Guiquan Zhang

Guangdong Provincial Key Laboratory of Plant Molecular Breeding,State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources,South China Agricultural University,Guangzhou 510642,Guangdong,China

Keywords:Gene pool Chromosome-segment substitution Breeding by design Breeding platform Rice

ABSTRACT Progress in plant breeding depends on the development of genetic resources,genetic knowledge,and breeding techniques.The core of plant breeding is the use of naturally occurring variation.At the beginning of the post-genomic era,a new concept of‘‘breeding by design”was proposed,which aims to control all allelic variation for all genes of agronomic importance.In the past two decades,we have applied a three-step strategy for research on rice breeding by design.In the first step,we constructed a singlesegment substitution line(SSSL)library using Huajingxian 74(HJX74),an elite xian(indica)rice cultivar,as the recipient in which to assemble genes from the rice AA genome.In the second step,we identified a series of desirable genes in the SSSL library.In the third step,we designed new rice lines,and achieved the breeding goals by pyramiding target genes in the HJX74-SSSL library.This review introduces the background,concept,and strategy of breeding by design,as well as our achievements in rice breeding by design using the HJX74-SSSL platform.Our practice shows that target chromosome-segment substitution is a way to breeding by design.

1.Introduction

Plant breeding is essential for crop production.New cultivars with improved productivity and quality can be developed by the genetic improvement of crops.Progress in plant breeding depends on the development of genetic resources,genetic knowledge,and breeding techniques.In the post-genomic era,advances in genomics will make it possible for breeders to select desirable genes throughout the genome and improve plant target traits as desired.With this goal,‘‘breeding by design”,which aims to control all allelic variation for all genes of agronomic importance,has been proposed[1].This concept will initiate a new era in plant breeding.Because of the great difference between the idea and the reality,putting it into practice is a challenge.

2.Rice genetic resources

The 23 species of the genus Oryza include two cultivated and 21 wild species.At least nine different genomes have been identified in the species Oryza.Nine of the wild species are tetraploid,and the remaining wild species as well as the cultivated species are diploid.Based on morphological traits and isolation barriers,the species can be divided into three main complexes.The wild species O.nivara,O.rufipogon,O.barthii(O.breviligulata),O.longistaminata,O.glumaepatula,and O.meridionalis,and the cultivated O.sativa and O.glaberrima,constitute the O.sativa complex.They share the AA genome,allowing gene transfer via conventional hybridization and selection[2-5].This is a useful gene pool for rice breeding[5].

The Asian cultivated rice O.sativa and the African cultivated rice O.glaberrima are thought to be an example of parallel evolution in crop plants.The Asian common wild rice,O.rufipogon,is the wild progenitor of O.sativa and shows a range of variation from perennial to annual types.The annual type,also given a specific name of O.nivara,was domesticated to become O.sativa.The African cultivated rice O.glaberrima was domesticated from the annual O.barthii,which in turn evolved from the perennial O.longistaminata.The Asian cultivated rice is grown worldwide,whereas the African cultivated rice is grown on a small scale in West Africa[4-5].

Domestication of Asian cultivated rice probably started about 10,000 years ago.The Chinese have recognized two rice cultivar groups,Hsien or Xian and Keng or Geng,since the Han dynasty[6-7].In the last century,these two groups of rice were named as two subspecies.Kato et al.[8]divided Asian cultivated rice intoO.sativa subsp.indica Kato and O.sativa subsp.japonica Kato.Ting[6,9]named the Hsien group O.sativa subsp.hsien Ting and the Keng group O.sativa subsp.keng Ting.O.sativa is a highly variable species with worldwide distribution.Approximately 780,000 rice accessions have been collected in gene banks worldwide[10].Glaszmann[11]genotyped 1688 rice cultivars from multiple Asian countries to determine allelic frequencies at 15 isozyme loci,finding that 95%of the cultivars fell into six groups.Garris et al.[12]genotyped 234 accessions of rice at 169 nuclear SSRs and two chloroplast loci,identifying five distinct groups corresponding to indica,aus,aromatic,tropical japonica,and temperate japonica rices.Agrama et al.[13]genotyped a core subset including 1763 accessions using 72 genome-wide SSR markers and found the core subset to be structured into the same five genetic groups.Wang et al.[7]analyzed 3010 diverse Asian cultivated rice genomes,obtaining results consistent with the five major groups previously recognized.Among the five groups,indica is genetically about equally distant from aus and aromatic but far from temperate japonica and tropical japonica,while tropical japonica is close to temperate japonica and aromatic but far from indica and aus[13].

Artificial domestication of Asian cultivated rice by breeding began in the last century.During a century of breeding,rice cultivars have experienced five generations of evolution[14].The first generation(1G)is the tall rice commonly used before 1960s.The second generation(2G)is the semidwarf rice commonly used since the 1960s.The intra-subspecific hybrid rice,including indica hybrid and japonica hybrid rice,is the third generation(3G),widely used since the 1970s.Inter-subspecific introgression rice,including japonica-introgressive indica inbred rice,indica-introgressive japonica inbred rice,japonica-introgressive indica hybrid rice,and indica-introgressive japonica hybrid rice,is the fourth generation(4G),which has been increasingly used in rice production since the end of the last century.The fifth generation(5G)is the intersubspecific hybrid rice of indica and japonica,which is coming[15].The five generations of cultivars and their breeding lines are the preferred genetic resources for rice breeding.

3.From genetic mapping to substitution mapping

Genetic variation is required for gene function analysis.There are two types of genetic variation:artificially induced and naturally occurring.Artificial variation is one of the primary tools in biological research.However,mutant phenotypes of genes for which the wild type carries a functionally null or a weak allele may not be detected.Another negative factor is that random mutagenesis is much more likely to produce loss-of-function mutations than to confer modified or new properties on targeted genes[16-18].

Natural variation can be found among and within natural populations.In contrast to artificial variation,it has the following characteristics.(1)Most of the variation is of a quantitative nature and controlled by quantitative trait loci(QTL).(2)It may arise not only from the presence or absence of allele function but from leaky or weakly functional alleles.(3)It allows researchers to identify gene function in cases where mutagenesis would result in lethality or redundancy.(4)It allows functional analysis of genes by functional rather than loss-of-function genetics.(5)It allows identifying genes that have subtle effects on overall phenotype and investigating the genetic architecture of complex traits[16-17].

The development of DNA markers and permanent mapping populations has made it possible to map QTL underlying complex traits.Primary mapping populations,such as F2,recombinant inbred lines(RILs)[19-20]and doubled haploid lines(DHLs)[21],are often used in genetic mapping of QTL to detect chromosomal regions controlling target traits.RILs and DHLs belong to the so-called permanent populations consisting of homozygous lines.The use of permanent mapping populations is preferred for QTL analysis because it allows replications and repeated analyses of the same population for any trait in multiple environments.RILs are superior to DHLs because of their higher recombination frequency in the population[22].However,these plant materials are insufficient for further analyses such as fine mapping and characterixation of target QTL.

Secondary mapping populations,such as introgression lines(ILs)[23-26]or chromosome-segment substitution lines(CSSLs)[27-28],allow the more precise analysis of target QTL.CSSL populations consist of genotypes that carry a small number of chromosome segments from a donor substituted into the genetic background of a recipient,and are obtained by recurrent backcrossing and marker-assisted selection(MAS)[29].CSSLs that carry a single substitution segment from a donor are called singlesegment substitution lines(SSSLs)[30-31],or near-isogenic lines(NILs)[32-33].

In genetic mapping in primary mapping populations,it has been difficult to determine the precise locations and gene actions of individual QTL.To characterize an individual QTL,it must be separated from the rest of the segregating loci for Mendelization of the QTL.The Mendelization of QTL is best accomplished by construction of SSSLs or NILs ideally differing only for the alleles in a small genomic region spanning a few centiMorgans(cM)around the QTL of interest.SSSLs or NILs allow further fine mapping and chromosome walking to start physical mapping of a target locus and finally accomplish positional cloning of genes at the QTL[16,33,34].

In summary,a secondary mapping population facilitates the genetic dissection of QTL for a complex trait.First,it can be used as an alternative mapping population to perform genome-wide substitution mapping.Second,it can be used to confirm QTL detected previously in a primary mapping population.Third,individual lines can be used as a starting point for the rapid Mendelization of particular QTL and for their fine-mapping and cloning(Fig.1).

4.From conventional breeding to breeding by design

Scientific technologies can greatly assist in the breeding of new cultivars,and new ideas combined with a strong element of intuition are key elements of plant breeding[35].The core of plant breeding is the selection of better types among variants in segregating populations[36].Based on available genetic resources,genetic knowledge,and breeding techniques,plant breeding over the centuries can be divided into four stages.

The first stage is selection breeding.At first,people selected ideal plants from natural populations containing spontaneous variation.Landraces are the result of centuries of local intuitive selection by farmers,and have become valuable resources for genetic improvement[37].

The second stage is hybridization breeding.After the laws of heredity were discovered in the nineteenth century,genetics became the basis of plant breeding.However,further advances in plant breeding took place after hybridization methodology was developed.Hybridization breeding(or cross breeding),based on crossing of different genotypes followed by trait selection,has become a common practice in plant breeding[38].In the 1960s,cross breeding promoted the‘‘Green Revolution”in the development of semi-dwarf wheat and rice cultivars[39].

The third stage is marker-assisted breeding.MAS using DNA markers instead of phenotypic assays reduces the cost and increases the precision and efficiency of subsequent selection steps applied in plant breeding[29,40,41].For example,MAS was used topyramid four bacterial blight resistance genes,Xa-4,xa-5,xa-13,and Xa-21 in rice.Pyramid lines with two,three,and four resistance genes were developed.Compared with lines with only a single resistance gene,the pyramid lines showed a wider spectrum and a higher level of resistance[42].

Fig.1.From genetic mapping to substitution mapping.P,parent;Q&q,quantitative trait locus;RIL,recombinant inbred line;MAS,marker-assisted selection.

The fourth stage is breeding by design.In the post-genomic era,it is possible for breeders to select desirable genes from the whole genome.By understanding the genetic basis of all agronomically important characters and the allelic variation at those loci,a breeder should be able to design superior genotypes in silico.This concept is called‘‘breeding by design”.The goal of breeding by design can be reached by a three-step approach:(1)mapping loci influencing all agronomically relevant traits;(2)assessing the allelic variation at those loci;and(3)breeding by design using desirable alleles of different loci[1].

5.Development of CSSLs as a genetic resource in crop

CSSLs,including those described as ILs,SSSLs,or NILs,are genetic stocks for the introgression of valuable genes from wild and distantly related species into cultivars and for dissecting QTL of complex traits.In the last three decades,more than 100 sets of CSSLs have been developed in 20 crops[43,44].

As an important food crop and model plant,rice(O.sativa)has a large number of this genetic resource.More than 50 sets of CSSLs have been developed in rice.Most were generated between indica and japonica subspecies using either indica[26,27,30,31,45-53]or japonica[27,28,52-64]as recipient parents.CSSLs or ILs were also created by introgression of chromosome segments from African cultivated rice(O.glaberrima)[65-72]or Oryza wild species such as O.nivara[73,74],O.rufipogon[75-85],O.barthii[86,87],O.meridionalis[88,89],and O.glumaepatula[81,87,90,91]into cultivars.In addition,a set of CSSLs has been created in weedy rice[92].

In hexaploid wheat(Triticum aestivum),three sets of chromosome substitution lines(CSLs)in the spring wheat cultivar Chinese Spring with whole-chromosomes from three donor cultivars[93]and a set of chromosome-arm substitution lines(CASLs)of wild emmer in common wheat[94]have been produced by cytogenetic methods.By MAS,several sets of ILs or CSSLs have been developed from Chinese endemic wheat[95],synthetic wheat[95,96],and Aegilops tauschii[97,98].As with hexaploid wheat,wholechromosome and chromosome-arm substitutions have been produced by cytogenetic methods in tetraploid cotton(Gossypium hirsutum)[99,100].ILs or CSSLs were developed from the crosses among tetraploid cotton species by MAS.Most of the CSSL sets were generated using G.barbadense as donor parent in the G.hirsutum genetic background[101,110].A total of 115 ILs have been obtained from two intraspecific populations of cultivated and feral cotton landraces[111].Two IL sets have been constructed from interspecific populations of G.hirsutum×G.tomentosum[112],and G.tomentosum×G.barbadense[113].In diploid maize(Zea mays),whole-chromosome substitutions were also first produced by cytogenetic methods[114].In comparison with hexaploid wheat,maize has a simpler genetic background.Most CSSL setshave been developed among inbred maize lines[115-121].Wang et al.[122]created a set of alien ILs from Z.mays ssp.mexicana.Mano and Omori[123]generated a set of interspecific ILs from teosinte(Z.nicaraguensis).

Tomato(Lycopersicon esculentum)is one of the earliest crops in which IL libraries were developed by MAS[23].Several libraries of introgression lines in cultivated tomato(L.esculentum)were obtained from the donor parents L.pennellii[23,25],L.hirsutum[32,124],L.pimpinellifolium[124],and Solanum lycopersicoides[125,126].Later,CSSL sets were widely generated in other crops,such as barley(Hordeum vulgare)[127-135],rye(Secale cereale)[136],pearl millet(Pennisetum glaucum)[137],soybean(Glycine max)[138-140],peanut(Arachis hypogaea)[141],pea(Pisum sativum)[142],rapeseed(Brassica napus)[143-146],cabbage(B.oleracea)[147],Chinese cabbage(B.rapa)[148],Indian mustard(B.juncea)[149],lettuce(Lactuca sativa)[150],melon(Cucumis melo)[151,152],cucumber(C.sativus)[153,154],ryegrass(Lolium perenne)[155],and strawberry(Fragaria vesca)[156].

Development of IL or NIL libraries for dissecting complex traits is necessary for functional genomic studies of Arabidopsis thaliana,a model plant in genomic research.Koumproglou et al.[157]generated a population of chromosome substitution strains(CSSs)by replacing chromosomes from the accession Columbia(Col-0)with homologous chromosomes from the accessions Landsberg erecta(L er)and Niederzenz(Nd)of Arabidopsis.Keurentjes et al.[33]created a population of 92 NILs carrying genome-wide chromosomal introgressions from the accession Cape Verde Islands(Cvi)into the L er background.Torjek et al.[158]developed a population of 140 reciprocal NILs,of which 78 were in the Col-0 background and 62 in the C24 background.Fletcher et al.[159]created a population of 75 NILs from an existing RIL population.This new genetic resource is valuable to the Brassicaceae research community.

These CSSLs are widely used to detect QTL and causal genes for crop improvement.Using this genetic resource,a large number of QTL and causal genes for a wide range of traits have been identified.Owing to the small number of CSSLs in each genetic background,however,the application of small CSSL sets in breeding by design is limited.There have been several reviews of the development and use of CSSLs as genetic resources for crop improvement,such as of the introgression of valuable genes from Oryza wild species into cultivated rice[43],of genetic improvement in upland cotton[100],of introgression breeding in food legumes[160],and of genetic improvement in crops[44].

In summary,the substitution of chromosomes or chromosome segments has long been an important method in plant genome research.Prior to the development of molecular markers,CSLs with a pair of alien whole-chromosome substitutions and CASLs with a pair of chromosome-arm substitutions were produced by cytogenetic methods in crops such as wheat[93,94],cotton[99,100],maize[114],and rice[65].The CSLs and CASLs were a valuable genetic resource for using alien chromosomes from wild species and dissecting genes in genomes.The emergence of molecular markers has made it possible to genotype whole genome.Accordingly,CSSLs or ILs with a few chromosome-segment substitutions[23-28]and SSSLs or NILs[30-33]with a single chromosomesegment substitution are widely produced by MAS in plants.They have become a new genetic resource for dissecting QTL for complex traits and for cloning genes(Fig.2).

6.Rice breeding by design via target chromosome-segment substitution

In order to investigate rice breeding by design via target chromosome-segment substitution,we have used a three-step strategy in the past two decades(Fig.3).The first step was to construct a SSSL library using Huajingxian 74(HJX74),an elite xian(indica)rice cultivar,as the recipient of gene resources from the rice AA genome.The second step was to identify a series of desirable genes in the SSSL library.The third step was to perform breeding by design by pyramiding target genes to develop new breeding lines in the HJX74 genetic background[161].

6.1.Construction of a SSSL library for collecting gene resources of the rice AA genome

6.1.1.The recipient and donor parents of the SSSL library

HJX74 was used as the recipient of the SSSL library.HJX74 is an elite cultivar developed by our lab.In 2000,the cultivar was released to farmers,and it is widely planted in southern China.The purpose of using HJX74 as recipient is to improve HJX74 using the desirable genes in the SSSL library.

The SSSL library is intended to collect as many gene resources as possible for rice breeding.So far,43 accessions from seven AA-genome species have been used as donors in three sub-libraries.In the Asian cultivated rice(O.sativa)sub-library,28 accessions have been used as donors,including indica and japonica subspecies originating worldwide.The African cultivated rice(O.glaberrima)sub-library has five donors originating in West Africa.Ten accessions of five wild species:O.nivara,O.rufipogon,O.barthii,O.glumaepatula,and O.meridionalis,have been used as donors to the wild rice sub-library[30,31,45,46,87,89].

6.1.2.Construction of the HJX74-SSSL library

Since 1998,donors have been crossed separately with the recipient HJX74.The F1hybrids were then backcrossed to HJX74.Each cross was screened with 300-400 polymorphic SSR markers to identify substitution segments and the genetic background in each segregating population.After backcrossing 3-7 times with MAS,plants carrying single substitution segments from the donor in the recipient genetic background were selected.Homozygous SSSLs were then developed by selfing[30,31,162].

To date,2360 SSSLs in the HJX74 genetic background have been developed from 43 donors of 7 rice species.There are 1610 SSSLs in the Asian cultivated rice sub-library,190 in the African cultivated rice sub-library,and 560 in the wild rice sub-library.The average length of substitution segments of SSSLs is about 19 cM.The total length of substitution segments of 2360 SSSLs was about 44,800 cM,which is about 29 rice-genome equivalents[30,31,45,46,87,89].Thus,the HJX74-SSSL library has collected abundant genes from the rice AA genome.

6.2.Genetic and molecular detection of naturally occurring variation in the SSSL library

6.2.1.Detection of QTL for complex traits

Like NILs,SSSLs carry a single substitution segment from a donor parent in the genetic background of recipient parent.Phenotype differences between SSSLs and the recipient parent may thus result from the single substitution segment.The SSSLs have been used to detect QTL for complex traits of agronomic importance,such as heading date[163],tiller number[164],blast resistance[165],sheath blight resistance[166],cold tolerance[167],seed dormancy[168],and stigma exsertion rate(SER)[169-171].Recently,a set of SSSLs derived from O.glumaepatula,a wild species,were used to identify QTL for SER.Seven QTL for SER were located on five chromosomes.qSER-3b on chromosome 3 and qSER-9 on chromosome 9 were fine-mapped by secondary substitution mapping[170].In another study,two closely linked QTL for SER were dissected from the respective substitution segments of chromosomes 2 and 3 by secondary substitution mapping[171].These results show that substitution mapping is a powerful tool for dissection of closely linked QTL of complex traits.

Fig.2.Genome composition of several types of substitution lines and their parents.CSL,Chromosome substitution line;CASL,chromosome-arm substitution line;CSSL,chromosome-segment substitution line;IL,introgression line;SSSL,single-segment substitution line;NIL,near-isogenic line.

6.2.2.Cloning of genes and assessment of allelic variation at target loci

SSSLs are powerful tools not only for identifying QTL but also for cloning QTL controlling complex traits.For example,a set of 153 SSSLs derived from Basmati 385,an indica Basmati variety,were used to detect QTL for grain size.Four QTL for grain length(GL)and four QTL for grain width(GW)were identified.The major grain-width locus on the long arm of chromosome 8(GW8)was then cloned.GW8 is synonymous with OsSPL16,and encodes a protein that is a positive regulator of cell proliferation.Sequence and function comparison of OsSPL16 revealed that the critical polymorphism between HJX74 and Basmati 385 is a 10-bp deletion in the OsSPL16 promoter region.gw8HJX74is a wild allele that shows high expression and promotes cell division and grain filling,with positive influence on grain width and yield in rice.In contrast,gw8Basmatiis a loss-of-function mutation associated with the formation of a slender grain and better quality of appearance.OsSPL16 contains an OsmiR156 target sequence and is regulated by miR156.Another allelic variant in the Iranian cultivar Amol3 has a 2 bp InDel at the miR156 target site in OsSPL16,and the gw8Amolallele functions as a loss-of-function mutation.The gw8Amolallele was associated with the increase in grain yield per plant relative to plants carrying the gw8Basmatiallele[172].Recently,the HJX74-SSSLs were used to clone two QTL of agronomic importance,GW6 controlling rice grain size and yield via the gibberellin pathway[173],and qDNR1(Dull Nitrogen Response 1)controlling nitrogen use efficiency and increasing grain yield[174].

For cloned genes,the SSSLs can be used to identify new alleles with different genetic effects.Using a set of SSSLs derived from O.glumaepatula,S23,a locus conferring hybrid male sterility of O.sativa and O.glumaepatula,was cloned.Sequence analysis showed that S23 was allelic with qHMS7,which conferred hybrid male sterility of O.meridionalis and O.sativa[175].However,the genetic effects were not identical.Hybrid male sterility caused by S23 was observed only under short-day condition[176].In another case,a new allele of OsHMA3 was cloned with SSSLs derived from BG367,an Asian cultivar.The new allele of OsHMA3 showed a complete loss of transport activity for cadmium,resulting in greatly increased cadmium transport to the rice shoots and grain[177].

Systematic assessment of allelic variation at loci of agronomic importance is necessary for molecular breeding.As a rich collection of gene resources,the SSSL library can be used to assess allelic variation systematically.For example,16 SSSLs carrying substitution segments with the Wx gene from various donors were selected from the library.From the 16 SSSLs,fvie alleles of the Wx gene:wx,Wxt,Wxg1,Wxg2and Wxg3,were identifeid.The apparent amylose contents(AAC)controlled by the Wx alleles varied from low to high[46,178].The identifciation of fvie alleles of Wx gene in the SSSLs laid a foundation for breeding of rice cultivars with differing AAC.

6.2.3.Detection of gene-by-gene and gene-by-environment interactions

Complex traits are controlled by multiple QTL and affected by environmental factors.Information about gene-by-gene and gene-by-environment interactions is essential for breeding by design.Using SSSLs,epistasis of QTL for heading date[179-181],an additive-by-environment interaction effect of QTL for panicle number[182]and for yield-component traits[183],and dynamic QTL for tiller number[184]were detected.Thus,SSSLs are powerful tools for detecting gene-by-gene and gene-by-environment interaction effects.

6.3.Breeding by design on the SSSL library platform

6.3.1.Breeding by design for complex traits

First,multiple QTL for targeted complex traits should be identified in the SSSL library.The SSSLs carrying target QTL are selected for trait design.The multiple target QTL located in different SSSLsare then pyramided to construct target complex traits.We are working on breeding by design for several complex traits,such as grain size,low chalkiness,high cooked-rice elongation,and high stigma exsertion rate.Some pyramid lines with target QTL are under development.

Fig.3.Construction of the HJX74-SSSL library and its application in rice breeding by design.MAS,marker-assisted selection.

6.3.2.Breeding by design for breeding lines

Breeding lines are usually necessary for variety breeding.For example,hybrid rice is developed using breeding lines:malesterile,maintainer,and restorer lines.We have developed a platform for breeding of cytoplasmic male-sterile(CMS)lines and maintainer lines.First,a maintainer line H121B carrying the rf3 and rf4 genes from XieqingzaoB in the HJX74 genetic background was developed.Three CMS lines with different sources of sterile cytoplasm,Zhenshan97A of WA-CMS,XieqingzaoA of DA-CMS,and HuanongA of YA-CMS,were used to develop isonuclear alloplasmic CMS lines in the HJX74 genetic background.Three isonuclear alloplasmic CMS lines with the genetic background of HJX74:WA-H121A,DA-H121A and YA-H121A,were developed using H121B as maintainer.H121B was then improved using the MADS50,gs3,and Wxtgenes in SSSLs to develop a new maintainer line H131B.Three new isonuclear alloplasmic CMS lines:H131A with WA-,DA-and YA-CMS,were developed using H131B as maintainer[185].In another effort,two maintainer lines HZB and HBB with the rf3 and rf4 genes from Zhenshan97B and BobaiB in the HJX74 genetic background were developed separately.Using HZB as the nuclear donor,three isonuclear alloplasmic CMS lines,HZA with WA-,DA-and YA-CMS,were developed.HXB and HBB were then improved using the genes Wxg1,ALK,fgr,gs3,and MADS50 in E5.Two new maintainer lines,E5-HXB and E5-HBB,were developed separately.Using E5-HXB and E5-HBB as maintainer lines,two new CMS lines,E5-HXA and E5-HBA,were developed separately.In comparison with the E5-HXA line carrying the rf3 and rf4 genes of XB,the CMS of the E5-HBA line carrying the rf3 and rf4 genes from BB was more sensitive to environmental conditions and more easily restored[186].

We have also developed a platform for breeding restorer lines.To assess allelic variation at the Rf3 and Rf4 loci,57 SSSLs carrying one of the loci in the substitution segments were selected from the library.In the set of SSSLs,four alleles were identified at each locus:Rf31,Rf32,Rf33,and Rf34and Rf41,Rf42,Rf43,and Rf44,respectively,ranging from weak to strong with respect to their restoring ability.HJX74 carried the Rf34allele at the Rf3 locus and a weaker allele at the Rf4 locus.One SSSL carried the Rf44allele and showed strong CMS-restoring ability[187].The SSSL with the Rf34and Rf44genotype was then improved by pyramiding of the gs3,gw8,Wxg1,and Alk genes in SSSLs,and a new restorer line H121R was developed.H121R was further improved by pyramiding of the blast resistance gene qBLAST11 in a SSSL,and a new restorer line H131R with blast resistance was developed[188].

6.3.3.Breeding by design for new varieties

The ultimate goal of breeding by design on the platform of the SSSL library is to breed new cultivars for rice production.Many ofthe practices of breeding by design have been conducted on this platform.

As one example,a SSSL carrying the Pb gene on the substitution segment of chromosome 4 from a donor was selected from the library.The SSSL showed almost the same phenotype as HJX74 except carrying the purple pericarp of the donor.After comprehensive evaluation,the SSSL become a new cultivar named Huaxiaohei 1 and was released to farmers in 2005.Huaxiaohei 1 was then improved by pyramiding of the Wxtand gs3 genes in SSSLs to develop a new cultivar,Huaxiaohei 2.In the same way,the pyramiding of desirable genes on this platform has been conducted step by step.Through breeding by design,a series of new cultivars or lines of‘‘black rice”will be developed on this platform.

A series of new cultivars or lines of‘‘red rice”have been developed on this platform.First,a SSSL carrying the Rc gene on the substitution segment of chromosome 7 from BG367 was selected from the SSSL library.The SSSL showed almost the same phenotype as HJX74 except with the red pericarp of the donor.As a new cultivar,the SSSL was named Huaxiaohong 1.Huaxiaohong 1 was then improved by pyramiding of the Wxg1,alk,gs3,and fgr genes from SSSLs to develop a new cultivar,Huaxiaohong 2.

A series of new cultivars or lines of‘‘white rice”have also been developed on this platform.Because HJX74 lacks high grain quality,it needs improvement to become a good-quality cultivar.According to the design for good grain quality,the pyramiding of three substitution segments carrying the four target genes Wxg1,alk,gs3,and gw8 in three SSSLs can meet the design requirements.The resulting pyramid line with three substitution segments in the HJX74 genetic background showed good grain quality,reaching the design goal.The pyramid line was named Huabiao 1 and released to farmers in 2009.Huabiao 1 was then improved further by pyramiding of the fgr aroma gene,and the resulting new good-quality variety with aroma was named Huabiao 3.

Thus,the HJX74-SSSL library is a powerful platform for rice breeding by design.Using target chromosome-segment substitution on the platform,a variety of complex traits,breeding lines,and new cultivars can be developed to meet the needs of rice breeding(Fig.3).

7.Concluding remarks and prospects

Progress in plant breeding depends on the integration and use of genetic resources,genetic knowledge,and breeding techniques.Although artificially induced variation has been widely used in plant genetic and physiological research,naturally occurring variation is a useful genetic resource for plant breeding.The core of plant breeding is the use of naturally occurring variation.

A large number of SSSLs in the HJX74 genetic background have been developed from genetically diverse rice AA-genome donors.The HJX74-SSSL library has assembled abundant genes from available rice genetic resources.In particular,the library has collected a large number of genes from wild species with the rice AA genome that retain genetic variation otherwise lost during the domestication of cultivated rice.The SSSLs are helpful for rapid screening of traits hidden in wild genomes,make a large amount of previously unexplored genetic variation rapidly available to plant breeders and geneticists,and make the genetic variation directly usable for breeding.The library represents a new resource that could greatly enrich conventional rice breeding.The development of the HJX74-SSSL library is a huge project that cost much time and labor.Obviously,the library is an unusual gene pool for genetic research and rice breeding.

The HJX74-SSSL library integrates the mining and utilization of desirable alleles in this gene pool.Once the alleles are detected in the library,they can be used directly for breeding.On the basis of the HJX74 variety,any deficiency of this variety can be corrected using the favorable alleles in the library via target chromosomesegment substitution.In this way,the variety can be improved repeatedly.As in an assembly line,designed breeding goals can be achieved step by step when favorable alleles are pyramided recurrently on the platform.Any rice products that we might design,such as complex traits,breeding lines,and new cultivars,can be developed on the platform.Thus,the HJX74-SSSL library is a powerful platform for rice breeding by design and provides an example of breeding by design in crops.

The HJX74-SSSL platform will integrate the technologies of genome editing,gene transfer,next-generation sequencing,and many other related technologies,to increase the potential of rice breeding by design.Genome-editing technology provides the means to quickly modify genomes in an accurate and predictable way,making it possible to directly introduce improvements into elite cultivars[189].Accordingly,this technology is believed to be a way to breeding by design[190].Although only genes with known functions in the genome can be edited,it can be used as an important additional technology for breeding by design.Thus,target chromosome-segment substitution is a way to breeding by design.Integration of other advanced technologies will further increase the design breeding ability of the HJX74-SSSL platform.

CRediT authorship contribution statement

Guiquan Zhang:Writing-review&editing.

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 Major Program of Transgenic New Variety Breeding of China(2009ZX08009005)and the National Natural Science Foundation of China(91435207 and 91735304).