A team from the J. Craig Venter Institute, the Salk Institute for Biological Studies, and elsewhere has shown that it's possible to glean gene expression and transcript sequence information from individual nuclei by sequencing the miniscule amounts of messenger RNA present in the organelles.
The approach, described in the early online version of the Proceedings of the National Academy of Sciences last month, hinges on the same sorts of complementary DNA synthesis methods already used for single-cell RNA sequencing. By applying such approaches to single nuclei from mouse neural progenitor cells or cells from a mouse brain sample, researchers showed that they could generate RNA sequence profiles similar to those found in individual cells or bulk samples from the same cell types.
The team picked up some 16,000 to 24,057 transcripts per nucleus in those mouse cells, on average. And comparisons with single cell and pooled cell or nucleus sequence suggest that the method provides an accurate peek at both gene expression profiles and transcript sequences, while uncovering subtle forms of transcriptome variability from one cell to the next.
"You can do single-cell transcriptomics on just the RNA that's in the nucleus," JCVI's Roger Lasken, a co-corresponding author on the study, told In Sequence. "It's a lot less RNA … but we have, now, a very gentle method of looking at the transcriptome."
It remains to be seen whether a similar RNA sequencing method can be used to assess frozen tissue samples — for instance, samples obtained from the brains of deceased individuals. But results so far are prompting enthusiasm about applying the single nucleus RNA sequencing approach in that realm and on other samples comprised of difficult-to-dissociate cells.
By having access to transcripts from nuclei alone, he and his co-authors noted that it may also be possible to get a better sense of the RNA regulation that takes places in the nucleus, along with a look at transcripts and non-coding RNAs that are over-represented in the nucleus.
In early transcriptomic studies, researchers typically collected, pooled, and extracted RNA from whole cells, Lasken explained. But with techniques introduced in recent years, it's become possible to convert increasingly small amounts of mRNA into cDNA, making it feasible to sequence ever-more-scant samples and even individual cells.
For their current study, Lasken and his colleagues took that a step further, applying existing methods for reverse transcription and second strand cDNA synthesis to picogram or smaller amounts of mRNA found in single nuclei.
The work stemmed, in part, from attempts to do transcriptomic sequencing on individual neurons, which proved difficult due to the complicated interconnections between such brain cells.
Also complicating the process was the fact that many standard protocols for teasing apart cells rely on mechanical manipulation and/or protease enzymes that are active at temperatures in which cells remain metabolically active.
"Two bad things are happening: you're damaging the cells, because they're connected to each other, and you're also treating them with proteases at elevated temperature," Lasken noted. "So those cells are highly stressed and are likely to be responding with an altered transcriptome."
"It's a reasonable concern that you're actually altering the very transcriptome that you want to look at," he said.
In an attempt to circumvent such potential confounders, he and his colleagues worked out a scheme for interrogating transcript patterns in single nuclei obtained by breaking down the cell membrane with Triton and grabbing the organelle by flow cytometry or micromanipulation.
Because the protocol does not require harsh treatment or elevated temperatures, Lasken noted, it should be applicable to nuclei nabbed from tissues that have been left on ice, minimizing sample preparation-related gene expression changes.
For the current study, the team first tried its hand at using quantitative PCR to look at the levels of mRNAs from select genes in eight flow cytometry-sorted single nuclei — and as many whole cells — originating from a fluorescently labeled mouse neural progenitor line.
After showing that detectable amounts of mRNA were present in the individual nuclei (albeit at slightly lower levels than in the single cells), the researchers took a crack at sequencing and comparing cDNA libraries made for three single mouse neural progenitor cells and three single nuclei isolated by micromanipulation.
In the process, they generated some 47 million Illumina reads per nucleus, on average, and 56 million reads for each single cell. The average number of distinct transcripts detected in the nuclei was more than 20,200 apiece, compared to the 20,065 transcripts identified from each individual cell, on average.
The transcript sets seen in the nuclei represented nearly 80 percent of the protein-coding transcripts documented in the NCBI's mouse reference sequence database, study authors noted.
Moreover, the range of expression levels described for specific single nucleus transcripts were on par with those found in whole cells or in pools of cells or nuclei from that cell type.
Even so, the single nucleus and single-cell sequence data both revealed transcriptional variability that was obscured when looking at pooled samples, highlighting the biological differences that exist between different cells within a given tissue.
When they generated 123 million sequence reads from cDNA libraries prepared with mRNA from two nuclei nabbed directly from mouse brain tissue, the investigators found transcriptome patterns that roughly corresponded with those in bulk samples from the same hippocampal region, though subtle differences were detected there as well.
In the present analysis, the team estimated that its sequencing method picked up transcripts that are present in between five and more than 25,000 copies in a given nucleus. Still, further analysis is needed to nail down the precise dynamic range of the approach, Lasken said.
He noted that the team is also considering whether the strategies used for isolating individual nuclei are ideal for future iterations of the RNA sequencing approach. For instance, additional research is needed to determine whether there are sample prep-related artifacts in the transcript data such as transcript leakage through the nuclear membrane that may affect transcript ratios detected.
Transcriptional variation stemming from experimental artifacts may be tough to ascertain with existing methods, Lasken added, particularly in light of the authentic biological variation that exists between cells from the same tissue types.
Nevertheless, results so far suggest that the transcript concentrations detected in individual nuclei are consistent with those observed in whole cells, both in terms of the extent of variation detected and the identity of genes showing high or low expression levels.
The researchers have performed their single nucleus RNA sequencing protocol with both SOLiD and Illumina sequencing instruments so far and believe it will be compatible with other sequencing technologies as well.
The approach does not seem to significantly bump up the time or cost investment compared to single-cell RNA sequencing, according to Lasken. On the contrary, he noted that there might actually be a slight time savings when preparing samples from the brain or other tissues that are tricky to tease into individual cells.
"In the case of brain tissue, it's actually quite a bit simpler, in terms of work and time, to [sequence] nuclei," he noted. "Our method is less work in terms of just lysing the cells and isolating the nuclei."
As such, he and his colleagues are gearing up to try out the single nucleus sequencing method on frozen tissue post-mortem. More generally, Lasken said that the approach should be applicable in any research application where individual gene expression or transcript sequence data is sought from individual cells that are highly interconnected and/or are difficult to isolate, including some plant samples.