Parent tRNA Modification Status Determines the Induction of Functional tRNA-Derived RNA by Respiratory Syncytial Virus Infection

Round 1

Reviewer 1 Report

The manuscript by Cho et al, focusses on tRNA derived fragments (tRFs) and their role during Respiratory Syncytial Virus (RSV) infection. The authors have previously shown that a specific subset of tRFs are induced by RSV infection in epithelial cells. Here they showed that increased tRFs can be detected in Nasopharageal swabs (NPS) from RSV-infected children. The authors then explore whether tRNA modification levels influence tRF levels using the MLC-Seq technique. They identify a single methylation site (A57) on mature GluCTC which decreases with RSV infection, and hypothesise about 2 other slight changes in molecular weight at modifications D19 and D20, but no evidence to support this claim is provided. The A57 site is known to be modified by AlkB Homolog 1, and loss of methylation here is known to influence tRF biogenesis. The authors then knockdown AlkB Homolog 1 and show that RSV replication is attenuated. The authors conclude that RSV infection influences AlkB Homolog 1 activity which increases 5’GluCTC tRF biogenesis, which in turn increases RSV replication. There are some interesting concepts here, but several points need to be addressed to substantiate the claims made.

The authors demonstrate that 3 tRFs are upregulated in NPS from RSV infected children but have only explored correlation analysis for GluCTC. Correlation analysis should be shown for all tRFs in the main figure.

The method used for qPCR of tRNA fragments should be added, and a description of how mock-infected cells are treated should be provided.

Fig 1D: The correlation between 5’Glu and RSV is strongly influenced by 1 single data point, it appears that 5’Cys and 5’Gly show are more consistent upregulation in RSV infected patients. The use of a logarithmic axis makes the results appear similar, but the axes are numbered differently (max 100 Vs 1000). The axes should be standardised, so comparison of the results is easier. The authors should explain why 5’Glu was chosen for further investigation over the other tRFs.

It is not clear why the 5’Glu tRF has not been precipitated with the probe to purify Glu tRNA as according to the diagram in Fig 2A the probe recognises the same sequence and 5’Glu tRF is upregulated in the samples used following RSV infection. Can the authors please explain the protocol and any size selection that was performed to explain why tRFs are not precipitated?

The MLC-seq method requires additional description, or perhaps a figure showing the workflow? The manuscript claims that there is a reduction in the molecular weight of 2 modifications but there is no data showing this. A figure displaying the output from the MLC-Seq would help demonstrate this finding. Including modifications that were detected but unchanged by RSV infection.

The authors have not investigated levels of the ribonucleases that generate tRFs, are these upregulated following infection in this system? They are known to be stress-responsive ribonucleases and viral infection will likely stimulate cell stress pathways.

There is no mechanism provided for how RSV infection could influence AlkB Homolog 1 activity, levels of AlkB Homolog 1 mRNA and protein should be quantified in RSV infected cells. The timeline of the experiments performed is not clear – could the virus be influencing levels of AlkB Homolog 1 directly? This should be investigated with a timecourse of AlkB expression following infection and tRF production to see if these levels inversely correlate.

The authors state that both 5’GluCTC and 5’GlyGCC are correlated with RSV replication which would suggest the mechanism may be sequence independent. The authors should discuss potential mechanisms for regulation of viral replication by tRFs. For example, regulation of retrotransposon replication by tRFs has been investigated by Schorn et al, 2017 (PMID: 28666125).

Author Response

1. The authors demonstrate that 3 tRFs are upregulated in NPS from RSV infected children but have only explored correlation analysis for GluCTC. Correlation analysis should be shown for all tRFs in the main figure.

Thanks for the valuable comment. We have included the requested figures (Figs 1E and F) in the revised manuscript.

2. The method used for qPCR of tRNA fragments should be added, and a description of how mock-infected cells are treated should be provided.

The related methods have been added in the revised manuscript: lines 98-105 and 108-109.

3. Fig 1D: The correlation between 5’Glu and RSV is strongly influenced by 1 single data point; it appears that 5’Cys and 5’Gly show more consistent upregulation in RSV-infected patients. The use of a logarithmic axis makes the results appear similar, but the axes are numbered differently (max 100 Vs 1000). The axes should be standardized, so comparison of the results is easier. The authors should explain why 5’Glu was chosen for further investigation over the other tRFs.

The logarithmic axis for 1A-C was used to demonstrate each dot in a clearly separated manner. Otherwise, some dots will be clustered together. As tRF5-GlyGCC expression from an individual was much higher than other samples, the program automatically set the max 1000. We tried different ways but thought it might be the way to present our data. We also discussed why 5’-Glu got our interest in lines 47-49 and why 5’-Glu was chosen for further investigation in lines 242-244.

4. It is not clear why the 5’Glu tRF has not been precipitated with the probe to purify Glu tRNA as according to the diagram in Fig 2A the probe recognises the same sequence and 5’Glu tRF is upregulated in the samples used following RSV infection. Can the authors please explain the protocol and any size selection that was performed to explain why tRFs are not precipitated?

We appreciate the constructive comment. tRFs were known from our previous studies to be derived from a very small proportion of parent tRNAs and their induction is indeed too little to be detected by RNA staining. Since this paper is mainly focused on the relationship between tRNA cleavage and modification changes in parent tRNA, we purified the tRNA band only. A separate manuscript on the role of tRF modification on RSV infection is being prepared separately. Size selection is now clarified in the revised manuscript.

5. The MLC-seq method requires additional description, or perhaps a figure showing the workflow? The manuscript claims that there is a reduction in the molecular weight of 2 modifications but there is no data showing this. A figure displaying the output from the MLC-Seq would help demonstrate this finding. Including modifications that were detected but unchanged by RSV infection.

We have included a workflow as Fig.3A and the output of key modifications and around have been shown in a supplementary excel file

6. The authors have not investigated levels of the ribonucleases that generate tRFs, are these upregulated following infection in this system? They are known to be stress-responsive ribonucleases and viral infection will likely stimulate cell stress pathways.

Our previous studies identified that the endonuclease angiogenin (ANG) is responsible for tRF induction by RSV (Figure. 7 of https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3594034/). Our unpublished data showed that ANG expression was not changed by RSV infection. However, the migration of ANG from the nuclear compartment to the cytosol compartment is enhanced by RSV. We are currently working on ANG-mediated cleavage and hope to have results together and out for publication in the very near future. Viral infection may not use stress pathways to produce tRFs, as hMPV, a close family member of RSV, cannot induce tRFs. In addition, the tRF types induced by viruses seem virus-specific.

7. There is no mechanism provided for how RSV infection could influence AlkB Homolog 1 activity, levels of AlkB Homolog 1 mRNA and protein should be quantified in RSV infected cells. The timeline of the experiments performed is not clear – could the virus be influencing levels of AlkB Homolog 1 directly? This should be investigated with a timecourse of AlkB expression following infection and tRF production to see if these levels inversely correlate.

The expression of ALKBH1 is checked and shown in Fig. 4E. The expression is overall not changed by RSV infection, suggesting RSV-regulated ALKBH1 activity is responsible. We will investigate this in the future.  

8. The authors state that both 5’GluCTC and 5’GlyGCC are correlated with RSV replication which would suggest the mechanism may be sequence independent. The authors should discuss potential mechanisms for regulation of viral replication by tRFs. For example, regulation of retrotransposon replication by tRFs has been investigated by Schorn et al, 2017 (PMID: 28666125).

Thanks for reviewer’s comments. This manuscript is a follow-up study to our previous studies on identifying the molecular mechanism used by tRF5-GluCTC to regulate RSV infection (by targeting LRP8 expression and subsequent its binding to the P protein of RSV to control RSV replication https://pubmed.ncbi.nlm.nih.gov/26156244/), and how tRF5-GluCTC finds its targets (https://pubmed.ncbi.nlm.nih.gov/33233493/), in addition to the endonuclease responsible for the cleavage (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3594034/). In another publication, we showed that not all induced tRFs regulate RSV replication (https://pubmed.ncbi.nlm.nih.gov/28708049/), supporting that sequence-dependent replication regulation by tRFs. This manuscript mainly focuses on its clinical implication and generates a model illustrating how a small portion of tRNAs are selected to be cleaved from the tRNA modification point of view.  

Reviewer 2 Report

The manuscript  "Parent tRNA Modification Status Determines the Induction of Functional tRNA-derived RNA by Respiratory Syncytial Virus Infection"  describes the role of post-transcriptional modifications of tRNAs in tRF biogenesis following RSV infection. The authors center their attention on tRNA GluCTC and on the effect of RSV infection on the modification status of this particular tRNA and on its cleavage and generation of tRF5-GluCTC that is highly impacted by RSV infection. The fact that viral infections can impact the tRNA epitranscriptome is highly relevant as modulation of tRNA modifications may constitute an antiviral strategy. Also, the authors present a very elegant method to perform tRF enrichment and isolation in samples. Although this is a relevant study, there are some issues that should be addressed and clarified prior to publication.

Major issues:

1) It is not clear from the manuscript how RNA was extracted from patient samples. Was RNA from the nasopharyngeal samples extracted with Trizol similarly to RNA extraction from the cell lines used in this study, or did the authors use another methodology?

2) Did the authors try to perform Northern Blotting of patient samples to further validate the qPCR data? Was the RNA amount too low that did not allow northern blotting analysis? Since all tRFs tested are increased upon infection, it would be relevant to have a qPCR for a tRF that does not change upon infection, to discard the hypothesis that RSV infection can be triggering indiscriminate tRNA fragmentation. Does RSV virus triggers tRF formation from tRNA GlyGCC or tRNA GlyTCC at similar levels as from tRNA GlyCCC, for example?

3) In Figure 2C it is not clear which samples are present in the northern blots (middle and right panels). The enriched tRNAs depicted in these panels are from control or infected cells? Why aren't both depicted?

4) The authors should present the mass spec profile in Figure 3 (or add it as a supplementary file together with the detection values). The mere representation of a % pie chart is not enough to convince that indeed there is a change in methylation at position 57. Also, it would be good to have the measurement of the levels of m1A at position 57 in patient samples, if possible.

5) The authors show that ALKBH1 expression is relevant for the m1A levels and consequent tRF5-GluCTC formation, but is ALKBH1 expression altered upon infection? This can be easily assessed by measuring ALKBH1 levels by qPCR and/or western blotting in mock and infected cells (and also in patient samples). This information is crucial to strengthen the authors conclusions.

6) The membrane in Figure 4B has 4 lanes and most probably lanes 1 and 3 are mock infected cells. Please add the indication of RSV infection in the picture, as in Figure 4A. How many replicates were performed for all the experiments depicted in figure 4? The quantification is missing in 4A and 4B and must be included.

7) ALKBH1 silencing led to a decrease in viral replication, but to directly correlate that with a decrease in tRF5-GluCTC formation, additional experiments must be performed, namely transfection of a tRF-GluCTC mimic and silencing of tRF5-GluCTC, and assess viral replication. ALKBH1 demethylates other RNAs, namely mRNAs, meaning that the effect observed may not be directly correlated with tRF formation but rather with alterations of methylation levels on relevant host (or even viral!) RNAs.

Minor:

Line 59 - I believe the authors meant "ALKBH1-mediated A demethylation", instead of methylation.

 

 

 

Author Response

Major issues:

• It is not clear from the manuscript how RNA was extracted from patient samples. Was RNA from the nasopharyngeal samples extracted with Trizol similarly to RNA extraction from the cell lines used in this study, or did the authors use another methodology?

Trizol extraction is a common way to extract RNAs. However, during the precipitation and washing steps, it is easy to lose the pallets. To reduce such an error, especially when patient samples are precious and limited in amount, and to obtain enough RNAs to maximize assays later, we chose to use column-based preparation using mirVanaTM Protein and RNA Isolation System from Invitrogen (#AM1556). The method has been updated in the revised manuscript (lines 82-86)

• Did the authors try to perform Northern Blotting of patient samples to further validate the qPCR data? Was the RNA amount too low that did not allow northern blotting analysis? Since all tRFs tested are increased upon infection, it would be relevant to have a qPCR for a tRF that does not change upon infection, to discard the hypothesis that RSV infection can be triggering indiscriminate tRNA fragmentation. Does RSV virus triggers tRF formation from tRNA GlyGCC or tRNA GlyTCC at similar levels as from tRNA GlyCCC, for example?

We thank the Reviewer for the constructive comment. Yes, the main reason we chose a qRT-PCR, instead of a Northern blot, is to use a limited sample amount to maximize the detection of tRF types of interest. To minimize sample preparation errors, we added a quantified synthetic RNA (cel-miR-39) for spike-in during RNA extraction procedures and subsequent normalization. Overall, cel-miR-39 signals were comparable among the samples, supporting the quality of RNA extraction/preparation and tRF quantification. We also included previously unshown data in the revised manuscript demonstrating similar tRF5-HisGTG expression in control and RSV patient samples (lines 202-203). Overall, this study, together with our previous studies (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3594034/), showed that tRFs are derived from a very small subset of tRNAs.

• In Figure 2C it is not clear which samples are present in the northern blots (middle and right panels). The enriched tRNAs depicted in these panels are from control or infected cells? Why aren't both depicted?

The revised Figure 2C has included samples from both control and infected cells.

• The authors should present the mass spec profile in Figure 3 (or add it as a supplementary file together with the detection values). The mere representation of a % pie chart is not enough to convince that indeed there is a change in methylation at position 57. Also, it would be good to have the measurement of the levels of m1A at position 57 in patient samples, if possible.

In the revised manuscript, we provided an excel file showing the relative abundance (counts) of A and mA at position 57.

• The authors show that ALKBH1 expression is relevant for the m1A levels and consequent tRF5-GluCTC formation, but is ALKBH1 expression altered upon infection? This can be easily assessed by measuring ALKBH1 levels by qPCR and/or western blotting in mock and infected cells (and also in patient samples). This information is crucial to strengthen the authors conclusions.

We thank the Reviewer for the valuable comment. We investigated ALKBH1 expression but did not see the expression change in both cytosol and nuclear compartments in the context of infection. Therefore, it is possible that RSV infection changes the ALKBH1 activity to regulate the methylation level of A57. We have included the result and related discussion in revised manuscript.

• The membrane in Figure 4B has 4 lanes and most probably lanes 1 and 3 are mock-infected cells. Please add the indication of RSV infection in the picture, as in Figure 4A. How many replicates were performed for all the experiments depicted in figure 4? The quantification is missing in 4B and must be included.

We have modified the figures as suggested. The silencing down vs RSV protein expression was repeated three times, and quantification was done accordingly.

• ALKBH1 silencing led to a decrease in viral replication, but to directly correlate that with a decrease in tRF5-GluCTC formation, additional experiments must be performed, namely transfection of a tRF-GluCTC mimic and silencing of tRF5-GluCTC, and assess viral replication. ALKBH1 demethylates other RNAs, namely mRNAs, meaning that the effect observed may not be directly correlated with tRF formation but rather with alterations of methylation levels on relevant host (or even viral!) RNAs.

The impact of mimic and antisense against tRF5-GluCTC on RSV replication has been shown in our previous publication https://pubmed.ncbi.nlm.nih.gov/23183536/

Minor:

Line 59 - I believe the authors meant "ALKBH1-mediated A demethylation", instead of methylation.

The error is corrected.

 

 

 

 

Round 2

Reviewer 1 Report

The authors have addressed my comments