NEW YORK (GenomeWeb News) – Researchers from the University of Cambridge and Applied Biosystems have published their work on single cell, whole transcriptome analysis using messenger RNA-sequencing. The paper appeared online in Nature Methods today.
By modifying a single-cell DNA amplification approach already in use, the team came up with a way to do digital gene expression analysis in a single cell using ABI SOLiD sequencing. When they applied this method to single mouse blastomeres, the researchers found that they could measure expression of almost all of the genes detected by microarray — along with nearly 5,200 genes not detected with microarrays. The method also turned up almost 1,800 previously undetected splice junctions.
Using the same approach, the researchers found thousands of genes that were up- or down-regulated in mouse eggs lacking either Dicer or Ago2, components of the microRNA processing machinery.
"This single-cell mRNA-Seq assay will greatly enhance our ability to analyze transcriptome complexity in individual cells during mammalian development, especially for early embryonic development and for stem cells, which are usually rare cell populations in vivo," senior author Azim Surani, a researcher at the University of Cambridge, and his co-authors wrote.
High-throughput transcriptome sequencing methods have been developed in the past, the authors noted, but most require relatively large amounts of RNA, which have to be collected from many cells. And such pooled analyses don't reflect what's happening in individual cells, which may be biologically relevant.
As reported in GenomeWeb Daily News sister publication In Sequence last November, Surani and his colleagues tackled this problem by selectively amplifying RNAs with poly-adenylated tails, which picks up mRNAs as well as a few non-coding RNAs. They also tweaked the way they constructed libraries to improve reproducibility before sequencing the RNAs and mapping the reads to transcripts in the mouse RefSeq database.
In the study published today, the team tested the mRNA-Seq method on individual mouse blastomeres — precursors to embryonic stem cells that were isolated at the four cell mouse embryo stage. They then mapped tens of thousands of 35 to 50 base reads to known RefSeq transcripts.
The results were very similar to those obtained when the researchers measured the expression of 80 four-cell mouse embryos pooled together with an Affymetrix array. The sequencing approach detected five or more reads for more than 94 percent of the genes identified by microarray.
But, the authors noted, the mRNA-Seq method turned up 5,270 genes that were missed by microarray, including more than a thousand transcripts for which the microarray did not have probes.
And while microarray analysis detected expression of about 400 genes not found using mRNA-Seq, the team's follow-up analysis by real-time PCR suggests many of these genes were false-positives or were only expressed in some of the pooled embryos evaluated by microarray.
"It is also possible that for some genes with low expression, their expression can be stochastically on or off in single cell, and these genes were probably not expressed in the individual cell analyzed by our mRNA-Seq assay," Surani and co-authors wrote.
Subsequent experiments revealed that mRNA-Seq was detecting expression of many genes known to play a role in embryonic development.
When the researchers honed in on their 50 base read data from single blastomeres, they found thousands of new splice junctions: about 6,700 junctions represented by at least two reads and 1,753 represented by five or more reads. In the blastomere, the researchers detected more than two different transcripts for about 19 percent of the expressed genes.
In individual mature mouse oocytes or eggs, the team found roughly 9,000 splice junctions with two or more reads and almost 2,100 with five or more reads.
Next, the team used their mRNA-Seq method to look at gene expression changes in mouse eggs lacking either Dicer or Ago2, components of the microRNA processing machinery, compared to wild type oocytes.
They found 1,696 transcripts that were up-regulated in the oocytes lacking Dicer. In the oocytes lacking Ago2, 1,553 genes were up-regulated, including 619 that were also more highly expressed in the Dicer mutant. Knocking out Dicer or Ago2 led to decreased expression of 1,571 or 1,121 genes, respectively, including 589 genes that were down-regulated in both mutants.
Although the team touted the new mRNA-Seq approach as an unbiased way to measure gene expression and find new splicing isoforms with extremely low background noise, they conceded that limitations remain.
For instance, they noted that since it employs poly(T) primers, the current mRNA-Seq approach only detects transcripts with poly(A) tails, which may miss histone and other mRNAs. They also noted that the method won't detect the 5' end of mRNAs that have more than 3,000 bases and can't distinguish between sense and antisense transcripts.