Introduction

Rice is one of the world’s most important staple crops. Because rice has the smallest genome size (420 Mb) among major cereal crops, it is the most popular model plant for agronomic, genetic, and physiological studies. Therefore, rice is important for deciphering the molecular control of agronomical traits and improving seed production and quality.

Since the 1970s, rice production has more than doubled (130%) due to the green revolution in cultivation technology and breeding. Breeding programs are large in scale and tend to require several years; as a consequence, data collection can be time-consuming and tedious. Although breeding has played a crucial role in the past, future challenges, such as climate change, natural resource depletion, and increasing population, will create an even higher demand for improved rice varieties.

Recently, various genetic resources for the functional analysis of the rice genome have been rapidly established, including T-DNA and transposon-tagged rice mutant populations. Among the most important outcomes is that large numbers of genes have been functionally characterized, many of which are directly related to rice traits. Most of these advances have been achieved through T-DNA insertional mutagenesis [1, 2]. In total, 154,391 insertional flanking sequence tags (FSTs) have been identified in the rice genome using T-DNA [3, 4]. POSTECH RISD has produced approximately 100,000 T-DNA-transformed lines and 20,889 genes disrupted by T-DNA insertion [3, 5]. In addition, RMD (Rice Mutant Database) and CIRAD-INRA have applied the Agrobacterium-mediated T-DNA insertion method to generate 31,892 and 27,870 FSTs harboring 6641 and 3602 genes, respectively [2]. Although T-DNA insertion has been used in the functional analysis of rice genes, it has several limitations. One disadvantage of this method is the potential deletion or chromosomal rearrangement of surrounding genomic DNA [6]. Additionally, the use of T-DNA causes social concerns because of the potential adverse effects of GM foods on human health and environmental safety [7].

Transposon insertion is another major strategy to obtain a large number of insertional mutants. [8, 9]. Transposable elements are divided into two groups: DNA-type (class II elements) elements that catalyze excision and reinsertion, i.e., the cut-and-paste mode, and retrotransposons (class I elements) that function in copy-and-paste mode via an RNA intermediate. In the rice genome, 32 families of retrotransposons have been identified [10], and Tos17 is the most active among active elements. Tissue culture-induced activation of Tos17 has been a useful tool for insertional mutagenesis and the functional analysis of genes.

Only one to five copies of Tos17 are present in various cultivars of rice; moreover, it is inherited stably, with heavy DNA methylation occurring under normal growth conditions and transposition only occurring during prolonged tissue culture [11]. ‘Nipponbare’ (NP) has two almost identical copies of Tos17 on chromosomes 7 (chr.07) and 10 (chr.10), and the only differences are a 90-bp insertion and 6 point mutations on chr.10 (Fig. 1a). Only Tos17 on chr.07 is transpositionally active, whereas the other copy is inactive [12]. Tos17 tends to insert into genic regions [13], although little is known about the occurrence and diversity of Tos17 within Korean rice cultivars.

Fig. 1
figure 1

Comparison of Tos17 in various Korean domestic rice cultivars to Tos17 in Nipponbare, a structure of Tos17 on chromosomes 7 (chr.07) and 10 (chr.10). Long terminal repeats (LTRs) are symbolized by open arrows, and open reading frames (ORFs) are indicated by black bars. Six red dots and one box indicate the point mutations and 90-bp insertion in chr.10. b Amplified Tos17 from various Korean domestic rice cultivars

Full size image

In addition to facilitating functional analyses of rice, the development of endogenous mutants can not only overcome social concerns but also contribute to the development of useful agronomic traits that suit the cultural and geological environments of selected cultivar lines. In a previous study, Miyao et al. [8] analyzed approximately 42,292 flanking sequences from 4316 mutant lines produced from 5-month tissue cultures of NP. Of these 42,292 flanking sequences, 3536 genes were aligned by BLASTX searches. A total of 53 types of abnormal phenotypes were observed from the seedling to harvest stages of these mutants [14]. However, this number of mutants is not sufficient to provide reasonable coverage of the total predicted 33,239 genes to develop rice varieties. Until now, the production of Tos17 mutants has been limited to the study of NP.

In this study, we generated Tos17 mutants using the Korean domestic rice cultivars O. sativa L. japonica ‘Ilmibyeo’ (IM) and ‘Baegjinju1ho’ (BJJ1) through a 1-month tissue culture. IM, which is representative of white rice in the Korean market, was developed from the three-way cross of Milyang `96//Milyang `95/Seomjinbyeo. BJJ1, which represents brown rice, was developed from a cross of Ilpum (MNU)-10-2-GH1-3 and ‘Seoanbyeo’. We analyzed 7608 flanking sequences of the newly transposed Tos17 mutants and produced 1672 and 843 mutants (M2 generation) from IM and BJJ1, respectively. In addition, we collected phenotypic data for the discovery of agronomically important varieties and genes, including mutants with a higher seed yield.

Materials and methods

Generation of Tos17 mutants

Peeled rice (IM and BJJ1) seeds were sterilized in 70% alcohol for 2 min, shaken in 50% Clorox for 15 min and washed with distilled H2O four times. Calli derived from embryos were grown in 2N6 media containing 4 g of Chu media (Duchefa Biochemie B.V.), 2 g of casamino acids, 0.5 g of proline, 0.5 g of glutamine, 30 g of sucrose, 2 mg of 2,4-D, and 2.5 g of Gelrite per liter, pH 5.8, at 28 °C for 1 month in the dark. Each round, hard, light-yellow callus with a diameter of 1–3 mm was chosen and transferred into fresh 2N6 medium. After growing at 28 °C for 4 days in the dark, the calli were transferred to fresh MSR media containing 4.4 g of MS salt (Duchefa Biochemie B.V.), 0.5 g of MSE, 5 mg of kinetin, 1 mg of NAA, 30 g of sucrose, and 4 g of Gelrite per liter, pH 5.8, and grown at 28 °C for 1 month under 12 h light/12 h dark conditions until greening. Then, the greening calli were grown into whole plants on MS0 media containing 4.4 g of MS salt (Duchefa Biochemie B.V.), 30 g of sucrose, 2.5 g of Gelrite per liter, pH 5.8, at 28 °C for 1 month under continuous light. In total, approximately 15,000 M0 plants were regenerated on MS0 media and sampled for genomic DNA extraction.

Amplification of flanking regions of Tos17 by adaptor-ligation PCR

Plant genomic DNA was extracted and purified from young leaves using NucleoSpin Plant II (Macherey–Nagel GmbH & Co. KG) according to the manufacturer’s protocol. Genomic DNA (500 ng) was digested with 2 U of restriction enzyme at 37 °C for 1 h and ligated with adaptors (50 pmol) by 5 U T4 DNA ligase (Takara, Japan) at 16 °C for 1 h in a 20-µl reaction volume. The first PCR was conducted with 5 µl of digested and ligated mixture, 0.5 pmol A1 (5′-GCGTAATACGACTCACTATAGCAATTAACC-3′) and T1 (5′-TGCTCTCCACTATGTGCCCTCCGAGCTA-3′) primers, and PCR premixture (Solgent, Korea) in a 20-µl reaction mixture with the following protocol: initial denaturation step at 95 °C for 5 min, 20 cycles at 94 °C for 30 s and at 72 °C for 1 min, and a final elongation step at 72 °C for 10 min. Then, the second PCR was conducted with 5 µl of the first PCR product using the A2 (5′-GACTCACTATAGCAATTAAC-3′) and T2 (5′-ACAAGTCGCTGATTTCTTCAC-3′) primers with the following conditions: an initial denaturation step at 94 °C for 5 min; 40 cycles of 94 °C for 30 s, 60 °C for 30 s, and 72 °C for one minute; and a final elongation step at 72 °C for 10 min. Amplified products were loaded on a 1% agarose gel, and the PCR products were purified using a HiYield Gel/PCR DNA Extraction Kit (RBC, Taiwan) and sequenced using an ABI3730XL system and the T2 primer.

Analysis of the FSTs of Tos17

The FSTs of Tos17 mutants were analyzed with the FSTVAL web tool [13]. We considered only alignments with an E-value of 1e−5 using the IRGSP-1.0 Rice Genome Annotation Database (http://rapdb.dna.affrc.go.jp/).

M2 generation in a paddy field

M1 seeds of the Tos17 mutants were obtained in the greenhouse from 2010 to 2015, and twelve lines of M1 seeds were planted in a paddy field located in Jeonju (35°N, 127°E), Korea, from 2016 to 2017. The M2 seeds were harvested and weighed.

Results and discussion

Determination of endogenous Tos17 position

To confirm the position of endogenous Tos17 on chr.07 and chr.10 in various Korean domestic rice cultivars, we used a Tos17-flanking primer in either chr.07 (PF07) or chr.10 (PF10) and a Tos17-common primer (PR1) and a Tos17-specific primer (PR10) in chr.10 based on the NP genome sequence (Fig. 1a). We isolated each genomic DNA from leaves of IM, BJJ1, ‘Nakdong’, ‘Dongin 1 ho’, ‘Sindongjin’, ‘Samkwang’, ‘Chucheongbyeo’ and ‘Hwayeongbyeo’.

Subsequently, 712-bp and 912-bp fragments were isolated using the PF10 primer and either the PR10 or PR1 primer, respectively, from the genomic DNA of the Korean domestic rice cultivars examined. In addition, an 873-bp fragment of the genomic DNA was amplified with the PF07 and PR1 primers but not with the PF07 and PR10 primers (Fig. 1b). The results of this sequencing showed that many Korean cultivars, including IM and BJJ1, harbor at least 2 copies of Tos17 on chr.07 and chr.10.

Generation of Tos17 mutant lines

To produce Tos17 mutant lines from calli, we chose IM and BJJ1 as representatives of white and brown rice in the Korean market, respectively. Each callus was induced from seed for 1 month on 2N6 media and regenerated on MS0 media (Additional file 1: Fig. S1). To determine the fraction of these putative newly transposed Tos17 copies, we extracted genomic DNA from the leaves of the M0 generation. In addition, we amplified the Tos17 flanking region from genomic DNA digested with the restriction enzyme MspI using the 3′ primer of Tos17 (T1 and T2 primers) and an adaptor-specific primer (A1 and A2 primers) (Additional file 1: Fig. S1). All of the PCR products exhibited two common fragments at 513 and 588 bp, which were derived from chr.07 and chr.10, respectively (Additional file 1: Fig. S2). The PCR products amplified from newly inserted Tos17 sequences were isolated from the gels and sequenced with the T2 primer (Additional file 1: Fig. S2). In this study, Tos17 was newly transposed as a low-copy-number region in the genome of IM and BJJ1 (0 ~ 3 copies).

Furthermore, we isolated 7608 FSTs. When we sequenced the newly transposed FSTs, the long terminal repeat (LTR) region of Tos17 was found to be identical to that of chr.07. As shown in a previous report, we confirmed that Tos17 in chr.07 is active only in tissue culture [12].

By analyzing the flanking sequences of Tos17, we found that all 4.2-kb fragments of Tos17 with the LTR region at both ends were cleanly inserted into the rice genome. In contrast, T-DNA tagging, another method used for insertional mutagenesis, involves incomplete insertion. The border sequence of the T-DNA is often broken before insertion, and in some cases, the T-DNA backbone is inserted into the genome without cleavage [15].

Analysis of Tos17 flanking sequences

To analyze the 7608 FSTs, we used FSTVAL, which was previously developed as a web tool to manage bulk flanking sequence tags [13]. Among the analyzed FSTs, 6959 were aligned to the rice genome of IRGSP1.0, and 279 FSTs (4%) exclusively matched the repeat region (Table 1). The frequencies of Tos17 insertions in genic and intergenic regions were 70% and 26%, respectively (Table 1), which indicates that the insertion of Tos17 was more frequent in genic than in intergenic regions.

Table 1 Insertion frequencies of Tos17 in genic and intergenic regions
Full size table

The distribution of the Tos17 insertions on the 12 rice chromosomes is shown in Additional file 1: Fig. S3 and Additional file 2: Table S1. Overall, three chromosomes (chr.01, chr.02, and chr.03) had high densities of Tos17 insertions, and chr.10 and chr.12 had relatively low densities of Tos17 insertions (Additional file 2: Table S1). Tos17 preferentially inserted into the distal ends of chromosomes rather than the centers, whereas Tos17 rarely inserted into centromeric regions (Additional file 1: Fig. S3).

To regenerate Tos17 mutants from the calli of IM and BJJ1, we independently harvested M1 seeds in a greenhouse and cultivated 1672 IM and 843 BJJ1 mutants (M2 generation) (Additional file 2: Table S2). The frequencies of Tos17 insertions in genic regions were 76% (1562 FSTs) and 74% (730 FSTs) in IM and BJJ1, respectively (Table 1). Among 1495 genes with FSTs in their genic regions, 1147 (76.7%) and 199 (13.3%) genes contained one site and two sites of Tos17 insertion, respectively (Additional file 2: Table S3).

Burr et al. [16] have reported that gene densities are generally highest at the distal ends of the chromosome arms and that fewer ESTs are found in centromeric regions. Rice has an average gene density of one gene per 9.9 kb, especially on chr.01, chr.02, and chr.03, which have high gene densities of one gene per 8.9, 9.1, and 8.7 kb, respectively [17]. These results support our observation that Tos17 predominantly inserted into genic regions (Table 1), which is one of the advantages of using Tos17 for insertional mutagenesis.

We found that 1533 genes were tagged by Tos17 insertion from 2515 mutants of the IM and BJJ1 cultivars. In a previous study, Miyao et al. [8] analyzed 3536 genes representing 42,292 NP mutants. A comparison of the differences among the mutant groups revealed that 830 genes among the Tos17 mutants (54%) were found only in the IM and BJJ1 mutants and not in the NP mutants.

Until now, a total of 65 Tos17 mutants from NP have been studied in 42 published papers, among which 29 domestic mutants with 13 mutated genes were obtained as IM and BJJ1 mutants. We produced 6 independent lines in which the Phosphate2 (PHO2, Os05g0557700) gene was knocked out. Among these lines, Y17MJ353 mutants showed the lesion mimic phenotype, which was similar to the phenotype of pho2 mutants reported by Lorieux et al. [18]. We also produced a BJJ002B07 mutant in which Tos17 is inserted into a SULTR-like phosphorus distribution transporter (SPDT, Os06g0143700). A previous study [19] reported that the concentration of phytate in brown rice was 25–32% lower in spdt mutants than in wild-type rice. In addition, by observing the phenotypes of the Tos17 mutants under biotic or abiotic stress conditions, such as pathogen, drought, cold, or salt stress, it will be possible to develop new rice lines with tolerance to various stresses.

Functional classification of Tos17 mutants via MapMan analysis

To functionally characterize genes with genic-region FSTs, we categorized 1533 genes from the IM and BJJ1 mutants with MapMan [20], excluding repeated genes. We also obtained 3280 genes from NP mutants and analyzed them with MapMan [3]. After mapping, 953 and 2269 genes were assigned to different MapMan terms (bin) in the IM/BJJ1 and the NP mutants, respectively (Additional file 2: Table S4). We subsequently compared the patterns of the categorized genes in the IM/BJJ1 and NP mutants (Additional file 1: Fig. S4B and C). Overall, 159 genes in the IM and BJJ1 mutants were mapped to the metabolism overview, with 16 bins (Additional file 1: Fig. S4B, Additional file 2: Table S4). As shown in Additional file 1: Fig. S4, the genes presented similar patterns in the overview of both metabolism and cell functions. In addition, we showed that Tos17 insertions were distributed throughout the genes of the rice genome without bias based on the assigned gene ontology (GO) terms by AgriGO analysis (Additional file 1: Fig. S5). To identify Tos17 mutants that could be useful for new variety development, it is important to examine whether Tos17 insertions were maintained over successive generations.

To confirm the accuracy and stability of the next generation of Tos17 insertion lines, 22 lines were selected from the various groups categorized by MapMan and then evaluated by tissue-direct PCR conducted on leaves from each line. We selected three genes each from the signaling and transport categories. Four genes from each of the stress, RNA, DNA and protein groups were also selected and examined (Additional file 1: Fig. S6). PCR fragments were amplified with gene-specific primers within a 1-kb region from the Tos17 insertion site and a Tos17 primer targeted to 239 bp from the 3′ end of Tos17. The PCR product size from IM (wild type) was approximately 1 kb, which was approximately 240 bp larger than that from the Tos17 insertion lines (+/+) (Additional file 1: Fig. S6). The insertion position of Tos17 confirmed by adaptor-ligation PCR in the M0 generation was identical to the genotyping PCR results from the M2 generation. We found that the Tos17 insertions were maintained through the M2 generations and did not appear in new positions without tissue culture (Additional file 1: Fig. S6).

Classification of phenotypes in 1000 lines of Tos17 insertion mutants at the vegetative stage

Plant heights are classified into two types: “semi-dwarf” and “long culm”. The “semi-dwarf” condition is characterized by plant heights that are 70 ~ 80% of the wild-type heights [14]. Y17MJ030 and Y17MJ818 showed phenotypes with heights of 75 and 50 cm, respectively (Fig. 2a). In the Y17MJ030 and Y17MJ818 lines, Tos17 was inserted into the introns of Os02g0131800 (Trivalent AI influx transporter) and Os06g0725100 (Lipase), respectively, as confirmed by the adaptor-ligation PCR method (Additional file 1: Fig. S7). In addition, Y17MJ223, which was 65 cm tall, also showed the semi-dwarf phenotype and had Tos17 inserted into Os02g0823000, peptidase A22B (Fig. 2a and Additional file 1: Fig. S7). The Y17MJ1016 line showed the long culm phenotype (Fig. 2b) and a Tos17 insertion that was not found in the IRGSP1.0 database but matched the Oryza sativa IM scaffold 750_cov161 (unpublished).

Fig. 2
figure 2

Representative phenotypes of Tos17 insertional mutants, a semi-dwarf, b long culms, c narrow leaves, d pale green leaves, e striped leaves, f lesion mimic, g weak growth, h low tillering, i high tillering, j dense panicles, k early heading, l long awns

Full size image

We observed a narrow leaf shape (Y17MJ166), in which Tos17 was inserted 14.326 kb upstream from Os04g0441600 (Fig. 2c), which encodes a protein similar to Androgen-induced 1; a pale green leaf (Y17MJ391), in which Tos17 was inserted in an exon of Os02g0828100 (Fig. 2d), which encodes a BRO1 domain containing protein; a striped leaf (Y17MJ660), in which Tos17 was inserted 12-bp downstream of Os06g0111300, which encodes a hypothetical conserved gene (Fig. 2e and Additional file 1: Fig. S7); and a lesion mimic phenotype (Y17MJ070), in which Tos17 was inserted 3.525 kb upstream of Os11g0187150 (hypothetical protein) (Fig. 2f and Additional file 1: Fig. S7).

A weak growth phenotype with slim seedlings presenting retarded growth (Y17MJ917), in which Tos17 was inserted into an exon of Os06g0705350 (pentatricopeptide repeat-containing protein) (Fig. 2g). We also observed low (Y17MJ911) and high (Y17MJ677) tillering mutants that contained new Tos17 insertions in an exon of Os11g0617700 (DUF594 domain-containing protein) and 264 bp downstream of Os06g0115300 (Acyl-CoA-binding protein), respectively (Fig. 2h and i).

Classification of phenotypes at the reproductive stage

Dense panicles (Y17MJ040) were observed in mutants in which Tos17 was inserted into an intron of Os02g0673500, which encodes a bHLH domain protein (Fig. 2j). Early heading phenotype and flowering occurred approximately 20 days earlier than the wild type (Y17MJ377) (Fig. 2k). In the long seed awn phenotype (Y17MJ192) (Fig. 2l), Tos17 was inserted into an exon of Os05g0380300, which encodes an NBS-LRR protein (Additional file 1: Fig. S7).

Among the Tos17 mutants, 1000 mutants from IM were grown and harvested, and the 100-grain weight from each of these mutants was measured for 5 events in each line (Additional file 1: Fig. S8A). The average 100-grain weight across the 1000 mutants was 2.46 g, which was similar to the average 100-grain weight of the IM wild type. Among the mutants, 10 lines were selected, and these samples showed a weight increase of 17–56% (2.9–3.9 g) compared with that of the IM wild type (Table 2, Additional file 1: Fig. S8B). In the Y17MJ368-1, Y17MJ380-1, and Y17MJ192-1 mutant lines, Tos17 was inserted into the exons of Os01g0110100 (phosphate transporter), Os02g0796600 (esterase/lipase/thioesterase domain containing protein), and Os05g0380300 (similar to NBS-LRR protein) (Fig. 3). Y17MJ940-4, Y17MJ600-2, Y17MJ760-1, and Y17MJ459-4 contained the newly transposed Tos17 in the introns of Os02g0118800 (similar to NBS-LRR disease resistance protein), Os03g0101100 (similar to palmitoyl-protein thioesterase-like), Os01g0178700 (similar to protein binding protein), and Os06g0702700, encoding butirosin biosynthesis (Fig. 3), respectively. In addition, the Tos17 insertions of Y17MJ644-1 and Y17MJ112-4 were positioned 5′ upstream of Os05g0137200 (similar to MDR-like ABC transporter) and Os02g0730000 (similar to mitochondrial aldehyde dehydrogenase), respectively. The Y17MJ550-1 line included two Tos17 insertions in the exons of the Os04g0632400 and Os02g0131850 genes, both encoding hypothetical proteins (Fig. 3). In the ten Tos17 mutants, the 100-grain weight and area of dehulled grain were greater than those in IM (Table 2). However, the area density of the mutants was similar to that of IM, with values of 97–106% (Table 2). Additionally, the Y17MJ192-1 mutant showed a high area density (117% of the IM area density). The area density is the degree of compactness of the deposition of nutrients, which determines the grain weight and the area of grain space [21]. A high area density may not always be related to the yield, rate of seedling growth, or earliness of plant development; however, this parameter is useful for producing new rice varieties.

Table 2 Ten Tos17 mutant lines showing an increase in weight of 16% (2.9 g) compared to that of Ilmibyeo (wild type)
Full size table
Fig. 3
figure 3

Insertion positions of Tos17 and grain phenotypes of Tos17 insertional mutants, Open and filled boxes indicate non-coding and coding regions, respectively. The insertion sites are shown by red triangles. Scale bar indicates 5 mm

Full size image

After analyzing 7608 flanking sequences, 1672 and 843 mutants were produced from IM and BJJ1, respectively (M2 generation). As an example, we identified Tos17 insertion mutants with high seed yields from phenotypic data. The production of a large number of Tos17 insertion mutants with insertion site information is a powerful method that can be performed in a short period of time compared to breeding programs using random mutations, such as chemical and radiation mutagenesis. Until now, the Tos17 insertion mutants of rice have been developed using NP as resources for the functional analysis of genes. In this study, we demonstrated the potential use of these mutants not only for functional analysis of genes related to agricultural trails but also for developing new rice varieties using commercial cultivars of brown and white rice in Korea. Furthermore, considering the natural mutation in Tos17 insertion, mutants with elite traits can be directly used in breeding. Such mutants can be used as a parental resource for breeding to produce secondary varieties with improved traits.