The DOF protein, SAD, interacts with GAMYB in plant nuclei and activates transcription of endosperm-specific genes during barley seed development


  • Isabel Diaz,

    1. Laboratorio de Bioquímica y Biología Molecular, Departamento de Biotecnología-UPM, E.T.S. Ingenieros Agrónomos, Ciudad Universitaria s/n, 28040 Madrid, Spain
    Search for more papers by this author
  • Manuel Martinez,

    1. Laboratorio de Bioquímica y Biología Molecular, Departamento de Biotecnología-UPM, E.T.S. Ingenieros Agrónomos, Ciudad Universitaria s/n, 28040 Madrid, Spain
    Search for more papers by this author
  • Ines Isabel-LaMoneda,

    1. Laboratorio de Bioquímica y Biología Molecular, Departamento de Biotecnología-UPM, E.T.S. Ingenieros Agrónomos, Ciudad Universitaria s/n, 28040 Madrid, Spain
    Search for more papers by this author
  • Ignacio Rubio-Somoza,

    1. Laboratorio de Bioquímica y Biología Molecular, Departamento de Biotecnología-UPM, E.T.S. Ingenieros Agrónomos, Ciudad Universitaria s/n, 28040 Madrid, Spain
    Search for more papers by this author
  • Pilar Carbonero

    Corresponding author
      (fax +34 913365695; e-mail
    Search for more papers by this author

(fax +34 913365695; e-mail


The DOF protein, SAD, previously shown to be a transcriptional activator in barley aleurone cells upon seed germination, also has an important role in gene regulation during endosperm development. mRNA was detected in early (10 days after flowering) developing barley seeds where it accumulated in the starchy endosperm, aleurone cells, nucellar projection, vascular tissues and the immature embryo, as shown by RT-PCR and in situ hybridization analyses. The SAD protein, expressed in bacteria, binds to oligonucleotides containing the prolamine box, 5′-A/TAAAG-3′sequence, derived from the promoter regions of the endosperm-specific genes Hor2 and Itr1, encoding a B-hordein and trypsin-inhibitor BTI-CMe, respectively. SAD competed for the same binding sites with another endosperm-expressed DOF protein, BPBF. Transient expression experiments in co-bombarded developing endosperms demonstrated that SAD trans-activated transcription from Hor2 and Itr1 promoters through binding to the intact DOF motifs. When the two DOF factors are co-bombarded together an additive effect was observed upon the expression of the Itr1 gene. In-frame fusion of the Sad ORF to the reporter green fluorescent protein gene directs the fluorescence expression to the nucleus in transiently transformed onion epidermal layers. The visualization of fluorescence in the nucleus of onion cells, using the bimolecular fluorescent complex (BiFC) approach, has shown the in vivo interaction between SAD and the R2R3MYB protein GAMYB. The interaction in plant cells has also been documented for the DOF protein BPBF and GAMYB, but nuclear interaction could not be detected between BPBF and SAD by this procedure.


In plants, the seed represents the plant structure linking parental and progeny generations. A seed consists of three distinct components: endosperm, embryo and seed coat. In cereal species, the endosperm is the major organ for storage compounds that subsequently are used by the germinating seedlings, whereas this function is performed by the embryo in most of the dicotyledonous plants (Olsen, 2004). During cereal seed maturation the endosperm is differentiated into two predominant tissue types, the starchy endosperm involved in the deposition of reserves, and the peripheral aleurone layer (Berger, 2003). The storage compounds accumulated during cereal seed development include carbohydrates, lipids and proteins, mainly prolamines (hordeins in barley), as well as several antimicrobial peptides and proteinaceous inhibitors of heterologous hydrolases putatively implicated in plant defence (Carbonero and Garcia-Olmedo, 1999; Carbonero et al., 1999).

The expression pattern of seed-specific genes, both in monocotyledonous and dicotyledonous plants (Lara et al., 2003), is highly regulated and this regulation involves both cis-motifs in promoters and trans-acting factors. Functional analysis of the endosperm-specific promoters in cereals has allowed the identification of conserved cis-motifs involved in such a regulation. To date, the most prominent of the cis-motifs described is the bi-partite endosperm box, containing two distinct protein-binding sites: the GCN4-like motif (GLM; 5′-ATGAG/CTCAT-3′) recognized by bZIP proteins of the Opaque2 subfamily (Albani et al., 1997; Oñate et al., 1999; Vicente-Carbajosa et al., 1998; Wu et al., 1998) and the prolamin box (PB; 5′-TGTAAAG-3′) bound by transcription factors of the DNA with one finger (DOF) class (Lijavetzky et al., 2003; Mena et al., 1998; Vicente-Carbajosa et al., 1997; Yanagisawa, 2002, 2004). A third motif, 5′-AACA/TA-3′, is the binding site of TFs of the R2R3MYB class (Díaz et al., 2002; Suzuki et al., 1998).

In barley, TFs belonging to these three classes have been reported as being involved in the regulation of genes encoding seed storage proteins, such as those encoding B-hordeins (gene Hor2), as well as some other genes related to defence functions such as that encoding trypsin inhibitor BTI-CMe (gene Itr1). BLZ1 and BLZ2 are members of the bZIP class that interact with the GLM motifs in Hor2 and Itr1 promoters through homodimer or heterodimer formation (Oñate et al., 1999; Vicente-Carbajosa et al., 1998). Both interact in vivo and synergistically trans-activate the appropriate reporter constructs in plant cells. BPBF, a TF of the DOF class that activated transcription of a native Hor2 promoter in barley endosperm through PB-recognition (Mena et al., 1998), was a repressor of the GA-induced thiol-protease cathepsin-B-like gene in aleurone cells upon germination (Mena et al., 2002). We have also demonstrated that GAMYB, a TF of the R2R3MYB class, is not only a regulator of GA-responsive genes upon seed germination (Gubler et al., 1995), but also activates gene expression during endosperm development through binding to the 5′-AACAA-3′ motif in native Hor2 and Itr1 promoters (Díaz et al., 2002).

SAD is a DOF transcription factor from barley that was previously shown to be expressed in barley aleurone cells following germination (Isabel-LaMoneda et al., 2003), where it recognized in vitro the pyrimidine box (5′-CTTT-3′) present in the tripartite GA-responsive complex (GARC) in the promoters of hydrolase genes. In transient expression experiments SAD trans-activated transcription from the cathepsin-B-like cysteine protease Al21 promoter in co-bombarded aleurone layers, in a manner similar to GAMYB (Isabel-LaMoneda et al., 2003).

In the present study, we report the expression of the SAD-encoding gene in developing barley seeds. In transient expression experiments in co-bombarded developing endosperms, SAD trans-activates native Hor2 and Itr1 promoters through binding to the 5′-TGTAAAG-3′ motif. In addition, SAD is capable of interacting with the GAMYB factor forming a bimolecular fluorescent complex in the nucleus of plant living cells, when each one of these transcription factor genes is fused to complementary green fluorescent protein (GFP)-encoding fragments. All these data, together with protein–DNA interaction experiments, strongly implicate SAD in the transcriptional activation of cereal genes expressed in the developing endosperm.


Sad is expressed in the developing barley seed

We have previously shown that the Sad gene had a relevant role as a transcription activator of hydrolase genes in germinating aleurone cells (Isabel-LaMoneda et al., 2003). To explore whether the Sad gene also had a role in seed development, we further investigated its expression in seed tissues. Total RNA was prepared from immature embryos, at 18 days after flowering (daf) and from endosperms isolated at four different stages after anthesis of barley cv. Bomi. As a control, RNA isolated from aleurone layers of 24-h-imbibed kernels was used. As the presence of Sad transcripts was almost undetectable by Northern blot analysis (data not shown), the time course of mRNA accumulation was analysed using semiquantitative RT-PCR techniques. As shown in Figure 1, at 10 daf, Sad mRNAs were detected in the developing endosperms, a peak being attained at 14 daf, declining thereafter to the point of being undetectable at 22 daf. Sad mRNAs were also present in immature embryos (18 daf) and in germinating aleurones (48 h post-imbibition). For comparative purposes, the expression of the barley Pbf gene encoding the DOF transcription factor BPBF was also studied. The Pbf gene showed a pattern of expression similar to Sad, although its message accumulation in immature embryos was undetectable. To know whether the pattern of expression of SAD was consistent with the possibility of being a regulator of seed-specific genes, the pattern of expression of the Itr1 and Hor2 genes encoding the CMe trypsin inhibitor and a B-hordein, respectively, was also analysed. Both mRNAs were abundantly expressed in the developing endosperm and that of Itr1 was also expressed in the immature embryo. Hor2 and Itr1 mRNAs were not detected in the post-germinating aleurone cells. The pattern of expression described in Figure 1 is thus compatible with SAD (and previously described BPBF) being a regulator of the expression of Hor2 and Itr1 during endosperm development.

Figure 1.

RT-PCR analysis of Sad expression in barley tissues. Total RNA was isolated from developing endosperm (En: 10, 14, 18 and 22 daf), immature embryos (iE, 18 daf) and aleurone (A) after 24 h of seed imbibition, and reverse transcribed in the presence of random hexamers. The first-strand cDNA was then amplified by PCR using specific primers for the Sad (SAD), Pbf (BPBF) Hor2 (HOR2) and Itr1 (CMe) transcripts. Amplification of a region of the 18S rRNA was used as the internal control (18S).

To localize the spatial expression of Sad within the developing seed, mRNA in situ hybridization studies were performed. In transversal sections of 20 daf developing endosperm, a clear signal with the antisense probe was observed, mainly in the starchy endosperm and in the aleurone layer, as well as in the nucellar projection and vascular tissues (Figure 2a). There was also a strong mRNA accumulation distributed throughout the embryo, as seen in the longitudinal section of 20 daf developing embryos (Figure 2b), that was especially prominent in the vascular tissues, the scutellum, the shoot meristem and the radicle. No signal above background was found when sections of developing seeds were hybridized with the sense probe used as negative control (Figure 2c,d).

Figure 2.

Expression of the Sad mRNA in developing barley seeds by in situ hybridization. Transverse sections of 20 daf developing endosperm (a, b) and longitudinal sections of 20 daf developing embryos (c, d). Hybridization was performed with the antisense Sad probe (a, c) or with the control sense probe (b, d). al, aleurone; e, endosperm; em, embryo; n, nucellar projection; p, pericarp; sc, scutellum; sm, shoot meristem; r, root meristem; vt, vascular tissues. Scale bars: (a, b) 200 μm; (c, d) 100 μm.

SAD, BPBF and GAMYB proteins are targeted to the nucleus

To explore the subcellular location of the SAD protein, a construct containing a translational fusion of its ORF to the GFP reporter gene under the control of the 35S promoter was used to bombard onion epidermal layers. The previously characterized transcription factors BPBF and GAMYB were also fused to GFP. As shown in Figure 3, the GFP fluorescence emitted by the GFP reporter was targeted to the nucleus of bombarded cells, as expected for the TFs included in the construct (Figure 3b,d,f). In contrast, the GFP fluorescence was distributed throughout the cells when the onion epidermal layers were bombarded with the 35S::GFP construct (Figure 3h). The tissue samples bombarded were photographed under light field (Figure 3a,c,e,g) and under BP 450/90 FT 510 LP520 Zeiss filter (Zeiss, Oberkochen, Germany) microscopy (Figure 3b,d,f,h).

Figure 3.

Sub-cellular location of TF proteins in onion epidermal cells. Epidermal onion cells were transiently transformed with 35S::SAD-GFP (a, b), 35S::BPBF-GFP (c, d), 35S::GAMYB-GFP (e, f) and 35S::GFP (g, h). After incubation for 24 h, cells were observed under a Zeiss Axiophot microscope under bright field (a, c, e, g) and under fluorescence (b, d, f, h) using BP450/90 FT 510 LP 520 filters. Arrows point to the fluorescent nucleus location.

SAD and BPBF interact with GAMYB in plant nuclei

The fluorophore of the GFP is produced by an autocatalytic cyclation reaction subsequent to protein folding (Tsien, 1998). To explore whether SAD interacted with BPBF and GAMYB in plant nuclei, we did reconstitution experiments of the GFP fluorophore. For this purpose, we used two ORF non-fluorescent fragments of the GFP that, when brought together by two interacting proteins fused to each one of the fragments, reconstituting the GFP fluorophore. We fused the two fragments of the GFP to the ORFs of SAD, BPBF and GAMYB, in all combinations, to detect possible homodimeric and heterodimeric interactions, and then we co-bombarded onion epidermal cells. Microscopic observations from the different TF combinations showed that the GFP fluorophore was reconstituted and targeted to the nucleus only when either a DOF transcription factor, SAD or BPBF, was allowed to interact with the R2R3MYB GAMYB, in the appropriated fusions to the GFP fragments (Figure 4a,d). No fluorescence was detected when the two DOF proteins, SAD and BPBF (Figure 4g), or when the same TF, were fused to the two complementary GFP fragments (data not shown). These results indicate that SAD and BPBF did not interact and that neither the DOF TFs nor GAMYB could form homodimers in plant nuclei. Reconstitution of the GFP observed was independent of which TF was fused to which portion of the GFP (the N-terminal 154 amino acid residues, or the C-terminal 85 residues). The nuclei were stained by 4,6-diamidine-2-phenylindol (DAPI) as controls (Figure 4b,e,h) and onion cells and their nuclei were also observed under bright field microscopy (Figure 4c,f,i). Fluorophore reconstitution by molecular association of the two GFP fragments was observed after 36 h of incubation and the green fluorescence had a weaker intensity than that observed when intact GFP fusions were carried out (Figure 3). As expected, no fluorescence was detected in reactions containing only one of the fragments, or when the GFP fragments were not fused to the TFs (data not shown).

Figure 4.

Nuclear location of reconstituted green fluorescent protein (GFP) complexes in transiently transformed onion epidermal cells. Nuclear fluorescence was visible following reconstitution of 35S::SAD-NGFP and 35S::GAMYB-CGFP (a) and 35S::GAMYB-NGFP and 35S::BPBF-CGFP (d) under a Zeiss Axiophot microscope using BP450/90 FT510 LP 520 filters. No fluorescence was observed with the 35S::SAD-NGFP and 35S::BPBF-CGFP combination (g). As control of sub-cellular location, DAPI staining using BP365 FT 395 LP397 filters (b, e, h) and bright field observations (c, f, i) were carried out. Arrows point to the nucleus location where the fluorophore is reconstituted.

SAD binds in vitro to DOF motifs in the promoter of endosperm genes

To evaluate whether SAD was capable of specifically binding in vitro to the 5′-A/TAAAG-3′ DOF motif (PB) in the promoters of Hor2 and Itr1 genes, electrophoretic mobility shift assays (EMSA) were performed. The SAD protein expressed as a GST fusion in Escherichia coli was incubated with four labelled-oligonucleotide probes, one (HD in Figure 5a) deduced from the corresponding region of the Hor2 gene promoter, and three (ItD1, ItD2 and ItD3 in Figure 5b) derived from the Itr1 gene promoter, that had three such motifs.

Figure 5.

Electrophoretic mobility shift assays (EMSA) of the recombinant SAD and BPBF protein fusions to GST, with oligonucleotides derived from the Hor2 and Itr1 gene promoters containing the AAAG motif.
(a) EMSA of the indicated recombinant SAD and BPBF proteins with the 32 bp 32P-labelled probe (HD) derived from the Hor2 gene promoter and its mutated version affected in the putative DOF-binding motif (hd). Competition experiments were performed using increasing amounts (20, 50 and 100×) of the indicated unlabelled HD and hd probes.
(b) EMSA of the indicated recombinant SAD and BPBF proteins with the 32P-labelled probes (ItD1, ItD2 and ItD3) derived form the Itr1 gene promoter. Sequence of the oligonucleotides used as probes is shown at the bottom of each panel. In the mutated version only the changed bases are indicated. The putative binding sites are underlined.

The HD probe was shifted only when incubated with the GST-SAD-enriched extract (+) and was not with a GST control protein (−) or when a mutated variant (hd) of the oligonucleotide, where the AAAG core sequence was changed to AgAc, was used in the EMSA. This binding was competed by a molar excess of the unlabelled wild-type probe but not by an excess of the mutated variant (Figure 5a). Besides, when the wild hordein-derived probe was incubated with a mixture of protein extracts of SAD and BPBF (Isabel-LaMoneda et al., 2003; Mena et al., 1998), both proteins competed for the same binding site, as a mixture of the patterns obtained from the TFs alone was observed (Figure 5a), instead of a super-shift. Similar results were observed using the three Itr1 promoter-derived probes, which contained core sequences of the DOF binding sites at −238, −182, and −163 nt from the ATG of the Itr1 gene (Royo et al., 1996). All of them were able to produce specific retardation bands when incubated with the SAD or with the BPBF DOF factors (Figure 5b). Binding specificity was demonstrated by competition titration with cold probes (data not shown).

SAD activates transcription from the pyrimidine box of the Hor2 and Itr1 promoters in co-bombarded barley endosperms

The functional relevance of the interaction observed in vitro between SAD and the binding core A/TAAAG for DOF proteins (Yanagisawa and Schmidt, 1999) present in the Hor2 and Itr promoters was further tested in plant cells by transient expression assays in co-bombarded barley developing endosperms.

Figure 6(a) shows schematically the constructs used in the assays with the promoter of the Hor2 gene. The effector constructs contained the complete cDNAs from the Sad, Pbf and GAMyb genes controlled by the CaMV35S promoter plus the first intron of the maize Adh1 gene, followed by the 3′non-coding region of the nopaline synthase gene (nos). The reporters used were pBhor and pBhor* described by Mena et al. (1998) and pBhor** reported by Díaz et al. (2002). pBhor contained the −560 bp promoter region upstream from the translation initiation codon of the Hor2 gene fused to the GUS-reporter gene and the nos terminator. The mutated versions pBhor* and pBhor** differ from the wild type only by the base changes at the DOF and GAMYB binding sites respectively (5′-TGTAAAG-3′→ 5′-TGTAgAc-3′ in pBhor* and 5′-TAACAAC-3′→5′-TgACAAg-3′ in pBhor**).

Figure 6.

Transient expression assays of the Hor2 gene promoter by SAD and BPBF in barley developing endosperm.
(a) Schematic representation of the reporter and effector constructs. The effector constructs contained the complete cDNAs of Sad, Pbf and GAMyb genes (SAD, BPBF and GAMYB respectively) under the control of the CaMV35S promoter (p35S), followed by the first intron of the maize AdhI gene (I-AdhI) and are downstream flanked by the 3′nos terminator (nos). The reporter constructs consisted of the uidA reporter gene (GUS) driven by the 560 bp of the Hor2 promoter, upstream of the ATG translation initiation codon (pBhor) or under its mutated versions (pBhor* and pBhor**) differing in the two indicated nucleotide changes (p and m) at the prolamin box (P) and at the MYB box (M) respectively.
(b) Transient expression assays by co-bombardment of developing barley endosperms (18 daf) with the indicated combinations of reporter and effector plasmids at a 1:1 molar ratio. β-glucuronidase (GUS) activity was detected by histochemical staining and subsequent counting of blue spots per endosperm, and was expressed as percentage of GUS activity relative to the control without effector (correlation coefficient with luminometer data 0.98). Transient GUS expression driven by the mutated pBhor* and pBhor** constructs relative to its wild-type pBhor are shown in the insert. In each experiment, sets of five endosperms were bombarded and four replicates were made. Standard errors are indicated.

Barley developing endosperms were transiently transformed by particle bombardment with the reporters alone or in combination with the effectors at a 1:1 molar ratio. When endosperms were co-bombarded with the pBhor construct and the SAD effector, a 10-fold increase in GUS activity was obtained compared with the activity measured when the pBhor reporter was bombarded without effector (Figure 6b). BPBF and GAMYB transcription factors also act as trans-activators of the B-hordein promoter by increasing the GUS activity eight and four times, respectively, and the mixture of SAD plus BPBF or SAD plus GAMYB showed a higher trans-activation effect than either one of them without having a synergistic effect (Figure 6b). Using the mutated reporter pBhor* construct, a trans-activation of the GUS expression by both DOF effectors was obtained at much lower levels (about twofold), probably due to the presence of at least a second pyrimidine box of lower affinity in the Hor2 promoter under consideration (Díaz et al., 2002). Meanwhile, the GAMYB protein failed to activate the GUS reporter although the MYB-binding site was intact in the pBhor*, indicating that a functional DOF-binding site and probably the conformational change obtained by binding of SAD/BPBF was essential for GAMYB to function. Finally, and as expected, the reporter construct pBhor** that contains a mutated MYB-binding motif was not activated by the GAMYB protein but it was activated by the two DOF proteins under consideration, SAD and BPBF (Figure 6b). Nevertheless, as shown in the insert in Figure 5(b), the mutation of the DOF and MYB-binding sites at the pBhor* and pBhor** plasmids, respectively, resulted in a lower basal transcriptional activity than that supported by the wild-type promoter in pBhor plasmid (65 and 45% respectively).

We further investigated the in vivo interaction between SAD and BPBF with the Itr1 gene promoter. The constructs are schematically represented in Figure 7(a). The effector constructs used were those described above. As reporters, serial fragment deletions of the Itr1 gene promoter fused to the GUS reporter gene (Royo et al., 1996) were chosen. The four Itr1 promoter constructs contained different fragments spanning to positions −343, −211, −179 and −83, respectively, from the translation initiation codon. The longest fragment included three putative DOF-binding sites, D1 (5′-ACTTTT3′), D2 (5′-AAAAGC-3′) and D3 (5′-CCTTTA′-3′) located at positions −238, −182 and −163, that were deleted one by one in the successive promoter deletions. The shorter fragment (−83) had no putative DOF motifs. The transient expression data are shown in Figure 7(b). The co-transfection of the longest promoter fragment (−343 bp) with SAD or BPBF, as effectors, resulted in eight- and four-fold increase in GUS activity, respectively, over that directed by the reporter alone. Similar percentage increases in GUS activation mediated by SAD and BPBF as effectors over background levels were obtained when the −211 and −179 bp promoter fragments were used as reporters, which still contained two and one DOF-binding motifs respectively. Only the complete deletion of DOF-binding sites in the shorter promoter fragment (−83 bp) led to lack of trans-activation by the SAD or BPBF transcription factors and in addition supported the lowest GUS expression levels (see insert in Figure 7b). Developing endosperms co-bombarded with the longest reporter construct plus an equimolar mixture of both effectors, SAD and BPBF, showed additive effects.

Figure 7.

Transient expression assays of the Itr1 gene promoter by SAD and BPBF in barley developing endosperm.
(a) Schematic representation of the reporter and effector constructs. The effector constructs were those already used in the experiments described in Figure 6. The reporter constructs consisted of the uidA gene (GUS) under the control of deletion series of the Itr1 promoter (−343, −211, −179 and −83 bp from the translation initiation ATG codon) described by Royo et al. (1996) and downstream flanked by the 3′nos terminator (nos). The boxes indicated as D1, D2 and D3 correspond to the putative DOF-binding sites. Their sequences and positions in the promoter (referred to the ATG translation initiation codon) are indicated.
(b) Transient expression assays by co-bombardment of developing barley endosperms (18 daf) with the indicated combinations of reporter and effector plasmids at a 1:1 molar ratio. β-glucuronidase (GUS) activity was assayed and expressed as in Figure 6. Transient GUS expression driven by the deletion series expressed as percentage relative to the value obtained with the longest promoter fragment considered as 100% is shown in the insert. In each experiment, sets of five endosperms were bombarded and four replicates of each experiment were made. Standard errors are indicated.

Taken together, all these results indicate that SAD, as previously shown for BPBF, is a trans-activator of the Hor2 and Itr1 gene promoters in barley developing endosperm.


Numerous studies of the expression control of cereal endosperm-specific genes have identified conserved cis-motifs within their promoters, responsible for this specificity that interact with trans-acting factors. In barley, binding sites recognized by bZIP, DOF and R2R3MYB transcription factors are important for this regulation. The bZIPs of the Opaque2 class (BLZ1 and BLZ2) were the first TFs shown to participate in this process through recognition of the GLM cis-motif (Oñate et al., 1999; Vicente-Carbajosa et al., 1998). BPBF was a DOF protein that activated transcription of a B-hordein gene (Hor2) during endosperm development and was a repressor of a hydrolase gene (Al21) in aleurone cells upon germination (Mena et al., 1998, 2002) through recognition of equivalent cis-motifs in the corresponding promoters: (i) the prolamin box (5′-TGTAAAG-3′) in Hor2 and (ii) the pyrimidine box (5′-CCTTTT-3′) in the Al21 promoters. HvGAMYB, a R2R3MYB transcription factor, first characterized in germinating aleurone cells (Gubler et al., 1995, 1999) also had an important role in gene regulation during endosperm development through recognition of the 5′-AACAA-3′ motif present not only in rice seed glutelin but also in other reserve protein gene promoters such as those of barley prolamine gene promoters (Díaz et al., 2002; Wu et al., 2000).

In this study we describe that a barley DOF protein, SAD, previously shown to be a regulator of hydrolase-encoding genes in post-germinating aleurone cells (Isabel-LaMoneda et al., 2003), also has a role as transcriptional activator of two genes, Hor2 and Itr1, expressed in barley endosperm during development.

As shown by RT-PCR and in situ hybridization analyses, the accumulation of the Sad transcripts in the developing endosperm preceded and overlapped with the expression of the Hor2 and Itr1 genes and this expression pattern is compatible with SAD being a regulator of these genes. Moreover, the in vitro binding assays of SAD to the prolamin-box (5′-TGTAAAG-3′), in the context of the Hor2 and the Itr1 promoters, supported a putative regulatory function. In vivo experiments, based on co-bombardment of barley developing endosperms, confirmed this hypothesis, showing that SAD behaved as an activator of the GUS reporter activity controlled by both barley gene promoters. This view is consistent with the observation that the pyrimidine box motifs, D in the Hor2 gene promoter and D1, D2 and D3 in the Itr1 gene promoter, are positive elements because their mutations or deletions diminished the controlled activity of these promoters (Forde et al., 1985; Royo et al., 1996).

In transient expression assays in the homologous barley tissue, an additive activation effect of the SAD/BPBF combination over the Itr1 expression was observed, when the two DOF factors were co-bombarded together, probably due to more than one DOF protein interacting with more than one of the three motifs present in the Itr1 promoter. Two direct evidences suggest that the activation capacity of SAD and BPBF might be indirectly regulated through their interaction with GAMYB: (i) in transient assays with the Hor2 promoter mutated in the prolamine box, the DOF proteins lost their role as activators, as expected, but the GAMYB trans-activation was also reduced, although the MYB-binding site was intact; (ii) fluorescence reconstitution (BiFC) assays in plant nuclei indicated that both DOF factors (SAD and BPBF) were able to interact with GAMYB, but reconstitution could not be achieved when SAD was fused to the N-GFP and BPBF to the C-GFP, or vice versa, indicating that these two DOF proteins did not interact.

Several approaches have been used to detect protein–protein interactions in vivo, the yeast two-hybrid system being the method of choice most widely used. Bracha-Drori et al. (2004) and Walter et al. (2004) reported the adaptation of the BiFC technique to detect protein–protein interactions at the sub-cellular level in plant systems. In comparison with these two recent reports, our approach differs in that we have used the GFP instead of the yellow fluorescent protein (YFP). Moreover, we have fused the TF genes upstream of either the N- or the C-terminal GFP fragments, as Walter et al. (2004) did. However, Bracha-Drori et al. (2004) reported that they could only detect fluorescence when the gene fusions were made downstream of the YFP fragments. The SAD/GAMYB and BPBF/GAMYB interactions reported here corroborate interactions previously reported with these factor combinations in the yeast two-hybrid system (Díaz et al., 2002; Isabel-LaMoneda et al., 2003). Neither in the yeast system nor using the BiFC approach could we detect interactions between the two DOF proteins SAD and BPBF.

Development and germination are the two main phases of the seed cycle separated by the dormancy period. The success of germination depends on the synthesis of reserve compounds during seed development; the endosperm in cereal seeds being the tissue where these compounds are stored. Seed maturation and germination are programmes under strict developmental control, characterized by antagonistic physiological processes (Raz et al., 2001; White et al., 2000) where different sets of genes are activated. Hor2 and Itr1 genes are strongly expressed during endosperm development while they are not detected in the aleurone cells upon germination, despite the fact that some of the transcription factors involved in their regulation, such are SAD or GAMYB, are expressed in both phases and in both cases behave as activators. However, other TFs have antagonistic roles. For example, maize VP1 and barley BPBF activate and repress, respectively, the maturation and germination programmes of expression (Hoecker et al., 1998; Mena et al., 1998, 2002). The physiological role of the SAD protein, described here, probably should be contemplated within a more complex combinatorial interaction of multiple TFs that would result in diverse programmes of gene regulation in both phases. This latter possibility seems to be applicable also to the GAMYB protein, previously described (Díaz et al., 2002; Gubler et al., 1995).

Experimental procedures

RNA preparation and RT-PCR analysis

Developing endosperms (10–22 daf) and immature embryos (18 daf) were prepared from developing barley seeds (Hordeum vulgare cv. Bomi), frozen in liquid nitrogen and stored at −70°C until used for RNA extraction. For the isolation of RNA from barley aleurones, cv. Himalaya seeds were used. Aleurone layers for transient assays were isolated 24 h after imbibition as described (Isabel-LaMoneda et al., 2003). Total RNA for PCR analysis was purified from several tissues of barley by the Rneasy protocol (Qiagen Inc., Hilden, Germany). Contamination by genomic DNA in the RNA preparation was avoided by DNase treatment using the DNA-free system (Ambion Inc., Austin, TX, USA). First-strand cDNA synthesis was primed with random hexamers and catalysed by M-MuLV Reverse Transcriptase according to the manufacturer's recommendations (Amersham Pharmacia Biotech, Freiburg, Germany). Specific probes for Sad and Pbf cDNAs were obtained by amplification of 497- and 460-bp fragments of the corresponding cDNAs using the primers previously described by Isabel-LaMoneda et al. (2003) and Mena et al. (2002) respectively. The complete cDNA encoding the barley trypsin inhibitor CMe (Itr1 gene) and a fragment of 896-bp of the B-hordein cDNA (Hor2 gene) were amplified using as forward and reverse primers, 5′-ATGGCGTTCAAGTACCAG-3′ and 5′-TTACAAGACCACTCC-3′ for Itr1 gene and 5′-GCAATCCGTGCAATCGTC-3′ and 5′-CACCGCTACATCGACA-3′ for the Hor2 gene. The 18S amplico was used as internal control using a mixture of 18S primers/competimers (Ambion Inc.) at a 2:8 molar ratio. The resulting PCR products were analysed by 2% agarose gel electrophoresis, visualized by ethidium bromide staining and the amplification sequenced to verify the specific amplification of the desired target messages.

In situ hybridization

Developing barley seeds (20 daf) were fixed, treated and sectioned as described by Díaz et al. (2002). Tissue sections were hybridized using a solution (100 μg ml−1 tRNA, 6x SSC, 3% SDS and 50% formamide) containing approximately 100 ng μl−1 antisense or sense DIG-labelled RNA probe corresponding to the 684-bp fragment (positions −456 to 228 from the ATG initiation codon) of the Sad gene. Hybridization was performed overnight at 52°C, followed by two washes in 2x SSC and 50% formamide for 90 min at the same temperature. Antibody incubation and colour detections were carried out according to the manufacturer's instructions (Boehringer, Mannheim, Germany).

Epidermal onion cell transformation

To create the translational fusions of Sad, Pbf and GAMyb genes to the GFP reporter gene, the corresponding cDNAs were PCR-amplified using the following primers:


The amplified products were independently cloned in-frame to the GFP gene in the plasmid psmRS-GFP containing the CaMV35S promoter and a soluble modified red-shifted GFP (Davis and Vierstra, 1998). The constructs were called 35S::SAD-GFP, 35S::BPBF-GFP and 35S::GAMYB-GFP. As control, the plasmid without TF insert (construct 35S::GFP) was used.

Inner epidermal layers of onion bulbs (Allium cepa) locally purchased were peeled and placed onto MS/2 agar-medium as described by Borrell et al. (2002). Particle bombardment was performed with a biolistic Helium gun device (PSP-1000; DuPont, Hercules, CA, USA) basically as described by Díaz et al. (2002), with the following modifications: each shot delivered 140 ng of DNA using rupture discs of 1100 psi and a distance between macrocarrier and sample of 9 cm. The GFP fluorescence was observed after 24 h of incubation at 22°C in the dark, under BP 450/90 FT 510 LP520 Zeiss filters in a Zeiss Axiophot microscope.

Bimolecular fluorescence complementation

To analyse interactions among SAD, BPBF and GAMYB proteins in plant cells, we used a modification of the technique based on the complementation between two non-fluorescent GFP fragments that can associate to form a BiFC, when they are brought together by the interaction of proteins fused to each one of the fragments (Hu et al., 2002).

First, the GFP gene was divided into two non-overlapping fragments by a PCR technique using two couples of primers. The sense and antisense oligonucleotides, #N-GFPs: 5′-CGAGGATCCTTGGCATGAGTAAAGGAGAAG-3′ and #N-GFPas: 5′-CGAGAGCTCTTATGCCGTGATGTATAC-3′, amplified the N-terminal region (nt 1–462) encoding the first 154 amino acid residues of the GFP protein. The forward and reverse primers #C-GFPs: 5′-CGAGGATCCTTGGCGACAAACAAAAGAATG-3′ and #C-GFPas: 5′-CGAGAGCTCTTATTTGTATAGTTCTCC-3′, amplified a 255-nt portion (nt 463–717) which included the 85 C-terminal residues. These fragments were cloned into the psmRS-GFP plasmid (Davis and Vierstra, 1998) by replacement of the GFP gene. Finally, the whole ORF of Sad, Pbf and GAMyb cDNAs were amplified by PCR using the oligonucleotides previously described for the sub-cellular location experiments and were fused in-frame with both the N- and C-terminal encoding fragments of GFP in the psmRS-GFP plasmid, generating the following constructions: 35S::SAD-NGFP; 35S::SAD-CGFP; 35S::BPBF-NGFP; 35S::BPBF-CGFP; 35S::GAMYB-NGFP; 35S::GAMYB-CGFP.

Inner epidermal layers of onions were prepared and co-bombarded, as described above, using the appropriated plasmid combination indicated in each case. The fluorescence emission was observed after 36 h of incubation at 22°C in the dark, under a BP 450/90 FT 510 LP520 Zeiss filter. As control of nuclei location, tissue samples were stained with DAPI (Serva, Heidelberg, Germany). All observations were made with a Zeiss Axiophot microscope.

Electrophoretic mobility shift assays

The SAD and BPBF proteins were fused to the GST vector and expressed in E. coli as previously (Isabel-LaMoneda et al., 2003; Mena et al., 1998). The probe containing the putative binding consensus DOF-binding site from the Hor2 promoter (HD) and its mutated version (hd) were generated as in Mena et al. (1998). Three probes (ItD1, ItD2 and ItD3) including a DOF-like binding site derived from the Itr1 promoter were produced by annealing the complementary single-stranded oligonucleotides, indicated below, to generate 5′-protruding ends. All probes were end-labelled with α32P-dATP by the fill-in reaction catalysed with the Kleenow (exo-free) DNA-polymerase (United States Biochemicals, Cleveland, OH, USA) and were purified from an 8% polyacrylamide gel (29:1 cross-linking).


DNA-binding reactions, competitions with the unlabelled or mutated probes and final analysis of DNA–protein complexes were performed as described by Díaz et al. (2002).

Transient expression assays in barley developing endosperm

The complete Sad, Pfb and GAMyb cDNAs under the control of the 35S promoter plus the first intron of the Adh1 gene (Gubler et al., 1995; Isabel-LaMoneda et al., 2003; Mena et al., 1998) were used as effector constructs. The reporter constructs pBhor, pBhor* and pBhor**, containing the specific promoter from the Hor2 gene and two mutated versions of it fused to the β-glucuronidase reporter (uidA gene) were previously described (Díaz et al., 2002; Mena et al., 1998). The reporter constructs containing the promoter from the Itr1 gene were obtained by linking a deletion promoter series to the uidA gene, as published by Royo et al. (1996). Gold particle coating and bombardment conditions were performed according to Díaz et al. (2002). After bombardment, the endosperms were incubated at 25°C for 24 h and GUS expression was determined histochemically, following Jefferson (1987). GUS activity was calculated as the mean value of blue spots per endosperm in each assay and expressed as percentage, considering 100% as the value obtained with the reporter constructs without effectors. The histochemical data were directly correlated with the GUS expression quantified by chemiluminiscence (GUS-light kit; Tropix, Bedford, MA, USA) per milligram protein with a correlation coefficient of 0.98 (data not shown).


Financial support from the Ministerio de Educación y Ciencia, Spain (project no. BMC 2003-06345) is gratefully acknowledged. M.M. and I.R.-S. are recipients of a Ramon y Cajal contract and a predoctoral fellowship, respectively, from the Ministerio de Educación y Ciencia of Spain. The technical assistance of Mar González is also acknowledged.