Genetic dissection of crown root traits and their relationships with aboveground agronomic traits in maize

SHA Xiao-qian, GUAN Hong-hui, ZHOU Yu-qian, SU Er-hu, GUO Jian, Ll Yong-xiang, ZHANG Deng-feng, LlU Xu-yang, HE Guan-hua, Ll Yu, WANG Tian-yu, ZOU Hua-wen, Ll Chun-hui#

1 College of Agriculture, Yangtze University, Jingzhou 434000, P.R.China

2 State Key Laboratory of Crop Gene Resources and Breeding, Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, Beijing 100081, P.R.China

3 Institute of Crops, Gansu Academy of Agricultural Sciences, Lanzhou 730070, P.R.China

4 Institute of Maize Research, Inner Mongolia Academy of Agricultural & Animal Husbandry Sciences, Hohhot 010031, P.R.China

Abstract The crown root system is the most important root component in maize at both the vegetative and reproductive stages.However, the genetic basis of maize crown root traits (CRT) is still unclear, and the relationship between CRT and aboveground agronomic traits in maize is poorly understood.In this study, an association panel including 531 elite maize inbred lines was planted to phenotype the CRT and aboveground agronomic traits in different field environments.We found that root traits were significantly and positively correlated with most aboveground agronomic traits, including flowering time, plant architecture and grain yield.Using a genome-wide association study (GWAS)coupled with resequencing, a total of 115 associated loci and 22 high-confidence candidate genes were identified for CRT.Approximately one-third of the genetic variation in crown root was co-located with 46 QTLs derived from flowering and plant architecture.Furthermore, 103 (89.6%) of 115 crown root loci were located within known domestication- and/or improvement-selective sweeps, suggesting that crown roots might experience indirect selection in maize during domestication and improvement.Furthermore, the expression of Zm00001d036901, a high-confidence candidate gene, may contribute to the phenotypic variation in maize crown roots, and Zm00001d036901 was selected during the domestication and improvement of maize.This study promotes our understanding of the genetic basis of root architecture and provides resources for genomics-enabled improvements in maize root architecture.

Keywords: maize, root, aboveground agronomic traits, GWAS, candidate genes

1.lntroduction

Maize (ZeamaysL.) is a major crop for food, feed,energy and forage, accounting for approximately 36%of the total worldwide cereal production (FAO 2020).Roots are crucial for plant development and growth by absorbing water and nutrients from soil, providing mechanical support, and interacting with rhizosphere microbial communities (Taiet al.2016), all of which are critical for ensuring high yield and yield stability in different environments.Previous studies have reported that root traits are significantly associated with stress tolerance.For example, roots in deeper soil can absorb more water and nitrates to enhance stress tolerance (Miet al.2010;Lynch 2013).A deeper root system can enhance drought tolerance in maize (Ribautet al.2009).A shallower angle of seed roots and crown roots leads to more phosphorus uptake (Zhuet al.2005).Therefore, the genetic improvement of root architecture has been considered a target for developing more productive and stress tolerant varieties (Kengkannaet al.2019).

Root system architecture (RSA) refers to the spatial distribution of all root components in a given growing environment (Paez-Garciaet al.2015).For RSA, the crown root system is the most important root trait for soil resource acquisition during the vegetative and reproductive stages in maize (Lynch 2013).The crown root angle (CRA) affects the vertical and horizontal distribution of roots in the soil, and it is related to root length.Steep CRAs are superior for water acquisition under drought stress in maize (Trachselet al.2013; Liet al.2018).Lynch (2013) proposed that an intermediate crown root number with a steep growth angle and few but long laterals might be ideal, which is beneficial for water and N acquisition in maize.Gao and Lynch (2016)documented that a low crown root number improves drought tolerance by increasing rooting depth and water acquisition from the subsoil in maize.Shenet al.(2019)found that the aboveground traits were good predictors of root traits; as the specific leaf area, dry matter,and nitrogen and phosphorus contents were strongly correlated with root tissue density and specific root length.Asefaet al.(2022) found that the specific leaf area was positively correlated with specific root length, while it was negatively correlated with root average diameter across various moisture levels.Hence, identifying quantitative trait loci (QTLs) or genes associated with crown rootrelated traits, and analyzing their genetic relationships with aboveground agronomic traits would contribute to the genetic improvement of stress tolerance in maize.

Crown root traits (CRTs) are complex quantitative traits that are controlled by multiple genes.Based on traditional linkage mapping, previous studies have conducted genetic dissection of crown root-related traits.For example, Omori and Mano (2007) identified 10 QTLs for CRA in two F2populations developed from different crossings between teosinte and maize inbred lines.Caiet al.(2012) conducted phenotyping for CRTs in a BC4F3population under field conditions, and identified five QTLs underlying crown root number at three maize developmental stages.Liet al.(2020) detected 25 QTLs related to CRA, diameter, and number in a recombinant inbred line population.With the rapid advancement of sequencing technology, genomewide association study (GWAS) has been effectively used to analyze the genetic mechanism of CRTs in maize.For example, Wanget al.(2021) used an association mapping panel including 316 diverse maize inbred lines for genotyping with 140 421 SNPs and phenotyping of three CRTs under three field trials at the maturity stage, and identified a total of 126 SNPs that significantly associated with three CRTs.Wuet al.(2022) detected 63 SNPs significantly associated with eight CRTs using an association mapping panel.Although some QTLs/QTNs (quantitative trait nucleotides) underlying CRTs have been identified in maize, only a few studies have dissected the natural variation in CRTs in the field, and a systematic phenotypic and genetic analysis is needed to reveal the relationships between CRTs and aboveground agronomic traits.

Previous studies have reported several genes underlying CRTs in maize.Based on the maize mutantrootless concerningcrownandseminalroots(rtcs), thertcsgene was cloned.It encodes a member of the plant-specific LATERAL ORGAN BOUNDARIES domain (LBD) protein that regulates crown root initiation in maize (Majeret al.2012).Fenget al.(2022) documented thatZmDRO1(homologous gene ofOsDRO1) could improve maize drought tolerance by altering the CRA under waterlimited conditions in the field.ZmCIPK15was significantly associated with CRA, and thecipk15mutant with steeper crown growth angles could improve nitrogen capturein silicoand in the field (Schneideret al.2022).Collectively, only a few genes underlying CRTs have been identified in maize.Therefore, it is urgent to mine the candidate genes regulating CRTs in order to improve the maize root architecture.

In this study, a total of 531 maize inbred lines were used to conduct phenotyping of the CRTs and agronomic traits in two different field environments at the maturity stage and to analyze the phenotypic relationships between CRTs and aboveground agronomic traits.Combined with highdensity SNPs obtained by resequencing, a GWAS was performed to detect the natural variation in three CRTs and eight agronomic traits.Furthermore, we identified high-confidence candidate genes of CRTs by integrating multiple methods, and examined common associations between CRTs and aboveground agronomic traits.

2.Materials and methods

2.1.Plant materials and field conditions

The association panel used in this study was comprised of 531 maize inbred lines, which were derived from a large-scale association mapping population previously reported by Liet al.(2022).The association panel consisted of temperate maize germplasm and dozens of tropical and subtropical lines from Southwest China and the International Maize and Wheat Improvement Center (CIMMYT).In addition, these inbred lines could be divided into seven groups (SS, NSS, Iodent, PA, PB,SPT and mixed) according to their genetic relationships.Phenotyping of crown root-related traits and aboveground agronomic traits was performed at two locations in 2018,Zhangye in Gansu Province and Bayannur in the Inner Mongolia Autonomous Region.For each location, all inbred lines were planted in an experimental field with a randomized block design including two replicates.Each line consisted of 13 individual plants grown in a plot that was 3 m in length with 0.6 m between rows.

2.2.Phenotypic data collection

To measure the CRTs, three representative individuals in each plot were selected at the maturity stage.A soil block of 25 cm in length, 25 cm in width and 30 cm in depth was excavated for individual plants.The soil was carefully washed away from the roots by a high-pressure water gun.Photographs of root architecture were collected from four different directions.Three root-related traits were measured, including CRA, crown root layer number(CRLN) and total crown root number (TCRN).CRA is the angle between the crown root and the vertical line of the shoot, and was assessed by REST Root Software based on the photographs of the root architecture (Colombiet al.2015).After the photographs were taken, crown roots were manually cut one by one with scissors.Then, CRLN and TCRN were calculated.

To investigate the relationships between aboveground traits and underground crown roots, a total of eight aboveground morphological traits were phenotyped,including days to silking (DTS), plant height (PH), ear height (EH), total leaf number (TLN) and leaf angle(LA).DTS was recorded for all plants in a plot.For plant architecture-related traits (PH, EH, TLN and LA),the measurements consisted of the average of three representative plants selected in each plot.To count the TLN of each inbred line, the 5th and 10th leaves of representative plants were marked using red paint.LA was measured as the angle of the first leaf above the primary ear from a plane defined by the stalk below the node subtending the leaf.After harvest, yield per plot(YPP), grain yield per plant (GYPP) and kernel number per ear (KNPE) were measured.The relevant phenotypic data are listed in Appendix A.

2.3.Linkage disequilibrium analysis

To estimate the linkage disequilibrium (LD) for the 531 lines, we calculated the squared correlation coefficient(r2) between pairwise SNPs using PopLDdecay Software(Zhanget al.2019).The program parameters were set as ‘-MaxDist 100 -MAF 0.01 -Miss 0.2’ to calculate the averager2between two SNPs in a 100-kb window.The distance of LD decay was represented as the physical distance over whichr2drops to 0.2.The overall LD decay distance among the 531 lines was 58 kb.

2.4.GWAS and candidate gene mining

Genotypes of the association panel were obtained from a previous study (Liet al.2022).After filtering with minor allele frequency (MAF)>0.05 and missing rate <80%, a total of 8 697 288 SNPs were identified for subsequent analysis.The kinship matrix of pairwise genetic similarities was calculated by EMMAX (Kanget al.2010), which was used as the variance–covariance matrix of the random effects.The average values of phenotypes from the two environments were used for the GWAS.The GWAS was performed using a linear mixed model with the EMMAX Software package.To determine the genome-wide significance threshold for the GWAS results, the number of genome-wide independent SNPs was estimated by PLINK(window size 100, step size 50, andr2≥0.2).Then, the threshold for significant trait-marker associations was set to 1.6×10–7(Bonferroni-corrected threshold of α=0.05) on the basis of 319 041 independent SNPs.Under the significant threshold of 1.6×10–7, only 65 significantly associated SNPs were obtained for the 11 traits.Given the rigor of the mixed linear model, we conservatively chose 1.0×10–5as the suggestive threshold to determine association signals for the subsequent GWAS results.The confidence intervals of the GWAS loci were determined by local LD block analysis where pairwiser2of the SNPs withP<1×10–5should be >0.5.Genes located directly in or within 58 kb(the genome-wide average distance of LD decay tor2=0.2)around the confidence interval were selected as the candidate genes for the GWAS loci.The phylogenes (http://www.phylogenes.org) database was used to search for homologous genes and construct the phylogenetic trees of candidate genes.The MaizeGDB (https://www.maizegdb.org), Rice Data (https://www.ricedata.cn/gene) and TAIR(https://www.arabidopsis.org) databases were used to annotate the functions of candidate genes.

2.5.Calculation of favorable alleles

Based on the associated SNPs, the allele types (reference allele or alternative allele) that confer better agronomic performance (i.e., smaller CRA, LA; lower EH; more CRLN,TCRN, TLN, YPP, GYPP and KNPE; higher PH and earlier DTS) were deemed favorable alleles.The number of favorable alleles was counted for each inbred line, and the correlations between the number of accumulated favorable alleles and trait values were analyzed.

2.6.Statistical analysis

IBM SPSS Statistics 21.0 Software was used for the statistical analysis of the 11 traits.Heritability was calculated using ANOVA in QTL IciMapping Software.Pearson correlation coefficients (r) were calculated for each trait using the packages “Hmisc” and “CorrPlot” in R Software.

2.7.Phylogenetic tree analysis

The phylogenetic tree was constructed using MEGA 5.05 Software.The method of tree construction was the proximity method, the tree shape was the original tree,and the number of tests was 1 000.

3.Results

3.1.Phenotypic variation and correlation analysis

For the association panel, we collected a total of 11 underground and aboveground traits, including three CRTs(CRA, CRLN and TCRN), one flowering time trait (DTS),four plant architecture traits (PH, EH, TLN, and LA) and three yield-related traits (GYPP, YPP and KNPE) (Table 1).The 11 traits showed wide phenotypic variation ranging from 5.6% for DTS to 45.2% for YPP.The phenotypic values of the 11 traits showed continuous and approximately normal distributions (Fig.1).All traits had a modest to high broadsense heritability (hB2) ranging from 43.0% for CRA to 94.0%for DTS.Furthermore, Pearson correlation analysis was performed to examine the phenotypic correlations between underground CRTs and aboveground agronomic traits.Significant and positive correlations were found among pairs of three CRTs (CRA, CRLN and TCRN).In addition,each of the three root traits was positively correlated with all aboveground agronomic traits, except for LA.

3.2.GWAS for crown root and agronomic traits

To dissect the genetic basis of the underground and aboveground traits, we performed GWAS for the 11 traits(Appendix B).Under the suggestive threshold of 1.0×10–5,a total of 1 933 associated SNPs were detected for the 11 traits (Appendix C).Of these, 388, 387, 642 and 576 associated SNPs were identified for CRTs, flowering time(FT), plant architecture-related traits (PART) and yieldrelated traits (YRT), respectively (Fig.2).We then defined an association locus as a chromosomal region in which the LD (r2) of pairwise associated SNPs was >0.5 and the distance between the adjacent pairs of associated SNPs was <500 kb.According to this definition, a total of 1 010 associated loci were identified for the 11 traits (Appendix D), including 115 loci for CRT, 316 loci for FT, 228 loci for PART, and 366 loci for YRT.

For the 1 010 associated loci, a total of 2 397 candidate genes were identified.Of these, 431, 519, 807 and 640 candidate genes were detected for CRT, FT, PART and YRT, respectively.Several known genes regulating agronomic traits and root development have been reported in previous studies.For example, Zm00001d010987(ZmRap2.7) and Zm00001d042315 (ZmMADS69), which are involved in regulating flowering time in maize (Salviet al.2007; Lianget al.2019), are associated with DTS(Appendix B-a).Zm00001d053371 (ZmVPS29) regulates the morphology of maize grains through an auxindependent process and its overexpression increases the yield per plant in different genetic backgrounds (Chenet al.2020), and it is associated with GYPP (Appendix B-b).Zm00001d043878 is associated with DTS and TCRN(Appendix B-a and c), encodes an AUX/IAA protein and has been namedrootlesswithundetectablemeristem1(rum1), which is involved in seminal and lateral root formation and development of the panicle and panicle axis inflorescences in maize (Wollet al.2005; Wang Y Jet al.2010; Zhanget al.2014; Baezet al.2020).We found that Zm00001d010255 (ZmGDIα-hel) associated with GYPP,CRLN and KNPE (Appendix B-b, d, and e), and it confers passive resistance to maize rough dwarf disease (Liuet al.2020).In addition, we also found that Zm00001d002562(ZmRAVL1) associated with DTS (Appendix B-a), and it was involved in maize leaf angle development and increased grain yield by editing itself (Tianet al.2019).

Table 1 Descriptive statistics of 11 traits and broad-sense heritability in the association panel

Fig.1 Phenotypic distribution and correlation analysis of the 11 traits.The plots of phenotypic distribution are on the diagonal.The graphs below the diagonal are the regression graphs, and each subgraph represents a regression of two corresponding traits on the diagonal.The numbers above the diagonal are the Pearson correlation coefficients, and each subgraph represents a correlation coefficient of two corresponding traits on the diagonal.CRA, crown root angle; CRLN, crown root layer number; TCRN,total crown root number; DTS, days to silking; PH, plant height; EH, ear height; LA, leaf angle; TLN, total leaf number; YPP, yield per plot; GYPP, grain yield per plant; KNPE, kernel number per ear.＊, ＊＊ and ＊＊＊ represent significances at the 0.05, 0.01 and 0.001 levels, respectively.

3.3.Aggregating effect of the associated SNPs

To understand the aggregating effect of the associated SNPs for each trait, we defined the allele types that confer better agronomic performance (i.e., smaller CRA, LA;lower EH; more CRLN, TCRN, TLN, YPP, GYPP and KNPE; higher PH and earlier DTS) as favorable alleles.We found that significant and modest to high correlation coefficients were observed between the phenotypes and the accumulated number of favorable alleles (Fig.3).Among the 11 traits, three traits (CRA, EH and LA) showed smaller phenotypic values of inbred lines containing more favorable alleles (Fig.3-A, F and G).The other eight traits, including CRLN, TCRN, TLN, YPP, GYPP, KNPE,PH and DTS, showed larger phenotypic values for inbred lines containing more favorable alleles (Fig.3-B–E and H–K).These results suggested that the QTNs detected in our study were accurate and the accumulation of favorable alleles might potentially increase yields in maize.

3.4.Common associations between CRTs and aboveground agronomic traits

Fig.2 Distribution of genome-wide associated SNPs.The tracks from the inner to outer layers are the distributions of associated SNPs for flowering time (FT), crown root traits(CRT), yield-related traits (YRT) and plant architecture-related traits (PART).Dots of different colors represent different traits.DTS, days to silking; CRA, crown root angle; CRLN, crown root layer number; TCRN, total crown root number; YPP, yield per plot; GYPP, grain yield per plant; KNPE, kernel number per ear; PH, plant height; EH, ear height; LA, leaf angle; TLN,total leaf number.

To determine the genetic relationships between crown root and aboveground traits, we identified whether the two flanking associated loci between CRTs and agronomic traits overlapped based on whether the positions of these two flanking QTLs were within the physical distances of 1 Mb.For the three CRT, we found that 55 (47.8%) out of 115 root-related trait QTLs overlapped aboveground agronomic traits (Appendix E).Among the 55 QTLs, the number of overlapping QTLs between CRT and flowering time was 24 out of 115, explaining 21% of the genetic variation in CRT.The number of overlapping QTLs between CRT and plant architecture-related traits was 22 out of 115, explaining 19% of the genetic variation in CRT.The number of overlapping QTLs between CRT and yield-related traits was 11 out of 115, explaining 9.5% of genetic variation in CRT.These results indicated that approximately onethird of the genetic variation in the CRTs was attributable to flowering time and plant architecture.

3.5.Root loci selected during maize domestication and improvement

Previous studies have reported that aboveground morphological features of maize were selected during domestication and improvement (Huffordet al.2012;Wanget al.2020).Root system architecture (RSA) buried in the ground is not easy for humans to directly select.However, by enhancing the uptake of nutrients and drought tolerance and increasing the yield of aboveground plants, RSA might have been one of the major targets indirectly selected during domestication and breeding.

To determine whether GWAS loci related to root architecture loci were selected, the physical loci of the crown root QTL were compared with those in reported selective sweeps during maize domestication and improvement (Chenet al.2022; Liet al.2022).We found that 103 (89.6%) out of 115 QTLs related to CRTs were located within known domestication- and/or improvementselective sweeps (Appendices F and G).Among the 103 loci, 30 (26.1%) and 102 (88.7%) out of 115 QTLs related to root traits were located within known domesticationselective sweeps and improvement-selective sweeps,respectively.These results indicated that the alleles of maize CRTs might have experienced selection in the process of maize domestication and breeding, which would provide valuable alleles for the genetic improvement of maize root architecture.

3.6.ldentification of crown root-related candidate genes

To further identify high confidence candidate genes for CRTs, we searched for candidate genes responsible for the variations in maize crown roots based on gene annotation information and the phylogenetic analysis of candidate genes with their homologs that have known functions.Finally, we were able to identify 22 high-confidence candidate genes, whose homologous genes in rice and/orArabidopsishave been functionally characterized as being associated with root development and/or stress resistance (Table 2).For instance,Zm00001d013159, which encodes a tubulin alpha-3 chain protein, was a CRLN candidate gene in this study.Its homologous gene in rice,TWISTEDDWARF1(Tid1), is involved in root development (Sunoharaet al.2009).Zm00001d010732 encodes a Rho-related protein,homologous toArabidopsisAtRAC7/ROP9which is specifically expressed in lateral roots and embryos(Nibauet al.2013), and was associated with TCRN.These candidate genes might be considered as the most promising candidates for maize crown root development.

To further evaluate the functions of these highconfidence candidate genes, we selected a candidate gene(Zm00001d036901) for deep analysis.Zm00001d036901 encodes a putative lectin-like receptor protein kinase,which is specifically expressed in roots.Phylogenetic analysis of Zm00001d036901 with homologous proteins from maize, wheat, rice andArabidopsisthalianashowed that Zm00001d036901 is homologous to salt intolerance 1(SIT1) in rice (Fig.4-A).OsSIT1is expressed in rice root epidermal cells and mediates salt sensitivity.It can promote the accumulation of reactive oxygen species(ROS), leading to growth inhibition and plant death under salt stress (Liet al.2014).In our study, a SNP(S_6_105966496), located in the promoter region of Zm00001d036901, was associated with CRA (Fig.4-B).Based on S_6_105966496, we classified the 531 maize genotypes into two allele groups (CC and TT).The accessions with the TT allele had a significantly smaller CRA than those with the CC allele (Fig.4-C).For example,the inbred line Tie7922 with the CC allele had a larger CRA than the inbred line H21 with the TT allele (Fig.4-D and E).Furthermore, the Zm00001d036901 expression data and root phenotypes of 35 lines from Renet al.(2022)were used to conduct a correlation analysis.There was a significant positive correlation between the CRA and the expression of Zm00001d036901 (Fig.4-F).Furthermore,we found that Zm00001d036901 was identified in a domestication-selective sweep selected between domesticated maize and its wild ancestor, teosinte, using the cross-population composite likelihood ratio (XP-CLR)approach (Fig.4-G), as well as an improvement-selective sweep selected during modern breeding of the female group (PA) and male group (SPT) in China (Fig.4-H).These results suggested that Zm00001d036901 may contribute to the phenotypic variation in maize CRA, so it may be potentially useful for molecular breeding.

4.Discussion

4.1.Crown root phenotyping under field conditions

Roots are important organs that absorb water andnutrients in plants.The maize root system consists of primary roots, seminal roots, crown roots and lateral roots arising from these axes (Hochholdingeret al.2004).For root phenotyping of maize, most previous studies focused on primary roots, seminal roots, and lateral roots at the seedling stage under laboratory and greenhouse conditions (Liet al.2015; Songet al.2016; Juet al.2018; Wanget al.2019).The plant cultivation methods used ranged from screening young seedlings grown on germination paper to direct excavation of greenhousegrown plants (Burtonet al.2014; Paceet al.2015).Although primary root and seminal root systems are crucial for seedling establishment, crown roots dominate resource acquisition during vegetative growth and reproductive development, and provide physical support for the shoot.In this study, using a shovelomics method,we measured three CRTs (CRA, CRLN and total crown root number) in large sets of genotypes after flowering under different field environments.Although shovelomics is a time- and cost-consuming process, and has a long field occupancy, the advantage of measuring CRTs in the field is ultimately relevant to maize yield performance.In addition, using root phenotypes obtained by shovelomics,stable QTLs identified across different environments could possibly be used for marker-assisted selection of root traits in maize breeding.

Fig.4 Identification of Zm00001d036901 underlying maize crown root angle.A, phylogenetic analysis of Zm00001d036901.B, Manhattan plots of the candidate gene.The black arrow indicates the physical positions of associated SNPs.C, Haplotype analysis.The values of crown root angle (CRA) for each allele group are displayed in the box plot, where n denotes the number of inbred lines belonging to each haplotype group.Statistical significance was determined using a two-sided t-test.D, image of representative inbred lines (Tie7922 and H21) with different haplotypes.E, statistics of the CRA differences between Tie7922 and H21.Values are represented as mean±SD.P-values of the two-sided t-test are shown.F, correlation analysis between CRA and the expression of Zm00001d036901.The data are from Ren et al.(2022).G, XP-CLR analysis of Zm00001d036901 during maize domestication.The grey dashed line indicates the physical position of Zm00001d036901.The data are from Chen et al.(2022).H, XP-CLR analysis of Zm00001d036901 during modern maize breeding.The grey dashed line indicates the physical position of Zm00001d036901.The data are from Li et al.(2022).

4.2.Comparison of known crown root loci with this study

In this study, a total of 115 loci associated with three CRTs were identified.Thirty-six (31.3%) of the 115 loci overlapped with known QTLs controlling crown roots(Appendix H).For example, nine loci associated with CRTs in this study were located withinqEAR5-1detected by using joint linkage analysis for effective brace root tier number (EBRTN) (Kuet al.2012).Liet al.(2020)identified hotspots (Chr6: 86.3–116.8 Mb) for CRTs, which were co-located with five QTLs for CRTs in this study.In addition, six of 22 high-confidence candidate genes underlying CRTs were co-located with QTLs reported by Liet al.(2020) and Guet al.(2017) (Appendix I).These results suggested that the QTLs detected in this study are potential genomic regions that could be selected to improve root architecture in maize breeding.

4.3.Common loci between crown roots and aboveground agronomic traits

We investigated the genetic relationships between CRTs and agronomic traits through QTL colocalization.When QTLs for two different traits colocalize, this might suggest the existence of a common regulator that controls the variation in both traits.In this study, we found that approximately one-third of the genetic variation in the crown roots was derived from flowering time and plant architecture.These QTLs are potentially key genetic loci that might regulate both root and aboveground traits.Breeders have always habitually selected for easy to observe agronomic traits, such as flowering time and plant architecture (Bomblies and Doebley 2006; Brownet al.2011; Liuet al.2012; Shanget al.2020; Zenget al.2022),which might result in the indirect selection of underground CRTs.Certainly, the possibility that colocalizing genomic regions contain genes that are closely linked but involved in different biological processes cannot be excluded.In the future, we could use transgenic or CRISPR-Cas9 technologies to validate the pleiotropic effects of candidate genes located within common loci between crown root and agronomic traits.

4.4.Candidate genes controlling CRTs in maize

Although previous studies have identified a host of QTL controlling CRTs in maize, few genes underlying crown roots have been identified and functionally validated (Guet al.2017; Zhanget al.2018; Moussaet al.2021).In this study, by integrating the candidate genes identified by GWAS and bioinformatics analysis, a total of 22 highconfidence candidate genes for maize crown roots,whose homologous genes in rice and/orArabidopsisare involved in root development and growth, were identified(Table 1).The homologous genes of Zm00001d010262 and Zm00001d048235 inArabidopsisencode different WD40 proteins, and these two homologous genes can interact with Damaged DNA Binding protein 1 (DDB1)to regulate the growth and development of plant roots(Bjerkan and Grini 2013; Dutilleulet al.2016); and the homologous genes of Zm00001d049167 and Zm00001d010732 inArabidopsispromote root growth by participating in signal transduction (Wanget al.2011; Nibauet al.2013).In addition, five of the 22 candidate genes were transcription factors involved in plant development and growth.For example,bZIP29was associated with CRLN and might regulate cell number in root meristems through the control of cell wall organization (Van Leeneet al.2016).

Lectin receptor-like kinases (LecRLKs) are a class of membrane proteins found in higher plants that are involved in diverse functions ranging from plant growth and development to stress tolerance (Vaidet al.2012,2013).In our study, a putative lectin-like receptor protein kinase (Zm00001d036901) was an important candidate in a list of 22 high-confidence candidate genes.The homologous gene of Zm00001d036901 in rice,OsSIT1, promotes the accumulation of ROS,leading to growth inhibition and plant death under salt stress (Liet al.2014).Haplotype analysis suggested that the alleles of associated SNPs in the promoter of Zm00001d036901 might contribute to crown root angle variation.Furthermore, expression analysis based on RNA-Seq data suggested that there was a significant and positive correlation between the crown root angle and the expression of Zm00001d036901(Fig.4-F).In addition, we found that Zm00001d036901 was an important candidate gene selected during the domestication and modern breeding of maize (Fig.4-G and H).These results suggested that Zm00001d036901 might be an important candidate for improving the maize crown root angle through molecular breeding.

5.Conclusion

This study identified many QTLs for crown root and aboveground agronomic traits by using an association panel with high-density SNP markers.CRTs were positively correlated with most aboveground agronomic traits.A total of 1 010 associated loci were identified for 11 traits, including 115 loci for CRTs, 316 loci for flowering time, 228 loci for plant architecture-related traits, and 366 loci for yield-related traits.Among these 1 010 loci,a total of 2 397 candidate genes including some known genes were identified for the 11 traits.The number of accumulated favorable alleles for these loci in inbred lines correlated well with the phenotypic values.Approximately the one-third of the genetic variation in the CRTs was attributable to flowering time and plant architecture.Of 115 loci related to CRTs, 103 (89.6%) were located within known domestication- and/or improvement-selective sweeps.A total of 22 high-confidence candidate genes underlying crown roots were further excavated.We selected the candidate gene Zm00001d036901 to conduct a systematic analysis, and found that the expression of Zm00001d036901 may contribute to the phenotypic variation in maize crown roots.Collectively, these loci and candidate genes might be applied in markerassisted selection or map-based cloning for the genetic improvements of crown roots in maize.

Acknowledgements

This work was supported by grants from the National Natural Science Foundation of China (31971891), the Guangxi Key Research and Development Projects, China(GuikeAB21238004), the Scientific Innovation 2030 Project, China (2022ZD0401703), and the Modern Agro-Industry Technology Research System of Maize, China(CARS-02-03).We thank Prof.Yang Xiaohong from China Agricultural University for providing XP-CLR data.We also thank Prof.Pan Qingchun from China Agricultural University for providing gene expression data.

Declaration of competing interests

The authors declare that they have no conflict of interest.

Appendicesassociated with this paper are available on https://doi.org/10.1016/j.jia.2023.04.022