OsGF14b modulates defense signaling pathways in rice panicle blast response

Shijun Yn ,Qing Liu b,Thoms Nk ,Wnji Hung ,Mngyu Chn ,Qin Kong ,Shng Zhng ,Wnyn Li ,Xun Li ,Qinjin Liu ,Jinyun Yng ,Alisir R.Frni ,*,Bin Liu

a Guangdong Key Laboratory for Crop Germplasm Resources Preservation and Utilization, Agro-biological Gene Research Center, Guangdong Academy of Agricultural Sciences, Guangzhou 510640, Guangdong, China

b Guangdong Key Laboratory of New Technology in Rice Breeding, Rice Research Institute, Guangdong Academy of Agricultural Sciences, Guangzhou 510640, Guangdong, China

c Max Planck Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Potsdam-Golm, Germany

d Proteomics and Metabolomics Facility, Institute of Biotechnology, Cornell University, 526 Campus Road, Ithaca, NY 14850, USA

e Guangdong Key Laboratory of New Technology in Plant Protection, Plant Protection Research Institute, Guangdong Academy of Agricultural Sciences, Guangzhou 510640,Guangdong, China

ABSTRACT Rice with panicle-blast resistance is needed for stable rice production.Although we have previously demonstrated that OsGF14b underlies a quantitative trait locus that positively regulates rice panicle blast resistance,the mechanism is still unknown.In this study,a multi-omics approach was used to investigate the possible downstream signaling pathway regulated by OsGF14b. OsGF14b both strongly activated the gibberellin biosynthetic pathway during pathogen infection and reprogrammed the lignin biosynthetic pathway.Reduced lignin accumulation was observed in glumes of OsGF14b-overexpressing plants in comparison with the wild type after pathogen inoculation.OsGF14b activated the auxin and jasmonic acid signaling pathways,but inactivated the salicylic acid signaling pathway.Auxin and jasmonic acid appeared to act independently on OsGF14b-mediated panicle blast resistance.The roles of gibberellin,lignin,and auxin were different from their roles in leaf blast,suggesting that different mechanisms underlie leaf and panicle blast resistance in rice.This study provides a comprehensive catalog of molecular changes that could be targets for future studies of rice panicle blast resistance.

Keywords:Rice Panicle blast resistance OsGF14b Gibberellin Phytohormone

1.Introduction

Rice blast,caused by the semi-biotrophic fungusMagnaporthe oryzae,is one of the leading causes of yield loss in rice worldwide.Employing host resistance is the most effective method to control blast disease.Blast infection can cause necrotic lesions on both leaves and panicles.Panicle blast,because it causes more yield loss,is more destructive than leaf blast[1].Generally,resistances to leaf blast and panicle blast are highly correlated.However,exceptions are found.In response to infection with the same isolate,a commercialindicarice variety Zhong 156 from China was resistant to leaf blast but susceptible to panicle blast [1].The two resistance genesPi24(t)andPi25(t),also showed different resistances to leaf and panicle blast,withPi24(t)conferring resistance to leaf blast only,butPi25(t)conferring resistance to both leaf and panicle blast.In our previous study to identify genes responding to leaf and panicle blast at the whole-genome level,at least 30% of responsive genes showed differential expression patterns after inoculation.Moreover,evaluation of leaf and panicle blast resistance in 182 rice breeding lines revealed[2]that several lines were resistant to leaf blast but not to panicle blast andvice versa.These reports suggest that there are different regulatory mechanisms for leaf and panicle blast resistance.Because all molecular studies of blast resistance to date have focused on leaf blast resistance,our knowledge of the mechanism of panicle blast resistance and even of the genes that function in such resistance remains limited.

Rice blast resistance can be classified into qualitative and quantitative resistance [3,4].Qualitative resistance conferred by one or more disease-resistance(R)genes is usually highly effective;however,it is generally race-specific and easily overcome owing to selection pressure and the rapid counter-evolution of the blast pathogen.In contrast,quantitative resistance,mediated by multiple genes or quantitative trait loci,is presumably racenonspecific and is considered to be more broad-spectrum and durable [3,4].For this reason,quantitative resistance is the preferred strategy for blast control,and breeding efforts should be applied to quantitative resistance genes in the development of new blastresistant cultivars [5].Unfortunately,quantitative resistance is always conferred by multiple genes having small effect and consequently difficult to isolate.Identifying such genes is thus an obstacle to the successful use of quantitative resistance to control rice blast disease.

In our previous study [6],we found that the geneOsGF14b,a member of the rice 14-3-3 gene family,quantitatively regulates blast resistance in rice.Constitutive overexpression ofOsGF14bincreased resistance to panicle blast but reduced resistance to leaf blast.By contrast,OsGF14bsilencing plants showed decreased resistance to panicle blast but increased resistance to leaf blast,indicating thatOsGF14bplays opposite roles in leaf blast and panicle blast resistance.OsGF14bappeared to positively regulate the expression of genes involved in the jasmonic acid (JA) signaling pathway but negatively regulate the expression of genes involved in the salicylic acid (SA) signaling pathway.Despite these intriguing results,we still have very limited knowledge of the detailed mechanism ofOsGF14b-mediated rice panicle blast resistance.The 14-3-3 family proteins are regulators of plant development and plant response to various stresses [7].

The object of the present study was to elucidate the mechanism by whichOsGF14bmediates panicle blast resistance in rice.For this purpose,proteomics,metabolomics,and lipidomics analyses were integrated.

2.Materials and methods

2.1.Plant materials and pathogens

The wild-typejaponicarice (Nipponbare) (Nip) and threeOsGF14b-overexpressing lines,including transgenic lines 2(OXGF14b-2),4 (OXGF14b-4),and 6 (OXGF14b-6) were used.Rice seeds were surface-sterilized,transferred to 0.5×MS medium,and incubated in a growth chamber for germination under light of 200 μmol m-2s-1with a 12-h photoperiod at 26.Seedlings were transplanted into soil and grown in a greenhouse.Magnaporthe oryzaeGD08-T13 was used for rice blast inoculation.Panicles at the initial heading stage of Nip and the OXGF14b plants were harvested at the same time point,with or without 24 h blast infection.They were immediately frozen in liquid nitrogen,with each biological replicate containing panicles pooled from 10 plants.

2.2.Protein extraction,isobaric tags for relative and absolute quantitation (iTRAQ) labeling,and LC-MS analysis

To identify mechanisms underlyingOsGF14b-based resistance to panicle blast,we first conducted an iTRAQ-based shotgun quantitative proteomics study using the wild type Nip andOsGF14boverexpressing plants (OXGF14b-4) sampled at the same time point with or without 24 h blast infection.An overview of the experimental design is presented in Fig.S1.Protein extraction from rice panicle samples followed Zhang et al.[8] with two replicates for each sample.A 100-μg sample of protein was reduced with 200 mmol L-1tris-(2-carboxyethyl)-phosphine (TCEP) at 37for 4 h,alkylated with 375 mmol L-1iodoacetamide for 30 min,and enzymatically digested with trypsin (Promega,Madison,WI,US)in a ratio of 1:50 at 37overnight.The digested peptide samples were labeled with iTRAQ following the instructions of the iTRAQ Reagents 4-plex kit(AB Sciex,Framingham,MA,USA).iTRAQ tags 114,115,116,and 117 were used to label the rice particle extracts of Nip without infection,Nip with 24 h infection,OXGF14b without infection,and OXGF14b with 24 h infection,respectively.The efficiency of iTRAQ labeling was further assessed by LC-MS.The labeled peptides were subjected to high-pH reverse-phase(hpRP)chromatography,and then high-resolution LC-MS(Orbitrap Fusion,Thermo-Fisher Scientific,San Jose,CA,USA) as described[9].All data were acquired with Xcalibur 2.1 software (Thermo-Fisher Scientific).

Tandem mass spectra were extracted,charge-state deconvoluted,and deisotoped with Proteome Discoverer version 1.4(Thermo-Fisher Scientific).MS/MS samples were processed with Mascot version 2.4.1 (Matrix Science,London,UK).Mascot was set up to search theJaponica_uniprot database (version 201410,99,936 entries) assuming the digestion enzyme trypsin.Mascot was searched with a fragment ion mass tolerance of 0.020 Da and a parent ion tolerance of 8.0 × 10-6.Carbamidomethylation of cysteine and iTRAQ 4-plex of lysine and the N-terminus were specified in Mascot as fixed modifications.Deamidation of asparagine and glutamine and oxidation of methionine were specified as variable modifications.Protein profiling is described in Table S1.Scaffold 4.4.3(Proteome Software Inc.,Portland,OR,USA)was used to validate MS/MS-based peptide and protein identifications.Peptide identifications were accepted if they achieved a false discovery rate(FDR)<1.0%.Protein identifications were accepted if they contained at least two identified peptides.Scaffold Q+4.4.3(Proteome Software Inc.,Portland,OR,USA) was used to perform label-based quantitation of peptide and protein [10].Proteins that displayed a fold change>1.3 and differing(P<0.05)abundance by both Mann-Whitney and permutation tests were assigned as differentially expressed.The R package Upset [11] was used for comparison of different sets.KOBAS 3.0 (http://kobas.cbi.pku.edu.cn/kobas3)[12] was used for gene ontology (GO) functional annotation and kyoto encyclopedia of genes and genomes (KEGG) enrichment analyses.

2.3.Metabolomic and lipidomic study using gas chromatography tandem mass spectrometry (GC–MS) and liquid chromatography tandem mass spectrometry (LC–MS)

Rice panicle samples,including the wild type and threeOsGF14b-overexpressing lines (OXGF14b-2,OXGF14b-4,and OXGF14b-6) were used for metabolomics and lipidomic experiments with five biological replicates for each combination of two factors (genotype and post-inoculation time).A 50-mg subsample of rice panicle powder from each sample was extracted as described previously [13,14].After centrifugation,two dried 150-μL aliquots of the polar phase in each sample were subjected to GC–MS (7850A-5975C,Agilent Technologies,Palo Alto,CA,USA)and LC–MS for metabolomic study.A dried 200-μL aliquot of the nonpolar phase from each sample was subjected to LC–MS for lipidomic study.Primary metabolite profiles were acquired by GC–MS with electron impact ionization (EI) using a previously described method [15].Specialized metabolites and lipid profiles were acquired by LC–MS with electrospray ionization operated in positive mode (ESI+) and in negative mode (ESI-) as previously described [13].

Fig.1.Differentially expressed proteins in the wild type (Nip)and the OsGF14b-overexpression line (OXGF14b)with or without 24 h blast infection.(a) UpSet plot to show numbers of up-and down-regulated proteins identified by pairwise comparisons and visualize the intersection of different pair sets,indicating which sets are participating in an intersection.(b)Venn diagrams showing numbers of differentially expressed proteins identified by pairwise comparisons.The areas represent the numbers of proteins.(c)Numbers of differentially expressed proteins associated with genotype and blast infection for 24 h.WOB,without blast infection;WB,with 24 h blast infection by M.oryzae.

The metabolomic data were further analyzed as described in our previous report [15].The Agilent MassHunter Qualitative Analysis B06.00 software (Agilent) and Agilent MassHunter Quantitative Analysis B.07.01 software (Agilent) were both used for GC–MS data analyses.The NIST (national institute of standards and technology) mass spectral library and an in-house mass spectral database established using authentic standards were used together for metabolite identification.Compound Discovery 3.1 software(Thermo-Fisher Scientific) and TraceFinder 3.3 software (Thermo-Fisher Scientific) were jointly used for LC–MS-based specialized metabolome data analysis,combining qualitative and quantitative analysis.Both an in-house library established using authentic standards and online databases including mzCloud (https://www.mzcloud.org/),HMDB (https://hmdb.ca/),and MassBank (http://www.massbank.jp/),were used for specialized metabolite identification.LipidSearch 5.0 software (Thermo-Fisher Scientific,Tokyo,Japan) was used for the identification of lipid molecular species.The metabolomic and lipidomic data were subjected to multivariate statistical analyses by SIMCA 13.0.3 software(Umetrics,Umea,Sweden)[16],such as principal component analysis(PCA),orthogonal partial least squares discrimination analysis (OPLS-DA).

2.4.Polyphenol and phytohormone quantification using LC–MS

Quantitative real-time PCR (qRT-PCR) analysis ofOsGF14bshowed that lines 4 and 6 showed high expression levels ofOsGF14b,and line 2 showed a medium expression level (Fig.S2).OPLS analysis of metabolomic data revealed that line 2 displayed a very different metabolic profile from line 4 and line 6 (Fig.2).For this reason,line 2 was not used in further experiments.Rice panicle samples from the wild type(Nip)and twoOsGF14b-overexpressing lines (OXGF14b-4 and OXGF14b-6) were used for polyphenol and hormone quantification by LC-MS (API4000,AB SCIEX).

Fig.2.Supervised multivariate statistical analysis of metabolomic data from rice panicle.Score plots of OPLS-DA derived from LC-MS data showed rice lipid metabolite differences between two genotypes with or without 24 h blast infection,including the wild-type japonica rice(Nip)and three OsGF14b gene-overexpressing transgenic lines(OXGF14b-2,OXGF14b-4,and OXGF14b-6).Score plots of OPLS-DA derived from LC-MS data show specialized metabolite differences between two genotypes with or without 24 h blast infection,including Nip,OXGF14b-2,OXGF14b-4,and OXGF14b-6 plants.Score plots of OPLS-DA derived from GC–MS data showed rice primary metabolite differences between two genotypes with or without 24 h blast infection,including Nip,OXGF14b-2,OXGF14b-4,and OXGF14b-6 plants.Numbers of metabolites identified by LC-MS and GC–MS.WOB,without blast infection;WB,with 24 h blast infection by M.oryzae.

Three replicates per sample were extracted for polyphenol determination as described in our previous report [17],and the extract was eluted following Stephens et al.[18].Polyphenol compounds were quantified in multiple-reaction monitoring (MRM)mode using optimized MS/MS conditions,listed in Table S2.

Three replicates per sample were extracted for quantifying the endogenous hormones auxin (IAA),JA,and SA in rice panicle samples following Pan et al.[19].JA,IAA,and SA were further quantified as described in our previous report[20].For full quantification of endogenous gibberellin acid (GA) content in rice panicle samples,GAs were extracted from the same samples with the solution and 17 internal standards described[21],and quantified by Nano-LC–MS(Bruker Daltonics,Bremen,Germany)in positive mode following Chen et al.[21].

2.5.qRT-PCR analysis

Total RNA of rice panicle samples,including the wild type and twoOsGF14b-overexpressing lines (OXGF14b-4 and OXGF14b-6),was extracted using TransZol Up(TransGen Biotech,Beijing,China)as described previously[6],and three replicates were analyzed for each sample.Gene-specific primers are listed in Table S3.

2.6.Histochemical staining and lignin analysis

Rice panicle samples,including the wild type and twoOsGF14boverexpression lines (OXGF14b-4 and OXGF14b-6),were fixed in formalin:acetic acid:alcohol(FAA)at 4°C,and sections were prepared following Jáuregui et al.[22].The sections were observed under bright-field lighting.Lignified tissue was stained bluegreen.Image acquisition was performed with Image-pro plus 6.0 software (Media Cybernetics,Inc.,Rockville,MD,USA).All experiments were repeated three times.

2.7.Phytohormone treatments and evaluation of disease resistance

To further investigate the effects of hormones on panicle blast resistance,whole rice panicles of the wild type and oneOsGF14boverexpressing line (OXGF14b-4) were wrapped with cotton,into which was injected 3–4 mL hormone solution or water solution as the control.The cotton was then wrapped with foil to maintain humidity.The cotton was removed from the panicle 24 h after treatment.The hormone-treated panicles were used for panicle blast inoculation as described [6].To analyze the effects of hormones on gene expression or other hormone accumulation,the same cotton-wrapping method was used and rice panicles were sampled at the same time point 24 h following blast infection or without blast infection.Both IAA and JA solutions were of 100 μmol L-1.All experiments were repeated twice.

3.Results

3.1.Changes in protein abundance associated with blast infection and OsGF14b-mediated panicle blast resistance

Of 49,044 unique peptides,7215 distinct proteins were identified as specific to rice panicle tissue,with an estimated FDR of 1%,of which 7130(98.8%)(Fig.S3;Table S1)were common to both biological replicate sets.We then checked for correlations in the protein abundances between genotypes and treatments to assess global responses to the pathogen challenge (Fig.S4).In OXGF14b vs.Nip comparisons of proteomic profiles,we observed no clear correlation (r=-0.02) (Fig.S4a;Tables S4-S5) in protein abundance between 24 h and infection and no-infection samples,indicating that many changes in protein expression were associated with genotype.However,comparisons of changes within individual genotypes between 24 h infection and without infection revealed a positive correlation (0.523) (Fig.S4b;Tables S6–S7),strongly suggesting that changes in the expression of many proteins in response to infection were similar between the two genotypes.More proteins were up-regulated than down-regulated in OXGF14b than in Nip plants after 24 h infection (Fig.1a;Table S4),or in OXGF14b plants without blast infection (Fig.1a;Table S6).

Among the altered proteins,235 were differentially abundant in both OXGF14b and Nip plants associated with 24 h infection,including 167 up-regulated and 68 down-regulated proteins(Fig.1a and b;Tables S6,S7),suggesting that those 235 regulated proteins were part of a common response to blast infection in both genotypes.However,108 proteins were up-regulated and 23 proteins were down-regulated exclusively in OXGF14b plants(Fig.1a and c;Table S6),revealing a set of proteins directly associated withOsGF14bexpression.A set of 169 down-regulated proteins were detected only in Nip plants,a finding that may also be closely associated with differences inOsGF14bexpressionmediated resistance to panicle blast (Fig.1a and c;Table S7).Finally,33 proteins were up-regulated and seven were downregulated in OXGF14b plants following 24 h of infection (Fig.1a and c;Table S4).

We further filtered these proteins to distinguish only those with a fold change>1.5 with both blast infection(OXGF14b_24 h infection vs.OXGF14b_without infection) andOsGF14b-mediated panicle blast resistance (OXGF14b_24 h infection vs.Nip_24 h infection).We identified 12 stress/defense-related proteins and 12 growth/photosynthesis-related proteins,including a cysteine proteinase inhibitor,a nonspecific lipid-transfer protein,a salt stress-induced protein,a cytochrome oxidase assembly protein,and photosystem I reaction center subunit V that were all significantly up-regulated in OXGF14b plants,in addition to 12 as-yet uncharacterized proteins (Table S8).GO enrichment analysis revealed that biological processes significantly enriched for upregulated proteins only in OXGF14b plants with 24 h infection included ‘‘oxidation–reduction process”,‘‘lipid transport”,‘‘response to oxidative stress”,‘‘defense response”,‘‘obsolete peroxidase reaction”,and ‘‘negative regulation of catalytic activity”(Fig.S5a and c).Five ribosomal proteins were down-regulated>2-fold in OXGF14b plants (Table S8),a finding consistent with the significant enrichment for down-regulated proteins in the ‘‘ribosome biogenesis translation”biological process in OXGF14b plants after 24 h blast infection (Fig.S5b and d).

KEGG pathway enrichment analysis of the above regulated proteins showed that up-regulated proteins were significantly enriched mainly in ‘‘ribosome”,‘‘photosynthesis-antenna proteins”,‘‘starch and sucrose metabolism”,and ‘‘pentose and glucuronate interconversions”(Table S9),while the down-regulated proteins were enriched mainly in the overlapping ‘‘phenylpropanoid biosynthesis pathway”and ‘‘flavonoid biosynthesis pathway”in OXGF14b plants,in comparison with Nip plants without blast inoculation (Table S10).However,up-regulated proteins were enriched only in ‘‘photosynthesis-antenna proteins”(Table S11),and down regulated proteins were enriched only in‘‘ribosome”proteins in OXGF14b plants compared to Nip plants with 24 h of infection (Table S12).Diterpenoid pathway proteins,upstream of phytoalexin and gibberellin acid (GA) biosynthesis[23],were up-regulated after 24 h infection in both OXGF14b and Nip plants (Tables S13,S14).For example,the abundances ofent-copalyl diphosphate synthase 2 (Q6Z5I0),momilactone A synthase (Q7FAE1),and gibberellin 2-beta-dioxygenase 2 (Q5W726)were significantly altered between the Nip and OXGF14b plants either with or without 24 h infection(Tables S4,S5).The‘‘phenylpropanoid biosynthesis pathway”was enriched in both the up-and down-regulated protein abundance lists of OXGF14b plants(Tables S13,S15),although this pathway was enriched only for up-regulated proteins in Nip plants (Tables S14,S16).Taken together,these proteomic results suggest the importance of the diterpenoid,phenylpropanoid,ribosome biogenesis,and photosynthesis pathways inOsGF14b-mediated rice panicle blast resistance.However,it is important to note that the expression levels of phenylpropanoid pathway proteins were not uniform throughout the pathway.

3.2.Changes in resistance-related metabolites detected in rice panicle

To determine the specific contributions of identified KEGG pathways toOsGF14b-mediated resistance,we next interrogated Nip plants and three OXGF14b transgenic lines (OXGF14b-2,OXGF14b-4,and OXGF14b-6) using GC–MS and LC–MS to identify differences in their metabolomic and lipidomic profiles associated withM.oryzaepanicle infection (Fig.S1).We generated a multiomics data matrix that included 157 features detected by GC–MS(Table S17),alongside 948 features detected in ESI+mode(Table S18) and 1363 features detected in ESI-mode (Table S19)by LC–MS-based metabolomics.We further obtained 704 features in ESI+mode (Table S20) and 444 features in ESI-mode by LC–MS-based lipidomics (Table S21).

A total of 73 metabolites in the GC–MS dataset (Fig.2d),88 metabolites in the LC-MS dataset (77 detected in ESI+mode and 24 detected in ESI-mode;Fig.2d),and 227 lipids (114 detected in ESI+mode and 115 in ESI-mode;Fig.2d)were identified based on comparison with standards or based on previously reported metabolite MS2spectral characteristics,including sugars,amino acids,organic acids,fatty acids,polyamines,polyphenols,phytoalexins,and phytohormones (Tables S17–S21).

OPLS-DA of the three omics datasets derived from GC–MS and LC–MS revealed major metabolomic changes between‘‘24 h infection”and ‘‘without infection”samples for both Nip and OXGF14b plants (Fig.2a–c).Among OXGF14b plants,lines 4 and 6 could be distinguished from line 2 without infection (Fig.2a–c),a finding consistent with the lowerOsGF14btranscriptional expression levels in line 2 (Fig.S2).S-plots generated from OPLS-DA models from each omics dataset (Figs.S6–S8) revealed that more lipids and specialized metabolites were increased both in the transgenic lines and the wild type with relatively few lipids and specialized metabolites showing a decrease associated with 24 h blast infection,and changes were more marked in the transgenic lines than in the wild type(Tables S22–S25).A similar,albeit less prominent,effect was also observed in the primary metabolites (Table S26).

Closer scrutiny of the lipidomic data revealed that,irrespective of the genotype,most of the up-regulated lipid molecule species after 24 h infection,including triacylglycerol (TG),diacylglycerol(DG),and phosphoglycerides (PC,PI,PE),contained greater amounts of polyunsaturated fatty acids,such as linolenic acid(LA;C18:3) (Fig.S9).Among them,the levels of TG(18:3/18:3/18:3),TG (16:0/18:3/18:3),TG (14:0/18:3/18:3),DG(18:3/18:3),DGDG (15:0/18:3),PI (18:3/18:3),PC (22:0/18:3),were much higher in OXGF14b plants than in Nip plants after 24 h infection (Fig.S9).These results suggest that LA also contributes to the resistance response to panicle infection as part of a coordinated regulation of specialized metabolites.Taken together,these metabolomic and lipidomic profiles suggest thatOsGF14b-mediated panicle blast resistance involves mainly specialized metabolite biosynthetic pathways.

3.3.OsGF14b modulates flux through the diterpenoid pathway during rice panicle blast infection

To test whether the changes in metabolites involved in the diterpenoid pathway,including the branches responsible for the biosynthesis of the phytohormone GA and phytoalexins,were in agreement with the proteomic results presented above,we next quantified phytoalexin and GA content.Pathogen inoculation significantly induced the accumulation of six phytoalexins:oryzalexin A,oryzalexin C,momilactone A,momilactone B,phytocassabes B,and phytocassabes C in both Nip and OXGF14b plants,based on their relative quantification result from the metabolomic data (Fig.3a).The contents of momilactone B and oryzalexin A were markedly higher in OXGF14b plants than in Nip plants both with and without 24 h blast infection,whereas lower contents of momilactone A,oryzalexin C,phytocassabes B,and phytocassabes C were observed in OXGF14b than in Nip plants after 24 h infection (Fig.3a).The relative content of momilactone A was markedly higher in OXGF14b plants without blast infection but lower after 24 h blast infection,compared to Nip plants(Fig.3a),suggesting that flux may be redirected to the GA branch inOsGF14b-overexpression plants.

We next quantified the absolute concentrations of GAs.As shown in Fig.3b,pathogen inoculation significantly induced the accumulation of 10 GAs:GA12,GA15,GA24,GA9,GA4,GA51,GA34,GA7,GA8,and GA5,but reduced the accumulation of GA53,GA44,GA20,and GA1both in wild-type and transgenic plants.Intriguingly,the contents of GA12,GA15,GA24,GA4,GA34,GA51,GA7,and GA9were markedly lower in transgenic than in wild-type plants before blast infection.Moreover,14 of 16 GAs were significantly accumulated in transgenic plants compared with wild-type plants after blast infection (Fig.3b).It has been reported [23] thatCPS1encodes anent-copalyl diphosphate(ent-CDP)synthase participating in the biosynthesis of GAs,but twoent-CDP synthase isoforms,CPS2 and CPS4,are involved in the biosynthesis of phytoalexins.To confirm the involvement of the GA signaling pathway and phytoalexin inOsGF14b-mediated panicle resistance at the transcript level,the expression levels of the genes encoding the six key enzymes for GA and phytoalexin biosynthesis were analyzed.Levels ofCPS1,CPS2andCPS4were increased 24 h after inoculation irrespective of genotype.However,the increase inCPS1was much higher,but those ofCPS2andCPS4considerably lower,in OXGF14b plants than in Nip plants(Fig.4a).These changes are in close agreement with the changes at metabolite and protein levels in the diterpenoid pathway (Figs.S10a,S11a).The changes of genes involved in phytoalexin and GA biosynthetic pathways,such as those for momilactone A synthase (MAS),also agreed closely with the determined metabolite levels (Fig.4a,S10b).Taken together,the results accumulated at the protein,metabolite and transcript level suggest thatOsGF14bmodulates the diterpenoid pathway by flux regulation during panicle blast inoculation in rice.In doing,so it partitions more flux toward the GAs branch and less to the phytoalexin branch.

3.4.OsGF14b-mediated panicle blast resistance involves reprogramming of the phenylpropanoid pathway

Our proteomic analysis revealed that the expression levels of several proteins,such as phenylalanine ammonia-lyases (PAL1,Q6K6Q1,B7EFQ4,and Q75HQ7),4-coumarate-CoA ligase 3(Q6ETN3),caffeoyl-CoA O-methyltransferase 1 (Q9XJ19) and cinnamoyl-CoA reductase 1 (Q6ERR4),involved in the phenylpropanoid pathway leading to lignin biosynthesis [25],were sharply reduced in OXGF14b plants compared with Nip plants either with or without 24 h blast infection(Fig.S11b).In agreement with the results of protein abundance,levels of metabolites involved in the lignin biosynthesis pathway,4-coumaric,4-hydroxycinnamic,ferulic,and vanillic acids,were also significantly reduced in OXGF14b plants after 24 h infection compared with Nip plants after 24 h infection (Fig.5 ;Tables S24,S25).As shown in Fig.5,metabolite levels in the phenylpropanoid pathway were in agreement with changes in both proteins and transcripts involved in this pathway(Fig.S10a and b),suggesting a strong correlation between metabolomic and proteomic abundances (Fig.S10).These results suggested the important role of lignin inOsGF14b-mediated panicle blast resistance.

To test this suggestion,some of the differentially abundant proteins that are key enzymes in lignin biosynthesis were selected for gene expression measurement.As shown in Fig.4b,the expression levels of twoPALgenes,PAL1andPAL2,were significantly suppressed in OXGF14b plants compared with Nip plants after 24 h infection,while the transcripts of two4-coumarate-CoA ligase(4CL) genes (4CL3and4CL5),twoperoxidase(PRX) genes (PRX1andPRX35),cationic peroxidase SPC4,cinnamoyl-CoA reductase 1(CCR1),cinnamyl alcohol dehydrogenase 2(CAD2),ferulic acid 5-hydroxylase 1(F5H1),andcaffeic acid/5-hydroxyferulic acid Omethyltransferase 1(COMT1) were more highly reduced in the transgenic plants than in the wild-type plants after 24 h infection,even thoughCCR1,PRX1,andPRX35showed higher transcription levels in transgenic than in wild-type plants without blast infection(Fig.4b).Benzidine blue staining of the stem cross sections of glumes showed that 24 h blast infection significantly induced the accumulation of lignin in wild-type but not in OXGF14b plants(Fig.4c and d).Taken together,these results indicate the important roles of lignin inOsGF14b-mediated panicle blast resistance.Unlike downstream phenylpropanoids,relative levels of the tested upstream phenylpropanoids such as flavonoids,iso-flavonoids,and flavonoid glucosides were increased in OXGF14b plants following 24 h of blast infection(Fig.5b).Absolute quantitation with authentic standards by MRM showed that results from the targeted analysis were consistent with those from the untargeted analysis(Fig.5c).

Fig.3.Changes of metabolites involved in the diterpenoid pathway response to rice panicle resistance.(a) Relative content of phytoalexin in the wild type (Nip) and two OsGF14b gene overexpressing transgenic lines(OXGF14b-4 and OXGF14b-6)with or without 24 h blast infection.(b)Absolute content of gibberellins(GAs)in Nip,OXGF14b-4,and OXGF14b-6 plants with or without 24 h blast infection.** represents the significant difference when P-value <0.01;* represents the significant difference when Pvalue <0.05.WOB,without blast infection;WB,with 24 h blast infection by M.oryzae.

Fig.4.Changes in expression levels of genes involved in diterpenoid and phenylpropanoid pathway responses to rice panicle infection.(a) Expression levels of key genes associated with the diterpenoid pathway in the wild type (Nip) and two OsGF14b gene-overexpressing transgenic lines (OXGF14b-4 and OXGF14b-6) with or without 24 h blast infection.(b) Expression levels of key genes associated with the phenylpropanoid pathway in Nip,OXGF14b-4,and OXGF14b-6 plants with or without 24 h blast infection.(c)Benzidine blue staining of stem cross sections of glumes from Nip and OXGF14b-4 plants with or without 24 h blast infection.(d)Relative content of lignin based on percent of stained area in Nip,OXGF14b-4,and OXGF14b-6 plants with or without 24 h blast infection.**represents significant difference when P-value<0.01;*represents significant difference when P-value <0.05.WOB,without blast infection;WB,with 24 h blast infection with M.oryzae.

Fig.5.Changes in metabolites involved in the phenylpropanoid-pathway response to rice panicle blast infection.(a)Model summarizing key metabolite changes involved in phenylpropanoid pathway with or without blast infection.Metabolites promoted by fungal invasion are shown in red boxes and those decreased by fungal invasion in green boxes.(b)Bar plots showing changes in relative metabolite levels between the wild type japonica rice(Nip)and two OsGF14b gene-overexpressing transgenic lines(OXGF14b-4 and OXGF14b-6).(c) Bar plots showing changes in absolute metabolite levels between the wild type japonica rice (Nip) and two OsGF14b gene-overexpressing transgenic lines (OXGF14b-4 and OXGF14b-6).** represents the significant difference when P-value <0.01;* represents the significant difference when P-value <0.05.WOB,without blast infection;WB,with 24 h blast infection by M.oryzae.

Fig.6.Effects of major phytohormones on rice panicle resistance.(a)Levels of three phytohormones(IAA,JA,and SA)and genes involved in their metabolism in the wild type(Nip) and two OsGF14b gene-overexpressing transgenic lines (OXGF14b-4 and OXGF14b-6) with or without 24 h blast infection.(b) Effect of exogenous IAA and JA on the panicle blast resistance phenotype in Nip and OXGF14b-4 plants 24 h after inoculation.(c) Effect of exogenous IAA and JA on the infected main axis length of Nip and OXGF14b-4 plants 24 h after inoculation.(d)Effects of exogenous IAA and JA on hormone accumulation in Nip and OXGF14b-4 plants 0,8,24,and 48 h after inoculation.(e)Effects of exogenous IAA and JA on pathogenesis-related (PR) gene expression levels in Nip and OXGF14b-4 plants 0,8,24,and 48 h after inoculation.** represents the significant difference when P-value<0.01;*represents the significant difference when P-value<0.05.WOB,without blast infection;WB,with 24 h blast infection by M.oryzae.

Beyond their roles in lignin biosynthesis,thePALgenes also encode key enzymes for SA biosynthesis,with phenylalanine also representing a SA precursor [6].The down-regulated expression of the PAL genes and the higher content of phenylalanine in OXGF14b plants suggested a hypothesis that the SA signaling pathway would also be suppressed in OXGF14b plants(Fig.4b;Fig.5a;Table S16).As expected,irrespective of genotype,the SA level was indeed decreased significantly after 24 h blast infection (Fig.6a).Lower SA levels were identified in OXGF14b plants than in Nip plants,whether or not they were injected (Fig.6a).

3.5.OsGF14b-mediated panicle blast resistance involves activation of IAA and JA signaling pathways

It has been reported[26]that reduced lignin levels result in elevated accumulation of JA,SA,and abscisic acid (ABA) whereas reduced deposition of cytokinin and IAA was associated with increased resistance to fungal infection in alfalfa(Medicago sativa).The altered accumulation of lignin and SA with 24 h infection in OXGF14b plants led us to speculate that the levels of other common phytohormones would also be altered in OXGF14b plants during blast infection.Quantification of endogenous levels of other hormones revealed that 24 h infection significantly increased the levels of JA and IAA,and that these levels were higher in OXGF14b plants than in Nip plants both with and without 24 h infection(Fig.6a).In agreement with the results of hormone quantification,the proteomic analysis indicated that pathogen infection strongly induced the expression of three lipoxygenase (LOX) enzymes(A3BUP8,LOX2 and LOX6) involved in JA biosynthesis in both Nip and OXGF14b plants,but that the abundances of these three proteins were significantly higher in OXGF14b plants than in Nip plants in the presence and absence of blast infection (Fig.S11c).The abundances of five proteins involved in the IAA signaling pathway also differed significantly between Nip and OXGF14b plants.As shown in Fig.S11c,flavin-containing monooxygenase 1(FMO1) involved in IAA synthesis was induced byM.oryzae,and its abundance was significantly higher in OXGF14b than in Nip plants both with and without 24 h infection.In contrast,the abundance of two Gretchen Hagen (GH3) proteins (GH3.8 and GH32),which function in reducing the free IAA level by conjugating amino acids to IAA [27],were markedly repressed in OXGF14b compared with Nip plants both with and without 24 h infection(Fig.6a).We also analyzed the expression of four genes encoding these proteins using qRT-PCR.The results showed good concordance with the proteomic data (Fig.6a),suggesting that riceOsGF14b-mediated disease resistance is associated with activation of the IAA and JA signaling pathways.

To further validate the positive roles of IAA and JA inOsGF14bmediated disease resistance,the panicles of Nip and OXGF14b plants were treated with exogenous IAA and JA before blast infection.Exogenous application of IAA and JA did not significantly reduce disease symptoms in wild-type plants.In contrast,both IAA-pretreated and JA-pretreated transgenic plants showed markedly increased disease resistance compared with non-treated OXGF14b plants(Fig.6b and c),suggesting that IAA and JA can help OXGF14b plants resist fungal blast at the heading stage.Interestingly,JA strongly induced the transcription of three pathogenesis-related genes (PR5,PR5-1,andPR10) in both Nip and OXGF14b plants,but the expression levels ofPRgenes were all higher in transgenic than in wild-type plants both with and without 24 h blast infection (Fig.S12).The transcription levels of the threePRgenes showed no apparent change in Nip plants after IAA treatment,whereas their expression was slightly reduced by IAA treatment in OXGF14b plants(Fig.S12).These results together indicate that IAA and JA can regulate the expression ofPRgenes in OXGF14b plants.

The simultaneous accumulation of JA and IAA led us to speculate that JA and IAA interact synergistically inOsGF14b-mediated panicle blast resistance.To test this hypothesis,we measured the expression of three JA synthesis-associated genes (LOX2,LOX6,andAOS2) in wild-type and transgenic plants after IAA treatment,and the expression of three IAA synthesis-associated genes(FMO1,Arabidopsis aldehyde oxidase 3(AAO3)andnitrilase1(NIT1))in wildtype and transgenic plants after JA treatment.IAA treatment did not influence the expression ofLOX2,LOX6,andAOS2,and JA treatment did not influence the expression ofFMO1,AAO3,andNIT1in either Nip or OXGF14b plants(Fig.S13).Similarly,the endogenous levels of JA were not significantly affected by IAA treatment and the endogenous levels of IAA were not significantly affected by JA treatment in either Nip or OXGF14b plants (Fig.6d).Because in a previous study,IAA increased the induction effect of JA on defense-related genes inArabidopsisresistance to necrotrophic pathogens [28],we also analyzed the expression ofPR5-1andPR10genes in transgenic panicles treated with 100 μmol L-1IAA,JA,or both.The induction ofPRgenes was not increased in panicles treated with both IAA and JA (Fig.6e).These observations suggested that JA and IAA do not interact with each other inOsGF14b-mediated panicle blast resistance in rice.

4.Discussion

Based on the above results,we present a working model ofOsGF14b(Fig.7).OsGF14b-mediated panicle blast resistance is involved in the activation of IAA and JA signaling pathways,accompanied by the reprogramming of the diterpenoid and phenylpropanoid pathways.First,OsGF14bmodulates the diterpenoid pathway through flux regulation during panicle blast inoculation in rice,resulting in a greater production of active GA4and reduced production of phytoalexins.Second,OsGF14baffects the accumulation of lignin precursors,and reduces the SA level,leading to panicle resistance.

Lignin,one of the endpoints of the phenylpropanoid pathway,is a major component of the plant cell wall and has been demonstrated to play a critical role in plant responses to pathogen attacks[24,28–35].In rice,a higher level of lignin was identified inlrd6-6mutants,which showed increased leaf blast and bacterial blight resistance,and increased lignin deposition at infection sites was found to be crucial for plant resistance toStriga hermonthica[24,36].Here,we identified expression changes in key genes involved in lignin biosynthesis at the transcript and protein levels,as well as changes of lignin content in theOsGF14b-overexpressing compared to wild-type plants after 24 h blast infection.These results suggest that lignin might play an important role in regulatingOsGF14b-mediated panicle blast resistance.This notion is seemingly in contrast to the positive role of lignin in modulating leaf blast resistance [36],indicating that different response mechanisms distinguish leaf and panicle blast resistance in rice.HCTantisense alfalfa plants that displayed reduced lignin levels also showed increased resistance to the fungusColletotrichum trifolii[25].Hormone levels were also modulated in theseHCTantisense plants,which displayed reduced IAA and GA but increased cytokinin,ABA,JA,and SA levels.In a recent study [34],silencing of cotton(Gossypium hirsutum)GhLac1expression resulted in low lignin levels,causing the accumulation of JA and specialized metabolites and conferring resistance toVerticillium dahliaeand cotton bollworm.Taken together,these reports imply an intimate connection between the lignin and the phytohormone signaling pathways in plant defense responses.We also identified higher levels of JA and IAA but lower levels of SA inOsGF14b-overexpressing plants than in wild-type plants with 24 h after blast infection.The lignin contents were also changed inOsGF14b-overexpressing plants compared to wild-type plants after 24 h blast infection,further supporting the interconnection of the lignin and plant hormone signaling pathways.

Fig.7.A working model of the role of OsGF14b in rice panicle resistance. OsGF14b-mediated panicle blast resistance is involved in the activation of IAA and JA signaling pathways,accompanied by the reprogramming of the diterpenoid and phenylpropanoid pathway.Red arrows denote increases and green arrows decreases in levels of metabolites or gene expression.

Recent studies [37–39] indicate that,in addition to the wellknown defense-related hormones(SA,JA,and ethylene),other hormones such as IAA,GA,cytokinin(CK),and ABA are also implicated in plant defense signaling pathways.A growing body of evidence has shown that the behavior of IAA in plant–pathogen interactions is complex and IAA appears to affect resistance differently depending on the pathogen.In most cases,IAA plays a negative role in regulating plant resistance to biotrophs via antagonistic crosstalk with SA [38–42] whereas IAA positively regulates plant resistance to necrotrophs via interaction with JA[27,43].However,in our study,a series of results suggest that IAA plays a positive role inOsGF14b-mediated panicle blast resistance in rice.First,the expression levels of the IAA synthesis-related proteins (genes)Q7XWZ6,B9G099,and Q6K306 were significantly higher inOsGF14b-overexpressing plants than in wild-type plants both with and without blast infection.Second,the endogenous concentrations of immediate products of IAA,tryptophan,and free IAA were all higher inOsGF14b-overexpressing plants than in wild-type plants both with and without 24 h blast infection.Third,exogenous application of IAA increased panicle blast resistance inOsGF14b-overexpressing plants but not in wild-type plants.Fourth,the transcription ofOsGF14bwas significantly induced after IAA treatment in both wild-type andOsGF14b-overexpressing plants.Taken together,these results support the proposal that the phytohormone IAA positively regulatesOsGF14b-mediated panicle blast resistance.However,it should be pointed out that the positive role of IAA in rice blast we identified in this study was identified in panicle tissues,whereas the previously reported negative role of IAA in blast was identified in leaf tissues[4].Given that IAA is distributed differently among plant tissues[44],these results suggest that the function of IAA in rice blast may also be dependent on the specific plant tissues in which the disease occurs.

OsGF14b-mediated panicle blast resistance also involved activation of the IAA signaling pathway and suppression of the SA signaling pathway.In contrast to IAA levels,endogenous levels of SA were significantly lower inOsGF14b-overexpressing plants than in wild-type plants both before and after pathogen inoculation,a finding consistent with negative crosstalk between SA and IAA pathways in plant defense responses.Given that the IAA and JA pathways usually interact positively with each other in plant disease resistance,we propose the hypothesis that positive crosstalk between these phytohormones occurs inOsGF14b-mediated panicle blast resistance.Given that previous reports [45,46] indicated that IAA can induce the expression of the JA biosynthesis geneAOC4and that methyl jasmonate can induce the biosynthesis of IAA by activation of IAA biosynthesis-related genes (ASA1,NIT3,andYUCCA2) inArabidopsis,a set of experiments were conducted to test whether IAA and JA could influence the biosynthesis of each other in both wild-type andOsGF14b-overexpressing plants.However,IAA biosynthesis was not influenced by JA treatment and JA biosynthesis was not influenced by IAA treatment,suggesting that IAA and JA signaling pathways function independently inOsGF14bmediated panicle blast resistance.However,we did not investigate the effects of IAA on other aspects of the JA pathway(such as transport,distribution,and signaling),or the effects of JA on other aspects of the IAA pathway.Thus,we cannot exclude the possibility that IAA and JA signaling pathways interact positively in aspects other than hormone biosynthesis inOsGF14b-mediated panicle blast resistance.Marked changes in GA content were also identified,especially for the bioactive GA1 and GA4.The levels of GA4 were strongly induced,while those of GA1 were reduced by blast inoculation in both the wild-type andOsGF14b-overexpressing plants.However,the concentrations of GA4 and GA1 were both higher inOsGF14b-overexpressing plants than in wild-type plants after blast infection,suggesting the positive role of GA in regulating rice panicle blast resistance.This result is in contrast to the negative roles of GA in plant resistance toM.oryzaein leaf tissue [47–49].Yang et al.[48]also reported that overexpression ofElongated uppermost internode(Eui) which encodes a GA-inactivating enzyme,led to increased levels of JA in rice plants.In our study,higher levels of both GA and JA were observed in panicles of the transgenic plants than in wild-type plants.Similarly to IAA,these differences may also imply differing mechanisms between leaf and panicle blast resistance in rice.

In summary,we have catalogued proteomic,metabolomic,lipidomic,and phytohormonal changes inOsGF14b-mediated panicle blast resistance.We propose thatOsGF14b-mediated blast resistance involves the activation of auxin and JA signaling pathways,accompanied by the reprogramming of the phenylpropanoid and diterpenoid pathways.The roles of GA,lignin,and IAA may differ between leaf and panicle blast resistance.Their specific roles in disease resistance may be plant tissue-dependent.These results provide additional leads for identifying components of the resistance response of the rice panicle to blast inoculation.

Data availability

All proteomic raw data and search results have been deposited in the iProX system (http://www.iprox.org/index) with the identifier ShijuanYan.The raw files may be accessed with user IDs IPX0001997000 and PXD017967.All metabolomic and lipidomic raw data have been deposited in the Metabolomics Workbench(https://www.metabolomicsworkbench.org/).The raw files may be accessed with DataTrackIDs ID1941 and ID1942,respectively.

CRediT authorship contribution statement

Shijuan Yan and Bin Liu:conceived the project.Qing Liu and Jianyuan Yang:performed blast inoculation and hormone treatment experiments.Wenjie Huang,Mengyu Chen,and Xuan Li:performed metabolomic and lipidomic experiments.Shijuan Yan:analyzed the metabolomic data.Mengyu Chen and Qian Kong:performed the proteomic experiment.Thomas Naake and Sheng Zhang:analyzed the proteomic data.Wenyan Li and Qinjian Liu:performed gene expression analysis.Shijuan Yan and Alisdair R.Fernie:integrated the multi-omics data.ShijuanYan and Qing Liu:wrote the manuscript.Shijuan Yan,Alisdair R.Fernie,and Bin Liu:revised the manuscript.

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 Dr.Jinglong Bei for providing technical support for the mass spectrometry analysis of proteins.This work was jointly supported by the National Transgenic Science and Technology Program(2019ZX08010003),National Key Research and Development Program of China (2018YFD0200302),the Special Fund for Scientific Innovation Strategy-Construction of High Level Academy of Agriculture Science (R2020PY-JX019 and R2020PY-JX001),the Presidential Foundation of Guangdong Academy of Agricultural Sciences (201611),the Innovation Team Project of Guangdong Modern Agricultural Industrial System (2018LM2150 and 2019KJ106),and the Science and Technology Program of Guangdong Province (2015A020209077).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.10.007.