Jin Li,Zheng Wng,Zhenyi Chng,Hng He,Xioyn Tng,Ligeng M*,Xing Wng Deng,*

a Peking University Institute of Advanced Agricultural Sciences,Weifang 261325,Shandong,China

b Shenzhen Institute of Molecular Crop Design,Shenzhen 518107, Guangdong,China

c College of Life Sciences,Capital Normal University,Beijing Key Laboratory of Plant Gene Resources and Biotechnology for Carbon Reduction and Environmental Improvement,Beijing 100048,China

Keywords:

A B S T R A C T TaMs1 encodes a non-specific lipid transfer protein(nsLTP)and is required for pollen development in wheat.Although MS1 is a Poaceae-specific gene,the roles of MS1 genes in other Poaceae plants are unknown,especially in rice and maize.Here,we identified one ortholog in rice(OsLTPg29)and two orthologs in maize(ZmLTPg11 and ZmLTPx2).Similar to TaMs1,both OsLTPg29 and ZmLTPg11 genes are specifically expressed in the microsporocytes,and both OsLTPg29 and ZmLTPg11 proteins showed lipid-binding ability to phosphatidic acid and several phosphoinositides.To determine their roles in pollen development,we created osltpg29 mutants and zmltpg11zmltpx2 double mutants by CRISPR/Cas9.osltpg29,not zmltpg11zmltpx2,is defective in pollen development,and only OsLTPg29,not ZmLTPg11,can rescue the male sterility of tams1 mutant.Our results demonstrate that the biological function of MS1 in pollen development differs in the evolution of Poaceae plants.

1.Introduction

In flowering plants,pollen grains represent male gametophytes consisting of two small sperm cells and a large vegetative cell and play important roles in sexual reproduction and life cycle[1].Mature pollen grains contain several structures with lipids as the main components:extracellular pollen wall,intracellular extensive membrane network and storage oil bodies[2].Great efforts have been made in studying the composition,biogenesis and function of the lipidic structure of pollen grains,and there have been many significant progresses[3–6].

The pollen wall is not only a barrier protecting male gametophytes from various biotic and abiotic stresses,but also an intermediary mediating pollen-stigma interaction and facilitating fertilization[7].Although the surface morphologies of pollen grains from different plants are various,the pollen wall is commonly comprised of three layers:the pollen coat,the outer exine layer and the inner intine layer[5,8].The pollen coat,also called‘‘pollenkitt”or‘‘tryphine”,is mainly composed of non-polar esters such as sterol esters and saturated acyl group[3,9];the exine layer is mostly made up of sporopollenin,a series of resistant biopolymer of polyhydroxylated aliphatic compounds and phenolics[3,5].The precursors of pollen coat and sporopollenin are synthesized and secreted by the tapetum[3].The intracellular membrane system of pollen grains,including extensive network of endoplasmic reticulum(ER),many membranous vesicles,and organelles including mitochondria and Golgi apparatus,is made up of phospholipids, namely phosphatidylcholine (PC), phosphatidylethanolamine(PE)and phosphatidylinositol(PI)[3].Phospholipid signaling is involved in pollen development,pollen germination and pollen tube elongation[4].The intracellular storage oil bodies whose dominant component is triacylglycerols(TAG)not only provide energy to the maturation of pollen grains after the loss of their major nutrient source-the tapetum,but also serve as energy reserves for pollen germination[3].

The process of pollen development involves considerable trafficking of lipids from tapetum into the pollen exine and exchanging of lipids between different intracellular membranes,which is mediated by transmembrane ABCG transporters or other lipid transporters such as non-specific lipid transfer protein(nsLTP)[10].nsLTPs which exist in all land plants were named for the character of non-specific binding to different lipids and the ability to facilitate transfer of phospholipids between membranesin vitro[11].nsLTP sequences are characterized by a conserved eight cysteine motif(8 CM)with the general pattern C-Xn-C-Xn-CC-Xn-CXC-Xn-CXn-C[12].Initially,based on sequence similarity and putative molecular weight of mature protein,nsLTPs from flowering plants are classified into two main families,Type 1(9 kDa)and Type 2(7 kDa)[13].With the identification of nsLTPs in lower plants which show rather low sequence similarity to those in flowering plants outside the 8 CM,a new classification system was developed based on the presence of glycosylphosphatidylinositol(GPI)modification site,the presence and position of introns,amino acid sequence identity and spacing between the cysteine residues.In this system,the two subfamilies established already,Types 1 and 2,are retained,while the other nsLTPs are classified into the subfamilies Types C–K[14].

It has been reported that nsLTPs play critical roles in pollen and anther development of many flowering plants.CaMF2fromChili pepper(Type C nsLTP according to Edstam’s classification system)is expressed specifically in anther with a peak at stage 4,and inhibition ofCaMF2resulted in shriveled pollen grains and low pollen germination ability[14,15].Two Type C nsLTP proteins fromArabidopsisare secreted to the locule of anther and become components of the pollen exine,and double RNAi knockdown of these two genes partially impaired the intine underneath the exine as well as viability of pollen on dehydration[16].

A point mutation ofZmMs44(a Type C nsLTP from maize)blocked the cleavage of signal peptide and impeded the secretion of protein from tapetal cells into the locule,resulting in dominant male sterility[17].LTPc3aandLTPc3b(two Type C nsLTPs from triticale)are expressed in the tapetal layer cells at the tetrad and uninucleate microspore stage,and the LTP3c-GFP translational fusion expressed in transgenicBrachypodium distachyonwas produced in the tapetum and secreted into the anther locule,indicating the possible functions in pollen development[18].In addition,TDNA insertion lines ofAtLTPG3andAtLTPG4(two Type G nsLTPs fromArabidopsis)showed deformed or collapsed pollen grains[19].TaMs1andTaMs5encode two Type G nsLTPs from wheat.TaMs1is expressed in secondary sporogenous cells and highly expressed in microsporocytes[20]while the peak ofTaMs5-Atranscripts present in pre-meiotic anthers[21].Bothtams1andtams5mutants bear aborted pollen grains in anthers,leading to male sterility[20–22].Furthermore,riceOsC6(a special nsLTP)is expressed predominantly in tapetum and weakly in microspores from stage 9 to stage 11 of anther development,and the RNAi lines displayed defective orbicule and exine development and reduced male sterility[23].

In this study,we identified one ortholog ofTaMs1from rice(OsLTPg29)and two orthologs from maize(ZmLTPg11andZmLTPx2).OsLTPg29andZmLTPg11are specifically expressed in the microsporocytes,and the corresponding proteins showed the same lipid-binding ability as TaMs1 protein.Theosltpg29mutants created by CRISPR displayed male sterility astams1mutants,while the double mutantszmltpg11zmltpx2did not.In addition,OsLTPg29could rescue the male sterility oftams1mutant,whileZmLTP11could not.These results indicate that althoughMS1genes exist extensively in the Poaceae lineage,their biological function may diverge during the evolution.

2.Materials and methods

2.1.Plant materials and growth condition

All plants were grown in a greenhouse at a white light intensity of 250 mmol m-2s-1or experimental station in Beijing.Bread wheat was grown under long-day conditions(16 h of light at 22–25 °C/8 h of dark at 15–20 °C),and rice and maize were grown under short-day conditions(12 h of light at 25–30 °C/12 h of dark at 22–25 °C).

2.2.Phylogenetic tree analysis

The protein sequences of 27 rice Type G nsLTPs(including OsLTPg29),26 maize Type G nsLTPs(including ZmLTPg11)plus ZmLTPx2,were download from the databases of wheat,rice and maize genome according to Wei et al.[29]and the protein sequence of ZmLTPx2 was corrected according to the actually amplified CDS sequence.Sequences were aligned with Clustal X 2.1,and the phylogenetic tree was constructed using Molecular Evolutionary Genetics Analysis(MEGA)10.1.7 software by the neighbor-joining method.A bootstrap analysis of 1000 replicates was performed to provide confidence estimates for the tree topologies.Full-length protein sequences of Type G nsLTPs were used for constructing the phylogenetic tree.

2.3.Synteny analysis

Synteny blocks of proteins were determined using jcvi which is the python version of MCscan[24].Parameters were applied by default settings(jcvi.compara.catalog ortholog--dbtype prot)and the software divided blocks according to the cscore filter(cscore≥0.70).

2.4.qRT-PCR and in situ hybridization

The developmental stages of the anthers were determined as reported previously[25].For gene expression analysis,rice tissues at the heading stage—including leaf,stem,root,glume,palea,lemma,pistil,and anthers at different developmental stages—were collected.Total RNA was isolated using TRI Reagent(Takara Bio Inc.,Japan),genomic DNA was removed with DNase I(Promega,Wisconsin,USA),and then cDNA was synthetized using a First-Strand cDNA Synthesis Kit(Thermo Fisher,Cleveland,OH,USA),according to the manufacturer’s instructions.qPCR was performed with the Applied Biosystems 7500 Real-Time PCR System(Thermo Fisher)using SYBR Premix Ex Taq II(Takara Bio Inc.,Japan).Each experiment was repeated biologically three times,each with three replicates.OsACTIN1andZmACTIN1were used as the normalizing reference gene.The relative expression levels were measured using the 2-ΔCt analysis method,and the results were represented as the means±SEM.The primers used for qRT-PCR are provided in Table S1.

In situhybridization was performed according to Wang et al.[20].Tissues were cut into 8-μm-thick sections and hybridization was performed overnight at 50 °C.A 832-bpOsLTPg29fragment and a 654-bpZmLTPg11fragment were amplified using primers OsLTPg29-ISH-F/R or ZmLTPg11-ISH-F/R(Table S1),respectively,and inserted into pEASY-T1 Simple Cloning Vector(TransGen Biotech,China)in both forward and reverse orientations.The vectors were linearized by digestion withHind III orEcoR I,and then used as a template to generate anti-sense and sense probes with T7 RNA polymerase.

2.5.CRISPR/Cas9-mediated mutation

The oligonucleotide pairs including targeting sequences(Table S2)fromOsLTPg29gene andZmLTPg11/ZmLTPx2gene were synthesized as primers,annealed and cloned intoBsaI-digested pHUN411 vector and pBUE411 respectively[26].The CRISPR/Cas9 vectors with different targeting sequence were delivered into rice protoplasts or maize protoplasts by PEG-mediated transfection,the PCR/RE assay was performed and the mutation rates were calculated[27].The CRISPR/Cas9 vector was introduced inAgrobacterium tumefaciensstrain AGL0 and transformed into rice ZH11 or maize Z31,as described previously[28].The positive mutants were selected from the transgenic plants by PCR/RE assay and were confirmed by sequencing.The target sequences and primers used in PCR/RE assay were provided in Table S2.

2.6.Complementation of the mutants

For functional complementation oftams1mutants,the casettesTaMs1p::gOsLTPg29orTaMs1p::gZmLTPg11were inserted into pAHC20 digested withHind III to yieldpAHC20-TaMs1p::gOsLTPg29orpAHC20-TaMs1p::gZmLTPg11,respectively.TaMs1pis a 2205 bp DNA fragment upstream of the start codon ATG ofTaMs1gene.gOsLTPg29is a 2504 bp genomic DNA fragment containing theOsLTPg29coding region.gZmLTPg11is a 1690 bp genomic DNA fragment containing theZmLTPg11coding region.

These constructs were transformed into the calli induced from the immature embryo of heterozygousms1gplants via particle bombardment,and the transgenic plants were selected and regenerated.The transgenic plants in thems1g/ms1gbackground were identified by PCR amplification and sequencing.

All the primers used for vector construction are listed in Table S1.

2.7.Protein-lipid overlay assay

MBP-TaMs1-His and MBP-His were expressed as described previously[20].To prepare the constructs expressing MBP-OsLTPg29-His and MBP-ZmLTPg11-His,primers OsLTPg29-His-F/R and ZmLTPg11-His-F/R were used,respectively,to amplify the CDS fragment encoding truncated proteins(lacking the signal peptide and C-terminal hydrophobic amino acid rich region)fused with 6×His at the C-terminus.

The sequences of the primers used for PCR are given in Table S1.Next,the PCR products were in-fused into pMal-c2x(New England Biolabs,Ipswich,MA,USA)digested withEcoR I andPstI.

PIP lipid strips(P-6001)and membrane lipid strips(P-6002)were purchased from Echelon Biosciences(Salt Lake City,UT,USA).Expression and purification of the fusion proteins inEscherichia coliand protein-lipid overlay assay was performed as described previously[20].

3.Results

3.1.Othologs of TaMs1 in rice and maize are specifically expressed in the microsporocytes

By BLAST analysis withTaMs1CDS,LOC_Os03g46110 was identified in rice genome with highest similarity,which encodes a protein containing 228 amino acid showing 59.5%–63.3% amino acid identity with the three TaMs1 proteins(Fig.1A,B)and has been named OsLTPg29 according to the classification system proposed by Edstam et al.[29].When the similar BLAST was performed in maize genome of B73,GRMZM2G151021 and GRMZM2G166484 were found,which have also been namedZmLTPg11andZmLTPx2[29].The BLAST in the genome of 12 other maize varieties,such as CML247,DK105,and EP1,also gave the same result,indicating that there are only these two genes in maize genome.ZmLTPg11 shows higher identity with three TaMs1 proteins(58.8%–60.4%)than ZmLTPx2(50.5%–50.8%)on account of a shorter C-terminal of ZmLTPx2(Fig.1A,B).All of the three proteins contain the conserved 8 CM,with signal peptide at N-terminal and hydrophobic amino acid rich region at C-terminal(Figs.1A,S1).To confirm whether OsLTPg29,ZmLTPg11,and ZmLTPx2 are orthologs of TaMs1,a phylogenetic tree was constructed using full-length protein sequences of all 27 Type G nsLTPs(including OsLTPg29)from rice,all 26 Type G nsLTPs(including ZmLTPg11)plus ZmLTPx2 from maize,and three TaMs1(Figs.1C,S2).OsLTPg29,ZmLTPg11,ZmLTPx2,and three TaMs1 were in the same cluster.Furthermore,synteny analysis of the regions where these genes are located in wheat,rice,and maize genome were carried out by the modified MCScan algorithm.The collinear blocks includingTaMs1and other gene pairs were found.The collinear block between rice and wheat contains 59 gene pairs,includingOsLTPg29.The collinear blocks between maize in 2 and 10 chromosome and wheat contain 20 and 21 gene pairs,includingZmLTPg11andZmLTPx2,respectively(Fig.S3).Phylogeny and synteny studies suggested thatOsLTPg29,ZmLTPg11,andZmLTPx2were reliable orthologs ofTaMs1in rice and maize.

Fig.1.Sequence alignment of TaMs1 protein and its orthologs in some grass species.(A)The sequence alignment of TaMs1 and its orthologs in rice and maize by software BioEdit 7.0.9.0.(B)Amino acid sequence identity between each other by Clustal Omega(http://www.clustal.org/omega/).(C)Phylogenetic tree of TaMs1,its orthologs and next similar Type G nsLTPs in rice and maize using MEGA 10.1.7 software.The numbers at the nodes show bootstrap values obtained for 1000 replicates.

RNA expression analysis by qRT-PCR revealed that the transcripts ofOsLTPg29were detectable in developing anthers,starting at stage 6(pre-meiosis)and peaking at stage 8(meiosis),but not detectable in the root,stem,leaf,glume,palea,lemma,pistil and anthers at stages 10–12(Fig.2A).To more precisely determine the temporal and spatial expression ofOsLTPg29,RNAin situhybridization was performed with wild type anthers.In the anthers of rice,the expression ofOsLTPg29gene was not detected at stage 4(Fig.2B).From stage 5 to stage 8b,strong expression ofOsLTPg29gene was detected in sporogenous cells and meiotic cell,not in the tapetum and epidermis(Fig.2C–E).At stage 9,no RNA expression was observed in the anther(Fig.2F).No signal was detected in the negative control using senseOsLTPg29transcript as a probe(Fig.2G).Similarly,qRT-PCR andin situhybridization were used to detect the expression pattern ofZmLTPg11andZmLTPx2.qRT-PCR showed thatZmLTPg11andZmLTPx2were specifically expressed in the anthers at pre-meiosis and meiosis stage(Fig.2H).Further,RNAin situhybridization result indicated thatZmLTPg11were expressed in sporogenous cells at premeiosis stage(Fig.2J)and meiotic cells at early meiosis stage(Fig.2K),but not in tetrads at late meiosis stage(Fig.2L).ZmLTPg11expression was also not detectable in unicellular microspore(Fig.2M),tapetum and epidermis(Fig.2I–M).

The expression pattern ofOsLTPg29,ZmLTPg11,andZmLTPx2is similar to that ofTaMs1previously observed[20]which suggested that OsLTPg29,ZmLTPg11 and ZmLTPx2 may also be involved in microspore development as TaMs1.

3.2.Both OsLTPg29 and ZmLTPg11 are phospholipid-binding proteins

Both OsLTPg29 and ZmLTPg11 have eight conserved crysteine residues(known as nsLTP domains)as their ortholog TaMs1 which has lipid-binding activity.To determine whether OsLTPg29 and ZmLTPg11 also bind lipids directly,we performed protein-lipid overlay assay.Recombinant MBP-OsLTPg29-His and MBPZmLTPg11-His were expressed in and purified from bacteria(Fig.3A,B).Commercially available strips spotted with lipids were incubated with recombinant proteins,then the proteins which bind to lipids were detected by immunoblotting using anti-MBP antibodies.As previously observed for TaMs1,both MBPOsLTPg29-His and MBP-ZmLTPg11-His showed lipid-binding ability to PA and several PIPs,including PI(3)P,PI(4)P,PI(5)P,PI(3,4)P2,PI(3,5)P2,PI(4,5)P2,and PI(3,4,5)P3,and no lipid-binding signal was detected for MBP-His control(Fig.3C,D).This result demonstrated that both OsLTPg29 and ZmLTPg11 have the same lipidbinding ability as TaMs1.

3.3.Mutation in OsLTPg29 by CRISPR/Cas9 leads to male sterility

To study whetherOsLTPg29is required for microgametogenesis,CRISPR/Cas9 system was used to disrupt endogenousOsLTPg29gene and createosltpg29mutants.

We designed four sgRNAs targeting exon 1 ofOsLTPg29(TS1–TS4)(Fig.S4A)and tested the abilities of these sgRNAs in rice protoplasts by PCR/RE assay.sgRNAs which targetOsLTPg29TS2 and TS3 displayed higher mutation efficiencies(12.9% and 10.0%)(Fig.S4B).Cloning and sequencing of the uncut bands revealed various mutations in the targetedOsLTPg29gene(Fig.S4C).

The vectors containing Cas9 and sgRNA targeting TS2 or TS3 were transformed into rice callus cells mediated byAgrobacterium tumefaciens.44 transgenic lines for TS2 target and 67 lines for TS3 target were obtained.PCR/RE assay was performed to identify the mutagenesis ofOsLTPg29in transgenic plants.InOsLTPg29TS2 target,22 mutation lines were identified,12 of which had biallelic mutations and 10 lines had mosaic or monoallelic mutations,and the mutation rate was 50.0%.InOsLTPg29TS3 target,a total of 27 mutation lines were identified,15 of which had biallelic mutation and 12 lines had mosaic or monoallelic mutations,and the mutation rate was 40.3%(Table 1).The biallelic mutations were confirmed by sequencing,showing deletions and insertions inOsLTPg29gene(Fig.4A).

Fig.2.Expression pattern of OsLTPg29,ZmLTPg11,and ZmLTPx2.(A)Quantitative RT-PCR analysis of OsLTPg29 expression in different organs of rice.OsACTIN1 served as a control.Data are shown as means±SEM(n=3).St8,stage 8;St10,stage10;St12,stage12;DAF,days after flowering.(B–G)in situ hybridization analysis of OsLTPg29 in anthers of rice.(B)An anther at stage 4 showing no OsLTPg29 expression.(C)An anther at stage 5 showing OsLTPg29 expression in sporogenous cells.(D)An anther at stage 7 showing OsLTPg29 expression in meiotic cell.(E)An anther at stage 8b showing OsLTPg29 expression in tetrads.(F)An anther at stage 9 showing no OsLTPg29 expression in microspores.(G)Hybridization with the sense probe produced no signal at stage 7.(H)Quantitative RT-PCR analysis of ZmLTPg11 and ZmLTPx2 expression in different organs of maize.ZmACTIN served as a control.Data are shown as means±SEM(n=3).PM,pre-meiosis;M,meiosis;UM,unicellular microspore stage;BP,bicellular pollen stage;MP,mature pollen stage;G,glume.(I–N)In situ hybridization analysis of ZmLTPg11 in anthers of maize.(I)An anther at secondary sporogenous cells formation stage showing no ZmLTPg11 expression.(J)An anther at pre-meiosis stage showing ZmLTPg11 expression in sporogenous cells.(K)An anther at early meiosis stage showing ZmLTPg11 expression in meiotic cells.(L)An anther at late meiosis stage showing no ZmLTPg11 expression in tetrads.(M)An anther at unicellular microspore stage showing no ZmLTPg11 expression.(N)An anther at early meiosis stage with the ZmLTPg11 sense probe.MC,meiotic cell;Msp,microspore;SP,sporogenous cell;T,tapetum;Tds,tetrads.Scale bars,20 μm.

Fig.3.Both OsLTPg29 and ZmLTPg11 bind phospholipids in vitro.(A)Coomassie blue staining of purified MBP-His,MBP-TaMs1-His,MBP-OsLTPg29-His and MBP-ZmLTPg11-His protein.M,marker.(B)Western blot analysis of purified MBP-His,MBP-TaMs1-His,MBP-OsLTPg29-His and MBP-ZmLTPg11-His protein with MBP antibody.(C,D)protein-lipid overlay assay.Purified MBP-His,MBP-TaMs1-His,MBP-OsLTPg29-His and MBP-ZmLTPg11-His protein were overlaid on an Echelon P-6001 lipid strip(C)or Echelon P-6002 membrane strip(D).CL,cardiolipin;Cs,cholesterol;DAG,diacylglycerol;LPA,lysophosphatidic acid;LPC,lysophosphocholine;PA,phosphatidic Acid;PC,phosphatidylcholine;PE,phosphatidylethanolamine;PG,phosphatidylglycerol;PI,phosphatidylinositol;PI(3)P,phosphatidylinositol-3-phosphate;PI(4)P,phosphatidylinositol-4-phosphate;PI(5)P,phosphatidylinositol-5-phosphate;PI(3,4)P2,phosphatidylinositol-3,4-bisphosphate;PI(3,5)P2,phosphatidylinositol-3,5-bisphosphate;PI(4,5)P2,phosphatidylinositol-4,5-bisphosphate;PI(3,4,5)P3,phosphatidylinositol-3,4,5-triphosphate;PS,phosphatidylserine;SM,sphingomyelin;S1P,sphingosine-1-phosphate;TAG,triacylglycerol.

To investigate whetherOsLTPg29is required for male fertility,the transgenic lines with biallelic mutations inOsLTPg29gene(osltpg29mutants)were further analyzed.When flowering,osltpg29biallelic mutants(26 out of 27)produced the anthers which were slightly smaller and indehiscent,without any other obvious differences in vegetative growth,inflorescence and flower compared with that of wild type(Fig.4B–E).I2-KI staining of pollen grains showed that the pollen grains of these biallelic mutants were shrunken and inactive(Fig.4F;Table 1).These results suggested that OsLTPg29 contributes to male fertility in rice.

Table 1Mutation rate of T0 plants and male fertility of osltpg29 mutants.

3.4.zmltpg11zmltpx2 double mutants created by CRISPR/Cas9 display normal male fertility

To investigate whetherZmLTPg11andZmLTPx2 are also required for male fertility,we used CRISPR/Cas9 technology to disruptZmLTPg11andZmLTPx2simultaneously to createzmltpg11zmltpx2double mutants.

Fig.4.The phenotypes of the osltpg29 mutants and zmltpg11zmltpx2 double mutants created by CRISPR/Cas9.(A)Representative mutations in OsLTPg29 gene from T0 transgenic plants.Green letters indicate the PAM sequences.Restriction sites are underlined.Deletions and insertions are indicated by dashes and blue letters,respectively.Numbers on the right side indicate types of mutation and numbers of nucleotides involved.(B)WT plant and osltpg29 mutant plants after bolting.Scale bar,10 cm.(C)Spikelets of WT and osltpg29 mutants with palea and lemma removed.Scale bar,1 mm.(D)Anthers of WT and osltpg29 mutants.Scale bar,0.5 mm.(E)Carpels of WT and osltpg29 mutants.Scale bar,0.5 mm.(F)Pollen grains of WT and osltpg29 mutants stained with I2-KI.Scale bar,100 μm.osltpg29-1 represents plant T0-44;osltpg29-2 represents plant T0-39 generated by vector pBUN422-OsLTPg29 TS3.(G)Representative mutations in ZmLTPg11 and ZmLTPx2 from the double mutants.(H)Tassels of WT and zmltpg11zmltpx2 double mutants.Scale bar,5 cm.(I)Anthers of WT and zmltpg11zmltpx2 double mutants.Scale bar,2 mm.(J)Pollen grains of WT and zmltpg11zmltpx2 double mutants stained with I2-KI.Scale bar,200 μm.

Firstly,we designed a sgRNA which targets exon1 and is strictly conserved in both genes(Fig.S5A).Then,the CRISPR/Cas9 vectorpBUE411-ZmLTPg11/ZmLTPx2was constructed and tested in maize protoplast.As shown in Fig.S5B and C,the mutations including insertion,deletion and substitution were achieved in bothZmLTPg11(21.5%)andZmLTPx2(26.7%).

Next,the vector was introduced into maize variety Z31 byAgrobacterium-mediated transformation.A total of 9 primary T0transgenic lines were regenerated,among which there were 7 transgenic lines carrying biallelic mutations in bothZmLTPg11andZmLTPx2(zmltpg11zmltpx2double mutant),and 2 transgenic lines carrying biallelic mutation inZmLTPg11and mosaic or monoallelic mutation inZmLTPx2(Fig.S6A,B).At florescence stage,only 8zmltpg11zmltpx2double mutant plants from 4 T0transgenic lines developed tassels and I2-KI staining of pollen grains showed no difference between wild type and these 8 plants.(Fig.S6C).In greenhouse,the T0plants except 5-1 and 6-1 developed tassels and ears at different periods,or didn’t develop tassel,so we had to pollinate the T0plants with pollen grains from wild type plants to keep the mutations.

A total of 197 F1(progenies of T0plants crossed with WT)or T1(progenies of self-bred T0plants)plants were examined using PCR/RE assay to identity the mutations inZmLTPg11andZmLTPx2,and were also genotyped for the stable insertion of sgRNA and Cas9 cassettes(Table S3).A total of 65 plants carrying bialleliczmltpg11mutations and bialleliczmltpx2mutations(zmltpg11zmltpx2double mutant)were identified from 8 T0transgenic lines(Table S3).I2-KI staining of pollen grains from these 65 double mutant plants showed that their pollen grains are as normal as that of wild type.

To eliminate the interference of sgRNA and Cas9 onZmLTPg11andZmLTPx2,progenies of five F1or T1plants(13,54,166,175,and 186)which were derived from three individual T0lines without sgRNA and Cas9 insertion(Table S3)were followed by genotyping and phenotyping.14zmltpg11zmltpx2double mutant plants were identified(Fig.4G;Table S4)and displayed normal pollen development(Fig.4H–J).According to the genotype and phenotype analysis of the three generations,it was concluded thatZmLTPg1andZmLTPx2are not required for male fertility in maize.

3.5.The function of OsLTPg29,not ZmLTPg11,is conserved with TaMs1

To obtain direct evidence whether the biological functions of OsLTPg29,ZmLTPg11 and TaMs1 are conserved,we performed functional complementation oftams1gmutant usingOsLTPg29genomic DNA orZmLTPg11genomic DNA driven byTaMs1promoter respectively.The two plasmids,TaMs1p::gOsLTPg29andTaMs1p::gZmLTPg11(Fig.5A),were individually introduced into immature embryos of heterozygoustams1gplants.The results showed thatTaMs1p::gOsLTPg29completely complemented the abnormal spikes and pollen developmental phenotypes oftams1gmutant(3 out of 10 lines),whileTaMs1p::gZmLTPg11could not(12 lines)(Fig.5B),indicating thatOsLTPg29plays a similar role in pollen development toTaMs1,whileZmLTPg11does not.

4.Discussion

TaMs1gene was first reported in wheat,which encodes a nonspecific lipid transfer protein and is required for microgametogenesis[20,22].We identified one orthologOsLTPg29in rice and two orthologsZmLTPg11andZmLTPx2in maize.OsLTPg29andZmLTPg11are both expressed in sporogenous cells and meiotic cells during pollen development asTaMs1gene.To study the function ofOsLTPg29andZmLTPg11/ZmLTPx2,osltpg29mutants andzmltpg11zmltpx2double mutants were created by CRISPR/Cas9 technology.As expected,osltp29mutants displayed male sterility;surprisingly,zmltpg11zmltpx2double mutants did not.Accordingly,OsLTPg29could restore the male sterility oftams1mutant,whileZmLTPg11can not.MS1is a newly evolved gene in the family Poaceae which is one of the largest and the most widely distributed groups of flowering plants[20,30].The phylogenetic relationships of Poaceae have been classified into two major clades,BEP clade(Bambusoideae,Ehrhartoideae,and Pooideae)and PACCMAD clade(Panicoideae,Arundinoideae,Chloridoideae,Centothecoideae,Micrairoideae,Aristidoideae,and Danthonioideae).Rice belongs to the subfamily Ehrhartoideae and wheat is included in the subfamily Pooideae,both of which are members of BEP clade;while maize belongs to Panicoideae,which is a member of PACCMAD clade.Therefore,rice(Ehrhartoideae)is more closely related to wheat(Pooideae)than maize(Panicoideae)[29].In term of divergence time,Panicoideae(maize)diverged from Pooideae(wheat)and Ehrhartoideae(rice)60 million years ago,while Pooideae(wheat)and Ehrhartoideae(rice)diverged from each other 50 million years ago[31].So,it can be speculated that althoughMS1originated specifically in the family Poaceae,its biological function involved in pollen development may not be acquired until the divergence of Pooideae(wheat)and Ehrhartoideae(rice)from Panicoideae(maize).Alternatively,it is just a special case that ZmLTPg11 and ZmLTPx2 were not involved in pollen development,because the function of other MS1 orthologs from PACCMAD clade except maize was unknown.Thus,it would be interesting to clarify whetherMS1contribute to male fertility in Bambusoideae,another subfamily in BEP clade,and other subfamilies in PACCMAD clade except Panicoideae.We noticed with interest that both ZmLTPg11/ZmLTPx2 and OsLTPg29 showed similar expression pattern,lipid-binding activity and sequence identity with TaMs1,butOsLTPg29notZmLTPg11could restore the male sterility oftams1mutant.In fact,it wasn’t hard to understand.After all,the expression pattern was mainly determined by promoter,and the lipidbinding activity of nsLTPs was conferred by the 8 CM.Although the sequence identity between ZmLTPg11/ZmLTPx2 and TaMs1 and that between OsLTPg29 and TaMs1 were similar,the identity between OsLTPg29 and ZmLTPg11/ZmLTPx2 was only 47.67%–54.19%(Fig.1B).The difference of some amino acids probably resulted in different spatial structure of the protein involved in different biological function.In addition,mutation ofOsLTPg29lead to male sterility,andzmltpg11zmltpx2mutant display normal pollen development,suggesting that in the evolution of maize,other nsLTPs,even other LTPs,have replaced the function in pollen development ofZmLTPg11andZmLTPx2.Of course,it is also possible that the lipid composition of maize pollen is different from that of wheat and rice,leading to the fact that ZmLTPg11 and ZmLTPx2 are not necessary for pollen development.

All MS1 proteins in rice(OsLTPg29),wheat(TaMs1),and maize(ZmLTPg11)were proved to possess phospholipid-binding ability.Phospholipids,such as PA,PC,PE,and PI,are the primary components in the intracellular membrane system of pollen grains,as in most cells[3]Phospholipid molecules not only serve as the major structural component of membranes,but also function as lipid signaling to regulate multiple processes of higher plants[4].PA,the simplest membrane phospholipid and the key precursor for the synthesis of other membrane lipids and storage lipids,exists in small amounts in biological membranes[32,33].PA can bede novosynthesized by Gro3P acyltransferase and 1-acylGro3P acyltransferase,or be produced from major membrane phospholipid classes by phospholipase D(PLD)or from diacylglycerol(DAG)by diacylglycerol kinase(DGK)[34].PA is located in membrane system of cells,including plasma membrane,ER,mitochondrial membrane,chloroplasts and nuclear membrane,etc.[33,35].The distribution of intracellular PA is dynamically changed,which is mediated via vesicular transport,lipid transfer proteins or PAproducing enzymes[33].PIPs are phosphorylated derivatives of PI with all combinations of one,two or three phosphates in the 3,4 and 5 positions of the inositol ring.They are produced by specific PI-kinases such as PI 3-kinase(PI3K),PI 4-kinase(PI4K),PI3P 5-kinase(PIKfyve)and PI4P 5-kinase(PI4P5K),and are degraded by specific phosphatase such as PTEN,SAC,5′-phosphatase,and PIPLC[36,37].Different PIPs are distributed in different organelles,with PI(3)P on membrane of early endosomes,late endosomes and vacuoles,PI(4)P on Golgi,PI(4,5)P2at the plasma membrane,and PI(3,5)P2on late endocytic organelles,etc.[36,38].

Fig.5.Functional complementation of wheat ms1g mutant using OsLTPg29 or ZmLTPg11.(A)Diagrams for the constructs of gOsLTPg29 or gZmLTPg11 driven by TaMs1 promoter.(B)Spikes of wild type,ms1g and transgenic lines containing TaMs1p::gOsLTPg29 or TaMs1p::gZmLTPg11 in ms1g background.Scale bar,1 cm.(C)Floral organs after removal of the palea and lemma of wild type,ms1g and transgenic lines containing TaMs1p::gOsLTPg29 or TaMs1p::gZmLTPg11 in ms1g background.Scale bar,1 mm.(D)Mature pollen grains stained with 1%I2-KI from wild type,ms1g and transgenic lines containing TaMs1p::gOsLTPg29 or TaMs1p::gZmLTPg11 in the ms1g background.Scale bar,200 μm.

In recent years,there are many reports on the roles of PA and PIPs in pollen development,pollen germination,and pollen tube growth.For instance,AtDGK2andAtDGK4,two key enzymes in PA synthesis,are highly expressed in pollen,anddgk2dgk4double mutant impairs the male gametophyte[35].AtVPS34 forms a class III PI3K complex responsible for production of PI(3)P from PI with AtVPS15 and AtATG6/AtVPS30.Mutation of eitherAtVPS34orAtVPS15gene affects vacuolar organization of pollens,and pollen grains carryingatvps34,atvps15oratatg6are defective in pollen germination,resulting in reduced male transmission[39–43].AtFAB1A and AtFAB1B are two PI3P 5-kinases responsible for the production of PI(3,5)P2,and thefab1a-1/fab1b-1pollen grains display an abnormal vacuolar phenotype during pollen development[44].AtVAC14,encoding a scaffold protein of PI(3,5)P2-metabolizing complex,is expressed in all cell layers of anthers including developing microspores.Microspores ofatvac14mutant display vacuolar fission after pollen mitosis I,which results in pollen abortion[45].AtPTEN mediates the degradation of 3-phosphorylated PI monophosphates and PI bisphosphates[46]andAtPTEN1is expressed specifically in pollen grains and RNAi silencing ofAtPTEN1causes pollen cell death during the late stage of development[47].Soon after we submitted this study,EPAD1gene was reported,which is the same gene ofOsLTPg29.Similar to our result,EPAD1gene is specifically expressed in male meiocytes and EPAD1 protein has lipid-binding activity.Furthermore,EPAD1 localizes to the plasma membrane and is required for primexine integrity and pollen exine patterning in rice[48].Therefore,OsLTPg29 may contribute to pollen development,probably through regulating the dynamic distribution of PA and PIs in the intracellular membrane system or plasma membrane of pollen grains.

CRediT authorship contribution statement

Jian Li:Investigation,Visualization,Data curation,Writingoriginal draft.Zheng WangInvestigation,Visualization,Data curation,Writing-original draft.Zhenyi Chang:Investigation,Data curation.Xiaoyan Tang:Investigation,Data curation.Hang He:Software,validation.Ligeng Ma:Conceptualization,Funding acquisition,Writing-review&editing.Xing Wang Deng:Conceptualization,Funding acquisition,Writing-review & editing.

Declaration of competing interest

Authors declare that there are no conflicts of interest.

Acknowledgments

This work was supported by Peking University Institute of Advanced Agricultural Sciences,and Beijing Municipal Government Science Foundation(IDHT20170513).

Appendix A.Supplementary data

Supplementary data for this article can be found online at http://doi.org/10.1016/j.cj.2020.xx.xxx.