New Loci Relating to Rice Tiller Angle Traits Discovery Through Genome-wide Association Analysis

Tiller angle (TA) is one of the most important agronomic traits which directly affected plant architecture, photosynthetic e�ciency, and planting density of rice. Although many quantitative trait loci (QTL) and associated genes for TA have been identi�ed, the detection of QTL for TA still required emphasis. In this study, 238 micro-core germplasm populations were used for genome-wide association analysis (GWAS) of the TA. The frequency distribution of TA was shown to be continuous in 2018 and 2019, �ve QTL designated as qTA1, qTA2, qTA6, qTA9, qTA12, and two QTL designated as qTA1, qTA1 were detected in 2018 and 2019, respectively. These QTL explained phenotypic variation ranged from 5.94–12.38%. Among them, qTA2 and qTA6 are new QTL for TA, and only qTA1 was detected repeatedly in two years. 253 germplasm from the 3K Rice Genome Project were used for candidate gene and haplotypes analysis, the candidate gene and haplotypes analyzed showed LOC_Os01g67770, LOC_Os02g35180 and LOC_Os09g38130 were the candidate gene of qTA1, qTA2, and qTA9, respectively. The results of this study provide insight into the genetic improvement of plant architecture in rice.


Introduction
Rice (Oryza sativa) is the staple food for more than half of the world's population and increasing rice production is crucial to ensuring global food security (Al-Tamimi et al. 2016).Rice TA is de ned as the angle between the vertical line and the side tillers with maximum inclination, re ects the spreading degree of rice plant architecture, is a key agronomic trait determining rice grain yield by affecting planting density (Wang et al. 2022;Cai et al. 2023).
Human long-term cultivation practice showed that the construction of rice ideotype can signi cantly improve rice yield (Wang et al. 2022).TA is affected by both genetic and environmental factors.TA QTL and gene mining is the basis for genetic improvement of plant architecture in rice (Wang et al. 2018; Huang et al. 2011;Zhao et al. 2011).
Previous studies showed that TA is a complex quantitative trait, and TA-related QTL has been explored in rice via different rice genetic populations.Dozens of QTL for TA were detected over recent decades, which are mainly located on chromosomes 1, 2, 3, 5, 7, 8, 9, 11, and 12 of rice, and some major QTL, such as qTAC8, qTAC9, and qTA3 were future ne mapping (Li et  shown that TA also is controlled by quality genes.TA-related genes have been explored in rice via different Mutant plants.Dozens of genes for TA were ne mapping over recent decades, la1, lazy1, Ostil1, tac2, lpa1, onac16-1D, fuct-1, CR-wox6 and CR-wox11, osarf17, la2, tac4 and lta1 (Takeshi et al. 2007 Zhang et al. 2018).Plant phytohormones also affected TA; there is evidence showing that gibberellin (GA) is also involved in regulating TA.OsGRF7 regulates rice plant architecture by regulating the synthesis and signaling pathways of auxin and gibberellin (Chen et al. 2020).All of these results indicated that the genetic regulation of TA is complex, the distribution of these QTL (genes) and excellent alleles of genes in germplasm is still not clear, and the application of cloned genes in production practice needs to be strengthened.Currently, GWAS is performed using core accessions with strong polymorphisms in target traits, has the advantages of extensive genetic variation, high reliability, high accuracy, and is an effective method to study TA, which provides information to analyze the natural variation and evolutionary change of TA (Wang et  In this study, GWAS was performed for TA using 238 micro-core germplasm resources, detected QTL and preliminarily analyzed candidate genes.Then, for new candidate genes, we selected 253 germplasm resources part of 3K Rice Genome Project to con rm the haplotype and the distribution of elite haplotypes of these candidate genes.A new candidate gene, qTA1, was discovered by GWAS in two consecutive years.The qPCR results indicated that cytokinin response regulator qTA1 signi cantly differential expression in rice stem base, TA of elite haplotype is larger.These results could help geneticists and breeders to better understand the distribution of elite haplotype.Meanwhile, this study also provides germplasm and genetic resources for the genetic improvement of TA in rice.

GWAS for TA in rice
Population genotype data was downloaded from Rice Diversity (http://www.ricediversity.org/),A total of 34,453 SNP markers were used for GWAS (missing data > 90%, minor allele frequency > 0.01), GWAS analysis was performed using the mixed linear model (MLM) of the association analysis software TASSEL 5.2 (Bradbury et al., 2007), with P < 1×10 − 3 as the threshold.There were two or more signi cant SNP loci in the physical location of 200 kb, which was identi ed as one QTL.The highest R 2 value of SNP loci in the QTL interval represents the contribution rate of the corresponding associated region (Zhao et al, 2011).Manhattan and Q-Q plots were drawn by R package CMplot.

Candidate gene screening
Referring to the Nipponbare genome annotation (https://rapdb.dna.affrc.go.jp/), 100 Kb upstream and downstream of the QTL locus were used as the candidate gene prediction interval, and the predicted genes, conserved hypothetical proteins, reducing transposons and transposons were deleted.At the same time, the relative expression levels of plant expression tissues and plant endogenous hormones in the functional candidate genes in the QTL interval predicted by the rice expression pro le database (https://ricexpro) were used to select genes with relatively higher expression levels as candidate genes.
the gene with the most signi cant hit within a local LD block constructed around the QTL was screened as the candidate gene (Shin et al., 2005).The R package 'LDheatmap' was used to draw the heatmap of pairwise LDs.All non-synonymous mutation SNPs in the 2Kb promoter region, exons, introns and downstream 1Kb promoter region of the gene were selected by using the comprehensive database of rice genomic variation and functional annotation (http://ricevarmap.ncpgr.cn/)published by Huazhong Agricultural University.SNP loci with primary allele frequency (Primary Allele Frequency) > 80% and deletion rate (DEL Frequency) > 8% were screened, high-density association and gene-based haplotype analyses were carried out to detect candidate genes of QTL for TA by using 253 accessions from 3K RGB germplasm.Haplotypes of each candidate gene carried by at least 15 accessions were used for multiple comparisons (Zhao et al. 2011).Duncan's multiple comparison tests (5% signi cance level) followed by one-way ANOVA were completed with SPSS 25.

RNA extraction and expression analysis
To verify whether the detected QTL-related genes affect the TA.Total RNAs were isolated from stem base of rice at the tillering stage bases using a TRIzol kit (Invitrogen) according to the user's manual.Total RNA was treated with DNase I and used to synthesize cDNA with the HiScript reverse transcriptase System.Quantitative reverse transcription qPCR was implemented with SsoFast EvaGreen Supermix Kit (Bio-Rad) on the CFX96 Real-time system (Bio-Rad) following the manufacturer's instructions.Rice OsActin gene (LOC_Os03g50885) was used as the internal reference gene and gene speci c primers are listed in (Supplement table A1).The 2 -△△ct method was used to calibrate relative expression levels of target genes with the reference gene (Livak et al. 2001).

phenotypic variation in TA among 238 micro-core germplasm populations
There were large variations in TA throughout the population in 2018 and 2019.The TA ranges from 9.63 °to 55.66 ° in 2018 and 14.59 ° to 47.76 ° in 2019, with an average of 27.86 ° and 25.82 °, the coe cient of variation (CV) was 24.23% and 25.43%, respectively.The absolute values of population skewness were 0.86 and 0.85, and the kurtosis was 1.42 and 0.67, respectively (Supplement Table A2), which indicated the frequency distribution of TA showed an approximately normal distribution (Fig. 2A, B).Correlation analysis showed that TA was signi cantly positive correlated in two years, and the correlation coe cient was 0.987(Fig.2C), indicated that genetics had a greater impact on TA than environmental factors.Statistical analysis of TA in different subgroups in the two years showed that there was no signi cant difference in other subgroups except for AUS and AROMATIC in 2018.

GWAS analysis for TA
GWAS analysis was performed of the 238 germplasm populations of TA in 2018 and 2019, based on the threshold (P value = 1.0×10 − 3 ), a total of 42 SNPs signi cantly associated with TA were identi ed, and these SNPs distribute on rice Chr 1, 2, 6, 9, and 12(Table 1).Based on these signi cant SNPs, we nally identi ed 6 QTL for TA.Among these QTL, qTA1, qTA2, qTA6, qTA9, qTA9-1, qTA12 were detected in 2018, qTA1 and qTA1-1 was detected in 2019 (Fig. 3).These QTL explained phenotypic variation ranged from 5.94-12.38%(Table 1).Among them, qTA2 and qTA6 are new QTL for TA Supplement Table 3 , and only qTA1 was repeated detected in 2018 and 2019, this indicated that qTA1 was stably expressed in different years.Referring to the previously study, TAC1, TAC3, and TIG1 were major genes controlling TA.The P value of the SNP anked by these genes were1.7×10− 3 , 3.2×10 − 2 and 3.9×10 − 1 , below the threshold (P = 1.0×10 − 3 ) in this study.We furtherly analyze the haplotypes, the results showed that there was no signi cant difference in TA between different haplotypes, the results showed that the SNPs anked by these genes did not in uence TA, this could be the cause of that we did not detect TAC1, TAC3, and TIG1 in this study (Supplement Fig. A3, Supplement Fig. A4, Supplement Fig. A5).To search for the candidate genes in six QTL, we choose the genes mainly expressed in stem or node based on the expression pro le in Rice Expression Database(http://expression.ic4r.org/),and then excluded the genes encoding retrotransposon or transposon protein based on their function annotations (https://www.rmbreeding.cn).12 annotated genes, including Zinc nger protein, auxin response factor and cytokinin response in qTA1, qTA2, qTA6, qTA9and qTA12, respectively.were preliminary excavated (Table 2).Then the high-density association and gene-based haplotype analyses were used for identifying the candidate genes by using 253 germplasm resources from 3K Rice Genome Project.We nally mining three candidate genes for qTA1 qTA2 and qTA9.Did not nd appropriate candidate genes for qTA2, qTA6 and qTA9-1 based on the results of haplotype analyses.In the region of qTA1, 7 SNPs were used for high-density association analysis.The annotated gene with the most signi cant hit was LOC_Os01g67770 (Fig. 4A).three major haplotypes were detected among 253 accessions based on three SNPs in the LOC_Os01g67770 promoter, ve SNPs in the coding region, and one SNP in the 3'-UTR region (Fig. 4B).Signi cant differences for TA were observed among the three haplotypes.Hap1, Hap2 and Hap3 contain 75, 70 and 90 accessions, respectively.Hap1, had the highest mean TA (25.60 °).And showed a signi cantly higher mean TA than the Hap2 (Fig. 4C and Fig. 4D).Hap1 was mainly composed of XI-1A and CA subgroups, and Hap2 was mainly composed of GJ-tmp and GJ-trp subgroups (Fig. 4E).In the region of qTA2, 8 SNPs were used for high-density association analysis.The annotated gene with the most signi cant hit was LOC_Os02g35180 (Fig. 5A).two major haplotypes were detected among 253 accessions based on three SNPs in the LOC_Os02g35180 promoter, three SNPs in the coding region, one SNP in encoding region, and one SNP in the 3'-UTR region (Fig. 5B).Signi cant differences for TA were observed among the three haplotypes.Hap1, Hap2 contain 112 and 92 accessions, respectively.Hap1, had the highest mean TA (23.62 °).And showed a signi cantly higher mean TA than the Hap2 (Fig. 5C and Fig. 5D).Hap1 were mainly composed of XI-1A, XI-1B, and XI-adm subgroups, and Hap2 were mainly composed of XI-1A, XI-1B, and XI-adm subgroups (Fig. 5E).In the region of qTA9, 8 SNPs were used for high-density association analysis.The annotated gene with the most signi cant hit was LOC_Os09g38130 (Fig. 6A).two major haplotypes were detected among 253 accessions based on four SNPs in the LOC_Os09g38130 promoter, one SNPs in the coding region, one SNP in encoding region, and one SNP in the 3'-UTR region (Fig. 6B).Signi cant differences for TA were observed among the three haplotypes.Hap1, Hap2 contain 112 and 92 accessions, respectively.Hap1, had the highest mean TA (27.03 °).And showed a signi cantly higher mean TA than the Hap2 and Hap3(Fig.6C and Fig. 6D).Hap1 and Hap2 is mainly composed of XI-1A, XI-1B, and XI-adm subgroups, Hap3 is mainly composed of GJadm, GJ-tmp and GJ-trp subpopulations (Fig. 6E).The above results showed that the elite haplotype of the TA mainly existed in the indica rice subgroup, and the TA haplotype of the indica rice subgroup was the elite haplotype.To further certify the LOC_Os01g67770, LOC_Os02g35180 and LOC_Os09g38130 are the candidate gene of qTA1, qTA2 and qTA9, qRT-PCR were used for analyzed the relative expression level of LOC_Os01g67770, LOC_Os02g35180 and LOC_Os09g38130 in ten accessions, the results showed that relative expression level of these genes are positively associated with TA, this indicated that LOC_Os01g67770, LOC_Os02g35180 and LOC_Os09g38130 may be the candidate genes of these QTL (Fig. 4, Fig. 5, and Fig. 6).OsActin gene (LOC_Os03g50885) was used as the internal reference gene, the 2 -△△ct method was used to calibrate relative expression levels of target genes with the reference gene.

Discussion
TA of rice affected by the genetic factor of accessions and environmental factors, genetic effects explained more of the variation, environment factors, such as the planting density per unit area, eld management, Light and temperature etc., also signi cantly affected TA (Wang et al. 2022).Plant type can be divided into plant morphology and plant type, the generalized plant type refers to the organic combination of the morphological characteristics of the plant body, including roots, stems, leaves, and ears, and the physiological functions directly related to the light energy utilization of the plant population; narrow plant type refers to the morphological characteristics and spatial arrangement of plants (Donald. 1968).The ideal plant type is to cultivate an excellent canopy structure, so that the crop can maximize the capture of solar energy, improve the photosynthetic e ciency of the population, and maximize the economic yield under the limited light conditions during the whole growth period (Wang et al. 2022).
Cultivating rice varieties with suitable TA, low shading, large photosynthetic area and high light energy utilization rate can effectively improve the yield of rice population.Natural variation is the form of variation that is signi cant in the evolution of developmental processes, genetic analysis of rice TA QTL derived from natural variation is of great signi cance for genetic improvement of rice plant type (

Abundant variations of TA in rice germplasm
In this study, our results showed that there was no signi cant difference in TA among different subpopulations in different years, and no signi cant difference in the variation amplitude and coe cient of variation of TA among the same subpopulations, this indicated that TA is greatly affected by environmental factors; we detected six QTL for TA in 2018, total explained about 47.79% of phenotypic variance, but just detected two QTL for TA in 2019, total explained 14% of phenotypic variance.these results showed that the expression of QTL for TA easily affected by environment factors.

Comparison of GWAS results with previously reported results
TA plays a vital role in rice production, it can directly affect plant architecture, photosynthetic e ciency, and planting density of rice, so appropriate TA is bene cial for improving rice production (Tan et al. 2008;Jin et al. 2008).Although signi cant progress in cloning of genes (QTL) involved in regulating TA has been made recently in rice, few of these genes was used in rice breeding for ideal architecture, and the distribution of alleles of TA in germplasm remain largely unknown.Then, we comparison of GWAS results with previously reported results of TA (Supplement Table A3), in our study, qTA1(id1024678), qTA1-1(id1024648), qTA2(id2008457), qTA6(id6000160), qTA9(id9007808), and qTA12(id12007801) for TA were detected in 2018 and 2019.and only qTA1 were repeatedly detected in two years, qTA1 explains 12.38% and 5.94% of phenotypic variance in 2018 and 2019, respectively.QTa1 which controls TA was rstly mapped between RG173-RG532 (Li et al. 1999), which locates on the same region with qTA1 and qTA1-1, suggesting that QTa1 may be an allele of qTA1 or qTA1-1.qTA-9, was mapped between RG662-CT100 (Qian et al. 2001), which locates on the same region with qTA9 and qTA9-1, suggesting that qTA-9 may be an allele of qTA9 or qTA9-1.qTA-12, was mapped between RG461-G148 (Qian et al. 2001), suggesting that qTA12 may be an allele of qTA-12.On the interval of qTA2 and qTA6, we did not nd any QTL for TA reported before, this indicated that qTA2 and qTA6 may be the new QTL for TA.

Identi cation of candidate genes for TA
It was reported that the plant endogenous hormones IAA, GA, BR, and SL can affected TA (Wang et al. 2022).D2, encoding a key enzyme in BR biosynthesis, positive regulates TA in rice (Dong et al. 2022).TAC1, LAZY1 and PROG1 regulate TA via involving in auxin polar transport (Yu et al. 2007;Jiang et al. 2007;Takeshi et al. 2007;Li et al. 2007;Yoshihara et al. 2007;Tan et al. 2008;Jin et al. 2008).In this study, consulting the relevant literature, the expression pro le in Rice Expression Database and the results of haplotype analyses, LOC_Os01g67770, was nally identi ed as the candidate gene of qTA1, it encodes cytokinin-signaling B-type response regulator, three SNPs in the promoter, ve SNPs in the coding region, and one SNP in the 3'-UTR region, caused signi cant difference in TA between three haplotypes.Hap1 is the superior haplotype conferring largest TA.LOC_Os02g35180, was identi ed as the candidate gene of qTA2, it encodes a cytokinin-responsive type-A response regulator.three SNPs in the promoter, three SNPs in the coding region, one SNP in encoding region, and one SNP in the 3'-UTR region, caused signi cant difference in TA between two haplotypes.Hap1 is the superior haplotype conferring largest TA.LOC_Os09g38130, identi ed as the candidate gene of qTA9, it encodes an auxin e ux carrier component, four SNPs in the promoter, one SNPs in the coding region, one SNP in encoding region, and one SNP in the 3'-UTR region caused signi cant difference in TA between three haplotypes.Hap1 is the superior haplotype conferring largest TA.
In the future, CRISPR/Cas9 could be applied, further verify the function of these genes and apply they in molecular breeding of ideotype.

Conclusions
We detected six QTL for TA by GWAS within 238 rice accessions, and screened 3 new candidate genes for qTA1, qTA2 and qTA9 vivo high-density association and gene-based haplotype analyses.Then we used the qRT-PCR to further validated the LOC_Os01g67770, LOC_Os02g35180 and LOC_Os09g38130 are the candidate gene of qTA1, qTA2 and qTA9, respectively.The identi cation of those candidate genes and their elite haplotypes will provide a promising source for molecular breeding of ideotype in rice.

Figure. 7
Figure.7 Expression veri cation the function of candidate genes.(A) Comparisons of TA values between accessions with larger TA and accessions with litter TA, *** p ≤ 0.001; (B), (C) and (D) relative expression level of LOC_Os01g67160, LOC_Os02g35180, and LOC_Os09g38130 in stem node of accessions.Rice

Figure 2 The
Figure 2

Table 1
List of QTL for TA identi ed by GWAS

Table . 2
Candidate genes and function annotations of QTL for TA Tan et al. 2008;Jin et al. 2008;Huang et al. 2011;Zhao et al. 2011).Compared with QTL mapping, GWAS enables the identi cation of causal loci at high-resolution and is a powerful tool for detecting favorable alleles in natural resources(Bai et al. 2011).