Automated and unbiased discrimination of ALS from control tissue at single cell resolution

Towards an integrated approach for understanding glia in Amyotrophic Lateral Sclerosis

Substantial advances in technology are permitting a high resolution understanding of the salience of glia, and have helped us to transcend decades of predominantly neuron-centric research. In particular, recent advances in 'omic' technologies have enabled unique insights into glial biology, shedding light on the cellular and molecular aspects of neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS). Here, we review studies using omic techniques to attempt to understand the role of glia in ALS across different model systems and post mortem tissue. We also address caveats that should be considered when interpreting such studies, and how some of these may be mitigated through either using a multi-omic approach and/or careful low throughput, high fidelity orthogonal validation with particular emphasis on functional validation. Finally, we consider emerging technologies and their potential relevance in deepening our understanding of glia in ALS.

Prediction of mechanistic subtypes of Parkinson's using patient-derived stem cell models

Parkinson's disease is a common, incurable neurodegenerative disorder that is clinically heterogeneous: it is likely that different cellular mechanisms drive the pathology in different individuals. So far it has not been possible to define the cellular mechanism underlying the neurodegenerative disease in life. We generated a machine learning-based model that can simultaneously predict the presence of disease and its primary mechanistic subtype in human neurons. We used stem cell technology to derive control or patient-derived neurons, and generated different disease subtypes through chemical induction or the presence of mutation. Multidimensional fluorescent labelling of organelles was performed in healthy control neurons and in four different disease subtypes, and both the quantitative single-cell fluorescence features and the images were used to independently train a series of classifiers to build deep neural networks. Quantitative cellular profile-based classifiers achieve an accuracy of 82%, whereas image-based deep neural networks predict control and four distinct disease subtypes with an accuracy of 95%. The machine learning-trained classifiers achieve their accuracy across all subtypes, using the organellar features of the mitochondria with the additional contribution of the lysosomes, confirming the biological importance of these pathways in Parkinson's. Altogether, we show that machine learning approaches applied to patient-derived cells are highly accurate at predicting disease subtypes, providing proof of concept that this approach may enable mechanistic stratification and precision medicine approaches in the future.

Random forest modelling demonstrates microglial and protein misfolding features to be key phenotypic markers in C9orf72-ALS

Clinical heterogeneity observed across patients with amyotrophic lateral sclerosis (ALS) is a known complicating factor in identifying potential therapeutics, even within cohorts with the same mutation, such as C9orf72 hexanucleotide repeat expansions (HREs). Thus, further understanding of pathways underlying this heterogeneity is essential for appropriate ALS trial stratification and the meaningful assessment of clinical outcomes. It has been shown that both inflammation and protein misfolding can influence ALS pathogenesis, such as the manifestation or severity of motor or cognitive symptoms. However, there has yet to be a systematic and quantitative assessment of immunohistochemical markers to interrogate the potential relevance of these pathways in an unbiased manner. To investigate this, we extensively characterised features of commonly used glial activation and protein misfolding stains in thousands of images of post-mortem tissue from a heterogeneous cohort of deeply clinically profiled patients with a C9orf72 HRE. Using a random forest model, we show that microglial staining features are the most accurate classifiers of disease status in our panel and that clinicopathological relationships exist between microglial activation status, TDP-43 pathology, and language dysfunction. Furthermore, we detected spatially resolved changes in fused in sarcoma (FUS) staining, suggesting that liquid-liquid phase shift of this aggregation-prone RNA-binding protein may be important in ALS caused by a C9orf72 HRE. Interestingly, no one feature alone significantly impacted the predictiveness of the model, indicating that the collective examination of all features, or a combination of several features, is what allows the model to be predictive. Our findings provide further support to the hypothesis of dysfunctional immune regulation and proteostasis in the pathogenesis of C9-ALS and provide a framework for digital analysis of commonly used neuropathological stains as a tool to enrich our understanding of clinicopathological relationships within and between cohorts. © 2022 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Axonal Length Determines Distinct Homeostatic Phenotypes in Human iPSC Derived Motor Neurons on a Bioengineered Platform

Stem cell-based experimental platforms for neuroscience can effectively model key mechanistic aspects of human development and disease. However, conventional culture systems often overlook the engineering constraints that cells face in vivo. This is particularly relevant for neurons covering long range connections such as spinal motor neurons (MNs). Their axons extend up to 1m in length and require a complex interplay of mechanisms to maintain cellular homeostasis. However, shorter axons in conventional cultures may not faithfully capture important aspects of their longer counterparts. Here this issue is directly addressed by establishing a bioengineered platform to assemble arrays of human axons ranging from micrometers to centimeters, which allows systematic investigation of the effects of length on human axonas for the first time. This approach reveales a link between length and metabolism in human MNs in vitro, where axons above a "threshold" size induce specific molecular adaptations in cytoskeleton composition, functional properties, local translation, and mitochondrial homeostasis. The findings specifically demonstrate the existence of a length-dependent mechanism that switches homeostatic processes within human MNs. The findings have critical implications for in vitro modeling of several neurodegenerative disorders and reinforce the importance of modeling cell shape and biophysical constraints with fidelity and precision in vitro.

Image-based deep learning reveals the responses of human motor neurons to stress and VCP-related ALS

AIMS

Although morphological attributes of cells and their substructures are recognised readouts of physiological or pathophysiological states, these have been relatively understudied in amyotrophic lateral sclerosis (ALS) research.

METHODS

In this study, we integrate multichannel fluorescence high-content microscopy data with deep learning imaging methods to reveal-directly from unsegmented images-novel neurite-associated morphological perturbations associated with (ALS-causing) VCP-mutant human motor neurons (MNs).

RESULTS

Surprisingly, we reveal that previously unrecognised disease-relevant information is withheld in broadly used and often considered 'generic' biological markers of nuclei (DAPI) and neurons ( β III-tubulin). Additionally, we identify changes within the information content of ALS-related RNA binding protein (RBP) immunofluorescence imaging that is captured in VCP-mutant MN cultures. Furthermore, by analysing MN cultures exposed to different extrinsic stressors, we show that heat stress recapitulates key aspects of ALS.

CONCLUSIONS

Our study therefore reveals disease-relevant information contained in a range of both generic and more specific fluorescent markers and establishes the use of image-based deep learning methods for rapid, automated and unbiased identification of biological hypotheses.