Non-canonical contribution of human Dicer helicase domain in antiviral innate immune response

  1. Morgane Baldaccini
  2. Léa Gaucherand
  3. Béatrice Chane-Woon-Ming
  4. Mélanie Messmer
  5. Floriane Gucciardi
  6. Sébastien Pfeffer

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

In mammalian somatic cells, the relative contribution of RNAi and the type I interferon response during viral infection is unclear. The apparent inefficiency of antiviral RNAi might be due to self-limiting properties and mitigating co-factors of the key enzyme Dicer. In particular, the helicase domain of human Dicer appears to be an important restriction factor of its activity. We studied the involvement of several helicase-truncated mutants of human Dicer in the antiviral response. All deletion mutants displayed a PKR-dependent antiviral phenotype against certain viruses and one of them, Dicer N1, acted in a complete RNAi-independent manner. Transcriptomic analyses showed that many genes from the interferon and inflammatory response were upregulated in Dicer N1 expressing cells. We could show that some of these genes appear to be controlled by NF-kB and that blocking this pathway abrogated the antiviral phenotype of Dicer N1. Our findings highlight the crosstalk between Dicer, PKR, and the NF-kB pathway, and suggest that human Dicer may have repurposed its helicase domain to prevent basal activation of antiviral and inflammatory pathways.

Article activity feed

  1. Version published to 10.1101/2023.08.23.554407v2 on bioRxiv
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1 (Evidence, reproducibility and clarity (Required)):

    The study largely focuses on the use of a 293 cell line that lacks a functional Dicer gene originally identified by the Cullen group. Baldaccini use this cell line, referred to as NoDice cells, to reconstitute various Dicer isoforms that have thus far been described in a variety of settings (e.g., stem cells and oocytes). Collectively, these data demonstrate the capacity of certain N-terminal truncations of Dicer to inhibit Sindbis virus and reduce the presence of viral dsRNA, supporting some of the observations made thus far concerning an antiviral role for mammalian Dicer. For other viruses, this impact was significantly more modest (SFV reduction is less than a log) or was not observed at all (VSV and SARS-CoV-2). The authors then go on to characterize the nature of the observed antiviral activity and ultimately implicate PKR and the induction of NF-kB in priming the cell's antiviral defenses. Importantly, the group also found that this antiviral activity neither required the nuclease activity of Dicer nor the kinase activity of PKR - providing evidence against antiviral RNAi in mammals. In all, the data would seem to suggest that Dicer can act as a dsRNA sensor and can mediate the activation of an NF-kB response - akin to what is observed in response to NOD or some TLR engagement. In all, it is the opinion of this reviewer that this work brings additional clarity to a concept that remains controversial in the field and therefore embodies something meaningful for the community.

    With that said, there are a few issues that require additional attention. The first of these is textual. The introduction of the paper accurately describes the evidence in support of mammalian RNAi but does not invest the same time in discussing the data to the contrary. For example, Seo et al demonstrated that virus infection results in poly-adp-ribosylation of RISC preventing RNAi activity (PMID: 24075860), Uhl et al showed that IFN-induced ADAR1 resolves dsRNA in the cell and prevents RNAi (PMID: 37017521), and Tsai et al showed that virus-derived small RNAs are not loaded into the RISC in a manner that would enable antiviral activity (PMID 29903832). None of this work is referenced in this manuscript and it generates an unbalanced introduction as it relates to the controversy surrounding the idea of RNAi in mammals.

    Reply: We thank the reviewer for their positive comments and suggestions. In the revised version of this manuscript, we will rewrite the introduction to take into account the published data that are not in favor of an antiviral role of RNAi in mammals and we will add the suggested references

    The second issue that would further strengthen this paper relates to the fact that the authors spend a considerable amount of time discussing the data of Figure 6 and 7 as conditions that are defined by a Dicer that can not be processive in its nuclease activity (WT) vs. one that can (N1). However, there seems to be little consideration about the fact that the introduction of WT Dicer into these cells also restores miRNA biology whereas N1 appears to remain only partially functional (based on the data of Fig 3E). Given this, it seems the authors should verify that the high baseline of NFkB signaling that is being observed when comparing WT to N1 is not a product of restored miRNA function in WT cells, in contrast to the hypotheses outlined in the manuscript. This could be addressed by silencing Drosha or DGCR8 in the Dicer knockout cells prior to their reconstitution of Dicer. In the opinion of this reviewer, this experimental control would significantly strengthen the conclusions the authors are making here.

    Reply: This would indeed be an ideal experiment to rule out the contribution of miRNAs in the observed phenotype. We believe however that this particular experiment would prove difficult to realize given that we reconstitute Dicer expression by lentiviral transduction and keep the cells under selection for a couple of weeks before using them for further experiments. This time frame is therefore not compatible with the use of siRNA to knock-down Drosha or DGCR8. Alternatively, we could knock them out by CRISPR-Cas9, but this would take too long and is not feasible in the frame of this work.

    We can however address the concern regarding the role played by miRNAs in the observed phenotype of the Dicer N1 cells. Indeed, we can determine the miRNA profile from our small RNA sequencing data and compare them between the Dicer WT and Dicer N1 cells. We have done this comparison and could not find striking differences in miRNA expression between the two cell lines. We will add this additional piece of evidence in our revised manuscript.

    Reviewer #1 (Significance (Required)):

    In the manuscript entitled, "Canonical and non-canonical contributions of human Dicer helicase domain in antiviral defense" Baldaccini et al. describe their findings concerning the ability of certain N-terminal deletion variants of Dicer in contributing to mammalian antiviral activity. The concept of a functional antiviral RNAi system in mammals is a contentious one with many publications including data both in support of its existence and opposing this idea. In this manuscript, Baldaccini et al. perform a wide range of well-controlled experiments to specifically address aspects of those reports to both provide clarity in what has been documented thus far and to expand on those concepts further.

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    Summary

    Whether RNAi is used as an antiviral mechanism in mammals has been a hotly debated issue. The research team previously published several papers on the roles of Dicer in siRNA/miRNA biogenesis and in antiviral responses. They have recently reported that the helicase domain of human Dicer specifically interacts with several proteins that are involved in the IFN response, including PKR. In this study, Baldaccini et al. investigated the involvement of Dicer in antiviral response using various mutants of human Dicer. They showed that deletion mutants of helicase domain exhibit antiviral activity that requires the presence of PKR. They further demonstrated that one of the mutants, N1-Dicer showed antiviral activity in an RNAi-independent manner but depending on the presence of either native PKR or kinase deficient mutants. Transcriptomic analysis revealed that numerous genes involved the IFN and inflammatory response were upregulated in the cells that express N1-Dicer, which is likely due to an increased activation of the NFκB pathway. Based on these findings, the authors propose that Dicer may act as antiviral molecule using its helicase domain, which representing a novel non-canonical function of Dicer.

    Major comments:

    1.The results from experiments with SARS-CoV2 are intriguing (Fig.2). The authors speculated that NFkb activation is in favor of the replication of this virus. It would be interesting to see the infection and replication of SARS-CoV2 in PKR deficient cells and cells expressing PKR mutants (as described in Fig.5). The results may prove/disapprove the authors' speculation and yield additional findings.

    Reply: We thank the reviewer for this suggestion. We have cells that are double knock-out for Dicer and PKR (NoDice/∆PKR) that were transduced to stably express Dicer WT or Dicer N1 and further transduced to express ACE2. We will infect those cell lines with SARS-CoV-2, which will allow us to see whether the difference in viral accumulation can still be observed in the absence of PKR. However, it might prove more difficult to reconstitute PKR expression (WT or mutants) in these cells since they are already transduced twice with two different constructs (Dicer and ACE2).

    Western blot analysis. In the method section, it is stated that proteins were quantified with Bradford method and equal loading was verified by Ponceau S staining. The members of also probed with gamma-tubulin (It was stated that antibodies against alpha-tubulin was used in the method section) as a loading control, however, the bend intensity of tubulin shows great variations among different lanes in several figures while Ponceau S staining is similar (Fig.s, 4, 5, and 8). The differences compromise the accuracy of the results.

    Reply: We apologize for the difference in Tubulin signal in some of our western blots. There are several possibilities to explain those inconsistencies between Ponceau staining and Tubulin blotting, including an effect of viral infection on Tubulin expression. To remove ambiguities around this issue, we could quantify the signal across several blot replicates and provide the quantification after normalization. In addition, we would like to stress that regarding quantification of the infection, we think that the plaque assay experiments are more reliable than quantification of western blot signals.

    3.RNA-seq analysis revealed that Dicer N1 cells have significantly increased expression levels of signaling molecules in type I IFN response even in uninfected cells. While this provides a potential explanation for the antiviral phenotype of N1-Dicer cells. I wonder why the expression levels of type I IFNs (probably the most potent antiviral molecules) were not analyzed in WT and Dicer N1 cells. Measurement of the levels of IFNα and IFNβ by ELISA in the cells before and after infection could provide the important and direct data to support their conclusion.

    Reply: This an interesting suggestion but unfortunately, we do not believe that it would possible to quantify IFNα and IFNβ by ELISA in the cell line that we used in our experiments. Indeed, the level of expression might just be too low to be able to measure something meaningful. We could measure the induction of IFNβ expression at the mRNA level by RT-qPCR though. However, we do not believe that the observed increased expression of genes that belong to the type I IFN response is solely the effect of an increased production of IFN. These genes are also under the control of other transcription factors, including NF-kB for some of them, and it might prove difficult to make a direct link with IFNα or IFNβ production.

    4.While the data presented in Fig. 5 provides convincing evidences that the antiviral activity of mediated by PKR against SINV is independent of its kinase activity in N1-Dicer cells. An interesting question is that whether antiviral activity associated with PKR is N1-Dicer dependent, which could be addressed by comparing the viral infection of NoDice∆PKR and NoDicer expressing PKR mutants.

    Reply: Yes indeed, we have generated NoDice/∆PKR cells expressing PKR WT or mutant and we will infect them with SINV to confirm whether the presence of Dicer N1 is needed for the observed phenotype.

    5.In the concluding paragraph of the discussion, the authors presented an oversimplified discerption of a complex model that involves a crosstalk between IFN-I and RNAi and Dicer-PKR interaction, which is difficult for the reader to compose a clear picture of mechanisms involved. It could be helpful to use a schematic illustration to summarize the action model of PKR incorporated with the canonical and non-canonical Dicer functions.

    Reply: We will add a schematic model in the revised version of our manuscript to summarize our main findings.

    Minor comments:

    1.It stated that NoDice FHA-Dicer WT #4 and NoDice FHA:Dicer N1 110 #6 are referred to as Dicer WT and Dicer N1 cells (p.6). For simplicity, Dicer WT and Dicer N1 cells should be used throughout manuscript, including in all figures. The labels in the figures are difficult to read and are confusing in some cases.

    Reply: This will be changed in the revised version to increase the clarity of the figures.

    2.It is to note that p-PKR was only detected at in N1-Dicer cells at 24 hpi (Fig.8A). This is an interesting observation that was not discussed. It appears that this could be due to a delayed viral replication since these cells are already in an elevated antiviral state. This possibility could be tested by examining viral replication and dsRNA accumulation at more time points in the experiments described in Fig.1.

    Reply: We have performed a kinetic of infection at more time points and we will incorporate these experiments in the revision.

    3.The authors may point out the limitations of the studies. For examples, all cells used in the study are engineered HEK cell lines and were tested with limited number of viruses. As such, the observations may reflect Dicer-PKR interaction under artificially overexpressed conditions, but how the model established from the current study applies to primary cells require further investigation.

    Reply: This is indeed important, we will add a sentence about this in the discussion.

    Reviewer #2 (Significance (Required)):

    The findings reported in this study shed some new light on a long-debated issue regarding the potential roles of RNAi as physiologically relevant antiviral mechanism in mammals. Identification of a new antiviral function of Dicer helicase domain via interaction with PKR is a new advancement of the field, and it also adds a new dimension to a complex subject that overlaps of innate immunity , RNA biology, and developmental biology associated with Dicer.

    Field of expertise: Innate immunity, cell signaling, cytokine biology

    Areas that that I do not have sufficient expertise to evaluate: Small RNA cloning, sequencing and, analysis.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    This work by Baldaccini et al. explores the interplay between Dicer and the antiviral protein PKR in the context of viral infection. It builds on a previous publication of the team which demonstrates that the Dicer helicase interacts with multiple RNA binding proteins, including PKR (see Montavon et al.). In this work from 2021, they demonstrate that an artificially-truncated form of Dicer (Dicer-N1) lacking part of the helicase is antiviral against RNA viruses in a PKR-dependent fashion. This was an interesting finding because the field largely assumed that Dicer-N1 performs its antiviral function via canonical dicing of dsRNA, as part of an antiviral RNAi pathway. The present manuscript follows up on this initial discovery and deciphers the specifics of Dicer-N1 antiviral phenotype, as well as delineates the interplay between Dicer's helicase and PKR. The authors main claims are as follow:

    1. i) Dicer-N1 antiviral effect does not require its catalytic activity, therefore is completely RNAi-independent.
    2. ii) Neither does it require canonical PKR activation, but relies instead on NF-kB-driven inflammation. The origin of this inflammation is not studied.
    3. ii) Truncated Dicers other than Dicer-N1 are antiviral through RNAi, but are also PKR-dependent. The authors claims are mostly supported by the data, although I suggest below some improvements regarding experimental approaches and data presentation. This work details in an interesting manner the interplay between the machinery of RNAi and the classical pathway of innate immunity (PKR). As explained by the authors, there is solid data in the literature demonstrating the mutual exclusivity of IFN and antiviral RNAi in differentiated cells. This mostly goes through the receptors LGP2, which inhibits dsRNA dicing by Dicer. The authors data suggest that, conversely, Dicer may play a role in preventing the unwanting activation of PKR (a non-canonical activation leading to inflammation). Given that PKR activation does not depend on virus, the authors discuss potential mechanisms of PKR triggering. This is an interesting topic that deserves further investigation (not necessarily within the frame of this work - it can be a follow-up). Another interesting piece of information is that different truncated Dicers behave differently with respect to implementing antiviral RNAi. Whilst Dicer-N1 isn't proficient in doing so, the other forms are. It shows that lab-generated truncations do not fully recapitulate what is observed with existing truncated Dicers (DicerO and aviD).

    Experimental design and data interpretation

    1. The authors should compare infection between different cell lines across a range of time points (ie, a virus growth curve). In Fig 4E for example, I worry that cells expressing or not PKR will reach the plateau of viral particle accumulation at different time points. One could imagine that cells lacking PKR do not show any differences in particle production at 24h, but do at earlier time points.

    Reply: This is an interesting suggestion, we can perform a kinetic experiment by looking at more time points to address this point. This will allow us to determine the time needed for every cell line to reach the plateau of infection.

    Western blots should be accompanied with proper quantifications plotted as bar graph with biological replicates (p-PKR, p-eIF2a and capsid).

    Reply: We have biological replicates for our western blot experiments, and we will quantify those to better determine the observed changes. However, in the case of p-eIF2a, we do not think it is pertinent to measure it since there are other kinases than PKR that are known to induce eIF2a phosphorylation upon SINV infection. It might therefore not prove very informative to precisely quantify this particular signal.

    Microscopy images should be properly quantified across biological replicates (Fig 1&2 for the J2 staining, for example).

    Reply: We could do a proper quantification of the J2 signal across replicates, but we do not think it would bring much to our message. Here, we mostly used J2 staining as a qualitative indication that the infection was impacted or not. We have a proper quantification of the effect with our plaque assay experiments, which are way more robust to determine the levels of infection between conditions.

    Confounding factors hinder the interpretation of siRNA accumulation (Suppl Fig 2): i) the efficiency of dsRNA dicing from different Dicers will generate different amounts of siRNAs from a given amount of dsRNA and ii) the higher antiviral response translates into decreased infection, so decreased dsRNA substrate. I suggest that the authors normalise the amount of viral siRNAs over the total amount of viral genomes. This should allow to assess if Dicer-N1 is better at dicing dsRNA than WT in these conditions.

    Reply: This is a valid concern and we agree that it is important to be able to normalize small RNA reads between conditions before reaching a conclusion. The problem is that there is no easy way to do this since we do not get a direct measurement of viral genomes accumulation from our small RNA sequencing data. To better compare the two conditions, we could normalize the individual viral siRNA to the total number of viral reads. Another problem that we face is that we are looking here at the AGO-loaded small RNAs, which makes it more difficult to assess dicing efficiency since not every generated siRNA might be loaded into an Argonaute protein. In fact, this has been proposed by the Cullen laboratory in a paper published in 2018 (Tsai et al. doi: 10.1261/rna.066332.118). They showed that although viral siRNAs were generated during IAV infection, those were inefficiently loaded and thus did not significantly impacted the infection.

    In Fig 8, the authors should verify that phospho-p65 increase depends on PKR by repeating the experiment in PKR KO cells.

    Reply: Yes, good point. We will check what happens to phosphorylation of p65 in PKR KO cells. In addition, we can also measure the effect on a known NF-kB target by RT-qPCR (e.g. PTGS2).

    Data representation

    1. Levels of phospho-PKR and eIF2a need to be normalised on the total amount of PKR and eIF2a, respectively. The authors should quantify the blots and present bar graphs with biological replicates and statistics.

    Reply: As mentioned above in our reply to point 2, we can add the quantification for phospho PKR, but we do not think it is pertinent to do it for eIF2a.

    Could the authors add the names of representative genes on the volcano plots of Fig 7?

    Reply: Yes, this will be done.

    Points of discussion

    1. In Fig 4C, catalytically-dead mutants of truncated Dicers (other than N1) do not display an antiviral effect. Presumably, such proteins implement canonical antiviral RNAi. Is there a reason why the authors interpret this data as Dicers being "partially" antiviral through RNAi (l. 92). This data instead suggest that is it totally dependent on RNAi.

    Reply: Indeed, and we do not say the contrary. It seems that some of this helicase-truncated Dicer proteins can act through RNAi. However, they also depend on PKR, so in the end it might be a combination of the two that allows their antiviral effect.

    Gurung et al. demonstrate that PKR is activated in Dicer KO mouse ES cells, which results in phosphorylation of eIF2a at steady-state. This is different from the authors' data, in which PKR activation does not affect eiF2a phosphorylation. Could the authors discuss this discrepancy?

    Reply: The problem that we face here is that SINV is known to also activate GCN2 and therefore eIF2a phosphorylation does not strictly rely on PKR in our experimental conditions. In addition, we did not check eIF2a phosphorylation in Dicer KO cells, but we always compare Dicer WT and Dicer N1 expressing cells.

    Do the authors expect that truncated Dicers other than N1 trigger an inflammatory response such as the one described for N1? Would it be possible to have this antiviral inflammatory response in conjunction with antiviral RNAi?

    Reply: This goes back to Point 1 mentioned previously. We think indeed that there might be a dual action of Dicer and that it will be important to check whether in other cellular systems or animal model such a phenomenon can be observed as well. This is a point that we did address in the discussion of our manuscript (line 522-525).

    Reviewer #3 (Significance (Required)):

    This is a study that conceptually advances the field of antiviral RNAi in mammals, including its interplay with the machinery of innate immunity. It is of interest for virologists and immunologists. My expertise is centered on the mechanisms of innate immunity in mammalian cells, including antiviral RNAi.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    This work by Baldaccini et al. explores the interplay between Dicer and the antiviral protein PKR in the context of viral infection. It builds on a previous publication of the team which demonstrates that the Dicer helicase interacts with multiple RNA binding proteins, including PKR (see Montavon et al.). In this work from 2021, they demonstrate that an artificially-truncated form of Dicer (Dicer-N1) lacking part of the helicase is antiviral against RNA viruses in a PKR-dependent fashion. This was an interesting finding because the field largely assumed that Dicer-N1 performs its antiviral function via canonical dicing of dsRNA, as part of an antiviral RNAi pathway. The present manuscript follows up on this initial discovery and deciphers the specifics of Dicer-N1 antiviral phenotype, as well as delineates the interplay between Dicer's helicase and PKR. The authors main claims are as follow:

    i) Dicer-N1 antiviral effect does not require its catalytic activity, therefore is completely RNAi-independent.

    ii) Neither does it require canonical PKR activation, but relies instead on NF-kB-driven inflammation. The origin of this inflammation is not studied.

    ii) Truncated Dicers other than Dicer-N1 are antiviral through RNAi, but are also PKR-dependent.

    The authors claims are mostly supported by the data, although I suggest below some improvements regarding experimental approaches and data presentation. This work details in an interesting manner the interplay between the machinery of RNAi and the classical pathway of innate immunity (PKR). As explained by the authors, there is solid data in the literature demonstrating the mutual exclusivity of IFN and antiviral RNAi in differentiated cells. This mostly goes through the receptors LGP2, which inhibits dsRNA dicing by Dicer. The authors data suggest that, conversely, Dicer may play a role in preventing the unwanting activation of PKR (a non-canonical activation leading to inflammation). Given that PKR activation does not depend on virus, the authors discuss potential mechanisms of PKR triggering. This is an interesting topic that deserves further investigation (not necessarily within the frame of this work - it can be a follow-up). Another interesting piece of information is that different truncated Dicers behave differently with respect to implementing antiviral RNAi. Whilst Dicer-N1 isn't proficient in doing so, the other forms are. It shows that lab-generated truncations do not fully recapitulate what is observed with existing truncated Dicers (DicerO and aviD).

    Experimental design and data interpretation

    1. The authors should compare infection between different cell lines across a range of time points (ie, a virus growth curve). In Fig 4E for example, I worry that cells expressing or not PKR will reach the plateau of viral particle accumulation at different time points. One could imagine that cells lacking PKR do not show any differences in particle production at 24h, but do at earlier time points.
    2. Western blots should be accompanied with proper quantifications plotted as bar graph with biological replicates (p-PKR, p-eIF2a and capsid).
    3. Microscopy images should be properly quantified across biological replicates (Fig 1&2 for the J2 staining, for example).
    4. Confounding factors hinder the interpretation of siRNA accumulation (Suppl Fig 2): i) the efficiency of dsRNA dicing from different Dicers will generate different amounts of siRNAs from a given amount of dsRNA and ii) the higher antiviral response translates into decreased infection, so decreased dsRNA substrate. I suggest that the authors normalise the amount of viral siRNAs over the total amount of viral genomes. This should allow to assess if Dicer-N1 is better at dicing dsRNA than WT in these conditions.
    5. In Fig 8, the authors should verify that phospho-p65 increase depends on PKR by repeating the experiment in PKR KO cells.

    Data representation

    1. Levels of phospho-PKR and eIF2a need to be normalised on the total amount of PKR and eIF2a, respectively. The authors should quantify the blots and present bar graphs with biological replicates and statistics.
    2. Could the authors add the names of representative genes on the volcano plots of Fig 7?

    Points of discussion

    1. In Fig 4C, catalytically-dead mutants of truncated Dicers (other than N1) do not display an antiviral effect. Presumably, such proteins implement canonical antiviral RNAi. Is there a reason why the authors interpret this data as Dicers being "partially" antiviral through RNAi (l. 92). This data instead suggest that is it totally dependent on RNAi.
    2. Gurung et al. demonstrate that PKR is activated in Dicer KO mouse ES cells, which results in phosphorylation of eIF2a at steady-state. This is different from the authors' data, in which PKR activation does not affect eiF2a phosphorylation. Could the authors discuss this discrepancy?
    3. Do the authors expect that truncated Dicers other than N1 trigger an inflammatory response such as the one described for N1? Would it be possible to have this antiviral inflammatory response in conjunction with antiviral RNAi?

    Significance

    This is a study that conceptually advances the field of antiviral RNAi in mammals, including its interplay with the machinery of innate immunity. It is of interest for virologists and immunologists. My expertise is centered on the mechanisms of innate immunity in mammalian cells, including antiviral RNAi.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Summary

    Whether RNAi is used as an antiviral mechanism in mammals has been a hotly debated issue. The research team previously published several papers on the roles of Dicer in siRNA/miRNA biogenesis and in antiviral responses. They have recently reported that the helicase domain of human Dicer specifically interacts with several proteins that are involved in the IFN response, including PKR. In this study, Baldaccini et al. investigated the involvement of Dicer in antiviral response using various mutants of human Dicer. They showed that deletion mutants of helicase domain exhibit antiviral activity that requires the presence of PKR. They further demonstrated that one of the mutants, N1-Dicer showed antiviral activity in an RNAi-independent manner but depending on the presence of either native PKR or kinase deficient mutants. Transcriptomic analysis revealed that numerous genes involved the IFN and inflammatory response were upregulated in the cells that express N1-Dicer, which is likely due to an increased activation of the NFκB pathway. Based on these findings, the authors propose that Dicer may act as antiviral molecule using its helicase domain, which representing a novel non-canonical function of Dicer.

    Major comments:

    1.The results from experiments with SARS-CoV2 are intriguing (Fig.2). The authors speculated that NFkb activation is in favor of the replication of this virus. It would be interesting to see the infection and replication of SARS-CoV2 in PKR deficient cells and cells expressing PKR mutants (as described in Fig.5). The results may prove/disapprove the authors' speculation and yield additional findings.

    1. Western blot analysis. In the method section, it is stated that proteins were quantified with Bradford method and equal loading was verified by Ponceau S staining. The members of also probed with gamma-tubulin (It was stated that antibodies against alpha-tubulin was used in the method section) as a loading control, however, the bend intensity of tubulin shows great variations among different lanes in several figures while Ponceau S staining is similar (Fig.s, 4, 5, and 8). The differences compromise the accuracy of the results. 3.RNA-seq analysis revealed that Dicer N1 cells have significantly increased expression levels of signaling molecules in type I IFN response even in uninfected cells. While this provides a potential explanation for the antiviral phenotype of N1-Dicer cells. I wonder why the expression levels of type I IFNs (probably the most potent antiviral molecules) were not analyzed in WT and Dicer N1 cells. Measurement of the levels of IFNα and IFNβ by ELISA in the cells before and after infection could provide the important and direct data to support their conclusion. 4.While the data presented in Fig. 5 provides convincing evidences that the antiviral activity of mediated by PKR against SINV is independent of its kinase activity in N1-Dicer cells. An interesting question is that whether antiviral activity associated with PKR is N1-Dicer dependent,, which could be addressed by comparing the viral infection of NoDice∆PKR and NoDicer expressing PKR mutants. 5.In the concluding paragraph of the discussion, the authors presented an oversimplified discerption of a complex model that involves a crosstalk between IFN-I and RNAi and Dicer-PKR interaction, which is difficult for the reader to compose a clear picture of mechanisms involved. It could be helpful to use a schematic illustration to summarize the action model of PKR incorporated with the canonical and non-canonical Dicer functions.

    Minor comments:

    1.It stated that NoDice FHA-Dicer WT #4 and NoDice FHA:Dicer N1 110 #6 are referred to as Dicer WT and Dicer N1 cells (p.6). For simplicity, Dicer WT and Dicer N1 cells should be used throughout manuscript, including in all figures. The labels in the figures are difficult to read and are confusing in some cases. 2.It is to note that p-PKR was only detected at in N1-Dicer cells at 24 hpi (Fig.8A). This is an interesting observation that was not discussed. It appears that this could be due to a delayed viral replication since these cells are already in an elevated antiviral state. This possibility could be tested by examining viral replication and dsRNA accumulation at more time points in the experiments described in Fig.1. 3.The authors may point out the limitations of the studies. For examples, all cells used in the study are engineered HEK cell lines and were tested with limited number of viruses. As such, the observations may reflect Dicer-PKR interaction under artificially overexpressed conditions, but how the model established from the current study applies to primary cells require further investigation.

    Significance

    The findings reported in this study shed some new light on a long-debated issue regarding the potential roles of RNAi as physiologically relevant antiviral mechanism in mammals. Identification of a new antiviral function of Dicer helicase domain via interaction with PKR is a new advancement of the field, and it also adds a new dimension to a complex subject that overlaps of innate immunity , RNA biology, and developmental biology associated with Dicer.

    Field of expertise: Innate immunity, cell signaling, cytokine biology

    Areas that that I do not have sufficient expertise to evaluate: Small RNA cloning, sequencing and, analysis.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    The study largely focuses on the use of a 293 cell line that lacks a functional Dicer gene originally identified by the Cullen group. Baldaccini use this cell line, referred to as NoDice cells, to reconstitute various Dicer isoforms that have thus far been described in a variety of settings (e.g., stem cells and oocytes). Collectively, these data demonstrate the capacity of certain N-terminal truncations of Dicer to inhibit Sindbis virus and reduce the presence of viral dsRNA, supporting some of the observations made thus far concerning an antiviral role for mammalian Dicer. For other viruses, this impact was significantly more modest (SFV reduction is less than a log) or was not observed at all (VSV and SARS-CoV-2). The authors then go on to characterize the nature of the observed antiviral activity and ultimately implicate PKR and the induction of NF-kB in priming the cell's antiviral defenses. Importantly, the group also found that this antiviral activity neither required the nuclease activity of Dicer nor the kinase activity of PKR - providing evidence against antiviral RNAi in mammals. In all, the data would seem to suggest that Dicer can act as a dsRNA sensor and can mediate the activation of an NF-kB response - akin to what is observed in response to NOD or some TLR engagement. In all, it is the opinion of this reviewer that this work brings additional clarity to a concept that remains controversial in the field and therefore embodies something meaningful for the community.

    With that said, there are a few issues that require additional attention. The first of these is textual. The introduction of the paper accurately describes the evidence in support of mammalian RNAi but does not invest the same time in discussing the data to the contrary. For example, Seo et al demonstrated that virus infection results in poly-adp-ribosylation of RISC preventing RNAi activity (PMID: 24075860), Uhl et al showed that IFN-induced ADAR1 resolves dsRNA in the cell and prevents RNAi (PMID: 37017521), and Tsai et al showed that virus-derived small RNAs are not loaded into the RISC in a manner that would enable antiviral activity (PMID 29903832). None of this work is referenced in this manuscript and it generates an unbalanced introduction as it relates to the controversy surrounding the idea of RNAi in mammals.

    The second issue that would further strengthen this paper relates to the fact that the authors spend a considerable amount of time discussing the data of Figure 6 and 7 as conditions that are defined by a Dicer that can not be processive in its nuclease activity (WT) vs. one that can (N1). However, there seems to be little consideration about the fact that the introduction of WT Dicer into these cells also restores miRNA biology whereas N1 appears to remain only partially functional (based on the data of Fig 3E). Given this, it seems the authors should verify that the high baseline of NFkB signaling that is being observed when comparing WT to N1 is not a product of restored miRNA function in WT cells, in contrast to the hypotheses outlined in the manuscript. This could be addressed by silencing Drosha or DGCR8 in the Dicer knockout cells prior to their reconstitution of Dicer. In the opinion of this reviewer, this experimental control would significantly strengthen the conclusions the authors are making here.

    Significance

    In the manuscript entitled, "Canonical and non-canonical contributions of human Dicer helicase domain in antiviral defense" Baldaccini et al. describe their findings concerning the ability of certain N-terminal deletion variants of Dicer in contributing to mammalian antiviral activity. The concept of a functional antiviral RNAi system in mammals is a contentious one with many publications including data both in support of its existence and opposing this idea. In this manuscript, Baldaccini et al. perform a wide range of well-controlled experiments to specifically address aspects of those reports to both provide clarity in what has been documented thus far and to expand on those concepts further.

  6. Version published to 10.1101/2023.08.23.554407v1 on bioRxiv