MYB96 recruits the HDA15 protein to suppress negative regulators of ABA signaling in Arabidopsis

Unlike activation of target genes in response to abscisic acid (ABA), how MYB96 transcription factor represses ABA-repressible genes to further enhance ABA responses remains unknown. Here, we show MYB96 interacts with the histone modifier HDA15 to suppress negative regulators of early ABA signaling. The MYB96-HDA15 complex co-binds to the promoters of a subset of RHO GTPASE OF PLANTS (ROP) genes, ROP6, ROP10, and ROP11, and represses their expression by removing acetyl groups of histone H3 and H4 from the cognate regions, particularly in the presence of ABA. In support, HDA15-deficient mutants display reduced ABA sensitivity and are susceptible to drought stress with derepression of the ROP genes, as observed in the myb96-1 mutant. Biochemical and genetic analyses show that MYB96 and HDA15 are interdependent in the regulation of ROP suppression. Thus, MYB96 confers maximal ABA sensitivity by regulating both positive and negative regulators of ABA signaling through distinctive molecular mechanisms.

need to carry out the CoIP assays in Arabidopsis using the native promoters driven constructs to make sure that the observed protein-protein interaction occurs under nature conditions.
2. More evidence is required to support the authors' claim that MYB96 recruit HDA15 to regulate gene expression. The authors may want to carry out ChIP-seq analysis to compare the global binding of these proteins, which will provide evidence whether they co-target to the overlapping set of genes.
3. The authors also need to generate and analyze the myb96 hda15 double mutant to further investigate the function and interaction of HDA15 and MYB96.
4. ChIP-qPCR analysis showing that MYB96 and HDA15 bind to ROS genes were carried out using 35S:MYB96-MYC and 35S:HDA15-GFP transgenic plants. The authors need to carry out the ChIP-qPCR analysis using the native promoters driven constructs to make sure that MYB96 and HDA15 bind to ROS genes under nature conditions. 5. In Figure 6, the authors can also co-transform p35S::MYB96 and p35S::HDA15 with their reporter constructs in both Col and had15 protoplasts in transient expression assays. This experiment will further confirm whether HDA15 affects MYB96-repressed genes only.
Minor points 1. Figure 1D The nuclear maker is required to support the claim that MYV96 and HDA15 interacts in the nucleus.

Figure 2 E-G
Gene expression was analyzed in seedlings treated with ABA for 6 h, I suggest that the authors may want to analyze more time points such as 12 and 24 h.
3. Figure 5 Since HDA15 affects both histone H3 and H4 deacetylation, the authors may also want to analyze whether H4ac accumulation at the ROP promoters in the myb96 and hda15 mutants was also affected. 4. Page 9 -"Reporter constructs, in which the ROP and KETOACYL-COA SYNTHASE (KCS) promoter sequences were fused with the minimal 35S promoter".
Please explain why the KCS promoter was selected for making reporter constructs. 5. Figure 6 Please explain what are pmKCS and pmROP.
Reviewer #3 (Remarks to the Author): The work from Lee and Seo describes the interaction between the transcription factor MYB96 and the Histone Deacetylase HDA15 and how this interaction regulates expression of ABA responsive genes as well plant growth under drought conditions. The premises are interesting, however the manuscript lacks of controls, some results are inconsistent and some conclusions are contradictory.    What about testing genes for the germination phenotype? If the opposite effect might be due to the dual function of MYB96 (repressor in some cases and activator in others) testing gene response is important for the model.  In figure A the authors look at the acetylation levels at the ROP transcriptional start sites in the myb96 and hda15 mutants and in their respective overexpressing lines. First there no increase in the acetylation levels for both mutants in control conditions compared to the wild type. This result is already difficult to interpret since in case of loss of histone de-acetylation you would expect an increase in acetylation. Usually histone acetylation positively correlates with gene expression, however in this case there is no correlation as in figure 3 we observe an increase in gene expression in both mutants. Furthermore for the overexpressing lines there is a decrease in acetylation which is consistent with the expression data, however this decrease occurs in different promoter regions, not just on the areas where the authors show binding of MYB96 and HDA15.
To further corroborate the model it would be interesting to see the effect in the double mutant.
Again, if the main point of this manuscript is the role of the interaction between MYB96 and HDA15, each experiment should provide data looking at the effect of both genes.

Responses to Reviewer #1:
The most significant problem is that the relationship between MYB96 and HDA15 is still not confirmed genetically. They used myb96 and HDA15 knockout mutants, or MYB96OX/hda15 transgenic plants, to show their relationship. Why don't you use myb96 hda15 double mutant for phenotypic analysis, gene expression and ChIP analysis? The double mutant will provide concrete evidence for the functional interactions between MYB96 and HDA15 in plants. It will be required for phenotypic analysis, gene expression and ChIP assays.
[Other comments] 1. In Fig. 1c and d, AD-MYB96 and BD-HDA15 were used in Y2H assay. Have you ever swapped AD and BD?
 We used the domain-swapped constructs, but it was difficult to see the results, because the BD-MYB96 fusion had strong self-transcriptional activation activity and masked the GAL4 activation induced by physical interactions between MYB96 and HDA15. We thus carried out alternative assays, including BiFC and Co-IP, to confirm the physical interactions. Please understand our reasoning. Fig. 1e, Co-IP showed in vivo interaction between MYB96 and HDA15. It is possible that ABA can affect their interactions. Authors should check that. And the resolution of image is too low. High resolution images should be used in this figure.

In
 We repeated the Co-IP assays and replaced the images with high resolution images (see Figure 1e). In addition, we also examined the interaction of MYB96 with HDA15 in the presence of different concentrations of ABA and found that they interact with each other in a dose-dependent manner. The data were newly included as Figure 3a.
3. In Fig. 4a, 35S-MYB96/hda15-1 should be required.  4. Methods should be described in detail, for example, how much concentrations, incubation/reaction time, etc. were used for experiments, especially for ChIP assay. Such information must be important to reproduce the results.

 We described our experiment protocols in more detail in the Materials and Methods section, as suggested.
[Minor points] 1. P.14, L311-316, descriptions about the MYB96-HDA15 interaction and light signaling is too strong.
 We revised and toned-down the sentences to avoid being misleading.
 We corrected the mistake.
 We included a relevant citation to provide information.

P.16, L382, What is pSATN?
 We included a relevant citation to provide information.

P17, What is the vector name used for transient expression analysis?
 We modified the descriptions and included a citation to provide information.

Responses to Reviewer #2:
Major points 1. CoIP assays for protein-protein interaction were carried out using N. benthamiana. The authors need to carry out the CoIP assays in Arabidopsis using the native promoters driven constructs to make sure that the observed protein-protein interaction occurs under nature conditions. 2. More evidence is required to support the authors' claim that MYB96 recruit HDA15 to regulate gene expression. The authors may want to carry out ChIP-seq analysis to compare the global binding of these proteins, which will provide evidence whether they co-target to the overlapping set of genes.
 We fully agree with the reviewer's comment. Global binding analysis of MYB96 and HDA15 and the direct comparison of genome-wide association would add valuable evidence for our conclusion. Genome-wide HDA15 binding in wild-type and myb96-1 backgrounds may also be relevant. Although we have attempted several times, it was difficult to get a comprehensive view from the global data. Instead, it was increasingly clear that the MYB96-HDA15 interaction is obvious at least at the ROP loci. We would like to focus more on this aspect in this manuscript. Please understand our reasoning.
3. The authors also need to generate and analyze the myb96 hda15 double mutant to further investigate the function and interaction of HDA15 and MYB96. 4. ChIP-qPCR analysis showing that MYB96 and HDA15 bind to ROP genes were carried out using 35S:MYB96-MYC and 35S:HDA15-GFP transgenic plants. The authors need to carry out the ChIP-qPCR analysis using the native promoters driven constructs to make sure that MYB96 and HDA15 bind to ROP genes under nature conditions.  We performed ChIP-qPCR analysis using transgenic plants expressing the native promoter constructs, as suggested. The results were newly included as Supplementary Figure S8. Figure 6, the authors can also co-transform p35S::MYB96 and p35S::HDA15 with their reporter constructs in both Col and had15 protoplasts in transient expression assays. This experiment will further confirm whether HDA15 affects MYB96-repressed genes only.

In
 Transient expression assays in the combinations suggested by the reviewer were carried out. We appreciate the suggestion and were able to add additional solid evidence that HDA15 selectively regulates MYB96-repressed genes. The data were newly included in Supplementary Figure S9.

Minor points
1. Figure 1D, The nuclear maker is required to support the claim that MYV96 and HDA15 interacts in the nucleus. 3. Figure 5, Since HDA15 affects both histone H3 and H4 deacetylation, the authors may also want to analyze whether H4ac accumulation at the ROP promoters in the myb96 and hda15 mutants was also affected.
 We examined both H3 and H4 deacetylation in myb96-1, hda15-1, and myb96-1hda15-1 mutants, as suggested. Please see Figure 6a and 6b. 4. Page 9 -"Reporter constructs, in which the ROP and KETOACYL-COA SYNTHASE (KCS) promoter sequences were fused with the minimal 35S promoter". Please explain why the KCS promoter was selected for making reporter constructs.
 The KCS genes are known to be directly activated by MYB96. Thus, we suspected that comparison of the KCS and ROP genes is a good model to show how MYB96 facilitates opposite roles in gene regulation. A concise explanation was provided to address this issue. Please see page 11, 2 nd paragraph. Figure 6, Please explain what are pmKCS and pmROP.         The MYB96-HDA15 complex primarily binds to the promoter regions. We thus included a gene body region as a negative control, as suggested. We also tested H3ac and H4ac levels at the control region for each locus. Please see Figure 5a, 5b, 6a, and 6b, and Supplementary Figure S8. Figure 5: The data in this figure are quite inconsistent with each other. In figure A the authors look at the acetylation levels at the ROP transcriptional start sites in the myb96 and hda15 mutants and in their respective overexpressing lines. First there no increase in the acetylation levels for both mutants in control conditions compared to the wild type. This result is already difficult to interpret since in case of loss of histone de-acetylation you would expect an increase in acetylation. Usually histone acetylation positively correlates with gene expression, however in this case there is no correlation as in figure 3 we observe an increase in gene expression in both mutants. Furthermore for the overexpressing lines there is a decrease in acetylation which is consistent with the expression data, however this decrease occurs in different promoter regions, not just on the areas where the authors show binding of MYB96 and HDA15. To further corroborate the model it would be interesting to see the effect in the double mutant.

5.
 Thank you very much for your suggestion. To overcome the inconsistency, we performed more experiments with 8 biological replicates. The original results were replaced with the final data. Since H4 acetylation levels are also related to gene expression and H4 acetylation is also influenced by HDA15, we also included data showing accumulation of H4 acetylation at each locus. We also analyzed the myb96-1hda15-1 double mutant, as suggested. In addition, we intensively compared the MYB96/HDA15-binding regions and control regions of the ROP loci to build a solid conclusion. Please see Figures 6a, 6b, 6c, and 7c.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): Authors significantly revised the manuscript and well answered to all points raised by this reviewer.
Reviewer #2 (Remarks to the Author): The revised manuscript has addressed most of my previous concerns. I have the following suggestions.
1. Figure 1b -Please indicate what the red bars and blue bars represent in the gene construct diagram. The authors claimed that an N-terminal fragment of MYB96 containing the R2R3-MYB DNAbinding domainis responsible for interaction with the N-terminal fragment of HDA15. Dose the Nterminal fragment of HDA15contin any specific domain that may be important for this interaction?
2. There seems to be additive effects between myb96 and hda15 on seed germination and plant survival phenotypes ( Fig. 3b and 3c). However, the additive effect was not observed in the expression and acetylation of ROPs (Fig 4, 6). This needs to be explained and discussed.
3. In Figure 6c, it is understandable that the H3ac level and expression of ROP genes are reduced in 35S:HDA15-GFP plants. However, why are the H3ac level and expression of ROP genes increased in 35S:HDA15-GFP/myb96-1 plants compared to wild type? Please explain. Figure 7c is redundant with Figure 6c and can be removed. 5. How pmKCS and pmROP mutant constructs were generated need to be described in the method section.

4.
6. Page 11, line 261 -"Taken together, the MYB96-HDA15 complex represses the ROPs, which are negative regulators of ABA signaling, to pro 262 mote ABA responses. The ABA-inducible MYB96 protein binds to the promoters of ROP genes and recruits HDA15 to facilitate H3 and H4 deacetylation. In the presence of high concentrations of ABA, the ROP genes are repressed, concomitant with histone deacetylation, in a MYB96-and HDA15-dependent pathway to ensure full activation of ABA responses (Fig. 8)". This part should be moved to the discussion section.

Reviewer #3 (Remarks to the Author):
The manuscript is much improved compared to the previous version and the authors addressed some of my previous concerns. However there is big discrepancy between the data submitted in the former and current version that the authors have not yet addressed. Figure 1: If you can't provide a western blot for the transfected cells, it will be informative to have a western blot for the yeast interaction. Figure 1E: I appreciate that the western has been repeated but an explanation for the previous results is still required, particularly for the presence of a double band. Figure 2: if the effects of ABA during germination and drought tolerance are both due to altered ABA sensitivity, it is possible that HDA15 and MYB96 are involved in different ABA response pathways. Therefore it is important to investigate the molecular mechanism in both pathways. It is not clear to me at what developmental stage the expression of RD22, RD29A and COR47 was assessed and how it relates with the different phenotypic responses. Figure 6: I still have major issues with this figure. First of all the authors replaced the original results with the final data without addressing the discrepancies with the previous version. Furthermore the data from the double mutant phenotype clearly showed an additive effect that is not visible in terms of gene expression in figure 4 and acetylation. Compared to the previous version where the authors investigated the acetylation levels of the regions A-B-C for each locus, now they show data for C-D region, ignoring the changes in acetylation that were previously observed in region B.
1. Figure 1b -Please indicate what the red bars and blue bars represent in the gene construct diagram. The authors claimed that an N-terminal fragment of MYB96 containing the R2R3-MYB DNA-binding domainis responsible for interaction with the N-terminal fragment of HDA15. Dose the N-terminal fragment of HDA15 contin any specific domain that may be important for this interaction?
 We provided more detailed information obtained from multiple public databases. Please see Figure 1b, figure legends, and revised manuscript (page 5,  2nd paragraph).
2. There seems to be additive effects between myb96 and hda15 on seed germination and plant survival phenotypes ( Fig. 3b and 3c). However, the additive effect was not observed in the expression and acetylation of ROPs (Fig 4, 6). This needs to be explained and discussed.
 MYB96 alone can regulate ABA-inducible genes (e.g. KCSs), and likewise HDA15 may also have additional functions in ABA responses indepedently of MYB96. This may explain additive effects of myb96-1hda15-1 on general ABA responses. However, at least in the regulation of ROP genes, both MYB96 and HDA15 are required and consistenly, no additive effects on the gene regulation were observed. We described the situation in the revised manuscript: please see page 9, 1st paragraph. Figure 6c, it is understandable that the H3ac level and expression of ROP genes are reduced in 35S:HDA15-GFP plants. However, why are the H3ac level and expression of ROP genes increased in 35S:HDA15-GFP/myb96-1 plants compared to wild type? Please explain.

In
 HDA15 regulation of acetylation and expression of ROPs depends on MYB96, because MYB96 specifies promoter regions of ROP loci. HDA15 would be less recruited in myb96-1 compared to wild type (low level of MYB96 will allow recruitment of HDA15 in WT) even under normal growth condition, and thus H3ac levels and transcript accumulation were further increased, similar to myb96-1 mutant (please see Figure 4c, 6a, and 6b). Figure 6c and can be removed.

Figure 7c is redundant with
 Figure 6c shows dependence of HDA15 on MYB96 in DNA binding, whereas Figure 7c shows dependence of MYB96 on HDA15 in histone deacetylation-based gene repression. While they look similar, we carefully think that they have different meaning. Therefore, we would like to maintain current organization. Please understand our reasoning.
5. How pmKCS and pmROP mutant constructs were generated need to be described in the method section.
 We provided details about pmKCS and pmROP constructs in the Method section and Supplemetentary Table S3. 6. Page 11, line 261 -"Taken together, the MYB96-HDA15 complex represses the ROPs, which are negative regulators of ABA signaling, to pro 262 mote ABA responses. The ABA-inducible MYB96 protein binds to the promoters of ROP genes and recruits HDA15 to facilitate H3 and H4 deacetylation. In the presence of high concentrations of ABA, the ROP genes are repressed, concomitant with histone deacetylation, in a MYB96-and HDA15-dependent pathway to ensure full activation of ABA responses (Fig. 8)". This part should be moved to the discussion section.
 We reorganized the descriptions in the Discussion section.
Responses to Reviewer 3 1. Figure 1: If you can't provide a western blot for the transfected cells, it will be informative to have a western blot for the yeast interaction.  Protoplasts transfected with the BiFC constructs were harvested to perform western blot analysis. Full-sized proteins were well-expressed, and similar protein levels were detected in each transfected sample. Please see Supplementary Figure S3.
2. Figure 1E: I appreciate that the western has been repeated but an explanation for the previous results is still required, particularly for the presence of a double band.
 Posttranslational modifications of the proteins might be involved, based on the fact that multiple bands are usually detected from epitope-tagged MYB96 and HDA15. Physical interactions most likely do not depend on certain types of protein modifications. Please understand our reasoning.
3. Figure 2: if the effects of ABA during germination and drought tolerance are both due to altered ABA sensitivity, it is possible that HDA15 and MYB96 are involved in different ABA response pathways. Therefore it is important to investigate the molecular mechanism in both pathways. It is not clear to me at what developmental stage the expression of RD22, RD29A and COR47 was assessed and how it relates with the different phenotypic responses.
 In general, if ABA signaling is activated at an upstream point of the pathways, delayed seed germination and enhanced drought tolerance both can be simultaneously induced, as exemplified by overexpression of ABA receptors and SnRKs or mutations in PP2C genes (abi). Considering that ROPs are upstream components of ABA signaling, which act with PP2Cs, the MYB96-HDA15-ROP module most likely regulates diverse ABA responses by promoting the upstream ABA signaling. Also, we included transcript accumulation of ABA signaling marker genes RD22, RD29A, and COR47 at the stage of early seedlings to support our claim that the MYB96-HDA15-ROP module is critical for enhancing ABA signaling, regardless of plant developmental stages (Please see Supplementary Figure S8). As the reviewer indicated, we cannot rule out that the MYB96 and HDA15 proteins may have additional, specific roles in ABA-dependent seed germination and stomatal movement for drought tolerance, but now we would like to emphasize this study that focuses on general ABA responses modulated by the MYB96-HDA15 complex acting at an upstream point. Figure 6: I still have major issues with this figure. First of all the authors replaced the original results with the final data without addressing the discrepancies with the previous version. Furthermore the data from the double mutant phenotype clearly showed an additive effect that is not visible in terms of gene expression in figure 4 and acetylation. Compared to the previous version where the authors investigated the acetylation levels of the regions A-B-C for each locus, now they show data for C-D region, ignoring the changes in acetylation that were previously observed in region B.  Regarding Figure 6, in the previous submission, we chose three promoter regions to check histone acetylation levels. As shown before, they all have similar trends, which may be due to expansion and propagation of chromatin contexts. To improve the quality of the data, we needed to provide negative controls (3'end of gene) to show that ABA-induced changes in H3ac/H4ac levels at ROP loci are particularly observed at the promoters. In addition, we also followed higher standard of ChIP assays by providing absolute values, rather than relative enrichment. By replacing the data, we could tell the regional specificity of histone modification with increased accuracy. This is why we updated the figures. To address the reviewer's comment, we included ChIP data conducted on another promoter region (which was included in previous submission, but updated with more replicates in the current manuscript). Please see Supplementary Figure S11.

4.
In addition, we would like to make interpretation about double mutant phenotypes. In addition to ROPs, MYB96 alone can regulate ABA-inducible genes (Figure 7 and other papers), and likewise HDA15 may also have additional functions in ABA responses indepedently of MYB96. This may explain additive effects of myb96-1hda15-1 on general ABA responses. However, at least in the 'regulation of ROP genes', both MYB96 and HDA15 are required and consistenly, no additive effect on the gene regulation was observed. It is reasonble that ROP genes are not sole regulatory targets of MYB96 and HDA15 for the control of ABA responses. We described the situation in the revised manuscript: please see page 9, 1st paragraph.