Elite sd1 alleles in japonica rice and their breeding applications in northeast China

Hnjing Sh, Hulong Liu, Gungxin Zho, Zhongmin Hn, Huilin Chng, Jingguo Wng,Hongling Zheng, Jifeng Zhng, Yng Yu,e, Yuqing Liu, Detng Zou, Shoujun Nie, Jun Fng,f,*

a Key Laboratory of Soybean Molecular Design Breeding, Northeast Institute of Geography and Agroecology, Chinese Academy of Sciences, Harbin 150081, Heilongjiang, China

b Key Laboratory of Germplasm Enhancement, Physiology and Ecology of Food Crops in Cold Region, Ministry of Education, Northeast Agricultural University,Harbin 150030, Heilongjiang, China

c Innovation Center, Suihua Branch of Heilongjiang Academy of Agricultural Sciences, Suihua 152052, Heilongjiang, China

d University of Chinese Academy of Sciences, Beijing 100049, China

e College of Life Science and Engineering, Shenyang University, Shenyang 110044, Liaoning, China

f The Innovative Academy for Seed Design, Chinese Academy of Sciences, Beijing 100101, China

Keywords:Semi-dwarf breeding Elite allele Pedigree analysis Geographic distribution Regional adaptability

ABSTRACT The Green Revolution gene sd1 has been used extensively in modern rice breeding, especially in indica cultivars.However,elite sd1 alleles and related germplasm resources used for japonica rice breeding have not been identified, and extensive efforts are needed for japonica rice breeding to obtain new dwarfing sources.Data from MBKbase-Rice revealed seven sd1 haplotypes in indica and four in japonica rice.Two new sd1 alleles were identified in indica rice.In 295 japonica accessions from northeast Asia,except for the weak functional allele SD1-EQ, sd1-r was the major allele, reducing plant height in comparison with SD1-EQ. Japonica germplasm resources carrying reported sd1 alleles were identified by genotype searching and further verified by literature search, genealogical analysis, and dCaps markers. Pedigrees and geographic distribution showed that sd1-r is an excellent allele widely used in northern China and Tohoku in Japan, and sd1-j is commonly used in east China and Kyushu in Japan.Dongnong- and Xiushui-series cultivars carrying sd1-r and sd1-j,respectively,are essential branches of the backbone parents of Chinese japonica rice, Akihikari and Ce21, with the largest number of descendants and derived generations.In semi-dwarf japonica rice breeding, sd1-d was introgressed into Daohuaxiang 2 (DHX2).Dwarf and semi-dwarf lines carrying sd1-d were selected and designated as 1279 and 1280,respectively,after withstanding typhoon-induced strong winds and heavy rains in 2020,and are anticipated to become useful intermediate materials for future genetic research and breeding.This work will facilitate the introduction, parental selection, and marker-assisted breeding, and provide a material basis for the next step in identifying favorable genes that selected together with the sd1 alleles in japonica backbone parents.

1.Introduction

Rice is a staple cereal for more than half of the world’s population[1].In recent years,extreme weather incidence has increased,particularly typhoon frequency in northeast China.Rice lodging at the late maturity stage caused by typhoons has attracted much attention[2-4].In rice breeding, reducing plant height has always been the main goal for increasing lodging resistance [2].In the 1960s,the Green Revolution in rice led to unprecedented increases in rice production,saving the world from imminent famine[5].Thesemi-dwarf 1(sd1) gene driving this revolution was introduced to prevent or mitigate lodging in rice even under high nitrogen fertilization [2,5,6].Although the gene has been used in breeding for decades,and severalsd1alleles have been deployed,elitesd1alleles and related germplasm resources used forjaponicarice breeding have remained long unidentified.

The recessive semi-dwarf genesd1from the Chinese variety Dee-Geo-Woo-Gen was identified [5,7-9] as the causal gene conferring the semi-dwarf phenotype in Taichung 1, IR8, and its derivatives.Subsequent genetic and biochemical analyses confirmed thatSD1encodes rice gibberellin 20 oxidase 2 (GA20ox-2), an enzyme that catalyzes the penultimate steps of the GA biosynthesis pathway, converting GA12/GA53to the bioactive GA precursors GA9/GA21[7,8,10].Recessivesd1is a deletion,substitution, or frameshift null functional allele of the dominant geneSd1.The other GA20ox protein encoded byGA20ox-1is expressed preferentially in flowers, ensuring normal flower development insd1plants and leading to semi-dwarfism without a seed yield penalty[8].In addition to rice plant height,sd1influences grain protein content, leaf angle, domestication, seed dormancy, tiller number,and spikelets per plant [1,11-14].

Although numerous dwarf mutants have been reported, independent semi-dwarf breeding programs worldwide use different alleles of the sameSD1gene [5].sd1alleles come from Dee-Geo-Woo-Gen, Ai-Jiao-Nan-Te, Zhaiyeqing 8, 93-11, Reimei, Calrose 76,Jikkoku,and Pusa 1652[8,9,15,16],and are abbreviated respectively assd1-d,sd1-AJNT,sd1-ZYQ8,sd1-9311,sd1-r,sd1-c,sd1-j,andsd1-bm(Fig.S1).sd1-dcontains a 383-bp deletion from exon 1 to exon 2, andsd1-AJNThas a GC deletion in exon 1.Both contain a premature stop codon due to frameshift mutations [8,9,15,17].sd1-9311andsd1-bmharbor different single nucleotide polymorphisms(SNPs)in exon 3,but both contain a premature stop codon[15,16].The foursd1-d,sd1-AJNT,sd1-9311,andsd1-bmalleles are thus considered loss-of-function types.The four weak-functional allelessd1-j, sd1-ZYQ8,sd1-c, andsd1-rcarry amino acid substitutions (G94V, P240L, L266F, and D349H) caused by SNPs in exons 1,2,2,and 3,respectively[5,8,9,15].Thejaponicacultivar Nipponbare carries a functional nucleotide polymorphism in SD1 (100E and 340Q), resulting in lower enzymatic activity than the SD1 in theindicacultivar Kasalath (100G and 340R) [18], and these were designatedSD1-EQandSD1-GR(orSd1JapandSd1Ind), respectively[18,19].Overall, unlike most modernindicacultivars carryingSD1loss-of-function alleles,japonicacultivars carry weak functional alleles[12,19].The null allelesd1-dhas been used in Japan to introduce a semi-dwarf phenotype injaponicarice cultivars, such as Kinuhikari and Dontokoi[15].However,this has not been reported in modernjaponicarice in China.Whether the null allelesd1-dcan be used in modernjaponicabreeding in northeast China is still a problem to be solved.

Owing to environmental changes, popularization of directseeded rice, and high-yield breeding strategies, lodgingresistance requirements are constantly increasing.Thesd1allele with weaker or null function has become more attractive thanSD1-EQinjaponicarice breeding.Recently, two major challenges have emerged in breeding semi-dwarf rice cultivars [20].The first challenge is that the development of semi-dwarfjaponicacultivars has proven to be more complex than that ofindicacultivars.The second is that the widespread use of thesd1allele reduces genetic diversity and introduces other associated adverse effects (for example,sd1-dreduces fertility under cool temperatures and increases drought sensitivity).For this reason,japonicarice breeding is in need of new dwarfing sources.

With the development of next-generation sequencing (NGS)technologies, it has become possible to sequence numerous rice accessions at relatively low cost.In the past decade, genotyping datasets for many rice accessions have been released, from which several loci associated with important agronomic traits have been identified [13,21].Several large-scale rice collections with sequencing and phenotypic data have provided valuable materials and knowledge for rice research and breeding projects[13,22-25].Recently, an integrated omics knowledgebase, MBKbase-rice(www.mbkbase.org/rice) was opened to the public.It integrates rice germplasm information, multiple reference genomes with a unified set of gene loci, population sequencing data, phenotypic data, known alleles, and gene expression data [21].The MBKbase-Rice database provides tools for the discovery of additionalsd1alleles and corresponding germplasm resources.

Germplasm resources provide a genetic basis for crop improvement [26].Since the early 1960s, rice production has achieved breakthroughs via the development of semi-dwarf cultivars.Semi-dwarf rice cultivars carrying thesd1allele have yielded many excellent offspring, such as MingHui 63 and 93-11, and have become backbone parents for rice breeding[27].Long-term breeding practices have shown that the replacement of crop cultivars depends on the discovery, creation,and effective use of new backbone parents [28].Recorded genealogical relationships and valuable germplasm resources derived from the genealogy are collected and preserved,laying the foundation for the use of these species for tracking the transmission of chromosomal blocks preserved in all members of the pedigree[29].Recently,some studies identified a large number of gene based on the genealogical relationships,genome resequencing technology and related data analysis methods[30,31].Therefore,trackingsd1alleles distribution in offspring derived from backbone parents provide reference and support for discovering favorable genes that have been strongly selected together with thesd1alleles since the Green Revolution.

To identify elitesd1alleles and related germplasm resources used forjaponicarice breeding, haplotype analysis of MBKbase-Rice andjaponicasubpopulation of northeast Asia was carried out.Japonicaaccessions carrying reportedsd1alleles were identified by genotype analysis, literature searching, pedigree analysis,and dCAPs marker identification.The pedigree relationships,derived progeny, and geographic distribution of accessions carryingsd1alleles were compared in Japanese and Chinesejaponicarice breeding programs.The application model ofsd1allele in futurejaponicabreeding of northeast China was present based on breeding practice.

2.Materials and methods

2.1.Plant material and trait measurement

A set of 295 temperatejaponicaaccessions from northeast Asia(Table S1)are conserved at the Rice Research Institute of Northeast Agricultural University (RRI, NEAU), as previously described by Li[32].Rice plants were planted at the Acheng Experimental Base of NEAU (45°32′N, 127°24′E) in 2018 and 2019.An eight-row ×20 plant block with a planting density of 30 × 13.3 cm was used for each cultivar.The plant heights of five plants in the middle of the fourth row were continuously monitored at the maturity stage.

Twenty-fourjaponicacultivars conserved in the RRI of NEAU were tested using a cleaved amplified polymorphic sequence(dCAPS) marker for thesd1-rallele (Table S2).Except for Nipponbare (SD1-EQ) used as a control, the other semi-dwarf cultivars were mainly from Heilongjiang and Jilin provinces.

To improve the lodging resistance of the high-quality cultivar DHX2 (carrying theSD1-EQallele), we introduced thesd1-dallele via conventional crossing.The chromosome segment substitution line (CSSL) SL2003 derived from a cross between thejaponicaKoshihikari and theindicaIR64 was provided by the Rice Genome Resource Center(RGRC),Rice Genome Project of the National Institute of Agrobiological Sciences (NIAS), Tsukuba, Japan.SL2003 has a Koshihikari (KS) genetic background and carries ansd1allele(sd1-d) from IR64; here it is referred to as KSsd1-d.SL2003 was crossed with DHX2 to generate an F1.Early-maturing dwarf lines in the F2segregating generation were selected for backcrossing with DHX2 to generate a BC1F1, followed by single-seed descent to BC1F7.Insertion/deletion (InDel) markers were used to identifysd1-din BC1F3.Two lines with dwarf and semi-dwarf phenotype were designated as 1279 and 1280 at the BC1F7generation in 2020.A stable semi-dwarf mutant line, DHX2-M3, was obtained by physical mutagenesis.These germplasms were planted at the Harbin experimental base of the Northeast Institute of Geography and Agroecology of Chinese Academy of Sciences (45°42′N,126°23′E) in 2020.Plant height was recorded for each line at the full-heading or maturity stage.

2.2.Mining for sd1 alleles and haplotype analysis

TheSD1gene sequences of 5459 rice accessions were retrieved from the recently completed rice sub-database of MBKbase.This sub-database reveals the relationships among germplasm, phenotype, and genotype by providing genotypes for each gene locus and their associated phenotypes in a population [21].Number of samples ≥10, ALT (alternative allele frequency)% ≥0.5, missing%≤25, ALT depth ≥1, and quality ≥3 were used as search criteria.A recent study [14] showed that plant height in rice is controlled mainly by the SD1 protein function but not the gene expression level, thus missense SNPs in the CDS region ofSD1mRNA(OsNipG0105661100.01) were extracted.A genotype defined as a group of SNPs and short InDels in MBKbase is equivalent to a haplotype when the identified SNP and InDel are homozygous in a genome [21].The subgroup components of each haplotype was classified by morphological features and SNP phylogenetic analyses.

Sequence reads of 295 temperatejaponicaaccessions from northeast Asia were retrieved from the NCBI Sequence Read Archive (SRA) website under accession ID PRJNA512109 [32].Reads were aligned to the Nipponbare reference genome IRGSP-1.0_genome from the Rice Annotation Project (RAP, https://rapdb.dna.affrc.go.jp), and the SortSam tool of the Picard package(https://broadinstitute.github.io/picard/) was used to sort the sequence alignment format (SAM) files by chromosome and position.PCR duplicates were removed using the MarkDuplicates tool in the Picard package with default parameters.SNPs and InDels were called using the HaplotypeCaller tool in GATK [33] by allele-specific annotation with all its usual standard annotations.SNPs that caused missense variants in theSD1exon region were used to define haplotypes.The haplotypes were reconstructed after missing and heterozygous SNPs were imputed with PHASE software [34].Haplotypes with an allele frequency ≥ 0.3% were included.

Statistical significance of difference between each haplotype was identified by one-way ANOVA combined Tukey’s post hoc test for plant height.

2.3.Identification of elite sd1 alleles in japonica rice germplasm

In addition to the accessions obtained by haplotype analysis,literature and database searches were used to identifyjaponicarice carrying thesd1allele.Two critical documents provide information on the distribution ofsd1alleles [15,35].MBKbase-rice permits searching germplasm resources by genotype.Owing to the screening conditions, some accessions may be filtered out during haplotype analysis.Accessions withsd1-r,sd1-j, andsd1-calleles were filtered by genotyping based on the Nipponbare reference, and the positions were selected as Chr.1 Pos 38385083 and genotype C forsd1-r, Chr.1 Pos 38382746 and genotype G forsd1-j, and Chr.1 Pos 38383363 and genotype T forsd1-c.Under these conditions, respectively 66, 16, and 7 accessions containingsd1-r,sd1-j,andsd1-cwere obtained.A similar method was used to filtered more than 500 high-quality sequences in RAP (https://ricegenome.dna.affrc.go.jp/) and yielded 6, 2, and 1 accessions containingsd1-r,sd1-j, andsd1-c, respectively.

2.4.Pedigree relationships of the sd1 allele in Japanese and Chinese japonica rice cultivars

The genealogy information was retrieved from the China Rice Data Center (CRDC, http://www.ricedata.cn/variety/) and Ineweb(https://ineweb.narcc.affrc.go.jp/hinsyu_top.html), a database that records breeding information of rice cultivars in Japan.

2.5.Development and validation of dCAPS and InDel markers based on sd1-r and sd1-d

Accessions with thesd1-rallele were found in the haplotype analysis to carry a missense mutation at Chr.1 Pos 38385083 (GC) and a synonymous mutation at Chr.1 Pos 38385082 (C-A).Based on this information, a derived cleaved amplified polymorphic sequence (dCAPS) marker dsd1-r was designed using dCAPS Finder 2.0 [36] and the recognition site (GAGCTC) was introduced into the reverse primer dsd1-R 5′- CGGTAGTGGCGCTGCG TGAAGCGCATGAGC T-3′for theSacI restriction enzyme.The forward primer dsd1-F was 5′-GGTATAAGAGCTGCCTGCACA-3′.Digestion withSacI of the amplified products of wild-typeSD1-EQand mutantsd1-ryielded respectively 149- and 181-bp fragments.

Based on the 383-bp deletion in thesd1-dallele, the following PCR primers were used to detect thesd1-dallele.Sd1-F1:5′-CAACACAGCGCTCACTTCTC-3′.Sd1-R1:5′- CGCCCCTACCACGAT CTTAT- 3′.The amplified fragments of the wild-typeSD1-EQand mutantsd1-dwere respectively 1138 and 755 bp in length.

2.6.DNA extraction, PCR, and gel electrophoresis

DNA was extracted using the cetyltrimethylammonium bromide method.To detectsd1-ralleles, a polymerase chain reaction(PCR) was performed in a 20-μL reaction mixture comprising 1× Es Taq MasterMix (Cwbiotech, Beijing, China), 0.4 mmol L-1forward and reverse primers, and 50-80 ng of DNA.The PCR program was as follows:initial denaturation at 94 °C for 2 min, 35 cycles of denaturation at 94 °C for 30 s, annealing at 58 °C for 30 s, extension at 72 °C for 10 s, and a final extension at 72 °C for 2 min.Ten microliters of PCR product were used for enzymatic analysis,and the rest was used as a control.After overnight digestion withSacI-HF(NEB,Beijing,China)at 37°C,10 μL of the enzymatic reaction solution and control product were used for electrophoresis.A 4% agarose gel was used to separate the fragments.For detectingsd1-dalleles, amplification of DNA (in 10 μL) was performed at a final concentration of 1× Super Pfx GC buffer,0.2 mmol L-1dNTPs,0.5 mmol L-1forward and reverse primers, 10-30 ng of DNA, and 0.5 U Super Pfx DNA polymerase(Cwbiotech).Initial denaturation was performed at 98 °C for 3 min.Thirty-five cycles were run, including denaturation at 98 °C for 10 s, annealing at 53.8 °C for 30 s, extension at 72 °C for 20 s,and final extension at 72°C for 10 min.Amplified products were electrophoresed on a 1.5% agarose gel.

3.Results

3.1.Mining and haplotype analysis of sd1 alleles from MBKbase-rice

Because the SNPs/InDels identified in this study were all homozygous, the genotype analysis based on MBKbase-rice is also called a haplotype analysis.Accordingly,the missense SNPs/InDels in the CDS region of theSD1gene were extracted,and 7 SNP/InDel missense sites and 11 haplotypes were retained(Tables 1,S3).The two newly identified SNP/InDel sites occurred at Chr.1 pos 38382761 and 38383144,with a 2-base insertion resulting in a frameshift mutation and a single SNP that induces an amino acid sub-N represents multiple continuous bases,which means there are multiple bases delition in the corresponding sample.In the haplotype row,the uppercase base letter indicates that this position variation is homozygous, - indicates missing reads, and 0 means the same as the reference.AF, allele frequency.stitution (C193S).SNPs/InDels at other sites have been reported previously(Fig.S1).The structural variation ofsd1-d(383 bp deletion) cannot be identified in all public databases at present owing to the effect of NGS read length.Because the sequence of hap 1 was the same as the Nipponbare reference, it was indentified here asSD1-EQ.The sequences of hap 2 and hap 9 have only one missense SNP (Q340R and D349H, respectively) and can be equivalent toSd1Indandsd1-r,respectively.The other haplotypes contained multiple missense SNPs or missing reads.

Table 1 Haplotype analysis of the sd1 allele from MBKbase-rice.

All haplotypes found in China and the percentages of other haplotypes accounted for more than 10% of the world’s total, except for hap 7 (Fig.1A).With respect to the functional SNP at Chr.1:38385057(Q340R),the haplotypes can be divided into thejaponicaandindicarice groups.Thejaponicarice group included four haplotypes:hap 1, hap 5, hap 9, and hap 11.The other seven haplotypes belonged to theindicarice group.This classification is consistent with the classification method based on phenotypic characteristics and SNP polymorphisms (Fig.1B, C).In the hap 1 and hap 5 subgroups,temperate and tropicaljaponicaaccounted for more than 45 and 20%, respectively.In hap 9 and hap 11, temperatejaponicarice had an absolute advantage (Fig.1C).Indicarice accounted for more than 70%of theindicarice group.Aus/boro respectively accounted for 10.9 and 16.5%in hap 2 and hap 4 subgroups,and the proportions in other ecotype types were relatively small (Fig.1C).

Differences in plant height among the 11 haplotypes were also evaluated(Fig.1D).Except for hap 11,the plant height of thejaponicarice group(including hap 1,hap 5,and hap 9)was significantly lower than that of hap 2 (strongSD1allele,Sd1Ind) by 22.4%, 20.2%, and 28.6%, respectively.Compared with hap 2, the plant height of hap 3 and hap 6 was significantly decreased by 23.2% and 19.6%, respectively, but there was no significant difference between hap 3 and hap 6.Thus, hap 3 and hap 6 could be regarded as newsd1alleles and were designatedsd1-New1andsd1-New2, respectively.This result also indicated that the effect of SNP C193S insd1-New2on plant height was masked by the 2 bp insertion at Chr.1 38382761.Owing to fewer plant height data and greater variation, heights of the weak or null functional haplotypes hap 7 and hap 10 were not significantly different from that of hap 2.Similar insignificant changes were observed between haplotypes hap 9(sd1-r)and hap 1(SD1-EQ).

3.2.sd1-r contributes to plant height variation of modern japonica rice in northeast Asia

Three missense SNPs in the CDS region ofSD1were extracted,and four haplotypes were retained(Tables 2,S1);these four haplotypes were detected in MBKbase-Rice.Here, hap J1 is the same as hap 1 in MBKbase-Rice and is referred to as theSD1-EQallele,which had already been artificially selected forjaponicarice breeding programs before the Green Revolution [18].hap J2, similar to hap 9 (sd1-r) in MBKbase-Rice and accounting for about 9% ofjaponicarice accessions, significantly reduced plant height by 10.1 and 8.1% in 2018 and 2019, respectively.hap J3 (identical to hap 2,Sd1Ind) and hap J4 (identical to hap 10) appeared to have been introgressed into a small number ofjaponicalandraces.

3.3.Identification and distribution of japonica rice cultivars carrying the sd1 allele in Japan and China

In the haplotype analysis of the two populations,sd1-jandsd1-cinjaponicarice reported earlier were not detected because of low allele frequency.Searching for germplasm by genotype in MBKbase-Rice and RAP yielded respectively 72, 18, and 8 accessions containingsd1-r,sd1-j,andsd1-c, withsd1-cbeing distributed mainly in the United States and Australia, which are not main rice-producing areas.We accordingly focused onsd1-randsd1-j.Literature search and haplotype analysis of the 295japonicaaccessions from northeast Asia yielded 103 accessions carryingsd1-rand 31 carryingsd1-j(Table S4).Fifteeen accessions in MBKbase-Rice come from multiple data sources, with only four accessions showing consistent sequencing results.For example,Aoyu 324 (also named Oou 324), Bing 03123 (approved name Xiushui 123), and Xiushui 128 were confirmed to carry thesd1-rallele, whereas Yangeng 5 hao carried thesd1-jallele.Bing 02-09(approved name Xiushui 09) has been verified to carrysd1-rin MBKbase-rice, as reported by Asano [15].Considering that thesd1-randsd1-jalleles originated from Reimei and Jikkoku in Japan,using breeding information from ineweb and CRDC, we can trace whether the ancestors of accessions with inconsistent results in MBK-base Rice came from these two cultivars.By genealogical analysis, 14 of 36 accessions from Japan can be traced to Reimei,and 12 of 13 cultivars can be traced to Jikkoku.However, owing to the lack of intermediate material information and unclear genealogy records, only 8 of the 40 cultivars from China can be traced to Reimei, and 6 out of 9 cultivars can be traced to Jikkoku(Table S4).

Table 2 Haplotype analysis of the sd1 allele from 295 japonica rice accessions and their plant height performance in different years.

To further test the authenticity of these cultivars,we developed a dCAPS marker forsd1-r.We screened 12 cultivars from MBKbase-Rice, 3 high-quality and lodging-resistant cultivars in Jilin, and 7 rice cultivars grown in large areas in different cumulative temperature zones of Heilongjiang province.The results of dCAPS marker detection showed thatsd1-ralleles were present in Longdao 21,Dongnong 427, Dongnong 419, Longdao 10, Mudanjiang 21, and Jigeng 101 (Fig.S2).

Fig 1.Haplotype analysis of the sd1 allele from MBKbase-rice.(A) Sample numbers of sd1 haplotypes in China and other countries.The subgroup component of each haplotype according to(B)morphological features and(C)SNP phylogenetic analyses.(D)Box plot of plant height of different sd1 haplotypes in rice.Tukey’s test was used to compare the effects of haplotypes.Ind,indica;TemJ,temperate japonica;TroJ,tropical japonica;Jap,japonica;Aus,Aus/boro;Int,intermediate;Bas,basmati/sadri;*,P<0.05;****, P < 0.0001.

The pedigree relationships of germplasm resources containingsd1alleles injaponicarice were drawn (Figs.S3, S4).According to the current pedigree information, thesd1alleles should have been derived for more than four generations in Japan,given that the currently identified materials were released in 2005 at the latest(Fig.S3), and thesd1-randsd1-jalleles have been separately derived for three and four generations, respectively, since they were introduced into China (Fig.S4).In addition, the Dongnong series of rice cultivars retain thesd1-rallele in the process of generation transmission,while the Xiushui series of rice cultivars containsd1-rorsd1-j.

From the perspective of breeding areas,japonicarice cultivars with thesd1-rallele in Japan are distributed mainly in Aomori,Iwate, and Yamagata in the Tohoku region.Those containing thesd1-jallele are found in the Kyushu region, while those with thesd1-dallele are distributed in the Chubu region.Japonicarice cultivars with thesd1-rallele in China are distributed mainly in northeast, east, and central China, while those withsd1-jalleles are distributed only in eastern China (Table S5).

3.4.Comparison of the progeny of japonica rice derived from sd1 alleles in Japan and China

We tracked the derived offspring of the reportedsd1alleles in Japan and China.According to our statistical data (Tables S6, S7),three original cultivars or earliest confirmed cultivars containingsd1alleles injaponicarice (Reimei, Jikkoku, and Kinuhikari) have been used to derive respectively 324, 346, and 368 cultivars in Japan.The offspring of Reimei were distributed mainly in Tohoku and Chubu, accounting for approximately 84% of the total.More than 76% of Jikkoku descendants were located in the Kyushu region, 68% of Kinuhikari descendants were distributed in the Kyushu and Chubu regions, and 21% were distributed in Shikoku and Chugoku.A similar situation also occurred in China:the derived offspring of Reimei were distributed mainly in northeast China, and those of Jikkoku were distributed mainly in eastern China.The distribution of derived descendants in the direction away from their origin gradually decreases.

Generally,cultivars with essential characteristics such as excellent comprehensive traits, high general combining ability, and wide adaptability are used to create more cultivars.Using pedigree analysis,we identifiedjaponicarice cultivars carrying thesd1allele and their derived progeny (>3) in Japan and China (Table S7).The numbere of first-generation cultivars derived from Reimei,Jikkoku,and Kinuhikari in Japan were 112, 16, and 135, respectively.The branches with the largest number of descendants for Reimei, Jikkkoku, and Kinuhikari were Akihikari (83), Mutsuhumore (10), and Maihime (33); Hoyoku (62), Reihou (39), Saiwaimochi (9), and Tsukuhiakamochi(8);Dontokoi(46),Ikuhikari(30),and Akidawara(11), respectively.Only nine descendants were directly derived from Reimei in China, and the branches with the largest number of derived generations and descendants were Akihikari(Qiuguang,46), Dongnong 419 (3), and Dongnong 423 (14).In China,sd1-jderived cultivars are derived mainly from the intermediate material Ce21 derived from Reihou (Fig.S4).The Xiushui 11 branch has yielded the largest generation numbers, which constitute the distribution ofsd1-jin east China.

3.5.Creating intermediate material for semi-dwarf japonicarice

DHX2 is very popular in China because of its excellent quality.However, owing to its tall (approximately 115 cm) plant stature,lodging often occurs in production [19].Thesd1allele in DHX2 was the same as in Nipponbare, both of which wereSD1-EQ.Accordingly, we used conventional breeding to introduce thesd1-dallele, or a physical mutagenesis strategy to improve lodging resistance (Fig.2A).Compared with DHX2, the plant height of a stable physically mutagenized line DHX2-M3 was reduced by approximately 15 cm,but this did not change its lodging tendency owing to successive typhoons affecting the northeast region.However, two stable lines, 1279 and 1280, carryingsd1-d(Fig.2B),withstood strong winds and heavy rains.Owing to the late growth of KS and KSsd1-dand the severe lodging of DHX2 and DHX2-M3,yield-related data were not available.For this reason,we compared only the plant height and internode composition of KS, KSsd1-d,1279, and 1280 (Fig.3).Compared with KS, except for the uppermost internode, all other internodes were shortened in KSsd1-d,and the plant height was reduced by 21.3%(Fig.3A,B,E).Although both lines carried thesd1-dallele, the 1279 line exhibited a dwarf phenotype (Fig.2), and its internodes were significantly shorter than those of 1280, especially the uppermost internode (Fig.3 C,D), which had a significantly lower contribution to plant height(Fig.3F).This implies that in addition to thesd1-dallele, the short stature of 1279 is likely caused by other genes controlling the uppermost internode’s length.

Fig 2.Phenotype of newly created semi-dwarf materials and molecular identification of the sd1-d allele.(A-C) Morphological comparison of DHX2, DHX2-M3, KS,KS sd1-d(SL2003), 1279, and 1280.The morphological comparison of DHX2 and DHX2-M3 was conducted on August 12 (before typhoon No.8, ‘‘Bavi”).DHX2-M3 was a stable physically mutagenized mutant line with the same heading date as DHX2 which headed on August 5.The morphological comparison of KS, KS sd1-d, 1279, and 1280 was made on September 7 (after typhoon No.10, ‘‘Poseidon”).KS sd1-d is a CSSL with a KS genetic background carrying the sd1-d allele from IR64, its original CSSL number is SL2003.KS and KS sd1-d are late-maturing materials;their heading date in Harbin is about September 5,one month later than DHX2.Two stable lines derived from the cross DHX2//(short culm and early heading lines in F2 of SL2003/DHX2)were designated 1279 and 1280 at the BC1F7 in 2020.The heading date of 1279 and 1280 was the same as DHX2.Scale bars, 10 cm.(D) Detection of the sd1-d allele by PCR.The amplified fragments of wild-type SD1-EQ and mutant sd1-d were respectively 1138 and 755 bp long.Lanes 1-24 display the differentiation of amplified fragments from the BC1F3 of DHX2//(SL2003/DHX2,F2).M,marker.1,KS;2,KS sd1-d(SL2003);3-24 independent lines of dwarf and semi-dwarf in BC1F3.1279 and 1280 were derived from the selfing of lines 12 and 15, respectively.

Fig 3.Gross morphology of internodes and panicles of KS,KS sd1-d, 1279, and 1280.(A-D)Appearances of the lengths of the panicle and internode lengths of KS, KS sd1-d,1279,1280.Scale bars,2 cm.(E)Comparison of panicle and internode lengths between KS,KS sd1-d,1279,and 1280.Values are means of the panicle and internode lengths of the main culms(n=5).Tukey’s test was used to compare panicle and internode lengths between adjacent materials.**,P<0.01,****,P<0.0001.(F)Contribution of panicles and each internode to plant height.P, panicle; I-VI indicate internodes from head to base.

4.Discussion

4.1.Opportunities and challenges for mining known gene haplotypes and corresponding germplasm

Recent advances in NGS have strongly influenced crop breeding,providing opportunities for large-scale population genetics and genomic analyses [13,21].High-coverage whole-genome sequencing(WGS)is used to characterize the genetic diversity of individuals, as well as to reduce sequencing costs.Long-read singlemolecule DNA sequencing makes it possible to obtain all alleles of all genes in the WGS population using high-throughput methods[21].MBKbase-Rice integrates public raw data sets, establishes a genotype-to-phenotype link, and provides a user-friendly and highly flexible tool for data visualization and mining.There was a 2-base insertion in two newly identifiedsd1alleles, hap 3 (sd1-New1) and hap 6 (sd1-New2),resulting in a frameshift mutation(Table 1).However, the effect of SNP C193S in hap 6 on plant height was masked by the 2-bp insertion at Chr.1 38382761(Table 1; Fig.1D).Therefore, the role of SNP C193S needs to be investigated further.Although multiplesd1alleles have been reported, newsd1alleles are still being mined [14,16]; mining the database is the most convenient way to obtain new alleles.

MBKbase-rice provides an advanced tool for searching germplasm resources by genotype, which differs from previous haplotype analysis.However, when we use genotypes to search for germplasm resources, we encounter greater challenges.Only four of 15 materials in MBKbase-Rice from multiple data sources showed a consistent sequencing result(Table S4).The inconsistent detection results may be affected by the quality of secondgeneration sequencing data.The reliability of data was affected by the bias of second-generation sequencing itself and the limitation of sequence length [38,39].To solve this problem without imposing filter conditions, we adopted three methods.Since thesd1-r and sd1-jalleles injaponicarice were first confirmed in Reimei and Jikkoku, tracing the pedigree may exclude some false positives.In addition,we adopted document tracking and molecular marker verification,given that some materials are intermediate materials or their genealogy is not well documented.We have obtained more comprehensive and reliable information onjaponicarice materials carryingsd1-randsd1-jalleles using the above methods (Table S5; Figs.S2-S4).

If offspring in the pedigree analysis carries the correspondingsd1allele, one parent from the original cultivar should also carry this allele.A recent study of improved rice cultivars in China over the past 40 years showed that Akihikari and Ce21 were the backbone parents of conventional rice in China[27].Sequencing results from different sources and pedigree relationships confirmed that Akihikari carries thesd1-rallele (Table S4; Fig.S3A).It can be inferred by pedigree analysis that Ce21 carries thesd1-jallele(Fig.S3B).The identification of backbone parents ofjaponicarice carryingsd1alleles provides references for parental selection and genetic research on the essential segments of backbone parents.

4.2.sd1-r is widely used in japonica rice both in northern Japan and China

Green Revolution genes have been widely used worldwide since the 1960s.Abundant evidence shows that mainly null functional alleles (sd1-d,sd1-AJNT, andsd1-9311) were used to modulate the plant height ofindicacultivars in modern breeding, while mostly weak functional alleles (Sd1JaporSD1-EQ) were used injaponicarice breeding [5,12,19].Given that the originaljaponicaalleleSD1-EQofSD1was artificially selected during domestication [18],it seems thatjaponicasemi-dwarf breeding was not as popular as inindicarice [12].However, Japan has made breakthroughs injaponicarice production by developing semi-dwarf cultivars in the early 1960s.The semi-dwarf sources of Reimei and Jikkoku constitute the basic framework of modern cultivars via crosses with other cultivars [37].The northeast is the mainjaponicariceproducing area in China, and the early breeding process relies on the introduced Japanese cultivars.Therefore, the corresponding alleles may be introgressed intojaponicarice cultivars in northeast China during the breeding process.As expected, thesd1-rallele was detected in the population from MBKbase-Rice and northeast Asianjaponicarice, butsd1-jwas not observed because of the low allele frequency (Tables 1, 2).Regardless of the cultivar carrying thesd1-rallele or its derived offspring,their geographical distribution was concentrated mainly in the Tohoku area of Japan and North China (Tables S5, S6).These results indicate thatsd1-ris an elite allele in these regions.Although the frequency ofsd1-jalleles was too low to be detected in the haplotype analyses,pedigree tracing revealed thatsd1-jwas distributed mainly in eastern China and Kyushu in Japan(Table S6).The results showed that both alleles were positively selected in the corresponding ecological regions.

4.3.Breeding strategies for creating new semi-dwarf japonica rice in northeast China

Rice production has achieved breakthroughs through the development of semi-dwarf cultivars since the early 1960s.Although many novel hormone-associated genes play a role in controlling plant height, their actual breeding value must be verified[5,19,20].Owing to the unfavorable phenotypes of other dwarf mutants (severe dwarfism,short panicle length,and small or sterile grain resulting in decreased yield), they are unsuitable for breeding [5].Although the widespread use ofsd1has brought about a series of problems,extending elite alleles to a broad range and introducing othersd1alleles is still the first choice for diverse ecological regions.

Our early breeding practices have shown that the more significant the gap between parental ecological types, the more difficult is selection and the lower is the probability of obtaining favorable variation suitable for cold environmental regions.Considering the reproductive barriers ofindica-japonicahybridization and the problem of choosing stable progeny lines, we used a chromosome segment substitution line (SL2003) as thesd1-ddonor to improve the lodging resistance of DHX2.SL2003 has a Koshihikari background, except for the first chromosome containing the IR64 segment.Given that Koshihikari and SL2003 are late-maturing accessions, their heading date in Harbin, Heilongjiang province is in early September.At this time,the temperature begins to decline,and they fail to mature naturally.Accordingly,the hybrid combination of SL2003 and DHX2 was deployed in Lingshui(Hainan).Earlymaturing dwarf lines in the F2segregation generation were selected for backcrossing with DHX2 to generate BC1F1, and then single-seed descent was performed for BC1F7.The single-seed descent of dwarf and semi-dwarf lines carryingsd1-dwere designated 1279 and 1280 at the BC1F7in 2020.In 2020, three typhoons severely affected crop production in northeast China, but the 1279 and 1280 lines withstood the strong winds and rainfall.The dwarf line 1279 has more tillers, short panicles, more kernels per panicle, and a higher seed setting rate than the semi-dwarf line 1280 (data not shown), but with a slightly enclosed panicle(Figs.2A,3).Based on the internode elongation pattern,rice dwarf mutants have been classified into five different types, dn-, dm-,d6-, nl- and sh-types [40].The internode elongation pattern of common cultivars controlled by thesd1allele is almost unchanged[41],and belongs to the dn-type.However,compared with KS,the uppermost internode length in KSsd1-d(SL2003) was increased.Thus,the uppermost internode’s length of KSsd1-dwas likely controlled bysd1-dand other genes,because approximately 16.43 Mb on chromosome 1 of SL2003 comes from IR64.Similar reasons can explain the variation of uppermost internode length for 1279 and 1280.The combination ofsd1-dwith other genes controlling the uppermost internode’s length reflects the possibility of usingsd1-dto create different plant architectures injaponicarice breeding.In short, the 1279 and 1280 lines carryingsd1-dalleles are excellent materials for breeding and genetic research.

Fig 4.The application model of the sd1 allele in future japonica breeding of northeast China.The current exceptional sd1-r allele is not sufficiently used and should be more widely used in the future.The sd1-j and sd1-d used in southern regions can be introduced into northern japonica rice by conventional breeding methods.New sd1 alleles(?)created by genome editing technology should also be used.

The distribution of thesd1allele and its derived descendants in the direction away from their origin gradually decreases(Table S6).This phenomenon reflects that after leaving the original ecological adaptation zone,the reuse of thesd1allele requires a long process of domestication, selection, and transformation.An earlier study[42] showed that the heading time gene determined the regional adaptability of Tongil-type short-culm rice cultivars developed in Republic of Korea, which carry thesd1-dallele from IR8.The critical genes or different gene combinations controlling early flowering in Heilongjiang province have been well identified [43,44].Hd2/Ghd7.1/OsPRR37plays a major role in controlling rice distribution in Heilongjiang, and theHd2/Ghd7.1/OsPRR37andHd4/Ghd7haplotypes became non-functional alleles in northward rice cultivation [43].Furthermore,indicaandjaponicahave developed distinct types for cold adaptability during domestication [45].During the domestication and breeding ofjaponicain cold climates,the favorableHAN1andCTB4aalleles were selected to confer cold tolerance at the seedling and booting stages, respectively [46].There is evidence [47] that the dwarfing gene from Dee-geowoo-gen(sd1-d)reduced fertility in cool temperatures at the booting stage.Thus, pyramiding of early-maturity and cold-tolerance genes should expand the ecological adaptation zone ofsd1-d.

Recently, genome editing has provided opportunities for crop improvement.However, the CRISPR/Cas9 system may not be as precise as expected in rice.The characterization and screening of targeted gene modification and nontarget DNA modification must be performed for generations before the CRISPR/Cas9 system can be extended from the laboratory to the field [48].Although we can change the negative impact of off-target effects through backcrossing, early molecular characterization and screening increase labor and costs,especially for multiple genes with multiple targets.

We propose a model to summarize the development of thesd1allele in future breeding in northeast China (Fig.4).The current exceptionalsd1-rallele is not sufficiently used and should be more widely used in the future.Thesd1-jandsd1-dused in southern regions can be introduced into northernjaponicarice by conventional breeding methods.Newsd1alleles created by genome editing technology should also be used.Overall, more work is needed to pave the way for the semi-dwarf breeding ofjaponicarice in northeast China.

5.Conclusions

This study identified elitesd1alleles and associated germplasm resources used forjaponicarice breeding.We described the full scope of the distribution and use ofsd1alleles in Japan and China’sjaponicarice breeding programs and developed dCAPS markers for thesd1-rallele.The results of this study will facilitate the selection and use of diversesd1alleles for rice MAS breeding and provide a material basis for the next step in identifying favorable genes selected together with thesd1alleles injaponicabackbone parents.

CRediT authorship contribution statement

Hanjing Sha:Conceptualization,Visualization,Writing-original draft, Writing - review & editing, Funding acquisition.Hualong Liu:Conceptualization, Writing - review & editing.Guangxin Zhao:Resources, Writing - review & editing.Zhongmin Han:Methodology,Writing-review&editing.Huilin Chang:Visualization,Writing-review&editing.Jingguo Wang:Writing-review&editing.Hongliang Zheng:Resource.Jifeng Zhang:Investigation.

Yang Yu:Review.Yuqiang Liu:Visualization.Detang Zou:Resource.Shoujun Nie:Conceptualization.Jun Fang:Conceptualization, Review, Funding acquisition, Project administration,Resources.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA24020301), Young Scientists Fund (CN) (31900423), Excellent Youth Foundation for Heilongjiang Scientific Committee (JC2017009), and Cooperative Innovation Extension System of Rice Modern Agricultural Industrial Technology in Heilongjiang province.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.05.005.