Yunsong Gu,Shichen Han,Lin Chen,Junyi Mu,Luning Duan,Yaxuan Li,Yueming Yan,Xiaohui Li

Key Laboratory of Genetics and Biotechnology, College of Life Science, Capital Normal University,Beijing 100048,China

Keywords:

ABSTRACT Reserve starch of cereal crop accounts for about 70% of grain endosperm and acts as an important human carbohydrate resource worldwide.Wheat reserve starch is synthesized by enzymatic machinery in endosperm cells.To identify genes involved in starch biosynthesis, we constructed 30 RNA-Seq libraries of 10 endosperm-development periods and performed expression and localization analyses.Of 166 endosperm-expressed homologs of starch biosynthesis-related genes, 74 showed expression correlated with reserve starch accumulation,including 26 with expected subcellular distribution and higher expression than their isoforms.The key proteins SUS3, UGP1, cAGPase, and Bt1-3 formed the main metabolic pathway and contributed the major substrates for starch processing in amyloplasts.Important isoforms, key pathway proteins, and the main carbon flux toward starch formation in the reserve starch biosynthesis pathway were identified.Based on a coexpression analysis, a library of 425 transcription factors was produced to screen for common regulators.TaMYB44 had features of transcription factors and bound to TaSUT1,TaSSIIIa, TaBEIIa, TaISA1, and TaBEIIb promoters in yeast, suggesting that the gene is a pathway regulator.This study sheds light on understanding the mechanism of reserve starch biosynthesis and will be helpful for increasing starch content in wheat endosperm via biotechnological strategies.

1.Introduction

Common wheat(Triticum aestivum L.,2n=6x=42,AABBDD)is one of three major global cereal crops and provides about 20%of the calories humans consume [1].As a predominant component, reserve starch accounts for 60%-75% of the wheat endosperm [2], and is composed of amylose and amylopectin in varying ratios among cultivars[3].

A research focus has been to explore the genetic mechanisms controlling the biosynthesis of reserve starch, and many starch biosynthesis-related genes have been identified in many plants [4-7], including ADP-glucose pyrophosphorylase (AGPase), starch synthase (SS), granule-bound starch synthase (GBSS), branching enzyme (BE), debranching enzyme (DBE), phosphorylase (PHO), and protein targeting to starch(PTST).

BRITTLE1 (Bt1)is targeted to the chloroplast and responsible for unidirectional transmembrane transport of ADPglucose [8].Suppressing TaBt1 in common wheat decreases kernel size, thousand-kernel weight, and total starch content[8].Knockdown of both BEIIa and BEIIb expression and knockout of BEIIa alone or with BEIIb causes an increase in amylose [9-12], whereas suppressing BEIIb alone fails to change amylose content [9].Three BE isoforms: BEI, BEIIa,and BEIIb,are phosphorylated in amyloplasts on Ser residues,and the activities of BEIIa and BEIIb are regulated by phosphorylation occurring at this site in amyloplasts [13].A missense mutation in the TaAGPL1-B gene (located on chromosome 1B) results in a marked decline in total starch and amylopectin contents, while highly expressed TaAGPL1 and TaAGPS1 increase starch content and kernel weight of common wheat [14-17].Knockout of all three GBSSI genes produces waxy wheat, in which the endosperm starch is composed almost exclusively of amylopectin[18,19].Interference with SSI expression in wheat kernels using RNAi leads to an increase in amylose content [20], and inactivation of SSIIa also increases the proportion of amylose [21-23].Mutated TaSSIVb-D results in a decline in the number of granules in leaves, but has no effect on reserve starch content in the endosperm[17,24].Grain starch and amylopectin contents are lower in agp.L-B1/ssIVb-D double mutants than in any of the single mutants[17].The recently reported PTSTs function as a recruiter of starch biosynthesis enzymes, and play a key role in starch granule initiation [25-27].Reduction in BGC1 (BGRANULE CONTENT 1)protein,an ortholog of PTST2 in wheat,results in a lack of B-type starch granules[28],while complete absence of BGC1 not only represses the initiation of B-type starch granules, but also severely disrupts A-type granule number and morphology[29].

Many studies have focused on the regulation of expression of single genes involved in starch biosynthesis.The MYC protein OsBP-5 binds specifically to the CAACGTG motif in the GBSSI promoter,and suppressing gene expression results in a decrease in amylose content of the mature seed of transgenic rice [30].Fu and Xue [31]showed that rice starch regulator 1(RSR1), a transcription factor (TF) from the APETALA2/ethylene-responsive element binding protein family, negatively regulated the expression of starch biosynthesis genes in RSR1 deficiency and overexpression assays.The OsbZIP58 regulates the expression of AGPL3,GBSSI,SSIIa,BEI,BEIIb,and ISA2 by binding directly to their promoters in rice [32].ZmbZIP91 shows a high expression correlation with eight key starch biosynthesis genes, and activates the AGPS1, SSI,and ISA1 promoters in maize leaf protoplasts and premature endosperm in a transient expression assay [33].The maize endosperm-specific TFs opaque2 (O2) and prolamine-box binding factor (PBF) regulate starch biosynthesis, as knockdown of PBF and knockout of O2 produce decreases of respectively 5% and 11% in starch content [34].ZmDof3 is expressed exclusively in the endosperm of the maize kernel,and knockdown of expression produces a defective kernel phenotype with reduced starch content and reduced expression levels of starch biosynthesis-related genes [35].ZmEREB156, which is localized in the nuclei of onion epidermal cells and possesses strong transcriptional activation activity, is a TF from the AP2-EREBP superfamily.This factor increases promoter activity of the Sh2 and SSIIIa genes when transiently co-transformed into maize endosperm [36].The ZmMYB14 TF shows an expression pattern similar to that of starch biosynthesis-related genes in maize kernels and activated the Bt1,Sh2,AGPS1,GBSSI,SSI,and BEI promoters in maize endosperm in transient gene overexpression assays[37].AGPL2,AGPS2,SSI,GBSSIIb,and BEI genes are upregulated in maize when ZmNAC36 is transiently overexpressed in developing endosperm [38].ZmNAC34 is highly expressed in maize endosperm and has features of a NAC TF.However,this factor is considered a negative regulator of starch biosynthesis, as its overexpression in rice results in a decrease in total starch accumulation and soluble sugar content [39].Another two NAC TFs, ZmNAC128 and ZmNAC130, are expressed specifically in maize endosperm, and suppressing their expression caused a reduction in starch with downregulated expression of AGPS1,SUS1,GBSSI,SSI,BEIIb,BEI,and PUL1[40].Starch biosynthesis is a complex metabolic pathway,and the identified TFs might cooperate with these unknown factors in controlling the synthesis of reserve starch in the endosperm of cereal crops.A co-expression analysis, based on the assumption that genes with similar expression patterns are likely to be functionally associated[31],has been successfully applied to investigate potential TFs of metabolic pathways[31,41-43].

Common wheat (AABBDD) is a hexaploid species with a genome of about 17 Gb.Many genes in common wheat are represented by three closely related homologs.The International Wheat Genome Sequencing Consortium (IWGSC) assembled a 14.5-Gb nearly completed genome and annotated 107,891 high-confidence genes [44].That study has created favorable conditions for elucidating an integrated starch biosynthesis pathway.The objectives of the present study were to establish a wheat starch biosynthesis pathway using analysis of 30 RNA-Seq libraries constructed from the developing endosperm of wheat and subcellular localization of major starch biosynthesis-related proteins, and to identify candidate regulators for the entire starch biosynthesis pathway using co-expression analysis.

2.Materials and methods

2.1.Plant materials and cultivation conditions

The spring wheat cultivar Chinese Spring(CS)was planted in the greenhouse of the Chinese Academy of Agricultural Sciences in Beijing.The culture conditions were 20 °C and a 16/8 h photoperiod at 15,000 lx (at 20 °C and 16/8 h photoperiod with supplementary lighting provided by highpressure sodium vapor lamps;Powertone SON-T AGRO 400 W;Philips Electronic, Leeds, UK).Seeds harvested individually from the central parts of the spikes of three plants at 5, 8, 11,14, 17, 20, 23, 26, 29, and 32 DPA (Fig.S1) were frozen immediately in liquid nitrogen and stored at ?80 °C.Embryos of the seeds were removed before RNA isolation.

2.2.RNA isolation, cDNA synthesis, and real-time quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted from developing endosperm tissues using the Trizol method [45].cDNA was synthesized with approximately 1.0 μg total RNA according to the manufacturer’s instructions (PrimeScript RT Reagent Kit, Takara,Japan).The primers listed in Table S1 were used for RT-qPCR analysis and were designed with the Primer3Plus online tool(http://www.primer3plus.com/cgi-bin/dev/primer3plus.cgi).A gene expression analysis was performed using GAPDH and autophagy related protein 8d (ATG8d) genes as internal references and the CFX96 RT-qPCR detection system (Bio-Rad Laboratories,Hercules,CA,USA)[46,47].

2.3.Library preparation and transcriptomic analysis

The quality of total RNA was examined using 1% agarose gel electrophoresis and an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA), and the concentration was determined with a Nanodrop2000 Spectrophotometer(Thermo Fisher Scientific,Wilmington,DE,USA).The total RNA was incubated with oligo (dT) attached magnetic beads.Purified mRNA was fragmented into ～200-nt pieces, which were reverse-transcribed to create cDNA libraries using a TruseqTM RNA sample prep Kit(Illumina,San Diego,CA,USA)following the manufacturer’s protocol.Sequencing was performed on an Illumina HiSeq 4000 platform (Illumina) by Beijing Novogene Bioinformatics Technology Co.Ltd.Each period was represented by three replications.

Sequencing of the libraries generated 53.69-75.49 million clean reads that were filtered from 55.28-77.79 million raw reads with Q20 scores of 94.79%-96.55% and Q30 scores of 88.07%-91.55%.The GC contents were 53.72%-61.75% (Table S2).The total number of clean reads mapped to the genome(IWGSC Refseq v1.0) with HISAT 2.0.4 software [48]with default parameters was 48.40-69.16 million per library;including 8.85 million multiple mapped reads and at least 45.12 million uniquely mapped reads.Pearson’s correlation calculated on whole-transcriptome data was used to assess the relationship strength between samples(Fig.S2).

The transcriptional levels of genes were evaluated by FPKM values (fragments per kilobase of exon per million fragments mapped) with HTSeq 0.6.1 [49].Using FPKM ＞1 as the threshold, transcripts of 52,717 genes from at least one of 10 stage databases were detected.Of these,44,966,40,546,37,913,36,121, 33,617, 31,370, 34,319, 31,663, 31,113, and 31,928 genes were expressed at the 5,8,11,14,17,20,23,26,29,and 32 DPA,respectively.Gene function was annotated using Blastx against Non-redundant (Nr) protein, Swissprot, and Kyoto Encyclopedia of Genes and Genomes(KEGG)databases with a threshold of E-value ＜0.00001.To identify and classify transcription factors, all the protein sequences of CS were analyzed using iTAK 1.2 [50].As a result, 12,221 transcripts marked as low-confidence in the reference annotation of IWGSC were identified with FPKM ＞1, of which 8756, 8126,7695, 6904, 6159, 5634, 5740, 4469, 5327, and 5590 appeared at the 5, 8, 11, 14, 17, 20, 23, 26, 29, and 32 DPA, respectively.A total of 3866 genes encoding TFs, including 232 lowconfidence TFs, were expressed in developing endosperm.The RNA-Seq data were deposited in the National Center for Biotechnology Information Sequence Read Archive (http://www.ncbi.nlm.nih.gov/sra/) under accession number PRJNA545291.

2.4.Phylogenetic analysis

Wheat candidates involved in starch biosynthesis were identified by Blastp alignment of homologous protein sequences derived from other model species against the CS database (http://202.194.139.32/).Wheat protein sequences that showed the highest identity with the query sequence were considered corresponding orthologs [51].ClustalW [52]was used to align the amino acid sequences of the starch biosynthesis-related proteins, and MEGA5 [53]was used to generate a neighbor-joining tree with 1000 replicates.

2.5.Subcellular localization

The coding sequences of the starch biosynthesis-related genes and TaMYB44-D were amplified by PCR from wheat cDNA and inserted into the 163hGFP vector(gift of Dr.Yimiao Tang, Beijing Academy of Agriculture and Forestry Sciences).The primers designed for PCR amplification are listed in Table S3.Arabidopsis protoplasts were isolated and transfected following Yoo et al.[54],and those of wheat protoplasts were treated using the method developed for barley [55].The 35S::TaMYB44-D-GFP plasmid was bombarded into onion epidermal cells at a helium pressure of 1100 psi using the PDS-1000/He system (Bio-Rad).Transformants were cultured on MS medium plates at 24 °C for 12 h in the dark.Fluorescent signals were observed using a Zeiss LSM 510 confocal laser microscope (Zeiss,Oberkochen,Germany).

2.6.Transcriptional activation assay

The complete and partial coding sequences of TaMYB44-D were cloned into the pGBKT7 vector with the primers listed in Table S4,and the generated pGBKT7-TaMYB44-D1-259, pGBKT7-TaMYB44-D1-69, pGBKT7-TaMYB44-D1-148, and pGBKT7-TaMYB44-D149-259were independently transfected into yeast strain AH109(Clontech,Palo Alto,CA,USA).The transformants were screened on SD/-Trp plates for three days at 30 °C.Three colonies were picked from the plates and propagated in 2.0-mL microtubes.The colonies were cultured on SD/-Trp/-His plates and those supplemented with X-α-gal at 28°C for three days were used to monitor the generation of blue color for transcriptional activation analysis.

2.7.Yeast one-hybrid assay

Fig.1– Phylogeny and expression analyses of 10 starch biosynthesis-related proteins in wheat.(a–j) SUT,SUS,UGP,AGPase,PGM,Bt1,PHO,SS,GBSS,and BE families.Neighbor-joining trees were constructed using amino acid sequences of starch biosynthesis-related proteins.Bootstrap values calculated for 1000 replicates were indicated at corresponding nodes.The maximum FPKM value from 10 periods was used to indicate the expression level of each starch biosynthesis-related gene.Hollow circles mean genes were undetectable(FPKM ＜1).Different colors were used to distinguish differentially expressed genes.Light gray,low level(1 ＜FPKM ＜10); dark gray,middle level(10 ＜FPKM ＜100);black,high level (100 ＜FPKM).

The TaMYB44-D coding sequence was cloned using cDNA prepared from 14 DPA endosperm samples of CS as the template, and then integrated into the pGADT7-Rec vector to construct prey plasmids.The promoter fragments amplified from CS gDNA were inserted into the pHIS2.1 vector with the Spe I and EcoR I sites to construct the bait plasmid.The Y187 yeast strain was co-transformed with pGADT7-Rec-TaMYB44-D and pHIS2.1-bait plasmids following the manufacturer’s instructions (Clontech).The resulting cells were plated on selective medium lacking leucine, tryptophan, and histidine(SD/-Leu/-Trp/-His) with an optimal concentration of 3-amino-1,2,4-triazole (3-AT).The primers are shown in Table S5.

2.8.Data analysis

Cluster 3.0 [56]was used to perform hierarchical clustering using the default parameters.The similarity metric was Euclidean distance, and hierarchical clustering employed the complete linkage method.A heatmap was constructed with TreeView [57].Pearson’s correlation values of the gene expression patterns were calculated with SPSS 19 (SPSS Inc.,Chicago,IL,USA)and a P＜0.01 was considered significant.

3.Results

3.1.Starch biosynthesis-related proteins in developing endosperm tissues of CS

Based on reported starch biosynthesis-related proteins in the model species rice, maize, and Arabidopsis, a phylogenetic analysis was performed to identify wheat starch biosynthesisrelated proteins, and 19 neighbor-joining trees (Figs.1, S3)were constructed.CS harbored 18 putative sucrose transporters (SUTs), which were clustered into five groups.Of the 11 detectable SUTs (FPKM ＞1), six SUTs from SUT1 and SUT2 groups showed a higher level of expression,with FPKM values＞10 in at least one stage dataset(Fig.1a).The SUT1 and SUT2 isoforms were localized in the cytomembrane of mesophyll cell protoplasts of wheat (Figs.2, S4).Nine putative alkaline/neutral invertases (INVs) were obtained from CS by Blastp using as queries INVs located in the cytosol of three model species [58,59].Of the five detectable INVs in wheat endosperm,three were expressed at a high level(10 ＜FPKM ＜100)(Fig.3a),and were localized in the cytosol(Fig.S4).Seven rice sucrose synthases (SUSs) were clustered into six groups, and the corresponding groups were also formed by 21 putative SUSs in wheat.There were marked expression differences among the 16 detectable SUSs.Transcriptional levels of three SUS3 members (TraesCS2A01G168200, TraesCS2B01G194200,and TraesCS2D01G175600)were 100-fold higher than those of SUS6s (TraesCS6D01G403800 and TraesCS6B01G466600) (Fig.1b).Interestingly, the three most highly expressed, SUS3,SUS2, and SUS4, were localized in the cytosol of wheat protoplasts (Figs.2, S4).Twenty-six putative hexokinases(HXKs) were found in CS, and most were expressed at a low level in the endosperm (Fig.S3b).Four detectable HXKs clustered with the three cytosol proteins OsHXK1, OsHXK7,and OsHXK8 [60], which were also localized in the cytosol of the wheat protoplast (Fig.S4).Five putative UDP-glucose pyrophosphorylases (UGPs) encoded in wheat were assigned to two clusters with the reported proteins[61].However,only the UGP1 members showed detectable expression in developing wheat endosperm, among which TraesCS5A01G353700 reached the highest level with FPKM ＞100 (Fig.1c).UGP1 was localized in the protoplast cytosol (Fig.2).In contrast to the two clusters in rice and maize [62], 18 putative fructokinases(FRKs) were clearly divided into three clusters, including 12 members detectable in wheat endosperm (Fig.S3c).The cytosolic distributions of TraesCS5A01G286200 (FRK1) and TraesCS7D01G315500 (FRK2) were confirmed by subcellular localization (Fig.S4).All 11 putative AGPase genes were expressed in wheat endosperm and were postulated to encode four subunits that assemble into cytoplasmic and plastidial heterotetramer enzymes according to the previous research results[63].The large TaAGPL1 subunit and the small TaAGPS1 subunit were localized in the cytosol,while TaAGPL2 and TaAGPS2 were in the chloroplast (Figs.2, S4), suggesting that the former two subunits formed the cytoplasmic heterotetramer, and the latter two formed the plastidial heterotetramer.Interestingly, transcriptional levels of the cytoplasmic subunits were about 10- to 100-fold higher than those of plastidial subunits (Fig.1d), and the proportion of cytosolic activity was 93%in wheat endosperm[64],indicating that reactions converting glucose-1-phosphate (glucose-1-P)and ATP to ADP-glucose during wheat endosperm development were controlled predominantly by the cytoplasmic TaAGPL1/S1 complex.Wheat endosperm expressed three cytosolic glucose-6-phosphate isomerase (PGI) genes (Figs.S3d, S4), which were possibly responsible for reversible isomerization between glucose-6 phosphate (glucose-6-P)and fructose-6 phosphate (fructose-6-P) in the cytosol.Only three cytosolic phosphoglucomutases (PGMs) were found in the CS database by Blastp being performed to query the plastidial and cytosolic PGMs, because they were clustered into the cytosolic group and located in the cytosol of the wheat protoplast (Figs.1e, 2).Wheat PGMs were expressed continuously during development of the endosperm.The results indicate that the interconversion of glucose-6-P and glucose-1-P in the cytosol was catalyzed by cytosolic PGM,while the component that catalyzed the counterpart reaction in the plastids remained unclear.In contrast to the single glucose-6-phosphate/phosphate translocator(GPT)isoform in rice, maize, and Arabidopsis, common wheat contained six GPTs and each was expressed at a similar level in developing endosperm(Fig.S3e).Wheat GPT1 and GPT2 were localized in the membrane of wheat chloroplasts(Fig.S4).Eight Bt1s were found in CS, and were classified into three distinct groups.The Bt1-3 group, including TraesCS6B01G210000, TraesCS6A01G175100,and TraesCS6D01G168200, showed an average 28-fold higher expression than the Bt1-1 members(Fig.1f).TaBT1-3-GFP fusion protein fluorescence exclusively targets the chloroplast of the wheat protoplast [8].High-level expression and chloroplast localization showed that the three Bt1-3s played a key role in transporting ADP-glucose into wheat plastids.

Fig.2–Subcellular localization of 11 starch biosynthesis-related genes in the wheat protoplast.All genes were inserted in front of hGFP in 163hGFP vectors.

Six candidates for disproportionating enzyme (DPE) genes were detectable at the transcriptional level in developing endosperm.These putative wheat DPEs and their homologs in the other three species were categorized into the DPE1 and DPE2 groups (Fig.S3f).Generally, DPE1 transfers two glucosyl units from maltose to glucan in the plastid,while DPE2 functions in the cytosol [65,66].The wheat DPE1 and DPE2 were also shown to accumulate in respectively the plastid and cytosol of the wheat protoplast(Fig.S4).Nine putative wheat PHOs were classified into the PHOH and PHOL groups.Among the six PHOH members,only three were detectable at the transcriptional level but at a low level, while all PHOLs showed relatively high expression in developing wheat endosperm (Fig.1g).The wheat PHOH was localized in the cytosol and PHOL was localized in the plastid(Figs.2, S4), in agreement with the subcellular distribution of reported PHOs [67].These results suggest that PHOL proteins encoded by TraesCS5A01G395200, TraesCS5B01G400000, and TraesCS5D01G404500 might form a complex with DPE1 for use of a broader range of substrates to increase the synthesis of larger maltooligosaccharides in the plastid as their homologous OsPho1-OsDpe1 acts in rice[68].Common wheat had six putative GBSSs, clustered into the GBSS I and II groups.GBSS II is responsible for transitory starch biosynthesis in chloroplasts[69,70]and its wheat homologs were expressed mainly at the early endosperm stage.Transcriptional levels of GBSS I members were on average 10- to 100-fold higher than those of GBSS II in developing endosperm(Fig.1i).GBSS I was localized in the plastid of the wheat protoplast (Fig.2).A total of 21 putative SSs were found in common wheat, including four groups: SSI, SSII, SSIII,and SSIV.SSII had all three isoforms SSIIa, SSIIb, and SSIIc, and SSIII had all two isoforms SSIIIa and SSIIIb,but SSIV had only one isoform SSIVb.The SSI and SSIIa members was on average 10-fold more highly expressed than the other SS members, but several-fold lower than GBSSI (Fig.1h).The distribution of SSI,SSIIa, SSIIb, SSIIc, SSIIIa, and SSIIIb in the plastid was demonstrated by subcellular localization analysis(Figs.2,S4).In view of the decrease in the short and intermediate chains of rice amylopectin resulting from the low transcriptional level of OsSSI and OsSSIIa [71,72], TaSSI and TaSSIIa members expressed at a high level and accumulated in plastids are suggested to play an important role in wheat amylopectin synthesis.Twelve putative BEs were encoded in common wheat,including BEI,BEIIa,BEIIb,and BEIII, whereas no BEIII members were detected at the transcriptional level in the endosperm (Fig.1j).The expressed BEIIa and BEIIb were localized to the plastid of the wheat protoplast(Figs.2,S4).Common wheat had two groups of DBEs,including three isoamylase(ISA)isoforms and pullulanase(PUL),which were expressed at low levels in developing endosperm.The ISA1 isoform,which showed the highest level of expression among ISAs(Fig.S3g),and PUL were localized in the plastid of the protoplast (Fig.S4).A Blastp search against the CS Refseq v1.0 data using PTSTs from rice, maize, and Arabidopsis, six putative PTSTs were obtained, including only the PTST1 and PTST2 isoforms.The PTST1 and PTST2 members showed low expression levels in the developing endosperm, although the former was expressed slightly more highly than the latter (Fig.S3i).PTST1-GFP fluorescence was visible in the plastid of the protoplast(Fig.S4).

Fig.3–Expression pattern of 89 starch biosynthesis-related genes.The expression level of a gene at a developmental period is indicated by the average FPKM value of triplicates.Cluster 3.0 was used to construct gene expression pattern based on RNASeq data of 10 periods and perform cluster analysis.Here are 10 periods: 5, 8,11, 14, 17,20, 23,26, 29, and 32 DPA.

3.2.Expression patterns of starch biosynthesis-related genes in developing endosperm of CS

RT-qPCR showed that the expression pattern of each gene was consistent with those based on RNA-Seq expression (Fig.S5).The RNA-Seq expression patterns of 89 starch biosynthesisrelated genes selected from 18 gene families having the characteristics of high-level expression in developing endosperm and expected subcellular localization in wheat protoplasts were clustered(Fig.3).These genes were classified into two groups.The group I members were expressed mainly from 8 to 20 DPA during endosperm development, corresponding to the important stage of reserve starch biosynthesis [73], while the group II members, including three HXKs,three SSIIcs, three SSIIIbs, and two of three SSIIbs, PGIs, and GPT1s, were expressed mainly at the early stage(5 DPA) and/or late stage (23-32 DPA).The expression levels of the three HXKs were markedly higher at the late stage than those of the others.Of the three SSIIbs, two showed higher expression levels at early and late stages, while the other was expressed at a higher level only at the early stage.The genes that were strongly expressed during the early stage of endosperm development are considered [73]to be involved in construction of fundamental cell machinery and initiation of starch granules.In contrast,the other 74 genes from 17 gene families showed similar expression patterns,expressed at particularly high levels during 8-20 DPA, suggesting that the group I members are essential for biosynthesis of reserve starch in developing endosperm.

3.3.Potential regulators of starch biosynthesis-related genes revealed by co-expression analysis

A total of 3517 putative TFs that could be detected in at least one library with FPKM values ＞1 were clustered with the 89 starch biosynthesis-related genes.Two to six TFs whose expression showed a significant correlation with that of the target gene were selected from the subgroup of each target gene.This procedure yielded 425 TFs involved in expression regulation of 89 starch biosynthesis-related genes (Table S6).These factors were classified into 36 TF families,of which the six largest were MYB,NAC,AP2-EREBP,bZIP,CCAAT,and C3H(Fig.S6).Interestingly, the expression pattern of MYB44 was highly correlated with those of multiple starch biosynthesisrelated genes,including SUT1,SUS3,AGPL1,AGPS1,PGM,Bt1-3, GBSSI, SSI, SSIIIa, BEIIa, BEIIb, ISA1, PUL, PHOL, and PTST1(Fig.S7,Table S6).MYB44 expression was similar to that of the starch biosynthesis-related genes, increasing sharply and later decreasing gradually between 8 and 23 DPA, suggesting that MYB44 is involved in the regulation of starch biosynthesis-related genes.

3.4.TaMYB44-D might be a regulator of starch biosynthesisrelated genes

Sequence alignment (Fig.S8) showed that MYB44 was highly conserved in Triticeae and belonged to the R2R3 subfamily on the basis of containing two tandem DNA-binding domains at the Nterminal region (Fig.S8).TaMYB44-D was highly expressed in wheat kernels,particularly in developing endosperm(Fig.4).

The transient expression of TaMYB44-D(TraesCS4D01G297900) in onion epidermal cells and Arabidopsis and wheat mesophyll protoplasts indicated that TaMYB44-D could be transported into nuclei (Fig.5).To ascertain whether TaMYB44-D had potential transcriptional activation ability, the complete coding sequence as well as the N-terminal and Cterminal coding regions of TaMYB44-D were fused to the GAL4 DNA binding domain in the pGBKT7 vector, and the resulting plasmids were transformed into yeast strain AH109.The results showed that yeast cells harboring the pGBKT7-TaMYB44-D1-259,pGBKT7-TaMYB44-D149-259or pGBKT7-GAL4-53 plasmid grew well on SD/-Trp/-His selection media and turned blue when treated with X-α-gal (Fig.6), indicating that TaMYB44-D is a transcriptional activator, and that its transcriptional activation domain was located in the C-terminal region.

To investigate the binding of TaMYB44-D to starch biosynthesis-related genes, the promoter sequences of 12 selected genes were cloned in front of the reporter gene HIS3 to construct the pHIS2.1-Pro plasmids.The plasmids pHIS2.1-Pro and pGADT7-Sec2-TaMYB44-D were co-transfected into the Y187 yeast strain to perform the Y1H assay.As shown in Fig.7,yeast cells that were transformed with pGADT7-Sec2-TaMYB44-D and any of the pHIS2.1, such as pHIS2.1-TaSUS1Pro, pHIS2.1-TaAGPL1Pro, pHIS2.1-TaBt1-3Pro, pHIS2.1-TaGBSSIPro, pHIS2.1-TaSSIPro, pHIS2.1-TaPULPro, and pHIS2.1-TaPTST1Pro did not grow well, while those carrying pGADT7-Sec2-TaMYB44-D and pHIS2.1-TaSUT1Pro, pHIS2.1-TaSSIIIaPro, pHIS2.1-TaBEIIaPro,pHIS2.1-TaISA1Pro,or pHIS2.1-TaPHOLPro survived on SD/-Lea/-Trp/3-AT.Thus,TaMYB44-D was able bind the TaSUT1,TaPHOL,TaSSIIIa,TaBEIIa,and TaISA1 promoters in yeast cells.

4.Discussion

4.1.Starch biosynthesis pathway in wheat endosperm cells

Using the RNA sequencing and subcellular localization results, we propose a scheme of starch biosynthesis processing during development of the wheat endosperm,including 26 starch biosynthesis-related proteins(Fig.8).

It is generally accepted that sucrose imported by SUTs is converted to fructose and glucose under the control of INVs in the cytosol.Although cytoplasmic INVs were expressed in the developing wheat endosperm, the expression patterns were different from those of most of the starch biosynthesisrelated genes and contrasted with the accumulation pattern of reserve starch for their high level of expression appearing at 5 DPA and the low-level of expression maintained from 8 to 32 DPA.In contrast, SUSs were expressed at high levels and the expected expression pattern corresponded with the accumulation of reserve starch.These findings indicate that sucrose conversion reactions were catalyzed mainly by SUSs rather than cINVs.Fructose,a hydrolysis product of sucrose,is subsequently phosphorylated by FRKs.Of the 12 detectable FRKs, three FRK2 members (TraesCS7D02G315500,TraesCS7B01G219800, and TraesCS7A01G319000) were highly expressed from 8 to 23 DPA, in accord with the accumulation pattern of reserve starch in developing endosperm[73].There were many detectable catalysts in the glycolytic metabolic pathway, including 20 putative phosphofructokinases (PFKs)and 19 aldolases (ALDs) (Table S7).Fructose-bisphosphate aldolase, which plays a key role in glycolysis and gluconeogenesis, is responsible for reversible cleavage of fructose-1,6-biophosphate into glyceraldehyde-3-phosphate and dihydroxyacetone phosphate.In this study, TraesCS3B01G422900(ALD), which reached the highest expression level and reached a peak at 8 DPA, was localized in the cytosol of the wheat protoplast (Fig.S9b).Thus, fructose-6-P would be converted to glucose-6-P by PGIs to maintain carbon flux toward starch formation, or fructose-1,6-biophosphate by PFKs for respiration during endosperm development in wheat.The cytoplasmic AGPase subunits were expressed 10-to 100-fold higher than plastidial subunits at the transcriptional level.GPT was highly expressed at 5 DPA and then stably expressed at a low level in other periods of endosperm development, a finding that is inconsistent with the high accumulation of starch during the 8-23 DPA stage.Furthermore, the sequence similarity search revealed that only cytosolic PGMs were found in common wheat, durum wheat, wild emmer wheat, Triticum urartu, Aegilops tauschii,and barley (Figs.2, S10).Thus, the low-level expression of plastidial AGPases,an unexpected expression pattern of GPTs,and the absence of plastidial PGMs suggest that conversion of glucose-6-P into ADP-glucose by the GPT-pPGM-pAGPase pathway was inferior to that of glucose-1-P into ADP-glucose by the cAGPase-Bt1 pathway, and glucose-6-P was converted to glucose-1-P in the cytosol mainly by cytoplasmic PGMs in the wheat endosperm.The hypothesis is further supported by the following experimental findings in previous reports: 1)conversion of glucose-1-P into ADP-glucose is catalyzed by AGPases in the cytosol of the wheat endosperm, as the cytosolic form accounts for more than 93% of total AGPase activity in the developing endosperm[64];2)a single missense mutation of cAGPL1 markedly inhibits starch biosynthesis, and highly expressed cAGPL1 and cAGPS1 increase the starch content of wheat [14,15,17]; 3) kernel size, thousand kernel weight, and grain total starch content were decreased in TaBt1-knocked down common wheat [8].Glucose-6-P could also be converted into 6-phosphoglucono-δ-lactone by glucose-6-phosphate dehydrogenases(G6PDs)for respiration,as suggested by the presence of 12 detectable G6PDs in the developing endosperm and the cytosolic distribution of the highly expressed TraesCS2B01G362700 in the wheat protoplast(Fig.S11).

Fig.4–Expression analysis of TaMYB44-D in CS.(a)Nine tissues.Except for the young spike,the tissues were collected from 14-DPA plants.(b),Nine endosperm development stages.GAPDH and ATG8d served as respective RT-qPCR controls(a)and(b)for normalization of the gene expression analyses.

Fig.5– Subcellular localization analysis of TaMYB44-D.(a) A schematic of the plasmid construct used for protoplast transformation;(b) onion epidermal cells;(c)Arabidopsis leaf protoplasts;(d) wheat leaf protoplasts.

Starch initiation is complex and still unexplained.The absence of maltotriose (G3) and maltotetraose (G4) in barley ptst1 mutants [27]suggests that PTST1 is involved in G3 processing in cereal endosperm.In rice endosperm,the PHO1-DPE1 complex uses a broad range of substrates,mainly G3,to synthesize larger maltooligosaccharides [68].TaPHOL and TaDPE1 were co-expressed in the developing endosperm with a similar expression pattern (Fig.3), in accord with the collaboration of PHO1 with DPE1 in starch initiation.PTST2 is essential for starch initiation in Arabidopsis leaves, and transfers selective maltooligosaccharide primers to SSIV [26].However,the homologs of PTST2 and SSIV in wheat displayed different expression patterns; the expression peak period of the former corresponded to the expression valley of the latter.It is possible that other isoforms of SS than SSIV were recruited to the maltooligosaccharide primers in the wheat endosperm.

Starch biosynthesis enzymes including GBSS, SS, BE, and DBE are involved in the elongation process of amylose and amylopectin.GBSS I is known [18,19,27]to be the key protein responsible for elongating amylose polymers after being recruited by PTST1 in endosperm, while GBSS II is required for transitory starch biosynthesis in chloroplasts [69,70].In our RNA-Seq data, GBSS I had one of the highest transcript abundances among all expressed starch biosynthesis-related genes,second only to AGPL1.In contrast,the expression level of GBSS II was too low to be detected at most developmental stages.Furthermore, GBSS I is localized in the plastid.These findings are consistent with the key role of GBSS I in amylopectin biosynthesis.Among the seven isoforms of SS,SSI and SSIIa, which showed high-level expression and plasmid localization, should have critical influences on amylopectin biosynthesis, as their rice orthologs determine amylopectin structure [71,72].The ISA1 was more important than the other isoforms owing to its high-level expression and rice ortholog interacting with FLO6(PTST2)in vitro[74].Except for BEIII, the BEs expressed at high levels in developing endosperm were considered key factors.

In summary sucrose is converted mainly into fructose and UDP-glucose with UDP under catalysis by SUS3s.Fructose flows to the starch synthesis pathway and is metabolized into glucose-1-P via the intermediate product glucose-6-P.The UDP-glucose product is directly converted into glucose-1-P by UGP1.Derivatives of glucose-1-P are converted by the AGPL1/S1 complex into ADP-glucose in the cytosol and this product is transported into the amyloplastid by Bt1-3.DPE1 and PHOL use short maltooligosaccharides (e.g., G3 and G5) as substrates to synthesize longer maltooligosaccharides during starch initiation.PTST2 recognizes a long maltooligosaccharide via its carbohydratebinding module(CBM48),and recruits SS isoforms for elongation.Large maltooligosaccharides are further processed into complex amylose and amylopectin molecules by starch biosynthesis enzymes,particularly GBSSI,SSI,SSIIa,BEI,BEIIa,and BEIIb.

Fig.6– Transcriptional activity analysis of TaMYB44-4D in the yeast AH109 strain.(a) The schematic diagram indicates the structure of the plastid.Gray frames represent DNA binding domains.(b) pGBKT7-GAL4 and pGBDKT7-53 were used as negative and positive controls,respectively.Transformants were screened on SD/-Trp media and an activation analysis was performed on SD/-Trp/-His medium and those supplemented with X-α-gal indicators.

4.2.Potential regulators of starch biosynthesis-related genes

Major starch biosynthesis-related genes including all of those highlighted in Fig.8 showed similar expression patterns, and maintained a particularly high level of expression during 8-20 DPA, in agreement with the high rate of endosperm starch accumulation [73].This finding suggests that these genes are regulated by common TFs.In fact, multiple starch biosynthesis-related genes are co-regulated by one or several common TFs[17,31-39].

These TFs,including OsBP-5,OsbZIP58,ZmbZIP91,O2,PBF,ZmDof3, ZmEREB156, ZmMYB14, ZmNAC36, ZmNAC128, and ZmNAC130 have been identified as positive regulatory factors of starch biosynthesis-related genes in rice or maize[17,30,32-38].A Blastp search of ZmbZIP91 against the CS Refseq v1.0 database, we obtained as the most similar wheat genes: TraesCS7A01G268700, TraesCS7D01G269300,TraesCS5A01G308400, TraesCS5B01G308800, and TraesCS5D01G315400.TraesCS7A01G268700 was ranked at the top of six TFs that potentially regulated TaAGPL2 and TaPHOL genes based on their expression similarity, and TraesCS7D01G269300 was considered a potential regulator of TaSSIIa and TaBEIIa(Table S6).However,the other three genes were undetectable at the transcriptional level owing to their low (＜1) FPKM values.Similarly, expression of three common O2 wheat homologs, named SPA [75], showed a high correlation with those of TaAGPS1, TaSSI, TaSSIIIa, TaBEIIb, TaISA1,and TaPUL, in general agreement with the findings of a previous study [43]; two wheat homologs of PBF and Dof-3 were co-expressed with TaAGPL1, TaAGPS1, TaGBSSI, TaSSI,TaSSIIIa, TaPUL, and TaPTST1, and three wheat homologs of ZmMYB14 were similar to TaSUT1, TaSSIIIa, and TaISA1 on the expression pattern (Table S6).Based on the Blastp results, TraesCS7A01G569300, TraesCS7D01G154200,TraesCS7A01G569100, TraesCS7D01G543500,TraesCS7A01G152500, and TraesCS7D01G543400 were assigned as wheat homologs of ZmNAC34, ZmNAC36,ZmNAC128, and ZmNAC130.Among the six genes,TraesCS7D01G154200, TraesCS7A01G569100,TraesCS7D01G543500, and TraesCS7A01G152500 appeared in the candidate TF lists of TaBEI and TaPTST2.These TFs were identified mainly in the MYB, NAC, bZIP, AP2-EREBP,and ABI3VP1 families, which were the largest families involved in the expression regulation of starch biosynthesis-related genes in this study (Fig.S6).Interestingly, TaMYB44 expression was correlated with those of multiple starch biosynthesis-related genes, and had features of a MYB transcription factor.It was speculated to regulate the expression of TaSUT1, TaPHOL, TaSSIIIa,TaBEIIa, and TaISA1 in developing wheat endosperm.These proteins are distributed in the starch biosynthesis pathway, including sucrose intake, starch initiation, and amylopectin synthesis processes,suggesting that TaMYB44 acts as a common regulator of starch biosynthesis pathway and preferentially promotes the biosynthesis of amylopectin in wheat.In maize, regulation by ZmMYB14 of ZmGBSSI and ZmSSI has been shown by transient expression and Y1H assays [37], whereas TaMYB44 did not activate the promoters of TaGBSSI and TaSSI in yeast.In contrast, the promoter of ZmSSIIIa is not recognized by ZmMYB14 in transient expression or Y1H assays[37],but its counterpart in wheat was bound by TaMYB44 in yeast.These results suggest that the MYB family contribute multiple regulators to starch biosynthesis and that different members are responsible for regulating different target genes.In wheat endosperm, gluten proteins and storage starch account for most of the dry weight of wheat endosperm.A R2R3 subfamily member TaGAMyb, carrying a typical DNA binding domain and two transcriptional activation domains, plays a dual role in the regulation of glutenin gene expression by directly binding to their promoters and by recruiting histone acetyltransferase TaGCN5 [76].Because of much difference in protein structure compared with TaGAMyb,whether TaMYB44 could regulate the expression of gluten genes awaits investigation.In summary, 425 TFs containing wheat homologs of reported TFs that positively regulate the expression of starch biosynthesis-related genes in rice and maize were putatively associated with wheat starch biosynthesis.

Fig.7–Yeast one-hybrid assay showing association between TaMYB44-D and promoters of starch biosynthesis-related genes.Promoter sequences were cloned from 12 genes,including TaSUS3,TaSUT1,TaAGPL1, TaBt1-3,TaGBSSI,TaSSI,TaSSIIIa,TaBEIIa,TaISA1,TaPHOL,TaPUL,and TaPTST1.The 180 mmol L?1 3-AT was used to inhibit low levels of HIS3 expressed in the absence of TaMYB44-D.

Fig.8– A starch biosynthesis model for wheat endosperm.Proteins with high-levels of expression and main flux are highlighted in bold.AGPL/S,ADP-glucose pyrophosphorylase large/small subunit;ALD,aldolase;BE, branching enzyme;Bt,brittle;DPE,disproportionating enzyme;FRK,fructokinase;G3, maltotriose;GBSS,granule bound starch synthase;G6PD,glucose-6-phosphate dehydrogenase;PFK,phosphofructokinase;PGM,phosphoglucomutase;GPT,glucose-6-phosphate/phosphate translocator; HXK,hexokinase;INV,invertase;ISA,isoamylase; PFK,phosphofructokinase; PGI,glucose-6-phosphate isomerase;PHO,phosphorylase;PTST,protein targeting to starch;PUL,pullulanase;SS, soluble starch synthase;SUS,sucrose synthase;SUT,sucrose transporter;UGP,UDP-glucose pyrophosphorylase.* is used to mark the targets of transcription factor TaMYB44.

Expressions of most biosynthesis-related genes in the starch metabolic pathway were correlated, suggesting their common regulation.In pools of TFs whose expression was highly correlated with that of single genes, TFs with a high frequency in the pool are most likely to be common regulators of the starch biosynthesis pathway.Their identification in the 425-TF library suggests the use of this strategy for identifying regulators of the pathway.

5.Conclusions

There are 200 homologs of starch biosynthesis-related genes in the genome of CS common wheat.Of 166 endosperm expression genes, 89 genes from 18 families, particularly the 74 genes closely associated with the accumulation of reserve starch,are suggested to determine reserve starch biosynthesis in developing endosperm.The key proteins SUS3, UGP1,cAGPase, and Bt1 form the main metabolic pathway and contribute major substrates for starch processing in amyloplasts.TaMYB44 activates the TaSUT1,TaSSIIIa,TaBEIIa,TaISA1,and TaBEIIb promoters in yeast cells,and may act as a common regulator of starch biosynthesis-related genes.The findings shed light on the function of starch biosynthesisrelated proteins and the processing of reserve starch in developing wheat endosperm.

CRediT authorship contribution statement

Xiaohui Li conceived and designed the research.Yunsong Gu conducted RNA-Seq libraries.Shichen Han performed TF analysis.Lin Chen, Junyi Mu, and Luning Duan contributed to subcellular localization.Yaxuan Li and Yueming Yan provided technical assistance and scientific discussion.Yunsong Gu and Xiaohui Li wrote the manuscript.All the authors read and approved the manuscript.

Declaration of competing interest

Authors declare that there are no conflicts of interest.

Acknowledgments

This research was supported the National Program on R&D of Transgenic Plants (2016ZX08009003-004), National Natural Science Foundation of China (31571652), and the Youth Innovative Research Team of Capital Normal University.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.08.002.