EMBO reportsembor.embopress.org/.../embr.201438643.reviewer-comments.pdfThe same reviewer also asks why the presence or absence of Foxd3 does not seem to influence the cells' response

EMBO reports - Peer Review Process File - EMBO-2014-38643

© European Molecular Biology Organization 1

Manuscript EMBO-2014-38643 Foxd3 suppresses NFAT-mediated differentiation to maintain self-renewal of embryonic stem cells Lili Zhu, Shiyue Zhang and Ying Jin Corresponding author: Ying Jin, Shanghai Institutes of Biological Sciences, Chinese Academy of Sciences/Shanghai JiaoTong University School of Medicine and Shanghai Stem Cell Institute, Shanghai JiaoTong University School of Medicine Review timeline: Submission date: 17 February 2014 Editorial Decision: 03 April 2014 Revision received: 10 July 2014 Editorial Decision: 21 August 2014 Additional correspondence (author): 27 August 2014 Additional correspondence (editor): 04 September 2014 Additional correspondence (author): 05 September 2014 Revision received: 30 September 2014 Accepted: 06 October 2014 Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.) Editor: Barbara Pauly

1st Editorial Decision 03 April 2014

Thank you for the submission of your research manuscript to our editorial offices. First of all, I would like to apologize for the delay in the review process of your manuscript. We have just now received the three enclosed reports on it. As you will see, while all referees agree on the potential interest of the manuscript, they have all raised a number of substantial concerns about the study and none of them is convinced that the data as they are presented at this point fully supports the hypothesis that Foxd3 regulates self-renewal of embryonic stem cells in the proposed way. As the reports are copied below I would like to outline the major issues only. All three referees raise concerns about the quality of the data, including both the quality of the microscopical images and of the results depicted in them, as well as the western blot data. For the latter, full blots of better quality should be shown with the relevant controls. Referee 1 also points out that the effects of Foxd3 on the expression of differentiation and stemness markers are rather mild and that it has to be shown that these effects are statistically significant. The same reviewer also asks why the presence or absence of Foxd3 does not seem to influence the cells' response to calcineurin/NFAT inhibition (figure 1E); this reviewer also feels that alternative explanations for some of the results should be discussed.



Referee 2 states that better proof for the existence of a Foxd3/NFAT/Tle4 complex in undifferentiated ESCs should be provided, including ChIP data on selected target genes. S/he also feels that the data on the role of Tle4 in counteracting NFAT-mediated gene expression should be strengthened by a loss-of-function approach. All reviewers also agree that the methods need to be explained in more detail. From the analysis of these comments it is clear that publication of your manuscript in our journal cannot be considered at this stage. On the other hand, given the potential interest of your study and the reviewers' constructive suggestions on how to improve it, we would like to give you the opportunity to address the referee concerns and would be willing to consider a revised manuscript with the understanding that the main issues raised by our reviewers must be fully addressed. I look forward to seeing a revised form of your manuscript when it is ready. Should you in the meantime have any questions, please do not hesitate to contact me. REFEREE REPORTS Referee #1: The paper describes a set of experiments aimed at showing that Foxd3, a transcription factor known to be important for the maintenance of pluripotency, functions through antagonizing the Calcineurin-NFAT signalling pathway. This is claimed to occur through a direct interaction between NFAT proteins and Foxd3 thereby repressing the activity of NFATs. The main issue with the paper is that in a number of cases the authors make claims for their data which are not supported upon close inspection. While the data is at times interesting and backed up by controls, the leaps made to make a complete story are too great for the dataset presented. Either the claims need toning down considerably or further experiments are required to obtain more solid evidence for the claims made. Another important deficiency is the quality of some of the figures. In particular the immunohistochemistry and brightfield images shown in figures 1 and 2 are poor and not fit for purpose. Further problems are with some of the blots such as in figure 3c and 5e where either purported bands are simply not visible or where non-specific signals mask the band of interest. Another major issue is that the experiments are poorly described with many of the important experiments not explained in sufficient or in some cases any detail. Specific points: The authors claim the conditional null results into a complete loss of expression. If so why is there a band in the qPCR after tamoxifen treatment in fig 1a? There is also still a band in the Western after tamoxifen treatment in fig 1a. If this is non-specific the authors need to comment? How specific is the FK506 inhibitor? A citation is required. The use of FK506 has the same effect on the cells whether Tamoxifen is added or not in fig 1e. This makes it hard to draw much of a conclusion as to whether the inhibitor is rescuing the Foxd3 null cells. The brightfield images in fig 2a and 2c are poor. The authors make the claim that fig 2a demonstrates that co-expression of NFATc3 and Foxd3 gives a different phenotype to that of NFATc3 alone. It is not clear to me from the images that this is the case. Furthermore, the claim that the removal of LIF is partially compensated for by overexpression of Foxd3 is not one I would make from looking at fig 2c. Why are these not AP stained?



The data leading to the claim for a suppression effect of Foxd3 on differentiation markers upon overexpression in fig 2d seems less than impressive. The effects are small on most of the markers shown and non-existent for T. The error bars for Igf2 are also so big that interpretation is difficult. The error bars for the -Lif Foxd3 overexpression are very big in Rex and Nanog samples in fig 2e. The interpretation is therefore difficult. The authors also claim that levels of Oct4 go up when Foxd3 is over-expressed but that is not the case. What is WCL in all blots ? Western control? The middle panel of figure 3a is the tightest cropped blot I've seen. This image has also been heavily adjusted in Photoshop. We need to see the rest of this blot in its original form The data in fig 3c is simply not good enough for publication. Where is the GST-Foxd3 on the lower panel? Where are the molecular weight markers? The cartoons in fig 3f need drawing approximately to scale. Why is there no IP with full length NFAT in fig 3g? It is needed for comparison of the strength of the IP. Is the size of NFATc-FH+C a little on the large size? How long is the LIF withdrawal in the experiments in figure 4d. Is the 3d suggesting 3 days? Is NFAT still expressed over these time frames or are levels changing with LIF withdrawal? Some q-PCR data to look show what is happening to NFAT would be informative. The data from fig 4c and 4d should be plotted on the same graph on the same scale so that we can see the relative actvities of the reporter in the 2 conditions. The authors claim that the data in figs 4e and f show that the NFAT binding truncation represses NFAT activity whilst the non-NFAT binding N-terminal part of Foxd3 does not. They go on to claim that this must mean that the interaction of Foxd3 and NFAT are required for the repression of NFAT activity. This is not necessarily the case. What about the situation in which the N-terminal region of Foxd3 is necessary for the repression via an interaction with another protein? The authors are very specific in their conclusions from these experiments and they are too strong. The EMSAs in fig 4g are of poor quality and are difficult to interpret. Surely if NFAT and Foxd3 were binding to the DNA to form a complex the complex would migrate with a larger size than the complex with either of the proteins alone? Are the authors suggesting that the complex forms off DNA? IF so they need to be much more specific. The anti-flag blot in fig. 5e is a mess. Also the whole figure is difficult to interpret as the + and - don't line up with the lanes (the same is true for 5f). How is there a signal in lane 1 of the top blot in fig. 5f? There is no HA-Foxd3 to IP so how can there be a Tle signal? The authors claim that the data is proof of a complex of 3 proteins? In the absence of step-wise IPs or other similar experiments it is just as likely that the data is a result of two separate complexes both being IP'd at the same time. Referee #2: In the manuscript entitled "Foxd3 counteracts NFAT-induced differentiation to maintain self-renewal of embryonic stem cells" Zhu et al. propose that an interaction between Foxd3 and NFATc3



recruits the corepressor Tle4 to target genes to maintain ES cell identity and suppress Nfat induced differentiation. This work aims at providing a mechanistic explanation for the phenotypes of Foxd3 deletion and Nfat overexpression which both induce differentiation in ES cells. The concept is of interest. However, the main conclusion that Foxd3 in complex with Tle4 counteracts Nfatc3 in undifferentiated ES cells needs to be substantiated by definitive experiments. Major points: 1. The hypothesis that Foxd3 represses differentiation in ES cells by counteracting Nfatc3 induced transcription via recruitment of Tle4 to differentiation genes requires the presence of a Foxd3/Nfatc3/Tle4 complex in undifferentiated ES cells. Presently evidence for a protein (or protein/DNA) complex containing Foxd3, Nfatc3 and Tle4 in ES cells is lacking. The only experiment indicating this is shown in Fig 3B, where CoIP shows pull down of Nfatc3 in differentiated ES cells transiently overexpressing flag-Foxd3. (Note: 3 days LIF withdrawal stated in the figure legend, but ESC extract stated in the main text! Please clarify - if interaction is only detectable during differentiation the interpretation must change). All other experiments describing protein interaction by Co-IP are performed by overexpression of multiple potential interaction partners in Cos7 cells. From these experiments no firm conclusions can be drawn about the situation in ES cells, or in fact about the presence of interactions in any unmodified cell type expressing endogenous protein levels. Ideally, the authors should perform key Co-IP experiments using Foxd3, Tle4 or Nfat3c specific antibodies; alternatively, if flag tagged proteins are used, near endogenous expression levels must be ensured and validated in stable ES cell lines to avoid artefacts. These experiments must be performed in undifferentiated ES cells. Binding in reciprocal co-IP should be shown. Furthermore co-recruitment of Foxd3, Nfatc3 and Tle4 on a selection of target genes needs to be shown by ChIP. Proper negative controls are needed, IgG is not sufficient (see also minor point 3). 2. The role of Tle4 in counteracting Nfatc3 induced activation of target genes has not been evaluated sufficiently. Fig5G and H show luciferase reporter assays after transient expression of a combination of Foxd3, Nfatc3, Tle4 and Aes in Cos7 and ES cells, respectively. However, Aes does not counteract only Tle4 function, therefore no conclusions about a specific role of Tle4 in the activation of Nfat target genes in the absence of Foxd3 can be drawn. A role for Tle4 could better be shown by a loss of function approach; either by siRNA or by genetic means. Furthermore, if Tle4 is indeed central for Foxd3 function in ES cells, then Tle4 depletion should produce a phenotype similar to that of Foxd3. This experiment should be performed, at least by siRNA, but more rigourously by gene mutation. Nfat expression induces differentiation in ES cells; therefore it is essential to show that the activity of the AP1 reporter is stable over the first days of ES cell differentiation. In the methods the authors state: "For NFAT:AP-1 reporter assays, LIF was removed on different days, as indicated in the text." I am unable to find these data in either text or figures. 3. The statement,"Foxd3 and NFAT displayed an opposite expression pattern during ESC differentiation" in the discussion is perplexing. Indeed the Western blot shown in Fig. 5c shows an absence of Nfatc3 protein in ES cells and a significant increase only at d7. If Nfatc3 functioned at d7 this would be downstream of the primary decision to differentiate and would make a direct role for Nfatc3 in promoting normal ESC differentiation unlikely. Furthermore, Foxd3 expression is lost early in differentiation. Therefore, according to data in Fig 5c Foxd3 and Nfatc3 are never coexpressed making an interaction unlikely and undermining the authors' proposition. In contrast to the results presented here, the authors showed in a previous paper (Cell Stem Cell, 2011, Fig. 2c) stable and clearly detectable expression of Nfatc3 by Western blot throughout the first 4 days of ESC differentiation after LIF withdrawal. They also show a clear Nfatc3 band in the Western shown in Fig3c in the current manuscript after 3 days of differentiation. The issue of when Nfatc3 is expressed must be clarified unambiguously. For this, as for other key points, pure populations of undifferentiated ES cells should be obtained by serum-free culture in inhibitors (2i) and LIF. Otherwise the point of action of Foxd3 or Nfatc3 will remain obscure. 4. No information about the statistical strength of qPCR results is given (number of replicates, technical or biological); this makes evaluation of these data near impossible. Also the reference conditions that were set to 1 should contain error bars. Minor points:



1. For key experiments the extent of transgene overexpression needs to be quantified on protein level. 2. Fig 1E: is 2uM a typo? 3. Fig1D and 4I: A specific action of Foxd3 on Src expression seems unlikely. This is for two reasons. Firstly, only a minor increase in Src expression is observed upon Foxd3 depletion. This could as well be secondary to increased differentiation. Secondly, the ChIP signal in Fig 4I is minimal. Proper controls are required. IgG is not sufficient. Negative genomic control regions must be used to exclude low background affinity of Foxd3 to the Src locus. 4. In the text the authors state "In comparison to control cells, ESCs overexpressing Foxd3 remained a partially colonial morphology (Figure 2C), higher levels of pluripotency-associated genes (such as Rex1, Oct4 and Nanog) and lower levels of lineage markers (Fgf5, T, Dab2 and Eomes) as well as EMT markers (Twist and Igf2) (Figure 2D and 2E)." Firstly, I cannot see maintenance of ES colony morphology in these images; eg Nanog/Oct4 IF staining could support this statement. Secondly, Foxd3 overexpression does not appear to delay differentiation significantly on marker expression level. Only Fgf5 expression seems to be slightly delayed, all other genes show an increase in differentiation regardless of increased Foxd3 levels. The error bars for Igf2 are too large to judge. The same is true for the expression of Rex1 and Nanog. Although levels seem increased upon Foxd3 overexpression, the kinetics of downregulation in differentiation are obfuscated by massive error bars. Oct4 expression is unchanged in all conditions. This experiment should be repeated, ideally in a differentiation time course, and results described accurately in the text. 5. Fig 6A: overexpression of Nfatc3 as well as deletion of Foxd3 will lead to differentiation. Therefore, apart from measuring gene specific functions global transcription analysis will also assay differentiation effects. This is very difficult to control but the limitation should be clearly indicated in the text. One possibility would be to attempt Foxd3 deletion and Nfat overexpression in 2iLIF medium, which allows maintenance of ES cells in the absence of several transcription factors that appear essential for ES cell self-renewal in FCS/LIF conditions. This might make it possible to discriminate cause from consequence and importantly also illuminate whether effects are manifest in truly undifferentiated ES cells. 6. Supplementary Fig 3. The authors state that they detect a "small" fraction of NFATc3 protein in the nuclear fraction in mouse ES cells. However, the Western blot in Supp Fig 3 shows very similar amounts in cytoplasmic and nuclear fractions. These results should be described accurately. 7. Materials and methods are not very detailed eg. details on ChIP protocol are missing; was it performed using Foxd3 antibody or a flag tagged protein? Cell culture and transfection experiments are not described in sufficient detail. 1. Does this manuscript report a single key finding? Yes. The main finding claimed by Zhu et al. is an interaction of Foxd3 with NFATc3 which ensures maintenance of ES cell identity by recruitment of the corepressor Tle4 to Nfat targets, thereby suppressing induction of differentiation. 2. Is the reported work of significance (YES), or does it describe a confirmatory finding or one that has already been documented using other methods or in other organisms etc (NO)? YES 3. Is it of general interest to the molecular biology community? YES 4. Is the single major finding robustly documented using independent lines of experimental evidence (YES), or is it really just a preliminary report requiring significant further data to become convincing, and thus more suited to a longer ¬format article (NO)? Preliminary, but a longer format is not required, just better experiments.



Referee #3: The manuscript describes an interesting study of the interaction between the pluripotency factor Foxd3 and the NFAT protein NFATc3. Overall, the results and conclusions drawn from them are broadly convincing and are sufficiently interesting to merit publication in EMBO Reports. My main caveats relate to the lack of detail about the experiments, both in the text, and to an even greater degree in the figure legends. This makes the paper a difficult read. Extensive revision is required to make it clearer. Specific points Figure 1D: Is the effect of FK506 on Src expression significant? How many replicates were performed to generate this result? This question also applies to Figures 1F and 2D and E. Figure 2C: Although the morphology of the LIF- Foxd3 overexpressing cells might well differ from the control cells, the differences between the control and Foxd3 overexpressing plates are so subtle that is difficult to exclude trivial causes such as differences in plating density. Figure 3B: The legend for this figure does not give enough information. What does WCL stand for? Figure 3C: Again the legend does not provide enough information. There is no indication of what the upper and lower panels are showing, or why two panels are shown. Page 9: "Western blot analysis showed that both Forkhead domain (110 amino acids) and C-terminus (251 amino acids) of Foxd3 were capable of binding with NFATc3 (Figure 3E)". It seems from Figure 3E that only Foxd3 N fails to bind. It would be helpful to state this in the text. It means that this deletion series was not particularly informative for defining the interacting regions and this should be stated in the text. The deletions of NFATC were more informative as they define the domain from 1-595 as the interacting region. Page 10: "Importantly, the activation was dramatically suppressed by co-transfection of Foxd3 with all NFATs but NFATc2 overexpressing cells (Figures 4A and B), revealing strong inhibition of Foxd3 on the transcriptional activities of NFATc1, 3 and 4." This sentence does not give a very clear explanation of the data in Figure 4A and B. Page 10: The difference between NFAT: K3-luc and NFAT:AP1-luc is not explained in the text or the legend to Figure 4, nor is it clear why they give different results. Giving references for the constructs is not sufficient. Figure 4: Asterisks next to histogram bars presumably indicate levels of significance (the reader does discover this when they reach the Methods) but no information about their meaning is given in the figure legend. Page 11: The EMSA experiment shown in Figure 4G and H is not well explained. Presumably, the IL2 control probe contains a previously characterized Nfatc3 binding site. However, this is not stated in the text or the figure legend. The binding patterns shown in Figure 4H are complex and it is unclear how the bands are identified as containing either or both proteins in the cells that have both co-transfected, without doing antibody supershifts. Figure 4I: The meaning of the double asterisk is not specified in the figure legend. Figure 5A: The labeling of the y axis should specify that expression is relative to expression of Foxd3. Figure 5C: Although western blotting of Tle4 shows a decrease to day 5 after LIF withdrawal, the level of the protein then seems to increase at day 7. This is not referred to or explained in the text.



1st Revision - authors' response 10 July 2014

Reviewer 1

COMMENTS FOR THE AUTHOR:

The paper describes a set of experiments aimed at showing that Foxd3, a transcription factor known

to be important for the maintenance of pluripotency, functions through antagonizing the

Calcineurin-NFAT signalling pathway. This is claimed to occur through a direct interaction between

NFAT proteins and Foxd3 thereby repressing the activity of NFATs.

The main issue with the paper is that in a number of cases the authors make claims for their

data which are not supported upon close inspection. While the data is at times interesting and backed

up by controls, the leaps made to make a complete story are too great for the dataset presented.

Either the claims need toning down considerably or further experiments are required to obtain more

solid evidence for the claims made.

Another important deficiency is the quality of some of the figures. In particular the

immunohistochemistry and brightfield images shown in figures 1 and 2 are poor and not fit for

purpose. Further problems are with some of the blots such as in figure 3c and 5e where either

purported bands are simply not visible or where non-specific signals mask the band of interest.

Another major issue is that the experiments are poorly described with many of the important

experiments not explained in sufficient or in some cases any detail.

Response: Thanks for pointing out these issues. We have toned down some of claims and also

conducted further experiments to provide solid evidence for our claims, as you will see in our

response to your comments. Moreover, we have improved the quality of figures and added more

details to methods.

Specific points:

Comment 1.The authors claim the conditional null results into a complete loss of expression. If so

why is there a band in the qPCR after tamoxifen treatment in fig 1a? There is also still a band in the

Western after tamoxifen treatment in fig 1a. If this is non-specific the authors need to comment?

Response 1: To answer the question, we verified the RT-PCR result shown in original Fig. 1a using

quantitative real-time RT-PCR. Consistently, Tamoxifen treatment led to an almost complete loss of

Foxd3 expression. Indeed, there seems trace amount of Foxd3 proteins remained. Therefore, we

have changed “complete loss” to “ an almost complete loss” in the text of the revised manuscript.

The protein band underneath the specific Foxd3 protein band is a non-specific band, which is

marked in the revised figure. Due to the addition of new data and limitation of space, data of

original figure 1a-1c are moved to Supplementary Fig S1 in the revised version of the manuscript

and new qRT-PCR data are shown in the current Fig 1A.

Comment 2. How specific is the FK506 inhibitor? A citation is required.



Response 2: FK506 is a specific inhibitor of NFAT pathway and a citation is added in the

manuscript text where FK506 was first used.

Comment 3. The use of FK506 has the same effect on the cells whether Tamoxifen is added or not

in fig 1e. This makes it hard to draw much of a conclusion as to whether the inhibitor is rescuing the

Foxd3 null cells.

Response 3: From gene expression analysis (current Fig 1D), FK506 did not alter marker gene

levels in the absence of Tamoxifen treatment, except for the level of Fgf5. In contrast, with

Tamoxifen treatment, FK506 clearly reduced the elevation in the differentiation-associated gene

expression induced by Foxd3 depletion. It appears that FK506 treatment made ESC colonies more

compact even in the absence of Tamoxifen. It could be due to the effect of FK506 on the possible

existence of the basal NFAT activity in ESCs. Although FK506 reduced the elevated differentiation

gene expression, the effect was neither statistically significant nor complete. This could be attributed

to the following two reasons. One could be relatively large variation in gene expression levels

among three independent experiments. Second, Foxd3 depletion might also lead to NFAT-

independent changes. The discussion is added to the revised manuscript.

Comment 4.The brightfield images in fig 2a and 2c are poor. The authors make the claim that fig 2a

demonstrates that co-expression of NFATc3 and Foxd3 gives a different phenotype to that of

NFATc3 alone. It is not clear to me from the images that this is the case. Furthermore, the claim that

the removal of LIF is partially compensated for by overexpression of Foxd3 is not one I would make

from looking at fig 2c. Why are these not AP stained?

Response 4: As suggested by the reviewer, we repeated the experiment in original Fig. 2a and

examined the alkaline phosphatase (AKP) activities by staining. Consistent results are obtained

(shown in Supplementary Fig. S1B in the revised manuscript). The result shows that overexpression

NFATc3 alone reduced the intensity of AKP staining, while cooverexpression of Foxd3 and

NFATc3 rescued the AKP intensity to certain extents. Importantly, gene expression analysis

supports that notion that Foxd3 coexpression could rescue the differentiation induced by NFATc3

significantly (Current Fig 1F).

We agree with the reviewer that the rescue effect of Foxd3 on LIF-withdrawal induced

differentiation was not as efficient as under NFATc3 overexpression condition, although Foxd3

could block LIF-withdrawal induced differentiation partially. This could be due to the involvement

of multiple factors, in addition to the activation of NFAT, in LIF-withdrawal induced ESC

differentiation. For better quality of the manuscript, we have removed original Fig. 1c, 1d and 1e as

well as related manuscript text from revised manuscript. I believe that this would not change the

main theme of the study, which focuses on the interaction between Foxd3 and NFAT signaling.

Comment 5.The data leading to the claim for a suppression effect of Foxd3 on differentiation

markers upon overexpression in fig 2d seems less than impressive. The effects are small on most of

the markers shown and non-existent for T. The error bars for Igf2 are also so big that interpretation

is difficult.

Response 5: The answer to this comment is the same as to comment 4.



Comment 6.The error bars for the -Lif Foxd3 overexpression are very big in Rex and Nanog

samples in fig 2e. The interpretation is therefore difficult. The authors also claim that levels of Oct4

go up when Foxd3 is over-expressed but that is not the case.

Response 6: The answer to this comment is the same as to comment 4.

Comment 7.What is WCL in all blots? Western control?

Response 7: I am sorry for missing the explanation for this abbreviation. The WCL stands for whole

cell lysate in all western blot figures. The explanation is included in all legends now.

Comment 8.The middle panel of figure 3a is the tightest cropped blot I've seen. This image has also

been heavily adjusted in Photoshop. We need to see the rest of this blot in its original form.

Response 8: We cropped the blot tightly for original Fig. 3A because we tried to avoid including

man-made marks in the figure. As suggested by the reviewer, we have carefully cropped the blot

and remade the figure, which is shown in Fig. 2A of revised manuscript. The whole blot in its

original form is provided below.

Comment 9.The data in fig 3c is simply not good enough for publication. Where is the GST-Foxd3

on the lower panel? Where are the molecular weight markers?Response 9: I am so sorry for this.

We have conducted the GST pull down assay again. The new result is shown in Fig. 2C of the

revised manuscript. The lower panel is the western blot result for GST and GST-Foxd3 proteins

used in the pull-down assay. GST-Foxd3 proteins have a specific band above 70 kDa marker, and a

degradation band around 70 kDa.



Comment 10.The cartoons in fig 3f need drawing approximately to scale.

Response 10: We have re-drawn the schematic cartoons for Foxd3 and NFATc3 domains. They are

shown in Figs. 2D and 2F, respectively, in the revised manuscript.

Comment 11.Why is there no IP with full length NFAT in fig 3g? It is needed for comparison of the

strength of the IP.

Response 11: We have conducted CoIP experiment again with full length HA-NFATc3. The result,

showing in new Fig. 2G, shows that NFATc3-FL, NFATc3-N and NFATc3-N+M can interact with

Foxd3.

Comment 12. Is the size of NFATc-FH+C a little on the large size?

Response 12: The “FH” domain in the original Fig. 3G should be “M” domain (middle domain).

The size of NFATc3-M+C is little larger than expected. This might be attributed to some kind of

posttranslational modifications.

Comment 13. How long is the LIF withdrawal in the experiments in figure 4d. Is the 3d suggesting

3 days? Is NFAT still expressed over these time frames or are levels changing with LIF withdrawal?

Some q-PCR data to look show what is happening to NFAT would be informative.

Response 13: Sorry for not explaining this clearly in our original manuscript. Reporter assays in

original Fig. 4D were conducted in ESCs without LIF for 4 days. “- LIF 4 days” is added to the

figure, shown in Fig. 3B of the revised manuscript. The expression pattern of NFAT was extensively

studied and clearly described shown our previous paper (Li and Zhu et al, Cell Stem Cell, 8:46-58,

2011). I am copying Figs. 2C and 2D of the paper below. It showes that NFATc4 protein level

increased during LIF withdrawal, while the protein level of NFATc3 was not changed. However,

proteins of NFATc3 were activated by LIF withdrawal, as they translocated from the cytoplasm into

the nucleus. Therefore, the expression level of NFATc3 is not changed after LIF withdrawal, rather

its subcellular localization changes during ESC differentiation. We normally determine its activity

through the reporter assay. An explanation is added to the revised manuscript.



Comment 14.The data from fig 4c and 4d should be plotted on the same graph on the same scale so

that we can see the relative activities of the reporter in the 2 conditions.

Response 14: This is a good suggestion. We have plotted the data from original Figs. 4D and 4D on

the same graph and the same scale. It is shown in Fig. 3B of the revised manuscript.

Comment 15.The authors claim that the data in figs 4e and f show that the NFAT binding

truncation represses NFAT activity whilst the non-NFAT binding N-terminal part of Foxd3 does

not. They go on to claim that this must mean that the interaction of Foxd3 and NFAT are required

for the repression of NFAT activity. This is not necessarily the case. What about the situation in

which the N-terminal region of Foxd3 is necessary for the repression via an interaction with another

protein? The authors are very specific in their conclusions from these experiments and they are too

strong.

Response 15: We have changed some words to tone down the claim, “suggesting that the physical

association with NFAT is probably important for the Foxd3 to inhibit NFAT’s transcriptional

activities.

Comment 16: The EMSAs in fig 4g are of poor quality and are difficult to interpret. Surely if

NFAT and Foxd3 were binding to the DNA to form a complex the complex would migrate with a

larger size than the complex with either of the proteins alone? Are the authors suggesting that the

complex forms off DNA? IF so they need to be much more specific.

Response 16: To improve the quality of EMSA result, we conducted the assay again with slightly



modified protocol and included the super-shift using antibodies against NFATc3 and Foxd3,

respectively, to show the specificity of the complex. The new result is shown in Fig. 3D. As the

reviewer pointed out, the size of NFATc3+Foxd3 proteins+DNA complex is larger than that of

Foxd3 with DNA complex or NFATc3 with DNA complex. In addition, we observed supershift

when NFATc3 or Foxd3 antibody was included in the reaction mixture. These data consolidate the

notion that NFATc3 and Foxd3 form protein complexes and bind the same sequence of the Src

promoter region sequence.

Comment 17: The anti-flag blot in fig. 5e is a mess. Also the whole figure is difficult to interpret as

the + and - don't line up with the lanes (the same is true for 5f).

Response 17: Sorry for the mess. We have conducted the CoIP experiments again and remade

figures. The quality of new data is markedly improved and the result is shown in Fig. 4B of the

revised manuscript.

Comment 18: How is there a signal in lane 1 of the top blot in fig. 5f? There is no HA-Foxd3 to IP

so how can there be a Tle signal?

Response 18: Thank you for pointing this out. You are right. There should not be a signal in lane 1

of the top blot in original Fig. 5f. The signal could be due to leaked proteins from the next lane on

the same gel, which happened to contain Flag-Tle4 proteins. In the revised manuscript, we have

used new data to show that Foxd3, NFATc3 and Tle4 form protein complexes in ESCs. The new

data are shown in Figs. 4C, 4D and 4E.

Comment 19: The authors claim that the data is proof of a complex of 3 proteins? In the absence of

step-wise IPs or other similar experiments it is just as likely that the data is a result of two separate

complexes both being IP'd at the same time.

Response 19: To address this issue, we conducted the CoIP experiments with antibodies against

Tle4, Flag-NFATc3, and Flag-Foxd3, respectively, in mouse ESCs and then immunoprecipitated

proteins were western blotted with antibodies against endogenous Foxd3, NFATc3 and Tle4,

respectively. In this way, we are able show that Foxd3 protein complexes contain NFATc3 and Tle4

proteins, that NFATc3 protein complexes contain Tle4 and Foxd3 proteins, and that Tle4 protein

complexes contain NFATc3 and Foxd3 proteins. Thus, it is likely that complexes containing all

three proteins exist in mouse ESCs.

Reviewer 2


In the manuscript entitled "Foxd3 counteracts NFAT-induced differentiation to maintain self-

renewal of embryonic stem cells" Zhu et al. propose that an interaction between Foxd3 and NFATc3

recruits the corepressor Tle4 to target genes to maintain ES cell identity and suppress Nfat induced

differentiation. This work aims at providing a mechanistic explanation for the phenotypes of Foxd3

deletion and Nfat overexpression which both induce differentiation in ES cells. The concept is of



interest. However, the main conclusion that Foxd3 in complex with Tle4 counteracts Nfatc3 in

undifferentiated ES cells needs to be substantiated by definitive experiments.

Specific points:

Comment 1: The hypothesis that Foxd3 represses differentiation in ES cells by counteracting

Nfatc3 induced transcription via recruitment of Tle4 to differentiation genes requires the presence of

a Foxd3/Nfatc3/Tle4 complex in undifferentiated ES cells.

Presently evidence for a protein (or protein/DNA) complex containing Foxd3, Nfatc3 and Tle4 in

ES cells is lacking. The only experiment indicating this is shown in Fig 3B, where CoIP shows pull

down of Nfatc3 in differentiated ES cells transiently overexpressing flag-Foxd3. (Note: 3 days LIF

withdrawal stated in the figure legend, but ESC extract stated in the main text! Please clarify - if

interaction is only detectable during differentiation the interpretation must change). All other

experiments describing protein interaction by Co-IP are performed by overexpression of multiple

potential interaction partners in Cos7 cells. From these experiments no firm conclusions can be

drawn about the situation in ES cells, or in fact about the presence of interactions in any unmodified

cell type expressing endogenous protein levels.

Ideally, the authors should perform key Co-IP experiments using Foxd3, Tle4 or Nfat3c specific

antibodies; alternatively, if flag tagged proteins are used, near endogenous expression levels must be

ensured and validated in stable ES cell lines to avoid artefacts. These experiments must be

performed in undifferentiated ES cells. Binding in reciprocal co-IP should be shown.

Furthermore co-recruitment of Foxd3, Nfatc3 and Tle4 on a selection of target genes needs to be

shown by ChIP. Proper negative controls are needed, IgG is not sufficient (see also minor point 3).

Response 1: To address the question of forming protein complexes containing endogenous

NFATc3, Foxd3 and Tle4 in undifferentiated and unmodified ESCs, we carried out CoIP

experiments using Tle4 specific antibody to immunoprecipitate endogenous Tle4, Foxd3 and

NFATc3 in unmodified undifferentiated ESCs. Western blot with specific antibody against Tle4,

Foxd3 and NFATc3 indicate that Tle4 antibody-immunoprecipitated protein complexes do contain

these three proteins. To further verify the existence of the protein complexes containing these three

proteins, reciprocal CoIP experiments were performed in Flag-Foxd3 or Flag-NFATc3 stably

expressed ESCs using specific antibody against Foxd3 or NFATc3 to precipitate protein complexes

containing Foxd3 or NFATc3. Subsequent western blot analysis using specific antibodies of

NFATc3 and Tle4 or Foxd3 and Tle4 shows that Foxd3 protein complexes contain NFATc3 and

Tle4 proteins and that NFATc3 protein complexes contain Foxd3 and Tle4 proteins. These data are

shown in current Figs.4C, 4D and 4E.

In addition, following the suggestion of reviewer 2, we conducted three endogenous ChIP

experiments in undifferentiated ESCs using specific antibodies against Foxd3, Tle4 or Nfat3c,

respectively. The results of current Fig 3F show that Foxd3, NFATc3 and Tle4 all can bind the same

Src promoter region, indicating that they might form a complex at the Src promoter to regulate Src

expression. Also, the negative control, which we used in this experiment, was a region of genomic

DNA between the GAPDH gene and the chromosome condensation-related SMC-associated protein



(CNAP1) gene. None of our three factors was enriched in this region. This result is shown in Fig.

3E-G.

Comment 2: The role of Tle4 in counteracting Nfatc3 induced activation of target genes has not

been evaluated sufficiently. Fig5G and H show luciferase reporter assays after transient expression

of a combination of Foxd3, Nfatc3, Tle4 and Aes in Cos7 and ES cells, respectively.

However, Aes does not counteract only Tle4 function; therefore no conclusions about a specific role

of Tle4 in the activation of Nfat target genes in the absence of Foxd3 can be drawn. A role for Tle4

could better be shown by a loss of function approach; either by siRNA or by genetic means.

Furthermore, if Tle4 is indeed central for Foxd3 function in ES cells, then Tle4 depletion should

produce a phenotype similar to that of Foxd3. This experiment should be performed, at least by

siRNA, but more rigourously by gene mutation.

Nfat expression induces differentiation in ES cells; therefore it is essential to show that the activity

of the AP1 reporter is stable over the first days of ES cell differentiation. In the methods the authors

state: "For NFAT:AP-1 reporter assays, LIF was removed on different days, as indicated in the text."

I am unable to find these data in either text or figures.

Response 2: To show the specific role of Tle4 in the activation of NFAT target genes, we used

specific siRNA against Tle4 to knock down the expression level of Tle4 in ESCs. As shown in

Supplementary Fig. S3A, two siRNAs against Tle4 both could significantly reduced Tle4 transcript

levels, while control siRNA did not affect the level of Tle4. When Tle4 siRNAs were introduced

into the NFAT reporter system in ESCs, we observed significant reduction in the capacity of Foxd3

to suppress NFAT transcriptional activities, while control siRNA did not have such a role. This

result suggests that Tle4 plays a specific role for Foxd3 to inhibit NFAT transcriptional activities.

Second, in order to test if Tle4 also plays a role for the maintenance of ESCs in an

undifferentiated state, Tle4 siRNAs were introduced into undifferentiated ESCs and marker gene

expression was examined by qRT-PCR. We found that knocking down of Tle4 expression led to

reduced expression of pluripotency-associated markers (Rex1 and Oct4) and increased expression of

differentiation-associated markers (Gata6, T and Fgf5). The results are shown in Supplementary

Figs. S3C and S3D.

Third, NFAT:AP1 Luc reporter is an effective tool to detect NFAT transcriptional activities as

it has a range of 50~1000 fold activation when NFATc3 is overexpressed in different cell types (Fig

3 and Fig 4). In revised Figure 3B, we show that the reporter had a 8-fold activation after LIF was

removed for 4 days. The extent of reporter activation is far lower than that in NFAT overexpression.

Last, sorry for not providing sufficient information in our original manuscript. In the revised

manuscript, we removed the sentence from method “For NFAT:AP-1 reporter assays, LIF was

removed on different days, as indicated in the text.”. Instead, we have provided necessary

information in the legend and text of the revised manuscript.

Comment 3: The statement,"Foxd3 and NFAT displayed an opposite expression pattern during

ESC differentiation" in the discussion is perplexing. Indeed the Western blot shown in Fig. 5c shows

an absence of Nfatc3 protein in ES cells and a significant increase only at d7. If Nfatc3 functioned at

d7 this would be downstream of the primary decision to differentiate and would make a direct role



for Nfatc3 in promoting normal ESC differentiation unlikely. Furthermore, Foxd3 expression is lost

early in differentiation. Therefore, according to data in Fig 5c Foxd3 and Nfatc3 are never

coexpressed making an interaction unlikely and undermining the authors' proposition.

In contrast to the results presented here, the authors showed in a previous paper (Cell Stem Cell,

2011, Fig. 2c) stable and clearly detectable expression of Nfatc3 by Western blot throughout the first

4 days of ESC differentiation after LIF withdrawal. They also show a clear Nfatc3 band in the

Western shown in Fig3c in the current manuscript after 3 days of differentiation. The issue of when

Nfatc3 is expressed must be clarified unambiguously. For this, as for other key points, pure

populations of undifferentiated ES cells should be obtained by serum-free culture in inhibitors (2i)

and LIF. Otherwise the point of action of Foxd3 or Nfatc3 will remain obscure.

Response 3: The question of when NFATc3 is expressed has been unambiguously described in our

previous paper (Cell Stem Cell, 2011). As shown in Fig. 2C of that paper, the steady-state level of

NFATc3 proteins is stable in ESCs either with or without LIF in ESCs, while NFATc4 protein

levels increases markedly after LIF withdrawal. However, NFATc3 proteins distribute

predominantly in the cytoplasm, being transcriptionally inactive in undifferentiated ESC (Fig. 2D in

Cell Stem Cell paper).

Up differentiation, NFATc3 is dephosphorylated by calcineurin and translocate into the nucleus.

Therefore, NFATc3’s function is regulated through its subcellular localization, but not its protein

level during ESC differentiation. It is expressed in the undifferentiated ESCs, but the majority of

them are phosphorylated and functionally inactive. Nevertheless, our western blot analysis using



fractionated cytoplasmic and nuclear proteins of ESCs suggests that there are certain amounts of

NFATc3 proteins in the nucleus of undifferentiated ESCs. Data are shown in Supplementary Fig. S5

in the revised manuscript. Consistently, in the current study, we did not detect marked changes in

the steady-state level of NFATc3 proteins during the first 5 days after LIF withdrawal. The lower

NFATc3 signal in original Fig. 5C, compared with the NFATc3 signal in Fig. 2C of our previous

Cell Stem Cell paper, is probably due to the shorter exposure time used when film was developed.

As pointed out by Reviewer 2, the statement that Foxd3 and NFAT displayed an opposite expression

pattern during ESC differentiation was not accurate. In fact, Foxd3 protein levels decreases

gradually and more NFATc3 proteins translocate from the cytoplasm into the nucleus during early

days of ESC differentiation. The interaction between NFATc3 and Foxd3 proteins take place both in

undifferentiated ESCs and early differentiated cells, which has been shown in the current Figs.

Moreover, our CoIP experiments among NFATc3, Foxd3 and Tle4 were also conducted with whole

cell lysates from undifferentiated ESCs (Figs. 4C, 4D and 4E), which further verifies that these three

proteins interact in undifferentiated ESCs. I hope that these lines of experimental evidence are

convincing. In order to avoid confusion, the top panel (NFATc3 western blot result) is removed

from the figure. The lower 3 panels (Tle4, Foxd3 and GAPDH western result) are shown in the

current Supplementary Fig S2C.

Regarding to conduct further experiments in 2i condition, our previous study showed that

inhibition of Erk1/2 signaling would inactivate the NFAT signaling. Therefore, it would be difficult

to study the interaction between Foxd3 and NFAT when Erk1/2 signaling is inhibited by 2i.

Comment 4: No information about the statistical strength of qPCR results is given (number of

replicates, technical or biological); this makes evaluation of these data near impossible. Also the

reference conditions that were set to 1 should contain error bars.

Response 4: Thank you for your criticism. All qPCR results were generated from three biological

replicates. In the revised manuscript, we have added necessary statistic information in every related

legend. Reference contains error bars too.

Minor points:

Comment 1: For key experiments the extent of transgene overexpression needs to be quantified on

protein level.

Response 1: As suggested, we include transgene protein levels for key experiments�such as Figs.

2A, 2B, 2E, 2G and Figs. 4A-E.

Comment 2: Fig 1E: is 2uM a typo?

Response 2: No, we added 2 mM of TM (Tamoxifen) to induce the deletion of Foxd3 in Foxd3

conditional knockout ESCs. To avoid the confusion, we simply mark the figure (original Fig. 1E,

current Fig. 1B) with – TM or + TM, respectively. The information of 2 mM is given in the legend.

Comment 3: Fig1D and 4I: A specific action of Foxd3 on Src expression seems unlikely. This is for

two reasons. Firstly, only a minor increase in Src expression is observed upon Foxd3 depletion. This

could as well be secondary to increased differentiation. Secondly, the ChIP signal in Fig 4I is



minimal. Proper controls are required. IgG is not sufficient. Negative genomic control regions must

be used to exclude low background affinity of Foxd3 to the Src locus.

Response 3: Thanks for pointing out this issue. We have added the statement “Nevertheless,

we do not rule out the possibility that Foxd3 depletion-induced ESC differentiation accounted for

the increased Src mRNA level.” To further address this comment, we conducted Foxd3 ChIP

experiment in ESCs again with a negative genomic control, which was a region of genomic DNA

between the GAPDH gene and the chromosome condensation-related SMC-associated protein

(CNAP1) gene. The genomic DNA of this region (cytogenetic location 12 p13.3) should not be

bound by transcription factors. Our qPCR result indicates that Foxd3 specifically associates with the

Src promoter sequence, but not the control sequence. The result is shown in the left panel of Fig.

3E.

Comment 4: In the text the authors state "In comparison to control cells, ESCs overexpressing

Foxd3 remained a partially colonial morphology (Figure 2C), higher levels of pluripotency-

associated genes (such as Rex1, Oct4 and Nanog) and lower levels of lineage markers (Fgf5, T,

Dab2 and Eomes) as well as EMT markers (Twist and Igf2) (Figure 2D and 2E)."

Firstly, I cannot see maintenance of ES colony morphology in these images; eg Nanog/Oct4 IF

staining could support this statement.

Secondly, Foxd3 overexpression does not appear to delay differentiation significantly on marker

expression level. Only Fgf5 expression seems to be slightly delayed, all other genes show an

increase in differentiation regardless of increased Foxd3 levels. The error bars for Igf2 are too large

to judge. The same is true for the expression of Rex1 and Nanog. Although levels seem increased

upon Foxd3 overexpression, the kinetics of downregulation in differentiation are obfuscated by

massive error bars. Oct4 expression is unchanged in all conditions. This experiment should be

repeated, ideally in a differentiation time course, and results described accurately in the text.

Response 4: In response to Reviewer 2’s comment, we repeated experiments of original Figs. 2D

and 2E and obtained consistent results with small error bars for tested genes. Indeed, the suppressive

effect of Foxd3 on LIF-withdrawal induced ESC differentiation is mild, although statistically

significant for certain tested marker genes. This could be due to the involvement with multiple

factors in LIF withdrawal-induced differentiation, in additional to NFAT activation. Based on this

consideration, we have removed the data of original Fig. 2D and Fig. 2E as well as related

manuscript text from the current manuscript and do not claim that Foxd3 could partially inhibit LIF

withdrawal-induced ESC differentiation. This would not affect the main scheme of the current

study, which focuses on the interaction between Foxd3 and NFATc3.

Comment 5: Fig 6A: overexpression of Nfatc3 as well as deletion of Foxd3 will lead to

differentiation. Therefore, apart from measuring gene specific functions global transcription analysis

will also assay differentiation effects. This is very difficult to control but the limitation should be

clearly indicated in the text. One possibility would be to attempt Foxd3 deletion and Nfat

overexpression in 2i/LIF medium, which allows maintenance of ES cells in the absence of several

transcription factors that appear essential for ES cell self-renewal in FCS/LIF conditions. This might

make it possible to discriminate cause from consequence and importantly also illuminate whether



effects are manifest in truly undifferentiated ES cells.

Response 5: Thanks for the suggestion. We have indicated the limitation of global transcription

analysis under the condition of overexpression of NFATc3 as well as deletion of Foxd3 in the

revised manuscript.

It would be interesting to compare global gene expression profiles between ESCs

overexpressing NFATc3 and deleting Foxd3 in the 2i/LIF medium and in the FCS/LIF medium.

However, the study would involve substantially more experiments and costs. In addition, our

previous study (Cell Stem Cell, 2011) demonstrated that Fgf/Mek/Erk1/2 and calcineurin signaling

are mutually dependent. Inhibiting one of the pathways will inactivate the other one. Our previous

data also showed that inhibitor of calcineurin alone or in combination with GSK3b inhibitor could

maintain ESC self-renewal under the serum-free condition in the absence of LIF (Cell Stem Cell,

2011). Therefore, Calcineurin-NFAT and Erk1/2 signaling should be both inhibited if ESCs would

be cultured under a 2i condition. We think that the serum/LIF culture condition offers a better

condition to study the interaction between Calcineurin-NFAT pathway and Foxd3 transcription

factor. Nevertheless, it is worth to explore the role of Foxd3 for ESC self-renewal under a 2i/LIF

condition. For the sake of limited revision time and research budget, we would like very much to

carry out such investigation in our next project.

Comment 6: Supplementary Fig 3. The authors state that they detect a "small" fraction of NFATc3

protein in the nuclear fraction in mouse ES cells. However, the Western blot in Supp Fig 3 shows

very similar amounts in cytoplasmic and nuclear fractions. These results should be described

accurately.

Response 6: Sorry for not providing sufficient experimental details for original Supplementary Fig.

3. It is true that the majority of NFATc3 proteins distribute in the cytoplasm, with a small fraction

being in the nucleus of undifferentiated ESCs. In order to use the similar time to develop the film,

we loaded 100 mg of the nuclear lysate and 30 mg of the cytoplasmic lysate onto the gel for

subsequent western blot analysis. Even under such condition, the intensity of NFATc3 protein in the

nucleus was lower than that in the cytoplasm (Supplemental Fig. S5 in the current version of the

manuscript). The information is provided in the revised figure legend.

Comment 7: Materials and methods are not very detailed eg. details on ChIP protocol are missing;

was it performed using Foxd3 antibody or a flag tagged protein? Cell culture and transfection

experiments are not described in sufficient detail.

Response 7: We have added necessary information in the legend and methods in the revised

manuscript.

Reviewer 3


The manuscript describes an interesting study of the interaction between the pluripotency factor

Foxd3 and the NFAT protein NFATc3. Overall, the results and conclusions drawn from them are

broadly convincing and are sufficiently interesting to merit publication in EMBO Reports. My main



caveats relate to the lack of detail about the experiments, both in the text, and to an even greater

degree in the figure legends. This makes the paper a difficult read. Extensive revision is required to

make it clearer.

Response: Thank you very much for the criticism. We have included all necessary information in

the revised manuscript.

Specific points:

Comment 1: Figure 1D: Is the effect of FK506 on Src expression significant? How many replicates

were performed to generate this result? This question also applies to Figures 1F and 2D and E.

Response 1: The suppressive effect of FK506 on Src gene expression after Tamoxifen treatment

was consistent in three biological replicates, but not statistically significant (Original Fig.1D is

moved to Fig.1B). This could be due to relatively big error bars. The same applies to original Fig.

1F (Fig. 1D in the current manuscript) except for gene Fgf5. We have added statistic information in

the legend and added the following statement to the manuscript text “Although FK506 reduced

Tamoxifen-induce differentiation gene expression, the effect was neither statistically significant nor

complete. This could be attributed to the following two reasons. One could be relatively large

variation in gene expression levels among three independent experiments. Second, Foxd3 depletion

might also lead to changes independent of NFAT signaling.”

Regarding to original Figs. 2D and 2E, we have omitted these figures in the revised manuscript

due to the mild role of Foxd3 overexpression in suppression of LIF-withdrawal induced ESC

differentiation, although the suppression for certain genes tested was statistically significant. This

could be due to the involvement with multiple factors in LIF-withdrawal induced differentiation, in

additional to NFAT activation. Thus, we do not claim that Foxd3 could inhibit LIF withdrawal

induced ESC differentiation. This would not affect the main scheme of the current study, which

focused on the interaction between Foxd3 and NFATc3.

Comment 2: Figure 2C: Although the morphology of the LIF- Foxd3 overexpressing cells might

well differ from the control cells, the differences between the control and Foxd3 overexpressing

plates are so subtle that is difficult to exclude trivial causes such as differences in plating density.

Response 2: For the same reason described above for Comment 1, original Fig. 2C has been

removed from the current manuscript.

Comment 3: Figure 3B: The legend for this figure does not give enough information. What does

WCL stand for?

Response 3: The “WCL” stands for whole cell lysate. This and other information have been

provided in the legend of current Fig. 2B.

Comment 4: Figure 3C: Again the legend does not provide enough information. There is no

indication of what the upper and lower panels are showing, or why two panels are shown.

Response 4: The original Fig. 3C is now changed to Fig. 2C in the revised manuscript. The upper

panel shows the western blot result with His antibody, suggesting that His-NFATc3 was pulled

down by GST-Foxd3, but not by GST alone. The lower panel is the western blot result using GST



antibody, showing the quantity and quality of GST control and GST-Foxd3 used in the GST pull

down assay. All necessary information is added in the legend.

Comment 5: Page 9: "Western blot analysis showed that both Forkhead domain (110 amino acids)

and C-terminus (251 amino acids) of Foxd3 were capable of binding with NFATc3 (Figure 3E)". It

seems from Figure 3E that only Foxd3 N fails to bind. It would be helpful to state this in the text. It

means that this deletion series was not particularly informative for defining the interacting regions

and this should be stated in the text. The deletions of NFATC were more informative as they define

the domain from 1-595 as the interacting region.

Response 5: Thanks for pointing this. We have added “only Foxd3-N domain failed to bind with

Foxd3” to our manuscript. The original Fig. 3E is shown in Fig. 2E of the current manuscript.

Comment 6: Page 10: "Importantly, the activation was dramatically suppressed by co-transfection

of Foxd3 with all NFATs but NFATc2 overexpressing cells (Figures 4A and B), revealing strong

inhibition of Foxd3 on the transcriptional activities of NFATc1, 3 and 4." This sentence does not

give a very clear explanation of the data in Figure 4A and B.

Response 6: In order to avoid the confusion and simplify the manuscript, we have removed the

original Figs. 4A and 4B from the current manuscript. Therefore, the role of NFATc1, c2 and c4 will

not be described, as NFATc3 is the focus of this study.

Comment 7: Page 10: The difference between NFAT: K3-luc and NFAT:AP1-luc is not explained

in the text or the legend to Figure 4, nor is it clear why they give different results. Giving references

for the constructs is not sufficient.

Response 7: Thanks for bringing this issue up. Both NFAT: K3-luc and NFAT:AP1-luc are widely

used NFAT reporter constructs. NFAT: K3-luc contains NFAT binding element only, while

NFAT:AP1-luc contains NFAT and AP1 binding elements. As our previous study (Cell Stem Cell,

2011) demonstrated that NFAT and AP1 interacts and functionally interdependent in ESCs, we have

omitted the data in original Fig. 4A and shown data of original Fig. 4B in Fig. 3A of current

manuscript, in order to avoid confusion and save the space. Thus, only NFAT:AP1 reporter is used

throughout the whole manuscript to make it a easy read.

Comment 8: Figure 4: Asterisks next to histogram bars presumably indicate levels of significance

(the reader does discover this when they reach the Methods) but no information about their meaning

is given in the figure legend.

Response 8: Sorry for missing the explanation. We have added the explanation for asterisk in the

legends.

Comment 9: Page 11: The EMSA experiment shown in Figure 4G and H is not well explained.

Presumably, the IL2 control probe contains a previously characterized Nfatc3 binding site.

However, this is not stated in the text or the figure legend. The binding patterns shown in Figure 4H

are complex and it is unclear how the bands are identified as containing either or both proteins in the

cells that have both co-transfected, without doing antibody supershifts.

Response 9: Firstly, we have removed the EMSA data for control IL2 probe in order to save space.

Second, we conducted EMSA experiment again with antibodies again Foxd3 and NFATc3,



respectively. The new result is showed in Figure 3D. We can see that the size of NFATc3+Foxd3

protein+DNA complex is larger than that of NFATc3+DNA or Foxd3+DNA complex. In addition,

we observed supershift when NFATc3 or Foxd3 antibody was included in the reaction mixture.

These data consolidate the notion that NFATc3 and Foxd3 form protein complexes and bind the

same sequence of the Src promoter region sequence.

Comment 10: Figure 4I: The meaning of the double asterisk is not specified in the figure legend.

Response 10: As suggested by the reviewer, we have added asterisk information in our legend.

Comment 11: Figure 5A: The labeling of the y axis should specify that expression is relative to

expression of Foxd3.

Response 11: Thanks for your correction. We have changed the label of the y axis as suggested in

current Figure S2A.

Comment 12: Figure 5C: Although western blotting of Tle4 shows a decrease to day 5 after LIF

withdrawal, the level of the protein then seems to increase at day 7. This is not referred to or

explained in the text.

Response 12: Thanks for pointing out this. We have added the description and provide the possible

explanation to this phenomenon. Tle4 is an important factor involving some vital signaling pathway

including Wnt pathway and other biological processes, in addition to its role in ESCs. Therefore, the

recovery of its protein level at day 7 of LIF withdrawal might reflect its involvement in lineage

commitment.

We truly appreciate the effort that all Reviewers and journal Editors have made and time they

spent on this study. The comments and criticisms are very constructive. I hope that our revision is

satisfactory and that the manuscript could be considered for publication in EMBO Reports.

2nd Editorial Decision 21 August 2014

Many thanks for the submission of your revised study to EMBO reports. The manuscript was sent back to the two of the original referees (#2 and 3) and while they overall appreciate that the study has been strengthened during revision, both still raise concerns that would need to be addressed before publication. While most of them are rather minor and include additional clarifications and controls, textual changes to better reflect the actual data, as well as better representation of existing data (e.g. comment 2 of referee 3), reviewer 2 still raises two major issues that are related to points brought up on the first version of the study: first, s/he feels that the hypothesis of a complex between Foxd3, NFATc3, and Tle4 should be strengthened by sequential Co-IPs and second, this referee also feels that ChIP-seq on more target genes should be performed. Given the overall interest the referees expressed in your study, I would like to give you the exceptional opportunity to revise the study a second time, with the understanding that the concerns of referee 2 and 3 concerns, especially the first one of referee 2, have to be addressed before the study can be published in EMBO reports. Formally, papers in EMBO reports have to be accepted within 6 months of the initial decision, which in your case would be October 3rd, 2014. We are, of course, still interested in publishing your study after this timepoint, but we would need to take the



novelty into account if your study can only be accepted after this date. Given that the final version of the manuscript might still need to be evaluated by one of the reviewers, please also outline briefly in a point-by-point manner how you have addressed the remaining referee concerns. If you submit your study at a later time we are, of course, still interested in publishing it if the remaining concerns have been adequately addressed, but we would need to assess its novelty afresh at this point. Please do let me know if you anticipate problems with this time-frame, as I am sure we can find a solution. REFEREE COMMENTS Referee #2: In the revised version Zhu et al have addressed most of the points raised. The authors now provide good evidence for protein-protein as well as some genetic interactions between Foxd3, NFATc3 and Tle4. Some indications are provided suggesting that Foxd3 represses Nfat induced activation and this repression is due to an interaction with Tle4. I appreciate that the authors have now focused on the molecular interaction of Foxd3 and Nfatc3 and toned down their claims of a specific involvement of the complex in cooperation with Tle4 in ES cell differentiation for which data was and is insufficient. Although the absence of data showing the functional relevance for the specific interactions shown limits the impact of this study, the presence of a complex of Foxd3, NFATc3 and Tle4 is interesting and might provide a basis for further functional studies. Some points remain to be addressed: Ad Specific point 1: Reciprocal IP has been performed in ES cells to show an interaction between Foxd3, Nfatc3 and Tle4. This data are of good quality. However, the clear proof for a complex containing all three proteins would be sequential CoIP. This could be done by eg. using the specific Tle4 AB and flag tagged Foxd3 or Nfatc3. Data is provided to show binding of Foxd3, Nfatc3 and Tle4 to the Src promoter. However, enrichment over a control region appears weak, despite appearing significant (max. 3 fold over control). To really judge if this data could reflect functional binding, more potential targets and more negative controls must be assayed. Ideally this should be done by ChIP-Seq. Ad Specific point 2: The authors have included siRNA experiments showing that Tle4 depletion can abrogate Fod3 mediated reduction of derepression after Nfatc3 overexpression. The authors also show that Tle4 KD results in increased expression of differentiation genes and a reduction in Oct4 and Rex1 expression. In case a genetic interaction exists as proposed by the authors Tle4 depletion should phenocopy the Foxd3 deficiency phenotype and induce ESC differentiation. Is this the case? Ad Specific point 3: Protein expression data during an ESC differentiation timecourse has been moved to Supplementary information and Nfatc3 levels have been removed. This is sensible as the increase in Nfatc3 levels at d7 is interesting but irrelevant for this study. Alternatively, and probably more informatively, it should be considered to include this data and show the expression timecourse only until d5. Ad Minor point 1: Here I was thinking mainly about AP reporter experiments as in co-transfection experiments of multiple interacting constructs stoichiometry will be important. Expression levels need to be controlled for. Ad Minor point 5: The limitation that partially differentiated cells are used for comparison is not discussed in the text.



Ad Minor point 6: "small fraction" is still misleading and very subjective. I would estimate 20-30% of Nfatc3 in the nucleus based on the data presented; this is a substantial portion of total levels. Additional new points: - The data presented does not justify the title, this should be changed to something more closely reflecting the results presented. -The "direct molecular and functional link between a pluripotency associated factor and an important ESC differentiation inducing pathway" suggested in the abstract is not really shown in the study. This needs toning down. -Fig1A: I suggest showing protein data here and qPCR in Supp information. -Fig2E: Why does the full length construct show multiple bands? -Fig3D: EMSA data is very weak, I cannot see any band indicating concommitant binding of Foxd3 and Nfatc3. For me, the interpretation here is that on the major proportion of the Src promoter only Foxd3 or Nfatc3 are bound. Cobinding appears rare. If sequential CoIP was successful this data could be removed. -Fig5F. Derepression upon Nfat OE seems minimal if compared to the amplitude change in early differentiation. Furthermore the effect of Aes in this experiment is very limited and only clear for Col3a1 expression. Do the authos have an explanation? Referee #3: I have read the revised manuscript by Zhu et al and I have looked at the replies to the comments by referee 1. The overall quality of the paper has improved substantially and I am satisfied that the authors have replied to my comments and those of reviewer 1. I have just two comments on the revised manuscript: Figure 2G: Pulldown band for NFATC3-N+M seems to b very weak despite there being a large amount of the protein in the extract. It would be a good idea for the authors to make some comment on this. Figure 4C: It is unclear why the blot in Figure 4C is quite so over exposed. The bands for input and Tle4 IP don't really look like bands at all but large blobs. Additional correspondence (author) 27 August 2014

Thanks for sending us your decision on our manuscript, in particular, for offering the exceptional

opportunity to revise the study a second time. My students and I have carefully read and discussed

your letter and reviewers’ comments on our revised manuscript. We appreciate the statement made

by the referee 3 that “the overall quality of the paper has improved substantially and I am satisfied

that the authors have replied to my comments and those of reviewer 1”. We are happy to address

two additional comments made by the referee 3 as well as the most comments made by referee 2.

However, I would like to discuss with you about the specific point 1 made by referee 2.



In the specific point 1, referee 2 requested two kinds of experiments.

About sequential Co-IP experiments. I agree with the referee, it is always better to have

additional data, and sequential CoIP is a one of the strategies used in protein complex verification.

However, it is not the only approach. The strategy was often used in the old days when advanced

strategies, such as ChIP-qPCR assay, were not widely utilized. Moreover, approach is very

antibody quality- and technique-demanding, and it is not a sensitive assay. The worst part of it is

that a negative result could not rule out the possibility of the presence of protein complexes. In the

initial comments, it was referee 2 who requested us to conduct reciprocal Co-IP assays in ESCs and

use specific antibodies. We have done all his requests to provide clear experimental data for the

presence of protein complexes containing Foxd3, NFATc3 and Tle4 in ESCs in the revised

manuscript. Even Referee 2 admitted, “Reciprocal IP has been performed in ES cells to show an

interaction between Foxd3, Nfatc3 and Tle4. This data are of good quality.”

In addition, in order to demonstrate the involvement of Tle4 in the function of Foxd3, we have

provided the multiple lines of evidence: 1) The distinct expression pattern of Tle4; ii) Tle4 interacts

with Foxd3 both in ESCs and in vitro; iii) Tle4 binds to the Src promoter, similarly to Foxd3; iv)

Functionally, Tle4 enhances Foxd3 to suppress NFAT transcriptional activity. In contrast, inclusion

of dominant-negative form of Tle4 or siRNA of Tle4 significantly reduces the suppression of Foxd3

on the NFAT activity. These data clearly show the specific role of Tle4 in Foxd3-mediated

suppression of NFAT activities.

Therefore, conducting the sequential Co-IP experiment could not add any more value to the

current study. Instead, this time-consuming and low sensitive assay will delay the publication,

risking the loss of the novelty. Therefore, I hope that you could reconsider this request.

About more target genes. In the initial comments, referee 2 asked us to conduct

ChIP assay, providing experimental evidence that all three proteins are recruited to a target

gene. We have done so in the revised manuscript. However, referee 2 criticized that

“enrichment over a control region appears weak, despite appearing significant (max. 3 fold

over control).” and requested us to conduct ChIP assays on more potential targets and more

negative controls in order to reflect functional binding. Firstly, Referee 2 asked us to carry

out additional studies as s/he concerned that 3-fold enrichment on the Src promoter is

weak. We have searched published data and found a few studies having NFAT or Foxd3

ChIP results on specific genomic locus （see figures below）. In these studies, weak

enrichment for either Foxd3 or NFAT was reported. Therefore, it seems reasonable that we

obtained consistent and statistically significant enrichment on the endogenous Src promoter

in ESCs for all three proteins.



Plank, J.L., et al., Transcriptional targets of Foxd3 in murine ES cells. Stem Cell

Res. 12(1): p. 233-40.

Lin, S.S., et al., Cav3.2 T-type calcium channel is required for the NFAT-

dependent Sox9 expression in tracheal cartilage. Proc Natl Acad Sci U S A. 111(19): p.

E1990-8.

Tardif, G., et al., NFAT3 and TGF-beta/SMAD3 regulate the expression of miR-

140 in osteoarthritis. Arthritis Res Ther. 15(6): p. R197.

Moreover, there is a technical problem to conduct ChIP assay for NFATc3 and Foxd3. Both

proteins have low DNA binding affinities, which explains their need of cofactors to facilitate DNA



binding and the absence of published genome-wide ChIP-on-ChIP or ChIP-seq study for them to

date, although they are very important factors in a variety of biological processes. We tried very

hard to conduct ChIP-seq experiments using antibodies against Foxd3 and NFATc3, respectively,

before we submitted this manuscript. However, we failed to obtain enough immunoprecipitated

genomic DNAs with either antibodies, which could be due to the known nature of these proteins to

bind DNA with a low affinity. With this technical difficulty, we conducted the microarray assays to

find potential target genes of Foxd3 and NFATc3. Unfortunately, the most candidates we identified

in this study are new targets. Thus, it is difficult to conduct ChIP-seq study without knowing the

binding region for the transcription factor on the target genes. Second, I need point out that Src is

the critical downstream target of NFATc3 to induce ESC differentiation, which has been clearly

demonstrated in our previous paper published in Cell Stem Cell. As pointed out by referee 2, “Data

is provided to show binding of Foxd3, Nfatc3 and Tle4 to the Src promoter”. In my opinion, study

on a key target gene should have significant impact for understanding how a pathway functions.

Third, if referee2 would like to have additional data to reflect functional binding, I have a

suggestion: qRT-PCR analysis of Src mRNA levels under the following conditions: i)

overexpression of NFATc3 in ESCs to induce Src expression; ii) co-overexpression of Foxd3 and

NFATc3 to test whether Foxd3 could prevent the induction of Src by NFATc3; iii) after knock

down of Tle4 or overexpression of dominant-negative Tle4, test whether Foxd3 could still suppress

the induction of Src by NFATc3. This experiment should be able to strengthen the notion that Tle4

is functionally required for Foxd3 to block induction of Src by NFATc3 in ESCs. In the future, with

the availability of better antibodies and ChIP-seq technology, we would like to continue working on

the whole genome ChIP-seq study of NFATc3 and Foxd3 in ESCs, which will further our

understanding how the balance of Foxd3 and NFAT controls the ESC fates at a whole genome scale.

Taken together, with already provided data of protein complexes containing endogenous

Foxd3, NFATc3, Tle4, and, more importantly, the multiple lines of evidence for the important role

of Tle4 in the Foxd3 function, as well as the previously proven critical role of Src in the function of

NFATc3 in ESCs, I wish that you would agree with me, and could reconsider the necessity to

experimentally address the first specific point of referee 2.

Thank you very much for your effort on this manuscript and look forward to hearing from

you soon.

Additional correspondence (editor) 04 September 2014

Many thanks for your patience while I was discussing your response with referee 2. I have now heard back from him and paste his/her answer below for your information: Referee feedback: "I do agree with the authors that sequential CoIP is tricky and of course a negative result does not mean that such a complex does not exist, given the intrinsic limitations of this experiment. However in case it worked it would be good proof of a trimeric complex - this is the reason why I have suggested it. I concur that reciprocal CoIP is good, and I would say sufficient, indication. Especially if corroborated by experiments showing genetic interaction as suggested by the authors themselves.



To your question: Sequential ChIP would have the same limitations as sequential CoIP, with the added difficulty that also the interaction with DNA must be maintained throughout the procedure. Concerning binding to the Src promoter, the question that the authors need to address is whether the binding of Foxd3, Nfat and Tle4 is specific (as opposed to low affinity background binding of transcription factors to open chromatin). As attempts to generate ChIP Seq data were not successful, I would suggest to analyse binding to more genomic loci to assess binding. One non-genic region is not an ideal negative control for this. To support their conclusion, it needs to be clearly shown that binding to the Src promoter is above background. A much better control would be promoters that are not bound by these factors. In the absence of ChipSeq data, these non-targets cannot be predicted but genes that do not respond to perturbation of Foxd3 or Nfat could be used. Some of the known Foxd3 targets in ES cells should be used as additional positive control to show that binding to Src shows a similar amplitude. This should be relatively simple experiments that can be performed using existing ChIP material." In essence, this reviewer agrees that sequential Co-IPs or sequential ChIP would be difficult and a negative result would not be very meaningful. However, s/he still feels that looking at more genomic loci (including both positive and negative controls) is necessary and I would like to kindly ask you to perform the suggested experiments, as without them, this referee is not convinced about the strength of the conclusions. Having said this, the requested experiments do not seem to be too difficult and I am therefore optimistic that you can perform them in a rather timely manner. Additional correspondence (author) 05 September 2014

Thank you very much for your reply and effort on this manuscript. I appreciate your and referee 2's understanding of our arguments. We will try our best to perform the suggested experiments and submit the revised manuscript as soon as possible. 2nd Revision - authors' response 30 September 2014

Reviewer 2


The authors now provide good evidence for protein-protein as well as some genetic interactions

between Foxd3, NFATc3 and Tle4. Some indications are provided suggesting that Foxd3 represses

Nfat induced activation and this repression is due to an interaction with Tle4. I appreciate that the

authors have now focused on the molecular interaction of Foxd3 and Nfatc3 and toned down their

claims of a specific involvement of the complex in cooperation with Tle4 in ES cell differentiation

for which data was and is insufficient.

Although the absence of data showing the functional relevance for the specific interactions shown

limits the impact of this study, the presence of a complex of Foxd3, NFATc3 and Tle4 is interesting

and might provide a basis for further functional studies.

Response: Thanks for your effect on our manuscript.

Specific points after discussion with the Editor and Reviewer:



Comment: I do agree with the authors that sequential CoIP is tricky and of course a negative result

does not mean that such a complex does not exist, given the intrinsic limitations of this experiment.

However in case it worked it would be good proof of a trimeric complex - this is the reason why I

have suggested it. I concur that reciprocal CoIP is good, and I would say sufficient, indication.

Especially if corroborated by experiments showing genetic interaction as suggested by the authors

themselves. Sequential ChIP would have the same limitations as sequential CoIP, with the added

difficulty that also the interaction with DNA must be maintained throughout the procedure.

Concerning binding to the Src promoter, the question that the authors need to address is

whether the binding of Foxd3, Nfat and Tle4 is specific (as opposed to low affinity background

binding of transcription factors to open chromatin). As attempts to generate ChIP Seq data were not

successful, I would suggest to analyse binding to more genomic loci to assess binding. One non-

genic region is not an ideal negative control for this. To support their conclusion, it needs to be

clearly shown that binding to the Src promoter is above background. A much better control would

be promoters that are not bound by these factors. In the absence of ChipSeq data, these non-targets

cannot be predicted but genes that do not respond to perturbation of Foxd3 or Nfat could be used.

Some of the known Foxd3 targets in ES cells should be used as additional positive control to show

that binding to Src shows a similar amplitude. This should be relatively simple experiments that can

be performed using existing ChIP material."

Response 1: As suggested by the reviewer, we conducted qPCR experiments using existing ChIP

materials obtained from previous Foxd3, NFATc3, and Tle4 ChIP assays. In addition to Src, we

examined the potential Foxd3 targets reported by Plank, J.L., et al., (Transcriptional targets of

Foxd3 in murine ES cells. Stem Cell Res. 12(1): p. 233-40.), including Sox15, Sox4, Foxb, Safb and

l5preB genes. We found that Foxd3, NFATc3 and Tle4 all could bind at the promoter of Sox15, at a

similar amplitude to their binding at the Src promoter. Interestingly, Foxd3 could also bind at the

promoter of l5preB gene, whereas NFATc3 and Tle4 did not. None of the factors bind at the

promoters of Sox4, Foxb and Safb genes. Thus, we show the specific association of Foxd3, NFATc3

and Tle4 with the promoter of Sox15 gene, in addition to the Src promoter in the revised manuscript

(Supplementary Figure S2). For additional negative control, the HPRT coding sequence served as a

negative control, as none of our three factors was enriched in this region. The ChIP qPCR data for

Src with two negative controls (the coding region of HPRT and cytogenetic location 12 p13.3) are

shown in the Figure 3E and Supplementary Figure S2, respectively, of the revised manuscript. I

hope that the reviewer is convinced now that the binding of these factors to Src is specific.

Additional points

Comment 2. The authors have included siRNA experiments showing that Tle4 depletion can

abrogate Foxd3 mediated reduction of derepression after Nfatc3 overexpression. The authors also

show that Tle4 KD results in increased expression of differentiation genes and a reduction in Oct4

and Rex1 expression. In case a genetic interaction exists as proposed by the authors Tle4 depletion

should phenocopy the Foxd3 deficiency phenotype and induce ESC differentiation. Is this the case?



Response 2: We observed that Tle4 depletion induced less robust ESC differentiation than Foxd3

did. Three possibilities exist. One could be the compensatory effect of the other members of the Tle

family. The Tle family has several members as showed in Supplementary Figure S3 and some of

them might substitute Tle4 and execute similar functions when Tle4 was depleted; the second reason

could be that the knock down was not complete. The third reason is that Foxd3 may have functions

independent of Tle4 to maintain ESCs self-renewal. Thus, silencing of Tle4 phenocopied Foxd3

deficiency partially.

Comment 3. Protein expression data during an ESC differentiation time course has been moved to

Supplementary information and Nfatc3 levels have been removed. This is sensible as the increase in

Nfatc3 levels at d7 is interesting but irrelevant for this study. Alternatively, and probably more

informatively, it should be considered to include this data and show the expression time course only

until d5.

Response 3: Thanks for your advice. We have added the NFATc3 protein levels back to the figure

and show all data until day 5 in the revised Supplementary Figure S3.

Ad minor points:

Comment 1. Here I was thinking mainly about AP reporter experiments as in co-transfection

experiments of multiple interacting constructs stoichiometry will be important. Expression levels

need to be controlled for.

Response 1: We agree with you. We have precisely controlled the DNA quantity for each construct

in co-transfection experiments. The quantity of each transfected construct was same and the whole

amount of constructs in different groups was also maintained equivalent.

Comment 2. The limitation that partially differentiated cells are used for comparison is not

discussed in the text.

Response 2: We are not clear about what the reviewer meant here. Could you please point out

which particular data in our manuscript used partially differentiated cells?

Comment 3. "small fraction" is still misleading and very subjective. I would estimate 20-30% of

Nfatc3 in the nucleus based on the data presented; this is a substantial portion of total levels.

Response 3: Thanks for bringing this issue up. We have modified relevant sentences in our

manuscript.

Ad new points:

Comment 1. The data presented does not justify the title, this should be changed to something more

closely reflecting the results presented.



Response 1: We have changed our title to “Foxd3 suppresses NFAT-mediated differentiation to

maintain self-renewal of embryonic stem cells.”

Comment 2. The "direct molecular and functional link between a pluripotency associated factor and

an important ESC differentiation inducing pathway" suggested in the abstract is not really shown in

the study. This needs toning down.

Response 2: We have removed “direct” from the sentence.

Comment 3. Fig1A: I suggest showing protein data here and qPCR in Supp information.

Response 3: We have done as the reviewer suggested.

Comment 4. Fig2E: Why does the full length construct show multiple bands?

Response 4: Those bands under the main band are degraded protein bands and this phenomenon can

also be seen in Figure 4B.

Comment 5. EMSA data is very weak, I cannot see any band indicating concommitant binding of

Foxd3 and Nfatc3. For me, the interpretation here is that on the major proportion of the Src

promoter only Foxd3 or Nfatc3 are bound. Cobinding appears rare. If sequential CoIP was

successful this data could be removed.

Response 5: A slower migrating band (lanes 4,6 and 7 in right panel) in Figure 3D above Foxd3 or

NFATc3 alone band (Lane 2 and Lane 3) indicates the concomitant binding of Foxd3 and NFATc3

with Src probe. Our EMSA data indicate that Foxd3 and NFATc3 bind on the Src promoter

together.

Comment 6. Fig5F. Derepression upon Nfat OE seems minimal if compared to the amplitude

change in early differentiation. Furthermore the effect of Aes in this experiment is very limited and

only clear for Col3a1 expression. Do the authos have an explanation?

Response 6: The experiments were conducted in ESCs, where Tle4 is highly expressed. The limited

derepression of NFAT for some genes could be due to a relatively low efficiency of transfected Aes

to interfere with the activity of endogenous Tle4.

Reviewer 3


I have read the revised manuscript by Zhu et al and I have looked at the replies to the comments by

referee 1.

The overall quality of the paper has improved substantially and I am satisfied that the authors have

replied to my comments and those of reviewer 1.



Response: Thank you very much for your comments.

Comment 1. Figure 2G: Pulldown band for NFATC3-N+M seems to b very weak despite there

being a large amount of the protein in the extract. It would be a good idea for the authors to make

some comment on this.

Response 1: This is an interesting question. It is likely that the M domain might impair the strength

of interaction between NFATC3-N+M and Foxd3. Or NFATC3-N+M protein might be easily

degraded in the process of IP, wash and elution, causing a weak band in the western blot.

Comment 2. It is unclear why the blot in Figure 4C is quite so over exposed. The bands for input

and Tle4 IP don't really look like bands at all but large blobs.

Response 2: This phenomenon could result from the fact that NFAT proteins have multiple

phosphorylation sites, leading to a smear- or blob- like band in the western blot. The experiments

have been repeated several times, consistent results are obtained.

3rd Editorial Decision 06 October 2014

Thanks a lot for submitting the final version of your study to our editorial office. I am now happy to accept it for publication here and thank you for your contribution and cooperation during the review process.

Documents

EMBO reportsembor.embopress.org/.../embr.201438643.reviewer-comments.pdfThe same reviewer also asks why the presence or absence of Foxd3 does not seem to influence the cells' response