Impaired SWEET-mediated sugar transportation impacts starch metabolism in developing rice seeds

Pei Li, Lihan Wang, Hongbo Liu, Meng Yuan

National Key Laboratory of Crop Genetic Improvement, National Center of Plant Gene Research (Wuhan), Huazhong Agricultural University, Wuhan 430070, Hubei, China

Keywords:Oryza sativa Sugar transporter SWEET Sugar Starch

ABSTRACT Sugar transportation and sugar-to-starch metabolism are considered important processes in seed development and embryo viability.A few plant SWEET proteins acting as sugar transporters have been reported to function in inflorescence and/or seed development.Here, we identified seven members of the 21 OsSWEET genes in rice that play essential roles in sugar transportation and sugar-to-starch conversion in seed development.Nineteen OsSWEET genes exhibiting different expression patterns during inflorescence and seed development were knocked out individually by CRISPR/Cas9.One third of the mutants showed decreased fertile pollen viability and shriveled mature caryopses, resulting in weakened seed traits.Grain fill-related genes but not representative grain shape-regulating genes showed attenuated expression in the mutants.Seed of each of these mutants accumulated more sucrose,glucose or fructose but less starch.Among all OsSWEET genes, OsSWEET4 and OsSWEET11 had major effects on caryopsis development.The sugar-to-starch metabolic pathway was significantly altered in ossweet11 mutants based on differential expression analysis in RNA sequencing assays, confirming that OsSWEET11 functions as a sugar transporter with a key role in seed development.These results help to decipher the multiple functions of OsSWEET genes and to show how they might be used in genetic improvement of rice.

1.Introduction

SWEET proteins are major transporters mediating sugar flux across cellular membranes in both prokaryotes and eukaryotes.Large numbers of SWEET proteins are present in all organisms ranging from bacteria and archaea to flora and fauna but with different members in each species.The number ofSWEETgenes within a species is not associated with its evolutionary complexity.Plants usually have moreSWEETgenes with at least seven members, even 59 members in wheat for example [1,2],whereas fauna such as mammals have only one [3].The larger number of members is suggestive of more diverse roles of SWEET proteins in plants than in animals.Gene duplication and fusion are considered the major driving forces during the evolution of plant SWEET proteins,whereas the key residues determining their function in substrate recognition and transport are highly conserved [1].

PlantSWEETgenes are reportedly involved in a range of biological activities throughout the entire growth cycle,including plantmicrobe interactions, phloem transport,nectar secretion,developing pollen nutrition, abiotic stress tolerance, seed germination,plant senescence, and plant hormone gibberellin (GA) transportation [3-5].A key feature of SWEET proteins is their function as transporters of sugars, including sucrose, glucose and fructose,and plant hormone GA.Most plant SWEET proteins transport one or several types of sugars, but currently only AtSWEET13, AtSWEET14, and OsSWEET3a are known to transport GA [6,7].

The riceSWEETgene family consists of 21 paralogs.The physiological functions of only eight paralogs,OsSWEET2b,OsSWEET3a,OsSWEET4,OsSWEET5,OsSWEET11,OsSWEET13,OsSWEET14andOsSWEET15, are partially known, whereas the roles of the other 13 paralogs are unclear[3,8].Among the eight functionally characterizedSWEETgenes in rice,some have been validated to transport sugars and to have roles in inflorescence and/or seed development.Suppressed or knock out plant mutants ofOsSWEET4,OsSWEET11,orOsSWEET14were sterile or showed reduced fertility with delayed reproductive development, reduced seed size, and shriveled caryopses [9-12].OsSWEET11plays an important role in sucrose transportation from maternal tissue to the maternal-filial interface during early caryopsis development.Knock out ofOsSWEET11led to decreased sucrose concentration in embryo sacs but increased sucrose content and attenuated starch levels in mature caryopses [12].OsSWEET15had a similar function toOsSWEET11in control of sucrose efflux across the nucellar epidermis/aleurone interface, andossweet11;15double knockout mutants accumulated starch in the pericarp but had less starch in the endosperm, causing shriveled caryopses [12,13].Similarly,OsSWEET4 functions as a glucose and fructose transporter, withossweet4knockout mutants exhibiting empty pericarp [14].These three rice SWEET proteins are involved in sugar transportation in developing caryopses, facilitating sugar-to-starch metabolism and are essential for fully filled mature seeds.It is currently unclear as to whether otherSWEETgenes in rice have similar roles in inflorescence and/or seed development.

Sucrose is produced by photosynthesis of plants in photosynthetically active tissues and is the primary source of energy for plant growth and development [15].After synthesis in source tissues, sucrose is actively loaded into the phloem and transported to sink organs, where it is either quickly hydrolyzed to glucose and fructose,or converted into relatively inert storage compounds like starch.After being transported to developing caryopses in rice,sucrose is hydrolyzed to glucose and fructose in the cytosol of endosperm cells followed by transformation into reserve starch in the amyloplasts.Reserve starch stored in endosperm of mature caryopses is the major carbohydrate providing energy for seed germination and seedling development, human and animal nutrition and various industrial applications [16].The processes enabling transportation of sucrose to the developing caryopses, hydrolyzation of sucrose to glucose and fructose,and then their transformation to reserve starch involve a group of catalytic pathways controlled by many genes.Disruption of any one of these genes can potentially disrupt the entire process,causing abnormal starch deposition and compromised embryo viability [16].Since inadequate sugar accumulation in developing caryopses could significantly alter sugar-to-starch conversion, and excess sugars in mature caryopses could markedly inhibit seed germination and seedling growth.Therefore, the concentration of sugars must be precisely controlled in developing and mature caryopses [4,16].

To uncover the potential roles of otherSWEETgenes in inflorescence and/or seed development we adopted a CRISPR/Cas9 gene knockout strategy to delete 19SWEETgenes that were expressed in the inflorescence and seeds of rice.We systematically evaluated their expression patterns during inflorescence and seed development, assessed the pollen viability, seed traits, and sugar and starch concentrations in each knockout mutant.The results indicated thatSWEETgenes have diversified expression patterns during various stages of inflorescence and seed development and have different effects on sugar-to-starch conversion due to their different sugars affinities.

2.Materials and methods

2.1.Generation of constructs and transgenic plants

OsSWEET7eandOsSWEET17were known to encode truncated functional domain-containing SWEET proteins.We therefore focused on the other 19OsSWEETgenes that encode intact SWEET proteins [1-3,5].CRISPR/Cas9 technology was used to separately knock out each of those genes in rice (Oryza sativassp.japonica)variety Zhonghua 11(ZH11).Two 20-bp sgRNA targeting the 5′UTR and/or the N-terminal exon of eachOsSWEETgene(Table S1)were cloned into the pYLCRISPR/Cas9 expression vector[17].These constructs were transformed intoAgrobacterium tumefaciensstrain EHA105, and further transformed into rice calli from mature embryos of ZH11 [18].More than 10 mutants were generated for eachOsSWEETgene and two or three mutants having a large deletion in each targeted sequence were selected for further analysis.The sequence of eachossweetknockout mutant was determined by PCR using gene-specific primers that amplified DNA fragments across the target sites.The PCR amplicons were directly sequenced and aligned with that in the wild type.

2.2.Pollen viability assays

The I2-KI staining assay was used to evaluate pollen viability.Ten anthers prior to anthesis from ten spikelets of each homozygous Cas9-free mutant grown in the field during the normal ricegrowing season (from April to October) were removed and placed on a glass slide.The anthers were crushed into a fine powder and stained with 1 mL 1% (V/V) I2in 3% (V/V) KI and observed with a light microscope.Pollen grains that were round and stained black were considered viable,shrunken and lightly stained or gray grains were assessed as non-viable [9].

2.3.Measurement of starch, lipid, and protein contents

The starch content in caryopses was extracted via the DMSO/HCl method and measured enzymatically using a starch assay kit according to the manufacturer’s protocol(Megazyme,Bray,Ireland).Amylose accumulation was assessed following the iodine-potassium iodide colorimetric method [19].Protein and lipid contents in the seeds were determined according to methods described previously[20].Twenty caryopses from each homozygous mutant were randomly sampled, and caryopses from twenty sibling plants were pooled for starch,lipid,and protein measurement.

2.4.Measurement of soluble sugar content

Twenty random caryopses from each of 20 homozygous mutant sibling plants were pooled for soluble sugar measurement.Fifty mg of caryopses were ground in liquid nitrogen,extracted with a solution of 800 μL methanol:chloroform:water (5:2:2, V/V/V) containing 10 μL of ribitol solution as an internal standard.After centrifugation at 12,000 r min-1for 10 min,200 μL of supernatant were collected and dried in a vacuum concentrator.The dried metabolites were derivatized with N,O-Bis(trimethylsilyl) trifluoroacetamide, and transferred to glass vials for GC-MS (Agilent 7200) analysis with the parameters of gas chromatography and mass spectrometry as reported previously [21].

2.5.Enzyme extraction and assays

Preparation of enzyme extracts and enzyme activity assessment followed Zhu et al.[22].Twenty caryopses at 15 days after fertilization from 20 spikelets of each Cas9-free homozygous mutant were sampled; these came from at least 10 different plants.About 200 mg of frozen caryopses were homogenized in a pre-cooled mortar containing 1 mL of extraction buffer (100 mmol L-1HEPES-NaOH,pH 7.6,5 mmol L-1MgCl2,5 mmol L-1DTT,2 mmol L-1EDTA, 12.5% glycerol, 5% insoluble polyvinylpyrrolidone 40).After centrifugation at 12,000×gfor 10 min, the supernatant was used for analysis of adenosine diphosphate glucose pyrophosphorylase (ADPGase), soluble starch synthase (SSSase), and sucrose synthase (SuSase) activities.All enzyme activities were expressed on a per mg protein basis.

2.6.Microscopy

Brown rice seeds were cut transversely and the ruptured transverse surfaces were coated with gold under vacuum conditions.Images were obtained with a JSM-6390LV scanning electron microscope(Jeol,Japan),and analysis was based on at least three biological replicates.

2.7.RNA-seq analysis

Rice spikelets at the booting stage were collected fromossweet11mutant and wild type with three biological replicates.RNA extraction, quantification, library construction, sequencing and data analysis were conducted by Novogene (Beijing) according to a published protocol.The DESeq R package was used for differential expression analysis, with relative expression |log2Fold change| > 1 andPadj< 0.01 considered as differentially expressed genes.Software tools Goatools and KOBAS were used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes(KEGG) analysis was performed with an adjustedPvalue cutoff of 0.05.

2.8.Gene expression analysis

Inflorescences at the heading stage were sampled for RNA extraction.Inflorescence developmental stages were described on a P1 to P6 scale with P1 indicating floral transition and floral organ development (panicle length 0-3 cm), P2 indicating the meiotic stage(panicle length 3-5 cm),P3 indicating the meiotic stage(panicle length 5-10 cm), P4 indicating the microspore stage (panicle length 10-15 cm), P5 indicating vacuolated pollen stage (panicle length 15-22 cm),and P6 indicating the mature pollen stage(panicle length 22-30 cm).Developing embryos or developing caryopses were sampled for RNA extraction at the grain fill stage represented by stages S1 to S5 with S1 indicating early globular embryos, S2 indicating middle and late globular embryos, S3 indicating embryo morphogenesis, S4 indicating embryo maturation,and S5 indicating dormancy and desiccation tolerance.Total RNA from different tissues was used for gene expression analysis by quantitative reverse transcription-PCR(qRT-PCR)as described previously [8].PCR primers are listed in Table S2.The riceactingene was used to standardize relative RNA measures.Each qRT-PCR assay was repeated at least twice with similar results, with each repetition having three replicates.

2.9.Phenotypic analyses

Grain length, grain width, grain thickness, and 1000-grain weight were measured when field-grown plants were mature.At least 100 grains from each plant were measured and measurements were repeated three times.Images of filled grains were taken using ImageScanner III (GE Healthcare, Chicago, IL, USA),and grain-related traits were obtained using a high-throughput rice phenotyping facility [23].

2.10.Statistical analysis

Statistical parameters(sample size and biological replicates)are reported in the figures and figure legends.Differences between samples were analyzed for statistical significance using twotailed Student’st-tests in Excel (Microsoft, Redmond, WA, USA).

3.Results

3.1.Preferential expression of rice SWEET genes during different stages of inflorescence development

We previously identified 21OsSWEETgenes and showed that they were expressed in various tissues, including roots, stems,leaves,leaf sheaths,and flowers;19 genes were expressed in flowers at diverse transcription levels[8].For more precise assessments we examined their expression profiles by qRT-PCR assays at specific stages of inflorescence development.Six genes (OsSWEET1b,OsSWEET2a,OsSWEET5,OsSWEET6b,OsSWEET7c, andOsSWEET13)showed increased expression levels, eight (OsSWEET2b,OsSWEET3b,OsSWEET4,OsSWEET6a,OsSWEET7a,OsSWEET7b,OsSWEET12, andOsSWEET16) had constant expression levels and two(OsSWEET11andOsSWEET14) showed decreased transcription levels during progression of inflorescence development.Expression ofOsSWEET1aincreased with inflorescence development and reached the highest level at the young microspore stage and then declined.OsSWEET3aandOsSWEET15had increased transcription levels from P1 to P3 or P4, but then decreased (Fig.1).These data indicate thatOsSWEETgenes have different expression patterns during rice inflorescence development.

Fig.1.Expression profiles of OsSWEETs during inflorescence development.Samples were collected from inflorescence of rice cultivar Zhonghua 11 at different developmental stages.P1,0-3 cm,floral transition and floral organ initiation;P2, 3-5 cm,meiotic stage;P3, 5-10 cm,meiotic stage; P4,10-15 cm,young microspore stage;P5,15-22 cm,vacuolated pollen stage; P6, 22-30 cm, mature pollen stage.Data are means±SD, n = 3.*, P < 0.05; **, P < 0.01 (Two-tailed Student’s t-tests).

3.2.Preferential expression of rice SWEET genes at different stages of embryo/seed development

As suppression or knockout ofOsSWEET11[3,12,13]and knockout ofOsSWEET14[10] caused severe seed-fill defects, we quantified the expression profiles of allOsSWEETgenes during embryo/seed development by qRT-PCR assay.ElevenOsSWEETgenes (OsSWEET2a,OsSWEET2b,OsSWEET3a,OsSWEET3b,OsSWEET6a,OsSWEET6b,OsSWEET7a,OsSWEET7b,OsSWEET7c,OsSWEET12, andOsSWEET15) had constant transcription levels, two (OsSWEET1b,andOsSWEET11) showed increasing transcription during embryo/seed development, five (OsSWEET1a,OsSWEET4,OsSWEET5,OsSWEET13, andOsSWEET16) showed decreasing levels, andOsSWEET14showed increased levels from S1 to S3, but decreased levels at later stages (Fig.2).The dynamic transcription levels of individualOsSWEETgenes during different stages of embryo/seed development suggested the important roles in embryo/seed development.

3.3.Generation of ossweet knockout mutants

To understand individualin vivofunctions ofOsSWEETgenes,CRISPR/Cas9 technology was used to knock out 19OsSWEETgenes.Two 20-nt sequences in the 5′UTR and/or N-terminal exons of eachOsSWEETgene were selected as target sites for Cas9 cleavage(Fig.S1).For eachOsSWEETgene,two or three mutants containing 166-bp to 1385-bp deletions detected by sequencing the target regions after PCR amplification were chosen for further analysis(Fig.S1).Alignment of corresponding mutant and wild type gene sequences showed that each mutant carried a truncated or altered sequence that was not transcribed(Fig.S2);that is,they were lossof-function mutants.To eliminate the potential off-target mutants we backcrossed each mutant with wild type plants and generated Cas9-free mutant and WT pairs with for eachOsSWEETgene for further analysis.

3.4.Several ossweet mutants had defective grain shape

Many the reported plantSWEETgenes are associated (or putatively associated) with reproductive development [3].Therefore,we mainly focused on reproduction-related performance of all currently identifiedossweetmutants.First,we investigated the pollen viability of allossweetmutants that were planted in the field.The mature pollen grains of plants withossweet1b,ossweet4,ossweet7c,ossweet11, andossweet14were shrunken and lightly stained or gray after I2-KI staining, with pollen viabilities of about 51%, 23%,66%,38%,and 63%,respectively,compared with 94%darkly stained(viable) pollen in the wild type (Fig.3A, B).The other 14 mutants had similar pollen morphology and comparable pollen viability to the wild type (Fig.S3A, B).

Fig.2.Expression profiles of OsSWEETs during embryo/seed development.Samples were collected from embryos/seeds of Zhonghua 11 at different developmental stages;S1,0-2 days after pollination,early globular embryos;S2,3-4 days after pollination,mid and late globular embryos;S3,5-10 days after pollination,embryo morphogenesis;S4,11-20 days after pollination, embryo maturation; S5, 21-30 days after pollination, establishment of dormancy and desiccation tolerance.Data are means±SD, n = 3.*,P < 0.05; **, P < 0.01 (Two-tailed Student’s t-tests).

Fig.3.Defectivepollenviabilityandseedtraitsinsomeossweetknockoutmutants.I2-KIstainingofpollengrainsofwildtypeandossweet1b,ossweet4,ossweet7c,ossweet11,andossweet14knockoutmutants.Darklystainedpollengrainswereassessedasviablewhereasthoseshrunkenorstaininggrayorgrayincolorwereclassedassterile.Bars,100μm.(B)Fertilepollenratesofwildtypeandossweet1b,ossweet4,ossweet7c,ossweet11,andossweet14knockoutmutants.(C)1000-grainweightsofwildtypeandossweet knockoutmutants.(D)Comparisonsofmaturecaryopsesofwildtypeandossweetknockoutmutants.Scale bars,1cm.Dataaremeans±SD,n=20.*,P<0.05;**,P<0.01(Two-tailedStudent’st-tests).

Fig.4.Relative transcription levels of genes affecting grain filling in ossweet knockout mutants.Data are means±SD, n = 3.*, P < 0.05; **, P < 0.01 (Two-tailed Student’s ttests).

We examined several seed traits in theossweetmutants,including seed set, grain length, grain width, grain thickness, and 1000-grain weight.Theossweet1b,ossweet4,ossweet7c,ossweet11, andossweet14plants had lower rates of seed setting than the wild type,consistent with their high levels of pollen inviability (Fig.3A)whereas the other 14ossweetmutants had similar seed setting rates to the wild type (Fig.S4).Most of theossweetmutants had similar grain length to the wild type, except for slightly longer grain length inossweet11and shorter grain length inossweet14(Fig.S4).Theossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutant plants had decreased grain width, whereas the other 13ossweetmutants were similar to the wild type.Theossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants had lower grain thickness than the wild type (Fig.S4).The 1000-grain weights of theossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants were 85.5%, 87.8%, 32.9%, 88.8%, 77.8%, 37.1%,and 67.5% less than the wild type, respectively; grain weights of the other 12 mutants were similar to the wild type (Fig.3C).Among the 19 mutants, theossweet4andossweet11mutants caused the most extreme effects, suggesting thatOsSWEET4andOsSWEET11played the most important roles in seed development.

3.5.Attenuated expression of grain filling genes in ossweet mutants

Many rice genes are involved in determination of grain traits[24],such as QTL for grain size on chromosome 3 (GS3) [25],grain width and weight on chromosome 5 (GW5) [26], grain width and weight on chromosome 2(GW2)[27]and an otubain-like protease gene (OsOTUB1) negatively regulating grain thickness on chromosome 8 [28].To evaluate whether any of these genes contributed to abnormal seed in theossweetmutants we traced the transcriptional activities of their markers in theossweetmutants.Young rice panicles were collected for RNA extraction and qRT-PCR.We found no evidence to suggest that the mutants affected any of these genes (Fig.S5).

Genes affecting grain filling or rate of grain filling also influence seed maturation [24].We determined the accumulation of representative genes in developing caryopses at 15 days after fertilization, includingGRAIN INCOMPLETE FILLING 1(GIF1),Nuclear Factor Y B 1(OsNF-YB1),andpyruvate kinase 3(OsPK3)which control grain filling, andGRAIN-FILLING RATE 1(GFR1) that controls the rate of grain filling [29-33].GIF1,OsNF-YB1, andOsPK3had significantly lower transcription levels inossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants, whereas the other 12 were unchanged relative to the wild type (Fig.4).GFR1had reduced expression levels inossweet4andossweet11mutants (Fig.4), whereas the other 17 were unchanged relative to the wild type(Fig.S6).Taken together,these results suggest that defective seed traits in at least sevenossweetmutants were caused by attenuated transcription of grain filling during caryopsis development rather than compromised transcription of grain shaperelated genes during panicle development.

3.6.Altered starch accumulation in ossweet mutants

Developing caryopses at 25 days after fertilization were sampled for analysis of starch, protein and lipid contents.There was significantly decreased starch accumulation inossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants compared with that in the wild type (Fig.5A).Scanning electron microscopy showed that the endosperms of all seven mutants comprised small, round or unregular, and loosely packed compound starch granules that clearly differed from the wild type with densely packed, polyhedral compound starch granules(Fig.5B).All seven mutants had less amylose accumulation(Fig.5A).Protein and lipid contents in the endosperms of all mutants were unchanged (Fig.S7).

We determined transcript levels of starch synthase genegranule bound starch synthase I(OsGBSSI) and amylopectin synthesis genesoluble starch synthase IIIa(OsSSIIIa) [19] in seeds from the above sevenossweetmutants.Expression of both genes was significantly lower in the mutants than in the wild type (Fig.S8).Taken together, attenuated starch accumulation in seeds of these sevenossweetmutants indicated abnormal sugar metabolism.

3.7.Sugar accumulation in ossweet mutants

Fig.5.Grain starch and amylose contents in ossweet knockout mutants.(A) Starch and amylose contents in rice seeds.Data are means±SD, n = 20.*, P < 0.05; **, P < 0.01(Two-tailed Student’s t-tests).(B) Scanning electron microscope images of transverse sections of seeds from selected ossweet knockout mutants.Scale bars, 10 μm.

We quantified the contents of sucrose, glucose, and fructose in mature seeds of all theossweetmutants.There were higher sucrose concentrations in seeds from theossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants, higher glucose contents inossweet1a,ossweet4,ossweet7c,ossweet11,andossweet14mutant seeds, and more fructose in seeds ofossweet1a,ossweet1b,ossweet4,ossweet6b, andossweet7cmutants than in wild type (Fig.6A).All otherossweetmutants had similar levels of these sugars to the wild type (Fig.S9).In parallel with higher sugar concentration and lower starch accumulation in mature seeds of theossweet1a,ossweet4,ossweet11, andossweet14mutants, the activities of adenosine diphosphate glucose pyrophosphorylase (ADPGase), soluble starch synthase (SSSase),and sucrose synthase (SuSase) in developing caryopses at 15 days after fertilization were significantly lower than the wild type(Fig.6B).We then assessed the expression levels of sucrose synthase genes,OsSUS3andOsSUS4[34].Compromised transcription levels were observed in theossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants (Fig.S10).These results demonstrated that sugar-to-starch metabolism was attenuated in all sevenossweetknockout mutants.

As theossweet1a,ossweet4,ossweet11, andossweet14mutants had lower grain weight, increased sugar content, and decreased starch accumulation, we carried out time-course assays to survey the dynamics of accumulation of both sugar and starch (Fig.6C).Sucrose accumulation in each mutant was lower than the wild type in caryopses at 5 and 10 days after fertilization, but higher at 15 days after fertilization.Starch contents were lower in the same mutants at 5,10 and 15 days after fertilization(Fig.6C).Inadequate transportation of sucrose from the leaves to developing caryopses might have contributed to lower sucrose accumulation in someossweetmutant caryopses at the early grain-fill stage and attenuated sugar-to-starch conversion might have caused the lower starch contents.

Fig.6.Altered sugar concentrations in mature seeds of ossweet knockout mutants and modified sugar-to-starch enzyme activity in developing caryopses of ossweet knockout mutants.(A) Sucrose, glucose and fructose contents in mature seeds of ossweet knockout mutants.(B) ADPGase, SSSase and SuSase activities in developing caryopses of ossweet knockout mutants at 15 days after fertilization(DAF).(C)Dynamic sucrose and starch accumulation in developing caryopses of ossweet knockout mutants.Data are means±SD, n = 10.*, P < 0.05; **, P < 0.01 (Two-tailed Student’s t-tests).

High concentrations of soluble sugars were implicated in inhibition of early seedling development and root growth [35].We cultured seeds of the above seven mutants and wild type on 0.5 MS medium for two weeks to evaluate effects on germination and early seedling development.Theossweet4,ossweet7c,ossweet11,andossweet14mutants had significantly reduced shoot length,root length, and fresh weight, whereas theossweet1a,ossweet1b, andossweet6bmutants had attenuated shoot length and fresh weight(Fig.7).These results indicated that the high soluble sugar contents had inhibitory effects on early seedling development.

3.8.Altered sugar-to-starch metabolism in the ossweet11 mutant

OsSWEET11was previously reported to be essential for reproductive development in rice [9,10,12], and foregoing results demonstrated thatossweet11mutant plants had decreased pollen viability and seed setting,shrunken mature caryopses,and attenuated starch and amylose accumulation.To further investigate the underlying mechanism, we performed RNA sequencing onossweet11mutant and wild type plant spikelets at the booting stage.A total of 314 differentially expressed genes (DEGs) were identified, 276 of which were up-regulated and 38 were downregulated with at least 2-fold changes inossweet11compared with the wild type (Fig.S11A).GO enrichment analysis indicated that most enriched GO terms for biological processes were regulation of macromolecule metabolism, macromolecule biosynthesis, transcription, and nucleic acid metabolism.The enriched GO terms for molecular functions were transcription factor activity and transcription regulatory activity (Fig.S11B).KEGG pathway enrichment analysis showed that the top enriched pathways were metabolic pathway and biosynthesis of secondary metabolites(Fig.S11C).

To integrate sugar accumulation and altered KEGG pathways in theossweet11mutant we focused on starch and sugar metabolism.There were three clades of enzymes with significantly higher activity inossweet11mutant than in wild type,including β-glucosidase(EC 3.2.1.21), sucrose synthase (EC 2.4.1.13), and trehalose phosphatase(EC 3.1.3.12)(Fig.S12).Among at least five genes encoding enzymes with β-glucosidase activityOs3bglu7,Os1bglu5,andOs4bglu18, but notOs3bglu8andOs9bglu33, had higher transcription levels in theossweet11mutant than in wild type(Fig.8A).Accordingly, there was higher β-glucosidase activity in developing caryopses ofossweet11mutant than in the wild type (Fig.8B).Since these enzymes catalyze β-D-glucosyl residues such as β-Dglucoside, cellobiose and cellodextrin to generate D-glucose, the higher β-glucosidase activity was consistent with higher glucose content inossweet11seeds.We simultaneously assessed expression levels ofOsTPPgenes encoding trehalose-6-phosphate phosphatases which catalyze conversion of trehalose-6-phosphate to trehalose [36].OsTPP7andOsTPP9had significantly higher transcriptional accumulation inossweet11mutant than in the wild type, whereas four otherOsTPPgenes had similar transcription levels in both theossweet11mutant and wild type(Fig.8C).In line with higher expression levels ofOsTPPsin caryopses of theossweet11mutant, there was 1.4- to 1.8-fold higher trehalose accumulation inossweet11mutant than in the wild type (Fig.8D).As trehalose is converted to glucose, accumulated trehalose resulted in abundant glucose inossweet11mutant seeds.

Fig.7.Effects of ossweet knockout mutants on seedling growth.Data are means±SD, n = 20.*, P < 0.05; **, P < 0.01 (Two-tailed Student’s t-tests).

4.Discussion

Plant SWEET proteins have roles in diverse developmental processes including inflorescence and seed development.OsSWEET4,OsSWEET11,OsSWEET15in rice, andZmSWEET4cin maize play key roles in seed-filling, largely through transport of sugar for endosperm development [12-14].The present results provide further insight into the roles of seven of 19SWEETgenes in sugar transportation and seed development based on gene editing and sugar-to-starch metabolism analysis.

4.1.SWEET genes have different roles in inflorescence and seed development

Previous work indicated that the 21 riceOsSWEETgenes had different expression patterns,most of them showing relatively higher transcriptional accumulation in panicles than in other tissues;only two were not detected in the panicles [8].Our current results on dynamic expression patterns during inflorescence and seed development indicated diverse specific roles.For example,OsSWEET4had constant expression during inflorescence development but decreased expression during seed development, whereasOsSWEET11had low expression during inflorescence development but higher expression during seed development.Although these two genes displayed different expression patterns their knockout mutants developed severely shrunken mature caryopses [12-14].Five of the 19ossweetmutants had greatly reduced pollen viability but displayed different expression levels during inflorescence development (Figs.1, 2).Although the other 14OsSWEETgenes showed decreased transcription levels during inflorescence development the respective mutants had normally stained pollen.However, effects on embryo sac and female gamete viability are quite likely and are yet to be investigated.Thus,we could not conjecture all their roles in inflorescence development simply based on expression patterns.The same fiveossweetmutants had significantly decreased seed setting and altered grain shape, resulting in decreased 1000-grain weight (Fig.3).Theossweet1aandossweet6bmutants similarly had decreased grain width and grain thickness but with normal pollen viability.Since there were apparently both conserved and divergent effects on inflorescence and seed development among the 21SWEETgene members we could not eliminate functional redundancy among them.For example,ossweet11plants had shrunken caryopses, andossweet15mutants had normal mature caryopses, butossweet11;15double mutants had more severely affected caryopses than mutants withossweet11alone [13].More double, triple even more complex homozygous mutants will need to be generated to determine whether otherOsSWEETgenes have functional redundancy during inflorescence and/or seed development.

4.2.SWEET-mediated sugar transportation is vital for starch metabolism

Some members of the rice SWEET family function as lowaffinity glucose and sucrose transporters in the HEK293T cell line andXenopusoocytes[37,38],and others complement sugar uptake deficiency in yeast mutants[4,8,14].However,only OsSWEET4 and OsSWEET11 transport glucose, sucrose or fructosein planta[12-14].Sevenossweetmutants (ossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11,ossweet14) accumulated higher sucrose in mature caryopses; among them thesweet1a,ossweet4,andossweet7cmutants also had more glucose and fructose, but theossweet11andossweet14mutants accumulated glucose but not fructose and theossweet1bandossweet6bmutants accumulated fructose but not glucose in mature caryopses (Fig.6).Sugar accumulation of indicates that the corresponding OsSWEET proteins act as sugar transporters.The different contents of glucose,sucrose, and fructose in mature caryopses indicate that individual OsSWEET proteins have specific affinity in selecting sugars for transportation.Whole-cell patch clamping experiments and 3D structural resolution of these OsSWEET proteins could assist in unraveling the situation.Although someOsSWEETgenes could not complement sugar uptake deficiency of the EBY4000 yeast mutant after adding glucose, fructose or sucrose as substrates [8],we were able to reach a conclusion regarding the specificity of OsSWEET proteins in sugar transportation.Similarly, someossweetmutants had normal caryopses, indicating minor, if any, effects on seed development, but they could be involved in sugar transportation among other tissues.

Fig.8.Altered representative genes and metabolism in ossweet11 knockout mutant seeds.(A)Relative transcription levels of representative genes with β-glucosidase activity.(B) Relative β-glucosidase activity in developing caryopses.(C) Relative transcription levels of OsTPP genes.(D) Trehalose content in developing caryopses.Data are means±SD, n = 3.*, P < 0.05; **, P < 0.01 (Two-tailed Student’s t-tests).

Conversion of sugars into inert storage compounds like starch is vital for plants where starch stores energy for cell metabolism[39].If sugars cannot be efficiently transported into endosperm and converted into starch there will be obvious consequences in regard to seed development and embryo viability.OsGCS1has roles in sucrose-to-starch conversion, andosgcs1knockout mutants have defective endosperm and embryos.The enlarged seed-like tissue contained sucrose but almost no starch [40].Among theossweetmutants low starch contents and increased sugar levels were observed in mature caryopses ofossweet1a,ossweet1b,ossweet4,ossweet6b,ossweet7c,ossweet11, andossweet14mutants.As these proteins have sugar transportation roles we speculate that sugars could not be efficiently converted into starch, resulting in sugars accumulated and starch compromised in these seeds.Altered expression of starch synthase genes also causes inadequate starch accumulation and shrunken caryopsis developments (Fig.3D).

4.3.OsSWEET4 and OsSWEET11 have major effects on sugar transportation in caryopses

Although sevenossweetmutants had significantly decreased pollen viability, abnormal seed traits and starch content, theossweet4andossweet11mutants had the lowest pollen viability and seed setting, lowest grain weight, and the minimum starch accumulation.Among the 19 genes investigated,OsSWEET4andOsSWEET11had the highest transcript accumulations in both inflorescences and seeds (Fig.S13).Consistent with the highest transcription levels, mature caryopses of theossweet4andossweet11mutants had more sucrose and glucose but less starch than all other mutants.This indicates thatOsSWEET4andOsSWEET11are essential for inflorescence and caryopsis development through transport of sugars to facilitate sugar-to-starch metabolism.

The finding that mature caryopses of theossweet4but not theossweet11mutant also stored more fructose hints that OsSWEET4 and OsSWEET11 have different affinities for fructose as a substrate for transportation.OsSWEET4 and OsSWEET11 encode membrane proteins consisting of seven transmembrane domains and belong to different phylogenetic clades[18].Consistent with roles in sugar transport, homo-oligomerization is necessary for the function of SWEET proteins [41].Structural biology revealed that SWEET proteins form homo-trimeric complexes that contain sugar-binding pockets across the membrane to efflux or influx sugars [42].Whether the amino acids differences between OsSWEET4 and OsSWEET11 determine different affinities for sugars is yet to be determined.

5.Conclusions

Previously published and present results suggest that riceSWEETgenes have essential roles in inflorescence and seed development, largely by transporting sugars to developing caryopses where they are converted to stored starch.However, the various riceSWEETgenes had diverse expression patterns and their translated proteins have diverse substrate affinity,resulting in different roles in caryopsis development.The possibility of additional specialized roles of these genes in transport of sugars between other tissues cannot be ignored.The present findings shed light on the roles ofSWEETgenes on sugar-to-starch conversion during caryopsis development and should contribute to the understanding ofSWEETgenes of other plant species.

CRediT authorship contribution statement

Pei Li and Hongbo Liu:Methodology.Lihan Wang:Visulization.Meng Yuan:Project administration, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by the National Natural Science Foundation of China(31821005,31822042,and 31871946),the Natural Science Foundation of Hubei Province(2020CFA058),and the Fundamental Research Funds for the Central Universities(2662019FW006).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.04.012.