To celebrate the 2-year anniversary of this blog, lets talk about the new Science paper in which the authors claim to performs in situ single cell, single molecule RNA sequencing. So what’s the big deal? Well, RNA sequencing (RNA-Seq) has become a very common method to study gene expression. In many cases, RNA-Seq uses cell extraction from an entire cell population -thus averaging the RNA content of each individual cells. In recent years, single-cell RNA-Seq is becoming more feasible (for example). In this case, cells are sorted via flow cytometry so that one can sort individual cells into designated wells in a multi-well plate. Thus, RNA from a single cell can be sequenced. Though this process is becoming both efficient and accurate, you loose information about the cellular localization of the RNA. FISH is a method that enables to determine the accurate localization of the target mRNA. However, FISH is limited to only a small number of mRNAs. Using color barcoding of the FISH probes can increase the complexity that is achieved (i.e. – one can simultaneously detect multiple types of RNAs) but these are still only a few compared to the entire transcriptome. This new paper in Science combines FISH with RNA sequencing to give Fluorescent in situ sequencing (FISSEQ). How did they do that? First, they generated cDNA by performing reverse-transcription in fixed cells with random primers. The cDNA was then circularized and then amplified. One of the nucleotides has a reactive group so that the cDNA is cross-linked to the surrounding macromolecules. Thus, the cDNA is localized to the same location as the RNA. The primer has an adapter sequence that can be used as template for sequencing or for FISH. They show some 3D FISH images of cells and tissues using a probe for this adapter. Pretty pictures, but not much info there. From here on, it gets trickier. For the sequencing, they used the SOLiD method. One problem was to reduce the spot density so that it will be sufficiently low to distinguish single molecules. If I understand them correctly, they modified the SOLiD sequencing probes so that the sequencing primers have mismatches that reduce the efficiency. The second problem was to identify auto fluorescence and other fluorescent artifacts in their images. Rather than fine-tuning a specific detection threshold, they relied on the fact that they are monitoring sequences. That means that the fluorescence at each spot should account for a known sequence (i.e. the colors pattern should change based on the sequence), whereas auto fluorescence should remain stable. So, they actually used no threshold at all.
They them showed a few applications of this method. One interesting feature was to show that the nucleus is enriched for non-coding and antisense RNAs compared to the cytoplasm that is enriched with mRNAs. Unfortunately, the nuclear/cytoplasm dichotomy was the only “localization” aspect in their paper. It would have been much more interesting to show mitochondrially-localized vs ER localized vs plasma membrane localized etc… Or, look at highly polarized cells like neurons and use FISSEQ to look at somatic vs dendritic or axonic RNAs. Another experiment they did was to look at the gene expression changes in response to a wound healing model. They found the expected increase in genes related to cell migration, and with some genes differentially expressed only in the migrating cells (at the “wound”) compared to contact inhibited cells that are close to them. Again, It would have been even better if they could show the subcellular of these mRNAs – are they at focal adhesion sites or other unique sites in the migrating cells? Their results on the whole seem convincing, but I didn’t check all the little technical details. One thing which is missing is a high resolution image showing the sequencing of just one RNA, compared to an auto fluorescence spot. All their images show one or many cells with multiple spots in different colors – but that is it. The authors say upfront that this is just a demonstration of their method, and that they expect it to improve in coming years, just like what happened with next-generation sequencing. We’ll see… Lee JH, Daugharthy ER, Scheiman J, Kalhor R, Yang JL, Ferrante TC, Terry R, Jeanty SS, Li C, Amamoto R, Peters DT, Turczyk BM, Marblestone AH, Inverso SA, Bernard A, Mali P, Rios X, Aach J, & Church GM (2014). Highly multiplexed subcellular RNA sequencing in situ. Science (New York, N.Y.), 343 (6177), 1360-3 PMID: 24578530 [post-pub note: this post was also published at the RNA-Seq blog, here].
Feb 2015 – the detailed protocol was published at Nature protocols.