Corrigendum: Spatiotemporal analysis of 3D human iPSC-derived neural networks using a 3D multi-electrode array.

[This corrects the article DOI: 10.3389/fncel.2023.1287089.].

Spatiotemporal analysis of 3D human iPSC-derived neural networks using a 3D multi-electrode array.

While there is a growing appreciation of three-dimensional (3D) neural tissues (i.e., hydrogel-based, organoids, and spheroids), shown to improve cellular health and network activity to mirror brain-like activity in vivo, functional assessment using current electrophysiology techniques (e.g., planar multi-electrode arrays or patch clamp) has been technically challenging and limited to surface measurements at the bottom or top of the 3D tissue. As next-generation MEAs, specifically 3D MEAs, are being developed to increase the spatial precision across all three dimensions (X, Y, Z), development of improved computational analytical tools to discern region-specific changes within the Z dimension of the 3D tissue is needed. In the present study, we introduce a novel computational analytical pipeline to analyze 3D neural network activity recorded from a "bottom-up" 3D MEA integrated with a 3D hydrogel-based tissue containing human iPSC-derived neurons and primary astrocytes. Over a period of ~6.5 weeks, we describe the development and maturation of 3D neural activity (i.e., features of spiking and bursting activity) within cross sections of the 3D tissue, based on the vertical position of the electrode on the 3D MEA probe, in addition to network activity (identified using synchrony analysis) within and between cross sections. Then, using the sequential addition of postsynaptic receptor antagonists, bicuculline (BIC), 2-amino-5-phosphonovaleric acid (AP-5), and 6-cyano-5-nitroquinoxaline-2,3-dione (CNQX), we demonstrate that networks within and between cross sections of the 3D hydrogel-based tissue show a preference for GABA and/or glutamate synaptic transmission, suggesting differences in the network composition throughout the neural tissue. The ability to monitor the functional dynamics of the entire 3D reconstructed neural tissue is a critical bottleneck; here we demonstrate a computational pipeline that can be implemented in studies to better interpret network activity within an engineered 3D neural tissue and have a better understanding of the modeled organ tissue.

Predicting Energetics Materials' Crystalline Density from Chemical Structure by Machine Learning.

To expedite new molecular compound development, a long-sought goal within the chemistry community has been to predict molecules' bulk properties of interest a priori to synthesis from a chemical structure alone. In this work, we demonstrate that machine learning methods can indeed be used to directly learn the relationship between chemical structures and bulk crystalline properties of molecules, even in the absence of any crystal structure information or quantum mechanical calculations. We focus specifically on a class of organic compounds categorized as energetic materials called high explosives (HE) and predicting their crystalline density. An ongoing challenge within the chemistry machine learning community is deciding how best to featurize molecules as inputs into machine learning models-whether expert handcrafted features or learned molecular representations via graph-based neural network models-yield better results and why. We evaluate both types of representations in combination with a number of machine learning models to predict the crystalline densities of HE-like molecules curated from the Cambridge Structural Database, and we report the performance and pros and cons of our methods. Our message passing neural network (MPNN) based models with learned molecular representations generally perform best, outperforming current state-of-the-art methods at predicting crystalline density and performing well even when testing on a data set not representative of the training data. However, these models are traditionally considered black boxes and less easily interpretable. To address this common challenge, we also provide a comparison analysis between our MPNN-based model and models with fixed feature representations that provides insights as to what features are learned by the MPNN to accurately predict density.

Automated Identification of Molecular Crystals' Packing Motifs.

Packing motifs-patterns in how molecules orient relative to one another in a crystal structure-are an important concept in many subdisciplines of materials science because of correlations observed between specific packing motifs and properties of interest. That said, packing motif data sets have remained small and noisy due to intensive manual labeling processes and insufficient labeling schemes. The most prominent labeling algorithms calculate relative interplanar angles of nearest neighbor molecules to determine the packing motif of a molecular crystal, but this simple approach can fail when neighbors are naively sampled isotropically around the crystal structure. To remedy this issue, we propose an optimization algorithm, which rotates the molecular crystal structure to find representative molecules that inform the packing motif. We package this algorithm into an automated framework-Autopack-which both optimally rotates the crystal structure and labels the packing motif based on the appropriate neighboring molecules. In this work, we detail the Autopack framework and its performance, which shows improvements compared to previous state-of-the-art labeling methods, providing the first quantitative point of comparison for packing motif labeling algorithms. Furthermore, using Autopack (available at https://ipo.llnl.gov/technologies/software/autopack), we perform the first large-scale study of potential relationships between chemicals' compositions and packing motifs, which shows that these relationships are more complex than previously hypothesized from studies that used only tens of polycyclic aromatic hydrocarbon molecules. Autopack's capabilities help pose next steps for crystal engineering research focusing not only on a molecule's adoption of a specific packing motif but also on new structure-property relationships.

A flexible 3-dimensional microelectrode array for in vitro brain models.

Three-dimensional (3D) in vitro models have become increasingly popular as systems to study cell-cell and cell-ECM interactions dependent on the spatial, mechanical, and chemical cues within the environment of the tissue, which is limited in traditional two-dimensional (2D) models. Although electrophysiological recordings of neuronal action potentials through 2D microelectrode arrays (MEAs) are a common and trusted method of evaluating neuronal function, network communication, and response to chemicals and biologicals, there are currently limited options for measuring electrophysiological activity from many locations simultaneously throughout a 3D network of neurons in vitro. Here, we have developed a thin-film, 3D flexible microelectrode array (3DMEA) that non-invasively interrogates a 3D culture of neurons and can accommodate 256 channels of recording or stimulation. Importantly, the 3DMEA is straightforward to fabricate and integrates with standard commercially available electrophysiology hardware. Polyimide probe arrays were microfabricated on glass substrates and mechanically actuated to collectively lift the arrays into a vertical position, relying solely on plastic deformation of their base hinge regions to maintain vertical alignment. Human induced pluripotent stem cell (hiPSC)-derived neurons and astrocytes were entrapped in a collagen-based hydrogel and seeded onto the 3DMEA, enabling growth of suspended cells in the matrix and the formation and maturation of a neural network around the 3DMEA probes. The 3DMEA supported the growth of functional neurons in 3D with action potential spike and burst activity recorded over 45 days in vitro. This platform is an important step in facilitating noninvasive electrophysiological characterization of 3D networks of electroactive cells in vitro.