BrrA02.LMI1 Encodes a Homeobox Protein That Affects Leaf Margin Development in Brassica rapa

Leaf margin morphology is an important quality trait affecting the commodity and environmental adaptability of crops. Brassica rapa is an ideal research material for exploring the molecular mechanisms underlying leaf lobe development. Here, we identified BrrA02.LMI1 to be a promising gene underlying the QTL qBrrLLA02 controlling leaf lobe formation in B. rapa, which was detected in our previous study. Sequence comparison analysis showed that the promoter divergences were the most obvious variations of BrrA02.LMI1 between parental lines. The higher expression level and promoter activity of BrrA02.LMI1 in the lobe-leafed parent indicated that promoter variations of BrrA02.LMI1 were responsible for elevating expression and ultimately causing different allele effects. Histochemical GUS staining indicated that BrrA02.LMI1 is mainly expressed at the leaf margin, with the highest expression at the tip of each lobe. Subcellular localization results showed that BrrA02.LMI1 was in the nucleus. The ectopic expression of BrrA02.LMI1 in A. thaliana resulted in a deep leaf lobe in the wild-type plants, and lobed leaf formation was disturbed in BrrA02.LMI11-downregulated plants. Our findings revealed that BrrA02.LMI1 plays a vital role in regulating the formation of lobed leaves, providing a theoretical basis for the selection and breeding of leaf-shape-diverse varieties of B. rapa.


Introduction
Leaves are essential organs that produce energy through photosynthesis in higher plants, playing an essential role in nutrient accumulation, gas exchange, light absorption, and water transport.Leaves are also the main edible organs of leafy vegetables and provide rich nutrients such as vitamins, soluble fiber, and mineral nutrition for humans.Leaves evolve different morphologies in the process of adapting to complex environments, which leads to functional variation in leaves [1].Leaf morphology is mainly determined by the outline of the leaf margin, with the leaf margin categorized as smooth, serrated, or lobed [2][3][4].Plants with deep leaf lobes have shown better adaptability to environmental stresses, such as drought, heat, and diseases, due to optimal canopy architecture [4].
Leaves originate from a mass of pluripotent cells at the flanks of the shoot apical meristem (SAM) [5].The process of leaf morphogenesis involves three stages: leaf initiation (I), primary morphogenesis (PM), and expansion and secondary morphogenesis (SM) [6].The leaf primordium forms via the lateral differentiation of cells from the peripheral region of the SAM.During the primary morphogenesis process, the basic form of the leaf, including leaf symmetry and the major subregions, is established [7].The expansion and secondary morphogenesis stage-during which the leaf surface area and volume increase a thousandfold through cell division and differentiation, accompanied by the appearance of typical characteristics of mature leaves such as stomata and epidermal hairs-is much longer than the primary morphogenesis stage.
Leaf development and morphogenesis require extremely complex physiological and biochemical processes and involve crosstalk among transcription factors, miRNAs, and hormones [8][9][10][11].The Class I KNOTTED1-LIKE HOMEOBOX (KNOX) family is an important transcription factor family that maintains meristem activity.In Arabidopsis, four members of the KNOX I family have been detected, SHOOTMERISTEMLESS (STM), BRE-VIPEDICELLUS (BP), KNOTTED1-LIKE2 (KNAT2), and KNOTTED1-LIKE6 (KNAT6), and the overexpression of either member leads to the appearance of lobed leaves [12].The DRM1/ARP (dormancy-associated protein 1/auxin-repressed protein) family is highly conserved in different species and antagonizes KNOXI family gene expression during leaf development.ASYMMETRIC LEAVES1 (AS1) and AS2 form a repressor complex that directly binds to specific promoter regions of the KNOXI family and represses KNOXI gene expression [13,14].CUP-SHAPED COTYLEDON (CUC) genes are expressed at the boundary of the leaves [15].The functional loss of CUC genes has been found to lead to less dissected leaves in many species, including Arabidopsis, tomato, pea, potato, Cardamine, and Aquilegia [16,17].CUC2 was found to be targeted and negatively regulated by miR164A, and the functional loss of miR164A has been shown to lead to increased leaf serrations in Arabidopsis.In addition, negative feedback loops involving PIN1 and CUC2 result in regular leaf serration in Arabidopsis [18].The overexpression of LEAFY (LFY) results in a deepened leaf lobe and complex leaf morphology in Cardamine hirsuta [19].The HD-Zip I transcription factor, LATE MERISTEM IDENTITY1 (LMI1), was first identified as a meristem identity regulator in Arabidopsis.LMI1 and its homologs in many other species have been identified as influencing leaf morphogenesis [20,21].The REDUCED COMPLEXITY (RCO) gene was found to be responsible for leaflet formation in C. hirsuta [20].Class II TEOSINTE BRANCHED1, CYCLOIDEA, and PRO-LIFERATING CELL FACTORS (TCP) genes have been reported to regulate leaf morphology by preventing cell proliferation [22,23].In addition to these gene families, some specific genes, including SIMPLE LEAF3 (SIL3) [24], BLADE-ON-PETIOLE1 (BOP1) and BOP2 [25,26], SAWTOOTH 1 (SAW1) and SAW2 [27], Squamosa-Promoter binding-like (SPL) [28,29], and JAGGED (JAG) [30,31], have been reported to be involved in leaf morphology.
Brassica species are rich in leaf margin morphology diversity and provide abundant genetic resources and ideal materials for exploring the molecular mechanism of leaf development.In B. napus, two tandemly duplicated LMI1-like genes, BnA10.LMI1 and BnA10.RCO, were reported to positively regulate lobed leaf formation [32,33].In Brassica oleracea, BoLMI1a, a homolog of LMI1, was also predicted as the candidate gene regulating leaf lobe formation.In addition, Feng et al. [34,35] demonstrated that BoALG10, which encodes a glycosyltransferase, plays a vital role in leaf lobe development.In Brassica rapa, the physical interval at the distal end of chromosome A10, which is composed of a rich quantitative trait locus (QTL) of the lobed leaf trait, is a research hotspot.However, the molecular mechanisms underlying leaf lobe development are poorly understood.In our previous studies, two regions related to leaf lobe formation, qBrrLLA10 and qBrrLLA02, were identified with bulked segregant analysis sequencing (BSA-seq) using a segregation population originating from the deeply lobed-leaf line MM and the serrated leaf line BY.We have already verified that BrrRCO was the causal gene underlying the qBrrLLA10 locus, which positively regulates lobed-leaf formation [36].However, the regulatory mechanism of the qBrrLLA02 locus in leaf lobe formation remains unknown.Here, we provide strong evidence that BrrA02.LMI1 is the candidate gene underlying the qBrrLLA02 locus, and BrrA02.LMI1 positively regulates leaf lobe formation independent of BrrRCO expression, providing novel insights into the regulatory network of lobed leaf formation in Brassica crops.

Candidate Gene Identification Underlying the qBrrLLA02 Locus
We have shown that two major QTLs, qBrrLLA02 and qBrrLLA10, are responsible for lobed leaf formation in B. rapa based on BSA-seq using an F 2 segregation population originating from the cross between a deeply lobed-leaf turnip inbred line MM and a serrated leaf Chinese cabbage inbred line BY.Our previous study indicated that BrrRCO was the causal gene underlying the qBrrLLA10 locus, and the molecular mechanism of BrrRCO in regulating the formation of lobed leaves was clearly elucidated.Here, we further tested whether the BraA02g001070.3C gene is the candidate gene underlying the qBrrLLA02 locus regulating leaf lobe development in B. rapa.
Among these annotated genes in the qBrrLLA02 interval on chromosome A02, the gene BraA02g001070.3C,which was functionally annotated as associated with leaf morphogenesis, caught our attention.We constructed a phylogenetic tree using BraA02g001070.3C and its putative homeologs in A. thaliana and other species (Figure 1A), which indicated that BraA02g001070.3C is a homolog of A. thaliana LMI1, which is required for leaf serration and bract formation [37]; therefore, we renamed BraA02g001070.3C as BrrA02.LMI1.The conserved domain analysis revealed that BrrA02.LMI1 contains a typical homeobox domain and a leucine zipper domain (Figure 1B).
Int. J. Mol.Sci.2023, 24, x FOR PEER REVIEW 3 of 12 expression, providing novel insights into the regulatory network of lobed leaf formation in Brassica crops.

Candidate Gene Identification Underlying the qBrrLLA02 Locus
We have shown that two major QTLs, qBrrLLA02 and qBrrLLA10, are responsible for lobed leaf formation in B. rapa based on BSA-seq using an F2 segregation population originating from the cross between a deeply lobed-leaf turnip inbred line MM and a serrated leaf Chinese cabbage inbred line BY.Our previous study indicated that BrrRCO was the causal gene underlying the qBrrLLA10 locus, and the molecular mechanism of BrrRCO in regulating the formation of lobed leaves was clearly elucidated.Here, we further tested whether the BraA02g001070.3C gene is the candidate gene underlying the qBrrLLA02 locus regulating leaf lobe development in B. rapa.
Among these annotated genes in the qBrrLLA02 interval on chromosome A02, the gene BraA02g001070.3C,which was functionally annotated as associated with leaf morphogenesis, caught our attention.We constructed a phylogenetic tree using BraA02g001070.3C and its putative homeologs in A. thaliana and other species (Figure 1A), which indicated that BraA02g001070.3C is a homolog of A. thaliana LMI1, which is required for leaf serration and bract formation [37]; therefore, we renamed BraA02g001070.3C as BrrA02.LMI1.The conserved domain analysis revealed that BrrA02.LMI1 contains a typical homeobox domain and a leucine zipper domain (Figure 1B).

Genotyping Analysis of BrrA02.LMI1 in B. rapa
To further identify whether BrrA02.LMI1 was the candidate gene underlying the qBrrLLA02 locus, we conducted a genomic sequence alignment of BrrA02.LMI1 between the two parental lines.We isolated and compared the genomic DNA segments, including the approximately 3 kb promoter sequence, gene body, and 2 kb 3′ flanking region of BrrA02.LMI1 between MM and BY.BrrA02.LMI1 encodes a protein comprising 227 amino acids.A 12 bp insertion and three nonsynonymous SNPs were detected in the coding sequence (Figure 2A, Figure S1A), which led to a four-amino-acid insertion and four-aminoacid substitution in the lobed-leaf MM compared with the corresponding positions of the

Genotyping Analysis of BrrA02.LMI1 in B. rapa
To further identify whether BrrA02.LMI1 was the candidate gene underlying the qBr-rLLA02 locus, we conducted a genomic sequence alignment of BrrA02.LMI1 between the two parental lines.We isolated and compared the genomic DNA segments, including the approximately 3 kb promoter sequence, gene body, and 2 kb 3 flanking region of BrrA02.LMI1 between MM and BY.BrrA02.LMI1 encodes a protein comprising 227 amino acids.A 12 bp insertion and three nonsynonymous SNPs were detected in the coding sequence (Figures 2A and S1A), which led to a four-amino-acid insertion and four-aminoacid substitution in the lobed-leaf MM compared with the corresponding positions of the serrated leaf BY (Figure S1B).In the promoter sequence, abundant polymorphisms, including 10 InDels and 28 SNPs, were identified between MM and BY (Figures 2A and S1C).We then investigated the expression patterns of BrrA02.LMI1 in parental lines.The fifth mature leaves of both parents were divided into two parts: leaf margin and leaf base, which were used for qRT-PCR assays.The results indicated that the BrrA02.LMI1 expression in different leaf segments of MM was significantly higher than that in BY, with the highest expression at the leaf margin of MM (Figure 2B).Therefore, we suspected that variations in the promoter sequence play a vital role in determining the activity of BrrA02.LMI1.To address this question, the promoter activity of BrrA02.LMI1 in MM and BY was detected using transient transcription activity assays.The luciferase assay showed that the promoter activity of the BrrA02.LMI1 of deeply lobed-leaf MM was significantly higher than that of the serrated leaf BY (Figure 2C).In light of these results, we speculated that the promoter variations of BrrA02.LMI1 may be responsible for increased expression levels and promoter activity, which induced the formation of leaf lobes in B. rapa.serrated leaf BY (Figure S1B).In the promoter sequence, abundant polymorphisms, including 10 InDels and 28 SNPs, were identified between MM and BY (Figure 2A, Figure S1C).We then investigated the expression patterns of BrrA02.LMI1 in parental lines.The fifth mature leaves of both parents were divided into two parts: leaf margin and leaf base, which were used for qRT-PCR assays.The results indicated that the BrrA02.LMI1 expression in different leaf segments of MM was significantly higher than that in BY, with the highest expression at the leaf margin of MM (Figure 2B).Therefore, we suspected that variations in the promoter sequence play a vital role in determining the activity of BrrA02.LMI1.To address this question, the promoter activity of BrrA02.LMI1 in MM and BY was detected using transient transcription activity assays.The luciferase assay showed that the promoter activity of the BrrA02.LMI1 of deeply lobed-leaf MM was significantly higher than that of the serrated leaf BY (Figure 2C).In light of these results, we speculated that the promoter variations of BrrA02.LMI1 may be responsible for increased expression levels and promoter activity, which induced the formation of leaf lobes in B. rapa.

Tissue-Specific Expression Patterns of BrrA02.LMI1 in B. rapa
To obtain the detailed tissue-specific expression patterns of BrrA02.LMI1, we constructed the vector proBrrA02.LMI1::GUS, fusing the promoter sequence of BrrA02.LMI1 MM to the GUS reporter gene and transformed it into wild-type Arabidopsis.The BrrA02.LMI1::GUS expression was observed in homozygous T3-positive transgenic lines.As shown in Figure 3A, BrrA02.LMI1::GUS expression was observed in all tissues during the two-leaf stage.In the slightly older seedling, BrrA02.LMI1::GUS expression was more visible in the proximal leaf margins, with the deepest stain in the serrations (Figure 3B).At the bolting stage, BrrA02.LMI1 was expressed in the margins and serrations of bracts (Figure 3C).In addition, we observed GUS signals in the margins of sepals, petals, and carpels in maturing flowers (Figure 3D).These findings indicated that BrrA02.LMI1 may be involved in the development of leaf margins and floral organs.

Tissue-Specific Expression Patterns of BrrA02.LMI1 in B. rapa
To obtain the detailed tissue-specific expression patterns of BrrA02.LMI1, we constructed the vector proBrrA02.LMI1::GUS, fusing the promoter sequence of BrrA02.LMI1 MM to the GUS reporter gene and transformed it into wild-type Arabidopsis.The BrrA02.LMI1::GUS expression was observed in homozygous T3-positive transgenic lines.As shown in Figure 3A, BrrA02.LMI1::GUS expression was observed in all tissues during the two-leaf stage.In the slightly older seedling, BrrA02.LMI1::GUS expression was more visible in the proximal leaf margins, with the deepest stain in the serrations (Figure 3B).At the bolting stage, BrrA02.LMI1 was expressed in the margins and serrations of bracts (Figure 3C).In addition, we observed GUS signals in the margins of sepals, petals, and carpels in maturing flowers (Figure 3D).These findings indicated that BrrA02.LMI1 may be involved in the development of leaf margins and floral organs.

Subcellular Localization of BrrA02.LMI1
We constructed a p35S:: BrrA02.LMI1-GFP vector containing the BrrA02.LMI1 coding sequence fused in frame with green fluorescent protein (GFP) under the control of the CaMV35S promoter to determine subcellular localization.Then, the recombinant and control plasmids were transferred into B. rapa protoplasts.As shown in Figure 4, the GFP signal of 35S:: BrrA02.LMI1-GFP was only visible in the nucleus, which is consistent with the function of BrrA02.LMI1 as transcription factors.

Function of BrrA02.LMI1 in Regulating Lobed Leaf Formation
To test the function of BrrA02.LMI1 in lobed leaf formation, we transferred the constructs 35S::BrrA02.LMI1 MM

Subcellular Localization of BrrA02.LMI1
We constructed a p35S:: BrrA02.LMI1-GFP vector containing the BrrA02.LMI1 coding sequence fused in frame with green fluorescent protein (GFP) under the control of the CaMV35S promoter to determine subcellular localization.Then, the recombinant and control plasmids were transferred into B. rapa protoplasts.As shown in Figure 4, the GFP signal of 35S:: BrrA02.LMI1-GFP was only visible in the nucleus, which is consistent with the function of BrrA02.LMI1 as transcription factors.

Function of BrrA02.LMI1 in Regulating Lobed Leaf Formation
To test the function of BrrA02.LMI1 in lobed leaf formation, we transferred the constructs 35S::BrrA02.LMI1 MM

Function of BrrA02.LMI1 in Regulating Lobed Leaf Formation
To test the function of BrrA02.LMI1 in lobed leaf formation, we transferred the constructs 35S::BrrA02.LMI1 MM and 35S::BrrA02.LMI1 BY overexpressing the coding sequence of the BrrA02.LMI1 of lobed-leaf MM and serrated leaf BY into Arabidopsis Col-0 plants under the control of the CaMV35S promoter.Approximately 22 and 15 independent 35S::BrrA02.LMI1 MM -and 35S::BrrA02.LMI1 BY -positive transgenic lines were obtained after hygromycin B resistance selection.The 35S::BrrA02.LMI1 MM (Figure 5A,B) and BrrA02.LMI1 BY (Figure S2A) lines both displayed a deep leaf lobe and an increased dissection index (Figures 5C and S2B).The expression of BrrA02.LMI1 in the overexpression lines was also significantly higher than that in the wild type (Figure 5D).These results demonstrated that the coding sequences of BrrA02.LMI1 are most likely functionally equivalent in MM and BY.
equivalent in MM and BY.
We further confirmed BrrA02.LMI1's function by using a turnip yellow mosaic virus (TYMV)-induced gene silencing (VIGS) method.The associated vector has been extensively used in Arabidopsis and Brassica species [38][39][40][41][42].To construct a pTY-BrrA02.LMI1 vector, the 80 -nt palindromic sequence specific to BrrA02.LMI1 was synthesized and ligated to the pTY vector.The recombinant pTY-BrrA02.LMI1 plasmids were used to inoculate seedlings of the lobed-leaf parent MM using a previously published method [36].In MM plants treated with pTY-BrrA02.LMI1, the expression of BrrA02.LMI1 was significantly downregulated in seedlings treated with pTY-BrrA02.LMI1 (Figure 5G, H), and no obvious expression level alteration of the BrrA02.LMI1 homologs, BrLMI1 and BrrRCO, was detected (Figure S3A, B).In addition, leaf lobe formation in BrrA02.LMI1-silenced plants was disturbed, and the leaves became rounder compared with those of the controls (Figure 5E, F).These findings showed that BrrA02.LMI1 plays a positive regulatory role in leaf lobe development in B. rapa.C,D) Arabidopsis overexpression lines had a higher expression of BrrA02.LMI1 and dissection index than that of the wild type.The ACTIN2 (At3g18780) gene in Arabidopsis was used as reference genes to normalize the expression levels of BrrA02.LMI1.Three biological replicates were used for each PCR, and for each replicate, three leaves were sampled.Data represent the mean ± SD (n = 3).Double asterisk denotes a statistically significant difference to the Col-0 wild type in the Student's t-test (p < 0.001).(E,F) BrrA02.LMI1-silenced plants showed a strongly reduced lobed-leaf phenotype.Bars = 1 cm; (G,H) BrrA02.LMI1-silenced plants had a downregulated expression of BrrA02.LMI1 and reduced dissection index.Three biological replicates were used for each PCR, and for each replicate, three leaves were sampled.The B. rapa GAPDH gene was used as a reference gene to normalize the expression levels of BrrA02.LMI1.Data represent the mean ± SD (n = 3).Double asterisk denotes a statistically significant difference to the wild type in the Student's t-test (p < 0.001).
We further confirmed BrrA02.LMI1's function by using a turnip yellow mosaic virus (TYMV)-induced gene silencing (VIGS) method.The associated vector has been extensively used in Arabidopsis and Brassica species [38][39][40][41][42].To construct a pTY-BrrA02.LMI1 vector, the 80 -nt palindromic sequence specific to BrrA02.LMI1 was synthesized and ligated to the pTY vector.The recombinant pTY-BrrA02.LMI1 plasmids were used to inoculate seedlings of the lobed-leaf parent MM using a previously published method [36].In MM plants treated with pTY-BrrA02.LMI1, the expression of BrrA02.LMI1 was significantly downregulated in seedlings treated with pTY-BrrA02.LMI1 (Figure 5G,H), and no obvious expression level alteration of the BrrA02.LMI1 homologs, BrLMI1 and BrrRCO, was detected (Figure S3A,B).In addition, leaf lobe formation in BrrA02.LMI1-silenced plants was disturbed, and the leaves became rounder compared with those of the controls (Figure 5E,F).These findings showed that BrrA02.LMI1 plays a positive regulatory role in leaf lobe development in B. rapa.

Discussion
Leaves are important vegetative organs in plants, and leaf margin morphology is not only a reflection of plant diversity but also an adaptation of plants to the environment.Leaf lobe traits have many advantages in practice.In our previous study, in order to identify the candidate genes controlling lobed leaf traits in Brassica rapa, a genetic population was constructed using a turnip inbred line MM with lobed leaves and a Chinese cabbage inbred line BY with serrated leaves [36].Two QTLs, qBrrLLA10 and qBrrLLA02, were detected using this F 2 population based on BSA-seq, and the molecular mechanism underlying the qBrrLLA10 locus was clearly elucidated.However, little is known about the candidate genes and regulatory mechanism underlying qBrrLLA02.In the present study, an LMI1 homolog from the European turnip line MM was identified as the causal gene underlying the qBrrLLA02 locus and was named BrrA02.LMI1.The sequence variations between the two parents, the spatiotemporal expression patterns, and the function of BrrA02.LMI1 were systematically analyzed, thereby helping to elucidate the genetic and molecular mechanisms underlying the involvement of the BrrA02.LMI1 gene in leaf lobe formation in B. rapa.
Many sequence variations were detected in the BrrA02.LMI1 promoter between the two parental lines.In addition, the expression level (Figure 2B) and promoter activity (Figure 2B) of BrrA02.LMI1 in lobed-leaf MM was substantially higher than that in serrated leaf BY.Therefore, we speculated that the promoter variation of BrrA02.LMI1 was responsible for different leaf margins in the two parents.The transgenic experimental results showed that the overexpression of the BrrA02.LMI1 allele from MM and BY in Arabidopsis both led to increased leaf complexity (Figures 5A-D and S2), which suggested that the function of the BrrA02.LMI1 coding sequence is equivalent in MM and BY.Collectively, these findings indicated that the expression level of BrrA02.LMI1 is positively related to the formation of leaf lobes in B. rapa, and that cis-regulatory divergences led to different leaf margin phenotypes between parental lines, which is consistent with previous reports in other species.In upland cotton, tandem repeat sequences in the GhLMI1-D1b promoter region enhance gene expression, leading to deep leaf lobes [21].In B. napus, promoter variations of BnA10.LMI1 determine the formation of lobed leaves in rapeseed [32,33].In ornamental kale, the higher expression level of BoLMI1 in the lobed-leaf parent was found to be due to numerous variations in the promoter region [26].In zucchini, a few variations in the promoter led to stronger CpDll promoter activity in the deeply lobed-leaf parent than in the entire-leaf parent [43].However, how promoter sequence variations affect gene expression and thus regulate the formation of leaf lobes still requires further investigation.To determine the mechanism of BrrA02.LMI1 more completely, the upstream factors regulating the expression of the BrrA02.LMI1 gene need to be further explored.In addition, the subcellular localization results showed that BrrA02.LMI1 is in the nucleus, which is consistent with transcription factor patterns and suggests that BrrA02.LMI1 may regulate leaf morphology by regulating the expression of downstream target genes.
The formation of both compound leaves and serrated leaf margins has a dosing effect [44].In both simple-leaf and compound-leaf species, the appearance of serrations or leaflets depends on the time of their initiation.If primordial initiation occurs after leaflet unfolding and formation, then serrations will form instead of leaflets [45].In Cardamine hirsuta, ChLMI1 expression is limited to the leaf margin, while RCO is expressed at the base of the leaf, which is the location of vigorous cell division in the leaves, thus increasing the number of leaflets.The different expression positions of ChLMI1 and RCO determine their respective biological processes during leaf development [20].In our study, BrrA02.LMI1 was mainly expressed at the margin of the leaf derived from the leaf primordium, and the expression region of BrrA02.LMI1 determined its involvement in leaf lobe formation rather than in the establishment of compound leaf patterns.
As a growth repressor, LMI1 participates not only in the formation of leaf serrations but also in the development of other plant organs.In Arabidopsis, LMI1 is a meristem identity regulator and participates in bract formation and the transformation of stipules to leaves [37,46].In pea, the homologous gene of LMI1, Tl, is expressed in tendrils, and the tendrils in tl mutants can be transformed into leaves [47].In our study, some BrrA02.LMI1 ectopic overexpression in Arabidopsis lines showed increased rosette leaves before bolting and more bracts in the reproductive stage (Figure S4), which underscores the idea that LMI1 and LMI1-like genes are pleiotropic in regulating plant growth and development.In addition, other members of Class I HD-Zip genes have also been proven to participate in leaf development.HAHB4 is a subclass I HD-Zip protein found in sunflower, and its overexpression in Arabidopsis transgenic plants showed shorter stems and internodes, rounder leaves, and enhanced drought resistance [48].The overexpression of ATHB13 can lead to abnormalities in the development of cotyledons and true leaves and incomplete connections between petioles and leaf sheaths [49].ATHB16-overexpressing plants were found to display larger leaves and delayed flowering times [50].Taken together with the findings of previous studies, our results revealed that the function of HD-Zip I genes involved in the formation of lobed leaves may be conserved across dicotyledons.However, it is necessary to investigate more HD-Zip I members and mine the endogenous and exogenous upstream factors that regulate the expression of HD-Zip I genes and downstream functional genes regulated by HD-Zip I genes.These results will provide a comprehensive understanding of the molecular mechanisms of this important gene family in regulating leaf lobe formation.
In conclusion, we identified BrrA02.LMI1 as the causal gene underlying the qBrrLLA02 locus and confirmed that BrrA02.LMI1 plays a positive regulatory role in controlling leaf lobe formation in B. rapa.The results will further enrich the understanding of the molecular mechanism of leaf lobe formation, providing a theoretical basis to facilitate the breeding of leaf-shape-diverse varieties of B. rapa and other Brassica crops.

Plant Materials
A lobed-leaf line MM and a serrated leaf line BY, which were described in previous reports [36], were used in this study.Lobed-leaf MM and Arabidopsis thaliana Col-0 were used as transformation receptors for virus-induced gene silencing (VIGS) and ectopic overexpression assays.The leaf dissection index was calculated according to the formula (perimeter squared)/(4π × area) [18].

RNA Extraction and Expression Analysis of BrrA02.LMI1
RNA samples from various leaf segments of MM and BY were collected.RNA extraction, first-strand cDNA synthesis, and qRT-PCR were performed as described previously [36].The B. rapa GAPDH [51] and ACTIN2 (At3g18780) [52] genes in Arabidopsis were used as reference genes to normalize the expression level of the candidate gene, and the 2 −∆∆Ct method was used to calculate the relative expression of the candidate gene.Three biological replicates and technical replicates were used for each PCR.

Phylogenetic Analysis of BrrA02.LMI1
To construct the phylogenetic tree and perform the multiple sequence alignment analysis, the homologs of BrrA02.LMI1 were searched against the NCBI database, TAIR databases, and Brassica database using the BrrA02.LMI1 protein as a query.All deduced protein sequences were aligned using DNAMAN 9 software.MEGA software (MEGA 6.0) [53] was used to construct the phylogenetic tree with the neighbor-joining method.The bootstrap repetition value was 1000.

Subcellular Localization of BrrA02.LMI1
We amplified the open reading frame (without a termination codon) of BrrA02.LMI1 in MM and fused them in frame with green fluorescent protein (GFP) to construct the p35S::BrrA02.LMI1 MM -GFP vector under the control of the 35S promoter.The p35S::BrrA02.LMI1MM-GFP and 35S::GFP control plasmids were then transformed into protoplasts of Chinese cabbage.We observed the GFP signals at 2 days after transfection using a laser-scanning confocal microscope (Zeiss LSM 710).

GUS Staining
To explore the tissue-specific expression patterns of BrrA02.LMI1, we cloned approximately 2 kb promoter sequences of BrrA02.LMI1 and inserted them into the PacI/XbaIlinearized binary vector pMDC163.The recombinant plasmid was then transformed into Arabidopsis Col-0 plants.For GUS staining, homozygous T3 transgenic positive lines were developed.Seedlings at different stages were sampled and immersed in GUS staining buffer at the same time and then incubated at 37 • C overnight in the dark.The treated samples were successively decolorized with 75% (v/v) ethanol and observed using a Nikon microscope (SMZ1500).

Functional Analysis of BrrA02.LMI1
The full coding DNA sequence (CDS) of BrrA02.LMI1 in deeply lobed-leaf MM and serrated leaf BY were amplified, and the purified products were inserted into the KpnI/SpeI-linearized binary vector pCAMBIA1301 driven by the CaMV 35S promoter.The 35S::BrrA02.LMI1 MM and 35S::BrrA02.LMI1 BY were then transformed into wild-type Arabidopsis, as previously described [36].Transgenic T1 plants were screened with Murashige and Skoog medium supplemented with 50 mg L −1 hygromycin, and homozygous T3 lines were used for further experiments.
The function of BrrA02.LMI1 was further confirmed via knockout assays using a TYMVbased VIGS system.To construct VIGS vector pTY-BrrA02.LMI1, an 80-nt palindromic exon sequence specific to BrrA02.LMI1 was synthesized and inserted into the SnaBI-linearized pTY vector with a T4 DNA ligase.The pTY-BrrA02.LMI1 and PTY (negative control) plasmids were then introduced into Escherichia coli Stb13 cells for plasmid extraction using the OMEGA Plasmid Giga Kit.The purified pTY-BrrA02.LMI1 and PTY plasmids were diluted to 300-500 ng/µL and then used to infiltrate the second to the fourth fully expanded true leaves of MM at the four-leaf stage.For the BrrA02.LMI1 function analysis, the leaf lobe phenotype was recorded and compared between treated and control plants.The silencing efficiency of BrrA02.LMI1 in MM plants was determined using a Qrt-PCR.
All the primers used in this study were listed in Table S1.

Figure 1 .
Figure 1.(A) Phylogenetic tree of BrrA02.LMI1 and LMI1-like proteins from various species.(B) Protein sequence alignment of LMI1 proteins from different species.

Figure 1 .
Figure 1.(A) Phylogenetic tree of BrrA02.LMI1 and LMI1-like proteins from various species.(B) Protein sequence alignment of LMI1 proteins from different species.

Figure 2 .
Figure 2. Gene structure, expression analysis, and promoter activity assays of BrrA02.LMI1 in lobedleaf MM and serrated leaf BY. (A) Schematic genetic variations of BrrA02.LMI1 between the two parents.(B) Transcriptional expression of BrrA02.LMI1 in leaf base and leaf margin from MM and BY.The B. rapa GAPDH gene was used as a reference gene to normalize the expression levels of BrrA02.LMI1.Data represent the mean ± SD (n = 3), **denotes p < 0.01.(C) Promoter activity assays of BrrA02.LMI1 in lobed-leaf MM and serrated leaf BY.

Figure 2 .
Figure 2. Gene structure, expression analysis, and promoter activity assays of BrrA02.LMI1 in lobed-leaf MM and serrated leaf BY. (A) Schematic genetic variations of BrrA02.LMI1 between the two parents.(B) Transcriptional expression of BrrA02.LMI1 in leaf base and leaf margin from MM and BY.The B. rapa GAPDH gene was used as a reference gene to normalize the expression levels of BrrA02.LMI1.Data represent the mean ± SD (n = 3).(C) Promoter activity assays of BrrA02.LMI1 in lobed-leaf MM and serrated leaf BY.
and 35S::BrrA02.LMI1 BY overexpressing the coding sequence of the BrrA02.LMI1 of lobed-leaf MM and serrated leaf BY into Arabidopsis Col-0 plants under the control of the CaMV35S promoter.Approximately 22 and 15 independent 35S::BrrA02.LMI1 MM -and 35S::BrrA02.LMI1 BY -positive transgenic lines were obtained
and 35S::BrrA02.LMI1 BY overexpressing the coding sequence of the BrrA02.LMI1 of lobed-leaf MM and serrated leaf BY into Arabidopsis Col-0 plants under the control of the CaMV35S promoter.Approximately 22 and 15 independent 35S::BrrA02.LMI1 MM -and 35S::BrrA02.LMI1 BY -positive transgenic lines were obtained

Figure 5 .
Figure 5. Phenotypic analysis of BrrA02.LMI1 overexpression in Arabidopsis plants and BrrA02.LMI1-silenced MM plants.(A) and (B) BrrA02.LMI1-overexpressing Arabidopsis plants showed deeply lobed leaves.Bars =1 cm; (C) and (D) Arabidopsis overexpression lines had a higher expression of BrrA02.LMI1 and dissection index than that of the wild type.The ACTIN2 (At3g18780) gene in Arabidopsis was used as reference genes to normalize the expression levels of BrrA02.LMI1.Three biological replicates were used for each PCR, and for each replicate, three leaves were sampled.Data represent the mean ± SD (n = 3).Double asterisk denotes a statistically significant difference to the Col-0 wild type in the Student's t-test (p < 0.001).(E) and (F) BrrA02.LMI1-silenced plants showed a strongly reduced lobed-leaf phenotype.Bars =1 cm; (G) and (H) BrrA02.LMI1-silenced plants had a downregulated expression of BrrA02.LMI1 and reduced dissection index.Three biological replicates were used for each PCR, and for each replicate, three leaves were sampled.The B. rapa GAPDH gene was used as a reference gene to normalize the expression levels of BrrA02.LMI1.Data represent the mean ± SD (n = 3).Double asterisk denotes a statistically significant difference to the wild type in the Student's t-test (p < 0.001).

Figure 5 .
Figure 5. Phenotypic analysis of BrrA02.LMI1 overexpression in Arabidopsis plants and BrrA02.LMI1silenced MM plants.(A,B) BrrA02.LMI1-overexpressing Arabidopsis plants showed deeply lobed leaves.Bars = 1 cm; (C,D) Arabidopsis overexpression lines had a higher expression of BrrA02.LMI1and dissection index than that of the wild type.The ACTIN2 (At3g18780) gene in Arabidopsis was used as reference genes to normalize the expression levels of BrrA02.LMI1.Three biological replicates were used for each PCR, and for each replicate, three leaves were sampled.Data represent the mean ± SD (n = 3).Double asterisk denotes a statistically significant difference to the Col-0 wild type in the Student's t-test (p < 0.001).(E,F) BrrA02.LMI1-silenced plants showed a strongly reduced lobed-leaf phenotype.Bars = 1 cm; (G,H) BrrA02.LMI1-silenced plants had a downregulated expression of BrrA02.LMI1 and reduced dissection index.Three biological replicates were used for each PCR, and for each replicate, three leaves were sampled.The B. rapa GAPDH gene was used as a reference gene to normalize the expression levels of BrrA02.LMI1.Data represent the mean ± SD (n = 3).Double asterisk denotes a statistically significant difference to the wild type in the Student's t-test (p < 0.001).