutionary Representation.

bioengineering↗

bioRxiv2024

The inflammatory microenvironment of the lung at the time of infection governs innate control of SARS-CoV-2 replication

SARS-CoV-2 infection leads to vastly divergent clinical outcomes ranging from asymptomatic infection to fatal disease. Co-morbidities, sex, age, host genetics and vaccine status are known to affect disease severity. Yet, how the inflammatory milieu of the lung at the time of SARS-CoV-2 exposure impacts the control of viral replication remains poorly understood. We demonstrate here that immune events in the mouse lung closely preceding SARS-CoV-2 infection significantly impact viral control and we identify key innate immune pathways required to limit viral replication. A diverse set of pulmonary inflammatory stimuli, including resolved antecedent respiratory infections with S. aureus or influenza, ongoing pulmonary M. tuberculosis infection, ovalbumin/alum-induced asthma or airway administration of defined TLR ligands and recombinant cytokines, all establish an antiviral state in the lung that restricts SARS-CoV-2 replication upon infection. In addition to antiviral type I interferons, the broadly inducible inflammatory cytokines TNF and IL-1 precondition the lung for enhanced viral control. Collectively, our work shows that SARS-CoV-2 may benefit from an immunologically quiescent lung microenvironment and suggests that heterogeneity in pulmonary inflammation that precedes or accompanies SARS-CoV-2 exposure may be a significant factor contributing to the population-wide variability in COVID-19 disease outcomes.

immunology↗

bioRxiv2024

Dihydroceramide desaturase governs endoplasmic reticulum and lipid droplet homeostasis to promote glial function in the nervous system

Dihydroceramide desaturases convert dihydroceramides to ceramides, the precursors of all complex sphingolipids. Reduction of DEGS1 dihydroceramide desaturase function causes pediatric neurodegenerative disorder hypomyelinating leukodystrophy-18 (HLD-18). We discovered that infertile crescent (ifc), the Drosophila DEGS1 homolog, is expressed primarily in glial cells to promote CNS development by guarding against neurodegeneration. Loss of ifc causes massive dihydroceramide accumulation and severe morphological defects in cortex glia, including endoplasmic reticulum (ER) expansion, failure of neuronal ensheathment, and lipid droplet depletion. RNAi knockdown of the upstream ceramide synthase schlank in glia of ifc mutants rescues ER expansion, suggesting dihydroceramide accumulation in the ER drives this phenotype. RNAi knockdown of ifc in glia but not neurons drives neuronal cell death, suggesting that ifc function in glia promotes neuronal survival. Our work identifies glia as the primary site of disease progression in HLD-18 and may inform on juvenile forms of ALS, which also feature elevated dihydroceramide levels.

developmental biology↗

bioRxiv2023

Learning sequence, structure, and function representations of proteins with language models

The sequence-structure-function relationships that ultimately generate the diversity of extant observed proteins is complex, as proteins bridge the gap between multiple informational and physical scales involved in nearly all cellular processes. One limitation of existing protein annotation databases such as UniProt is that less than 1% of proteins have experimentally verified functions, and computational methods are needed to fill in the missing information. Here, we demonstrate that a multi-aspect framework based on protein language models can learn sequence-structure-function representations of amino acid sequences, and can provide the foundation for sensitive sequence-structure-function aware protein sequence search and annotation. Based on this model, we introduce a multi-aspect information retrieval system for proteins, Protein-Vec, covering sequence, structure, and function aspects, that enables computational protein annotation and function prediction at tree-of-life scales.

bioinformatics↗

bioRxiv2023

Circadian regulation of MGMT expression and promoter methylation underlies daily rhythms in TMZ sensitivity in glioblastoma

BackgroundGlioblastoma (GBM) is the most common primary brain tumor in adults. Despite extensive research and clinical trials, median survival post-treatment remains at 15 months. Thus, all opportunities to optimize current treatments and improve patient outcomes should be considered. A recent retrospective clinical study found that taking TMZ in the morning compared to the evening was associated with a 6-month increase in median survival in patients with MGMT-methylated GBM. Here, we hypothesized that TMZ efficacy depends on time-of-day and O6-Methylguanine-DNA Methyltransferase (MGMT) activity in murine and human models of GBM. Methods and ResultsIn vitro recordings using real-time bioluminescence reporters revealed that GBM cells have intrinsic circadian rhythms in the expression of the core circadian clock genes Bmal1 and Per2, as well as in the DNA repair enzyme, MGMT. Independent measures of MGMT transcript levels and promoter methylation also showed daily rhythms intrinsic to GBM cells. These cells were more susceptible to TMZ when delivered at the daily peak of Bmal1 transcription. We found that in vivo morning administration of TMZ also decreased tumor size and increased body weight compared to evening drug delivery in mice bearing GBM xenografts. Finally, inhibition of MGMT activity with O6-Benzylguanine abrogated the daily rhythm in sensitivity to TMZ in vitro by increasing sensitivity at both the peak and trough of Bmal1 expression. ConclusionWe conclude that chemotherapy with TMZ can be dramatically enhanced by delivering at the daily maximum of tumor Bmal1 expression and minimum of MGMT activity. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=164 SRC="FIGDIR/small/557630v1_ufig1.gif" ALT="Figure 1"> View larger version (36K): org.highwire.dtl.DTLVardef@2e9058org.highwire.dtl.DTLVardef@1069917org.highwire.dtl.DTLVardef@2cf000org.highwire.dtl.DTLVardef@40abc9_HPS_FORMAT_FIGEXP M_FIG C_FIG

cancer biology↗