Dyna-vivo-seq unveils cellular RNA dynamics during acute kidney injury via in vivo metabolic RNA labeling-based scRNA-seq

Kun Yin, Yiling Xu, Ye Guo, Zhong Zheng, Xinrui Lin, Meijuan Zhao, Dong He, Dianyi Liang, Zhi Zhu, Junhua Zheng, Shichao Lin, Jia Song, Chaoyong YangNature Communications 2024 · DOI

TL;DR

Researchers have developed Dyna-vivo-seq, a novel method that combines in vivo metabolic RNA labeling with single-cell sequencing to track gene expression dynamics in living mice. By distinguishing "new" from "old" RNA, the team discovered that kidney cells with higher metabolic turnover are significantly more susceptible to acute injury, providing a high-resolution temporal map of disease progression that standard "snapshot" sequencing misses.

What Problem Does This Paper Address?

Standard single-cell RNA sequencing (scRNA-seq) is inherently destructive; it provides a static snapshot of a cell's state at the moment of capture. While researchers use "RNA velocity" (analyzing the ratio of unspliced to spliced RNA) to infer future states, this method is often noisy and limited by the slow pace of natural splicing.

Metabolic labeling—where cells are fed a chemical tag like 4-thiouridine (4sU) that gets incorporated into newly synthesized RNA—offers a more direct way to measure dynamics. However, this has historically been restricted to in vitro cell cultures. When cells are removed from an organism and cultured, their physiology changes, meaning the "dynamics" measured in a dish may not reflect what happens in a living body. Dyna-vivo-seq bridges this gap by enabling high-sensitivity metabolic labeling directly within a living animal.

Methods

The Dyna-vivo-seq workflow consists of several innovative steps:

  1. In Vivo Labeling: Mice receive a tail-vein injection of 4sU, a biocompatible analog that rapidly diffuses into tissues (like the kidney) and integrates into nascent RNA.
  2. Single-Cell Capture: After a specific incubation period (optimized at 90 minutes), the tissue is dissociated, and cells are captured using the Well-Paired-Seq2 (WPS2) platform, a high-throughput microwell system.
  3. Chemical Conversion: Captured RNA is treated with iodoacetamide (IAA). This triggers a chemical transition where the 4sU-tagged "new" RNA exhibits a T-to-C mutation during reverse transcription.
  4. Bioinformatics: By sequencing the RNA and looking for these T-to-C substitutions, researchers can computationally separate the "New RNA" matrix from the "Old RNA" matrix for every single cell.

Datasets

The study primarily generated and analyzed new datasets:

  • Normal Mouse Kidney: A profile of 24,237 cells to establish a baseline of RNA turnover across 20 different cell clusters.
  • Acute Kidney Injury (AKI) Model: A time-course dataset of 27,590 cells following ischemia-reperfusion injury (IRI) at multiple intervals (pre-ischemia, ischemia, and various reperfusion stages).
  • Validation Data: Comparative datasets of in vitro vs. in vivo labeling to prove the superiority of the in vivo approach.
  • Public Access: Data is available via the Gene Expression Omnibus (GEO) under accession GSE253836.

Key Insights and Ideas

  • Metabolic Heterogeneity: The most striking finding was that cells of the same type (e.g., proximal tubular cells) are not metabolically identical. The researchers identified "high" and "low" metabolic labeling populations that were indistinguishable using standard total RNA counts.
  • The Vulnerability Link: Cells with "high" metabolic labeling—meaning they are churning through RNA faster—were found to be much more susceptible to injury during AKI. This suggests that a cell's baseline metabolic "speed" is a primary determinant of its resilience.
  • Superior Temporal Resolution: By focusing on "new RNA," the team identified 90 transcription factors (TFs) that responded rapidly to injury. Many of these (like Jun and Cebpb) showed significant activity changes within 5.3 hours, changes that were either delayed or completely invisible when looking at "old" or total RNA.
  • In Vivo vs. In Vitro: The paper demonstrates that in vitro labeling significantly underestimates RNA turnover (0.6% T-to-C substitution vs. 2.0% in vivo), likely due to the stress of cell dissociation and the loss of the native microenvironment.

Key Results

  • Sensitivity: Dyna-vivo-seq achieved a ~20.7% new RNA labeling ratio in vivo, providing a robust signal for dynamic analysis.
  • Biological Discovery: Identified that high-metabolic PT cells exhibit heightened activation of pathways related to RNA splicing and metal ion response, which may explain their injury susceptibility.
  • RNA Velocity Accuracy: Using metabolic labeling data produced "velocity" flows that were much more coherent and biologically plausible than standard splicing-based models, especially in capturing the transition from ischemia to reperfusion.
  • Cost-Effectiveness: The authors highlight that the workflow is accessible, costing less than $0.10 per cell for library preparation.

Limitations and Future Work

  • Labeling Window: The 90-minute window is a "sweet spot" for kidney injury, but different biological processes (like slow development or ultra-fast signaling) might require different 4sU exposure times.
  • Organ Specificity: While successful in the kidney, the diffusion and uptake of 4sU may vary in other organs (e.g., the brain due to the blood-brain barrier), requiring further optimization.
  • Spatial Integration: The authors suggest that the next frontier is combining Dyna-vivo-seq with spatial transcriptomics. Knowing which cells are changing is powerful, but knowing where they are located in the tissue architecture during those changes would provide a complete 4D map of disease.

Summary updated

Abstract

A fundamental objective of genomics is to track variations in gene expression program. While metabolic RNA labeling-based single-cell RNA sequencing offers insights into temporal biological processes, its limited applicability only to in vitro models challenges the study of in vivo gene expression dynamics. Herein, we introduce Dyna-vivo-seq, a strategy that enables time-resolved dynamic transcription profiling in vivo at the single-cell level by examining new and old RNAs. The new RNAs can offer an additional dimension to reveal cellular heterogeneity. Leveraging new RNAs, we discern two distinct high and low metabolic labeling populations among proximal tubular (PT) cells. Furthermore, we identify 90 rapidly responding transcription factors during the acute kidney injury in female mice, highlighting that high metabolic labeling PT cells exhibit heightened susceptibility to injury. Dyna-vivo-seq provides a powerful tool for the characterization of dynamic transcriptome at the single-cell level in living organism and holds great promise for biomedical applications.
BibTeX
@inproceedings{yin2024dyna,
  author = {Kun Yin and Yiling Xu and Ye Guo and Zhong Zheng and Xinrui Lin and Meijuan Zhao and Dong He and Dianyi Liang and Zhi Zhu and Junhua Zheng and Shichao Lin and Jia Song and Chaoyong Yang},
  title = {{Dyna-vivo-seq unveils cellular RNA dynamics during acute kidney injury via in vivo metabolic RNA labeling-based scRNA-seq}},
  booktitle = {Nature Communications},
  year = {2024},
  doi = {10.1038/s41467-024-54202-4},
}
Figure