Roy Parker recently sent a “Letter to the Editor“, published in RNA journal, in which he suggested that the MS2 system might not be best suited for live imaging of mRNA in budding yeast. According to Parker, the MS2 system inhibits the function of Xrn1, the major cytoplasmic 5′ to 3′ RNA exonuclease in budding yeast, causing us to image mostly the remaining 3’UTR fragments. Thus, he claims, it is possible that interpertation of mRNA localization data using this system in yeast can be faulty. We wrote a response to his letter which just opened the debate even further.
But lets start with his Letter:
The Letter presents Northern analysis of MS2-tagged mRNAs in yeast. Briefly, they tested 4 mRNA species: QCR8, PGK1, MFA2 and ASH1 mRNAs, which were tagged by MS2 loops in the 3’UTR. Their Northern analysis, of steady-state mRNA levels, show the accumulation of what appears to be a smear of 3’UTR fragments, in cells that express the MS2 coat protein (MCP) fused to GFP (actually, 3xGFP), but not in cells which do not express MCP-GFP or in cells where Xrn1 protein is absent. They also claim that the U1A-system does not inhibit Xrn1 processivity and therefore can be used as an alternative.
Generally, it does look as if Xrn1 is being arrested at the loops, causing accumulation of fragments. Their results indicate what could be a real problem for RNA imaging in yeast.
But there are several problems with this paper. I will not go into details of all the problems – we summarized most of them in our response. Parker, in the commentary that he wrote in reponse to our response, waved most of the issues we raised as “irrelevant”. I like Roy; I met him a few times in meetings and he was always nice; he shares his plasmids and strains and I know he was a nice reviewer to a previous lab member’s paper. But calling all of our comments “irrelevant” seems to me unprofessional. Ok, rant over.
Our two major problems were: 1. No imaging and 2. Over expression.
For this first issue, we were truely surprised that for a paper that makes claims about quality of imaging experiments, there was not even an attempt at imaging these mRNA (at least, none recorded). The second issue was that for all of the mRNAs tested, but one, it was always with over expression. The only endogenously expressed mRNA, PGK1, was actually underexpressed compared to the wild-type.
The reason we decided to write a response to Parker was becasue we felt that it disqualifies many (most, all?) papers that use this system in yeast, and this is our bread & butter. By “our”, I mean both the Gerst lab and my previous lab – the Singer lab who developed this system. At first we wanted to publish a collaborative response, but decided to split it (mainly because my data was ready earlier and I wanted it out already). See at the end of this post for some more about that.
So, we focused on the main issue of over expression. We first used pre-existing (unpublished) data that we already had from a few years ago, to ask if we can see accumulation of 3′-end fragments when using this system with genomically-tagged genes. Instead of doing Northern analysis, we used RNA-seq data of MS2-labeled mRNAs that were pulled-down from whole-population cell lysates using the RaPID system: the MCP is fused to GFP and to sterptavidin-binding peptide (MCP-GFP-SBP) and affinity purified on streptaviding beads.
What we found is that for 6 out of 10 mRNAs, there is some accumulation of 3′-end fragments (based on increased RNA-seq reads of this region). But we did not detect such accumulation in the other 4 mRNAs. This suggested to us that Parker is right that such fragments can exist, but wrong in generalizing his findings, whereas we think that this should be examined on a case by case basis.
I then takcled the issue of over-expression and imaging together. We already had strain in the lab that Dima made for his studies, of the MFA2 mRNA tagged with MS2 endogenously or over-expressed using the GAL1 promoter (single copy, integrated into the genome) or on a multi-copy plasmid. What I found really exemplified why we should always try to work as close to endogenous levels as possible. Over-expression cause an accumulation of huge RNA granules which typically co-localized with a P-body marker (P-bodies are considered to be sites of mRNA decay).I used FISH-Quant for analyzing the RNA granules and it seems they contain many many copies of the “mRNA”. I suspect that these are accually the “fragments” detected by Northern. Such huge granules were rarely seen in endogenously labeled mRNAs.
By accident, I also found that growth conditions (e.g. growth media) can exaggerate this difference between endogenous to even mild over-expressed mRNAs. We also did RT-qPCR analysis to verify the existance of fragments by another means. We found more fragments when the mRNA is over-expressed. We were also surprised that with or without expression of MCP-GFP made no difference, contrary to Parker’s claims.
We therefore concluded that the large amount of fragments detect by Parker in his Northern blots is probably due to over-expression of his mRNAs.
Our response was obviously sent to Parker, which asked for our strains. In the name of advancing science, we were glad to supply them. We were certain that his analysis will confirm our analysis.
Indeed, his Northen analysis of our MFA2-MS2 strains confirmed that over-expression results in more fragments. We are glad that we agree on that.
However, to our dismay, he did also found fragments in all of our endogenously-expressed mRNAs, including those that did not show accumulation in our RaPID-RNA-seq data. Since these are our strains, and as Parker claims – similar growth conditions – I can only assume that this differnce stems from the different methods of analysis. It could be that RaPID-RNA-seq is not detecting some of the fragments, or that the differnce lies with the MCP (fused to 3xGFP or GFP-SBP can make a differnce, maybe). At the moment I don’t have an answer. I also cannot explain why they do see a difference between with or without expression of MCP in our MFA2-MS2 strains.
Interesting point that was not mentioned in either our response or their commentary – it looks like for some mRNAs, expression of MCP in the cells causes an increase in the level of the full-length transcript. Is it just stabilization of the transcript (how?) or due to increased transcription? (Maybe due to some weird effect of Xrn1 not completing the degradation of the mRNA, leading to changes in the feedback to the nucleus. This might be a nice system to further test this mechanism. I should mention that to my PhD mentor. Hmmm…).
Because we feel the debate is still on, we wanted to solve this issue by doing co-FISH: labeled the mRNA by two sets of probes – one for the MS2 loops and the other for the coding sequence, in the presence or absence of MCP in the cells. ?Then see if we get an over-representation of MS2-spots only as Parker’s work suggests, or we will get most spots to co-localize.
I already did it once for one mRNA. Alas, after I imaged the slides I realized that the Cy5 filter of our brand new Zeiss microscope was defective, so I could not rely on the MS2 spot analysis. It took a month to replace the filter. By that time, our Response, along with Parker’s commentary, were already accepted. I also had more pressing experiments to do. So I put that on hold at the moment.
Meanwhile, two members of the Singer lab, Maria & Evelina, also started to explore this, mostly by doing co-FISH. I don’t want to discuss unpublished results, but I was given permission to say that they are now working on improving the MS2 system in yeast. Bin Wu from the Singer lab already made improvments to the MS2 system, by creating a version with synonymous changes to reduce the chance of recombination or other events that occur when you have repeating sequences. This version, called MBSv5 allows, for instance, for better co-localization of MS2 and coding sequence in co-FISH experiments (see Fig 1I in this paper). The next version of MS2 system should, hopefully, allow Xrn1 to properly degrade the mRNA and will not create these 3′-end fragments, while still allowing accurate, single-molecule imaging of mRNA in live cells.
Garcia JF, & Parker R (2015). MS2 coat protein bound to yeast mRNAs block 5′ to 3′ degradation and trap mRNA decay products: implications for the localization of mRNAs by MS2-MCP system. RNA (New York, N.Y.) PMID: 26092944
Haimovich G, Zabezhinsky D, Haas B, Slobodin B, Purushothaman P, Fan L, Levin JZ, Nusbaum C, & Gerst JE (2016). Use of the MS2 aptamer and coat protein for RNA localization in yeast: A response to “MS2 coat proteins bound to yeast mRNAs block 5′ to 3′ degradation and trap mRNA decay products: implications for the localization of mRNAs by MS2-MCP system”. RNA (New York, N.Y.), 22, 660-666 PMID: 26968626
Garcia, J., & Parker, R. (2016). Ubiquitous accumulation of 3′ mRNA decay fragments in mRNAs with chromosomally integrated MS2 arrays RNA, 22 (5), 657-659 DOI: 10.1261/rna.056325.116
Wu B, Miskolci V, Sato H, Tutucci E, Kenworthy CA, Donnelly SK, Yoon YJ, Cox D, Singer RH, & Hodgson L (2015). Synonymous modification results in high-fidelity gene expression of repetitive protein and nucleotide sequences. Genes & development, 29 (8), 876-86 PMID: 25877922