[
International Worm Meeting,
2017]
C. elegans has been fundamental to our current understanding of aging and the molecular mechanisms that play a role in determining healthspan and longevity. Aging studies, however, pose technical challenges that are typically solved by labor-intensive picking and transfer of populations, or by the use of reproduction-inhibiting agents. In this work, we are developing microfluidic platforms that allow performing aging studies in the absence of reproduction-inhibitng agents, and with far less manual involvement. Our platforms are focused on two main areas which have not been fully develepoed due to these limitations. First, we have developed a platform that allows for lifelong longitudinal, high-resolution (sub-cellular) phenotyping of C. elegans populations. This platform allows tracking subcellular processes in the same population for hundreds of animals, and we are currently aiming to quantify the morphological changes that neurons and synapses undergo during the aging process in a variety of conditions. In our second platform, our goal is to perform genetic screens in aged individuals (i.e., post-reproduction). We aim to identify mutations that result in "late-onset" phenotypes, which are only exhibited late in life. Our platform allows monitoring of individual mutagenized animals, cultured in individual microfluidic chambers, while isolating their progeny by interfacing with a deep-well system. Our platforms dedicated to challenging aging studies can result in the identification of genes and morphological changes associated with the aging process under natural conditions.
[
International Worm Meeting,
2019]
Aging is known to trigger decline in neuronal function and morphology. Age-induced changes in PVD neuron, a multidenderic sensory neuron responsible for mechanosensation and thermosensation have been extensively characterized in C. elegans. Some hallmarks of aging in PVD neurons include disorganized and dislocated dendrites, as well as neurite beading and protrusions. In a majority of these aging studies, the analysis has been limited to qualitative assessments of these phenotypes. However, considering the intricate nature of PVD neurons, qualitative evaluation of images is not sufficient to track complex age-induced changes. Previous work has identified neurite beading at old age, which is affected by expression of antimicrobial peptides. In this work, we sought to quantitatively analyze the intricate aging phenotypes of neurite beading in PVD using computer-assisted quantitative analysis. Due to the complexity of these phenotypes, traditional image processing techniques lack the power for accurate segmentation. To tackle this problem, we sought to integrate state of art machine learning techniques to conduct semantic segmentation using a Convolutional Neuronal Network algorithm. This approach enabled us to perform high-throughput, unbiased, and accurate image segmentation. To train this model, 19 ground truth masks with more than 2000 positive objects were created from images acquired throughout confocal fluorescence microscopy of a population's lifespan. The accuracy of the model (True positive/ (True Positive+ False Negative)) was ~ 91.5% based on validation using 12 test images. The output of the model was then used to extract and characterize metrics such as number of beads, mean bead surface area, bead localization, and mean distance between beads. In this work we were able to address the main issues of qualitative assessment in studying age-induced phenotypes in PVD neuron by integrating cutting edge semantic segmentation using a Convolutional Neuronal Network. By integrating this tool, we were be able to quantify the dynamic morphological changes in an aging population, which revealed interesting bead formation and disappearance patterns. Using this novel method for deep phenotyping complex aging morphologies, we will also be able to characterize the neurodegenerative effects induced by other factors such as acute cold shock, oxidative stress, dietary restriction, and genetic manipulation on PVD neurite beading process and compare the phenotypic profiles caused by each of these stressors.