The advent of RNA sequencing has enabled researchers to probe the transcriptome and capture a snapshot of life in a multicellular sample. Although many insights have been gained from unravelling these patterns of transcription, the resolution of these findings has left many questions unanswered. For example: when a gene’s expression is measured, does this reflect a constant level in all cells or is it an average of a wide range of expression levels? (Genome Biology reported from a recent single-cell analysis conference that 99 percent of cells are transcription poor and one percent are transcription rich at any one point in time).
Single-cell transcriptomics promises to answer these questions by examining gene expression in single cells, rather than across a bulk population. Rapid progress is currently being made in methods for single-cell transcriptomic analyses, making this approach ever more powerful and affordable. In a Research Highlight in Genome Biology, Itai Yanai from the Technion-Israel Institute of Technology, Israel, and colleagues discuss two new methods for single-cell transcriptomics whose novelty lies in their ability to probe transcripts within their native environments. These methods progress transcriptomics yet another step toward a more faithful capturing of biology’s complexity – not only can we now profile the transcriptome within single cells, we can do so in situ.
Here Yanai expands on his Research Highlight’s discussion of single-cell transcriptomics, and shares his thoughts on what exciting avenues of research can now be explored with the two new in situ methods.
Why has single-cell transcriptomics recently energized the genomics community?
Single-cell transcriptomics has actually been around for many years, but it wasn’t very practical economically and there were concerns about its accuracy. With the widespread use of high-throughput sequencing, both of these obstacles have been cleared – making this approach ready for prime time. It’s now possible to affordably examine a population of cells and characterize its cellular heterogeneity, allowing for the detection of very rare cells whose pooling would have been virtually impossible.
Recent years have seen a number of advances in methods for single-cell transcriptomics, including your own contribution of CEL-Seq. Do you think we are still at the technology development stage or is sequencing cost now the strongest prohibitive factor?
At the moment there is no one single-cell transcriptomics method that answers all of the available features: our CEL-Seq method is multiplexed but sequences only the 3’ ends of transcripts, so it is limited in providing information on alternative splicing.
There is still definitely room for new methods. For example, current methods all depend upon an initial reverse transcription step which turns out to be inefficient, leaving many transcripts undetected.
The Fluidigm C1 instrument is enabling a way to automate single-cell procedures and it will be very exciting to see how it facilitates new methods.
What is clear though is that the actual sequencing step, while not cheap, is not the most expensive step!
The new methods described in your Research Highlight, TIVA and FISSEQ, allow the high-throughput study of transcripts in their natural environment. Are there any new avenues of research now possible that you think might prove particularly interesting?
For embryologists, these methods will prove incredibly useful. For example, in my lab we previously studied the temporal program of gene expression in the C. elegans embryo (Dev Cell. 2012, May 15, 22(5):1101-8). Using TIVA, individual cells can now be selected for transcriptome analysis. If this were repeated for all cells, it would enable the definition of the transcriptome of all cells throughout the development of an animal.
FISSEQ is incredible and promises to solve most if not all of the problems of single-cell sequencing.
How easy do you think it will be for other labs to implement TIVA and/or FISSEQ?
The FISSEQ paper has made a very good proof-of-principle argument for in situ RNA-seq. However, it seems that extensive expertise will be required for any lab to learn this method (understatement!). Therefore the future for FISSEQ may instead unfold as a commercial venture that performs FISSEQ using a dedicated instrument.
TIVA, on the other hand, seems to have a far less steep learning curve. But it does require a lot of hands on work, as it is not multiplexed: one slide is required for each cell. It will be interesting to see whether cells can somehow be barcoded as part of the TIVA method to enable multiplexed libraries for sequencing.
FISSEQ uses the SOLiD sequencing platform, however a recent summary of state-of-the-art sequencing in Biome noted the demise of SOLiD. Do you think in situ transcriptome sequencing might herald renewed interest in SOLiD’s ‘color space’ sequencing technology?
Regarding a SOLiD comeback, it seems it’s too early to tell. In principle, FISSEQ should be adaptable to work on the Illumina platform. However, if it’s already up and running for SOLiD and can be made to work in other labs, then this may indeed be an important unique ability for SOLiD and herald a twist of fate for Illumina.
Your Research Highlight notes a transcriptomic heterogeneity within cell populations that has been hinted at in single-cell studies. However RNA-seq is generally only performed on the same individual cell once, not over time. Do you think transcriptomic heterogeneity within a population is merely sampling seen within any given cell throughout its lifetime, or does it point to intrinsic differences between cells?
This is a really interesting question and there are methods now in place to test it! Imagine taking a population of cells and identifying the transcriptome of each cell. A principal components analysis (PCA) would then potentially identify the precise stage of each cell in its cell cycle. With this known, you could then first identify any cell-cycle related changes in gene expression.
Also, you could take cells from the same stage and ask how much heterogeneity at that point. Thus, without looking at different time points, you could actually deconvolve the time perspective out of it. Another more traditional approach would be to sample cells from populations at different times and identify the population heterogeneity at each time.
I think the bottom line is that while we don’t have the answers to these exciting questions yet, it is very much plausible to address them now.
Transcriptome studies of cell populations have examined how RNA abundances and isoform choices relate to the genome and epigenome. How soon do you expect we will see advances in these areas at the level of single cells, and what impact do you think this will have on the wider field of biology?
Extending single-cell methods beyond just the transcriptome is indeed an area of a lot of progress. One field that seems like it is about to blossom is single-cell HiC, where methods are well in the works to map the structure of the genome in individual cells. My guess is that we’ll see the first wave of these methods very soon and then a refining of them with time. A brave new world!
Questions from Naomi Attar (@naomiattar), Senior Editor for Genome Biology.