Natural variants and editing events provide insights into routes for spike architecture modification in common wheat

Min Fan,Xiaoyu Zhang,Ragupathi Nagarajan,Wenxuan Zhai,Yahya Rauf,Haiyan Jiaa,,Zhengqiang Ma,Liuling Yan,*

a Crop Genomics and Bioinformatics Center and National Key Laboratory of Crop Genetics and Germplasm Enhancement,Nanjing Agricultural University,Jiangsu,Nanjing 210095,Jiangsu,China

b Department of Plant and Soil Sciences,Oklahoma State University,Stillwater,OK 74078,USA

Keywords:Q gene Spike shape miR172 Gene editing Wheat

ABSTRACT Spike architecture is an indicative trait of grain yield in common wheat(Triticum aestivum).A segregating population was generated for mapping genes contributing to spike morphometric traits by crossing the two common wheat cultivars’CItr 17600’with branching spikes and’Yangmai 18’with normal spikes.A major quantitative trait locus for spike length was mapped to the Q5A region of chromosome 5A.Yangmai 18 carried a Q5Ab allele for short spikes,which harbored one SNP in the last intron,and a 1-bp InDel in the 720-bp fragment from the start codon,compared to Q5Aa in Chinese Spring.CItr 17600 harbored a q5Ab allele for long spikes,which has a 6-bp deletion compared to the reported q5Aa allele that was involved in the binding site of microRNA 172(miR172).This 6-bp deletion in immediately upstream of this binding site was involved in changes of four amino acids.The natural q5A allele appeared to be rare in common wheat but frequent in tetraploid T.turgidum accessions with branching spikes.The CRISPR/Cas9 technology was used to edit the upstream region involving in the miR172 binding site in Yangmai 18 and identified two independent editing events,one with a 1-bp insertion in Q5A and the other with a 2-bp deletion in Q5D,resulting in several shapes of spikes in the transgenic progeny.In addition to the effects of natural q5A allele and the edited Q5A genes,this study indicated the regeneratability and transformability of Yangmai 18 as an elite cultivar.Altogether,this study provides insight into future modification and engineering of spike architecture in common wheat.

1.Introduction

Hexaploid common wheat(Triticum aestivum L.,2n=6x=42,AABBDD genomes)is one of the most important cereal crops in the world[1–3].Spikes,the most important reproductive organs that influence grain yield,develop from the apex meristem of the main stem or a fertile tiller.Each spike consists of stacked spikelets that comprise grain-forming florets[4–8].Modifying wheat spike architecture is thus a key strategy to increase grain yield potential[9–11].

Spike architecture is highly diverse among members of the grass family(Poaceae).Tropical species,e.g.,maize(Zea mays)and rice(Oryza sativa),show highly branched tassels or panicles,whereas temperate grass species,e.g.,wheat and barley(Hordeum vulgare),typically possess a branchless spike[5,12–15].Wheat spikes are composed of a varying number of spikelets,and each spikelet sits on a node along the spike rachis.The distance between two neighboring spikelet nodes on the spike rachis determines the shape of a spike[16–18].Common wheat presents a normal spike that looks like a square head,which results in the underlying gene to be called‘Q’[9,19].The square spike is also called a subcompact spike(i.e.,partially compact),as it is longer relative to the compact spike that is typical for club wheat(T.aestivum ssp.compactum),but shorter compared to the lax or spelt spikes found in spelt(T.aestivum ssp.spelta)[20–26].In addition,some wheat cultivars initiate supernumerary spikelets that may develop into branching spikes or advance to florets,forming short or long ramified spikelets[27]and four-rowed wheat spikes[28].Wheat cultivars with supernumerary spikes or spikelets are rare and this trait were largely regulated by environments[5],but the diverse spike shapes in diploid wheat and tetraploid wheat species or subspecies(ssp.)provide a wealth of genetic resources to optimize spike architecture in common wheat[11,26,29].

The Q genes have played a key role in the domestication of bread wheat.The Q allele responsible for the square spike is partially or incompletely dominant over the q allele specifying the spelt-type spike in wheat[30–33].As Q genes and their alleles are represented by three homoeologs in common wheat,spike shape is determined by dosage effects between Q5A/q5A,Q5B/q5B,and Q5D/q5D[31].Four or fewer doses of the q allele were reported to result in plants with speltoid spikes,while five doses or more of Q genes are necessary to produce a square-headed spike[19].Q5A confers a stronger influence on spike architecture than Q5D,while Q5B is a pseudogene,making a comprehensive description of the relative contribution of each Q gene/allele challenging[9,24,30,31].

The Q5A gene was cloned by using a positional approach and shown to encode an APETALA2(AP2)-type transcription factor[9,30].Subsequent studies reported on the multiple Q and q alleles harbored by wheat cultivars,but only a few have been functionally characterized.One allele carried a single nucleotide polymorphism(SNP)in the coding sequence,resulting in a substitution of a valine(Val)residue at position 329 in the q protein by isoleucine(Ile)at the same position in the Q protein[30].This point mutation is associated with the domesticated spike phenotype[30].A second allele was characterized as a SNP in a presumed miR172 binding site in exon 10 that contributed to the domestication of spike architecture[11,12,32–34].A third allele consisted of a 161-bp transposon insertion in exon 5 of the Qtallele from Tibetan semiwild wheat,resulting in spelt spikes[26,35].While this allelic variation is associated with spike shape,the mechanisms underlying the pleiotropic effects of the Q/q alleles are not known.

Transgenic approaches have been employed to validate the functions of Q and its regulatory site by miR172.After Q5A from the durum wheat(Triticum durum)variety‘Langdon’(LDN)was transformed into the common wheat germplasm‘Bobwhite’[30],several transgenic wheat progenies showed compactoid or subcompactoid spikes due to different transgene copy numbers,thus affecting the expression level of the transgene.Additional mutations in the miR172 target site of Q genes were associated with higher Q expression and compact spike architecture[33].Decreased miR172 activity and increased expression of Q genes in MIM172 transgenic plants were associated with compact spike[32].In addition,a sham ramification phenotype was associated with 5QA loss-of-function alleles in the previous studies.

The primary research objective of this study was to map genes that can be used to modify spike architecture in wheat by using two wheat cultivars with different spike shapes.Accordingly,we mapped a major quantitative trait locus(QTL)associated with spike length to the Q5A region that is involved in spike architecture.We used CRISPR/Cas9 genome editing to validate the functions of the identified Q5A locus.In both mapping populations and transgenic editing progeny,we observed several spike shapes,providing new insights into the application of natural variants and editing events for the modification of spike architecture in wheat.

2.Materials and methods

2.1.Phenotypes of CItr 17600×Yangmai 18 mapping populations

CItr 17600,Yangmai 18,and their derived F3populations were tested in field at the Stillwater Research Station in the 2016–2017 growing season.The F3lines were arranged in a replicatesin-sets design for measuring spike traits in the field experiments.Each line was planted in one row,and the single-row plots were 1 m long,spaced 0.5 m apart.Spike length from the F3population was used for mapping.A F4line that showed segregation in genotypes in the targeted region and in spike length in the field was tested in a greenhouse at 20–25 °C and a long-day photoperiod(16 h light/8 h dark)without vernalization.A total of 336 plants,including 186 F2,82 F3and 68 F4,was screened for crossovers between two flanking markers,and nine recombinants were identified.Six of these lines with spring growth habit were tested for phenotypes.

2.2.Development of GBS markers

The CItr 17600×Yangmai 18 F2population was genotyped using genotyping-by-sequencing(GBS)approach,and a total of 5438 GBS markers were organized into genetic linkage groups for mapping[36].The GBS marker sequences were ordered along each chromosome based on the genomic sequence of Chinese Spring IWGSC RefSeq v1.0(https://urgi.versailles.inra.fr/blast/?dbgroup=wheat_iwgsc_refseq_v1_chromosomes).

2.3.Allelic variation in Q5Ab and q5Ab and development of a PCR marker for distinguishing the two alleles

The genomic TraesCS5A02G473800 sequence was PCR-amplified from Yangmai 18 and CItr 17600 as three separate overlapping fragments using primers QPF1 with QPR1 for fragment 1,QPF4 with QAR2 for fragment 2,and QAF1 with QAR2 for fragment 3.The sequences are provided in Figs.S1,S2.

2.4.PCR marker of Q5Ab and q5Ab alleles

The information for the primers QAF4 and QAR3 is provided in Table S1 with the expected sizes of PCR product.PCRs for the above markers were performed in Quickload Master Mix(BioLabs)under the following conditions:denature at 94°C for 5 min,amplification for 35 cycles at 94°C for 30 s,56°C for 30 s,and 72°C for 1 min per cycle,and final extension at 72 °C for 10 min.The PCR products digested with restriction enzyme Bsa I were run on a 2% agarose gel,showing polymorphic fragments between the q5Ab allele(CItr 17600,166 bp)and Q5Ab allele(Yangmai 18,249 bp).

2.5.Development of KASP markers flanking QSl.osu-5A

The allele-specific forward primers were designed to carry the standard FAM(5′-GAAGGTGACCAAGTTCATGCT-3′)and HEX(5′-G AAGGTCGGAGTCAACGGATT-3′)tails with the targeted SNP at the 3′end,and a specific reverse primer was designed(Table S1).PCRs were performed in a Real-Time PCR Cycler(ABI-7500),and PCR products were read in a fluorescence scanner following manufacturer instruction.PCR cycling was performed using the following protocol:denature at 94 °C for 5 min,followed by ten cycles of touchdown PCR(94 °C for 20 s;touchdown at 61 °C initially and decreasing by-0.6°C per cycle for 60 s),followed by 40 additional cycles(94 °C for 20 s;55 °C for 60 s).

2.6.Transcription levels of Q5Ab/q5Ab

Quantitative reverse transcription polymerase chain reaction(qRT-PCR)was used to determine the transcript levels of Q5A(Q5A-EX-F2 and Q5A-EX-R2)and Actin(Actin-F2 and Actin-R2)from the same cDNA samples.Total RNA was extracted from the leaves and immature spikes using TRIzol reagent(Invitrogen.Carlsbard,CA,USA).The cDNA was synthesized from 1 μg RNA treated with Deoxyribonuclease I by a SuperScriptTM II Reverse Transcriptase kit X and with oligo(dT)20primer(Invitrogen).

Quantitative RT-PCR(qRT-PCR)was carried out on a 7500 realtime PCR System(Applied Biosystems,Foster City,CA,USA)using iQTM SYBR Green Supermix(Bio-Rad laboratories,Hercules,CA,USA),with actin used as an endogenous control.Gene transcript levels were calculated by the 2–ΔΔCT method,where CT is the cycle threshold.There is a three-step cycling program consisting of an initial denaturation step at 95°C for 3 mins,followed by 39 cycles at 95°C for 15 s,57°C for 30 s,and 72°C for 30 s.The primers for the expression are listed in Table S1.

2.7.Construction of CRISPR/Cas9-Q-sgRNA for gene editing

The sgRNA Scorer 1.0.33 program was used to design a TraesCS5A02G473800 sgRNA with chromosome A,B and D matches,which was cloned into pBUN421(harboring the codon-optimized maize Cas9 gene)and pGTR14(containing gRNA-tRNA units)for genome editing[37].The gene-editing construct was also transformed into Yangmai 18.Primers Q5A-Cas-F and Q5A-Cas-R(Table S1)were designed to anneal oligos.The program for annealing oligos is 95 °C for 5 mins,decrease 1 per min and from 95 °C down to 25 °C for 70 mins,10 °C for holding.

Transform pBUN421 vector into DH10B and plate on LB+Kanamycin(50 mg L-1);After cloning and transformation,check by colony PCR with following condition:The Primers:TaU3-F1(5′-GGGAAACGACAATCTGATCCAAG-3′)/R2(5′-GACAGA CAGACATTTCCCAGAACG-3′)Homemade Taq,regular 20 μL rxn.95 °C(6 min),{94C °C(30 s),53 °C(30 s),72 °C(1 min)}×30 cycles,72°C(1 min).The size of positive clone is 928 bp;The size of negative clone is 2130 bp.Particle-bombardment-mediated CRISPR/SpCas9 system(TaU3 promoter drive single-guide RNA)with pBUN421-target fragment.

2.8.Transformation with Yangmai 18

We used Yangmai 18 as the receptor.When it’s 12 days after flowering,the embryos should be 0.5–1 mm long.These embryos were used for particle-bombardment-mediated CRISPR/SpCas9 transformation.Then primers(TaU3F1/TaU3R2 and ZCas9F/ZCas9R)were applied to check the positive plants,as well as primers(Q-sgR-CK-F1/QA-R1,Q-5B-F1/Q-5BD-R1 and Q-5D-F1/Q-5BD-R1)designed to amplify the editing region on chromosome 5A,5B and 5D(primers are shown in the Table S1).The genotypes of T2and T3population checked with primers(Q5A-GE-KASP-F1/F2/R and Q5D-GE-KASP-F1/F2/R)were shown in supporting Tables S2,S3.The transgenic plants were tested in the same greenhouse as used for the recombinant lines of normal wheat.

3.Results

3.1.Identification of a major quantitative trait locus for spike length

Yangmai 18 is a common wheat cultivar developed in Jiangsu,China.This cultivar was reported to carry numerous alleles for high grain-setting rates in both apical spikelets and basal spikelets forming a square spike[38](Fig.1A).CItr 17600 is another common wheat cultivar developed by the International Maize&Wheat Improvement Center.This cultivar was reported to have a super capacity for spikelet development,forming the type of branching spikes/spikelets previously observed in the tetraploid wheat cultivar‘Miracle’[11,39,40](Fig.1A).We set up a single cross between these two common wheat cultivars with spring growth habit,and the resulting F1progeny was self-pollinated to generate a population of 186 F2plants for genotyping in a photoperiod temperaturecontrolled greenhouse.We tested each F2line in a complete randomized block design in field at Stillwater Research Station.We then phenotyped the F3population for several traits including branching spikelets,the number of spikelets per spike,and spike length.We observed different spike shapes in CItr 17600×Yangmai 18 F2and F3progeny plants(Fig.S3).A previous study reported the trait of branching spikes was controlled by the wheat FRIZZY PANICLE gene(WFZP-A1)on chromosome arm 2AS,which encodes an APETALA2/ethylene response transcription factor[41].However,no QTL for branching spikes was not mapped to be associated with any markers at the WFZP1 locus in our population,and few plants showed complete branching spikes comparable to those seen in cultivar CItr 17600.The distribution pattern of spike length in the F3population suggested that this trait might be genetically regulated by multiple genes(Fig.S4A).We focused on one QTL for spike length in this study.

We genotyped the 186 Yangmai 18×CItr 17600 F2plants using genotyping-by-sequencing(GBS).We obtained 5438 GBS markers that we assembled into 36 linkage groups,which we assigned to the 21 chromosomes of hexaploid common wheat.Interval mapping(IM)analysis identified a major QTL for spike length mapping to the long arm of chromosome 5A(Table S2),which we refer to here as QSl.osu-5A(Fig.1B).This QTL had a logarithm of the odds(LOD)value at the peak position of 10.8 and explained 28.6 % of the total standing phenotypic variation in the F3population tested in the field.Average spike length showed significant differences between the three possible genotypes at this locus(Fig.1C).Spike length was 10.7±0.06 cm(n=49)in plants homozygous for the CItr 17600 allele,which was significant higher that 8.5±0.27 cm(n=35)in plants homozygous for the Yangmai 18 allele P=3.5E-09)and 9.8±0.18 cm(n=70)in heterozygous plants(P=7.2E-05).Spike length in plants homozygous for the Yangmai 18 allele was significantly higher than in plants homozygous for the CItr 17,600 allele in heterozygous plants(P=1.4E-06).

3.2.Genetic validation of QSl.osu-5A

We designed two experiments to validate the contribution of QSl.osu-5A to the phenotype.First,we selected the F3line WF121,as it displayed clear segregation for the spike length trait and was heterozygous at the QSl.osu-5A locus,to generate an F4population for phenotyping in the greenhouse.We genotyped each plant with a PCR marker representing the QSl.osu-5A locus(Fig.S4B).We determined that plants homozygous for the CItr 17600 allele had significantly longer spikes(12.6±0.12 cm,n=117)than those homozygous for the Yangmai 18 allele(9.8±0.08 cm,n=191)(P=1.01×10E-60).Spike length for plants homozygous for the CItr 17600 allele was also significantly different from that of plants heterozygous for the Q5A gene(12.1±0.2 1 cm,n=59)(P=0.01)(Fig.S4C).These results confirmed the existence of a major QTL for spike length at the QSl.osu-5A locus.

Second,we selected six recombinant F4lines at the QSl.osu-5A locus to generate F5populations.We developed two kompetitive allele-specific PCR(KASP)markers,Q5A-KASP1 at position 649,784,619 bp(Fig.1D)and Q5A-KASP10 at position 650,844,959 bp(Fig.1E),flanking QSl.osu-5A for fine-mapping.We screened 336 F2:4plants with the two markers,leading to the identification of nine crossover events.We focused on six lines with crossovers to generate their corresponding F5populations for spike length phenotyping in the same greenhouse.Again,the spikes of these plants were significantly longer in those homozygous for the CItr 17600 allele compared to those homozygous for the Yangmai 18 allele(CItr 17600 allele=9.6 cm;Yangmai 18 allele=8.02 cm,P=5.62×10–18)but not relative to heterozygous plants(9.00 cm,P=0.065)(Fig.S4D).The results from these six F5populations further supported the existence of a major QTL for spike length at the QSl.osu-5A locus.We narrowed the mapping interval for QSl.osu-5A to a 1,060,340-bp region on chromosome 5A between the two flanking markers(Fig.1D,E).

Fig.1.Mapping of a new q5A allele for spike length in a common wheat population.(A)Spike architecture of two parental lines,CItr 17600 with Miracle spike on the left and Yangmai 18 with a normal or semi-square spike on the right.(B)Mapping of QSl.osu-5A,associated with spike length.Phenotypic data from the F3 population in field condition was integrated with the GBS markers of a linkage group on chromosome 5A.The physical locations of the GBS markers were determined based on the IWGSC Chinese Spring genomic sequence.The horizontal dashed line represents a threshold log of the odds(LOD)value of 3.0.(C)Genetic effects of QSl.osu-5A on spike length.(D,E)Development of two KASP markers flanking the Q gene on chromosome 5A.(D)Q5A-KASP1.(E)Q5A-KASP10.KASP-FAM showing the CItr 17600 allele and KASP-HEX showing the Yangmai 18 allele are indicated in different colors,and NTC(non-template control)used for controls is indicated in red dots.(F,G)Comparison of transcription levels of Q5A/q5A.(F)Q5A/q5A transcription levels in leaves.(G)Q5A/q5A transcription levels in young spikes.Q5A from Yangmai 18,q5A from CItr 17600.

According to the reference genome for the cultivar Chinese Spring(CS)from the International Wheat Genome Sequencing Consortium(IWGSC)[40],this interval contained 30 annotated genes,from TraesCS5A01G473400 to TraesCS5A01G476500.Notably,TraesCS5A02G473800 encoding an AP2-like transcription factor is Q5A.Rather than screening for more recombinants to further narrow down the candidate region,we analyzed Q5A for differences between its CItr 17600 and Yangmai 18 alleles.

3.3.Allelic variation at Q5A

Q5A consists of 10 exons and 9 introns,is 3229 bp in length for its Q5A allele,and encodes a protein of 447 amino acids in CS.We sequenced the complete Q gene from the translation start codon to the stop codon as well as 720 bp upstream of the start codon from CItr 17600(Fig.S1)and from Yangmai 18(Fig.S2).To distinguish from previously published alleles,which are herein designated Q5Aa and q5Aa,the Yangmai 18 allele was named Q5Ab and the CItr 17600 allele q5Ab.We identified 27 SNPs or insertion/deletions(InDels)between Q5Ab and q5Ab.These SNPs and InDels distributed along the promoter,introns,and exons between the two alleles(Figs.S1,S2).To determine if any of the polymorphisms in the regulatory regions affected the transcription of Q5A,we performed RT-qPCR for Q5A in CItr 17600 and Yangmai 18.We observed no significant difference in flag leaves between Q5A in Yangmai 18 and q5A in CItr 17600(P=0.448)(Fig.1F).However,the Q5A allele from Yangmai 18 was more highly expressed than the q5A allele from CItr 17600 in immature spikes(P=0.001)(Fig.1G).

We then turned to the Q5A/q5A coding region:the deduced Q5Ab protein in Yangmai 18 differed from the deduced q5Ab protein in CItr 17600 by an isoleucine residue at position 329 in place of valine in q5Ab.The I329V polymorphism was shown associated with the difference in spike shape,as the I329 isoform produced square spikes in Chinese Spring,while the V329 variant produced a spelt spike in T.dicoccoides(accession TA106)[30].Compared to Q5Aa in Chinese Spring,Q5Ab harbored one SNP in the last intron,and a 1-bp InDel in the 720-bp fragment from the start codon.Compared to q5Aa in TA106,q5Ab showed ample variation over the gene region,with a major difference in exon 10(Fig.2A).

To identify the origins of q5Ab,we performed a Basic Local Alignment Tool(BLAST)search of the genomic sequences deposited in GenBank using q5Ab as a query.We analyzed totally 84 sequences,including 76 complete Q5A sequences from different wheat accessions,four accessions MK423900 to MK423903,which did not include the first 100 bp in exon 1 of Q5A,Q5B and Q5D as controls,plus CItr 17600 and Yangmai 18(Table S3).The q5Ab allele was characterized by the polymorphism in exon 8 for the I329V substitution and the SNP in exon 10 for the miR172 binding site;we detected this combination of SNPs in seven out of 82 accessions analyzed here(Fig.2A),indicating that the q5Ab allele is rare.Of these seven accessions,four belonged to two subspecies of common wheat,Yunnan wheat(T.aestivum ssp.yunnanense)(MK101278 and MK101279),spelt(T.aestivum ssp.spelta)(MK443263 and AY714341).AY714341 accession was a European DS 5A accession in T.aestivum subsp.spelta,which was predicted to harbor a q5A allele for spelt spike[30],although the allele was not experimentally validated.Two other accessions were from emmer wheat(T.turgidum)(GenBank # MK423902 and MK423903),which were reported be associated with branching spikes in the tetraploid wheat species[11].Thirty-two accessions from different diploid,tetraploid,and hexaploid wheat species and subspecies showed the same signature sequence in exon 8 and exon 10 as the q5Aa allele(Fig.2A).

Fig.2.Diverse alleles of Q gene on chromosome 5A.(A)Multiple DNA sequence alignment.Only signature sequence in exon 8 and exon 10 of Q5A/q5A are aligned for comparison with q5Ab in Yangmai 18 and Q5A in CItr 17600.The groups do not represent haplotypes.Number on the right side represents the number of sequence accessions that are deposited in GenBank and provided in Table S3.The sequence of the miR172 binding site is indicated with a square in red.Two SNPs within the miR172 binding site are indicated with an arrow.The sequences of sgRNA and PAM for gene editing are underlined.(B)A possible crossover event within Q/q in two accessions.X represents a crossover.(C)Deduced amino acid sequences in exon 8 and exon 10 of Q5Ab and q5Ab.The two amino acids deletion and two other amino acid substitutions that are resulted from the 6-bp are indicated in red.

The signature sequence of the Q5Ab allele of Yangmai 18 at exon 8 and exon 10 was shared by 30 accessions,which represented different species and subspecies of tetraploid and hexaploid wheat but not diploid wheat(Fig.2A).Furthermore,the Q5Ab allele exhibited the same sequence over the entire gene in the Polish tetraploid T.turgidum ssp.polonicum CItr 191826(GenBank#AY714339)and the Iranian T.aestivum ssp.spelta DS 5A accession(GenBank #AY714340).The latter two accessions were predicted to carry a Q5A allele for the square spike[30],but this allele was not functionally validated.

Interestingly,the signature sequence at exon 8 and exon 10 identified several deviations from the a and b alleles of Q and q genes.Indeed,nine accessions,designated Q5c1 to Qc4[23],harbored four unique polymorphisms in the miR172 binding site(Fig.2A).In addition,we detected a possible crossover event within Q/q in two accessions(Fig.2B).Accession KJ541510 from T.aestivum ssp.sphaerococcum displayed the I329 polymorphism in exon 8 characteristic of the Q5Aa allele,but it presented a C at nucleotide 3139 of exon 10 typically associated with the q5Aa allele.Accession JX524750 from T.aestivum exhibited the opposite genotype,with exon 8 encoding the V329 variant from q5Aa and exon 10 showing a T at nucleotide 3139 normally seen with the Q5Aa allele.

Compared to the other Q/q alleles,q5Ab showed two significant changes at the miR172 binding site in exon 10.One change consisted of the two SNPs G3121T and C3139T that disrupt the miR172 binding site.The second change was a 6-bp InDel immediately upstream of the miR172 binding site resulting in the gain/loss of two amino acids and another two amino acid substitutions(Fig.2C).As it was not clear whether the phenotypic differences between the Q5Ab and q5Ab alleles resulted from changes in transcription or in protein sequence,we employed CRISPR/Cas9-mediated genome editing to validate the functions of Q5Aa.

3.4.Effects of edited Q genes by CRISPR/Cas9

We designed a 20-bp single guide RNA(sgRNA)to edit Q genes in wheat.This sgRNA included 10 bp of the miR172 binding site;importantly,the sgRNA target site was identical between the three homoeologous genes Q5A,Q5B,and Q5D(Fig.2A).We introduced the resulting CRISPR/Cas9-sgRNA-Q construct into immature Yangmai 18 embryos by the gold particle bombardment method.From 38 surviving T0plants out of 1000 transformed embryos,three plants were positive for the construct.Luckily,one positive T0plant(Q-ED11)appeared to have editing events at several Q genes,as evidenced by mismatching sequence signals obtained from PCR products covering the sgRNA target sites.We then cloned and sequenced individual PCR products to define the editing events in the Q genes.We discovered that both Q5A and Q5D were edited in the Q-ED11 plant,but Q5B was not edited in the sequence.At the Q5A locus,we detected a 1-bp insertion of a C nucleotide 5 bp downstream of the protospacer adjacent motif(PAM)(Fig.3A).In the same plant,Q5D harbored a 2-bp deletion downstream of the PAM(Fig.3B).Due to the 1-bp insertion in edited Q5A and the 1-bp insertion in edited Q5D,their reading frames were shift,coding regions were extended,and new stop codons were generated(Fig.3C).The resulting Q5A-ED11 protein differed from Q5A after amino acid 410,with the replacement of the C-terminal 38 amino acids found in wild-type Q5A by 61 new residues from another reading frame(Fig.3D).The resulting Q5D-ED11 protein was predicted to consist of an 86-amino acid C-terminal extension after residue 412 that replaces the C-terminal 39 amino acids present in wild-type Q5D(Fig.3D).

Fig.3.Editing events of Q5A and Q5D.(A)1-bp insertion in Q5A.(B)2-bp deletion in Q5D.The sequences of sgRNA binding site are in bold,PAM sequence is in purple,and the miR172 sequences are underlined(A,B).(C)Comparison of DNA sequences of edited Q5A and edited Q5D.The splicing site AG at the 3′end of intron 9 is indicated with black rectangle,and the sequences from the last exon(exon 10)in Q5A and Q5D are aligned.The sgRNA sequences are indicated with blue rectangle,and the original strop codons are indicated with red rectangle.Due to the 1-bp insertion in Q5A and the 2-bp deletion in Q5D,reading frame was changed,resulting in extended DNA and new stop codon that are indicated with black square.(D)Comparison of DNA sequences of edited Q5A and edited Q5D.The deduced protein sequence start from amino acids encoded by exon 10.Due to 1-bp insertion in Q5A and 2-bp deletion in Q5D,reading frame was changed,resulting in extended DNA that are translated into amino acids after the original stop codons.The deduced amino acids included 61 new residues in edited Q5A and 86 new residues in edited Q5D that are not indicated with red rectangle.

We collected 33 T1seeds from the Q-ED11 T0plant,of which only four T1plants harbored the editing events described above.Two of the four T1plants produced a head but no seeds and died for unknown reasons.The remaining two plants,designated QED11-13(Fig.3C)and Q-ED11-23(Fig.3D)showed spelt spikes and enhanced plant height in the Yangmai 18 background.QED11-13 and Q-ED11-23 plants were self-pollinated to produce T2seeds.We collected 22 T2seeds for Q-ED11-13 and 11 T2seeds for Q-ED11-23.We planted all the seeds and genotyped each plant using primers for the CRISPR/Cas9 construct.None of the T2progeny from Q-ED11-23 had the construct,while the T2progeny from Q-ED11-23 segregated for the CRISPR/Cas9.To test which progeny plant had Q genes edited,we developed marker KASP-Q5A-ED11(Fig.S5A)to query the editing state at Q5A and marker KASP-Q5D-ED11(Fig.S5B)for the editing state at Q5D.We observed that the edited Q5A locus was sufficient to reverse the square spike of wild-type Yangmai 18 into a spelt spike(Fig.4A,B).By contrast,the edited Q5D locus did not appear to affect the square spikes of Yangmai 18 but was instead involved in the shortening of spikes,thus altering spike shape(Fig.4C).Spike length was positively correlated with plant height(Fig.4D).However,the segregation of spike shape and plant height in the T2progeny did not completely align with the editing status at Q5A and Q5D.Surprisingly,we detected both short ramified spikelets and long ramified spikelets in the T2progeny(Fig.4E).Three other traits,grain threshability,glume toughness and rachis fragility,were also associated with the edited Q5A allele(Fig.5).

Fig.4.Effects of edited Q5A-ED11 and edited Q5D-ED11.(A)The whole view of untransformed Yangmai 18(left)and Q-ED11-13(right).(B)The whole view of Q-ED11-23.(C)The whole view of spike shapes in different plants,Q-ED11-13–10(wild-type for both Q5A and Q5D),Q-ED11-23–12(Edited Q5A and wild-type Q5D),Q-ED11-13–12(wildtype Q5A and edited Q5D),and Q-ED11-13–17(Edited Q5A and edited Q5D).(D)The whole view of different plants,Yangmai 18,Q-ED11-23–12(edited Q5A and wild-type Q5D),and Q-ED11-13–18(wild-type Q5A and edited Q5D)(E)The spike phenotype of three spikes for only Q5A-edited plant.

4.Discussion

The primary objective of this study was to use two parental lines with different spike architectures to explore the underlying genes by QTL mapping.The identification of the q5Ab allele will facilitate the selection of elite genes/alleles for ideal plant and spike architectures.Notably,we observed several spike shapes in the CItr 17600×Yangmai 18 progeny,indicative of transgressive segregation for spike traits.The q5Ab allele was also associated with long spelt spikes in the progeny,resulting in the reversion and de-domestication of domesticated traits in common wheat.Previous works[30–32,34]have identified two main polymorphisms that differentiate domesticated 5QA from wild 5qA alleles:one that generates a V(q)to I(Q)substitution(position 329 aa);and a silent C(q)to T(Q)(position 3139 bp)in the miR172 target site that affects miR172 regulation.This study reports the Q5Ab and q5Ab alleles that modulates spike architecture in various genetic backgrounds.

The q5Ab allele had a new G-to-T polymorphism in the miR172 binding site located at 3121 bp in exon 10,in addition to the previously reported T-to-C polymorphism in q5Aa at 3139 bp[32,33].The T3139C polymorphism in the miR172 binding site is sufficient to increase Q5A transcript levels by preventing miR172-dependent mRNA cleavage and/or translational repression[32].The Q5Ab and q5Ab alleles were expressed at significantly different levels in immature spikes,but not in leaves.However,it remains unclear whether the dramatic changes observed for multiple traits were caused by higher Q transcript levels alone,as the Q5A and q5A proteins differed at residue 329,at which position the V329I substitution results in compacted spikes and reduced plant height in domesticated wheat.The regulation of Q protein abundance and function by tae-miR172 and the transcriptional co-repressor TOPLESS(TaTPL)were shown to result in changes in wheat spike architecture[43].The q5Ab allele carried a 6-bp deletion that removed two amino acids and affects two other amino acids.While we cannot distinguish the relative contribution of Q/q alleles from that of the miR172 binding site,it will be important to test if Q5Ab and q5Ab interact with distinct partners in future studies.

We detected the q5Ab allele in only 6 out of 80 wheat accessions whose genomes were deposited in GenBank,indicating that q5Ab is rarely used for the breeding of modern cultivars.Accessions with the q5Ab allele included two of hexaploid T.aestivum ssp.yunnanense,which are considered a sub-origin site for cultivated common wheat[39],as well as the two European hexaploid T.aestivum ssp.spelta accessions MK443263 and AY714341.The European DS 5A accession of T.aestivum subsp.spelta(AY714341)also harbored the q5Ab allele,which was predicted to be a q5A allele for spelt spikes[6],although the allele was not functionally validated.Notably,the q5Ab allele was identical to qdel-A or haplotype 4 in TR119165(GenBank#MK423902)and differed by one SNP from haplotype 3 in TRI3261(GenBank #MK423903)[11].Forty-two out of 44 T.turgidum accessions with branching spikes presented the qdel-A allele,indicating that this allele is used frequently to obtain branching spikes in modern cultivars.

Fig.5.Effects of edited Q5A and edited Q5D on grain threshability and glume toughness.(A)Free threshability.At spikelet level,Q5A-ED spikes were composed of hard glumes,which also tightly covered the grains and prevented free threshing.During threshing,the spikelets were broken with the attached rachis and the glumes had to be manually removed to get the grains from the Q5A-ED spikes.To calculate the free threshability trait,individual spikes were counted for number for spikelets attached to the rachis before and after threshing.For threshing,a single dried spike was passed through a 2-cm wide rubber tube and crushed between hands for 5 times by applying equal pressure.The spikelets attached to the rachis after the crushing were counted and the percentage was calculated.Q5A-ED significantly reduced the threshability.(B)Glumes toughness.Glumes were mostly open and exposing the matured seeds in wild-type and Q5D-GE spikelets of the center of spike,whereas the glumes were tightly covering the seeds in the Q5A-ED spikelets(Upper).Some of these edited plants also showed abnormal central floret growth in each spikelet,which were mostly sterile and did not produce any seeds(Middle).The florets of the lower most spikelets were mostly paired and coved with extremely hard glumes(Lower).(C)Rachis fragility.The rachis of the Q5A-ED spikes was longer and comparatively fragile than the wild-type and Q5D-ED spikes.

The existence of the q5Ab/qdel-A allele in tetraploid T.turgidum and hexaploid T.aestivum ssp.yunnanense and European T.aestivum subsp.spelta accessions provides clues as to the evolution and domestication of modern wheat cultivars.Q5A alleles,including Q5Aa,which is most frequently used in modern cultivars and is represented by four Qcalleles[23],all have polymorphisms in the miR172 binding site,suggesting that the nucleotides within this small 21-bp region are often targeted by mutation in response to selection pressure for different spike architecture in cultivated common wheat.Branching spikes are not heavily selected for by breeders,possibly because this trait depends on the genetic background.It is possible that this trait is also regulated by TEOSINTE BRANCHED1(TB1)and FRIZZY PANICLE(FZP)[11,17,44].Most supernumerary spikelets/florets in the extended spikelets of cultivar CItr 17,600 were sterile in its progeny,but most of the branching spikes in T.turgidum are fertile[11],which is consistent with the notion that several elite genes are masked in hexaploid wheat[45].It will be important to identify suppressors that prevent grain fertility in common wheat and to test if grain fertility rates can be increased by crossing common wheat with branching spikes and wheat cultivars carrying genes promoting high grain fertility[8,46].

We employed CRISPR/Cas9-mediated gene editing to create new mutations in the Q5 loci,whereby the native miR172 binding sites were intact but the encoded proteins were altered.We had expected to observe longer spikes in transgenic Yangmai 18 plants with their Q5Ab edited.Surprisingly,these transgenic plants showed different spike shapes rather than changes in spike length.Another unexpected trait in Q5Ab-edited transgenic Yangmai 18 plants was sham ramification,which was clearly associated with editing at Q5A and not with Q5D,although both loci were edited in this background.Sham ramification entails the formation of ectopic florets and spikelets,resulting in a deviation from the canonical spike and spikelet structures seen in wheat[5,11,33,47,48].However,the sham ramification trait was not inherited in a clear genetic segregation in Q5Ab edited plants,raising a possibility that the unexpected occurrence of these traits might be linked to off-target editing,though no other sequences were identical to the designed sgRNA in the established IWGSC reference sequence for bread wheat[41].The sham ramification trait is caused by a mechanism(s)that is likely more complicated than first anticipated[32,33,43].In a mostly recent study,spike length was regulated by TaCol-B5 on chromosome 7BL,which was cloned using the same CItr 17600×Yangmai 18 population[49].The independent study on TaCol-B5 demonstrated that this gene modified spike architecture and enhanced field-based grain yield[49].The TaCol-B5 gene could be introduced into new varieties that increase the yield potential of cereals[50].

In summary,this study identified the q5A allele as a natural variant in common wheat.This allele is masked in cultivar CItr 17600 with Miracle-type spikes,but its segregation enabled the observation of various spike shapes.Genome editing of Q5A via CRISPR/Cas9 engineered a new spike architecture in wheat.These natural variations and editing events in Q genes provide new insights into the modification of spike architecture in wheat.

CRediT authorship contribution statement

Min Fan:Investigation,Data curation,Formal analysis,Supervision.Xiaoyu Zhang:Investigation,Data curation,Formal analysis.Ragupathi Nagarajan:Investigation.Wenxuan Zhai:Investigation.Yahya Rauf:Investigation.Haiyan Jia:Investigation,Supervision.Liuling Yan:Writing the 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 grants from the USDA National Institute of Food and Agriculture(NIFA)(2017-67007-25932 and 2022-68013-36439)and the Oklahoma Center for Advanced Science and Technology(OCAST,AR17-020-03).Min Fan received scholarships from the China Scholarship Council,and‘‘Programme of Introducing Talents of Discipline to Universities”Project of China(B08025)and a Short-term Exchange Fund in Agricultural College of Nanjing Agricultural University for her Ph.D.research work at the Oklahoma State University.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2022.04.009.