Characterization of evolutionarily distinct rice BAHD‐Acyltransferases provides insight into their plausible role in rice susceptibility to Rhizoctonia solani

Plants produce diverse secondary metabolites in response to different environmental cues including pathogens. The modification of secondary metabolites, including acylation, modulates their biological activity, stability, transport, and localization. A plant‐specific BAHD‐acyltransferase (BAHD‐AT) gene family members catalyze the acylation of secondary metabolites. Here we characterized the rice (Oryza sativa L.) BAHD‐ATs at the genome‐wide level and endeavor to define their plausible role in the tolerance against Rhizoctonia solani AG1‐IA. We identified a total of 85 rice OsBAHD‐AT genes and classified them into five canonical clades based on their phylogenetic relationship with characterized BAHD‐ATs from other plant species. The time‐course RNA sequencing (RNA‐seq) analysis of OsBAHD‐AT genes and qualitative real‐time polymerase chain reaction (qRT‐PCR) validation showed higher expression in sheath blight susceptible rice genotype. Furthermore, the DNA methylation analysis revealed higher hypomethylation of OsBAHD‐AT genes that corresponds to their higher expression in susceptible rice genotype, indicating epigenetic regulation of OsBAHD‐AT genes in response to R. solani AG1‐IA inoculation. The results shown here indicate that BAHD‐ATs may have a negative role in rice tolerance against R. solani AG1‐IA possibly mediated through the brassinosteroid (BR) signaling pathway. Altogether, the present analysis suggests the putative functions of several OsBAHD‐AT genes, which will provide a blueprint for their functional characterization and to understand the rice–R. solani AG1‐IA interaction.


INTRODUCTION
Plant secondary metabolites play an important role in diverse processes involved in growth and development and resistance to various biotic and abiotic stresses (Erb & Kliebenstein, 2020). Although a few pathways are involved in the biosynthesis of major classes of plant secondary metabolites, the modifications and the expansion of the basic skeleton result in the immense diversity of the plant's secondary metabolites (Wang et al., 2019). These modifications and expansion processes involve various biochemical reactions such as decarboxylation, hydroxylation, glycosylation, methylation, oxidation and reduction, and acylation (Wang et al., 2019). Among these biochemical reactions, acylation is the most prevalent modification that expands the diversity of phenolic compounds synthesized through various pathways (Bontpart et al., 2015). The acylation of the metabolites influences their properties, functions, transport, and storage. For instance, acylation of anthocyanins modulate their solubilization, stability and localization, and color properties (Gomez et al., 2009;Mugford et al., 2009;Nakayama et al., 2003). The acylation is catalyzed by acyltransferases, where activated acyl donor, such as acyl-sugars, acylated-proteins, or acylated-CoA thioesters, donate their acyl group to modify the secondary metabolites (Bontpart et al., 2015). The acyltransferases belong to the large protein family and are broadly categorized into two major classes based on their specificity for the activated acyl donors: BAHD-ATs responsible for acyl-CoA dependent acylation, while the Serine carboxypeptidaselike-acyltransferases are responsible for 1-O-β-glucose esters dependent acylation (Bontpart et al., 2015).
The BAHD-AT name is derived from the first four characterized enzymes of the family (namely, benzylalcohol Oacetyltransferase, BEAT; anthocyanin O-hydroxycinnamoyl transferase, AHCT; anthranilate N-hydroxycinnamoyl/ benzoyltransferase, HCBT; and deacetylvindoline 4-O-acetyl transferase, DAT) (Dudareva et al., 1998;Fujiwara et al., 1998;St-Pierre et al., 1998;Yang et al., 1997). All the biochemically characterized BAHD-ATs, hitherto, are of molecular weight ranging from 48 to 55 kDa and known to function as monomeric enzymes (Bontpart et al., 2015;D'Auria, 2006). Though the sequence similarity among the members of the BAHD-ATs family is only 25-34% at protein level (Ma et al., 2005), members with functional similarity may share the sequence identity of up to 90% (Boatright et al., 2004). Despite low sequence similarity, the BAHD-AT members share two motifs: a conserved HXXXD(G) motif involved in substrate binding and catalysis and a less conserved DFGWG motif with a structural role in the catalytic activity of BAHD-ATs (Unno et al., 2007). The HXXXD(G) motif of BAHD-ATs is essential for the catalytic activity as determined using site-directed mutagenesis (Bayer et al., 2004).
All BAHD-ATs are plant-specific and the members of this family have some unique role in plants. For example, Peng et al. (2016) showed the involvement of rice (Oryza sativa L.) BAHD-ATs in the synthesis of an aromatic amine conjugate named phenolamides, which are hydroxycinnamoyl acylated products of the phenolic acid. The natural diversity in the phenolamides is derived from the allelic variation of the BAHD-AT genes responsible for their biosynthesis. A member of the rice BAHD-AT genes fmaily, OsAT1, imparted resistance responses against bacterial blight and blast diseases in rice (Mori et al., 2007). The rice line overexpressing OsAT10 exhibited an alteration in the content of ferulic acid and paracoumaric acid (hydroxycinnamic acids) linked to the cell wall polysaccharide matrix and led to the enhanced saccharification (20-40%) (Bartley et al., 2013). The plant BAHD family enzymes have complex substrates (a variety of acyl-donors and acceptors) and different products. It is very difficult to predict the substrate specificity of uncharacterized BAHD family enzymes because (a) of the minimal sequence similarity among the members of BAHD gene family and (b) the multisite acylation pathways of BAHD-family acyltransferases are poorly understood (Bontpart et al., 2015;Chiang et al., 2018;Wang et al., 2020). The first BAHD gene to be functionally identified was glossy2 (Zea mays L.)/CER2 [Arabidopsis thaliana (L.) Heynh.], which was associated with extending epicuticular waxes to limit water loss and protect against pathogenic invasion (D'Auria, 2006). Recently, Xu et al. (2021) identified and characterized the expression of the BAHD family during development, ripening, and stress response in banana (Musa acuminata Colla). These studies suggest the diverse roles of BAHD-ATs and offer opportunities to explore their new roles and functions.
Sheath blight is an emerging and destructive disease in rice, which is caused by a soil-borne fungal pathogen Rhizoctonia solani AG1-IA. It is a necrotrophic fungal pathogen with a broad host range that can infect plant species from 32 taxonomic groups (Molla et al., 2020). The lack of significant resistance against sheath blight in wild and cultivated rice species limits the conventional plant breeding or genetic techniques to impart sheath blight resistance in rice. Sheath blight appears to be a quantitative trait that is controlled by the combinatorial action of several genes with minor effects . Though several minor quantitative trait loci (QTL) for resistance to sheath blight were mapped in rice, their validity is still not confirmed, and no major resistant gene has been identified to date (Channamallikarjuna et al., 2010;Yuan et al., 2019). In addition, several factors, including plant architecture and environmental conditions, also influence the disease phenotype (Jia et al., 2013) and therefore, confound the trait dissection. Identification of susceptibility genes could offer a promising solution to engineer rice for enhanced tolerance to sheath blight, but a lack of thorough understanding of mechanisms of rice-R. solani AG1-IA interaction at the molecular level makes this task difficult. The recent advancement in the genomic tools, RNA-sequencing (RNA-seq) and methylome analysis, helps to provide various datasets that can be comprehensively integrated to develop the functional footprint of phenotype and the underlying mechanisms of hostpathogen interactions (Karre et al., 2017(Karre et al., , 2019A. Kumar et al., 2016). Owing to the diverse role of the BAHD-AT gene family in plants, here, we performed genome-wide identification, comparative phylogenetic analysis, and expression profiling of rice BAHD-ATs in response to an Indian isolate of R. solani AG1-IA in two rice genotypes with varying level of tolerance to sheath blight. In addition to using whole-genome bisulfite sequencing, we also performed the cytosine DNA methylation profiling of rice BAHD-AT genes. Based on the expression profiling, the putative candidate genes were identified and their plausible role in rice-R. solani AG1-IA interaction is discussed.

Data retrieval
The complete rice genome and the protein sequence data sets were downloaded from Rice Genome Annotation Project

Identification of BAHD-ATs gene family
We retrieved the Hidden Markov Model profile of transferase domain (PF02458.15) from Pfam v31.0 and used it to identify the putative BAHD-AT proteins in rice, Arabidopsis, soybean,

Core Ideas
• OsBAHD-AT genes involved in acylation of secondary metabolites identified in the rice genome. • OsBAHD-AT genes differentially expressed in their response to R. solani AG1-IA infection. • Higher expression of OsBAHD-AT genes in susceptible genotype suggested role in susceptibility. • The epigenetic regulation via cytosine DNA methylation of OsBAHD-AT genes was revealed.
sorghum, maize, and B. distachyon protein sequence datasets. For the identification, the 'hmmsearch' of HMMER V3.2 was performed, with a cut-off e-value 1 × 10 −4 . The presence of transferase domain, as well as other domains, in the identified putative BAHD proteins was determined by using the PfamScan program (parameters: -e_seq 1e −04 -e_dom 1e −04 -clan_overlap). The putative BAHD proteins were then further subjected to 'find individual motif occurrences' search for the presence of two signature motifs-HXXXDG and DFGWG-with an e-value of 1 × 10 −3 . Parallelly, we also used the genetically or biochemically characterized BAHD-AT sequences from the previously published data (D'Auria, 2006) for local BLASTp search against the rice protein dataset to identify any left-out rice BAHD-AT member.

Chromosomal mapping, nomenclature, gene duplication, structural analysis, and in silico characterization of rice BAHD-ATs
We retrieved a general feature format (gff) file from The Rice Annotation Project to extract the genomic coordinates of rice BAHD-AT encoding genes. The extracted genomic coordinates were then mapped on the 12 rice chromosomes using the MapInspect program (http://mapinspect.apponic.com/). Since the NCBI and other databases have obscure nomenclature of rice BAHD-AT proteins, the identified rice BAHD-AT proteins were named by adding the suffix 'OsBAHD' to designate rice BAHD-AT protein and numbered according to the genomic coordinates of the corresponding gene model from top to bottom on 1 to 12 rice chromosomes (Table 1). For the duplication analysis of the rice BAHD-AT family, the result of all-vs.-all BLASTp analysis of total rice protein sequences was used to identify the collinear pairs of BAHD-AT genes using the MCScanX tool with default parameters (Y. . The extracted genomic coordinates were also used to schematically represent the intron-exon structure of OsBAHD-ATs using Gene Structure Display Server (gsds.cbi.pku.edu.cn/). The subcellular localization  of OsBAHD-AT gene family members was predicted by using web servers, CELLO v.2.5 (Yu et al., 2006) and WoLF PSORT (Horton et al., 2007).

Phylogenetic and sequence analysis
The complete protein sequences of identified BAHD-AT gene family members of rice, Arabidopsis, soybean, sorghum, maize, and B. distachyon were used to determine their evolutionary relationship in different plant species. The multiple sequence alignment of complete protein sequences was performed through ClustalW2 with default parameters, and the phylogenetic tree was constructed using MEGA v6.06 (https://www.megasoftware.net). Tree topology was inferred by the neighbor-joining method with 1,000 bootstrap replicates.

Plant material and inoculation with R. solani AG1-IA
Two rice genotypes, that is, a germplasm line designated as PAU-ShB8, a moderately resistant genotype (hereinafter referred to as ShB) and PR114, a susceptible genotype (hereinafter referred to as PR) with varying levels of susceptibility to R. solani AG1-IA (Aggarwal et al., 2019), were grown in pots containing sterile soil mix. The fungal inoculum was prepared by placing a single sclerotium (compact mass of hardened mycelium) of R. solani AG1-IA on a potato dextrose agar (HiMedia) plate and incubating it at 28˚C under dark conditions until the plate was full of fungal sclerotia (∼4 d). The 7-8-week-old rice plants were inoculated at the second top-most collar of the tiller by placing a single sclerotium on the inner side of the sheath (Ghosh et al., 2017). The inoculated plants were maintained at 28˚C temperature, 80% relative humidity, and 16/8 h light/dark photoperiod inside a growth chamber (Conviron). The sheath samples containing both the healthy and lesion contacting part ∼1 cm up and down the site of inoculation were collected at 0, 24, 48, and 72 h postinoculation (hpi). The samples were collected in three biological replicates for each genotype, where a replicate consisted of a pooled sheath from four to five tillers of a rice plant, immediately frozen in liquid nitrogen, and stored at −80˚C until further use.

RNA-seq analysis
The 100-mg tissue samples were used for total RNA isolation using the RNeasy Plant Mini Kit (Qiagen) as per the manufacturer's instructions. To remove the DNA contamination, an on-column DNase treatment was given to the isolated total RNA. The NanoDrop UV-VIS spectrophotometer and denaturing formaldehyde agarose gel were used to determine the concentration and the integrity of isolated RNA, respectively. The RNA-seq libraries for all time intervals were prepared using TruSeq RNA sample preparation kit v2 (Illumina Inc.) following the manufacturer′s instructions. The library quantification and the insert size determination were performed using Qubit 2.0 fluorometer (Life Technologies) and Agilent Bioanalyzer DNA 1000 chip (Agilent Technologies), respectively. The prepared libraries were sequenced for paired-end (2 by 100) sequencing using NovaSeq 6000 system (Illumina Inc.) at Council of Scientific and Industrial Research-Institute of Himalayan Bioresource Technology, Palampur. The raw reads were filtered using Trimmomatic v 2.39 (Bolger et al., 2014), and the filtered high-quality reads were then mapped to the rice reference genome (https://rapdb.dna.affrc.go.jp). The raw reads data is submitted to NCBI Sequence Read Archive (SRA) as a BioProject with accession number PRJNA702874. The data were analyzed for differential gene expression using TopHat and Cufflinks tools (Trapnell et al., 2012). The value of transcript abundance was presented as fragments per kilobase of exon model per million mapped reads (FPKM). The log 2 fold change in the FPKM values was employed for the heat map construction with hierarchical clustering by using the MeV v4.4.1 software package (http://mev.tm4.org).

qRT-PCR analysis
The 1.0 μg of total RNA was converted to complementary DNA (cDNA) using the Verso cDNA synthesis kit as per manufacturer instructions (Thermo Fisher Scientific). The cDNA was 10-fold diluted before performing the quantitative real-time polymerase chain reaction (qRT-PCR) assays in a 20-μl reaction using SYBR green PCR master mix (Thermo Fisher Scientific, USA), on a real-time PCR thermocycler (Agilent, USA). Each qRT-PCR assay was performed with three independent biological replicates, each consisting of three independent technical replicates. The transcript abundance of ß-tubulin was used to normalize the qRT-PCR data before calculating the relative fold-change employing the delta-delta C t (cycle threshold) method (Livak & Schmittgen, 2001). The primer sequences used in the present study are listed in Supplemental Table S1.

Bisulfite sequencing and analysis
The genomic DNA was isolated using GenElute Plant Genomic DNA Miniprep Kit (Sigma-Aldrich) and fragmented to 200-300 bp via sonication using Adaptive Focused Acoustics (Covaris

Identification and annotation of BAHD-AT gene family members in different plant species
The 'hmmsearch' for the transferase domain (PF02458.15) resulted in the identification of a total of 145 rice proteins. Among the identified 145 proteins, the pfamscan, with 1 × 10 −4 e-value cutoff, confirmed the presence of transferase domain in only 92 rice proteins. Subsequently, the 92 candidate proteins were then subjected to 'find individual motif occurrences' search, with 1 × 10 −3 e-value cutoff, to identify the signature motifs, that is, HXXXDG and DFGWG, which reduced the total putative BAHD-ATs to 85. Parallelly, the BLASTp search of already characterized BAHD-AT proteins against the rice protein dataset resulted in the identification of 86 putative rice BAHD-AT proteins, where the subsequent pfamscan showed the absence of BAHD transferase domain in one protein and thus excluded from the analysis. By combining the results of 'hmmsearch' program and BLASTp, we were able to identify a total of 85 rice BAHD-AT proteins encoded by 85 rice gene models. The rice BAHD-AT gene annotations reported in the public databases (NCBI and UniProt) are highly disordered; therefore, we assigned a uniform nomenclature to the 85 rice BAHD-AT genes. The rice BAHD-AT proteins were designated as OsBAHD, followed by Arabic numbers 1-85 according to the position of their encoding genes on linkage groups (LGs) 1-12 in 5′ to 3′orienta-tion. Similar criteria of nomenclature have been employed for the NAC proteins in potato (Solanum tuberosum L.) (Singh et al., 2013), HyPRP proteins in rice (Kapoor et al., 2019), and MADS-box proteins in apple (Malus domestica (Suckow) Borkh.) (G. Kumar et al., 2016a). The average length of identified rice OsBAHD proteins was ∼450 amino acids that range from 150 to 623 amino acids (Table 1). The average molecular weight of OsBAHD proteins was 48.4 kDa, where the lowest and the highest molecular weights were 16.19 and 66.67 kDa, respectively (Table 1). In addition to rice, we also identified BAHD-ATs in two dicot plants (Arabidopsis and soybean) and three monocot plants (Sorghum, maize, and B. distachyon) using a similar approach. A total of 41, 102, 83, 75, and 74 BAHD-AT encoding genes were identified in Arabidopsis, soybean, sorghum, maize, and B. distachyon, respectively. The list of BAHD-AT genes identified in these plant species, along with their characteristics, is provided as Supplemental Table S2.

OsBAHD-AT chromosomal distribution, duplication events, structural analysis, and subcellular localization of encoded proteins
The genomic coordinates for genes encoding 85 OsBAHD-ATs were used to map them on 12 rice LGs. The 85 members of the OsBAHD-AT gene family showed non-random distribution on 12 rice LGs, where more than 76% of OsBAHD-AT genes were found to be localized on six LGs (LG1, LG2, LG4, LG6, LG8, and LG10) with the highest of 16 OsBAHD-AT genes on LG10 (Figure 1). Often, the clustered distribution on the chromosomes is observed for the genes that belong to a family. The clustering of OsBAHD-AT genes showed 15 clusters comprised of a total of 51 OsBAHD-AT genes distributed on eight rice LGs (Figure 1). The LG2 and LG10 contain the largest clusters comprised of seven and eight genes, respectively. Additionally, the LG6 and LG10 have a maximum number of three clusters, each with 10 and 13 genes, respectively, while LG1, LG2, LG4, LG9, and LG11 have either one or two gene clusters.
The analysis of the rice complete genome sequence has revealed the ancient whole-genome duplication and recent segmental duplication in the rice genome (Yu et al., 2005). From the duplication analysis of the rice genome, we identified 12,127 (17.89%) genes as tandemly duplicated and 5,654 (8.34%) genes as segmental duplicated. These duplication events in the rice genome may also lead to the expansion of the OsBAHD-AT gene family in rice. Among all the OsBAHD-AT genes, 10 were located on the segmentally duplicated regions of LG1, LG2, LG4, LG5, and LG8 (Supplemental Figure S1). The three OsBAHD-AT pairs were present in duplicated segments on each LG1 and LG5, followed by two OsBAHD-AT genes on LG4 and one each on LG2 and LG8. Additionally, F I G U R E 1 Chromosomal mapping and the clustering of 85 OsBAHD-AT encoding genes on 12 rice linkage groups (LGs). The genes shown in boxes indicate their presence within the 200 Kbp region to form a gene cluster. The clade to which genes present in the cluster belongs is shown in parenthesis 23 OsBAHD-AT genes were found to be tandemly duplicated (Supplemental Figure S2). Therefore, these results suggested that the duplication events in the rice genome have contributed to the expansion of the OsBAHD-AT gene family. Results from the gene structure analysis of OsBAHD-AT genes showed that the majority of them (36 genes, 42%) have either single intron (37 genes, 43.5%) or intronless (36 genes, 42.3%), while the remaining genes (12 genes, 14%) have introns that vary from two to three in number (Supplemental Figure S3).
The OsBAHD-ATs were further predicted for their localization in the different subcellular components such as chloroplast, cytoplasm, and plasma membrane using CELLO and WoLF PSORT (Supplemental Table S3). According to CELLO prediction, OsBAHD-ATs scored relatively higher for their localization in the chloroplast, whereas WoLF PSORT predicted OsBAHD-AT gene localization mainly in the cytoplasm. Several OsBAHD-ATs are potentially localized in different subcellular compartments and also showed multiorganellar localization in Golgi bodies and plastids (dual), cytoplasm, and nucleus (dual), chloroplast, nucleus, mitochondria, peroxisome, and vacuoles, etc. In total, the differences in the prediction of OsBAHD-ATs were notable between the two prediction programs (Supplemental Table S3). To date, no prior information is known about the characterization of molecular function and subcellular localization of OsBAHD-ATs in rice. These results indicate the possible function of BAHD-ATs in different organelles including the cytosol. These findings were concurrent with the findings of Yu et al. (2009) andTuominen et al. (2011) in Populus and Arabidopsis. The prediction of BAHD-ATs in different subcellular compartments apart from the cytosol suggests the diverse functionality of BAHD-ATs in the biosynthesis and modification of a repertoire of metabolites, primarily the plants' secondary and specialized metabolites (Peng et al., 2016). Similarly, Bontpart et al. (2015) stated that the concomitant activities of different BAHD-ATs in different compartments may synergistically promote the accumulation of the metabolites. Also, acylated defense molecules synthesized in the cytosols are possibly toxic to the plant, so vacuolar and multicellular localization and trafficking activities may offer an alternate means of coping with this scenario at various stages of growth, tissues, and environmental conditions.
The OsBAHD-ATs sequences were used to generate the phylogenetic relationship with the characterized BAHD -ATs from other plant species (Bontpart et al., 2015;D'Auria, 2006). The phylogenetic analysis helped us to classify the OsBAHD-ATs in five different clades from I to V (Figure 2). This classification may indicate the possible functional similarity of OsBAHD-ATs with that of characterized BAHD-ATs. Most of the OsBAHD-ATs were grouped in clade V with 38 OsBAHD-AT members, followed by 31 members to clade The Plant Genome F I G U R E 2 Phylogenetic relationship and the conserved motif composition of OsBAHD-ATs. (A) The phylogenetic tree of OsBAHD-AT proteins and biochemically characterized BAHD-AT proteins of other plants. The multiple sequence alignment and the phylogenetic tree construction were carried out using MEGA v6.06 using the maximum likelihood method with 1,000 bootstrap replicates. The tree topology revealed the five classical phylogenetic clades (I to V). (B) The MEME analysis revealed the conserved motifs in OsBAHD-ATs and their schematic representation. Each motif shown here is represented by a colored box with a consensus given at the bottom I, 12 members to clade IV, and four members were classified in clade II (Figure 2). The clade III of the characterized BAHD-ATs contains none of the BAHD-AT members from rice and other monocots, and this suggests the clade III members are dicot specific and may have some exclusive func-tion in dicot species. Interestingly, clade IV, which contains only one characterized BAHD-AT from barley (AA073071), has 12 members of OsBAHD-ATs with significant sequence similarity as determined through high bootstrap value (Figure 2).

F I G U R E 3
Phylogenetic relationship of OsBAHD-ATs with that of Arabidopsis, G. max, Z. mays, S. bicolor, and B. distachyon. The amino acid sequence of mature BAHD-AT proteins was used for sequence alignment and subsequent construction of the phylogenetic tree employing MEGA v6.06. The different shapes with color codes are given to the BAHDs of different plant species. The sphere indicates the monocots, while the diamond indicates the dicots. The color codes in different shapes are as follows: red for rice, orange for S. bicolor, blue for Z. mays, magenta for B. distachyon, lawn-green for Arabidopsis, and cyan for G. max. The tree was divided into 11 major clades named sequentially from A to K To analyze the phylogenetic relationship of rice BAHD-ATs with that of other closely related monocots (Sorghum, maize, and B. distachyon) and some distant dicots (Arabidopsis and soybean) species, we constructed an unrooted phylogenetic tree for the OsBAHD-ATs and BAHD-AT proteins of other plant species. The phylogenetic tree classifies 85 OsBAHD-ATs into 11 distinct phylogenetic clades (A to K) along with their orthologs in other monocot and dicot plant species (Figure 3). Although the BAHD-ATs were found to be distributed uniformly among the major clades, it is apparent that the BAHD-ATs of monocots and dicots clustered into distinct subclades of major phylogenetic clades. Exceptionally, clade I contains only monocot-specific BAHD-ATs, while clade H contains only the dicot-specific BAHD-ATs (Figure 3). The subclade I(b) contains BAHD-AT members that belong to the clade IV of the phylogenetic tree containing characterized BAHD-AT members (Figures 2 and 3). Interestingly, I subclade contains none of the dicot BAHD-AT members, and this observation indicates that subclade I specifically belong to the monocot species (Figure 3).
Among the genetically and biochemically characterized BAHD-ATs (D'Auria, 2006), there is no apparent specificity for the acyl -donors and -acceptors among the members of a given clade. Broadly, the members of clade I use malonyl-CoA as an acyl donor to acylate the phenolic compounds. Similarly, clade V members mainly utilize benzoate-CoA and hydroxycinnamoyl-CoA to acylate various metabolites including terpenoids, alkaloids, and medium-chain aliphatic and aromatic esters (D'Auria, 2006) (Bontpart et al., 2015). However, the hydroxycinnamoyl-CoA has been reported as an acyl donor for members of multiple clades (D'Auria, 2006). Mutagenesis coupled crystallographic structure analysis of two different BAHD-ATs revealed the structural diversity among the binding sites for acyl-donor and -acceptor molecules (Unno et al., 2007). It also provides plausible cause for the utilization of vast diversity of acceptor molecules and relatively limited acyl-donors by different BAHD-ATs, thus, a potential challenge for phylogeny-based inference on the functional similarity of the other BAHD-AT family members.

Expression profiling of OsBAHD-AT genes in response to R. solani AG1-IA identified several putative candidate genes
The tissue type has a profound effect on gene expression, therefore our expression experiment had the tissue type from the rice sheath where the fungus inoculation and the infection was established. This strategy helped to identify the localized as well the global response of the fungus infection. The log 2 fold-change in the OsBAHD-AT genes expression vs. 0 hpi is given as a heatmap (Figure 4). Out of a total of 85 OsBAHD-AT genes, 33 exhibited expression in all the samples collected after R. solani AG1-IA inoculation. The RNAseq data suggest that the expression of OsBAHD-AT gene family members in germplasm line ShB was mostly downregulated or remained unchanged upon R. solani AG1-IA inoculation. In contrast, 29 out of 33 OsBAHD-AT genes were upregulated and only four were downregulated at one or other time-point after inoculation in susceptible variety PR. However, the genes, namely OsBAHD28, OsBAHD53 of clade I, and OsBAHD63 of clade V, exhibited upregulated expression in ShB during the later stages of postinoculation. Interestingly, the OsBAHD-AT genes exhibiting dysregulation of expression (change in expression vs. control) mostly belonged to clade I, and the majority of them (12 out of 20 genes) were either downregulated or had reduced expression in ShB upon R. solani AG1-IA inoculation. Importantly, in contrast, these genes of clade I exhibited higher expression in susceptible genotype PR. The OsBAHD38, OsBAHD57, and OsBAHD80 genes had higher expression in ShB after F I G U R E 4 The heat map representation for the transcript abundance level of OsBAHD-AT genes in R. solani AG1-IA challenged rice genotypes viz. PR and ShB at different time intervals. The number of an asterisk (*) next to the gene name indicates the clade to which a particular gene belongs R. solani AG1-IA inoculation; however, the expression of these genes was comparatively more in PR. This data suggests the importance of clade I OsBAHD-AT genes in rice sheath blight disease. Also, the genes of clade IV, namely OsBAHD18 and OsBAHD62, except OsBAHD31, had lower expression in the susceptible vs. the resistant genotype. To validate the RNA-seq-based expression profiling, the qRT-PCR assay was conducted for the randomly selected genes, that is, OsBAHD01, OsBAHD09, OsBAHD11, OsBAHD44, OsBAHD51, OsBAHD56, and OsBAHD57 in the R. solani AG1-IA challenged samples of ShB and PR. Since qRT-PCR and RNA-seq data of the selected genes shows a common pattern of expression in the respective rice genotypes, the qRT-PCR data corroborate the RNA-seq data ( Figure 5).
In one of the previous studies, the expression of the OsBAHD57 was found to be upregulated by 3.7-and 2.8-fold in two susceptible rice genotypes (PB1 and TP309, respectively) challenged with R. solani AG1-IA (Ghosh et al., 2018). In the present study, the OsBAHD57 has similar upregulated expression (>1.9-fold) in the susceptible genotype PR upon fungal inoculation, while its expression remains unchanged in resistant genotype ShB (Figures 4 and 5). Altogether, it can F I G U R E 5 The qualitative real-time PCR (qRT-PCR)-based expression of selected OsBAHD-AT genes in R. solani AG1-IA inoculated rice genotypes. All the qRT-PCR assays were performed with three biological and their respective technical replicates. Error bars represent the standard error in the replicates be concluded that the OsBAHD-AT gene family has an important role in rice-R. solani AG1-IA interaction. The BAHD-ATs acylate the flavonoids and the other secondary metabolites to diversify their biological role in mitigating the environmental stresses (Bontpart et al., 2015), the higher expression of BAHD-AT gene family members in susceptible rice genotype PR suggests their negative role in rice sheath-blight resistance ( Figure 4). One of the previous studies reported that a mutant of a novel member of the BAHD-AT gene family, eps1 in Arabidopsis, had compromised resistance against Pseudomonas syringae because of reduced pathogen-induced salicylic acid (SA) accumulation and expression of pathogenesis-related genes (Zheng et al., 2009); where the exogenous SA application restored the disease resistance that suggests EPS1 works upstream of SA. However, it was also observed that Arabidopsis overexpressing ABS1, another novel BAHD-AT, has the resemblance of brassinosteroid (BR) deficient or signaling mutant with dwarf phenotype, which can be rescued by BR treatment, thus state the function of ABS1 upstream to BR (M. . Likewise, the BRASSINOS-TEROID INSENSITIVE 1 (BIA1) and BIA2 BAHD-AT-like proteins were found to be involved in the inactivation of BR to maintain the BR homeostasis in Arabidopsis (Gan et al., 2020;Roh et al., 2012;Zhang & Xu, 2018). These studies thus F I G U R E 6 Change in the methylation level at individual methylated cytosine site of OsBAHD-ATs in different sequence contexts, in the control and R. solani AG1-IA inoculated rice genotypes, is shown via box plot. Each box shows the interquartile range for the number of methylated cytosines established the link between BAHD-ATs and BR biosynthesis and signaling. The exogenous application of brassinolide (a most common and important BR) in tobacco and rice was observed to enhance the disease resistance in tobacco and rice against biotrophic fungi and suggested that the BRmediated disease resistance works independently of the systemic acquired resistance (Nakashita et al., 2003). It is also evident that an exogenous application of BRs has protective effects against a broad range of plant pathogens including fungi, bacteria, and viruses (Bajguz & Hayat, 2009). In a challenge to this prevailing view that BRs positively regulated the disease resistance in rice, a study showed that rice pathogen Pythium graminicola used BRs as a virulence factor to infect rice possibly mediated through negative cross-talk with SA and gibberellic acid pathways (De Vleesschauwer et al., 2012). Similarly, it was recently found that rice BRI1 mutant d61-1 (BR signaling mutant) and the mutant of BR biosynthesis gene d2 are highly resistant to R. solani (Yuan et al., 2018). Though these pieces of evidence are conflicting to define the role of BR in rice sheath blight resistance, the detailed study in Arabidopsis has demonstrated both the antagonistic and synergistic effect of BR in plant's innate immunity and is a function of BR homeostasis (Albrecht et al., 2012;Belkhadir et al., 2012). Additionally, it was suggested that the variation in the plant disease resistance response through pattern-triggered immunity depends on the relative levels of BR and its receptor proteins BRI1 and BAK1 (BRI1 associated kinase 1) (Wang, 2012). This relative wealth of information on BAHD-ATs and BRs suggests that the upregulation of OsBAHD-AT genes in PR may perturb the BR homeostasis and eventually BR-mediated disease resistance, thus explain the plausible enhanced susceptibility to sheath-blight disease in PR. However, conclusive experimental evidences on how exogenous BR treatment affect the BAHD expression and eventually disease tolerance are required to validate this hypothesis.

Methylation status of OsBAHD-ATs in R. solani AG1-IA challenged rice
Epigenetic modification is a well-known strategy used by plants to adapt to various environmental changes without changing the genetic sequence. DNA cytosine methylation is an epigenetic modification that affects gene expression and is thus associated with transcriptional regulation in various plants (Gallego-Bartolomé, 2020; G. Kumar et al., 2016b). Therefore, in the present study, we analyzed the dynamics of cytosine DNA methylation in the OsBAHD genes and their 2-kb flanking regions at three sequence contexts, that is, CG, CHG, and CHH. The analysis of the methylation level of OsBAHD-AT genes showed a higher percentage of methylated cytosine in the CG context, followed by CHG and CHH in all the analyzed samples of both genotypes (Figure 6). Interestingly, the percentage of methylated cytosines of OsBAHD-AT genes in resistant genotype ShB gradually increased after fungal inoculation in all the sequence contexts (CG, CHG, and CHH). In contrast, the overall reduction in methylated cytosines was observed in susceptible genotype PR.
The analysis of differential methylated regions (DMRs) of OsBAHD-AT genes showed more hypomethylation at all the time points postinfection (24, 48, and 72 hpi) in PR, whereas in ShB the number of hypo-and hypermethylated regions remained nearly constant (Figure 7). In the comparison between genotypes, ShB vs. PR, the higher number of DMRs were hypomethylated in PR genotype at corresponding time points as compared with hypermethylated regions. The reduction in total methylation level ( Figure 6) and hypomethylation of the OsBAHD-AT genes (Figure 7) may explain their corresponding upregulated and higher expression in the PR genotype ( Figure 5). Though no previous reports are available on epigenetic modification of BAHD-AT genes, from our results it may be inferred that the expression of OsBAHD-ATs encoding genes is epigenetically regulated in R. solani challenged sheath blight susceptible genotype PR.

CONCLUSIONS
The present study identified and characterized the BAHD-AT gene family comprised of 85 members. The genome organization of OsBAHD-AT gene members showed nonrandom distribution and localized as clusters on all 12 rice LGs. The duplication analysis showed that the expansion of the rice BAHD-AT gene family has mostly occurred through tandem duplication. The phylogenetic analysis of OsBAHD-AT genes along with other characterized BAHD-AT proteins of other plant species enabled us to classify the OsBAHD-ATs in five distinct clades, where none of OsBAHD-AT members belongs to clade III. The high-throughput RNA-seq analysis allowed us to determine the expression of OsBAHD-AT genes, where the expression of 33 members was detected in R. solani AG1-IA challenged rice genotypes having differential tolerance to sheath blight disease. The changes in cytosine methylation level and DMR indicates the epigenetic regulation of rice BAHD-AT gene expression in response to R. solani. The population-wide DMR studies for BAHD gene loci may develop the confidence of using OsBAHD-AT genes as epiallele for rice crop improvement strategies. We also discussed the plausible link in the OsBAHD-AT gene expression, BR homeostasis, and the sheath blight resistance; however, further experimental validation is required. This systematic study of OsBAHD-AT genes provided a foundation for future gene functional characterization and their potential applications in rice genetic improvement for sheath blight resistance. Altogether the present study indicated an importance of OsBAHD-AT genes to devise strategies for the development of sheath blight disease-resistant rice cultivars.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.