Ectopic expression of a male fertility gene,LOGL8,represses LOG and hinders panicle and ovule development

Shuifu Chen,Sulin Lou,Xiucai Zhao,Shijuan Zhang,Letian Chen,Ping Huang,Guane Li,Yingying Li,Yao-Guang Liu,*,Yuanling Chen,*

a State Key Laboratory for Conservation and Utilization of Subtropical Agro-bioresources,South China Agricultural University,Guangzhou 510642,Guangdong,China

b Key Laboratory of Plant Functional Genomics and Biotechnology of Guangdong Provincial Higher Education Institutions,Guangzhou 510642,Guangdong,China

c Guangdong Provincial Key Laboratory of Protein Function and Regulation in Agricultural Organisms,South China Agricultural University,Guangzhou 510642,Guangdong,China

d College of Life Sciences,South China Agricultural University,Guangzhou 510642,Guangdong,China

e College of Life Sciences and Oceanography,Shenzhen University,Shenzhen 518060,Guangdong,China

f Shenzhen Key Laboratory of Marine Bioresource and Eco-Environmental Science,Shenzhen 518060,Guangdong,China

g Guangdong Provincial Key Laboratory for Plant Epigenetics,Shenzhen 518060,Guangdong,China

Keywords:Cytokinin-activating enzyme LONELY GUY LOGL8 Pollen Ovule

ABSTRACT Grain number and seed-setting rate are components of crop yield.Cytokinin influences grain yield.However,emerging studies suggest that high cytokinin signals often lead to reduced branching or seed-setting rate,leading to reduced grain yield,although the mechanisms remain unclear.In this study,we identified and characterized the rice(Oryza sativa L.)gene LONELY GUY-LIKE 8(LOGL8),based on analysis of the LOGL8-pm(promoter mutant of LOGL8)mutant,which harbors a T-DNA insertion in the promoter of this gene.The mutation in LOGL8-pm causes ectopic hyperexpression of LOGL8 in inflorescence organs,resulting in plants with smaller panicles and defective ovules lacking archesporial cells and integuments.Knockout of LOGL8 caused pollen abortion,leading to a reduced seed-setting rate.LOGL8 encodes a putative cytokinin-activating enzyme.Our results showed that LOGL8 directly catalyzes the biosynthesis of bioactive cytokinins.Therefore,we propose that the ectopic expression of LOGL8 disrupts cytokinin spatiotemporal distribution and causes inhibition of LONELY GUY(LOG),which affects panicle branching and female organ development.These findings reveal the important role of LOGL8 in male development,and highlight the delicate balance of local cytokinin levels during panicle branching and female organ development.

1.Introduction

Panicle traits such as grain number per panicle and seed-setting rate are determinants of yield in cereals.Grain number per panicle is determined mainly by the number and activity of lateral branch meristems initiated from inflorescence meristems.Seed-setting rate depends primarily on the proper function of the reproductive organs needed to accomplish fertilization[1,2].Elucidating the regulatory mechanisms of panicle development is important for genetic improvement of yield in crops.

The phytohormone cytokinin functions in panicle development by promoting cell proliferation and differentiation in the inflorescence meristem,gynoecium development,and male and female gametogenesis[3,4].The homeostasis of cytokinin levels in planta is maintained by the dynamic balance of cytokinin biosynthesis,reversible conjugation,irreversible degradation,and activation.Cytokinins are synthesized mainly by isopentenyl transferase and LONELY GUY(LOG)-family enzymes,which catalyze the first rate-limiting step and the final activation step,respectively[5,6].Cytokinin oxidase/dehydrogenase(CKX)enzymes carry out the irreversible degradation of cytokinins via cleavage of the cytokinin side chain[7].Histidine kinase(HK)receptors perceive cytokinin signals,leading to the histidine phosphotransferase-mediated activation of cytokinin response regulator(RR)genes,which in turn regulate the expression of downstream cytokinin-responsive genes[8].

Several cytokinin-associated genes that determine panicle branching have been characterized in rice.Grain Number 1a(GN1a)encodes OsCKX2,which degrades cytokinin.Reduced OsCKX2 expression leads to the accumulation of cytokinin in the inflorescence meristem,thereby increasing spikelet number per panicle[9].Both the F-box protein LARGER PANICLE(LP)[10]and the zinc-finger transcription factor DROUGHT AND SALT TOLERANCE(DST)[11]negatively regulate panicle branching and spikelet number per panicle by activating OsCKX2 expression,while the Dof transcriptional activator Short Panicle 3(SP3)positively regulates panicle traits by modulating cytokinin levels[12].

Cytokinin-associated genes affect reproductive organogenesis and fertility.In Arabidopsis thaliana,CYTOKININ INSENSITIVE(CKI)encodes an HK that invokes the cytokinin response.Loss of function of CKI results in female gametophytic lethality[13].AtRR1 and AtRR10 directly activate the carpel identity-defining gene AGAMOUS(AG)to control carpel formation[14].Cytokinin also affects male fertility[15-17].In maize(Zea mays),overaccumulation of ZmCKX1 transcript in male organs results in male sterility[18].In rice,mutation of OsRR24 causes male sterility resulting from the defective meiosis of pollen mother cells[19].However,our knowledge of the regulation of male development by cytokinin-mediated processes is limited.

LOG-family(LOG-like,LOGL)proteins are thought to fine-tune the local distribution and concentrations of bioactive cytokinins via their tissue/organ-specific expression patterns[20].In Arabidopsis,AtLOG4 participates in the activation of cytokinins in the epidermis of the shoot apical meristem(SAM)to define the expression domain of WUSCHEL(WUS)[21].In rice,LOG is expressed mainly in the central tips of the SAM,panicle branch meristem,floral meristem,and epidermis of the ovule primordium.Null mutation of LOG severely reduced panicle size and hindered floral organ development owing to premature termination of meristem activity[6].Ovules of the weak mutant log-3 undergo abortion and lack archesporial cells and integuments[22].

Although rice contains 11 LOG-like homologs,few of them have been functionally studied.Rice An-2/LABA-1/LOGL6 is expressed in the epidermis of the awn primordium and controls awn elongation and panicle branching[23,24].Recent studies[12,23,25,26]have shown that elevating cytokinin signaling by overexpressing LOG,LOGL5,LOGL6,LOGL10,or SP3 or knocking out OsCKX11 leads to similar phenotype of small panicles and/or a low seed-setting rate,but the underlying mechanism remains unclear.We previously[27]identified a rice T-DNA insertion mutant with small panicles and reduced seed-setting rate but normal male fertility,yet the causal gene was not identified.In the present work,we determined that the T-DNA tagged in the promoter of a LOG-like gene LOGL8,and accordingly named the mutant LOGL8-pm(promoter mutant of LOGL8).To elucidate the mechanism underlying the impaired panicle branching and female development in LOGL8-pm,we first confirmed the causal gene via genetic experiments of mutant reoccurrence and overexpressing LOGL8;we further determined the biological function of LOGL8 by creating gene-knockout mutants,analyzing the spatiotemporal expression pattern of LOGL8,and testing the enzymatic activity of LOGL8 proteins.Furthermore,we analyzed the potential role of LOG on the malformed development in LOGL8-pm.Our findings reveal the important role of LOGL8 in male development,and highlight the delicate balance of local cytokinin levels during panicle branching and female organ development.

2.Materials and methods

2.1.Plant materials

The LOGL8-pm mutant was isolated from a rice T-DNA insertion population derived from a japonica cultivar Zhonghua 11(ZH11).The plants were grown in a greenhouse or paddy field at South China Agricultural University(Guangzhou,China)under normal growing conditions.

2.2.Isolation of the T-DNA flanking sequence

The T-DNA flanking sequence and insertion site in the LOGL8-pm mutant were identified by thermal asymmetric interlaced(TAIL)PCR[28].The primers used for TAIL PCR are listed in Table S1.

2.3.Sequence alignment and phylogenetic analysis of LOG-family proteins

The full-length amino acid sequences of LOG-family proteins were aligned with BioXM 2.7.120(https://cbi.njau.edu.cn/BioXM/).A phylogenetic tree was constructed with MEGA 5.05(https://www.megasoftware.net/)by the neighbor-joining method.

2.4.Histological observation and RNA in situ hybridization

Floral organs were examined and photographed under a dissecting microscope(Olympus,Tokyo,Japan).Mature pollen grains were stained with 1% I2-KI solution and observed under a light microscope.Developing panicles and spikelets were fixed in 0.1 mol L-1phosphate solution(pH 7.0)with 4%(w/v)paraformaldehyde and 0.25%glutaraldehyde for 16 h at 4°C,dehydrated through an ethanol gradient,and embedded into Paraplast Plus(Sigma-Aldrich,St.Louis,MO,USA).Each sample was cut into 8-μm-thick sections using a rotary microtome(Leica,Weztlar,Germany),and stained pollen grains and sections were imaged under an inverted microscope(Zeiss,Oberkochen,Germany).To observe ovule structure,the sections were stained with 0.25% toluidine blue.

Specific sequences of LOGL8 and LOG cDNA(293 and 897 bp,respectively)were cloned into the pGEM-T easy vector(Promega,Fitchburg,WI,USA)using the primers listed in Table S1.In vitro synthesis of RNA probes was performed using a DIG RNA Labeling Kit(Roche,Basle,Switzerland).Probe signals were detected using a DIG Nucleic Acid Detection Kit(Roche).RNA in situ hybridization was performed as described by Lou et al.[29].

2.5.RNA extraction and qRT-PCR

Total RNA was extracted from various rice tissues with TRIzol reagent(Thermo Fisher Scientific,Waltham,MA,USA),and reverse-transcribed into cDNA using M-MLV Reverse Transcriptase(Promega).qRT-PCR was performed using an iQ SYBR Green Supermix kit(Bio-rad,Hercules,CA,USA)and the Real-time PCR Detection System(Bio-rad)with three biological replicates.ACTIN 1(LOC_Os03g50885)was used as an internal reference.Relative expression was calculated by the 2-ΔΔCT and 2-ΔCT methods[30].The primers used for qRT-PCR are listed in Table S1.

2.6.Vector construction and rice transformation

Recombinant vectors for rice transformation were constructed using the Gibson Assembly method[31].A 5776-bp genomic DNA fragment of LOGL8 including the 1451-bp region upstream from the start codon,3914-bp genomic coding sequence,and 411-bp region downstream from the stop codon was cloned into the binary vector pCAMBIA1305.1 to generate the recurrent mutant construct.The coding sequence of LOGL8 driven by the maize Ubiquitin promoter was used to generate the overexpression vector.The design of primers for the target sequence and a vector for CRISPR/Cas9-mediated knockout of LOGL8 expression,and the decoding of targeted mutation events,were performed with CRISPR-GE(http://skl.scau.edu.cn/)[32].The CRISPR/Cas9 binary vector pYLCRISPR/Cas9Pubi-H was constructed as described by Ma et al.[33].The resulting recombinant constructs were introduced into ZH11 by Agrobacterium-mediated transformation[34].The primers used to generate the transformation vectors are listed in Table S1.

2.7.Subcellular localization

The coding sequence of LOGL8 was cloned into the N terminus of the green fluorescent protein(GFP)gene in the pLYd1GFP vector.The recombinant vector was co-expressed with the P35S::mCherry construct or the nucleus-localized construct P35S::mCherry-NLS in rice protoplasts.The fluorescent signals of the transformed cells were imaged under a confocal microscope(Zeiss).The primers used to generate the subcellular localization vectors are listed in Table S1.

2.8.Detection of cytokinin-activating enzyme activity

The coding sequences of LOG and LOGL8 were cloned into the pCold-I vector and used to express His-tagged recombinant proteins.The primers used to create prokaryotic expression vectors are listed in Table S1.The BL21(DE3)strain was used as the E.coli host.Recombinant protein expression in E.coli in LB medium was induced by adding 1 mol L-1isopropyl-β-D-thiogalactoside(IPTG)and incubating for 24 h at 15 °C.The E.coli cells were then subjected to ultrasonication.The recombinant proteins(LOG-His and LOGL8-His)were purified from cell-free extracts using Ni-NTA Superflow(Qiagen,Hilden,Germany)resin according to the manufacturer’s protocol.

The 100μL reaction mixture,including 3μg purified LOG or LOGL8 proteins,50μmol L-1iPRMP substrate,50μmol L-1Tris-HCl(pH 7.0),and 1μmol L-1MgCl2,was incubated at 30 °C for 2 h.After addition of 10μL 20% acetic acid to terminate the reaction,the reaction mixture was centrifuged at 16,000×g at room temperature for 20 min.The supernatant was filtered through a 0.22-μm micropore membrane filter.The reaction products were separated and monitored on a reverse-phase column(Waters,Milford,MA,USA)using high-performance liquid chromatography(HPLC)(Waters).The detailed steps were performed as described in Kuroha et al.[35].

2.9.Quantification of cytokinins

Fig.1.The LOGL8-pm mutant shows reduced branching and female sterility.(A)The structure of LOGL8/LOC_Os05g51390 and the position of the T-DNA insertion in its upstream promoter region.The black and white boxes indicate the coding sequence and untranslated regions,respectively.(B)Plant architectures of wild type(WT)and LOGL8-pm.Scale bar,15 cm.(C)Panicle architectures of WT and LOGL8-pm.Scale bar,5 cm.(D-H)LOGL8-pm shows dramatically reduced primary and secondary branch number per panicle,spikelet number per panicle,and seed-setting rate,compared to WT.Data are mean±SD(*,P＜0.05;**,P＜0.01,n=10).(I)Mature pollen grains of WT and LOGL8-pm.The pollen grains were stained with 1%I2-KI solution.Like those of the WT,LOGL8-pm pollen grains were fertile.Scale bars,50μm.(J)Ovules of WT and LOGL8-pm.The LOGL8-pm ovule appears as a nucellar-like cell clump without an archesporial cell or integuments.ar,archesporial cell;ii,inner integument;nc,nucellus-like cell clump;oi,outer integument.Scale bars,50μm.(K)Frequency of ovules with or without archesporial cells in WT and LOGL8-pm plants(n=100).ar,archesporial cell.

Cytokinins were extracted from young leaves of 30-day-old seedlings.Endogenous cytokinin levels were quantified by HPLC-MS/MS.The experiment was performed in three biological replicates.Extraction and measurement steps were performed by Greensword Creation Technology Co.Ltd.(Wuhan,Hubei,China).

3.Results

3.1.Characterization of LOGL8-pm

We previously identified a T-DNA insertion mutant with reduced panicle branches and seed-setting rate in rice[27].The T-DNA insertion site is identified to be located 1451 bp upstream of the start codon of LOC_Os05g51390 by TAIL PCR[28].This gene,which is named LOGL8,is the nearest homolog of LOG and is predicted to encode a putative cytokinin-activating enzyme(Figs.S1,S2).Based on the location of the T-DNA in the promoter region,we designated the mutant as LOGL8-pm(Fig.1A).

The LOGL8-pm mutant showed fewer tillers than the wild type(WT;Fig.1B),as well as small panicles(Fig.1C)with fewer panicle branches and spikelets per panicle(Fig.1D-G).The mean seedsetting rate of LOGL8-pm was dramatically reduced to 43.5%,compared with 82.0%in WT(Fig.1H).Given that the pollen viability of LOGL8-pm was normal(Fig.1I),the reduced seed-setting rate of this mutant might be attributed to abnormal female organ development.Further histological investigation confirmed that most ovules of the mutant failed to develop archesporial cells and integuments,instead forming nucellus-like cell clumps(Fig.1J-K).

3.2.Confirmation of the causal gene

To determine whether LOGL8 was indeed the causal gene of LOGL8-pm,we introduced a rice LOGL8 genomic sequence containing the 1451-bp upstream promoter regulatory region,the entire exon-intron region(3914-bp),and a 411-bp downstream region into WT plants.Ten independent transgenic lines were generated,and they showed markedly reduced tiller number(Fig.2A),primary and secondary branch numbers per panicle(Fig.2B-E),spikelet number per panicle(Fig.2F),and seed-setting rate(Fig.2G).The transgenic plants showed normal pollen viability(Fig.2H)and nucellus-like ovules lacking archesporial cells and integuments(Fig.2I),resembling the phenotypes of LOGL8-pm.We accordingly designated these lines as LOGL8-pm-recurrent(LOGL8-pm-R).We hypothesized that the T-DNA insertion disrupted the upstream regulatory region of LOGL8 and impaired its expression pattern,resulting in the abnormal phenotypes of the mutant.

3.3.The LOGL8-pm mutation causes ectopic hyperexpression of LOGL8 in inflorescence organs

To find out whether the spatiotemporal expression pattern of LOGL8 is different between in WT and LOGL8-pm plants,we performed qRT-PCR analysis of various plant organs including roots,leaf blades,leaf sheaths,developing panicles,and anthers.The result showed that LOGL8 was expressed predominantly in WT anthers(Fig.3A),which is consistent with the information in the Rice Expression Profile Database(https://ricexpro.dna.affrc.go.jp/)(Fig.S3).LOGL8 was expressed in WT anthers mainly during the early microspore stage(Fig.S4).However,LOGL8 was expressed at higher levels in all tested organs of LOGL8-pm plants(Fig.3A).

To compare the expression patterns of LOGL8 between the WT and LOGL8-pm in more detail,we performed RNA in situ hybridization of LOGL8 transcript in young panicles and anthers.In agreement with the qRT-PCR results,in the WT,low levels of LOGL8 transcript were detected in most tissues examined,but clear signals were observed in the tapetum and tetrads of anthers(Fig.3B-F).By contrast,in LOGL8-pm,substantially stronger LOGL8 transcript signals were detected in all tissues examined(Fig.3B-F).Similarly,the expression level of LOGL8 in LOGL8-pm-R was drastically upregulated(Fig.S5).

Fig.2.Recurrence of the LOGL8-pm phenotype.(A)Plant architectures of WT and LOGL8-pm-recurrent plants(LOGL8-pm-R).The LOGL8-pm-R construct contained a 5776-bp genomic sequence of LOGL8(including the 1451-bp upstream promoter region,3914-bp genomic coding sequence,and 411-bp downstream region).Scale bar,15 cm.(B)Panicle architectures of WT and LOGL8-pm-R.Scale bar,5 cm.(C-G)Compared to WT,LOGL8-pm-R plants showed markedly reduced primary and secondary branch number per panicle,spikelet number per panicle,and seed-setting rate,similar to LOGL8-pm.Values are mean±SD(**,P＜0.01,n=10).(H)Mature pollen grains of WT and LOGL8-pm-R.The pollen grains were stained with 1%I2-KI solution.Similar to those of WT,LOGL8-pm-R pollen grains were fertile.Scale bars,50μm.(I)Ovules of WT and LOGL8-pm-R.Compared to WT,LOGL8-pm-R ovule showed a nucellus-like cell clump without an archesporial cell or integuments,similar to LOGL8-pm ovule.ar,archesporial cell;ii,inner integument;nc,nucellus-like cell clump;oi,outer integument.Scale bars,50μm.

Fig.3.Expression pattern of LOGL8.(A)Comparison of LOGL8 expression in various tissues via qRT-PCR.ACTIN 1 served as an internal reference.Values are mean±SD(**,P＜0.01,n=3).(B-F)In situ hybridization of LOGL8 transcripts in young panicles,ovaries,and anthers of WT(top)and LOGL8-pm(bottom).(B)Primary branch primordium stage.(C)Secondary branch primordium stage.(D)Florets at the ovule primordium stage.(E)Ovaries at the megaspore mother cell stage.(F)Anthers at the meiosis stage.In WT,a strong expression signal was observed in anthers,whereas no signals were detected in panicles or ovaries.However,in LOGL8-pm,strong signals were generally detected in panicle branches,ovule primordia,carpels,stamens,and anthers.ca,carpel;e,epidermis;ov,ovule;pb,primary branch;sb,secondary branch;st,stamen;sti,stigma;t,tapetum;tds,tetrads.Scale bars,100μm.

We accordingly speculated that the mutant phenotype was probably caused by ectopic expression of LOGL8 in LOGL8-pm and LOGL8-pm-R,so we generated transgenic rice plants overexpressing LOGL8 driven by the maize Ubiquitin promoter.The phenotypes of these plants indeed mimicked the phenotypes of LOGL8-pm(Fig.S6).

3.4.LOGL8 primarily functions in pollen development rather than in panicle branching or ovule development

To investigate the role of LOGL8 in plant growth and development,we constructed LOGL8 knockout mutants in a background of a japonica cultivar ZH11 via CRISPR/Cas9-mediated genome editing[33].The editing site was designed to target the fifth exon of LOGL8(Fig.S1).Three T0lines with biallelic frameshift mutations in LOGL8(disrupting the encoding domain of the cytokininactivating enzyme)were generated(Table S1).Compared to WT plants,the LOGL8 knockout mutants(T1)showed markedly reduced seed-setting rate due to malformed anthers containing partially stained pollen grains(indicating pollen abortion)but normal ovule development(Fig.4A-F).No significant differences in panicle length,primary and secondary branches per panicle,or spikelets per panicle were observed between WT and mutant plants(Fig.4G-J).Thus,LOGL8 functions in male development rather than in panicle branching or female organ development.

3.5.LOGL8 has cytokinin-activating enzyme activity,like LOG

To identify the biological function of LOGL8,we investigated its subcellular localization.We fused the coding sequence of LOGL8 with the green fluorescent protein(GFP)gene at the N terminus and transformed rice protoplasts with this P35S::LOGL8-GFP construct and P35S::mCherry or P35S::mCherry-NLS,encoding a nucleus-localized protein.The LOGL8-GFP fusion protein localized to the cytosol and nucleus(Fig.5A).

A region of LOGL8 shares high amino acid sequence similarity with the core cytokinin-activating domain of LOG(Fig.S2B).We accordingly examined whether LOGL8,like LOG,has phosphoribohydrolase enzyme activity,which converts cytokinin riboside 5′-monophosphates(such as iP riboside 5′-monophosphate,iPRMP)to their bioactive forms(such as isopentenyladenine,iP)[6].We performed an in vitro enzyme assay using His-tagged recombinant LOG and LOGL8 proteins.Similarly to LOG,LOGL8 was capable of catalyzing the conversion of iPRMP to iP(Fig.5B).

To confirm the effect of the disrupted expression pattern of LOGL8 in vivo,we measured endogenous cytokinin concentrations in young leaves of 30-day-old seedlings.Compared to the WT,the levels of cZR(cZ riboside),DHZR(DHZ riboside),iPR(iP riboside),tZR(tZ riboside)and their bioactive forms(cZ,cis-zeatin;DHZ,dihydrozeatin)were significantly reduced in LOGL8-pm(Fig.5C).We also measured the expression levels of cytokininresponsive OsRR genes in the seedlings.Five of the nine OsRR family genes were significantly downregulated in LOGL8-pm compared to WT(Fig.5D).

Fig.4.Knockout of LOGL8 causes pollen sterility.(A)Plant architectures of WT and LOGL8-knockout(logl8-1)plants.Scale bar,15 cm.(B)Panicle architectures of WT and logl8-1.Scale bar,5 cm.(C)Florets of WT and logl8-1.Scale bars,1 mm.(D)Mature pollen grains of WT and logl8-1.The pollen grains were stained with 1% I2-KI solution.Compared to WT,logl8-1 showed pollen abortion,as revealed by shallow staining.Scale bars,50μm.(E)Ovules of WT and logl8-1.No visible differences in ovules were observed between WT and logl8-1.ar,archesporial cell;ii,inner integument;oi,outer integument.Scale bars,50μm.(F-J)Comparison of seed-setting rate(F),panicle length(G),primary and secondary branch number per panicle(H-I),and spikelet number per panicle(J)between WT and logl8-1.Values are mean±SD(**,P＜0.01,n=10).

These results indicate that LOGL8 has cytokinin-activating activity that directly controls the biosynthesis of bioactive cytokinins.Disrupting the spatiotemporal expression of LOGL8 via a T-DNA insertion(or transgenic overexpression)altered the distribution of bioactive cytokinins.

3.6.Ectopic expression of LOGL8 induced inhibition of LOG activity in LOGL8-pm

Local accumulation of cytokinin is necessary for cell fate specification and meristematic activity[36,37].LOG positively controls panicle branching and ovule development by determining the local cytokinin contents in the central tip of the panicle branch primordium and in the epidermis of ovule primordium[6,22],which are located adjacent to the ectopically expressed regions of LOGL8 in LOGL8-pm.To investigate the relationship between LOGL8 and LOG and their roles in regulating cytokinin levels,we performed qRT-PCR analysis of LOG transcripts in young panicles at several developmental stages.LOG was dramatically downregulated in these tissues in LOGL8-pm relative to the WT(Fig.6A).

The reduced LOG expression in young panicles and ovules of LOGL8-pm was confirmed by RNA in situ hybridization(Fig.6BE).In the WT,strong LOG transcript signals were detected in the central tips of primary and secondary branch primordia(Fig.6BC),the ovule epidermis,and the inner side of the carpel opposite the ovule(Fig.6D).Even during the archespore and integument formation stage,strong signals still persisted on the inner side of the carpel,while reduced signals were observed in the ovule(Fig.6E).By contrast,much weaker LOG transcript signals were detected in all corresponding tissues of LOGL8-pm(Fig.6B-E).These results suggest that an inhibition of LOG expression due to the LOGL8-triggered misdistribution of active cytokinins is responsible for the smaller panicles and malformed ovules in LOGL8-pm.

4.Discussion

Cytokinins play crucial roles in plant development through influencing cell division and differentiation,including in reproductive organogenesis and fertility[19,38].Growing studies have demonstrated that disruption of cytokinin homeostasis hinders female and male gametophyte development[13-17].In this study,we demonstrate that LOGL8 mainly participates in pollen development directly through producing active cytokinins(Figs.4,5B).But intriguingly,overexpression of LOGL8 transcripts in pollens does not affect pollen fertility(Figs.1I,3),suggesting that excess cytokinin in pollens may not interfere with pollen development.

Fig.5.LOGL8 has phosphoribohydrolase activity against cytokinin nucleoside 5′-monophosphates.(A)Subcellular localization of LOGL8 protein.The LOGL8-GFP construct was co-expressed with the mCherry construct or the nucleus-localized mCherry-NLS construct in rice protoplasts.Scale bars,20μm.(B)The enzymatic activity of LOGL8 for converting inactive iPRMP(iP riboside 5′-monophosphate)substrate into bioactive iP(isopentenyladenine),as revealed by HPLC assay,with LOG used as a positive control.(C)Comparison of cytokinin concentrations in young leaves of WT and LOGL8-pm.cZ,cis-zeatin;cZR,cZ riboside;DHZ,dihydrozeatin;DHZR,DHZ riboside;FW,fresh weight;iPR,iP riboside;tZ,trans-zeatin;tZR,tZ riboside.Data are mean±SD(*,P＜0.05;**,P＜0.01,n=3).(D)Comparison of OsRR gene expression in young leaves of WT and LOGL8-pm using qRT-PCR.ACTIN1 served as an internal reference.Values are mean±SD(*,P＜0.05;**,P＜0.01,n=3).

Fig.6.LOG expression is downregulated in the LOGL8-pm mutant.(A)Comparison of LOG expression in young panicles of WT and LOGL8-pm at various developmental stages using qRT-PCR.ACTIN 1 served as an internal reference.Values are mean±SD(**,P＜0.01,n=3).(B-E)In situ hybridization of LOG transcripts in inflorescences and ovules of WT(top)and LOGL8-pm(bottom).(B)Primary branch primordium stage.(C)Secondary branch primordium stage.(D)Florets at the ovule primordium stage.(E)Ovaries at the archesporial cell stage.In WT,strong expression signals of LOG were detected in the central tips of primary branch primordia,secondary branch primordia,ovule primordium epidermis,and the inner side of the carpel opposite the ovule.Clear signals were still detected in ovules at the archesporial cell stage.By contrast,in LOGL8-pm,much weaker signals were detected in the corresponding regions.ca,carpel;ov,ovule;pb,primary branch;sb,secondary branch.Scale bars,100μm.

Cell fate specification depends on the local accumulation of cytokinin[36,37,39,40].For example,the Arabidopsis HK cytokinin receptors establish a chalazal-localized cytokinin signal that helps specify functional megaspore fate[36].Arabidopsis CKI activates the cytokinin response in the embryo sac,and its loss-of-function mutation causes antipodal and central cells to adopt an egg cell fate[13].LOG determines the localization of bioactive cytokinins in the central tip of the panicle branch primordium and the epidermis of the ovule primordium[6,22],i.e.,near the regions of ectopic expression of LOGL8 in LOGL8-pm.Our findings support the notion that the ectopic hyperexpression of LOGL8 in panicle branches and ovaries is the molecular basis of the LOGL8-pm phenotypes,which disturbs the spatiotemporal distribution of bioactive cytokinins in these tissues.Likewise,overexpressing LOG,LOGL5,LOGL6,LOGL8,or LOGL10 causes a reduction of panicle branching,similar to that of loss-of-function mutants of LOG(Fig.S6)[6,22,23,25].

These findings suggest that the common abnormalities of plants overexpressing various LOG homologs may be due to the impaired LOG function rather than the intrinsic functions of these homologs.We speculate that the perturbed cytokinin distribution in these plants hinders the establishment of cytokinin localization in the proximal position of the panicle,likely because that the ectopic expression of LOGL8 induces an inhibition of LOG expression.Consistent with this notion,cytokinin and LOG transcript levels were indeed reduced in LOGL8-pm(Figs.5C,6).Additionally,LOGL8 is expressed predominantly in anthers in WT plant(Figs.3,S3),and the function-knockout mutant of LOGL8 showed mainly pollen abortion without overlapping visible phenotypes with LOGL8-pm(Figs.1,4).These results might also help explain the low fertility of plants with strong cytokinin signals in panicles,such as plants overexpressing SP3[12]and the osckx11 mutant[26].The establishment of LOG-mediated cytokinin localization in the ovule epidermis may prove to be a prerequisite for the specification of archesporial cells and integuments,a notion inviting further investigation.

Overexpressing LOGL8 in japonica rice cultivar ZH11 had little influence on awn development(Fig.S6C),but produced a phenotype similar to that of the transgenic japonica rice cultivar Nipponbare overexpressing LOG or LOGL6[23].In contrast,overexpressing LOGL6 in the indica rice cultivars Guangluai 4 and Kasalath markedly promoted awn elongation[23],suggesting that the diverse effects of LOG-family genes on awn development could be attributed to the diverse genetic backgrounds of these subspecies.LOG-family genes function redundantly in Arabidopsis;no visible phenotype has been observed in any single or double null mutant,despite the differing spatiotemporal expression patterns of these genes[35,41].Similarly,knockout of the cytokininbiosynthesis gene IPT7 in Arabidopsis did not produce a visible phenotype,except when other ipt mutations were combined with ipt7[42].However,in Medicago truncatula,MtLOG1 is expressed specifically in the nodule primordium,where it regulates nodule development and lateral root formation[43].It thus appears that the biochemical functions of LOG-family members are conserved,but that their roles may vary,with distinct tissue-specific patterns.

Disrupting cis-regulatory sequences may alter gene expression and thus induce phenotypic changes.A T-DNA insertion in the 6292-bp promoter region of RICE FLORICAULA/LEAFY(RFL)causes the ectopic expression of this gene specifically in floral meristems and floret organs,leading to the formation of multiple ovaries[29].A deletion of the inverted sequence in the promoter proximal region of Oryza sativa REPRODUCTIVE MERISTEM 20(OsREM20)alters its expression level,increasing grain yield[44].Deep dissection of the upstream regulatory region of LOGL8 and the related trans-factors may reveal the mechanism that leads to the distinct expression pattern of this gene in male organs.

Here,we present a working model(Fig.7)to summarize the function of LOGL8 and explain how LOGL8-induced inhibition of LOG expression leads to the LOGL8-pm phenotype,with small panicles and female sterility.Our findings provide new knowledge into cytokinin-mediated male development and suggest how the delicate regulation of local cytokinin homeostasis affects panicle branching and female organogenesis in rice.Meanwhile,although LOGL8-pm is a T-DNA insertion mutant and seems to be an artificial phenomenon,considering that there are many active transposons in plants,if transposon-insertions locate in promoter regions of genes,similar effects on affecting gene expression patterns and phenotypic variations may occur naturally.Evolutionarily,disruption of promoter or upstream regulatory region causing alteration of gene expression is a common way that leads to complex interactions between genes.An additional reminder of our results is that overexpression experiments should be used carefully to determine gene functions when studying hormone-related genes and genes that may cause secondary effects by its product,because in these cases the phenotype of overexpression of a gene is not correlated with that of its loss-of-function mutant,and thus may not reflect the original function of the genes.

Fig.7.A working model for LOGL8-LOG-mediated panicle branching and reproductive organ development in LOGL8-pm.Homeostasis of cytokinin distribution is required for plant growth and development.In general,LOG positively regulates panicle branching and ovule development by directly controlling local cytokinin accumulation(black arrows),while LOGL8,the nearest homolog of LOG,controls primarily male fertility(red arrows).However,the ectopic expression of LOGL8 in panicles and pistils due to disruption of the promoter-regulatory region of this gene by the presence of a T-DNA insertion(or the LOGL8-transgene with a truncated promoter-regulatory region;blue text and line)results in disturbed cytokinin distribution.This leads in turn to inhibition of LOG expression in the proximal position of the panicle and ovule,causing abnormal panicle and ovule development.

CRediT authorship contribution statement

Shuifu Chen:Conceptualization,Data curation,Funding acquisition,Investigation,Validation,Writing-original draft.Sulin Lou:Conceptualization,Data curation,Investigation,Validation,Writing-original draft.Xiucai Zhao:Data curation,Investigation,Validation.Shijuan Zhang:Data curation,Investigation,Validation.Letian Chen:Writing-review & editing.Ping Huang:Data curation,Investigation,Validation.Guande Li:Data curation,Investigation,Validation.Yingying Li:Data curation,Investigation,Validation.Yao-Guang Liu:Conceptualization,Funding acquisition,Project administration,Writing-review & editing.Yuanling Chen:Conceptualization,Funding acquisition,Project administration,Writing-review & editing.

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(31991222,32030080,31271301),the Guangdong Basic and Applied Basic Research Founation(2020A1515110207),and the China Postdoctoral Science Foundation(2020M682726).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2022.02.009.