A phosphorylated transcription factor regulates sterol biosynthesis in Fusarium graminearum

Sterol biosynthesis is controlled by transcription factor SREBP in many eukaryotes. Here, we show that SREBP orthologs are not involved in the regulation of sterol biosynthesis in Fusarium graminearum, a fungal pathogen of cereal crops worldwide. Instead, sterol production is controlled in this organism by a different transcription factor, FgSR, that forms a homodimer and binds to a 16-bp cis-element of its target gene promoters containing two conserved CGAA repeat sequences. FgSR is phosphorylated by the MAP kinase FgHog1, and the phosphorylated FgSR interacts with the chromatin remodeling complex SWI/SNF at the target genes, leading to enhanced transcription. Interestingly, FgSR orthologs exist only in Sordariomycetes and Leotiomycetes fungi. Additionally, FgSR controls virulence mainly via modulating deoxynivalenol biosynthesis and responses to phytoalexin.

The manuscript by Liu et al describes the identification and detailed characterisation of a here-to unrecognised pathway that regulates sterol biosynthesis. After providing evidence that Fusarium homologues of known sterol regulatory pathways from other organisms did not have this role in F. graminearum, the authors undertook an elegant loss-of-function genetic screen to find transcription factors that contribute to basal tolerance to sterol biosynthesis inhibitors. The authors went on to characterise how the protein functions and the signalling and chromatin remodelling complexes involved in its regulation of sterol biosynthesis. Although the bulk of the paper (and title) focuses on the role of the identified transcription factor and signalling pathways in sterol biosynthesis regulation the work provides tantalising evidence that the transcription factor has a broader role in regulation of transcription associated with the pathogenesis process. For this reason the title could be made slightly broader.
I have relatively minor comments Line 136: "restore the sensitivity of yeast Upc2 and Ecm22 mutants to azole compounds" could be interpreted that the Upc2 and Ecm22 mutants are actually resistant to these compounds which I suspect is not the intended meaning. Perhaps "restore" should be replaced with "complement". Likewise in the supplementary figure 2 legend "restore sensitivity" implies the mutants are actually resistant. Line 200: The link between the Y2H result and the Ssk2 mutant result does not read very well. Perhaps "Among them, we observed that the mutant of Fgssk2 also showed increased sensitivity to…" could be changed to "One interacting protein was the FgSsk2 MAPKK kinase and mutants in the encoding gene showed increase sensitivity to…" Line 206: In Figure 3e it appears the difference between expression of CyP51A under tebuconazole treatment is only slightly reduced in the ssk2 mutant compared to Ph1. Should this sentence on line 206 read "to 89.2%" The analysis of deoxynivalenol production in the FgSR mutant seemed an obvious course of investigation given the in plant a phenotype of the mutant not spreading from the inoculation point in wheat and the shared upstream pathways for sterol and deoxynivalenol biosynthesis which were shown to be regulated by FgSR including via direct binding to the promoters. Likewise the results of the analysis of DNA damage stress also seemed logical for inclusion given the ChIP-seq data supported enrichment of FgSR at the promoters of some of the genes involved in the response to these stresses. It is less clear why the sensitivity to BOA was tested and how FgSR regulates the expression of the detoxification genes given these were not identified in the ChIP-seq data. Whilst the data are interesting around BOA a better justification of why these genes were investigated and an explanation of how FgSR might regulate their expression should be included. Figure 2: promoter is spelled incorrectly in the vertical axes of part b. It might also be useful to align the order of the different regions assayed in the promoters with the bars as they are currently in opposite orientation in the promoter diagram to the graphs below. In figure 2a, delta-erg4 induced ergosterol deficiency clearly induces cyp51a expression. However, in Supp Fig 5, cyp51a does not seem to be induced in the delta-erg4::FgSR-GFP strain which should functionally equivalent to the delta-erg4 strain used in figure 2a. Can the reason for this discrepancy be explained? Line 408: vegetative not vegetable Overall the work is a very convincing characterisation of the novel sterol regulation system in Fusarium and related ascomycete fungi. I suspect this paper will be quite influential in the field as the findings are highly novel. The statistical analyses have been performed with quantitative data but the tests used to identify groups that differ (defined on bar charts by letters) have not been explained in the figure legends or the methods. Having said that the data is still very convincing so this is just a technicality. Donald Gardiner Reviewer #2 (Remarks to the Author): Fusarium graminearum is an important plant pathogenic fungus. In this study, the authors showed that six orthologs of known regulators of ergasterol biosynthesis were dispensable. The FgSR mutant was then found to have increased sensitivity to tebuconazole. Enrichment of FgSR at the promoter of Cyp51A/B was observed. The authors then screened yeast two-hybrid library and identified FgSsk2 and Swp73 as two of the FgSR-interacting proteins. Five putative phosphorylation sites were identified and characterized in FgSR. Mutations at these five sites affected the interaction of FgSR with Swp73. The authors then showed that the SNF complex may be recruited to the Cyp51A promoter via FgSR. They also found that FgSR was enriched in other genes related to ergasterol synthesis by ChIP-seq. Putative FgSR binding sites were identified. Interestingly, FgSR proteins were only found in Sordariomycets and Leotiomycetes. Based on these data, the authors proposed that FgSR recruits the chromatin remodeling factors to regulate fungal sterol biosynthesis.
To this reviewer, some of the conclusions were not well-supported by data presented in this manuscript.
Major concerns: 1. The subcellular localization and enrichment of FgSR on the CYP51 promoters were not affected by tebuconazole treatment. One major experimental evidence to show that phosphorylation of FgSR recruited the SNF complex to these promoters is that 5A mutation affected the interaction between FgSR and Swp73. Because this results is critical the conclusions of this manuscript, it is important to prove the interaction between FgSR and Swp73 and the effect of 5A mutations on their interactions by co-IP assays and BiFC assays. A simple yeast two-hybrid assays is not sufficient. Although FgSsk2 was dispensable for tebuconaole induced enrichment, Swp73 and Arp9 were. Is that somewhat contradictory? According to the descriptions, phosphorylation of FgSR by FgSssk2 and other kinase is important for recruiting the SNF complex to the FgSR bound promoters.
2. I have serious doubt about the interaction of FgSsk2 with FgSR and the direct role of FgSsk2 in the phosphorylation of FgSR. The authors concluded that FgSsk2 co-localized with FgSR to the nucleus based on Fig. 3d. However, Figure 3d clearly showed that FgSsk2 localized to the cytoplasm and FgSR localized to the nucleus under normal growth conditions. (Without activation by stresses, FgSsk2 is likely not in the nucleus) The authors used the same culture conditions for co-IP assays. It is questionable how can the authors detected such a strong interaction between FgSsk2 and FgSR by co-IP assays even though they differed so significantly in subcellular localization. Will stress or fungicide treatment affect their interactions? The authors should use BiFC assays to show their co-localization. (BiFC was used to show the dimerization of FgSR) Line 226-228. This conclusion about phosphorylation of FgSR by FgSsk2 is inappropriate. The authors did not present any direct evidence to show that FgSsk2 phosphorylates FgSR. The effect of FgSsk2 deletion on FgSR phosphorylation could be indirect because the normal function of FgSsk2 is in the Ssk2-Pbs2-Hog1 pathway. Is the activated or unactivated FgSsk2 that phosphorylates FgSR? Does the role of FgSsk2 in the activation of FgSR involves Pbs2 and Hog1?. The authors could tested this hypothesis by doing similar assays with the downstream MAPKK or MAPK mutants.
3. Regarding the putative cis element identified and characterized in this study, to this reviewer, the authors did not present convincing or direct evidence to show it is the binding site of FgSR. The in vivo effects of their deletion or mutation in the Cyp51A promoter may be caused by other factors that bind to or recognize these elements (not by FgSR per se). The authors should conduct EMSA assays to show the direct binding of FgSR to these cis elements and mutational effects. Also, why only five of these genes, not majority of the sterol synthesis related genes, have these elements? (I don't know how reliable the ChIP-seq data generated by the authors) 4. I am also concerned with the five phosphorylation sites described in the manuscript. It was not clearly described how the authors identified these five sites. Is there any way to predict phosphorylation sites of MAPKKK on non-MAPKK targets? Also, because of the differences between FgSsk2 deletion and 5A mutations, the authors should do mutations at individual sites to identify the FgSsk2 phosphorylation sites. (Deletion of FgSSK2 had no effect on tebuconazole induced expression of Cyp51A but five phosphorylation sites were important for that). Another experiment is to generate the 5D mutations to mimic activation. Regarding Figure 3h, if the authors could identified the FgSR phosphorylation sites by phosphoproteomics analysis, they should be able to use the same method to identify changes in the phosphorylation sites of FgSR in the FgSsk2 deletion mutant. 5. Fusarium graminearum has two SREBP genes and four Upc2 orthologs. It is possible that some of these genes may have overlapping functions. Based on phenotypical characterizations of the single mutants of these genes, the authors could not concluded that these genes are not important for regulating sterol synthesis. In fact, the FgSR deletion mutant was only reduced approximately 50% in growth rate under normal growth conditions, suggesting FgSR is not an important regulator of sterol synthesis because of the essential functions of ergasterol. This transcription factor FgSR has been characterized in an earlier study (Son et al., 2011. PLoS Pathogens). Deletion of this transcription factor had pleiotropic defects, including increased sensitivity to oxidative stress and reduced virulence. The authors identified approximately 50 putative FgSR-interacting proteins in yeast two-hybrid assays. FgSsk2 and Swp71 were just two of them. Somehow, both of them are related to sterol synthesis. ???
Minor points: Deletion of ML likely affects protein folding, which may affect many things other than dimerization. Which domain is responsible for DNA binding or binding to the cis elements identified in this study?

NCOMMS-18-00653
This manuscript deals with the mayor plant pathogen <i>Fusarium graminearum</i> and the discovery of a novel regulator of sterol biosynthesis. From mammals to yeast, sterol biosynthesis is regulated by a rather conserved transcription factor called Upc2/SREBP. In this manuscript, a so far not described description factor FgSR is described that provides a novel mechanism for regulating sterol biosynthesis. Most importantly, this regulation differs very much from the regulation by the conserved SREBP/Upc2 orthologs. For example, FgSR is located in the nucleus independent of ergosterol starvation. Further FgSR protein levels do not alter by sterol treatment, and the protein level bound to the promoters of target genes is not increased under steroldeprived conditions. Finally, FgSR phosphorylation regulates its transcriptional activity via recruiting the SWI/SNF complex, which is responsible for chromatin remodeling. The manuscript is highly complex and provides a huge wealth of information. For example, the data are presented in nine figures that overall display 47 subfigures. However, the overall finding of this manuscript is restricted to a few ascomycetous groups, namely the Sordariomycetes and the Leotiomycetes. With this, a rather specific regulatory mechanism is described. Overall, it is a very ambitious paper and I have some points, which should be clarified before publication.
1. Line 189ff.: The authors claim that phosphorylated FgSR recruites the SWI/SNF complex. A mayor finding in this chapter is the interaction of FgSR with FgSSK2, encoding a mayor MMAPKK kinase. The authors predict that FgSR contains five predicted amino acid sites (line 217), however I do not see the real data for this. For example, Fig. 3h is a general phosphorylation experiment, and Fig. S6 only shows two phosphorylation sites as evidence by LC-MS/MS. Further in this chapter, the authors found an interaction of FgSwp73, a component of the SWI/SNF complex. This is solely based on a Y2H approach. They further provide other data, for example enrichment of FgArp9 at the FGCYP51A promoter or the enrichment of H3 at the promoter of FgCYP51A. All these are to my mind rather indirect evidences for the statements made by the authors that the SWI/SNF complex is recruited by FgSR and is involved in the transcriptional regulation of sterol biosynthesis genes (line 247-251). Fig. 2 (line 832): PH-1 transformed with GFP served as a negative control. Maybe I do not understand this comment, but I do not see it in the figures. 3. Fig. 6A: The authors show the ChIPseq data for several of the ergosterol biosynthesis genes. However, three are missing. Explain why no data were received. 4. Fig. 7: The MEME analysis is shown for five genes. I do not understand why only five target genes were selected, although much more target genes were identified in the corresponding chapter (line 302-309). 5. Fig. 7: There are four promoter derivatives shown (P1-P4). However, P4 was not checked <i>in vivo</i> ( Fig. 7C, D, E), this should be briefly explained. Further, some of the capital letters (A, B) are not explained in the legend. 6. The discussion contains many repetitions of the results, for example in line 494 or 482. 7. Further, in line 409 the authors mention that the binding-cis element for FgSR also exists in the promoters of other ascomycetes fungi. Has this been shown somewhere? Can they provide some references?

Author responses:
We thank editor and all reviewers for their greatly comments and suggestions, which have helped us to substantially improve the manuscript. Here are point-by-point to the comments raised by editor and the reviewers Reviewer #1 (Remarks to the Author): The manuscript by Liu et al describes the identification and detailed characterisation of a here-to unrecognised pathway that regulates sterol biosynthesis. After providing evidence that Fusarium homologues of known sterol regulatory pathways from other organisms did not have this role in F. graminearum, the authors undertook an elegant loss-of-function genetic screen to find transcription factors that contribute to basal tolerance to sterol biosynthesis inhibitors. The authors went on to characterise how the protein functions and the signalling and chromatin remodelling complexes involved in its regulation of sterol biosynthesis.
We thank the reviewer for the positive comments.
1. Although the bulk of the paper (and title) focuses on the role of the identified transcription factor and signalling pathways in sterol biosynthesis regulation the work provides tantalising evidence that the transcription factor has a broader role in regulation of transcription associated with the pathogenesis process. For this reason the title could be made slightly broader.
Re: Thank you for your suggestion. It is a good point. We added "pathogenicity" in the title.
2. Line 136: "restore the sensitivity of yeast Upc2 and Ecm22 mutants to azole compounds" could be interpreted that the Upc2 and Ecm22 mutants are actually resistant to these compounds which I suspect is not the intended meaning. Perhaps "restore" should be replaced with "complement". Likewise in the supplementary figure 2 legend "restore sensitivity" implies the mutants are actually resistant.
3. Line 200: The link between the Y2H result and the Ssk2 mutant result does not read very well. Perhaps "Among them, we observed that the mutant of Fgssk2 also showed increased sensitivity to…" could be changed to "One interacting protein was the FgSsk2 MAPKK kinase and mutants in the encoding gene showed increase sensitivity to…" Re: Thank you for your suggestion. We revised this sentence to make this part go smoothly (lines 205-207). 5. It is less clear why the sensitivity to BOA was tested and how FgSR regulates the expression of the detoxification genes given these were not identified in the ChIP-seq data. Whilst the data are interesting around BOA a better justification of why these genes were investigated and an explanation of how FgSR might regulate their expression should be included.
Re: Our study found that deletion of FgSR reduced approximately 50% growth, but caused a dramatically decrease in virulence (Fig. 9a,b). We were therefore interested in exploring mechanism for the reduced virulence in FgSR. Kettle and his collogues have reported that degradation of the benzoxazolinone class of phytoalexins is important for virulence of Fusarium. Hence, we tested the sensitivity of FgSR to multiple phytolexins including gramine, 2-benzoxazalinone and tryptamine, and found that FgSR was specifically hypersensitive to BOA, but not to other phytolexins (Image A below).
The transcriptions of three genes (FgFDB1,-2, and -3) that were reported to be involved in BOA detoxification were down-regulated in FgSR, but these three genes were not identified in the ChIP-Seq assay, indicating that FgSR might indirectly regulate expression of the three genes. Until now, the knowledge about the mechanism of BOA action and detoxification is very limited. To explore the potential target of BOA, we deleted 22 of the FgSR target genes identified by ChIP-Seq assay that are potential related to stress responses, and determined sensitivity of these mutants to BOA. But none of the mutants displayed changed sensitivity to BOA compared with the wild type (Image B below). Thus, molecular mechanisms of FgSR in regulating transcription of FgFDB1,-2,-3 remain unclear.
6. Figure 2: promoter is spelled incorrectly in the vertical axes of part b. It might also be useful to align the order of the different regions assayed in the promoters with the bars as they are currently in opposite orientation in the promoter diagram to the graphs below.
Re: Thank you for your reminder. We redrew the promoter diagram according the orders of ChIP-qPCR assays (Fig. 2b). And we corrected "promoter" in 9. The statistical analyses have been performed with quantitative data but the tests used to identify groups that differ (defined on bar charts by letters) have not been explained in the figure legends or the methods. Having said that the data is still very convincing so this is just a technicality.
Re: We added the information for statistical analysis in corresponding figure legends.
Reviewer #2 (Remarks to the Author): Fusarium graminearum is an important plant pathogenic fungus. In this study, the authors showed that six orthologs of known regulators of ergasterol biosynthesis were dispensable. The FgSR mutant was then found to have increased sensitivity to tebuconazole. Enrichment of FgSR at the promoter of Cyp51A/B was observed. The authors then screened yeast two-hybrid library and identified FgSsk2 and Swp73 as two of the FgSR-interacting proteins. Five putative phosphorylation sites were identified and characterized in FgSR. Mutations at these five sites affected the interaction of FgSR with Swp73. The authors then showed that the SNF complex may be recruited to the Cyp51A promoter via FgSR. They also found that FgSR was enriched in other genes related to ergasterol synthesis by ChIP-seq. Putative FgSR binding sites were identified. Interestingly, FgSR proteins were only found in Sordariomycets and Leotiomycetes. Based on these data, the authors proposed that FgSR recruits the chromatin remodeling factors to regulate fungal sterol biosynthesis.
To this reviewer, some of the conclusions were not well-supported by data presented in this manuscript.
Major concerns: 1. The subcellular localization and enrichment of FgSR on the CYP51 promoters were not affected by tebuconazole treatment. One major experimental evidence to show that phosphorylation of FgSR recruited the SNF complex to these promoters is that 5A mutation affected the interaction between FgSR and Swp73. Because this results is critical the conclusions of this manuscript, it is important to prove the interaction between FgSR and Swp73 and the effect of 5A mutations on their interactions by co-IP assays and BiFC assays. A simple yeast two-hybrid assays is not sufficient.
Although FgSsk2 was dispensable for tebuconaole induced enrichment, Swp73 and Arp9 were. Is that somewhat contradictory? According to the descriptions, phosphorylation of FgSR by FgSssk2 and other kinase is important for recruiting the SNF complex to the FgSR bound promoters.
Re: We conducted the Co-IP assays to detect the interaction of FgSR and FgSwp73.
As shown in Fig. 4b, FgSR interacted with FgSwp73, and this interaction was enhanced under tebuconazole treatment (lines 254, 255). As you suggested, we performed BiFC assay to verify the interaction of FgSwp73 and FgSR, but did not observed the interaction signal. Previous studies have shown that BiFC assays may give false negatives, which may result from the conformation or activity of the tagged proteins are changed by tags (Kerppola, 2008;Lee et al, 2011). In this study, FgSsk2 phosphorylates FgSR. We therefore further tested the interaction strength of FgSsk2 and FgSR by Co-IP assays. As shown in Fig. 3c, this interaction was enhanced by tebuconazole treatment (lines 211, 212). Consistently, the high levels of the phosphorylated isoforms of FgSR in the wild type by tebuconazole treatment were 6 reduced in FgSSK2 mutant (Fig. 3f). In addition, we found that phosphorylation of FgSR was critical for its interaction with FgSwp73, and the enrichment of FgSwp73 and FgArp9 at the promoter of FgCYP51A (Fig. 4). Lee Re: Thank you for insightful suggestions. The IP samples of Co-IP assays were enriched by GFP-Trap agarose beads. Thus, the interaction signal could be enlarged in Co-IP assay. Regarding the subcellular localization of FgSsk2, because the RFP fluorescence signal of PH-1::FgSsk2-RFP was weak, and the native untagged FgSsk2 may affect localization of the FgSsk2-RFP that was introduced into the wild type (in the original Fig. 3d ), we knocked out the native FgSSK2 in this revision, and found that the FgSsk2-RFP obviously localized to both the nucleus and cytosol in FgSsk2::FgSsk2-RFP strain (Fig. 3d). Again, we conducted BiFC, and did not observe the interaction signal. Therefore, we further performed immunofluorescence assays, and found that both FgSsk2 and FgSR are clearly present in the nucleus (Fig.   3e). Moreover, the Co-IP assay showed that FgSR-FgSsk2 interaction was enhanced by tebuconazole treatment (Fig. 3c). These results indicate that FgSsk2 is able to interact with FgSR in the nucleus.
As you suggested, we tested the interaction of FgSR with FgPbs2 and FgHog1.
The yeast two-hybrid assays showed that FgSR does not interact with FgPbs2 and FgHog1, but with FgSsk2 (Fig. S6). These results indicate that FgSsk2, but not FgPbs2 and FgHog1, is involved in phosphorylating FgSR ( Fig. S6; lines 219-223).
3. Regarding the putative cis element identified and characterized in this study, to this reviewer, the authors did not present convincing or direct evidence to show it is the binding site of FgSR. The in vivo effects of their deletion or mutation in the Cyp51A promoter may be caused by other factors that bind to or recognize these elements (not by FgSR per se). The authors should conduct EMSA assays to show the direct binding of FgSR to these cis elements and mutational effects. Also, why only five of these genes, not majority of the sterol synthesis related genes, have these elements? (I don't know how reliable the ChIP-seq data generated by the authors) Re: We have conducted the yeast one-hybrid and the result showed that FgSR is able to bind the promoter of FgCYP51A via two repeats of CGAA element directly. Many previously reports have shown that the yeast one-hybrid is a reliable approach to determine direct binding of a protein to DNA element in a heterologous system (Hens In addition, we also tried to perform EMSA, but failed in purification of FgSR-His via the E. coli expression system because FgSR-His was mainly present the inclusion body. We tried to optimize different conditions, such as temperature, prokaryotic expression vectors, but FgSR-His was still present the inclusion body (Image below).
In this study, the ChIP-Seq analysis provided a putative cis-element (revised Fig.   S8e). To get an accurate cis-element that is bound by FgSR. We selected five ergosterol-related genes, which were down-regulated in RNA-Seq and targeted by FgSR in the ChIP-Seq assay. By using the MEME (Motif-based sequence analysis tools), we found that the promoters of all these five genes contain a predicted 16-bp cis-element bearing two CGAA-repeated sequences (Fig. 7a). Further analysis revealed that this cis-element existed in the promoters of 64 (54%) FgSR target genes identified from the ChIP-Seq assay including 20 sterol biosynthesis genes ( Fig. 6a; Table S3). 4. I am also concerned with the five phosphorylation sites described in the manuscript. It was not clearly described how the authors identified these five sites. Is there any way to predict phosphorylation sites of MAPKKK on non-MAPKK targets? Also, because of the differences between FgSsk2 deletion and 5A mutations, the authors should do mutations at individual sites to identify the FgSsk2 phosphorylation sites.
(Deletion of FgSSK2 had no effect on tebuconazole induced expression of Cyp51A but five phosphorylation sites were important for that). Another experiment is to generate the 5D mutations to mimic activation.
Regarding Figure 3h, if the authors could identified the FgSR phosphorylation sites by phosphoproteomics analysis, they should be able to use the same method to identify changes in the phosphorylation sites of FgSR in the FgSsk2 deletion mutant.
Re: Five sites of FgSR were predicated by using the program NetPhos (http://www.cbs.dtu.dk/services/NetPhos/). Further, liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis verified that two of them were phosphorylated under tebuconazole treatment. Hence, we further studied the function of these five sites.
Based on your suggestion, we constructed the 5D mutant (phosphorylated mutant) and found that FgSR 5D could restore the sensitivity of FgSR to tebuconazole and the transcription of FgCYP51A with tebuconazole treatment (Fig.3a,f). We also obtained the mutants containing individual phosphorylation site mutated from S/T to A (dephosphorylated mutation), and found these mutants did not exhibit increased sensitivity to tebuconazole, indicating the function of these five phosphorylation sites may be redundant. Please see lines 241-245, and Fig. S7b.
In Fig. 3f, we directly marked the relative expression of FgCYP51A in PH-1 under tebuconazole treatment with digit 78.05 to avoid the graph looking too high.
The relative expression of FgCYP51A in FgSsk2 mutant is 8.47 in comparison with that of PH-1 without treatment. Therefore, deletion of FgSSK2 decreased the expression level of FgCYP51A by 89.15% under tebuconazole treatment. We reconstructed Fig. 3f to make the results clearly. Furthermore, to identify the sites of FgSR potentially phosphorylated by FgSsk2, we detected the phosphoproteomics in FgSsk2::FgSR-GFP treated by tebuconazole in two independent LC-MS/MS analyses. The result showed that S305 of FgSR site was not phosphorylated in the mutant, but it was phosphorylated in the wild type. 5. Fusarium graminearum has two SREBP genes and four Upc2 orthologs. It is possible that some of these genes may have overlapping functions. Based on phenotypical characterizations of the single mutants of these genes, the authors could not concluded that these genes are not important for regulating sterol synthesis. In fact, the FgSR deletion mutant was only reduced approximately 50% in growth rate under normal growth conditions, suggesting FgSR is not an important regulator of sterol synthesis because of the essential functions of ergasterol.
This transcription factor FgSR has been characterized in an earlier study (Son et al., 2011. PLoS Pathogens). Deletion of this transcription factor had pleiotropic defects, including increased sensitivity to oxidative stress and reduced virulence.
The authors identified approximately 50 putative FgSR-interacting proteins in yeast two-hybrid assays. FgSsk2 and Swp71 were just two of them. Somehow, both of them are related to sterol synthesis?
Re: Thanks for your constructive comments. We conducted additional experiments to confirm that SREBP and Upc2 orthologs were not involved in ergosterol biosynthesis.
Second, FgCYP51A still could be up-regulated by tebuconazole treatment in the SREBP and UPC2 single or double deletion mutants (Fig. S1c). Third, the transcription of each SREBP or Upc2 ortholog was not altered by tebuconazole treatment or deletion of another SREBP or Upc2 ortholog (Fig. S1d). Taken together, SREBP and Upc2 orthologs in F. grainearum are not involved in ergosterol biosynthesis.
Erg4 is the last enzyme catalyzing ergosterol biosynthesis. Our previous study has reported that FgERG4 deletion mutant displayed 50% growth relative to the wild type although the mutant could not synthesize ergosterol (Liu et al, 2013). In S. cerevisiae, deletion of ERG4 also is viable and ERG4 shows similar growth compared with the wild type although ERG4 cannot biosynthesize ergosterol (Zweytick et al, 2001). Moreover, deletion of the master sterol regulator in Aspergillus fumigatus or Cryptococcus neoformans only caused a slight decrease in growth under the normal conditions (Willger et al, 2008;Chun et al, 2009). Taken together, it is reasonable to draw the conclusion that FgSR is the master regulator of sterol biosynthesis in F. graminearum although FgSR only exhibited approximately 50% growth relative to the wild type under the normal conditions.
We apologized for making a mistake in numbering FgSR-interacting proteins obtained by screening cDNA library of F. graminearum. In fact, total 67 FgSR-interacting proteins were identified. Among them, 44 potentially ergosterol-related proteins were knocked out. Determining the sensitivity of 44 mutants to tebuconazle, we found that 41 mutants showed similar sensitivity, only kinase FgSsk2 (FGSG_00408), kinase FgCdc15 (FGSG_10381) and FgSR (FGSG_01176) deletion mutants displayed increased sensitivity in comparison with the wild type (Image below). Further, yeast two-hybrid assays confirmed that FgSR interacted with the full length of FgSsk2, but not the full length of FgCdc15. Therefore, we focused on the function of FgSsk2 for regulating FgSR in this study. 1. Fig. 3B. It appears to this reviewer that the images for the negative control and FgSR or FgSR5A in Fig. 3b were not the original images or images of low quality.
Re: We replaced the images in Fig. 3b with images having high resolution from another repeated experiment. 3. Deletion of ML likely affects protein folding, which may affect many things other than dimerization. Which domain is responsible for DNA binding or binding to the cis elements identified in this study?
Re: Thank you for your good suggestion. According to the NCBI (https://www.ncbi.nlm.nih.gov/) and SMART (http://smart.embl-heidelberg.de/) analysis, FgSR contains a zinc finger (ZF) domain which is a typical DNA binding domain (Lee et al, 1989;Persikov et al, 2015). We conducted experiments to study the function of this zinc finger domain. As shown in revised Fig. 5d-f, the strain lacking the ZF domain (FgSR-C ZF ) displayed the defects in response to azole compounds, ergosterol biosynthesis, induction of FgCYP51A expression by tebuconazole treatment, and the binding ability to FgCYP51A promoter (lines 291-296). These results indicate that the zinc finger domain is responsible for DNA binding.
Previous studies have found that the dimerization domain of zinc finger transcription factor is primarily located in the C-terminal next to the DNA binding domain (Näär et al, 2009). In this study, deletion of the middle linker (ML) domain disrupted the formation of FgSR homodimer (Fig. 5a-c). Further, similar to FgSR and FgSR-C ZF , the strain lacking the ML domain ( FgSR-C ML ) could not bind to the promoter of FgCYP51A, further displayed the defects in response to azole compounds, ergosterol biosynthesis, and induction of FgCYP51A expression by tebuconazole treatment (Fig. 5d-f). Moreover, ChIP-qPCR assay showed that similar to FgSR-C ZF , FgSR ML could not bind to the promoter of FgCYP51A (Fig. 5h).
Therefore, we proposed that the homodimerization of FgSR mediated by the ML domain is required for its binding to the target gene promoter (lines 290-300). This manuscript deals with the mayor plant pathogen <i>Fusarium graminearum</i> and the discovery of a novel regulator of sterol biosynthesis. From mammals to yeast, sterol biosynthesis is regulated by a rather conserved transcription factor called Upc2/SREBP. In this manuscript, a so far not described description factor FgSR is described that provides a novel mechanism for regulating sterol biosynthesis. Most importantly, this regulation differs very much from the regulation by the conserved SREBP/Upc2 orthologs. For example, FgSR is located in the nucleus independent of ergosterol starvation. Further FgSR protein levels do not alter by sterol treatment, and the protein level bound to the promoters of target genes is not increased under sterol-deprived conditions. Finally, FgSR phosphorylation regulates its transcriptional activity via recruiting the SWI/SNF complex, which is responsible for chromatin remodeling. The manuscript is highly complex and provides a huge wealth of information. For example, the data are presented in nine figures that overall display 47 subfigures. However, the overall finding of this manuscript is restricted to a few ascomycetous groups, namely the Sordariomycetes and the Leotiomycetes. With this, a rather specific regulatory mechanism is described. Overall, it is a very ambitious paper and I have some points, which should be clarified before publication.
Thank you very much for the positive comments. Based on your suggestion, we constructed the 5D mutant FgSR-C 5D (a mimic phosphorylated isoform) and found that the mutant could restore the sensitivity of FgSR to tebuconazole and the transcription of FgCYP51A with tebuconazole treatment (Fig.3a,f). We also obtained the mutants containing individual non-phosphorylated mutation (from S/T to A), and found these mutants did not exhibit increased sensitivity to tebuconazole, indicating the function of these five sites may be redundant. Please see lines 241-245, and Fig. S7b. 2. Further in this chapter, the authors found an interaction of FgSwp73, a component of the SWI/SNF complex. This is solely based on a Y2H approach. They further provide other data, for example enrichment of FgArp9 at the FGCYP51A promoter or the enrichment of H3 at the promoter of FgCYP51A. All these are to my mind rather indirect evidences for the statements made by the authors that the SWI/SNF complex is recruited by FgSR and is involved in the transcriptional regulation of sterol biosynthesis genes (line 247-251).
Re: The Co-IP assay was conducted to confirm that FgSR interacts with FgSwp73.
Moreover, the Co-IP assay showed that the interaction of these two proteins was Re: In order to confirm that enrichment determination of FgSR-GFP the promoter of FgCYP51A is reliable, and not caused by the GFP protein, we determined the GFP enrichment at FgCYP51A promoter in the negative control strain PH-1::GFP, and found that the GFP enrichment was undetectable in the promoter of FgCYP51A (down and right panel in Fig. 2b), indicating that GFP cannot bind the FgCYP51A promoter. 4. Fig. 6A: The authors show the ChIP-seq data for several of the ergosterol biosynthesis genes. However, three are missing. Explain why no data were received.
Re: According to the ChIP-Seq analysis, FgSR showed the significant enrichment at the promoters of 20 ergosterol biosynthesis genes, but not at the promoters of 8 genes in ergosterol biosynthetic pathway. We therefore did not display ChIP-Seq data for the eight genes in Fig. 6a Chip-seq and in vivo transcriptome analyses of the Aspergillus fumigatus SREBP SrbA reveals a new regulator of the fungal hypoxia response and virulence. Plos Pathogens, 10(11), e1004487. 5. Fig. 7: The MEME analysis is shown for five genes. I do not understand why only five target genes were selected, although much more target genes were identified in the corresponding chapter (line 302-309).
Re: Indeed, the ChIP-Seq analysis provided a putative cis-element (revised Fig. S8e).
To get an accurate cis-element that was bound by FgSR. We selected five ergosterol-related genes, which were down-regulated in RNA-seq and the promoters of these genes are bound by FgSR in ChIP-Seq assay. Thus, by using the MEME (Motif-based sequence analysis tool) program, we found the promoters of all these five genes have the 16-bp cis-element containing two CGAA-repeated sequences (Fig.   7a), which is consistent with the putative cis-element identified from ChIP-Seq analysis. In addition, this element is present in the promoters of 64 (54%) FgSR target genes.
6. Fig. 7: There are four promoter derivatives shown (P1-P4). However, P4 was not checked <i>in vivo</i> (Fig. 7C,  1. I still have doubts about the interaction of FgSsk2 with FgSR and the direct role of FgSsk2 in the phosphorylation of FgSR. For the direct interaction between FgSsk2 and FgSR, in the previous submission, the authors showed the localization of FgSsk2 to the cytoplasm and localization of FgSR to the nucleus although co-IP data showed they interacted in hyphae cultured under normal conditions. In the revised manuscript, the authors generated a new transformant and showed strong FgSsk2-RFP signals in hyphae cultured under normal conditions. The explanation that the endogenous FgSsk2 may interfere the localization of FgSsk2-RFP in the previous submission did not make sense to me (It should not because of co-IP with the epitope tag). More importantly, my understanding is that localization of activated MAP kinases or MAP kinase complexes to the nucleus is a dynamic process. Assuming the localization of FgSsk2 to the nucleus with FgPbs2 and FgHog1, should the localization of FgSsk2-RFP to the nucleus occurred dynamically (in-and-out, transiently) ONLY after this pathway was activated? Why do all the nuclei have strong RFP signals under normal culture conditions, assuming FgSsk2 and its downstream kinases are not activated? Also, if this assumption of the FgSsk2-FgPbs2-FgHog1 complex is correct, should the authors be able to detect the interaction of FgPbs2 or FgHog1 with FgSR if they could detect the FgSsk2-FgSR interaction by co-IP? It is puzzling to this reviewer that FgSsk2-RFP localized to the nucleus under normal culture conditions (not activated). Should FgSsk2, FgPbs1, or FgHog1 be mainly in the cytoplasm before this pathway is activated? Is it possible that the localization of FgSsk2-RFP to the nucleus under normal culture conditions is due to overexpression of the fusion construct? Along the same line, the strong interaction between FgSR and FgSsk2 detected by co-IP may be due to the overexpression of FgSR fusion construct. The interaction between a kinase and its substrate is also a transient process. It also sounds somewhat odd for the explanation given by the authors about the BiFC assays. They showed the dimerization of FgSR by BiFC and localization of FgSsk2-RFP to the nucleus. Does that mean fusion with split GFP or RFP had no effect on their localizations to the nucleus? Somehow, the interaction of Swp71 with FgSR also could not be visualized by BiFC.
2. Regarding phosphorylation of FgSR by FgSsk2, the authors still did not present any direct evidence to show that FgSsk2 directly phosphorylates FgSR in the revised manuscript. The effect of FgSsk2 deletion on FgSR phosphorylation could be indirect because the normal function of FgSsk2 is in the Ssk2-Pbs2-Hog1 pathway. Deletion of FgSsk2 will affect the activation of FgPbs2 and FgHog1. Is the FgHog1 or FgPbs2 mutant defective in FgSR activation? Based on a quick literature search, the Fgssk2, FgPbs2, and FgHog1 mutants had the same phenotypes. If these mutants of the same MAP kinase pathway phenocopied each other, how can FgSSK2 have a direct phosphorylation target as important as FgSR in F. graminearum? The authors may need to discuss about this point.
The authors provide a revised version of a previous manuscript and addressed all concerns, made in the reviews for the initial manuscript. I appreciate that major concerns regarding phosphorylation sites of the transcription factor were addressed. In particular, the predicted phosphorylation sites were identified using NetPhos. And this explains in part the follow-up experiments. (see also the critics of reviewer #2, point2) Another concern was related to the identified cis-element that putatively was bound by transcription factor FgSR. Although the authors gave a lot of significant answers, concerning this cis-element, an important experiment is still missing. They were unable to demonstrate the binding of the cis-element by FgSR using an EMSA assay (see reviewer #2, point 3 and the corresponding answer). All other points were addressed and most of the sloppy spelling mistakes were corrected.