Biology subjects

Xiong, Z.

Publications and source records attributed to Xiong, Z..

70 records · Page 4Linked to original sources

bioRxiv2021

Construction of a neural network energy function for protein physics

Classical potentials are widely used to describe protein physics, due to their simplicity and accuracy, but they are continuously challenged as real applications become more demanding with time. Deep neural networks could help generating alternative ways of describing protein physics. Here we propose an unsupervised learning method to derive a neural network energy function for proteins. The energy function is a probability density model learned from plenty of 3D local structures which have been extensively explored by evolution. We tested this model on a few applications (assessment of protein structures, protein dynamics and protein sequence design), showing that the neural network can correctly recognize patterns in protein structures. In other words, the neural network learned some aspects of protein physics from experimental data.

biophysics↗

bioRxiv2021

Crowdsourced Identification of Multi-Target Kinase Inhibitors for RET- and TAU- Based Disease: the Multi-Targeting Drug DREAM Challenge

A continuing challenge in modern medicine is the identification of safer and more efficacious drugs. Precision therapeutics, which have one molecular target, have been long promised to be safer and more effective than traditional therapies. This approach has proven to be challenging for multiple reasons including lack of efficacy, rapidly acquired drug resistance, and narrow patient eligibility criteria. An alternative approach is the development of drugs that address the overall disease network by targeting multiple biological targets ( polypharmacology). Rational development of these molecules will require improved methods for predicting single chemical structures that target multiple drug targets. To address this need, we developed the Multi-Targeting Drug DREAM Challenge, in which we challenged participants to predict single chemical entities that target pro-targets but avoid anti-targets for two unrelated diseases: RET-based tumors and a common form of inherited Tauopathy. Here, we report the results of this DREAM Challenge and the development of two neural network-based machine learning approaches that were applied to the challenge of rational polypharmacology. Together, these platforms provide a potentially useful first step towards developing lead therapeutic compounds that address disease complexity through rational polypharmacology. Author SummaryMany modern drugs are developed with the goal of modulating a single cellular pathway or target. However, many drugs are, in fact, dirty; they target multiple cellular pathways or targets. This phenomenon is known as multi-targeting or polypharmacology. While some strive to develop cleaner therapeutics that eliminate secondary targets, recent work has shown that multi-targeting therapeutics have key advantages for a variety of diseases. However, while multi-targeting drugs that affect a precisely-defined profile of targets may be more effective, it is difficult to computationally predict which molecules have desirable target profiles. Here, we report the results of a competitive crowdsourcing project (the Multi-Targeting Drug DREAM Challenge), where we challenged participants to predict chemicals that have desired target profiles for cancer and neurodegenerative disease.

pharmacology and toxicology↗

bioRxiv2021

Muscle-specific Cavin4 interacts with Bin1 to promote T-tubule formation and stability in developing skeletal muscle

The cavin proteins are essential for caveola biogenesis and function. Here, we identify a role for the muscle-specific component, Cavin4, in skeletal muscle T-tubule development by analyzing two vertebrate systems: mouse and zebrafish. In both models Cavin4 localized to T-tubules and loss of Cavin4 resulted in aberrant T-tubule maturation. In zebrafish, which possess duplicated cavin4 paralogs, Cavin4b was shown to directly interact with the T-tubule-associated BAR domain protein, Bin1. Loss of both Cavin4a and Cavin4b caused aberrant accumulation of interconnected caveolae within the T-tubules, a fragmented T-tubule network enriched in Caveolin-3, and an impaired Ca2+ response upon mechanical stimulation. We propose a role for Cavin4 in remodeling the T-tubule membrane early in development by recycling caveolar components from the T-tubule to the sarcolemma. This generates a stable T-tubule domain lacking caveolae that is essential for T-tubule function.

cell biology↗

bioRxiv2020

In vivo proteomic mapping through GFP-directed proximity dependent biotin labelling in zebrafish

Protein interaction networks are crucial for complex cellular processes. However, the elucidation of protein interactions occurring within highly specialised cells and tissues is challenging. Here we describe the development, and application, of a new method for proximity-dependent biotin labelling in whole zebrafish. Using a conditionally stabilised GFP-binding nanobody to target a biotin ligase to GFP-labelled proteins of interest, we show tissue-specific proteomic profiling using existing GFP-tagged transgenic zebrafish lines. We demonstrate the applicability of this approach, termed BLITZ (Biotin Labelling In Tagged Zebrafish), in diverse cell types such as neurons and vascular endothelial cells. We applied this methodology to identify interactors of caveolar coat protein, cavins, in skeletal muscle. Using this system, we defined specific interaction networks within in vivo muscle cells for the closely related but functionally distinct Cavin4 and Cavin1 proteins.

cell biology↗

bioRxiv2020

The role of gene dosage in budding yeast centrosome scaling and spontaneous diploidization

Ploidy is the number of whole sets of chromosomes in a species. Ploidy is typically a stable cellular feature that is critical for survival. Polyploidization is a route recognized to increase gene dosage, improve fitness under stressful conditions and promote evolutionary diversity. However, the mechanism of regulation and maintenance of ploidy is not well characterized. Here, we examine the spontaneous diploidization associated with mutations in components of the Saccharomyces cerevisiae centrosome, known as the spindle pole body (SPB). Although SPB mutants are associated with defects in spindle formation, we show that two copies of the mutant in a haploid yeast favors diploidization in some cases, leading us to speculate that the increased gene dosage in diploids rescues SPB duplication defects, allowing cells to successfully propagate with a stable diploid karyotype. This copy number-based rescue is linked to SPB scaling: certain SPB subcomplexes do not scale or only minimally scale with ploidy. We hypothesize that acquisition of lesions in structures with incompatible allometries such as the centrosome may drive changes such as whole genome duplication, which have shaped the evolutionary landscape of many eukaryotes. Author SummaryPloidy is the number of whole sets of chromosomes in a species. Most eukaryotes alternate between a diploid (two copy) and haploid (one copy) state during their life and sexual cycle. However, as part of normal human development, specific tissues increase their DNA content. This gain of entire sets of chromosomes is known as polyploidization, and it is observed in invertebrates, plants and fungi, as well. Polyploidy is thought to improve fitness under stressful conditions and promote evolutionary diversity, but how ploidy is determined is poorly understood. Here, we use budding yeast to investigate mechanisms underlying the ploidy of wild-type cells and specific mutants that affect the centrosome, a conserved structure involved in chromosome segregation during cell division. Our work suggests that different scaling relationships (allometry) between the genome and cellular structures underlies alterations in ploidy. Furthermore, mutations in cellular structures with incompatible allometric relationships with the genome may drive genomic changes such duplications, which are underly the evolution of many species including both yeasts and humans.

genetics↗

bioRxiv2020

Facing small and biased data dilemma in drug discovery with federated learning

Artificial intelligence (AI) models usually require large amounts of high-quality training data, which is in striking contrast to the situation of small and biased data faced by current drug discovery pipelines. The concept of federated learning has been proposed to utilize distributed data from different sources without leaking sensitive information of these data. This emerging decentralized machine learning paradigm is expected to dramatically improve the success of AI-powered drug discovery. We here simulate the federated learning process with 7 aqueous solubility datasets from different sources, among which there are overlapping molecules with high or low biases in the recorded values. Beyond the benefit of gaining more data, we also demonstrate federated training has a regularization effect making it superior than centralized training on the pooled datasets with high biases. Further, two more cases are studied to test the usability of federated learning in drug discovery. Our work demonstrates the application of federated learning in predicting drug related properties, but also highlights its promising role in addressing the small data and biased data dilemma in drug discovery.

pharmacology and toxicology↗

bioRxiv2020

Automated design and optimization of multitarget schizophrenia drug candidates by deep learning

Complex neuropsychiatric diseases such as schizophrenia require drugs that can target multiple G protein-coupled receptors (GPCRs) to modulate complex neuropsychiatric functions. Here, we report an automated system comprising a deep recurrent neural network (RNN) and a multitask deep neural network (MTDNN) to design and optimize multitargeted antipsychotic drugs. The system successfully generates novel molecule structures with desired multiple target activities, among which high-ranking compound 3 was synthesized, and demonstrated potent activities against dopamine D2, serotonin 5-HT1A and 5-HT2A receptors. Hit expansion based on the MTDNN was performed, 6 analogs of compound 3 were evaluated experimentally, among which compound 8 not only exhibited specific polypharmacology profiles but also showed antipsychotic effect in animal models with low potential for sedation and catalepsy, highlighting their suitability for further preclinical studies. The approach can be an efficient tool for designing lead compounds with multitarget profiles to achieve the desired efficacy in the treatment of complex neuropsychiatric diseases. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=87 SRC="FIGDIR/small/999615v1_ufig1.gif" ALT="Figure 1"> View larger version (27K): org.highwire.dtl.DTLVardef@17046d8org.highwire.dtl.DTLVardef@3b4c2aorg.highwire.dtl.DTLVardef@10590e9org.highwire.dtl.DTLVardef@6cc609_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2020

An Endocytic Capture Model for Skeletal Muscle T-tubule Formation

The skeletal muscle T-tubule is a specialized membrane domain essential for coordinated muscle contraction that shows dysmorphology in a number of genetically inherited muscle diseases. However, in the absence of genetically tractable systems the mechanisms involved in T-tubule formation are unknown. Here, we have used the optically transparent and genetically tractable zebrafish system to probe T-tubule development in vivo. By combining live imaging with three-dimensional electron microscopy we derived a four-dimensional quantitative model for T-tubule formation. To elucidate the mechanisms involved in T-tubule formation in vivo we developed a quantitative screen for proteins that associate with and modulate early T-tubule formation including an overexpression screen of the entire zebrafish Rab protein family. We propose a new endocytic capture model involving i) formation of dynamic endocytic tubules at transient nucleation sites on the sarcolemma ii) stabilization by myofibrils/sarcoplasmic reticulum and iii) delivery of membrane from the recycling endosome and Golgi complex.

cell biology↗

bioRxiv2020

Novel culture method to diagnose fungal infections with biopsy tissues

Biopsy tissue is a difficult sample to gain for the identification of potential pathogen in the clinical laboratory. At present, there are no effective culture methods for small-sized biopsy tissue. In this study, we propose a novel tissue biopsy culture method based on Myco/F lytic liquid culture system, namely, tissue was grinded before injected into culture bottles. More types and numbers of fungi are cultivated using this culture system than sputum or BALF culture. A few of mycobacteria were also detected with ground tissues. The method may be a promising alternative to or supplement for the traditional plate culture in clinical practice.

microbiology↗

bioRxiv2019

DeepSeqPanII: an interpretable recurrent neural network model with attention mechanism for peptide-HLA class II binding prediction

Human leukocyte antigen (HLA) complex molecules play an essential role in immune interactions by presenting peptides on the cell surface to T cells. With significant progress in deep learning, a series of neural network based models have been proposed and demonstrated with their good performances for peptide-HLA class I binding prediction. However, there still lack effective binding prediction models for HLA class II protein binding with peptides due to its inherent challenges. In this work, we present a novel sequence-based pan-specific neural network structure, DeepSeaPanII, for peptide-HLA class II binding prediction. Compared with existing pan-specific models, our model is an end-to-end neural network model without the need for pre- or post-processing on input samples. Besides state-of-the-art peformance in binding affinity prediction, DeepSeqPanII can also extract biological insight on the binding mechanism over the peptide and HLA sequences by its attention mechanism based binding core prediction capability. The leave-one-allele-out cross validation and benchmark evaluation results show that our proposed network model achieved state-of-the-art performance in HLA-II peptide binding. The source code and trained models are freely available at https://github.com/pcpLiu/DeepSeqPanII.

bioinformatics↗

bioRxiv2019

Replaying the evolutionary tape to investigate subgenome dominance in allopolyploidBrassica napus

Interspecific hybridization and allopolyploidization merges evolutionarily distinct parental genomes (subgenomes) into a single nucleus. A frequent observation is that one subgenome is "dominant" over the other subgenome, having a greater number of reatined duplicate genes and being more highly expressed. Which subgenome becomes dominantly expressed in allopolyploids remains poorly understood. Here we "replayed the evolutionary tape" with six isogenic resynthesized Brassica napus (rapeseed) allopolyploid lines and investigated subgenome dominance patterns over the first ten generations. We found that the same subgenome was consistently more dominantly expressed in all lines and generations. Furthermore, DNA methylation differences between subgenomes mirrored the observed gene expression bias towards the Brassica oleracea derived C subgenome in all lines and generations. These differences in gene expression and methylation were also found when comparing the progenitor genomes, suggesting subgenome dominance is related to inherited parental genome differences rather than a byproduct of allopolyploidization. Gene network analyses indicated an enrichment for network interactions and several biological functions for C subgenome biased pairs, but no enrichment was observed for A subgenome biased pairs. These findings demonstrate that "replaying the evolutionary tape" in allopolyploids results in repeatable and predictable subgenome expression dominance patterns based on preexisting genetic differences among the parental species. These findings have major implications regarding the genotypic and phenotypic diversity observed following plant hybridization in both ecological and agricultural contexts.

evolutionary biology↗

bioRxiv2019

CD200 immune checkpoint reversal at the site of tumor vaccine inoculation: A novel approach to glioblastoma immunotherapy

PurposeAdvances in immunotherapy have revolutionized care for some cancer patients. However, current checkpoint inhibitors are associated with significant toxicity and yield poor responses for patients with central nervous system tumors, calling into question whether cancer immunotherapy can be applied to glioblastoma multiforme. We determined that targeting the CD200 activation receptors (CD200AR) of the CD200 checkpoint with a peptide inhibitor overcomes tumor-induced immunosuppression. We have shown the clinical efficacy of the peptide inhibitor in a trial in companion dogs with spontaneous high-grade glioma; adding the peptide to autologous tumor lysate vaccines significantly increased overall survival relative to tumor lysate alone (median survival of 12.7 and 6.36 months, respectively).\n\nExperimental design: This study was developed to elucidate the mechanism of the CD200ARs and develop a humanized peptide inhibitorWe developed macrophage cell lines with each of four CD200ARs knocked out to determine their binding specificity and functional responses. Using proteomics, we developed humanized peptide inhibitors to explore their effects on cytokine/chemokine response, dendritic cell maturation and CMV pp65 antigen response in CD14+ cells. GMP-grade peptide was further validated for activity.\n\nResultsWe demonstrated that the peptide specifically targets the CD200AR complex to induce an immune response. Moreover, we developed and validated our humanized peptides for inducing chemokine response, stimulating immature dendritic cell differentiation and significantly enhancing an antigen-specific response. We also determined that the use of the peptide downregulated the expression of CD200 inhibitory and PD-1 receptors.\n\nConclusionThese results support consideration of a CD200 peptide ligand as a novel platform for immunotherapy against multiple cancers including glioblastoma multiforme.\n\nTranslational relevanceThis report evaluates the ability to modulate the CD200 immune checkpoint by employing synthetic peptides directed as ligands to its paired immune activation receptor. We previously reported the presence of CD200 in the sera and tumor vasculature of patients with glioblastoma multiforme (GBM). We have also shown that a canine CD200 activation receptor ligand extends the lives of companion dogs with high grade glioma. The data we present here show that the human peptide ligand (hCD200ARL) directed to the CD200 activation receptor on CD14+ cells activates immune upregulation through induction of a cytokine response and dendritic cell differentiation. In addition, hCD200ARL downregulates the inhibitory CD200 and PD-1 receptors. These findings provide a basis to evaluate hCD200ARL as a novel immune therapy for patients with GBM. Downregulation of PD-1 suggests that hCD200ARL may also obviate the need for PD1 and PD-L1 directed therapies for GBM and other malignancies.

cancer biology↗

bioRxiv2019

Novel genetic loci affecting facial shape variation in humans

The human face represents a combined set of highly heritable phenotypes, but knowledge on its genetic architecture remains limited despite the relevance for various fields of science and application. A series of genome-wide association studies on 78 facial shape phenotypes quantified from 3-dimensional facial images of 10,115 Europeans identified 24 genetic loci reaching genome-wide significant association, among which 17 were previously unreported. A multi-ethnic study in additional 7,917 individuals confirmed 13 loci including 8 unreported ones. A global map of polygenic face scores assembled facial features in major continental groups consistent with anthropological knowledge. Analyses of epigenomic datasets from cranial neural crest cells revealed abundant cis-regulatory activities at the face-associated genetic loci. Luciferase reporter assays in neural crest progenitor cells highlighted enhancer activities of several face-associated DNA variants. These results substantially advance our understanding of the genetic basis underlying human facial variation and provide candidates for future in-vivo functional studies.

genomics↗

bioRxiv2018

DeepSeqPan, a novel deep convolutional neural network model for pan-specific class I HLA-peptide binding affinity prediction

Interactions between human leukocyte antigens (HLAs) and peptides play a critical role in the human immune system. Accurate computational prediction of HLA-binding peptides can be used for peptide drug discovery. Currently, the best prediction algorithms are neural network based pan-specific models, which take advantage of the large amount of data across HLA alleles. However, current pan-specific models are all based on the pseudo sequence encoding for modeling the binding context and depend on the available HLA protein-peptide bound structures. In this work, we proposed a novel deep convolutional neural network model (DCNN) for HLA-peptide binding prediction, in which the encoding of the HLA sequence and the binding context are both learned by the network itself without requiring the HLA-peptide bound structure information. Our DCNN model is also characterized by its binding context extraction layer and dual outputs with both binding affinity output and binding probability outputs. Evaluation on public benchmark datasets shows that our DeepSeqPan model without HLA structural information in training achieves state-of-the-art performance on a large number of HLA alleles with good generalization capability. Since our model only needs raw sequences from the HLA-peptide binding pairs, it can be applied to binding predictions of HLAs without structure information and can also be applied to other protein binding problems such as protein-DNA and protein-RNA bindings. The implementation code and trained models are freely available at https://github.com/pcpLiu/DeepSeqPan.

bioinformatics↗

bioRxiv2017

Scalable volumetric imaging for ultrahigh-speed brain mapping at synaptic resolution

We describe a new light-sheet microscopy method for fast, large-scale volumetric imaging. Combining synchronized scanning illumination and oblique imaging over cleared, thick tissue sections in smooth motion, our approach achieves high-speed 3D image acquisition of an entire mouse brain within 2 hours, at a resolution capable of resolving synaptic spines. It is compatible with immunofluorescence labeling, enabling flexible cell-type specific brain mapping, and is readily scalable for large biological samples such as primate brain.

neuroscience↗

bioRxiv2017

Genetics and epigenetic alterations of hexaploid early generation derived from hybrid between Brassica napus and B. oleracea

Good fertility was observed previously in hexaploid derived from hybrid (ACC) between calona Zhongshuang 9(Brassica napus, 2n = 38, AACC) and kale SWU01 (B. oleracea var. acephala, 2n = 18, CC). However, the mechanism to underlying the character is unknown. In the present study, genetic and epigenetic alterations of S0, 6 S1, and 18 of their S2 progenies with hexaploid chromosome conformation (20A + 36C) were selected to compare with ACC and their parental species. 13.08% and 26.45% polymorphism alleles different from two parental species were identified in ACC via 58 SSR (simple sequence repeats) and 14 MSAP (methylation sensitive amplified polymorphism), respectively. 33.74% new alleles in DNA methylation, but not in DNA sequence were detected in S0 after chromosome doubling of ACC. DNA profilling revealed a little genetic but much epigenetic differences among S0, S1 and S2 generations. Genetic alteration was relatively stable, because only 8.09% and 3.21% alleles inheriated from ACC were changed in S2 and S1, respectively. While on average of 52.44 {+/-} 5.32% DNA methylation site inherited from ACC were detected in S1, and 41.52 {+/-} 9.04% in S2 due to dramatic epigenetic variance among early generations. New DNA methylation sites occurred in S0 would inheritated into the successive generations, but the frequency was decreased because some new site might be recovered. It demonstrated that much DNA methylation but a little DNA sequence variance was occurred in hexaploid early generation.

genetics↗