Geminiviruses employ host DNA glycosylases to subvert DNA methylation-mediated defense

DNA methylation is an epigenetic mechanism that plays important roles in gene regulation and transposon silencing. Active DNA demethylation has evolved to counterbalance DNA methylation at many endogenous loci. Here, we report that active DNA demethylation also targets viral DNAs, tomato yellow leaf curl China virus (TYLCCNV) and its satellite tomato yellow leaf curl China betasatellite (TYLCCNB), to promote their virulence. We demonstrate that the βC1 protein, encoded by TYLCCNB, interacts with a ROS1-like DNA glycosylase in Nicotiana benthamiana and with the DEMETER (DME) DNA glycosylase in Arabidopsis thaliana. The interaction between βC1 and DME facilitates the DNA glycosylase activity to decrease viral DNA methylation and promote viral virulence. These findings reveal that active DNA demethylation can be regulated by a viral protein to subvert DNA methylation-mediated defense.

In this paper, Xiaojian Gui and colleagues provide compelling evidence that active viral demethylation is critical to promote virulence in coinfections by geminiviruses and satellite DNA. They show that the βC1 protein of the satellite TYLCCNB interacts with ROS1L and DEMETER (DME) glycosylases in Nicotiana benthamiana and Arabidopsis thaliana, respectively, to decrease viral DNA methylation and promote viral accumulation and virulence. This is a very nice, elegant and convincing paper. Very well done and very well written. I don't see any flaws and the data is of great significance in the field. I only have a few minor comments to do. 1) L76. In these experiments we can see the effect of TYLCCNB on TYLCCNV-infected plants that have been previously infiltrated with TRV-GUS or TRV-NbDMLs. I miss however a TRV-free control for comparison (as shown in Fig. 3). This control could be important as it has been shown that TRV interferes with DNA de/methylation (Diezma-Navas et al. Mol Plant Pathol. 2019 Oct;20(10):1439-1452. doi: 10.1111/mpp.12850. Epub 2019. Is there any possible scenario by which TRV itself could influence the outcome of subsequent infections in this experiment?
2) L124. Without a proper TYLCCNV-infected control (this control is lacking) they cannot discard a transcriptional activation of NbDML in TYLCCNV-infected plants, and that NbDML activation be further modulated when TYLCCNB is co-infecting.
3) Interestingly, viral accumulation in TYLCCNV+B-infected plants remains much higher than controls (TYLCCNV-infected plants) when βC1 carries the V17A mutation ( Fig. 3C) or when DMLs are silenced (Fig. 1C). It seems likely that ROS1L and βC1 partially retain their functions in both cases. However, methylation levels are comparable to that observed in TYLCCNV-infected controls. Is it possible that additional mechanisms (perhaps involving other DMLs) may assist βC1 in promoting virulence? 4) Viral DNA methylation in the presence of βC1(V17A) is similar to that observed in the TYLCCNV control. This mutation does not compromise the interaction between βC1 and SAHH. This seems to suggest that βC1-mediated suppression of DNA methylation via SAHH interaction is negligible (only CHH methylation is slightly reduced). However, is it possible that suppression of DNA methylation by SAHH and active demethylation by ROS1L/DME are coupled in a way that the reduced methylation observed TYLCCNV/BSCTV+B-infected plants is a consequence of both activities? 5) Please, the 35S::RFP-H2B transgenic N. benthamiana plants are not mentioned in the main text. Please, describe briefly and indicate what they were used for. In general, details provided in the methods section are insufficient to reproduce the experiments.

Reviewer #2 (Remarks to the Author):
This is review of the manuscript "Geminiviruses hijack host DNA glycosylases to subvert DNA methylation-mediated defense" submitted by Gui and colleagues to Nature Communications. In this study, the authors report that active DNA demethylation can target vial DNAs to promote their virulence. They also found that βC protein could directly interact with NbROS1L and DME from Arabidopsis, and enhanced the DNA glycosylase activity in vitro. In addition, they demonstrated that the protein interaction between βC and DME was essential for promote vial virulence. This study suggested a novel function of active DNA demethylation in plants. However, the evidence to support their hypothesis is not enough so far.
Specific comments: 1. The DNA methylation changes are not significant enough to support their hypothesis. The methylation results in Fig. 1d, Fig. 3d, Fig. 4e, and Fig. 5f are misleading since the DNA methylation only occurs in several clones (Fig. S3,7,10,and 13). The DNA methylation pattern is also weird. Furthermore, subclones from PCR should be removed during bisulfite sequencing (e.g. Fig. S13).
2. Since βC specifically interact with NbROS1L, but not other member of the family, the authors should specifically silencing NbROS1L by VIGS in Fig.1.
3. The authors should use DME knock down (RNAi) mutant in Arabidopsis to test the vial virulence to further support their hypothesis.
4. So far, the authors still can not fully exclude the possibility that active DNA demethylation promote vial virulence through targeting the endogenous genes. In the same, βC may interact with other unknown components to enhance vial virulence.

Gui et al., 2021
The manuscript presented by Gui and colleagues describe that a geminiviral protein (βC1 from TYLCCNB, a DNA satellite associated to tomato yellow leaf curl China virus, TYLCCNV) interacts with the host DNA glycosylases, NbROS1L and AtDME and increase the demethylation activity of the latter in vitro and in vivo. The authors showed that active DNA demethylation is important for the virulence of TYLCCNV/TYLCCNB and for the level of DNA methylation at the IR. The presence of the satellite DNA induces viral accumulation and reduces the level of viral DNA methylation, whereas these observations disappear if the plant´s DME-like (DMLs) DNA glycosylases are silenced. The manuscript shows convincing data for the interaction of βC1 with the N-terminal of both DNA demethylases, NbROS1 and AtDME, by two different approaches (BiFC and Co-IP) and demonstrate that Valine 17 from βC1 is essential for these interactions. The decrease in viral DNA methylation induced by the DNA satellite or the presence of βC1 is lost in a βC1V17A mutant (DNA methylation comes to levels similar to the when βC1 is not present). Moreover, the authors showed that βC1 promotes AtDME activity both in vitro and in vivo at the viral genome, and consequently, plants that overexpress AtDME and contain the tandem TYLCCNV/TYLCCNB accumulate more viral (and satellite) DNA and its genome nearly losses all DNA methylation. The findings of this work, which I believe are strongly supported by the way the data is presented, are relevant and interesting for the scientific community. Therefore, I feel that the manuscript is suitable for publication in Nature Communications but the following concerns should be addressed before publication: 1) The introduction is a bit short and significant information is missing, so it should be improved.
-Authors should include the findings from Stroud et al., 2013 (Cell) (which surprisingly is not mentioned in the text) and explain more precisely how DNA methylation is stablished and maintained. In the actual version, just a few lines are used to explain this complex process (lines 37-42) -Is reference 13 correctly used in line 51? To my knowledge Lister at al., 2008 (Cell) or Stroud et al., 2013 (Cell) will be more appropriate.
-line 54: "central cells" of? Explain better -Although the word "geminivirus" is in the title of the paper, it is not mentioned in the introduction. The authors talk about "plant viruses" and begomoviruses and give many references related to geminivirus but the section in the introduction about geminivirus should be completed and improved. Moreover, a paragraph indicating the multifunctionally of βC1 protein should be included 3) Figure 1b suppl.: Is DML6 expressed in seeds? It is not observed at cycle 30. Please show a better image or mention in the text that this transcript is not detected 4) From the information in Table1-Supplemental, I understand that the TRV-NbDMLs used in Figure 1 contains a fragment that is able to silence the six DME-like (DMLs) DNA glycosylases from Nicotiana benthamiana. If this is the case, it should be specifically indicated in the text (line 85). Authors should include in Figure 2a-supplementary an alignment of the fragment cloned in TRV and show and the putative sRNA hits in the six genes according to VIGS tools from Solgenomics (screen capture will be enough).
5) The statement "TYLCCNB promotes TYLCCNV infection…" (lines 100-101) has already been demonstrated. Please rewrite this paragraph and indicate the reference. 10) βC1 interacts with NbROS1L and AtDME through their N-terminal domain and the interaction between βC1 and AtDME and NbROS1L is lost in the βC1V17A mutant. Therefore, the authors should mention in the discussion that the lack of interaction between βC1V17A and At ROS1, AtDML2 and/or At DML3 could also be involved in the results obtained in 12) The statement on line 207 is not accurate: "overexpression of DME enhances TYLCCNV infection" as the presence of βC1 or TYLCCNB is also needed. Please rephrase and explain rigorously.
13) The authors showed that βC1V17A mutant is able to interact with SAHH but is this mutant able to suppress gene silencing? Are there any other functions of βC1 affected in this mutant? Authors should mention this in the discussion 14) Authors do not show if the interaction between βC1and AtDME or NbROS1 have any impact on the host genes controlled by these demethylases. Therefore, the verb "hijack" in the tittle seems too pretentious to me. As the authors have not shown that the interaction between BC1 and DME alters its activity in the host genome, they should rephrase the tittle of the manuscript.
15) The authors abuse of the use of "short sentences" throughout the manuscript that do not help the readers to follow the rationale behind some of the statements or experiments. Here are some examples but the authors should review the whole manuscript and make longer comprehensive sentences. We appreciate the constructive comments made by the reviewers. We have provided additional data and revised our manuscript to address the concerns raised by the reviewers. The revisions are highlighted. We wish the revisions are sufficient and the manuscript is now acceptable for publication. Point-bypoint responses are listed below.

Reviewer #1
In this paper, Xiaojian Gui and colleagues provide compelling evidence that active viral demethylation is critical to promote virulence in coinfections by geminiviruses and satellite DNA. They show that the βC1 protein of the satellite TYLCCNB interacts with ROS1L and DEMETER (DME) glycosylases in Nicotiana benthamiana and Arabidopsis thaliana, respectively, to decrease viral DNA methylation and promote viral accumulation and virulence. This is a very nice, elegant and convincing paper. Very well done and very well written. Response: We thank the reviewer for the comment. We have added the TRVfree control (inoculated with pCambia1300) in Fig. 1 and Supplementary Fig. 3.
The symptoms and molecular phenotypes developed by TRV-GUS-inoculated plants were exactly the same as those developed by pCambia1300-inoculated plants, suggesting that TRV itself does not influence the outcome of TYLCCNV and TYLCCNA+B infections. Our results are consistent with previous findings (Li et al., 2014). Diezma-Navas et al found that the up-regulation of ROS1 could be observed 7 days after TRV infection (Diezma-Navas et al., 2019). However, our results showed that the expression levels of DNA demethylases, including NbROS1, were comparable between TRV-GUS-and mock-infiltrated plants ( Supplementary Fig. 3a). We speculated that the difference could result from the difference in the plant tissues used. Systemically infected leaves were used in our experiment, while inflorescences were harvested for gene expression analyses by Diezma-Navas and colleagues.
2) L124. Without a proper TYLCCNV-infected control (this control is lacking) they cannot discard a transcriptional activation of NbDML in TYLCCNV-infected plants, and that NbDML activation be further modulated when TYLCCNB is coinfecting.

Response:
We agree with the reviewer's concern and TYLCCNV-inoculated control has been added in the experiment ( Supplementary Fig. 4a).
3) Interestingly, viral accumulation in TYLCCNV+B-infected plants remains much higher than controls (TYLCCNV-infected plants) when βC1 carries the V17A mutation (Fig. 3C) or when DMLs are silenced (Fig. 1C). It seems likely that ROS1L and βC1 partially retain their functions in both cases. However, methylation levels are comparable to that observed in TYLCCNV-infected controls. Is it possible that additional mechanisms (perhaps involving other DMLs) may assist βC1 in promoting virulence?

Response:
We thank the reviewer for the comment. As shown in the original figures, the methylation (especially CHH methylation) levels in TYLCCNV+Binfected plants were reproducibly, albeit mildly, lower than TYLCCNV-infected plants when βC1 carries the V17A mutation (Fig. 3d) or when NbDMLs are silenced (Fig. 1d). We have repeated the experiments three more times and the new data confirmed previous results (new Fig. 1d and Fig. 3d). Considering that βC1 can target SAHH to suppress DNA methylation, we speculate that this is because SAHH activity is suppressed in TYLCCNV+B V17A -inoculated plants or TRV-NbDMLs/TYLCCNV+B-inoculated plants, but not in TYLCCNV-inoculated plants. Our results suggest that βC1-mediated activation of NbROS1L activity and suppression of SAHH activity both contribute to βC1-mediated suppression of DNA methylation.
TYLCCNV DNA methylation is lower in TYLCCNV+B V17A -inoculated plants or TRV-NbDMLs/TYLCCNV+B-inoculated plants than in TYLCCNV-inoculated plants. This could lead to the results that viral accumulation remains higher in TYLCCNV+B V17A -inoculated plants or TRV-NbDMLs/TYLCCNV+B-inoculated plants than in TYLCCNV-inoculated plants (Fig. 1b, c, Fig. 3b, c). Considering that βC1 is a multifunctional virulence factor that can target different pathways to subvert host defense, we speculate that other factors may also assist βC1 in promoting virulence.
4) Viral DNA methylation in the presence of βC1(V17A) is similar to that observed in the TYLCCNV control. This mutation does not compromise the interaction between βC1 and SAHH. This seems to suggest that βC1-mediated suppression of DNA methylation via SAHH interaction is negligible (only CHH methylation is slightly reduced). However, is it possible that suppression of DNA methylation by SAHH and active demethylation by ROS1L/DME are coupled in a way that the reduced methylation observed TYLCCNV/BSCTV+B-infected plants is a consequence of both activities?
Response: Please see our response to this reviewer's comment #3. to mark the nucleus. We only showed the YFP channel and all the BiFC data will be uploaded as source data. We have provided more details in the methods.

Reviewer #2
This is review of the manuscript "Geminiviruses hijack host DNA glycosylases to subvert DNA methylation-mediated defense" submitted by Gui and colleagues to Nature Communications. In this study, the authors report that active DNA demethylation can target vial DNAs to promote their virulence. They also found that βC protein could directly interact with NbROS1L and DME from Arabidopsis, and enhanced the DNA glycosylase activity in vitro. In addition, they demonstrated that the protein interaction between βC and DME was essential for promote vial virulence. This study suggested a novel function of active DNA demethylation in plants. However, the evidence to support their hypothesis is not enough so far.
Specific comments: 1. The DNA methylation changes are not significant enough to support their hypothesis. The methylation results in Fig. 1d, Fig. 3d, Fig. 4e, and Fig. 5f are misleading since the DNA methylation only occurs in several clones (Fig. S3, 7, 10, and 13). The DNA methylation pattern is also weird. Furthermore, subclones from PCR should be removed during bisulfite sequencing (e.g. Fig.   S13).

Response:
We thank the reviewer for the comment. To address the concern that the percent of methylated cytosines is affected by the number of clones that are sequenced, we amplified the IR regions of TYLCCNV or BSCTV after bisulfite treatment and the PCR products were subjected to library construction  (Fig. 1d, Fig. 3d, Fig. 4e and Fig. 5f). We agree with the reviewer that the CG, CHG, and CHH methylation levels of 2. Since βC specifically interact with NbROS1L, but not other member of the family, the authors should specifically silencing NbROS1L by VIGS in Fig.1.

Response:
We thank the reviewer for the suggestion. However, the sequences of NbROS1L and other members of the family are highly similar, especially high sequence similarity between NbROS1L and NbROS1 ( Supplementary Fig. 2a), making it very difficult to silence NbROS1L only.

The authors should use DME knock down (RNAi) mutant in Arabidopsis to
test the vial virulence to further support their hypothesis.

Response:
We agree with the reviewer that using DME knock down (RNAi) mutant in Arabidopsis to test viral virulence can further support our hypothesis.
We have tried several strategies to knock down DME in Arabidopsis. However, we have not yet obtain RNAi lines in which DME is efficiently knocked down, very likely because DME is essential for endosperm development (Choi et al., 2002).
4. So far, the authors still cannot fully exclude the possibility that active DNA demethylation promote vial virulence through targeting the endogenous genes.
In the same, βC1 may interact with other unknown components to enhance vial virulence.

Response:
We agree with the reviewer that we cannot exclude the possibility that active DNA demethylation could also target endogenous genes to promote vial virulence and that βC1 may interact with other unknown components to enhance vial virulence. Our study provides strong evidence that NbROS1L could be employed to promote virulence through demethylating TYLCCNV IR DNA region in the presence of βC1. We have observed that NbDMLs knockdown leads to consistent decreased virulence accompanied by increased DNA methylation of TYLCCNV, that overexpression of DME leads to increased virulence accompanied by decreased TYLCCNV DNA methylation, that βC1 interacts with NbROS1L as well as DME, and that βC1 promotes the demethylase activity of DME. All of these point to the conclusion that DML could be employed by the geminivirus to facilitate the infection. Furthermore, the requirement of βC1 on DML virulent function can be supported by the loss of interaction between NbROS1L and βC1 V17A , the absence of NbROS1Lmediated virulence change and DNA methylation changes with inoculation of TYLCCNV only, the strongly attenuated NbROS1L-mediated virulence change and DNA methylation changes accompanied by mutated βC1 V17A when compared to wild-type βC1.
To determine how many endogenous genes were affected when βC1 was expressed, we performed whole-genome bisulfite sequencing and compared the DNA methylation profiles of Col-0 and βC1-expressing plants. We identified 1140 hypo-differentially-methylated regions (hypo-DMRs) in βC1-expressing plants (Fig.1a in this response). Distribution analysis of these hypo-DMRs revealed that they were highly enriched in CDS, intron and intergenic region ( Fig.1b in this response). GO analysis of hypo-DMR-associated genes showed that biological processes associated with these genes were not relevant to viral virulence ( Fig.1c in this response), suggesting that βC1 and active DNA demethylation promote vial virulence less likely through targeting many endogenous genes. However, it is still possible that hypomethylation of some genes contributes to βC1-mediated promotion of viral virulence and further research is needed to investigate this possibility. We have added some discussion about this in the revised manuscript and will investigate this possibility in another study.
βC1 V17A significantly compromised the reduction of viral DNA methylation as compared to βC1 (Fig. 3c, Fig. 4e), suggesting that the interaction between βC1 and NbROS1L or DME contributed much to the reduction of viral DNA methylation and the increase in viral virulence. As we discussed in response to Reviewer #1's comments #3 and #4, βC1 may also suppress SAHH activity to reduce viral DNA methylation and increase viral virulence. 5. Line 182, produce β-and β,δ-elimination products.

Response:
We thank the reviewer for pointing out this mistake, which has been corrected.

Response:
We thank the reviewer for pointing out this mistake, which has been corrected.

Reviewer #3
Gui et al., 2021 The manuscript presented by Gui and colleagues describe that a geminiviral The findings of this work, which I believe are strongly supported by the way the data is presented, are relevant and interesting for the scientific community.
Therefore, I feel that the manuscript is suitable for publication in Nature Communications but the following concerns should be addressed before publication: 1) The introduction is a bit short and significant information is missing, so it should be improved. -

Response:
We tried several times to detect the expression of the DNA demethylases by western blot but did not yield strong signals for full-length proteins. We thus used RT-PCR instead to show that they were expressed at comparable levels ( Supplementary Fig. 4c). Response: The data have been provided in Supplementary Fig. 9a in the original manuscript, and now they are shown in Supplementary Fig. 8a.

7) Authors
10) βC1 interacts with NbROS1L and AtDME through their N-terminal domain and the interaction between βC1 and AtDME and NbROS1L is lost in the βC1V17A mutant. Therefore, the authors should mention in the discussion that the lack of interaction between βC1V17A and At ROS1, AtDML2 and/or At DML3 could also be involved in the results obtained in figure 4.

Response:
Our BiFC results revealed that βC1 does not interact with AtROS1, AtDML2 and/or AtDML3, Thus, loss of interaction between βC1 V17A and AtROS1, AtDML2 and/or AtDML3 could not account for the results obtained in

Response:
We thank the reviewer for pointing out this, and they have been adjusted.
12) The statement on line 207 is not accurate: "overexpression of DME enhances TYLCCNV infection" as the presence of βC1 or TYLCCNB is also needed. Please rephrase and explain rigorously.

Response:
We agree with the reviewer that the presence of βC1 or TYLCCNB is also needed. The second half of the sentence in the original manuscript indicated this.
13) The authors showed that βC1V17A mutant is able to interact with SAHH but is this mutant able to suppress gene silencing? Are there any other functions of βC1 affected in this mutant? Authors should mention this in the discussion Response: Our bisulfite sequencing results showed that viral DNA methylation in TYLCCNV+B V17A -inoculated plants was lower than that in TYLCCNVinoculated plants (Fig. 3d) and viral DNA methylation in βC1 V17A -OX lines was lower than that in Col-0 (Fig. 4e), suggesting that βC1 V17A can still suppress DNA methylation by interacting with SAHH. We have added more discussions about this in the revised manuscript. It remains unclear whether other functions of βC1 are affected.
14) Authors do not show if the interaction between βC1and AtDME or NbROS1 have any impact on the host genes controlled by these demethylases.
Therefore, the verb "hijack" in the tittle seems too pretentious to me. As the authors have not shown that the interaction between BC1 and DME alters its activity in the host genome, they should rephrase the tittle of the manuscript.

Response:
We agree with the reviewer and we have rephrased the title.
15) The authors abuse of the use of "short sentences" throughout the manuscript that do not help the readers to follow the rationale behind some of This a second review of a previously submitted manuscript by Xiaojian Gui and colleagues entitled Geminiviruses employ host DNA glycosylases to subvert DNA methylation-mediated defense. I was very much enthusiastic about the first version of this paper, but I agree that the comments made by the reviewers and the manner these comments have been addressed in this revised version have contributed to improve greatly the overall quality of the work. Particularly, I am very much satisfied with their response to my comments and the correction made when necessary. I feel they have also provided solid arguments in response to the other reviewers' concerns. Again, I don't see any relevant flaws in the data analysis and interpretation, and the conclusions are fully supported by their results. Now, the methodology is comprehensive. In conclusion, I fully support this paper for publication.
Reviewer #2 (Remarks to the Author): The authors have successfully addressed all of my concerns.
Reviewer #3 (Remarks to the Author): The manuscript has clearly been improved after the changes introduced by the authors, following the reviewers' comments. I just have a minor comment from a new sentence that the authors have included in the final version of the introduction: Line 66-67: interfere with the RNA silencing pathway (REF 22,23) As βC1 can suppress DNA methylation through inactivating SAHH (REF 22) but not by interfering with the RNAi machinery that leads to DNA methylation, I believe the sentence in line 66-67 should be corrected as follows: interfere with post-transcriptional and transcriptional gene silencing (REF 23,22) In my opinion, this final form of the manuscript could be published in Nature Communications.
We appreciate the constructive comments made by the reviewers. We wish the revisions are sufficient and the manuscript is now acceptable for publication.
Point-by-point responses are listed below.

Reviewer #1
This a second review of a previously submitted manuscript by Xiaojian Gui and colleagues entitled Geminiviruses employ host DNA glycosylases to subvert DNA methylation-mediated defense. I was very much enthusiastic about the first version of this paper, but I agree that the comments made by the reviewers and the manner these comments have been addressed in this revised version have contributed to improve greatly the overall quality of the work. Particularly, I am very much satisfied with their response to my comments and the correction made when necessary. I feel they have also provided solid arguments in response to the other reviewers' concerns. Again, I don't see any relevant flaws in the data analysis and interpretation, and the conclusions are fully supported by their results. Now, the methodology is comprehensive. In conclusion, I fully support this paper for publication.

Response:
We thank the reviewer for his/her positive assessment of our manuscript.

Reviewer #2
The authors have successfully addressed all of my concerns.

Response:
We thank the reviewer for his/her positive assessment of our manuscript.

Reviewer #3
The manuscript has clearly been improved after the changes introduced by the authors, following the reviewers' comments. I just have a minor comment from a new sentence that the authors have included in the final version of the introduction: Line 66-67: interfere with the RNA silencing pathway (REF 22,23) As βC1 can suppress DNA methylation through inactivating SAHH (REF 22) but not by interfering with the RNAi machinery that leads to DNA methylation, I believe the sentence in line 66-67 should be corrected as follows: interfere with post-transcriptional and transcriptional gene silencing (REF 23,22) In my opinion, this final form of the manuscript could be published in Nature

Response:
We thank the reviewer for his/her positive assessment of our manuscript. We agree with the reviewer and we have changed the sentence in line 66-67 as follows: The multifunctional protein βC1 can block MAPK signaling, interfere with post-transcriptional and transcriptional gene silencing and manipulate jasmonic acid signaling to subvert host defense.