Biology subjects

Choi, M. L.

Publications and source records attributed to Choi, M. L..

9 recordsLinked to original sources

bioRxiv2026

Intrinsic dimensionality of deep learning representations reveals cell death-associated heterogeneity in Parkinsons disease iPSC-derived neurons

As AI is increasingly used to stratify heterogeneous Parkinsons disease, it is essential to determine whether learned representations preserve disease-relevant variation within diagnostic or genetic groups. In iPSC-derived neurons from three familial PD cases, deep learning representations encoded an independent measure of cell death, and their intrinsic dimensionality increased with death burden. Latent dimensionality offers a way to examine heterogeneity before AI-derived clusters are interpreted as disease subtypes.

cell biology↗

bioRxiv2025

Organelle-Aware Representation Learning Enables Label-Free Detection of Mitochondrial Dysfunction in Live Human Neurons

Mitochondrial dysfunction is a convergent hallmark of neurodegenerative diseases and represents a promising biomarker for early diagnosis and therapy. However, current in vitro assays rely on fluorescence or electron microscopy, which are invasive, low-throughput, and incompatible with longitudinal analysis. Here, we present a noninvasive framework by integrating label-free optical diffraction tomography (ODT) with organelle-aware representation learning to detect subtle mitochondrial dysfunction in live human induced pluripotent stem cell (hiPSC)-derived neurons. Through virtual staining of the nuclei, lysosomes, and mitochondria, we establish two complementary and interpretable classification pipelines: a deep learning model with organelle-aware encoder and a logistic regression model on morphometric descriptors. Both models achieve approximately 85% accuracy: the deep model provides end-to-end prediction with minimal feature engineering, whereas the logistic regression model offers a more interpretable, feature-based approach. To our knowledge, this is the first demonstration of ODT-based organelle-resolved virtual staining in live human neurons, establishing a scalable, non-invasive platform for mitochondrial disease modeling, drug discovery, and neurodegeneration research.

neuroscience↗

bioRxiv2025

Autonomous neuronal resilience under metabolic stress highlights fundamental differences between hESC and hiPSC-derived neurons

Human induced pluripotent stem cells (hiPSCs) aim to replicate the developmental and functional capacity of human embryonic stem cells (hESCs). Here, we identify neuronal resilience under metabolic stress as a critical parameter for benchmarking equivalence. Without medium replenishment, hESC-derived cortical neurons underwent neural resurrection--a spontaneous recovery marked by increased cell density, preserved morphology, and sustained calcium signaling and mitochondrial function for 23 days. In contrast, hiPSC-derived neurons survived but showed reduced adaptability, deteriorating within 10-14 days despite higher initial densities. Our findings introduce a physiologically relevant assay for stress resilience and highlight the need to optimize hiPSC differentiation to achieve hESC-like performance, improving their translational value for disease modeling and regenerative therapy.

cell biology↗

bioRxiv2024

Alpha-synuclein aggregates trigger anti-viral immune pathways and RNA editing in human astrocytes

RNA editing is a post transcriptional mechanism that targets changes in RNA transcripts to modulate innate immune responses. We report the role of astrocyte specific, ADAR1 mediated RNA editing in neuroinflammation in Parkinsons disease. We generated hiPSC-derived astrocytes, neurons and co-cultures and exposed them to small soluble alpha-synuclein aggregates. Oligomeric alpha-synuclein triggered an inflammatory glial state associated with TLR activation, viral responses, and cytokine secretion. This reactive state resulted in loss of neurosupportive functions, and the induction of neuronal toxicity. Notably, interferon response pathways were activated leading to upregulation, and isoform switching of the RNA deaminase enzyme, ADAR1. ADAR1 mediates A-to-I RNA editing, and increases in RNA editing were observed in inflammatory pathways in cells, as well as in post-mortem human PD brain. Aberrant, or dysregulated, ADAR1 responses and RNA editing may lead to sustained inflammatory reactive states in astrocytes triggered by alpha-synuclein aggregation, and this may drive the neuroinflammatory cascade in Parkinsons.

neuroscience↗

bioRxiv2024

GLP-1 receptor agonism ameliorates Parkinsons disease through modulation of neuronal insulin signalling and glial suppression

Neuronal insulin resistance is linked to the pathogenesis of Parkinsons disease through unclear, but potentially targetable, mechanisms. We delineated neuronal and glial mechanisms of insulin resistance and glucagon-like 1 peptide (GLP-1) receptor agonism in human iPSC models of synucleinopathy, and corroborated our findings in patient samples from a Phase 2 trial of a GLP-1R agonist in Parkinsons (NCT01971242). Human iPSC models of synucleinopathy exhibit neuronal insulin resistance and dysfunctional insulin signalling, which is associated with inhibition of the neuroprotective Akt pathways, and increased expression of the MAPK-associated p38 and JNK stress pathways. Ultimately, this imbalance is associated with cellular stress, impaired proteostasis, accumulation of -synuclein, and neuronal loss. The GLP-1R agonist exenatide led to restoration of insulin signalling, associated with restoration of Akt signalling and suppression of the MAPK pathways in neurons. GLP-1R agonism reverses the neuronal toxicity associated with the synucleinopathy, through reduction of oxidative stress, improved mitochondrial and lysosomal function, reduced aggregation of -synuclein, and enhanced neuronal viability. GLP-1R agonism further suppresses synuclein induced inflammatory states in glia, leading to neuroprotection through non cell autonomous effects. In the exenatide-PD2 clinical trial, exenatide treatment was associated with clinical improvement in individuals with higher baseline MAPK expression (and thus insulin resistance). Exenatide treatment led to a reduction of -synuclein aggregates, and a reduction in inflammatory cytokine IL-6. Taken together, our patient platform defines the mechanisms of GLP-1R action in neurons and astrocytes, identifies the population likely to benefit from GLP-1R agonism, and highlights the utility of GLP-1R agonism as a disease modifying strategy in synucleinopathies.

neuroscience↗

bioRxiv2022

Protein aggregation and calcium dysregulation are the earliest hallmarks of synucleinopathy in human midbrain dopaminergic neurons

Mutations in the SNCA gene cause autosomal dominant Parkinsons disease (PD), with progressive loss of dopaminergic neurons in the substantia nigra, and accumulation of aggregates of -synuclein. However, the sequence of molecular events that proceed from the SNCA mutation during development, to its end stage pathology is unknown. Utilising human induced pluripotent stem cells (hiPSCs) with SNCA mutations, we resolved the temporal sequence of pathophysiological events that occur during neuronal differentiation in order to discover the early, and likely causative, events in synucleinopathies. We adapted a small molecule-based protocol that generates highly enriched midbrain dopaminergic (mDA) neurons (>80%). We characterised their molecular identity using single-cell RNA sequencing and their functional identity through the synthesis and secretion of dopamine, the ability to generate action potentials, and form functional synapses and networks. RNA velocity analyses confirmed the developmental transcriptomic trajectory of midbrain neural precursors into mDA neurons using our approach, and identified key driver genes in mDA neuronal development. To characterise the synucleinopathy, we adopted super-resolution methods to determine the number, size and structure of aggregates in SNCA-mutant mDA neurons. At one week of differentiation, prior to maturation to mDA neurons of molecular and functional identity, we demonstrate the formation of small aggregates; specifically, {beta}-sheet rich oligomeric aggregates, in SNCA-mutant midbrain immature neurons. The aggregation progresses over time to accumulate phosphorylated aggregates, and later fibrillar aggregates. When the midbrain neurons were functional, we observed evidence of impaired physiological calcium signalling, with raised basal calcium, and impairments in cytosolic and mitochondrial calcium efflux. Once midbrain identity fully developed, SNCA-mutant neurons exhibited bioenergetic impairments, mitochondrial dysfunction and oxidative stress. During the maturation of mDA neurons, upregulation of mitophagy and autophagy occured, and ultimately these multiple cellular stresses lead to an increase in cell death by six weeks post-differentiation. Our differentiation paradigm generates an efficient model for studying disease mechanisms in PD, and highlights that protein misfolding to generate intraneuronal oligomers is one of the earliest critical events driving disease in human neurons, rather than a late-stage hallmark of the disease.

neuroscience↗

bioRxiv2022

Structural conversion of α-synuclein at the mitochondria induces neuronal toxicity

Aggregation of -Synuclein (-Syn) drives Parkinsons disease, although the initial stages of self-assembly and structural conversion have not been captured inside neurons. We track the intracellular conformational states of -Syn utilizing a single-molecule FRET biosensor, and show that -Syn converts from its monomeric state to form two distinct oligomeric states in neurons in a concentration dependent, and sequence specific manner. 3D FRET-CLEM reveals the structural organization, and location of aggregation hotspots inside the cell. Notably multiple intracellular seeding events occur preferentially on membrane surfaces, especially mitochondrial membranes. The mitochondrial lipid, cardiolipin triggers rapid oligomerization of A53T -Syn, and cardiolipin is sequestered within aggregating lipid-protein complexes. Mitochondrial aggregates impair complex I activity and increase mitochondrial ROS generation, which accelerates the oligomerization of A53T -Syn, and ultimately causes permeabilization of mitochondrial membranes, and cell death. Patient iPSC derived neurons harboring A53T mutations exhibit accelerated oligomerization that is dependent on mitochondrial ROS, early mitochondrial permeabilization and neuronal death. Our study highlights a mechanism of de novo oligomerization at the mitochondria and its induction of neuronal toxicity.

neuroscience↗

bioRxiv2022

Deep learning in human neurons predicts mechanistic subtypes of Parkinson's

Parkinsons disease (PD) is a common, devastating, and incurable neurodegenerative disorder. Several molecular mechanisms have been proposed to drive PD, with genetic and pathological evidence pointing towards aberrant protein homeostasis and mitochondrial dysfunction. PD is clinically highly heterogeneous, it is likely that different mechanisms underlie the pathology in different individuals, each requiring a specific targeted treatment. Recent advances in stem cell technology and fluorescent live-cell imaging have enabled the generation of patient-derived neurons with different mechanistic subtypes of PD. Here, we performed multi-dimensional fluorescent labelling of organelles in iPSC-derived neurons, in healthy control cells, and in four different disease subclasses. We generated a machine learning-based model that can simultaneously predict the presence of disease, and its primary mechanistic subtype. We independently trained a series of classifiers using both quantitative single-cell fluorescence variables and images to build deep neural networks. Quantitative cellular profile-based classifiers achieve an accuracy of 82%, whilst image based deep neural networks predict control, and four distinct disease subtypes with an accuracy of 95%. The classifiers achieve their accuracy across all subtypes primarily utilizing the organellar features of the mitochondria, with additional contribution of the lysosomes, confirming their biological importance in PD. Taken together, we show that machine learning approaches applied to patient-derived cells are able to predict disease subtypes, demonstrating that this approach may be used to guide personalized treatment approaches in the future.

cell biology↗

bioRxiv2020

Highly enriched hiPSC-derived midbrain dopaminergic neurons robustly models Parkinson's disease.

The development of human induced pluripotent stem cells (hiPSC) has greatly aided our ability to model neurodegenerative diseases. However, generation of midbrain dopaminergic (mDA) neurons is a major challenge and protocols are variable. Here, we developed a method to differentiate hiPSCs into enriched populations (>80%) of mDA neurons using only small molecules. We confirmed the identity of the mDA neurons using single-cell RNA-sequencing and detection of classical markers. Single-cell live imaging demonstrated neuronal calcium signalling and functional dopamine transport. Electrophysiology measures highlighted the ability to form synapses and networks in culture. Patient-specific hiPSC lines differentiated to produce functional mDA neurons that exhibit the hallmarks of synucleinopathy including: aggregate formation, oxidative stress as well as mitochondrial dysfunction and impaired lysosomal dynamics. In summary, we establish a robust differentiation paradigm to generate enriched mDA neurons from hiPSCs, which can be used to faithfully model key aspects of Parkinsons disease (PD), providing the potential to further elucidate molecular mechanisms contributing to disease development.

neuroscience↗