University of Michigan researchers have come up with complementary sequencing-based methods — known as Bru-Seq and BruChase-Seq, respectively — that make it possible to track not only the formation of new RNA molecules, but also their stability within the cell.
The techniques involve replacing one of the four building blocks of RNA — uridine — with another form of the base, called bromouridine.
For the Bru-Seq step, the researchers let cells carry out their transcriptional duties using bromouridine rather than uridine for a spell. Then, by targeting only bromouridine-containing RNAs for deep sequencing, they are able to look specifically at transcripts produced during the labeling period.
By removing the bromouridine and adding back uridine — the strategy behind BruChase-Seq — the team can also see the fraction of newly formed transcripts that remain after a given stretch of time.
"If you take a cell line and use our techniques to kind of describe it, you get a list of all the 20,000 genes ranked in order of their rate of RNA synthesis — how frequently each gene is transcribed," Mats Ljungman, a radiation oncology and environmental health sciences researcher at the University of Michigan, told In Sequence.
"[With BruChase-Seq], then you also get a list of all the relative stabilities of these transcripts," he continued.
Together, the Bru-Seq and BruChase-Seq datasets make it possible to work out the changing RNA synthesis and stability profiles of different cell types or in cells subjected to various mutations, treatments, or environmental exposures.
In a proof-of-principle study published online recently in the Proceedings of the National Academy of Sciences, for instance, Ljungman and his colleagues demonstrated the utility of Bru-Seq and BruChase-Seq approaches for tracing RNA synthesis and stability in human fibroblast cells before and after treatment with tumor necrosis factor — a cytokine that produces a fairly well-characterized inflammatory response.
That analysis indicated that the steady-state gene expression profiles previously associated with TNF exposure likely reflect a complex combination of altered transcription patterns at certain genes as well as shifts in the relative stability of specific transcripts.
"We applied the techniques to something that is well known to have a huge effect on gene expression," Ljungman explained. "We're really sorting out the transcription part and the stability part of this response."
Indeed, the methods themselves were motivated by an interest in understanding not only steady-state transcript levels — the sort of information that can be gleaned from standard RNA-sequencing or microarray-based gene expression analyses — but also the transcriptional dynamics influencing this overall RNA content.
"It's the equilibrium between synthesis and stability that regulates the steady-state RNA," Ljungman said. "We were interested in looking at those two terms of the equation."
On the synthesis side, he explained, the team has taken to tracking freshly made RNA by "feeding" bromouridine to living cells for around half an hour. Because bromouridine is specifically incorporated into newly produced RNA molecules, this treatment tags transcripts being synthesized during that time frame.
These molecules can then be snatched out of a total RNA solution with bromouridine-specific antibodies, so that only nascent RNAs are interrogated in the subsequent deep sequencing step.
"We isolate the Bru-containing RNA and then we throw away all the older RNA that's not labeled," Ljungman said. "We want to see, 'What is the cell synthesizing right now?'"
Generally speaking, the type of information obtained by Bru-Seq is similar to that found with other sequencing-based methods developed to measure newly formed, or 'nascent,' RNA levels, he explained, such as global run-on sequencing, or GRO-Seq (see GWDN 12/4/2008), native elongating transcript sequencing, also called NET-Seq, or nascent RNA sequencing (Nascent-Seq).
Where the team's strategy differs, though, is in the application of an additional step, BruChase-Seq, which offers a look at not only the newly made RNA transcriptome, but also the RNA stabilome — the set of RNAs that stick around over a given time frame.
With BruChase-Seq, Ljungman said, "we basically can age this population of RNA that we labeled with bromouridine."
In their PNAS study, for instance, the researchers applied Bru-Seq and BruChase-Seq to human fibroblast cells both before and after stimulating the cells with TNF.
For those analyses, the team used an initial bromouridine labeling period of 30 minutes, coupled with a six-hour uridine chase, though Ljungman noted that the group has had success using shorter bromouridine exposure and chase times.
In the untreated fibroblast cells, the investigators saw transcription across roughly one-third of the genome, identifying sets of transcribed RNAs that corresponded quite well with nascent RNA patterns found through GRO-Seq.
Generally speaking, the representation of gene exons compared to introns jumped after a six-hour uridine chase, consistent with the notion that transcripts destined to code for proteins are somewhat more stable than the intervening sequences that are spliced out. Still, the team reported, transcript stability also varied depending on the identity of the gene considered.
Following an inflammation-inducing TNF treatment, meanwhile, the researchers saw shifts at the transcriptome and stabilome levels, with some genes showing higher or lower expression, changes in transcript stability, or both.
On the synthesis side, for instance, they identified 472 genes with higher-than-usual expression after an hour-long TNF treatment. Another 204 genes showed muted transcription during the same time frame.
For more than 150 transcripts, TNF exposure corresponded to enhanced stability compared to the untreated fibroblast cells, while 58 transcripts had lower-than-usual stability after the treatment.
"Usually we think that if we do something and RNA [levels] go up, we automatically think that a transcription factor got activated and now there's more transcription from that gene and that's why we have more RNA," Ljungman noted.
"But this paper illustrates that it's not only transcription," he said. "A lot happens at the level of stability of the transcripts. There's another layer of regulation that cells have."
Together, the collection of transcripts exhibiting TNF-related alterations in synthesis and/or stability should serve as a resource for members of the community working on inflammation-related research, Ljungman argued, since different biological mechanisms may contribute to transcriptional and post-transcriptional arms of the inflammation response.
Along with the information generated for nuclear transcripts, study authors explained, the Bru-Seq and BruChase-Seq methods offer insights into the synthesis and stability of ribosomal and mitochondrial transcripts, too.
Since any newly formed ribosomal or mitochondrial transcripts will be present in the collection of bromouridine-labeled RNAs targeted for sequencing, the ability to track the dynamics of those transcripts depends on the ability to map the reads appropriately — something Ljungman and his colleagues did by developing ribosomal and mitochondrial reference sequences akin to "extra" chromosomes.
That analysis is just one part of a larger computational pipeline that the researchers developed to deal with the Bru-Seq and BruChase-Seq data. These computational approaches may continue to be tweaked as more and more data are produced.
For instance, Ljungman noted that the group is now generating additional time course data to get a better sense of the dynamics behind RNA synthesis and stability, as well as the kinetics of RNA splicing.
Though the experiments described in the new PNAS study were performed using Illumina's HiSeq 2000, both Bru-Seq and BruChase-Seq are expected to be compatible with any of the existing sequencing platforms. "You're just isolating RNA," Ljungman said. "And then you can go to whatever platform you're comfortable with."
The price tag for the Bru-Seq/BruChase-Seq experiments is currently on the order of $400 to $500 per sample when researchers multiplex three samples per HiSeq lane.
At the moment, Ljungman estimated that library preparation costs are coming in at around $80 to $100 per sample, though he noted that it may be possible to pare that down a bit by scaling up the process.
Because the approaches hinge on active incorporation of bromouridine into newly formed transcripts, he cautioned that both Bru-Seq and BruChase-Seq are limited to cells that can be grown or cultured in the lab.
Nevertheless, the researchers have had some success in preliminary experiments using the techniques to test fresh human blood samples and eventually hope to apply the approaches to understanding RNA synthesis and stability in a clinical context.
For the time being, though, they are focusing on a combination of validation experiments and studies exploring a broad range of basic research questions. In particular, Ljungman said the group is interested in applying the methods to understanding everything from cellular differentiation and aging to cancer and DNA damage/stress response.
The researchers are also keen to see what they can learn about non-coding RNAs and microRNAs using RNA synthesis and stability data gleaned from Bru-Seq and BruChase-Seq experiments.
"We see all the primary transcripts for miRNAs. So that's another project we have in mind, is to start mapping and annotating the transcripts for miRNA," Ljungman said.