2023 年 73 巻 5 号 p. 421-434
Fusarium wilt is a significant disease in radish, but the genetic mechanisms controlling yellows resistance (YR) are not well understood. This study aimed to identify YR-QTLs and to fine-map one of them using F2:3 populations developed from resistant and susceptible radish parents. In this study, two high-density genetic maps each containing shared co-dominant markers and either female or male dominant markers that spanned 988.6 and 1127.5 cM with average marker densities of 1.40 and 1.53 cM, respectively, were generated using Genotyping by Random Amplicon Sequencing-Direct (GRAS-Di) technology. We identified two YR-QTLs on chromosome R2 and R7, and designated the latter as ForRs1 as the major QTL. Fine mapping narrowed down the ForRs1 locus to a 195 kb region. Among the 16 predicted genes in the delimited region, 4 genes including two receptor-like protein and -kinase genes (RLP/RLK) were identified as prime candidates for ForRs1 based on the nucleotide sequence comparisons between the parents and their predicted functions. This study is the first to use a GRAS-Di for genetic map construction of cruciferous crops and fine map the YR-QTL on the R7 chromosome of radish. These findings will provide groundbreaking insights into radish YR breeding and understanding the genetics of YR mechanism.
Radish (Raphanus sativus L., 2n = 2x = 18), a member of the family of Brassicaceae, is an important root vegetable that is very popular worldwide. It is consumed in a variety of ways such as root vegetable, salad, leafy vegetable, etc., and it is an important source of minerals and vitamins in the human diet. Radish production is seriously threatened by Fusarium wilt (FW), a disease caused by the soil-borne fungus Fusarium oxysporum f. sp. raphani (FOR). The pathogen attacks radishes by penetrating through mechanical damage on the root surface, extends through the vascular tissues, and spreads throughout the internal tissues, thereby causing stunted growth, wilting and yellowing of leaves, which finally results in the plant’s death (Garibaldi et al. 2006, Gordon 2017). Radish yellows was first discovered in California, the United States, in 1934 (Kendrick and Snyder 1936), and then spread to Wisconsin (Pound 1953), and Washington (Du Toit and Pelter 2003). In Japan, it was discovered in Wakayama Prefecture in 1952, and then spread throughout Japan (Hida and Ashizawa 1985). Radish has a large production area worldwide, which makes it difficult to control FW with agrochemicals (Hida and Ashizawa 1985), and conventional farming methods are not very effective because the spores remain in the soil for a long period of time without a host (Gordon 2017). Therefore, the best method for its control is the use of Fusarium yellows resistance (YR) cultivars (Ma et al. 2021, Pu et al. 2012).
Over the past decade, Quantitative Trait Loci (QTLs) and fine mapping analyses had been conducted in crops of the family Brassicaceae for identification of YR genes and elucidating resistance mechanisms to FW. In QTL and segregation studies on Brassica oleracea, Pu et al. (2012) identified a YR-QTL, FocBo1, as a single dominant gene that confers resistance to FW caused by F. oxysporum f. sp. conglutinans, and was mapped to the linkage group, C7. Lv et al. (2013) identified a FOC1 gene on cabbage chromosome C6 as the YR gene to F. oxysporum f. sp. conglutinans, based on phenotypic assay of 160 double haploid (DH) lines, derived from the F1 of a cross between the resistant and susceptible cabbage parents. In B. rapa, Bra012688 and Bra012689 genes were identified as candidate genes for YR in Chinese cabbage (B. rapa var. pekinensis) by Shimizu et al. (2014) through differential gene expression analysis, using RNA sequencing. Thereafter, in B. oleracea, the YR gene was map-based cloned as an orthologous gene of Bra012688 (Lv et al. 2014, Shimizu et al. 2015).
In radish, several studies have identified YR-QTLs in different regions of the genome. Kaneko et al. (2007) discovered a YR-QTL on R7, which was subsequently identified as corresponding to the long arm of chromosome 1 in Arabidopsis thaliana by Shirasawa et al. (2011). Yu et al. (2013) identified 8 QTLs (qFW1-qFW8) responsible for YR, and were distributed on 5 LGs (2, 3, 4, 6 and 7). Of these loci, 3 (qFW3, qFW4 and qFW8) were constantly detected in three independent tests over 2 years. Since qFW4, mapped on LG3 (chromosome R5), accounted for most of the phenotypic variations (14.63%), the locus was recognized as a major QTL and designated as Fusarium wilt Resistance locus 1 (Fwr1) (Yu et al. 2013). In the later study, Yu et al. (2020) fine-mapped the Fwr1 to a region of 139.8 kb between 2 DNA markers FM82 and FM87 and identified ORF4 as a YR candidate gene that encodes a serine/arginine-rich protein kinase. In another QTL analysis of radish, Ma et al. (2021) detected 2 YR-QTLs, FoRsR7.1 and FoRsR9.3 on R7 and R9 chromosomes, respectively, and further argued five putative resistance-related genes in the region of FoRsR7.1. Lee et al. (2021) reported that the genome-wide association study (GWAS) analysis using 225 radish accessions identified 44 single nucleotide polymorphisms (SNPs) and twenty candidate genes associated with YR. They also did QTL analysis using an F2 population derived from a YR line and a susceptible line, and consequently, four QTLs were identified, one of which collocated with the SNP on R7 detected in the GWAS study.
Genetic maps constructed to identify YR-QTLs have been laborious so far, but recent progress of the next generation sequencing (NGS) technology has made it possible to generate linkage maps quickly and efficiently. Genotyping by Random Amplicon Sequencing-Direct (GRAS-Di), one of the Genotyping-by-Sequencing technologies, is a recently developed technique that uses random primers to amplify amplicons to be sequenced (Enoki 2019, Enoki and Takeuchi 2018, Miki et al. 2020). This method can be applied to thousands of samples by using primer sets containing different index sequences (Miki et al. 2020). Use of GRAS-Di maps has been applied in several cultivars because of the high reproducibility of this method and the very low percentage of missing values (Fekih et al. 2023, Miki et al. 2020, Moriya et al. 2021, Takeshima et al. 2022, Umeda et al. 2021). However, there are no reports of using this method for linkage mapping in cruciferous crops.
As mentioned above, one of the major YR-QTLs in radishes have been identified on the R7 chromosome (Kaneko et al. 2007, Lee et al. 2021, Ma et al. 2021), and several candidate genes have been proposed based on the co-localization of the identified QTLs. However, it is not known whether the QTLs detected in these different research groups are same or different from each other because of the lack of detailed molecular information of each locus, such as sequence information of the DNA markers that were used to define the loci. Regarding this, in this study, we tried to identify the QTLs responsible for YR in radish using GRAS-Di and candidate genes in ForRs1, one of the 2 QTLs detected, then the results were compared to those of previous studies to establish solid basis for YR breeding in radish.
F1 plants were produced from a cross between the parents, YR RK15-1 and susceptible AKM radish inbred lines (Supplemental Fig. 1). The parental inbred lines were supplied by Nanto Seed Co., Ltd., Nara, Japan. F1 seeds were planted and later self-pollinated. The resulting F2 was divided into two groups, one for a comparative test of two inoculation methods, pipetting and direct seed sowing methods, and the other for GRAS-Di map construction and QTL analysis. The 132 F2 plants were subjected to GRAS-Di analysis (see “Linkage map construction using GRAS-Di technology” for details) and further self-pollinated to produce the F2:3 (F3) populations. Among them, 126 and 130 F2:3 populations were used for QTL analysis conducted in 2021 and 2022, respectively. Ten seedlings of F3 per F2:3 population were used at each of the two periods of the inoculation tests that were done in the greenhouse. Okute-Sakurajima (JP 27228), the cultivar from which a radish reference genome sequence that was mainly used in this study, RSAskr_r1.0, was established (Shirasawa et al. 2020), was provided from the National Agriculture and Food Research Organization/NARO genebank, Japan to check its susceptibility to FOR.
Inoculation testAll inoculation tests were performed using FOR MAFF731043 that was provided by NARO genebank, Japan. The potato sucrose broth (PSB) medium for fungus maintenance and inocula preparation was prepared using the protocol of Pu et al. (2012). Two methods of inoculation tests were employed in this study, the pipetting and the direct seed sowing methods (Supplemental Fig. 1). For the pipetting method, the experiments were conducted twice by Nanto Seed Co., Ltd., in Nara prefecture, Japan, in mid-June of 2021 and 2022. The inocula (2 × 105 cfu) was added to each gram of soil, after transplanting ten-day old seedlings into plastic pots, and the plants were grown in the greenhouse where the minimum temperatures were 19.1°C and 16.0°C; the maximum temperatures were 32.9°C and 33.0°C; the average temperatures were 25.0°C and 23.5°C during the tests in 2021 and 2022, respectively. For the direct seed sowing method, the method followed the protocol of Pu et al. (2012), in which inocula was made by mixing healthy soil with diseased soil, in the ratio of 9:1, respectively. Thereafter, seeds were directly sown in the diseased soil, and the plants were grown in the growth chamber with temperatures controlled to a maximum range of 28°C with 16 h of light and 8 h of dark conditions. The phenotyping for QTL analysis was conducted using 10 F3 plants for each F2:3 population. Since this investigation required a large-scale experiment, the pipetting method was taken because it is less labor intensive than the direct seeding method. Inoculation trials of the parents and another set of the F2 plants were conducted using both methods to determine whether the results of the two methods were comparable to each other. In all the inoculation tests, RK15-1 was used as resistant control while AKM was used as susceptible control.
Disease severity index was taken at 28 days after inoculation (DAI) for all lines. The disease index (DI) ranged from 0 to 3, DI = 0, no disease symptom; DI = 1, showing slight yellowing of leaves; DI = 2, showing deep yellowing and loss of leaves/atrophy; and DI = 3, plant death (Fig. 1). Plants showing DI of 0 and average DI of 1 and below were classified as resistant, while others showing above 1 were classified as susceptible. When DI was calculated in phenotyping of F2:3 lines regarding QTL analysis, percentage of DI (PoDI) was estimated as follows, DI = Σ (DI × X) × 100/N × 3, where X = number of plants with symptom in each line, N = total number of seedlings. For the other inoculation test, the mean DIs (DI) were calculated in each experimental plot.
Inoculation test of fusarium yellows. Disease index score: the disease index (DI) ranged from 0 to 3; DI = 0 no disease symptom; DI = 1 showing slight yellowing of leaves; DI = 2 showing deep yellowing and loss of leaves/atrophy; and DI = 3 plant death. 0 ≤ DI ≤ 1.00 and 1.00 < DI ≤ 3.00 were classified as fusarium yellows resistance and susceptibility, respectively.
For the construction of the linkage map using GRAS-Di, genomic DNA samples were extracted from the parents and the 132 F2 plants using NucleoSpin® Plant II Kit (Takara Bio, Japan). After that, the first PCR was performed using arbitrary random primers containing Nextera adapter sequences and 3-base random oligomers that randomly bind to genomic DNA. Details of the PCR procedure can be found in Hosoya et al. (2019). The second PCR was conducted to add an Illumina multiplexed 8-base dual index to each amplicon pool. As a result, one sequencing library, including amplicons from 132 F2 plants and two replicates of each parent, were constructed and subsequently sequenced on the HiSeq 4000 platform (Illumina, USA) with 150-bp paired-end mode. Markers were identified with GRAS-Di software v 1.0.5 (Toyota, Aichi, Japan). The GRAS-Di procedure was conducted by Genebay Co., Ltd. (Kanagawa, Japan).
Sequencing the bulked GRAS-Di library using HiSeq 4000 technology generated 776 million reads, equivalent to 116 Gb of data, or 835 Mb per sample. Subsequent analysis using the GRAS-Di software resulted in a total of 3827 dominant markers, of which 31, 14, 1264, 2424, and 94 markers were classified into quality categories A, B, C, D, and E, respectively. Of these markers, those not classified as E (3733 markers) were further filtered based on the large skewness from the 3:1 ratio. The GRAS-Di software algorithm provides co-dominant polymorphisms between parents, as well as dominant polymorphisms specific to each parent. Since either type can be used for linkage-map construction, we constructed three genetic maps using these polymorphisms as markers: a map using only the segregation information of co-dominant markers, and two maps using both co-dominant markers and dominant markers from either the male or the female parent.
To generate F2 linkage maps, we used the Antmap ver.1.2 (Iwata and Ninomiya 2006) and excluded co-dominant and dominant markers with skewed segregation ratios at X2 test (P < 0.01 and P ≤ 0.05, respectively). For the construction of linkage groups, we employed the “all combinations” method and the “recombination” criterion, using a recombination threshold of 0.3 for the co-dominant and male-derived marker maps and 0.27 for the female-derived marker map. The parameters for locus ordering were “Kosambi map function”, “ordering SARF”, and “targeted groups all”.
QTL detection and validationQTL analysis was performed by composite interval mapping (CIM, Zeng 1994) using the Windows QTL cartographer v.2.5 (Wang et al. 2012) by combining the genotype (GRAS-Di) and phenotype (F2:3 inoculation test) data. CIM analysis was done using the Zmapqtl standard model 6, with a window size of 10 cM and a walking speed of 1 cM. To estimate a genome wide LOD threshold score for a QTL at 95% confidence level (P = 0.05), a 1000-permutation test was performed by shuffling the phenotypic means with the genotype (Doerge and Churchill 1996).
For validation of the two detected YR-QTLs on R2 and R7 chromosomes (designated as ForRs2 and ForRs1), we selected four F2 plants (F2-13, F2-80, F2-50, and F2-95) with heterozygous genotypes at one QTL but homozygous at the other QTL. To determine the genotypes at the two QTLs, we referred to the genotypes of the flanking GRAS-Di marker information, AMP0000754 and AMP0003886 for ForRs1 and ForRs2, respectively. Their F3 seeds ranging from 50 to 100 of each of the four genotypes were sown in the diseased soil for phenotyping (direct seed sowing inoculation method). At the same time, using seedling cotyledons or leaves, their genotypes were analyzed using the CAPS markers, AMP0000754-CAPS and AMP0003886-CAPS (Supplemental Table 1). The phenotypic data was analyzed using the Tukey-Kramer test of Excel Toukei 2010 for Windows (Social Survey Research Information Co., Ltd., Tokyo, Japan).
Whole genome re-sequencing of the parents and genotypingThe parents were re-sequenced as follows. Their genomic DNAs were extracted using the NucleoSpin Plant II Kit (Takara Bio, Japan), then the DNA was sequenced by Genebay Co., Ltd. (Kanagawa, Japan) using the HiSeq 4000 device. The obtained reads were mapped to the radish reference genome sequence RSAskr_r1.0 (Shirasawa et al. 2020) by BWA (Li and Durbin 2010), and the nucleotide polymorphisms between the parent lines extracted from BAM files were identified using GATK (McKenna et al. 2010). The nucleotide polymorphisms between the parents were also visually identified using IGV software (Broad Institute). Based on the detected polymorphisms, CAPS and InDel markers located on the ForRs1 region were designed for fine mapping (ForRs1-specific DNA markers; designated as #15-37, Supplemental Table 1).
For genotyping, one cm diameter pieces of fully expanded leaves or cotyledons were collected and frozen with liquid nitrogen for DNA extraction. The CTAB (Cetyl Trimethyl Ammonium Bromide) protocol, according to Murray and Thompson (1980) with modifications according to Sato et al. (2019), was used for extracting genomic DNA. The PCR genotyping was conducted as follows using DNA markers designed from the detected DNA polymorphisms between parents. The PCR mixtures were composed of 0.5 μl of plant DNA, 2.5 μl of EmeraldAmp PCR Master Mix (Takara Bio Inc., Japan), 0.2 μl of a mixture of forward and reverse primers (10 pmol each), and 1.8 μl of distilled water. The mixture was incubated at 94°C for 5 minutes, followed by 40 cycles of 94°C for 30 seconds (sec), 55°C–60°C for 30 sec, and 72°C for 30 sec and one cycle of 72°C for 5 min. For CAPS markers, the PCR products were digested with the appropriate restriction enzymes (Supplemental Table 1) before electrophoresis. Gel electrophoresis was conducted using either 1–2% agarose gel or 13% acrylamide gel to visualize the DNA fragments under a UV transilluminator. Gelstar solution (Takara Bio Inc., Japan) at a 10,000-fold dilution was used to stain acrylamide gels.
Fine mapping of ForRs1 using F2:3 populationsFrom 132 F2 plants, 7 plants (F2-15, 28, 46, 52, 61, 69, and 68) having recombination breakpoints within the ForRs1 region were selected based on genotypes of two GRAS-Di markers, AMP0000754 and AMP0010720, which flank the ForRs1 on R7 chromosome. The selected F2 plants were self-pollinated to produce F3 seeds. Next, 15 to 81 F3 plants of each F2:3 population were inoculated with FOR for a period of 28 days. At approximately 10 days after germination, each F3 plant was genotyped using ForRs1-specific DNA markers and only those having homozygous recombinant chromosomes were used for the subsequent fine-mapping study. To compare our identified ForRs1 region with the location of the candidate genes identified by Ma et al. (2021), the reference genome of R. sativus cv. WK10039, Rs2.0, was used (Cho et al. 2022).
Nanopore and RNA-sequencingNanopore sequencing of the YR parent, RK15-1 was performed to obtain precise genome sequence information of the ForRs1 region of the resistant accession, as follows. DNA of RK15-1 was isolated using NucleoBond HMW DNA isolation kit (Takara Bio Inc., Japan), then sequencing adapters were ligated using Ligation Sequencing kit LSK110 according to the 1D protocol from Oxford Nanopore. The resulting library was then sequenced on MinION R9.4 flow cells (Oxford Nanopore Technologies, U.K.). The generated nanopore long reads were assembled into contigs using NECAT (Chen et al. 2021) and then were polished with RK15-1 Illumina short reads using NextPolish (Hu et al. 2020).
RNA-seq was also performed for both parents. The seedlings of each parent grown on the diseased and healthy soils were sampled at two different time points (8 and 11 days after sowing), then immediately immersed into liquid nitrogen and stored at –80°C until use. From each of the 8 samples, total RNA was extracted using RNeasy Mini Kit and RNA-Free DNase Kit according to manufacturer’s protocol (QIAGEN, Netherlands). For RNA-seq, RNAs of three seedlings at the same time point were pooled for each genotype. The resulting 8 total RNA samples were sent to Genebay Co., Ltd. (Kanagawa, Japan). The extracted RNA was purified using Sera-mag Magnetic Oligo (dT) Beads (Thermo Fisher Scientific, USA). The poly(A) selected RNA samples were used to prepare cDNA libraries and eight cDNA libraries were sequenced using Illumina HiSeq 4000 platform generating 150-bp paired-end reads. All RNA-seq reads from RK15-1 were pooled and mapped to the obtained nanopore contigs of RK15-1 using HISAT2 (v2.0.4), and subsequently analyzed by StringTie (v 1.2.3) (Pertea et al. 2016) to annotate transcripts and estimate their expression levels from the mapping data. Next, we also determined expression levels of the transcripts in each of the 8 samples using HISAT2 and StringTie. BLASTP searches were conducted using the predicted amino acid sequences of the annotated transcripts as queries against 2 databases, the TAIR 10 Proteins database from The Arabidopsis Information Resource (TAIR) (https://www.arabidopsis.org/index.jsp) and the radish reference genome RSAskr_r1.0 (https://plantgarden.jp/ja/index).
NGS sequence data have been submitted to the DDBJ database (https://www.ddbj.nig.ac.jp) under the accession number DRA016541 and DRA016542 for DNA-seq and RNA-seq, respectively.
The two inbred lines, RK15-1 and AKM, along with their F1 hybrid and subsequent F2 population, were examined for susceptibility to FOR (Supplemental Fig. 2). RK15-1 showed DIs of 0.50 ± 0.25 and 0.50 ± 0.22 using the pipetting and direct seed sowing methods, respectively, indicating resistance to FOR. AKM, on the other hand, showed DIs of 2.70 ± 0.25 and 3.00 ± 0.00 using the pipetting and direct seed sowing methods, respectively, indicating susceptibility to FOR. The F1 plants were not stably resistant (DI = 1.18 ± 0.20) using the direct seed sowing method, suggesting the partial dominance of YR in this cross combination. The F2 plants showed bimodal frequency distribution of disease index in both methods. Additionally, cv. Okute-Sakurajima, the cultivar from which the reference genome sequence RSAskr_r1.0 was established (Shirasawa et al. 2020), was found to be susceptible to FOR (Average DI = 3.0 using 18 plants). Okute-Sakurajima exhibited wilt symptoms and the immediate death occurring 7 to 8 DAI. Conversely, AKM seedlings were moderately susceptible, displaying yellowing 13–16 DAI, followed by wilt and death (data not shown).
Linkage map constructionGRAS-Di was conducted using 132 F2 plants between the parents. The obtained GRAS-Di markers, of which quality ranked from A to D, were further filtered to eliminate those exhibiting skewness from the Mendelian segregation within the F2. As a result, 3030 markers were classified into three categories: 363 co-dominant markers, 1345 dominant markers derived from RK15-1, and 1322 dominant markers derived from AKM. Using co-dominant markers, a map spanning 843.9 cM, with an average marker density of 2.32 cM was obtained (Table 1, Supplemental Fig. 3). The longest LG was 132 cM corresponding to R5, and the shortest LG was 63.3 cM corresponding to R3 and R7. The highest number of markers (59) was anchored on R2, followed by R5 (58), and the lowest number of markers (24) was identified on R7. The consensus linkage groups of the map were confirmed by aligning the GRAS-Di amplicon (Supplemental Table 2) to the reference genome of R. sativus (RSAskr_r1.0). Hereafter, the map will be referred to as “co-dominant derived map” in this article.
Linkage group data for the F2 population resulting from the crossing of RK15-1 and AKM. Length in centimorgan and number of markers are indicated at the top and bottom of each row for each linkage group, respectively.
| Maps | Linkage group | Total | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| R1 | R2 | R3 | R4 | R5 | R6 | R7 | R8 | R9 | ||
| Co-dominant map | 74.6 | 100.7 | 63.3 | 123.4 | 132.0 | 114.4 | 63.3 | 64.8 | 107.4 | 843.9 |
| (32) | (59) | (30) | (30) | (58) | (46) | (24) | (41) | (43) | (363) | |
| RK15-1-derived map + co-dominant marker | 95.5 | 120.9 | 90.2 | 138.1 | 145.3 | 130.4 | 80.8 | 77.7 | 109.7 | 988.6 |
| (52) | (96) | (71) | (87) | (100) | (100) | (57) | (79) | (63) | (705) | |
| AKM-derived map + co-dominant marker | 80.0 | 139.1 | 158.3 | 132.4 | 135.6 | 130.6 | 117.0 | 100.4 | 134.1 | 1127.5 |
| (57) | (99) | (100) | (81) | (84) | (94) | (61) | (85) | (77) | (738) | |
Although the co-dominant derived map has sufficient marker density for QTL analysis, to construct a map with higher density of markers, we attempted to use the information of dominant markers together with the co-dominant markers. Since combining RK15-1 (female)- and AKM (male)-derived markers into one map resulted in distorted marker order, such as large gaps and incorrect assignment of markers, possibly due to limitations of the mapping algorithm (Jansen 2009, Knapp et al. 1995, Tan and Fu 2007, Tan et al. 2017), the segregation data of RK15-1- and AKM-derived markers were separately combined with that of co-dominant markers. After the reduction of the numbers of the dominant markers that were mapped at the same position, the resulting RK15-1- and AKM-derived maps spanned 988.6 and 1127.5 cM, with average marker densities of 1.40 and 1.53 cM, respectively (Table 1, Supplemental Fig. 4). Hereafter, the 2 maps will be referred to as “RK15-1-derived map” and “AKM-derived map” in this article, respectively. The AKM-derived map was longer than the other two maps due to the increase of the length of the specific linkage groups such as the R3 and R7. The comparison between the three maps using co-dominant markers as anchor markers revealed good collinearity among them, although the order of some markers was reversed between the genetic maps (Supplemental Fig. 5). It is remarkable that AKM- and RK15-derived maps effectively positioned multiple dominant markers between two co-dominant markers, specifically, in the ForRs1 region on the R7 chromosome, flanked by the co-dominant markers AMP0000754 and AMP0010720. This is particularly valuable for selecting recombinants with distinct chromosomal recombination points within this region for map-based cloning purposes.
QTL analysis and confirmation of QTL effectNext, we generated F2:3 populations derived from the 132 F2 plants that were used in the GRAS-Di analysis. Among them, 126 and 130 F2:3 populations were used for inoculation tests conducted in mid-June 2021 and 2022 for QTL analysis. This population showed a continuous distribution of PoDI with a slight susceptibility shift between the two inoculation tests (Fig. 2). The distribution was skewed to higher PoDI in 2022 than in 2021, suggesting that symptoms were more severe in 2022 than in 2021.
Frequency distribution of the percentage of disease severity index (PoDI) in the F2:3 populations. The results of the inoculation tests conducted in 2021 (left) and 2022 (right) were shown, respectively. The pipetting method was employed in these analyses.
QTL analyses using the phenotype data of 2021 and co-dominant marker map detected two QTLs at R2 and R7 chromosomes, with LOD values of 4.7 and 7.3, respectively (Fig. 3). The QTL on R7 chromosome was repetitively detected when the phenotype data of 2022 was used, with LOD value of 11.6 (Fig. 3). This QTL on R7 accounted for 18.1 and 33.0 % of the phenotypic variation in 2021 and 2022, respectively (Table 2). Negative values for the additive effect of the two QTLs indicate that the RK15-1 allele in each of the R2 and R7 QTLs confers resistance to FOR. The major (R7) and minor (R2) QTLs were named ForRs1 and ForRs2, respectively. Additional minor QTL on R5, together with the two QTLs, was detected in the 2021 trial when the RK15-1- and AKM-derived maps were used (Supplemental Fig. 6).
QTL profiles for YR derived in the F2:3 populations. Pink and gray lines in each graph represent data of 2021 and 2022, respectively. (A) LOD plots for YR along the whole chromosomes are shown. (B) LOD plots of R7 that harbors ForRs1 are indicated. On the bottom of each graph, names of chromosomes (left) and GRAS-Di markers (right) are indicated. The horizontal lines in those graphs represent the threshold LOD values (3.6).
Summary of QTLs detected for yellows resistance to Fusarium oxysporum f. sp. raphani
| Name | Linkage Group | Marker interval | 1 LOD confidence interval (cM) | LOD | R2 (%)a | Additive effectb | Dominance effect | |
|---|---|---|---|---|---|---|---|---|
| ForRs2 | R2 | 1st Test | AMP0010176–AMP0013639 | 2.54–15.88 | 4.7 | 11.5 | –8.9 | –6.2 |
| ForRs1 | R7 | 1st Test | AMP0000754–AMP0009342 | 41.1–54.8 | 7.3 | 18.1 | –12.1 | –2.7 |
| 2nd Test | AMP0000754–AMP0010720 | 41.1–50.7 | 11.6 | 33.0 | –17.5 | –4.2 |
a Proportion of the phenotypic variance explained by each QTL.
b Additive effects of RK15-1 allele.
To confirm the associations between 2 QTLs and YR, we chose four F2 plants (F2-13, F2-80, F2-95, and F2-50) with heterozygous genotypes at either one of the two QTLs, while homozygous at the other QTL. Two DNA markers, AMP0000754-CAPS and AMP0003886-CAPS, designed from GRAS-Di markers and RSAskr_r1.0 genome sequence that were found to be closely linked to each ForRs1 and ForRs2, respectively, were used to predict the genotypes at the two QTLs (Supplemental Table 1). As a shorthand notation, parental allele genotypes at the ForRs1 were assigned A and a for the resistance (RK15-1) and susceptibility (AKM) alleles, respectively. Similarly, parental alleles of ForRs2 locus were assigned B (RK15-1) and b (AKM). Based on the analysis, the genotypes of F2-13, F2-80, F2-50, and F2-95 were identified as Aabb, AaBB, aaBb and AABb, respectively. These four F2 plants were then self-pollinated to produce segregating F3 for subsequent analysis. Different patterns of segregation in YR phenotypes were observed in the 4 F3 populations (Fig. 4). For F2:3-13 (Aabb) and F2:3-80 (AaBB) progenies segregating at the ForRs1 locus, most of the plants with homozygous RK15-1 allele (AA) were resistant, but the plants with AKM homozygous allele (aa) were susceptible, regardless of the genotypes of ForRs2 (Fig. 4). The susceptibility of the plants with heterozygous allele (Aa) in F2:3-13 and F2:3-80 fluctuated, and the respective average disease indexes (DIs) were 1.41 and 1.32 (Fig. 4). This result indicated that the RK15-1 allele of ForRs1 (A) conferred higher resistance in a dominant or partial dominant manner, while the AKM allele (a) showed almost complete susceptibility. No contribution of the AKM allele of ForRs1 (a) to YR was also supported by the observation of the F2:3-50 line (aaBb), in which all progenies showed susceptibility regardless of the genotypes at the ForRs2 locus (B/b). Contribution of RK15-1-derived allele (B) at ForRs2 to YR was observed in the F2:3-95 line (AABb), in which all plants were AA at ForRs1. These suggested epistatic interaction between the 2 loci.
Relationship between YR phenotypes and genotypes at ForRs1 and ForRs2. The values of DI of the F3 of the 4 investigated F2:3 populations, in which genotype segregation at either ForRs1 or ForRs2 locus occurs, are represented as box plots. At the top of each graph, the name of the F2:3 population and the F2 genotype are indicated. The genotypes at the segregating locus and the numbers of the investigated plants are indicated at the bottom of each graph. *, ** and NS indicate statistically significant at P ≤ 0.05, 0.01 and not significant, respectively.
F2 plants with chromosomal recombination within the ForRs1 region (recombinants) were identified among the 132 F2 plants using the ForRs1 flanking GRAS-Di markers, AMP0000754 and AMP0010720. As a result, 18 F2 plants having recombinants at the locus were found. Among them, 7 F2 recombinants (F2-15, 28, 46, 52, 61, 69, and 68) with distinct recombination points were used for the fine-mapping. The selected F2 recombinants had a non-recombinant (parental) and recombinant chromosome pair at ForRs1 and were self-pollinated to produce the F3 progenies (Supplemental Fig. 7). Exceptionally, the F2-61 plant contained two recombinant chromosomes derived from both male and female gametes of the F1. Some of the F3 populations showed low variance in DIs, while others showed segregation (Supplemental Fig. 7). Simultaneously, genotypes in the ForRs1 region of the F3 plants were determined using ForRs1-specific DNA markers designed from the re-sequenced data of the parents (#15-37, Fig. 5A, Supplemental Table 1). Based on the genotype data, we selected F3 plants having homozygous genotypes at all investigated markers for fine mapping, because they showed greater phenotypic stability of YR compared to those having heterozygous genotypes (Supplemental Fig. 7). The instability could be explained by the partial dominance of the resistant (RK15-1) to susceptible (AKM) allele that was inferred from the unstable resistant phenotype observed in the F1 (Supplemental Fig. 2). The resulting graphical genotypes of each F2:3 line, together with their DI, suggested that the ForRs1 region could be narrowed down to the region between #23 and #31 markers (Fig. 5B).
Fine mapping of ForRs1 locus. (A) ForRs1 overlaps FoRsR7.1. The schematic figure shows comparison between 3 alleles of the ForRs1 locus of RK15-1 (contig-bctg00000035, this study), Okute-Sakurajima RSAskr_r1.0 (Shirasawa et al. 2020), and WK10039 (Rs2.0, Cho et al. 2022). Each of the 3 lines represents each allele. Positions and names of the DNA markers used in this study are indicated on RSAskr_r1.0, and the corresponding positions on the other 2 alleles were indicated with green dotted lines extending from them. Positions and the names of the genes associated with FoRsR7.1, that were identified by Ma et al. (2021), are indicated on Rs2.0. When the allelic sequences of the FoRsR7.1 associated genes exist on RK15-1, their positions are indicated with dotted lines of blue and black extending from them as indicated on RS2.0. The genes highlighted in black and blue indicate the marker genes flanking FoRsR7.1 and the candidate genes predicted by Ma et al. (2021), respectively. Please note that RK15-1 is resistant and Okute-Sakurajima is susceptible to FOR (this study), while YR phenotype of WK10039 is unknown. (B) Fine-mapped ForRs1 region. Graphical genotypes and disease index (DI) of the selected recombinants in the F2:3 progenies are shown. Marker names are indicated at the top of the row. Homozygotes of the RK15-1-derived resistant allele and homozygotes of the AKM-derived susceptible allele are represented by orange and blue boxes, respectively. Regions of unknown genotypes due to marker intervals are shown as gray boxes. Name of lines and DIs are shown on the left and right side, respectively. The fine-mapped ForRs1 region is indicated with a bi-directional line with arrow at the bottom. (C) Genes annotated in the fine-mapped ForRs1 region of RK15-1. The genes in the region between #23 and #31 markers, with an estimated length of 195 kb, were represented as boxes.
As described earlier, Okute-Sakurajima, the cultivar used to establish the radish reference sequence RSAskr_r1.0 (Shirasawa et al. 2020) which was mainly used in this study, was susceptible to FW. Since it was not clear whether the reference genome of the susceptible accession was useful for identification of YR candidate genes in the ForRs1 region of RK15-1, we conducted long-read whole genome sequencing of RK15-1 using Nanopore technology to obtain the genome sequence information of the resistant accession. From the analysis, we obtained 666 contigs with the maximum and minimum contig sizes of 12.9 Mbp and 0.53 Mbp, respectively (Supplemental Table 3). The total assembly size of 489 Mbp was roughly equal to the estimated radish genome size of 504.5 Mb (Shirasawa et al. 2020). The completeness of the assembly evaluated using BUSCO (Manni et al. 2021) was 99.2%, high enough to be used for the downstream analyses. From the contigs, contig-bctg00000035 (4.5 Mb) was identified to contain both sequences of AMP0000754 and AMP0010720 that flank ForRs1. The contig showed high level of sequence collinearity with the allelic region of RSAskr_r1.0 (Fig. 5A), again suggesting accuracy of the sequence assembly. Based on the contig-bctg00000035, the length of the fine-mapped ForRs1 region (between #23 and #31 markers) was estimated as 195 kb (Fig. 5B, 5C).
Previously, Lee et al. (2021) conducted QTL analysis using an F2 population derived from a YR line and a susceptible line and reported that the qFWR3 detected on R7 was co-located with the SNP located in Rs404770 gene, which encodes the glycosyltransferase family protein. The blastn search using Rs404770 sequence as a query revealed that this gene is not located in the ForRs1 region of this study that is sandwiched by AMP0000754 and AMP0010720. Ma et al. (2021) detected a YR-QTL, FoRsR7.1 on the chromosome R7, which was defined by a gene Rs382960 and a DNA marker BRPGM1176 on a radish reference genome sequence Rs1.0. They identified 5 R-genes in this region, including 3 candidate genes for YR. Therefore, we tried to know relative positions of ForRs1 and FoRsR7.1. Since the sequence of BRPGM1176 marker was not available, we used Rs379750, a gene close to BRPGM1176, to determine the FoRsR7.1 region. Fig. 5A showed high level of collinearity between the sequences of Rs2.0 (newer version of Rs1.0, Cho et al. 2022), RSAskr_r1.0 (Okute-Sakurajima), and contig-bctg00000035 (RK15-1) at ForRs1 region. This comparison revealed that the fine-mapped ForRs1 region was completely included in the FoRsR7.1 region (between Rs382960 and Rs379750). However, 2 out of the 3 candidate genes in FoRsR7.1, Rs382940 and Rs382950 were outside of the fine-mapped ForRs1 region. Furthermore, the other candidate gene in FoRsR7.1, Rs382200, was not present in the contig-bctg00000035 of RK15-1. These findings suggest that YR genes in the ForRs1 region of this study could be different from any of the 3 candidate genes in FoRsR7.1 (Ma et al. 2021), even though the two regions are completely overlapped.
In the fine-mapped ForRs1 region on contig-bctg00000035 of RK15-1, 16 genes were annotated based on the RNA-seq data and included catalytic enzymes, receptor-like proteins, a transmembrane protein of unknown function, and various proteins belonging to different families or with specific roles such as transcription factors and ribosomal proteins (Fig. 5C, Supplemental Table 4). We excluded 8 genes (RK15.25438, RK15.25439, RK15.25443, RK15.25445, RK15.25446, RK15.25447, RK15.25449, RK15.25450) from possible candidate genes of ForRs1 because they showed no sequence variation of the coding regions between the parents, and they were stably expressed in both RK15-1 and AKM (Supplemental Table 5, Supplemental Fig. 8).
Among the remaining eight candidate genes (Table 3), in AKM, the RK15.25441 mRNA, annotated as a RING/U-box superfamily protein (E3 ligases), exhibits an abnormally truncated structure (Supplemental Fig. 9) compared to that of RK15-1. This suggests that while the gene mutation in the susceptible AKM parent may be linked to YR-susceptibility, the CDS sequence of the susceptible Okute-Sakurajima gene (RSAskr1.0R7g74443) with the same annotation perfectly matches RK15-1 CDS. This indicates that the RK15.25436 gene may be excluded from the list of candidates. Compared to RK15-1, two AKM esterase genes (RK15.25436, RK15.25437) and one AKM casein lytic proteinase gene (RK15.25440) expressed truncated mRNAs. These genes lack R-gene specific motifs, suggesting that their role as hydrolytic enzymes makes it unlikely for them to function as resistance genes.
Annotation, parent polymorphism, and expression level in the candidate genes located in the delimited region of ForRs1
| Genes | Annotation | Mutations found in AKM referred to the amino acid sequences of RK15-1 | TPMa | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Structure of transcript | Premature stop codon | No. of substituted amino acid residues | No. of deleted/inserted amino acid residues | RK15-1 | AKM | ||||
| CONT | INF | CONT | INF | ||||||
| RK15.25435 | Receptor like protein 15 | Normal | Yes (by frameshift at F507fs) | 19 (prior to the pre-mature stop codon)b | 2 (W85_A86insAR) 1 (E398_S399insI) |
0.65 | 1.59 | 0.09 | 0.77 |
| RK15.25436 | Encodes a member of the glycerophosphodiester phosphodiesterase (GDPD) family | Abnormalc | No | 14 (L7S T16N G33S F92A T93I N98D S102N N180S G193S K221R T240I A290T E315Q QS371T) | 8 (N238_L239insLIDSPKVL) | 19.37 | 25.46 | 9.34 | 15.43 |
| RK15.25437 | Encodes a member of the glycerophosphodiester phosphodiesterase (GDPD) family | Abnormal | No | 8 (V20G S118F M120L Y149F P191S A238D V269M H386L) | 0 | 4.39 | 3.15 | 4.03 | 4.79 |
| RK15.25440 | Encodes ClpB1, which belongs to the casein lytic proteinase/heat shock protein 100 (Clp/Hsp100) family | Normal | No | 1 (Q59A)d | 1 (E58_Q59insT) | 4.66 | 4.86 | 5.54 | 9.79 |
| RK15.25441 | RING/U-box superfamily protein | Abnormal | No | 4 (P28A S108N N134S Q142E) | 12 (F13_K14insKSMGEDTVIEPI) 1 (V137del) 1 (D147del) |
3.26 | 3.75 | 0.43 | 0.12 |
| RK15.25442 | Tetratricopeptide repeat (TPR)-like superfamily protein | Normal | No | 1 (S10F) | 0 | 3.83 | 4.00 | 4.16 | 2.71 |
| RK15.25444 | Encodes a putative transcription factor (MYB95) | Normal | No | 1 (N131S)d | 0 | 0.29 | 0.15 | 1.91 | 1.15 |
| RK15.25448 | Protein kinase superfamily protein. PBL29, PBS1-LIKE 29 | Normal | No | 2 (E30Q L116P) | 0 | 0.77 | 0.75 | 0.13 | 0.00 |
a TPMs (Transcripts per million) were obtained in the RNA-seq analysis using RNAs collected at the 11 days after sowing from three seedlings of each parent grown on the diseased (INF) and healthy (CONT) soils.
b Substituted amino acid residues are indicated in Supplemental Fig. 12.
c Visible structural abnormality observed using IGV, as exemplified in Supplemental Fig. 9.
d Similarity percentage of the proteins between RK15-1 and AKM is 100%.
The remaining 4 genes encode proteins that were assumed to be related to pathogen resistance. RK15.25435 (RLP) possesses R-protein-like motifs such as leucine-rich repeats (LRRs), where many SNPs were found between RK15-1 and the two susceptible lines, AKM and Okute-Sakurajima. Consequently, premature stop codons were generated in the coding sequences (CDSs) of the two susceptible lines (Supplemental Fig. 10, Table 3). RK15.25448 encodes a receptor-like kinase (RLK) which possesses a leucine-rich repeats (LRRs) and a kinase domain. A nonsynonymous substitution (E>Q) was found in the exon 1 of this gene between RK15 and the two susceptible lines. Additionally, another nonsynonymous substitution (L>P) was found in exon 2 of AKM allele. RK15.25442 encodes a tetratricopeptide repeat (TPR)-like superfamily protein, and RK15.25444 encodes a putative transcription factor MYB95 which contains a Myb DNA-binding domain. Those protein families are known to regulate various physiological processes including pathogen resistance. In these two genes, nonsynonymous substitutions were found between RK15-1 and the two susceptible lines (Supplemental Fig. 10). As a result, these four genes remained as the best candidates for ForRs1.
The GRAS-Di software algorithm provides co-dominant polymorphisms between parents, of which the number is limited, whereas it provides thousands of dominant markers (Enoki 2019, Enoki and Takeuchi 2018, Miki et al. 2020). Thus, use of dominant markers to map construction is important for GRAS-Di technology. However, it is reported that estimating recombination frequencies between dominant markers is least efficient in F2 progeny, particularly in repulsion dominant markers (Jansen 2009, Knapp et al. 1995, Tan and Fu 2007, Tan et al. 2017). To overcome this challenge, Miki et al. (2020) and Fekih et al. (2023) mapped the GRAS-Di amplicon sequences to the reference genome sequence to revise irregular marker orders, and thereafter, the recombination frequencies of the dominant markers were estimated by R/qtl software. As a result, such a reference genome mapping-based genotyping method successfully generated high-quality reference genomes. Alternatively, to address the challenge posed by the unsuitability of repulsion dominant markers for linkage maps, Knapp et al. (1995) suggested that dominant markers can be split into two groups and mapped, one with dominant alleles from the male and one with dominant alleles from the female, thereby creating two pure-coupling F2 maps. Therefore, we created separate maps, and consequently, successfully created a set of high-density GRAS-Di maps by separately including RK15-1 (female)-derived and AKM (male)-derived markers in co-dominant markers. Those maps effectively positioned multiple dominant markers between two co-dominant markers in the ForRs1 region on the R7 chromosome, which was very useful for selecting recombinants with distinct recombination points in the target region. It is thought that the use of GRAS-Di would be promoted by effectively using dominant markers. In fact, in the F1 pseudo-testcross method applied to heterozygous perennials, where the use of uniparental dominant markers is preferred over co-dominant markers, high-density male and female linkage maps using GRAS-Di markers have been facilitated in sugarcane (Umeda et al. 2021) and apple (Moriya et al. 2021).
Comparison of the previously identified YR-QTLs in radishSeveral studies have discovered YR-QTLs in different genomic regions of radish. However, the absence of shared markers among researchers makes it difficult to confirm the identical nature of the identified QTLs. To address this issue, we used the consensus linkage group (Shirasawa and Kitashiba 2017). Regarding the minor YR-QTLs, in the study conducted by Ma et al. (2021), two minor YR-QTLs, namely FoRs9.1 and FoRs9.2, were identified on R9, appearing only once across three studies. Yu et al. (2013) reported the discovery of eight YR-QTLs (qFW1-8) during three trials spanning a two-year period. Among these, a major YR-QTL was identified on LG3 (R5), while the remaining seven QTLs, qFW6 on LG4 (R4), qFW3 on LG3 (R5), qFW1 and qFW2 on LG2 (R6), qFW5 and qFW8 on LG7 (R8), and qFW7 on LG6 (R9), were classified as relatively minor YR-QTLs based on the R2 values. In another study by Lee et al. (2021), a total of four YR-QTLs were identified, with three QTLs on R3, R6, and R8 being categorized as minor YR-QTLs. Interestingly, the syntenic analysis revealed a correspondence between FoRs9.1 and qFW7 on LG6 (R9) (Ma et al. 2021). Thus, the reported minor YR-QTLs distributed in the chromosomes spanning R3 to R9 indicated that our study could discover a novel minor YR-QTL, ForRs2, located on R2.
Regarding the major YR-QTLs, Kaneko et al. (2007) found a YR-QTL on LG1 (R7), corresponding to the long arm of chromosome 1 in A. thaliana. Yu et al. (2013) discovered qFW4 on LG3 (R5), which explained the majority of the phenotypic variation. Ma et al. (2021) identified two major YR-QTLs, FoRsR7.1 and FoRsR9.3, on R7 and R9 chromosomes, respectively. Lee et al. (2021) performed a QTL analysis and associated one of the identified QTLs with an SNP on R7 from their genome-wide association study (GWAS). Thus, major YR-QTLs conferring high resistance to FOR have been detected in R5, R7, and R9, with R7 being repeatedly identified by different investigators. Comparison of the location of the major YR-QTLs is also difficult due to the lack of common markers among the investigators. Even so, in this study, we showed that the region of ForRs1 overlapped with FoRsR7.1 of Ma et al. (2021), suggesting the possibility that the region could be one of the major loci for YR in radish. However, since the candidate genes identified by the 2 groups differ from each other, together with the allelic structural diversity at the locus that was observed as the presence-absence variation of a gene (Rs382200) between alleles, it is not clear if the YR genes in ForRs1 and FoRsR7.1 are the same or different. Plant disease resistance loci often show diversity in allelic structure resulting from rapid genomic changes such as deletions, duplications, and gene losses/exchanges, which may have contributed to the rapid evolution of the disease resistance genes (Kuang et al. 2005, Read et al. 2020). Therefore, despite the availability of several radish reference genome sequences, if they were derived from susceptible accessions such as Okute-Sakurajima of RSAskr_r1.0, it is unknown whether their information is useful to molecularly determine the resistance gene. This is because the susceptible allele of the resistance gene cannot be annotated when it has accumulated nonsense mutations or when the gene sequence is completely lost in the susceptible allele. Regarding this, we believe that it is quite important to have a long contig sequence of the genomic region of interest for the resistant accessions/varieties which would be used in specific mapping study. In this study, we used Nanopore sequencing technology to obtain the long genome sequenced of the ForRs1 region in the YR parent, RK15-1.
Identification of the candidate genes controlling YRIn this study, ForRs1 region fine-mapping and RNA-seq analysis was performed, resulting in the identification of 16 transcripts. By utilizing the RNA-seq data and nucleotide sequences of each potential gene, eight genes that exhibited no sequence variation between YR RK15-1 and susceptible AKM were excluded from consideration as candidate genes. Among the remaining eight candidate genes listed in Table 3, after excluding the genes expressing hydrolase and ligase enzymes, four R-gene analogue genes with significant non-synonymous SNPs between RK15-1 and AKM emerged as the most promising candidates.
RK15.25448 encodes an RLK domain similar to the A. thaliana resistance gene PBS1-Like 29 (AT1G74490). PBS1-Like proteins are known to associate with RPS5 (RESISTANCE TO PSEUDOMONAS SYRINGAE 5) and confer resistance against certain strains of Pseudomonas syringae in A. thaliana (Pottinger and Innes 2020). The PBS1-Like genes belong to the Arabidopsis receptor-like cytoplasmic kinase (RLCK) subfamily VII, which consists of 46 members. Among the PBS1-Like orthologous genes in the radish genome, it is necessary to investigate whether RK15.25448, the ortholog of PBS1-Like 29, exhibits specificity in recognizing F. oxysporum f. sp. raphani. Another candidate resistance gene, RK15.25435, encodes an RLP (receptor-like protein) with a leucine-rich repeat (LRR) motif. Interestingly, the YR radish candidate gene (Fwr1) identified on chromosome R5 (Yu et al. 2020) is also an RLK. Notably, albeit B. oleracea and B. rapa are closely related to radish, their TIR-NBS-LRR genes FocBo1 and FocBr1, which are not RLP/RLK genes, confer Fusarium wilt resistance, respectively (Shimizu et al. 2014, 2015).
The gene RK15.25438, which encodes a protein interacting with phytosulfokine, was not considered a candidate gene due to the identical coding sequence (CDS) in RK15-1 and AKM, as well as their nearly identical mRNA expression levels. Despite its exclusion, RK15.25438 possesses an LRR domain similar to PSKR1, which is involved in plant immunity and interacts with a danger-associated molecular pattern, phytosulfokine (Hu et al. 2023). The alignment of the amino acid sequences of the LRR domains of PSKR1 and RFO2 (encoded by the fusarium resistance gene RFO2 in A. thaliana) showed a high similarity, and RFO2 acts as a decoy receptor for PSKR1 (Shen and Diener 2013). The alignment of amino acid sequences, including RK15.25438, revealed a high similarity among these three gene products (Supplemental Fig. 11). Thus, the RK15.25438 gene might have an undisclosed role in disease resistance reactions. As just mentioned, AKM has the functional allele of this gene, but Okute-Sakurajima (FW susceptibility) does not because of the deletion of the region containing this gene, which was confirmed by the sequence analysis of the ForRs1 region of Okute-Sakurajima. We identified three RLP/RLK genes in the ForRs1 locus, including this gene. Duplicated R-genes are reported to commonly co-localize with disease resistance QTL. Previous studies have shown that clusters of NBS-LRR genes can provide effective resistance against rice blast (Ashikawa et al. 2008) and Peronospora parasitica in Arabidopsis (Sinapidou et al. 2004). Furthermore, it has been found that the rice CC-NLR pair RGA4 and RGA5, derived from the tandemly repeated genes, form a heterodimer in response to the Avr-Pia effector of Magnaporthe oryzae (Cesari et al. 2014).
We also identified RK15.25442 (tetratricopeptide repeat (TPR)-like superfamily protein) and RK15.25444 (putative transcription factor MYB95) in the delimited region. The former protein has a protein-protein interaction domain by which the protein can interact with and regulate the activity of key immune proteins, such as resistance (R) proteins. In fact, MoChia1, a chitinase secreted by rice pathogen M. oryzae, suppresses plant immune response by binding chitin. Rice protein OsTPR1 interacts with MoChia1, allowing chitin accumulation and re-establishing immunity, demonstrating how plants counteract fungal chitinase (Yang et al. 2019). Regarding MYB transcription factor, in sorghum infected by the fungal pathogen Colletotrichum sublineolum, accumulation of 3-deoxyanthocyanidin phytoalexins at the site of primary infection is regulated by a MYB transcription factor (Ibraheem et al. 2015). The RK15.25442 and RK15.25444 genes exhibited polymorphism between RK15-1 and AKM. Therefore, it is necessary to further examine whether RK15.25442 and RK15.25444 genes qualify as candidate genes.
The present study pioneers the utilization of a GRAS-Di map for constructing the genetic map of cruciferous crops and achieving a detailed mapping of the YR-QTL on the R7 chromosome of Japanese radish. As a result, we successfully narrowed down the ForRs1 locus to a 195 kb region and identified four genes including RLP/RLK genes as best candidate genes. These findings will provide groundbreaking insights into radish YR breeding and understanding the genetic mechanism of YR. Further studies are necessary to determine the responsible genes for YR at the ForRs1 locus.
C.S.A.E., J.S., E.F. and K.O. conceived and designed the research. C.S.A.E., S.K., M.K. and K.O. conducted the experiments and prepared samples. T.K., M.S., M.N. and N.M. assisted in NGS library preparation and sequencing. C.S.A.E., T.K., M.S, M.N. and K.O. analyzed the data. C.S.A.E., C.O.E., E.F. and K.O. wrote the manuscript. E.F. and K.O. supervised the experiments. K.O. contributed by providing funding and also managed the whole project. All authors contributed to the development of this manuscript.
The authors sincerely thank NARO Genebank, Japan, for kindly providing a radish cv. Okute-Sakurajima (JP27228) and a strain of F. oxysporum f. sp. raphani (MAFF731043). This research was partially supported by the Honjo International scholarship foundation (Honjo scholarship) for C.S.A. Ezeah. This work was partially supported by JST Adaptable and Seamless Technology transfer Program through Target-driven R&D (A-STEP) Grant Number JPMJTM20DG to K. Okazaki.
