Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher.
Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?
Some links on this page may take you to non-federal websites. Their policies may differ from this site.
Drug-induced liver injury (DILI) remains a leading cause of drug attrition and acute liver failures, partly due to the inadequacy of animal models to accurately predict human clinical outcomes, which necessitates the utilization of in vitro models of the human liver. However, primary human hepatocytes (PHHs) are in short supply for routine drug screening. In contrast, induced pluripotent stem cells (iPSCs)-derived hepatocyte-like cells (HLCs) are a nearly unlimited cell source but display a fetal-like (versus adult-like) phenotype when differentiated using conventional protocols on tissue culture plastic or glass adsorbed with 2D extracellular matrix (ECM) proteins. Electrospinning can produce porous nanoscale 3D fibers that have a large surface area and present a high density of receptor ligands to modulate cell phenotype. However, the application of electrospinning to generate 3D liver-derived ECM substrates for HLC differentiation remains unexplored. Therefore, here we developed methods to a) electrospin nanofibers of different porosities and diameters using porcine liver ECM (PLECM) with or without type I collagen and b) use these fibers to determine functional modulation in iPSC-derived HLCs while using PHHs as a control cell type relative to conventional adsorbed ECM substrates.Free, publicly-accessible full text available October 13, 2023
Decellularized Liver Nanofibers Enhance and Stabilize the Long‐Term Functions of Primary Human Hepatocytes In Vitro
Owing to significant differences across species in liver functions, in vitro human liver models are used for screening the metabolism and toxicity of compounds, modeling diseases, and cell‐based therapies. However, the extracellular matrix (ECM) scaffold used for such models often does not mimic either the complex composition or the nanofibrous topography of native liver ECM. Thus, here novel methods are developed to electrospin decellularized porcine liver ECM (PLECM) and collagen I into nano‐ and microfibers (≈200–1000 nm) without synthetic polymer blends. Primary human hepatocytes (PHHs) on nanofibers in monoculture or in coculture with nonparenchymal cells (3T3‐J2 embryonic fibroblasts or primary human liver endothelial cells) display higher albumin secretion, urea synthesis, and cytochrome‐P450 1A2, 2A6, 2C9, and 3A4 enzyme activities than on conventionally adsorbed ECM controls. PHH functions are highest on the collagen/PLECM blended nanofibers (up to 34‐fold higher CYP3A4 activity relative to adsorbed ECM) for nearly 7 weeks in the presence of the fibroblasts. In conclusion, it is shown for the first time that ECM composition and topography synergize to enhance and stabilize PHH functions for several weeks in vitro. The nanofiber platform can prove useful for the above applications and to elucidate cell‐ECM interactions in the human liver.
We give a simple, fast algorithm for hyperparameter optimization inspired by techniques from the analysis of Boolean functions. We focus on the high-dimensional regime where the canonical example is training a neural network with a large number of hyperparameters. The algorithm --- an iterative application of compressed sensing techniques for orthogonal polynomials --- requires only uniform sampling of the hyperparameters and is thus easily parallelizable. Experiments for training deep neural networks on Cifar-10 show that compared to state-of-the-art tools (e.g., Hyperband and Spearmint), our algorithm finds significantly improved solutions, in some cases better than what is attainable by hand-tuning. In terms of overall running time (i.e., time required to sample various settings of hyperparameters plus additional computation time), we are at least an order of magnitude faster than Hyperband and Bayesian Optimization. We also outperform Random Search 8x. Additionally, our method comes with provable guarantees and yields the first improvements on the sample complexity of learning decision trees in over two decades. In particular, we obtain the first quasi-polynomial time algorithm for learning noisy decision trees with polynomial sample complexity.