To identify the SWEET genes in B. rapa genome, amino acid sequences of 17 A. thaliana SWEET (AtSWEET) genes were used as the bait to search against the B. rapa database (BRAD¹) by performing a BLASTP analysis. The physicochemical parameters, including molecular weight (kDa) and isoelectric points (pI), were calculated according to the pI/Mw tool at online ExPASy² database. Proteins containing MtN3/saliva domain, the typical transmembrane domains of eukaryotic SWEET, were defined to belong to the B. rapa SWEET gene family (BrSWEETs). BrSWEETs paralogous genes to the AtSWEET was named by adding a suffix (a, b, c…) according to E values from high to low, representing sequence similarity level to the corresponding AtSWEET.

To better understand the evolutionary relationships among A. thaliana, B. napus, and B. rapa SWEETs gene family, multiple sequence alignment of all BrSWEET sequences was conducted using the ClustalX program. An unrooted phylogenetic analysis was performed using the neighbor-joining (NJ) method of the MEGA5 program using the full-length SWEET amino acid sequences of A. thaliana, B. napus, and B. rapa. A bootstrap test was performed with 1000 replicates.

All identified SWEET genes were mapped to B. rapa chromosomes using Mapchart (Version 2.1) software based on their physical position available at the B. rapa genome database¹. Gene Structure Display Server (GSDS)³ was used to draw the gene structure diagram of the BrSWEET family genes according to the genomic sequences and the corresponding coding sequence of each BrSWEET gene. To further confirm the coding sequence, all BrSWEETs were queried against on line EST database deposited in NCBI and sequences generated from our lab RNA-seq data of B. rapa (Chen et al., 2016) with BLAST.

TMHMM Server 2.0⁴ was utilized to forecast the protein structure of BrSWEET family. We performed the MEME (Multiple Expectation maximization for Motif Elicitation) analysis on the predicted BrSWEET with the conditions: (1) optimum motif width was set to 6 and 50; (2) maximum number of motifs was designed to 10 motifs; (3) the iterative cycles were set by default.

A pair of Chinese cabbage near-isogenic lines (NIL), carrying either the CR or clubroot-susceptible (CS) allele at the CRb locus, were used in our study. Briefly, CR-NIL was bred by introgressing CRb gene from clubroot-resistant line ‘CR Shinki DH’ to the susceptible inbred line “BJN3-2” based on marker-assisted selection (Zhang et al., 2012). All the plants were maintained in the culture room under a 16 h photoperiod at 25°C. The single-spore isolate (Pb4) of P. brassicae was propagated in CS Chinese cabbage. Resting spores were isolated from homogenized clubbed roots and diluted to a density of 10⁷/mL with sterile distilled water until inoculation. Thirteen-day-old seedlings of CR and CS NILs were inoculated by injecting 1 mL of resting spore suspension into the soil around each plant (Zhang et al., 2012). Plants supplemented with the same volume of water were used as the control. Leaves, roots and hypocotyls from 10 individuals were collected at 0, 0.5, 1, 1.5, 2, 3, 4, 6, and 9 days post inoculation (dpi) for gene expression analysis, respectively. Also, sugar contents of the leaves and roots were determined. To verify the successful infection, a pair of NILs was maintained at the culture room for 30 days. Three biological replicates were included for each treatment, and each replicate contained 10 plants.

For analyzing clubroot formation in Arabidopsis, T-DNA insertion mutant sweet11 (SALK_073269.20.35.X) from the Salk collection and wild type Col-0 were grown for 3 weeks in a chamber with 12 h light/12 h dark cycle. The phenotype of Atsweet11, including plant height, leaf size, leaf color, and flowering showed no difference to wild type Col-0. P. brassicae was infected and the gall formation was checked after 3, 4, and 5 weeks. Three biological replicates were conducted for each treatment, and each replication contained 10 plants. The disease rate (DR) and the disease index (DI) was evaluated after inoculation. Disease symptoms were scored as follows: 0, no symptoms; 1, a few small clubs on the lateral roots; 2, larger clubs on the lateral roots; 3, swelling of the main roots; 4, severe galling of tissues of both lateral and main roots. The DI was calculated according to the formula DI = [nw] × 100/4T, where n is the number of plants in each class, w is disease symptoms (0–4), and T is the total number of plants tested (Chaube and Singh, 1991; Siemens et al., 2010). Data on the phenotype of disease index of the tested lines were analyzed for statistical significance using analysis of variance (ANOVA) conducted using SPSS 17.0.

For pathogen DNA quantification, total DNA was extracted from 1 g roots of the Arabidopsis sweet11 mutants and wild type Col-0 by DNAsecure plant kit (Tiangen, Beijing, China), according to the manufacturer’s instructions. DNA quantity was estimated with Nanodrop instrument and final concentration was adjusted to 10 ng/μl for each PCR reaction. The Arabidopsis F-box protein gene mentioned by Lemarie et al. (2015) was used for normalization and the P. brassicae target gene was described by Faggian et al. (1999). Quantitative real-time PCR reaction was performed following by Lemarie et al. (2015). The relative DNA quantities were expressed as the ratio between the DNA quantities of P. brassicae and the plant DNA value. Three replications were performed for each reaction. Significant differences between the mean values among Arabidopsis wild type and mutants were determined according to ANOVA analysis at P < 0.01 using SPSS 17.0.

For SWEET gene expression analysis, total RNA was extracted from the roots, hypocotyls and leaves of CR and CS NILs by using the RNA prep Pure Plant Total RNA Extraction Kit (Tiangen, Beijing, China) according to the manufacturer’s instructions. cDNA was synthesized from 1 μg RNA after removement of genomic DNA. The quantitative real-time PCR (qRT-PCR) analysis was performed as described by Jain et al. (2006). Briefly, the cDNA samples were used as template and mixed with 200 nmol of each primer and SYBR Green PCR Real Master Mix (Tiangen, Beijing, China) for real-time PCR analysis using ABI 7500 Real Time PCR System and Software 7500 ver. 2.0.3 (Applied Biosystems, United States) according to the manufacturer’s instructions. The thermal cycling was as following: 95°C for 3 min; 40 cycles of 95°C for 30 s, 57°C for 30 s, and 68°C for 1 min. The fluorescence signal was collected during the elongation at 68°C of every cycle. The gene specific primer sequences were listed in Supplementary Table S2. The relative expression levels for each of the SWEET genes were quantified with respect to the internal standard, Actin and 18sRNA. The experiments were repeated at least three times.

Soluble sugars were extracted from the roots and leaves of CR and CS NILs. After grinding the tissues, 1 g of powder samples were dissolved in 80% ethanol and were kept at 80°C for 1 h. The extracts were concentrated and dissolved with 1 ml deionized and distilled H₂O. Sucrose, glucose and fructose were assayed by high performance liquid chromatography (HPLC) using a NH₂ Analytical HPLC column (Waters, Milford, MA, United States) with 75% acetonitrile and 25% double-distilled H₂O as solvent, the velocity of 1.0 ml⋅min^-1 and refractometer Parallax detection, according to manufacturer’s directions (Siemens et al., 2010). Soluble sugars were extracted from three biological replicates at each time point as aforementioned treatments. HPLC analysis were repeated three times for each treatment.

To test the effects of P. brassicae infection on sugar translocation, the contents of soluble sugars (glucose, fructose, and sucrose) in the leaves and roots of CS and CR NILs were examined before and after infection. The contents of sugars had no significant differences between CS and CR NILs without P. brassicae inoculation, but showed variance after infection (Figure 1).

Upon P. brassicae inoculation, an increase of glucose contents was observed in the leaves and roots of both two genotypes (Figures 1A,B). The contents of glucose were slightly increased in the leaves of CS and CR NILs at 1 dpi, while increased from 2 dpi. Compared to the CS-NIL, CR-NIL showed dramatically high content of glucose in the leaves from 2 to 9 dpi. However, the contents of glucose were dramatically increased from 12 to 49% in the roots of the CS-NIL from 1 to 9 dpi, and much less increase of the glucose contents occurred in the roots of the CR-NIL (Figures 1A,B).

In the leaves of un-inoculated plants, fructose contents gradually increased in the leaves of both CS and CR NILs at 1 dpi (Figure 1C). Generally, the fructose contents were lower in the leaves of both CS-NIL and CR-NIL than that of un-inoculated plants at each time point after P. brassicae inoculation. It was noticed that significant decreases of fructose contents were observed in the leaves of CS-NIL compared to CR-NIL from 2 dpi. However, significant increases were observed in the roots of both NILs during the infection time course, especially in CS-NIL. The fructose contents in roots of CS-NIL were higher about 35–42% than that in CR-NIL from 1 to 9 dpi (Figure 1D). It could be observed that patterns of fructose and glucose accumulations in the roots of both CS and CR-NILs are similar.

Sucrose contents showed no changes in leaf or root tissues between un-inoculated CS and CR NILs (Figures 1E,F). After P. brassicae infection, lower sucrose contents were detected at the early infection time points (2 dpi) in the leaves of the CS-NIL than that of the CR-NIL. Thereafter, significant increases of sucrose contents were observed in the leaves of both CS and CR NILs, but no significant differences between two NILs. In the root tissues, sucrose contents showed an increase tendency in inoculated and un-inoculated CR-NILs at the late infection time points (from 4 to 9 dpi). However, much less increases of sucrose contents were detected in the roots of the CS-NIL than that of the CR-NIL from 2 dpi.

In summary, glucose and fructose levels reduced in the leaves of CS-NIL, but increased in the roots after P. brassicae infection. Also, sucrose content of CS-NIL was lower than CR-NIL in roots. These results imply that P. brassicae infection may activate translocation of sugar from the leaves to roots.

To analyze the possible sugar transporter involved in long distance transport, we firstly identified SWEET genes in B. rapa genome. Based on the similarities with the 17 Arabidopsis AtSWEETs, totally 32 SWEETs were identified in B. rapa, and named BrSWEET1a to BrSWEET17b on the base of their identity to AtSWEETs (Supplementary Table S1). 28 BrSWEET genes are multicopy either duplication or triplication of corresponding AtSWEET, except for four single copy genes (BrSWEET8, -9, -10, and -13). In addition, the orthologous gene of AtSWEET6 was not found in the B. rapa genomes. Majority of BrSWEET genes were about 1500 base pairs (bp), and amino acids residues are from 178 to 316. Genes longer than 7.5 kb were also found in BrSWEET15b and BrSWEET17b. BrSWEETs contained 3–7 TM helices (Supplementary Figure S1) harboring one or two conservative MtN3/saliva domains (Supplementary Figure S2). The isoelectric points (pI) value of each BrSWEETs was higher than 7.50. Among them, pI of 19 BrSWEETs was higher than 9.0, and others were ranged between 7.59 and 8.97. Noticeably, no single SWEET gene was observed to be closely associated with the CRb locus (Zhang et al., 2014) on the B. rapa genome.

Further, to better understand the evolutional relationships of BrSWEETs, amino acid sequences of the 117 SWEET proteins (17, 32, 68 from A. thaliana, B. rapa, and B. napus, respectively) were aligned and an unrooted phylogenetic tree was constructed (Figure 2). All 32 BrSWEET family members contain two conserved regions, and they might have evolved to a closer relationship during evolution (Supplementary Figure S2). Seven sister pairs of genes were identified in the phylogenetic trees with very strong bootstrap support (100%) (Figure 2). Most of the gene pairs had short branch lengths, suggesting their recent divergence (Figure 2).

All of 32 BrSWEET genes, together with other 85 members were grouped into 4 major clades, namely clade I, II, III, and IV (Figure 2). Gene families in clade III had relative higher molecular weight than others (Supplementary Table S1). All BrSWEET and BnSWEET were clustered closely with their AtSWEET orthologs.

According to the gene loci information, all BrSWEET genes were assigned to the 9 chromosomes of B. rapa with the exception for BrSWEET13 (Figure 3 and Supplementary Table S1). Duplication event of AtSWEET genes was found in B. rapa genomes. 8 AtSWEET genes (AtSWEET1, -2, -3, -4, -7, -12, -16, and -17) were duplicated and 4 genes (AtSWEET5, -11, -14, and -15) are triplicated in B. rapa. Most of BrSWEET paralogous genes were dispersed to different chromosomes. However, tandem array of 2 BrSWEET5 and 2 BrSWEET11 was found on the chromosome A02 and A06, respectively. Other 3 BrSWEET genes (BrSWEET14, -16 and -17) were found to be involved in the duplication of chromosomal fragments.

Exon-intron organizations for each BrSWEET gene were analyzed by alignment of full-length cDNA sequences to the genomic DNA sequences of B. rapa. Characterization of the gene structure diversity revealed that most BrSWEET genes contained 6 exons (Figure 4). Five BrSWEET genes (BrSWEET3b, -7a, -7b, -14c, and -15c) appeared to have exon-intron loss variations, probably due to the fact that loss/gain of one or two intron/exon were happened during the evolution of B. rapa. Most genes had similar exon lengths, while the length of intron varied obviously for each gene. Among which, BrSWEET11c, -15b, -17a, and -17b, had extremely long intron.

The common domains for cytosolic localization, apoplastic localization, and transmembrane anchorage were found in all BrSWEET genes (Supplementary Figure S1), indicating that BrSWEET genes are composited with membrane spanning, intracellular and extracellular regions, and responsible for sugar transport. Most of BrSWEETs had 2 conservative MtN3/saliva domains with 7 TM helices (Supplementary Figures S1, S2), the typical domain of SWEET in eukaryote. However, genes (BrSWEET3b, 7b, and 15c) with 3 or 4 TM helices were found to have only one complete MtN3/saliva domain.

To investigate the structural diversity further, all BrSWEET genes were submitted to MEME program for the conserved motifs structure analysis. As the results, 10 conserved motifs were identified, among which 6 motifs (motifs 1–6) were annotated as MtN3/saliva domains of the SWEET genes (Supplementary Figure S3). All genes contained motif 1, 2, 5, and 6 in the C terminal of BrSWEETs However, the order of motif arranged in the N terminal was different. Motif 1-4/5-6-7 was found in clade I, II, and IV, but motif 7-2-3-5-6 in clade III. Additionally, loss of some motifs was observed in a few genes, including motif 1-2-5 in the C terminal of BrSWEET3b and BrSWEET7b, and motif 3-5-6 in the N terminal of BrSWEET15c. Gene families in clade I and II contained a specific motif 4 except for 3 genes in clade II. Motifs 8 and 10 were specific to clade III. The variation of the structural motifs among BrSWEET family suggested functional diversity of the SWEET genes.

Two conservative structures of MtN3/saliva domain with TM1 (motif1/2/3), TM2 (motif4/5) and TM3 (motif6), and TM5 (motif1), TM6 (motif2/5) and TM7 (motif6), respectively, were present in all genes except for BrSWEET3b, BrSWEET77b, and BrSWEET15c. These 2 conservative structure domains were consisted of about 85 amino acids, and were located in the similar position (Supplementary Table S1 and Supplementary Figure S3). TM4 is structured with motif 7 or motif 8, and it plays the role of contact between two MtN3/saliva domains.

To investigate expression patterns of BrSWEET genes in B. rapa, semi-quantitative RT-PCR analysis was carried out in leaf, hypocotyl and root tissues of uninfected plants of CS-NIL. The semi qRT-PCR data revealed that the transcripts were detected for the majority of BrSWEET genes among all tissues (Figure 5). Half of the genes were expressed in all tissues. Transcript of four genes, including three duplicated genes (BrSWEET3b, -5a and -5b) and one single copy genes (BrSWEET8) were undetectable in any given tissues tested. High transcript abundance of six genes (BrSWEET7b, -10, -14a, -14b, -15a, and -15c) were expressed in the roots and hypocotyls, and two genes (BrSWEET3a and BrSWEET7a) were predominantly expressed in the roots and leaves. Several genes were expressed in only one kind of tissue, including BrSWEET4b in leaves, BrSWEET5c and BrSWEET9 in hypocotyls, and BrSWEET16b in roots. The expression profiling of BrSWEET genes suggests their possible involvements in broad biological functions during vegetative growth and development of B. rapa.

To study the expressions of BrSWEET gene responsive to the P. brassicae infection, the expression patterns were determined among 3 different tissues (leaf, hypocotyl, and root) at different time points post P. brassicae inoculation by qRT-PCR. The heatmap results revealed that the expression patterns among BrSWEET genes were variable depending on the tissues and treatments (Figures 6, 7 and Supplementary Figure S4).

BrSWEETs expressed differentially in the roots of CS and CR NILs after infection of P. brassicae (Figure 6). In comparison with un-inoculated plants, among all BrSWEET genes 14 (BrSWEET1a, -1b, -2a, -3b, -7a, -7b, -11a, -14a, -14b, -15b, -15c, -16a, -16b, and -17a) and 15 (BrSWEET1a, -2a, -3a, -4a, -10, -11a, -11c, -12a, -12b, -14a, -14b, -14c, -15c, -16a, and -17b) were up-regulated in CS-NIL and CR-NIL respectively. Transcripts of other genes were undetectable or down-regulated during P. brassicae infection. BrSWEET3b, -4b, -5b, -5c, -8, and -9 were not expressed in un-inoculated plants, but were induced expression in the P. brassicae-infected roots of both CS and CR NILs. Among them, BrSWEET3b, -8 and -9 were strongly induced in CS-NIL and BrSWEET4b, -5b, and -5c were significantly up-regulated in CR-NIL. By comparing between CS and CR NILs after P. brassicae infection, expression levels of 15 genes (BrSWEET1a, -1b, -2a, -3b, -7a, -7b, -8, -9, -11a, -14b, -15b, -15c, -16a, -16b, and -17a) were higher in the CS-NIL, but expression levels of other 16 genes were relatively higher in the CR-NIL. Most notably, the expression of BrSWEET1a, -2a, -9, -11a, and -15c was strongly up-regulated (over 1000-folds) in the roots of CS-NIL, and the expression of BrSWEET4a, -5c, -12a, and -12b was significantly up-regulated in the roots of CR-NIL (Figure 6). Clustering and heatmap analysis further revealed that among up-regulated BrSWEET genes in CS-NIL, several genes (BrSWEET7a, -7b -8, -9, -15b, and -15c) showed transiently induced expression at the early infection time points (0.5–1.5 dpi), whereas induced expressions of other genes (BrSWEET1a, -1b, -2a, -3b, -11a, -16b, and -17a) maintained up to the late infection time points at 6–9 dpi (Figure 6).

In the hypocotyls, expressions of BrSWEET3a, -3b, -4b, -5b, -7a, -8, and -16b were strikingly induced on both CR and CS-NILs by the infection of P. brassicae. Eleven genes (BrSWEET4a, -5b, -8, -10, -11a, -11b, -12b, -14c, -15a, -17a, and -17b) exhibited down-regulation, and other 20 genes were up-regulated in the hypocotyls of CS and CR NILs after infection of P. brassicae. Remarkably, 2 SWEET genes (BrSWEET9 and -16a) were found to be significantly up-regulated specifically in hypocotyls of CS-NIL after infection of P. brassicae (Supplementary Figure S4).

In leaf tissues, 9 BrSWEET genes (BrSWEET3b, -5a, -5b, -5c, -7b, -14a, -14b, -15a, and -16b) showed undetectable levels of expression on both CR and CS NILs before or after P. brassicae infection, while 4 genes (BrSWEET8, -9, -10, and -15c) showed enhanced expressions on both CR and CS NILs upon P. brassicae infection. Among these four P. brassicae-induced genes, BrSWEET9 had higher expression levels in CR-NILs than CS-NILs, whereas BrSWEET8, BrSWEET10 and BrSWEET15c showed similar expression patterns between CR and CS NILs. Most of the expressed genes either in CR or CS NILs showed differential regulations at different infection time points. Six SWEET genes (BrSWEET2b, -14c, -15b, -15c, -16a, and -17a) were found to be the most significantly up-regulated in the leaves of CS-NIL (Figure 7).

AtSWEET11 in Arabidopsis mediates sucrose efflux in phloem parenchyma cells, promoting sucrose loading into the phloem by SUC2 for long distance transport (Chen et al., 2012). Our results revealed that P. brassicae-infected plants showed reduced sucrose contents in the leaves and roots, while increased glucose and fructose contents in the roots of B. rapa. Moreover, increased expression of BrSWEET11a gene occurred in root tissues upon P. brassicae infection. These results suggest that SWEET11 gene might be involved in host sugar transport into diseased clubroot tissues. To investigate possible involvement of SWEET11 during P. brassicae pathogenesis, Arabidopsis sweet11 mutant (sweet11) and its corresponding wild type Col-0 line were inoculated with P. brassicae. Disease development and gall formation were examined at 3, 4, and 5 weeks post inoculation (wpi) with P. brassicae resting spores. At 3 and 4 wpi, sweet11 mutants showed significantly lower disease incidence than that of the wild-type plants (Figure 8). After 5 wpi, all plants including mutant and the wild-type produced disease symptoms with various levels of disease severities. The disease index (DI) value of sweet11 mutants was significantly lower than that of the wild-type at all time points post inoculation (Figure 8C and Supplementary Table S3). Furthermore, we detected the relative pathogen DNA quantities of both wild type and sweet11 mutants after inoculation. The results showed that the P. brassicae content was continuously increased in both sweet11 mutants and Col-0 of Arabidopsis from 3 to 5 wpi (Figure 8D). Meanwhile, the DNA content of P. brassicae in the sweet11 mutants was significantly lower than wild type Col-0 at each time point (Figure 8D and Supplementary Table S3). All the results indicated that the progression of clubroot development in the sweet11 mutant was significantly delayed when compared to that of the wild-type plants. Thus, Arabidopsis SWEET11 gene is suggested to be responsive to the P. brassicae infection, and contributes to host susceptibility to the disease.

The number of SWEET genes is largely variable among green plant species. For example, unicellular and green algae have only 1–3 copies of SWEET genes, but 18–23 genes in monocots, and 15–68 genes in dicots (Chong et al., 2014; Patil et al., 2015; Jian et al., 2016). Arabidopsis, the closely related species to Brassica genus, contains 17 SWEET genes (Kühn et al., 2010). In consistence with the genome triplication of Brassica species since its divergence from Arabidopsis (Wang et al., 2011), much higher numbers of SWEET genes were identified in B. rapa. These suggest that there may be functional redundancy or divarication between the respective SWEET members. However, the occurrence of gene loss during polyploid speciation was also found in the B. rapa genome corresponding to 5 Arabidopsis AtSWEET genes, i.e., the AtSWETT6 homolog was absent in B. rapa genome and AtSWETT8, AtSWETT9, AtSWETT10, as well as AtSWETT13 have only one respective orthologous gene in the B. rapa genome. The expansion or loss of some SWEET genes was also found in potato (Manck-Götzenberger and Requena, 2016), tomato (Feng et al., 2015), rice (Yuan and Wang, 2013) and B. napus (Jian et al., 2016). The expansion of some SWEET family genes of the Brassica genome suggests their possible functional differentiations in response to the environmental conditions.

In order to better understand the role of SWEET gene family during colonization of B. rapa by P. brassicae, we firstly performed a gene identification in genomic level. The phylogenetic study revealed that the BrSWEET family genes were classified into 4 clades, an agreed feature defined in other higher plants as well (Kühn et al., 2010; Patil et al., 2015). Proteins from clade III were highly conserved among Arabidopsis and Brassica species, as shown by the strong support of branches separating the three protein groups, whereas in the other clades the identification of orthologous genes between species was more difficult. In this research, 32 BrSWEET genes, most of the closely related genes in family exhibit similar motif compositions, and these motifs were arranged in the same order in the C terminal of all BrSWEETs. It suggested that there are functional similarities in the SWEET family genes. However, the order of motif arranged in the N terminal was different. Motif 1-4/5-6-7 was found in clade I, II, and IV, but motif 7-2-3-5-6 in clade III. Previous studies of sugar transport properties showed that the SWEET genes in Clade I, II, and IV were monosaccharide (glucose, fructose, galactose, and so on) transporters, the Clade III SWEET genes were sucrose (disaccharide) transporter (Chen et al., 2010, 2012; Lin et al., 2014). So, we speculated that the different motifs of N terminal MtN3 domain between Clade I, II, IV, and Clade III result in the specificity of the SWEET genes in transporting different sugars. In other word, the variation of the structural motifs among BrSWEET family suggested functional diversity of the SWEET genes.

Most of the characterized or predicted SWEET genes encoded membrane proteins with 7 TM helices harboring 2 MtN3/saliva domains (Chen et al., 2010). In consistency with this, the majority of SWEET genes identified in B. rapa contained 7 TM helices. In spite of the variation of structural motifs, the two MtN3/saliva domains structured with TM1-3 and TM5-7 are conserved among BrSWEETs. However, protein sequences of SWEET genes with 4 (BrSWEET7b) or 6 (BrSWEET2b, -4b, -8, -14c, and -17b) TM helices distributed in the 2 MtN3/saliva domains were found. Interestingly, protein sequences of BrSWEET3b and BrSWEET15c with respective 3 and 4 TM helices were also found, which were previously only reported in bacteria and archaea (Wang et al., 2014). Expression profiling revealed that BrSWEET genes containing less than 7 TM helices were expressed in various tissues, indicating that they are possible functional genes although their substrate molecules and specific functions remains to be further studied.

In this study, a total of 28 paralogous BrSWEET genes corresponding to 15 AtSWEETs were found in B. rapa. Expression profiling revealed that the paralogous genes of 6 BrSWEET (BrSWEET1, -2, -11, -12, -13, -17) gene families transcribed in all tested three tissues: roots, hypocotyls, and leaves, implying their functional redundancy and important role in vegetative growth stages. All clade III members of SWEETs were highly expressed in flowers, leaves and seeds, but not in roots in B. napus (Jian et al., 2016). In our result, all clade III BrSWEETs were transcribed in roots except BrSWEET9. Additionally, four genes (BrSWEET4a, -14c, -15b, and -16a) were also constantly expressed in all tissues, whereas their corresponding paralogs showed different expression patterns, indicating a possible gene function divergence which is commonly found in Brassica species (Chen et al., 2012; Jian et al., 2016). The remaining paralogous genes were either expressed in a tissue-specific manner or not expressed. For example, among 3 paralogous genes of BrSWEET5, BrSWEET5a and 5b were not expressed in all tested tissues, except BrSWEET5c expressed in the hypocotyls. However, the induction of BrSWEET5b and BrSWEET5c were observed in the roots and hypocotyls of both CR and CS NILs after P. brassicae infection. Similarly, ortholog of SWEET5b in potato was also induced in response to fungal infection, but not expressed in control tissues (Manck-Götzenberger and Requena, 2016). BrSWEET3b, -4b, -7a, -15c, and -16b were also P. brassicae-inducible, while showed different expression profiling to their paralogs. Coincident with BrSWEET3b and -7a, induction for SWEET3b and SWEET7a orthologs in rice and potato was also observed in response to mycorrhiza colonizations (Fiorilli et al., 2015; Manck-Götzenberger and Requena, 2016). These findings further suggested that the different orthoglous BrSWEET genes may play the similar roles for sugar transport in response to the biotic interactions during plant evolutionary process.

Several studies revealed that pathogens competed for sugars with infected host cells by hijacking host sugar efflux systems for disease development, being accompanied by highly induced expression of sugar transporter genes (Siemens et al., 2006; Doidy et al., 2012; Lemoine et al., 2013). P. brassicae also needs sugars from hosts for completing its life cycle, thereby inducing gall formation as an additional sink. Thus, regulation of those sugar transporters might determine the consequence of the plant-pathogen interactions.

To investigate the sugar redistribution during the B. rapa–P. brassicae interaction, we firstly detected the content of three soluble sugars in both leaves and roots of CS and CR NILs. In higher plants, sucrose is the main form of long distance transportation of carbohydrates from the source to the sink. In A. thaliana, it was found that sucrose was accumulated in uninfected leaves, but not in the leaves of P. brassicae-infected plants (Evans and Scholes, 1995), that is consisted with our results. Additionally, we also detected a decrease of sucrose together with an increase of glucose and fructose in clubroot galls in accordance with previous metabolic analysis in galls of both A. thaliana and Brassica species during development of clubroot disease (Brodmann et al., 2002; Mitchell and Rice, 2010), as well, contents of glucose and fructose in leaves of CS-NILs dramatically decreased. All these suggested that sucrose produced in P. brassicae-infected plants was exported from the leaves, probably rapidly metabolized, and further transferred into the clubroot pathogens.

Some SWEET genes showed no expression in normal growth/developmental conditions, but were transcriptionally induced by P. brassicae infection, including six genes (BrSWEET3b, -4b, -5b, -5c, -8, and -9) in the roots, seven genes (BrSWEET3a, -3b, -4b, -5b, -7a, -8, and -16b) in the hypocotyls, and three genes (BrSWEET8, -9, and -15c) in the leaves. Furthermore, inducted expression of several BrSWEET genes was at much higher levels in the CS-NIL than that of CR-NIL upon P. brassicae infection either in leaves (BrSWEET2b, -14c, -15b, -15c, -16a, -17a), hypocotyls (BrSWEET9, -16a) or roots (BrSWEET1a, -2a, -9, -11a, -15c). It is expected that these genes are responsible for transporting sugars from the leaf to the sink root tissues associated with P. brassicae colonization. Some other studies have demonstrated that increasing of SWEET expressions is to facilitate sugar transports to the non-plant sinks created by the phytopathogen infection or sugar secretions into the soil as carbon supply for pathogen growth (Patrick, 1989; Chen et al., 2010). Most of those genes belong to Clade III SWEETs which function as sucrose transporters responsible for long distance sugar transportation (Chen et al., 2010, 2012). Several studies have reported the involvement of Clade III members in various plant pathogenic systems, special SWEET11 orthologous genes in canola (B. napus) (Jian et al., 2016) and rice (Chen et al., 2010; Cernadas et al., 2014; Chandran, 2015). In our study, Arabidopsis sweet 11 mutants showed significantly lower disease index value and the P. brassicae DNA content compared to that of wild-type plants after P. brassicae inoculation, revealing that the BrSWEET11 played a crucial role during clubroot disease development. Hence, the enhanced sucrose transport from the leaf source tissues to the clubroot galls might be the main carbon transfer pathway in clubroot disease establishment, and the Clade III SWEETs play important roles in this progress. Whereas, notably some other BrSWEET genes were highly expressed in CR-NIL. Whether enhanced expressions of these genes contribute to host defense responses or are required to maintain carbon supplies for plant growth and development remains to be further investigated.

Competition for sugars at the plant–microbe interface could be controlled by the SWEET family transporters, and modulation of these transporter activities may determine the consequence (susceptibility or resistance) of the interaction (Toyofuku et al., 2000). However, this study did not examine the possible involvement of SWEET genes as potential susceptibility factors to the P. brassicae pathotypes. Presumably, SWEET genes of P. brassicae may also participate in competing for sugars in a manner of coevolving between host varieties and pathogen specializations (pathotypes), which is worthy of further study.