(I-348) Studying Temporal Changes in Autophagy and Zika Virus Replication using Image-Based Profiling

Introduction:: Autophagy is a dynamic intracellular process by which cellular constituents such as damaged organelles and misfolded proteins are degraded and recycled to maintain cellular homeostasis. Numerous stress conditions cause an upregulation in autophagy, ranging from nutrient starvation to viral infection. It is also possible to regulate autophagy using chemical drug treatments.

Several diseases have dysregulated autophagy levels, such as neurodegenerative diseases, infectious diseases, and cancer. This has driven the development of pharmacological agents to modulate autophagy as treatment options. One notable example of such a disease is congenital Zika syndrome, which comprises a range of birth defects caused by Zika virus (ZIKV) infection and linked to dysregulated autophagy. Its effect on millions of people in the 2015-2016 epidemic combined with the lack of an approved vaccine and links to birth defects made it an important research subject matter.

ZIKV infection induces autophagy, consequently resulting in an increase in the formation of autophagy proteins and vesicles. However, the challenge of autophagy measurement mainly lies in its dynamic nature, and its complexity increases when studied alongside perturbations such as virus infection. Image-based profiling provides a holistic view of cellular responses and phenotypes and can allow perturbations to be characterized. This study aims to gain insights on the interplay between ZIKV and autophagy by leveraging highly quantitative fluorescence microscopy and image-based profiling, which are tools that can measure and describe autophagy dynamics with high temporal and single-cell resolution.

Materials and Methods::

We generated a stable Huh-7 cell line expressing fluorescent protein mKate2 (a red fluorescent protein) on autophagy vesicles (Huh7-dCas9-mKate2-LC3) by lentivirus transduction. Huh-7 cells positive for mKate 2 signal were sorted as a bulk population using a cell sorter.

We seeded the Huh-7 reporter cell line on 96-well glass bottom plates and infected it with a fluorescent reporter Zika virus expressing Venus (a green fluorescent protein) for 36 hours. We imaged the cells under a Nikon ECLIPSE Ti2 wide field fluorescence microscope fitted with an incubated chamber and automated stage. Images were captured at hourly intervals on both the red and green channels. We performed the experiments in technical triplicates.

For image analysis, we generated cell masks through a custom-trained model using the Cellpose algorithm. The cell masks were used alongside the bTrack algorithm to obtain single-cell measurements. We only analyzed cells with complete tracks (present through the whole duration of experiments). We generated puncta masks for autophagy vesicles using the spot detection tool within the NIS Elements imaging software package. The generated masks were used for feature extraction. We extracted features using the skimage Python package for general features, while Haralick features and Zernike moment features were extracted using the mahotas Python package. Descriptive population statistics for features were also calculated. We applied data preprocessing practices such as discarding NaN values and standardization through median Z-scores. In total, ~900 features were extracted for each cell to build a morphological profile using this image analysis pipeline.

Results, Conclusions, and Discussions:: Through the image analysis pipeline, we characterized morphological changes in infected cells. We observed that cells infected with different MOI (multiplicity of infection) of ZIKV exhibited different morphological profiles. Overall, median GFP intensity increased, corresponding to the accumulation of ZIKV viral proteins in the system. The accumulation increased with high MOI as expected. Furthermore, the median number of autophagy vesicles was observed to be higher following infection, which increased with higher MOI. The trends of these results are consistent with literature. Single-cell information also confirms these trends in ZIKV accumulation and autophagy induction, and revealed heterogeneity in cellular responses to infection. Following infection, we also observed changes in other features such as cell area, mean cell intensity, and autophagy vesicle area. Cell area increased after ZIKV infection at all MOIs, mean intensity decreased following infection at all MOIs, while autophagy vesicle area only increased briefly at high MOI.

In the future, we will further analyze the morphological profiles through clustering techniques to identify groups of similarly behaving cells and conditions. Clustering can also help distinguish between infected and uninfected cells and cells at different stages of infection, which can provide insights on the kinetics of infection progression. Additionally, we can train machine learning models for classification using the morphological profiles to identify infected cells, which results in a high-throughput method of detecting and discovering disease phenotypes. Lastly, we can use these morphological profiles to identify potential biomarkers that may lend insight on involved mechanisms in autophagy and its role in virus replication, or biomarkers that can be used to characterize cellular responses to virus infection. This study showcases the potential of image-based profiling in acquiring deeper understanding of cellular processes and how they respond to changes in their environment.

Acknowledgements (Optional): :

References (Optional): :