Genetic Diversity of Genes Controlling Unilateral Incompatibility in Japanese Cultivars of Chinese Cabbage

In recent years, unilateral incompatibility (UI), which is an incompatibility system for recognizing and rejecting foreign pollen that operates in one direction, has been shown to be closely related to self-incompatibility (SI) in Brassica rapa. The stigma- and pollen-side recognition factors (SUI1 and PUI1, respectively) of this UI are similar to those of SI (stigma-side SRK and pollen-side SP11), indicating that SUI1 and PUI1 interact with each other and cause pollen-pistil incompatibility only when a specific genotype is pollinated. To clarify the genetic diversity of SUI1 and PUI1 in Japanese B. rapa, here we investigated the UI phenotype and the SUI1/PUI1 sequences in Japanese commercial varieties of Chinese cabbage. The present study showed that multiple copies of nonfunctional PUI1 were located within and in the vicinity of the UI locus region, and that the functional SUI1 was highly conserved in Chinese cabbage. In addition, we found a novel nonfunctional SUI1 allele with a dominant negative effect on the functional SUI1 allele in the heterozygote.


Introduction
Most Japanese cultivars of Chinese cabbage (Brassica rapa L.) are F 1 hybrids. Traditionally, their seeds have been produced using the Brassica self-incompatibility (SI) system. The SI system in Brassica is sporophytically controlled by a single S-locus with highly variable, multiple alleles [1]. The S-locus region contains two genes, SRK and SP11/SCR, which correspond to female and male S determinants, respectively [2]. SRK encodes a transmembrane receptor kinase, which is expressed specifically in stigma, and SP11/SCR encodes a small cysteine-rich ligand for SRK, which is localized on the pollen coat [3][4][5]. The S-haplotype-specific interaction of SP11 and the extracellular domain of SRK induces the SI reaction, in which the self-pollen fails to germinate or penetrate into the stigma [6]. The number of S-haplotypes has been estimated to be more than 100 in B. rapa [7][8][9]. Advanced understanding of the S-haplotype diversity, including dominance relationships between the haplotypes [10], is important for the efficient production of high-quality F 1 -hybrid seed in Brassica crops.
In addition to SI, we reported an interesting incompatibility relationship between Turkish and Japanese populations of B. rapa [11,12]. Pollen of the Turkish line was rejected on the stigma of the Japanese line, although crossing in the reverse direction showed compatibility. This cross-incompatibility operating in one direction, unilateral incompatibility (UI) occurred within species, in contrast to the UI that is known to occur in interspecies crossing [13,14]. Our molecular genetic studies of intraspecies UI in B. rapa revealed that it was controlled by the stigma-expressed gene, stigmatic unilateral incompatibility 1, SUI1, encoding an SRK-like receptor kinase and the pollen-expressed gene, pollen unilateral incompatibility 1, PUI1, encoding an SP11-like small cysteine-rich ligand. SUI1 and PUI1 are tightly linked and are considered to originate from a duplication event of the SRK-SP11 region in Brassica [12]. The S locus is located on chromosome A07, while the UI locus (containing SUI1 and PUI1) is on chromosome A04 of B. rapa [12]. From our further analysis of genetic diversity and distribution of the PUI1 and SUI1 genes in B. rapa, a functional PUI1-1 allele was found only in the Turkish lines and not in the Japanese lines, while the three functional SUI1 alleles (SUI1-1, -2, and -3) were found in Japanese wild populations and some cultivated varieties. Thus, loss of function of SUI1 in Turkish lines and PUI1 in Japanese lines might have resulted in the unidirectional pollen-stigma incompatibility in B. rapa [12].
The physiological pollen-rejection phenotype of the intraspecies UI is similar to that of SI and is consistent with the involvement of M-locus protein kinase (MLPK) in UI, which may function in SRK-mediated SI signal transduction [15,16]. It is noteworthy that the incompatibility response of UI is almost as strong as in the rigid SI phenotype in B. rapa. Thus, UI may have an effect on the SI-dependent breeding process in B. rapa. In this study, we extensively analyzed SUI1 and PUI1 alleles in Japanese cultivated lines of Chinese cabbage (Brassica rapa var. pekinensis). The results presented here give new insight into the historical relationship between UI and the breeding system of Chinese cabbage in Japan.

Cultivars of Chinese Cabbage Produced by Japanese Seed Companies
The UI phenotype observed on the stigma (stigma-side UI phenotype) was originally identified in the Japanese commercial hybrid variety 'Osome' of Japanese mustard spinach, Komatsuna (B. rapa var. perviridis), from the Takii seed company [11]. To understand the role of SUI1 in Japanese B. rapa cultivars, here we examined 52 commercial cultivars of Chinese cabbage (B. rapa var. pekinensis) from 16 Japanese seed companies (listed in Table 1) to determine their SUI1 and PUI1 alleles in addition to their stigma-side UI phenotype. All the cultivars used in this study, except 'Kashinhakusai' (#8), are F 1 hybrids. Because functional SUI1 alleles behave as dominant over nonfunctional alleles [11], they can be analyzed to predict the UI phenotype on the stigma side of hybrid varieties.

Discussion
Highly controlled pollen-stigma incompatibility is important for F 1 hybrid seed production of Brassica cultivars. The molecular mechanism of SI in Brassica has been studied for many years and is used in F 1 breeding. The recently discovered UI system, regulated by SUI1 and PUI1, can potentially provide another mechanism to control pollination in B. rapa. Therefore, determination of the UI genotype is considered as important as the SI genotype in the breeding of this major Japanese vegetable, Chinese cabbage. In this study, we determined the SUI1 and PUI1 allelic diversity of 22 and 48 cultivars, respectively, of Chinese cabbage in Japan. In addition, we confirmed the stigma-side UI phenotype of 47 cultivars. This revealed that most of the cultivars showed a stigma-side UI phenotype with a functional SUI1 allele (SUI1-2), whereas no functional PUI1 allele (PUI1-1) was found. We also searched the re-sequence data of B. rapa lines that are stocked at Chungnam National University and found a functional SUI1-2 allele in a South Korean population (data not shown). The fact that functional SUI1 alleles are present in Japanese and South Korean cultivars should be taken into consideration in breeding programs for B. rapa. UI may be beneficial as the additional incompatibility, which could be used in breeding programs by the introduction of PUI1-1 to the pollen donor.
To the best of our knowledge, there is no report that traits important for Chinese cabbage are mapped to flanking regions of the UI locus in chromosome A04. Thus, for an unknown reason, the functional SUI1-2 has been selected, and its sequence has been conserved during the breeding of Chinese cabbage cultivars in Japan. It would be interesting to investigate whether SUI1 itself strengthens SI and thus increases the efficiency of F 1 seed production.
In our previous study, we isolated nine intact alleles of SUI1 and showed that SUI1-1, SUI1-2, and SUI1-3 are incompatible with PUI1-1/PUI1-1 pollen [12]. SUI1-1 was originally isolated from a Japanese commercial hybrid variety of Komatsuna (B. rapa var. perviridis), and SUI1-2 and SUI1-3 were found in Japanese wild populations of B. rapa [12]. In the current study, we isolated three novel intact SUI1 alleles; one (SUI1-10) belongs to the functional clade (with SUI1-1, SUI1-2, and SUI1-3) and the other two alleles (SUI1-11 and SUI1-12) belong to the nonfunctional clade ( Figure 2). The fact that SUI1-10/SUI1-10 homozygote is stigmatic UC indicates that SUI1-10 is a nonfunctional allele ( Table 3). The Cys-413 residue of SUI1-2 is the last of the 12 highly conserved cysteine residues in the SUI1 extracellular domain and is located within the PAN_APPLE domain, which is the C terminal region of the extracellular receptor region. It has been clarified that homodimerization of SRK in Brassicaceae is essential for ligand interaction [17]. The PAN_APPLE domain of SRK has been shown to be important for ligand-independent dimer formation of SRKs and is responsible for correct intracellular trafficking [18][19][20][21]. It has been reported that the last Cys residue of SRK is predicted to form an intramolecular disulfide bond [20,21]. Thus, although the SUI1-10 sequence is similar to the functional SUI1-2, the C413Y mutation of SUI1-10 might cause structural disruption of SUI1 and breakdown of incompatibility through unusual dimer formation.
A feature of the sporophytic regulation of SI is the dominance relationship between S-haplotypes [10,22,23]. The molecular mechanism of the pollen-side dominance relationship has been well studied and revealed that mono-allelic gene expression of the dominant SP11 haplotype is controlled by small RNA-based epigenetic regulation [24][25][26]. On the stigma side, there is a complex allelic interaction that is as yet unexplained [10]. It was presumed that the SRK protein itself determines the dominance relationship rather than differences in SRK gene expression [23], and Naithani et al. [18] noted that the stigma-side dominance relationship may result from an increased tendency for heterodimer formation in some SRK pairs [18]. On the other hand, the existence of dominant negative alleles of receptor kinases that function as receptor complexes in many situations during plant development is widely known [27][28][29]. In most of these, the formation of a receptor complex with abnormal receptor proteins or receptor-related proteins encoded by dominant negative alleles causes disruption of signaling pathways. Thus, one possible explanation for the dominant negative effect of SUI1-10 may be an increase of SUI1-2/SUI1-10 heterodimer on the stigma surface and competitive inhibition of the interaction with the PUI1 ligand. We also found a dominant negative effect of SUI1-10 to SUI1-3, which has four aa substitutions (R322H, I326L, R363H, and V364D) compared to the extracellular domain of SUI1-2, suggesting that these four residues are not important for the effect.
In this study, it was found that the PUI1 gene of Japanese cultivars of Chinese cabbage showed very low diversity. Among six PUI1 alleles, of which only PUI1-1 from a Turkish strain can induce UI [12], only two patterns of genotype (pui1-3/pui1-4 or pui1-3/pui1-4/pui1-6) were observed, and no cultivars with a functional PUI1-1 allele could be found. Interestingly, the pui1-3/pui1-4 genotype might consist of two linked pui1-3 and pui1-4 genes ( Figure S1). Similarly, the pui1-3/pui1-4/pui1-6 genotype might consist of three linked pui1-3, pui1-4, and pui1-6 genes ( Figure S1). Such duplication and triplication of nonfunctional PUI1 have complicated the UI locus region. Although such PUI1 duplication or triplication cannot be found in the reference genome information of B. rapa inbred line Chiifu (B. rapa reference genome version 3.0, https://brassicadb.cn, accessed on 1 April 2021), de novo genomic sequence assembly of these Chinese cabbage cultivars using next-generation sequencing technology, including long-read sequencing, would provide new insights into the genomic structure of the UI locus [30]. In fact, we can find the two duplicated PUI1 genes on the UI locus of the genome sequence of B. rapa Z1(version 1.0, https://brassicadb.cn, accessed on 19 October 2021, Figure S2) [31].
Further analysis of the genetic diversity of the UI locus in B. rapa other than Chinese cabbage (subsp. pekinensis), such as turnips (subsp. rapa), leafy Brassica crops (subsp. chinensis, periridis), and field mustard (subsp. oleifera) will not only contribute to the discovery of novel alleles but also provide new insights into the genomic structure of the pollen-side factor and the dominant recessive interaction of the stigma-side factor. It will also be interesting to determine whether the UI locus has a multi-allelic structure like the S locus.

Plant Material
The plant material consisted of 52 commercial cultivars of Chinese cabbage, B. rapa ssp. pekinensis (Table 1). All except one, 'Kashinhakusai,' were F 1 hybrid cultivars. To produce self-pollinated progeny, bud pollination was performed. Petals and stamens were removed from a young flower bud (2-4 d before flowering), and the immature pistil was pollinated. The pollinated pistil was then covered with a paper bag until the seed was harvested. Plant materials were vernalized at 4 o C for 4 weeks in a refrigerator and then grown in a greenhouse.

Test Pollination
Flower buds were cut at the peduncle and pollinated. After pollination, they were stood on 1% solid agar for about 24 h under room conditions. Then, pistils of the pollinated flowers were softened in 1N NaOH for 1 h at 60 • C and stained with basic aniline blue (0.1 M K 3 PO 4 , 0.1% aniline blue). Samples were mounted in 50% glycerol on slides and observed by UV fluorescence microscopy ( Figure S3) [32]. At least three flowers were used from each cross combination, and observations were generally replicated at least three times on different dates for each cross combination. For the determination of the stigma-side UI phenotype, PUI1-1/PUI1-1 homozygous plants (S 24 t, S 40 t, and S 21 t) were used as the pollen donor in test pollinations (S 21 t was produced for this study) [16].

Cloning, Sequencing, and Genotyping of SUI1and PUI1 Alleles
Total DNA was extracted from young leaf tissue of B. rapa by the procedure of Murray and Thompson (1980) or using a DNeasy plant mini kit (Qiagen) [33]. For molecular cloning of full-length SUI1 and PUI1 genes, genomic PCR was performed using KOD-Plus-Neo DNA polymerase (TOYOBO) according to the manufacturer's instructions. PCR primers SUI1cDNA_F3 and SUI1_gR2 for SUI1 and PCP-like1-F1 and PCP-like1-R1 for PUI1 were used (Table S3). All amplified fragments were detected as a single band in the gel electrophoresis. PCR products were modified by adding 3 -A overhangs using A-attachment mix (TOYOBO) and cloned into a vector, pTAC-2, using DynaExpress TA PCR Cloning kit (Biodynamics). The nucleotide sequence was determined with a 3500 or 310 Genetic Analyzer using Big Dye Terminator version 3.1 or 1.1 Cycle Sequencing Kit (Applied Biosystems); in the case of SUI1, the SUI1-specific sequencing primers, SUIcDNA_F3, SUI_gR2, SUIinter_cF1, SUIinter_cF2, SUIinter_cF3, SUIinter_GF1, SUI1inter_cF4, and SUIinter_cF5 ( Figure 1 and Table S3), were used. GENETYX version 13 software package (GENETYX Corp.) was used for the sequence comparison and alignment. For the segregation analysis, we determined the genotype of SUI1 and PUI1 alleles by direct sequencing of PCR products. SUI1-1 and SUI1-10 alleles were amplified using primers SUI1_2-10typeSDF and SUI1_2-10typeSDR (Table S3). Each PUI1 allele was amplified using the primer pair for the PUI1 second exon region, PUI1-3.4.6-F, and PUI1-3.4.6-R (Table S3, Figure S4). For discrimination of PUI1 alleles by PCR-RFLP, amplified DNA fragments were cut by restriction enzyme (BamHI, SalI, or BsrI), followed by checking on an electrophoresed agarose gel ( Figure S4). For the direct sequencing marker, amplified fragments were purified from the electrophoresed agarose gel and sequenced as described above.

Phylogenetic Analysis
Phylogenetic analysis was performed on the Phylogeny.fr platform (http://www. phylogeny.fr/, accessed on 21 October 2021) [34]. Full-length amino acid sequences were aligned with MUSCLE (version 3.7) configured for the highest accuracy. Accession numbers of SRKs and SUI1s are listed in Table S4. After alignment, ambiguous regions were removed with Gblocks (version 0.91b). The phylogenetic tree was reconstructed using the PhyML program (version 3.0 aLRT). The default substitution model was selected assuming an estimated proportion of invariant sites and 4 gamma-distributed rate categories to account for rate heterogeneity across sites. The reliability of internal branches was assessed using the bootstrapping method (100 bootstrap replicates). The tree was represented with TreeDyn (version 198.3).