Comments
Write a commentNo comments have been published yet.
PREreview of Dissecting translation elongation dynamics through ultra-long tracking of single ribosomes
- Published
- License
- CC BY 4.0
The authors develop an imaging technique to monitor translation on single ribosomes in vivo. This work builds on previous efforts by extending the authors’ SunTag system for visualizing new proteins with ‘infinite’ translation of circularized mRNAs which lack stop codons. The authors confirm the accuracy of their technique by observing the effects of various inhibitors of translation, initiation, and elongation and by recapitulating translation times observed for problematic mRNA sequences. They then apply this approach to reveal a potential novel role for eIF4A as a regulator of elongation. They also reveal heterogeneity in the intrinsic translation rates of ribosomes within cells, suggesting differences in ribosome composition as an underlying explanation. This work presents a substantial methodological advance and will enable fundamental insights into translation and its regulation in vivo, as well as precise measurement of the effects of pharmaceuticals.
Overall, we found the paper to be impressive and the demonstration of the technique quite convincing. Most of our comments pertain to dissecting sources of ribosomal heterogeneity, which we see as a key contribution of this paper.
The authors report heterogeneity in the elongation rates of individual ribosomes and propose several possible causes, including heterogeneous ribosome composition or rRNA modification, ribosome damage, mutations in the socRNAs, differences between cells, or different subcellular localizations (line 364). The authors provide experimental evidence against the latter three causes. We are convinced that the apparent heterogeneity in translation rates is greater than the experimental noise, but we are curious whether mRNA modification should be mentioned in the text as another potential source of variation2,3. Some mRNA modifications can also be dynamic during their lifetime, like m6A (N6-methyladenosine) methylation, which enhances or represses translation by affecting mRNA stability4. It would be interesting to see whether the ribosome speed can change because of such mRNA modifications. If the assumption is that similar mRNA sequences undergo similar modifications and thus its role is negligible, we would suggest citing evidence to support this explicitly within the text.
To study translation by single ribosomes, the authors generate circularized mRNAs using the Tornado system1, which uses a pair of ribozymes to generate an mRNA that is ligated by an endogenous RNA ligase. Ribozymes often display some degree of off-target splicing, which in this case could conceivably produce socRNAs with different lengths. The paper that reported the Tornado system also appeared to display some variation in lengths of splice products. Given that the authors performed sequencing of socRNAs purified from cells, we suggest that the authors address this point directly within the text. The authors also note that all socRNAs consist of either 5 or 10 SunTags- was any meaningful difference in elongation rate noted between the two different constructs? It would be helpful to know whether length of the mRNA is an important confounding factor, and if so to what degree it contributes to heterogeneity.
To identify pauses in elongation, the authors fit hidden Markov models to their fluorescence data using a procedure adapted from FRET literature. As positive and negative controls for benchmarking their model, they generate synthetic data with and without pauses by multiplying pause sequences by a constant value. In figure 5N, the authors show probabilities of identifying pauses for their positive and negative controls at a single cutoff. It might be more effective to display performance of the model across a range of possible cutoffs, as is common for classification tasks in machine learning literature, given this is a tunable parameter. Additionally, we would like to clarify whether 50% corresponds to the expected probability of identifying a pause based on how the synthetic data was generated, or if this means that only 50% of synthetic pauses were correctly identified. Additionally, on line 400 and elsewhere, the ‘h’ in ‘hidden’ and the ‘m’ in ‘modeling’ should not be capitalized in “Hidden Markov Modeling.”
In an attempt to rule out the possibility that rare stochastic pauses could account for elongation heterogeneity between single ribosomes, the authors conduct a simple correlation analysis of translation rate between the first half and second half time windows. They then state that significant correlation is found, yet the coefficient of determination is only 0.45. What are the potential explanations for the low signal? Furthermore, if the aim of the correlation analysis is to point out that the pause time is not the primary determinant of a single ribosome’s elongation rate, then a scatterplot showing correlation between amount of time spent in a “pause state” as classified by the HMM against the average slope of its translation trace for each ribosome should be shown. If there is strong correlation, this would suggest that the differences in translational dynamics are due to variation in ribosome pause rate, in which case the question would become why certain ribosomes stall more, as opposed to inherent differences in the enzymatic rate of elongation.
References
4. Wang, X., Zhao, B. S., Roundtree, I. A., Lu, Z., Han, D., Ma, H., Weng, X., Chen, K., Shi, H., & He, C. (2015). N(6)-methyladenosine Modulates Messenger RNA Translation Efficiency. Cell, 161(6), 1388–1399. https://doi.org/10.1016/j.cell.2015.05.014
Competing interests
The authors declare that they have no competing interests.