Evolutionary analysis and functional characterization of SiBRI1 as a Brassinosteroid receptor gene in foxtail millet

Brassinosteroids (BRs) play important roles in plant growth and development. Although BR receptors have been intensively studied in Arabidopsis, those in foxtail millet remain largely unknown. Here, we show that the BR signaling function of BRASSINOSTEROID INSENSITIVE 1 (BRI1) is conserved between Arabidopsis and foxtail millet, a new model species for C4 and Panicoideae grasses. We identified four putative BR receptor genes in the foxtail millet genome: SiBRI1, SiBRI1-LIKE RECEPTOR KINASE 1 (SiBRL1), SiBRL2 and SiBRL3. Phylogenetic analysis was used to classify the BR receptors in dicots and monocots into three branches. Analysis of their expression patterns by quantitative real-time PCR (qRT-PCR) showed that these receptors were ubiquitously expressed in leaves, stems, dark-grown seedlings, roots and non-flowering spikelets. GFP fusion experiments verified that SiBRI1 localized to the cell membrane. We also explored the SiBRI1 function in Arabidopsis through complementation experiments. Ectopic overexpression of SiBRI1 in an Arabidopsis BR receptor loss-of-function mutant, bri1-116, mostly reversed the developmental defects of the mutant. When SiBRI1 was overexpressed in foxtail millet, the plants showed a drooping leaf phenotype and root development inhibition, lateral root initiation inhibition, and the expression of BR synthesis genes was inhibited. We further identified BRI1-interacting proteins by immunoprecipitation (IP)-mass spectrometry (MS). Our results not only demonstrate that SiBRI1 plays a conserved role in BR signaling in foxtail millet but also provide insight into the molecular mechanism of SiBRI1. Supplementary Information The online version contains supplementary material available at 10.1186/s12870-021-03081-8.

the interaction between SiBRI1 and SiBAK1 [25], but SiBRI1 in foxtail millet has not yet been characterized.
To obtain insights into the functions of BRs in the C4 model species foxtail millet, we conducted an evolutionary and functional examination of foxtail millet BR receptors. We identified four putative BR receptor genes in the foxtail millet genome, SiBRI1, SiBRL1, SiBRL2 and SiBRL3, and analysed their expression patterns and roles in the BR signaling pathway. Our findings showed that SiBRI1 could rescue the dwarf phenotype of bri1-116, and enhance the dephosphorylation of BZR1 in vivo to activate the BR signaling pathway. When SiBRI1 was overexpressed in foxtail millet, the plants showed droopy leaves, root development inhibition, and the expression of BR synthesis genes was inhibited. Furthermore, We also found 128 SiBRI1 interacting proteins through IP-MS, which laid the foundation for further study of SiBRI1 function.

Phylogenetic analysis of BRI1 family genes
We investigated whether the foxtail millet genome encodes BR receptors of the canonical BR signal transduction pathway. We identified foxtail millet BR receptors by using the AtBRI1 and OsBRI1 protein sequences in Phytozome and the Setaria italica Functional Genomics Database with the BLASTP algorithm [26]. We found four putative BR receptor genes in the foxtail millet genome, which were named SiBRI1, SiBRL1, SiBRL2 and SiBRL3 (Table S1).
We performed a phylogenetic analysis including SiBRI family members, the published BRI family genes (Table  S1), and the representative plants that had relatively complete annotated genome data and phylogenetic relationship information by the Angiosperm Phylogeny Group (APG) taxonomy [27] ( Figure S2). Because BRI1 belongs to the RLK superfamily, searching with a single hidden Markov model (HMM) cannot exclude the considerable redundancy. Therefore, the published BRI1 protein sequences were used to construct an HMM, and HMMER V 3.3 [28] was used to search for candidate genes with the complete protein sequence data of the species that hadn't indentified the BRIs family genes before. Notably, we found BRI1 family genes only in angiosperms. Finally, 98 BRI1 gene family members were identified and used for phylogenetic analysis. To map the phylogenetic relationships among these members, MUS-CLE alignment methods and neighbour-joining (NJ) phylogenetic inference methods were employed. In addition, the full-length BRI1 sequences were analysed separately (Fig. 1).
The plant BRI1 gene family members identified in the present study (Fig. 1A) could be divided into three , an neighbor-joining (NJ) phylogenetic tree was constructed using the full-length protein sequence alignments of BRI1 genes identified using MUSCLE in MEGAX. Bootstrap supports were indicated by the colour of the branch. The OTUs are labelled as follows: Dicotyledons (blue); Monocotyledons (red); Amborella trichopoda (black). The colour blocks indicate the types, with type I (pink), type II (green), and type III (yellow) denoted. (B), Gene structures of the BRI or BRL genes. The lengths of rectangles and lines are scaled according to the mRNA lengths. CDSs (green rectangles), UTRs (yellow rectangles), and introns (black line) are denoted subgroups: type I, type II and type III. The type I BRI1 family gene subgroup was the largest branch and accounted for 38.8% of the genes observed, the type II subgroup accounted for 33.7% of the genes, and the type III subgroup accounted for the remaining 27.5% of the genes. BRI1 belonged to type I, BRL1 and BRL3 belonged to type II, and BRL2 belonged to type III. The predominance of type I BRI1 genes may be related to the fact that these genes help regulate the whole process of plant growth and development, while BRL1 and BRL3 genes function mainly in the root [29]. Notably, BRL1 and BRL3 belong to the same type, while BRL2 belongs to a separate type. This result corresponds to the finding that the extracellular domain of BRL2 cannot interact with BR. In addition, the monocots and dicots were further divided into two clusters in each type (Fig. 1A).
SiBRI1 was predicted to have 53.96% identity with AtBRI1 and 79.40% identity with OsBRI1, its homologues in Arabidopsis and rice. Besides OsBRI1, ZmBRI1b, SbBRI1, ZmBRI1a, BdBRI1, HvBRI1, TaBRI1a, HvBRI1b belonged to a sub-branch of the phylogenetic tree with SiBRI1, and the homology is 86.8%, 86.7%, 83.2%, 81.3%, 81.5%, 81.5%, 80.3%, respectively. The protein sequence lengths of the BRI1 genes varied from 827 aa (CsBRI1 in Cucumis sativa) to 1288 aa (ZmBRL1 in maize), and most BRI1 genes had no introns (Fig. 1B). After cloning and sequencing of the flanking regions of SiBRI1 in Setaria italica (cultivar Yugu-1) with specific primers, the full-length cDNA was predicted to contain a long open reading frame that encoding a protein of 1118 aa. Alignment analysis indicated that SiBRI1 shares a conserved signal peptide, LRR, transmembrane (TM) domain and kinase domain with AtBRI1 and OsBRI1, and the critical amino acids for the kinase activity of AtBRI1 were also conserved in SiBRI1, suggesting that SiBRI1 is an active kinase ( Figure S3).

Tissue and subcellular localization of SiBRI1 and its orthologues
We used quantitative real-time PCR (qRT-PCR) to determine the expression patterns and transcript abundance of SiBRI1 and its orthologues in foxtail millet. SiBRI1 was universally expressed in the leaves (L), stems (S), darkgrown seedlings (DGS), roots (R), non-flowering spikelets (NFS), and dry seeds (DS) of foxtail millet ( Fig. 2A). SiBRL1, SiBRL2 and SiBRL3 were highly expressed in root, stems and dry seeds ( Fig. 2A). We speculate that SiBRI1 and its paralogues genes may play different roles in different tissues.
As mentioned above, a signal peptide in the N-terminus and a TM domain were predicted to be present in SiBRI1 ( Figure S3). To determine the subcellular localization of SiBRI1, we constructed a fusion protein, SiBRI1::eGFP, driven by a maize ubiquitin promoter and transformed Error bars indicate the mean ± standard deviation (SD). N = 3. Statistically significant differences are indicated by different lowercase letters (p < 0.05, one-way ANOVA with Tukey's significant difference test). (B), Confocal images indicate the localization of SiBRI1-eGFP in the roots of 3-day-old dark-grown seedlings overexpressing SiBRI1 with a GFP tag at the C-terminus. Scale bar = 20 µm it into foxtail millet. Then, we detected clear fluorescence signals on the plasma membrane in root meristem cells by laser confocal microscopy (Fig. 2B). The results showed that SiBRI1 is a cell membrane protein.

Functional analysis of SiBRI1 in Arabidopsis
To verify whether SiBRI1 encodes a BR receptor, we transformed the SiBRI1-encoding sequence under the control of the CaMV 35S promoter into Arabidopsis Columbia-0 (Col-0) and a BR-insensitive stunted Arabidopsis mutant, bri1-116. bri1-116 is produced by a point mutation of Glutamate at site 583 in the 21st LRR before the AtBRI1 island domain. This mutation results in early termination of the peptide chain and a phenotype with complete AtBRI1 deletion, severe plant dwarfism, shortened petioles, and shrunken and rounded leaves [14]. After SiBRI1 overexpression, the plant height, silique length, and leaf blade morphology in the bri1-116 (SiBRI1/bri1-116) transgenic lines were observed to be similar to Col-0, and the leaf curling and elongation exhibited by BR-activated plants were observed after SiBRI1 overexpression in Col-0 (SiBRI1/Col) plants ( Fig. 3A and S4). Previous studies have shown that when BR levels are low, BIN2 phosphorylates and inactivates BZR1 to inhibit plant growth; BRs promote growth by inducing the dephosphorylation of BZR1, a hallmark of active BR signaling pathway [11]. Therefore, we detected the phosphorylation levels of AtBZR1 in different transgenic plants. In Col-0, a weak band was observed for unphosphorylated AtBZR1, whereas in bri1-116, the band for unphosphorylated AtBZR1 was almost undetectable. When SiBRI1 was transferred into the Col-0 background or into bri1-116, the strength of the band for unphosphorylated AtBZR1 increased significantly (Fig. 3B), indicating that the BR signal had been activated.
Exogenous application of low levels of bioactive brassinolide (BL) promotes plant growth. To examine the responsiveness of SiBRI1 overexpression seedlings to exogenous BL application, we grew seeds on 1 / 2 Murashige and Skoog (MS) medium supplemented with BL at a series of concentrations (0, 5, 10 and 100 nM) for seven days and measured the primary root length. Without BL treatment, the root length of SiBRI1/bri1-116 was significantly greater than that of bri1-116 but approximately 38% less than that of Col-0, and the root length of SiBRI1/Col was shorter than that of Col-0. SiBRI1 /Col exhibited a phenotype of slightly shorten primary roots, similar to that generated following BR application, as BR treatment inhibited the lengthening of plant roots (Fig. 4A). SiBRI1/Col seedlings grown on 1 / 2 MS media containing 5 and 10 nM BL showed greater sensitivity to BL than Col-0 seedlings, and SiBRI1/bri1-116 plants showed greater sensitivity to BL than bri1-116 plants under 5, 10 and 100 nM BL treatment in a concentration dependent manner (Fig. 4B). These results indicated that the SiBRI1 over expression plants exhibited enhanced sensitivity to BL.
Propiconazole (PCZ) is a specific inhibitor of BR biosynthesis and inhibits hypocotyl elongation under both light and dark conditions in Arabidopsis and maize [30]. To determine the responses of SiBRI1 over expression seedlings to PCZ, we cultivated germinated seeds on 1 / 2 MS medium containing 0.25 µM PCZ in the dark for seven days. Under dark conditions, 0.25 µM PCZ decreased the hypocotyl length of wild-type seedlings by 67%. The hypocotyls of SiBRI1 over expression seedlings had a wavy, twisted phenotype. The hypocotyl length of bri1-116 seedlings was very short and showed no significant difference between the PCZ and non-PCZ treatments. SiBRI1/bri1-116 and SiBRI1/Col seedlings showed greater insensitivity to PCZ than Col-0 seedlings ( Fig. 4C, D, and S4). We also tested the phosphorylation levels of AtBZR1 in different transgenic plants and found that the phosphorylation of AtBZR1 in the transgenic plants was inhibited (Fig. 4E). All these results indicate that the SiBRI1 transgene successfully complements the BR-insensitive phenotype in bri1-116.

SiBRI1 regulates BR signaling in foxtail millet
To test whether SiBRI1 regulates BR signaling in foxtail millet, we overexpressed full-length SiBRI1 in Ci846 which is an easily transformed variety of S. italica and generated a pUbi:SiBRI1-eGFP transgenic (SiBRI1-OX) plant through callus transformation. At the 4-leaf stage, SiBRI1-OX seedlings in two independent lines, OX23 and OX14, had larger leaf angles and droopier leaves than Ci846 seedlings (Fig. 5A); notably, large leaf angles and drooping leaves have been reported to be specific phenotypes controlled by BR in foxtail millet [25]. Both of these phenotypes were dependent on the protein expression level of SiBRI1 (Fig. 5B). Notably, the transcript levels of SiBRI1 were downregulated after BR treatment in both leaves and roots ( Figure S5), consistent with our published RNA-Seq data [31].
Activation of the BR signaling pathway is known to inhibit the expression of BR synthesis genes via a feedback mechanism. To verify that SiBRI1 is the receptor of BR, a qRT-PCR assay was used to detect the responses of the BR synthesis genes SiCPD, SiD2, SiDWARF and SiDWF4 in the roots of different SiBRI1-OX transgenic lines in Ci846 ( Figure S6). The qRT-PCR results showed that the transcript levels of these genes were reduced by BR regulation and that the expression of these genes was lower in SiBRI1-OX plants than in Ci846 plants (Fig. 5C). Taken together, these results indicate that SiBRI1 overexpression activates the BR signaling pathway in foxtail Error bars indicate the mean ± standard deviation (SD). Statistically significant differences are indicated by different lowercase letters (p < 0.05, one-way ANOVA with Tukey's significant difference test) millet and that SiBRI1 is a receptor in the conserved BR signaling pathway.

SiBRI1 affects root growth and lateral root development in foxtail millet
As BRI1 is a vital positive modulator in the BR signaling pathway, BRI1-OX plants are sensitive to the BL growth response, i.e., the inhibition of root development in Arabidopsis and rice [32], as described above. To determine whether SiBRI1 similarly plays a positive regulatory role in foxtail millet, we investigated the responses of two independent SiBRI1-OX/Ci846 lines differing in their expression levels of SiBRI1 to 0.01 and 0.1 µM BL ( Fig. 6A and S6). BL at a concentration of 0.01 µM significantly inhibited root and leaf growth in OX23 and OX4 plants compared with wild-type plants. When the BL concentration was increased to 0.1 µM, leaf and primary root growth were also significantly inhibited in wild-type plants, but the inhibition was more obvious in OX23 and OX4 plants than in wild-type plants (Fig. 6A, B).
We also counted lateral roots in SiBRI1-OX/Ci846 plants following treatment with different concentrations of BL. SiBRI1-OX/Ci846 plants had more lateral roots than Ci846 plants (Fig. 6C). Seedlings grown on medium containing 0.01 µM BL showed short primary roots but had more lateral roots than control seedlings (no BL), whereas primary root length and lateral root number both decreased with increasing BL concentration in Ci846 plants (Fig. 6B-C). These results indicated that the sensitivity of lateral root development to the BR response differs from that of primary root elongation; specifically that primary root elongation is more sensitive than lateral root development to BL. However, the number of lateral roots in SiBRI1-OX/Ci846 plants was drastically decreased and was much lower than that in control plants when the media contained 0.01 µM BL (Fig. 6C). Therefore, we concluded that the SiBRI1-OX/ Ci846 plants were more sensitive to exogenous BR than the control plants. These findings also support that BRI1 as a positive regulator of BR signal transduction in foxtail millet.
PLETHORA 1 (PLT1) in Arabidopsis play an important role in regulating root development [33], we found that the expression of PLETHORA-LIKE 1 (SiPLT-L1, based on the homology with AtPLTs) in foxtail millet roots decreased after BR treatment, and the transcript level of SiPLT-L1 in SiBRI1-OX/Ci846 was significantly lower than Ci846 (Fig. 7). LATERAL ORGAN BOUND-ARIES DOMAIN16 (LBD16) play pivotal role in lateral root initiation [34], we also test the transcription of SiLBD16 (based on the homology with AtLBD16) under BL in the roots of SiBRI1-OX/Ci846 and Ci846 by qRT-PCR, and found that SiLBD16 significantly dropped under BL treatment, and the transcription of SiLBD16 in SiBRI1-OX/Ci846 was lower than Ci846, which was corresponding to the phenotype of SiBRI1-OX/Ci846 under BL (Figs. 7 and 6C). In conclusion, BR may effect root growth and lateral root development via SiPLT-L1 and SiLBD16.

The interaction proteins of SiBRI1 in foxtail millet
To understand the functions of SiBRI1 and the mechanisms underlying SiBRI1-mediated BR signaling specificity, we performed a proteomics study of SiBRI1-interacting proteins in foxtail millet. Microsomal proteins were extracted from the SiBRI1-OX/Ci846 plants treated with BL or not, or the non-transgenic Ci846 as negative control, and SiBRI1-eGFP and associated proteins were immunoprecipitated using GFP-Trap. Liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis identified 128 proteins that were co-immunoprecipitated in the SiBRI1-eGFP sample but not in the non-transgenic Ci846 control sample, 48 proteins that were co-immunoprecipitated from the SiBRI1-OX/Ci846 plants without BL treatment, 54 proteins that were co-immunoprecipitated only from the SiBRI1-OX/ Ci846 plants with BL treatment, and 26 proteins that  Table S2-4).
Gene Ontology (GO) annotation and enrichment analysis was performed for the SiBRI1 interaction proteins with BL treatment or not separately. To analyse the effective information of proteins more accurately, GO annotation was performed online using AgriGO (http:// bioin fo. cau. edu. cn/ agriGO/). For the 74 interaction proteins that found in the SiBRI1-OX/Ci846 plants without BL treatment, the proteins related with development, hormone, peptide metabobic and biosynthetic was enriched. When the BL treatment to the seedling, the SiBRI1 interaction proteins changed to related with nucleotide phosphorylation, glycolytic process, hormone, light stimulus, abiotic stimulus and so on (Fig. 8B, Table S5). We also found SiBSK7, a homologous gene of AtBSK7 in Arabidopsis, and BSKs has been reported to be the direct substrate of BRI1 in the BR signaling pathway [8]. In conclusion, our work lays a foundation for future studies on the function of SiBRI1 in foxtail millet.

Discussion
Foxtail millet has helped nourish human civilizations throughout Asian history, and it is still considered a staple food today. Unlike other staple crops (e.g., rice, wheat, and maize), foxtail millet can grow in arid or semiarid areas where the water supply and soil fertility are limited. With the increasing impact of global warming and loss of arable land due to human activity, foxtail millet is an excellent substitute crop that can help secure food safety for human societies in the future. However, compared with other popular staple crops, the average yield of foxtail millet is low. This leaves great opportunity for future improvement by using modern inbreeding technology.
In general, leaf angle, effective tiller number, number of grains per panicle, and thousand-grain-weight are the most important traits that determine the average yields of cereal crops. Studies of rice indicate that BRs play an important role in regulating these productivity-related traits. Although BR signaling has been extensively studied in several species, BR signaling in foxtail millet is still largely uncharacterized.
Identifying BR receptors in more plants and deciphering their functions are important initial steps towards deciphering BR signaling networks and understanding their evolution. In this study, we cloned the foxtail millet BR receptor gene SiBRI1 and confirmed that it functioned as a BR receptor in Arabidopsis and foxtail millet at the physiological, genetic, and molecular levels. We also identified three other foxtail millet BR receptor genes, SiBRL1, SiBRL2 and SiBRL3. Phylogenetic analysis was used to classify the BR receptors in angiosperms into three types: I (BRI1), II (BRL1 and BRL3) and III (BRL2).
Our phylogenetic analysis also provides a basis for naming BRI1 family genes of other species in the future. BRI1 and its homologous genes play roles in different tissues in Arabidopsis thaliana and rice [21,35]. Our qRT-PCR experiment revealed that the expression patterns of SiBRI1 and its paralogues genes differed in different tissues, indicating that these genes may play different roles in different tissues. Due to the importance of BRI1, the roles of BRI1 in regulating plants development has been commendable studied recently, the literature on the specific roles of BRLs are relatively scare. AtBRL3 has been reported to be involved in plant drought defense and ROS homeostasis [36,37]. The study of BRLs involved in growth and development pathways in plants, improving the understanding of the specific roles of different BRI family genes.
To date, homologues of the BR receptor BRI1 have been identified and studied in five monocot species: rice, maize, barley, wheat and B. distachyon [18,20,21,24,38,39]. The functions of the BRI1 homologues in B. distachyon and maize have been revealed in studies on plants in which BdBRI1 and ZmBRI1, respectively, were silenced by RNAi [18,20]. In addition to OsBRI1, OsBRL1 and OsBRL3 have been characterized in rice. And different defective alleles of BRI1 showed a canonical BR-defective phenotype in rice and barley [21,24].
However, prior studies on B. distachyon had failed to determine whether the monocot homologues of BRI1 are able to function as BR receptors in dicot plants [38]. Furthermore, in monocots, the end product of BR synthesis and the more active BR seems to be castasterone rather than BL. The structure of BRI1 was found to be conserved between monocot and dicot plants, so we hypothesized that the function of BRI1 was conserved in these plants. To test our hypothesis, we cloned the gene and heterologously expressed it in bri1-116 plants. We found that the full-length coding sequence of SiBRI1 completely rescued the dwarf phenotype of bri1-116. bri1-5 is a weak mutant allele of AtBRI1 containing a point mutation that results in a C69Y amino acid substitution in the extracellular domain of BRI1 [15]. Although BRI1 in some dicot plants, such as GmBRI1b, GmBRI1a and MtBRI1, has been demonstrated to restore the phenotype of bri1-5 plants [16,17,19], whether the phenotype of bri1-116, the null allele of AtBRI1, can be resuced by ectopic expression is unclear. Our study proved that SiBRI1 could completely rescue the null allele of AtBRI1. These findings provide solid evidence that SiBRI1 can completely replace the function of AtBRI1. Our results demonstrate the conserved functions and structures of BRI1 between dicot and monocot plants and broaden our understanding of the BR response in crop plants.
The BR signaling pathway is both conserved and specific in different plants. Although we identified two key components involved in BR signaling in foxtail millet, SiBRI1 and SiBZR1 [31], the phenotypes after BR treatment differ between dicot and monocot plants. Dicot plants show increases in hypocotyl length and plant height, and reductions in root length under light conditions. For monocot plants, such as rice and maize, the main results of BR treatment are shortened roots and increased leaf angles [18,32]; droopy leaves and increased lateral root numbers were also found in foxtail millet in our study. BR signaling mutants have been found to exhibit smaller leaf angles and more erect leaf habits than wild-type plants, whereas BR gain-of-function mutants show increased lamina joint bending [32]. These findings lay the foundation for further studies on BR signaling pathways in monocots.
The phosphorylation of BRI1, as an RLK, is very important for precise control of BR signals, and its phosphorylation sites have been reported in Arabidopsis and tomato [40,41]. BRI1 comprises an extracellular domain, a TM domain, and a cytoplasmic domain [14]. The cytoplasmic domain contains the juxtamembrane region (JM), a serine/threonine/tyrosine kinase domain (KD), a the C-terminal (CT) domain. Most of the phosphorylation sites of AtBRI1 are located in the KD, such as Tyr-956, Thr-1039, Thr-1049, Ser-1044, and Thr-1045, and exhibit strong functions in BR signaling and plant growth [42][43][44]. Thr-1050 in the KD of tomato SlBRI1 negatively regulates the activity of SlBRI1 [40]. We also found 20 phosphorylation sites of OsBRI1 by in vitro kinase assay and mass spectrometry ( Figure S2). Thr-963, Ser-966, Thr-969, Ser-1087 and Ser-1091 are conserved in Arabidopsis and rice. These phosphorylation sites were first found in OsBRI1, but their biological functions need further study. We also attempted to identify the phosphorylation sites of SiBRI1 in vivo by immunoprecipitation combined with mass spectrometry. Unfortunately, we did not find any post-translational modifications of SiBRI1 under BL treatment; this topic may warrant further study. We also found 128 SiBRI1 interaction proteins, we found SiBSK7, the homologous gene of AtBSKs, which helpful to analyze the components of BR signaling pathway in foxtail millet. In addition, RESISTANCE TO PSEU-DOMONAS SYRINGAE PV. MACULICOLA 1 (SiRPM1) which response to defense [45], interacts with SiBRI1 through STING analysis [46], suggesting that SiBRI1 may be involved in disease resistance stress. To analyze the mechanism between SiBRI1 and interacting proteins is of great significance for elucidating the biological functions of SiBRI1 involved in multiple signaling pathways.
Overall, in this study, we identified the BR receptors in foxtail millet and divided the BRI1 family genes into three branches by phylogenetic analysis. Transformation of SiBRI1 restored the phenotype of bri1-116, and this complementation was a consequence of BR signaling pathway restoration. These results support SiBRI1 as a novel BRI1 gene in foxtail millet. Additionally, overexpression of SiBRI1 was found to modulate root development. Further investigations are needed to confirm the relationship between SiBRI1 and plant architecture, which may lead to the development of strategies to improve foxtail millet yield.
For genome version information, see Table S1.
The HMM for the characteristic domain of BRI1 proteins was downloaded from the Pfam database (http:// pfam. xfam. org) [49]. HMMER V 3.3 [28] was used to search for candidate genes with the whole protein sequences of each different species. Because the BRI1 family belongs to the RLK superfamily and because different members of the superfamily have similar domain compositions, it is difficult to search and identify BRI1 members in other species directly with the existing Pfam model. We analysed the results of the two multiple alignment methods (ClustalW and MUS-CLE) and used three phylogenetic inference methods (NJ, maximum likelihood (ML), and minimum evolution (ME) in MEGA X [50] with 1000 bootstrap replicates to choose stable phylogenetic trees. Because of the need for functional divergence analysis and positive selection analysis, whole protein sequences were used to construct phylogenetic relationships among BRI1 gene family members. Among the plant species discussed, only angiosperms expressed the BRI1 gene in a strict sense.

Plant growth and BL treatment
We received the Arabidopsis lines bri1-116 from Zhiyong Wang (Carnegie Institution for Science, Stanford) [ After the foxtail millet seeds were sterilized with chlorine gas (100 mL of 5% NaClO and 4 mL of HCl for 6-8 h), they were scattered on 1 / 2 MS medium (Phyto-Technology Laboratories, Overland Park, KS) with 1% sucrose and 0.3% Phytagel (PhytoTechnology Laboratories, Overland Park, KS), pH = 5.7. The plates were maintained under a 16L/8D cycle at 28 °C for 2 days. The seedlings with consistent growth were transferred to 6 × 20 cm (height) glass bottles containing one of several concentrations of BL in MS medium for further growth for 6 days before taking photos, removed for root length and lateral root number measurement, or cleared for an Asana microscopic observation. The lengths of the primary roots were measured using Image J software. 8-dayold pUbi:SiBRI1-eGFP/Ci846 seedlings were immersed in 1 μM BL for 1 h, and leaves and roots were placed into liquid nitrogen immediately after treatment. Then, RNA was extracted to detect BR synthesis gene expression.
After the Arabidopsis seeds were sterilized with 75% ethanol and 0.1% Triton X-100 for 5 min, they were grown on 1 / 2 MS medium with 1% sucrose and 0.45% Phytagel containing different concentrations of BL and PCZ. The plates were maintained under a 16L/8D cycle at 22 °C for 7 days. Photographs were taken, and the lengths of the primary roots were measured using Image J software. For clearly present the measurement data, we shown the relative length. We set the lengths of the primary roots under 0 nM BL or hypocotyls under 0 μM PCZ as 1, and the 5-100 nM BL or 0.25 μM PCZ group calculated the relative value respectively.

qRT-PCR
Total RNA was extracted according to Zhao (2021) [31]. First-strand cDNA was synthesized from approximately 1 µg of total RNA using M-MLV Reverse Transcriptase (Takara Bio, Inc., Otsu, Japan). qRT-PCR was performed according to a standard protocol using a Bio-Rad CFX Connect Real-Time PCR machine (Bio-Rad Laboratories, Hercules, CA, United States) and the SYBR Premix Ex Taq ™ system (Takara Bio, Inc.). The primers used are listed in Table S6, and SiActin was used as an internal reference. The average value from at least three biological replicates is presented.

Subcellular localization of SiBRI1
pUbi:SiBRI1-eGFP was constructed by using a recombinant cloning technique. First, the SiBRI1 CDS was cloned with a pCR ™ 8/GW/TOPO ® TA Cloning Kit (Invitrogen) and then cloned into the modified binary vector pCAM-BIA1305-eGFP via an LR enzyme (Invitrogen). pCAM-BIA1305-eGFP contains a maize ubiquitin promoter and a C-terminal enhanced GFP tag. The expression vectors were incorporated into Agrobacterium tumefaciens strain EHA105 and transformed into Ci846 with callus-based gene transformation procedures.
The pUbi:SiBRI1-eGFP transgenic plants were grown under a 16L/8D cycle at 28 °C on 1 / 2 MS medium for 3 days. The root tips were collected, and an FV3000 confocal microscope was used to stimulate GFP signals at a wavelength of 488 nm. The emission signals were collected and transmitted at a wavelength of 518 nm.

Immunoblotting
One-week-old and two-week-old Arabidopsis seedlings and foxtail millet leaves were ground to a fine powder in liquid nitrogen. Immunoblotting was then performed using anti-GFP (ProteinFind ® Anti-GFP Mouse Monoclonal Antibody, TRANS, Beijing, China) and anti-BZR1 (Cat: YKZPK82, Youke Biotechnology, Shanghai, China) antibodies.

Genotype identification of bri1-116 plants
After three weeks of growth of bri1-116 plants, DNA was extracted. The PCR primers BRI1-seq3 (CCA AAT CTCT-gCCAgAACCC) and BRI1-116GT-R (TAC CTC ATC AGG AAT CGA ACCAG), which produce an amplicon 852 bp in size, were used. The Col-0 product was cut into two bands (347 bp and 505 bp) by the MssI enzyme. The bri1-116 product could not be cut by the MssI enzyme due to point mutation; thus, the product remained 852 bp in size.

Immunoprecipitation and LC-MS/MS identification
Ci846 seeds were germinated and then grown in doubledistilled water at 28 °C under LD conditions for 4 days. Seedlings with similar root and coleoptile lengths were selected and transferred to 1 / 2 strength Hoagland solution, and allowed to grow for 4 more days. The leaves were then immersed in 1 / 2 strength Hoagland solution with 1 μM BL or not for 2 h, and leaves were placed into liquid nitrogen immediately after treatment and ground to a fine powder in liquid nitrogen. The tissues were extracted using a Grinding buffer containing 25 mM HEPES, 0.6% PVP, 5 mM VC, 5 mM EDTA, 25 mM NaF, 1 mM Na 2 MoO 4 , 1 mM Na 3 VO 4 , 5 mM DTT, 2 mM imidazole and protease inhibitor cocktails (Sigma) at 4 °C, and centrifuged at 4 °C for 10 min and 2000 × g. The homogenate was centrifuged at 10,000 g for 45 min and the deposit was extracted using a NEB buffer containing 20 mM HEPES (pH 7.5), 40 mM KCl, 1 mM EDTA, 1% TritonX-100 and protease inhibitor, then incubated with GFP-Trap Agarose (Ychromotek, gta) for 2 h at 4 °C. The Beads were washed on a column with 20 bed volumes of NEB buffer and then eluted with SDS loading buffer (0.125 M Tris-HCl (pH 6.8), 2% β-Mercaptoethanol, 4% SDS, 20% glycerol, 0.25% bromophenol blue). Proteins eluted from the GFP-Trap were separated by 8-20% Precast-GLgel Tris-Glycine (BBI-Rad, E919DA0304), and each line was cut into four pieces according to the molecular weight of the proteins. Protein in-gel digestion and LC-MS/MS identification was performed according to our published method [51,52].