Continuous-flow stimulation for high-resolution temporal analysis of single-cell transcriptome data
Loading...
Files
Date
Authors
Publisher
Abstract
Since its first attempts more than 15 years ago, single-cell sequencing has become a powerful technique for the analyses of cellular behavior. Compared to classical bulk sequencing, a considerable advantage is the possibility to study cellular dynamics on single-cell level, for example in differentiation and signal transduction. For this, the cal-culation of pseudotime based on similarities in the expression profile between the ana-lyzed cells has become a frequently applied technique and can be used to visualize dy-namical processes in single-cell experiments.
A prototypical and well-characterized intracellular signaling pathway in the immune response is the NF-κB signaling system. In this system, it was shown that stimulus speci-ficity of the immune response is encoded by dynamical features like peak amplitude, duration of the signal, or oscillation both on protein and mRNA level. However, in almost all recent single-cell time-series experiments with stimulated cells, multiple samples are taken at specific time points during the experiment and the trajectories in between are imputed. To not depend on mathematical predictions and to increase the temporal resolution of the analyses by real data, a continuum of cells with theoretically all possi-ble stimulation times would be required. So far, no experimental approach exists for the generation of such a sample.
In this work, a new experimental setup to create such a continuum for single-cell sequencing was developed from scratch and termed µCONST (microfluidic continuous-flow stimulation). The well-described NF-κB signaling system was chosen due to its rele-vance in autoinflammatory diseases and autoimmune disorders. During the develop-ment of the µCONST setup, different approaches were tested and critical steps were optimized. Finally, using microfluidics, millions of THP-1 cells could be stimulated with TNF continuously for a total stimulation time of five hours. From these cells, a dataset with more than 18,000 cells and approximately 65,000 reads per cell could be generated after using the BD Rhapsody™ platform for encapsulation and barcoding. By fractionized sample collection and multiplexing during sequencing, information on real-time was available as reference for cellular pseudotime. Preliminary analyses showed promising findings and the potential for further investigations but also unspecific stimulus-independent effects requiring further explorations.
A high degree of biological and technical noise as well as only slight transcriptional changes between the cells prevented the application of published algorithms for pseudotime analyses and prediction of temporal trajectories. By adaption of the estab-lished bioinformatic method in parallel with the development of new mathematical ap-proaches the full potential of the µCONST dataset could be exploited in future analyses. With this knowledge, the inherent dynamics of the NF-κB signaling system and also other pathways could be investigated in more detail compared to “classical” time-series ex-periments. This might broaden our understanding of transcriptional dynamics and would open the possibility for new therapeutic approaches against autoinflammatory diseases and autoimmune disorders.
Review
Metadata
License
Except where otherwise noted, this item's license is described as Attribution 4.0 International