Hu Yun,Zhengyn Xu,Xueqin Tn,Peng Go,Mengy Jin,Wencheng Song,Shigung Wng,Yunhi Kng,Peixiong Liu,Bin Tu,Yuping Wng,Peng Qin,Shigui Li,Bingtin M,*,Weiln Chen,*

a State Key Laboratory of Crop Gene Exploration and Utilization in Southwest China,Rice Research Institute,Sichuan Agricultural University,Chengdu 611130,Sichuan,China

b Guangdong Key Laboratory of New Technology in Rice Breeding,Rice Research Institute,Guangdong Academy of Agricultural Sciences,Guangzhou 510640,Guangdong,China

Keywords:Rice Panicle length Grain number per panicle Quantitative trait locus TAW1

ABSTRACT Grain number per panicle (GNP) is a complex trait controlled by quantitative trait loci (QTL),directly determining grain yield in rice.Identifying GNP-associated QTL is desirable for increasing rice yield.A rice chromosome segment substitution line (CSSL),F771,which showed increased panicle length and GNP,was identified in a set of CSSLs derived from a cross between two indica cultivars,R498 (recipient) and WY11327(donor).Genetic analysis showed that the panicle traits in F771 were semidominant and controlled by multiple QTL.Six QTL were consistently identified by QTL-seq analysis.Among them,the major QTL qPLN10 for panicle length and GNP was localized to a 121-kb interval between markers N802 and N909 on chromosome 10.Based on quantitative real-time PCR and sequence analysis, TAWAWA1(TAW1),a known regulator of rice inflorescence architecture,was identified as the candidate gene for qPLN10.A near-isogenic line,NIL-TAW1,was developed to evaluate its effects.In comparison with the recurrent parent R498,NIL-TAW1 showed increased panicle length (14.0%),number of secondary branches (20.9%) and GNP (22.0%),and the final grain yield per plant of NIL-TAW1 was increased by 18.6%.Transgenic experiments showed that an appropriate expression level of TAW1 was necessary for panicle development.Haplotype analysis suggested that the favorable F771-type (Hap 13) of TAW1 was introduced from aus accessions and had great potential value in high-yield breeding both in indica and japonica varieties.Our results provide a promising genetic resource for rice grain yield improvement.

1.Introduction

Rice is one of the most important staple foods in the world,and continually improving rice yield is an enduring goal of breeders.Rice grain yield is determined by four components:grain number per panicle (GNP),panicle number,seed setting rate,and 1000-grain weight [1,2].GNP is a complex quantitative trait and contributes the most to grain yield.GNP is determined largely by panicle length and numbers of primary and secondary branches [3].

Several genes or quantitative trait loci (QTL) have been cloned and found to be involved in panicle development.For example,LAX PANICLE 1(LAX1) andLAX2are required for the initiation or maintenance of axillary meristems in rice panicles,andlax1andlax2mutants show defects in producing lateral branches in panicles [4–6].DENSE AND ERECT PANICLE 1(DEP1),DEP2,DEP3,Short panicle 1(SP1),andSP3have been reported to influence mainly panicle length in rice;mutants in both genes have short panicles[7–11].Panicle branches are key determinants of GNP,and several genes have been reported to regulate GNP by influencing the numbers of primary or secondary branches,includingGrain number 1a(Gn1a)[12];LARGER PANICLE(LP)[13];DROUGHT AND SALT TOLERANCE(DST)[14];NARROW LEAF 1(NAL1)[15,16];Growth Regulating Factor 6(OsGRF6) [17];Grain Number per Panicle 1(GNP1) [18];NUMBER OF GRAINS 1(NOG1) [19];FRIZZY PANICLE(FZP) [20–22];andIdeal Plant Architecture 1(IPA1) [23–25].

Usually,a longer duration of panicle differentiation is beneficial for the formation of larger panicles and more panicle branches,which lead to higher GNP.Several genes with pleiotropic effects on grain number,plant height,and heading date have been cloned,includingGrain number,plant height,and heading date 2(Ghd2)[26],Ghd7[27],Ghd7.1[28,29],andGhd8[30,31].Increased expression of these genes under long-day conditions results in delayed heading and increased plant height and grain number.The development of multi-floret spikelets has recently been proposed [32–35] for increasing GNP and grain yield in rice.LATERAL FLORET 1(LF1) encodes a class III homeodomain-leucine zipper (HD-ZIP III)protein,and a mutation in the microRNA165/166 target sequence ofLF1leads to ectopic expression of the meristem maintenance geneOSH1and induces initiation of lateral floral meristems [33].DOUBLE FLORET 1(DF1) encodes a lipase that functions directly in the regulation of spikelet determinacy [32].Both thelf1anddf1mutants generate multi-floret spikelets,which have the potential to increase GNP and rice yield.

TAW1,encoding an ALOG family protein of unknown function,is a unique regulator of inflorescence meristem activity and spikelet meristem phase transitions in rice [36].The dominant gain-offunction mutanttaw1-Dwith increased expression ofTAW1shows increased numbers of panicle branches and spikelets.In contrast,loss-of-function mutant and RNA interference plants ofTAW1produce small inflorescences with reduced panicle branches,indicating thatTAW1is a positive regulator of inflorescence development [36].However,natural alleles ofTAW1that could be directly applied to breeding have not been identified.

In this study,we identified a CSSL,F771,with increased panicle length and GNP,derived from a cross between twoindicacultivars R498(recipient)and WY11327(donor).Based on QTL-seq and phenotypic analysis of recombinants,one major QTLqPLN10for panicle length and GNP was localized to a 121-kb interval on chromosome 10.By qRT-PCR and sequence analysis,TAW1was confirmed as the candidate gene ofqPLN10.NIL-TAW1showed increased panicle length,GNP,and grain yield.Thus,the natural allele ofTAW1identified here can contribute to high yield in rice breeding.

2.Materials and methods

2.1.Plant materials and growth conditions

The rice CSSL F771,with increased panicle size and GNP,was identified in a set of CSSLs (BC4F5) derived from a cross between an eliteindicacultivar,Shuhui 498 (R498),as the recipient parent and a particularly large-panicleindicacultivar,Waiyin-11327(WY11327),as the donor parent.F771 was crossed with R498 as recurrent parent to generate a segregating BC5F2population for primary QTL mapping.Heterozygous recombinants were selected from the BC5F2population and selfed to generate BC5F3homozygous recombinants with different genotypes for fine mapping ofqPLN10.Simultaneously,a near-isogenic line,NIL-TAW1,which harbored only a short chromosomal segment containingTAW1from F771 without other segments,was developed from the BC5F3population by marker-assisted selection (MAS).All plants were grown in experimental fields of Sichuan Agricultural University either at Wenjiang (Sichuan) or at Lingshui (Hainan) under normal cultivation conditions.A schematic diagram showing the construction of plant materials and planting locations and years is shown in Fig.S1.

2.2.Phenotypic investigation

For the BC5F2population,the main panicle of each plant was harvested at the maturity stage for phenotypic investigation,including panicle length and GNP.F771 and NIL-TAW1plants were grown in a randomized block design with at least three replicates.Each block comprised three rows with 10 plants per row,and the transplant spacing was 16.7 × 26.7 cm.At maturity,five middle plants in each block were randomly harvested and bulked for recording agronomic traits including plant height,panicle length,numbers of primary and secondary branches,GNP,1000-grain weight,seed setting rate,tiller number per plant,and grain yield per plant.The 1000-grain weight was measured using SC-G software(Wanshen Detection Technology Co.,LTD.,Hangzhou,China).Significance tests were performed using Student’st-tests.All data were analyzed in Microsoft Excel 2019 (Microsoft,Redmond,WA,USA).

2.3.Scanning electron microscopy (SEM)

Fresh young panicles from R498 and F771 were collected and fixed in 2.5%glutaraldehyde overnight at 4°C and then dehydrated in a graded ethanol series from 30% to 100%.The samples were then dried and sputter-coated as previously described [37],and examined by SEM (JSM-7500F,JEOL,Tokyo,Japan).

2.4.QTL-seq analysis

QTL-seq analysis was performed as previously described[38].In brief,35 BC5F2individuals with extremely large panicles (LP) and small panicles (SP) were selected,and genomic DNA of each selected plant was extracted and bulked in equal proportions to generate LP and SP bulks.Each DNA bulk was then subjected to whole-genome resequencing with an MGISEQ-2000 sequencer(MGI Tech Co.Ltd,Shenzhen,China).The SNP index in each bulk was calculated as previously described [39],the Δ(SNP index)was defined as (SNP index of LP bulk)– (SNP index of SP bulk),and a graph showing the relationship between Δ(SNP index) and genomic position was plotted to identify putative genomic regions associated with the LP phenotype.

2.5.Development of InDel and SNP markers

For fine-mapping ofqPLN10,17 polymorphic markers were developed according to the sequence differences obtained from QTL-seq,including 15 InDel and two SNP markers.InDel markers were designed using Primer Premier 5.0 (www.PremierBiosoft.-com),and SNP markers for amplification-refractory mutation system (ARMS) PCR were identified with the online PRIMER1 system(http://primer1.soton.ac.uk/primer1.html).To identify NIL-TAW1,one ARMS-PCR marker (N973) was developed based on the polymorphic site–1863 in the promoter ofTAW1between R498 and F771,for use in MAS in breeding.All primer sequences are listed in Table S1.

2.6.Quantitative realtime PCR (qRT-PCR)

Total RNA was extracted from tissues using Plant Total RNA Isolation Kit (FOREGENE,Chengdu,China).First-strand cDNA was synthesized from 500 ng of total RNA using RT Easy II (with gDNase)(FOREGENE).qRT-PCR analysis was performed using KAPA SYBR FAST qPCR Kit (KAPA,Boston,MA,USA) on a qTOWER3G Real-Time PCR thermocycler (Analytik Jena AG,Jena,Germany).OsActin1was used as internal control.Three replicates were used for each assay.The primers for qRT-PCR assays are listed in Table S2.

2.7.Vector construction and transformation

Two fragments,containing the 2 kb promoter region,5′untranslated region (UTR),entire coding region and 0.8 kb 3′UTR ofTAW1were amplified separately from F771 and R498,and then subcloned into the binary vector pCAMBIA1300 using ClonExpress II One Step Cloning Kit(Vazyme,Nanjing,China).The resulting vectors were introduced intoAgrobacterium tumefaciensstrain EHA105 and then transformed intojaponicavariety ZH11.The primers for vector construction are listed in Table S3.

Fig.1.ComparisonofphenotypesbetweenR498andF771.(A)PlantarchitectureofR498andF771.Scalebar,10cm.(B)PaniclearchitectureofR498andF771.Scalebar,3cm.(C)ComparisonofgrainnumberperpaniclebetweenR498andF771.Scalebar,3cm.(D,E)Statisticalanalysisofpaniclelength(D)andgrainnumberperpanicle(E)of R498andF771.Valuesaremeans±SD(n=15plants).P-valueswerecalculatedusingStudent’st-tests;**,significantdifferenceatP<0.01.(F–I)Scanningelectronmicrographs ofyoungpaniclesatearlydevelopmentalstagesinR498andF771.(F,H)Youngpaniclesof0.1–0.5mm.(G,I)youngpaniclesof0.5–1.5mm.PBM,primarybranchmeristem;SBM,secondarybranchmeristem;fm,floralmeristem;le,lemma;pa,palea;sl,sterilelemma;rg,rudimentaryglume.

2.8.Dualluciferase transient assays

To compare the R498-type and F771-type promoter activity ofTAW1,the 2.1 kb promoter fragments upstream of theTAW1translation start site were separately amplified from R498 and F771 and then subcloned into a pGreenII 0800-LUC vector containing the LUC reporter gene.The resulting vectors were transfected into rice protoplasts for transient expression,andRenillaluciferase (REN)was used as internal control.Relative luciferase activity LUC/REN was measured using the Dual Luciferase Reporter Gene Assay Kit(Beyotime,Shanghai,China).The primers for vector construction are listed in Table S3.

2.9.Haplotype analysis

SNP and InDel variation information ofTAW1in rice accessions was obtained using Rice Variation Map 2.0 (http://ricevarmap.ncpgr.cn),and haplotype analysis was performed based on the variation.Only haplotypes found in more than five accessions were selected.A phylogenetic tree of these haplotypes based on their nucleotide sequence was constructed with MEGA 5.1 (http://www.megasoftware.net) using the neighbor-joining method with default parameters and 1000 bootstrap replications.

3.Results

3.1.Identification and characterization of F771

To identify QTL for GNP in rice,a set of CSSLs was generated from twoindicacultivars R498 and WY11327 as recipient and donor parents(Fig.S1),respectively.Among the CSSLs,F771,which showed increased plant height,panicle size,and GNP (Fig.1A–C),was identified in the BC4F5population(Fig.S1).In comparison with the recurrent parent R498,the panicle length and the GNP were increased by respectively 27.9% and 39.1% (Fig.1D and E).

Fig.2.Fine mapping and candidate gene analysis of qPLN10.(A) Graphical genomic region of qPLN10 identified by QTL-seq analysis.Black and white bars indicate genome fragments from F771 and R498,respectively.CEN,centromere.(B) Primary mapping of qPLN10 by screening recombinants.Numbers of recombinants are shown under the bar.(C)Fine mapping of qPLN10 based on phenotypes of homozygous recombinants.PL,panicle length;GNP,grain number per panicle.Values are means±SD(n=10 plants).** i ndicates significant difference at P <0.01 (Student’s t-test).Black and white rectangles indicate the homozygous F771 and R498 genotypes,respectively.(D) Putative candidate genes in the mapping region according to the Rice Genome Annotation Project(http://rice.plantbiology.msu.edu/).(E)Comparison of expression levels of TAW1 in R498 and F771.YP1 and YP2 indicate young panicles with lengths of 1 cm and 2 cm,respectively.Expression levels in R498 were set to one.Values are means ± SD (n=3 replicates).** i ndicates significant difference at P <0.01 (Student’s t-test).(F) A diagram showing the sequence differences of TAW1 between R498 and F771.The black rectangle indicates the exon,white rectangles indicate 5′ and 3′ UTRs,and black lines indicate introns and promoter.Positions of polymorphic sites are shown in the diagram.(G) Schematic diagram of two reporter constructs for dual-luciferase transient assays.(H) Relative luciferase activity was measured in rice protoplasts transfected with the reporter construct.Values are means ± SD (n=3 replicates).** in dicates significant difference at P <0.01 (Student’s t-test).

Fig.3.Characterization of NIL-TAW1.(A) Plant architecture of R498 and NIL-TAW1.Scale bar,10 cm.(B) Panicle architecture of R498 and NIL-TAW1.Scale bar,3 cm.(C)Comparison of grain number per panicle between R498 and NIL-TAW1.Scale bar,3 cm.(D)Comparison of the relative expression level of TAW1 in R498 and NIL-TAW1.YP1 and YP2 indicate young panicles with lengths of 1 and 2 cm,respectively.Expression levels in R498 were set to one.Values are means ± SD (n=3 replicates).** i ndicates significant difference at P <0.01 (Student’s t-test).(E–N) Statistical analysis of plant height (E),panicle length (F),number of primary branches (G),number of secondary branches(H),grain number per panicle(I),1000-grain weight(J),seed setting rate(K),number of tillers per plant(L),grain weight per panicle(M)and grain yield per plant(N) of R498 and NIL-TAW1.Values are means ± SD (n=6 replicates).** i ndicates significant difference at P <0.01 (Student’s t-test).

To investigate the cytological mechanism of panicle development,we performed SEM (Fig.1F–I).The panicle development of F771 was markedly delayed.For example,when panicles of R498 and F771 were 0.1–0.5 mm in length,the secondary branches and spikelet primordia of R498 had clearly emerged (Fig.1F),F771,however,was still in the primary branch primordia differentiation period (Fig.1H).Spikelet differentiation was initiated in R498 when F771 was still in the stage of secondary branch primordia differentiation (Fig.1G and I).These results suggested that F771 might have delayed and prolonged branch formation and spikelet differentiation period,leading to increased spikelet number.

3.2.Genetic analysis and fine-mapping of qPLN10

F771 was crossed with the recurrent parent R498 to generate BC5F1progeny,and a BC5F2segregating population comprising 585 plants was developed by self-pollination of BC5F1(Fig.S1).As F771 showed differences in panicle length and grain number(Fig.1D and E),these traits were investigated in the BC5F2population.GNP was positively correlated with panicle length (Fig.S2A),suggesting that the panicle traits of F771 were controlled by the same genes.In the BC5F2population,panicle length and grain number showed normal distributions (Fig.S2B and C),suggesting that these traits in F771 were controlled by multiple QTL.Most plants showed panicle length and grain number intermediate between their parents (Fig.S2B and C),and the panicle length and GNP of BC5F1plants were between those of R498 and F771(Fig.S3).These results indicated that the QTL for panicle traits of F771 were semidominant.

For rapid location of this panicle trait QTL,the whole-genome resequencing-based QTL-seq [38] was performed.Six putative QTL were identified on chromosomes 2,6,7,9,10,and 12 based on the Δ(SNP index) plot (Fig.S4A and B;Table S4),in agreement with the genetic analysis showing that the panicle traits of F771 were controlled by multiple QTL (Fig.S2B and C).Among them,qPLN10(Panicle Length and Grain Number 10),located on the long arm of chromosome 10,showed the largest Δ(SNP index) values(Fig.S4A;Table S4).To narrow the region ofqPLN10,several polymorphic markers were developed using the QTL-seq results.By identification of recombinants in the BC5F2population,qPLN10was first mapped within the interval between N800 and N788(Fig.2A and B).Eight heterozygous recombinants were selfed to generate BC5F3homozygous recombinants with different genotypes (Fig.2C).The panicle length and GNP of L1,L2,L3,and L5 were similar to those of R498,however,the panicle length and GNP of L4,L6,L7,and L8 were significantly greater than those of R498.Based on these results,qPLN10was further localized to a 121-kb region between markers N802 and N909 (Fig.2C).

3.3.Candidate gene analysis for qPLN10

According to the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/),nine predicted genes were located in the 121-kb region (Fig.2D;Table S5).Five of these showed extremely low or no expression in inflorescences (Table S5),including two En/Spm-like transposon proteins (LOC_Os10g33740andLOC_Os10g33750),a no apical meristem protein(LOC_Os10g33760),a leucoanthocyanidin reductase (LOC_Os10g33774) and a MYBrelated protein (LOC_Os10g33810).These genes are unlikely to be involved in regulating panicle development.The remaining four genes were preferentially considered as possible candidate genes,including a hypothetical protein (LOC_Os10g33770),a reported inflorescence architecture regulatorTAW1(LOC_Os10g33780) [36],a heat shock protein (LOC_Os10g33790),and a cytoplasmic malate dehydrogenase (LOC_Os10g33800).

Fig.4.Characterization of transgenic plants of TAW1.(A) Relative expression levels of TAW1 in ZH11 and transgenic plants. TAW1F771 and TAW1R498 indicate plants transformed with F771-type and R498-type TAW1,respectively.Values are means ± SD (n=3 replicates).** i ndicates significant difference at P <0.01 (Student’s t-test).(B)Plant architecture of ZH11 and transgenic plants at maturity.Scale bar,10 cm.(C) Panicle morphology of ZH11 and transgenic plants.Scale bar,3 cm.The white arrow indicates degenerated spikelets.(D–M) Statistical analysis of plant height (D),panicle length (E),number of primary branches (F),number of secondary branches (G),grain number per panicle (H),1000-grain weight (I),number of tillers per plant (J),seed setting rate (K),grain weight per panicle (L),and grain yield per plant (M) of ZH11 and transgenic plants.Values are means ± SD (n=4 replicates).* and ** respectively indicate significant difference at P <0.05 and P <0.01 (Student’s t-test).

Comparison by qRT-PCR of expression levels of the four candidate genes in developing young panicles between R498 and F771 showed that onlyTAW1was significantly up-regulated in F771(Figs.2E,S5).Sequence comparison showed no differences in the coding region ofTAW1between R498 and F771,but revealed 11 polymorphisms in the promoter region ofTAW1,including one 3-bp deletion and 10 base substitutions (Fig.2F).Consistently,dual-luciferase transient expression assays in rice protoplasts showed that the F771-type promoter ofTAW1had higher relative activity than the R498-type(Fig.2G and H).Thus,TAW1was likely the candidate gene forqPLN10,and polymorphisms in the promoter ofTAW1could account for the phenotypic difference.

3.4.Development and characterization of NIL-TAW1

To evaluate the effect ofTAW1,we developed the near-isogenic line NIL-TAW1from the BC5F3population using MAS(Figs.3A,S1).NIL-TAW1harbored only a short chromosomal segment containingTAW1from F771 and lacked other segments.Expression analysis confirmed thatTAW1was significantly up-regulated in NIL-TAW1(Fig.3D).The plant height and architecture of NIL-TAW1were similar to those of R498 (Fig.3A and E),but the panicle length of NIL-TAW1was increased by 14.0%(Fig.3B and F).There was no significant difference in the number of primary branches,but the number of secondary branches was significantly increased by 20.9% (Fig.3G and H),and the GNP was correspondingly significantly increased by 22.0% (Fig.3C and I).The 1000-grain weight and seed setting rate of NIL-TAW1were slightly decreased by 4.8% and 3.8%,respectively (Fig.3J and K).The tiller number per plant was comparable to R498(Fig.3L).The grain weight per panicle and grain yield per plant of NIL-TAW1were significantly increased by 21% and 18.6% (Fig.3M and N),respectively.

In view of the changes in panicle architecture and GNP in NILTAW1,we investigated the expression levels of several cloned genes associated with panicle architecture or GNP in developing young panicles.Four of these genes were significantly up-regulated in NIL-TAW1(Fig.S6).OsMADS22andOsMADS55,which were positive regulators of GNP and function downstream ofTAW1[36],were increased by respectively 2.1 and 5.3 times in NIL-TAW1.

3.5.Functional analysis of TAW1 in transgenic lines

Fig.5.Expression pattern analysis of TAW1.(A)Expression pattern of TAW1 obtained from ePlant Rice(http://bar.utoronto.ca/eplant_rice/).(B)Expression of TAW1 in various tissues investigated by qRT-PCR.Root was sampled at the seedling state;leaf blade 1 and leaf blade 2 indicate leaf blades at the seedling and booting stages,respectively;leaf sheath and stem were sampled at the booting stage;numbers for inflorescence and hull indicate inflorescence length (cm);numbers for caryopsis indicate days after fertilization.Values are means ± SD (n=3 replicates).

To further verify the functions of different types ofTAW1,two fragments,containing the 2 kb promoter region,5′untranslated region(UTR),entire coding region,and 0.8 kb of the 3′UTR ofTAW1were separately amplified from F771 and R498 and introduced intojaponicacultivar ZH11.Two independent transgenic lines with increased expression ofTAW1were obtained and investigated(Fig.4A).Compared with the wild type (WT),all transgenic lines showed decreased plant height (Fig.4B and D);the panicle length was significantly decreased by 33.2% on average (Fig.4C and E);and parts of the apical spikelets were degenerated (Fig.4C).The lengths of primary branches were markedly decreased,and the secondary branches were clustered on the primary branch(Fig.4C).The numbers of primary and secondary branches were reduced by 28.4%and 16.7%(Fig.4F and G),respectively,and thus the GNP was significantly reduced by 51.9% (Fig.4H).The 1000-grain weight was slightly but significantly reduced by 5.9%(Fig.4I).The tiller number and seed setting rate were comparable to those of the wild type(Fig.4J and K).The grain weight per panicle and grain yield per plant were significantly reduced by 56.2% and 47.2% (Fig.4L and M),respectively.

Expression pattern analysis ofTAW1using the Rice eFP Browser(http://www.bar.utoronto.ca/efprice/cgi-bin/efpWeb.cgi) and the Rice Expression Profile Database (https://ricexpro.dna.affrc.go.jp/)showed thatTAW1was highly expressed mainly in root,young leaf,leaf sheath,stem,and caryopsis;however,the expression in inflorescences was relatively low(Figs.5A,S7).The expression patterns ofTAW1in various organs was further confirmed by qRT-PCR(Fig.5B).Combined with the panicle phenotype of transgenic plants(Fig.4C),these results suggested that an appropriate expression level ofTAW1is necessary for its function in regulating panicle development.

3.6.Haplotype analysis of TAW1

To investigate the haplotype ofTAW1,we used the Rice Variation Map 2.0 (http://ricevarmap.ncpgr.cn) to identify natural variation ofTAW1in 4262 germplasm accessions,including 2492indica,198aus,1421japonica,and 151 intermediate accessions.Of 64 polymorphisms identified,40 were located in the 2-kb promoter region,10 in the 5′UTR,four in the intron,and 10 in the coding region (Table S6).These accessions were divided into 17 haplotypes (Hap 1–17) based on 43 polymorphisms (Fig.6A).Hap 1–3 and Hap 9–12 were identified mainly injaponicaaccessions,Hap 4–8 and Hap 15–17 inindicaaccessions,and Hap 13 and Hap 14 inausaccessions (Fig.6B).Phylogenetic analysis showed that the 17 haplotypes were divided into two clades:theindicaandausaccessions belonged mainly to clade I and thejaponicaaccessions belonged mainly to clade II (Fig.6C),indicating divergence ofTAW1betweenindicaandjaponicaaccessions.R498 and F771 harbored Hap 7 and Hap 13,respectively (Fig.6C).Frequency analysis showed that the favorable Hap 13 was present mainly inausaccessions and a small number ofindicaaccessions,but was almost completely absent injaponicaaccessions (Fig.6D).

Fig.6.Haplotype analysis of TAW1 in rice germplasms.(A) Natural variations and haplotypes of TAW1 in 4262 germplasms.Sites polymorphic between R498 and F771 are highlighted in red.(B)Distribution of haplotypes.(C)Phylogenetic tree of haplotypes was constructed based on nucleotide sequence using the neighbor-joining method with MEGA 5.1.Numbers indicate bootstrap support based on 1000 replications.(D) Frequency analysis of Hap 13 of TAW1.

4.Discussion

4.1.Identification of a major QTL for panicle length and GNP

GNP,one of the direct determinants of grain yield in rice,is a complex quantitative trait determined largely by panicle architecture,including panicle length,and numbers of primary and secondary branches [1].To date,multiple genes/QTL associated with GNP have been cloned and characterized.In this study,the CSSL F771 with increased panicle length and GNP was identified from a set of CSSLs,which were derived from a cross between twoindicacultivars R498 (recipient) and WY11327 (donor).Panicle traits in F771 were controlled by multiple QTL,as six QTL for panicle length and grain number were identified on six different rice chromosomes,namely,qPLN2,qPLN6,qPLN7,qPLN9,qPLN10,andqPLN12(Fig.S4).Among them,qPLN10exhibited the largest Δ(SNPindex) values (Table S4),and was localized to a 121-kb interval between markers N802 and N909 based on the phenotypes of homozygous recombinants (Fig.2C).Nine predicted genes were located in this interval,according to the Rice Genome Annotation Project(Fig.2D;Table S5).Through qRT-PCR and sequence analysis(Figs.2E,S5),TAW1(LOC_Os10g33780),a known inflorescence architecture regulator [36],was identified as the likely candidate gene forqPLN10.

4.2.Genetic background is important for the utilization of TAW1

In a previous study [40],NIL-taw1(BC5F5),carryingtaw1-D2(a mutation ofTAW1),was developed in thejaponicacultivar Koshihikari background and its effects on yield and yield-related traits were characterized.In the present study,a near-isogenic line NIL-TAW1,harboring theTAW1from F771,was developed in the eliteindicacultivar R498 background to evaluate its effects(Figs.3,S1).In agreement with the previous study[40],our results showed thatTAW1did not affect tiller number per plant(Fig.3L),butTAW1increased GNP owing to an increased number of secondary branches rather than primary branches (Fig.3G–I).NIL-TAW1showed no significant change in seed setting rate (Fig.3K),and the final grain yield per plant of NIL-TAW1increased by 18.6%(Fig.3K).However,Fukushima et al.[40]found that the grain yield of NIL-taw1was significantly lower than that of Koshihikari,and speculated that this was because Koshihikari did not provide a suitable genetic background for high yield owing to late heading in the cold northern region of Japan,leading to markedly decreased seed set in NIL-taw1.Thus,the genetic background may be an important factor for effective use ofTAW1in high-yield breeding.

4.3.Appropriate expression level of TAW1 is essential for panicle development

Several studies [41–44] have demonstrated that appropriate expression levels are important for the functions of genes such asOsSPL7,OsSPL14,OsSPL16,OsSPL17,OsSPL18,D2/SMG11,andGNP6/MOC1.In a previous study,taw1-D1heterozygous andtaw1-D2homozygous mutants,which showed increased expression ofTAW1,showed an increased number of secondary branches.In contrast,loss-of-functiontaw1-3mutants and RNA interfere lines ofTAW1showed small panicles with reduced branches [36],indicating thatTAW1is a positive regulator of panicle size and panicle branches.In the present study,the expression ofTAW1was increased by respectively 2.3 and 1.6 times in F771 and NILTAW1in comparison with R498 (Figs.2E,3D),resulting in large panicles with increased secondary branches (Figs.1B,3B and H).

However,an excessively high expression level ofTAW1led to defective phenotypes,such as in homozygoustaw1-D1mutants,which showed suppressed stem elongation and inflorescences that could not emerge from the leaves but formed an agglomerate with multiple undifferentiated meristems[36].Consistently,in the present study,transgenic plants with increased expression ofTAW1(more than 300 times higher than that of ZH11) showed reduced plant height and panicle length (Fig.4A–E),and both primary and secondary branch numbers were significantly decreased (Fig.4F and G),resulting in a severe reduction of GNP(Fig.4H).Moreover,parts of the apical spikelets were degenerated(Fig.4C).Expression pattern analysis showed thatTAW1was highly expressed mainly in root,leaf sheath and stem,and relatively weakly expressed in inflorescences (Fig.5),suggesting that high expression was not needed during panicle development.These results indicate that appropriate expression ofTAW1is essential for panicle development.Promoter editing ofTAW1using the CRISPR/Cas9 system may be an alternative strategy for high grain number,such as has been successfully applied inWaxy[45–47].

4.4.Natural variation of TAW1 has potential value in breeding

AlthoughTAW1is a major regulator of inflorescence architecture in rice [36],natural alleles ofTAW1that could be used in breeding have not been identified.In this study,we characterized the natural variation ofTAW1in 4262 germplasm accessions and performed haplotype analysis(Table S6).As a result,17 haplotypes were obtained and divided into two clades,which reflected mainly a divergence ofTAW1betweenindicaandjaponicaaccessions(Fig.6A and C).R498 and F771 carried Hap 7 and Hap 13(Fig.6C),respectively.This result is interesting,considering that the F771-type promotor ofTAW1had higher transcriptional activity than the R498-type(Fig.2H),and the expression level ofTAW1was higher in F771 and NIL-TAW1than in R498(Figs.2E,3D),indicating that Hap 13 was a favorable haplotype contributing to high grain number and grain yield.Frequency analysis showed that Hap 13 was present mainly inausaccessions and a fewindicaaccessions and almost absent injaponicaaccessions(Fig.6D),suggesting that the favorable Hap 13 was introduced fromausaccessions and little used inindicaandjaponicacultivars.These results provide a new strategy for high-yield breeding in rice by introducing Hap 13 ofTAW1fromausaccessions.

Sequence alignment and phylogenetic analysis showed that the orthologs ofTAW1in multiple plant species were highly conserved,although it evolved independently in dicotyledons and monocotyledons(Fig.S8),indicating a possible conserved function in different species.Consistently,TtTAW1-1A,aTAW1-homologous gene inTriticum turgidum,has been revealed to function directly in regulating the transition of indeterminate branch meristem fate into determinate spikelet meristem fate [48].A rachis-length QTL,qSbRL1.2067,was colocalized withSobic.001G219400,the ortholog ofTAW1in sorghum[49].These results indicated thatTAW1had a conserved function in regulating inflorescence development in grasses,suggesting the potential value of characterizing natural variation ofTAW1in other plants.

5.Conclusions

A major rice QTL,qPLN10,for panicle length and GNP was identified and delimited to a 121-kb interval on chromosome 10.Based on qRT-PCR and sequence analysis,the known regulator of rice inflorescence architectureTAW1was identified as the candidate gene forqPLN10.NIL-TAW1showed increased panicle length,GNP,and grain yield.These results provide a natural allele ofTAW1,which could be used to develop high-yield breeding in rice through conventional crossing followed by MAS.

CRediT authorship contribution statement

Hua Yuan,Bingtian Ma,and Weilan Chen:conceived and designed the experiments.Zhengyan Xu,Xueqin Tan,Peng Gao,Mengya Jin,Yunhai Kang,and Peixiong Liu:performed the experiments.Hua Yuan,Wencheng Song,Shiguang Wang,Bin Tu,Yuping Wang,Peng Qin,and Shigui Li:analyzed the data.Hua Yuan:wrote the manuscript.All authors have read and approved the final manuscript.

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 National Transgenic Science and Technology Program(2016ZX08001004-002)and the National Key Research and Development Program of China(2016YFD0100406).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.11.004.