A unique landrace of kimchi cabbage, jombaechoo (B. rapa), and an accession from collections of the landrace gaetmoo (R. sativus), were used as parents to develop a new, stable baemoochae line. They were supplied by The Center of Germplasm, The Rural Development Administration in Korea with the introduction no. of PI204163 for B. rapa and IT216097 for R. sativus. They were known to be maintained high content of disulfide constituent of sulforaphene in an analysis. Jombaechoo is much similar to radish in the leaf morphology, an enlarged terminal lobe and smaller lateral lobes. Gaetmoo is a typical wild type in the root shape with numerous thin roots from the slender and little bit small root. Jombaechoo and gaetmoo were self-pollinated two and three times, respectively, to produce more uniform lines. Bud pollination was performed on jombaechoo pistils. Anthers from jombaechoo were removed 1 day before being pollinated by gaetmoo pollen [11] . Ten-day-old immature hybrid ovules were cultured in darkness at 25˚C on modified B₅ medium (Duchefa Co.) [12] supplemented with 600 mg/L CaCl₂2H₂O, 800 mg/L glutamine, 100 mg/L L-serine, 0.1 mg/L NAA, 0.1 mg/L BAP, 0.1 mg/L 2,4-D, and 30 g/L sucrose. The ovule-cul- ture derived plants were transplanted to soil for acclimatization. A 0.3% colchicine solution was applied to the growing points of established plants for 3 days, using absorbent cotton, to double chromosomes. The size and wave edge of leaves were different to both parents and flowers were white in color, as if the inter-generic hybrid. The F₁ plants capable of producing pollen were self-polli- nated and cross-pollinated reciprocally to BB#1. Three to five fresh flowers and approximately 10 buds on the same flower branch were self-pollinated, and non-pollinated buds were removed. This procedure was repeated on several branches to acquire sufficient seed. Cross pollination to BB#1 was managed in the same way as the self pollination with flowers. The plant 07OV-127 was subjected to microspore mutagenesis. Liquid NLN 13 medium [13] , supplemented with 0.1 mg/L BA, was added to the microspore culture. A stock solution of n-nitroso n-methyl urethane (NMU) was added to the culture solution containing a microspore density of 10⁴ per mL to a final concentration of 0.01 μM. Embryos derived from microspores were transplanted to MS₂ medium supplemented with 2% sucrose to MS medium for plant regeneration [14] . Normal plants with well-developed shoots and roots were transplanted to pots for acclimatization. Abnormal plants were cut into pieces of about 5.0 ´ 5.0 mm and sub-cultured in MSk medium (1.0 mg/L BA, 0.2 mg/L NAA and 2% sucrose added to basic MS medium) to induce normal plants [14] . Acclimated plants were vernalized in a low-temperature greenhouse, in which the temperature was about 2˚C to 7˚C in the night time and 10˚C to 12˚C in the day time for 25 days, without colchicine treatment for chromosome doubling, followed by exposure to 12˚C - 30˚C for seed production.

Data collected from pollinations in earlier generations were abbreviated for shortening the volume of the manuscript and stated in the text for understand the flow of the story.

Meiotic chromosome behavior of BB#5 from prophase I to anaphase II was observed as described by [15] [16] , with minor modifications. An unstable line, BB#4, and the stable line BB#1 were compared. Chromosome abnormalities were scored and analyzed in 500 pollen mother cells (PMCs) using a fluorescence microscope (Olympus BX53, Japan) equipped with a CCD camera (Leica DFC365 FS, Germany).

Genomic in situ hybridization (GISH) analysis, using blocking DNA of R. sativus and probe DNA of B. rapa, was applied to BB#5 to determine its meiotic chromosomal constitution [17] . The CTAB-extraction method was used to extract and purify total genomic DNA from young leaves of B. rapa and R. sativus [18] . Genomic DNA from B. rapa was labeled with dig-11-dUTP (Roche, Germany) by nick translation and used as a probe, while the unlabeled DNA of R. sativus was fragmented at 100˚C for 7 min and used as a block. During GISH analysis, the slides were pretreated with RNase A buffer (100 µg・ml ⁻¹ RNase A in 2´ SSC) at 37˚C for 1 h and then treated with 100 µg/ml pepsin (Sigma, USA) in 0.01 M HCl at 37˚C for 10 min. After washing for 5 min with 2´ SSC, the slides were post-fixed with 4% paraformaldehyde in 2´ SSC, washed with 2´ SSC, and dehydrated in an ethanol series (70, 90, and 100%) for 2 min each. A hybridization mixture (50% formamide, 10% dextransulfate, 2´ SSC, 5 ng・µl⁻¹ unlabeled genomic DNA, and 50 ng・µl ⁻¹ of labeled genomic DNA as a probe) was prepared and then adjusted to 40 µl per slide with nuclease-free water. The GISH mixture was denatured at 90˚C for 10 min and immediately stored on ice for at least 5 min prior to mounting on slides. Chromosomes were then co-denatured at 80˚C for 3 min on a ThermoBrite (Fisher Scientific, USA), then hybridized overnight at 37˚C. The slides were washed in 2´ SSC for 30 min at RT, 0.1´ SSC for 30 min at room temperature and finally in 2´ SSC for 15 min at 42˚C. A dig-labeled probe was detected using an anti-dig-FITC conjugate (Sigma, USA) at 37˚C for 1 h. The slides were dehydrated in an ethanol series (70, 90, and 100 %) for 3 min each, air-dried, and counterstained with 1 μg/ml DAPI (Roche, Germany) in Vectashield (Vector Lab., Inc., USA). Images were captured with an Olympus BX53 fluorescence microscope equipped with a Leica DFC365 FS CCD camera, and processed using Cytovision version 7.2 (Leica Microsystems, Germany) software. Further image enhancements were performed using Adobe Photoshop CC.

Seed contents of total glucosinolate and glucoraphenin of BB#5, which is the precursor of sulforaphene, were measured according to the optimized official method, ISO 9167-1(1992) [19] . Two hundred milligrams of dried seeds were mixed with 10-ml boiled 70% methanol and incubated at 70˚C for 20 min to inactivate plant tissue enzymes. Following centrifugation at 3000 g for 15 min at 4˚C, the supernatant was concentrated to 3 ml under N₂ gas. Residue was re- suspended with 1 mL of 20 mM sodium acetate buffer and loaded onto a DEAE Sephadex-A25 column. Purified sulfatase (75 µL, 150 U/mL) was also loaded onto a Sephadex-A25 column and stored at 37˚C for 12 h. Desulfo-glucosino- lates were eluted 1.5mL ddH₂O (HPLC grade). The total eluted solution was concentrated to 0.5 mL for HPLC analysis. The sample was filtered through a 0.45-µm PVDF syringe filter before injection. The samples were analyzed on Varian LC-900 and separated on C-18 reverse phase column (Zorbax Eclipase XDB-C18, 4.6 ´ 250-mm i.d., 5-µm particle size, Macherey-nagel, Germany) at column temperature of 35˚C. Compounds were detected by UV detector set at 229 nm and separated by following program with a water-acetonitrile gradient at a constant 1 mL・min⁻¹. A gradient of water (A) and acetonitrile (B) was employed to separate the compounds by increasing the proportion of solvent B from 2% to 20% over 25 min and then switched to an additional linear gradient of 20% to 100% of solvent B up to 35 min. The glucosinolate analysis was replicated three times.

Multiple plants were produced from the cultured hybrid ovules (F₁) between jombaechoo and gaetmoo. Eight (F₁) of 13 plants from 07OV-124 to 07OV-136 were transplanted into soil after acclimation, grown in the greenhouse, and treated with colchicine. Most plants produced large amounts of pollen. One plant, 07OV-125, produced only one seed (F₂) from self-pollination and the other seven plants produced no F₂ seed. However, 15 hybrid seeds from 1 to 4 of 9 mates designated F₁F₁ were produced by reciprocally cross-pollinating the plants with the cultivar BB#1. One plant, 07OV-127, subjected to microspore mutagenesis for stabilization, produced a few embryos and two plants (designated F₁Mn₁) were acclimated.

All seeds harvested from self-pollinating the F₁ plants (one seed, F₂) and cross- pollinating the F₁ plants with BB#1 (15 seeds, F₁F₁) were planted and grown together with the two microspore-derived plants (F₁Mn₁). All plants produced large amounts of pollen. However, one F₂ plant did not produce any seed from self-pollinations made on more than 100 flowers and cross-pollinations made on 45 flowers distributed over 10 branches over a period of 6 d. It was determined to be female sterile. Two plants (07OV-127Mn-1 and 07OV-127Mn-2) derived from microspore mutagenesis did not produce seed from self-pollinations despite a total of 1,099 pollinations. However, these plants produced some many seeds in reciprocal crosses with several accessions of ´Brassicoraphanus. These results indicated that both 07OV-127Mn-1 and 07OV-127Mn-2 were self sterile, and not stable. Only 5 of 15 F₁F₁ plants produced self seeds (F₁F₂). Seed yields were poor, at 1 - 3 seeds per plant, with the exception of (BB#1x07OV-128)-3, which produced 29 seeds. On the contrary, the yield of hybrid seeds (F₁F₁BC₁) of three plants crossed with BB#1 was generally as high as 6 - 17 seeds per combination (Table 1). The reason that fewer seeds were produced from self-pollina- tion than cross-pollination is unknown.

Mn: Microspore culture with mutagen NMU of 0.01 μM. Pollination was done to more than 15 flowers for each mate.

An F₁F₁ plant, (BB#1x07OV-128)-4, produced the greatest amount of seed from self-pollinations and was subjected to microspore mutagenesis. Twenty seven and 38 plants from plots treated with and without mutagen, respectively, were grown until flowering. Only 7 and 8 plants, among the 21 and 28 pollen- producing plants, respectively, produced seeds from self-pollination. However, only one plant derived from mutagen treatment and two plants derived without mutagen treatment in each plot were selected as being highly fertile, as they produced more than 10 seeds per plant and 1 seed per pollination on the highest branch (Table 2).

In the succeeding generation, two of the three high-seed-yielding plants were grown separately in a different net case of 1-m length ´ 1-m width ´ 2-m height and pollinated via bees for 20 d. From this trial, only one line originating from the NMU treatment produced an average of 35g (about 7.000 grains) of seeds per plant (Table 3). This seed yield was similar to that for BB#1. When grown in the field during the fall and spring seasons, the one line was highly uniform and different from BB#1 in morphology. Leaves were dark green and glossy, and longer and fewer in number than BB#1 leaves. Bolting time in spring was also much later than for BB#1 (Figure 1). This line was labeled BB#5 and designated for a cytology study, glucosinolate analysis and variety protection. Plant performance was stable for more than six subsequent generations. Microspore muta-

MC: Microspore culture. Mn: Microspore culture with mutagen NMU of 0.01 μM. The line, BB#1 is a stable cultivar and the plant 07OV-128 is a newly synthesized unstable allopolyploid in ×Brassicoraphanus. ※: Stabilized.

MC: Microspore culture. Mn: Microspore culture with mutagen, NMU of 0.01 μM. Fertility in next generation: pollinated with bees in the net case of 1 m length ´1 m width ´2 m high for 20 d.

genesis with NMU was confirmed as a practical method of stabilizing allopolyploids between B. rapa and R. sativus. BB#5 was registered after 2 year’s examination in 2015.

The unstable line BB#4 and stable cultivar BB#1 were compared cytologically with BB#5. Various irregular divisions in meiosis of chromosomes were observed in all lines. However, the numbers of cells undergoing abnormal division were different between stable and unstable lines. The stable BB#1 and BB#5 lines were 10% and 12%, respectively, in terms of irregularities from prophase to anaphase, whereas the unstable line BB#4 was 57.4% irregular from prophase to anaphase (Table 4). GISH analysis identified 19 chromosome pairs, 10 from the AA genome of B. rapa and 9 from the RR genome of R. sativus (Figure 2). These results indicated that the BB#5 of baemoochae undergoes diploid-like division during meiosis [2] .

The total glucosinolate content of BB# 5 was slightly lower than that of BB#1 (4.64 and 4.94 mg・g⁻¹ dry weight [D.W.], respectively). The content of glucoraphenin, a precursor of sulforaphene, of BB#5 was similar to that of BB#1, at 1.57 and 1.52 mg・g⁻¹ D.W., respectively (Table 5).