Kapazoglouetal.BMCPlantBiology2012,12:166 http://www.biomedcentral.com/1471-2229/12/166 RESEARCH ARTICLE Open Access The study of two barley Type I-like MADS-box genes as potential targets of epigenetic regulation during seed development Aliki Kapazoglou1, Cawas Engineer1, Vicky Drosou1, Chrysanthi Kalloniati3, Eleni Tani1, Aphrodite Tsaballa2, Evangelia D Kouri3, Ioannis Ganopoulos2, Emmanouil Flemetakis3 and Athanasios S Tsaftaris1,2* Abstract Background: MADS-box genes constitute a large family of transcription factors functioning as key regulators of many processes during plant vegetativeand reproductive development. Type II MADS-box genes have been intensively investigated and are mostly involved invegetativeand floweringdevelopment. A growing number of studies ofType I MADS-box genes inArabidopsis, have assigned crucial roles for these genes ingamete and seed development and have demonstrated that a numberof Type I MADS-box genes are epigenetically regulated by DNA methylation and histone modifications. However, reports on agronomically important cereals such as barley and wheat are scarce. Results: Here wereport the identification and characterization oftwo TypeI-like MADS-box genes, from barley (Hordeum vulgare),a monocot cereal crop of high agronomic importance. Protein sequence and phylogenetic analysis showed that theputative proteins are related to Type IMADS-box proteins, and classified them ina distinct cereal clade.Significant differences in gene expression among seed developmental stages and betweenbarley cultivars withvarying seed size were revealed for both genes. One of these genes was shown to be induced by the seed development- and stress-related hormonesABA and JA whereas in situ hybridizations localized theother gene to specific endosperm sub-compartments.The genomic organization ofthe latter has highconservation withthe cereal TypeI-like MADS-box homologues and thechromosomalposition ofboth genes is closeto markers associated with seed quality traits. DNA methylation differences are present in theupstream and downstream regulatory regions of the barley TypeI-like MADS-box genesin two different developmental stages and inresponse to ABA treatment which may be associated with gene expression differences. Conclusions: Two barley MADS-box genes were studied that are related to Type I MADS-box genes. Differential expression in different seed developmental stages as well as in barley cultivars withdifferent seed sizewas evidenced for both genes. The two barley TypeI MADS-box genes were found to be induced by ABA and JA. DNA methylation differences indifferent seed developmental stages and after exogenous applicationof ABA is suggestive of epigenetic regulation ofgene expression. The study of barley Type I-like MADS-box genes extendsour investigations of gene regulation during endosperm and seed development ina monocot crop likebarley. Keywords: MADS-Box, Epigenetic regulation, Chromatin, DNA methylation, Histonemethylation, Seed development, Endosperm,Retrotransposon, Barley *Correspondence:[email protected] 1InstituteofAgrobiotechnology(INA),CERTH,Thermi-ThessalonikiGR-57001, Greece 2DepartmentofGeneticsandPlantBreeding,AristotleUniversityof Thessaloniki,ThessalonikiGR-54124,Greece Fulllistofauthorinformationisavailableattheendofthearticle ©2012Kapazoglouetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsofthe CreativeCommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse, distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited. Kapazoglouetal.BMCPlantBiology2012,12:166 Page2of22 http://www.biomedcentral.com/1471-2229/12/166 Background manner where the paternal allele is expressed whereas In angiosperms, the endosperm of the developing seeds the maternal allele is silenced [19,20]. Both DNA and is formed as a result of the double fertilization event. histone methylation are responsible for these silencing Fertilization of the egg cell by a sperm cell from the events [19-21]. The PRC2 Polycomb group complex is male gametophyte generates the diploid embryo from partially responsible for keeping the maternal allele si- which the tissues, organs, and shoot meristems of the lent in the female gametophyte and in the seed after plant will be generated. Fertilization of the adjacent cen- fertilization[19-21].Thisisachievedthroughthehistone tral cell by a second sperm cell forms a triploid endo- methylation activity of the PRC2 complex, conferred by sperm which supports embryo growth and development one of its subunits, the histone methyltransferase by producing storage proteins, lipids and starch [1,2]. MEDEA (MEA) with histone 3 lysine 27 trimethylation During this process, a large number of genes are acti- activity (H3K27me3). The PRC2 complex also restricts vated. Epigenetic regulatory considerations become im- the expression of PHE1 to the chalazal domain of the portant in relation to which parental allele will be endosperm after fertilization. Arabidopsis mea mutants expressed and in which reproductive tissue, a factor that show upregulated PHE1 expression and form defective ultimately governs among other things the size of the seed-like structures before fertilization and endosperm endosperm[3-6]. overproliferation after fertilization [19,20,22]. Further- Developmental transition of plants from the vegetative more, a distantly located region downstream of paternal to the reproductivestage and from the floral stage to the PHE1 was found to have a DNA methylation require- seed stage relies on the activity of MADS-box transcrip- mentfor PHE1expression[22]. tion factors. With the completion of the Arabidopsis Additional Type I MADS-box genes have been impli- genome, more than 100 genes encoding for MADS-box cated in gamete and seed development. The AGA- transcription factors were uncovered which can be MOUS-LIKE23 (AGL23) gene was found to regulate phylogenetically classified in five clades, termed MIKC, female gametophyte formation and normal embryo de- Mα, Mβ, Mγ and Mδ (Arabidopsis Genome Initiative, velopment [23]. AGAMOUS-LIKE 80 (AGL80) was 2000) [7,8]. The Mα, Mβ, Mγ subfamilies comprise the shown to be critical for central cell and endosperm de- Type I lineage, whereas the MIKC and Mδ constitute velopment in Arabidopsis [24]. In agl80 mutants the for- the Type II lineage. Similarly, phylogenetic analysis mation of the central cell is defective, and endosperm based on the completed rice genome revealed 44 Type II development fails to initiate after fertilization. Similarly, MADS-box genes, of the MIKC and Mδ clades, and 35 AGAMOUS-LIKE 61 (AGL61) also termed DIANA, plays Type I MADS-box genes of the Mα, Mβ, Mγ clades [9]. an important role in central cell and endosperm forma- The two lineages are proposed to have arisen by a gene tion, in Arabidopsis [16,25]. Like agl80, agl61 mutants duplication that took place in the common eukaryotic haveaberrantcentralcellmorphology whichdegenerates ancestor more than a billion years ago [10-13]. MIKC before fertilization occurs, and endosperm development genes harbor the characteristic MADS-box domain (M), does not take place post fertilization [16,25]. Both genes the Intervening domain (I), the Keratin-like domain (K) are expressed exclusively in the central cell and early and the C-terminal domain (C). In contrast, the Type I endosperm and this along with their similar mutant genes contain only the MADS-box domain (M). Al- phenotype suggests that AGL61 and AGL80 proteins though theType I genes represent about 60% of the total may function as heterodimers within the central cell. In MADS genes in Arabidopsis, they have only recently agreement to this, AGL61 and AGL80 proteins were started to be studied, whereas the Type II genes have found to interact in yeast two-hybrid assays [16,25]. Of been extensively investigated both structurally and func- equal interestis AGL62with high expression in the early tionally [8,14-17]. This is due mostly to the fact that nuclear endosperm and sharp decline right before cellu- Type I MADS genes function during early gametogen- larization [26]. In agl62 mutants, the endosperm cellu- esis, embryogenesis and seed formation and conse- larizes prematurely indicating that the AGL62 is quently homozygous mutants are lethal. In addition required for repression of precocious cellularization dur- their expression is very low during these developmental ing the syncytial phase. AGL62 is under the epigenetic stages[8,18].AmongthefewTypeIfamilymembersthat control of PRC2 genes, as in prc2 mutants AGL62 fails have been studied is PHERES 1 (PHE1) [also known as to become silent and endosperm cellularization is AGAMOUS-LIKE37 (AGL37)], a Mγ-type gene with an arrested [26]. Thus, AGL62 seems to regulate the timing important role in Arabidopsis gamete and seed develop- of endosperm cellularization, which is triggered epigen- ment [19]. Epigenetic regulatory mechanisms have been etically by PRC2-mediated AGL62 silencing. Likewise, a implicated in the transcriptional control of this gene. recent study on AGL36 demonstrated that the expres- The PHE1 gene is expressed transiently at high levels sion of the AGL36 maternal allele is epigenetically con- immediately after fertilization in a parentally imprinted trolled in a sequential manner, firstly upregulated by the Kapazoglouetal.BMCPlantBiology2012,12:166 Page3of22 http://www.biomedcentral.com/1471-2229/12/166 function of DEMETER (DME), a DNA glycosylase en- three of the ArabidopsisType I MADS-box genes, PHE1, zyme responsible for demethylating MEDEA, and then AGL62, and AGL36 are under epigenetic regulation downregulated during endosperm development by the mediated, in part, by the chromatin repressive enzymatic PRC2 complex [27]. Understanding the genetic and epi- complex PRC2. Our group has recently characterized genetic processes controlling the timing of endosperm barley genes encoding a putative PRC2 Polycomb group cellularization could be of particular interest in agricul- complex [30]. To further focus our research efforts in ture,asprematureordelayedcellularizationisassociated this area and study potential barley targets of a PRC2 with smallandlargeseed sizeandweight,respectively. complex, we report here the identification and structural An extensive genome-wide study of 60 Type I MADS- characterization of two barley Type I-like MADS-box box genes in Arabidopsis uncovered a cell-type-specific genes. Their expression has been studied in different tis- expression pattern during female gametophyte and early sues and seed developmental stages, in two cultivars endosperm development [18]. Most genes are expressed with varying seed size, and after exogenous application in the central cell and antipodal cells in the female gam- of the developmental- and stress-related phytohormone etophyte and in the chalazal and peripheral endosperm ABA. In addition, their genomic organization was exam- of 1–2 days after fertilization (DAF) developing seeds. ined and compared to their cereal homologues. Finally, Thesedata areinagreementwiththeresultsfromearlier the 5’upstream regions were analyzed for conserved cis functional studies of individual genes like PHE1, AGL23, regulatory elements and the DNA methylation patterns AGL61, AGL62 and AGL80 and propose a role for Type of upstream and downstream regions were investigated I MADS-box genes in female gametophyte and endo- in two tissues with differential Type I-like MADS-box spermdevelopment. gene expression. Considerably less is known about Type I MADS-box genes in other plant species, especially monocots. In Methods wheat, 42 MADS-box genes were identified in silico, of Plantmaterial which 8 were classified as Type I [28]. In rice, a global- Commercial barley cultivars, Caresse, Byzantio, and scalemicroarrayexpression analysisidentifiedanexpres- Ippolytos differing in seed size and weight were planted sion pattern for all Type II and Type I MADS-box genes in the field and were the source oftotal RNA for expres- during vegetative and reproductive development [9]. sion analysis. For Caresse, the weight of 1000 grains is Overall Type I MADS-box genes had lower expression 50–55 gr, and 98% of seeds have diameter longer than levels than Type II MADS-box genes. Certain Type I 2.5 mm, for Byzantio the weight of 1000 grains is 36–42 genes were expressed throughout development, whereas gr and 75% of seeds have diameter longer than 2.5 mm, others exhibited more specific expression in particular whereas for Ippolytos, seeds weight 25–31 gr per 1000 tissues. For example, four rice Type I genes, of the sub- grains and only 35–45% of seeds have diameter longer class Mα, were found predominately expressed in seeds than 2.5mm(www.cerealinstitute.gr). 5–20 days after fertilization suggesting a role for these genes in seed development. Recently, a rice Type I Hormonaltreatment MADS-box gene, OsMADS87, was reported to be mater- Seven-day-old seedlings (Caresse) grown in a growth nally expressed in rice endosperm and associated to chamber (16 hours (h) light, 8 h darkness, at 22°C) were endosperm developmental transitions caused by inter- sprayed with 100 μM ABA, (abscisic acid +/− cis, trans- specific hybridization[29]. ABA, SIGMA) and 100 μM JA (methyl jasmonate, Cereal crops account for about 50% of global human ALDRICH). Aerial parts of plants were collected at 6 h calorific intake with the endosperm of cereal seeds being and 24 h after treatment and immediately stored in li- one of the most important sources (faostat.fao.org). quid nitrogen. Aerial parts from five plants were pooled Chief among these monocots are barley, wheat, rice and togetherforRNAextractionforeachtimepoint.Control maize. Besides human nutrition, cereal crops also repre- plantsweresprayedwithwater plus0.2% Tween. sent major sources of feedstock, fiber and recently bio- fuel substrates. Contrary to Arabidopsis and other dicots RNAisolationandfirststrandcDNAsynthesis where the endosperm is consumed during seed develop- Total RNA wasisolatedfrom roots,shoots,apical meris- ment, in monocots, such as barley and other cereals, the tems, first leaves of seedlings, flowers before fertilization endosperm persists and constitutes the nutritional part (immature flowers), seeds 1–3, 3–5, 5–10, 10–15, 15–20 oftheseedcontainingstorageproteinsandstarch. days after fertilization (DAF), and aerial parts after hor- Considering the implications of Type I MADS-box monal treatments, respectively, using TRI REAGENT genes on seed development and the agronomic import- (SIGMA) according to the instructions of the manufac- ance of cereal endosperm, we set out to identify and turer. First strand cDNA synthesis was performed using characterize Type I MADS-box genes in barley. At least 1.0 μg total RNA, 0.5 μg 3’ RACE Adapter primer, Kapazoglouetal.BMCPlantBiology2012,12:166 Page4of22 http://www.biomedcentral.com/1471-2229/12/166 5’-GGCCACGCGTCGACTAGTAC (T) -3’ (Invitrogen), OPA-2009-3 H [37], the Barley, OPA123-2008, Consen- 17 1 mM dNTPs and 200U of Superscript II (Invitrogen) in sus-Hordeum-OPA123-2008-3 H and the Barley, OWB, 20 μL total volume, according to the specifications of OPA2008-Hordeum-OWB-OPA2008-3 H [38] using the themanufacturer. HarvEST tool and the comparative map viewer (Cmap) availableatGramene(http://www.gramene.org/cmap/). Proteinsequenceanalysis 1895 viridiplantae proteins sequences from UniProt with ExpressionanalysisofbarleyTypeI-likeMADS-boxgenes a statistically significant hit for the MADS-box domain Qualitative RT-PCR and quantitative real-time RT-PCR (PF00319: SRF-type transcription factor (DNA-binding wasperformed withcDNA synthesized from1 μg oftotal and dimerization domain) were collected from the Pfam RNA from roots, stems, meristems, leaves, immature database [31]. After sequence fragments were filtered flowers, seeds 1–3 DAF, 3–5 DAF, 5–10 DAF, 10–15 out, the resulting set was the subject of an all-against-all DAF and 15–20 DAF and aerial parts of seedlings similarity detection step using the BLASTalgorithm. We after ABA and JA treatment. For real-time PCR, each used the bidirectional best hit approach [32] for disco- sample reaction was set up in a PCR reaction mix vering putative orthologous proteins; we also asked for (20 μl) containing 5 μl of the 1:50 diluted cDNA, any orthologous sequence to be one of at least twenty 0.25 μM of each primer and 1X Platinum SYBR Green sequences with the MADS-box domain from the corre- qPCR Supermix-UDG (Invitrogen, Paisley, UK) and sponding species, so that we reduce the chance of best using the Corbett Rotor Gene 6000. Each reaction was hits due to lack of sequence context. The multiple align- performed in triplicates. General thermocycler condi- ment was created with MAFFT [33] and edited with Jal- tions were50°C for 2 min, 95°C for 2 min,then 42 cycles View version 2.5 [34] after which the phylogenetic tree of 95°C for 15 sec, annealing [53°C for HvOS1 and 57°C was constructed in the same software with the Neighbor for HvOS2] for 20 sec, extension 72°C for 20 sec, then Joining method using a BLOSUM62-based distance 72°C for 5 min. To identify the PCR products a melting measure. The phylogenetic tree was calculated using curve was performed from 65°C to 95°C with observa- MEGA 3.1 software [35] by the Neighbor-Joining tions every 0.2°C and a 10-s hold between observa- Method with p-distance correction [36]. Bootstrap tions. Relative quantification was performed using actin values wereobtainedfrom1000bootstrapreplicates.Ac- as the reference gene and HvActinF / HvActinR as pri- cession numbers for sequences used for alignments and mers. The barley genes HVA22 and HvADC (arginine de- phylogenetic analysis areindicated inTable 1. carboxylase 2) which are known to be induced by ABA and JA, respectively [39,40], were used as positive con- Genomicorganization trols. All primers used in expression analysis corres- The genomic sequences of Brachypodium distachyon _ pond to non-conserved regions and are shown in Bradi2g59120, Brachypodium distachyon_ Bradi2g59190, (Additional file 1: Table S2). Oryza sativa_OsMADS65_Q9XJ61 (Os01g0922800), and Zeamays_ZmMADS1_GRMZM2G171650_B4FML Insituhybridization 1, were downloaded from the Phytozome database In situ hybridization experiments were performed as (http://www.phytozome.net/). Thesequencesofthetwo it has been described previously [41]. Briefly, seeds full length barley Type-like MADS-box ESTs were used were fixed in 4% (w/v) paraformaldehyde supplemen- to interrogate the barley database http://webblast.ipk- ted with 0.25% (v/v) glutaraldehyde in 10 mM so- gatersleben.de/barley/index.php for detection of gen- dium phosphate buffer (pH 7.4) for 4 h in a vacuum omic sequences. Genomic organization of exons and aspirator. Fixed tissues were block-stained in 0.5% (w/v) introns was obtained using the mRNA-to-genomic safranin, dehydrated through ethanol series, embed- alignment Spidey tool, in NCBI (http://www.ncbi. ded in paraffin and cut into 8 mm-thin sections. nlm.nih.gov/spidey). Detection of retroelements was Antisense RNA probes labelled with digoxigenin-11- performed with the MASiVE and LTRharvester tools rUTP (Boehringer Mannheim, Mannheim, Germany) (http://tools.bat.ina.certh.gr/masive/) (http://tools.bat. were originated from PCR-generated templates incorpo- ina.certh.gr/ltrharvester/) and homology was visua- ratingT3 polymerase sites. The probe was designed close lized with Circoletto (http://tools.bat.ina.certh.gr/cir- tothe3’-UTRofthegeneanditslengthwas202bp.Sec- coletto/), all three tools developed in-house at INA, tions were prepared for hybridization as described before by the Bioinformatics Analysis Team (BAT). [42]andhybridizedovernightat42°Cin50%(v/v)forma- mide, 300 mM NaCl, 10 mM Tris–HCl pH 7.5, 1 mM Mappinginsilico EDTA, 0.02% (w/v) Ficoll, 0.02% (w/v) polyvinylpyrroli- An in silico visual comparative analysis was performed done,0.025%(w/v)bovineserumalbumin(BSA),10%(v/v) against the Barley, OPA 2009, Consensus-Hordeum- dextran sulfate and 60 mM DTT. After hybridization, the Kapazoglouetal.BMCPlantBiology2012,12:166 Page5of22 http://www.biomedcentral.com/1471-2229/12/166 Table1TypeIMADS-boxsequencesusedforalignmentsandphylogenetictreeconstruction Organism Genename Accessionnumber TypeI Cerealtype-I-like Hordeumvulgare HvOS2 HM130526.1TC178280 Triticumaestivum TaAGL33 ABF57950.1Q1G159 Brachypodiumdistachyon Bradi2g59190 Bradi2g59190 Hordeumvulgare HvOS1 HM130525.1BG365393 Triticumaestivum TaAGL42 ABF57942.1Q1G167 Brachypodiumdistachyon Bradi2g59120 Bradi2g59120 Zeamays ZmB4FML1 NP_001140218.1 Orysasativa OsMADS65 Os01g0922800Q9XJ61 Mα Arabidopsisthaliana AtAGL62 NP_200852.1Q9FKK2 Arabidopsisthaliana AtAGL61 NP_850058.1A5AZX6 Arabidopsisthaliana AtAGL23 NP_176715.1 Arabidopsisthaliana AtAGL28 NP_171660.1 Poplartrichocarpa AGL62-like XP_002313197.1B9HN34 Vitisvinifera AGL62-like A5AZX6 Ricinuscommunis AGL62-like B9S7W9 Orysasativa OsMADS71 Os06g22760.1 Orysasativa OsMADS78 Os09g02830.1 Orysasativa OsMADS79 Os01g74440.1 Mβ Arabidopsisthaliana AtAGL47 NP_200380.1 Arabidopsisthaliana AtAGL82 NP_200697.1Q9FIM0 Poplartrichocarpa PtMADS-I-like XP_002301840 Orysasativa OsMADS90 Os07g04170 Orysasativa OsMADS91 Os01g11510.1 Orysasativa OsMADS96 OsO1g67890 Mγ Arabidopsisthaliana PHE1AGL37 NP_176712.1O80805 Arabidopsisthaliana PHE2AGL38 NP_176709.2Q7XJK8 Arabidopsisthaliana AGL80 NP_199678.1Q9FJK3 Orysasativaindica OsAGL80-like EAY73584.1Q75IC5 Orysasativajaponica OsjAGL35-related BAD81343.1 Orysasativa OsMADS82 Os04g24800 Orysasativa OsMADS83 Os04g24810 Orysasativa OsMADS85 Os04g25920 Poplartrichocarpa PtAGL80-like XP_002329991.1B9I7J4 Petuniahybrid PhAGL80-like B6DT62 Ricinuscommunis RcAGL80-like B9S273 Vitisvinifera VvAGL80-like A5BJU9 sections were treated with a solution containing visualized with anti-digoxigenin antibodies conjugated 500 mM NaCl, 1 mM EDTA, 10 mM Tris–HCl and with alkaline phosphatase. Images were processed using 50μg/mlRNaseA.Finally,sectionswerewashedseveral Photoshop 7 software (Adobe Systems Inc., San Jose, times in a 2xSSC solution. Hybridization signals were CA, USA). Kapazoglouetal.BMCPlantBiology2012,12:166 Page6of22 http://www.biomedcentral.com/1471-2229/12/166 Genomewalkingand5’upstreamsequenceanalysis MADS-box domain and the absence of a K-domain, Genome walking experiments were conducted on leaf characteristic of Type I MADS-box proteins. These can- genomic DNA obtained from the barley Caresse cultivar. didates were: BG365393 and TC178280 (TIGR database, The Clontech genome walking kit was used and proce- www.tigr.org) (see Materials and Methods). During the dureswereconductedaccordingtomanufacturer’sspecifi- course of our investigations another study characterizing cations. Sequence information was obtained by doing these sequences with respect to their role in the threeprogressiveroundsofgenomewalkingusingESTse- vernalization process was reported [43]. These authors quencedataasstartingpoints.Statisticallysignificantpre- designated these sequences as ODDSOC1(HvOS1) and diction for CpG islands was performed using the online ODDSOC2(HvOS2), respectively. In our recent study on predictor,whichispartofthesequencemanipulationsuite the identification and characterization of two SOC1-like at http://www.bioinformatics.org/SMS/index.html. The gene homologues from barley [44] it was shown that prediction of the putative cis acting elements was accom- two HvSOC1-like proteins containing the K-box domain plished using theTSSP /Prediction of PLANT Promoters are closely related to the Type II MADS-box proteins algorithm (Using RegSite Plant DB, Softberry Inc.) in the whereas they are more distantly related to ODDSOC1 SoftBerry database (http://linux1.softberry.com/cgi-bin/ and ODDSOC2. However, since Greenup et al. (2010) programs/promoter/tssp.pl) and PlantCARE (http://bio [43] were the first to report these genes we will utilize informatics.psb.ugent.be/webtools/plantcare/html/). The the nomenclature introduced by these authors and here- positionoftheputative5’upstreamregulatoryelementsis after will be referring to the sequences BG365393 and indicatedaccordingtothestartcodon(ATG). TC178280 as HvOS1 and HvOS2, respectively. HvOS1 is composed of 834 nt encoding a putative pro- DNAmethylationassays tein of 167 aa, and harboring a 5’UTR of 76 nt and a Genomic DNA was prepared from immature flowers 3’UTRof253nt.HvOS2iscomposedof1319ntencoding andfrom1–3DAFseeds(Caresse)withQiagencolumns a putative protein of 159 aa, and containing a 5’UTR of following the protocol of the manufacturer (Qiagen 382 nt, and a 3’UTR of 460 nt. An alignment of MADS- Plant genomic DNA kit). Cytosine DNA methylation box proteins (Type I and Type II) from different plants is was analyzed by restricting 1 μg of genomic DNA from shown in Figure 1. The two barley putative protein each sample with the methylation-dependent enzyme sequencesHvOS1 andHvOS2showhighsimilarity tothe McrBC (NEB Biolabs), according to the manufacturer’s wheat sequences that have been previously classified as instructions, and PCR-amplifying equal quantities of Type I MADS-box sequences [28]. HvOS1 has 96% iden- McrBC-treated and untreated samples. Primers used are tity with the wheat TaAGL-42 sequence and HvOS2 shownin(Additionalfile1:TableS2). shares 92% identity with the wheat TaAGL-33 sequence. HvOS1 and HvOS2 putative protein sequences share Results 88.9% identity with each other. In Type II MADS-box IdentificationofTypeI-likeMADS-boxgenesinbarley proteins the characteristic three subdomains of the andproteinsequenceanalyses K-region,K1,K2andK3areevidentasshowninFigure1. Initial efforts to identify orthologues of Type I MADS- The barley HvOS1 and HvOS2, wheat TaAGL-33 and box genes in barley EST databases via BLASTand other TaAGL-42, as well as their closest relatives rice sequence similarity-based approaches failed to result in OsMADS65 and maize ZmB4FML1 harbor no regions significant hits. Additional efforts using degenerate pri- with similarity to K regions, as expected for Type I mers designed using Arabidopsis sequence data and bar- MADS-box proteins. A separate alignment of the barley ley cDNA from multiple cultivars did not result in target HvOS1 and HvOS2 and different Type I MADS-box gene isolation either (data not shown). Hence a more sequences was constructed for more clarity (Figure 2). rigorous bioinformatics approach was employed and a Most of the similarity between the barley HvOS1 and monocot-specific C-terminus probe was generated using HvOS2 and the other Type I MADS-box sequences is existing sequence data for MADS-box putative proteins. concentrated in the MADS-box domain (first 80 aa), Thesequenceofthisprobe wasthefollowing: whereas the rest of the sequences show a large degree of GAGXXVNGXOXXXNXDXXXXOOQXXLKEIAXWX divergence. A phylogenetic tree was constructed from a XQNNAOXXDANOLEKLEOLLTOALRNTKXKKMLXO largenumberofMADS-boxproteinsfromdifferentplants ONXG, where X is any amino acid and O is a possible (Figure 3). Three of the subclasses of Type I, Mα, Mβ gap. This probe was used to isolate several cereal Type and Mγ, are shown grouping out into separate clades. I-like MADS-box sequences from available EST data- The barley HvOS1 and HvOS2 putative proteins, and bases and the GenBank non-redundant CDS translated their putative cereal orthologues, namely, wheat TaAGL- database. Of particular interest were two hits from the 33 and TaAGL-42, brachypodium Bradi2g59190 and barley EST database, chosen based on the presence of a Bradi2g59120, rice OsMADS65, and maize ZmB4FML1, Kapazoglouetal.BMCPlantBiology2012,12:166 Page7of22 http://www.biomedcentral.com/1471-2229/12/166 MADS-box domain I II I II K1 domain K2domain K3domain I II Figure1AminoacidsequencecomparisonsbetweenTypeIMADS-boxproteinsandTypeIIMADS-boxproteinsfromdifferentplants. ThenameofeachsequenceconsistsofitsUniprotIDorGenBankaccessionnumber,followedbythespeciesabbreviationaccordingtoUniprot. TheK-boxdomains,K1,K2andK3oftheTypeIIMADS-boxproteinsareshowninboxesandindicatedwithredbars.TheMADS-boxdomainis alsoindicatedwitharedbar.NoK-boxispresentintheType-Iproteins.Identicalaminoacidsareshownindarkblueandsimilaraminoacidsin lightblue. cluster together forming a distinct subclass (Figure 3). Ippolytos examined. Conversely, HvOS2 transcripts As mentioned above due to the absence of a typical are present in roots, shoots, apical meristem, leaves, K-domain which is the hallmark of Type-I proteins, we and immature flowers in both Caresse and Ippolytos named these barley sequences as Type I-like HvMADS- (Figure 4A). box sequences. Real time PCR analysis was employed to examine and compare the expression of barley Type I-like HvMADS- ExpressionanalysisoftheTypeI-likeHvMADS-boxgenes box genes at different seed developmental stages and in indifferenttissuesandduringseeddevelopment different cultivars, Caresse (a large-seed cultivar) and End point PCR analysis showed that HvOS1 tran- Ippolytos(a small-seedcultivar) (Figure 4B). scripts were detected only in immature flowers in For HvOS1, an induction of approximately 8 fold in Caresse, whereas they were present in all tissues of Caresse 1–3 DAF and 3–5 DAF seed, and a decrease Kapazoglouetal.BMCPlantBiology2012,12:166 Page8of22 http://www.biomedcentral.com/1471-2229/12/166 MADS-box domain Figure2AminoacidsequencealignmentsbetweenbarleyHvOS1,HvOS2,andTypeIproteinsfromdifferentplants.Thenameofeach sequenceconsistsofitsGenBankaccessionnumber,alsoshowninTable1.Identicalaminoacidsareshownindarkblueandsimilaraminoacids inlightblue.TheMADS-boxdomainisindicatedwitharedbar. thereafter, was observed. In Ippolytos, HvOS1 was DAF seeds whereas they increased slightly in 10–15 induced in 1–3 and 3–5 DAF seeds by approximately 2 DAF as compared to immature flowers. In Caresse 15– fold. Transcript levels dropped by about 2 fold in 5–10 20 DAF seeds transcript levels increased by about 2 fold, Kapazoglouetal.BMCPlantBiology2012,12:166 Page9of22 http://www.biomedcentral.com/1471-2229/12/166 SEPALATA-AGAMOUS-SOC1 (Type II) 99 HvOS2 96 TaAGL33 41 Bradi2g59190 Cereal 68 OsjMADS65 Type I-like Bradi2g59120 MADS-box 100 73 HvOS1 99 TaAGL42 MADS box ZmMADS-I-like 94 100 OsMADS78 51 OsMADS79 OsMADS71 74 100 AtAGL28 Ma AtAGL23 100 AtAGL61 AtAGL62 94 98 PtAGL62-like OsMADS90 100 OsMADS91 100 AtAGL82 Type I 49 AtAGL47 MADS-box PtMADS-like OsMADS96 96 AtAGL80 100 PtAGL80-related 100 31 AtPHE1 100 AtPHE2 99 100 OsiAGL80-related OsjAGL35-related 100 OsMADS83 OsMADS82 100 100 OsMADS85 0.1 Figure3PhylogenetictreeofMADS-boxproteinsfromdifferentplants.PhylogenetictreeshowingthedifferentcladesofMADS-boxTypeI familyproteins.TheTypeIIMADS-boxSEPALATA,AGAMOUSandSOC1familiesareshownasacondensedbranch.Thesequencesusedandtheir accessionnumbersareshowninTable1.BarleyHvOS1andHvOS2membersareinbold.Numbersindicatebootstrapvalues(1000=100%). whereas in Ippolytos 15–20 DAF they decreased by 15–20 DAFseeds (Figure 4B). Expression of HvOS2 was about 5fold (Figure 4B). examined in another large-seed cultivar, Byzantio. A marked increase of approximately 8 fold in HvOS2 Similar to Caresse, HvOS2 transcript levels were signifi- transcript accumulation was observed in Caresse seeds cantly lower in 5–10, 10–15 DAF seeds as compared to 1–3 DAF and 3–5 DAF, as compared to immature Ippolytos(AdditionalFile2). unfertilized flowers. Expression levels dropped thereafter More detailed expression of the HvOS2 gene within in 5–10, 10–15 and 15–20 DAF seeds to levels compar- the seed was examined by in situ localizations using able to those of unfertilized flowers. In the small-seed crosssectionsofseedsandDIG-labelledantisenseprobe. cultivar, Ippolytos, HvOS2 exhibited substantial tran- Strong hybridization signal was observed in the aleurone script accumulation in seeds 1–3 DAF and 3–5 DAF. In layer and the first layer of endosperm cells adjacent to 5–10 DAF and 10–15 DAF seeds HvOS2 had a pro- thealeuronelayerthatwasabsentfromthesense-control nounced increase of approximately 10–15 fold, in con- (Figure 4C), suggesting cell-specific expression in the trast to Caresse. In Caresse seeds 15–20 DAF there was endosperm. In addition, strong hybridization signal was a slight increase of HvOS2 transcript. Conversely, a sub- observed in embryo cells between sections hybridized stantial decrease of about 10 fold was seen in Ippolytos with antisense as well as sense probes suggesting no Kapazoglouetal.BMCPlantBiology2012,12:166 Page10of22 http://www.biomedcentral.com/1471-2229/12/166 A) R AM YS L IF HvOS1 HvOS2 Caresse HvACTIN HvOS1 HvOS2 Ippolytos HvACTIN B) HvOS1 o ati 12 * r 10 * n o i 8 s s e 6 * * r p 4 * x e * 2 e v i 0 t a l C5 C6 C7 C8 C9 C10 IP5 IP6 IP7 IP8 IP9 IP10 e r seed developmental stage o i HvOS2 t a r 30 r n * o 25 si s 20 pre 15 * * ex 10 * * * e v 5 ti a 0 el r C5 C6 C7 C8 C9 C10 IP5 IP6 IP7 IP8 IP9 IP10 seed developmental stage C) antisense sense e e a a p p Figure4(Seelegendonnextpage.)