Translating the information encoded in mRNAs into proteins is one of the most basic processes in biology. The mechanism requires a machinery (i.e. ribosomes) and components (mRNA template, charged tRNAs, regulatory factors, energy) that are shared by all organisms on Earth. We’ve learned a great deal about translation over the last century. We know how it works, how it is being regulated at many levels and under varuious conditions. We know the structures of the components. We have drugs that can inhibit translation. With the emergance of next-gen sequencing, we can now perform ribosome profiling and determine exatly which mRNAs are being translated, how many ribosomes occupay each mRNA species and where these ribosomes “sit” on the mRNA, on average. New biochemical approaches like SILAC and PUNCH-P can quantifiy newly synthesized proteins & peptides. Yet, all of that information comes from population studies, typically whole cell populations. Rarely, whole transcriptome/ribosome analysis of a single cell is performed. Still, there is no dynamic information of translation, since cells are fixed and/or lysed. And there is no spatial information regarding where in the cell translation occurs (poor spatial information can be determined if cell fractionation is performed, which is never a perfect separation of organelles/regions and we are still not at the stage of single organelle sequencing).
Imaging translation in single cells is intended to provide both spatial and dynamic information on translation at the single cell and, hopefully, single mRNA molecule resolution. Recently, four papers were published (on the same day!) providing information on translation of single mRNAs. Here is a summary of these papers.
There have been several attempts at imaging translation, which I’ve discussed in the post about TRICK. TRICK was the closest anyone got to image translation of a single mRNA in live cells, but it only applies to the first round of translation. Now, four groups who used similar approches have followed the translation dynamics of single mRNAs for many rounds and for long durations (up to two hours). The four groups are those of Rob Singer (my former PI); Tim Stasevich – a new lab in Colorado state U. (in collaboration with several groups from HHMI Janelia Reseach Center); Marvin Tanenbaum (also new lab, I think, a former postoc of Ron Vale. They developed the Suntag); and Xiaowei Zhange (who develpoed STORM super-resolution imaging, among other achievments).
Here is a summary of the systems that the four groups used:
So, lets break it down:
All groups used common human adherent cell lines to establish the system. two groups also looked at localized translation in neurons.
There is a variation in the method of expression (introducing the DNA, promoters, UTRs).
Three groups used Suntag. Suntag is a repeating peptide sequence (an epitope of the yeast GCN4 protein). This multimerized peptide chain (24xrepeats) can be fused to the protein of interest. A single-chain variable fragment (scVF) antibody against this epitope, fused to GFP, is also expressed in the cell. The scFV-GFP antibodies bind the Suntag epitope very efficienty. Thus, the Suntag-ed protein will be “decorated” by 24scFV-GFP and should shine bright against the GFP background. The three groups who used Suntag realized that this also allows detection of the nascent peptide, while it is being translated, since the Suntag-antibody interaction is very fast and long-lasting. Furthermore, if the mRNA is translated by multiple ribosomes (polysome), then the signal co-localized with the mRNA (labeled by the MS2 or PP7 systems) will be stronger than that of the single, mature protein. This is indeed the case as demonstrated by all the labs.
The 4th group used the Spaghetti-monster (SM; multimerized FLAG or HA peptides) instead of the Suntag, but the basic principle is the same. Here, the anti-FLAG antibody fragment (Fab), labeled by a small fluorescent dye (e.g. Cy3) was not expressed in the cell, but exogenously loaded on the cells. I was surprised to see that the MS2-coat protein (fused to Halotag and pre-labeled with a fluorescent dye) was also exogenously loaded onto the cells. On the one hand, this may seem less elegant than the Suntag system where all components are in the cell and there is no need to protein purification, for instance. But there are a few advantages here: 1. the SM tag is smaller than the Suntag. 2. the Cy3-labeled Fab is also smaller, I think, than the scFV-GFP fusion protein. Thus, this system might be less disruptive to the translation process and the labeled protein. 3. One can easily control the levels of the Fab, and it is probably more uniform throughout the cell population than expressing it in the cells. 4. Greater variability for imaging – no need to engineer a new scFV construct for each color (and then infect cells. Then select for cells with appropriate expression levels). The Fab can be easily labeled by dyes of any color. 5. The ease of Fab labeling and loading allowed for dual imaging of the translation of two distinct protein tags (FLAG-SM and HA-SM) in the same cell.
Two of the groups tried to reduce the background fluorescence of the mature protein, and prevent sequestration of the scFV-GFP by the mature protein, by degrading it. One group used a protein, ODC, that has a very fast turn-over, and is actualy one of the mammalian proteins with shortests half life (~30min). The other group used an inducible degron. This allows greater versatility, but requires additional components to the system.
All groups verified that they are detecting translation events, by inhibiting translation with various drugs. This allowed also to obtain more data on translation dynamics and mRNA mobility.
One groupd tethered the mRNA to the plasma membrane (using the PP7 coat protein). This provided stability (i.e. the mRNA ddi not move much), which allowed longer imaging of single molecules to infer translation dynamics. Also, the mature protein diffused away from the membrane, thus reducing background in the area close to the mRNAs.
Two groups added ER-targeting sequences to the proteins. This allowed the groups to measure the translation rate and mRNA mobility of cytoplasmic vs. ER-tethered mRNA.
Two groups tested how some physiological conditions or unique cis elements affect translation. Thus, one group tested how translation responds to two types of stress. As expected, translation was inhibited upon stress. This allowed the researchers to analyze the signal transduction pathway that leads to this inhibition. By adding cis-elements that control the translation of a stress-response protein (ATF4), they measured the dynamics of that response under stress.
A second group looked at other cis-elements that affect translation dynamics. For instance, they found that 5-10% of the translating mRNAs remained bright even after prolonged hh treatment (a drug that inhibits initiation but allows elongating ribosomes to complete translating). The behavior of this subset of mRNAs was similar to mRNAs in cells treated with the drug 4NQO that causes a chemical damage to the mRNA. Thus, it seems that 5-10% of the mRNAs are chemically defective. Still, these mRNAs did not disapear, indicating that no-go decay machinery did not target these mRNAs. Why were those mRNAs spared from decay? Is the stall eventually resolved? How? Can we use this system to isolate the mRNAs with stalled ribosomes for molecular or biochemical analysis? – there are no answers here but these are very interesting questions. To further investigate stalled ribosomes, they introdues a strong ribosome stalling sequence from the mRNA encoding Xbp1 protein. This is a regulatory element that affects the localization of the mRNA. Indeed, this element caused ribosome pausing on the reporter mRNA. Surprisingly, most ribosomes are only briefly delayed, and only a small subset is stalled for a long time. This experiment shows, again, the power of single molecule analysis. The question remains – why only a small subset is stalled for a long time?
The same group also examined the effect of alternatively splices 5’UTR on the translation of the cell-cycle regulated gene Emi1. They found that the long form of the 5’UTR represss translation of the mRNA. Here, again, there was a small subset of mRNAs (~2%) that was heavily translated. How dothese mRNAs escape the repression? How does that affect the cell?
These findings demonstrate how such a tool is very usefull to study translation behavior at the single cell level and the effect of cis-elements under different conditions. But even more important, this allows studying the behavior of individual mRNAs, with very interesting and unforseen results. I was surprised that none of the groups looked at the translation dynamics of an IRES. This seems like a perfect tool for that. Hopefully in future studies.
The power of these systems is to measure translation dynamics of individual mRNAs and extract data, that is sumarized in the next table:
To cover the highlights:
- Not all mRNAs are being translated, and the % of translating mRNAs differ from gene to gene.
- The elongation rate is in the same order of magnitude for all 4 studies, ranging from ~3amino acids/sec to ~10aa/sec. This probably depends on the method used (e.g. Suntag vs Spaghetti monster), and the transalation efficiency of differnt sequences (e.g. length, secondary structures, codon usage, regulatory cis-elements and proteins, epi-transcriptomic modifications and more).
- Similarly, the calculated initiation rate and distance between ribosomes is similar among all 4 groups.
The papers also contain a lot of data on mRNA mobility at different scenarios. Bottom line – translation per se does not seem to affect the diffusion rate of “free” cytoplasmic mRNAs, although the mRNP is havier with all the ribosomes. However, tethering due to translation of specific elements (e.g. ER targeting) obviously affects mobility of the translated mRNA.
I could go on and on to detail more data – there’s a large collection of data in those papers, but I won’t – so just go read. And think how to utilize this system for your research.
Update 9/2016: A new paper was published, using a similar Suntag system and showing more or less similar results but at differnt scenarios. In this paper, from Bertrand’s lab, they increased the Suntag to 56 repeats, included an intron and selectable marker. Their first construct – of the Ki67 mRNA, was also labeled with MS2 loops at the UTR, but other mRNAs (POLR2A and dynein heavy chain) were not. Though similar, their result of average elongation rate was the fastest – 13.2 to 13.8 a.a. /second. They have some interesting results on localization of translation. Another nice paper.
Yan X, Hoek TA, Vale RD, & Tanenbaum ME (2016). Dynamics of Translation of Single mRNA Molecules In Vivo. Cell, 165 (4), 976-89 PMID: 27153498
Wang C, Han B, Zhou R, & Zhuang X (2016). Real-Time Imaging of Translation on Single mRNA Transcripts in Live Cells. Cell, 165 (4), 990-1001 PMID: 27153499
Morisaki, T., Lyon, K., DeLuca, K., DeLuca, J., English, B., Zhang, Z., Lavis, L., Grimm, J., Viswanathan, S., Looger, L., Lionnet, T., & Stasevich, T. (2016). Real-time quantification of single RNA translation dynamics in living cells Science DOI: 10.1126/science.aaf0899
Wu, B., Eliscovich, C., Yoon, Y., & Singer, R. (2016). Translation dynamics of single mRNAs in live cells and neurons Science DOI: 10.1126/science.aaf1084
Pichon X, Bastide A, Safieddine A, Chouaib R, Samacoits A, Basyuk E, Peter M, Mueller F, & Bertrand E (2016). Visualization of single endogenous polysomes reveals the dynamics of translation in live human cells. The Journal of cell biology PMID: 27597760