Mutation of OsGUN4 Uncoupled the Sugar-dependent Signals to Regulate Starch Biosynthesis in Rice

Starch biosynthesis requires plastid-to-nucleus signals to ordinate the ow of carbon, which is partly mediated by tetrapyrrole intermediates. We previously revealed that mutation of the Genomes Uncoupled 4 (OsGUN4) would affect tetrapyrrole intermediates, but the underlying mechanisms for regulatory roles of OsGUN4 on starch biosynthesis remains largely unknown. PGM: phosphoglucomutase; PhANGs: photosynthesis-associated nuclear genes; PUL: pollulanase; RSR: RICE STRACH REGULATOR; SBE: starch branching enzyme; SnRK1s: Snf1-related protein kinases; SS: sucrose synthase; SSS: soluble starch synthase; SPP: sucrose phosphate phosphatase; SPS: sucrose phosphate synthase; SuS: sucrose synthase; SUT: Sucrose transporter; UGP: UDP-glucosepyrophosphorylase.


Results
In this study, we revealed that the OsGUN4 mutation not only retarded the carbon ow from sucrose to starch but also disabled the sensitive response to exogenous feeding sucrose. Moreover, extra addition of nor urazon (NF) would aggravate insensitivity to the sucrose-dependent induction in gun4 epi , especially exhibiting collapse declines of the starch biosynthetic enzymes' activities. However, genes encoding starch synthetic enzymes performed discordance with the activities of starch biosynthesis associated enzymes upon on the scarce expression of OsGUN4 in gun4 epi . These results indicated that OsGUN4 played regulatory roles on biosynthetic genes and enzyme activity in starch biosynthesis. Furthermore, we also concluded that 1 O 2 derived from GUN4/protoporphyrin IX (proto) might be responsible for the sugardependent signals to regulate starch biosynthesis, due to positive correlations between the singlet oxygen ( 1 O 2 ) and many starch biosynthetic genes that were subjected to the control of three reported transcriptional factors (TFs) during starch biosynthesis. Eventually, the OsGUN4 mutation would also relieve the repression of glucose on Snf1-related protein kinases (SnRK1) but seem to negatively mediate the functioning of ADP-Glc pyrophosphorylase (AGPase).

Conclusion
In summary, we demonstrated that OsGUN4 serve as a broker to activate 1 O 2 -mediated signals in the sugar signaling cascade, possibly functioning upstream through TFs, and that OsGUN4 played roles in the SnRK1A-mediated signals, partly through the accumulations of sugars, e.g. glucose or sucrose.

Background
Plants assimilate atmospheric CO 2 during photosynthesis using light energy to produce sugars and chemical energy (ATP) for plant growth (Graf and Smith 2011). In leaves, sugars are partly retained in chloroplasts during the day to synthesize transitory starch for short term storage, and then exported to non-photosynthetic organs during the subsequent night for long-term storage (Sulpice et al. 2009). Starch is the major storage carbohydrate in higher plants, with essential physical functions and economical importance. As a major factor for plant growth, starch biosynthesis buffers metabolism and growth against the daily light/darkness alternation to avoid a shortfall of carbon at the end of the dark period (Smith and Stitt 2007; Sitt and Zeeman 2012). Besides, transient starch is photosynthetic synthesized during the day to provide carbon and energy under inactive photosynthesis (Bahaji et al. 2014).
Leaf starch mainly accumulates in the photosynthetic palisade and mesophyll (M) cells (Tsai et al. 2009;Van Bel 2003;Geiger 2011), and major mesophyll cells in mature leaves are source for sucrose transport into sink tissues (Fig. 1). Moreover, plastids of the photosynthetic organs are responsible for the temporarily synthesize of starch in leaves (Bahaji et al. 2014). While enzymatic functionality of the respective plastids depends largely on its own specialized proteome, and corresponding shifts of these proteome determine the transitions of different plastid types along with changes from environmental conditions and tissues (Enami et  Development from undifferentiated proplastid to functional plastid is coordinately achieved between plastid and nucleus, requiring cooperation between nucleus-to-plastid antegrade signaling and plastid-tonucleus retrograde signaling (Chan et al. 2016). The GUN (genome uncoupled) proteins were identi ed for plastid-to-nucleus signaling studies (Susek et al. 1993). Thereinto, GUN4 have been found to be involved in the retrograde signaling pathway in Arabidopsis (Larkin et al. 2003) and rice ). Besides, the mutation of OsGUN4 in rice have also been revealed to deregulate transcription of photosynthesisassociated nuclear genes (PhANGs) depending on disruption of singlet oxygen ( 1 O 2 )-induced signaling pathway . This model suggested that accumulation of heme in active chloroplast can activate a mechanism to induce the expression of PhANGs (Larkin 2016). Interestingly, the plastid-tonucleus retrograde signals is also revealed to regulate expression of nuclear starch biosynthetic genes, which is partly mediated by tetrapyrrole intermediates, i.e., heme (Enami et al. 2011). Besides, the mutation of OsGUN4 in rice have also been revealed to greatly affect tetrapyrrole intermediates, including heme, protoporphyrin IX (proto) and Mg-Proto ). Above on, retrograde signaling may play important roles in starch biosynthesis of leaves, but the underlying mechanism remains largely unknown.
In previous studies, we revealed that the OsGUN4 mutation greatly deregulated biosynthesis of tetrapyrrole intermediates and functioning in 1 O 2 -induced signaling pathway in rice Li et al. 2017). Here, we further employed the rice epi-genetic mutant gun4 epi to examine carbon metabolites, starch biosynthetic enzymes, genes involved in starch biosynthesis and eventually plan to disclose the roles of OsGUN4 on starch biosynthesis during vegetative stages. In conclusion, these ndings would con rm that OsGUN4 plays regulatory roles in starch biosynthesis.

Plant Materials
The following materials were used in this study: wild-type (Longtepu B, LTB) and its gamma ray-induced

Quanti cation of Proto
Brie y, 0.3 g ne powder from grinded leaves were immediately mixed with pre-cold alkaline acetone containing 0.1 N NH 4 OH (9:1; v/v), and centrifugated at 16,000 × g for 5 min. Subsequently, the supernatants were collected for extraction of Proto (Papenbrock and Grimm 2001). The concentrations of proto was determined using 0.3 g fresh leaf tissues with the commercial enzyme linked immunosorbent assay kits (ELISA) method following the manufacturer's instructions.

Determination of Singlet Oxygen Contents
The concentrations of 1 O 2 was determined using the SOSG (singlet oxygen sensor green) method (Hideg et al. 2002) with 300 mg leaf samples. The uorescence spectra were detected at excitation of 485 nm and emission of 520 nm using a uorescence spectrophotometer (359S, Lengguang Tech., China).

Analysis of Metabolites
The concentrations of amylose, starch, and protein were determined with the methods described previously (Han et al. 2012). Sucrose, fructose, and glucose were analyzed following the methods described previously (Tang et al. 2016).

Quantitative Real-time PCR Analysis
Quantitative real-time PCRs were performed with the methods as previously described ).
Relative gene expression was calculated in relation to the rice Ubiquitin gene using the 2 -ΔΔCt method (Livak and Schmittgen 2001). Genes were subjected to RT-PCR analysis by using gene-speci c primers (Additional le 1: Table S1).

RNA Sequencing
The cDNA libraries were constructed with RNA extracted from seedlings of 35 DAG and sequenced on an Illumina Hiseq 2000 platform (Beijing Novogene Bioinformatics Technology Co., Ltd. Beijing, China). For mapping, after removing adapter sequences from the raw reads, the Tophat v2.0.9 program (Trapnell et al. 2012) with E-value 10 -5 as cut-off point was employed to align the cleaned data to the reference's genome sequences (www.gramene.org). Simultaneously, the DESeq package (ver 2.1.0) was used with a false discovery rate (FDR) ≤ 0.005 and the absolute value of the log2 (fold change) with RPKM ≥1 as the threshold to detect differentially expressed genes (DEGs). Besides, the GOseq R package with P 0.05 and WEGO software were used to conduct gene ontology (GO) enrichment, while KEGG pathways were employed for DEG analysis with a FDR ≤ 0.05 as signi cant levels.

Statistical Analysis
Values were expressed as means ± standard deviations (n=6) and analyzed using two-way ANOVA test followed by the Tukey's Multiple Comparison Test with P < 0.05.

Mutation of OsGUN4 Performed Aberrant Starch Metabolism
Previous studies revealed the positive effects of OsGUN4 mutation on photosynthetic capacity during vegetative stages (Zhou et al. 2006; Wu et al. 2007), but no detail was focused on the relationship between photosynthetic products, e.g. sucrose, and starch biosynthesis. To determine the effects of OsGUN4 mutation on starch biosynthesis during vegetative stage, the carbon metabolites and relative starch biosynthetic enzymes were investigated in seedlings (Fig. 2, Fig. 3, Additional le 1: Table S2 and  Table S3). Compared to the wild-type, both of the sucrose and amylose contents were increased in gun4 epi (Fig. 2), while the fructose, glucose, total starch and protein contents were decreased (Fig. 2), suggesting that the OsGUN4 mutation promoted to accumulate the sucrose but decrease starch synthesis.
To explain why the OsGUN4 mutation led to abnormal starch metabolism existed, the seedlings were treated with exogenous sucrose. In the wild-type, contents of the sucrose and amylose were reduced, whereas the fructose, glucose, total starch and protein contents were increased after treatment (Fig. 2, Additional le 1: Table S2). However, in gun4 epi , no difference was detected before and after treatment, but both of sucrose and amylose concentration were higher than that in the wild-type, whereas the fructose, glucose, total starch and protein contents were lower, indicating the retardative transformation of sucrose to starch in gun4 epi leaves (Fig. 2).
We next treated seedlings with exSuc added with NF (the agent for blocking photosynthesis, causing the gun phenotype). After exposed to exSuc + NF , the contents of carbon metabolites were little induced compared with the control, but greatly inhibited in relative to the single sucrose treatment in wild-type, indicating that NF blocked the sucrose-induced signaling (Fig. 2). Nevertheless, no difference was still detected between the control and the sucrose treatment in gun4 epi , but there were signi cant differences for the treatment of sucrose added with NF compared to the other two treatment (Fig. 2). All these results suggested that OsGUN4 mutation in uenced the starch biosynthesis in leaves.

Mutation of OsGUN4 Deregulated Activities of Starch biosynthetic Enzymes
Dynamic activity changes of enzymes involved in starch biosynthesis were in accordance with the contents of carbon metabolites (Fig. 3, Additional le 1: Table S3). In consistent with the results as shown in Fig. 2, activities of AGPase, SSS and SBE showed signi cant increases, but activities of SS, SPS and GBSS were increased in gun4 epi (Fig. 3).
After exposed to exSuc, signi cant increased activities of AGPase, SSS and SBE, but decreased activities of SS, SPS and GBSS were showed in wild-type, whereas no difference was detected in gun4 epi , suggesting the retardative accumulation of starch from sucrose in gun4 epi (Fig. 3). However, after exposed to exSuc + NF , activities of the related enzymes (Fig. 3), was little induced compared with the control, but greatly inhibited in relative to the single sucrose treatment in wild-type, indicating that NF blocked the sucrose-induced signaling. Still, no difference was detected between the control and the exSuc treatment in gun4 epi , however, the sucrose added with NF treatment greatly affected the enzyme activities in relative to other treatments (Fig. 3). All these results suggested that OsGUN4 mutation in uenced activities of the starch biosynthetic enzymes in leaves.

Differently Expressed Genes Related to Starch Biosynthesis Revealed by RNA-seq
To analyze the detailed regulation of OsGUN4 on starch biosynthesis in vegetative leaves, RNA-seq was performed in the wild-type and gun4 epi . According to the mapping results using the metabolism overview pathways in MapMan, a total of 468 DEGs were identi ed between gun4 epi and WT by RNA-seq, with 203 genes being up-regulated and 265 down-regulated in gun4 epi (Fig. 4a, Additional le 1: Table S4).
Furthermore, the exogenous sucrose induced the gene expression for AGPase, GBSS, SSS, SBE, whereas reduced the expression of genes for SS and SPS in the wild-type (Fig. 4-6). After exposed to sucrose added with nor urazon, the gene expression for SS, SPS and GBSS still remained higher expression than that of sucrose treatment (Fig. 5-6), whereas gene expression for APGase, SSS and SBE were little increased in the wild-type (Fig. 6). However, sucrose treatment induced no signi cant difference with the control in gun4 epi (Fig. 5-6). But the sucrose supplement with nor urazon treatment intensi ed the trend of gene expression changes, and showed more enhanced dynamics in gun4 epi than that in the wild-type (Fig. 5-6). All these results were consistent with the results as shown in Fig. 2 and Fig. 3, suggesting the regulatory role of OsGUN4 on expression of starch biosynthetic genes.
Effects of OsGUN4 Mutation on Some Reported Sugar-dependent signals OsGUN4 is localized in plastid in our previous studies, so it is impossible for OsGUN4 to regulate gene expression as transcriptional factors (TFs) in nucleus. Thus, to clarify the OsGUN4-meidatd signals from plastid to nucleus for gene expression, the reported singlet oxygen- ) and sucrose-mediated (Lu et al. 2007) signals in rice were used for further investigation (Fig. 7, Additional le 1: Table S6). In WT, there were signi cant accumulation of 1 O 2 after exSuc treatments in relation to LL, but no obvious difference was detected between exSuc and exSuc + NF treatments (Fig. 7a). However, in gun4 epi , the accumulation of 1 O 2 showed no changes after exSuc treatments, but additional NF made lower 1 O 2 contents than that in WT, indicating that OsGUN4 might function in response to sucrose via 1 O 2 -mediated signals (Fig. 7a). Moreover, accumulations of proto in WT showed positive relation to the 1 O 2 contents, whereas more proto accumulated in gun4 epi , especially after exposed to exSuc + NF (Fig. 7b). These results suggested GUN4 and proto was essential for the generation of 1 O 2 -mediated signals. Furthermore, to clarify the possible connection of 1 O 2 -mediated signals to TFs, three reported TFs, including starch biosynthetic bZIP58, and sucrose-dependent NAC36 and MYB14, were further investigated in seedlings. Expression of OsbZIP58 was signi cantly decreased in gun4 epi , but, after sucrose treatment, there was no obvious expression difference of bZIP58 in gun4 epi (Fig. 7c). However, in wild-type, expression of bZIP58 was signi cantly induced by sucrose, while showed no obvious changes with CK after addition of NF (Fig. 7c). Moreover, exSuc induced enhanced the expression of NAC36 and MYB14, but additional NF greatly relieved these changes in WT (Fig. 7d and Additional le 1: Fig. S4). Nonetheless, in the absence of OsGUN4, expression of NAC36 and MYB14 performed no response to exSuc, and the phenomenon would be aggravated with supplement of NF in gun4 epi (Fig. 7d and Additional le 1: Fig. S4). Thus, it can be seen that expression of bZIP58, NAC36 and MYB14 showed consistent trends with contents of 1 O 2 and proto, suggesting the possible connection of 1 O 2 and these three examined TFs.
However, genes encoding Snf1-related protein kinases (SnRK1s) and its down-stream TF of MYBS1, were signi cantly down-regulated after exSuc treatments in WT, but could be restored to higher expression with the additional NF ( Fig. 7e and f). The similar phenomenon was also showed in gun4 epi , while the expression of SnRK1A and MYBS1 in gun4 epi were greatly induced compared to that of WT under the exSuc + NF treatments ( Fig. 7e and f). These results indicated that no direct evidence suggested the connection of 1 O 2 -mediated signals to SnRK1A-mediated signals, but that GUN4 performed roles in the SnRK1A-mediated signals, possibly via the accumulations of sugars, e.g. glucose or fructose.

OsGUN4 is Involved in Regulation of Starch Biosynthesis
In plant cells, plastids display a high morphological and functional variations, and include four major forms of etioplast, chloroplast, chromoplast and amyloplast (Liebers et al. 2017). Despite displaying diverse and tissue-dependent functions, each differentiated form of plastid shares a set of genomes (Lopez-Juez and Pyke 2005). Chloroplasts are the location for photosynthesis and biosynthesis of transient starch (Bahaji et al. 2014). Also, the aberrant chloroplasts usually would cause abnormal photosynthesis and starch metabolism (Bahaji et al. 2014;Liebers et al. 2017). The OsGUN4 mutation performed aberrant chloroplast morphology as reported previously (Zhou et al. 2006), and also indeed reduced the accumulation of starch at here (Fig. 2). Nonetheless, the mutation of OsGUN4 did not cause the decrease of sucrose derived from photosynthesis in gun4 epi , and inversely, the OsGUN4 mutation led to the accumulation of sucrose (Fig. 2), which is partly related to the positive effects of OsGUN4 mutation on photosynthetic capacity during vegetative stages (Zhou et al. 2006;Wu et al. 2007). On the other hand, this is due to the deregulated enzyme activities involved in starch biosynthesis (Fig. 3). For example, in gun4 epi , the enhanced activities of SS and SPS was responsible for the accumulation of sucrose (Fig. 3), whereas the decreased AGPase, SSS and SBE activities made neglect effects on starch synthesis (Fig. 3). All these results suggested that OsGUN4 mutation blocked the accumulation of starch from sucrose in leaves.
Generally speaking, over-accumulations of starch biosynthetic intermediates, i.e. ADPglucose and sucrose would result in photo-oxidative stresses (Ragel et al. 2013;Guo et al. 2017). For example, the mutation of TaSSIVb-D in wheat induced the reduction of starch granule number and photosynthetic e ciency, this may be attributed to high contents of the substrate ADPglucose (Ragel et al. 2013;Guo et al. 2017). This is consistent with the results as shown in Fig. 4 and Fig. 6, which could also explain the enhanced AGPase activity and increased expression of OsSSIV in gun4 epi . Besides, the addition of exogenous sucrose also indicated that GUN4 played a role in the normal synthesis of starch. Exogenous sucrose could greatly promote the transient starch and protein biosynthesis in wild-type, but could not induce the accumulation of sucrose and amylose in gun4 epi (Fig. 2). This was due to the dynamic activity changes of enzymes involved in starch biosynthesis, which were in accordance with the contents of metabolites (Fig. 3).
Furthermore, NF treatment is usually used for explore the uncoupled phenomenon of PhANGs transcriptional levels from chlorophyll accumulation in Arabidopsis (Susek et al. 1993) and C. reinhardtii (Formighieri et al. 2012). Our previous studies also revealed that the mutation of OsGUN4 deregulate transcription of PhANGs depending on disruption of 1 O 2 -induced signaling pathway in rice ). Here, after NF treatment, the induction by exogenous sucrose were nearly eliminated or weakened in the wild-type, whereas the OsGUN4 mutation aggravated no response to sucrose signals in gun4 epi (Fig. 2-3, Additional le 1: Fig. S3-5). All these results suggested OsGUN4 functions in response to sugar signals during starch biosynthesis.

Roles of OsGUN4 in the Regulation of Starch Biosynthesis
Functioning of the plastids requires cooperation of plastid genes and nuclear genes, which could reach the balance of photosynthesis and starch biosynthesis (Chan et al. 2016). Although GUN4 have been revealed in the plastid-to-nucleus signaling pathway ), it has not yet been reported its similar functions on starch biosynthesis. As is shown in preceding part of the text, OsGUN4 indeed function in the starch biosynthesis, and the OsGUN4 mutation also greatly deregulated many genes for key enzymes in starch biosynthesis (Fig. 4-5, Additional le 1: Fig. S5). Thus, we can conclude that OsGUN4 may regulate the genes encoding key enzymes in starch biosynthesis. There are mainly two ways to employ the regulation, by tetrapyrrole intermediates and by sucrose signals.
Inhibitors of plastid gene expression could repress amyloplast differentiation and starch biosynthesis in tobacco (Nicotiana tabacum) Bright Yellow-2 (BY2) cultured cells (Enami et al. 2011). This indicated a plastid-to-nucleus retrograde signals from plastid gene expression to the regulation for expression of nuclear starch biosynthesis genes, partly mediated by tetrapyrrole intermediates, i.e., heme (Enami et al. 2011). In our previous studies, the OsGUN4 mutation greatly affect tetrapyrrole intermediates, including heme, Mg-Proto and Proto in rice ). The blocking of photosynthesis and starch biosynthesis in gun4 epi also illustrated this from Suc added with NF treatment (Fig. 1-3, Additional le 1: Fig. S3-5), suggesting the suppressive signals from plastid to nucleus to promote starch biosynthesis.
Regulation of sucrose signals on starch biosynthesis can realized via transcription factors in cereal crops, e.g. NAC36 ), MYB14 (Zhang et al. 2014). OsGUN4 is localized in plastid in our previous studies, so it is impossible for OsGUN4 to regulate gene expression as TFs in nucleus. Instead, OsGUN4 might function in regulation of genes participating in starch biosynthesis via transcription factors, e.g. bZIP58 (Fig. 7). RNA-seq and RT-qPCR assays revealed that many genes for key enzymes in starch biosynthesis were signi cantly down-regulated in gun4 epi , including including the transcription factor of bZIP58 and target genes of OsBEIIb and OsSSI, which are vital for the formation of amylopectin and starch granules, while displayed up-regulated expression of OsSSIIIa and OsGBSSI that promotes the formation of amylose (Fig. 3, Fig. 6 and Additional le 1: Fig. S3). Thus, the retardative transformation from sucrose to starch mostly depends on the deregulated expression of genes for starch biosynthetic enzymes.

OsGUN4 Might Serve as a Broker of the Sugar-dependent Signals to Regulate Starch Biosynthesis
Sugars not only play key roles in metabolism and structural constituents of plant cells but also serve as essential components of signaling pathway in sugar responses (Rolland et al. 2006;Lu et al. 2007). Here, as shown in above, the OsGUN4 mutation led to accumulation of sucrose (Fig. 2), attributing to the seriously shrinking activities of starch biosynthetic enzymes, especially for AGPase (Fig. 3), which is the key regulatory enzyme of starch biosynthesis (Tiessen et al. 2003). The reduced activity of AGPase would retard the carbon ow from sucrose to starch in the scarce presence of OsGUN4 (Fig. 8). This could be concluded from results of exSuc +NF in Fig. 3. Although extra NF caused the decreased AGPase activity in both of WT and gun4 epi , while it made the much more seriously downregulated activities of AGPase in gun4 epi (Fig. 3). However, genes encoding the large subunit of AGPase, e.g. AGPL1, AGPL3 and AGPL4, performed enhanced expression in gun4 epi compared to wild-type (Fig. 6), displaying the uncoupled phenomenon that the inconsistent performance of chloroplast and nucleus. Correspondingly, many genes for SS, SPS, SSS and SBE also showed similar phenomenon (Fig. 5-6 and Additional le 1: Fig. S3 Interestingly, all of them were suggested to be as the regulator of genes encoding AGPase in rice (Wang et al. 2013) or in maize Zhang et al. 2014). Thus, genes encoding bZIP58, as well as genes for the orthologs of NAC36 and MYB14 in rice, were also investigated at here (Fig. 7 and Additional le 1: Fig. S4). All of bZIP58, NAC36 and MYB14, not only showed insensitive response to the sucrose but also would be aggravated immunity to sucrose with additional NF in gun4 epi (Fig. 7d and Additional le 1: Fig. S4). Correspondingly, downstream genes of the examined TFs, including AGPL2, SSI, AGPS1 and OsBEIIb, also displayed similar expression trends (Fig. 6). Therefore, we can conclude that OsGUN4 plays regulatory roles in starch biosynthesis, possibly via TFs, but the details of OsGUN4 mediated mechanism underlying sugar-regulated transcription remain mostly unclear (Fig. 8).
The tetrapyrrole intermediates, e.g. heme, proto, have been reported to be function as constitutes for signals from chloroplast to nucleus (Larkin, 2016;Tabrizi et al. 2016). We also previously reported one light-dependent singlet oxygen-mediated signal for regulating PhANGs in rice, attributing to the collision of GUN4 and proto . Here, we also demonstrated that 1 O 2 derived from GUN4 and proto might be also responsible for the sugar-dependent signals to regulate starch biosynthesis. On the one hand, accumulations of proto in WT showed positive relation to the 1 O 2 contents, whereas more proto accumulated in gun4 epi , especially after exposed to exSuc +NF (Fig. 7b), suggesting GUN4 and proto was essential for the generation of 1 O 2 -mediated signals. On the other hand, expression of bZIP58, NAC36 and MYB14 showed positive and consistent trends with contents of 1 O 2 and proto (Fig. 7), indicating the possible connection of 1 O 2 and these three examined TFs.
Additionally, SnRK1A functions upstream from the interaction between MYBS1 and αAmy3 in the sugar signaling cascade in rice (Lu et al. 2007). In this study, after exposed to exSuc, genes encoding SnRK1A and MYBS1 were signi cantly down-regulated both in WT and gun4 epi mutant, but could be restored to higher expression with the additional NF ( Fig. 7e and f), with the decreased sucrose contents but accumulated glucose (Fig. 2). The repression of glucose on SnRK1A was consistent with the results as shown in Lu et al. 2007. However, the expression of SnRK1A and MYBS1 were greatly induced in gun4 epi compared to that of WT under the treatment of exSuc +NF (Fig. 7e and f), depending on the greatly decreased glucose (Fig. 2). Nonetheless, feeding of sucrose have been reported to prevent AGPase redox inactivation in potato tubes (Tiessen et al. 2003), and we here also found that feeding of sucrose did cause the enhanced AGPase in WT, but it seemed to be diminished in gun4 epi mutant, especially after the additional feeding of NF (Fig. 3). Thus, it can be concluded that GUN4 also performed roles in the SnRK1A-mediated signals, possibly via the accumulations of sugars, e.g. glucose or sucrose (Fig. 8).

Conclusion
In summary, we demonstrated that OsGUN4 serve as a broker to activate 1 O 2 -mediated signals in the sugar signaling cascade, functioning upstream from three examined TFs, e.g. bZIP58, NAC36, MYB14, and that OsGUN4 performed roles in the SnRK1A-mediated signals, possibly through the accumulations of sugars, e.g. glucose or sucrose. Consequently, OsGUN4 plays regulatory roles in starch biosynthesis via mediating biosynthetic gene expression and enzyme activity.