Transcriptome changes in seeds during coleorhiza hair formation in rice

To Song,A.N.M.Ruiyth Bin Rhmn,Detosh Ds,Neng-Hui Ye,Feng Yng,Fu-Yun Zhu,Mo-Xin Chen,e,*,Jin-Hu Zhng*

a Co-Innovation Center for Sustainable Forestry in Southern China,College of Biology and the Environment,Nanjing Forestry University,Nanjing 210037,Jiangsu,China

b Shenzhen Research Institute,The Chinese University of Hong Kong,Shenzhen 518057,Guangdong,China

c Southern Regional Collaborative Innovation Center for Grain and Oil Crops in China,College of Agriculture,Hunan Agricultural University,Changsha 410128,Hunan,China

d Department of Biology,Hong Kong Baptist University,Kowloon 999077,Hong Kong,China

e CAS Key Laboratory of Quantitative Engineering Biology,Shenzhen Institute of Synthetic Biology,Shenzhen Institutes of Advanced Technology,Chinese Academy of Sciences,Shenzhen 518055,Guangdong,China

f School of Life Sciences and State Key Laboratory of Agrobiotechnology,The Chinese University of Hong Kong,Shatin 999077,Hong Kong,China

Keywords:Embryo Endosperm Transcriptomics Metabolomics Hormone Coleorhiza hair

ABSTRACT Coleorhiza hairs are hairlike structures in seeds of the grass family(Poaceae).The molecular mechanisms underlying its formation are largely unknown,study on this topic will expand our understanding of the effects of water status on germination during rice (Oryza sativa L.) direct seeding.Seeds of Nipponbare were treated under two water conditions:in one,half of the seed surface was immersed in water and the other half was embryo side in air (EIA),and in the other,the whole seed was covered by water(CBW).Coleorhiza hairs formed only in EIA samples.Transcriptomics was used to identify the gene regulation during coleorhiza hair formation in EIA (vs.CBW) embryos and endosperm.Embryos displayed more transcriptome modulation even though smaller in size than the endosperm.Differentially expressed genes (DEGs) were enriched in both primary and secondary metabolism and showed changes in abscisic acid,auxin,jasmonic acid,and salicylic acid signatures.Metabolites enrichment data were positively correlated with gene expression changes in the affected metabolic functional pathways.The presence of shorter coleorhiza hairs in an OsRHL1(Os06g0184000,a coleorhiza hair formation regulation candidate gene) knockout mutant suggested that root hair-associated DEGs share molecular regulators that control the formation of coleorhiza hairs.

1.Introduction

The grass family includes crops such as rice (Oryza sativa L.),maize (Zea mays L.),wheat (Triticum aestivum L.),and barley (Hordeum vulgare L.) that provide most of the calories consumed by humans[1].Seeds of the grass family have coleorhiza hairs,which grow on the epidermis of the coleorhiza[2].Coleorhiza hairs arise from specific epidermal cells of the developing coleorhiza at the beginning of germination and are thought to have waterabsorbing and anchoring functionality [3–5].Coleorhiza hairs of pasture species exude cohesive substances into the soil [6].The anchoring function of coleorhiza hairs in many cereal crops is dispensable because these seeds are usually covered by soil soon after seeding [4].Their water-absorbing function has more significance in these cereal crops.There are few reports on their role in seed germination.

Mechanization of land preparation and harvesting is impractical for rice establishment because paddy fields are small and scattered[7].Direct seeding of rice can reduce water input,labor requirements,and greenhouse gas emissions,and this alternative to conventional transplanting is conducive to sustainable development in China [7,8].Mechanical direct seeding is increasingly widely applied in China with the help of herbicides.However,germination of mechanical direct seeding is still unsatisfactory,especially in arid regions.

Omics technologies have been widely used to study molecular physiological processes and characteristics [9–13].The transcriptome provides information that can be used to catalog all expressed genes [14].Transcriptome analysis has been used to study molecular mechanisms in root hair proliferation induced by water deficiency in maize [15] and reveal the role of root hairs as an environmental sensor to maintain physiological function in barley under water-deficiency conditions [16].Transcriptome analysis has also provided useful clues about root hair development in rice[17].Combined transcriptomic and metabolomic analyses have been widely used to study multiple phenomena in recent years [18–20].The objectives of this study were to use transcriptome analysis to identify the regulation mechanism of coleorhiza hair formation and metabolomics to validate findings from the transcriptome analyses.

2.Materials and methods

2.1.Plant materials and growth conditions

To break seed dormancy,ripened rice seeds were dried in an oven at 50 °C for three days.

In the first experiment,hulled Nipponbare (japonica) seeds were glued (with super glue) to the middles of slides (25.4 mm× 76.2 mm) with 10 seeds per slide,and six slides were placed in a slide box (Fig.1A).The seeds were then germinated under two different water exposure conditions.In the first,half of the seed surface was immersed in water and the other half underwent embryo in air (EIA) positioning.In the second,whole seeds were covered by water (CBW).Each treatment included three boxes(18 slides) for replication,and treatments lasted for two days (on average it took two days after sowing for rice to form coleorhiza hairs).

In the second experiment,to investigate the effects of root hair length-regulating genes on coleorhiza hair formation,lowland rice Kasalath (wild type,indica) and a rice root hair mutant rhl1–1 (an ethyl methane sulfonate-generated mutant in the background of Kasalath,and with root hair length approximately 1/10 of that in Kasalath) [21] donated by professor Chuanzao Mao (Zhejiang University,Hangzhou,Zhejiang,China) were also used with the above treatments.

In the third experiment,to confirm the importance of the embryo in coleorhiza hair formation,some Nipponbare seeds were dissected into embryo and endosperm before germination.Thirty embryos (10 embryos in each replication) were held in 98% air humidity without direct contact with water,and 30 embryos were covered with water.Both treatments were applied for two days to determine whether coleorhiza hairs formed.

2.2.RNA sequencing and data analysis

2.2.1.RNA extraction and quality checking

Once hairs formed on the embryos,the seeds in experiment one were dissected into embryo and endosperm separately for both treatments.RNA from both tissues in seeds that received either treatment (EIA or CBW) was sequenced.

EIA-and CBW-treated embryos and endosperm in Nipponbare were collected when coleorhiza hairs developed in the EIA treatment(approximately 40 h after the beginning of treatment).Samples (three replicates per sample) were collected as shown in Fig.1A.RNA was extracted using an E.Z.N.A.Plant RNA Kit(Omega Biotek,Norcross,GA,USA)and quantified with a Kaiao K5500 Spectrophotometer (Kaiao,Beijing,China).RNA integrity and concentration were assessed with an RNA Nano 6000 Assay Kit in a Bioanalyzer 2100 (Agilent Technologies,Palo Alto,CA,USA).The RNA concentration for library preparation was measured and then diluted to 1 μg μL-1.

2.2.2.Library preparation for RNA sequencing

To generate sequencing libraries,2 μg of total RNA was input to the NEBNext Ultra RNA Library Prep Kit for Illumina(New England BioLabs,Ipswich,MA,USA) and libraries were constructed as follows:poly-T oligo-attached magnetic bead purification of mRNA from input total RNA;mRNA fragmentation by addition of divalent cations under heating;first strand cDNA synthesis with random hexamer primers;RNAse H degradation of residual RNA;and second strand cDNA synthesis and purification followed by terminal repair,A-tailing,and adapter addition.PCR amplification was performed to finish library preparation.

Fig.1.The seed-germination experimental setup and phenotypes of seeds subjected to EIA and CBW.(A) The setup used in all experiments.(B) Seed treated with EIA (top panel) and profuse elongated hairs on coleorhiza (including epiblast and ventral scale) that have formed.Bottom panel in (B) shows seeds treated with CBW where no coleorhiza hair was observed.(C) Ccoleorhiza hair formation on separated embryo under EIA and CBW.(D) Principal component analysis (PCA) of transcriptome samples.

2.2.3.Library inspection,clustering,and sequencing

Insert size in the library was evaluated with a StepOnePlus Real-Time PCR System (valid library concentration ＞10 nmol L-1).The clustering of the index-coded samples was performed with the HiSeq PE Cluster Kit v4-cBot-HS (Illumina,San Diego,CA,USA).Libraries were sequenced on the Illumina NovaSeq 6000 platform.Read counts are described in Table S1.

2.2.4.Transcriptome analysis

A quality check of the reads was performed with FastQC [22],and trimming/adapter removal (Cutadapt 1.15) [23] was performed to obtain clean reads (Table S1).Filtering criteria:1) Cutadapt was used to remove adapter sequences[23],and 2)reads with quality scores greater than Q20 were retained.These reads were mapped to the reference genome IRGSP-1.0,INSDC Assembly GCA_001433935.1 [24] with HISAT2 2.0.5 [25].The read count for each gene in each sample was obtained with HTSeq 0.6.0[26],and after normalization of read counts,FPKM (fragments per kilobase million mapped reads) was calculated to estimate the expression levels of genes in each sample.DESeq2 1.6.3 [27]was used to identify differential gene expression by estimating the gene expression level by linear regression,calculating fold changes for sample comparisons(EIA vs.CBW),P-value with Wald test and corrected P-value (q-value) following Benjamini and Hochberg adjustment.Genes with q-value at most 0.05 and absolute value of log2fold-change at least 1 were identified as differentially expressed genes (DEGs).Functional enrichment analyses were performed using Gene Ontology(GO)[28]and Kyoto Encyclopedia of Genes and Genomes (KEGG) [29] pathway analyses.All comparisons shown in the transcriptome data compare EIA-vs.CBW-treated seeds.

2.2.5.Hormonometer and MapMan analysis

Hormonometer was used to correlate transcriptome responses with publicly available gene expression data for hormone treatment responses[30].Rice protein sequences(corresponding to DEGs and extracted from IRGSP-1.0_protein_2021-05-10.fasta.gz as obtained from RAPDB [31] were aligned against the Arabidopsis proteome(Araport11_genes.201606.pep.fasta.gz as obtained from TAIR [32]using the NCBI blast command-line pipeline[33]to find putatively homologous genes.These Arabidopsis gene IDs were then used as an input in hormonometer analysis.A heat map of the hormonometer results was drawn with gplots package in R software[34].

MapMan [35] was used to visualize rice functional groups represented in the RNA-Seq samples.log2Fold-change values were mapped onto the heat maps of pathways in MapMan.

For the transcription factor (TF) hypergeometric enrichment test,the Microsoft Excel (Microsoft Corporation,Redmond,WA,USA) function HYPGEOM.DIST (number of specified TFs in the DEG list,total number of TFs in the DEG list,number of specified TFs in the genome,total number of TFs in the genome) was used to obtain the enrichment P-value of a specified TF in the DEG list,and a P-value ＜0.05 was taken to represent enrichment of that TF in the DEG list.

2.3.Metabolic profiling

Ground embryo sample (0.05 g) and 800 μL of 100% methanol were used to conduct metabolite extraction following Niu et al.[36].The samples were vortexed for 30 s and centrifuged at 12,000 r min-1at 4 °C for 15 min.Clear supernatant (200 μL)was then transferred to sample vials for liquid chromatography mass spectrometry (LC-MS) analysis.Metabolites were detected using an LC-Q/TOF-MS analysis platform (Agilent 1290 Infinity LC,6530 UHD and Accurate-Mass Q-TOF/MS) at Shanghai Sencichip Infotech Co.Ltd.(Shanghai,China) following Tong et al.[37].Chromatography was performed on an Agilent C18 column (Agilent,100 mm × 2.1 mm,1.8 μm).The column temperature was maintained at 40 °C and samples were eluted at a flow rate of 0.4 cm3min-1.The mobile phase consisted of 0.1% formic acid solution (A) and acetonitrile (B).The gradient elution process was as follows:0–2 min,5% B;2–13 min,5%–95% B;13–16 min,95%B.The injection volume was 4 μL and automatic injector temperature 4 °C.The MS system was operated in both positive-ion(ESI+) and negative-ion (ESI-) modes.The capillary voltages in positive mode and negative mode were 4 and 3.5 kV,respectively.The remaining parameters were the same in positive and negative mode.The drying gas flow was 11 L min-1,the gas temperature was 350 °C,the nebulizer pressure was 0.31 MPa,the fragment voltage was 120 V,and the skimmer voltage was 60 V.Scans were collected from 100 to 1000 m/z.Reference ions were injected to monitor the accuracy of the mass axis in the process of mass spectrometry.The reference ions were of respectively 121.0509 and 922.0098 m/z and 119.0363 and 966.0007 m/z in positive and negative ion modes.Metabolite depletion or enrichment was performed by comparison of EIA-with CBW-treated seeds.Metabolites with ‘‘variable importance in projection” (VIP)values ＞1.0 and P-values ＜0.05 were considered as differentially accumulated metabolites (those with log2Fold-change ＞0 were identified as up-accumulated metabolites and those with log2Foldchange ＜0 as down-accumulated metabolites).KEGG databases were used to construct enrichment pathways of metabolites.

2.4.Hormone measurement

Hormone measurement was conducted following He et al.[38].Aliquots of 1.0 g of each sample were rapidly frozen in liquid nitrogen,ground to powder,and extracted with 0.5 mL of a solution of methanol,water and formic acid(15:4:1,v/v/v)at 4°C.The extract was vortexed for 10 min and centrifuged at 14,000 r min-1at 4°C for 5 min.The supernatant was collected,and the extraction steps repeated.The combined extracts were evaporated to dryness under a stream of nitrogen gas,reconstituted in 80% methanol(v/v),ultrasonicated for 1 min,and filtered (PTFE,0.22 μm;Anpel,Shanghai,China).Quantification of endogenous indoleacetic acid(IAA),abscisic acid (ABA) and salicylic acid (SA) was performed using an ultra-performance liquid chromatography-tandem mass spectrometry (LC–MS/MS) system (UPLC,Shim-pack UFLC SHIMADZU CBM30A system,Kyoto,Japan;MS,Applied Biosystems,Foster City,CA,USA).The contents of IAA,ABA and SA were determined using standard curves and expressed as ng g-1fresh weight(FW).Standard curves representing 0.01,0.05,0.1,0.5,1,5,10,50,100,200,and 500 ng mL-1of IAA,ABA,and SA were used to quantify each compound.Three biological replications were performed.

2.5.Statistical analysis

Analysis was conducted on IBM SPSS Statistics version 18(SPSS Inc.,IL,USA)employing Duncan’s multiple range test post hoc test(significance assigned for P-value less than or equal to 0.05).

3.Results

3.1.Coleorhiza hairs formed specifically in seed embryos exposed to air

Both EIA-and CBW-treated seeds germinated normally within 2 days.Even though germination was similar in both cases,the EIA-treated seeds developed a singular phenotype:profuse hairlike structures originating from the epidermis of the entire embryo surface(coleorhiza,epiblast,and ventral scale)just before root emergence,which were designated as coleorhiza hairs (Fig.1B).No coleorhiza hairs were observed on the CBW-treated seeds.Coleorhiza hair formed in separated embryos held at 98%air humidity,while similar to the phenotype of intact seeds under CBW treatment,coleorhiza hair was not formed in those separated embryos covered by water(Fig.1C).This observation suggested dissection of the seeds into embryo and endosperm for transcriptome analysis,given that coleorhiza hair-forming signals seemed to be derived sufficiently from the embryo alone without any endosperm.We hypothesized that while some signals,such as carbohydrate mobilization,could contribute to the growing embryo via the endosperm,the coleorhiza hair-regulating signals were confined mostly within the embryo.

3.2.Tissue specificity was the major factor contributing to coleorhiza hair formation

The replicates for each tissue-treatment combination clustered closely with each other,suggesting robust replication in the transcriptome profiling (Fig.1D).Along PC1,major variation (82%)was due to tissue.PC2 showed minor variation(15%)among treatments but only in embryos,suggesting that treatments affected the transcriptome of embryos differentially in comparison with endosperm.The CBW-treated samples in the embryos and endosperm were close to each other,whereas the EIA-treated samples were located farthest apart,suggesting a large effect of EIA treatment on transcriptome modulation in both embryos and endosperm.

3.3.Embryos displayed strong transcriptome modulation between different water conditions

As suggested by the PCA,more DEGs were found in embryos than in endosperm.Whereas 4478 genes were upregulated and 2396 genes downregulated in embryos,only 734 genes were upregulated and 609 downregulated in endosperm (Fig.2A).The overlap between these datasets is illustrated in Fig.2B.There were 416 common upregulated DEGs and 365 common downregulated DEGs in embryo and endosperm.Only 12 genes downregulated in embryos were upregulated in endosperm.Only 23 genes upregulated in embryos were downregulated in endosperm.More than half of DEGs(up-or downregulated)in endosperm were also DEGs in embryos,suggesting less distinct transcriptomic signals in the endosperm.

3.4.Functional classification of DEGs by GO and KEGG pathway analysis

GO enrichment tests identified biological processes overrepresented in each tissue(Fig.2C).In the embryo,the top enriched biological processes belonged to metabolic and growth processes:‘‘carbohydrate metabolism”,‘‘response to oxidative stress”,‘‘phenylpropanoid biosynthesis”,‘‘cell wall organization and biogenesis”,and ‘‘sulfur metabolism”.This finding suggested that genes involved in energy use derived from carbon utilization for growth and secondary metabolic processes associated with response to abiotic stress are prevalent in embryos.However,in the endosperm,stimuli response-and metabolism-related processes including ‘‘defense response”,‘‘carbohydrate catabolism”,and ‘‘lignin metabolism” were specifically enriched.Processes common to the two tissues were ‘‘oxidation-reduction processes”,‘‘secondary metabolic processes”,‘‘carbohydrate metabolism”,and‘‘cell wall organization and biogenesis”.

KEGG overrepresentation analysis (Fig.2D) suggested enrichment of photosynthesis,phenylpropanoid biosynthesis specifically in embryos,and glycolysis/gluconeogenesis in both tissues.Other terms associated with secondary metabolism,fatty acid metabolism,and sulfur and amino acid metabolism were enriched in embryo-and carbon fixation-associated processes in endosperm only.

3.5.Carbohydrate metabolism-related pathways are strongly affected

DEGs in the embryos were enriched in carbohydrate metabolism-related pathways including starch and sucrose metabolism,glycolysis/gluconeogenesis,pyruvate metabolism,and tricarboxylic acid (TCA) cycle pathways,as observed in the KEGG enrichment analysis (Fig.3).More specifically,in the starch and sucrose metabolism pathway,27 genes were upregulated and 4 downregulated;in the glycolysis/gluconeogenesis pathway,20 genes were upregulated and 15 downregulated;and in the pyruvate metabolism pathway,10 genes were up-and 3 downregulated.In TCA cycle pathways,all 7 genes were upregulated.

DEGs in the endosperm were enriched similarly in carbohydrate metabolism-associated pathways,such as starch and sucrose metabolism,glycolysis/gluconeogenesis,and pyruvate metabolism pathways(Fig.S1).In the starch and sucrose metabolism pathway,3 genes were upregulated;in the glycolysis/gluconeogenesis pathway,5 genes were upregulated and 10 downregulated;in the pyruvate metabolism pathway,3 genes were upregulated and 1 downregulated.

3.6.Metabolomic measurements supported gene expression changes in affected functional pathways

Because many metabolic changes were observed in both tissues,the overall metabolic changes in rice embryos during these conditions were assessed by MapMan gene expression(Fig.4A).Secondary metabolism was highly upregulated.Lipid metabolismassociated genes,cell wall-modifying genes,and nucleotide metabolism degradation genes were upregulated.Overall,high upregulation of all metabolic genes(among DEGs)was observed.

There were 53 up-and 49 down-accumulated metabolites in the embryo (Table S3),and 48 up-and 23 down-accumulated metabolites in the endosperm(Table S4).These differentially accumulated metabolites were enriched in a) secondary metabolites such as phenylpropanoids,b)nucleotides(aminoacyl-tRNA biosynthesis),c) fatty acids (sphingolipids),d) sugars (galactose),and e)amino acids (Fig.4B).

3.7.Responses of hormone and hormone-associated genes in the embryos and endosperm

Most DEGs in embryos were enriched in the auxin,ABA,and SA signal transduction pathways (Fig.5A).In endosperm,DEGs were enriched in the auxin signal transduction pathway (Fig.S2).

The hormonometer analysis is depicted in Fig.5B.This correlation analysis suggested moderate upregulation of responses to JA,ABA,cytokinin,brassinosteroid (BR),and SA and downregulation of responses to auxin,gibberellic acid (GA) and ethylene in the embryos.In the endosperm,similar but faint responses were observed,while strong opposite regulation for early auxin and late GA responses was observed.Mapping hormone-specific genes that were DEGs in our datasets suggested similar upregulation of ABA,JA and SA responses and a mixed response to auxin,ethylene,cytokinin,and GA (Fig.S3).

Tissue hormone measurements in the samples used for RNASeq suggested that while ABA increased in both tissues,representing a general response during seed germination,auxin was reduced specifically in embryos and increased in endosperm,while SA was specifically decreased only in embryos (Fig.5C).

3.8.Profile of TF expression in EIA vs.CBW in the embryos and endosperm

Fig.2.Changes in gene expression of CBW and EIA.(A) Number of up-and downregulated DEGs in the two tissues for EIA vs.CBW transcriptome comparison.(B) Venn diagram showing the intersection of up-and downregulated DEGs sets as shown in A.GO (C) and KEGG (D) analysis of DEGs in EIA vs. CBW comparison for embryo and endosperm.Scale denotes negative log10FDR and higher value represents greater enrichment.EIA,embryo in air;CBW,covered by water.

In total,383 differentially expressed TFs (241 up-and 142 downregulated) (Table S5) were identified in embryo,and 74 (33 up-and 41 downregulated)differentially expressed TFs were identified in endosperm (Fig.6A).These TFs belonged mainly to the bHLH (basic helix-loop-helix),bZIP (basic region-leucine zipper),MYB (v-myb avian myeloblastosis viral oncogene homolog),NAC(NAM/ATAF/CUC),WRKY (WRKY protein) and ERF (ethylene responsive factor) families.The Venn intersections of TF DEGs in the embryo and endosperm revealed 16 common upregulated TFs and 20 common downregulated TFs,again suggesting that the TF induction overlap between the two tissues was small(Fig.6B).TA hypergeometric enrichment test suggested that the CCAAT box binding factor was highly upregulated and that HAP2(NF-YA) family,Aux/IAA proteins,and bHLH proteins were highly enriched in embryo-upregulated DEGs (Fig.6C).Some DEGs,such as WRKY domain proteins,C2C2 (Zn) DOF family proteins,and AP2/EREBP proteins coding genes,were enriched in both embryo up-and downregulated DEGs.Specifically,for embryodownregulated DEGs,C2C2 (Zn) CO-like proteins and MADS box proteins coding genes were enriched.In the endosperm,upregulated DEGs were enriched in WRKY domain protein-encoding genes,and downregulated DEGs were enriched in JUMONJI and pseudo-ARR protein-encoding genes.

3.9.Root hair formation-related genes were upregulated exclusively in embryos

As coleorhiza hair showed some similarities to root hair in morphology and function,we focused on DEGs with previously implicated roles in root hair formation.We found 7 such genes in the embryo,but none in the endosperm (Table S6),and all of them were upregulated.We obtained a mutant of one candidate gene in Table S6 (OsRHL1,Os06g0184000),here called rhl1–1.The mutant showed a very short root hair length compared to wild type Kasalath (Fig.7).Coleorhiza hairs formed under EIA treatments,and as expected,rhl1–1 showed shorter coleorhiza hairs than Kasalath (Fig.7).

4.Discussion

4.1.Carbohydrate metabolism pathways may be involved in coleorhiza hair formation

Fig.3.Carbohydrate metabolism-associated pathways of embryo in CBW vs.EIA comparison.(A) Starch and sucrose metabolism.(B) Glycolysis/gluconeogenesis.(C)Pyruvate metabolism.(D)tricarboxylic acid(TCA)cycle.Red means upregulated and green downregulated.Gene IDs can be found in Table S2 according to the numbers in the squares.EIA,embryo in air;CBW,covered by water.

Germination is an efficient process,which is accompanied by metabolic and physiological changes,such as mobilization of stored reserves in seeds followed by energy production for growth of new tissues[39].In previous studies[40–42],water deficits during pollination disrupted carbohydrate metabolism in maize ovaries,mature maize leaves,and seedlings of wheat cultivars indicating that water conditions strongly influence carbohydrate metabolism status during multiple stages of plant growth and development.During germination,seeds of drought-tolerant genotypes maintained their ability to use stored starch reserves via higher amylase activity and anaerobic respiration and displayed higher rates of ethylene production and lower peroxidase activity[43].In the GO analysis,embryo-associated DEGs were enriched in growth-and metabolism-related processes probably required to fuel growth,whereas endosperm-associated DEGs were enriched in defense response and primary/secondary metabolism(Fig.2C).This finding suggests the priority of growth responses over defense responses in the growing embryo and vice versa in the established endosperm.As expected,in this study,DEGs in embryo and endosperm were enriched in starch and sucrose metabolism,glycolysis/gluconeogenesis,and pyruvate metabolism pathways.We know that α-amylase in the aleurone layer secretes hydrolysis enzymes to the starchy endosperm,and the embryo utilizes sugars released by this starch degradation for its growth[44].Thus,energy and intermediate metabolites produced during this metabolic cooperation of the embryo and endosperm with respect to pathways of carbohydrate metabolism may be necessary for coleorhiza hair formation.

4.2.Phytohormones such as ABA,IAA and their master TFs may regulate coleorhiza hair formation

Fig.4.Differentially expressed metabolites at transcriptome and metabolomics levels(CBW vs.EIA comparison).(A)MapMan image for metabolites at transcriptome level.Blue indicates up-and yellow down-regulation.(B) Metabolites enriched processes heat map at metabolomics levels.Scale gives negative log10P-value and greater means more enrichment.EIA,embryo in air;CBW,covered by water.

Previous studies [45–47] have revealed that plant hormones participate in the response to abiotic stresses during seed germination.Many studies [48–51] have shown that phytohormones play important roles in regulating seed germination,including positive (GA) and negative (ABA) regulation.In this study,DEGs were upregulated in the ABA,JA and SA signal transduction pathways,while those belonging to auxin and ethylene were downregulated.Among them,DEGs belonged mostly to auxin,ABA and SA signal transduction pathways in the embryo.In the endosperm,DEGs were associated mainly with the auxin signal transduction pathway,a finding also reflected in the hormone measurement study.Upon seed imbibition,ABA levels decrease to allow embryos to germinate and develop into seedlings.However,under abiotic stress conditions,these ABA levels remain high [52].Auxin,in contrast,functions in seed germination and in plant responses to biotic and abiotic stresses [51,53–54].SA increases the seed capacity to mount adaptive responses to environmental water stress during early germination stages[55].These findings suggest that coleorhiza hair formation is regulated by multiple hormones.The roles of ABA,auxin and ethylene remain to be investigated.

TFs act as upstream regulators of major gene expression changes during any treatment-induced phenotype change;they have been reported [56–58] to regulate response to abiotic stress during seed germination.Many TFs regulate seed germination and seedling development.In the TF enrichment test,embryo DEGs consisted mainly of CCAAT box binding factor (HAT2),implicated in development from embryogenesis to vegetative stages,and bHLH and MYB domain protein-encoding genes (also involved in plant development and in response to drought stress).Auxinassociated Aux/IAA proteins and AP2/EREBP (ethylene-responsive element-binding protein)-encoding genes were enriched,supporting roles for these hormones (Fig.6C).Overexpression of AtMYC2(bHLH) and AtMYB2 (MYB) TFs,in addition to producing an ABAhypersensitive response,improved the osmotic stress tolerance of transgenic plants [59–60].NPR1 acts in SA signal transduction,and bZIP TFs interact with their active monomeric forms[61].ERFs are the major downstream regulatory factors of the ET signaling pathway in stress responses [62].The ERF family has also been reported to participate in JA-regulated stress responses [63–64].

Fig.5.Hormone traits in CBW vs.EIA comparison.(A)Hormone signal transduction pathways of embryo.Red means upregulated and green downregulated.Gene IDs can be found in Table S2 according to the numbers in the squares.(B)Hormone-correlational analysis.Blue signifies strong correlation towards induction and yellow signifies strong correlation to repression.(C) Hormone contents.EIA,embryo in air;CBW,covered by water.

WRKY domain protein-encoding genes were found among both up-and downregulated DEGs.The WRKY 70 TF is a key component mediating the antagonistic interaction between JA and SA [65].These master regulators of the described hormones may affect hormone regulation of coleorhiza hair formation.

4.3.Coleorhiza hair and root hair formation may share similar genetic mechanisms in increasing seed water retention capacity

Being sessile in nature,plants have evolved to respond to changing environmental conditions with adaptive developmental plasticity.Adaptive behavior may start at seed germination and continue during the entire life cycle [66].It has been suggested[67] that root hairs are important structures that help increase water and nutrient absorption areas;thus,genetic factors promoting root hair growth may increase interactions with soils and water and nutrient uptake,thereby increasing stress tolerance.Indeed,studies [68–69] investigating the role of root hair in water retention showed that water uptake efficiency was reduced to 64% in a hairless mutant brb than in the wild type.In a similar study[70]investigating the role of root hairs under normal vs.stress conditions and comparing wild type with root-hairless lines(NR23)in Arabidopsis,NR23 showed a 47% reduction in water absorption and decreased drought tolerance.We propose that coleorhiza hairs formed in EIA seeds help to absorb moisture from the air,supporting germination under insufficient water conditions such as drought.In the present study,seven genes involved in root hair development were upregulated in the embryos of EIA seeds and thus potentially contributed to coleorhiza hair formation(Table S6).

Fig.6.Changes in TF expression of CBW and EIA.(A)Number of up-and downregulated TFs in the two tissues for EIA vs.CBW transcriptome comparison.(B)Venn diagram showing the intersection of up-and downregulated TFs sets as shown in A.(C) Hypergeometric enrichment test to find the most enriched TF families in up-and downregulated DEGs of embryo and endosperm for EIA vs.CBW transcriptome comparison.Darker colors represent higher enrichment.EIA,embryo in air;CBW,covered by water.

Fig.7.Coleorhiza hair formation in a root hair mutant.(A) Coleorhiza hairs in Kasalath (wild type).(B) Coleorhiza hairs in rhl1–1 (an ethyl methane sulfonategenerated OsRHL1 mutant in the background of Kasalath).(C) Coleorhiza hairs and root hairs in Kasalath.(D) Coleorhiza hairs and root hairs in rhl1–1.

Arabidopsis AUX1/LAX auxin influx carrier genes function in many auxin-mediated developmental programs either independently or collaboratively,as in root hair development [71].OsLAX1 (Os01g0856500) is an auxin transport-associated gene in rice and the OsAUX1 gene family is encoded by five closely related OsAUX1/LAX genes [72].Whereas wild type roots displayed root hairs that were three times longer,the mutant Osaux1 was not responsive to low phosphate-induced root hair length promotion,showing that OsAUX1 promoted root hair growth in low-phosphate conditions. Thus,OsLAX1(Os01g0856500) and auxin transport appear essential for coleorhiza hair length regulation.

OsSNDP1 (Os10g0122600) was identified based on map-based cloning and sequencing of a short-root hair mutant [73],and the mutation arose from a single amino acid substitution in a Sec14-nodulin domain protein.The root hairs of Ossndp1 were only 1/6 as long as those of wild type plants and complementation lines produced root hairs that had lengths similar to those of wild type plants.OsSNDP1 (Os10g0122600) may also be involved in coleorhiza hair formation.

OsCSLD1 (Os10g0578200) was also shown to function in root hair development,with mutants in this gene showing root hair lengths 1/4–1/3 of those of wild type plants [74].OsCSLD1(Os10g0578200) belongs to a group of cellulose synthase like-1(CSL) protein genes that may regulate root hair growth independent of and in parallel with OsSEY1 (Os11g0582300,also named OsRHD3)[74].OsRHD3(Os11g0582300),encoding protein that regulate root hair response,was up-regulated under moderate water stress [75].OsbHLH125 (Os01g0111500,also named OsRSL1) and OsbHLH127 (Os06g0496400,also named OsRSL3) belong to the ROOT HAIR DEFECTIVE 6-like(RSL)class I family.Osrsl1 and Osrsl2 single mutants showed normal root hair formation,whereas an Osrsl1/Osrsl2 double mutant showed root hair lengths approximately 60% those of wild type plants.OsRSL1 (Os01g0111500)and OsRSL3(Os06g0496400)overexpressing lines showed root hair lengths 1.4 those of wild type plants.Complementation of the Arabidopsis rhd6 short root-hair mutant with RSL1/2/3 rescued root hair length to that observed in wild type plants [76].We propose that OsCSLD1 (Os10g0578200),OsSEY1 (Os11g0582300),OsbHLH125 (Os01g0111500),and OsbHLH127 (Os06g0496400)function in coleorhiza hair formation independently or interactively.

Fig.8.Model of transcriptome regulation involved in coleorhiza hair formation under EIA.

A novel transcription factor with a bHLH domain was found[21]to be involved in rice root hair development in an indica rice species (Kasalath),and the root hair length of an Osrhl1 mutant was approximately 1/10 of that in the wild type.Because root hair length decreased the most in the OsRHL1 mutant,we used mutant rhl1–1 for scoring coleorhiza hair formation in comparison to Kasalath wild type plants.Indeed,as expected,the mutant developed very short coleorhiza hairs in comparison with the wild type(Fig.7).

All these potential candidates may be useful for detailed study of coleorhiza hair development at the cell level using microscopic imaging and molecular biology techniques.

4.4.Model regulation involved in coleorhiza hair formation

We propose that in the EIA treatment,deregulated hormone signals from ABA and auxin in the embryo control transcriptional regulation of root hair-controlling genes via control of transcription factors (Fig.8).Massive carbohydrate metabolism changes in the embryos may fuel this growth,and these signals may also be derived from the endosperm.In the endosperm,in contrast,both ABA and auxin levels increased.However,despite these overall hormone and carbohydrate metabolic dynamics in the embryo,there may be even finer local changes that control coleorhiza hair formation.We observed decreased auxin in embryos under coleorhiza hair-forming conditions,and it has been suggested that auxin positively regulates root hair formation.Future root hairspecific expression studies may identify specific signals for coleorhiza hair development.

5.Conclusions

Coleorhiza hair formation may involve many players involved in root hair formation,as energy processes in primary as well as secondary metabolism are regulated during hair formation,and major root hair formation hormones such as IAA and ABA,which responds to water status,are potential regulators of hair formation.Genetic studies using root hair-defective mutants and pharmacological studies involving the above hormones may confirm their association with root hair formation and reveal genetic mechanisms specific to this process in grass species.

CRediT authorship contribution statement

Tao Song:Investigation,Data curation,Writing–original draft.A.N.M.Rubaiyath Bin Rahman:Conceptualization,Investigation.Debatosh Das:Investigation,Data curation,Writing -review &editing.Neng-Hui Ye:Investigation.Feng Yang:Investigation.Fu-Yuan Zhu:Investigation.Mo-Xian Chen:Supervision,Funding acquisition,Project administration.Jian-Hua Zhang:Supervision,Funding acquisition,Project administration.

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

We thank professor Chuanzao Mao (Zhejiang University,Zhejiang,China) for donating the rhl1–1 and Kasalath rice seeds.This work was supported by the National Key Research and Development Program of China (2017YFE0118100),the Natural Science Foundation of Hunan Province (2021JJ40247),Science Technology and Innovation Committee of Shenzhen(JSGG20170822153048662,GJHZ20190821160401654),Platform Funding for Guangdong Provincial Enterprise Key Laboratory of Seed and Seedling Health Management Technology(2021B1212050011),the National Natural Science Foundation of China (32001452),the Natural Science Foundation of Jiangsu Province(SBK2020042924),Jiangsu Agricultural Science and Technology Innovation Fund (CX (21) 2023),and Hong Kong Research Grant Council (GRF 14177617).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.09.009.