Arabidopsis FHY3 and FAR1 integrate light and strigolactone signaling to regulate branching.

Branching/tillering is an important parameter of plant architecture and is tightly regulated by both internal factors (such as plant hormones) and external factors (such as light conditions). How the various signaling pathways converge to coordinately regulate branching is not well understood. Here, we report that in Arabidopsis, FHY3 and FAR1, two homologous transcription factors essential for phytochrome A-mediated light signaling, and SMXL6/SMXL7/SMXL8, three key repressors of the strigolactone (SL) signaling pathway, directly interact with SPL9 and SPL15 and suppress their transcriptional activation of BRC1, a key repressor of branching, thus promoting branching. In addition, FHY3 and FAR1 also directly up-regulate the expression of SMXL6 and SMXL7 to promote branching. Simulated shade treatment reduces the accumulation of FHY3 protein, leading to increased expression of BRC1 and reduced branching. Our results establish an integrated model of light and SL coordinately regulating BRC1 expression and branching through converging at the BRC1 promoter.

Xie et al. dissected the involvement of FHY3 and FAR1 in the regulation of branching in response to light quality change. fhy3/far1 has reduced branching under white light but diminished the difference in branching under white light versus EOD-FR shown for wild type. Inside to the story, FHY3 and FAR1 inhibit the binding of SPL9 and SPL15 to the BRC1 promoter, and SPL9 and SPL15 directly activates BRC1 expression. Meanwhile, three SMXL proteins also interact with SPL9 and SPL15 but rather repress their transactivation activity. Finally, FHY3 and FAR1 bind the first exons of SMAXL6 and SMAL7 and activate their expression. A model is followed to present their overall discoveries.
Major comments 1) In Figure 3d, another control, MYC plus FHY3:HA, is needed for in vivo co-IP on the right panel. In addition, in vivo co-IP with FAR1 may show a low affinity toward either SPL9 or SPL15. The in vivo results with FAR1 may partially explain the subtle phenotype of far1. In Figure 6d, FAR1 appears to have a weak transactivation activity for SMXL6 and 7 compared to that of FHY3. 2) In Seven-day-old seedlings grown under WL conditions with EOD-FR treatment, both FHY3 and FAR1 proteins accumulated much less as shown in Figure 1d and 1e. Similar results were also shown for adult plants of four-week old in Figure S1. In Figure 4, EOD-FR may affect the ChIP of SPL9 and SPL15 to the BRC1 promoter. The authors may choose either wild type or FHY3 OX plants and variable ages to detect the best responses. 3) In Figure 6, the model suggests that "simulated shade treatment reduces the accumulation of FHY3 protein, leading to increased expression of BRC1 and reduced branching". As EOD-FR affects FHY3 and FAR1 accumulation, EOD-FR may affect the expression of SMXL6 and SMXL7 in either wild type or FHY3 OX. Data with EOD-FR treatment added in this figure plus experiments suggested in 2) would strengthen the model.
Minor points 1) In line 705, revise as both FHY3 and FAR1 protein level rapidly declined in seedlings treated with EOD-FR. 2) In line 454, fill the missing parts for "2_-pBRC1 or 2_-pBRC2".
Reviewer #2 (Remarks to the Author): This manuscript describes some potentially very interesting results regarding the integration of light signalling into the shoot branching pathway in Arabidopsis. The authors propose that FHY3 and FAR1 regulate branching both by blocking SPL9:SPL15 regulation of BRC1, and by promoting transcription of SMXL678, which also block SPL9:SPL15 regulation of BRC1. Along the way, this involves showing that SPL9/SPL15 regulate BRC1 in Arabidopsis (not shown before, but homologous to rice), and that SMXL678 act through SPL9/SPl15 (again homologous to rice, not previously shown in Arabidopsis). So there is potentially a lot of interesting work here Each part of the manuscript uses multiple different approaches to support its conclusions, and superficially, the manuscript seems very convincing. However, the manuscript lacks a substantial amount of methodological detail regarding what was actually done, and how it was done. I have major concerns about some of the results, because although it is clear what the result shows it is not clear how the results were obtained.
In particular, the authors claim to have expressed full length SMXL6 protein using a simple bacterial expression system. This is widely acknowledged to be impossible among labs working on strigolactone signalling. I simply don't believe that the authors expressed the protein as described in the manuscript. The SMXL6 protein doesn't appear to be the correct size either -I'd expect a full-length His-SMXL6 from pET28a to be about 111 kDa (107.8 kDa for SMXL6 plus 3.4 kDa extra for the His-tag and linker residues), but the band here is under 100 kDa. It also isn't particularly pure (none of the "pure" proteins are), and the coomassie gel looks to me like it might have been altered to make the contaminant bands less prominent compared to the desired bands.
To credit the authors purification work, I would want to see: 1) the exact mass and sequence of the SMXL6 fusion protein they express 2) more information on their expression methods ie. growth medium, extraction method, extraction buffer (plus any additives) 3) raw Coomassie gels of their expression with ladders, including induced vs uninduced cells, and soluble vs insoluble fractions for both, plus any westerns they might have done (eg anti-His) with ladders 4) raw Coomassie gels of their purification with ladders, including crude, flow through, wash and elution fractions, plus any westerns they might have done (eg anti-His) with ladders 5) confirmation whether they did any secondary purifications prior to that gel, and any related evidence The same applies to some extent for the other protein work: the supporting evidence and methods are simply not up to scratch.
I would also highlight that the genetic crosses the authors perform to support their protein work are largeky uninformative, and do not provide any evidence for genetic interactions. 1) BRC1 vs SPL9/15 The brc1 brc1 x rSPL9 OE is additive, and BRC1OE spl9 spl15 is epistatic -no evidence for interaction here. Indeed, the evidence is more in favour of non-interaction.
2) FHY3/FAR1 vs SPL9/SPL15 Again, all the crosses of fhy3 far and FHY3OE x spl9 spl15 demonstrate is that spl9 spl15 is epistatic to FHY3. That in itself isn't good evidence for an interaction.
3) SMXL678 vs SPL9/SPL15 Again, the results just show that SPL9/SPL15 are epistatic to SMXL678. The authors state that "while SMXL6D-OE/ SPL9-OE had more rosette branches than SPL9-OE", but that is absolutely not what the data in  Response: Thanks for helpful suggestions. As suggested, we re-performed the Co-IP assay with the new MYC control plus FHY3-HA (See new Fig. 3d). We also performed the Co-IP assay for the in vivo interaction between FAR1 and SPL9/SPL15 and provided the data in the revised manuscript (See new Fig. 3d). In Figure 6d, we also quantified the luciferase activity by the LUC/REN assay with internal reference (35S::REN) (See Supplementary Fig. 18), which shows that indeed FAR1 has a weaker transactivation activity for SMXL6 and SMXL7 than FHY3.

Q2)
In Seven-day-old seedlings grown under WL conditions with EOD-FR treatment, both FHY3 and FAR1 proteins accumulated much less as shown in Figure 1d and 1e.
Similar results were also shown for adult plants of four-week old in Figure S1. In Response: Thanks. We corrected the sentence in Line 705 as "the FHY3 protein level rapidly declined in seedlings treated with EOD-FR".

Q1) the exact mass and sequence of the SMXL6 fusion protein they express
Response: We understand the reviewer's concern and indeed we also found that SMXL6 protein is hard to be induced and expressed in E. coli. We apologize for not having provided the necessary details of protein expression and purification. We now  Fig. 23; Supplementary data set). We re-purified all proteins used in this study, including GST, MBP, GST-SPL9SBP, GST-SPL15SBP, 6xHis-FHY3-I and 6xHis-SMXL6, and re-ran their coomassie gels and conducted western blotting as requested ( Supplementary Fig. 22). All these data verified the identities of the purified target proteins.
As a note, we firstly used Escherichia coli both BL21 (DE3) and Tuner stains to induce SMXL6 protein expression with different incubation temperatures (16 and 37 o C) and different working concentrations of isopropyl β-D-thiogalactoside (IPTG) (0.1, 0.5, 1 and 5 mM), however, no obvious expression of SMXL6 proteins was detected under these conditions. Then we turned to the E. coli Transette stain (modified from Rosette stain) to induce SMXL6 protein expression with various concentrations of IPTG at 16 and 37 o C, respectively. We found that expressed SMXL6 protein only existed in pellet but not in the supernatant. Finally, we set 4 o C as the induction temperature (the actual temperature in shaker is around 5-6 o C) and used 0.1 mM IPTG to induce SMXL6 protein expression. We incubated the bacteria for more than 20 h with gentle rotation (80 rpm) and then harvested the cells. After sonication and purification, we obtained recombinant SMXL6 proteins used for all the assays (coomassie staining, western blot analysis, and LC-MS/MS and Q-TOF assays).

Q2) more information on their expression methods ie. growth medium, extraction method, extraction buffer (plus any additives)
Response: See reply to Q1.

Q3) raw Coomassie gels of their expression with ladders, including induced vs uninduced cells, and soluble vs insoluble fractions for both, plus any westerns they might have done (eg anti-His) with ladders
Response: Thanks for helpful suggestions. As suggested, we re-purified the protein and re-ran raw coomassie gels including crude, flow through, wash and elution proteins and conducted western blotting. We supplied the data in Supplementary Fig.   22.

Q4) raw Coomassie gels of their purification with ladders, including crude, flow through, wash and elution fractions, plus any westerns they might have done (eg anti-His) with ladders
Response: See our reply to Q3. Q5) confirmation whether they did any secondary purifications prior to that gel, and any related evidence Response: Thanks. We did not perform any secondary purifications in this study.
The same applies to some extent for the other protein work: the supporting evidence and methods are simply not up to scratch.
I would also highlight that the genetic crosses the authors perform to support their protein work are largeky uninformative, and do not provide any evidence for genetic interactions.

Q1) BRC1 vs SPL9/15
The brc1 brc1 x rSPL9 OE is additive, and BRC1OE spl9 spl15 is epistatic -no evidence for interaction here. Indeed, the evidence is more in favour of non-interaction.

Response:
To verify the genetic relationship between SPL9/15 and BRC1, we grew brc1 brc2, spl9 spl15, BRC1-OE, rSPL9-OE, and their high order mutants in large scale (more than 50 plants for each genotype) and counted their rosette-leaf branch number. The results showed that that there was no significant difference in the rosette-leaf branch number between brc1 brc2/ rSPL9-OE and brc1 brc2, and no significant difference between BRC1-OE/ spl9 spl15 and BRC1-OE (new Supplementary Fig. 4). In addition, we performed RT-qPCR assay, which showed that the BRC1 expression levels in spl9 spl15 and brc1 brc2/ rSPL9-OE plants were significantly lower than that in wild type, while the BRC1 expression levels in rSPL9-OE and BRC1-OE/ spl9 spl15 plants were significantly higher than that in wild type ( Supplementary Fig. 5). These data strongly suggest that SPL9/15 act upstream of BRC1 to inhibit branching. To further confirm this, as the editor suggested, we generated the constructs containing the BRC1 coding regions driven by the wild type BRC1 promoter (pBRC1wt::gBRC1) or BRC1 promoter with the SBP binding site mutated (pBRC1m::gBRC1) and transformed these constructs into brc1 brc2, spl9 spl15/ brc1 brc2, rSPL9-OE/ brc1 brc2 backgrounds and examined their effects on branch numbers. We found that the branch number in the brc1 brc2 plants harboring the pBRC1wt::gBRC1 transgene was restored to wild type level, whereas the brc1 brc2 plants harboring the pBRC1m::gBRC1 transgene had significantly more branches than the wild type plants (similar to brc1 brc2). We also found that the spl9 spl15/ brc1 brc2 plants carrying the pBRC1wt::gBRC1 or pBRC1m::gBRC1 transgenes had similar branch number as the spl9 spl15/ brc1 brc2 plants ( Supplementary Fig. 8). Moreover, we found that rSPL9-OE/ brc1 brc2 plants harboring pBRC1wt::gBRC1 transgene had significantly fewer branches than brc1 brc2 and wild type plants. However, no significant difference was detected between rSPL9-OE/ brc1 brc2 plants harboring the pBRC1m::gBRC1 transgene and the brc1 brc2 mutant plants (Supplementary Fig. 8). Collectively, these results strongly support the notion that SPL9 and SPL15 act to regulate branch number through regulating BRC1 expression via binding to the the SBP core motif in the BRC1 promoter.

Q2) FHY3/FAR1 vs SPL9/SPL15
Again, all the crosses of fhy3 far and FHY3OE x spl9 spl15 demonstrate is that spl9 spl15 is epistatic to FHY3. That in itself isn't good evidence for an interaction.

Response:
We believe that our multiple molecular assays (including yeast two-hybrid, BiFC, LCI, pull-down, Co-IP) provided substantial evidence to support the notion that  Supplementary Fig. 12). To test a possible effect of FHY3/FAR1 on the transcription of SPL9/15, we conducted yeast one-hybrid assay to test the binding of FHY3/FAR1 to the promoters of SPL9 and SPL15. The result showed that no binding between FHY3/FAR1 and the promoters of SPL9/15 (See Supplementary Fig. 19). This observation is consistent with the observation that there is no typical FHY3/FAR1 binding site (FBS, 5'-CACGCGC-3') in the promoters of SPL9/15, suggesting that FHY3/FAR1 do not bind to the promoters of SPL9/15. Further RT-qPCR assay showed that there is no significant difference in the transcript levels of SPL9 or SPL15 between wild type, fhy3far1 and FHY3-OE plants (See Supplementary Fig. 12 Supplementary Fig. 16), supporting the notion that SPL9 acts downstream of SMXL6/7. To test whether SMXL6/7 regulate SPL9/15 expression, we also conducted the yeast one-hybrid assay to test whether SMXL6/7 bind to the promoters of SPL9 and SPL15. The result showed that no binding between SMXL6/7 and the promoters of SPL9/15 was detected (See Supplementary Fig. 19), suggesting that SMXL6/7 do not bind to the promoters of SPL9/15 and transcriptionally regulate their expression.
Further, no significant difference was detected in the expression levels of SPL9 or SPL15 between wild type, smxl6/7/8 and SMXL6D-OE plants (See Supplementary   Fig. 16). Together, all results support the notion that SMXL6/7 regulate BRC1 expression through direct physical interaction with SPL9/15.
To test the possibility whether FHY3/FAR1 and SMXL6/7 directly regulate BRC1 expression, we also conducted two sets of additional experiments. First, we tested whether FHY3/FAR1 and SMXL6/7 could directly bind to the BRC1 promoter by bypassing SPL9/15. However, yeast one-hybrid assay revealed that there was no direct binding of FHY3/FAR1 or SMXL6/7 to the BRC1 promoter (See Supplementary Fig. 19). This result suggests that it is unlikely FHY3/FAR1 and SMXL6/7 directly regulate BRC1 transcription. Second, we tested whether FHY3/FAR1 and SMXL6/7 could regulate BRC1 activity through protein-protein interaction. Our yeast two-hybrid assay showed that no direct interaction was detected between BRC1 and FHY3/FAR1 or SMXL6/7 proteins (See Supplementary Fig. 20).
These results are consistent with our proposition that both FHY3/FAR1 and SMXL6/7 do not directly regulate BRC1 gene expression/activity.
We also noted that the SMXL6D-OE/ SPL9-OE plants had slightly more (~15.7%) branches than SPL9-OE plants, suggesting that SMXL6D may affect branching number through additional targets (such as SPL15) besides SPL9. This observation is still consistent with our proposition that SMXL6/7 act upstream of SPL9/15 to regulate branching. To be more accurate, we modified the sentence "while SMXL6D-OE/ SPL9-OE had more rosette branches than SPL9-OE" as "while SMXL6D-OE/ SPL9-OE had slightly more (~15.7%) rosette branches than SPL9-OE".
The authors have addressed my comments.
Reviewer #2 (Remarks to the Author): In this revised manuscript, the authors have addressed, or claimed to have addressed, all my criticisms from my previous review.
However, I continue to have concerns about some of experiments and the incomplete documentation provided by the authors. Here is a sample of concerns: 1) The manuscript uses a huge number of different transgenic Arabidopsis lines. How were these transformed? What strategy did they use to make stable T3 homozygous lines? How many independent transgenic lines did they examine for each construct? What percentage showed the phenotype? Which line did they pick for the final analysis, and why?
Based on the speed at which they generated new constructs for this manuscript (e.g. pBRC1wt:gBRC1 and pBRC1m:gBRC1) --less than 6 months --the authors must have used T1 lines for this analysis. Thus each "sample" of a transgenic line would have actually consisted of separate, independent T1 transgenic lines, with different genomic insertions and therefore effects on gene expression. Some of the lines would have no phenotype at all. It is simply not appropriate to perform phenotypic experiments on a collection of independent T1 lines.
Is this also the case for all the other transgenic lines in the paper? Did the authors perform the experiments on a collection of freshly generated, independent T1 lines? Or did they use T3 lines? If I asked the authors for seed from the lines shown in the paper, would they actually be able to send me homozygous lines?
All of this is completely standard experimental practice when using transgenic Arabidopsis, and important methodological information to provide.

2) Statistics
The majoriy of the statistical methodology provided in the paper just says: "Different letters indicate significant differences by two-way ANOVA (p<0.01). But ANOVA does not indicate where the differences occur, only that there are significant differences. So how did the authors assign these letters?
Did the authors check their data for normality and use the appropriate non-parametric test where needed? (a core requirement for publishing in Nature journals).

3) Figure manipulation
I would also draw the authors attention to Figure 5B, where in the top row, exactly the same leaf has apparently been infiltrated with SMXL7-cLUC AND SMXL8-cLUC, in exactly the same splotch pattern, but with different signal intensities. Of course, these are just the same image with different exposures, masquerading as separate experiments. This is not a freak mistake, because on the lower row of Figure 5B, exactly the same leaf has apparently been infiltrated with SMXL6-cLUC AND SMXL7-cLUC, in exactly the same splotch pattern, but again with different signal intensities. Again, the same image with different exposures is being used to cover two experiments.

Point-to-point response to reviewers' comments:
Reviewer #1: The authors have addressed my comments.

Reply:
We wish to thank reviewer #1 for his/her support towards this work. Reply: Thanks for reminding us to clarify these issues. All the stable transgenic lines were generated using the floral dip method as described in Clough and Ben 1 . Then the seeds of the transformed plants (designed as T 0 ) were harvested and sown on half strength Murashige and Skoog (1/2MS) media plates supplemented with 1% sucrose, 50 g ml -1 cefalexin (for inhibiting the growth of agrobacterium) and proper concentration of antibiotic suggested by the vendors.

Reviewer
The survived seedlings (designated T 1 generation, typically we grew more than 30 plants for each construct, which are treated as independent lines) with green cotyledons and long root were transferred into soil to grow for T 1 seeds.
More than 100 T 1 seeds from individual T 1 line were sown on the 1/2 MS media plates containing antibiotic. After 7-10 days' growth, the seedlings (T 2 ) with yellow cotyledons and short or no roots (presumably transgene-negative control seedlings) were counted and the ratio of antibiotic-sensitive to antibiotic resistant seedlings was calculated. From the T 1 line with the segregation ratio ~1:3 (indicating a single transgene locus), more than 20 antibiotic resistant T 2 seedlings were transferred into soil to grow for T 2 seed harvest and their seeds were harvested separately from individual T 1 plant. The T 2 seeds from individual T 2 plant were again sown on the 1/2 MS media plates with antibiotic and that the lines did not show segregation again were deemed homozygous. Meanwhile, we also plated the T 2 seeds from individual T 2 plant on the 1/2 MS media plates without antibiotic as duplicates and the homozygous seedlings grown on 1/2 MS media plates without antibiotic were used for further assay (to avoid effect of antibiotics). T 3 seeds are harvested from selfed homozygous T 2 plants and used for phenotypic assays. For the smxl6/7/8 CRISPR vector, we totally obtained 57 individual T 1 transgenic lines. After PCR detection with the special primer pairs, however, we found there was no obvious deletion for SMXL6 gene in most of these lines, although there were obvious or complete deletion for both SMXL7 and SMXL8 in 12 individual line. We sequenced the PCR products amplified with SMXL6 primer pair using these 12 lines as template and found that two lines (Line #2 and Line #16) contain mutations in the SMXL6 gene (Supplementary Figure 15). These two T 2 lines were propagated to obtain the homozygous smxl6/7/8 mutant from their T 3 generation and used for further study.
Based on their phenotype and expression level in T 1 heterozygous plants, we usually selected 3-5 independent overexpression lines for each construct to obtain homozygous lines.
During the period we also checked their phenotype. We re-analyzed their phenotype and expression levels in their homozygous lines to make sure that the expression level is associated with the phenotype. Based on the phenotypic analysis and expression detection, the line with the highest expression level and obvious phenotype is selected for further study. Actually, we used 2-3 lines for each construct at the beginning to do the phenotype analysis.
To save time, we used the T 1 lines of the new generated constructs for analyzing the roles of BRC1 wild type promoter and mutated promoter, a practice used in previously published studies 2 . We are also willing to share our reported materials in this manuscript upon request.

Q2. Statistics
The majoriy of the statistical methodology provided in the paper just says: "Different letters indicate significant differences by two-way ANOVA (p<0.01). But ANOVA does not indicate where the differences occur, only that there are significant differences. So how did the authors assign these letters? Did the authors check their data for normality and use the appropriate non-parametric test where needed? (a core requirement for publishing in Nature journals).
Reply: Thanks for valuable comments. We added a paragraph about the statistical analysis in the Methods section (See Page 23, Lines 645-653). We used the SPSS software (version IBM SPSS Statistics 22.0) and the two-sided least significance difference (LSD) test for multiple comparisons. We provided the analysis result in the Source Data file together with their raw data.
To show the difference between two group, we used letters not symbol * since the same symbol can not tell all the differences between groups. We assign and label these letters according to the analysis result. In brief, we assign the highest value as "a". The second highest value will be assigned as "b" if there are significant differences between the highest value and the second highest value, otherwise assigned as "a" and so forth.

Q3. Figure manipulation
I would also draw the authors attention to Figure 5B, where in the top row, exactly the same leaf has apparently been infiltrated with SMXL7-cLUC AND SMXL8-cLUC, in exactly the same splotch pattern, but with different signal intensities. Of course, these are just the same image with different exposures, masquerading as separate experiments. This is not a freak mistake, because on the lower row of Figure 5B, exactly the same leaf has apparently been infiltrated with SMXL6-cLUC AND SMXL7-cLUC, in exactly the same splotch pattern, but again with different signal intensities. Again, the same image with different exposures is being used to cover two experiments.
Reply: Thanks for pointing this out. It is a careless mistake and we apologize for that (we noticed it ourselves during the checking process and corrected it when we sent a revised manuscript to the editor to answer his quires last time). As suggested by the editor, we also provided the biological replicates as many as possible for all the fluorescence images including the luciferase activity assay and luciferase complementation imaging assay (See the Supplementary pdf file in the Source Data file).