DNA methylation repels targeting of Arabidopsis REF6

RELATIVE OF EARLY FLOWERING 6 (REF6/JMJ12), a Jumonji C (JmjC)-domain-containing H3K27me3 histone demethylase, finds its target loci in Arabidopsis genome by directly recognizing the CTCTGYTY motif via its zinc-finger (ZnF) domains. REF6 tends to bind motifs located in active chromatin states that are depleted for heterochromatic modifications. However, the underlying mechanism remains unknown. Here, we show that REF6 preferentially bind to hypo-methylated CTCTGYTY motifs in vivo, and that CHG methylation decreases REF6 DNA binding affinity in vitro. In addition, crystal structures of ZnF-clusters in complex with DNA oligonucleotides reveal that 5-methylcytosine is unfavorable for REF6 binding. In drm1 drm2 cmt2 cmt3 (ddcc) quadruple mutants, in which non-CG methylation is significantly reduced, REF6 can ectopically bind a small number of new target loci, most of which are located in or neighbored with short TEs in euchromatic regions. Collectively, our findings reveal that DNA methylation, likely acting in combination with other epigenetic modifications, may partially explain why REF6 binding is depleted in heterochromatic loci.

I credit the authors for performing the key experiment, which is to perform ChIP-seq of REF6 in the ddcc mutant. Unfortunately, the results do not support that DNA methylation is the major deterministic reason for REF6 binding as 1198/1220 or 98.2% of peaks did not exhibit ectopic binding whereas only 1.8% did. As shown in Figure 4a, the vast majority of ectopic binding events are the weakest binding events given their position in the plot (bottom left in upper quadrant). This further weakens the importance of the methylated sites being important to controlling REF6 binding activity. Why is the number of sites only reported for one arm of one chromosome? They should be reported for all possible sites.
Furthermore, the rare ectopically bound sites in the ddcc mutant could be due to indirect effects. For example, new regions of open chromatin could make it easier for REF6 to bind DNA or differential chromatin states could also facilitate REF6 binding. Therefore, the ddcc data in this case doesn't provide enough ectopic events to convince the reader that DNA methylation is the causal region for differential binding.
Overall, if there are only few ectopically bound REF6 sites in the ddcc mutant, which loses all nonCG methylation, then what is the importance of this mechanism given that it is already proven that some DNA binding proteins are sensitive to DNA methylation levels. In conclusion, it is of my opinion that the results are an incremental advance that will be of interest to a highly specialized audience.
As an aside, there are also thousands of K27me3 regions in the genome that are differential K27 methylated in different cell types and due to different environments. If REF6 binds the same motif why are all regions not demethylated at the same time since they all possess the same motif? It likely doesn't happen because the REF6 motifs are not the only factor that determines demethylation. There must be some other sequence specific factors that mediate REF6's recruitment to target genes.
Reviewer #2 (Remarks to the Author): The manuscript by Qiu et al. investigated the potential role of DNA methylation in preventing heterochromatin binding of REF6, a JmjC-domain-containing H3K27me3 histone demethylase. This is a further extension of previous work from the authors' lab demonstrating that REF6 acts as an H3K27me3 histone demethylase (Lu et al., Nature genetics, 2011) and recognizes the CTCTGYTY DNA sequences specifically in the euchromatic regions, but not in the heterochromatin (Cui et al., Nature genetics, 2016). In this study, the authors showed that REF preferred to bind unmethylated DNAs both in vitro and in vivo and concluded that DNA methylation (particularly CHG methylation) repels REF6 targeting. The data presented are interesting and will provide a potentially mechanistic insight into the interplay between DNA and histone methylation in plants. The topic is important and should have a broad appeal to the readers of Nature Communications. The following are specific comments to further improve this manuscript.
My major concern is the lack of sufficient in vivo data supporting the key conclusion that DNA methylation repels REF6 chromatin targeting. While the in vitro EMSA binding and crystal structural analyses are solid, the supporting in vivo data is week. The authors performed a correlative data analysis and found that REF6 binding sites are negatively correlated with DNA methylation particularly in the heterochromatin (Figure 1). This is not surprising given the euchromatic enrichment nature of REF6. It is important to compare the CTCTGYTY motifs only in the euchromatin with and without REF6 binding. The authors should also provide more details regarding the data analysis. For example, within the 3464 REF6 binding regions, how many of them contain CTCTGYTY motifs? What about the number of CTCTGYTY motifs without REF6 binding? Also, it is beneficial to the general readers to provide more explanations on the purposes and results of the genomic analysis.
Along the same line, my another concern is the biological significance of the anti-correlation between heterochromatic DNA methylation and REF6 binding. Since REF6 is a H3K27me3 demethylase and H3K27me3 is mainly in the body of euchromatic genes. It doesn't make too much sense for me that the authors draw strong conclusion based on the correlation from the DNA methylation in heterochromain, unless the authors can demonstrate a distinct role of REF6 in heterochromatin independent of its H3K27me3 demethylase activity. Thus, it is critical to only focus the analysis on the biological relevant regions. The authors had previously generated H3K27me3 ChIP-seq in ref6 mutants and generated a metaplot in Fig S1d. It is unclear what the purpose and how this metaplot is used in supporting their conclusion. It is important to correlate the hyper H3K27me3 regions in ref6 with the CTCTGYTY motifs for further analysis.
Investigating the REF6 genome-wide occupancy in ddcc mutants is very nice (figure 4). The authors noted ectopic binding of REF6 to short TEs. But, why the number (22 out of 1220) is so low? Are 1220 the total REF6 ectopic binding peaks? Are they mostly in the heterochromatin? Of the 98% (~1200) REF6 ectopic binding peaks that don't overlap with TEs, what are they and what about their DNA methylation levels?
I noticed that the REF6 ChIP-seq signals appear to be general higher in ddcc compared to Col (e.g. CUC1 control loci in Fig. 4b). The details regarding the REF6 ChIP-seq reads and replicates are missing. It is unclear whether the REF6 ChIP-seq has been biologically repeated, which is important as a certain variation in peak identification is expected between replicates. It is more critical to perform extensive normalization analysis for ChIP-seq data to normalize the overall signal to noise ratio between ddcc and Col samples. Although not required for this manuscript, this reviewer highly recommended normalizing across samples using spiked-in methods.
As stated above that REF6 is a H3K27me3 demethylase and H3K27me3 is mainly in the gene body regions, mostly with CG methylation. In Fig.1b, the CG DNA methylation level had a most obvious change among the three DNA methylation contexts. Thus, it makes more sense to test the role of CG methylation in REF6 binding (e.g REF6 ChIP-seq in met1 or ddm1 mutants). The authors may also consider (not required) perform REF6 ChIP-seq in imb1 or ros1dme1dme2 triple mutants (hypermethylation mutants) to provide more evidence.
The manuscript is well written in general, but may benefit to have more details in the results as well as the method sections and more discussions. It is also important to include a table describing the genomic sequencing data. L156, change "significant" to "great" as no significant test has been done in the ITC experiments.
Reviewer #3 (Remarks to the Author): In the current submission "DNA methylation repels REF6 targeting in Arabidopsis" Qiu et al. demonstrate that the Zinc Finger domain of this H3K27 demethylase is sensitive to DNA methylation both in vitro and in vivo and reveal the structural basis for this sensitivity. Furthermore, they demonstrate that by altering the DNA methylation landscape in the ddcc mutant background, REF6 localizes to new target sites. While there is a small, but growing number of TFs shown to be sensitive to DNA methylation based on in vitro analyses, there remains little evidence regarding the effects of such sensitivities in vivo. The data by Qiu et al represent an excellent in vivo example in which DNA methylation affects chromatin association and provide what may be the first example wherein a methylation sensitive DNA binding module connects two epigenetics modifications (DNA methylation and H3K27 methylation). The findings of this manuscript are well supported by the data and I only have minor comments (see below). minor comments: line 63, the sentence about the role of DNA methylation in gene silencing is redundant with the second sentence of the same paragraph and could be removed. As part of the last paragraph of the introduction the authors should also mention the recent work from the Ecker lab in looking at the methylation sensitivity of many TFs via DAPseq assays. The phrase "intrinsic DNA methylation unfavorable DNA binding activity" is quite awkward. Perhaps "methylation sensitive DNA binding activity" would be better? The authors use two different nomenclatures to refer to methylated cytosines (mC and 5mC). As these are the same thing, the authors should stick to a single naming system to avoid unnecessary confusion. In Figure 3e, I believe that the C5 and G5 labels are switched. line 164-65 the names of DRM2 and CMT3 were already defined in the introduction and thus do not need to be redefined here. Information regarding the locations of the REF6 peaks from the ChIP experiments (wt, ddcc, etc) should be provided as well as general information about the ChIP libraries (coverage, mapping etc). If available additional information regarding the effects of the new REF6 targets sites in the ddcc background, like H3K27me levels or gene expression changes, would make an excellent addition to this manuscript. In figure 4a, the legend refers to "the thick red dot" but there are many red dots. In figure 4d, the legend refers to chromosome 1 as "chromosome I".
Reviewer #4 (Remarks to the Author): In this paper, Qiu et al report an "antagonistic mechanism" between REF6 and DNA methylation. This interesting finding is supported by structural and seq-based genomic profiling studies. Overall, this story is conceptually important and the regulatory mechanism is interesting. I would like to recommend publication of this manuscript if the following concerns are properly addressed.
Major point: 1) According to ITC assays (Figure 3f), the DNA methylation only results in about twofold reduction of the binding affinity. REF6 binds to methylated DNA strands at KD=77-173 nM, which is still a very strong binding event. In this case, I do not hold the view that "5mCs lead to significant reduction in the binding affinity to the CTCTGYTY motif" in vitro. According to the EMSA data (Figure 2), it is hypermethylation of cytosine but not single site 5mC that functions to repel REF6. Therefore, the authors should carefully revisit their conclusion and squarely conclude their observations. An ITC titration using hypermethylated DNA substrate should be performed. If proven true, it's better to change the current title to "DNA hypermethylation repels REF6 targeting in Arabidopsis". Similarly, the hypermethylation state of the "CTCTGYTY" motif should be examined and confirmed in vivo. This is can be explored by single-base-resolution sequencing analysis of the genome DNA.
2) Additional perturbation studies should be performed. For example, based on the structural analysis, the authors should be able to design REF6 mutants that can tolerate 5mC. It would be nice to investigate the functional impact if the mutant REF6 is introduced in plant. Such an effort will significantly improve the quality of the story.
Minor points: 1) A complete ITC fitting parameters should be provided.
2) Figure 3e, C5 and G5' are mislabeled. In addition, a role of W1311 should be confirmed by mutagenesis studies.
3) The proposed mechanistic module in Figure 5 is not supported by the present data. In fact, the binding affinity of REF6 to methylated CTCTGYTY motif is not weak, and even corresponding complex crystal structures have even been determined. The red blocking arrow does not properly reflect the experimental observations. 4) In supplementary Table 1, the unit cell angles should be fixed to integral numbers, e.g. (90, 90, 90). These values are not measured ones for the current space groups. Also please double check the space group of ZnF2-4-NAC004-5mC1. The unit cell angles of (90, 90, 120) is not consistent with the P4(3) space group.
This paper describes how methylcytosines can repel binding of the histone demethylase REF6. The evidence to support these conclusions is from 1) the depletion of REF6 binding to methylated motifs from genome-wide ChIP data, 2) EMSA on two target loci, 3) binding constants from ITC and 4) ectopic binding in a mutant that has no nonCG methylation. Response: We thank the reviewer for pointing out this issue. The DAP-seq is a great method to identify potential TF-binding sites. Nevertheless, DAP-seq is an in vitro method using recombinant TFs to pull-down purified DNA, which doesn't reflect the binding affinity in vivo. Comparing the DAP-seq and ChIP-seq binding sites of the same TFs, we found the DAP-seq tend to get much more binding peaks than ChIP-seq (data not shown). Therefore, we believe that the DAP-seq data cannot fully reflect TFs-binding features in vivo. We use both in vivo and in vitro methods to demonstrate how DNA methylation affects REF6 recruitment, which is of good interest in understanding the interplay between DNA methylation and H3K27me3 beyond TF binding specificity along.
The data in this study suggests that although DNA methylation is associated with differential binding it is clearly not deterministic. In the REF6 binding data there are ~15% of binding events that are methylated (line 108). This shows that REF6 can bind methylated DNA. One key experiment missing would be to perform ChIP-BS-seq using REF6 in Col vs ddcc. How many methylated reads are identified in the Col-0 data. This would be a direct in vivo measurement of differential sensitivity to DNA methylation.

Response
Response: This is a great suggestion to improve our manuscript. To further confirm that REF6 preferentially bind to unmethylated DNA motifs in vivo, Table 1 and Supplementary   Fig. 2a). The results showed an anti-correlated profile between REF6 binding signal and DNA methylation level at REF6 binding peaks ( Fig. 2a and Supplementary Fig. 2). DNA methylation level of REF6 binding peaks identified by ChIP-BS-seq in Col are as low as that in ddcc, indicating that there is not significant difference between wild-type Col and ddcc for differential sensitivity to non-CG DNA methylation (Fig. 2b). Moreover, REF6 bound DNA showed lower methylation level compared to that in WGBS data, indicating that REF6 bound DNA was depleted for DNA methylation while the methylation at REF6 binding sites seen in WGBS data may come from DNA without REF6 binding in some cell types (Fig. 2b).

we performed REF6 ChIP-bisulfite-sequencing (ChIP-BS-seq) (Statham, et al. Genome Research. 2012) in Col, compared with ddcc, in which non-CG methylation is completely lost (Supplementary
These results give direct evidence supporting that REF6 prefers to bind hypomethylated DNA in the Arabidopsis genome. I credit the authors for performing the key experiment, which is to perform ChIP-seq of REF6 in the ddcc mutant. Unfortunately, the results do not support that DNA methylation is the major deterministic reason for REF6 binding as 1198/1220 or 98.2% of peaks did not exhibit ectopic binding whereas only 1.8% did. As shown in Figure 4a  for REF6 to bind DNA or differential chromatin states could also facilitate REF6 binding. Therefore, the ddcc data in this case doesn't provide enough ectopic events to convince the reader that DNA methylation is the causal region for differential binding.
Response: We thank the reviewer to point this out. We previously showed that REF6 binding affinity was affected by chromatin states, and open chromatin with exposed DNA sequence tended to promote REF6 binding.
We believe that other epigenetic modifications may also participate in REF6 targeting to specific regions which causes the relatively low number of ectopic binding in ddcc mutant. However, we do believe that the ddcc data demonstrated that DNA methylation is the cause of differential REF6 binding at some of the regions in the genome. As this can be recaptured in the in vitro EMSA result showing that REF6-ZnF can directly bind to AT4G11710 in the absence of DNA methylation, but cannot bind to methylated AT4G11710 (Fig. 3). Taking together these in vitro and in vivo data, we believe that DNA methylation is the cause for differential binding of REF6 at some of the loci, although the number is small.

Figure 3. Cytosine methylation in CTCTGYTY motifs decreases DNA-binding affinity of REF6-ZnF in vitro.
EMSA with AT1G02230 and AT4G11710 probes. REF6-ZnF specifically bound the unmethylated probes, but had significantly lower (or no) affinity for probes containing one or more methylated cytosines.
Overall, if there are only few ectopically bound REF6 sites in the ddcc mutant, which loses all nonCG methylation, then what is the importance of this mechanism given that it is already proven that some DNA binding proteins are sensitive to DNA methylation levels. In conclusion, it is of my opinion that the results are an incremental advance that will be of interest to a highly specialized audience.
Response: We thank the reviewer to point this out. REF6 is the major player in removing Polycomb repressive marks, H3K27me3. Here, we provide evidence and mechanism that REF6 can only function outside heterochromatin due to DNA methylation, revealing a clear interplay between two major epigenetic marks which we believe will be of general interest in the field of epigenetics. Arabidopsis has relatively simple heterochromatin and small amount of DNA methylation which might be the reason why the number is relatively small. However, we believe the mechanism is conserved in other species with more complex heterochromatin.
As an aside, there are also thousands of K27me3 regions in the genome that are differential K27 methylated in different cell types and due to different environments. If REF6 binds the same motif why are all regions not demethylated at the same time since they all possess the same motif? It likely doesn't happen because the REF6 motifs are not the only factor that determines demethylation.
There must be some other sequence specific factors that mediate REF6's recruitment to target genes. My major concern is the lack of sufficient in vivo data supporting the key conclusion that DNA methylation repels REF6 chromatin targeting. While the in vitro EMSA binding and crystal structural analyses are solid, the supporting in vivo data is week.
The authors performed a correlative data analysis and found that REF6 binding sites are negatively correlated with DNA methylation particularly in the heterochromatin  Table 1 and Supplementary Fig. 2a). The results showed an anti-correlated profile between REF6 binding signal and DNA methylation level at REF6 binding peaks (Fig. 2a and Supplementary Fig. 2). DNA methylation level of REF6 binding peaks identified by ChIP-BS-seq in Col are as low as that in ddcc, indicating that there is not significant difference between wild-type Col and ddcc for differential sensitivity to non-CG DNA methylation (Fig. 2b).

Moreover, REF6 bound DNA showed lower methylation level compared to that
in WGBS data, indicating that REF6 bound DNA was depleted for DNA methylation while the methylation at REF6 binding sites seen in WGBS data may come from DNA without REF6 binding in some cell types (Fig. 2b). These results give direct evidence supporting that REF6 prefers to bind hypomethylated DNA in the Arabidopsis genome. to contain motifs as many as possible.

Figure 2. ChIP-BS-seq shows REF6 prefers to bind hypomethylated regions in
Along the same line, my another concern is the biological significance of the anti-correlation between heterochromatic DNA methylation and REF6 binding. Here we proved that CHG DNA methylation is at least one of the factors that prevents REF6 from binding to heterochromatin and helps REF6 to find its targets globally. It will be very interesting to dissect the reason why REF6 shall be repelled from heterochromatin as suggested by the reviewer in the future which we believe is out of the scope of the current manuscript.
The authors had previously generated H3K27me3 ChIP-seq in ref6 mutants and generated a metaplot in Fig S1d. It is unclear what the purpose and how this metaplot is used in supporting their conclusion. It is important to correlate the hyper H3K27me3 regions in ref6 with the CTCTGYTY motifs for further analysis. I noticed that the REF6 ChIP-seq signals appear to be general higher in ddcc compared to Col (e.g. CUC1 control loci in Fig. 4b). The details regarding the REF6 ChIP-seq reads and replicates are missing. It is unclear whether the REF6

Response
ChIP-seq has been biologically repeated, which is important as a certain variation in peak identification is expected between replicates. It is more critical to perform extensive normalization analysis for ChIP-seq data to normalize the overall signal to noise ratio between ddcc and Col samples. Although not required for this manuscript, this reviewer highly recommended normalizing across samples using spiked-in methods.  As stated above that REF6 is a H3K27me3 demethylase and H3K27me3 is mainly in the gene body regions, mostly with CG methylation. In Fig.1b, the CG DNA methylation level had a most obvious change among the three DNA methylation contexts. Thus, it makes more sense to test the role of CG methylation in REF6 binding (e.g REF6 ChIP-seq in met1 or ddm1 mutants). The authors may also consider (not required) perform REF6 ChIP-seq in imb1 or ros1dme1dme2 triple mutants (hypermethylation mutants) to provide more evidence.

Response
Response: This is a great idea! To explore the role of CG methylation in REF6 targeting, we performed REF6 ChIP-seq in met1-1 mutant, which loss CpG methylation in gene body (Catoni et al, EMOB J., 2017). However, we found no obvious changes in REF6 binding between met1 and Col (Supplementary Figure 1.1 below). Therefore, we concluded that gene body The manuscript is well written in general, but may benefit to have more details in the results as well as the method sections and more discussions. It is also important to include a table describing the genomic sequencing data.
Response: Thank you for the suggestions to help improving our manuscript. We have provided the information in our modified manuscript and the supplementary table describing the genomic sequencing data. L156, change "significant" to "great" as no significant test has been done in the ITC experiments.

Supplementary
Response: Thank you for pointing out this mistake. We have changed "significant" to "great".
Reviewer #3 (Remarks to the Author) In the current submission "DNA methylation repels REF6 targeting in The phrase "intrinsic DNA methylation unfavorable DNA binding activity" is quite awkward. Perhaps "methylation sensitive DNA binding activity" would be better?
Response: Thank you for the wording. We have changed it into "methylation sensitive DNA binding activity".
The authors use two different nomenclatures to refer to methylated cytosines (mC and 5mC). As these are the same thing, the authors should stick to a single naming system to avoid unnecessary confusion.
Response: Thank you for pointing out this mistake. We have changed all mC into 5mC.
In Figure 3e, I believe that the C5 and G5 labels are switched.
Response: Thank you for pointing out this mistake. We have corrected it.
line 164-65 the names of DRM2 and CMT3 were already defined in the introduction and thus do not need to be redefined here. In figure 4a, the legend refers to "the thick red dot" but there are many red dots.

Response
Response: Thank you for pointing out this mistake. We have changed it into "the thick red dots".
In figure 4d, the legend refers to chromosome 1 as "chromosome I".
Response: Thank you for pointing out this mistake. We have changed it as Reviewer #4 (Remarks to the Author): In this paper, Qiu et al report an "antagonistic mechanism" between REF6 and DNA methylation. This interesting finding is supported by structural and seq-based genomic profiling studies. Overall, this story is conceptually important and the regulatory mechanism is interesting. I would like to recommend publication of this manuscript if the following concerns are properly addressed.
Major point: 1) According to ITC assays (Figure 3f), the DNA methylation only results in about twofold reduction of the binding affinity. REF6 binds to methylated DNA strands at KD=77-173 nM, which is still a very strong binding event. In this case, I do not hold the view that "5mCs lead to significant reduction in the binding affinity to the CTCTGYTY motif" in vitro. According to the EMSA data ( Figure   2), it is hypermethylation of cytosine but not single site 5mC that functions to repel REF6. Therefore, the authors should carefully revisit their conclusion and squarely conclude their observations. An ITC titration using hypermethylated DNA substrate should be performed. If proven true, it's better to change the current title to "DNA hypermethylation repels REF6 targeting in Arabidopsis". completely abolished the protein-DNA interaction (Fig. 4), which is consistent with our EMSA result. We repeated this experiment several times and got consistent result among replicates. In view of this result, we further performed ITC assay with 5mC 1 +5mC 3 probe and found that 5mC 1 +5mC 3 completely abolished the protein-DNA interaction, which is also consistent with our EMSA result. Therefore, we replaced the old data with these new data in the revised manuscript.  Table 1). The results showed an anti-correlated profile between REF6 binding signal and DNA methylation level at REF6 binding peaks (Fig. 2a).

DNA methylation level of REF6 binding peaks identified by ChIP-BS-seq in Col
are as low as that in ddcc, indicating that there is not significant difference between wild-type Col and ddcc for differential sensitivity to non-CG DNA methylation (Fig. 2b). Moreover, REF6 bound DNA showed lower methylation level compared to that in WGBS data, indicating that REF6 bound DNA was depleted for DNA methylation while the methylation at REF6 binding sites seen in WGBS data may come from DNA without REF6 binding in some cell types (Fig. 2b). These results give direct evidence supporting that REF6 prefers to bind hypomethylated DNA in the Arabidopsis genome. As to the CHG hypermethylation in CTCTGYTY motif, no REF6 binding is found. 2) Additional perturbation studies should be performed. For example, based on the structural analysis, the authors should be able to design REF6 mutants that can tolerate 5mC. It would be nice to investigate the functional impact if the mutant REF6 is introduced in plant. Such an effort will significantly improve the quality of the story.
Response: This is a great suggestion. We expressed REF6-ZnF with a Trp1311 to Ala mutation (W1311A) and performed EMSA and ITC with NAC004 probe. We found that W1311A mutation abolished the interaction between REF6-ZnF and DNA probe, indicating Trp1311 is important for REF6-ZnF binding to DNA. Therefore, design of such a mutation version of ZnF is still challenging even we know W1311 is the methylation-sensitive site.
But this is a promising direction that will improve our understanding for the function of REF6 in the future.    3) The proposed mechanistic module in Figure 5 is not supported by the present data. In fact, the binding affinity of REF6 to methylated CTCTGYTY motif is not weak, and even corresponding complex crystal structures have even been determined. The red blocking arrow does not properly reflect the experimental observations.
Response: Thank you for pointing out this. According to our EMSA and new ITC results, 5mC 1 +5mC 3 and 5mC 5 completely abolished the protein-DNA interaction. Therefore, we thought that the proposed mechanistic model can reflect the experimental observations. Table 1, the unit cell angles should be fixed to integral numbers, e.g. (90, 90, 90). These values are not measured ones for the current space groups. Also please double check the space group of ZnF2-4-NAC004-5mC1. The unit cell angles of (90, 90, 120) is not consistent with the P4(3) space group.

4) In supplementary
Response: Thank you for pointing out this. In Supplementary Table 1, we have changed the unit cell angles to integral numbers, and changed the space group of ZnF2-4-NAC004-5mC 1 to P3 1 .