The BnTFL1–BnGF14nu–BnFD module regulates flower development and plant architecture in Brassica napus

Jinjun Wng, Chi Zhng, Youpeng Chen, Ynn Sho, Meifng Lio, Qin Hou, Weitng Zhng,Yng Zhu, Yun Guo, Zijin Liu, Christin Jung*, Mingxun Chen,*

a State Key Laboratory of Crop Stress Biology for Arid Areas, National Yangling Agricultural Biotechnology & Breeding Center, Shaanxi Key Laboratory of Crop Heterosis, and College of Agronomy, Northwest A&F University, Yangling 712100, Shaanxi, China

b College of Agriculture and Biotechnology, Zhejiang University, Hangzhou 310058, Zhejiang, China

c Plant Breeding Institute, University of Kiel, D-24098, Kiel, Germany

Keywords:BnTFL1 CRISPR/Cas Flower development Plant architecture Seed yield

ABSTRACT Flower development and plant architecture determine the efficiency of mechanized harvest and seed yield in Brassica napus.Although TERMINAL FLOWER 1 (AtTFL1) is a regulator of flower development in Arabidopsis thaliana, the function and regulatory mechanism of TFL1 orthologs in B.napus remains unclear.Six BnTFL1 paralogs in the genome of the B.napus inbred line‘K407’ showed steadily increasing expression during vernalization.CRISPR/Cas-induced mutagenesis of up to four BnTFL1 paralogs resulted in early flowering and alteration of plant architecture,whereas seed yield was not altered in BnTFL1 single, double, or triple mutants.Six BnTFL1 paralogs, but not BnaA02.TFL1, showed an additive and conserved effect on regulating flowering time, total and terminal flower number, and plant architecture.BnaA10.TFL1 regulates flower development by interacting with BnaA08.FD through the protein BnaA05.GF14nu, resulting in the transcriptional repression of floral integrator and floral meristem identity genes.These findings about the regulatory network controlling flower development and plant architecture present a promising route to modifying these traits in B.napus.

1.Introduction

Rapeseed(Brassica napus L.,2n=38)is a vegetable oil crop that can also be used as a vegetable, livestock feed, biodiesel, green manure, honey source, and ornamental plant [1].The transition from the vegetative to the reproductive stage of a plant is a complex biological process,including flowering and floral organ development, as well as the establishment of plant architecture.Early flowering supports maximum reproductive success by allowing the plant to evade various environmental stresses during seed ripening [2].Delayed flowering and anomalous floral organs may significantly decrease rapeseed yield, especially under stress conditions [3].Plant architecture, controlled mainly by plant height,inflorescence architecture, and branch number, is a determinant of the efficiency of mechanized harvest and seed yield [2,4].Indeterminate B.napus cultivars are generally taller, their flowering period is more extended, and their siliques mature asynchronously, resulting in yield loss during mechanized harvesting[5,6].Therefore,understanding the genes controlling flower development and/or plant architecture would provide potential targets for molecular breeding in B.napus.

In the model plant Arabidopsis thaliana,floral transition and floral organ specification, two main aspects of flower development,are controlled by multiple pathways [7].During floral transition,the vegetative shoot apical meristem is converted into an inflorescence meristem producing floral meristems [8,9], where a small number of genes trigger floral organ production.Class A genes(APETALA1, AtAP1 and AtAP2), class B genes (AtAP3 and PISTILLATA,AtPI), class C gene AGAMOUS (AtAG),class D genes (AGAMOUS-LIKE 1, AtAGL1, AtAGL5, and AtAGL11), and class E genes (SEPALLATA1,AtSEP1, AtSEP2, AtSEP3, and AtSEP4) are functionally redundant in controlling the specification of the four floral organs:sepals,petals,stamens, and carpels [10].The development of flowers and inflorescences is mediated by TERMINAL FLOWER 1 (AtTFL1) and FLOWERING LOCUS T (AtFT), both of which are members of the phosphatidylethanolamine-binding protein (PEBP) family [11,12].In addition to AtTFL1 and AtFT, this protein family contains four more members: BROTHER OF FT AND TFL1 (AtBFT), ARABIDOPSIS THALIANA CENTRORADIALIS (AtATC), TWIN SISTER OF FT (AtTSF),and MOTHER OF FT AND TFL1 (AtMFT).They form three subfamilies,the AtTFL1-like(AtTFL1,AtBFT,and AtATC),the AtFT-like(AtFT and AtTSF), and the AtMFT-like subfamily [13].The AtTFL1 polypeptide shares high similarity with AtFT,despite their antagonistic functions [11,14–18].The functional divergence between AtTFL1 and AtFT is conferred by five critical amino acid residues and a divergent external loop [15,17,18].Previous studies[8,9,19] indicated that AtTFL1 is predominantly transcribed in the inner part of the central zone of the shoot apical meristem during floral transition, and resulting protein then moves to the outer layer of the shoot apical meristem.AtTFL1, as a repressor of flowering, inhibits the expression of many floral meristem identity genes,including LEAFY(AtLFY),FRUITFULL(AtFUL),AtAP1,and LATE MERISTEM IDENTITY2 (AtLMI2), through its interacting partner of the bZIP transcription factor family AtFD, while AtFT competes with AtTFL1 for the interaction with AtFD, thus activating the expression of these genes to promote flowering [11,12,14,16,19–21].AtTFL1 also acts as a regulator of inflorescence and floral organ development [8,9,11,19].It restricts the spatial expression of the floral meristem identity genes AtAP1, AtLFY, and CAULIFLOWER(AtCAL) to the periphery of the shoot apical meristem to maintain inflorescence indeterminacy and architecture [9,19].Besides,AtTFL1 also controls endosperm cellularization timing and seed size by stabilizing ABSCISIC ACID INSENSITIVE 5 [22].

AtTFL1 orthologs have been cloned from other plant species:Malus domestica [23], Glycine max [24], Oryza sativa [25,26],Gossypium hirsutum[27],Linum usitatissimum[28],Cucumis sativus[29,30],B.juncea[31],and Dendrobium orchids[32],and act in regulating flower development and/or inflorescence architecture similar to AtTFL1.There is increasing evidence[5,6,33–35]that BnTFL1 regulates flowering and/or inflorescence architecture in B.napus.However, the redundancy and molecular mechanism of BnTFL1 governing flower development and plant architecture are still largely unknown.

In this study, we uncovered a critical role of BnTFL1 in flower development and plant architecture.We demonstrated that BnaA10.TFL1 modulates flower development by interacting with BnaA08.FD through the protein BnaA05.GF14nu,ultimately resulting in the repression of floral integrator and floral meristem identity genes by BnaA08.FD.Notably, seed yield was not reduced despite a lower number of flowers per plant in BnTFL1 single,double, and triple homozygous mutants.

2.Materials and methods

2.1.Plant materials and growth conditions

The A.thaliana ecotype Col-0 and the B.napus winter-type inbred line ‘K407’ were used.The T-DNA insertion mutant fd-3(SALK_054421) was in the Col-0 background.All A.thaliana and Nicotiana benthamiana plants were cultured in a greenhouse under long-day conditions with 16 h light (white fluorescent light at 160 μmol m-2s-1) and 8 h dark at 22 °C.For phenotyping experiments,B.napus plants were grown in the greenhouse under longday conditions with 16 h light(white fluorescent light at 300–400 μmol m-2s-1)at 25°C and 8 h dark at 18°C.After 4 weeks,plants were vernalized at 4 °C for 5 weeks under long-day conditions.Plants of similar size were then selected and grown under longday conditions (16 h light at 25 °C and 8 h dark at 18 °C).

2.2.Nucleic acid isolation, primer design, gene cloning, target site design, and plasmid vector construction

The sequences of AtTFL1, AtFT, PRODUCTION OF ANTHOCYANIN PIGMENT 1(AtPAP1),and TRANSPARENT TESTA 8(AtTT8)genes were retrieved from the TAIR database (https://www.arabidopsis.org/)and BnTFL1, BnaA08.FD, and BnaA05.GF14nu genes were retrieved from the Brassica napus pan-genome information resource(BnPIR)database (https://cbi.hzau.edu.cn/bnapus/index.php).Specific primers were designed using the Prime-BLAST tool at National Center for Biotechnology Information (NCBI, https://www.ncbi.nlm.nih.-gov/).Template genomic DNA was isolated from‘K407’leaves with a SteadyPure Plant DNA Extraction Kit (Accurate Bio, Changsha,Hunan,China).Total RNA samples were isolated from shoot apices of A.thaliana Col-0 and‘K407’with the MiniBEST Plant RNA Extraction Kit (TakaRa Bio, Dalian, Liaoning, China).Template cDNA was synthesized from total RNA with a PrimeScript RT Kit(TakaRa Bio).The PCR products were cloned into the pMD19-T vector (TaKaRa Bio), and six randomly selected single colonies were sequenced by Sangon Biotechnology (Shanghai, China).All primers used for gene cloning are listed in Table S1.

To generate BnTFL1 CRISPR/Cas9 constructs,two specific singleguide RNA (sgRNA) sequences, namely sgRNA-1 and sgRNA-2,were designed to target BnTFL1 genes using CRISPRdirect(https://crispr.dbcls.jp).A BLAST search against the B.napus cultivar ‘ZS11’ reference genome revealed no off-targets.Pairs of oligonucleotides of sgRNA-1 and sgRNA-2 were synthesized by Sangon Biotechnology (Shanghai, China) and annealed to generate dimers, which were cloned into the AarI restriction site of the CRISPR/Cas9 vector between the U6-26 promoter and the sgRNA scaffold.

To generate recombinant plasmids with the GFP reporter gene,the coding domain sequences (CDSs) of BnaA02.TFL1, BnaA03.TFL1, BnaA10.TFL1, BnaC03.TFL1, and BnaC09.TFL1 without stop codons were separately amplified and cloned into the pCAMBIA-1300–35S–GFP vector under the control of the CaMV35S(35S)promoter.Similarly, the CDSs of BnaA08.FD and BnaA05.GF14nu without stop codons were separately amplified and cloned into the pGreen–35S–GFP vector under the transcriptional control of the 35S promoter.To generate recombinant plasmids carrying the 6HA reporter gene,the CDSs of BnaA10.TFL1 and BnaA08.FD without stop codons were separately amplified and cloned into the pGreen–35S–6HA vector under the transcriptional control of the 35S promoter.The CDS of BnaA08.FD without stop codons was amplified and cloned into the pGBKT7 vector (Clontech, USA) to generate the pGBKT7–BnaA08.FD construct.

To generate recombinant plasmids carrying the luciferase (LUC)reporter gene, the CDSs of AtPAP1, BnaA10.TFL1, and BnaA08.FD without stop codons were separately amplified and cloned into the JW771 (nLUC) vector and the CDSs of AtTT8, BnaA08.FD, and BnaA05.GF14nu without stop codons were separately amplified and cloned into the JW772 (cLUC) vector.Similarly, the CDSs of BnaA10.TFL1, BnaA08.FD, and BnaA05.GF14nu without stop codons were amplified and cloned into the pGEX–4T-1, pMAL–c2x, and pET28a vectors to generate the glutathione S-transferase (GST)–BnaA10.TFL1, maltose-binding protein (MBP)–BnaA08.FD, and 6xHis–BnaA05.GF14nu constructs, respectively.The CDSs of BnaA10.TFL1 and BnaA08.FD were separately amplified and cloned into pGreenII 62-SK vector under the transcriptional control of the 35S promoter, while the promoters of BnaA03.SOC1, BnaA05.SOC1-1, BnaC03.SOC1, BnaA06.LFY, BnaC03.LFY, BnaCnn.LFY,BnaA02.FUL, BnaA09.FUL, BnaC07.FUL, BnaA07.AP1-1, BnaC01.AP1,BnaC06.AP1, BnaA07.MYB41, BnaA09.MYB41, BnaA08.CAL, and BnaC03.CAL were separately amplified and cloned into pGreenII 0800-LUC vector.All primers used for plasmid construction are listed in Table S1.

2.3.Protein sequence and phylogenetic analysis

The BnPIR database was used to localize the BnTFL1 paralogs on the B.napus chromosomes (cultivar ‘ZS11’).Locations of BnTFL1 paralogs on chromosomes of the B.napus cultivar ‘ZS11’ were drawn using Mapchart.The protein sequences were analyzed using DNASTAR Lasergene 11(DNAStar Co.,Madison,WI,USA).Multiple sequence alignment of the protein sequences of AtFT, AtTFL1, and BnTFL1 was performed with MUSCLE (https://www.ebi.ac.uk/Tools/msa/muscle/).The conserved PEBP protein domain was predicted with the Conserved Domain Search program (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi).Phylogenetic trees were constructed with MEGA7 using the neighbor-joining method with the p-distance model,and bootstrap values were calculated at 1000 replicates.The proteins used in the phylogenetic analysis are listed in Table S2.

2.4.Plant transformation and mutation detection

The 35S:BnaA08.FD–GFP construct was introduced into Agrobacterium tumefaciens strain GV3101, which was then used to transform A.thaliana fd-3 plants via the floral dip method [36].Transgenic T1plants were selected with Basta (Bayer, RZ0160) on soil and confirmed by PCR using the specific primers 35S-P/35S:BnaA08.FD–GFP-Xma I_R.T2and T3seeds were screened on 1/2 MS medium (pH 5.7, 1% sucrose, 1% agar) containing 10 μg mL-1glufosinate-ammonium.To generate transgenic B.napus plants,hypocotyls of ‘K407’ were infected with A.tumefaciens (strain GV3101) carrying recombinant CRISPR-cas or GFP plasmids.T1transformants were selected with hygromycin and confirmed by DNA genotyping using PCR.Rooted T1plants were transplanted into soil and then transferred into the greenhouse after acclimation.DNA genotyping using Cas9-F and Cas9-R primers confirmed the presence of the transgene.To identify putative mutations, the genomic DNA of selected mutants was amplified with primers binding to the respective paralogs.The PCR products were Sanger sequenced (Sangon Biotechnology, Shanghai,China).PCR products indicative of mutations were cloned into the pMD19-T vector(TaKaRa Bio), and six randomly selected single colonies were Sanger-sequenced.Primers used for identifying mutations caused by CRISPR/Cas9 are listed in Table S1.

2.5.Transient expression in N.benthamiana leaves

For subcellular localization assay, recombinant plasmids with the GFP reporter gene were transformed into A.tumefaciens strain GV3101.The bacteria were injected into the leaves of 3-week-old N.benthamiana seedlings.Transient expression was observed as previously described [37].Images were collected using a laser scanning confocal microscope (ZEISS LSM 700, Germany) at 72 h after agroinfiltration.For transient dual-LUC reporter assay, effector and reporter constructs were individually transformed into A.tumefaciens strain GV3101 carrying the pSoup-P19 vector (Weidi Biotechnology, Shanghai, China).They were transiently expressed in the leaves of 3-week-old N.benthamiana seedlings with different effector/reporter combinations.Firefly LUC and renilla luciferase(REN)activities were measured as previously described[37].Relative REN activity was used as an internal control, and LUC/REN ratios were calculated.Each construct was expressed in six different plants.

2.6.Transcriptional activity assay

The recombinant plasmid with GAL4 binding domain and the empty vector pGBKT7 were separately transformed into the Y2HGold yeast strain containing the His and LacZ reporter genes.The transformed Y2HGold cells were cultured on SD medium(Coolaber, China) with or without Trp, Ade, and His at 30 °C for 3 days.

2.7.Plant phenotyping

Transgenic A.thaliana plants were grown in the greenhouse under long-day conditions(16 h light/8 h dark)at 22°C.The number of rosette leaves formed on the main shoot 55 days after sowing (DAS) was measured as a proxy indicator of flowering time.Flowering time of B.napus was measured as the number of days from the end of vernalization to the first flower.Also measured were plant height (distance between the stem base and the tip of the main inflorescence at maturity), branching height (distance between the stem base and the base of the first primary branch),primary branch number(the number arising from the main stem),and secondary branch number (the number arising from the primary branch).Developing seeds and embryos 21, 28, and 35 days after pollination and mature seeds were collected from the main inflorescence and then selected randomly to be photographed by a SZ61 stereomicroscope(Olympus).Seed diameter was measured with ImageJ software and 100 seeds of each line were tested.Inflorescences, floral organs, siliques, and whole plants were selected randomly and photographed with a Nikon camera.For phenotyping, an average of 12 plants per line were used.

2.8.RNA-seq and quantitative real-time PCR (RT-qPCR)

The shoot apices from 12 plants (‘K407’ and s1-1) 63 DAS,and the non-bolting inflorescences from 12 ‘K407’ plants 83 DAS and 12 s1-1 plants 75 DAS were used for an RNA-seq experiment.Plants were grown in the greenhouse under long-day conditions(16 h light at 25 °C and 8 h dark at 18 °C) after vernalization(4 °C, 16 h light/8 h dark, 5 weeks).Three independent biological replicates for each genotype were sequenced by Majorbio Technology Inc.(Shanghai, China) following the standard protocol(https://cloud.majorbio.com/page/flow/view.html?cmd_id = 181).The ‘ZS11’ genome was used as a reference genome (https://www.ncbi.nlm.nih.gov/genome/203?genome_assembly_id =335272).The local BLASTP program was applied to obtain the corresponding A.thaliana orthologs of each B.napus differentially expressed gene (DEG).The annotation data for A.thaliana (TAIR_-Data_20140331) were retrieved from TAIR and used to predict the Gene Ontology (GO) annotations of B.napus DEGs.The Excel add-in for significance analysis of RNA-seq was used to identify DEGs between ‘K407’ and s1-1 plants.DEGs with |log2ratio| ≥1 and P adjust ≤0.05 are listed in Tables S3-S7.

Shoot apices from the B.napus inbred line ‘K407’ at 28, 35, 42,49, 56, 63, 70, 77, and 84 DAS, and young leaves and shoot apices from A.thaliana plants were used to analyze gene expression.The shoot apices at 63 DAS from‘K407’and from two BnaA10.TFL1 single homozygous mutants(s1-1 and s2-1)were used to verify RNAseq results.RT-qPCR was performed for three independent biological replicates and two technical replicates with SYBR Green Master Mix (Cofitt, Hongkong, China) on a Quant Studio 7 Real-Time System.PCR was performed in a total volume of 20 μL containing 200 nmol L-1of each primer and 50 ng cDNA templates using the following cycling conditions: denaturation at 95 °C for 2 min followed by amplification by 40 cycles of 95 °C for 15 s, 58 °C for 30 s,and 72°C for 30 s.Relative expression values were calculated by normalizing against AtEF1αA4 or BnGAPDH with the 2-ΔΔCTmethod.The primers used for gene expression analysis are listed in Table S1.

2.9.Protein interaction assays

For pull-down assay, Escherichia coli BL21 cells harboring the recombinant plasmids and the empty pGEX-4T-1 and pMAL-c2x vectors as controls were incubated at 37 °C for 2 h, followed by 12 h at 22 °C after addition of 1 mmol L-1isopropyl b-D-1-thiogalactopyranoside (Sigma-Aldrich, V900917).All proteins with various tags were purified using the GST-tag Protein Purification Kit, His-Tag Protein Purification Kit (Beyotime, China), or MBPTag Protein Purification Kit (Abbkine, China).The purified 6xHis–BnaA05.GF14nu fusion protein was mixed with the GST or GST–BnaA10.TFL1 fusion protein in 400 μL binding buffer (50 mmol L-1Tris-HCl pH 7.5, 200 mmol L-1NaCl, 1 mmol L-1EDTA, 1%[v/v] NP-40, 1 mmol L-1dithiothreitol (DTT), 10 mmol L-1MgCl2and 1× protease inhibitor, pH 8.0), and then incubated with the GST-tag Purification Resin (Beyotime, China) at 4 °C for 4 h to detect the interaction between BnaA10.TFL1 and BnaA05.GF14nu.The purified 6xHis–BnaA05.GF14nu fusion protein was mixed with MBP or MBP–BnaA08.FD fusion protein in 400 μL binding buffer,and then incubated with the MBP-tag Resin (Abbkine, China) at 4 °C for 4 h to detect the interaction between BnaA08.FD and BnaA05.GF14nu.The purified MBP–BnaA08.FD fusion protein was mixed with GST, GST–BnaA10.TFL1, GST with 6xHis–BnaA05.GF14nu, or GST–BnaA10.TFL1 with 6xHis–BnaA05.GF14nu fusion protein, respectively, and then incubated with the GST-tag Purification Resin(Beyotime,China)at 4°C for 4 h to detect interactions among BnaA10.TFL1, BnaA05.GF14nu, and BnaA08.FD.The resins were washed five times with washing buffer (50 mmol L-1Tris-HCl pH 7.5, 400 mmol L-1NaCl, 1 mmol L-1EDTA, 1 mmol L-1DTT, pH 8.0) and then boiled with SDS-polyacrylamide gel electrophoresis (SDS–PAGE) loading buffer at 94 °C for 5 min.The eluted proteins were separated by SDS–PAGE and immunoblotting was performed with anti-GST, anti-MBP, or anti-His antibodies(Immunoway, 1:2000 dilution).

For split-LUC assay, the recombinant vectors carrying the LUC reporter gene were separately transformed into A.tumefaciens strain GV3101.The bacteria were mixed and injected into the leaves of 3-week-old N.benthamiana seedlings.The infiltrated plants were grown for 72 h and the infiltrated leaves were then sprayed with 1 mmol L-1luciferin (Yuanye, China) and kept in the dark for 5 min.LUC activity was imaged with a low-light cooled charge-coupled device imaging apparatus (Tanon 4600, China).

For co-immunoprecipitation(Co-IP)assay,leaves of 3-week-old N.benthamiana seedlings were injected with A.tumefaciens(strain GV3101),harboring the 6HA and GFP fusion constructs.After 72 h following injection, the leaves were collected and ground in liquid nitrogen and total protein was extracted using lysis buffer(50 mmol L-1Tris-HCl pH 8.0, 150 mmol L-1NaCl, 2 mmol L-1EDTA,10%[v/v]glycerol,5 mmol L-1DTT,1%[v/v]NP-40,1 mmol L-1PMSF and 1× protease inhibitor).After centrifugation at 12,500×g at 4 °C for 15 min, the supernatant was collected and transferred to a new tube.An aliquot of 50 μL of each protein extract was kept as an input, and the remaining protein extract was immunoprecipitated for 4 h by anti-HA magnetic beads(Thermo Fisher, Cat# 88837) at 4 °C with gentle shaking.Next,the beads were washed three times with washing buffer(50 mmol L-1Tris-HCl pH 8.0,150 mmol L-1NaCl,2 mmol L-1EDTA,10%[v/v]glycerol,1 mmol L-1DTT,0.15%[v/v]Triton X-100,1 mmol L-1PMSF and 1× protease inhibitor) and then boiled with SDS–PAGE loading buffer at 94 °C for 5 min to elute the proteins.Finally,immunoprecipitated and input proteins were separated by SDS–PAGE, and immunoblotting was performed with anti-HA and anti-GFP antibodies (Immunoway, 1:2000 dilution).

3.Results

3.1.The rapeseed genome encodes six BnTFL1 paralogs

A search for ‘TERMINAL FLOWER 1’ genes in the BnPIR database revealed six paralogs located on six chromosomes in the genome of B.napus cultivar‘ZS11’(Fig.S1;Table S8).We constructed paralogspecific primers binding upstream of the start codon and downstream of the stop codon(Fig.S2)and amplified the six BnTFL1 paralogs using genomic DNA of the B.napus inbred line ‘K407’.All amplicons carried the same CDS as ‘ZS11’, except for BnaC02.TFL1,which differed only by two single-nucleotide polymorphisms(SNPs, Fig.S3).

The polypeptide encoded by TFL1 from the B.napus inbred line‘K407’shared more than 86%amino acid identity with AtTFL1 and more than 84% identity with the other five BnTFL1 proteins(Fig.S4).The amino acid sequences of BnaA02.TFL1 and BnaC02.TFL1 are identical (Fig.S4).Like AtTFL1, the four BnTFL1 proteins contained a highly conserved PEBP domain and five amino acid residues critical for TFL1 function [15,18].However, BnaA10.TFL1 and BnaC09.TFL1 differed from the other sequences by an Arg instead of a His at position 88 (Fig.S4A).A previous study [17]had shown that the divergent external loop essential for TFL1 function varied in size by 14–17 amino acid residues between different plant species.Consistently,the BnTFL1 protein’s divergent external loop ranged from 15 to 17 aa,with 10 of these conserved across all proteins (Fig.S4A).

In the phylogenetic tree incorporating the BnTFL1 sequences with 110 members of the PEBP family from nine plant species(Table S2), the 116 PEBP proteins were divided into the TFL1-like clade,the FT-like clade,and the MFT-like clade(Fig.S5).All BnTFL1 proteins clustered with TFL1 proteins from other plant species.BnaA03.TFL1 and BnaC03.TFL1 displayed the highest similarity to AtTFL1 from A.thaliana (Fig.S5).

Taken together, the polypeptide structure suggests that all six BnTFL1 proteins might have similar functions as AtTFL1.

3.2.Expression patterns of BnTFL1 paralogs

We measured the dynamic expression of BnTFL1 paralogs in‘K407’ shoot apices at multiple growth stages by RT-qPCR.We designed paralog-specific primers to distinguish between the six transcripts except between BnaA02.TFL1 and BnaC02.TFL1 and between BnaA10.TFL1 and BnaC09.TFL1, which could not be separated owing to the similarity of their sequences.BnTFL1 expression peaked at the end of vernalization at 63 DAS except in BnaA03.TFL1,whose expression peaked at 56 DAS(Fig.S6A).The BnTFL1 expression profiles were consistent with previous studies [6,34,35], further supporting their function in flower development.

Previous studies[11,19]have indicated that AtTFL1 functions as a transcriptional cofactor without a DNA-binding domain.Multiple sequence alignments of TFL1 proteins indicated that the six BnTFL1 proteins also lack a DNA-binding domain (Fig.S4A).We further determined the subcellular localization of the BnTFL1 proteins by transiently expressing BnTFL1-GFP fusion products in N.benthamiana leaf epidermal cells.Across all constructs used, a fluorescent signal was visible in the cytoplasm and nucleus (Fig.S6B).

In brief, these results implied that BnTFL1 proteins, located in the cytoplasm and nucleus, might function in the shoot apical meristem to regulate flower development in B.napus.

3.3.BnTFL1 is critical for flower development in B.napus

In the next step,we investigated the BnTFL1 function by a gene knock-out approach in the inbred line ‘K407’.We found two regions highly conserved between the six BnTFL1 paralogs suitable as sgRNA targets.The first region (sgRNA-1) lies in the first exon with 100% sequence identity between four BnTFL1 paralogs,whereas two paralogs displayed an SNP.The second region(sgRNA-2) in exon 4 is 100% identical to BnaA10.TFL1 and BnaC09.TFL1 (Fig.S2).The constructs were introduced into the inbred line‘K407’by Agrobacterium-mediated hypocotyl transformation.After hygromycin selection, 52 sgRNA-1 and 15 sgRNA-2 transgenic plants(termed the T1generation)were obtained.Sixteen T1plants(#1 – #16) showed a terminal flower phenotype in contrast to untransformed ‘K407’ controls, which showed indeterminate growth.

We searched for mutations by sequencing PCR amplicons encompassing the target region.All T1plants showed gene editing in up to three paralogs (Fig.1A; Table S9).Altogether, 20 different mutant alleles with nucleotide insertions and deletions were identified(Fig.1A;Table S9)together with non-edited alleles indicating chimerism of the T1plants.Four plants (#1, #12, #14, and #15)were homozygous for mutant alleles(Fig.1A;Table S9).Seeds from 16 T2populations were generated by selfing the T1plants.For each T2plant,the presence of the transgene was determined by PCR.The BnTFL1 target sites were sequenced with paralog-specific primers.All T1mutant alleles were found in the T2generation(Fig.1A).Consistent with the presence of the Cas9 transgene in T2populations,new mutant alleles appeared in nine T2populations(progeny of T1plants#5,#6,#9,#10,#11,#12,#13,#15,and#16)which had not been detected in their T1parents (Fig.1A).Complex editing patterns were found even in T2plants, illustrating that Cas9 activity gave rise to de novo mutations in this generation(Fig.1A).At least two plants of each T3population were tested for the presence of the transgene, and T-DNA-free plants were identified.The BnTFL1 paralogs enclosing the target region were amplified from nontransgenic T3plants and the amplicons were sequenced.Four single, five double, two triple, and two quadruple homozygous nontransgenic mutants were identified(Fig.1).Fifteen mutations were predicted to cause frame shifts and result in non-functional proteins, whereas three mutations (s1-3, s1-4, and s1-9) resulted in the deletion or substitution of one or more amino acids (Fig.1B;Table S10).

We produced T4populations where non-transgenic homozygous plants for the BnTFL1 mutations were selected for phenotypic analysis.We selected 13 primary T4populations to measure the effect of the mutations on flower development.The T4homozygous mutants flowered between 79 and 86 DAS, whereas the control flowered at 92 DAS.Compared with ‘K407’, all 13 homozygous T4knockout mutants showed early flowering(Fig.2A, D).The double mutants homozygous for the BnaA03.TFL1 and BnaC03.TFL1 (s1-4),and BnaC02.TFL1 and BnaC03.TFL1 alleles (s1-5) flowered earlier than the single mutant homozygous for the BnaC03.TFL1 allele(s1-2;Fig.2A,D),indicating that BnaA03.TFL1 and BnaC02.TFL1 also suppresses flowering in B.napus.In contrast to the donor line, all homozygous mutants displayed terminal flowers at the apex(Fig.2B), which showed floral organ aberration: pedicel absence,sepal,petal,and stamen reduction and/or absence,and carpel connection and/or split(Fig.2C).But they still produced some normal flowers on the basal parts of inflorescences before the floral termination(Fig.2B, C,F).There was a correlation between the number of mutant alleles and the terminal flower effect.The more mutations a plant carried,the earlier it flowered(Fig.2A,D).The number of total flowers and the proportion of normal flowers on the main inflorescence decreased with the number of mutant alleles(Fig.2E,F).In contrast, the proportion of terminal flowers on the main inflorescence increased (Fig.2G).

In the next experiment, we overexpressed BnaA10.TFL1 and BnaC03.TFL1 in the B.napus line ‘K407’.These sequences were selected because they showed respectively the lowest and highest amino acid identities with AtTFL1 (Fig.S4).For each construct(Fig.S7A), at least 10 independent T1transgenic plants were obtained after hygromycin selection and PCR amplification using specific primers (Table S1).Finally, three independent T3homozygous for each transgene were selected for hygromycin tolerance and further confirmed by PCR amplification with specific primers(Fig.S7B,C),and the expression of the BnTFL1 genes was measured by RT-qPCR.The expression levels of the two BnTFL1 paralogs were much higher than in‘K407’(Fig.S7D,E).Overexpression of BnaA10.TFL1 or BnaC03.TFL1 delayed flowering by 11 and 13 days, respectively, relative to ‘K407’ (Fig.S8A, C) without changes in floral organ morphology, terminal flower formation, plant height, total flower number (main inflorescence), or primary and secondary branch number, but with greater branching height (Fig.S8A, B,D–H).

In summary, the knockout and overexpression studies suggested that all six BnTFL1 paralogs but BnaA02.TFL1 are repressors of flowering in B.napus and function in an additive manner.

3.4.BnTFL1 affects plant architecture, but not seed yield

On the hypothesis that BnTFL1 mutations would also affect plant architecture and seed yield, 13 homozygous T4mutant lines were grown in the greenhouse with ‘K407’ controls.Plants homozygous for BnTFL1 mutations were shorter and lower across all T4knock-out mutant lines (Fig.3A;Table S11).Flowering characters,plant height,and branching height were correlated with the number of mutant alleles, suggesting that all six BnTFL1 paralogs except for BnaA02.TFL1 regulate these characters too in an additive manner.

The shoot number, a major component of plant architecture, is controlled by multiple factors, including hormones and various developmental signals [38].All homozygous mutants showed a decrease in primary branch number and an increase in secondary branch number (Fig.3A; Table S11).All carried fewer siliques per plant than the controls.The homozygous quadruple mutants displayed significantly fewer siliques per plant than the other mutants (Table S11).Although all homozygous mutants showed increased seed size, seed embryo size, seed weight, and silique length and width, there were no significant differences among them(Fig.3B–D;Table S11).All showed a decrease in seed number per plant in comparison with the controls, and in particular the homozygous quadruple mutants produced fewer seeds per plant than the other mutants, because BnTFL1 mutations did not affect seed number per silique(Table S11).Except for the BnTFL1 quadruple homozygous mutant, the seed yield per plant of homozygous BnTFL1 mutants did not differ from ‘K407’ (Table S11).

In summary,these results indicate that BnTFL1 modulates plant architecture mainly by forming terminal flowers.In contrast,yield components were altered in neither the single, double, or triple homozygous mutants nor the overexpression lines.

3.5.BnaA10.TFL1 regulates the expression of genes involved in flower development

Fig.1.CRISPR/Cas9-induced mutagenesis of BnTFL1 genes in the B.napus inbred line ‘K407’.(A)Pedigree of BnTFL1 mutant generations derived from 16 different T1 plants.Each box represents a single plant.Red and black boxes indicate respectively transgenic and non-transgenic plants with a terminal flower phenotype.The number of plants used to test for the presence of the transgene in the offspring of T1 and T2 plants is shown in blue.Seed codes are shown above each genotype.(B) Sequence alignment between wild type and BnTFL1 homozygous mutants from the T3 generation containing the target sites.The mutants generated at target sites sgRNA-1 and sgRNA-2 were named s1 and s2.Protospacer-adjacent motif(PAM)sites are shown with blue letters,InDels are marked in red,and red‘-’symbols represent deletions.Letter codes A,B,C,D,E, and F indicate alleles of BnaA02.TFL1, BnaA03.TFL1, BnaA10.TFL1, BnaC02.TFL1, BnaC03.TFL1, and BnaC09.TFL1, respectively.The wild type allele is indicated by the suffix h,whereas edited alleles are numbered according to their mutation types (Table S10).Homozygous plants used for phenotyping were obtained in the T3 generation.

Fig.2.BnTFL1 modulates flower and inflorescence development in B.napus.(A–C) The onset of flowering 16–28 days after vernalization (A), the formation of determinate inflorescences(B),and terminal flowers(C).Red arrows indicate the inflorescence apex.Normal flowers are those produced before floral termination.Terminal flowers refer to a cluster of flowers produced at the apex of the inflorescence with anomalous floral organs.(D–G)Days to flowering time(D),number of total flowers(E),percentages of normal flowers(F),and terminal flowers(G)on the main inflorescence.Values are means±SD of 12 plants.Different lowercase letters among‘K407’and various homozygous mutants indicate differences at P ≤0.05 (one-way ANOVA with Tukey’s test).T4-mutants homozygous for BnTFL1 mutations and the inbred line ‘K407’ were grown in the greenhouse under long-day conditions (16 h light at 25 °C and 8 h dark at 18 °C) after vernalization (4 °C, 16 h light/8 h dark, 5 weeks).

We wished to determine how BnaA10.TFL1 regulates flower development in B.napus in concert with other genes.Accordingly,we performed a transcriptome analysis using the ‘K407’ and BnaA10.TFL1 single homozygous mutant s1-1.Respectively 2690 and 6304 transcripts were differentially expressed in shoot apices and non-bolting inflorescences (Table S3).In shoot apices, 1059 DEGs were upregulated (Tables S3, S4), and 1631 downregulated(Tables S3, S5), whereas in non-bolting inflorescences, 3196 DEGs were upregulated (Tables S3,S6) and 3108 downregulated(Tables S3,S7).Sets of 48(4.53%)upregulated genes and 45(2.76%)downregulated genes influenced flowering time, whereas 90 (8.50%)upregulated genes and 30(1.84%)downregulated genes influenced floral organ development in shoot apices (Tables S3-S5).In nonbolting inflorescences, 59 (1.85%) upregulated genes and 42(1.35%) downregulated genes influenced flowering time, and 272(8.51%) upregulated genes and 215 (6.92%) downregulated genes influenced floral organ development (Tables S3, S6, S7).

In the next experiment,we measured the effect of BnTFL1 mutations on floral integrator and floral meristem identity genes known to exert pleiotropic effects[39,40].The expression of 17 floral integrator and floral meristem identity genes with at least one G-box cis motif in their promoter regions was measured by RT-qPCR.Their expression patterns were consistent with those identified by RNA-seq (Fig.4B).

A homolog of one of these genes, AtSOC1, integrates multiple flowering signals to regulate floral transition and organ development in A.thaliana [41].The expression of its B.napus ortholog,BnSOC1, was much higher in the shoot apices of s1-1 and s2-1 mutants than in ‘K407’, whereas it was not altered in non-bolting inflorescences (Fig.4; Tables S4, S6, S7).Five other genes, AtLFY,AtFUL, AtAP1, AtLMI2, and AtCAL, control flowering time and floral meristem identity in A.thaliana [39,40].We investigated the expression profiles of their B.napus orthologs.Three BnLFY, four BnFUL, three BnAP1, two BnMYB41 (homologous to AtLMI2), and two BnCAL were significantly upregulated in the shoot apices of s1-1 and s2-1 mutants.Their expression levels were not changed in non-bolting inflorescences, except for the downregulation of BnaCnn.LFY(Fig.4;Tables S4,S6,S7).When the genes were classified according to the ABCDE floral-development model,the class A(BnAP1), B (BnAP3 and BnPI), C (BnAG), and E (BnSEP1, BnSEP2,

BnSEP3, and BnSEP4) genes were significantly upregulated in the shoot apices of s1-1 mutant relative to ‘K407’ (Fig.4A; Table S4).In contrast, no different expression of these genes mentioned above was observed in non-bolting inflorescences of the s1-1 mutant,except for the upregulation of BnaA06.SEP4(Fig.4A;Tables S6, S7).The expression levels of class D (BnAGL1 and BnAGL11)genes were not altered in the shoot apices of the s1-1 mutant compared to ‘K407’, but were upregulated in non-bolting inflorescences (Fig.4A; Tables S4-S6).

In conclusion,the expression patterns suggest that BnaA10.TFL1 regulates, in a tissue-specific manner, downstream targets from multiple regulatory pathways involved in flower development.

3.6.BnaA10.TFL1 interacts with BnaA08.FD via BnaA05.GF14nu

In A.thaliana, AtTFL1 works as a transcriptional cofactor, interacting with the bZIP transcription factor AtFD and the 14-3-3 protein AtGF14v, which serves as a bridge protein [12,14].We speculated that BnTFL1 is recruited to the downstream genes by crosstalk with BnFD via the B.napus 14-3-3 protein.A search for‘FD’and‘14-3-3-like protein’genes in the BnPIR database revealed six BnFD paralogs and 50 14-3-3 paralogs in the genome of the B.napus cultivar ‘ZS11’.For further investigation, we focused on the BnFD and 14-3-3 paralogs BnaA08.FD and BnaA05.GF14nu,respectively, because they have the highest amino acid identities with AtFD and AtGF14v (Fig.S9).

First,we investigated the function of BnaA08.FD by overexpression in the late-flowering A.thaliana fd-3 mutant [42].At least 10 independent T1transgenic plants were obtained by Basta selection and the presence of the transgene was determined by PCR amplification with specific primers(Table S1).Three T3lines homozygous for the transgene were selected and a high expression of the BnaA08.FD gene was further measured in these transgenic plants by RT-qPCR (Fig.S10).The ectopic expression of BnaA08.FD in the fd-3 mutant resulted in early flowering (Fig.S11A, B), suggesting that BnaA08.FD acts similarly to FD from A.thaliana in accelerating flowering time.

In the next experiment, we investigated the subcellular localization of BnaA08.FD in N.benthamiana leaf epidermal cells where a 35S:BnaA08.FD–GFP construct was transiently expressed.We observed GFP within the nucleus (Fig.S11C).Yeast cells transformed with pGBKT7–BnaA08.FD fusion construct activated the expression of the reporter genes and survived in the selective medium (SD/-Trp/-Ade/-His; Fig.S11D).These findings support the hypothesis that BnaA08.FD functions as a transcription factor like AtFD from A.thaliana [20].BnaA10.TFL1–GFP too was localized in the cytoplasm and the nucleus (Fig.S6B), suggesting that BnaA08.FD might interact with BnaA10.TFL1 in the nucleus.

To investigate how the B.napus TFL1, GF14nu, and FD proteins interact,we conducted a series of experiments in E.coli and N.benthamiana.Pull-down assays in E.coli showed that BnaA10.TFL1 and BnaA08.FD interact with BnaA05.GF14nu in vitro(Fig.5A,B).GST–BnaA10.TFL1 and the MBP–BnaA08.FD did not coprecipitate when incubated in the absence of 6xHis–BnaA05.GF14nu,but did coprecipitate in the presence of 6xHis–BnaA05.GF14nu, indicating that BnaA05.GF14nu serves as a bridge protein to form the BnaA10.TF L1–BnaA05.GF14nu–BnaA08.FD module (Fig.5C).

Next,split-LUC and Co-IP experiments were performed by transient expression of BnaA10.TFL1, BnaA05.GF14nu, and BnaA08.FD in N.benthamiana leaf epidermal cells.Strong LUC signals were observed when any two proteins of BnaA10.TFL1,BnaA05.GF14nu,and BnaA08.FD were co-expressed in N.benthamiana leaves(Fig.5D–F).Consistently, the Co-IP results revealed that BnaA05.GF14nu and BnaA08.FD were separately immunoprecipitated with BnaA10.TFL1, whereas BnaA05.GF14nu was immunoprecipitated with BnaA08.FD, further supporting the in vivo interactions of any two proteins of BnaA10.TFL1, BnaA05.GF14nu, and BnaA08.FD in N.benthamiana leaves (Fig.5G–I).BnaA10.TFL1 interacted with BnaA08.FD (Fig.5F, I), suggesting that at least one functional homolog of BnaA05.GF14nu was expressed and displayed conserved functions with BnaA05.GF14nu in N.benthamiana leaves.

Thus,interaction between BnaA10.TFL1 and BnaA08.FD requires the BnaA05.GF14nu protein to form the BnaA10.TFL1–BnaA05.G F14nu–BnaA08.FD module.

3.7.The BnaA10.TFL1–BnaA05.GF14nu–BnaA08.FD module inhibited the expression of floral integrator and floral meristem identity genes

The transcriptome study revealed many DEGs associated with flower development (Fig.4; Tables S4-S7).We investigated the effect of the BnaA10.TFL1–BnaA05.GF14nu–BnaA08.FD module on the transcription activities of selected floral integrator and floral meristem identity genes,using dual-LUC reporter assays in N.benthamiana leaves.We generated reporter constructs containing the B.napus SOC1,LFY,FUL,AP1,MYB41,and CAL promoter-driven firefly LUC gene, and the 35S promoter-driven renilla LUC gene(Pro35S:REN; internal control).The pGreenII 62-SK empty vector and the pGreenII 62-SK recombinant vector containing the BnaA10.TFL1 or BnaA08.FD genes were used as effectors (Fig.6A).The BnaA10.TFL1 or BnaA08.FD genes and the LUC gene under the transcriptional control of the B.napus SOC1, LFY, FUL, AP1, MYB41,or CAL promoters were transiently co-expressed in N.benthamiana leaves.If only one gene construct was used for transformation,there were no significant differences in the LUC/REN ratio compared to the empty effector(Fig.6B).However,the LUC/REN ratios of these reporters were markedly reduced when BnaA10.TFL1 and BnaA08.FD were simultaneously expressed, except for the reporters driven by BnaA05.SOC1-1,BnaC03.SOC1,and BnaA07.AP1-1 promoters (Fig.6B).

Fig.4.BnaA10.TFL1 regulates the expression of key genes related to flower development.(A)Heat map of differentially expressed genes involved in flowering time and floral organ development in shoot apices and non-bolting inflorescences of the BnaA10.TFL1 single homozygous mutant (s1-1) relative to the wild type (‘K407’).The numerical values for the red to blue gradient bars represent log2 fold change(FC)in the expression of genes in the s1-1 mutant versus those in the wild type(‘K407’).(B)RT-qPCR with floral integrator and floral meristem identity genes expressed in shoot apices of‘K407’and two BnaA10.TFL1 single homozygous mutants(s1-1 and s2-1)63 days after sowing.Results were normalized against the expression of BnGAPDH as an internal control.Values are means ± SD of three biological samples along with two technical repeats.Asterisks indicate significant differences in gene expression levels between mutants and ‘K407’ (two-tailed paired Student’s t-test, P ≤0.05).

Fig.5.BnaA10.TFL1 interacts with BnaA08.FD through BnaA05.GF14nu as a bridge protein.(A, B) In vitro pull-down assay with BnaA10.TFL1 and BnaA05.GF14nu (A), and between BnaA08.FD and BnaA05.GF14nu(B).(C)In vitro pull-down assay to detect the formation of the BnaA10.TFL1-BnaA05.GF14nu-BnaA08.FD module.The proteins were purified in E.coli.(D–F) N.benthamiana split-LUC assay of the interaction between BnaA10.TFL1 and BnaA05.GF14nu (D), BnaA08.FD and BnaA05.GF14nu (E), and between BnaA10.TFL1 and BnaA08.FD(F).The interaction between AtPAP1 and AtTT8 was a positive control[60].A vector containing only cLUC was used as a negative control.cLUC,C-terminal luciferase; nLUC, N-terminal luciferase.(G–I) Co-IP assay of the interaction between BnaA10.TFL1 and BnaA05.GF14nu (G), BnaA08.FD and BnaA05.GF14nu (H),and between BnaA10.TFL1 and BnaA08.FD (I).Total protein was extracted from agroinfiltrated N.benthamiana leaves and immunoprecipitated by anti-HA antibodies.

Thus, a protein complex formed by BnaA10.TFL1, BnaA05.GF14nu, and BnaA08.FD acts in regulating flower development by inhibiting the expression of floral integrator and floral meristem identity genes (orthologs of SOC1, LFY, FUL, AP1, MYB41, and CAL;Fig.7).

4.Discussion

Rapeseed is a major source of vegetable oil and protein-rich livestock feed.Given that flower development and plant architecture strongly influence seed yield, investigating the genetic mechanism underlying these traits could improve rapeseed breeding.

Fig.6.Transient dual-LUC reporter assay of transcriptional repression of floral integrator and floral meristem identity genes by the BnaA10.TFL1–BnaA05.GF14nu–BnaA08.FD module.(A) Schematic presentation of vector constructs showing effector genes with and without BnaA10.TFL1 or BnaA08.FD and the reporter genes under the control of various B.napus genes written as BnX (BnaA03.SOC1, BnaA05.SOC1-1, BnaC03.SOC1, BnaA06.LFY, BnaC03.LFY, BnaCnn.LFY, BnaA02.FUL, BnaA09.FUL, BnaC07.FUL, BnaA07.AP1-1,BnaC01.AP1, BnaC06.AP1, BnaA07.MYB41, BnaA09.MYB41, BnaA08.CAL, and BnaC03.CAL).(B) Transient dual-LUC reporter assay.Each reporter construct was combined with control (empty), or BnaA10.TFL1, or BnaA08.FD, or both BnaA10.TFL1 together with BnaA08.FD effector construct and transiently expressed in N.benthamiana leaves.The expression of renilla luciferase (REN) was used as an internal control, and the LUC/REN ratio represents relative promoter activity.Values are means ± SD of six plants.Lowercase letters indicate differences among the control (empty) and the effector constructs at P ≤0.05 (one-way ANOVA with Tukey’s test).

Fig.7.A proposed model of the BnaA10.TFL1–BnaA05.GF14nu–BnaA08.FD module regulating flower development in B.napus.During vernalization,BnaA10.TFL1 expression steadily increases in young seedling shoot apices,and its encoded protein cooperates with BnaA08.FD through the protein BnaA05.GF14nu to form the BnaA10.TFL1–BnaA05.GF14nu–BnaA08.FD module, thus inhibiting the expression of floral integrator and floral meristem identity genes.

In the present study, we aimed to characterize the redundancy and underlying regulatory mechanism of TFL1 orthologs in B.napus.In A.thaliana, loss of AtTFL1 function accelerated flowering and altered inflorescence and plant architecture,resulting in terminal flowers at the apex of the inflorescence and increased seed size[8,11,22].The allopolyploid B.napus genome resulted from an interspecific hybrid between B.rapa and B.oleracea.One gene in A.thaliana corresponds to multiple orthologs in the B.napus genome, most of which are duplicated and functionally similar and redundant [43].Accordingly, there are six BnTFL1 paralogs in the B.napus inbred line ‘K407’ genome (Fig.S4).Knocking out one or more paralogs of BnTFL1 resulted in early flowering, altered inflorescence and plant architecture, and increased seed size (Figs.2,3; Table S11).Overexpression of either BnaA10.TFL1 or BnaC03.TFL1 in B.napus delayed flowering and increased branching height(Fig.S8A–D).Our findings agree with a previous finding [44] that Brassica TFL1 proteins generally carry an Arg instead of a His at position 88.At position 88,BnaA10.TFL1 and BnaC09.TFL1 proteins carry an Arg instead of a His (Fig.S4A), which, however, does not affect their functions as flowering-time, plant-architecture, and seed-size regulators.The finding that the six BnTFL1 proteins shared high amino acid sequence similarity and contained the highly conserved PEBP protein domain (Fig.S4) confirms their close evolutionary relationship with AtTFL1.These findings indicate that BnTFL1 paralogs and AtTFL1 play a conserved role in regulating flower development, plant architecture,and seed size in B.napus.

In a previous study [33], a BnTFL1-2 (BnaA10.TFL1) mutation (a one amino acid substitution,Gly105Arg)slightly delayed flowering and increased seed yield per plant in the B.napus line‘Express 617’.A mutation of either Bnsdt1(BnaA10.TFL1) or Bnsdt2 (BnaC09.TFL1)did not influence flowering but caused the formation of determinate inflorescences in the B.napus line ‘4769’ [5,35].BnA10.TFL1 rescued the determinate inflorescences in the B.napus line ‘517’[6].In the B.napus cultivar ‘Westar’, knockout of any of the five BnTFL1 paralogs except for BnaC02.TFL1 resulted in the alteration of plant architecture and a decrease in silique length and seed number per silique, whereas a single BnaC03.TFL1 mutation promoted flowering[34].Five of six BnTFL1 paralogs suppressed flowering in an additive manner (Fig.2A, D), and inhibited silique development, but did not affect seed number per silique in the B.napus inbred line ‘K407’ (Fig.3D; Table S11).The BnaA10.TFL1 mutation did not alter seed yield per plant in the B.napus inbred line ‘K407’ (Table S11).Numerous sequence variations have been reported in the promoter and/or coding region of BnFT, BnFLC,and BnCO from winter, semi-winter, and spring ecotypes of B.napus,which resulted in a different expression of these genes,thus leading to the divergence of flowering[45,46].These together indicate that the BnTFL1-mediated regulation of flowering, silique development, and seed yield (Figs.2A, D, 3D; Table S11) [5,33–35] depends on the genetic background and may be caused by sequence variations in the promoter and or/coding region of the respective ecotypes.Seed yield per plant is determined by silique number per plant,seed number per silique,and single seed weight[47].Seed yield per plant is limited by sink size during silique development [48], and a larger silique with more or larger seeds resulted in higher seed yield[2].In agreement with these findings,in our study seed yield per plant was unaltered in BnTFL1 single,double, and triple homozygous mutants, whereas it was reduced in the BnTFL1 quadruple homozygous mutant, indicating that the larger seeds and siliques could compensate for reduced seed number in the single,double,and triple homozygous mutants but not in the quadruple homozygous mutants (Table S11).

How does BnTFL1 control flowering and floral organ development at the transcriptional level?In A.thaliana,the floral transition is controlled by regulatory networks composed mainly of photoperiod, vernalization, ambient temperature, age, autonomous, and gibberellin pathways[49].The coordinated spatiotemporal expression of genes in these pathways determines flower development[50].CYCLING DOF FACTOR 1 (AtCDF1), AtCDF3, AtCDF5, and SUPPRESSOR OF PHYA-105 3 (AtSPA3) repress flowering via the photoperiod pathway in A.thaliana [51,52].The miR156-SQUAMOSA PROMOTER BINDING LIKEs (AtSPLs) module defines the age pathway, and AtSPL3/5/10 promote flowering by activating the expression of AtLFY, AtAP1, and AtFUL in A.thaliana [53,54].FLOWERING PROMOTING FACTOR 1(AtFPF1)was proposed[55]to promote flowering via the gibberellin pathway in A.thaliana.Thus, the reduced expression of BnCDF1, BnCDF3, BnCDF5, and BnSPA3 and increased expression of BnFPF1,BnSPL3,BnSPL5,BnSPL8-1,and BnSPL10 in the shoot apices of the s1-1 mutant(Tables S4,S5)may account for the modulation by BnaA10.TFL1 of flowering in B.napus.

The signals perceived by the above six pathways of the onset of flowering ultimately converge on floral integrator genes and then activate the floral meristem identity genes,thereby inducing flowering and specifying the inflorescence and floral meristem [50].The floral integrator gene AtSOC1 and the floral meristem identity genes AtLFY,AtFUL,AtAP1,AtLMI2,and AtCAL function in the formation of inflorescences and floral organs[39,40,56].AtSOC1 is a transcriptional regulator of AtLFY, AtFUL, and AtAP1 [41,56].AtLFY controls the expression of AtAP1, AtLMI2, and AtCAL, and in turn,is activated by AtAP1 and AtLMI2 [40,57,58].AtAP1 induction also depends on AtLMI2 and AtCAL [40,58].Previous studies[11,14,25,26] with A.thaliana and O.sativa indicated that TFL1 interacts with the bZIP transcription factor FD via the 14-3-3 bridge protein; however, in C.sativus, neither CsTFL1 nor CsTFL1d interacts with CsFD or Cs14-3-3 proteins, but both interact with CsNOT2a [29,30], implying that TFL1 interacting partners are not conserved among plant species.In our study,BnaA10.TFL1 cooperated with BnaA08.FD through the bridge protein BnaA05.GF14nu to form the BnaA10.TFL1–BnaA05.GF14nu–BnaA08.FD module(Fig.5C).The transcript levels of the floral integrator gene of BnSOC1 and the floral meristem identity genes of BnLFY, BnFUL,BnAP1,BnMYB41,and BnCAL were increased in s1-1 and s2-1 shoot apices (Fig.4; Table S4).The BnaA10.TFL1–BnaA05.GF14nu–Bna A08.FD module suppressed the expression of these floral integrator and floral meristem identity genes (Fig.6B).The finding that the module did not repress the expression of the floral integrator genes BnaA05.SOC1-1 and BnaC03.SOC1 or the floral meristem identity gene BnaA07.AP1-1 (Fig.6B) suggests that in the BnaA10.TFL1 loss-of-function mutant, the upregulation of BnaA05.SOC-1,BnaC03.SOC1, and BnaA07.AP1-1 may be mediated by alterations of flowering pathways of photoperiod, age, and gibberellins, and by the expression of BnSPLs, BnSOC1, BnLFY, BnMYB41, and BnCAL(Tables S4, S5).However, a role of BnaA08.FD or FD PARALOG paralogs cannot be ruled out [11,20].

After the specification of inflorescences and floral meristems,the identities of floral organs in each floral whorl are shaped by the ABCDE genes [10,50].Given that abnormal expression of ABCDE genes causes abnormal development of floral organs[10,50], the higher expression of class A (BnAP1), B (BnAP3 and

BnPI), C (BnAG), D (BnAGL1 and BnAGL11), and E (BnSEP1, BnSEP2,BnSEP3, and BnSEP4) genes (Fig.4A; Tables S4, S6) could be one cause of floral organ aberrations in the s1-1 mutant.The finding that class A, B, C, and E genes were upregulated in shoot apices(Fig.4A; Table S4), whereas the class D gene was upregulated in non-bolting inflorescences of the s1-1 mutant (Fig.4A; Table S6)suggests that BnaA10.TFL1 sculpts the patterning, differentiation,and formation of floral organs in a tissue-specific manner.

Do TFL1 play differing roles in the floral transitions of B.napus and A thaliana? Exposing A.thaliana seedling to a short period of vernalization induces flowering, whereas winter-type rapeseed needs a prolonged period of vernalization.BnTFL1 expression in shoot apices of young‘K407’plants increased during vernalization and decreased after vernalization (Fig.S6A).This finding, together with the earlier flowering of the BnTFL1 mutants (Fig.2A, D) suggests that BnTFL1 is required to prevent flowering when young winter-type B.napus plants are exposed to vernalization.Expression of BnSPL genes was increased in the shoot apices of the s1-1 mutant (Table S4).However, AtSPL genes were not regulated by AtTFL1 in A.thaliana[14,16].We propose that,unlike in A.thaliana,BnTFL1 mediates an age-dependent vernalization response in winter-type B.napus.

We also investigated how BnTFL1 controls plant architecture.Plant architecture is controlled by a complex regulatory network involving hormones and various genes [59].The AtTFL1 mutation affects plant architecture mainly via the formation of determinate inflorescences caused by the occurrence of terminal flowers[8,9,11,16,19].AtTFL1 also controls branch fate by repressing AUXIN RESPONSE FACTOR 5 (AtARF5), LONELY GUY 5 (AtLOG5), ABI FIVE BINDING PROTEIN 2(AtAFP2),and SMAX1-LIKE 8(AtSMXL8),thereby regulating plant architecture [12,16].This suggests that except for the formation of terminal flowers, the increased expression of

BnARF5, BnLOG5, BnSMXL8, and BnAFP2 in the s1-1 mutant (Tables S4,S6) supports the BnTFL1 regulatory role in shaping plant architecture in B.napus.The finding that primary and secondary branch numbers did not vary among homozygous mutants (Fig.3A;Table S11) implies that BnTFL1 integrates signals from different regulatory networks.

In summary,the study describes the function of TFL1 paralogs in rapeseed and how BnTFL1–BnGF14nu–BnFD regulates flowering,plant architecture, and seed yield.This could open new perspectives to modify plant architecture.CRISPR/Cas9-induced mutation of BnTFL1 will permit shortening growing periods and confer improved rapeseed architecture, which may provide targets towards defining potential breeding strategies aiming at addressing longer growing periods in the face of subsequent crop cultivation and adverse environmental conditions.

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.

CRediT authorship contribution statement

Jianjun Wang:Conceptualization, Formal analysis, Investigation,Visualization, Writing-original draft-review & editing.Chi Zhan:Formal analysis, Investigation, Visualization.Youpeng Chen:Formal analysis, Visualization.Ya-nan Shao:Visualization.Meifang Liao:Visualization.Qian Hou:Visualization.Weitang Zhang:Visualization.Yang Zhu:Writing-original draft.Yuan Guo:Investigation, Validation.Zijin Liu:Investigation, Validation.Christian Jung:Writing-original draft-review & editing, Supervision.Mingxun Chen:Conceptualization, Resources, Project administration, Funding acquisition, Writing-original draft-review & editing,Supervision.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2023.10.002.