Pingxi Wng,Yun Yng,Dongdong Li,Jiling Xu,Riling Gu,Jun Zheng,Junjie Fu,d,Jinhu Wng,*,Hongwei Zhng,d,*

a Institute of Crop Sciences,Chinese Academy of Agricultural Sciences,Beijing 100081,China

b Center for Seed Science and Technology,College of Agronomy and Biotechnology,China Agricultural University,Beijing 100193,China

c College of Life Science and Technology,Henan Institute of Science and Technology,Xinxiang 453003,Henan,China

d National Nanfan Research Institute(Sanya),Chinese Academy of Agricultural Sciences,Sanya 572024,Hainan,China

Keywords:Maize Plant height ZmAMP1 Green revolution

ABSTRACT Gene resources associated with plant stature and flowering time are invaluable for maize breeding.In this study,using an F2:3 population derived from a natural semi-dwarf mutant grmm and a normal inbred line Si 273,we identified a major pleiotropic QTL on the distal long arm of chromosome 1(qPH1_dla),and found that qPH1_dla controlled plant height,flowering time,ear and yield traits.qPH1_dla was finemapped to a 16 kb interval containing ZmAMP1,which was annotated as a glutamate carboxypeptidase.Allelism tests using two independent allelic mutants confirmed that ZmAMP1 was the causal gene.Realtime quantitative PCR and genomic sequence analysis suggested that a nonsynonymous mutation at the 598th base of ZmAMP1 gene was the causal sequence variant for the dwarfism of grmm.This novel ZmAMP1 allele was named ZmAMP1_grmm.RNA sequencing using two pairs of near isogenic lines(NILs)showed that 84 up-regulated and 68 down-regulated genes in dwarf NILs were enriched in 15 metabolic pathways.Finally,introgression of ZmAMP1_grmm into Zhengdan 958 and Xianyu 335 generated two improved F1 lines.In field tests,they were semi-dwarf,early-flowering,lodging-resistant,and high-yielding under high-density planting conditions,suggesting that ZmAMP1_grmm is a promising Green Revolution gene for maize hybrid breeding.

1.Introduction

Plant height(PH)shows strong correlations with grain yield and biomass[1,2],and is a key target trait in crop breeding.Modifying PH has been suggested as one major approach to increasing planting density and yield per unit area[3,4].However,in comparison with other crops such as rice and wheat,maize PH has shown little decrease over its breeding history[5],suggesting that there are great opportunities for harnessing PH variation to improve maize varieties.

Cloning genes controlling the genetic variation of agronomic traits is the basis for designing future crop varieties[6,7].Several genes associated with PH have been cloned in crop species,such as Ghd7 and Hd1 in rice[8,9]and Rht1 in wheat[1,10].Several genes controlling PH have been functionally characterized in maize,including Vegetative to generative transition 1(Vgt1)[11],ZmGA3ox2[12],Brachytic2[13],and ZmTE1[14].Vgt1 regulates PH and flowering time(FT)by regulating ZmRap2.7 expression[11],whereas ZmGA3ox2,Brachytic2,and ZmTE1 control PH by regulating gibberellin production or auxin transport[2,14,15].However,applications of these genes in maize breeding are scarce according to our literature review,suggesting a need to clone other genes controlling maize PH.

In Arabidopsis,the Altered Meristem Program 1(amp1)mutant showed a variety of phenotypes including enlarged meristem,increased cotyledon size,early flowering,and short stature[16–18].The Arabidopsis AMP1 gene encoded a putative glutamate carboxypeptidase,and showed expression in all surveyed tissues[19].In maize,ZmAMP1 was associated not only with PH and internode length[20],but with other traits such as leaf size,preharvest sprouting,and anthocyanin accumulation.Two extremely dwarf mutants,d2003 and m34,carried mutant alleles of ZmAMP1[21,22],but the severe phenotypic defects limited their applications in maize breeding.

With their advantages of high harvest index,resistance to lodging,and tolerance to high planting density,semi-dwarf varieties have greatly increased the yield of major crops since the 1960s[1,2,23].The well-known Green Revolution gene SD1,encoding a GA biosynthetic GA20 oxidase,was the causal gene for the dwarfism of the semi-dwarf rice varieties[2].The rice sd1 mutation increased the accumulation of DELLA proteins[2,24],negative regulators of GA signaling and plant growth[25].The wheat Green Revolution Rht-B1b and Rht-D1b alleles encoded gain-of-function mutant DELLA proteins,and conferred GA-insensitive dwarfism[1].Successful applications of these well-known genes in crop breeding suggested that manipulating functional genes could be an effective way to improve crop varieties.

In this study,a natural semi-dwarf and early-flowering maize mutant grmm(an abbreviation of‘‘Green Revolution mutant of maize”)was crossed to a normal inbred line Si 273 to construct an F2:3segregation population.To clone the causal gene underlying the dwarfism of grmm,we performed preliminary mapping of 12 traits using the F2:3population,and find a pleiotropic locus controlling PH and several other traits was detected on the distal long arm of chromosome 1(qPH1_dla).Next,we used fine-mapping and allelism tests to confirm the causal gene ZmAMP1 Then,we performed RNA sequencing(RNA-seq)analysis to explore the molecular mechanism that was associated with the dwarfism of grmm.To test whether the mutant allele of ZmAMP1 had breeding value,we introgressed the mutant allele into two popular hybrid varieties,Zhengdan 958 and Xianyu 335,and test the field performance of the improved F1lines.

2.Materials and methods

2.1.Plant materials and field experiments

Two maize inbred lines,Si 273 and grmm,were studied(Fig.1A).Si 273 is an ordinary maize inbred line used in other studies[26],and grmm is a natural mutant inbred line showing semidwarf plant stature and early-flowering phenotype.In the summer of 2014,grmm was crossed with Si 273 to produce F1seeds in Shunyi(Beijing,China,40.2′N,116.6′E).In the winter of 2014,F2and BC1F1seeds were obtained using grmm as the recurrent parent in Yuanjiang(Yunnan province,China,23.6′N,102.0′E).A total of 197 F2:3families were developed in Shunyi in the summer of 2015.At the same time,BC2F1seeds were obtained by backcrossing BC1F1lines to grmm.In the winter of 2015,BC2F2and BC3F1seeds were produced in Sanya(Hainan province,18.4′N,109.2′E).In the summer of 2016,BC2F3and BC3F2seeds were produced in Shunyi.A flow chart of population development is presented in Fig.S1.In every season,the planting density was 66,660 plants per hectare.

2.2.Phenotypic evaluation

The 197 F2:3families were evaluated in two environments:Sanya in the winter of 2015 and Shunyi in the summer of 2016.In each environment,the tested traits included PH traits[PH,ear height(EH),and PH above ear(PHAE)],FT traits[heading date(HD),pollen-shedding date(PSD),and silking date(SD)],ear traits[ear length(EL)and ear diameter(ED)],and yield traits[ear row number(ERN),kernel number per row(KNPR),hundred-kernel weight(HKW),and yield per plant(YPP)].The field experiment was arranged with two replications in a completely randomized design.Broad-sense heritability(H2)was calculated following the formula[27]:

where δ2g,δ2geand δ2are the variances of genotype,genotype by environment interaction,and residual error,respectively,and n and r are the numbers of environments and replications,respectively.For each tested trait of the F2:3population,best linear unbiased prediction(BLUP)values were estimated following the linear model.

where yijmis the phenotype of the ith（i=1，2···，197）genotype in the jthenvironment（j=1，2）,and the mth（m=1，2）replication.μ is the mean of the population,g is the genotype effect,l is the environment effect,gl is the effect of genotype-by-environment interaction,δ is the replication effect nested in environment effect,and ε is the residual error.All effects in the model were treated as random effects.The model was solved using the R package lme4[28].The BLUP data are presented in Table S1.

2.3.DNA extraction and genotyping

Genomic DNA was extracted from seedling leaves of two parents and the 197 F2plants.DNA samples were genotyped at Beijing Compass Biotechnology Co.,Ltd.(Beijing,China)using DNA chips containing 5179 single-nucleotide polymorphisms(SNPs)[29].SNPs showing no polymorphism between the two parents or showing segregation distortion in the F2population were excluded.The genotypic data are presented in Table S2.A linkage map was constructed using QTL IciMapping software 4.0[30]with default settings.The inclusive composite interval mapping method was used to identify genetic loci controlling each trait.The LOD threshold was determined from 1000 permutations.Additive effects(AE)and phenotypic variance explained(PVE)by QTL were estimated by the software.

2.4.Fine-mapping of qPH1_dla

Based on the primary mapping results,BC2F1lines that were heterozygous at the target locus were selected.The BC2F2families developed from the selected BC2F1lines were used for finemapping.In the summer of 2016,a population containing～4200 BC2F2plants was sown in Shunyi,and 970 extremely short plants were used for fine mapping.To further localize the target locus,a large population containing 8940 BC2F3plants was sown in Sanya in the winter of 2016,and 2120 extremely short plants were used for fine mapping.The insertion/deletion markers used for fine mapping are described in Table S3.

2.5.Cytological analysis of near-isogenic lines

To investigate whether the dwarfism of grmm was caused by cell number or cell size,we selected a pair of near-isogenic lines(NILs)from a BC3F3family(of which grmm was the recurrent parent)in Sanya in the winter of 2016(Fig.S1).NILs carrying the grmm and Si 273 chromosome segements were designated as NILgrmmand NILSi273,respectively.At the tasseling stage of NILgrmmand NILSi273,transverse and longitudinal sections of the sixth and seventh internodes were stripped and fixed in 2.5%glutaraldehyde.They were then were washed three times in phosphate buffer,successively dehydrated in 30%,50%,70%,and 90%ethanol,and finally dried in liquid CO2.The samples were then fixed on metal brackets,coated with gold particles with a SCD040 Sputter coater machine(Balzers Union,Liechtenstein,Switzerland),and observed by using an Olympus SZX7 scanning electron microscope(Hitachi,Tokyo,Japan).The cell area was calculated by dividing the scanned area by the number of cells in the area.The length of cells was calculated by dividing the length of the scanned area by the number of cells along the length of the area.The area and length of cells were the means of three replications.Differences between NILs were tested by t-test.

Fig.1.Comparison of the parent lines and the NILs.(A)The appearance of Si 273 and grmm at the mature stage.(B)The stem(above)and the ear(below)of NILSi 273 and NILgrmm.(C,D)Scanning micrographs of transverse sections of sixth internodes of NILSi 273(C)and NILgrmm(D).(E,F)Scanning micrographs of longitudinal sections of sixth internodes of NILSi 273(E)and NILgrmm(F).(G,H)Comparison of cell areas in transverse sections(G)and cell lengths in longitudinal sections(H)of parenchyma cells in sixth and seventh internodes of NILSi 273 and NILgrmm.N6,sixth internode;N7,seventh internode.Gray and black columns indicate NILSi 273 and NILgrmm,respectively.‘‘ns”indicates no significant difference by t-test.

2.6.Allelism test

To confirm that ZmAMP1 was the causal gene underlying qPH1_dla,two Mu-insertion mutants of ZmAMP1(UFMu-10012 and UFMu-05780)were obtained from the Maize Stock Center(https://www.maizegdb.org/data_center/stock).The genotypes of the Mu-insertion mutants were identified using primers designed based on the Mu and ZmAMP1 genome sequences(Table S3).Heterozygous Mu-insertion mutants were crossed with grmm to produce F1seeds in the winter of 2017 in Sanya.PH of F1plants was evaluated in the summer of 2018 in Shunyi,and the segregation of pH in the F1populations tested by χ2test to confirm whether the segregation of tall and short plants followed a 1:1 ratio.

2.7.RNA extraction,qPCR,and RNA sequencing

To compare the expression of ZmAMP1 in the two parents and NILs,the internodes of grmm,Si 273,NILgrmmand NILSi273(V6 stage)were sampled for RNA extraction.For each line,three biological replications were sampled,with each replication containing the internodes of five plants.Total RNA was extracted with an RNA prep Pure Plant Kit(Tiangen,Beijing,China).The RNA of the four NILs(d2003 and K36;NILgrmmand NILSi273)was reversetranscribed into cDNA using TransScript II One-Step gDNA Removal and cDNA Synthesis SuperMix Kit(TransGen,Beijing,China).Quantitative real-time PCR(qPCR)was performed in a 7300 Sequence Detection System(Applied Biosystems,Waltham,MA,USA)using TransStart Green qPCR SuperMix(TransGen).The significance of expression differences between NILs was tested by t-test in Microsoft Excel 2010(https://www.microsoft.com/).The primers used for qPCR were designed from the sequence of the first exon of ZmAMP1,and are described in Table S3.ZmGAPDH was used as the reference control.

A natural dwarf mutant,d2003,was identified from maize inbred line K36[20],and a single-base insertion in the first exon of ZmAMP1 resulted in a premature termination of translation in d2003.d2003 and K36 were used as another group of NILs.The internodes of d2003 and K36 at V6 stage were sampled for RNA extraction.Three biological replications were sampled,with each replication containing the internodes of five plants.The RNA samples of the four NILs were quantified with a NanoDrop2000 spectrophotometer(Thermo Scientific,Waltham,MA,USA).RNA samples were sent to Beijing Berrygenomics Technology Co.,Ltd.(Beijing,China)for RNA-seq.The pipeline for analyzing DEGs distinguishing each pair of NILs was described in our previous report[31].Genes with adjusted P-value＜0.05 were identified as DEGs.KEGG pathway enrichment analysis of the DEGs was performed with KOBAS(https://kobas.cbi.pku.edu.cn)[32].

2.8.Comparing ZmAMP1 genome sequences of grmm and other inbred lines

To identify causal genome variants for the dwarfism of grmm,the genome sequences of ZmAMP1 were amplified from grmm and Si 273,and aligned with DNAMAN software(https://dnaman.software.informer.com/).The ZmAMP1 genome sequences of B73,CML247,EP1 and F7 were retrieved from MaizeGDB(https://www.maizegdb.org/)and compared with the genomic sequence of ZmAMP1 in grmm to identify specific sequence variants in grmm.To determine whether other inbred lines contained the sequence variant found in grmm,we amplified the genome sequences containing the sequence variant from 42 randomly chosen maize inbred lines used in a previous study[26].The primers used for amplifying the sequence variant are described in Table S3.

2.9.Evaluation of ZmAMP1_grmm in maize breeding

Zheng 58 and Chang 7-2 are respectively the female and male parents of Zhengdan 958,and PH6WC and PH4CV are respectively the female and male parents of Xianyu 335.Zhengdan 958 and Xianyu 335 are the most popular maize hybrid varieties in China.In the winter of 2016 in Sanya,grmm(donor parent)was crossed to four recurrent parents(Zheng 58,Chang 7-2,PH6WC,and PH4CV).BC1F1and BC2F1plants were produced in the summer and winter of 2017,respectively.BC3F1and BC4F1seeds were produced in the summer and winter of 2018,respectively.In the summer of 2019,we harvested BC4F2seeds(carrying homozygous ZmAMP1_grmm alleles)from the BC4F1plants developed from Zheng 58 and Chang 7-2 as recurrent parents,naming these plants Zheng 58_grmm and Chang 7-2_grmm respectively.At the same time,we harvested BC5F1seeds from the BC4F1plants derived from PH6WC and PH4CV as recurrent parents.In the winter of 2019,the BC5F1plants were self-pollinated to produce BC5F2plants(carrying homozygous ZmAMP1_grmm alleles),which were named respectively PH6WC_grmm and PH4CV_grmm depending on the recurrent parents.

In the winter of 2019 in Sanya,Zheng 58_grmm and Chang 7-2_grmm were used as respectively female and male parents to develop an improved F1line,which was named Zhengdan 958_grmm.In the winter of 2020 in Sanya,PH6WC_grmm and PH4CV_grmm were used as respectively female and male parents to develop another improved F1line,which was named Xianyu 335_grmm.Because Zhengdan 958_grmm and Xianyu 335_grmm carried the ZmAMP1_grmm allele,both improved F1lines were shorter and more early-flowering than their control counterparts,and were expected to be lodging-resistant.We evaluated the field performance of Zhengdan 958_grmm at a planting density of 133,320 plants per hectare,and those of Zhengdan 958 at a normal planting density of 66,660 plants per hectare in the summer of 2020(environment 1)and 2021(environment 2)in Langfang(Hebei province,China,39.5′N,116.6′E),and in the summer of 2021 in Shunyi(environment 3).Meanwhile,we evaluated the field performance of Xianyu 335_grmm at 133,320 plants per hectare and that of Xianyu 335 at 66,660 plants per hectare in the summer of 2021 in Langfang(environment 2).In each environment,we grew five replications,with each replication containing four adjacent rows.The tested traits included PH,FT,grain yield,and lodging rate(LR).Grain yield per plot was adjusted to grain yield per hectare.Lodging rate was calculated by dividing the number of lodged plants by the total number of plants at the harvesting stage.Differences between line means were tested by t-test.

3.Results

3.1.Preliminary mapping identified the pleiotropic large-effect QTL qPH1_dla

For each tested trait,there were significant differences between the parents(Table S4).The coefficients of variation of these traits ranged from 3.13%to 25.27%.Their broad-sense heritability ranged from 0.55 to 0.89,indicating that they were controlled mainly by genetic factors.PH was significantly correlated with FT,ear,and yield traits.The latter also showed significant correlation with FT and ear traits(Fig.S2).

Eleven QTL controlling PH traits,30 QTL controlling FT traits,11 QTL controlling ear traits,and 15 QTL controlling yield traits were identified(Table S5).The QTL located at～276 cM on the distal long arm of chromosome 1 was associated with all tested traits(Table 1).Because this locus had the largest LOD value,PVE,and additive effect for PH(Tables 1,S5),it was named qPH1_dla.

3.2.Morphological and cytological analysis of NILgrmm and NILSi 273

There were significant phenotypic differences between NILSi273and NILgrmm(Fig.1B;Table S6),and the internode length of NILgrmmwas significantly shorter than that of NILSi273(Fig.S3),indicating that this dwarfism of grmm was associated with its short internode.

In the transverse section of the sixth internode,the vascular bundle and parenchyma cells had similar sizes in NILgrmmand NILSi273(Fig.1C,D).In longitudinal sections of the sixth internodes,the sizes of internode cells were also similar in NILgrmmand NILSi273(Fig.1E,F).Statistically,the cell areas or cell lengths showed no differences between the sixth or seventh internodes of NILgrmmand NILSi273(Fig.1G,H).

3.3.Fine mapping of qPH1_dla

We used PH as the target trait for fine-mapping of qPH1_dla,which was located in the interval 273–277 cM(Fig.2A),and this position corresponded approximately to the physical position from 281.8 to 289.3 Mb according to maize B73 reference genome version 3(https://www.maizegdb.org/gbrowse/maize_v3).We performed a preliminary screen of recombinants using flanking markers D2 and D25 described in Table S3,the positions of which were 281.2 and 289.9 Mb,respectively.Using the 970 extremely short plants of the BC2F2population,the QTL region was narrowed to the interval between G2 and R18(Fig.2B;Table S3),with a length of 227 kb according to the B73 reference genome version 3(https://www.maizegdb.org/gbrowse/maize_v3).Using the 2120 extremely short plants of the BC2F3population,the QTL region was further narrowed to a 16-kb interval between R14 and G13(Fig.2B;Table S3).Two candidate genes,GRMZM2G010353 and GRMZM2G011385,were found in this interval(Fig.2B).

By searching the maize genetics and genomic database(https://www.maizegdb.org),we found that GRMZM2G011385 is and rarely expressed in shoot apical meristem(SAM)or internodes,and GRMZM2G010353 was a homolog of Arabidopsis AMP1,and accordingly designated ZmAMP1.ZmAMP1 has 10 exons(Fig.2C)and encodes a putative glutamate carboxypeptidase.Two sequence variations of ZmAMP1 have been reported,including the 33rd nucleotide insertion within the first exon leading to a premature stop codon[20]and a G-to-A mutation at the 1606th nucleotide resulting in a change of amino acid[22].We did not detect the two variants in grmm(Fig.S4)and designated this new allele as ZmAMP1_grmm.

3.4.An allelism test confirmed that ZmAMP1 was the causal gene

To confirm whether ZmAMP1 was the causal gene,we performed an allelism test using two independent Mu-insertion mutants of ZmAMP1(UFMu-10012 and UFMu-05780;Fig.2C).For each Mu-insertion mutant,because homozygous mutants carrying Mu-insertion alleles were embryo-aborted[21],only heterozygous mutants were available.As expected,in the offspring population produced by self-pollinating the heterozygous Muinsertion mutants,the number of plants carrying the wild type and the heterozygous genotypes followed a segregation ratio of 1:2(Table S7).The heterozygous UFMU-10012 and UFMU-05780plants were crossed separately to grmm to produce F1families.The numbers of short and tall plants followed a segregation ratio of 1:1 in F1families(Table 2).Generally,the allelism test supported a conclusion that ZmAMP1 was the causal gene underlying qPH1_dla.

Table 1 qPH1_dla was associated with all tested traits in the tested F2:3 population.

Table 2 Allelism test using UFMu-10012(+/-)and UFMu-05780(+/-).

Fig.2.Fine mapping of qPH1_dla.(A)qPH1_dla on chromosome 1.(B)Fine mapping of qPH1_dla using BC2F2 and BC2F3 populations.In total 970 extremely short plants in the BC2F2 generation and 2120 extremely short plants in the BC2F3 generation were used to delimit qPH1_dla to a 16-kb region.Vertical lines represent positions of DNA markers and numbers below the linkage map are numbers of recombinants.(C)The gene structure of ZmAMP1 and the approximate insertion sites of two Mu-insertion mutants UFMu-05780 and UFMu-10012.Black boxes and lines indicate exons and introns,respectively.

3.5.A nonsynonymous mutation was probably the causal variant

qPCR analysis showed that there were no significant differences between NILgrmmand NILSi273,and between parental lines(Fig.3A),suggesting that the causal variants of ZmAMP1_grmm was not in the promoter region,and did not influence ZmAMP1 expression.When the ZmAMP1 genomic sequence of grmm(3′to the ATG transcription start codon)was compared to that of Si 273,70 sequence variants were found between the two alleles,including 53 nucleotide substitutions,8 insertions,and 9 deletions(Fig.S4).To identify the causal variant among the 70 variants,we compared the ZmAMP1 genomic sequence of grmm to those of B73,CML247,EP1 and F7(https://www.maizegdb.org/),and found that only one variant was specific to grmm(Figs.3B,S4).The 598th base in the first exon of ZmAMP1 was a T in grmm and a C in Si 273,B73,CML247,EP1,and F7.

To further investigate whether the sequence variant at the 598th base was specific to grmm,we sequenced 42 randomly chosen inbred lines and found that the 598th base was a cytosine in all these lines(Fig.3C).This sequence variant caused a nonsynonymous mutation,leading to a change of arginine to cysteine at the 200th amino acid of the ZmAMP1_grmm protein in the proteaseassociated superfamily domain(Fig.3B).This change altered the polarity of the amino acid.We inferred that the nonsynonymous mutation at the 598th base was probably the causal variant for the dwarfism of grmm.

Fig.3.Analysis of ZmAMP1 expression and identification of the causal genomic variant.(A)Comparison of ZmAMP1 expression between parental lines and between NILs;bars indicate mean±standard deviation(n=3);P-values were obtained by t-test;‘‘ns”indicates nonsignificance.(B)Alignment of the genomic sequence of ZmAMP1 containing the 598th base from grmm,Si 273,and four public inbred lines;PA superfamily,protease-associated superfamily;TFR dimer,transferrin receptor-like dimerization domain.(C)The genomic variant at the 598th base was specific to grmm.Comparison was performed among grmm and 42 randomly chosen inbred lines.The red arrow indicates the 598th base.

3.6.RNA-seq identified DEGs between NILs

Two pairs of NILs(d2003 versus K36;NILgrmmversus NILSi273)were used for RNA-seq analysis.The quality of RNA-seq data was reliable(Table S8).RNA-seq analysis detected 2270 DEGs between d2003 and K36 and 2921 DEGs between NILgrmmand NILSi273(Fig.4A).Of these 84 and 68 DEGs were commonly up-and down-regulated in the dwarf NILs(d2003 and NILgrmm),respectively(Fig.4A;Table S9).KEGG analysis showed that these DEGs were enriched in 15 main KEGG pathways(Fig.4B).The most significantly enriched pathway was arginine and proline metabolism(4 genes),followed by metabolic pathways(21 genes),and alanine,aspartate and glutamate metabolism(3 genes).These pathways may be involved in the dwarfism of grmm.

3.7.Field evaluation of the improved F1 lines carrying homozygous ZmAMP1_grmm alleles

Zhengdan 958_grmm flowered approximately one week earlier than Zhengdan 958 in all tested environments(Fig.5A,E,I),and the PH of Zhengdan 958_grmm was significantly lower than that of Zhengdan 958(Fig.5B,F,J).The LR of Zhengdan 958_grmm was 0 in two of the three tested environments(Fig.5C,K).There were differences in LRs of Zhengdan 958 among the three environments(Fig.5C,G,K)that may have been due to variation in environment factors,especially wind power.Compared with the phenotypes of Zhengdan 958 in environments 2 and 3,the low LR of Zhengdan 958 in environment 1 reduced yield loss(Fig.5D).Zhengdan 958_grmm had higher grain yield than Zhengdan 958 when LR was high(Fig.5H,L).Comparing the phenotypes of Xianyu 335 and Xianyu 335_grmm revealed that the improved line was early-flowering(Fig.5M),short(Fig.5N),lodging resistant(Fig.5O),and high-yielding(Fig.5P).

4.Discussion

Fig.4.DEGs common to two pairs of NILs.(A)Common DEGs identified by comparing two pairs of NILs.The blue ellipse indicates the DEGs differentiating d2003 and K36,and the yellow ellipse indicates the DEGs differentiating NILgrmm & NILSi 273.Compared to the tall NILs(K36 and NILSi 273),the expressions of 84 and 68 DEGs(the overlapping region of the two ellipses)were commonly up-or down-regulated in the dwarf NILs(d2003 and NILgrmm).(B)KEGG pathway enrichment analysis of the common DEGs.The color scale indicates the P-value and the‘‘rich factor”indicates the proportion of DEGs in all unigenes enriched in a pathway.The size of the dot indicates the number of genes enriched in the pathway.

In Arabidopsis,AMP1 influenced a variety of traits,including seed dormancy[33],the formation of shoot apex meristems[34]and leaves[35].Similarly,mutations in ZmAMP1 also caused various abnormal phenotypes in maize,including preharvest sprouting,defective embryo,delayed onset of anthocyanin accumulation in mutant seeds,delayed vegetative phase change,accelerated rate of seedling growth,and PH[36,37].Our results also supported that ZmAMP1 had multiple phenotypic effects based on the following findings:(1)In contrast to the extreme phenotypes caused by the two mutant alleles[20,22],ZmAMP1_grmm allele conferred semi-dwarfism(Fig.1A,B);(2)qPH1_dla controlled 12 traits,including PH,FT,ear,and yield traits(Table 1);(3)NILgrmmand NILSi273displayed differences in PH,internode length,and flowering time(Fig.S3;Table S6),and there were differences in cell number between NILs owing to the results that there are no differences in cell area or length in NILs(Fig.1C–H).The pleotropic effect of ZmAMP1 indicates that it is possible to select a favorable mutant allele for crop breeding.

Studies of the mammalian homolog of AMP1 may provide clues on the importance of the enzymatic domain.In mammals,glutamate carboxypeptidase II hydrolyzes the neuropeptide Nacetylaspartylglutamate to N-acetylaspartate and glutamate[38],suggesting that the enzymatic activity is important for the biological function of AMP1 protein in plants.Considering that nonsynonymous mutation caused amino acid changes in the proteaseassociated superfamily domain,this mutation might influence ZmAMP1 function by altering its enzymatic function.Elucidating the biochemical function of ZmAMP1 and identifying the substrate of this glutamate carboxypeptidase may reveal its biological function.

The mechanism by which ZmAMP1 regulated PH in maize could be speculated from the 84 and 68 genes up-and down-regulated in the dwarf NILs.GRMZM2G050137 was up-regulated in both dwarf NILs,and its gene homolog OsglHAT1(a histone H4 acetyltransferase)in rice regulated rice grain weight,yield,and plant biomass[39].Another gene,GRMZM2G117961,was also up-regulated in both dwarf NILs,and overexpressing its homolog OsMADS26 in rice conferred semi-dwarfism[40,41].It is possible that ZmAMP1 regulates PH and yield-related traits by regulating GRMZM2G050137 and/or GRMZM2G117961.

Varieties carrying Green Revolution genes have short and strong stalks,improved lodging resistance,and increased tolerance to high planting density.In this study,we cloned a new allele of ZmAMP1,and found that the maize hybrid lines containing this allele were shorter,more resistant to lodging,and more tolerant to high planting density than the control lines,suggesting that ZmAMP1_grmm could be used as a Green Revolution gene in maize hybrid breeding.Short stature,dense-planting tolerance,and early maturation are among the main breeding objectives in maize hybrid breeding in China,given that short plant stature increases lodging resistance,dense planting increases the number of plants per unit area,and early maturation reduces kernel water content at the harvest stage.Considering that the F1lines carrying ZmAMP1_grmm displayed these characteristics,ZmAMP1_grmm should be a desirable gene for maize breeders.

CRediT authorship contribution statement

Pingxi Wang:Project administration,Investigation,Visualization,Writing–original draft,Funding acquisition.Yuan Yang:Investigation,Visualization,Writing–original draft.Dongdong Li:Investigation.Jialiang Xu:Investigation.Riliang Gu:Writing–review & editing.Jun Zheng:Resources.Junjie Fu:Resources,Funding acquisition.Jianhua Wang:Supervision,Resources,Funding acquisition.Hongwei Zhang:Conceptualization,Supervision,Resources,Writing–review & editing,Funding acquisition.

Fig.5.Field evaluation of improved F1 lines carrying ZmAMP1_grmm.(A–L)show comparisons of field performance of Zhengdan 958 and Zhengdan 958_grmm in three environments.(A–D)show flowering time(A),plant height(B),lodging rate(C),and grain yield(D)in environment 1(Langfang,2020).(E–H)show flowering time(E),plant height(F),lodging rate(G),and grain yield(H)in environment 2(Langfang,2021).(I–L)show flowering time(I),plant height(J),lodging rate(K),and grain yield(L)in environment 3(Shunyi,2020).(M-P)show comparisons of field traits of Xianyu 335 and Xianyu 335_grmm in environment 2.The traits were flowering time(M),plant height(N),lodging rate(O)and grain yield(P).FT,flowering time;PH,plant height;LR,lodging rate.**,P＜0.01;***,P＜0.001;ns,no significance.

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 research was supported by the Hainan Provincial Joint Project of Sanya Yazhou Bay Science and Technology City(320LH043),the Key Scientific and Technological Research Project in Henan Province(222102110091),the China Agriculture Research System(CARS-02-13),the Hainan Yazhou Bay Seed Laboratory(B21HJ0223),and the Chinese Academy of Agricultural Sciences(CAAS)Innovation Project(CAAS-ZDRW202004).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2022.06.001.