Genome-Wide Characterization of DGATs and Their Expression Diversity Analysis in Response to Abiotic Stresses in Brassica napus

Triacylglycerol (TAG) is the most important storage lipid for oil plant seeds. Diacylglycerol acyltransferases (DGATs) are a key group of rate-limiting enzymes in the pathway of TAG biosynthesis. In plants, there are three types of DGATs, namely, DGAT1, DGAT2 and DGAT3. Brassica napus, an allotetraploid plant, is one of the most important oil plants in the world. Previous studies of Brassica napus DGATs (BnaDGATs) have mainly focused on BnaDGAT1s. In this study, four DGAT1s, four DGAT2s and two DGAT3s were identified and cloned from B. napus ZS11. The analyses of sequence identity, chromosomal location and collinearity, phylogenetic tree, exon/intron gene structures, conserved domains and motifs, and transmembrane domain (TMD) revealed that BnaDGAT1, BnaDGAT2 and BnaDGAT3 were derived from three different ancestors and shared little similarity in gene and protein structures. Overexpressing BnaDGATs showed that only four BnaDGAT1s can restore TAG synthesis in yeast H1246 and promote the accumulation of fatty acids in yeast H1246 and INVSc1, suggesting that the three BnaDGAT subfamilies had greater differentiation in function. Transcriptional analysis showed that the expression levels of BnaDGAT1s, BnaDGAT2s and BnaDGAT3s were different during plant development and under different stresses. In addition, analysis of fatty acid contents in roots, stems and leaves under abiotic stresses revealed that P starvation can promote the accumulation of fatty acids, but no obvious relationship was shown between the accumulation of fatty acids with the expression of BnaDGATs under P starvation. This study provides an extensive evaluation of BnaDGATs and a useful foundation for dissecting the functions of BnaDGATs in biochemical and physiological processes.


Introduction
Triacylglycerol (TAG), the major component of vegetable oils, consists of three fatty acids esterified to a glycerol backbone. In plants, TAG is not only mainly stored in seeds, functioning as an energy reservoir to facilitate germination and early seedling growth, but also provides precursors for membrane biosynthesis and lipid signaling, which are crucial for normal plant growth and development [1][2][3]. DGAT (diacylglycerol acyltransferase, EC2. 3.1.20), the key rate-limiting enzyme of the Kennedy pathway for TAG biosynthesis, sponding genome databases (Table S2). All sequences of the CDS, genomic DNA and protein of DGATs in B. napus and other typical plants in this study are presented in Tables S3-S6.
The amino acid sequences of BnaDGAT genes were characterized: the protein length ranged from 317 AA to 510 AA, and the molecular weight (MW) ranged from 35.64 kDa to 57.96 kDa (Table 2). Moreover, the isoelectric point (pI) values ranged from 7.75 to 8. 90, which showed that these proteins are alkaline (Table 2). Moreover, the subcellular localization signals of four BnaDGAT1s and four BnaDGAT2s were detected in the endoplasmic reticulum, as were the DGAT1s and DGAT2s of the other plants in this study (Table 2 and  Table S7). No subcellular localization signals of DGAT3s were detected in B. napus or other plants ( Table 2 and Table S7).
Identity analysis of BnaDGATs, BraDGATs and BolDGATs was performed by the Clustal W method based on nucleotide sequences and amino acid sequences (Table S8). The results showed that the three DGAT subfamilies shared very low identity with each other subfamily (8.2% to 17.2%), while the similarities among eight DGAT1s, eight DGAT2s and four DGAT3s were high (72.4% to 100%) at the amino acid sequence level. The similarity between every BnaDGAT in the A subgenome and its orthologous one in the C subgenome was higher than 96%. The identity of DGATs based on nucleotide sequences was similar to that based on amino acid sequences (Table S8).

Chromosomal Location and Collinearity Analysis
Chromosomal location (Table 2 and Figure S1A) analysis showed four BnaDGAT1s distributed on chromosomes A07, A09, C07 and C09; four BnaDGAT2s distributed on chromosomes A01, A03, C01 and C07 and that two BnaDGAT3s were distributed on chromosomes A08 and C08. In B. rapa, five BraDGATs were located on B. rapa chromosomes A07, A09, A01, A03 and A08 ( Figure S1B) and in B. oleracea, five BolDGATs were located on B. oleracea chromosomes C07, C09, C01, C07 and C08 ( Figure S1C). Five BnaDGAT loci on the B. napus A subgenome were highly parallel with five BraDGAT loci on the B. rapa A genome, and five BnaDGAT loci on the B. napus C subgenome were highly parallel with five BolDGAT loci on the B. oleracea C genome, which suggested that B. napus inherited and retained all the DGAT genes of B. rapa and B. oleracea.
The results of gene synteny analysis ( Figure 1A) showed that four BnaDGAT1s, four BnaDGAT2s and two BnaDGAT3s might be duplicated genes, suggesting that BnaDGAT genes were frequently duplicated during oilseed rape evolution. The results of comparative synteny of DGAT gene pairs among A. thaliana, B. oleracea, B. rapa, B. napus and B. nigra ( Figure 1B) showed that AtDGAT1, AtDGAT2 and AtDGAT3 had collinearity relationships with two BolDGAT1s/two BraDGAT1s/four BnaDGAT1s/two BniDGAT1s, two BolDGAT2s/two BraDGAT2s/four BnaDGAT2s/two BniDGAT2s and one BolDGAT3/one BraDGAT3/two BnaDGAT3s/one BniDGAT3, respectively.
Moreover, the synonymous mutations (Ks), nonsynonymous mutations (Ka) and Ka/Ks ratios of the orthologous DGAT gene pairs between B. napus and A. thaliana and the paralogous BnaDGAT gene pairs were evaluated (Table 3). The results showed that all of the Ka/Ks ratios of these gene pairs were lower than 1 ( Table 3), indicating that these gene pairs experienced strong purifying selective pressure. The duplication time of these gene pairs was presumed using the formula T = Ks/2R, with R (1.5 × 10 −8 ) representing neutral substitution per site per year. The results showed that the duplication times of the orthologous DGAT gene pairs between B. napus and A. thaliana ranged from 10.67 MYA to 20.11 MYA (Table 3), with an average value of 15.06 MYA. These results indicated that DGATs of B. napus diverged from A. thaliana~16 MYA, which was consistent with the recent whole-genome triplication event that occurred approximately 9-15 MYA or even 28 MYA [55]. The corresponding duplication times of the paralogous BnaDGAT gene pairs ranged from 1.37 to 12.78 MYA, with an average value of 8.36 MYA (Table 3). Two peaks of duplication times were observed in B. napus: one peak (1.37-3.57 MYA) represented the duplication time of these genes, which occurred during the divergence of the A genome and C genome of Brassica, and the other peak (10.95-12.78 MYA), representing a duplication time of~10 MYA, corresponded to the Brassica whole-genome triplication event (9-15 MYA) [55]. Therefore, the processes of B. napus speciation and Brassica whole-genome triplication likely played important roles in the divergence of the BnaDGAT duplicated genes in B. napus. genes were frequently duplicated during oilseed rape evolution. The results of compara tive synteny of DGAT gene pairs among A. thaliana, B. oleracea, B. rapa, B. napus and B nigra ( Figure 1B) showed that AtDGAT1, AtDGAT2 and AtDGAT3 had collinearity rela tionships with two BolDGAT1s/two BraDGAT1s/four BnaDGAT1s/two BniDGAT1s, tw BolDGAT2s/two BraDGAT2s/four BnaDGAT2s/two BniDGAT2s and one BolDGAT3/on BraDGAT3/two BnaDGAT3s/one BniDGAT3, respectively.

Evolutionary Relationship and Exon/Intron Gene Structure Analysis of BnaDGATs
The phylogenetic tree ( Figure 2A) showed that three well-supported clades were grouped as the DGAT1 subfamily, DGAT2 subfamily and DGAT3 subfamily. Within the DGAT1 subfamily and DGAT2 subfamily, four BnaDGAT1s and four BnaDGAT2s, respectively, were grouped more closely with the corresponding DGAT subfamilies of the other five Brassica U's triangle species (B. rapa, B. oleracea, B. nigra, B. juncea and B. carinata) and Arabidopsis than to those of the other dicot or monocot plants. Within the DGAT3 subfamily, two BnaDGAT3s were grouped more closely to those of B. oleracea, B. rapa, B. nigra, B. carinata and Arabidopsis than to the corresponding subfamilies of the other dicot or monocot plants. The DGAT1s, DGAT2s and DGAT3s of the four monocot plants were always grouped together, rather than with those of the twelve dicot plants. These results indicated that the duplication events of DGAT genes occurred after the divergence of dicot and monocot plants.
The results of the exon/intron gene structures ( Figures2B and S2B and Table S7) revealed that the gene structures of most DGAT1s, DGAT2s and DGAT3s were highly and independently conserved during evolution but confirmed distinct differences among the three DGAT subfamilies, suggesting that the three DGAT subfamilies evolved separately from three different ancient genes. All four BnaDGAT1s have 16 exons as well as DGAT1s from most other plants, except BjuB029654 (13 exons), BjuB028615 (14 exons) and BraA07g001370.3C (14 exons). Most DGAT2s are generally composed of seven to ten exons. In addition, BnaA01G0206700ZS had seven exons and BnaA03G0420700ZS had six exons in the B. napus database BnPIR (Table S7), but we cloned these genes and confirmed that they (BnaA.DGAT2.a and BnaA.DGAT2.b) had nine and eight exons, respectively. Members of DGAT3s from B. napus and the selected plants in this paper contain two exons, except BcaC07g37153 (four exons), BjuA014363 (six exons) and BjuB045147 (five exons).

The Conserved Domains and Motif Analyses
To investigate the conserved motifs and domains, the DGAT proteins from B. napus and the other selected plants were predicted by both the Conserved Domain Database (CDD) with Batch CD-Search in NCBI [57] and the MEME program (https://meme-suite. org/meme/tools/meme (accessed on 21 October 2021)) ( Additionally, the top 20 conserved motifs were identified in the DGAT proteins ( Figure S3). All the DGAT1 homologues shared one motif 2, one motif 3, one motif 7, one motif 8, one motif 9, one motif 12 and one motif 15. All DGAT1 homologues had one motif 1, except BjuB029654 and BjuB028615. All DGAT1 homologues had one motif 4, except BjuB029654 and 29912.m005373. All DGAT1 homologues had one motif 20, except BjuB029654 and Zm00001d036982. All DGAT1 homologues, except BjuA046403, had one motif 13. All DGAT1 homologues of six Brassica species, A. thaliana, and A. hypogaea shared one motif 19. All DGAT2 homologues had one motif 5, one motif 6, one motif 10, one motif 11, one motif 12 and one motif 14. All DGAT2 homologues had one motif 17, except BraA03g045590.3C, AH13G38440.1 and AH05G01580.1. All DGAT2 homologues had one motif 16, except BraA03g045590.3C and AH13G38440.1. All DGAT3 homologues had one motif 18 at their C-terminus, except BjuB045147 and BjuA014363. All DGAT3 homologues shared one or two motifs 15 at their C-terminus, except Os05g04620.1, Seita.3G054300.1 and Zm00001d024765. These results supported the hypothesis that the DGAT1, DGAT2 and DGAT3 subfamilies evolved separately during eukaryote evolution, as demonstrated by the phylogenetic tree and gene structure ( Figure 2).

Putative Transmembrane Domains of DGAT Proteins
The putative transmembrane domains (TMDs) of DGAT proteins from B. napus and other plants were predicted by TMHMM (https://services.healthtech.dtu.dk/service.php? TMHMM-2.0 (accessed on 10 April 2021)) ( Figure 3 and Table S7) [58]. Each of the DGAT1s was predicted to harbor seven to ten putative TMDs. For B. napus, BnaA.DGAT1.a and BnaC.DGAT1.a had eight putative TMDs, and BnaA.DGAT1.b and BnaC.DGAT1.b shared nine putative TMDs. For most DGAT2s, one to four putative TMDs were detected. Each of four BnaDGAT2s had two putative TMDs at the N-terminus with a large cytosolic C-terminal domain. There were no putative TMDs in any of the examined DGAT3s, consistent with their soluble nature.

Oil Droplets in S. cerevisiae H1246 Overexpressing BnaDGATs
DGAT is the rate-limiting enzyme in the last step of TAG synthesis, and TAG is the main component of oil bodies. Therefore, the yeast mutant H1246 (MATα are1-Δ::HIS3, are2-Δ::LEU2, dga1-Δ::KanMX4, lro1-Δ::TRP1 ADE2) [59] that are defective in TAG biosynthesis were used to determine whether the ten BnaDGATs are able to complement the mutated enzymes and allow the synthesis and accumulation of TAG. As shown in Figure  4, many obvious oil bodies were detected in yeast H1246 overexpressing BnaDGAT1s, while no obvious oil bodies were detected in H1246 overexpressing BnaDGAT2s and BnaDGAT3s, suggesting that only the four BnaDGAT1s were able to re-establish TAG synthesis in yeast H1246. Predicted transmembrane domains (TMDs) of BnaDGATs. Each BnaDGAT1s had eight or nine putative TMDs, and each BnaDGAT2s showed two putative TMDs. However, neither BnaDGAT3 had putative TMDs. The putative TMDs of BnaDGATs were predicted using TMHMM Server v. 2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0 (accessed on 10 April 2021)). Regions of BnaDGATs predicted to be located inside or outside the membrane are shown in bold blue and pink lines, respectively.

Oil Droplets in S. cerevisiae H1246 Overexpressing BnaDGATs
DGAT is the rate-limiting enzyme in the last step of TAG synthesis, and TAG is the main component of oil bodies. Therefore, the yeast mutant H1246 (MATα are1-∆::HIS3, are2-∆::LEU2, dga1-∆::KanMX4, lro1-∆::TRP1 ADE2) [59] that are defective in TAG biosynthesis were used to determine whether the ten BnaDGATs are able to complement the mutated enzymes and allow the synthesis and accumulation of TAG. As shown in Figure 4, many obvious oil bodies were detected in yeast H1246 overexpressing BnaDGAT1s, while no obvious oil bodies were detected in H1246 overexpressing BnaDGAT2s and BnaDGAT3s, suggesting that only the four BnaDGAT1s were able to re-establish TAG synthesis in yeast H1246.

Fatty Acid Profiles in Yeast H1246 and INVSc1 Expressing BnaDGATs
In this study, the fatty acids in S. cerevisiae H1246 and INVSc1 (MATa his3Δ1 leu2 trp1-289 ura3-52/MATα his3Δ1 leu2 trp1-289 ura3-52) expressing BnaDGATs were quantitatively analyzed by GC-MS to evaluate the effect of BnaDGATs on the accumulation of fatty acids in B. napus ( Figure 5). As shown in Figure 5A, in H1246, the expression of BnaDGAT1s resulted in a slight accumulation of C10:0 and C12:0 fatty acids and a significant accumulation of C14:0, C14:1, C16:0, C16:1n7, C18:0 and C18:1n9 fatty acids and increased the content of total fatty acids by one-two times but significantly decreased the Oil droplets were observed in yeast H1246 expressing BnaDGAT1s but were not detectable and/or were very weak in yeast H1246 expressing BnaDGAT2s and BnaDGAT3s. Bar = 10 µm.

Fatty Acid Profiles in Yeast H1246 and INVSc1 Expressing BnaDGATs
In this study, the fatty acids in S. cerevisiae H1246 and INVSc1 (MATa his3∆1 leu2 trp1-289 ura3-52/MATα his3∆1 leu2 trp1-289 ura3-52) expressing BnaDGATs were quantitatively analyzed by GC-MS to evaluate the effect of BnaDGATs on the accumulation of fatty acids in B. napus ( Figure 5). As shown in Figure 5A, in H1246, the expression of BnaDGAT1s resulted in a slight accumulation of C10:0 and C12:0 fatty acids and a significant accumulation of C14:0, C14:1, C16:0, C16:1n7, C18:0 and C18:1n9 fatty acids and increased the content of total fatty acids by one-two times but significantly decreased the content of C18:1n7 fatty acids. The expression of BnaDGAT2s and BnaDGAT3s promoted the accumulation of both C14:0 and C14:1 fatty acids. As shown in Figure

Expression Analysis of BnaDGATs under Different Stresses
Six-week-old B. napus seedlings were used to investigate the expression patterns of BnaDGATs in roots, stems and leaves under abiotic stresses, including P starvation, low N, drought and salinity, by qRT-PCR.
Under P starvation, compared with mock, the expression levels of BnaDGAT1s, BnaA.DGAT2.b and BnaC.DGAT2.b in roots, stems and leaves were lower; the expression patterns of BnaA.DGAT2.a and BnaC.DGAT2.a in roots, stems and leaves were not changed significantly; the expression levels of two BnaA.DGAT3s in leaves were decreased, while their expression levels in roots and stems were not changed significantly (Figure 7).
During low N, compared with mock, the expression patterns of four BnaDGAT1s in stems and leaves and BnaA.DGAT1.b in roots were not changed, while the expression levels of BnaA.DGAT1.a and BnaC.DGAT1.a in roots were lower. For BnaDGAT2s, the expression patterns of BnaA.DGAT2.a and BnaC.DGAT1.a in roots and stems under low N were not changed, while their expression levels in leaves were lower; the expression levels of BnaA.DGAT2.b in leaves and stems were lower, while its expression levels in roots were higher; the expression levels of BnaC.DGAT2.b in stems were lower, while its expression levels in leaves and roots were higher. For the two BnaDGAT3s, their expression patterns in roots and stems were quite consistent with those in mock, while their expression levels in leaves decreased at the late stages of low N treatment (Figure 7).
During salt stress, compared with mock, the expression levels of four BnaDGAT1s in the stems were higher, while their expression levels in roots were lower. For BnaDGAT2s, the expression levels of BnaA.DGAT2.a and BnaC.DGAT2.a in roots, stems and leaves were higher; the expression levels of BnaA.DGAT2.b in leaves and stems were lower, while its expression levels in roots were higher; the expression levels of BnaC.DGAT2.b in stems were lower, but its expression levels in roots were higher. For BnaDGAT3s, the expression levels of BnaA.DGAT3 in stems and leaves and BnaC.DGAT3 in roots, stems and leaves were lower than those in mock with 12-48 h stress (Figure 7).
During drought stress, compared with mock, the expression levels of four BnaDGAT1s in leaves at 12 h of treatment and BnaC.DGAT1.a BnaA.DGAT1.b and BnaC.DGAT1.b in roots at 48 h of treatment were higher, while the expression patterns of BnaA.DGAT1.a and BnaC.DGAT1.a in stems were quite consistent. For BnaDGAT2s, the expression levels of BnaA.DGAT2.a and BnaC.DGAT2.a in stems and leaves were higher; the expression levels of BnaA.DGAT2.b in leaves and stems were lower, while its expression levels in roots were higher; the expression patterns of BnaC.DGAT2.b in leaves and stems were quite consistent with those in mock, while its expression levels in roots were higher than those in mock. For BnaDGAT3s, the expression levels of BnaA.DGAT3 in stems and roots and BnaC.DGAT3 in roots were higher than those in mock, while the expression levels of BnaC.DGAT3 in leaves treated for 12-48 h were quite low (Figure 7). N, drought and salinity, by qRT-PCR.
Under P starvation, compared with mock, the expression levels of BnaDGAT1s, BnaA.DGAT2.b and BnaC.DGAT2.b in roots, stems and leaves were lower; the expression patterns of BnaA.DGAT2.a and BnaC.DGAT2.a in roots, stems and leaves were not changed significantly; the expression levels of two BnaA.DGAT3s in leaves were decreased, while their expression levels in roots and stems were not changed significantly (Figure 7).   Under P starvation, compared with mock, the content of total fatty acids in roots at 6 h to 48 h, in stems at 3 h to 48 h and in leaves at 1 h to 48 h significantly increased, as well  Figure 8D).

Cis-Elements in BnaDGAT Promoters and Transcription Factors and miRNA Regulating BnaDGATs
To gain more insights into the potential function and regulatory mechanism of ten BnaDGATs, we analyzed the cis-regulatory elements in their putative promoters by using the plantCARE database. The putative promoters of ten BnaDGATs were cloned as 1759 bp, 1488 bp, 1657 bp, 836 bp, 775 bp, 2151 bp, 1973 bp, 1660 bp and 1459 bp from ZS11 ( Table 2 and Table S6). The putative promoters of three AtDGATs were derived from TAIR (Table S9). The cis-acting regulatory elements in the 776-1500 bp upstream promoter regions of BnaDGATs and AtDGATs were displayed ( Figure 9 and Table S10).
As the putative promoters of DGAT1s, they all shared some light-responsive elements. All P AtDGAT1 , P BnaA.DGAT1.a , P BnaC.DGAT1.a , and P BnaC.DGAT1.b shared anaerobic induction elements and gibberellin-responsive elements. P AtDGAT1 , P BnaA.DGAT1.a and P BnaC.DGAT1.a possessed MeJA-responsive elements and defense and stress-responsive elements. Both P AtDGAT1 and P BnaA.DGAT1.a contained a circadian control element. Both P BnaA.DGAT1.b and P BnaC.DGAT1.b had a low-temperature-responsive element. P AtDGAT1 contained a droughtinducibility element. P BnaC.DGAT1.a contained an SA-responsive element. P BnaA.DGAT1.a contained two GCN4_motifs and three Skn-1_motifs, which are endosperm-expressive elements. In addition, P BnaA.DGAT1.a contained a seed-specific RY element. There was only one Skn-1_motif in P BnaC.DGAT1.a .  The putative promoters of AtDGAT2 and BnaDGAT2s also contained some lightresponsive elements, anaerobic induction elements and circadian control elements. P AtDGAT2 , P BnaA.DGAT2.a and P BnaA.DGAT2.b shared a drought-responsive element. P AtDGAT2 , P BnaA.DGAT2.b and P BnaC.DGAT2.b contained gibberellin-responsive elements. P BnaC.DGAT2.a , P BnaA.DGAT2.b and P BnaC.DGAT2.b shared a Skn-1_motif and an ABA-responsive element. Both P BnaA.DGAT2.b and P BnaC.DGAT2.b had a low-temperature-responsive element. Both P BnaA.DGAT2.a and P BnaC.DGAT2.a shared an elicitor-responsive element. Both P BnaA.DGAT2.b and P BnaC.DGAT2.b contained one MeJA-responsive element. P AtDGAT2 contained an auxin-responsive element and an SA-responsive element. P BnaC.DGAT2.a contained a heat-stress-responsive element and a wound-responsive element. P AtDGAT3 , P BnaA.DGAT3 and P BnaC.DGAT3 , also contained some light-responsive elements and ABA-responsive elements. Both P BnaA.DGAT3 and P BnaC.DGAT3 contained MeJAresponsive elements. Both P AtDGAT3 and P BnaA.DGAT3 shared a defense-and stress-responsive element. Both P AtDGAT3 and P BnaC.DGAT3 contained low-temperature-responsive elements. P AtDGAT3 contained a gibberellin-responsive element and an auxin-responsive element. P BnaA.DGAT3 contained an endosperm expression element and an anaerobic induction element. P BnaC.DGAT3 contained a drought-responsive element.
Transcription factors (TFs) regulate the precise initiation of gene transcription by binding the cis-acting elements of gene promoters. Therefore, we identified the target TFs putatively regulating the expression of BnaDGATs using the PlantRegMap server, and a total of 209 relationships were identified (Table S11). Three BnaDGAT gene families may be regulated by different TFs, such as the B3 family, dehydration-responsive element-binding protein (DREB), GATA, ethylene response factor (ERF), WRKY family, MYB and bZIP family, indicating that these TFs regulate plant development and stress responses.

Discussion
B. napus is an allotetraploid (AACC) crop that originated from the hybridization of two diploid progenitors, B. rapa (AA) and B. oleracea (CC) [51,52]. In this study, ten BnaDGATs were cloned and identified from B. napus and grouped into three subfamilies-BnaDGAT1, BnaDGAT2 and BnaDGAT3-based on their homology. Systematic analyses of chromosome location, gene synteny, physicochemical properties, phylogenetic tree, exon/intron gene structure, conserved domain and motif compositions, TMDs and the distribution of ciselements in the promoters were performed. The functions of BnaDGATs were detected in yeast H1246 and INVSc1. Moreover, qRT-PCR and prepublished RNA-seq data were analyzed to determine the expression patterns of BnaDGATs. These results provide an extensive evaluation of BnaDGATs and a useful foundation for dissecting the functions of BnaDGATs in biochemical and physiological processes.

Gene Duplication and Functional Diversification of DGAT Family Members
The phylogenetic tree showed that BnaDGATs were grouped into three well-supported clades: the DGAT1 subfamily, the DGAT2 subfamily and the DGAT3 subfamily (Figure 2A). Identity analysis showed that the three DGAT subfamilies shared very low identity with each other subfamily (Table S8). Gene synteny analysis showed that there was no gene synteny among the three DGAT subfamilies (Figure 1). The analysis of gene structures revealed distinct differences among the three DGAT subfamilies in exon/intron gene structure ( Figures 2B and S2B). Conserved domain and motif analyses showed that the three DGAT subfamilies had different conserved domains and motifs (Figures 2C and S3B). Each DGAT1 was predicted to harbor seven to ten putative TMDs, one to four putative TMDs were detected in most DGAT2s and there were no putative TMDs in any of the examined DGAT3s (Figure 3 and Table S7). The expression patterns of the three BnaDGAT gene families were different among diverse tissues ( Figure 6). All of these results showed that the DGAT1, DGAT2 and DGAT3 gene subfamilies showed apparent differences and indicated that they are divergent genes and may have a distinct origin, consistent with what is described in soybeans [65] and upland cotton [25]. In addition, the results of this study showed that three BnaDGAT gene subfamilies were frequently duplicated during the speciation and evolution of B. oleracea, B. rapa and B. napus and suggested that B. napus inherited and retained all the DGAT genes of B. rapa and B. oleracea (Figures 1 and S1), consistent with what is described in tetraploid cotton [25].

Role of BnaDGATs in Oil Biosynthesis
DGAT1 has been functionally confirmed in oil biosynthesis in Arabidopsis, soybean, oilseed rape, and so on [3]. Expression analysis revealed that DGAT1 was abundantly expressed in the developing embryos in several oilseed crops, and its transcript level was correlated with oil accumulation in developing seeds [49]. Analysis of the putative promoters showed that P BnaA.DGAT1.a contained two GCN4 motifs (endosperm expressive elements), three Skn-1 motifs (endosperm expressive elements) and one seed-specific RYelement and that P BnaC.DGAT1.a possessed one Skn-1 motif (Figure 9). In this study, the expression levels of BnaA.DGAT1.a and BnaC.DGAT1.a in seeds and embryos gradually increased during seed and embryo development (Figure 6), which corresponded with the rapid oil accumulation stage in canola seeds, indicating that BnaA.DGAT1.a and BnaC.DGAT1.a were important in TAG biosynthesis in canola seeds. The expression levels of BnaA.DGAT1.b and BnaC.DGAT1.b in seeds first gradually increased and then gradually decreased during seed development, and the expression levels of BnaA.DGAT2.a and BnaC.DGAT2.a in seeds, silique walls and embryos first gradually increased and then gradually decreased during the development of seeds, silique walls and embryos (Figure 6), which implied that BnaA.DGAT1.b, BnaC.DGAT1.b, BnaA.DGAT2.a and BnaC.DGAT2.a may also play an important role in TAG biosynthesis in canola seeds. Moreover, there were high expression levels of BnaA.DGAT1.b and BnaC.DGAT1.b in the anthers and stamens of canola, indicating that BnaDGAT1 might be involved in the reproductive development of oilseed rape.
Some studies have focused on the role of DGAT3 in TAG biosynthesis [10,25,46,47,72]. In this study, almost all DGAT3 homologues were found to share the TRX_Fd_family domain ( Figure 2C). To date, AtDGAT3 and CsDGAT3 have been confirmed as metalloproteins involved in TAG biosynthesis in plants [46,72]. DGAT3 homologues were not found in mossy or algal species [73], indicating that they may have arisen during plant evolution. The expression levels of BnaDGAT3s were significantly higher than those of BnaDGAT1s and BnaDGAT2s during canola seed development (Figure 6), consistent with what is described in upland cotton and soybean [65,74]. In this study, overexpressing BnaDGAT3s could not restore TAG synthesis in yeast H1246 and did not promote the accumulation of fatty acids (Figures 4 and 5). Therefore, the function of BnaDGAT3s in synthesizing TAGs needs further testing using oilseed rape or other protein expression systems.

The Response of BnaDGATs to Abiotic Stresses
Phosphorus starvation can increase the accumulation of oils in most microalgae [75][76][77][78][79][80] as well as in the vegetative tissues of plants, such as Arabidopsis, tomato, tobacco and barnyard grass (Echinochloa crusgalli) [81,82]. In this study, we found that the accumulation of total fatty acids was greatly promoted ( Figure 8A), but the expression levels of BnaDGAT1s, BnaDGAT2s and BnaDGAT3s were not enhanced in the roots, stems and leaves of B. napus seedlings under P starvation (Figure 7). In Arabidopsis, the expression of genes involved in TAG synthesis, such as AtDGAT1, AtDGAT2, AtPDAT1, AROD1, AtLPCAT2, AtBCCP2 and PDH-E1a, was not increased under P starvation [81,82]. The above studies showed that the expression levels of DGAT1s were not directly related to the increase in TAG accumulation under P starvation. In previous studies, a self-inhibiting motif in the N-terminal region of BnaDGAT1 bound PA and shifted BnaDGAT1 to a higher activity state [83,84]; PLDZ2 (phospholipase D Z2) degraded phospholipids into PA and was induced by P starvation [85,86]. SnRK1 (Snf1-related kinase 1) inhibited DGAT1 activities by phosphorylating S/T in the SnRK1 targeting motif [23,52,84], and the activity of Arabidopsis SnRK1 was reduced and its catalytic subunit AKIN11 was degraded under P starvation [87]. Therefore, the enhancement of TAG accumulation may be related to the higher activities of DGAT1s regulated by the increase in PA and the decrease in SnRK1 activity under P starvation. In addition to TAG, P starvation can also promote the accumulation of DAG, MGDG, DGDG and SQDG by inducing the expression of NPC4, NPC5, PLDZ2, PAH1, PAH2, MGD2, MGD3, DGD1, DGD2, SQD1 and SQD2 in Arabidopsis seedlings [81,82]. This result suggested that the increase in total fatty acids is a consequence of the accumulation of TAG, DAG, MGDG, DGDG and SQDG in B. napus seedlings under P starvation, which may be caused by gene expression changes similar to those in Arabidopsis.
Many studies have shown that most microalgae can accumulate high levels of oil under N starvation [75,[88][89][90][91][92][93][94]. For plants, N and C are closely coordinated to affect chloroplast lipid metabolism and TAG content [36,[95][96][97]. This study found that low N (5 mM) promoted the accumulation of total fatty acids in roots at 1 h to 6 h and 24 h and in stems at 24 h and 48 h, but inhibited the accumulation of total fatty acids in stems at 6 h and in leaves at 48 h ( Figure 8B). In previous reports, low N (0.1 mM and 0.65 mM) promoted TAG accumulation, especially 0.1 mM N with 50 mM sucrose, by inducing the expression of AtDGAT1, AtDGAT2, AtPDAT1 and AtOLEOSIN1 in Arabidopsis seedlings [36,96]. In this study, low N (5 mM) did not induce the expression of four BnaDGAT1s, BnaA.DGAT2.a, BnaC.DGAT2.a and two BnaDGAT3s in roots, stems and leaves; BnaA.DGAT2.b in leaves and stems and BnaC.DGAT2.b in stems but promoted the expression of BnaA.DGAT2.b in roots and BnaC.DGAT2.b in leaves and roots (Figure 7). In previous studies, Arabidopsis seedlings were cultured under low N (0.1 mM) using Murashige and Skoog (MS) solid medium, while B. napus seedlings were cultured under low N (5 mM) in this study using Hoagland solution without carbon sources. In normal 1/2 Hoagland solution, the nitrogen content was 7.5 mM. Therefore, it may not be sufficient to induce the expression of BnaDGATs in B. napus under 5 mM N (50 times that under 0.1 mM).
In this study, the accumulation of total fatty acids in roots and leaves at 3 h and 6 h and in stems at 1 h to 6 h was reduced but was promoted in stems at 12 h and 24 h using 15% PEG for stress treatment ( Figure 8C). Previous studies showed that drought stress reduced phospholipids (PC, PE, PG), glycolipids (MGDG and DGDG) and total fatty acids in B. napus and increased neutral lipids (mainly TAG) [98][99][100][101], which is consistent with maize [102], soybean [103] and cotton [104]. Therefore, the changes in total fatty acids in the roots, stems and leaves of B. napus seedlings under drought stress were the combined result of a decrease in phospholipids and an increase in neutral lipids. In this study, drought stress for 24 h increased the expression levels of four BnaDGAT1s in leaves, BnaA.DGAT2.a and BnaC.DGAT2.a in stems and leaves; BnaA.DGAT2.b and BnaC.DGAT2.b in roots; BnaA.DGAT3 in stems and roots and BnaC.DGAT3 in roots, but decreased the expression levels of BnaA.DGAT2.b in leaves and stems and BnaC.DGAT3 in leaves (Figure 7). A previous study showed that the expression of AtDGAT1 was promoted by ABI4 and ABI5 under drought stress [38]. Therefore, it is necessary to further test whether ABI4 and ABI5 participate in the regulation of BnaDGAT expression in B. napus under drought stress. Furthermore, drought-inducible elements (MBS) were found in the potential promoters of BnaA.DGAT2.a, BnaA.DGAT2.b, BnaC.DGAT2.b, BnaA.DGAT3 and BnaC.DGAT3 ( Figure 9 and Table S10). In Arabidopsis, MBS is generally present in the promoter of drought-inducible genes and bound by AtMYB2 in response to drought stress [105][106][107]. The expression of AtMYB2 was induced by drought, salt or ABA treatment in Arabidopsis seedlings [105]. Therefore, the expression levels of BnaA.DGAT2.a, BnaA.DGAT2.b, BnaC.DGAT2.b, BnaA.DGAT3 and BnaC.DGAT3 may be regulated by BnaMYB2 under drought stress, which needs to be further tested.
For most microalgae, the appropriate salinity (20-40 g/L, i.e., 340-680 mM) can increase the accumulation of lipids, while lipid accumulation will be negatively affected when the salinity is excessive [75]. In Arabidopsis, TAG accumulation and the expression of DGAT1 were promoted by enhancing the expression of ABI4 and ABI5 under salt stress (100 mM NaCl) [38]. In this study, it was found that the accumulation of total fatty acids was reduced in roots at 1 h to 24 h and in stems at 1 h to 6 h and 48 h, while it was elevated in leaves at 6 h and 48 h under salt stress (150 mM NaCl; Figure 8D). Moreover, the expression levels of four BnaDGAT1s in stems; BnaA.DGAT2.a and BnaC.DGAT2.a in roots, stems and leaves and BnaA.DGAT2.b in roots were increased, but the expression levels of four BnaDGAT1s in roots; BnaA.DGAT2.b in leaves and stems; BnaC.DGAT2.b in stems; BnaA.DGAT3 in stems and leaves and BnaC.DGAT3 in roots, stems and leaves at 12-48 h of stress decreased (Figure 7). Therefore, it is necessary to further test and explore the relationship between the accumulation of fatty acids and the expression of BnaDGATs in B. napus under salt stress.

Identification of DGAT Family Members in B. napus and in Other Plants
To identify candidate DGAT family members in B. napus, the CDS and peptide sequences of the three AtDGATs from the A. thaliana genome database (http://www. arabidopsis.org/ (accessed on 25 March 2020)) with corresponding Gene IDs (At2G19450, At3G51520 and At1G48300) were retrieved and used as queries to perform BLAST searches in two B. napus Genome Databases (BnPIR, http://cbi.hzau.edu.cn/bnapus (accessed on 2 March 2021), and GENOSCOPE, https://www.genoscope.cns.fr/brassicanapus/ (accessed on 25 March 2020)) with the default parameters. The CDSs and genomic DNAs of BnaDGATs were cloned from B. napus ZS11 using primers (Table S1) (Table S2). Then, the theoretical molecular weight (MW) and isoelectric point (pI) were predicted using the ProtParam tool (https://web.expasy.org/protparam/ (accessed on 22 October 2021)) on the basis of their amino acid sequences. The subcellular location pattern of each BnaDGAT was evaluated via ProtComp v.9.0 in softberry (http://linux1.softberry.com/ (accessed on 22 October 2021)). Multiple sequence alignments of DGATs of B. napus, B. oleracea and B. rapa were performed based on full-length proteins and full-length CDSs by ClustalW, and their identities were evaluated by Sequence Distances in MegAlign software.

Chromosomal Location and Gene Synteny Analysis
The detailed chromosome locations of DGATs in B. napus, B. rapa and B. oleracea were acquired from the GFF genome files downloaded from the B. napus genomic database (Bn-PIR, http://cbi.hzau.edu.cn/bnapus (accessed on 11 October 2021)) and Brassica Database (BRAD, http://brassicadb.cn/#/ (accessed on 11 October 2021)), respectively, and the predicted locations on the chromosomes were mapped by using TBtools software [62], with red-colored gene names indicated as relative positions. Gene synteny analyses of DGATs in A. thaliana, B. oleracea, B. rapa, B. napus and B. nigra were carried out by using TBtools software [62]. In addition, we calculated the synonymous (Ks), nonsynonymous mutations (Ka) and Ka/Ks ratio at each codon by Tbtools [62]. In addition, the DGAT gene pair duplication time was presumed using the formula T = Ks/2R, with R (1.5 × 10 −8 ) representing neutral substitution per site per year [108].

Phylogenetic, Gene Structure, Conserved Domain and Motif Analyses
The MUSCLE program in MEGA11 software was used to perform DGAT alignments with amino acid sequences by default parameters, and then the rectangular phylogenetic tree was constructed using the neighbor-joining (NJ) method with 2000 bootstrap replications. The rectangular phylogenetic tree Newick format was saved for constructing the gene structure, conserved domain and motifs. The genomic and coding sequences of DGAT genes were obtained from their corresponding genome databases (Table S2) and rendered in the Gene Structure Display Server (GSDS2.0; http://gsds.gao-lab.org/ (accessed on 20 October 2021)) [56] to construct their gene structures. Amino acid sequences of DGATs were submitted to the MEME program (Version 5.4.1, https://meme-suite.org/meme/tools/ meme (accessed on 21 October 2021)) with the maximum motif search set to 20 and other parameters set to default to identify the conserved protein motifs [109]. A functional search of the conserved domains was performed using the Conserved Domain Database (CDD) with Batch CD-Search in NCBI. The conserved motifs and domains were visualized by using TBtools software [62]. The putative transmembrane domains of DGATs were predicted using TMHMM Server v. 2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0 (accessed on 10 April 2021)). Multiple sequence alignments of DGATs were performed based on their full-length proteins by ClustalW in MegAlign software and then crested with CLC Sequence Viewer 6.8 software.

Nile Red Staining and Microscopy
Lipid droplets in the transformants of yeast H1246 were stained with Nile Red and then visualized on a Leica TCS SP5 (Leica Microsystems, Wetzlar, Germany) laser scanning confocal microscope, as previously described [92].

Analysis of Fatty Acids in Yeast Transformants and B. napus Seedlings by GC-MS
The transformants of yeast H1246 and INVSc1 first cultured with liquid SC-ura medium containing 2% (w/v) glucose at 30 • C overnight were then diluted to OD 600 = 0.1 with liquid SC-ura medium containing 2% (w/v) galactose and 1% (w/v) raffinose and shaken at 30 • C and 250 rpm for 72 h. The yeast cells were harvested by centrifugation and dried at 55 • C, and then the pellets were ground to a fine powder. For B. napus seedlings, samples of roots, stems and leaves were also ground to a fine powder under liquid nitrogen and then freeze-dried. For fatty acid extraction, 50 mg of yeast powder or 15 mg of freeze-dried powder of every B. napus sample was incubated in 3 mL of 7.5% (w/v) KOH in methanol for saponification at 70 • C for 5 h. Then, the pH was adjusted to 2.0 with HCl, and the fatty acid was subjected to methyl-esterification with 2 mL of 14% (w/v) boron trifluoride in methanol at 70 • C for 2 h. A phase separation was produced by adding 2 mL of 0.9% (w/v) NaCl and 4 mL of hexane. The upper phase was dried under a nitrogen gas flow and resuspended in acetic ether. Analysis of fatty acid methyl esters (FAMEs) was performed by GC-MS (GC-QQQ, 7890A-7001B, Agilent Technologies, Santa Clara, CA, USA) equipped with a capillary column (HP-FFAP, 30 mm × 0.25 mm ID, 0.25 µm; Agilent Technologies). Hydrogen was used as the carrier gas at a flow rate of 1.0 mL/min. The injection port, transmission line and ion source temperatures were held at 220 • C, 230 • C and 230 • C, respectively. The temperature of the column oven was programmed from 60 to 180 • C at 10 • C/min, then from 180 to 210 • C at 3 • C/min, and finally from 210 to 220 • C at 5 • C/min and held for 15 min. The FA content was quantified using heptadecanoic acid (C17:0, Sigma, St. Louis, MO, USA) as an internal standard added to samples prior to extraction. All experiments were performed in biological triplicates.

B. napus Seedling Treatments and Sampling
The seedlings of B. napus Westar were hydroponically cultured in 1/2 Hoagland solution [110] for six weeks with a 16 h day/8 h night cycle at 23 • C and then were used for the stresses of P starvation (KH 2 PO 4 replaced by an equimolar amount of KCl), low N (KNO 3 replaced with an equimolar amount of KCl), drought (15% w/v PEG6000) and salinity (150 mM NaCl). Seedlings cultured in normal 1/2 Hoagland solution were used as the control group (mock). Roots, stems and leaves of seedlings were sampled at seven time points: 0 h, 1 h, 3 h, 6 h, 12 h, 24 h and 48 h. The collected samples were immediately dipped in liquid nitrogen and then stored at −80 • C for the relative expression levels of BnaDGATs and fatty acid analysis. All experiments were performed in biological triplicates.

Genomic DNA Extraction, Total RNA Isolation, Primary cDNA Synthesis and qRT-PCR
Genomic DNA of B. napus ZS11 was extracted using a method modified from a CTABbased protocol [111] for cloning genomic DNA of BnaDGATs. Total RNA of B. napus was extracted using an RNAprep Pure Plant Kit (DP432, TIANGEN BIOTECH Co., Ltd., Beijing, China), and B. napus complementary DNA was synthesized using TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix (AT311-03, TransGen Biotech, Beijing, China), according to the manufacturer's instructions. For qRT-PCR, 2 µg of total RNA was first used for the synthesis of first-strand cDNA using oligo(dT) 18 as a primer, and then qRT-PCR was performed on a CFX Connect TM Real-Time PCR system (Bio-Rad, Hercules, CA, USA) using EvaGreen 2× qPCR MasterMix (MasterMix-S, Abm, Vancouver, BC, Canada), according to the manufacturer's instructions. The specificity of primers (Table  S3) for qRT-PCR was confirmed by separating the products on agarose gels and clone sequencing. BnaACT7 was used as an internal reference gene, and the relative expression levels of each BnaDGAT to BnaACT7 in leaves, stems and roots at 0 h were set to 1. All qRT-PCRs were performed in biological triplicates.

Analyses of Transcription Factors and miRNAs Targeting BnaDGATs and Cis-Acting Elements in BnaDGAT Promoters
Transcription factors regulating BnaDGATs were predicted using PlantRegMap [112], with B. napus as the target. The full-length cDNA sequences of BnaDGAT homologues were submitted to the psRNATarget website (https://www.zhaolab.org/psRNATarget/ (accessed on 25 November 2021)) [113] for a potential miRNA search against the B. napus miRNA database. For cis-element analysis, the regions 776-1500 bp upstream of the start codon of BnaDGATs and AtDGATs were subjected to the plantCARE database (http://bioinformatics. psb.ugent.be/webtools/plantcare/html/ (accessed on 20 September 2021)) [114] for ciselement searching.

Conclusions
In summary, ten BnaDGATs were identified and grouped into three different DGAT subfamilies. These BnaDGATs are derived from three different ancestors and evolved separately during plant evolution, as proposed by analyzing physicochemical properties, chromosome location, gene synteny, phylogenetic tree, exon/intron gene structure, conserved domain and motif compositions and TMDs. BnaDGAT1s possess the ability to introduce the fatty acids C10, C12, C14, C16 and C18 into TAG in S. cerevisiae and a higher preference for the fatty acids C16:0, C16:1n7, C18:0 and C18:1n9 than C10:0, C12:0, C14:0, C14:1 and C18:1n7. BnaDGAT1s are the main diacylglycerol acyltransferases that synthesize TAGs in B. napus. The role of BnaDGAT2s and BnaDGAT3s in TAG synthesis in B. napus needs to be further clarified. Some B3, bZIP, MYB-like transcription factors and other transcription factors involved in the response to light signals may regulate the expression of BnaDGATs. P starvation increased fatty acid accumulation in B. napus seedlings. The relationships between the expression of BnaDGATs and the accumulation of lipids in B. napus under low N, drought and salt conditions remain to be further confirmed. Overall, the findings of this study contribute to the understanding of BnaDGAT genes in fatty acid biosynthesis and abiotic stress responses in oilseed rape.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/plants11091156/s1, Figure S1: Schematic representation of DGAT gene distribution on Brassica napus, Brassica rapa, and Brassica oleracea chromosomes; Figure S2: Gene structures of DGAT family members in plants; Figure S3: The conserved motifs of DGAT family members in plants; Table S1: Primer sequences for PCR and RT-PCR analysis in this study; Table S2: The corresponding genome databases of B. napus and other plant species in this paper; Table S3: The sequences of DGAT proteins derived from the corresponding genome databases of B. napus and other plant species in this study; Table S4: The CDSs of DGAT1 genes downloaded from the corresponding genome databases of B. napus and other plant species in this study; Table S5: The genomic DNA sequences of DGAT1 genes downloaded from the corresponding genome databases of B. napus and other plant species in this study; Table S6: The sequences of the CDSs, genomic DNAs, proteins and putative promoters of BnaDGAT gene members cloned from B. napus ZS11 in this study; Table S7: Identification of DGAT1 subfamily genes in plants; Table S8: Identity matrix of DGATs in B. napus and three diploid Brassica species based on full-length proteins and CDSs; Table S9: The sequences of the putative promoters of AtDGAT1, AtDGAT2 and AtDGAT3; Table S10: The cis-acting regulatory elements in the putative promoters of BnaDGATs and AtDGATs were analyzed by Plant-CARE; Table S11: Transcription factors regulating BnaDGATs were predicted using PlantRegMap; Table S12: miRNAs target B. napus, B. rapa and A. thaliana DGAT genes. References [115][116][117][118][119][120][121][122][123][124][125][126][127][128][129][130][131][132][133]