Structural variation of GL1 gene determines the trichome formation in Brassica juncea

A 448 kb region on chromosome B02 was delimited to be associated with trichome trait in Brassica juncea, in which the BjuVB02G54610 gene with a structural variation of 3 kb structure variation (SV) encoding a MYB transcription factor was predicted as the possible candidate gene. Mustards (Brassica juncea) are allopolyploid crops in the worldwide, and trichomes are essential quality attributes that significantly influence its taste and palpability in vegetable-use cultivars. As important accessory tissues from specialized epidermal cells, trichomes also play an important role in mitigating biotic and abiotic stresses. In this study, we constructed a F2 segregating population using YJ27 with intensive trichome leaves and 03B0307 with glabrous leaves as parents. By bulked segregant analysis (BSA-seq), we obtained a 2.1 Mb candidate region on B02 chromosome associated with the trichome or glabrous trait formation. Then, we used 13 Kompetitive Allele Specific PCR (KASP) markers for fine mapping and finally narrowed down the candidate region to about 448 kb in length. Interestingly, among the region, there was a 3 kb sequence deletion that located on the BjuVB02G54610 gene in the F2 individuals with trichome leaves. Genotyping results of F2 populations confirmed this deletion (R2 = 81.44%) as a major QTL. Natural population re-sequencing analysis and genotyping results further validated the key role of the 3 kb structure variation (SV) of insertion/deletion type in trichome development in B. juncea. Our findings provide important information on the formation of trichomes and potential target gene for breeding vegetable mustards.


Introduction
Mustards (Brassica juncea) belonging to the family Brassicaceae are often used as sources of condiment, vegetable or edible oil worldwide (Yang et al. 2018). The presence of trichomes influences its pest resistance and palpability as well. And previous studies show that the trichomes densities are inducible by herbivore in B. juncea, and thus, trichomes are supposed to be a defense mechanism to prevent insects (Mathur et al. 2012).
Trichomes are some kinds of epidermal projection which locate on the surface of any part of the plant body, have different shapes, sizes, structures and functions depending on species, and can be differentiated into persistent and ephemeral; unicellular or multicellular; simple or branched; and non-glandular or glandular (Fambrini and Pugliesi 2019). Trichomes play an important role in plant growth and development by protecting them from biotic and abiotic stresses because of its structure traits and chemical reinforcement . Some specialized trichomes become the substances that human beings rely on for survival, such as cotton fibers (Lee et al. 2007) and some glandular secreting trichomes, which can also be considered "chemical factories" as they synthesize and secret many economically important compounds such as artemisinin (Han et al. 2022; Yiqing Meng and Xiaolong Lyu have contributed equally. Pattanaik et al. 2014). For some kinds of vegetable crops and fruits, especially leafy vegetables, trichome traits are considered as the most important external quality traits which affect the taste and consumers' choice . However, trichomes are also undesirable in some crop production because of their negative impact on harvesting Zhang et al. 2012).
Trichomes are found on species within the Brassicaceae, but the density and localization vary among the Brassicaceae (Nayidu et al. 2014). The trichome morphology is also used in determining phylogenetic affinities among Brassicaceae taxa (Beilstein et al. 2008). Like in other plants, trichomes in Brassicaceae affect the adaptation to the environment and protect plants from attack by insect herbivores. For example, in Chinses cabbage, trichomes are important in the regulation of foliar Pb uptake and translocation (Gao et al. 2021). And it is reported that on B. villosa (B. villosa) and on mustard lines, the modest dense trichomes coverage on leaves provide protection against the crucifer flea beetle (Brian and Dawson 2002;Gruber et al. 2006;LAMB and R. 1980;Palaniswamy and Bodnaryk 1994). Similar to other Brassicaceae crops, trichomes in mustards are singlecell and non-glandular and have no branch. Most of the trichomes appear on leaves and stem in the vegetative growth period and disappeared in the reproductive growth period in varieties with trichome. However, in the past, there was not much in-depth research on trichomes of B. juncea, and the application of it mainly focused on variety identification (Kumari et al. 2020), insect resistance identification (Mathur et al. 2011(Mathur et al. , 2012 and detoxification mechanism (Salt et al. 1995).
In Arabidopsis thaliana (A. thaliana), trichomes are found on most aerial parts of the plants, including rosette leaves, stem and cauline leaves but not on the hypocotyl and the cotyledons (Schellmann and Hulskamp 2005). Trichomes originate from the protodermal cells of the developing leaf primordia, which is the same as leaf epidermal cells but differentiate in different morphology (Szymanski et al. 2000). It is found that at the base of young leaves, single cells are spaced out at regular distances in an area of apparently equivalent protodermal cells development into trichomes (Hülskamp et al. 1994;Larkin et al. 1996). Those protodermal cells destined to become trichomes stop mitotic cell divisions and turn to endoreduplication cycles, by which the trichome cells increase in size and change its direction of growth to perpendicular to the leaf surface. The trichome cells undergo at least four rounds of endoreduplication cycles, which result in a DNA content of 32C, accompanied by rapid cell enlargement, form mature trichome with two to three branches (Hülskamp et al. 1994;Melaragno and Coleman 1993;Schnittger and Hulskamp 2002). The mechanism of trichome production and development in A. thaliana is classified as de novo pattern and the activator-inhibitor model produced by over 70 different genes (Khan et al. 2021). According to this model, activators and inhibitors work together to regulate the distribution of trichomes (Pesch and Hulskamp 2009). Three classes of interacting regulators are postulated to form a combinatorial regulators complex, which are the R2R3-MYB transcription factor encoded by GLABRA1 (GL1), the basic helix-loop-helix (bHLH) proteins encoded by GLABRA3 (GL3) and ENHANCER OF GLABRA3 (EGL3), as well as and the WD40-protein encoded by TRANSPARENT TESTA GLABRA1 (TTG1) (Zhao et al. 2008). GL1 plays a central role in the production of trichome. Its loss-of-function mutants exhibit an almost complete loss of trichome initiation, and its overexpression inhibits trichome development by unknown leaf epidermal inhibition program (LEIP) (Larkin et al. 1994;Szymanski et al. 1998). GL1 encodes a member of the MYB family of transcriptional regulators, which contains two repeats of the MYB DNA binding domain at the N terminus and an acidic C terminus with no similarity to other MYB proteins (Larkin et al. 1993(Larkin et al. , 1994Oppenheimer et al. 1991). By analyzing phenotype of various mutants in Arabidopsis, it has been found that GL3 and EGL3 function in a redundant manner (Zhang et al. 2003), while GL3 and EGL3 mediate interaction between MYB and TTG1 (Payne et al. 2000). TTG1 encodes a WD40 PROTEIN and can move freely between epidermal cells and accumulate in cells having high levels of GL3 (Balkunde et al. 2011;Bouyer et al. 2008;Walker et al. 1999). In addition to these genes, a group of R3 MYBs which typically lack the activation domain are also involved in trichome development, and act as negative regulators. Most of these proteins compete with GL1 to bind GL3/EGL3, form a repressor complex, and affect the expression of downstream target, include TRIPTY-CHON (TRY ) (Schellmann et al. 2014;Schnittge et al. 1999), CAPRICE (CPC) (Wada et al. 1997), ENHANCER OF TRY , and CPC 1 (ETC1, ETC2 and ETC3) (Kirik et al. 2004a(Kirik et al. , 2004bWester et al. 2009). TRICHOMELESS 1 (TCL1 and TCL2) acts as a negative regulator of trichome development by affecting the expression of GL1, as well as competing with GL1 for binding to GL3 (Pattanaik et al. 2014). It has also been demonstrated that CPC and ETC3 can move from cell-to-cell to competitively replace the GL1 within the activation regulators complex and stop trichome formation in the neighboring cells (Pattanaik et al. 2014). Collectively, trichome development depends on a group of activators and repressors that form a regulatory loop and tune the expression of downstream gene targets (Pattanaik et al. 2014).
Bulked segregant analysis (BSA-seq) has become a popular approach for rapid gene mapping and is applied to many different species especially in plants (Huang et al. 2022). With the application of next-generation sequencing technology, the precision and accuracy of plant genomes have been greatly improved, which lay the foundation for BSA-seq. Several high-coverage genomes of mustards have been released based on different sequencing strategies (Yang et al. 2016Paritosh et al. 2021). The assemble of these high-quality genomes and pan-genomes accelerates gene mapping of traits and assesses the potential role of SVs in genotype-phenotypic variation.
In this research, we conducted BSA in the F 2 population derived from YJ27, an oil-use variety, and 03B0307, a stem-use variety as parents, to explore the genomic regions responsible for the trichome or glabrous trait. The candidate region was narrow down by KASP markers, and the candidate gene BjuVB02G54610 included a 3 kb structural variation (SV) as confirmed genotyping results. It was homologous with AT3G27920, which encodes GL1, a MYB transcription factor required for induction of trichome. Our finding provides important gene resource and insights into structural variation associated with trichome or glabrous trait formation in B. juncea.

Plant materials
Two inbred line, YJ27 which had trichome leaves, and 03B0307 which had leaves with glabrous adaxial surface, were used as parental lines to generate F 2 for BSA and fine mapping. A total of 1304 F 2 individuals were grown in 2019 and used for BSA-seq. To confirm the phenotypes, 128 natural varieties were plant in 2022; each variety planted three individuals as biological repetition. The majority of experiments were conducted under greenhouse condition at Zhejiang University, Hangzhou, China. For greenhouse experiments, the plants were germinated in 50 well planting trays and grown at 14 h/10 h (light/dark, photoperiod) and 22 °C (temperature). The upper diameter of the well was 48 mm, the bottom was 18 mm, and the depth was 93 mm. Standard practices for irrigation, fertilization, and control of pests and diseases were followed.

Evaluation of trichome or glabrous phenotypes
Trichome number of two parental lines on every leaf was calculated during 2017; after taking the photographs of leaves, we manually calculated the number of trichomes in Photoshop software through the mouse-clicking counting tool provided by the software. The F 2 population was evaluated in 2019, and the number of trichome on the second leaf of each F 2 individual was used to analyze the trichome or glabrous trait. The trichome number frequency distribution histogram of the 1304 F 2 individuals was drawn by GraphPad Prism 8 (https:// www. graph pad-prism. cn/) using the method of nonlinear regression (curve fit), with a group spacing of 5. For the natural population, the trichome number of each biological repetition was evaluated, and the average of them was regarded as the trichome number of this variety.

Bulked segregant analysis sequencing
In order to apply BSA-seq in this research, two F 2 pools and two parental pools were constructed. Each F 2 pools contained DNA from 50 individuals with trichome leaves or glabrous leaves, respectively. Re-sequencing of four pools was completed with the help of Novogene Company. According to their process, qualified DNA samples were randomly broken into fragments with a length of 350 bp by Covaris crusher. After treating, the constructed libraries were sequencing by illumina HiSeq™ PE150. Raw reads were trimmed to obtain high-quality clean reads by removing reads pair with adapters, paired reads which the content of N contained in the single-end sequencing read exceeded 10% of the read, and which the number of low-quality (Q ≤ 5) bases contained in a single-end sequencing read exceeds 50% of the length of the read. Clean reads were aligned and mapped to the B. juncea reference genome 2.0 (http:// brass icadb. cn/#/ Downl oad/) using BWA software (https:// github. com/ lh3/ bwa) with the parameter of mem − t 4 -R. Then, the read groups were added, sorted and indexed by SAMtools (https:// github. com/ samto ols/ samto ols) with the parameter of -@ 4. SNPs in multiple samples were detected by GATK software (https:// github. com/ broad insti tute/ gatk/ relea ses), and filtered using VariantFiltration, and the filtering parameter was -filter-expression "QD < 2.0 || FS > 60.0." The filtered VCF file in table format was used as an input file for analyzing the significance of QTLs and plotted Manhattan plots using the method of QTL-seq analysis and G'analysis in R studio developed by Mansfeld et al. (Mansfeld and Grumet 2018), by which SNPs were filtered by "refAllele-Freq = 0.20, minTotalDepth = 10, maxTotalDepth = 400, minSampleDepth = 5, minGQ = 50." G' analysis was calculated with a 1 Mb slide window; QTL-seq analysis was calculated with the sliding window whose size also 1 Mb and repeated 1000 times for each read depth to generate confidence intervals; thereby, the 95% and 99% confidence level were selected as the screening threshold. All commands and codes for BSA-seq analysis are available at https:// github. com/ bmans feld/ qtlse qr.

DNA isolation
Fresh leaves were collected from each plant, including parents, F 1 and F 2 population, and natural population. Genomic DNA was extracted using the CTAB method (CIMMYT 2005). For BSA-seq, DNA of F 2 individuals and parents was extracted separately, and then treated by RNase. The purity and integrity of DNA were analyzed by agarose gel electrophoresis. The purity of DNA was also detected by Nano drop with OD 260/280 ratio. And DNA concentration was quantified accurately by Qubit with the help of Novogene Company. For genotyping and cloning, the concentration of DNA of F 2 individuals, parents, F 1 individuals and natural varieties was adjusted to 50 ng/μL.

Marker development and genotyping
KASP markers were development based on the SNPs in the candidate interval from BSA-seq to narrow down the region. All these SNP markers were genotyped in the mapping population using competitive allele-specific PCR assays based on KASP™ technology (LGC genomics, Teddington, Middlesex, UK) at LGC genomics. A total of 13 successful markers were designed. In addition, two markers based on the SV we found in the region were also design with Primer 5 (https:// www. biopr ocess online. com/ doc/ primer-premi er-5design-progr am-0001), one of which was flanking the SV, and another was on the SV. The primer sequences of KASP markers and SV markers developed in this research are presented in Table S1 and Table S2. 212 F 2 individuals, as well as parental individuals, were genotyped with the KASP markers for fine mapping. Later, 292 F 2 individuals, parental individuals and one F 1 individuals were genotyped with the SV markers. To demonstrate the genotyping result of F 2 individuals, the re-sequencing data of several varieties with or without trichomes were analyzed and visualized with IGV software. One hundred and twenty-eight natural varieties were also genotyped with the SV markers.

Phenotypic differences of trichomes in Brassica juncea accessions
Trichomes on the adaxial surface were easy to observe and count. However, because of the characteristics of leaf and trichome, it was always that only trichomes on one half of the leaf divided along the midrib could be count through photography. The leaves number were named one by one according to the growth order, that is, the first true leaf of the plant was first leaf, the second true leaf is second leaf, and so on. Phenotypes of the parents of the mapping population, i.e., YJ27 and 03B0307, are shown in Fig. 1a, b, c, respectively. The adaxial surface of leaves on 03B0307 was glabrous during the whole life, while YJ27 had trichome in its vegetative growth period. YJ27 was an early flowering variety without vernalization in B. juncea, and it would start to enter the reproductive growth period when the third true leaf grow out, and flower bud differentiation started when the fifth leaf grow out. After the fifth true leaf grow out, the plant entered the reproductive growth stage, and the leaves produced thereafter had no trichome on their adaxial surface. Trichome number on leaves of YJ27 increased first and then decreased with the number of leaves; at the second leaf, the number was the most (Fig. 1b). Based on these realities, the number of trichomes on half of the second leaf was used to represent this individual in this study. Trichome number was counted in the parental lines, 62 F 1 individuals, 1304 F 2 individuals, and 128 natural varieties. The average number of trichome in YJ27 was 116.9, in F 1 was 86.76, and in 03B0307 was 0, indicating that the trait with trichomes should be a dominant trait (Fig. 1c). The segregation of trichome trait in the F 2 population tend to be complex: 196 individuals were glabrous, and others exhibited a continuous frequency distribution with a range of 1-177 (Fig. 1d). Among the 128 natural varieties in this study, 56 varieties were glabrous, and 72 varieties had trichome range of 1 to 286 (Fig. 5b).

BSA-seq analysis anchors a candidate region on B02 chromosome for trichomes
Based on the result of F 2 population, 50 plants with glabrous leaves and 50 plants with the most intensive trichome leaves were selected and pooled as glabrous pools and trichome pools, respectively. The whole-genome re-sequencing data from YJ27, 03B0307, glabrous and trichome pools were aligned with the B. juncea genome 2.0. A total of 53.0, 46.7, 13.4 and 9.8 billion of raw data were generated from YJ27, 03B0307, glabrous and trichome pools, respectively. The GC content of raw reads ranged from 38.16 to 38.44%. And the summary of the sequencing data was given in Table S3. The data volume, sequencing quality and GC distribution of the pools indicated that the database was successfully established, while the coverage of the reference genome indicated that the alignment results were normal, and the results could be used for variant detection and correlation analysis. After trimming and filtering, a total of 2,868,669 polymorphic marker loci were identified, and 2,017,379 SNPs were used for QTL identification through BSA-seq analysis. As shown in Table S4 and Fig. 2, the BSA-seq analysis detected one putative region between 56.25 and  (Fig. 2d), in which the number of SNPs was 5207. And the putative regions of QTL-seq were between 56.47 Mb and 57.62 Mb, and between 57.73 Mb and 58.18 Mb on Chromosome B02, in which the numbers of SNPs were 2627 and 825 (Fig. 2c).

Identification of candidate gene via fine mapping and genotyping approaches
For fine mapping of trichome trait, KASP markers between 56.25 Mb and 58.40 Mb on Chromosome B02 were developed based on the SNPs in the candidate interval from BSAseq. The results showed that 13 KASP markers exhibited polymorphism. Twenty-two recombinant plants of 212 F 2 individuals were analyzed for fine mapping of trichome trait. As shown in Fig. 3a, F 2 -211, F 2 -234 and F 2 -244 had the same glabrous leaves as 03B0307, and F 2 -270 had the trichome leaves like YJ27. The right region of M3 had genotypes similar to 03B0307 on F 2 -211. On F 2 -234, the region between M1 and M3 and the right region of M8 had the similar genotypes to 03B0307. And on F 2 -270, the region between M1 and M3 had the similar genotypes as 03B0307. These results indicated that the potential genes would not at the left region of M8. F 2 -244 had the similar genotypes with 03B0307 at the left region of M10, suggesting the potential genes might be at the left region of M10. These results narrowed down the candidate region between M8 and M10 that was 57,098,172 ~ 57,546,265 on Chromosome B02. Moreover, the genotyping results of another 18 recombinant plants supported the results to narrow down the region to between M8 and M10. Therefore, the region was delimited to an interval of approximately 448 kb flanked by markers M8 and M10 in which there were 84 genes. Gene information of this region is listed in Table S5.
Based on the re-sequencing data of parents, a gene named BjuVB02G54610 in which there was a 3 kb SV was found in the 448 kb candidate region (Fig. 3b). This gene was annotated as MYB domain protein 0 in Brassica database and was homologous with AT3G27920, which encodes GL1, a MYB-like protein required for induction of trichome. For this reason, we believed that this gene was the most likely gene for trichome trait in B. juncea. In order to prove, markers were designed for genotyping and cloning based on the 3 kb SV of this gene. The genotyping results of 03B0307, YJ27 and F 1 using the full-length marker of the SV (Fig. 3c) and unique marker on the SV (Fig. 3d) confirmed that this SV existed. This SV was homozygous insertion in 03B0307 and represented as II, was homozygous deletion in YJ27, and represented as DD. In F 1 , it was therefore denoted by ID, which means it was heterozygous. Sequence alignment revealed that this SV located from 310 bp of the gene in the second exon to 4101 bp in the second intron, which resulted in the deletion of at least two amino acids in the second exon (Fig. 3e). Specific sequencing results near the 3 kb SV of 03B0307 and YJ27 are provided in Figure. S1.
Genotyping results of 292 F 2 individuals with and without trichome confirmed the association with the SV and the trait. As shown in Fig. 4, all the glabrous individuals were genotype II, while individuals with trichome were ID or DD, except for three individuals with few trichomes. And with the increase in the number of trichome, the proportion of DD gradually increased. Regression analysis showed that this SV determined 81% of the trait variation, and there were significant differences in the trichome number among the three genotypes (p < 0.0001).
In order to investigate whether the variability of trichome in natural population can be explained on the basis of this gene, the re-sequencing results of eight varieties with and without trichome were analyzed and showed similar segregation to 03B0307 and YJ27 (Fig. 5a). The genotyping results of natural population also showed a high correlation between this SV and the trait. Most of the varieties with II genotype were glabrous, most of the varieties with ID or DD genotype have trichome, and the R 2 was 48.8% (Fig. 5b). The result of significance analysis showed that there was a significant difference in the trichome number between II and DD (p < 0.0001).

Discussion
Trichomes on vegetable-or oil-mustards are important to the resistance of biotic and abiotic stresses (Mathur et al. 2011(Mathur et al. , 2012, and are widely used in variety identification (Kumari et al. 2020). Owing to the natural environment and artificial selection, the trichome trait appears different in varieties of mustards as shown in this research (Fig. 1), while their trichomes are all single-and non-glandular cells like those in A. thaliana, except that they have no branches. Our observations showed that different from A. thaliana, most of the trichomes in mustards occurred during their vegetative growth phase, and disappeared after they turn to reproductive growth. It is reported that trichomes develop in a regular spacing pattern in young leaves of A. thaliana (Pesch and Hulskamp 2004), and they have the same origin as leaf epidermal cells, but differentiate into different morphologies Fig. 2 BSA results of trichome trait formation. (a) Graphs of the SNP index of the trichome pool, (b) graphs of the SNP index of the glabrous pool, and (c) the ΔSNP index values used for the association analysis. The x-axis and y-axis represent the 18 mustard chromosomes in two sub-genomes and the SNP index, respectively. The dotted curve line indicates the fitted SNP index or ΔSNP index. The dotted horizontal line indicates the association threshold of 99% confidence interval. (d) Major quantitative trait loci for trichome trait formation detected by G prime method ◂ which require a balance of cell proliferation, differentiation, intercellular communication, and morphogenesis control (Szymanski et al. 2000). However, studies about trichomes regarding the genes controlling trichome formation and development in Brassicaceae are less reported compared with Arabidopsis (Gruber et al. 2006;Nayidu et al. 2014). In Brassica napus, BnaA.GL1.a, BnaC.SWEET4.a, BnaC. WAT1.a and BnaC.WAT1.b have different SNP patterns and different transcript levels in hairy-and glabrous-varieties, and these genes imply their roles for sugar and auxin signaling in leaf trichome development in B. napus (Xuan et al. 2020). The QTL analyses of trichomes on Brassica rapa (B. rapa) reveal that multiple genes determine the number of trichomes, and major QTLs and genes are located on LG A4, A6, A7 and A9 (Feng et al. 2009;Kawakatsu et al. 2017;Kubo et al. 2010;Song et al. 1995). Another research shows that Bra025087 encoding a cyclin family protein, Bra035000 encoding an ATP-binding protein/kinase/protein kinase/ protein serine/threonine kinase and Bra033370 encoding a WD-40 repeat family protein influence the formation of trichomes by participating in trichome morphogenesis . The analysis of genetic locus indicates that BoTRY can be the candidate gene of leaf trichome formation in Brassica oleracea (Mei et al. 2017). In this study, we used BSA-seq and fine mapping tools and thus found a 448 kb candidate region on chromosome B02 associated with trichome formation in B. juncea (Fig. 2).
Interestingly, in the candidate region, a 3 kb SV was found to be associated with trichome formation (Fig. 3). SV is an important hidden layer for genetic variations and is exploited only recently with high-quality genome sequencing technologies. Extensive SVs have been discovered to be responsible for a major source of genetic diversity. These variants may have function consequences by disrupting or modifying regulatory elements, altering gene structure or altering copy number (Marroni et al. 2014). Some studies have shown that CNVs contribute to differences in metal tolerance (Maron et al. 2013), flowering (Rosloski et al. 2010), disease resistance (Hurwitz et al. 2010;McHale et al. 2012;Muñoz-Amatriaín et al. 2013) and glyphosate resistance (Gaines et al. 2010). Depending on where it exists, PAVs may affect gene expression, such as changes in grape color (Kobayashi et al. 2004); changes of anthocyanin content in blood orange (Butelli et al. 2012), or disrupt genes, leading to a non-functional protein, to a lack of translation of the defective mRNA to polypeptide, or to an alteration of expression of the defective mRNA, for example, the production of tallow flesh in peach (Falchi et al. 2013). With the development of sequencing technology, it has become feasible to confirm the existence of various SVs based on NGS data (Alkan et al. 2011) as employed in our current research, although it has the limitation of short read lengths, making it challenging to identify variants in repetitive regions (Treangen and Salzberg 2011).
In our research, the 3 kb SV, which have strong connection with trichome trait formation, was located on Bju-VB02G54610, which is a homolog gene of AtGL1 in A. thaliana. GL1 plays a central role in the production of trichome, and its loss-of-function mutants exhibit an almost complete loss of trichome initiation (Marks 1997). As a member of the MYB family, GL1 contains two repeats of the MYB DNA binding domain at the N terminus and an The red lines represent II genotype, the yellow lines represent ID genotype, and green lines represent DD genotype, each line represents one variety. The height of the line represents the trichome number of this variety (colour figure online) acidic C terminus with no similarity to other MYB proteins (Larkin et al. 1994(Larkin et al. , 1993Oppenheimer et al. 1991). It forms a complex with TTG1 and GL3 or EGL3 to mediate the trichome formation and is a positive regulator in regulatory model (Zhao et al. 2008). Like other patterning proteins except TTG1, it is initially expressed in all epidermal cells, with increased levels in initial trichomes, but at higher levels of leaf development, its expression is generally restricted to trichomes (Larkin et al. 1993;Oppenheimer et al. 1991). And analysis of GL1 shows that it acts locally rather than over long distances (Hülskamp et al. 1994). Inhibitors such as TRY compete with GL1 for the binding region of the complex to inhibit trichome production (Pesch and Hulskamp 2004). According to the activator-inhibitor model, activators such as GL1 can activate their own inhibitors and upregulate their own expression through a positive feedback loop, allowing the system to amplify random fluctuations to break the initial balance between cells, then inhibitors move to adjacent cells to inhibit them (Meinhardt and Gierer 1974;Turing 1952). We found that although the SV was associated with the presence or absence of trichome in most F 2 individuals, not all the F 2 individuals with genotype II show glabrous phenotype. We also found this phenomenon when we performed the genotyping in natural population indicating the trait of presence or absence of trichome in B. juncea is not regulated by a single gene, but the GL1 plays a dominant role. We also found that the trichome number distribution of those F 2 individuals presenting a normal distribution, suggesting the dense of trichomes is a quantitative trait. Therefore, the presence/absence of trichomes and the density of trichomes are two different traits and regulated by different genes in mustard. This phenomenon was also found in other Brassicas. In Arabidopsis, GL1 and TTG1 regulate the initiation and development of trichomes, while other genes, such as EGL3, ETC1, ETC2 and ETC3, act in regulating the density and quantity of trichomes (Khan et al. 2021). Since trichomes are both beneficial to pest resistance and adverse to eating attribute, it is crucial to understand the laws of its production and development for achieve desirable breeding targets. As B. juncea crops are widely grown as vegetable, condiment and oil source, researches of trichomes will contribute to the improvement of its stress resistance, flavor and eating quality.
In conclusion, our research shows that a gene named BjuVB02G54610 encoding a MYB transcription factor is most likely to affect the trichome trait in alloploidy Brassica juncea crops, and a structural variation of 3 kb in this gene is highly associated with the trait formation. And our successful fine mapping of the gene associated with trichome formation provides valuable gene resource for further directional breeding via gene editing approach.