Biology subjects

Lim, H.

Publications and source records attributed to Lim, H..

36 records · Page 2Linked to original sources

bioRxiv2022

Transcriptomics reveals amygdala neuron regulation by fasting and ghrelin thereby promoting feeding

The central amygdala (CeA) consists of numerous genetically-defined inhibitory neurons that control defensive and appetitive behaviors including feeding. Transcriptomic signatures of cell types and their links to function remain poorly understood. Using single-nucleus RNA-sequencing, we describe nine CeA cell clusters, of which four are mostly associated with appetitive and two with aversive behaviors. To analyze the activation mechanism of appetitive CeA neurons, we characterized Serotonin receptor 2a (Htr2a)-expressing neurons (CeAHtr2a) that comprise three appetitive clusters and were previously shown to promote feeding. In vivo calcium imaging revealed that CeAHtr2a neurons are activated by fasting, the hormone ghrelin, and the presence of food. Moreover, these neurons are required for the orexigenic effects of ghrelin. Appetitive CeA neurons responsive to fasting and ghrelin project to the parabrachial nucleus (PBN) causing inhibition of target PBN neurons. These results illustrate how the transcriptomic diversification of CeA neurons relates to fasting and hormone-regulated feeding behavior. HighlightsTranscriptomics reveals four appetitive, two aversive and three unknown CeA clusters Ghrelin and fasting activate CeAHtr2a neurons comprising three appetitive clusters Ghrelins orexigenic effects require the activity of appetitive CeA neurons CeA[->]PBN inhibition is necessary for ghrelin/fasting induced feeding

neuroscience↗

bioRxiv2022

Next-generation large-scale binary protein interaction network for Drosophila

Generating reference maps of the interactome networks underlying most cellular functions can greatly illuminate genetic studies by providing a protein-centric approach to finding new components of existing pathways, complexes, and processes. Here, we applied state-of-the-art experimental and bioinformatics methods to identify high-confidence binary protein-protein interactions (PPIs) for Drosophila melanogaster. We performed four all-by-all yeast two-hybrid (Y2H) screens of >10,000 Drosophila proteins, resulting in the FlyBi dataset of 8,723 PPIs among 2,939 proteins. As part of this effort, we tested subsets of our data and data from previous PPI datasets using an orthogonal assay, which allowed us to normalize data quality across datasets. Next, we integrated our FlyBi data with previous PPI data, resulting in an expanded, high-confidence binary Drosophila reference interaction network, DroRI, comprising 17,232 interactions among 6,511 proteins. These data are accessible through the Molecular Interaction Search Tool (MIST) and other databases. To assess the utility of the PPI resource, we used novel interactions from the FlyBi dataset to generate an autophagy interaction network that we validated in vivo using two different autophagy-related assays. We found that deformed wings (dwg) encodes a protein that is both a regulator and a target of autophagy. Altogether, the resources generated in this project provide a strong foundation for building high-confidence new hypotheses regarding protein networks and function.

cell biology↗

bioRxiv2022

Imaging unlabeled axons in the mouse retina by second harmonic generation

We describe a novel microscopy for studying the network of axons in the unlabeled fresh wholemount retina. The intrinsic radiation of second harmonic generation (SHG) was utilized to visualize single axons of all major retinal neurons, i.e., photoreceptors, horizontal cells, bipolar cells, amacrine cells, and the retinal ganglion cells. The cell types of SHG+ axons were determined using transgenic GFP/YFP mice. New findings were obtained with retinal SHG imaging: Muller cells do not maintain uniformly polarized microtubules in the processes; SHG+ axons of bipolar cells terminate in the inner plexiform layer (IPL) in a subtype-specific manner; a subset of amacrine cells, presumably the axon-bearing types, emits SHG; and the axon-like neurites of amacrine cells provide a cytoskeletal scaffolding for the IPL stratification. To demonstrate the utility, retinal SHG imaging was applied for testing whether the inner retina is preserved in glaucoma, using DBA/2 mice as a model of glaucoma and DBA/2-Gpnmb+ as the non-glaucomatous control. It was found that the morphology of the inner retina was largely intact in glaucoma and the pre-synaptic compartments to the retinal ganglion cells were uncompromised. It proves retinal SHG imaging as a promising technology for studying the physiological and diseased retina in 3D.

neuroscience↗

bioRxiv2020

Tracking pre-mRNA maturation across subcellular compartments identifies developmental gene regulation through intron retention and nuclear anchoring.

To globally assess the distribution and processing of gene transcripts across subcellular compartments, we developed extensive RNA-seq datasets of both polyA+ and total RNA from chromatin, nucleoplasm and cytoplasm of mouse ESC, neuronal progenitors, and neurons. We identified protein-coding genes whose polyadenylated transcripts were more abundant in chromatin than cytoplasm. We defined introns exhibiting cotranscriptional splicing, complete intron retention in cytoplasmic RNA, and many introns retained in nucleoplasmic and chromatin RNA but absent from cytoplasmic RNA, including new introns controlled during neuronal development. In particular, we found that polyadenylated Gabbr1 transcripts are expressed in mESC but remain sequestered on chromatin until neuronal differentiation when they are processed and released to the cytoplasm. This developmental regulation of splicing and chromatin association demonstrates that the abundance of polyadenylated RNA is not always an indicator of functional gene expression. Our datasets provide a rich resource for analyzing many other aspects of mRNA maturation.

molecular biology↗

bioRxiv2020

A deep learning framework for elucidating whole-genome chemical interaction space

Molecular interaction is the foundation of biological process. Elucidation of genome-wide binding partners of a biomolecule will address many questions in biomedicine. However, ligands of a vast number of proteins remain elusive. Existing methods mostly fail when the protein of interest is dissimilar from those with known functions or structures. We develop a new deep learning framework DISAE that incorporates biological knowledge into self-supervised learning techniques for predicting ligands of novel unannotated proteins on a genome-scale. In the rigorous benchmark studies, DISAE outperforms state-of-the-art methods by a significant margin. The interpretability analysis of DISAE suggests that it learns biologically meaningful information. We further use DISAE to assign ligands to human orphan G-Protein Coupled Receptors (GPCRs) and to cluster the human GPCRome by integrating their phylogenetic and ligand relationships. The promising results of DISAE open an avenue for exploring the chemical landscape of entire sequenced genomes.

bioinformatics↗

bioRxiv2020

Exploration of Chemical Space with Partial Labeled Noisy Student Self-Training for Improving Deep Learning: Application to Drug Metabolism

MotivationDrug discovery is time-consuming and costly. Machine learning, especially deep learning, shows a great potential in accelerating the drug discovery process and reducing its cost. A big challenge in developing robust and generalizable deep learning models for drug design is the lack of a large amount of data with high quality and balanced labels. To address this challenge, we developed a self-training method PLANS that exploits millions of unlabeled chemical compounds as well as partially labeled pharmacological data to improve the performance of neural network models. ResultWe evaluated the self-training with PLANS for Cytochrome P450 binding activity prediction task, and proved that our method could significantly improve the performance of the neural network model with a large margin. Compared with the baseline deep neural network model, the PLANS-trained neural network model improved accuracy, precision, recall, and F1 score by 13.4%, 12.5%, 8.3%, and 10.3%, respectively. The self-training with PLANS is model agnostic, and can be applied to any deep learning architectures. Thus, PLANS provides a general solution to utilize unlabeled and partially labeled data to improve the predictive modeling for drug discovery. AvailabilityThe code that implements PLANS is available at https://github.com/XieResearchGroup/PLANS

bioinformatics↗

bioRxiv2020

Genome-wide Prediction of Small Molecule Binding to Remote Orphan Proteins Using Distilled Sequence Alignment Embedding

Endogenous or surrogate ligands of a vast number of proteins remain unknown. Identification of small molecules that bind to these orphan proteins will not only shed new light into their biological functions but also provide new opportunities for drug discovery. Deep learning plays an increasing role in the prediction of chemical-protein interactions, but it faces several challenges in protein deorphanization. Bioassay data are highly biased to certain proteins, making it difficult to train a generalizable machine learning model for the proteins that are dissimilar from the ones in the training data set. Pre-training offers a general solution to improving the model generalization, but needs incorporation of domain knowledge and customization of task-specific supervised learning. To address these challenges, we develop a novel protein pre-training method, DIstilled Sequence Alignment Embedding (DISAE), and a module-based fine-tuning strategy for the protein deorphanization. In the benchmark studies, DISAE significantly improves the generalizability and outperforms the state-of-the-art methods with a large margin. The interpretability analysis of pre-trained model suggests that it learns biologically meaningful information. We further use DISAE to assign ligands to 649 human orphan G-Protein Coupled Receptors (GPCRs) and to cluster the human GPCRome by integrating their phylogenetic and ligand relationships. The promising results of DISAE open an avenue for exploring the chemical landscape of entire sequenced genomes.

systems biology↗

bioRxiv2020

PtrR (YneJ) is a novel E. coli transcription factor regulating the putrescine stress response and glutamate utilization

Although polyamines, such as putrescine (Ptr), induce envelope stress for bacteria, they are important as nitrogen and carbon sources. Ptr utilization in Escherichia coli involves protein glutamylation, and glutamate stands at a crossroads between catabolism and anabolism. This communication reports that the transcription factor YneJ, here renamed PtrR, is involved in the regulation of a small regulatory RNA gene, fnrS, and an operon, yneIHGF, encoding succinate-semialdehyde dehydrogenase, Sad (YneI), glutaminase, GlsB (YneH), and several other genes. The yneI promoter is activated during putrescine utilization under nitrogen/carbon starvation conditions, and we show that PtrR is important for the putrescine stress response. It is also a repressor of fnrS gene expression, involved in the cascade regulation of mRNA synthesis for the marA and sodB genes, involved in antibiotic responses. PtrR transcriptional regulation of fnrS leads to a regulatory cascade induced by this small RNA that affects mRNA levels of ompF and the multidrug resistance regulator, MarA. We propose that PtrR functions as a dual activator/repressor, and that its regulation is important for the responses to different stress conditions involving L-glutamine/L-glutamate and putrescine utilization. IMPORTANCEPutrescine is an important source of nitrogen for many organisms, but it also induces stress. Although its metabolism has been studied extensively, the regulatory mechanisms that control the stress response are still poorly understood. This study reveals that the HTH-type transcriptional regulator, YneJ in Escherichia coli, here re-named PtrR, is important for the putrescine stress response, in part because it plays a role in outer membrane porin regulation as a sensor in a regulatory cascade. Direct PtrR transcriptional regulation of the fnrS, yneI (sad), gltS and ptrR genes is documented and rationalized, and nine PtrR binding sites were identified using ChIP-Exo. A ptrR mutant exhibited altered resistance to a tetracycline group of antibiotics under microaerophilic conditions, suggesting that PtrR indirectly controls expression of porin genes such as ompF.

microbiology↗

bioRxiv2020

Analysis of circulating-microRNA expression in lactating Holstein cows under summer heat stress

Korean peninsular weather is rapidly becoming subtropical due to global warming. In summer 2018, South Korea experienced the highest temperatures since the meteorological observations recorded in 1907. Heat stress has a negative effect on Holstein cows, the most popular breed of dairy cattle in South Korea, which is susceptible to heat. To examine physiological changes in dairy cows under heat stress conditions, we analyzed the profiles circulating microRNAs isolated from whole blood samples collected under heat stress and non-heat stress conditions using small RNA sequencing. We compared the expression profiles in lactating cows under heat stress and non-heat stress conditions to understand the regulation of biological processes in heat-stressed cows. Moreover, we measured several heat stress indicators, such as rectal temperature, milk yield, average daily gain, and progesterone concentration. All these assessments showed that pregnant cows were more susceptible to heat stress than non-pregnant cows. Particularly, progesterone concentrations known to have maternal warming effects were at similar levels in non-pregnant cows but significantly increased in pregnant cows under heat stress conditions. The differentially expressed miRNAs and their putative target genes were analyzed in pregnant cows. Interestingly, we found that differentially expressed miRNAs (bta-miR-146b, bta-miR-20b, bta-miR-29d-3p, bta-miR-1246) specifically targeted progesterone biosynthesis (StAR) and the function of corpus luteum-related genes (CCL11, XCL), suggesting that pregnant cows with elevated progesterone concentrations are more susceptible to heat stress. In addition, we found the differential expression of 11 miRNAs (bta-miR-19a, bta-miR-19b, bta-miR-30a-5p, and several from the bta-miR-2284 family) in both pregnant and non-pregnant cows under heat stress conditions. In target gene prediction and gene set enrichment analysis, these miRNAs were found to be associated with the cytoskeleton, cell junction, vasculogenesis, cell proliferation, ATP synthesis, oxidative stress, and immune responses involved in heat response. These miRNAs can be used as potential biomarkers for heat stress.

genomics↗

bioRxiv2020

Multiplex generation, tracking, and functional screening of substitution mutants using a CRISPR/retron system

We developed a clustered regularly interspaced short palindromic repeats (CRISPR)/retron system for multiplexed generation of substitution mutations by co-utilization of a retron system that continuously expresses donor DNA and a CRISPR/Cas9 cassette that induces cleavage at target genomic loci. Our system efficiently introduces substitution mutation in the Escherichia coli genome in a high-throughput manner. These substitution mutations can be tracked by analysis of retron plasmid sequences without laborious amplification of individual edited loci. We demonstrated that our CRISPR/retron system can introduce thousands of mutations in a single experiment and be used for screening phenotypes related to chemical responses or fitness changes. We expect that our system could facilitate genome-scale substitution screenings.

bioengineering↗

bioRxiv2019

A Bayesian approach to accurate and robust signature detection on LINCS L1000 data

LINCS L1000 dataset produced by L1000 assay contains numerous cellular expression data induced by large sets of perturbagens. Although it provides invaluable resources for drug discovery as well as understanding of disease mechanisms, severe noise in the dataset makes the detection of reliable gene expression signals difficult. Existing methods for the peak deconvolution, either k-means based or Gaussian mixture model, cannot reliably recover the accurate expression level of genes in many cases, thereby limiting their robust applications in biomedical studies. Here, we have developed a novel Bayes theory based deconvolution algorithm that gives unbiased likelihood estimations for peak positions and characterizes the peak with a probability based z-scores. Based on above algorithm, a pipeline is built to process raw data from L1000 assay into signatures that represent the features of perturbagen. The performance of the proposed new pipeline is rigorously evaluated using the similarity between bio-replicates and between drugs with shared targets. The results show that the new signature derived from the proposed algorithm gives a substantially more reliable and informative representation for perturbagens than existing methods. Thus, our new Bayesian-based peak deconvolution and z-score calculation method may significantly boost the performance of invaluable L1000 data in the down-stream applications such as drug repurposing, disease modeling, and gene function prediction.

bioinformatics↗

bioRxiv2019

Electronic Health Records Based Prediction of Future Incidence of Alzheimers Disease Using Machine Learning

Nationwide population-based cohort provides a new opportunity to build a completely automated risk prediction model based on individuals history of health and healthcare beyond existing risk prediction models. We tested the possibility of machine learning models to predict future incidence of Alzheimers disease (AD) using large-scale administrative health data. From the Korean National Health Insurance Service database between 2002 and 2010, we obtained de-identified health data in elders above 65 years (N=40,736) containing 4,894 unique clinical features including ICD-10 codes, medication codes, laboratory values, history of personal and family illness, and socio-demographics. To define incident AD two operational definitions were considered: "definite AD" with diagnostic codes and dementia medication (n=614) and "probable AD" with only diagnosis (n=2,026). We trained and validated a random forest, support vector machine, and logistic regression to predict incident AD in 1,2,3, and 4 subsequent years. For predicting future incidence of AD in balanced samples (bootstrapping), the machine learning models showed reasonable performance in 1-year prediction with AUC of 0.775 and 0.759, based on "definite AD" and "probable AD" outcomes, respectively; in 2-year, 0.730 and 0.693; in 3-year, 0.677 and 0.644; in 4-year, 0.725 and 0.683. The results were similar when the entire (unbalanced) samples were used. Important clinical features selected in logistic regression included hemoglobin level, age, and urine protein level. This study may shed a light on the utility of the data-driven machine learning model based on large-scale administrative health data in AD risk prediction, which may enable better selection of individuals at risk for AD in clinical trials or early detection in clinical settings.

bioinformatics↗

bioRxiv2019

Single-cell analysis of a mutant library generated using CRISPR-guided deaminase

CRISPR-based screening methods using single-cell RNA sequencing (scRNA-seq) technology enable comprehensive profiling of gene perturbations from knock-out mutations. However, evaluating substitution mutations using scRNA-seq is currently limited. We combined CRISPR RNA-guided deaminase and scRNA-seq technology to develop a platform for introducing mutations in multiple genes and assessing the mutation-associated signatures. Using this platform, we generated a library consisting of 420 sgRNAs, performed sgRNA tracking analysis, and assessed the effect size of the response to vemurafenib in the melanoma cell line, which has been well-studied via knockout-based drop-out screens. However, a substitution mutation library screen has not been applied and transcriptional information for mechanisms of action was not assessed. Our platform permits discrimination of several candidate mutations that function differently from other mutations by integrating sgRNA candidates and gene expression readout. We anticipate that our platform will enable high-throughput analyses of the mechanisms related to a variety of biological events.

bioengineering↗

bioRxiv2018

Rational Discovery of Dual-Action Multi-Target Kinase Inhibitor for Precision Anti-Cancer Therapy Using Structural Systems Pharmacology

Although remarkable progresses have been made in the cancer treatment, existing anti-cancer drugs are associated with increasing risk of heart failure, variable drug response, and acquired drug resistance. To address these challenges, for the first time, we develop a novel genome-scale multi-target screening platform 3D-REMAP that integrates data from structural genomics and chemical genomics as well as synthesize methods from structural bioinformatics, biophysics, and machine learning. 3D-REMAP enables us to discover marked drugs for dual-action agents that can both reduce the risk of heart failure and present anti-cancer activity. 3D-REMAP predicts that levosimendan, a drug for heart failure, inhibits serine/threonine-protein kinase RIOK1 and other kinases. Subsequent experiments confirm this prediction, and suggest that levosimendan is active against multiple cancers, notably lymphoma, through the direct inhibition of RIOK1 and RNA processing pathway. We further develop machine learning models to identify cancer cell-lines and patients that may respond to levosimendan. Our findings suggest that levosimendan can be a promising novel lead compound for the development of safe and effective multi-targeted cancer therapy, and demonstrate the potential of genome-wide multi-target screening in designing polypharmacology and drug repurposing for precision medicine.\n\nAuthor SummaryMulti-target drug design (a.k.a targeted polypharmacology) has emerged as a new strategy for discovering novel therapeutics that can enhance therapeutic efficacy and overcome drug resistance in tackling multi-genic diseases such as cancer. However, it is extremely challenging for conventional computational tools that are either receptor-based or ligand-based to screen compounds for selectively targeting multiple receptors. Existing multi-target drug design mainly focuses on compound screening against receptors within the same gene family but not across different gene families. Here, we develop a new computational tool 3D-REMAP that enables us to identify chemical-protein interactions across fold space on a genome scale. The genome-scale chemical-protein interaction network allows us to discover dual-action drugs that can bind to two types of targets simultaneously, one for mitigating side effect and another for enhancing the therapeutic effect. Using 3D-REMAP, we predict and subsequently experiments validate that levosimendan, a drug for heart failure, is active against multiple cancers, notably, lymphoma. This study demonstrates the potential of genome-wide multi-target screening in designing polypharmacology and drug repurposing for precision medicine.

systems biology↗

bioRxiv2018

Risk factors associated with Parkinson’s disease: An 11-year population-based South Korean study

ObjectiveTo validate various known risk factors of Parkinsonism and to establish basic information to formulate public health policy by using a 10-year follow-up cohort model.\n\nMethodsThis population based nation-wide study was performed using the National Health Insurance Database of reimbursement claims of the Health Insurance Review and Assessment Service of South Korea data on regular health check-ups in 2003 and 2004, with 10 years follow-up.\n\nResultsWe identified 7,746 patients with Parkinsonism. Old age, hypertension, diabetes, depression, anxiety, taking statin medication, high body mass index, non-smoking, non-alcohol drinking, and low socioeconomic status were each associated with an increase in the risk of Parkinsonism (fully adjusted Cox proportional hazards model: hazard ratio (HR) 1.259, 95% confidence interval (CI) 1.194-1.328 for hypertension, HR 1.255, 95% CI 1.186-1.329 for diabetes, HR 1.554, 95% CI 1.664-1.965 for depression, HR 1.808, 95% CI 1.462-1.652 for anxiety, and HR 1.157, 95% CI 1.072-1.250 for taking statin medication).\n\nConclusionsIn our study, old age, depression, anxiety, and a non-smoker status were found to be risk factors of Parkinsonism, in agreement with previous studies. However, sex, hypertension, diabetes, taking statin medication, non-drinking of Alcohol, and lower socioeconomic status have not been described as risk factors in previous studies and need further verification in future studies.

epidemiology↗

bioRxiv2017

ANTENNA, a Multi-Rank, Multi-Layered Recommender System for Inferring Reliable Drug-Gene-Disease Associations: Repurposing Diazoxide as a Targeted Anti-Cancer Therapy

Existing1drug discovery process follows a reductionist model of \"one-drug-one-gene-one-disease,\" which is not adequate to tackle complex diseases that involve multiple malfunctioned genes. The availability of big omics data offers new opportunities to transform the drug discovery process into a new paradigm of systems pharmacology that focuses on designing drugs to target molecular interaction networks instead of a single gene. Here, we develop a reliable multi-rank, multi-layered recommender system, ANTENNA, to mine large-scale chemical genomics and disease association data for the prediction of novel drug-gene-disease associations. ANTENNA integrates a novel tri-factorization based dual-regularized weighted and imputed One Class Collaborative Filtering (OCCF) algorithm, tREMAP, with a statistical framework that is based on Random Walk with Restart and can assess the reliability of a specific prediction. In the benchmark study, tREMAP clearly outperforms the single rank OCCF. We apply ANTENNA to a real-world problem: repurposing old drugs for new clinical indications that have yet had an effective treatment. We discover that FDA-approved drug diazoxide can inhibit multiple kinase genes whose malfunction is responsible for many diseases including cancer, and kill triple negative breast cancer (TNBC) cells effectively at a low concentration (IC50 = 0.87 M). The TNBC is a deadly disease that currently does not have effective targeted therapies. Our finding demonstrates the power of big data analytics in drug discovery, and has a great potential toward developing a targeted therapy for the effective treatment of TNBC.

bioinformatics↗

bioRxiv2017

Committed hemopoietic progenitors, not stem cells, are the principal responders to Hox gene transduction

As hemopoietic stem cells differentiate, their proliferative lifespan shortens by unknown mechanisms. Homeobox cluster (Hox) genes have been implicated by their enhancement of self-renewal when transduced into hemopoietic cells, but gene deletions have been inconclusive because of functional redundancy. Here we enforced HOXB4 expression in purified precursor stages, and compared responses of early stages expressing the endogenous genes with later stages that did not. Contrary to the prevalent view that transduced Hox genes enhance the self-renewal of hemopoietic stem cells, stem cells or their multipotent progeny expressing the endogenous genes showed little response. Instead, immortalization, extensive self-renewal and acquired reconstituting potential occurred in committed erythroid and myeloid progenitors where the endogenous genes were shutting down. The results change our understanding of the stages affected by exogenous HOX proteins and point to shutdown of the endogenous genes as a principal determinant of the shortened clonal lifespans of committed progenitor cells.

cell biology↗

bioRxiv2016

Ploidy tug-of-war: evolutionary and genetic environments influence the rate of ploidy drive in a human fungal pathogen

Variation in baseline ploidy is seen throughout the tree of life, yet the factors that determine why one ploidy level is selected over another remain poorly understood. Experimental evolution studies using asexual fungal microbes with manipulated ploidy levels intriguingly reveals a propensity to return to the historical baseline ploidy, a phenomenon that we term ploidy drive. We evolved haploid, diploid, and polyploid strains of the human fungal pathogen Candida albicans under three different nutrient limitation environments to test whether these conditions, hypothesized to select for low ploidy levels, could counteract ploidy drive. Strains generally maintained or acquired smaller genome sizes in minimal medium and under phosphorus depletion compared to in a complete medium, while mostly maintained or acquired increased genome sizes under nitrogen depletion. Surprisingly, improvements in fitness often ran counter to changes in total nuclear genome size; in a number of scenarios lines that maintained their original genome size often increased in fitness more than lines that converged towards diploidy. Combined, this work demonstrates a role for both the environment and genotype in determination of the rate of ploidy drive, and highlights questions that remain about the force(s) that cause genome size variation.

evolutionary biology↗