Genome-wide indel markers shared by diverse Asian rice cultivars compared to Japanese rice cultivar 'Koshihikari'.

Insertion-deletion (indel) polymorphisms, such as simple sequence repeats, have been widely used as DNA markers to identify QTLs and genes and to facilitate rice breeding. Recently, next-generation sequencing has produced deep sequences that allow genome-wide detection of indels. These polymorphisms can potentially be used to develop high-accuracy polymerase chain reaction (PCR)-based markers. Here, re-sequencing of 5 indica, 2 aus, and 3 tropical japonica cultivars and Japanese elite cultivar 'Koshihikari' was performed to extract regions containing large indels (10-51 bp) shared by diverse cultivars. To design indel markers for the discrimination of genomic regions between 'Koshihikari' and other diverse cultivars, we subtracted the indel regions detected in 'Koshihikari' from those shared in other cultivars. Two sets of indel markers, KNJ8-indel (shared in eight or more cultivars, including 'Khao Nam Jen' as a representative tropical japonica cultivar) and C5-indel (shared in five to eight cultivars), were established, with 915 and 9,899 indel regions, respectively. Validation of the two marker sets by using 23 diverse cultivars showed a high PCR success rate (≥95%) for 83.3% of the KNJ8-indel markers and 73.9% of the C5-indel markers. The marker sets will therefore be useful for the effective breeding of Japanese rice cultivars.

Introduction

Japanese lowland rice cultivars belong to the temperate japonica group, and have specific phenotypes that are well-adapted to particular environmental conditions in Japan. Numerous new cultivars with improved yield and eating quality have been developed by changing the haplotypes in the rice genome through breeding (Yamamoto , Yonemaru ). Recently, climate change has forced us to look for ways to increase the genetic resistance and/or tolerance of rice to both biotic and abiotic stresses. To this end, it is necessary to uncover the most effective alleles for improving agronomic traits, and to develop a wide range of genetic resources for use in breeding. Quantitative trait locus (QTL) analysis and subsequent introduction of a new allele by marker-assisted selection is one way to facilitate improvement of rice cultivars (Miura , Yamamoto ). The introgression of a favorable allele found in a diverse Asian rice population into Japanese rice cultivars has been widely used to improve elite Japanese cultivars (http://www.naro.affrc.go.jp/genome/database/ine/index.html). Because of almost even genetic distances from each accession (Kojima , Shomura ), ten cultivars (five from the indica group: ‘Bei Khe’, ‘Bleiyo’, ‘Deng Pao Zhai’, ‘Nona Bokra’, and ‘Qiu Zhao Zhong’; two from the aus group: ‘Kasalath’ and ‘Tupa 121-3’; and three from the tropical japonica group: ‘Basilanon’, ‘Khao Nam Jen’, and ‘Khau Mac Kho’) were used as donors to produce chromosomal segment substitution lines (CSSLs) in our rice improvement project (Fukuoka ). Such regional variations in the rice genome may reflect functional variations of the genes located in these chromosomal regions. CSSLs make it easier to detect weak allelic effects due to weaker genetic interactions, and lines with favorable phenotypes can be used as near-isogenic lines for breeding.

DNA markers are important tools for genetic dissection and improvement of target traits, and many types of DNA markers have been developed for rice genomes. After completion of the rice genome sequencing project, the sequence data was used to develop simple sequence repeat (SSR) markers (McCouch ). SSR markers are particularly useful because they are frequently multi-allelic, usually contain polymorphic insertion-deletion (indel) sites, and can be easily detected by polymerase chain reaction (PCR) and gel electrophoresis (i.e., without the need of a specific detection apparatus). Next-generation sequencing (NGS) has provided sequence data for multiple rice cultivars. Several Japanese rice cultivars have been also re-sequenced to provide genome-wide single-nucleotide polymorphisms (SNPs) for the analysis of genome constitutions in the Japanese rice population (Arai-Kichise , 2014, Nagasaki , Takano , Yamamoto , Yonemaru ). If the flanking sequences of an SNP allele have a recognition site for a restriction enzyme, the SNP can be detected by using a cleaved amplified polymorphic sequence (CAPS) assay; however, many SNP alleles require complex detection methods, such as derived CAPS and PCR with confronting two-pair primers (Hamajima 2001). High-throughput genotyping for large numbers of SNP markers has been achieved by using a high-density array (Appleby ), and genome-wide SNPs can be located and genotyped by using NGS (Spindel ).

Without abundant SNP markers, a relatively small number of indel markers would be enough to discriminate genotypes for rough mapping of a QTL, and marker-assisted introgression of the QTL into recipient cultivars. SSR markers have been widely used for this purpose, but they require screening for polymorphisms between each pair of parental cultivars due to unpredictable indel sizes, and small indels (≤5 bp) are difficult to discriminate by gel electrophoresis. Therefore, indels that have a predictable size and substantial polymorphic region (≥10 bp) are ideal for genotyping. NGS has not only provided SNPs but also indels for multiple rice cultivars. However, large indels (10–100 bp) have rarely been verified for marker use.

Such indels would be promising targets for the design of PCR-based markers to be used for QTL analysis between Japanese rice cultivars and diverse Asian rice cultivars. Additionally, these indel markers would be available for the introgression of a favorable genome segment derived from diverse Asian cultivars into Japanese rice cultivars. To enhance the success rate in PCR amplification and the detection of polymorphisms between Japanese rice cultivars and diverse Asian cultivars, we need information about indels with large polymorphisms that are shared by many diverse Asian cultivars compared to Japanese rice cultivars.

Here, we aimed to establish indel markers that can be used to discriminate between the genome regions of the Japanese elite cultivar ‘Koshihikari’ and 10 rice cultivars as described previously. By mapping NGS re-sequencing data and the lowland cultivar (temperate japonica) ‘Koshihikari’, indel regions that were common to many diverse rice cultivars but not ‘Koshihikari’ were detected and experimentally validated. All indel information is summarized in the supplementary materials.

Materials and Methods

Detection of genome-wide indel regions by means of Illumina short-read sequencing

To detect indel regions across the rice genome, we selected 10 cultivars for short-read Illumina re-sequencing. We have already obtained re-sequencing data for ‘Koshihikari’ (Yamamoto ), but we obtained new re-sequencing data to maintain the quality of the sequence information.

Genomic DNA was extracted from leaves of each cultivar by the CTAB method (Murray and Thompson 1980). By using the Illumina HiSeq 2000 system (Illumina Inc, CA, USA), Illumina short reads for each cultivar were obtained with total sequence lengths ranging from 14.2 to 26.1 Gbp (Table 1). Low-quality bases (phred quality score, Rice Genome Sequencing Project (IRGSP) v.1 reference genome by using version 0.6.2 of the BWA software with its default settings (Li and Durbin 2009). Only uniquely mapped reads with a mapping quality score of ≥20 were sorted and indexed, by using version 0.1.18 of the SAMtools software (Li ). To improve the raw alignment around indels, local re-alignment was performed by using version 1.5.0 of the GATK software (DePristo ). PCR duplicates were removed by using version 1.6.3 of the Picard software (http://picard.sourceforge.net). Indels and SNPs were identified individually for each sample by using the mpileup algorithm of SAMtools, with a filtering threshold (Q = 20).

Table 1

Sequence information for the 11 cultivars obtained by re-sequencing analysis using Illumina short reads

Cultivar	Category	Sequencing data from all reads		Mapped sequence data within the genome			Mapped sequence data within the genic regions
		Total nucleotides	Depth (fold)	Total bases	Coverage (%)	Depth (fold)	Total bases	Coverage (%)	Depth (fold)
‘Basilanon’	tropical japonica	17,740,676,971	45.9	12,738,101,906	85.2	40.1	1,995,801,715	95.2	38.7
‘Bei Khe’	indica	16,631,917,251	43.0	11,080,842,144	81.4	36.5	1,832,672,990	94.0	36.0
‘Bleiyo’	indica	22,550,774,596	58.4	10,983,117,212	81.5	36.1	1,741,246,934	93.5	34.4
‘Deng Pao Zhai’	indica	17,936,229,769	46.4	10,754,343,950	81.6	35.3	1,727,564,789	93.9	34.0
‘Kasalath’	aus	14,219,078,182	36.8	8,696,383,968	81.8	28.5	1,501,708,865	94.3	29.4
‘Khao Nam Jen’	tropical japonica	21,006,070,809	54.4	15,194,549,019	90.3	45.1	2,285,688,311	97.2	43.4
‘Khau Mac Kho’	tropical japonica	22,905,210,790	59.3	17,688,217,567	88.9	53.3	2,634,014,052	96.0	50.6
‘Koshihikari’	temperate japonica	19,910,363,512	51.5	14,205,341,196	93.5	40.7	2,459,122,055	98.2	46.2
‘Nona Bokra’	indica	26,101,851,416	67.6	13,343,845,531	82.6	43.3	2,124,784,352	94.2	41.6
‘Qiu Zao Zhong’	indica	17,301,402,545	44.8	10,666,117,459	81.0	35.3	1,559,351,856	92.7	31.0
‘Tupa 121-3’	aus	16,752,431,002	43.4	11,151,194,610	82.2	36.4	1,716,190,899	94.1	33.7

Primer design for common indel markers

To design indel markers for detection of indel polymorphisms by electrophoresis, we extracted only indel regions with both a large size (≥10 bp) and a high sequencing depth (DP, ≥5 fold) from each indel list for the 11 cultivars. Because these indel regions also included SSR polymorphisms, some coincided with the positions of published SSR markers (International Rice Genome Sequencing Project 2005, McCouch 2002). Primer pairs for the selected indel regions were automatically designed by using a Perl script to control the Primer3 core program (Rozen and Skaletsky 2000). PCR product size ranged from 80 to 150 bp.

We screened primer pairs for duplication of sequences to maintain specificity. When the sequence of a primer pair matched that of another primer pair, the corresponding pairs were eliminated because they were considered redundant. If either sequence of a primer pair included all or a part of the masked sequences, it was classified as having a low uniqueness and eliminated. The indel markers were retained after elimination of primer pairs with the same marker positions as those of ‘Koshihikari’.

To demonstrate the general utility of these indel variations, we compiled two sets of indel markers with different criteria. The first set (KNJ8-indel) was developed mainly to distinguish chromosomal segments derived from tropical japonica cultivars from those of Japanese lowland rice cultivars. In particular, we mainly intended to discriminate the genomic regions between lowland rice cultivars and upland rice cultivars genetically close to tropical japonica. This indel marker set was based on indel regions common to at least eight cultivars, including the tropical japonica cultivar, ‘Khao Nam Jen’, but distinct from the temperate japonica cultivar, ‘Koshihikari’, and consisted of 915 indel markers (Supplemental Table 1). The second set (C5-indel) was established to distinguish chromosomal segments of lowland rice cultivars (temperate japonica) from all other cultivars except temperate japonica. These indel sets were aimed at discriminating genomic segments among the genetically diverse rice cultivars. The indel regions that were common to five to eight cultivars but were distinct from ‘Koshihikari’ were selected for a total set of 9,899 indel markers (Supplemental Table 2).

Validation of the designed indel markers

To validate the two marker sets, we tested the primer pairs for 120 KNJ8-indel markers and 785 C5-indel markers by means of gel electrophoresis. Markers were randomly chosen so as to cover as much as possible of the entire genome, but were randomly chosen within each portion of the genome. The indel primer sets were validated by using a survey population consisting of 23 cultivars (Table 2), 11 of which were the cultivars used for detection of the indel regions. All 23 cultivars were donor or recipient cultivars of CSSLs or other analytical populations from our previous research (Fukuoka ). Extraction of genomic DNA from these cultivars followed the same protocol that we used for re-sequencing.

Table 2

List of the 23 cultivars used for validation of the indel markers

Name	Category	Origin
‘Basilanon’	tropical japonica	the Philippines
‘Bei Khe’	indica	Cambodia
‘Bleiyo’	indica	Thailand
‘Davao 1’	indica	the Philippines
‘Deng Pao Zhai’	indica	China
‘Hayamasari’	temperate japonica	Japan
‘Hitomebore’	temperate japonica	Japan
‘IR64’	indica	the Philippines
‘Kasalath’	aus	India
‘Khao Nam Jen’	tropical japonica	Laos
‘Khau Mac Kho’	tropical japonica	Vietnam
‘Koshihikari’	temperate japonica	Japan
‘LAC 23’	tropical japonica	Nigeria
‘Muha’	aus	Indonesia
‘Naba’	indica	India
‘Nipponbare’	temperate japonica	Japan
‘Nona Bokra’	Indica	India
‘Owarihatamochi’	tropical japonica	Japan
‘Qiu Zhao Zhong’	Indica	China
‘Silewah’	tropical japonica	Indonesia
‘Takanari’	Indica	Japan
‘Toboshi’	Indica	Japan
‘Tupa 121-3’	Aus	Bangladesh

The 120 primer sets from KNJ8-indel were PCR amplified in approximately 5 μL of reaction mixture consisting of the master mix from the KAPA2G Fast PCR Kit (Nippon Genetics Co. Ltd., Japan) containing 1.7 to 3.3 pmol of each primer, and about 2 ng of the genomic DNA template. The conditions were an initial denaturation step for 1 min at 95°C, then 35 cycles of 10 s at 95°C, 10 s at 55°C, and 1 s at 72°C, followed by a final extension for 30 s at 72°C. The 785 primer sets from C5-indel were PCR amplified in approximately 8 μL of reaction mixture consisting of a GoTaq Colorless Master Mix (Promega, USA), 3.2 pmol of each primer, and a low amount of genomic DNA added with a plastic pin. The conditions were an initial denaturation step for 5 min at 94°C, then 35 cycles of 15 s at 94°C, 30 s at 55°C, and 30 s at 72°C, followed by a final extension for 2 min at 72°C. PCR products were analyzed by means of gel electrophoresis on 3–4% agarose gels in Tris/borate/EDTA buffer. After staining with ethidium bromide, the band patterns on the agarose gels were photographed under ultraviolet light for verification of indel polymorphisms. We classified the band patterns into four categories based on the number of bands: 0, no bands; 1, a band similar to the reference allele (‘Nipponbare’ IRGSP v.1); 2, an alternative allele type; 3, an unexpected allele type; and 4, multiple bands.

Results

Number of indel regions detected by NGS

The sequence depths estimated from sequencing data for all reads of the 11 cultivars ranged from 36.8 (‘Kasalath’) to 67.6 (‘Nona Bokra’) (Table 1). The mapped sequence depths were lower than those of the total sequence, and ranged from 28.5 (‘Kasalath’) to 53.3 (‘Khau Mac Kho’). The highest and lowest coverages of a reference genome by input reads were observed in ‘Koshihikari’ (93.5%) and ‘Qiu Zhao Zong’ (81.0%), respectively. The genic regions that were annotated by RAP-DB (Sakai ) had higher coverage (92.7 to 98.2%) than the overall genome (Table 1).

The total number of indels, including heterozygous alleles, detected in the 11 cultivars was less than one-fifth of the number of SNPs (Table 3, Supplemental Table 3). The number of indels ranged from 24,898 for ‘Koshihikari’ to 341,433 for ‘Bei Khe’, with intermediate values of ca. 100,000 for ‘Khao Nam Jen’ and ‘Khau Mac Kho’. ‘Koshihikari’ had very few indels compared to the other cultivars. The number of selected indels was about one-tenth of the total number of indels in all 11 cultivars.

Table 3

Number of indels compared to the reference genome (‘Nipponbare’ IRGSP v.1) detected in the 11 cultivars

Cultivar	Category	Indels	Indels (selected)
‘Basilanon’	tropical japonica	234,431	22,815
‘Bei Khe’	indica	341,433	32,514
‘Bleiyo’	indica	336,533	34,069
‘Deng Pao Zhai’	indica	334,157	32,953
‘Kasalath’	aus	334,087	29,871
‘Khao Nam Jen’	tropical japonica	107,797	9,462
‘Khau Mac Kho’	tropical japonica	125,815	11,755
‘Koshihikari’	temperate japonica	24,898	1,648
‘Nona Bokra’	indica	347,557	35,921
‘Qiu Zao Zhong’	indica	324,021	31,215
‘Tupa 121-3’	aus	323,633	29,388

Number and distribution of common indel markers

As a result of designing primers, about half of the indel regions were eliminated. Few indel markers were detected between ‘Koshihikari’ and ‘Nipponbare’ (752 markers), so the removal of these markers did not greatly influence the number of indel markers. In total, we selected 121,248 indel markers (with redundancy) and 36,766 indel markers (without redundancy) for the 10 other cultivars (i.e., non-’Koshihikari’ cultivars). In terms of marker efficiency, the available markers to detect indels that were common to all 10 cultivars but not ‘Koshihikari’ were effective but few (283 markers, Fig. 1). The largest number was observed for solitary indel regions, which represent indel markers common to ‘Koshihikari’ and only one other cultivar (12,198 markers). The number of indel markers decreased as the number of cultivars sharing the same primer sequences (termed here “common cultivars”) increased (Fig. 1).

Fig. 1

Numbers of indel markers shared between various numbers of cultivars. Common cultivars are defined as those sharing the same primer sequences for the indel markers.

The KNJ8-indel markers had an uneven distribution within the rice genome (Fig. 2A). In particular, markers were scarce in the long arm of chromosome 8, and in the short arms of chromosomes 11 and 12. In contrast, the C5-indel set was distributed more evenly throughout the rice genome (Fig. 2B).

Fig. 2

Distribution of the two indel marker sets, (A) KNJ8-indel and (B) C5-indel, within the rice genome. The horizontal bars in each chromosome show the positions of all indel markers (KNJ8-indel, total number = 915; C5-indel, total number = 9,899). Bars that extend to the left of the chromosome indicate the indel markers that were selected for use as validation markers (KNJ8-indel, validation number = 120; C5-indel, validation number = 785).

Experimental validation of the two indel marker sets

As an example of validation, clear single bands were detected around the expected PCR product sizes for two markers in the KNJ8-indel set: KNJ8-indel83 and KNJ8-indel335. The size differences between the amplified fragments could be discriminated among the 23 cultivars (Fig. 3). We used these categories to genotype all markers for the validation sets for the KNJ8-indel and C5-indel markers; the resulting data are presented in Supplemental Tables 4, 5, respectively.

Fig. 3

Electrophoretograms for two indel markers from the KNJ8-indel marker set: KNJ8-indel83 and KNJ8-indel335. M, Molecular marker (a 100-bp DNA ladder). Cultivars: 1, ‘Nipponbare’; 2, ‘Koshihikari’; 3, ‘Hayamasari’; 4, ‘Owarihatamochi’; 5, ‘LAC 23’; 6, ‘Khao Nam Jen’; 7, ‘Khau Mac Kho’; 8, ‘Deng Pao Zhai’; 9, ‘Naba’; 10, ‘Bleiyo’; 11, ‘Qiu Zhao Zong’; 12, ‘Bei Khe’; 13, ‘Koshihikari’; 14, ‘Basilanon’; 15, ‘Muha’; 16, ‘Tupa 121-3’; 17, ‘Toboshi’; 18, ‘Davao 1’; 19, ‘IR64’; 20, ‘Nona Bokra’; 21, ‘Kasalath’; 22, ‘Takanari’; 23, ‘Silewah’; 24, ‘Hitomebore’.

The availability of two sets of indel marker in different cross combination types of cultivar groups is summarized in Table 4. The percentages of the markers with a success rate of 95% or more for PCR were 83.3% and 73.9% in the KNJ8-indel and C5-indel marker sets, respectively. The percentage for the KNJ8-indel set with different genotypes between ‘Nipponbare’ and ‘Koshihikari’ (9.2%) showed a similar percentage with that of the C5-indel set (12.5%). A polymorphism between ‘Nipponbare’ and ‘Khao Nam Jen’ could be detected for 78.3% of the markers in the KNJ8-indel set, whereas the C5-indel set had a lower percentage of different alleles (11.3%) than in the KNJ8-indel set. A similar difference was observed for ‘Nipponbare’ vs. tropical japonica (60.8% for KNJ8-indel versus 10.6% for C5-indel). The KNJ8-indel and C5-indel sets had more similar percentages of polymorphisms (81.7% and 62.9%, respectively) for ‘Nipponbare’ vs. aus than those for ‘Nipponbare’ vs. tropical japonica. In ‘Nipponbare’ vs. indica, relatively high percentages of polymorphisms were observed in both marker sets (85.8% for KNJ8-indel versus 74.9% for C5-indel). A comprehensiveness index (CI), an index of the availability of markers for detection of allelic polymorphisms between the Japanese rice cultivars belonging to the temperate japonica group and cultivars belonging to other cultivar groups, showed that the KNJ8-indel set had higher availability (45.8%) than the C5-indel set (7.0%).

Table 4

Summary statistics for the validation of the two indel marker sets. Values in brackets represent the percentage (%) of the total

	Success rate (>95%)a	‘Nipponbare’ vs ‘Koshihikari’b	‘Nipponbare’ vs ‘Khao Nam Jen’c	‘Nipponbare’ vs tropical japonicad	‘Nipponbare’ vs ause	‘Nipponbare’ vs indicaf	CIg
KNJ8-indel	100 (83.3)	11 (9.2)	94 (78.3)	73 (60.8)	98 (81.7)	103 (85.8)	55 (45.8)
C5-indel	580 (73.9)	98 (12.5)	89 (11.3)	83 (10.6)	494 (62.9)	588 (74.9)	55 (7.0)

The number and percentage of markers with a PCR success rate of 95% or more (i.e., PCR without null or multiple bands).

The number and percentage of markers showing different genotypes between ‘Nipponbare’ and ‘Koshihikari’.

The number and percentage of markers showing different genotypes between ‘Nipponbare’ and ‘Khao Nam Jen’.

The number and percentage of markers showing different genotypes between ‘Nipponbare’ and three or more of the six tropical japonica cultivars.

The number and percentage of markers showing different genotypes between ‘Nipponbare’ and two or more of the three cultivars of the three aus cultivars.

The number and percentage of markers showing different genotypes between ‘Nipponbare’ and five or more of the 10 indica cultivars.

The comprehensiveness index (CI) of the KNJ8-indel and C5-indel markers requires that the conditions in footnotes a, c–f, and d–f be satisfied and that there is the same genotype between ‘Nipponbare’ and ‘Koshihikari’.

The relationships between the CI of the indel markers and the number of common cultivars are illustrated in Fig. 4. The CI of the C5-indel set, for the five to eight common cultivars, ranged from 3.8 to 10.4%. The CI of the KNJ8-indel set was higher, with values of 34.8%, 43.9%, and 63.6% for 8, 9, and 10 common cultivars, respectively. The chi-square test for differences in CI values among the various numbers of common cultivars was significant for both marker groups (both P < 0.05).

Fig. 4

Relationships between the comprehensiveness index (CI) of the indel markers (see Table 4) and the numbers of common cultivars in the two indel marker sets, KNJ8-indel and C5-indel. Common cultivars are defined as those sharing the same primer sequences for the indel markers.

Discussion

When detecting nucleotide polymorphisms, including indels, by using a re-sequencing strategy with NGS, high accuracy depends on the precision of the reference genome. Fortunately, in the rice genome project, the Sanger sequencing method was used to sequence a minimum set of genome segments across the entire ‘Nipponbare’ genome and then precise reference sequences were established (International Rice Genome Sequencing Project 2005). For re-sequencing with short reads, the sequencing depth affects the accuracy of polymorphism detection. Although both the sequencing read length and the number of end sequences (single-end or paired-end) were different from those in our study, sequencing depth reached a plateau of 15-fold in ‘Koshihikari’ in a previous study (Yamamoto ). Here, the sequencing depth of the mapped sequences was at least 28.5-fold (Table 1), which is twice the average depth (~14-fold) achieved in the most current sequencing project, “The 3000 Rice Genomes Project” (The 3000 Rice Genomes Project 2014). The results of our study suggest that this sequencing depth is sufficient to accurately detect nucleotide polymorphisms. Indeed, it appears there is no relationship between the genome coverage and the sequencing depth of the 11 cultivars (Table 1). The differences in genome coverage among the 11 cultivars reflect their similarity to the ‘Nipponbare’ reference sequence, and this similarity was coincident with the genetic distance in the phylogenetic tree in other studies (Fukuoka , Kojima ). Three cultivars, ‘Koshihikari’, ‘Khao Nam Jen’, and ‘Khau Mac Kho’, which belong to the japonica group (temperate and tropical japonica), showed higher coverage than the other cultivars, which belong to the aus and indica groups. ‘Basilanon’, which is a large genetic distance from other japonica cultivars, was intermediate in coverage between japonica and aus or indica. As expected, the number of polymorphic sites (SNPs and indels) in the genomes of the 11 cultivars showed the opposite trend to the genome coverage of them (Table 1, Supplemental Table 3).

Because re-sequencing with short reads cannot physically detect indels that are longer than the reads themselves, we only found insertions and deletions of less than 35 bp and 92 bp, respectively (data not shown). Indels with a larger size (>100 bp) are undoubtedly located all over the genome. Copy number variations, which are defined as very large indels (>1 kb) (Feuk ), have been found at high recombination positions and are thought to have been recently generated within breeding populations (Yu ). The construction of pseudomolecule sequences of ‘Kasalath’ also uncovered 7,393 large indels (>100 bp) between ‘Kasalath’ and ‘Nipponbare’ (Sakai ). The development of new sequencing technology, such as single-molecule, real-time sequencing technology, which allows us to read lengths of 5 kb or longer (Eid ), will facilitate the detection of large indels and the establishment of indel markers at these regions.

Several systems have been used in other studies for the development of indel markers in rice: 4–40 bp indels in 120–480 bp PCR amplicons separated on 8% denaturing polyacrylamide gels (Zeng ), 30–100 bp indels in 100–350 bp PCR amplicons separated on 3% agarose gels (Wu ), and 51–2000 bp indels in PCR amplicons of up to ten times the size separated on 3% agarose gels (Yamaki ). In the two latter cases, by using 3% agarose gels for the electrophoresis, indel size was set at 10% or more of the PCR amplicon size. Here, we found that indels of ≥10 bp in PCR amplicons of ~100 bp could be easily detected by separation on 3–4% agarose gels, and we designed the markers accordingly. Indeed, we observed a high PCR success rate for the two marker sets (Table 4), and the indel markers could be clearly used to discriminate allelic differences by electrophoresis.

As expected, there was a tendency toward increased CI as the number of cultivars sharing the markers increased in both sets of markers (Fig. 4, Table 4). For the C5-indel marker set, we observed 7.0% CI in the 785 validated markers, which corresponds to 693 markers in the total of 9,899 designed markers. There was an average of 548 kb between the designed markers, which is sufficient for the identification of rough map positions of QTLs detected between diverse Asian rice cultivars and Japanese rice cultivars.

To take advantage of successive introgression, backcross breeding has been widely used to improve elite Japanese cultivars. DNA markers greatly increase the efficiency of backcross breeding because they eliminate the need to wait for plants to reach maturity before they reveal favorable or unfavorable phenotypes (Fukuoka ). The two indel marker sets established in our study are expected to help the introgression of the alleles derived from diverse rice cultivars into Japanese elite cultivars and to accelerate rice breeding in Japan.