BRI1-mediated removal of seed coat H3K27me3 marks is a brassinosteroid-independent process

  1. Rishabh Pankaj
  2. Rita B. Lima
  3. Guan-Yu Luo
  4. Sinah Ehlert
  5. Gerardo del Toro-de León
  6. Heinrich Bente
  7. Pascal Finger
  8. Hikaru Sato
  9. Duarte D. Figueiredo

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Seed development in angiosperms starts with double fertilization, where two paternal sperm cells fertilize the maternal gametes. This leads to the formation of the embryo and of the endosperm. These fertilization products are enveloped by the maternally-derived seed coat, the development of which is inhibited prior to fertilization by the Polycomb Repressive Complex 2 (PRC2). This complex deposits the repressive histone mark H3K27me3, whose removal is necessary for seed coat formation. Here, we show that JUMONJI-type (JMJ) histone demethylases are expressed in the seed coats of Arabidopsis thaliana (Arabidopsis) and are necessary for its formation. We propose that JMJ activity is coupled to Brassinosteroid (BR) function, as BR effectors physically recruit JMJ proteins to target loci. Consistent with this, we show that loss of BR biosynthesis and signaling leads to seed coat defects, and that loss of the main BR receptor, BRI1, results in H3K27me3 hypermethylation. Moreover, our data points to BRI1 mediating H3K27me3 removal independently of BRs, while a different receptor, BRL3, likely regulates seed coat formation in a BR-dependent manner. We thus propose a model where seed coat development relies on canonical and non-canonical functions of BR receptors.

Article activity feed

  1. Review Commons

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

    Learn more at Review Commons

    Reply to the reviewers

    We thank both reviewers for their detailed and critical assessment of our work. Below we provide a step-by-step response to your concerns.

    Reviewer #1

    Evidence, reproducibility and clarity

    The manuscript presents data demonstrating the function of BRI1 in removing the H3K27me3 epigenetic marks in genes involved in seed coat development in Arabidopsis. The results support that BRI1 may function here independently of brassinosteroid. The work combines genetics with a large panel of mutant lines, phenotyping by quantitative microscopy and chemical treatment, H3K27me3 profiling by CUT&TAG, and data mining for published gene profiling. The introduction is adequately informative, complete and explaining the state-of-the-art to the readers. The result part may be a bit lengthy (especially the first part) and some parts may be a bit repetitive.

    Thank you for your positive assessment of our work and for the constructive criticism. Below we respond to each of your points.

    Major

    1. Seed size is dependent of multiple factors. And few are explained here, notably the number of seeds per silique, the number of ovule per silique, the position of the silique of the branch (related to the age of the meristem), the total number of produced siliques (fertilised flowers) by the inflorescence meristem and the plant. And maybe if produced by the main and lateral branches. Were the authors consistent in the evaluation of analyzed siliques coming from the same type of branches, same age of the meristem, etc? Especially as some of the analysed mutants are dwarf, which is a sign of different plant fitness compared to WT.

    This is a valid point. We did aim to analyze seeds coming from the main inflorescences of the plants and at similar stages of shoot development. This was harder to achieve in some genotypes, as indeed some BR and JMJ mutants have different plant architectures. However, we did repeat those experiments multiple times and always found consistent differences between the WTs and the mutants. See also our response to your next point and to the first point raised by Reviewer 2, as well as our new Fig. S6.

    1. The seed perimeter measurements in BR mutant seeds (Figure S6) are variable. Are you sue the ovule size does not have any influence? What about presenting the relative size as earlier in the text?

    Yes, this is particularly true for the Col-0 vs dwf5 comparison. The reason for this is that a different growth chamber was used for this experiment (greenhouse vs a climate chamber). We have observed that absolute seed growth phenotypes can change depending on the environmental conditions, which is something we are currently studying. However, importantly, we do not see changes in relative growth of the mutants when compared to the WT, independently of the growth conditions. That is, BR mutants produce consistently smaller seeds than the WT, independently of the conditions in which the plants are grown. To illustrate this point, we now add a new Figure, Fig. S6, where we show four independent biological replicates of assays comparing seed size between WT and det2 or bri1. These replicates were done in different growth chambers.

    Indeed, presenting the data as relative size would solve this issue, but we worried that we would be hiding the "real" values by doing so. However, if the Reviewer and Editor deem it necessary, we could replot the data as relative to WT.

    1. The number of evaluated samples is often {plus minus} n = 30, sometimes less, meaning less than what a silique contains of seeds. Did the authors evaluate the variability and reproductibility of their measurements, e.g, how many siliques per plant, how many plants, how many biological repeats? For example, in Figure S6, the number of measured ovules were as low as 16, which could be the reason why no significant difference in size were observed (low statitical strength). The variation in the Col WT is already visible. Is this variation significant?

    On average we pooled seeds from 6-10 siliques coming from 2-3 different plants of the same genotype. We then took microscopic photos of 60 to 100 random seeds in those pools. Out of those, 30 random photos were used for the measurements. You are right this is an important point. We now added this information to the Methods section.

    Moreover, we did calculate whether the sample size we were using provided enough statistical power. For the differences that we see, of around 50 um in perimeter, 26 samples would have been enough to achieve 80% statistical power, which most studies use as standard. In most of our experiments we used closer to 30 samples, which gives us 95% power.

    Indeed, the left-most panel on Fig S6B is the exception. With that plot we mostly wanted to test if ovules produced by BR mutants were smaller than those of WT plants. That does not seem to be the case, even if the sample size is small. However, if deemed necessary, we can repeat those measurements with a higher sample number.

    1. You indicate (line 149) that REF6 is not expressed in the gametophyte but GFP signal is observed in the cytoplasm for the central cell in Fig 1. The same goes for the expression pattern with the GUS line in Figure S2. (Line 290) One can not exclude expression in the endosperm or embryo with the presented pictures, or in the seed coat in older seeds.

    We interpreted those diffuse signals in the cytoplasm of the gametophyte as background noise, as REF6 should be nuclearly localized. But we could be wrong. We therefore made changes to the text in lines 150-152 to reflect this.

    And you are right that REF6 is expressed in the endosperm and embryo in later stages of development. We mention this in lines 157-159.

    1. Make sure that you do not overstate your result conclusions, or add a reference to some of the statements. For example, line 185, for the choice of 3 DAP time point and the fact that seed coat development is based on cell expansion and interaction with the endosperm. Another example, in line 262, is where it is stated that the jmj mutants are compromised in ovule and pollen development. This was not assessed. You only checked the reduced seed set, not the fitness of the gametophytes. Or in line 337, where you indicate that KLUH is not expressed in all integument layers.

    Thank you for pointing this out. For the claim that seed size at early time points is dictated only by the seed coat and endosperm, and not by the embryo, we added the appropriate reference. For the claim that jmj mutants are compromised in ovule development, this was based on our observations of Fig. S3C. We do see malformed or absent megagametophytes in jmj mutants. For pollen development, you are correct that we did not formally address this. We rephrased the sentence to reflect that. For the statement that KLU is not expressed in all integument cell layers, we added the reference.

    1. Another example of this is in line 289 where you stated "a sporophytic function of JMJs at early stages of seed development, [..] and to a zygotic function at the later stages of seed development". I am not sure on what data do you base this conclusion as in all three categories (endosperm, embryo, seed coat) in Fig 2 and S5, genes are expressed in pre-globular stages. And again in line 475: "seed coat growth genes are expressed independentlyof fertilization". Do you have any evidence, a reference?

    The evidence for a sporophytic JMJ function at early stages of seed development, and zygotic function at later stages, comes from our observations that jmj seed phenotypes are maternal in origin at early stages, but become zygotic later in development. But you are correct that we have to be careful with this interpretation. We now modified that sentence accordingly.

    For the data of Fig 2E and Fig S5, we cannot rule out that some putative REF6 target genes are also expressed even when in the absence of REF6. The expression of those genes is also likely controlled by other factors. The point we wanted to make with those plots is that REF6 may have different target genes in different seed tissues, thus potentially regulating different developmental processes in a tissue-specific manner. We mention this in lines 288-290.

    For your second point, we added the adequate reference.

    1. (around lines 461) I understand that using a 35S promoter is not a good strategy as it would affect many other tissues. Did you consider using a tissue-specific approach as presented in Figure 4?

    We suppose you mean the 35S::ELF6 construct. Yes, this makes sense and we did spend quite some time trying to come up with a good strategy. However, we failed to find a suitable promoter. The issue is that we would need a promoter that is active in all (or most) seed coat layers, but only after fertilization. There are promoters like those of TT genes which are active post-fertilization, but only in one cell layer, and thus likely not useful for our purpose. And there are other promoters, like those of STK or ANT, which are expressed in most integument cell layers, but are also expressed during integument development, and not just after fertilization. So they would have the same issue as the 35S promoter. Unfortunately, so far we have not identified a promoter that would be useful for this kind of experiment, which is why we went with a constitutive promoter, but which is specific to the sporophytic tissues.

    1. You observed that the triple swn clf bri mutant is less dwarf than bri1 mutant and stated in line 483 that it is larger, has more leaves, grow tallerm and flower later and longer. Do you have any qunatitative data? If not, I would state that these observations are qualitative from growing plant aside.

    You are correct that this was based on qualitative assessments, rather than on quantitative data (as it was not the point of the manuscript). We now indicate this in lines 489-490.

    Minor:

    1. The title should precise the studies species, here Arabidopsis thaliana. Also the title of one of the part could be rephrased. "in a zygotic manner" sounds strange.

    We modified both title and subtitle, as suggested.

    1. Scale bars are missing in many figures.

    Fixed.

    1. The font size in the graphs is small. The authors may use the empty space of the figures to increase the size of the graphs for clarity. Guidelines could be found here https://tpc.msubmit.net/html/TPC_Detailed_Figure_Guidelines.pdf, as example of good practices.

    You are right. We revised all the figures and increased the font size, especially in the plot labels.

    1. Be consitent in the mutant name, e.g., brz1-D is also presented as brz1-d.

    Fixed.

    1. Figure legend S1: I would not use the word "extremely" while you still have 30% seed set. Extremely would qualifiy for

    We suppose you mean Fig. S3. We corrected the legend.

    1. Figure S8 is missing the WT control for comparison.

    Fixed.

    1. Figure S12, stats are missing

    Fixed.

    1. I would recommend to add a line in the Supplemental tables with the name as this name disappears from the file name during upload. It would help the readers to navigate the data.

    We now made it so the top line is static and is always visible.

    1. Methods: Are all the lines listed used in the study? SR2200 is missing for the method, and please indicate the selection marker for each of the generated lines for open-access of the data if other researchers later use your lines.

    You are right that some references had been left over from a previous document. We now updated the list of lines.

    And indeed, we forgot to mention the use of SR2200. It is now added to the Methods section. We also added the information on the selection markers for the lines we generated.

    1. You have a duplicate for reference Vukašinovíc et al.

    Fixed.

    1. Line 393, remove "s" in embryo and endosperm, in coat (line 674), in size (lines 684, 686

    Fixed.

    1. Line 410, write RPS5A in upper case.

    Fixed throughout the manuscript.

    1. LIne 676, the sentence "...H3K27me3 to be removed from the integuments." I would recomend to be more precise. For example "H3K27mme3 marks to be removed from genes to be expressed in the integuments" or something like that.

    We rephrased this sentence to "We thus hypothesized that BR signaling would be required for JMJ function, allowing for H3K27me3 to be removed from genes necessary for seed coat formation."

    Significance

    The authors provide novel information on the step-wise regulation of seed coat development and its influence on seed size. This is a topic of general interest, beyond the plant model Arabidopsis, especially in the context of reduced seed set caused by (a)biotic stress. The results of this study are valuable to understand seed size regulation in differnet growth context or species. The group previously showed that the auxin phytohormone is necessary after fertilization to initiate seed coat differentiation by inhibiting PRC2. However, as seed coat develops mainly as cell elongation, the epigenetic marks are not diluted by cell division and needs to be actively removed. This study provides insight into this process by identifcation 2 JMJ proteins responsible for removing H3K27me3 marks in the seed coat after fertilization to initiation seed coat development and regulating seed size. BRI1, BES1 and BZR1 are involved in this process, indepently of brassinosteroid, to guide JMJ to their target loci. While the study bring some genetic evidence of this process, molecular insight is still missing. Notably the identification of the target genes and how BRI1 is regulated/activated upon fertilization. Or how auxin and BRI1 co-regulate the process. These questions appear how of scope of this current study.

    Thank you for the assessment. Indeed, the identification of BRI1 downstream genes is out of scope of this work. As you point out earlier in the review, the manuscript is already quite long, and adding such data would make it even more so.

    Reviewer #2

    In this study, Pankaj et al. investigate the role of brassinosteroids and H3K27me3 in seed development, particularly in controlling seed size. They demonstrate that defects in these pathways affect seed size control and suggest that this control occurs in the maternal seed coat. This paper presents novel findings that merit publication and would be of interest to the plant community. However, the data interpretation and presentation could be improved. Additionally, I have a few comments that necessitate further analysis and revision.

    Thank you for the careful and critical assessment of our work. Below we respond to each of the points you raised.

    Major Comments

    1. My main concern is the use of seed size measurement as a proxy for seed coat development. Mature seed size measurements can vary significantly with growth conditions, so it is crucial that the authors present at least three independent experiments (wild type and mutant grown in parallel) in a single box plot to ensure data reliability. Additionally, due to the high number of seeds analyzed, significant changes are often observed, though they are not always reproducible. The authors should standardize their seed measurements, using either seed area or seed perimeter.

    You are right that we do see some variation in seed size between experiments. And, indeed, we suspect this is due to slightly different plant growth conditions, for example when different growth chambers are used. As you suggest, we now show data from four independent biological replicates of seed size comparisons of WT and BR mutants. This is in the new Fig. S6. As you can see, although we do see variations in absolute seed sizes, depending on the growth conditions, there is a consistent difference between WT and mutant seeds across experiments.

    1. It would be beneficial to include data on cell division and cell elongation in the seed coat if the authors aim to extend the seed size phenotype to a seed coat phenotype.

    This is indeed a good point. However, we already showed in a previous publication that seed coat growth is driven by cell elongation and not cell division (https://elifesciences.org/articles/20542). But you are right that this is important to point out. We mention it in lines 66-67.

    1. It is challenging to be fully convinced by the seed coat specificity of the phenotype, as the authors observe variations in total seed set and phenotypic differences in self-crosses and when the mutants are used paternally. Some of the observed phenotypes do not support their hypothesis. In all mutant analyses, the authors should complement their phenotype analysis using seed coat-specific promoters and include heterozygote measurements, as done in some figures.

    We assume you mean the effect of jmj mutations. For BR mutants, we do show data supporting a seed coat effect (Fig. 4). For PRC2 mutants, that has also been previously described (doi.org/10.7554/eLife.20542 and doi.org/10.1073/pnas.1117111108).

    For the JMJ mutants, you are right that we cannot be 100% sure that their effect is purely sporophytic. We now modified the text accordingly to reflect this (see also the response to point 6 of Reviewer 1). We indeed show that REF6 and ELF6 are expressed in the sporophytic tissues of the ovule and that the double mutant has seed coat defects (smaller seed coats and defects in accumulation of proanthocyanidins). And although we can say that those defects are maternal in nature, we can not 100% conclude that they are simply due to the effect of those JMJs in the sporophyte. There may be gametophytic effects that we cannot rule out, even though we do not see either protein expressed in embryo sacs. Thank you for pointing this out.

    Doing a tissue-specific rescue of these phenotypes would be very informative indeed, but also very hard. As we mention in the response to point 7 of Reviewer 1, we do not currently have suitable promoters for this. So we simply cannot run such experiments in a reasonable time frame.

    Overall, we now tried to be more careful in our conclusions and avoid claiming that the effect of JMJs is purely sporophytic. We can make that argument for the BR machinery and for PRC2, but not necessarily for JMJs. You are correct in that assessment.

    1. The authors need to include a fluorescent reporter for ELF6; tissue-specific expression cannot be conclusively determined with the GUS reporter.

    We did obtain an ELF6::GFP line from Caroline Dean's lab (https://www.pnas.org/doi/full/10.1073/pnas.1605733113), but could not see much expression during endosperm or seed coat development. As you can see from that publication, even in embryos and in roots the expression of ELF6:GFP is very blurry. It seems ELF6 is simply expressed at very low levels. We therefore used the GUS reporter, as a more sensitive means to visualize where ELF6 is expressed. You are right that the results are not as precise as that obtained with a fluorescent reporter. However, note that we simply claim that ELF6 is expressed in the integuments and seed coat (line 155). This can be clearly seen in Fig. 1B. The blue product of the β-glucuronidase reaction should be immotile and not travel between tissues (also note that there are no plasmodesmata between endosperm and seed coat). Therefore, we believe that GUS is a suitable reporter to test the seed coat expression of ELF6.

    1. Text editing: In some places, the text is unclear and could benefit from simplification. The authors should replace the term "seed coat formation," as developmentally, integuments are already present before fertilization. The authors are not studying the formation of the seed coat but rather its growth. They should also clarify the term "PRC2 removal." It is unclear whether the authors mean PRC2 lack of expression in the integument, PRC2 eviction from chromatin, or removal of H3K27me3.

    Thank you for noting that. It is very important to us that the text is clear to the reader. If you could indicate where the text is unclear, we are happy to simplify it.

    Regarding the wording, we refer to "seed coat formation" because the seed coat only indeed forms after fertilization. Before fertilization, the sporophytic tissues that cover the megagametophyte are called integuments, and not seed coat. Therefore, we see the seed coat as "forming" from the integuments (i.e., the integuments become seed coat via growth and differentiation).

    With PRC2 removal we indeed mean reduction of expression of PRC2 components. We now make this clear in lines 54-55.

    Minor Comments

    1. L151: Is REF6 expressed in zygotic tissues?

    Reviewer 1 also raised this question. We now added this information to lines 148-150.

    1. Confirm mutant complementation with the different reporter lines.

    All mutant lines that we used have been previously described to be either loss-of-function or hypomorphic mutants. We did not use any mutant line that has not been previously described. We added all references to the corresponding publications in the Methods.

    1. Confirm by qPCR that JMJ13 is indeed not expressed in seeds.

    We tested JMJ13 as a possible factor involved in H3K27me3 removal in the seed coat due to it being described, together with ELF6 and REF6, as one of the three main H3K27 demethylases. But there are, in fact, transcriptomic datasets showing that the expression of JMJ13 is indeed very low or absent in seeds: see RNAseq data in Table S3 in doi.org/10.3389/fpls.2022.998664. Moreover we checked CPMs on published seed scRNAseq datasets (doi.org/10.1038/s41477-021-00922-0) and JMJ13 (AT5G46910) has zero transcript counts in these datasets.

    Because of these two independent instances showing that the expression of JMJ13 is extremely low in seeds (or even totally absent), together with the analysis that we did of the fluorescent reporter line, we believe this is sufficient evidence that this JMJ is specific to the pollen during reproductive development. Note that the reporter that we used is strongly expressed in pollen grains, as had been previously described (doi.org/10.1038/s41556-020-0515-y).

    Even so, if the Reviewer and the Editor deem it necessary that we check JMJ13 expression by qPCR, we can of course do so.

    1. Fig1a and Fig1b: Align the panels in the figure.

    Done.

    1. L183-189: This section is unclear.

    I am sorry that the section is not clear. If you direct us to the points that need to be cleared, we are happy to make changes.

    1. There may be a PDF artifact, but most figures have unattractive misaligned boxes.

    We went through every figure and made slight modifications to avoid such artifacts. We hope they now appear more clear in the new version.

    1. Change the color in Fig 2a.

    Fixed.

    1. The introduction is heavily self-cited. The authors should try to include a broader range of literature.

    It is not clear to us why the Reviewer sees it like that. We only refer to three of our publications in the Introduction. One review manuscript and two research manuscripts. We cite almost 40 manuscripts in the introduction. Therefore, citing three of our works does not seem out of line to us, especially since those manuscripts laid the foundation for this work.

    1. Fig3F: Typo in "microM."

    Fixed.

  2. 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

    Review of "BRI1-mediated removal of seed coat H3K27me3 marks is a brassinosteroid-independent process" In this study, Pankaj et al. investigate the role of brassinosteroids and H3K27me3 in seed development, particularly in controlling seed size. They demonstrate that defects in these pathways affect seed size control and suggest that this control occurs in the maternal seed coat. This paper presents novel findings that merit publication and would be of interest to the plant community. However, the data interpretation and presentation could be improved. Additionally, I have a few comments that necessitate further analysis and revision.

    Major Comments

    • My main concern is the use of seed size measurement as a proxy for seed coat development. Mature seed size measurements can vary significantly with growth conditions, so it is crucial that the authors present at least three independent experiments (wild type and mutant grown in parallel) in a single box plot to ensure data reliability. Additionally, due to the high number of seeds analyzed, significant changes are often observed, though they are not always reproducible. The authors should standardize their seed measurements, using either seed area or seed perimeter.

    • It would be beneficial to include data on cell division and cell elongation in the seed coat if the authors aim to extend the seed size phenotype to a seed coat phenotype.

    • It is challenging to be fully convinced by the seed coat specificity of the phenotype, as the authors observe variations in total seed set and phenotypic differences in self-crosses and when the mutants are used paternally. Some of the observed phenotypes do not support their hypothesis. In all mutant analyses, the authors should complement their phenotype analysis using seed coat-specific promoters and include heterozygote measurements, as done in some figures.

    • The authors need to include a fluorescent reporter for ELF6; tissue-specific expression cannot be conclusively determined with the GUS reporter.

    • Text editing: In some places, the text is unclear and could benefit from simplification. The authors should replace the term "seed coat formation," as developmentally, integuments are already present before fertilization. The authors are not studying the formation of the seed coat but rather its growth. They should also clarify the term "PRC2 removal." It is unclear whether the authors mean PRC2 lack of expression in the integument, PRC2 eviction from chromatin, or removal of H3K27me3.

    Minor Comments:

    • L151: Is REF6 expressed in zygotic tissues?

    • Confirm mutant complementation with the different reporter lines.

    • Confirm by qPCR that JMJ13 is indeed not expressed in seeds.

    • Fig1a and Fig1b: Align the panels in the figure.

    • L183-189: This section is unclear.

    • There may be a PDF artifact, but most figures have unattractive misaligned boxes.

    • Change the color in Fig 2a.

    • The introduction is heavily self-cited. The authors should try to include a broader range of literature.

    • Fig3F: Typo in "microM."

    Cross-commenting:

    I think our reviews highlight the same issues. For me, the first point is definitely the most critical.

    Significance

    This paper presents novel findings that merit publication and would be of interest to the plant community. However, the data interpretation and presentation could be improved. Additionally, I have a few comments that necessitate further analysis and revision.

  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 #1

    Evidence, reproducibility and clarity

    The manuscript presents data demonstrating the function of BRI1 in removing the H3K27me3 epigenetic marks in genes involved in seed coat development in Arabidopsis. The results support that BRI1 may function here independently of brassinosteroid. The work combines genetics with a large panel of mutant lines, phenotyping by quantitative microscopy and chemical treatment, H3K27me3 profiling by CUT&TAG, and data mining for published gene profiling. The introduction is adequately informative, complete and explaining the state-of-the-art to the readers. The result part may be a bit lengthy (especially the first part) and some parts may be a bit repetitive.

    Major comments:

    1. Seed size is dependent of multiple factors. And few are explained here, notably the number of seeds per silique, the number of ovule per silique, the position of the silique of the branch (related to the age of the meristem), the total number of produced siliques (fertilised flowers) by the inflorescence meristem and the plant. And maybe if produced by the main and lateral branches. Were the authors consistent in the evaluation of analyzed siliques coming from the same type of branches, same age of the meristem, etc? Especially as some of the analysed mutants are dwarf, which is a sign of different plant fitness compared to WT.

    2. The seed perimeter measurements in BR mutant seeds (Figure S6) are variable. Are you sue the ovule size does not have any influence? What about presenting the relative size as earlier in the text?

    3. The number of evaluated samples is often {plus minus} n = 30, sometimes less, meaning less than what a silique contains of seeds. Did the authors evaluate the variability and reproductibility of their measurements, e.g, how many siliques per plant, how many plants, how many biological repeats? For example, in Figure S6, the number of measured ovules were as low as 16, which could be the reason why no significant difference in size were observed (low statitical strength). The variation in the Col WT is already visible. Is this variation significant?

    4. You indicate (line 149) that REF6 is not expressed in the gametophyte but GFP signal is observed in the cytoplasm for the central cell in Fig 1. The same goes for the expression pattern with the GUS line in Figure S2. (Line 290) One can not exclude expression in the endosperm or embryo with the presented pictures, or in the seed coat in older seeds.

    5. Make sure that you do not overstate your result conclusions, or add a reference to some of the statements. For example, line 185, for the choice of 3 DAP time point and the fact that seed coat development is based on cell expansion and interaction with the endosperm. Another example, in line 262, is where it is stated that the jmj mutants are compromised in ovule and pollen development. This was not assessed. You only checked the reduced seed set, not the fitness of the gametophytes. Or in line 337, where you indicate that KLUH is not expressed in all integument layers.

    6. Another example of this is in line 289 where you stated "a sporophytic function of JMJs at early stages of seed development, [..] and to a zygotic function at the later stages of seed development". I am not sure on what data do you base this conclusion as in all three categories (endosperm, embryo, seed coat) in Fig 2 and S5, genes are expressed in pre-globular stages. And again in line 475: "seed coat growth genes are expressed independentlyof fertilization". Do you have any evidence, a reference?

    7. (around lines 461) I understand that using a 35S promoter is not a good strategy as it would affect many other tissues. Did you consider using a tissue-specific approach as presented in Figure 4?

    8. You observed that the triple swn clf bri mutant is less dwarf than bri1 mutant and stated in line 483 that it is larger, has more leaves, grow tallerm and flower later and longer. Do you have any qunatitative data? If not, I would state that these observations are qualitative from growing plant aside.

    Minor comments:

    1. The title should precise the studies species, here Arabidopsis thaliana. Also the title of one of the part could be rephrased. "in a zygotic manner" sounds strange.

    2. Scale bars are missing in many figures.

    3. The font size in the graphs is small. The authors may use the empty space of the figures to increase the size of the graphs for clarity. Guidelines could be found here https://tpc.msubmit.net/html/TPC_Detailed_Figure_Guidelines.pdf, as example of good practices.

    4. Be consitent in the mutant name, e.g., brz1-D is also presented as brz1-d.

    5. Figure legend S1: I would not use the word "extremely" while you still have 30% seed set. Extremely would qualifiy for <5%, I guess.

    6. Figure S8 is missing the WT control for comparison.

    7. Figure S12, stats are missing

    8. I would recommend to add a line in the Supplemental tables with the name as this name disappears from the file name during upload. It would help the readers to navigate the data.

    9. Methods: Are all the lines listed used in the study? SR2200 is missing for the method, and please indicate the selection marker for each of the generated lines for open-access of the data if other researchers later use your lines.

    10. You have a duplicate for reference Vukašinovíc et al.

    11. Line 393, remove "s" in embryo and endosperm, in coat (line 674), in size (lines 684, 686

    12. Line 410, write RPS5A in upper case.

    13. LIne 676, the sentence "...H3K27me3 to be removed from the integuments." I would recomend to be more precise. For example "H3K27mme3 marks to be removed from genes to be expressed in the integuments" or something like that.

    Cross-commenting:

    I have been comparing our peer-review reports of the manuscript and found much similarity on our assessment:

    1. The seed size assemment and how this relates to seed coat development

    2. The GUS expression of ELF6 is not sufficient for the provided conclusion of the ELF6 expression

    3. The same would be for REP6

    4. Use of tissue-specific (seed coat specific) promoters to confirm the conclusion.

    Significance

    The authors provide novel information on the step-wise regulation of seed coat development and its influence on seed size. This is a topic of general interest, beyond the plant model Arabidopsis, especially in the context of reduced seed set caused by (a)biotic stress. The results of this study are valuable to understand seed size regulation in differnet growth context or species. The group previously showed that the auxin phytohormone is necessary after fertilization to initiate seed coat differentiation by inhibiting PRC2. However, as seed coat develops mainly as cell elongation, the epigenetic marks are not diluted by cell division and needs to be actively removed. This study provides insight into this process by identifcation 2 JMJ proteins responsible for removing H3K27me3 marks in the seed coat after fertilization to initiation seed coat development and regulating seed size. BRI1, BES1 and BZR1 are involved in this process, indepently of brassinosteroid, to guide JMJ to their target loci. While the study bring some genetic evidence of this process, molecular insight is still missing. Notably the identification of the target genes and how BRI1 is regulated/activated upon fertilization. Or how auxin and BRI1 co-regulate the process. These questions appear how of scope of this current study.

    My filed of expertise: hormones, plant reproduction, Arabidopis, oilseed rape, microscopy, transformation

  4. Version published to 10.1101/2023.12.07.569203v3 on bioRxiv
  5. Version published to 10.1101/2023.12.07.569203v2 on bioRxiv
  6. Version published to 10.1101/2023.12.07.569203v1 on bioRxiv