A localization screen reveals translation factories and widespread co-translational RNA targeting

Racha Chouaib, Adham Safieddine, Xavier Pichon, Arthur Imbert, Oh Sung Kwon, Aubin Samacoits, Abdel-Meneem Traboulsi, Marie-Cécile Robert, Nikolay Tsanov, Emeline Coleno, Ina Poser, Christophe Zimmer, Anthony Hyman, Hervé Le Hir, Kazem Zibara, Marion Peter, Florian Mueller, Thomas Walter, Edouard Bertrand

Posted on: 1 June 2020 , updated on: 2 June 2020

Preprint posted on 21 May 2020

Article now published in Developmental Cell at http://dx.doi.org/10.1016/j.devcel.2020.07.010

mRNA localization is a heterogeneous riddle: translation might be the solution

Selected by Mafalda Pimentel

Categories: bioinformatics, cell biology

Background

Much progress has been made since the first report on localization of a specific mRNA in 1983 (non-muscle actin in the egg of the ascidian Styela¹). Meanwhile, several transcripts were shown to localize in different cell types and species, usually through sequence-specific mechanisms involving RNA binding proteins and molecular motors. Some mRNA localization mechanisms are linked to translation: they can either be repressed during transport to ensure local delivery of proteins, can localize mRNAs by linkage of the ribosomes to nascent localization peptides.

Recent studies deployed “spatial transcriptome” in situ approaches and observed global patterns of distribution for various mRNAs, but lacked mechanistic insight. By screening the localization of 523 GFP-tagged transcripts, the authors of this preprint aimed to understand how local translation affects overall cellular mRNA distribution.

Key findings

Importantly, the use of a library of 523 HeLa cell lines containing GFP-tagged genes in bacterial artificial chromosomes (BAC)², allowed to:

use a single set of single-molecule FISH probes (GFP) to recognize the different mRNAs
look at the GFP-tagged protein localization
expect that all the regulatory sequences (promoter, enhancer, introns, UTRs) will faithfully reproduce the behavior of endogenous mRNAs

The authors initially screened mRNAs encoding motor proteins, and validated the observed localizations for some endogenous transcripts. They proceeded to analyze the remaining library, ending up with six patterns of localization: at the “cell edge” or specifically in “protrusions”; aggregated in cytoplasmic “foci”; in the “perinuclear” region, surrounding the “nuclear edge” or even “intranuclear”; and at the centrosome of “mitotic” cells.

Figure 1 – A) Schematic of the localization patterns detected in interphase cells. B) Examples of some localized mRNAs (red) and respective GFP-tagged proteins (green). Source: figure 3 of the preprint.

Overall, 6% of the analyzed transcripts showed a specific cytoplasmic localization. The accumulation of mRNAs in foci was the pattern most often observed. While the majority of these mRNAs colocalized with P-bodies, the transcripts of four genes did not (BUB1, DYNC1H1, CTNNB1/β-catenin and ASPM) thus constituting distinct structures.

To perform a more quantitative and unbiased analysis, the authors adapted an mRNA detection algorithm to include mixed patterns of localization³. Through machine learning, the localization features of 27 mRNAs in 9710 cells were thoroughly calculated. The results were similar to the manual annotations, but in addition revealed a high degree of intra and intercellular heterogeneity for all studied mRNAs.

Of the screened mRNAs, 2% colocalized with the GFP-tagged protein. Treatment with Puromycin (translation inhibitor that releases nascent peptide) disrupted the localization patterns for most of these mRNAs as well as the colocalization with the encoded protein. Local translation was also demonstrated (using a SunTag) for the four mRNAs accumulated in non-P-body foci, which completely disassembled when translation was inhibited and were therefore entitled as “translation factories”.

Interestingly, the few cells that did not show β-catenin translation factories, had high levels of nuclear β-catenin-GFP. Since this protein was described to enter the nucleus upon Wnt signaling, the authors stimulated cells with WNT3A and observed disassembly of translation factories. Furthermore, they showed that the degradation/oligomerization proteins Axin and APC were present in these foci and that their depletion led to its disassembly. These results suggest that the β-catenin mRNA clusters at co-translational protein degradation foci.

What I like about this work

First of all, I really like the experimental approach. It is known that splicing and mRNA-RBP stoichiometry can affect transcript localization. The use of genome integrated BACs allows the screening of multiple mRNA and protein localizations in conditions very similar to endogenous expression. I also appreciate the adaptation and use of an automated quantification algorithm to show the heterogeneity in mRNA distribution. This is an obvious feature that is often overlooked, although it may be important to understand how mRNA localization varies depending on the specimen and environment. This work also provides an extensive list of mRNAs that localize through translation dependent mechanisms, which can be very useful to the community.

Standing questions for the future

It is interesting to see that kinesin mRNAs accumulate at protrusions, particularly Kif1c, since they are plus-end motors. Could this reflect its role as an integrin recycling transporter?

Given the extended periods of ASPM polysomes at the nuclear envelope, could these represent stalled/poised translation events related to the cell cycle status?