Introduction

The genus Diospyros L. consists of 400 species, among which oriental persimmon (Diospyros kaki Thunb.) is the most economically important. The origin of oriental persimmon is believed to be China, and it is a popular fruit particularly in countries of East Asia such as China, Korea, and Japan (Yonemori et al. 2000). Recently, persimmon has become popular in many other countries. According to FAO (Food and Agriculture Organization of the United Nations) statistics, worldwide persimmon production in 2014 was 5,200,000 t in total, with 71.9% from China, 8.3% from Korea, 4.7% from Spain, 4.6% from Japan, and 3.5% from Brazil. Because many persimmon cultivars have strong astringency owing to soluble tannin in the flesh, deastringency treatment such as postharvest treatment with carbon dioxide is frequently used to make the flesh palatable. Depending on the effect of seed formation on the natural astringency loss in flesh at harvest time, persimmon is classified into four types: (1) pollination-constant non-astringent (PCNA), (2) pollination-variant non-astringent (PVNA), (3) pollination-variant astringent (PVA), and (4) pollination-constant astringent (PCA) (Kajiura 1946). PCNA fruit naturally lose astringency at maturity and become edible without deastringency treatment. However, PVNA fruit sometimes retain astringency in the flesh, because loss of astringency in the whole flesh is dependent on seed formation. PVA fruit lose astringency only in a small portion of flesh around the seeds. PCA fruit always retain astringency in the whole flesh irrespective of the number of seeds. In contrast to the non-PCNA (i.e., PVNA, PVA and PCA) types, which can retain astringency at harvest time, PCNA type accessions lack the ability to accumulate large amounts of soluble tannins during fruit development (Yonemori et al. 2000); this natural astringency loss is highly desirable for commercial production. Therefore, breeding superior PCNA cultivars with high eating quality and productivity is a major goal of our breeding program.

The PCNA trait is recessive to the non-PCNA trait and is qualitatively inherited (Ikeda et al. 1985). Expression of the PCNA trait is under the control of a single locus, ASTRINGENCY (AST), and requires the presence of the recessive ast allele at all copies of the AST locus on each of the six corresponding chromosomes (Akagi et al. 2009; Akagi et al. 2010). D. kaki cultivars are hexaploid (2n = 6× = 90), with the exception of a few nonaploid (2n = 9× = 135) cultivars, such as ‘Hiratanenashi’ and ‘Miyazakitanenashi’ (Tamura et al. 1998; Zhuang et al. 1990). AST and ast alleles generally show autohexaploid inheritance, but segregation of these alleles is sometimes slightly distorted from the ratio expected from the autohexaploid inheritance model (Akagi et al. 2012; Kanzaki et al. 2008; Kono et al. 2016; Mitani et al. 2014b). Although it had been previously believed that PCNA trait originated only in Japan, Wang (1982) reported ‘Luotian-tianshi’, the PCNA cultivar of Chinese origin. Subsequently, other PCNA cultivars, including ‘Tianbaogai’ (former name, ‘Baogai Tian Shi’), were found in the area where ‘Luotian-tianshi’ was found (Yonemori et al. 2005). The PCNA trait of Chinese origin (C-PCNA) is genetically dominant to the non-PCNA traits, and the locus responsible for the C-PCNA trait is different from the AST locus (Ikegami et al. 2004, 2006).

Owing to the recessive inheritance of the PCNA trait in the hexaploid persimmon, breeders need to mainly cross PCNA accessions to each other to efficiently acquire PCNA offspring. So far, except for bud sports, only 18 local PCNA cultivars that have been found in the central part of Japan are preserved at the Grape and Persimmon Research Station, NIFTS (Yamada et al. 2012), while more than 1000, 900, and 180 cultivars mostly consisting of non-PCNA types have been reported in Japan, China, and Korea, respectively (Agricultural Research Station 1912; Cho and Cho 1965; Wang et al. 1997). In the persimmon breeding program at the National Agriculture and Food Research Organization (NARO) in Japan, repeated crosses within a small PCNA gene pool resulted in inbreeding depression, represented by reduced fruit weigh, vigor, and productivity (Yamada 1993; Yamada et al. 1994). To avoid inbreeding, we have incorporated non-PCNA cultivars, whose genetic background differs from that of PCNA cultivars (Guo and Luo 2011; Kanzaki et al. 2000; Naval et al. 2010; Parfitt et al. 2015), into the breeding program.

Our strategy is based on a pseudo-backcross (Bouquet 1986; Ruengphayak et al. 2015), in which F1 non-PCNA offspring from a cross between non-PCNA (Aaaaaa, AAaaaa, AAAaaa, AAAAaa, AAAAAa, or AAAAAA) and PCNA (aaaaaa) cultivars are backcrossed to a PCNA cultivar. However, this strategy still yields only a limited proportion of PCNA offspring in the BC1 generation: the expected proportion of PCNA offspring from a non-PCNA F1 parent with Aaaaaa, AAaaaa, or AAAaaa is 50, 20, or 5%, respectively, under an autohexaploid model with non-chromatid segregation (Allard 1960). To avoid unnecessary cultivation of non-PCNA offspring in the selection field, we apply DNA marker-assisted selection (MAS) for the PCNA trait. Kanzaki et al. (2010) identified the sequence of an AST- and ast-linked region by screening a genomic library with a 5R probe that showed restriction fragment length polymorphisms (RFLPs) between PCNA and non-PCNA cultivars. Finally, they produced sequence-characterized amplified region (SCAR) markers for this region (termed the ‘5R adjacent region’). The reliability of a SCAR marker that amplifies the 5R adjacent region has been demonstrated in practical MAS for offspring of ‘Taiten’ × ‘Kanshu’. Discrepancy between phenotypic evaluation by a sensory test and estimated genotype was found in only 3 out of 251 offspring (Mitani et al. 2014a), confirming tight linkage of the region to the AST locus. Therefore, we call this region the AST locus-linked region. Multiplex PCR using a set of three primers (AST-F, PCNA-F, and 5R3R) that hybridize to the 5R adjacent region simultaneously amplifies both AST and ast allele-linked DNA (Kanzaki et al. 2010). We have tested more than 5000 individuals by multiplex PCR (Sato and Yamada 2016) and have found this to be a labor-saving and reliable system for MAS. In the pseudo-backcross strategy with MAS, identifying the copy number of AST and ast alleles in non-PCNA parents is particularly important because this critically determines the segregation rate of PCNA offspring. We can obtain PCNA offspring even at the F1 generation when a non-PCNA parent with a genotype of AAAaaa, AAaaaa, or Aaaaaa is used. In contrast, pseudo-backcross is necessary to obtain PCNA offspring when a non-PCNA parent with the genotype of AAAAaa, AAAAAa, and AAAAAA is used. Therefore, determination of the copy number of AST and ast alleles among a wide range of non-PCNA germplasms would accelerate efficient PCNA breeding.

Akagi et al. (2010) reported a method to directly estimate the copy number of AST and ast alleles based on quantitative real-time polymerase chain reaction (qPCR) of the ast allele-linked region. They estimated quantitative genotypes at the AST locus in 63 non-PCNA cultivars/selections, among which 15 cultivars were either AAAaaa, AAaaaa, or Aaaaaa. Detection of one copy difference of ast allele by qPCR, especially for accessions with higher copy number of ast allele, would require high levels of technical skill. This methodological limitation of qPCR resulted in significant calculation errors in measurements for some accessions. Furthermore, the method elaborately uses three genomic regions to standardize quantity of amplified products linked to ast allele. However, inconsistency of the estimated allele dosage among the references was observed for 32% (20/63) of the accessions. This implies that copy number in the genome and/or sequence of the references can be slightly different among the accessions used. Therefore, estimating exact copy number for any persimmon accessions would be technically difficult even by this ingenious method. As another option to estimate allele dosage at the AST locus, we have previously presented a fragment analysis that detects high polymorphisms in fragment size at the AST locus-linked region (Kono et al. 2016), technically less labor-intensive method than qPCR. We found a total of 12 AST allele-linked fragments of various sizes caused mainly by a simple sequence repeat (SSR) in 14 non-PCNA cultivars. The detected number of AST fragments indicated the minimum AST allele copy number, because we could not determine which allele(s) occurred in more than one copy when the number of detected alleles was fewer than six in hexaploid persimmon. Segregation data of the AST allele and/or progeny genotypes would be required to identify the precise copy number of AST alleles in non-PCNA individuals. However, we considered that the high-throughput fragment analysis using a capillary sequencer could be informative enough to find non-PCNA parents with fewer AST and more ast alleles from a wide range of non-PCNA germplasms, because an accession with a highly polymorphic AST locus-linked region would display more alleles.

In addition to allele dosage estimation, the fragment size analysis described above could be applicable to cultivar identification because of high polymorphism at the AST locus-linked region (Kono et al. 2016). Various types of DNA markers have been applied to persimmon cultivar identification, e.g., random amplified polymorphic DNA (RAPD; Badenes et al. 2003; Luo et al. 1995; Yamagishi et al. 2005), RFLP (Maki et al. 2001), retrotransposon-based (Du et al. 2009), and SSR (Naval et al. 2010) markers. A marker technology that reveals a high amount of polymorphism with minimum cost is optimal for practical cultivar identification. Fragment size analysis using multiplex PCR (Kanzaki et al. 2010) could potentially be such as system because it can detect various alleles simply by single multiplex PCR and fragment size analysis (Kono et al. 2016).

The objectives of the present study were (1) to determine the sizes of AST- and ast-allele-linked fragments (hereafter termed AST and ast fragments) at the AST locus-linked region in 237 persimmon accessions of widely varying genetic backgrounds and (2) to test whether AST and ast fragment size analysis can be applied successfully for cultivar identification. During the fragment analysis, we identified non-PCNA accessions with a single ast fragment peak, which enabled us to identify the sequence of each ast fragment. Based on the fragment sizes and sequence polymorphisms, we discuss the possible origin of ast alleles.

Materials and methods

Plant materials

Supplemental Table S1 shows the 237 D. kaki Thunb. accessions used in this study: these comprise 148 non-PCNA, 3 C-PCNA, and 86 PCNA accessions. The sample population comprised 202 local cultivars, 18 crossbred cultivars, and 17 selections. Bud mutants and synonyms were each counted as a separate accession. The origins of the accessions were Brazil (1), China (19), Israel (2), Italy (12), Japan (193), Japan–China (1), Korea (4), New Zealand (2), Spain (1), and Turkey (2). We treated the selection 310-24 (C-PCNA) as a Japan–China hybrid, because its origin is both Japan (‘Taishuu’; PCNA) and China (‘Luotian-tianshi’; C-PCNA). All plant materials were obtained from the Grape and Persimmon Research Station, NIFTS, Hiroshima, Japan.

DNA extraction

Approximately 1.0-cm2 leaves or one to five dormant buds, whose outer scales were peeled by one or two layers (Kono et al. 2018), were frozen in liquid N2 and stored at − 20 °C until use. Samples were homogenized under liquid N2, and 500 μL of pre-wash buffer (0.1 M HEPES-NaOH (pH 8.0), 0.1% (w/v) soluble polyvinylpyrrolidone, and 10 mM dithiothreitol) was added and mixed thoroughly by vortexing; dithiothreitol was added to the buffer just before use. After centrifugation at 20,000×g at 4 °C for 5 min, the supernatants were discarded. Pellets were resuspended in 1 mL of pre-wash buffer and centrifuged at 20,000×g at 4 °C for 3 min; the process was repeated three times in total. The final pellets were subjected to DNA extraction with a Nucleon Phytopure Plant extraction kit (GE Healthcare UK Ltd., Little Chalfont, UK). DNA concentration was measured by using a Qubit 3.0 fluorometer (Thermo Fisher Scientific Inc., Waltham, MA, USA).

Fragment size analysis

Multiplex PCR of the AST locus-linked region was performed with two fluorescent-labeled forward primers HEX-AST-F (5′-GTTGCATCGCATAGCGGGTTTGAGG-3′), FAM-PCNA-F (5′-CCCCTCAGTGGCAGTGCTGC-3′), and unlabeled reverse primer 5R3R (5′-GAAACACTCATCCGGAGACTTC-3′) (Kanzaki et al. 2010). The 10-μL PCR reaction mixture contained 5 μL of 2× Gotaq Master Mix (Promega, Madison, WI, USA), 0.2 μM of the two forward primers and the reverse primer, and 10–20 ng of genomic DNA. The PCR cycling conditions were as follows: 95 °C for 3 min for initial denaturation, followed by 27 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, polymerization at 72 °C for 1 min, and final extension at 72 °C for 10 min. We increased the number of PCR cycles to 31 when amplification was insufficient. Amplified PCR products were separated by using an ABI 3130xl Genetic Analyzer (Thermo Fisher Scientific Inc.). The size of each amplified fragment was calculated based on the GeneScan 500 ROX Dye Size Standard (Thermo Fisher Scientific Inc.) by using the GeneMapper ver. 5.0 software (Thermo Fisher Scientific Inc.).

For additional SSR markers, we first screened 19 markers (Naval et al. 2010) developed by Soriano et al. (2006) and selected six (ssrdk10, 14, 16, 17, 29, and 30) that showed clear peaks with few stutter bands. PCR was performed by using a specific forward primer for each marker with an M13(-21) tail at the 5′-end, the universal FAM-labeled M13(-21) primer (Schuelke 2000), and a specific reverse primer with a pigtail (5′-GTTTCTT-3′) added to its 5′-end to reduce non-adenylated products (Brownstein et al. 1996). The 10-μL PCR reaction mixture contained 5 μL of 2× Gotaq Master Mix (Promega), 0.05 μM of the specific forward primer, 0.2 μM of the specific reverse primer and the universal primer, and 5–10 ng of genomic DNA. PCR conditions were as follows: 95 °C for 2 min for initial denaturation, followed by 26 cycles of denaturation at 95 °C for 30 s, annealing at 50 °C for 45 s, polymerization at 72 °C for 45 s, and final extension at 72 °C for 10 min. We increased the number of PCR cycles to 31 when amplification was insufficient. Separation of PCR products and size analysis were performed as described for the multiplex PCR for the AST locus-linked region. PCR amplification and fragment analysis of AST locus-linked and ssrdk markers were carried out at least twice to ensure the reproducibility of the produced bands.

Direct sequencing

Genomic DNA of 48 non-PCNA accessions (45 genotypes) that showed only one-size peak of ast fragment and two PCNA accessions of ‘Gosho’ (a347a349a355) and ‘Gosho-Gose’ (a349a355) were used. To avoid the production of chimeric fragments from similar template sequences during PCR, which was observed in our previous study on grape (Kobayashi et al. 2001), we did not clone ast fragments from accessions with variably-sized ast fragments. PCR to generate ast fragments was performed using forward primer PCNA-F and reverse primer 5R3R (Kanzaki et al. 2010). The 30-μL PCR reaction mixture contained 1.5 mM MgCl2, 0.2 mM dNTPs, 0.2 μM each primer, 1× ExTaq buffer, 0.25 U ExTaq polymerase (Takara, Kyoto, Japan), and 10–20 ng DNA. The PCR cycling conditions were the same as described for multiplex PCR of the AST-linked region in the Fragment size analysis section. Amplified ast fragments were precipitated with ethanol and dissolved in 10 μL of TE. Aliquots (1 μL) of purified PCR products were sequenced by using a BigDye terminator v3.1 cycle sequencing kit (Thermo Fisher Scientific Inc.) with either PCNA-F or 5R3R primer. For the direct sequence analysis of the ast fragments of ‘Gosho’ and ‘Gosho-Gose’, we used a349 (5′-TAAAGCAATGAACCTTTTGG-3′)- and a353 (5′-TAAAGCAATGAACCTTTTGT-3′)-specific forward primer.

For the direct sequence analysis of ast fragments of ‘Aizumishirazu’, ‘Ichidagaki’, and ‘Zenjimaru’, which have both a349 and a353 alleles, we performed PCR using either a349-specific forward primer, as described above, and reverse primer 5R3R or a353-specific forward primer, as described above, and reverse primer 5R3R. Amplified fragments were purified and sequenced using a349-specific forward primer, a353-specific forward primer, or the 5R3R reverse primer. Nucleotide sequences were analyzed by using GENETYX v. 9 software (GENETYX Corp., Tokyo, Japan).

Data analysis

Polymorphism information content (PIC) value is widely used to measure the information content of molecular markers based on the number and frequency distribution of alleles. The PIC value for each marker having l alleles was calculated as follows:

$$ \mathrm{PIC}=1-\left({\sum}_{i=1}^l{P_i}^2\right)-{\sum}_{i=1}^{l-1}{\sum}_{j=i+1}^l2{P}_i^2{P}_j^2, $$

where P i and P j are the population frequencies of the ith and jth alleles (Botstein et al. 1980). P i or P j for the AST marker was calculated as the occurrence of the given allele divided by the sum of occurrences of AST and ast alleles in 184 genotypes.

Results

The AST locus-linked region of 237 persimmon accessions (Supplemental Table S1) was analyzed by using the multiplex PCR system. Since our samples may include unknown bud mutants and/or synonyms, we additionally analyzed these accessions with six SSR markers (ssrdk10, 14, 16, 17, 29, and 30; Naval et al. 2010) to uncover unique genotypes and thus avoid complexity of the sample population. Multiplex PCR for the AST locus-linked region and uniplex PCR for each of the six ssrdk markers yielded a total of 101 different alleles; the total allele number per individual accession was 19–32 (Table 1). The combination of the AST locus-linked marker and ssrdk markers identified 184 different genotypes (127 non-PCNA, 3 C-PCNA, and 54 PCNA) (Table 1). Accessions sharing identical genotypes at the set of 7 markers were categorized into 21 groups (Table 2). These groups consisted of known bud sports having the same origin and/or morphologically similar cultivars with different names, most likely synonyms. Of the 12 accessions introduced from Italy, only ‘Shogatsu-Italy’ and ‘Cioccolatino’ exhibited unique genotypes, while the others were categorized into either group 3, 4, 8, 12, 16, or 19 (Table 2). The Korean cultivar ‘Chung Nam’ showed the same genotype as the Japanese cultivar ‘Okayamaokugosho’ (group 6). ‘Flat Fuyu’ and ‘Fuyu-New Zealand’, which were introduced from New Zealand, showed the same genotype as the Japanese cultivars ‘Okugosho’ (group 15) and ‘Fuyu’ (group 14), respectively (Table 2). The other non-Japanese accessions, Brazil (1), China (19), Israel (2), Japan–China (1), Korea (3), Spain (1), and Turkey (2), showed unique genotypes among the 237 accessions.

Table 1 Summary of the AST locus-linked marker and six ssrdk markers and their discrimination ability for 127 non-PCNA, 3 C-PCNA, and 54 PCNA persimmon genotypes
Full size table
Table 2 Cultivars with identical genotypes in the AST locus-linked marker and all six ssrdk markers
Full size table

Among the 184 unique genotypes, 21 AST fragments (size, 220–254 bp) and 5 ast fragments (size, 347–355 bp) were detected by the multiplex PCR system (Tables 3 and 4). Given the tight link between the 5R adjacent region and AST locus in previous studies (Mitani et al. 2014a; Kanzaki et al. 2010), we regarded polymorphisms of the amplified fragments as polymorphisms of AST and ast alleles at the AST locus. We detected 12 rare AST (A220, A230, A232, A234, A236, A238, A242, A244, A248, A250, A252, and A254) alleles and one rare ast (a347) allele (frequency, < 0.02; Naval et al. 2010). At least one rare allele were found in 14 out of 19 Chinese (IDs: 17, 72, 74, 75, 77, 78, 88, 89, 113, 114, 115, 122, 124, 129), 3 out of 4 Korean (IDs: 29, 107, 190), and 1 out of 1 Brazilian accessions (ID: 63) (Table 3; Supplemental Table S1), suggesting uniqueness of Chinese, Korean, and Brazilian accessions among the 237 accessions. Cultivars from Israel, Italy, New Zealand, Spain, and Turkey did not have any rare alleles (Tables 3 and 4).

Table 3 Detected AST and ast alleles in 127 non-PCNA and 3 C-PCNA persimmon genotypes
Full size table
Table 4 Detected ast alleles in 54 PCNA persimmon genotypes
Full size table

The total number of different AST and ast alleles per individual genotype was 2–7 with an average of 4.4 (Table 1). Our samples included a few nonaploid cultivars: e.g., ‘Miyazakitanenashi’ and ‘Hiratanenashi’ (Tamura et al. 1998; Zhuang et al. 1990). A total of seven different AST and ast alleles were detected in ‘Miyazakitanenashi’ (A224A231A233A240a349a353a355) only, and a total of six different AST and ast alleles were observed for ‘Hiratanenashi’ (A224A228A231A250a349a353), its four known bud sports, and ‘Hiratanenashi-Italy’ and ‘Tone Hiratanenashi’ (group 8 in Tables 2 and 3). Among the 182 genotypes with hexaploidity, 31 non-PCNA genotypes showed a total of six AST and ast fragments, suggesting the presence of one copy of each allele in these genotypes (Table 3). The C-PCNA locus differs from the AST locus (Ikegami et al. 2004), and three C-PCNA accessions, 310-24 (Kono et al. 2016), ‘Tianbaogai’, and ‘Luotian-tianshi’ (Akagi et al. 2010) had AST alleles (ID: 19, 40, and 72 in Table 3). In this study, we focused on the genotype at the AST locus-linked region, and thus, we treated the three C-PCNA accessions the same as non-PCNA type. Among the 127 genotypes of hexaploid and nonaploid non-PCNA accessions and 3 genotypes of C-PCNA accessions, the number of AST alleles was 1–6 with an average of 3.4 (Table 3). We identified 40, 13, and 6 genotypes having 4, 5, and 6 different AST alleles per individual, respectively. ast alleles were detected not only in PCNA genotypes, but also in a wide range of non-PCNA genotypes. Of the 127 non-PCNA and 3 C-PCNA genotypes, 109 had at least one ast allele (Table 3), which is consistent with the prevalence of ast alleles in the non-PCNA population studied in Akagi et al. (2010). We identified 19 genotypes composed of 21 accessions (IDs: 2–4, 19–28, 67–71, 111, 190, 203) with 3 different ast alleles per individual. Because ast alleles were less polymorphic than AST alleles, the detected number of ast alleles per individual was less than that for the AST allele: the number of ast alleles in non-PCNA accessions was 0–3 with an average of 1.5 (Table 3).

The AST locus-linked marker in this study showed a PIC value of 0.90 (Table 1), which is considered highly informative according to the criteria of Botstein et al. (1980) (highly informative, PIC > 0.5; moderately informative, 0.25 < PIC < 0.5; and relatively uninformative, PIC < 0.25), suggesting that multiplex PCR of the AST locus-linked region could be used for cultivar identification. Of the 127 non-PCNA and 3 C-PCNA genotypes, 113 showed a unique AST and ast allele pattern, whereas only 5 out of 54 PCNA genotypes showed a unique ast allele pattern (Table 1). Table 1 shows the discrimination ability of the AST locus-linked marker compared with six ssrdk markers developed by Soriano et al. (2006). A total of 75 alleles were detected by the six ssrdk markers. Single ssrdk markers had 9 to 20 alleles, including 2 to 11 rare alleles. Observed allele number per individual for each ssrdk marker was 1–5, 1–6, or 2–6 with an average of 2.5 to 4.0. PIC values ranged from 0.70 to 0.86. As a result, a unique genotype based on a single ssrdk marker was found for only 11 to 62 out of 127 non-PCNA and 3 C-PCNA genotypes and 0 to 14 out of 54 PCNA genotypes (Table 1). The set of six ssrdk markers discriminated all genotypes except for two non-PCNA and four PCNA genotypes that could be discriminated by the AST locus-linked marker (Table 1): ‘Tamopan’ (A224A226A229A236a349) had A236, whereas ‘Mopanshi’ (A224A226A229A238a349) had A238 (Table 3); ‘Misatogosho’ (a347a349a355) did not have a353, which was present in ‘Fukurogosho’, ‘Hazegosho’, and ‘Zennosuke’ (a347a349a353a355) in group 17; and ‘Gosho’ (a347a349a355) had a347, but this was absent in ‘Gosho-Fukushima’, ‘Gosho-Gose’, ‘Izushikogosho’, and ‘Kaibaragosho’ (a349a355) in group 18 (Table 4).

Analysis of ast fragments identified 45 non-PCNA genotypes with only one-size peak of ast fragment (22, 2, 19, and 2 genotypes for a349, a351, a353, and a355, respectively) (Tables 3 and 5). To gain insight into the ast allele, we directly sequenced the amplified ast fragments of the 48 accessions with these genotypes. We found that differences in the sizes of the ast fragments were caused by different numbers of AT-dinucleotide SSRs beginning at the 259th nucleotide of the fragment (Fig. 1), which corresponds to the region that was previously reported to cause allele size polymorphisms at AST fragments (Kono et al. 2016). The sequences of all a349 fragments from 19 Japanese and 5 Chinese accessions were identical (Table 5). Interestingly, the a353 fragment contained several single nucleotide polymorphisms (SNPs), resulting in four types of a353 (a353-1, a353-2a, a353-2b, and a353-2c; Fig. 1). The allele a353-1 was found in six Japanese, one Turkish, one Israeli, and one Chinese accession (Table 5). The alleles a353-2a, a353-2b, and a353-2c were found in eight, four, and one accessions, respectively, all of which were Japanese (Table 5). Sequences of the a351 fragment from two Chinese accessions, ‘Raotianhong’ (A222A226A229a351) and ‘Gong-cheng-shui-shi’ (A228A230A231A242A250a351), were identical, and sequences of the a355 fragment from two Japanese accessions, ‘Hegurogaki’ (A225A228A231A240a355) and ‘Yukineri’ (A226A228A231A233a355), were identical (Table 5). The sequence difference between a353-2a and a351, and between a353-2b and a355 fragments, was one AT repeat of the SSR (Figs. 1 and 2). Fragments a351, a353-1, a353-2a, a353-2b, a353-2c, and a355 (a353 group) were more similar to each other than to a349, which differs from the a353 group at > 10 SNPs (Fig. 2). Consequently, the above seven ast alleles were categorized into two groups based on their sequence similarity to a349 (Fig. 2).

Table 5 Summary of direct sequence analysis of ast fragments for 48 non-PCNA accessions with one ast fragment peak and 3 non-PCNA local cultivars having ancient origin with two ast fragment peaks
Full size table
Fig. 1
figure 1

Schematic representation of the AST- and ast-allele-linked regions isolated from genomic libraries of ‘Nishimurawase’ and ‘Jiro’, respectively (Kanzaki et al. 2010), and multiple nucleotide sequence alignment of seven ast fragments amplified with PCNA-F and 5R3R primers. Black boxes indicate the 5R probe that was used for library screening. The gray box indicates a large insertion named Indel-3 in the ast allele-linked region. Arrows indicate the positions of primers used for the multiplex PCR. Numbers indicate the positions from the 5′-end of the 5R region. Our fragment analysis detected AST allele-linked fragments amplified with AST-F and 5R3R and/or ast allele-linked fragments amplified with PCNA-F and 5R3R. Note that non-proofreading DNA polymerase adds an adenylate to the 3′ end of the PCR fragment, resulting in a longer PCR fragment than the actual size of the amplified region. Because the mobility of the labeled single-stranded fragments depends on several factors (i.e., length, sequence, running conditions), size inconsistency between sequencing and fragment analysis can occur (Pasqualotto et al. 2007). For simplicity, we used “called” sizes in this study

Full size image
Fig. 2
figure 2

Diagram of the relationship between ast alleles. Sequence differences between ast allele-linked fragments connected by double-headed arrows are indicated. Relationship between a347 and a349 is indicated by a gray double-headed arrow because the complete sequence of the a347 fragment was not obtained

Full size image

Sequences of the a353 fragments of ‘Amahachiya’ (A224A226A228a353), ‘Omidanshi’ (A224A225A228A231a353), and ‘Shimofuri’ (A224A225A228A231a353) were consistent with a353-2a and a353-2b except for the 212th nucleotide, where sequence chromatograms showed heterozygous C/T (Fig. 3a). Since the difference between a353-2a and a353-2b is one SNP (C or T) at this position (Fig. 1), these alleles can be notated as ‘Amahachiya’ (A224A226A228a353-2aa353-2b), ‘Omidanshi’ (A224A225A228A231a353-2aa353-2b), and ‘Shimofuri’ (A224A225A228A231a353-2aa353-2b) (Table 5). The a353 fragment of ‘Otani’ (A224A225A226a353) was consistent with a353-2a except for the 101th nucleotide, where sequence chromatograms again showed heterozygous C/T (Fig. 3b). We denoted the a353 fragment with C nucleotide at 101th position as a353-2c; hence, the allele present in ‘Otani’ would be A224A225A226 a353-2aa353-2c (Table 5). Sequences of the a353 fragments of ‘Rojo Brillante’ (A224A228a353) and ‘Kanzo’ (A224A231a353) showed heterozygous chromatograms at several positions (data not shown), indicating the existence of at least two types of a353 fragment in these accessions. The sequences of these a353 fragments could not be determined because SNPs at multiple sites gave several possible sequence patterns (Table 5).

Fig. 3
figure 3

Sequencing chromatograms of ast allele-linked fragments. Black arrows indicate the locations of heterozygous nucleotides. a ‘Amahachiya’, ‘Omidanshi’, and ‘Shimofuri’ show heterozygosity at the 212th nucleotide of a353, whereas ‘Kubo’ and ‘Dejima’ show homozygosity. b ‘Otani’ shows heterozygosity at the 101th nucleotide of a353, whereas ‘Kubo’ shows homozygosity. c Sequencing of ‘Gosho’ and ‘Gosho-Gose’ with a349- and a353-specific primers. Note that a349- and a353-specific primer harbors an a349- and a353-group-specific SNP at its 3′-end. In ‘Gosho’, sequencing with an a349 specific primer shows heterozygosity beginning from the 275th nucleotide. The other three chromatograms show homozygosity in this region

Full size image

‘Gosho’ (a347a349a355) and ‘Gosho-Gose’ (a349a355) (and others in group 18) shared the same alleles for all six ssrdk markers, but not the AST locus-linked marker (Table 4); furthermore, the morphological traits of ‘Gosho’ were very similar to those of the accessions in group 18 (data not shown). Therefore, ‘Gosho’ could be derived from one of the accessions in group 18 if a mutation of either a349 or a355 gave rise to a347. To determine the origin of the a347 allele, we directly sequenced ast fragments amplified with PCNA-F and 5R3R by using either an a349- or a353-specific sequence primer; these primers, respectively, harbor an a349- or a353-group-specific SNP at their 3′-end (G for a349 or T for a353 group at 47th nucleotide; Fig. 1). The sequence chromatogram for ‘Gosho’ with the a349-specific primer showed heterozygosity from the 275th nucleotide onward, while that with the a353-specific primer showed homozygosity in this region (Fig. 3c). This result is consistent with the situation where the a349-specific primer yields the sequences of both the a349 fragment and a related fragment with an AT dinucleotide deletion in the AT repeat region. Chromatograms for ‘Gosho-Gose’ sequenced with either the a349-or a353-specific primer showed homozygosity. These results clearly indicate that a347 originated from a349 (Fig. 2).

Fragment analysis revealed that the set of ast alleles in the local non-PCNA cultivars having known ancient origin contained either a349 or a353 or both: ‘Aizumishirazu’ (originated in the 1300s; A224A228A231A240a349a353) ‘Ichidagaki’ (originated around 600 years ago; A224A228A231A233a349a353), ‘Saijo’ (originated in 1239; A222A224A226A231A233a349), and ‘Zenjimaru’ (originated in 1214; A224A228A240a349a353) (Yamada 1996a, b, c, d) (Table 3). We therefore reasoned that revealing the sequences of the a349 fragment of ‘Saijo’ and the a349 and a353 fragments of the other three ancient cultivars may shed light on the development of ast alleles in Japan. Assuming that the a349 and a353 fragments of ‘Aizumishirazu’, ‘Ichidagaki’, and ‘Zenjimaru’ have a349- and a353-specific SNPs at the 47th nucleotide, respectively (Fig. 1), we conducted PCR with two primer combinations for each cultivar: a349-specific forward and 5R3R reverse primers and a353-specific forward and 5R3R reverse primers. The size of the PCR products corresponded to that of the region between the specific forward primers and the 5R3R reverse primer (data not shown), supporting the presence of both a349 and a353-group fragments. Direct sequencing analysis revealed that the a349 fragments of the three cultivars were the same as the corresponding regions of the a349 fragment of the 24 accessions in Table 5. The sequences of the a353 fragments derived from ‘Ichidagaki’ and ‘Zenjimaru’ were the same as that of the corresponding region of a353-1, and those of ‘Aizumishirazu’ were the same as that of the corresponding region of a353-2a (Table 5).

Discussion

Information on the AST and ast allele dosage in non-PCNA accessions is important for efficient breeding of PCNA progenies from non-PCNA parents, because parents with fewer AST and more ast alleles have a higher chance of producing PCNA offspring. By conducting high-throughput fragment analysis of the AST locus-linked marker in 237 persimmon accessions using capillary sequencer, we identified a total of 21 AST and 5 ast fragments of different sizes. The number of different-sized fragments indicates the minimum copy number of AST and/or ast alleles. We identified 21 non-PCNA accessions (19 genotypes) with 3 different-sized ast fragments per individual; these included 17 local non-PCNA cultivars (IDs: 2–4, 20–23, 25–27, 67–71, 190, 203), 3 F1 crossbred cultivars/selection (IDs: 19, 24, and 28) from PCNA × non-PCNA crosses, and 1 nonaploid seedless cultivar (ID: 111). Compared with other non-PCNA cultivars, these 17 local cultivars have great potential for use as parents in our pseudo-backcross strategy, because crosses between any of these cultivars and a PCNA cultivar would produce non-PCNA progenies with fewer AST and more ast alleles and even PCNA progenies at the F1 generation. Among these 17 local cultivars, ‘Aosa’, ‘Chung Nam’, ‘Hiroshimashimofuri’, ‘Sakushumishirazu’, ‘Shogatsu-Italy’, ‘Okayamaokugosho’, ‘Tenryubo’, and ‘Yoshidagosho’ bear relatively large fruit (around 300 g) with fewer appearance defects on fruit skin and thus would be primary candidates for parents in the pseudo-backcross strategy. Furthermore, according to AFLP (amplified fragment length polymorphism) analysis, ‘Hiroshimashimofuri’, ‘Sakushumishirazu’, and ‘Tenryubo’ have a different genetic background to that of PCNA cultivars (Parfitt et al. 2015), and so their use would avoid inbreeding depression. On the other hand, a total of 59 non-PCNA genotypes with four or more different AST alleles were identified. If one of these accessions were to be used in the initial non-PCNA × PNCA cross in the pseudo-backcross strategy, the breeder would need to choose F1 non-PCNA individuals with fewer AST and more ast alleles to backcross to the PCNA parent to efficiently produce PCNA offspring at the BC1 generation.

The minimum numbers of AST and ast alleles determined here were in good accordance with the allele numbers estimated in 63 non-PCNA accessions by qPCR analysis of an ast allele-linked marker (Akagi et al. 2010); of these 63 accessions, 34 were also included in the current study. The numbers of AST and ast allele fragments were equivalent to or smaller than the estimated number of the corresponding alleles in Akagi et al. (2010) with two exceptions: ‘Heixinshi’ (AAAAAa in Akagi et al. 2010; A226A254a349a353 in current study) and ‘Omidanshi’ (AAAaaa in Akagi et al. 2010; A224A225A228A231a353 in current study). A possible explanation for this inconsistency is that a one-ast-allele difference could occur in the quantitative genotyping owing to methodological limitation and depending on the reference site used (Akagi et al. 2010). One of the possible genotypes for ‘Heixinshi’ and ‘Omidanshi’ deduced from our fragment analysis is AAAAaa, which is one ast allele more, and one ast allele less, respectively, than the genotypes determined for these cultivars by Akagi et al. 2010. Since we identified a353-2a and a353-2b fragments in ‘Omidanshi’ in the sequence analysis, the genotype of ‘Omidanshi’ would be A224A225A228A231a353-2aa353-2b.

The AST and ast allele data obtained by the sequencer-based fragment analysis is limited in that it represents minimum copy number rather than actual allelic copy number. Excluding the 2 genotypes of 8 nonaploid accessions, 31 genotypes with a total of 6 different AST and ast alleles were assumed to have one copy of each allele, because D. kaki is generally hexaploid. However, for the other 151 genotypes, which showed fewer than six different-sized fragments, the method could not reveal which alleles were duplicated. One solution to this problem is to examine the segregation of the alleles in the F1 generation of a non-PCNA of interest × PCNA cross. We previously successfully estimated the copy number of the AST allele of six non-PCNA parents by assessing segregation of each AST allele and F1 progeny genotypes in the non-PCNA × PCNA F1 population (Kono et al. 2016). A different approach is necessary to determine ast allele copy number because the presence of the same ast alleles in both non-PCNA and PCNA parents makes it difficult to deduce copy number and inheritance of the ast allele in the F1 generation. Crossing PCNA or non-PCNA cultivars of interest with a non-PCNA parent that has no ast alleles would be one solution; here, we successfully identified 23 non-PCNA accessions (21 genotypes) that have no ast alleles. In the same way as for the AST allele, one could estimate ast allele copy number and inheritance in a parent of interest by assessing segregation of each ast allele and/or progeny genotype.

Our results indicate that fragment analysis with the AST locus-linked marker would be applicable for cultivar identification of non-PCNA accessions. The AST locus-linked marker discriminated 87% (113/130) of non-PCNA genotypes, whereas single ssrdk markers discriminated 8% (11/130) to 48% (62/130) of them. PIC value, the number of total and rare alleles, and the mean allele number per individual of the AST locus-linked marker explain its higher discrimination power compared to ssrdk markers. Additionally, the AST locus-linked marker displayed allele size polymorphisms within groups of cultivars with different names but practically indistinguishable morphological traits, most likely synonyms: ‘Tamopan’ and ‘Mopanshi’, ‘Gosho’ and cultivars in group 18, and ‘Misatogosho’ and cultivars in group 17. It would be possible that a clonal individual derived from the original one was named differently and independently accumulate nucleotide mutations that do not alter phenotype significantly. The fact that the above cultivars shared identical genotypes in the ssrdk markers, but not in the AST locus-linked marker, is consistent with the high discrimination power of the AST locus-linked marker. We previously revealed that polymorphisms in the AST fragment consist of AT repeats, indels, and SNPs (Kono et al. 2016), and here we observed that polymorphisms in the ast fragment consisted of AT repeats and SNPs. These results suggest that the AST locus-linked region has a high mutation rate. In practical germplasm identification, breeders would be interested in verifying the identity of a germplasm with the original cultivar. Partly because persimmons are relatively new to non-Asian countries, germplasm exchange and frequent bud mutation have resulted in mislabeling and confusion of persimmon cultivars (Badenes et al. 2003; Yonemori et al. 2000). Under these conditions, highly polymorphic and cost-effective DNA markers would be an ideal tool for cultivar identification. Distinguishing the size differences of AST and ast alleles by using a capillary sequencer is reproducible and simple. The allele data presented in this study, which covers many persimmon accessions, should work well for screening target accessions. In contrast to the AST allele, the ast allele revealed only five different sizes, suggesting that it is of more recent origin. Low polymorphism of the ast allele limits discrimination of PCNA accessions by the AST locus-linked marker, and thus addition of other markers, such as ssrdk markers, is indispensable. The genotyping data for each of the six ssrdk markers presented in Supplemental Table S2 are readily available.

There are many synonym groups among persimmon genetic resources (Agricultural Research Station 1912; Yonemori et al. 2000). Here, by using the AST locus-linked marker and the six ssrdk markers, we identified several possible synonyms, most of which were suggested as synonyms in previous studies based on morphological traits and isozyme analysis (Fruit Tree Experiment Station of Hiroshima Prefecture 1979; Sugiura et al. 1990; Tao and Sugiura 1987; Tao et al. 1989). Based on AFLP analysis, Yonemori et al. (2008) reported high similarity between Italian accessions ‘Brazzale’, ‘Moro’, ‘Rispoli’, and the Japanese accession ‘Zenjimaru’ (group 4) and between the Japanese accession ‘Amahyakume’ and the Italian accession ‘Kaki Tipo’ (group 3). Since many Italian cultivars are likely to have been imported directly or indirectly (through North America) from Japan in the nineteenth century (Bellini and Giordani 2005), some Italian accessions could be categorized into the same group as the original Japanese accession. We here newly report the following possible synonyms: ‘Chung Nam’ and ‘Okayamaokugosho’ (group 6), ‘Edoichi’ and ‘Kurokuma’ (group 7), ‘Kubogataobishi’ and ‘Shoujyou’ (group 11), ‘Mizushimagosho-Italy’ and ‘Shogatsu’ (group 12), ‘Yamatogaki’ and ‘Yamatohyakume’ (group 13), and ‘Giant Fuyu’ and ‘Mikado’ (group 16). Similar morphological traits within the above groups support the notion that group members are synonyms to each other (data not shown). Except for bud sports, no accession was categorized into a group from which its morphological features differed substantially. All known bud mutants of ‘Hiratanenashi’, ‘Fuyu’, and ‘Jiro’ were categorized into the same group as the original cultivar (groups 8, 14, and 19, respectively).

Our results provide insights into the origin of the ast allele. Polymorphisms in ast alleles have been reported (Akagi et al. 2010; Akagi et al. 2012; Kono et al. 2016), but there has been no systematic analysis of the distribution and sequences of these alleles among persimmon genetic resources. Here, we found five ast fragments of different sizes among 237 accessions; ast alleles were found not only in Japanese accessions, but also in some accessions from China, Korea (Akagi et al. 2010), and other countries. Excluding accessions with the same allele composition at the AST locus-linked marker and six ssrdk markers as Japanese accessions, Chinese local cultivars in particular had ast alleles; the ast fragment sizes in the Chinese cultivars were either 349, 351, or 353, which is the same as those found in Japanese cultivars. Our finding that the sequences of the a349 allele of 24 accessions (19 Japanese and 5 Chinese) were identical suggests that the a349 allele has a common origin in these accessions. The presence of a353-1 in ‘Huo-shi’ (China), ‘Harbiye’ (Turkey), and ‘Triumph’ (Israel) as well as in six Japanese accessions also implies that a353-1 shares the same origin in these accessions.

Depending on sequence similarities of the ast fragment, the eight ast alleles found in this study could be categorized into two groups: the a349 group (a347 and a349) and the a353 group (a351, a353-1, a353-2a, a353-2b, a353-2c, and a355). We did not find any ast alleles that had a moderately similar sequence to both the a349 and a353 group; therefore, future work to finding the missing link between these two allelic groups is required. In conjunction with the two separated groups of ast alleles, the limited existence of the a349 and a353 groups in the four Japanese local cultivars with ancient origin suggests that a bottleneck event occurred during ast allele development. A possible scenario is that ast alleles first arose in China and then developed several mutations and that Chinese cultivars with ast allele(s) from the a349 and a353 groups were by chance introduced to Japan. The same sequence for a349 in Japanese and Chinese accessions, and for a353-1 in Japanese and Chinese accessions, is consistent with this scenario. To assess this notion, we need to screen many Chinese accessions and characterize their ast alleles. Fragment analysis of the AST locus-linked marker would be useful to screen for ast allele-containing accessions from among the more than 900 local non-PCNA cultivars in China (Wang et al. 1997).

In this study, we present comprehensive data for the minimum number and characteristics of AST and ast alleles in 237 persimmon accessions. Based on allelic size polymorphisms, we identified non-PCNA accessions with the larger than average numbers of ast alleles. These accessions are important genetic resources that could be used to accelerate PCNA breeding using our pseudo-backcross strategy and to extend the genetic background of PCNA cultivars to prevent inbreeding. In addition, we have demonstrated that the displayed allele data is useful for identification of non-PCNA cultivars. Fragment size and direct sequence analysis of ast fragments suggest that ast alleles likely originated in China. We conclude that the data presented here will help breeders select non-PCNA parents with more ast alleles to more efficiently breed PCNA offspring and that the data provides new insights into how the ast allele has developed.