Biology subjects

Hsu, C.

Publications and source records attributed to Hsu, C..

26 records · Page 2Linked to original sources

bioRxiv2022

Chromatin Remodeling in Patient-Derived Colorectal Cancer Models

Patient-Derived Organoids (PDO) and Xenografts (PDX) are the current gold standards for patient derived models of cancer (PDMC). Nevertheless, how patient tumor cells evolve in these models and the impact on drug response remains unclear. Herein, we compared the transcriptomic and chromatin accessibility landscapes of six matched sets of colorectal cancer (CRC) PDO, PDX, PDO-derived PDX (PDOX), and original patient tumors (PT) and discovered two major remodeling axes. The first axis delineates PDX and PDO from PT, and the second axis distinguishes PDX and PDO. PDOX were more similar to PDX than they were to PDO, indicating that the growth environment is a driving force for chromatin adaptation. Using bivariate genomic footprinting analysis, we identified transcription factors (TF) that differentially bind to open chromatins between matched PDO and PDOX. Among them, KLF14 and EGR2 footprints were enriched in all six PDOX relative to matched PDO, and silencing of KLF14 or EGR2 promoted tumor growth. Furthermore, EPHA4, a shared downstream target gene of KLF14 and EGR2, altered tumor sensitivity to MEK inhibitor treatment. Altogether, patient-derived CRC cells undergo both common and distinct chromatin remodeling in PDO and PDX/PDOX, driven largely by their respective microenvironments, which results in differences in growth and drug sensitivity and needs to be taken into consideration when interpreting their ability to predict clinical outcome.

cancer biology↗

bioRxiv2022

Learning inverse folding from millions of predicted structures

We consider the problem of predicting a protein sequence from its backbone atom coordinates. Machine learning approaches to this problem to date have been limited by the number of available experimentally determined protein structures. We augment training data by nearly three orders of magnitude by predicting structures for 12M protein sequences using AlphaFold2. Trained with this additional data, a sequence-to-sequence transformer with invariant geometric input processing layers achieves 51% native sequence recovery on structurally held-out backbones with 72% recovery for buried residues, an overall improvement of almost 10 percentage points over existing methods. The model generalizes to a variety of more complex tasks including design of protein complexes, partially masked structures, binding interfaces, and multiple states.

systems biology↗

bioRxiv2021

MAP4K3 inhibits Sirtuin-1 to repress LKB1-AMPK to promote amino acid dependent activation of mTORC1

mTORC1 is the key rheostat controlling cellular metabolic state. Of the various inputs to mTORC1, the most potent effector of intracellular nutrient status is amino acid supply. Despite an established role for MAP4K3 in promoting mTORC1 activation in the presence of amino acids, the signaling pathway by which MAP4K3 controls mTORC1 activation remains unknown. Here we examined the process of MAP4K3 activation of mTORC1 and found that MAP4K3 represses the LKB1-AMPK pathway to prevent TSC1/2 complex inactivation of Rheb. When we sought the regulatory link between MAP4K3 and LKB1 inhibition, we discovered that MAP4K3 physically interacts with the master nutrient regulatory factor Sirtuin-1 and phosphorylates Sirtuin-1 to repress LKB1 activation. Our results reveal the existence of a novel signaling pathway linking amino acid satiety with MAP4K3-dependent suppression of SIRT1 to inactivate the repressive LKB1-AMPK pathway and thereby potently activate the mTORC1 complex to dictate the metabolic disposition of the cell.

cell biology↗

bioRxiv2021

Nanopore callers for epigenetics from limited supervised data

Nanopore sequencing platforms combined with supervised machine learning (ML) have been effective at detecting base modifications in DNA such as 5mC and 6mA. These ML-based nanopore callers have typically been trained on data that span all modifications on all possible DNA k-mer backgrounds--a complete training dataset. However, as nanopore technology is pushed to more and more epigenetic modifications, such complete training data will not be feasible to obtain. Nanopore calling has historically been performed with Hidden Markov Models (HMMs) that cannot make successful calls for k-mer contexts not seen during training because of their independent emission distributions. However, deep neural networks (DNNs), which share parameters across contexts, are increasingly being used as callers, often outperforming their HMM cousins. It stands to reason that a DNN approach should be able to better generalize to unseen k-mer contexts. Indeed, herein we demonstrate that a common DNN approach (DeepSignal) outperforms a common HMM approach (Nanopolish) in the incomplete data setting. Furthermore, we propose a novel hybrid HMM-DNN approach, Amortized-HMM, that outperforms both the pure HMM and DNN approaches on 5mC calling when the training data are incomplete. Such an approach is expected to be useful for calling 5hmC and combinations of cytosine modifications, where complete training data are not likely to be available.

bioinformatics↗

bioRxiv2021

A multifaceted role of progranulin in regulating amyloid-beta dynamics and responses

Haploinsufficiency of progranulin (PGRN) is a leading cause of frontotemporal lobar degeneration (FTLD). PGRN polymorphisms are associated with Alzheimers disease (AD). PGRN is highly expressed in the microglia near A{beta} plaques and influences plaque dynamics and microglial activation. However, the detailed mechanisms remain elusive. Here we report that PGRN deficiency reduces human APP and A{beta} levels in the young male but not female mice. PGRN deficient microglia exhibit increased expression of markers associated with microglial activation, including CD68, Galectin-3, TREM2 and GPNMB, specifically near A{beta} plaques. In addition, PGRN loss leads to up-regulation of lysosome proteins and an increase in the nuclear localization of TFE3, a transcription factor involved in lysosome biogenesis. Cultured PGRN deficient microglia show enhanced nuclear translocation of TFE3 and inflammation in response to A{beta} fibril treatment. Taken together, our data revealed a sex- and age-dependent effect of PGRN on amyloid metabolism and a role of PGRN in regulating lysosomal activities and inflammation in plaque-associated microglia, which may contribute to the disease mechanism associated with PGRN polymorphisms in AD.

neuroscience↗

bioRxiv2021

Combining evolutionary and assay-labelled data for protein fitness prediction

Predictive modelling of protein properties has become increasingly important to the field of machine-learning guided protein engineering. In one of the two existing approaches, evolutionarily-related sequences to a query protein drive the modelling process, without any property measurements from the laboratory. In the other, a set of protein variants of interest are assayed, and then a supervised regression model is estimated with the assay-labelled data. Although a handful of recent methods have shown promise in combining the evolutionary and supervised approaches, this hybrid problem has not been examined in depth, leaving it unclear how practitioners should proceed, and how method developers should build on existing work. Herein, we present a systematic assessment of methods for protein fitness prediction when evolutionary and assay-labelled data are available. We find that a simple baseline approach we introduce is competitive with and often outperforms more sophisticated methods. Moreover, our simple baseline is plug-and-play with a wide variety of established methods, and does not add any substantial computational burden. Our analysis highlights the importance of systematic evaluations and sufficient baselines.

synthetic biology↗

bioRxiv2019

A novel exosome biogenesis mechanism: multivesicular structures budding and rupturing at the plasma membrane

Exosomes are small vesicles secreted by the cells, which mediate intercellular signalling and systemic physiological processes. Exosomes are known to originate from the intraluminal vesicles of the multivesicular endosome that fuses with the plasma membrane. We found that the non-small cell lung cancer (NSCLC) cell lines, HCC15 and A549, secreted exosomes with typical morphology and protein contents. Unexpectedly, transmission electron microscopy images indicated that the cells formed multivesicular structures that protruded from the plasma membrane and ruptured to release the exosomes. There were smooth multivesicular structures surrounded by an ordinary looking membrane, multivesicular structures coated by an electron dense layer with regular spacing pattern, and intermediate forms that combined elements of both. Electron microscopy images suggested that exosomes are release from these structures by burst events and not by the conventional fusion process. The molecular details of this novel mechanism for membrane association, deformation and fusion is to be unveiled in the future.

cell biology↗

bioRxiv2017

Exome chip meta-analysis elucidates the genetic architecture of rare coding variants in smoking and drinking behavior

BackgroundSmoking and alcohol use behaviors in humans have been associated with common genetic variants within multiple genomic loci. Investigation of rare variation within these loci holds promise for identifying causal variants impacting biological mechanisms in the etiology of disordered behavior. Microarrays have been designed to genotype rare nonsynonymous and putative loss of function variants. Such variants are expected to have greater deleterious consequences on gene function than other variants, and significantly contribute to disease risk.\n\nMethodsIn the present study, we analyzed [~]250,000 rare variants from 17 independent studies. Each variant was tested for association with five addiction-related phenotypes: cigarettes per day, pack years, smoking initiation, age of smoking initiation, and alcoholic drinks per week. We conducted single variant tests of all variants, and gene-based burden tests of nonsynonymous or putative loss of function variants with minor allele frequency less than 1%.\n\nResultsMeta-analytic sample sizes ranged from 70,847 to 164,142 individuals, depending on the phenotype. Known loci tagged by common variants replicated, but there was no robust evidence for individually associated rare variants, either in gene based or single variant tests. Using a modified method-of-moment approach, we found that all low frequency coding variants, in aggregate, contributed 1.7% to 3.6% of the phenotypic variation for the five traits (p<.05).\n\nConclusionsThe findings indicate that rare coding variants contribute to phenotypic variation, but that much larger samples and/or denser genotyping of rare variants will be required to successfully identify associations with these phenotypes, whether individual variants or gene- based associations.

genetics↗