Biology subjects

Liu, L.

Publications and source records attributed to Liu, L..

At least 415 records · Page 23Linked to original sources

bioRxiv2023

Unbiased Complete Estimation of Chloroplast Number in Plant Cells Using Deep Learning Methods

Chloroplasts are essential organelles in plants that are involved in plant development and photosynthesis. Accurate quantification of chloroplast numbers is important for understanding the status and type of plant cells, as well as assessing photosynthetic potential and efficiency. Traditional methods of counting chloroplasts using microscopy are time-consuming and face challenges such as the possibility of missing out-of-focus samples or double counting when adjusting the focal position. Here, we developed an innovative approach called Detecting- and-Counting-chloroplasts (D&Cchl) for automated detection and counting of chloroplasts. This approach utilizes a deep-learning-based object detection algorithm called You-Only-Look-Once (YOLO), along with the Intersection Over Union (IOU) strategy. The application of D&Cchl has shown excellent performance in accurately identifying and quantifying chloroplasts. This holds true when applied to both a single image and a three-dimensional (3D) structure composed of a series of images. Furthermore, by integrating Cellpose, a cell-segmentation tool, we were able to successfully perform single-cell 3D chloroplast counting. Compared to manual counting methods, this approach improved the accuracy of detection and counting to over 95%. Together, our work not only provides an efficient and reliable tool for accurately analyzing the status of chloroplasts, enhancing our understanding of plant photosynthetic cells and growth characteristics, but also makes a significant contribution to the convergence of botany and deep learning. One-sentence summaryThis deep learning-based approach enables the accurate complete detection and counting of chloroplasts in 3D single cells using microscopic image stacks, and showcases a successful example of utilizing deep learning methods to analyze subcellular spatial information in plant cells. The authors responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell/) is: Zhao Dong (dongzhao@hebeu.edu.cn), Shaokai Yang, (shaokai1@ualberta.ca), Ningjing Liu (liuningjing1@yeah.net), and Qiong Zhao (qzhao@bio.ecnu.edu.cn).

cell biology↗

bioRxiv2023

Melon: metagenomic long-read-based taxonomic identification and quantification using marker genes

Long-read sequencing holds great potential for characterizing complex microbial communities, yet taxonomic profiling tools designed specifically for long reads remain lacking. Here we introduce Melon, a novel marker-based taxonomic profiler that capitalizes on the unique attributes of long reads. Melon employs a two-stage classification scheme to reduce computational time and is equipped with an expectation-maximization-based post-correction module to handle ambiguous reads. Melon achieves superior performance compared to other existing tools in both mock and simulated samples. Using wastewater metagenomic samples, we demonstrate the applicability of Melon by showing it is capable in providing reliable estimates of overall genome copies, and species-level taxonomic profiles.

bioinformatics↗

bioRxiv2023

Conformational Space Profile Enhances Generic Molecular Representation Learning

The molecular representation model is a neural network that converts molecular representations (SMILES, Graph) into feature vectors, that carries the potential to be applied across a wide scope of drug discovery scenarios. However, current molecular representation models have been limited to 2D or static 3D structures, overlooking the dynamic nature of small molecules in solution and their ability to adopt flexible conformational changes crucial for drug-target interactions. To address this limitation, we propose a novel strategy that incorporates the conformational space profile into molecular representation learning. By capturing the intricate interplay between molecular structure and conformational space, our strategy enhances the representational capacity of our model named GeminiMol. Consequently, when pre-trained on a miniaturized molecular dataset, the GeminiMol model demonstrates a balanced and superior performance not only on traditional molecular property prediction tasks but also on zero-shot learning tasks, including virtual screening and target identification. By capturing the dynamic behavior of small molecules, our strategy paves the way for rapid exploration of chemical space, facilitating the transformation of drug design paradigms.

bioinformatics↗

bioRxiv2023

Discovery of anthelmintic small molecules in the Medicines for Malaria Ventures COVID and Global Health Priority Boxes using an infrared-based assay for Caenorhabditis elegans motility

BackgroundParasitic nematodes are a public health problem globally, and an economic burden on animal and plant agricultural industries. With their ability to generate drug resistance, new anthelmintic compounds must be constantly sourced. MethodsUsing the free-living nematode, Caenorhabditis elegans, in an infrared-based motility assay, we screened 400 compounds from two open-source, small-molecule collections distributed by the Medicines for Malaria Venture, namely, the COVID Box and Global Health Priority Box. The screening assay was first validated for worm number, DMSO concentration and final volume. ResultsPrimary and secondary (time- and concentration-dependent) screens of both boxes, identified twelve compounds as hits; nine of which were known anthelmintics. Three novel anthelmintic hits, flufenerim, flucofuron and indomethacin were identified with EC50 values ranging from 0.211 to 23.174 {micro}M. Counter toxicity screens with HEK293 cells indicated varying degrees of toxicity with EC50 values ranging from 0.453 to >100 {micro}M. ConclusionsA C. elegans motility assay was optimized and used to screen two recently-released, small molecule libraries. One or more of these three novel active compounds might serve as starting points for the development of new anthelmintics.

microbiology↗

bioRxiv2023

Enhancing Recognition and Interpretation of Functional Phenotypic Sequences through Fine-Tuning Pre-Trained Genomic Models

Decoding high-quality human genomic sequences requires comprehensive analysis of DNA sequence functionality. Through computational and experimental approaches, researchers study the genotype-phenotype relationship and generate important datasets that help unravel complicated genetic blueprints. This study explores the use of deep learning, particularly pre-trained models like DNA_bert_6 and human_gpt2-v1, in interpreting and representing human genome sequences. We meticulously construct multiple datasets linking genotypes and phenotypes to fine-tune pre-trained models for precise DNA sequence classification. Furthermore, we specifically focused on the human endogenous retrovirus (HERV) dataset with commendable classification performance (both binary and multi-classification accuracy and F1 values above 0.935 and 0.888, respectively). We evaluate the influence of sequence length on classification results and analyze the impact of feature extraction in the models hidden layers using the HERV dataset. To further understand the phenotype-specific patterns learned by the model, we perform enrichment, pathogenicity and conservation analyzes of specific motifs in the HERV sequence with high average local representation weight (LRAW) scores. Overall, the generated datasets further provide numerous additional genotype-phenotype datasets for evaluating the performance of genomic models. The findings highlight the potential of large models in learning DNA sequence representations, particularly when utilizing the HERV dataset, and provide valuable insights for future research. This work represents an innovative strategy that combines pre-trained model representations with classical omics methods for analyzing the functionality of genome sequences, fostering cross-fertilization between genomics and advanced AI. The source code and data are available at https://github.com/GeorgeBGM/Genome_Fine-Tuning.

bioinformatics↗

bioRxiv2023

Dorsal hippocampal oxytocin receptor regulates adult peer bonding in rats

Healthy social relationships are beneficial whereas their breakdown is often linked to psychiatric disorders. Parental care and bonding between sexual partners have been well studied both at the level of behavioral analysis and underlying neuronal mechanisms. By contrast, little is known about the neural and molecular basis of peer bonding, defined as social bonds formed between unrelated individuals of the same sex, due to the lack of a suitable experimental paradigm. We found that adult Sprague Dawley (SD) rats of the same sex form strong peer bonds with each other following co-housing. Peer bonded rats exhibit affiliative displays toward their cagemates who are distressed whereas they exhibit agonistic behaviors toward strangers in these situations. Using innovative, genetic strategies in rats, we show that both oxytocin receptor (OXTR) bearing neurons and Oxtr signaling in the dorsal hippocampus are essential for peer bonds to form. Together, we have developed a new platform for studying peer bonding and demonstrate a neural pathway that governs this behavior.

neuroscience↗

bioRxiv2023

AKAP8 interacting with DDX5 to regulate R-loop balance involves in lung carcinoma cell growth

Triple-nucleotide chain of RNA:DNA hybrid structure R-loop is formed during transcription process which is closely associated with transcriptional regulation, transcriptional termination, epigenetic modifications, and structure of chromatin. Dysregulation of R-loop formation or resolution may disturb normal DNA replication or RNA transcription, leading to associated progress of disease. Herein, we found A-kinase-anchoring protein 8 (AKAP8) bind with R-loop structure in cell. Knock down of AKAP8 showed perturbation of balance of genomic R-loop formation and gene transcription. Evidences was shown that AKAP8 interacted with R-loop resolution protein ATP-dependent DEAD box RNA helicase DDX5 and increased chromatin associated DDX5 level. Finally, AKAP8 promoted UCP2 transcription and resolved R-loop level of its promoter region, and contributed to cell growth of adenocarcinomic human alveolar basal epithelial cell line A549, which may provide clues for poor prognosis of lung adenocarcinoma in high AKAP8 expression patients.

cell biology↗

bioRxiv2023

XBB.1.5 monovalent mRNA vaccine booster elicits robust neutralizing antibodies against emerging SARS-CoV-2 variants

COVID-19 vaccines have recently been updated with the spike protein of SARS-CoV-2 XBB.1.5 subvariant alone, but their immunogenicity in humans has yet to be fully evaluated and reported, particularly against emergent viruses that are rapidly expanding. We now report that administration of an updated monovalent mRNA vaccine (XBB.1.5 MV) to uninfected individuals boosted serum virus-neutralization antibodies significantly against not only XBB.1.5 (27.0-fold) and the currently dominant EG.5.1 (27.6-fold) but also key emergent viruses like HV.1, HK.3, JD.1.1, and JN.1 (13.3-to-27.4-fold). In individuals previously infected by an Omicron subvariant, serum neutralizing titers were boosted to highest levels (1,504-to-22,978) against all viral variants tested. While immunological imprinting was still evident with the updated vaccines, it was not nearly as severe as the previously authorized bivalent BA.5 vaccine. Our findings strongly support the official recommendation to widely apply the updated COVID-19 vaccines to further protect the public.

immunology↗

bioRxiv2023

Identifying cooperative genes causing cancer progression with dynamic causal inference

Cancer progression is driven by complex gene interactions, which are not fully understood due to the dynamic and multifaceted nature of the disease. Traditional methods for identifying genes driving cancer typically focus on detecting mutational events, often neglecting other genetic alterations and the temporal dynamics of the cancer development. This study introduces a novel approach that expands our understanding of cancer by identifying cooperative networks of cancer drivers through dynamic causal inference, complementing classical mutation-based methods. We developed a data-driven causal inference method that integrates a temporal dimension to gene expression data through pseudotime analysis. By modelling cancer progression as a dynamic system, the method evaluates gene interactions and their causal relationships using causal kinetic models (CKMs). This approach enables detecting cancer driver genes regardless of their mutational status. The approach was applied to both single cell and bulk sequencing datasets of breast cancer to identify and rank cancer driver genes based on their stability on the causal model. The proposed method effectively identifies and ranks cancer driver genes by considering their dynamic interactions during cancer progression. It systematically identified cancer driver genes in both categories: those with single nucleotide variants (SNVs) and those with other alterations. The method demonstrated a significant overlap with known cancer driver genes from the Cancer Gene Census, validating its effectiveness. Additionally, the method uncovered novel driver genes that act cooperatively to influence cancer development, offering a more comprehensive view of the genetic mechanisms underlying cancer. Our approach, along with the scripts for the experiments and datasets used, can be found at https://github.com/AndresMCB/DynamicCancerDriverKM

bioinformatics↗

bioRxiv2023

14-3-3 binding motif phosphorylation disrupts Hdac4 organized condensates to stimulate cardiacreprogramming

Cell fate conversion is associated with extensive epigenetic and post translational modifications (PTMs) and architectural changes of sub-organelles and organelles, yet how these events are interconnected remains unknown. We report here the identification of a phosphorylation code in 14-3-3 binding motifs (PC14-3-3) that greatly stimulates induced cardiomyocyte (iCM) formation from fibroblasts. PC14-3-3 was identified in pivotal functional proteins for iCM reprogramming, including transcription factors and epigenetic factors. Akt1 kinase and PP2A phosphatase were a key writer and eraser of the PC14-3-3 code, respectively. PC14-3-3 activation induces iCM formation with the presence of only Tbx5. In contrast, PC14-3-3 inhibition by mutagenesis or inhibitor-mediated code removal abolished reprogramming. We discovered that key PC14-3-3 embedded factors, such as Hdac4, Mef2c, Nrip1, and Foxo1, formed Hdac4 organized inhibitory nuclear condensates. Notably, PC14-3-3 activation disrupted Hdac4 condensates to promote cardiac gene expression. Our study suggests that sub-organelle dynamics regulated by a post-translational modification code could be a general mechanism for stimulating cell reprogramming and organ regeneration. HighlightsO_LIA PC14-3-3 (phosphorylation code in 14-3-3 binding motifs) is identified in pivotal functional proteins, such as HDAC4 and Mef2c, that stimulates iCM formation. C_LIO_LIAkt1 kinase and PP2A phosphatase are a key writer and a key eraser of the PC14-3-3 code, respectively, and PC14-3-3 code activation can replace Mef2c and Gata4 in cardiac reprogramming. C_LIO_LIPC14-3-3 activation disrupts Hdac4 organized condensates which results in releasing multiple 14-3-3 motif embedded proteins from the condensates to stimulate cardiac reprogramming. C_LIO_LISub-organelle dynamics and function regulated by a post-translational modification code could be a general mechanism in stimulating cell reprogramming and organ regeneration. C_LI Graphic abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=150 SRC="FIGDIR/small/567913v1_ufig1.gif" ALT="Figure 1"> View larger version (52K): org.highwire.dtl.DTLVardef@1fec3aforg.highwire.dtl.DTLVardef@e13b16org.highwire.dtl.DTLVardef@ee354borg.highwire.dtl.DTLVardef@ecd58e_HPS_FORMAT_FIGEXP M_FIG C_FIG

cell biology↗

bioRxiv2023

Heparan sulfate dependent phase separation of CCL5 and its chemotactic activity

Secreted chemokines form concentration gradients in target tissues to control migratory directions and patterns of immune cells in response to inflammatory stimulation; however, how the gradients are formed is much debated. Heparan sulfate (HS) binds to chemokines and modulates their activities. In this study, we investigated the roles of HS in the gradient formation and chemoattractant activity of CCL5 that is known to bind to HS. CCL5 and heparin underwent liquid-liquid phase separation (LLPS) and formed gradient, which was confirmed using CCL5 immobilized on heparin-beads. The biological implication of HS in CCL5 gradient formation was established in CHO-K1 (wild type) and CHO-677 (lacking HS) cells by Transwell assay. The effect of HS on CCL5 chemoattractant activity was further proved by Transwell assay of human peripheral blood cells. Finally, peritoneal injection of the chemokines into mice showed reduced recruitment of inflammatory cells either by mutant CCL5 (lacking heparin binding sequence) or by addition of heparin to wild type CCL5. Our experimental data propose that co-phase separation of CCL5 with HS establishes a specific chemokine concentration gradient to trigger directional cell migration. The results warrant further investigation on other heparin binding chemokines and allows for a more elaborate insight into disease process and new treatment strategies.

cell biology↗

bioRxiv2023

Asymmetrical Bi-antennary Glycans Prepared by a Stop-and-Go Strategy Reveal Receptor Binding Evolution of Human Influenza A Viruses

Glycan binding properties of respiratory viruses have been difficult to probe due to a lack of biological relevant glycans for binding studies. Here, a stop-and-go chemoenzymatic methodology is presented that gave access to a panel of 32 asymmetrical bi-antennary N-glycans having various numbers of N-acetyl lactosamine (LacNAc) repeating units capped by 2,3- or 2,6-sialosides resembling structures found in airway tissues. It exploits that the branching enzymes MGAT1 and MGAT2 can utilize unnatural UDP-2-deoxy-2-trifluoro-N-acetamido-glucose (UDP-GlcNTFA) as donor. The TFA moiety of the resulting glycans can be hydrolyzed to give GlcNH2 at one of the antennae that temporarily blocks extension by glycosyl transferases. The N-glycans were printed as a microarray that was probed for receptor binding specificities of evolutionary distinct human A(H3N2) and A(H1N1)pdm09 viruses. It was found that not only the sialoside type but also the length of the LacNAc chain and presentation at the 1,3-antenna of N-glycans is critical for binding. Early A(H3N2) viruses bound to 2,6-sialosides at a single LacNAc moiety at the 1,3-antenna whereas later viruses required the sialoside to be presented at a tri-LacNAc moiety. Surprisingly, most of the A(H3N2) viruses that appeared after 2021 regained binding capacity to sialosides presented at a di-LacNAc moiety. As a result, these viruses agglutinate erythrocytes again, commonly employed for antigenic characterization of influenza viruses. Human A(H1N1)pdm09 viruses have similar receptor binding properties as recent A(H3N2) viruses. The data indicates that an asymmetric N-glycan having 2,6-sialoside at a di-LacNAc moiety is a commonly employed receptor by human influenza A viruses.

biochemistry↗

bioRxiv2023

Poisoning medical knowledge using large language models

Biomedical knowledge graphs constructed from scientific literature have been widely used to validate biological discoveries and generate new hypotheses. Recently, large language models (LLMs) have demonstrated a strong ability to generate human-like text data. While most of these text data have been useful, LLM might also be used to generate malicious content. Here, we investigate whether it is possible that a malicious actor can use LLM to generate a malicious paper that poisons scientific knowledge graphs and further affects downstream biological applications. As a proof-of-concept, we develop Scorpius, a conditional text generation model that generates a malicious paper abstract conditioned on a promoting drug and a target disease. The goal is to fool the knowledge graph constructed from a mixture of this malicious abstract and millions of real papers so that knowledge graph consumers will misidentify this promoting drug as relevant to the target disease. We evaluated Scorpius on a knowledge graph constructed from 3,818,528 papers and found that Scorpius can increase the relevance of 71.3% drug disease pairs from the top 1000 to the top 10 by only adding one malicious abstract. Moreover, the generation of Scorpius achieves better perplexity than ChatGPT, suggesting that such malicious abstracts cannot be efficiently detected by humans. Collectively, Scorpius demonstrates the possibility of poisoning scientific knowledge graphs and manipulating downstream applications using LLMs, indicating the importance of accountable and trustworthy scientific knowledge discovery in the era of LLM.

bioinformatics↗

bioRxiv2023

Ablation of parasympathetic cholinergic innervation of the liver prevents diet-induced obesity and hepatic steatosis in mice.

Hepatic lipid metabolism is regulated by the autonomic nervous system of the liver, with the sympathetic innervation being extensively studied, while the parasympathetic efferent innervation is less understood despite its potential importance. In this study, we investigate the consequences of disrupted brain-liver communication on hepatic lipid metabolism in mice exposed to obesogenic conditions. We found that a subset of hepatocytes and cholangiocytes are innervated by parasympathetic nerve terminals originating from the dorsal motor nucleus of the vagus. The elimination of the brain-liver axis by deleting parasympathetic cholinergic neurons innervating the liver prevents hepatic steatosis and promotes browning of inguinal white adipose tissue (ingWAT). The loss of liver-innervating cholinergic neurons increases hepatic Cyp7b1 expression and fasting serum bile acid levels. Furthermore, knockdown of the G protein-coupled bile acid receptor 1 gene in ingWAT reverses the beneficial effects of the loss of liver-innervating cholinergic neurons, leading to the reappearance of hepatic steatosis. Deleting liver-innervating cholinergic neurons has a small but significant effect on body weight, which is accompanied by an increase in energy expenditure. Taken together, these data suggest that targeting the parasympathetic cholinergic innervation of the liver is a potential therapeutic approach for enhancing hepatic lipid metabolism in obesity and diabetes.

physiology↗

bioRxiv2023

Bioinformatics Illustrations Decoded by ChatGPT: The Good, The Bad, and The Ugly

Emerging studies underscore the promising capabilities of large language model-based chatbots in conducting fundamental bioinformatics data analyses. The recent feature of accepting image-inputs by ChatGPT motivated us to explore its efficacy in deciphering bioinformatics illustrations. Our evaluation with examples in cancer research, including sequencing data analysis, multimodal network-based drug repositioning, and tumor clonal evolution, revealed that ChatGPT can proficiently explain different plot types and apply biological knowledge to enrich interpretations. However, it struggled to provide accurate interpretations when quantitative analysis of visual elements was involved. Furthermore, while the chatbot can draft figure legends and summarize findings from the figures, stringent proofreading is imperative to ensure the accuracy and reliability of the content.

bioinformatics↗

bioRxiv2023

Healthy adults favor stable left/right hand choices over performance at an unconstrained reach-to-grasp task

Reach-to-grasp actions are fundamental to the daily activities of human life, but few methods exist to assess individuals reaching and grasping actions in unconstrained environments. The Block Building Task (BBT) provides an opportunity to directly observe and quantify these actions, including left/right hand choices. Here we sought to investigate the motor and non-motor causes of left/right hand choices, and optimize the design of the BBT, by manipulating motor and non-motor difficulty in the BBTs unconstrained reach-to-grasp task We hypothesized that greater motor and non-motor (e.g. cognitive/perceptual) difficulty would drive increased usage of the dominant hand. To test this hypothesis, we modulated block size (large vs. small) to influence motor difficulty, and model complexity (10 vs. 5 blocks per model) to influence non-motor difficulty, in healthy adults (n=57). We hypothesized that healthy adults with high non-dominant hand performance in a precision drawing task should be more likely to use their non-dominant hand in the BBT. Our data revealed that increased motor and non-motor difficulty led to lower task performance (slower speed), but participants only increased use of their dominant hand only under the most difficult combination of conditions: in other words, participants allowed their performance to degrade before changing hand choices, even though participants were instructed only to optimize performance. These results demonstrate that hand choices during reach-to grasp actions are more stable than motor performance in healthy right-handed adults, but tasks with multifaceted difficulties can drive individuals to rely more on their dominant hand. Statements and DeclarationsDr. Philip and Washington University in St. Louis have a licensing agreement with PlatformSTL to commercialize the iPad app used in this study.

neuroscience↗

bioRxiv2023

Alasemenia, the earliest ovule with three wings and without cupule

The ovules or seeds (fertilized ovules) with wings are widespread and especially important for wind dispersal. However, the earliest ovules in the Famennian of the Late Devonian are rarely known about the dispersal syndrome and usually surrounded by a cupule. From Xinhang, Anhui, China, we report a new taxon of Famennian ovules, Alasemenia tria gen. et sp. nov. Each ovule possesses three integumentary wings evidently extending outwards, folding inwards along abaxial side and enclosing most part of nucellus. The ovule is borne terminally on smooth dichotomous branches and lacks a cupule. Alasemenia suggests that the integuments of the earliest ovules without a cupule evolved functions in wind dispersal and probable photosynthetic nutrition. It indicates that the seed wing originated earlier than other wind dispersal mechanisms such as seed plume and pappus, and that three- or four-winged seeds were followed by seeds with less wings. Mathematical analysis shows that three-winged seeds are more adapted to wind dispersal than seeds with one, two or four wings under the same condition.

paleontology↗

bioRxiv2023

The fatty liver disease-causing protein PNPLA3-I148M alters lipid droplet-Golgi dynamics

Non-alcoholic fatty liver disease (NAFLD), recently renamed metabolic dysfunction-associated steatotic liver disease (MASLD), is a progressive metabolic disorder that begins with aberrant triglyceride accumulation in the liver and can lead to cirrhosis and cancer. A common variant in the gene PNPLA3, encoding the protein PNPLA3-I148M, is the strongest known genetic risk factor for MASLD to date. Despite its discovery twenty years ago, the function of PNPLA3, and now the role of PNPLA3-I148M, remain unclear. In this study, we sought to dissect the biogenesis of PNPLA3 and PNPLA3-I148M and characterize changes induced by endogenous expression of the disease-causing variant. Contrary to bioinformatic predictions and prior studies with overexpressed proteins, we demonstrate here that PNPLA3 and PNPLA3-I148M are not endoplasmic reticulum-resident transmembrane proteins. To identify their intracellular associations, we generated a paired set of isogenic human hepatoma cells expressing PNPLA3 and PNPLA3-I148M at endogenous levels. Both proteins were enriched in lipid droplet, Golgi, and endosomal fractions. Purified PNPLA3 and PNPLA3-I148M proteins associated with phosphoinositides commonly found in these compartments. Despite a similar fractionation pattern as the wild-type variant, PNPLA3-I148M induced morphological changes in the Golgi apparatus, including increased lipid droplet-Golgi contact sites, which were also observed in I148M-expressing primary human patient hepatocytes. In addition to lipid droplet accumulation, PNPLA3-I148M expression caused significant proteomic and transcriptomic changes that resembled all stages of liver disease. Cumulatively, we validate an endogenous human cellular system for investigating PNPLA3-I148M biology and identify the Golgi apparatus as a central hub of PNPLA3-I148M-driven cellular change. Significance StatementFatty liver disease affects nearly a quarter of the worlds population and has both environmental and genetic risk factors. A mutation in the gene PNPLA3 that converts Ile 148 to Met is the strongest known genetic risk factor for developing fatty liver disease. Using a series of techniques to track endogenous PNPLA3 and PNPLA3-I148M biogenesis and localization, we reveal new insights into how the mutation changes cellular dynamics. Although previous reports focus on its role on lipid droplets, we reveal that PNPLA3-I148M also functions at the Golgi apparatus, an organelle critical for protein transport into and out of the cell and lipid signaling. PNPLA3-I148M causes altered Golgi morphology and drives changes reminiscent of liver disease.

cell biology↗