The transcription factor IbNAC29 positively regulates the carotenoid accumulation in sweet potato

Abstract Carotenoid is a tetraterpene pigment beneficial for human health. Although the carotenoid biosynthesis pathway has been extensively studied in plants, relatively little is known about their regulation in sweet potato. Previously, we conducted the transcriptome database of differentially expressed genes between the sweet potato (Ipomoea batatas) cultivar ‘Weiduoli’ and its high-carotenoid mutant ‘HVB-3’. In this study, we selected one of these candidate genes, IbNAC29, for subsequent analyses. IbNAC29 belongs to the plant-specific NAC (NAM, ATAF1/2, and CUC2) transcription factor family. Relative IbNAC29 mRNA level in the HVB-3 storage roots was ~1.71-fold higher than Weiduoli. Additional experiments showed that the contents of α-carotene, lutein, β-carotene, zeaxanthin, and capsanthin are obviously increased in the storage roots of transgenic sweet potato plants overexpressing IbNAC29. Moreover, the levels of carotenoid biosynthesis genes in transgenic plants were also up-regulated. Nevertheless, yeast one-hybrid assays indicated that IbNAC29 could not directly bind to the promoters of these carotenoid biosynthesis genes. Furthermore, the level of IbSGR1 was down-regulated, whose homologous genes in tomato can negatively regulate carotene accumulation. Yeast three-hybrid analysis revealed that the IbNAC29-IbMYB1R1-IbAITR5 could form a regulatory module. Yeast one-hybrid, electrophoretic mobility shift assay, quantitative PCR analysis of chromatin immunoprecipitation and dual-luciferase reporter assay showed that IbAITR5 directly binds to and inhibits the promoter activity of IbSGR1, up-regulating carotenoid biosynthesis gene IbPSY. Taken together, IbNAC29 is a potential candidate gene for the genetic improvement of nutritive value in sweet potato.


Introduction
Carotenoids are pigments, widely distributed in nature, and are divided into two groups: (i) carotenes including lycopene and α/β/γ -carotene, and (ii) xanthophyll like lutein, zeaxanthin, and violaxanthin [1]. Over 750 natural carotenoids have been found in plants, algae, fungi, and bacteria [2,3]. Interestingly, carotenoids not only are crucial in these organisms that can synthesize them, but also in animals and humans. Humans must obtain carotenoids in their diet because their body cannot synthesize them [4,5].
The plant NAC (NAM, ATAF1/2, and CUC2) protein family is involved in diverse biological processes, including lateral root formation, secondary cell wall synthesis, and vegetative organ and fruit development [13,14]. Overexpression of SlNAC1 decreases the levels of β-carotene, lycopene, and total carotenoid, while increasing the lutein content in tomato (Solanum lycopersicum) [14]. In SlNAC4 RNA interference transgenic fruits, the total carotenoid level is significantly reduced after the break (B) stage [13,15]. Similarly, in the B + 3 and B + 10 stages of the NAC transcription factor SlNAC3 mutant, nor-like1, carotenoid levels are also significantly decreased [16]. On the contrary, overexpression of SlNAC-NOR significantly accelerates the fruit ripening process and produces higher carotenoid levels [17].
Besides, the MYB transcription factors also are important in regulating carotenoid biosynthesis. Based on the number of MYB conserved domains, MYBs are divided into (i) R1-(including one MYB domain); (ii) R2R3-(including two MYB domains); and (iii) R1R2R3-type MYB (including three MYB domains) subgroups. At present, mostly R2R3-type MYBs are reported to regulate carotenoid biosynthesis. For example, overexpressing AdMYB7 causes the accumulation of carotenoids and chlorophyll in kiwifruit [18]. Conversely, downregulating the R2R3-MYB transcription factor RCP1 (Reduced carotenoid pigmentation 1) expression reduces carotenoid content in Mimulus lewisii flowers [19]. Overexpression of CrMYB68 (Citrus reticulate) negatively regulates the expression of NbBCH2 and NbNCED5 to suppress the transformation of α-and β-branch carotenoids in tobacco leaves [20]. Moreover, MYB transcription factors form complexes with other proteins to participate in pigment biosynthesis. In Medicago truncatula, the MtWP1-MtTT8-MtWD40-1 complex regulates f lower pigmentation via the anthocyanin and carotenoid biosynthesis [21].
According to previous studies, STAY-GREEN (SGR) is an evolutionarily conserved chloroplast-targeted protein in higher plants which works in carotenoids biosynthesis, chlorophyll degradation and senescence [22,23]. Silencing the LeSGR1 (Lycopersicon esculentum) expression inhibits chlorophyll degradation in the leaves and fruits of tomato. Interestingly, SlSGR1 regulates lycopene and βcarotene accumulation by interacting directly with SlPSY1, a key carotenoid biosynthesis enzyme gene [24].
Sweet potato (Ipomoea batatas (L.) Lam. [2n = B 1 B 1 B 2 B 2 B 2 B 2 = 6x = 90]) provides carbohydrates and carotenoids for humans and is one of the most important food crops across the world. Sweet potato, especially the orange-f leshed cultivars, contains high levels of β-carotene, which could combat vitamin A deficiency [25,26]. In this study, we found that overexpression of IbNAC29 significantly increases the carotenoid content. We also demonstrated that IbNAC29 participates in the carotenoid biosynthesis by forming a regulatory module with IbMYB1R1 (R1-type MYB) and IbAITR5. Moreover, IbAITR5 represses the transcription of IbSGR1. Our results further indicated that IbNAC29 might enhance this repression, thus resulting in the carotenoid accumulation.

Results
IbNAC29 is a potential candidate gene for regulating the carotenoid biosynthesis pathway Previously, we performed RNA sequencing analyses on sweet potato cultivar Weiduoli and its high-carotenoid mutant HVB-3 ( Fig. 1a) to identify the differentially expressed genes [27]. Among these genes, the expressions of three NAC transcription factor genes, including IbNAC29, IbNAC74, and IbNAC87, were upregulated in HVB-3 [27]. As shown in Fig. 1b, IbNAC29 is homologous to the NAC transcription factor SlNOR-like1. Regulation of carotenoid biosynthesis by SlNOR-like1 in tomato has been reported recently [16]. Furthermore, IbNAC29 was widely expressed in the leaf, stem, and root tissues of HVB-3 (Fig. 1c). Quantitative real-time PCR (qRT-PCR) analysis showed mRNA level in the storage roots of HVB-3 was ∼1.71-fold higher than Weiduoli, thereby showing its potential link with carotenoid biosynthesis (Fig. 1d). Therefore, we selected IbNAC29 for subsequent analyses.
The coding sequence of IbNAC29 was 849 bp and contained three exons and two introns, encoding a protein of 282 amino acids (Fig. 1e). Based on the NCBI's Conserved Domains Database [28], N-terminal region of IbNAC29 contains a highly conserved NAM DNA-binding domain (Fig. 1f).

IbNAC29 is nuclear-localized and can function in transcriptional activation
To further study the subcellular localization of IbNAC29, we expressed the IbNAC29-GFP fusion protein in protoplasts. As a control, empty GFP plasmid was transfected into protoplasts. As shown in Fig. 2, GFP itself was distributed in the nucleus and the cytoplasm as expected, whereas the fusion protein IbNAC29-GFP was nuclear-localized (Fig. 2a). Furthermore, the position of green f luorescence from the IbNAC29-GFP fusion protein merged with the red f luorescence from the nuclear marker ARF1-mCherry [29], suggesting that IbNAC29 localizes to the nucleus.
Next, we used the transient expression system to investigate whether IbNAC29 acts as a transcriptional activator. We coexpressed the effector and reporter vectors in protoplasts, and quantified the luciferase activity after 16 h incubation. The results showed that the luciferase activity is significantly increased when IbNAC29 is co-expressed (Fig. 2b), thus indicating that IbNAC29 is a transcriptional activator.

Overexpression of IbNAC29 enhances carotenoid levels in the storage roots of sweet potato
To further investigate whether IbNAC29 regulates carotenoids in sweet potato, we generated IbNAC29-overexpression (IbNAC29-OE) plants by Agrobacterium-mediated transformation of sweet potato variety Lizixiang (Figs S1 and S2, see online supplementary material). After examining the IbNAC29 mRNA levels in these plants using qRT-PCR, we selected three lines (OE-2, OE-7, and OE-23) with the up-regulated IbNAC29 mRNA levels for further study (Fig. S1, see online supplementary material). Cross-sectional f lesh samples of the transgenic lines were slightly yellower and had orange spots relative to the wild type (WT) (Fig. 3a).
Because carotenoids are stored in plastids [30,31], we next analysed the plastids in the storage roots of transgenic IbNAC29-OE using transmission electron microscopy (TEM). The number of carotenoid globules in the IbNAC29-OE plants was significantly increased compared with the WT (Fig. 3a), suggesting the high levels of carotenoids accumulation in the storage roots of IbNAC29-OE plants.

Carotenoid biosynthesis-related genes are upregulated in IbNAC29-OE plants
Next, we used qRT-PCR assays to determine the expression of carotenoid biosynthesis-related genes in IbNAC29-OE plants and WT at storage root expansion stage. The carotenoid biosynthesis pathway is shown in Fig. 4a. In this study, we observed elevated mRNA levels of IbDXS, one MEP pathway gene (Fig. 4b) and four carotene biosynthesis genes (IbGGPPS, IbPSY, IbLCYE, and IbLCYB) ( Fig. 4c-f).
Previous research has shown that CYP97A (cytochrome P450 monooxygenase) works synergistically with CYP97C to hydroxylate α-carotene into lutein [32][33][34]. Both IbCYP97A3 and IbCYP97C1 are elevated in IbNAC29-OE. Therefore, the upregulated IbCYP97A3 and IbCYP97C1 might lead to the lutein accumulation in IbNAC29-OE plants ( Fig. 3c and 4g-h). Interestingly, we found increased zeaxanthin and capsanthin levels, but a decreased violaxanthin level. The expression of IbBCH, IbZEP, and IbCCS was also activated in IbNAC29-OE. We thus proposed that the decreased violaxanthin level may be because of its conversion to capsanthin under the high IbCCS expression ( Fig. 3e-g and 4i-k). Therefore, our results suggested that the upregulation of carotenoid biosynthesis genes causes the carotenoid accumulation in the storage roots of transgenic IbNAC29-OE sweet potato.

IbNAC29 could not bind to the promoters of carotenoid biosynthesis-related genes
When IbNAC29 was overexpressed in the sweet potato, the genes for carotenoid biosynthesis were significantly elevated in IbNAC29-OE. Next, we performed the yeast one-hybrid (Y1H) experiment to investigate the potential relationship of IbNAC29 and the promoters of the above genes. The promoter fragments of IbGGPPS, IbPSY, IbLCYE, and IbLCYB were independently amplified by PCR using genomic DNA as the template and cloned into the pLacZi2μ vector. The yeast activation domain (AD) was fused with the coding sequence of IbNAC29 to form the effector 42 AD-IbNAC29 construct. Both the reporter constructs IbNAC29-GFP was co-transformed with ARF1-mCherry, which was used as a nuclear marker. Bar = 10 μm. b Transactivation assay of IbNAC29 in protoplasts. The GAL4 BD empty vector was used as a negative control. The expression level of REN was used as an internal control. Error bars indicate SD (n = 3). * * indicates a significant difference from that of pBD at P < 0.01, by Student's t-test. and the effector 42 AD-IbNAC29 were cotransformed into yeast; 42 AD alone as a negative control. As shown in Fig.  S4 (see online supplementary material), IbNAC29 protein did not bind to these promoters. These results remind us that IbNAC29 may indirectly inf luence carotenoid biosynthesis via other factors.

IbNAC29 forms a regulatory module with IbMYB1R1 and IbAITR5
To investigate the possible interacting partners of IbNAC29 involved in carotenoid biosynthesis, we screened the sweet potato yeast two-hybrid (Y2H) library. Among these potential interacting proteins, we identified an R1-type MYB1 protein IbMYB1R1. Previous studies have shown that R2R3-type MYB, along with other factors, form a regulatory complex which affects anthocyanin biosynthesis [21,35,36]. Through yeast two-hybrid library screening, we isolated a IbMYB1R1-interacting protein IbAITR5. IbAITR5 belongs to a novel family of transcription factors, working as a member of ABA-induced transcription repressors (AITRs). The Y2H assays revealed that although IbNAC29 and IbAITR5 individually interacted with IbMYB1R1, there was no interaction between IbNAC29 and IbAITR5 (Fig. 5a). Using the yeast three-hybrid (Y3H) assays, we also observed that IbNAC29, IbMYB1R1, and IbAITR5 apparently formed a regulatory module (Fig. 5b). These interactions among IbNAC29, IbMYB1R1, and IbAITR5 were verified in the leaf epidermal cells of Nicotiana benthamiana using bimolecular f luorescence complementation (BiFC) assays. We observed a sharp yellow f luorescence in the nucleus when IbNAC29-nYFP or IbAITR5-nYFP was co-expressed with IbMYB1R1-cYFP, while negative controls showed no YFP f luorescence signal (Fig. 5c). Furthermore, we found that the IbMYB1R1 and IbAITR5 proteins were localized in the nuclei of the protoplasts (Fig. S5, see online supplementary material), which was consistent with the location of IbNAC29, thereby suggesting that IbNAC29, IbMYB1R1, and IbMYB1R1 may form a regulatory module and function in the nucleus.
Next, we used co-immunoprecipitation (co-IP) assays to investigate the IbNAC29-IbMYB1R1 and IbMYB1R1-IbAITR5 interactions in vivo. We isolated the total proteins co-expressed by IbMYB1R1-Myc with HA-IbNAC29 or HA-IbAITR5 in the leaf epidermal cells of Nicotiana benthamiana, and incubated them with anti-c-Myc agarose beads. We detected HA-IbNAC29 and HA-IbAITR5 in the immunoprecipitated proteins, but not in the negative control ( Fig. 5d-e). These experiments further indicated that IbMYB1R1 physically interacts with IbNAC29 and IbAITR5 in planta, confirming the previous results.
Taken together, these results confirmed that IbNAC29 could interact with IbMYB1R1, which forms an intermediate bridge with IbAITR5 to potentially form the IbNAC29-IbMYB1R1-IbAITR5 regulatory module.

IbAITR5 directly binds to the IbSGR1 promoter and represses its transcript activity
We first examined the relative mRNA level of the SGR1homologous gene IbSGR1 in IbNAC29-OE plants using qRT-PCR. qRT-PCR analysis revealed that relative IbSGR1 mRNA level was strongly reduced in the IbNAC29-OE plants (Fig. S6a, see online supplementary material), suggesting that IbNAC29 may negatively regulate IbSGR1.
To test the hypothesis, we conducted Y1H assays to explore the relationship between the IbNAC29-IbMYB1R1-IbAITR5 regulatory module and the IbSGR1 promoter. Interestingly, we found that IbAITR5, rather than IbNAC29 and IbMYB1R1, directly binds to the IbSGR1 promoter (Fig. 6a). Then, we used the dual-luciferase reporter assays to assess the luciferase activity of IbSGR1 driven by the IbAITR5. These results revealed that when IbSGR1pro:LUC was co-transformed with IbAITR5, IbAITR5 inhibited the IbSGR1 promoter activity. Therefore, our data demonstrated that IbAITR5 represses the IbSGR1 promoter activity by binding to its promoter (Fig. 6b).
Next, we used the electrophoretic mobility shift assay (EMSA) and chromatin immunoprecipitation-quantitative polymerase chain reaction (ChIP-qPCR) assays to validate whether IbAITR5 could bind to the IbSGR1 promoter. In the EMSA assay, IbAITR5-GST bound to a 28 bp fragment of IbSGR1 in vitro (Fig. 6c). Additionally, the ChIP-qPCR assays confirmed that IbAITR5 also binds in vivo to the IbSGR1 promoter (Fig. 6d). Thus, our results collectively suggested that IbAITR5 represses IbSGR1 transcription by directly binding to its promoter.

IbNAC29-IbMYB1R1-IbAITR5 regulatory module regulates carotenoid biosynthesis
To further investigate how IbNAC29, IbMYB1R1, and IbAITR5 affected the transcriptional activity of IbSGR1, we conducted the dual-luciferase reporter assays. As shown in Fig. 7a  IbSGR1pro vector co-transient in protoplasts. However, in the presence of IbMYB1R1, IbNAC29 enhanced the inhibitory activity of IbAITR5 on the IbSGR1 promoter (Fig. 7a). It has been reported that SlSGR1 inf luences the SlPSY1 expression pattern in tomato [24]. Furthermore, the dualluciferase assays revealed that the IbSGR1 also inf luences the expression of IbPSY (Fig. 7b). The repression of the IbPSY1 gene in the presence of IbSGR1 expression is in accordance with previous studies [24]. Therefore, our results suggest that the IbNAC29-IbMYB1R1-IbAITR5 regulatory module potentially regulates carotenoid biosynthesis via the regulation of IbPSY1.

Discussion
Carotenoids are tetraterpenoids molecules that play pivotal roles in photosynthesis, pigmentation, and development. Despite an in-depth mechanistic basis for understanding the carotenoid biosynthesis, relatively little is known about how this pathway is transcriptionally regulated. Previously, we conducted the transcriptome database of differentially expressed genes between the Weiduoli and its high-carotenoid mutant HVB-3 [27]. Among these genes, NAC transcription factors IbNAC29, IbNAC74, and IbNAC87 were upregulated in HVB-3. In this study, we selected and characterized demonstrated overexpression of IbNAC29 increased the levels of various carotenoids in the storage roots, including α-carotene, lutein, β-carotene, zeaxanthin, and capsanthin (Fig. 3). Indeed, the carotenoid biosynthetic gene expression (IbDXS, IbGGPS, IbPSY, etc) was also up-regulated in IbNAC29 transgenic plants. This could potentially explain why carotenoid accumulation is elevated. Previous reports have suggested that overexpression of PmDXS and IbGGPS increased the carotenoid content in Arabidopsis [37,38]. Furthermore, overexpressing LCYE elevates the carotenoid lutein level in Arabidopsis leaves [39]. Also, overexpression of IbLCYB2 increases the carotenoid content in the sweet potato's storage roots [40]. In plants, the SGR gene encodes the key enzyme for chlorophyll degradation [23]. In tomato, SlSGR1 reportedly regulates chlorophyll degradation [22,24]. Silencing SlSGR1 inhibits chlorophyll degradation, resulting in the retention of a green phenotype. As a matter of fact, SlSGR1 regulates the lycopene accumulation in tomato by directly inhibiting the activity of a key carotenoid biosynthesis enzyme, SlPSY1 [24]. Overexpression of CsPSY enhances carotenoid accumulation in Hongkong kumquat [41]. Both CsSGRa and CsSGRb interact with CsPSY1 to inhibit the citrus carotenoid biosynthesis, chlorophyll degradation and carotenoid biosynthesis, which are highly conserved processes in plants [42]. Similarly, the overexpression of CsPSY enhances carotenoid accumulation in Hongkong kumquat [41]. Therefore, our result suggested that the upregulation of carotenoid biosynthesis genes might cause the accumulation in the carotenoids.
Previous studies have reported that the tomato NAC transcription factor SlNOR-like1 directly binds to the SGR1 promoter, thus regulating fruit ripening and carotenoid accumulation [16]. However, Y1H assay indicated IbNAC29 could not directly bind to the promoters of carotenoid biosynthesis-related enzymes. To explore the possible mechanism of IbNAC29 involved in carotenoid biosynthesis, we screened the sweet potato yeast twohybrid (Y2H) library. Among these potential interacting proteins, we identified an R1-type MYB1 protein IbMYB1R1. Previous studies have shown that R2R3-type MYB, along with other factors, form a regulatory complex which affects anthocyanin biosynthesis [21,35,36]. Through yeast two-hybrid library screening, we isolated a IbMYB1R1-interacting protein IbAITR5. In our study, the results showed that IbAITR5 could directly bind to the IbSGR1 promoter, inhibiting the expression of the IbSGR1 (Fig. 6). The mRNA level of IbSGR1 is down-regulated in IbNAC29-OE, which is consistent with its negative role in carotenoid accumulation. Although we detected enhanced carotenoids accumulation in the IbNAC29-OE storage roots (Fig. 3), we did not find any direct interaction between IbNAC29 and the IbSGR1 promoter (Fig. 6a). Therefore, our results suggested that IbNAC29 might have a different regulatory mechanism with SlNOR-like1, possibly because they belong to different clades in the evolutionary tree.
Through Y3H, EMSA, ChIP-qPCR, and dual-luciferase assay analyses, our study demonstrated that the IbNAC29-IbMYB1R1-IbAITR5 regulatory module mediates the carotenoids biosynthesis via protein-protein interactions to regulate the downstream target gene expression in sweet potato. It has been reported that AITRs are transcription repressors in plants [43], and we found that the IbAITR5 mRNA level in the IbNAC29-OE plants was upregulated (Fig. S6b, see online supplementary material). Thus, we proposed that IbNAC29 enhances the inhibitory activity of IbAITR5 by affecting its transcriptional activity. This leads to reduce the expression of the IbSGR1 (Figs 2a and 7a), resulting in further alleviation of the inhibition of IbSGR1 on mRNA level of the key carotene biosynthesis gene IbPSY. Up-regulated expression of IbPSY might lead to enhanced carotenoids accumulation in the storage roots (Fig. 8).
Altogether, our findings unveil the mechanism underlying the regulation of the carotenoids accumulation and provide new insights for genetic improvement in the sweet potato. To further understand the mechanisms that regulate carotenoid biosynthesis in staple crops, we will further identify the direct targets of IbNAC29 by combining transcriptome analysis with chromatin immunoprecipitation analysis in the future. Moreover, we will attempt to use the CRISPR/Cas9-based gene editing approach to further understand its role in the development of sweet potato.

Plant materials and growth conditions
Sweet potato cultivar Weiduoli with orange-f lesh and its high carotenoid mutant HVB-3 were used for RNA sequencing analyses. Sweet potato cultivar Lizixiang was used as the recipient for Agrobacterium-mediated transformation, which is a pale-yellow f lesh with low carotenoid content. Transgenic test-tube seedlings were grown on Murashige and Skoog medium at 28 • C with a 13-hlight/11-h-dark cycle. The transgenic plants were cultivated in the field of the experimental stations of China Agricultural University and adhered to normal agricultural practice.

Gene identification and sequence analysis
Total RNA was extracted using TRIzol reagent (Invitrogen, USA). Complementary DNAs (cDNA) were obtained using HiFiScript gDNA Removal cDNA Synthesis Kit (CwBio, Beijing, China) according to the manufacturer's protocol. The RACE (rapid amplification of cDNA ends) experiment was used to obtain the full-length cDNA sequence of IbNAC29. According to the EST sequence obtained from previous studies [27], the coding sequences of IbMYB1R1, IbAITR5, and IbSGR1 were obtained from Lizixiang using the homologous cloning method. DNAMAN software, MEGA 7.0 software, and the Splign tool were used to analyse amino acid sequence alignments, exon-intron, and phylogenetic relationships, respectively.

Sweet potato transformation and qRT-PCR analysis
The embryogenic suspension cultures of Lizixiang were transformed with the pCAMBIA1300-35S-IbNAC29-GFP vector via Agrobacterium-mediated transformation [44]. The transgenic sweet potato plants were selected using hygromycin as a selection marker. The plants were transferred to a greenhouse, planted in the nutrient vegetative soil, and then transplanted to the field for phenotype observation. The IbActin gene of sweet potato (AY905538) was used as the internal control for expression analysis by qRT-PCR assays [45,46]. The mRNA levels of genes were calculated by comparative CT method [47]. The experiment was conducted using three biological replicates consisting of pools of three plants. Values are means ± SD of three biological repeats.

Measurement of carotenoid contents
Carotenoids were extracted as described previously [37]. Three independent storage roots from each freshly harvested WT and IbNAC29-OE transgenic plant were mixed, respectively. Carotenoids and the relative contents were measured as previously described [48].

Transmission electron microscope
The storage roots of IbNAC29-OE and WT were fixed as previously described [40]. The number of carotenoid globules was observed using TEM (JEM-1230, Tokyo, Japan).

Yeast assays
In the Y1H assay, the open reading frames of IbNAC29, IbMYB1R1, and IbAITR5 sequences were separately cloned into the pB42AD vector. The promoter sequences of IbGGPPS, IbPSY, IbLCYB, IbLCYE, and IbSGR1 genes from Lizixiang were cloned separately into the pLacZi2μ vector. In short, various LacZ reporter plasmids were cotransformed with the pB42AD fusion constructs into EGY48 yeast strain. The pLacZi2μ reporter and pB42AD were co-transformed as negative controls. Transformants were grown on SD/−Trp-Ura dropout plates containing 5-bromo-4chloro-3-indolyl-β-D-galactopyranoside (X-Gal) for blue color development.

BiFC assay
Empty pSPYNE-35S or the pSPYCE-35S vector cloned with the IbNAC29, IbMYB1R1, and IbAITR5 coding sequences were transformed into the Agrobacterium tumefaciens strain EHA105. Combinations of pSPYNE and pSPYCE vectors, together with P19, were infiltrated into the Nicotiana benthamiana leaf epidermal cells. The YFP signal was observed by using a laser confocal scanning microscope at an excitation wavelength of 488 nm after 48 h growth (Olympus, Tokyo, Japan).

Dual-luciferase assay
Rice shoot protoplasts were isolated and used for the dualluciferase assays, as described previously [50]. For the transcriptional activity assay, the empty pBD vector was used as the negative control to measure the transcriptional activity of IbNAC29.
For the DNA-promoter interaction assay, the IbNAC29, IbMYB1R1, IbAITR5, and IbSGR1 coding sequences were cloned separately into the pGreenII 62-SK vector. The IbSGR1 and IbPSY promoters were cloned separately into the pGreenII0800-LUC vector. Firef ly luciferase (LUC) and Renilla luciferase (REN) activity levels were measured using a dual-luciferase reporter assay system (Promega, USA). Four technical replicates were conducted in the experiments.

EMSA
EMSA was performed according to the manufacturer's instructions (Thermo Fisher Scientific, USA). Glutathione beads purified recombinant GST-labeled IbAITR5 protein expressed in Escherichia coli Transetta (DE3). The NACRS element containing biotin-labeled probes synthesized by Tsingke (Beijing) were used as binding probes, while unlabeled probes were used as competing probes.

ChIP-qPCR analysis
The ChIP assay was carried out as described previously [46]. The plants of pSuper1300-IbAITR5-GFP were cut into pieces and immediately fixed with 1% (v/v) formaldehyde solution. Next, the samples were ground into fine powders under liquid nitrogen. StepOnePlus™ was used to analyse the enrichment of immunoprecipitated DNA. IbSGR1 promoter P2 fragment contained a NACRS element (sequence is ACGTGA), while P1 having no NACRS element served as the negative control. Four technical replicates were conducted in the experiments using. All the above primer sequences are shown in Table S1 (see online supplementary material).