A common wild rice-derived BOC1 allele reduces callus browning in indica rice transformation

Callus browning, a common trait derived from the indica rice cultivar (Oryza sativa L.), is a challenge to transformation regeneration. Here, we report the map-based cloning of BROWNING OF CALLUS1 (BOC1) using a population derived from crossing Teqing, an elite indica subspecies exhibiting callus browning, and Yuanjiang, a common wild rice accession (Oryza rufipogon Griff.) that is less susceptible to callus browning. We show that BOC1 encodes a SIMILAR TO RADICAL-INDUCED CELL DEATH ONE (SRO) protein. Callus browning can be reduced by appropriate upregulation of BOC1, which consequently improves the genetic transformation efficiency. The presence of a Tourist-like miniature inverted-repeat transposable element (Tourist MITE) specific to wild rice in the promoter of BOC1 increases the expression of BOC1 in callus. BOC1 may decrease cell senescence and death caused by oxidative stress. Our study provides a gene target for improving tissue culturability and genetic transformation.

I n contrast to japonica rice (Oryza sativa ssp. japonica), Agrobacterium tumefaciens-mediated transformation of indica rice (Oryza sativa ssp. indica) is hindered by callus browning. Callus browning is a common feature of many plant species that presents problems for in vitro culture, a process required for transgenic breeding, resulting in decreased regenerative ability, poor growth, and even death 1 . The use of antioxidants [2][3][4] , adsorbing agents 5 , low salt concentrations 6 , and growth regulators 7 may lessen the effects of callus browning to a certain degree, but because some species do not tolerate these treatments, there is no universal solution to this problem.
The physiological and biochemical mechanisms of callus browning have been extensively studied. For instance, reducing the activity of polyphenol oxidase (PPO), which catalyzes the oxidation of phenols to quinones that in turn polymerize to form brown pigments 8 , can mitigate the browning of callus 9 . Browning calli have lower antioxidant enzyme activity than non-browning calli and oxidative stress is associated with callus browning 10,11 . Endogenous 12 and exogenous 13 ethylene (ET) cause browning in plant cell cultures. In addition, reduced nitrite reductase activity leads to callus browning 14 .
Different periods of in vitro rice tissue culture are controlled by many complex genes, respectively 15 . Several quantitative trait loci (QTLs) for culturability traits in rice have been identified [16][17][18][19][20][21][22][23][24] . Using positional cloning, a major QTL gene encoding ferredoxinnitrite reductase (NiR) that determines regeneration ability in rice was identified 19 . QTLs involved in callus browning have been identified on all 12 rice chromosomes 17,[21][22][23][24] . However, to our knowledge, there have been no reports on the isolation of genes related to callus browning in any plant species. The molecular mechanisms underlying physiological processes inducing browning in plant cell culture still remain unclear.
Genetic analyses of rice culturability have mainly been performed on cultivated rice, and few studies have employed common wild rice (Oryza rufipogon Griff.). In this study, we isolate and characterize a gene responsible for mitigating callus browning in rice, named BROWNING OF CALLUS1 (BOC1). The BOC1 allele is specific to common wild rice. BOC1 is an allele of a rice homolog of SRO (SIMILAR TO RCD ONE), termed OsSRO1c. The appropriate upregulation of BOC1 significantly reduces callus browning during callus proliferation. BOC1 is upregulated in common wild rice callus due to the presence of the Tourist MITE in the promoter region of BOC1. Additionally, BOC1 improves the genetic transformation efficiency and frequency of Hygromycin (Hyg)-resistant calli following gene transfer. BOC1 may regulate oxidative stress and programmed cell death (PCD), which may repress the callus browning. Our findings not only elucidate the mechanisms of callus browning during tissue culture, but also provide important insights into establishing a suitable genetic transformation system for the molecular breeding of rice.

Results
Phenotypic analysis of introgression line YIL25. To identify the gene responsible for callus browning in rice, we constructed a set of introgression lines using a Yuanjiang common wild rice (YJCWR, O. rufipogon) accession that is relatively resistant to browning (Fig. 1b) as a donor and an elite indica cultivar Teqing (O. sativa) that is more susceptible to callus browning (Fig. 1a) as the recipient. We screened the introgression line YIL25 by inoculating mature seeds on unimproved NB (UINB) medium (NB basal medium without additives) for 7 days, and transferred the resulting scutellum-derived calli to UINB medium for 3 weeks of subculture (Fig. 1c). The callus browning rate (CBR) and callus browning index (CBI) were lower in YIL25 than Teqing (Fig. 1j, k). YIL25 contained three YJCWR chromosomal segments on the short arm of chromosomes 2, 3, and 5 ( Supplementary Fig. 1). We, therefore, chose YIL25 as a reduced-browning line to cross with Teqing. The F 1 plants showed similar phenotypes of CBR and CBI to YIL25 ( Supplementary Fig. 2). These results indicated that the reduced-browning trait is controlled by a dominant gene. The histological examination indicated that the cell arrangement of YIL25 calli was more even and the cells were smaller and denser compared to Teqing (Fig. 1d, e). Observations using scanning electron microscopy (SEM) revealed that YIL25 calli consisted of many globular nodules (Fig. 1h) and the cells of YIL25 were turgid (Fig. 1i), whereas Teqing calli had fewer globular nodules (Fig. 1f) and the cells of Teqing were flaccid (Fig. 1g).
Map-based cloning of the callus browning trait gene. To clone the gene for the callus browning trait, we constructed a F 2 population of 198 segregating plants derived from a cross between YIL25 and Teqing. We used the CBI as a phenotypic index as it reflects the frequency and degree of callus browning. First, we measured the CBI on the seed-derived callus of the 198 individual plants. We then performed QTL analysis using simple sequence repeat (SSR) markers. The callus browning trait QTL, qCBT3, which is located between SSR markers RM3131 and RM3766 on the short arm of chromosome 3, had the largest effect, explaining 14% of the phenotypic variance associated with decreased CBI (Fig. 2a; Supplementary Table 1).
For high-resolution mapping, we selected F 2 individuals that were heterozygous at the qCBT3 locus (RM3131-RM3766), containing one chromosomal segment from Teqing and one from YJCWR by self-crossing to construct a segregating population to eliminate gene interference on introgression segments of chromosomes 2 and 5 (Fig. 2b). We identified plants with recombination events between markers RM3131 and RM3766 by genotyping the 6377 individuals in the population and evaluated the CBI of these recombinant individuals (Fig. 2b). We then developed single nucleotide polymorphism (SNP) markers and delimited qCBT3 to an 18.6 kb region between SNP markers SN11 and SN13. This region contained only one predicted gene (LOC_Os03g12820) in the Nipponbare reference genome (The Rice Genome Annotation Project Database, Fig. 2c). Sequence analysis of the coding region in this interval revealed no sequence variation between Teqing and YIL25. Next, we compared the promoter sequences upstream of the translation start site between Teqing and YIL25 and identified one 337 bp deletion, one 1 bp deletion, and three SNPs in Teqing ( Fig. 2d; Supplementary Fig. 3). Therefore, we chose LOC_Os03g12820 as the candidate gene for BROWNING OF CALLUS1 (BOC1).
Functional analysis of BOC1. To elucidate the function of LOC_Os03g12820, we constructed an RNA interference construct targeting LOC_Os03g12820 and transformed it into introgression line YIL25 (designated as pRi-BOC1-YIL25). Eighteen transgenic lines were obtained. Reverse transcription-quantitative PCR (RT-qPCR) analysis showed that LOC_Os03g12820 was expressed at lower levels in the pRi-BOC1-YIL25 positive transgenic lines than in the negative (non-transformed) plants (Fig. 3e). In addition, RNAi suppression of LOC_Os03g12820 in the pRi-BOC1-YIL25 positive transgenic lines to the level of Teqing caused serious callus browning compared to the non-transformed controls ( Fig. 3a-e).
Furthermore, we introduced a complementation vector containing the promoter, open reading frame and 3′-UTR of LOC_Os03g12820 from YIL25 into Teqing (designated as pCPL-BOC1-Teqing) and obtained 15 positive transgenic plants.
RT-qPCR analysis showed that the relative expression levels of LOC_Os03g12820 were significantly elevated in the pCPL-BOC1-Teqing positive plants compared to the negative plants (Teqing) (Fig. 3j). The CBI was reduced in these positive lines when the expression level of LOC_Os03g12820 was elevated at least 1.8 fold compared with the negative plants ( Fig. 3f-j). This finding indicates that LOC_Os03g12820 mitigates the callus browning phenotype. These results suggest that LOC_Oso3g12820 is BOC1 regulates callus browning in rice.
Next, we generated transgenic Teqing plants overexpressing the LOC_Os03g12820 coding sequence of YIL25 (as shown in Supplementary Fig. 3, the rose red and red underline represent the nucleotide sequence of the overexpression construct, designated as pOE-BOC1-Teqing) and obtained 22 positive transgenic lines. RT-qPCR indicated that BOC1 was strongly upregulated in the pOE-BOC1-Teqing positive transgenic lines compared to the negative plants (Fig. 3o). Nevertheless, when BOC1 expression levels were 8.8-times higher in the tested pOE-BOC1-Teqing positive lines compared to the negative plants, callus browning was more severe (Fig. 3k- Fig. 4b).
Transcriptional characterization of BOC1. To elucidate the expression profiles of BOC1, we analyzed the expression pattern of BOC1 in different organs by RT-qPCR. The expression level of BOC1 was highest in YIL25 calli after 21 days of subculture, followed by the tiller base, roots, culms, and leaves, and it was lowest in young panicles ( Supplementary Fig. 5). BOC1 expression increased gradually in Teqing and YIL25 throughout tissue culture. BOC1 expression did not significantly differ between Teqing and YIL25 during the early stage of tissue culture, but significant differences were observed during the later stage; BOC1 expression was notably higher in YIL25 than in Teqing in callus after 21 days of subculture (Fig. 4a). We further examined the expression of BOC1 in calli after 21 days of subculture by RNA in situ hybridization and observed that the hybridization signals were stronger in YIL25 than in Teqing, confirming the notion that BOC1 is expressed at higher levels in YIL25 callus (Fig. 4b).
Sequencing analysis revealed one 337 bp deletion, one 1 bp deletion, and three SNPs in the promoter region of BOC1 between Teqing and YIL25 (Fig. 4c). To determine whether these deletions and SNPs in the BOC1 promoter region directly contribute to its expression, we generated various constructs by inserting different combinations of mutant promoter fragments of YIL25 into binary vector pGreenII 0800-LUC. We introduced these constructs into rice protoplasts produced from callus that had been cultured for 21 days and examined the expression of the luciferase reporter gene (LUC). As shown in Fig. 4c, no significant differences in activity were observed between the YIL25 promoter fragment and mutated promoter fragments containing three SNPs (constructs M2-M4) and one 1 bp deletion (construct M5). However, the 337 bp deletion (construct M1) significantly decreased the promoter activity of BOC1 to a level similar to that of Teqing (Fig. 4c). These results suggest that the 337 bp deletion is responsible for the reduced expression level of BOC1 in Teqing calli. Using the RepeatMasker Web Server (http://www. repeatmasker.org/cgi-bin/WEBRepeatMasker), we examined the rice transposable element database using the 337 bp sequence as a query and determined that it is the Tourist-like miniature inverted-repeat transposable element (Tourist MITE), which belongs to the PIF/Harbinger superfamily. In addition, using PLACE (Plant Cis-acting Regulatory DNA Elements; http://www. dna.affrc.go.jp/htdocs/PLACE/), we identified three DOFCOREZM (AAAG, DOF TFBS) 25 , one MYB1AT (TAACCA, MYB TFBS) 26 , one MYCCONSENSUSAT (CAGATG, MYC TFBS) [26][27][28] , and one ACGTATERD1 (ACGT, ACGT TFBS) 28 in this sequence; these elements are associated with plant stress responses ( Supplementary  Fig. 6). Subcellular localization analysis indicated that the BOC1-GFP fusion protein was specifically targeted to the nucleus in rice protoplasts (Fig. 4d).
BOC1 influences genetic transformation efficiency. To determine if BOC1 enhances the transformation efficiency of rice due to decreased callus browning, we transformed the binary vector pCAMBIA1300 into YIL25 (YIL25 1300), Teqing (Teqing 1300), and Teqing 1300 as a control. The pCAMBIA1300-BOC1 complementation construct from YIL25 was transformed into Teqing (Teqing pCPL) by Agrobacterium tumefaciens-mediated transformation and cultured on UINB medium for hygromycin (Hyg)   selection. The selection frequency of Hyg-resistant calli was more than 2.7-fold higher in YIL25 1300 than in Teqing 1300 ( Fig. 5a, b, d), and the transformation efficiency was 2.5-fold higher in YIL25 1300 than in the Teqing 1300 control (Fig. 5e), indicating that YIL25 containing BOC1 is an excellent receptor material for genetic transformation. Meanwhile, the selection frequency of Hyg-resistant calli was 2.1-fold better (Fig. 5a, c, d) and the transformation efficiency was 1.9-fold better in Teqing pCPL compared to the Teqing 1300 control (Fig. 5e). These results indicate that BOC1 improves the selection frequency of Hygresistant calli and the genetic transformation efficiency of rice by reducing callus browning during Agrobacterium-mediated transformation.
Physiological indices related to oxidative stress. To further explore the molecular mechanisms of callus browning, we measured various physiological and biochemical indices in calli at 21 days of subculture. Compared to Teqing, the H 2 O 2 contents were lower in both YIL25 and pCPL calli and higher in pOE (pOE-BOC1-Teqing) calli ( Fig. 6a; Supplementary Fig. 7a (Fig. 6b). MDA (malondialdehyde) is a biomarker of lipid peroxidation and an indicator of reactive oxygen species (ROS) production. Compared to Teqing, MDA contents were notably lower in the YIL25 and pCPL lines but higher contents in the pOE lines ( Fig. 6c; Supplementary Fig. 7b), reflecting less damage from H 2 O 2 in YIL25 and pCPL plants. Moreover, catalase (CAT), glutathione reductase (GR), and glutathione S-transferase (GST) activity and antioxidant glutathione (GSH) contents were higher in YIL25 and pCPL vs Teqing callus ( Fig. 6d-g). However, GST activity and GSH contents were lower in pOE lines than in Teqing callus although CAT and GR activity did not significantly differ among lines ( Supplementary Fig. 7c-f). Finally, the levels of the plant hormones salicylic acid (SA) and ethylene (ET), as well as the direct ET precursor, 1-aminocyclopropane-1-carboxylic acid (ACC), were significantly lower in YIL25 compared to Teqing (Fig. 6h-j). Collectively, these findings suggest that BOC1 may inhibit PCD caused by oxidative stress, thereby reducing callus browning.
Among the UUD genes, 98 genes were involved in stress responses and cell death (Fig. 7c). Notably, these included 26 pentatricopeptide repeat protein genes, whose dysfunction leads to ROS accumulation; 29,30 these genes were upregulated in YIL25 Whiskers extend to data no more than 1.5 times the interquartile range, and the remaining data are indicated by dots. P-values of (d, e) were determined by two-tailed Student's t-tests. Source data are provided as a Source Data file.
and pCPL but downregulated in pOE ( Fig. 7c; Supplementary  Fig. 8a). GST conjugates with GSH to scavenge ROS in order to protect cellular components from damage. In the current study, 3 GST genes present in the UUD DEG set and absent in the DDU DEG set ( Fig. 7c; Supplementary Fig. 8c). In addition, compared to Teqing, GST activity and GSH contents were higher in YIL25 and pCPL callus but lower in pOE callus (Fig. 6f, g; Supplementary Fig. 7d,f). Among the UUD genes were four aldehyde dehydrogenase genes; aldehyde dehydrogenase enhances tolerance to various abiotic stresses by scavenging cytotoxic aldehydes ( Supplementary Fig. 8d), and MDA contents were higher in of Teqing and pOE callus compared to YIL25 and pCPL callus ( Fig. 6c; Supplementary Fig. 7b). LOC_Os01g67980, which negatively regulates the senescence process, encodes senescence-associated protein OsSAG12-1. RNAi transgenic lines declined the expression of OsSAG12-1 and developed early senescence 31 ( Fig. 7c; Supplementary Fig. 8f). In addition, the UUD genes included transcription factor genes associated with stress and defense responses, which might function in maintaining ROS homeostasis ( Fig. 7c; Supplementary Fig. 8b, g). Twelve DDU genes were associated with oxygenase and oxidase; cytochrome P450 has monooxygenase activity as a source of ROS 32 (Supplementary Fig. 8h). The DDU group also included genes encoding aspartic proteinases, peptidases, and nuclease (involved in cell degradation), as well as genes encoding polygalacturonase, endoglucanase, and glycosyl hydrolases (involved in cell wall degradation) ( Fig. 7c; Supplementary  Fig. 8i). Emerging evidence suggests that aspartic proteases are involved in plant development and cell death 33 . Finally, five DDU DEGs encode positive regulators of the senescence process (Fig. 7c). These DEGs include LOC_Os01g53260, encoding WRKY transcription factor OsWRKY23, which positively regulates the senescence process 34 , and LOC_Os12g36850, encoding a pathogenesis-related bet v I family protein similar to probenazole-inducible protein PBZ1, which induces cell death in rice, tobacco, and Arabidopsis via its RNase activity inside the cell 35 (Supplementary Fig. 8j). When we treated callus that was subcultured for 7 days with 0.5% H 2 O 2 for 18 h, PBZ (LOC_Os12g36850) was upregulated compared to the untreated control ( Supplementary Fig. 8k). Taken together, these findings suggest that the appropriate expression of BOC1 may be beneficial for rice cells to be in reducing state, prevented various stress, and then inhibited cell death.  Data 4). Therefore, we performed an association test of callus browning after 21 d of subculture vs. sequence variations in the O. rufipogon accessions and Asian rice cultivars. The association test of the common wild rice revealed the strongest association signals at the Tourist MITE and the four SNPs (Fig. 8a). Furthermore, we investigated the CBI of the 50 accessions of common wild rice and determined that this value was lower in the 19 accessions with the Tourist MITE than in the 31 accessions without this insertion (Fig. 8b). Moreover, the BOC1 gene was expressed at higher levels in the 19 accessions with the Tourist MITE than in the 31 accessions without the Tourist MITE ( Fig. 8c; Supplementary Data 3). In addition, we obtained previously reported the genome resequencing data for 446 Oryza rufipogon accessions 36  III (closer to japonica) and 7 were in group II (the intermedia group between indica and japonica) (Supplementary Fig. 9a). By contrast, the 178 accessions without this MITE insertion were distributed among all the three groups including 65 in group I (closer to indica), 49 in group II, and 64 in group III (Supplementary Fig. 9a).
Nucleotide alignment of the 74 Asian rice cultivars revealed 9 Indels and 27 SNPs (Fig. 8d; Supplementary Data 4). To ascertain whether BOC1 was responsible for callus browning in Asian rice cultivars, firstly, an association analysis was carried out between callus browning and sequence variations of 74 Asian rice cultivars. The result indicated that the Indel8 (−433, T indel)  in the promoter of BOC1 was the strongest signals (Fig. 8d). However, this site was no polymorphic in the 50 accessions of O. rufipogon (Supplementary Data 3, 4). We further found that the CBI in 15 varieties with T insertion in the promoter of BOC1 was lower than that in 59 varieties without this insertion ( Fig. 8e;  Supplementary Data 4). We also analyzed the expression levels in these varieties and the result showed that the expression levels of BOC1 was higher in the 15 varieties with T insertion in the promoter of BOC1 than that in 59 varieties without this insertion ( Fig. 8f; Supplementary Data 4). The above results suggest that the sequence variation in the BOC1 promoter influence callus browning by affecting the expression level of BOC1.
Finally, we analyzed the fixation index (F ST , the level of population differentiation), on chromosome 3 in 1083 cultivated indica and japonica rice varieties 36 . The F ST level in BOC1 was 0.21, indicating that no significant differences in BOC1 exist between the indica and japonica subspecies ( Supplementary  Fig. 9b). Therefore, we propose that the effect of BOC1 on callus browning is not related to the differences between the indica and japonica subspecies.

Discussion
Here, we demonstrated that the proper expression of BOC1 decreases callus browning and increases the transformation efficiency of rice. BOC1 encodes an SRO (SIMILAR TO RCD ONE) protein that is allelic to OsSRO1c, a regulator of oxidative stress responses 37 . RCD1 plays an important role in regulating cell death by controlling oxidative stress responses in Arabidopsis responses [38][39][40][41][42][43] . OsSRO1c is the major stress-responsive gene in the SRO family. Both ossro1c-1 mutants and amiRNA (artificial microRNA)-OsSRO1c plants are hypersensitive to drought stress at the seedling stage. In addition, OsSRO1c-overexpressing rice plants showed more severe damage under drought stress than wild-type plants, whereas plants overexpressing SNAC1, encoding a NAC transcription factor that directly regulates OsSRO1c expression, showed increased drought resistance 29 . In the current study, RNA interference and BOC1 overexpression plants showed a callus-browning phenotype similar to or more severe than that of Teqing, whereas lines harboring the complementation vector exhibited reduced callus browning, indicating that the appropriate upregulation BOC1 reduces callus browning.
Transposable elements, among the most variable components of the genome, can replicate and integrate into positions, which affects the expression levels of adjacent genes, thereby resulting in genetic diversity 44 . For example, the terminal-repeat retrotransposon in miniature (TRIM) elements insertion in the promoter of MS2 allele in the male-sterile ms2 mutant resulted in an anther-specific expression pattern in wheat 45 . In maize, an 82 bp miniature inverted-repeat transposable element (MITE) into the promoter of ZmNAC111 represses its expression; this MITE is significantly associated with natural variation 46 . In the current study, the presence of Tourist MITE in the BOC1 promoter significantly increased BOC1 expression in rice callus.
Callus cultured in vitro is subjected to complex growth conditions compared to cells in plants. Callus cells are exposed to the oxygen in air, which tends to result in oxidative stress. Oxidative browning is a prevalent problem in plant tissue culture that can cause cell death, resulting in brown tissues 47 . A moderate increase in the levels of the ROS H 2 O 2 helps protect plant cells from subsequent exposure to more severe abiotic or oxidative stress 48 . However, the excessive accumulation of H 2 O 2 might be involved in regulating PCD [38][39][40][49][50][51] . Moreover, high levels of H 2 O 2 induce the accumulation of SA, resulting in severe oxidative damage 39,52 ; in turn, SA pretreatment mediates the accumulation of H 2 O 2 53 . Endogenous ET in in vitro culture may inhibit the synthesis of protective enzymes, such as CAT, for scavenging H 2 O 2 , which is involved in senescence, browning, and necrosis of callus 54 . In the current study, YIL25 callus contained decreased levels of H 2 O 2 and the phytohormones ET and SA, as well as increased levels of CAT, GR, and GST activity compared to Teqing. The RNA-seq data analysis revealed that BOC1 influences the expression of genes associated with ROS production and scavenging, stress response and cell death. In recently report, RCD1 coordinates chloroplast and mitochondrial functions through interaction with transcription factors. RCD1 participates in regulation of mitochondrial respiration and chloroplast ROS processing and may contribute to the survival of plant under a changing environment 55 . However, as shown in Supplementary  Fig. 4b, RCD1 and OsSRO1c have different evolutionary histories which may have contrasting functions in the regulation of ROS related processes. In the current study, we use the calli subcultured for 21 d in the dark and is not necessary for photosynthesis by origin of carbon from medium. Therefore, the interaction and phenotype of BOC1 may be different from seedling stage, which will be an interesting research topic.
Common wild rice, the wild progenitor of cultivated rice, has an important gene pool that includes some alleles no longer present in cultivated rice 56 . The insertion of the 337 bp Tourist MITE in the BOC1 promoter increases BOC1 expression. This insertion is only detected in certain accessions of wild rice, but not in cultivated rice. Therefore, wild rice could be used to provide alleles for improving the tissue culture characteristics and genetic transformation efficiency of cultivated rice. Moreover, the genetic manipulation of the BOC1 promoter in cultivated rice might also improve tissue culture ability. Perhaps the BOC1 allele of wild rice and even similar alleles in other crops of the Gramineae family could be used to create recipient varieties more amenable to genetic transformation. The use of BOC1 in in vitro rice callus propagation might help improve genetic transformation efficiency and selection to facilitate the molecular breeding of rice.  Tissue culture procedure. Mature, healthy, dehusked seeds were sterilized by immersion in 70% ethanol for~2 min, followed by 15% sodium hypochlorite solution for 15 min with shaking, and rinsed three or four times with sterile water on an ultraclean workbench. Unimproved NB (UINB) medium (NB medium containing N6 macronutrient components 58 , B5 micronutrient components, and organic components 59 , supplemented with 2 mg/L of 2,4-D and 30 g/L of sucrose) was used for callus induction and subculture. The pH of the medium was adjusted to 5.8 with 1 N KOH, and 3 g/L of Phytagel was added before autoclaving at 121°C for 20 min. Approximately 90 mature seeds per line, evenly distributed among three Petri dishes, were incubated in induction medium for 7 days at 28°C in the dark. The induced calli were transferred to subculture (UINB) medium and incubated in the dark at 28°C for 3 weeks. The traits of the calli in each Petri dish were recorded.

Methods
Phenotype observations. The tendency for callus browning was categorized into four levels: (0) less than 1/10 of the callus tissue was brown (recorded as no browning); (1) 1/10~1/3 of the callus tissue was brown, considered light browning; (2) 1/3~2/3 of the callus tissue was brown, considered medium browning; (3)  Luciferase assays. The promoter sequences upstream of the BOC1 translation start site of YIL25 and Teqing were amplified using specific primers designed with site-directed mutations at the InDels and SNPs. The fragments were inserted into the pGreenII 0800-LUC vector to drive the expression of the firefly luciferase (LUC) gene. The effector plasmid DNA was transformed into rice protoplasts prepared from rice calli that had been subcultured for 21 days in the dark using the polyethylene glycol-mediated method 62 . The enzyme solution contained the following:  (5): where J is the analyte H 2 O 2 flux (unit: pmol·cm −2 ·s −1 ), dc is its concentration gradient, dx is the distance between the two points (30 μm), and D 0 is its diffusion constant (unit: cm −2 ·s −1 ).
Measurement of lipid peroxidation. MDA was measured using the thiobarbituric acid (TBA) assay 64 . The calli (100 mg) were homogenized in 1 ml 0.1% (w/v) TCA in an ice bath. After centrifuging the homogenate at 12,000 × g for 10 min, 0.5 ml of the supernatant was combined with 0.5 ml 0.6% thiobarbituric acid (TBA) and incubated at 95°C for 15 min. The samples were cooled immediately in an ice bath and the absorbance of the supernatant measured at 450, 532, and 600 nm.
Absorbance was recorded at 412 nm 65 .
Enzyme activity assays. The enzyme activity assays were carried out by Shanghai Sanshu Biotechnology Co., Ltd (Shanghai, China). Catalase (CAT) activity was measured based on the consumption of H 2 O 2 and the decrease in absorbance at 240 nm 66 . Glutathione reductase (GR) was assayed based on the decrease in A 340 due to nicotinamide adenine dinucleotide phosphate (NADPH) oxidation. The reaction mixture of GR contained 1.0 ml of 0.1 M potassium phosphate buffer (pH 8.0) including 0.5 mM EDTA, 0.5 mM MgCl 2 ,1 mM NADPH, 10 mM oxidized glutathione (GSSG), and 0.1 ml enzyme extract 67 . The glutathione S-transferase (GST) was assayed using the substrate 1-chloro 2,4-dinitrobenzene (CDNB), and the activity was detected spectrophotometrically based on the change in absorbance at 340 nm 68 .
Endogenous hormone level measurements. The calli were ground into a powder in liquid nitrogen in a mortar and each sample (0.05 g) were transferred to a 1.5 ml screw-cap tube. After adding 50 μl of internal standards and 500 μl of extraction solvent (2-propanol: H 2 O: HCl = 2:1: 0.002) to each tube, the sample was incubated on a shaker at 900 × g for 30 min at 4°C. After adding 1 ml trichloromethane, the sample was shaken for 30 min and centrifuged at 14,000 × g for 5 min at 4°C to form two phases. A 1.2 ml sample of solvent from the lower phase was dried in a nitrogen evaporator, dissolved in 0.1 ml methanol and endogenous SA levels quantified by high-performance liquid chromatography-mass spectrometry (HPLC-MS/MS) 69 . Ethylene was analyzed according to a published method 70 with minor modifications. The calli were placed into 100-mL vials that were promptly sealed with a rubber plug and incubated at 28°C for 48 h in the dark. Ethylene content in 2 ml of the headspace gas was analyzed with a gas chromatograph (GC-17A, Shimadzu, Japan).
RNA-seq analysis. Total RNA was isolated from Teqing and YIL25, pCPL and Teqing, and pOE and Teqing calli (subcultured for 21 days) with three biological replicates. Paired-end libraries were constructed and sequenced on the Illumina HiSeq 2500 platform at the Novogene Company (China). The RNA-seq reads were mapped to the reference genome (Os-Nipponbare-Refrence-IRGSP-1.0, MSU7) using TopHat2 with default parameters, and the FPKM (fragments per kilobase of exon per million mapped reads) of each gene and the DEGs (fold change ≥ 2, FDR < 0.001) between YIL25 and Teqing, pCPL and Teqing, and pOE and Teqing were calculated 71 .
Sequencing and data analysis. The~4.6 kb genomic fragment covering the entire BOC1 gene (2740 bp), the 1,452 bp 5′-flanking region, and the 380 bp 3′-flanking region were amplified using five PCR primer pairs and sequenced using the Sanger sequencing approach. A previously published dataset in 1,083 diverse rice accessions was used to calculate the fixation index (F ST ) across chromosome 3 between indica and japonica varieties using VCFtools 0.1.15, with a 100-Kb window size 36,72 .
Data analysis. A two-tailed Student's t-test was performed using SPSS version 16 (SPSS Inc., Chicago, IL, USA). QTL analysis was performed using MapManager QTXb 20 73 . Amino acid sequences homologous to BOC1 were downloaded from the National Center for Biotechnology Information website (http://blast.ncbi.nlm. nih.gov/). Phylogenetic trees were constructed using the neighbor-joining algorithm in MEGA version 7 with 1000 bootstrap replications 74 . The association tests were analyzed by R v3.6.1.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Data supporting the findings of this work are available within the paper and its Supplementary Information files. A reporting summary for this Article is available as a Supplementary Information file. The datasets generated and analyzed during the current study are available from the corresponding author upon request. RNA-seq data for callus that had been subcultured for 21 days from Teqing and YIL25, Teqing and the complementation line pCPL-BOC1-Teqing, and Teqing and the overexpression line pOE-BOC1-Teqing was deposited in the National Center for Biotechnology Information (NCBI) under GEO accession number GSE142094. The source data underlying Figs. 1j, 1k, 2a, 2b, 3e, 3j, 3o, 4a, 4c, 5, 6, 8b, 8c, 8e, and 8f, as well as Supplementary Figs. 2, 5, 7, 8k, and 9 are provided as a Source Data file.