Skip to main content
Premium Trial:

Request an Annual Quote

RNA Timestamps Method Clocks Age of Individual RNAs in Single-Cell Experiments


NEW YORK – A new method can provide data on the age of individual transcripts evaluated in single-cell RNA sequencing experiments.

RNA timestamps, developed by researchers at the Broad Institute and the Massachusetts Institute of Technology, uses RNA editing to tag transcripts with a reporter motif that recruits a protein fusion that causes adenosine to inosine edits to accumulate over time. The method, described Monday in Nature Biotechnology, offers hour-scale accuracy on the age of the RNA, the authors said.

"By combining observations of multiple timestamped RNAs driven by the same promoter, we can determine when the promoter was active," said Fei Chen, a core faculty member of the Broad and a senior author on the paper.

RNA timestamps require an engineered human adenosine deaminase and are therefore limited in their use to genetically modified cell lines and organisms. Chen said his team is now working on developing a mouse model whose cells include the editing constructs. Also, the method targets specific genes and their products, not the entire transcriptome. It works with both plate- and droplet-based single-cell sequencing methods, including 10x Genomics' Chromium platform, offering 1-hour resolution over a timeframe of 12 to 16 hours.

RNA timestamping is a "clever strategy," said Hao Wu, a researcher at the University of Pennsylvania who has developed his own method for studying gene expression dynamics and who was not involved in the study.

"The major advance that this method brings to the table is, it will be able to continuously record the messenger RNA age, whereas previous technology, including metabolic labelling or RNA velocity alone, can only give you one time point or discrete temporal information," he said. "Cellular age will not be reflected, but it might be useful for looking at some cells responding to external stimuli."

The RNA timestamp project started in 2017, but Chen said he drew on his experience as a grad student in Ed Boyden's lab at MIT, which he collaborated with for this paper. "For a long time, I and many others had been interested in building a molecular recorder that encoded neuronal dynamics at the millisecond level into DNA or RNA," Chen said. "That is still of interest to many of us, but that's a very challenging problem. Along the way, we realized you could encode the dynamics of transcription in this way. And actually, it was very clear from the first couple experiments that you could make something like this work."

The method makes use of an engineered human adenosine deaminase acting on RNA 2 catalytic domain (ADAR2cd) that is fused to the MS2 capsid protein. The enzyme targets RNAs tagged with MS2 binding sites with high specificity and edits adenosine-rich sequences. "Edits in this region can subsequently be identified as A-to-G mutations in high-throughput sequencing of the timestamps," the authors wrote. The final enzyme used was selected from several options to edit on the order of hours.

The team showed proof-of-concept in a HEK293T cell line expressing the editing construct and the timestamp was placed in the 3' untranslated region. To pair RNA timestamps with 10x Genomics' single-cell transcriptomics assays, the researchers developed a way to use dial-out PCR, a tagging-based method to retrieve DNA molecules with desired sequences, to associate the stamp with the 10x barcode. "It's roughly similar to 10x's feature barcoding technology," a method for running other assays in parallel to 10x's single-cell gene expression, Chen said.

RNA timestamps joins several methods available to study gene expression dynamics in single cells, including single-cell thiol-linked alkylation for the metabolic sequencing of RNA (scSLAM-seq), new transcriptome alkylation-dependent single-cell RNA sequencing (NASC-seq), 5-ethynyl-uridine in single cells sequencing (scEU-seq), sci-fate, and single-cell metabolically labeled new RNA tagging sequencing (scNT-seq).

Chen said the method also follows in the tradition of gene editing-based cell reporter systems, such as those developed by MIT synthetic biology researcher Tim Lu using CRISPR-Cas9.

Analyzing the transcriptional dynamics of specific promoters is a particular application mentioned in the paper. The authors developed an algorithm that can analyze the resulting RNAs from a promoter and infer "the transcriptional program that is most consistent with a particular set of timestamps," they wrote.

"You could imagine doing timestamps at very high throughput if you had a promoter library," Chen said. "You could reconstruct the dynamics of many promoters, not on a single-cell level, but in a pool."

He added that his lab is still interested in using the method to study neuronal activity, especially the transcription of early genes. Wu suggested that transcription patterns following pathogen-host interactions were another event that could be studied with the method.

Chen said the costs of doing RNA timestamps was dominated by the single-cell preparation and sequencing costs, with the cost of introducing or activating the recorder adding on the order of cents per cell, when measuring thousands of cells.

In addition to building a mouse carrying the RNA timestamp machinery, Chen said his lab is looking for ways to extend the recording window beyond 12 to 16 hours. Because the edits accumulated, after that amount of time the sites become saturated, "so you start to lose temporal resolution."

Addressing this issue could be key for RNA timestamping to gain traction. "One thing we always hear people ask is if we can make the recording window longer," Chen said.