Japanese researchers have come up with a simplified single-cell RNA sequencing strategy, dubbed Quartz-seq, to reproducibly document non-genetic forms of messenger RNA diversity in single cells — an approach that they described in a study appearing online in Genome Biology in April.
There, the team outlined tweaks to the whole-transcript amplification, or WTA, and mRNA to complementary DNA conversion steps that make it possible to do quantitative single-cell mRNA sequencing in fewer steps than are usually needed for poly-adenine tailing-based transcriptional sequencing.
By focusing on ways of dialing back the confounding results caused by PCR byproducts that might otherwise be mistaken for authentic heterogeneous transcripts, the study's authors explained, Quartz-seq provides a peek at the genuine transcriptional variability between cells, including cells from the same tissue.
In their current analysis, for instance, they demonstrated the feasibility of finding gene expression variability in cells from a given cell type, including mouse embryonic stem cells collected specifically from the G1 stage of the cell cycle. They also used the method to distinguish between distinct cell stages and cell types.
"The method detects different cell types and different cell-cycle phases of a single cell type," study authors wrote. "Moreover, this method can comprehensively reveal gene expression heterogeneity between single cells of the same cell type at the same cell-cycle phase."
Based on results so far, Quartz-seq also appears to compare favorably with other methods for gauging transcription in single cells, co-first authors Yohei Sasagawa and Itoshi Nikaido noted in an email message to In Sequence.
Both researchers were affiliated with the Riken Center for Developmental Biology when research for the study was performed. They are now based in the bioinformatics research unit at Riken's Advanced Center for Computing and Communication.
Past studies hint that between-cell differences in mRNA and protein profiles are relatively common, even within cells from the same tissue type.
That heterogeneity can spring up under different developmental processes or physiological conditions, Sasagawa and Nikaido noted. But it may also extend to situations in which embryonic stem cells are exposed to the same culture conditions, perhaps reflecting the ultimate fate of these cells.
Despite the potential biological relevance of non-genetic gene expression heterogeneity, though, the process has been tricky to quantify or characterize comprehensively.
For their part, Sasagawa, Nikaido, and colleagues began tackling this problem around 2009. At that time, those interested in doing transcriptome sequencing on single cells primarily turned to a whole-transcript amplification method described in Nucleic Acids Research in 2006.
That method was originally developed for doing array-based analyses of single cells, Sasagawa and Nikaido noted, and involves the addition of multiple adenine nucleotides to the initial cDNA, known as poly-A tailing.
Since then, several approaches have been proposed for interrogating transcriptomes in individual cells. For instance, the Smart-seq (CSN 7/25/2012) and STRT-seq strategies used a template switching reaction to help tag transcripts at the cDNA stage, while a multiplexed method called CEL-seq relies on in vitro transcription rather than PCR for amplification.
For their part, the Japanese team focused on finding ways to streamline the poly-A-based WTA process, optimizing the enzymes and reaction conditions used so that it can be performed using just one PCR tube per cell while curbing interference by PCR byproducts.
In particular, researchers switched up the polymerase enzyme employed during a PCR step used to produce cDNA, employing the MightyAmp DNA polymerase for that step, Sasagawa and Nikaido noted.
They also made modifications to the reverse-transcription and second-strand synthesis reactions steps, the pair added, explaining that "[o]ur simplified method enabled us to consistently obtain highly reproducible cDNA that was optimized for RNA-seq."
For the current analysis, for instance, the researchers first tried out their single Quartz-seq method using 10 picograms of DNA from mouse embryonic stem cells. From there, they looked at transcript profiles in several other cell types and cell cycle stages.
To look at gene expression fluctuations within a given cell cycle, for instance, the group did Quartz-seq on cells from two mouse embryonic stem cell sets — one comprised of a dozen cells and another of eight cells.
Within each set, individual stem cells were nabbed from the G1 phase of the cell cycle by DNA staining and a fluorescence-activated cell sorting application called FACSvantage and placed in its own PCR tube, study authors explained.
From there, they converted mRNAs into first-strand cDNA using a reverse transcription primer that was subsequently digested with an exonuclease enzyme to curb byproduct synthesis — a step introduced to duck the need for gel purification.
After adding a poly-A tail to the first-strand cDNA, the researchers did second strand-synthesis with a tagging primer before bumping up cDNA levels by PCR amplification using a suppression primer, again designed to deter the production of byproducts.
Once amplified, researchers explained, the cDNA can be fragmented, prepped, and sequenced — in this case using Illumina's HiSeq 1000 or HiSeq 2000 instruments and a paired-end sequencing approach — or funneled into microarray analyses, or Quartz-Chip, experiments.
In those single-cell experiments, the investigators reproducibly detected gene expression fluctuations across the G1 cells — results that they verified using a quantitative PCR approach that did not require amplification.
Their analyses showed that it's possible to use Quartz-seq data to discern between mouse embryonic stem cells at different stages of the cell cycle, too.
"When used in the global gene expression analysis of real single cells, the single-cell Quartz-seq approach successfully detected gene expression heterogeneity even between cells of the same cell type and/or between different cell-cycle phases," Sasagawa, Nikaido, and colleagues wrote.
"This observed gene expression heterogeneity was found to be highly reproducible in two independent experiments and can be distinguished from experimental errors, which were measured through technical replicates of pooled samples," they added. "Therefore, single-cell Quartz-seq is a useful method for the comprehensive identification and quantitative assessment of cellular heterogeneity."
Similarly, the approach produced distinct gene expression clusters when researchers compared the Quartz-seq data from mouse embryonic stem cells and from more differentiated primitive endoderm cells.
When they compared it to other single-cell transcriptome sequencing methods, meanwhile, they found that Quartz-seq edged out conventional RNA sequencing and single-cell RNA sequencing approaches when it came to quantifying single-cell mRNA patterns in a reproducible and sensitive manner.
For example, in samples containing 10 picograms of total RNA, the team reported more robust reproducibility in replicate Quartz-seq experiments than in replicate experiments done using other single-cell RNA sequencing approaches or using RNA sequencing approaches that don't rely on WTA.
The group reported a rise in sensitivity with Quartz-seq, too. With 10 picograms of total RNA from mouse embryonic stem cells, for instance, the new method reliably picked up more than 81 percent of transcripts — 7,739 of the 9,500 or so believed to make up the mRNA mixture. In contrast, the Smart-seq approach detected just over 63 percent of transcripts in the stem cells.
Similarly, the researchers detected more transcripts in Quartz-seq-based analyses of mouse embryonic stem cells or Caenorhabditis elegans cells than they did when they used CEL-seq to test comparable total RNA concentrations from the samples.
In is current iteration, the Quartz-seq protocol includes WTA and library preparation steps that are designed to be compatible with Illumina sequencing instruments, researchers noted. But the general strategy is expected to jibe with other next-generation sequencing platforms as well.
For their part, Sasagawa and Nikaido noted that the quantitative features of Quartz-seq rely on sufficient numbers of read or tag numbers, suggesting that it might also work especially well on a relatively low-cost platform such as Life Technologies' Ion Proton.
The cost per cell for the current study came in at around $130, researchers said, though they noted that that price will vary depending on the type of sequencing services or facilities available.
Of that, $5 per cell went toward WTA. Library preparation costs came in at around $25 per cell and sequencing expenses were around $100 per cell.
The ability to routinely measure transcriptome profiles in single cells could be beneficial in the clinic, researchers noted, for assessing heterogeneous cancer cell responses to treatments, for instance, or managing collections of induced pluripotent stem cells in a regenerative medicine setting.
And Quartz-seq could have more basic research applications, too. In particular, Sasagawa and Nikaido said it might find favor with those interested in assessing gene expression patterns and/or gene expression heterogeneity during carefully timed biological processes, such as embryonic development or in tissues that are subject to circadian rhythms.
It is also expected to prove useful for studies designed to characterize single-cell transcription kinetics or biological noise.
The researchers are currently fine-tuning the technique with an eye to adding in a layer of genomic information on top of the transcript profiles already being generated. They are also working with collaborators to begin applying it to studies of several different cell types and are interested in finding ways to boost the throughput of the approach while further reining in costs.