Live seq: Live-cell transcriptome analysis 
with single-cell cytoplasmic biopsies

In the study by Chen. et al. (2022), the authors showed that biopsies represent reliable transcriptional snapshots of a single cell. [2] In the publication, they demonstrated that a single cell transcriptome can be obtained from a biopsy, and that, by comparing this to conventional scRNA-seq methods, it had minimal effect on the recorded transcriptome. [2] In a next step, they showed that, with sequential sampling on the same cell, Live seq can provide temporal readouts, of both rapid and slow cellular state transitions, and that this enables detection of minor differential expression events within populations. All this because biopsies keep the cells alive. This crucial advancement in temporal transcriptomics is supported by four key experiments from the work of Chen. et al. (2022) that are outlined below. [2]

To establish Live-seq as a tool for time resolved gene expression analysis, as a first step, the authors set out to show that the method for taking a biopsy preserves cell viability without disruption of the physiology of the cell studied. To start off, the viability after taking a biopsy was assessed for all the cell types used: primary mouse adipose stem and progenitor cells (ASPCs) before and after differentiation, IBA (Interscapular Brown Adipose) cells (fibroblasts) and from quiescent and LPS stimulated macrophage-like RAW264.7 cells. They observed consistent results across three cell types, with viability percentages falling between 85% to 89% for extraction volumes ranging between 1.2 pL to 5.0 pL (mean values) (Table 1). The fact that a cell is resilient enough to have a small biopsy taken from its cytoplasm was already shown by Guillaume-Gentil et al. 2016. [1] In this paper, the authors examined the cell’s survival as a function of the cell mean extracted volume. They showed that after taking up to 4.0 pL from the cytoplasm of HeLa cells, 82% remained viable for over 5 days and behaved similarly compared to non-extracted neighborin cells. (Table 1) In this case, the authors hypothesized that the observed lethality of the method likely resulted from the high extraction volume that represented a large proportion of the cells mean volume of 4.4pL. In addition, they showed that a HeLa cell following the extraction of 2.9 pL, divided normally after ~30 hours, similarly to its non-extracted neighboring cells. [1]

The recovery behavior of cells after extraction is important when measuring (short) cell state transitions. Therefore, the authors further corroborated on the impact that a biopsy had on macrophage-like RAW264.7 cells. Using time-lapse microscopy, they monitored the growth dynamics of RAW cells after sampling. They discovered that the extracted RAW cells regained their initial volume within 100 to 320 minutes from the time that the biopsy was taken. They showed also that the cells continued to exhibit growth patterns similar to their non-biopsied counterparts. These findings confirm that cells can swiftly recover and continue to proliferate after undergoing a cytoplasmic biopsy.

To understand cellular identities, Live-seq was used to analyze full-length transcripts across various cell types and states. Biopsies and the corresponding transcriptomes were collected from different cell types: primary mouse adipose stem and progenitor cells (ASPCs) before and after differentiation, IBA (Interscapular Brown Adipose) cells (fibroblasts) and from quiescent and LPS stimulated macrophage-like RAW264.7 cells (Figure 2a). t-SNE analysis of the obtained transcriptomes revealed five distinct clusters, reflecting the cells’ identities and activation states. This demonstrated the prowess of Live seq to distinguish cell types and states. To further investigate these cellular clusters, the researchers conducted a comprehensive gene expression analysis and Gene Ontology enrichment. Their findings supported the notion that each cluster's distinct characteristics aligned with the inherent features of the cells it represented.  

To validate the reliability of Live-seq, gene expression profiles from cytoplasmic biopsies were compared with those obtained from whole-cell Smart-seq2 assays and integrated using t-SNE plots (Figure 2b, 2c). Although conventional scRNA-seq resulted in a higher average detection of genes per cell, the integration of both data sets revealed a strong correlation between cells with similar types and states (Figure 2f). Ultimately, the researchers were able to confirm the accuracy of cell type and state classifications, regardless of the sampling technique. Remarkably, the Live seq approach offered the added advantage of preserving the cells' viability, eliminating the need for sacrificing them in the process.

These findings revealed that, for both rapid and slow transition models, Live-seq data can be harnessed to accurately track the true trajectory of cells and capture dynamic, transcriptomic changes of individual cells in real-time. 

Circling back to the LPS-activation experiment where we started at, the final experiment intended to connect the molecular states of individual RAW cells with their downstream response to LPS. This is particularly important because macrophages, including RAW cells, display a heterogeneous reaction to LPS that remains unexplained. In response to LPS, the macrophages express TNF-alpha, which was used as a proxy to assess activation level during this experiment. For this, RAW cells were used and expressed mCherry fused to a TNF-alpha promoter. In that way, highly activated macrophages could be identified from their level of mCherry expression. The typical heterogeneity in the macrophage population was already evident from the variation in the mCherry expression in the ground state (Figure 4).

First, the ground-state transcriptomes of individual RAW cells were recorded. Then, after exposing these same cells to LPS, mCherry expression was measured. With this, they set out to measure both the dynamics and amplitude to which an individual cell responded to LPS activation. One of the striking findings was the anticorrelation of the gene Nfkbia-Tnf, a key negative regulator of the LPS-NF-κB signaling pathway by suppressing NF-κB and thus activation in macrophages (Figure 5a). This finding could have only been possible with Live-seq’s capacity to measure both ground-state gene expression and downstream phenotypic response. In contrast, conventional “endpoint” scRNA-seq data revealed a positive correlation between Nfkbia and Tnf-mCherry, likely because of averaging data in a highly heterogeneous cell population (Figure 5b).

[1] O. Guillaume-Gentil, R.V. Grindberg, R. Kooger, L. Dorwling-Carter, V. Martinez, D. Ossola, M. Pilhofer, T. Zambelli & J.A. Vorholt. Tunable Single-Cell Extraction for Molecular Analyses. (Jul 2016) Cell, 166(2), 506-516. doi: 10.1016/j.cell.2016.06.025.

[2] W. Chen, O. Guillaume-Gentil, P. Yde Rainer, C. G. Gäbelein, W. Saelens, V. Gardeaux, A. Klaeger, R. Dainese, M. Zachara, T. Zambelli, J. A. Vorholt & B. Deplancke. Live-seq enables temporal transcriptomic recording of single cells. (Aug 2022) Nature, doi:10.1038/s41586-022-050