Introduction

Gibberellins are plant hormones directly related to important plant growth and development events including seed germination, flowering and ovule development (Sun 2010; Gupta and Chakrabarty 2013; Gomez et al. 2016; Voegele et al. 2011). The gibberellin receptor GIBBERELLIN-INSENSITIVE DWARF1 (GID1) and DELLA proteins, key negative regulators of gibberellins, act on the signal transduction pathway of gibberellin in Arabidopsis thaliana and rice (Griffiths et al. 2006; Ueguchi-Tanaka et al. 2005). The binding of gibberellin to the GID1 receptor causes a conformational change in the N-terminal extension of GID1, which inhibits gibberellin action (Sun 2010). The formation of the gibberellin–GID1–DELLA complex results in the rapid degradation of DELLA proteins, releasing the action of gibberellin by destabilizing and degrading DELLA proteins (Davière and Achard 2013; Hartweck and Olszewski 2006; Sun 2010). The GID1 gene was first identified in rice and encodes a protein with similarity to hormone-sensitive lipases (HSL) (Ueguchi-Tanaka et al. 2005). Overexpression analysis of GID1 in rice shows a hypersensitive phenotype to gibberellin, growing taller, containing long light-green leaves and with reduced fertility when compared with the control, while the gid1 mutant presents insensitivity to gibberellin, suggesting that GID1 is the only gibberellin receptor in rice (Ueguchi-Tanaka et al. 2005). DELLA proteins participate in the correct formation of ovule integument and seed development in Arabidopsis, as shown in the della global mutant (the quintuple gaiT6 rgaT2 rgl1-1 rgl2-1 rgl3-1 mutant). Short integuments and no distinction between the inner and outer cell layers of the integuments caused defects in the testa formation of della global mutant seeds. Moreover, the number of abortions increased and seed density per fruit was reduced (Gomez et al. 2016).

In angiosperms, ovule development starts with the formation of the ovule primordia from the placental tissue of the ovary. During early ovule development, a single sub-epidermal cell differentiates from the nucellus, leading to the formation of the megaspore mother cell (MMC). The inner and outer integuments begin to develop as the MMC undergoes meiosis, to give rise to four haploid megaspores. Three of them degenerate and the only one to survive becomes a functional megaspore, usually the one closest to the chalazal region. After three rounds of mitosis and cellularization, an embryo sac (ES) of the Polygonum-type, with three antipodal cells, a diploid central cell, two synergids, and one egg cell is formed, typical of plants of sexual reproduction (Schneitz et al. 1995; Grossniklaus and Schneitz 1998; Yang et al. 2010).

In Brachiaria, a genus of forage grass of the Poaceae family, reproduction may be sexual or asexual by apomixis, the latter resulting in progenies that are genetically identical to the mother plants (Carneiro et al. 2006). The apomictic reproduction by apospory predominates in these plants. In apospory, nucellar cells near the MMC differentiate into aposporous initial cells, which will undergo two mitosis events to form unreduced ES of the Panicum-type, with a diploid central cell, two synergids, and one egg cell. The MMC degenerates during meiosis, or complete meiosis resulting in four megaspores, which will degenerate (Araujo et al. 2000; Dusi and Willemse 1999).

In Brachiaria brizantha (Syn. Urochloa brizantha), a putative GID1-like gene of 606 nucleotides was previously detected by RNA-Seq analysis, as highly expressed in early stages of ovary development. In the model plant Arabidopsis, plant hormones were related to a fundamental role in the initiation of the ovule primordia and in the correct determination of the number of ovules (Bartrina et al. 2011; Cucinotta et al. 2014). Three genes were identified in these plants and considered as orthologous to the rice GID1, by a database search and gibberellin receptor activity of their encoded protein: AtGID1a, AtGID1b and AtGID1c (Nakajima et al. 2006). Functional analysis of double mutant plants gid1a gid1b showed reduced fertility; gid1a gid1c also reduced fertility and showed fruits with reduced length; gid1b gid1c did not show any phenotypic difference from wild-type plants (Gallego-Giraldo et al. 2014; Griffiths et al. 2006). The triple gid1a gid1b gid1c mutant showed to be extremely dwarf-sized and was insensitive to gibberellin, resulting in complete infertility. These results suggest that GID1 is the only gibberellin receptor in Arabidopsis and that this gene participates in reproductive development (Griffiths et al. 2006). The expression patterns of the three genes of Arabidopsis at anthesis, characterized using pAtGID1::AtGID1-GUS fusion constructs, showed a GID1a expression signal in all tissues, mainly surrounding the ES; GID1b only appeared in the chalaza region of the ovule and GID1c only in valves, with no expression in ovules (Gallego-Giraldo et al. 2014). Phenotypic traits of the gid1-1 rice mutant, such as dwarf phenotype and insensitivity to gibberellin, were restored to normal with the overexpression of AtGID1a, AtGID1b and AtGID1c (Nakajima et al. 2006), showing the conserved nature of GID1 in these species. In Arabidopsis, the role of GID1 is associated with seed-set and fruit growth (Gallego-Giraldo et al. 2014); however, the role of GID1 during ovule development is still unclear. Studies related to GID1 might contribute to clarifying the mechanisms underlying the action of gibberellin in ovule development, as well as its participation in the early differential ES development of sexual and apomictic plants.

Our study aimed to evaluate the expression of GID1 during ovary development and, particularly, in the onset of ES of sexual and apomictic plants. The B. brizantha gene, named BbrizGID1, showed higher expression during ovule development at megasporogenesis. We detected BbrizGID1 transcripts in the MMC of sexual and apomictic B. brizantha and in the surrounding nucellar cells of the apomicts. In an attempt to mimic BbrizGID1 expression in Arabidopsis ovules, we investigated the results of overexpressing AtGID1a under the control of the ovule-specific SEEDSTICK promoter (pSTK) (Kooiker et al. 2005) and the 35S constitutive promoter of Cauliflower mosaic virus (p35S) (Benfey et al. 1990). Morphological characterization of reproductive organs, in particular ovules, in the transformed plants suggested that AtGID1a is involved in the differentiation of additional MMC-like cells from nucellar cells near the MMC and in the integument development, suggesting either a delayed megasporogenesis or an earlier growth of integuments in the transgenic plants. Our findings on the GID1 expression during ovule development provide valuable information on the role of GID1 in the ovule development of sexual and apomictic plants.

Materials and methods

BbrizGID1 sequence analysis

To verify the similarity of the BbrizGID1 sequence previously identified in RNA-Seq libraries of sexual and apomictic ovaries of B. brizantha, basic local alignment search tool (BLAST) (Altschul et al. 1990, 1997) was performed against the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov) database. BLASTn and BLASTx were used to compare nucleotides and translated sequences, respectively. Sequences showing the most significant nucleotide and amino acid sequence homology were used in multiple sequence alignments with ClustalW2 (Chenna et al. 2003). The cutoff e value of ˃ 2e− 57 and identity above 90% was used to define homologs. The Arabidopsis Information Resource database (TAIR, http://arabidopsis.org) was also used for comparative analyses between the BbrizGID1 sequence and the sequences from the Arabidopsis model plant. The genomic sequences of BbrizGID1 from sexual and apomictic genotypes were obtained searching genomic libraries under construction. Restriction map analysis of BbrizGID1 from sexual and apomictic genomes was performed with Webcutter 2.0.

Plant material

Two accessions of B. brizantha from Embrapa’s germplasm collection, grown in the field, were used in this work: BRA 002747, a sexual diploid (2n = 2x = 18), and BRA 000591, a facultative apomictic tetraploid (2n = 4x = 36) named B. brizantha cv. Marandu, with 98% of apospory (Araujo et al. 2007). Ovaries and anthers of both accessions were dissected from flowers of different plants before anthesis (I, II, III and IV), at different stages of development, under a Zeiss Stemi SV11 stereomicroscope. Ovaries at stages I and II contain ovules in megasporogenesis, whereas stages III and IV contain ovules in megagametogenesis (Araujo et al. 2000). The occurrence of nucellar cells and MMC characterizes ovules at stage I, in both sexual and apomictic plants. At stage II, pistils present ovules with degenerated tetrads, and in apomictic plants, aposporous initials can be visualized. At stage III, ovules with coenocytes are observed. Ovaries at stage IV contain mature ES, before anthesis (Araujo et al. 2000; Rodrigues et al. 2003). Arabidopsis wild-type (ecotype Columbia) and the pAtGID1a::AtGID1a-GUS, pAtGIDb::AtGID1b-GUS and pAtGID1c::AtGID1c-GUS transgenic lines (Gallego-Giraldo et al. 2014; Suzuki et al. 2009), provided by Dr. Perez-Amador, were cultivated in a growth room at 22 °C under long-day (16 h light per 8 h dark) conditions.

Southern hybridization analysis

Genomic DNA was extracted from young leaves of B. brizantha using a Dellaporta procedure (Dellaporta et al. 1983) modified by Leblanc et al. (1997). To estimate the copy number of BbrizGID1, 30 µg of genomic DNA from each genotype was digested with EcoRI, HindIII and XbaI at 2.5 U/µg of DNA. A 300 bp fragment PCR amplified with the primers GGACAGGGACTGGTACTGGAAG and CTGAGGAAGTCGGCGATCTC for BbrizGID1 without internal restriction sites for the enzymes used to digest the genomic DNA was used as probe. Labeling the probe with 32P –αdCTP, hybridization and radioactive detection assays were performed as described (Sambrook and Russell 2001). The hybridized membrane was placed in the cassette containing Imaging Plate (IP) and the image generated on the device Phosphorimager™ FUJIFILM FLA-3000 SERIES, using the Image Reader FLA-3000 V1.8 series programs and image display by Multi Gauge V3.0.

RNA isolation and RT-qPCR analyses

In sexual and apomictic B. brizantha plants, total RNA was extracted in three biological replicates, each constituted of a pool of 250 ovaries, at each of the four stages, using TRIZOL® (Invitrogen) as previously described (Rodrigues et al. 2003). Reverse transcriptase reaction was performed with SuperScript III (Invitrogen) and 2 ug of RNA according to the manufacturer’s protocol. RT-qPCRs were done using SYBR Green PCR Master Mix Green (Applied Biosystems). Oligonucleotide pairs for BbrizGID1 were designed using Primer 3.0 program (Rozen and Skaletsky 2000): forward (TCAAGCTCGTGTACCGCGAG) and reverse (CTGAGGAAGTCGGCGATCTC). The analysis of data was performed by the method ΔΔCt, with BbrizUBCE as reference gene, encoding the ubiquitin (forward GGTCTTGCTCTCCATCTGCT) and (reverse CGGGCTGTCGTCTCATACTT), as previously used (Silveira et al. 2009).

In Arabidopsis plants, total RNA was extracted from 1 to 2 inflorescences in three biological replicates of pSTK::AtGID1a transgenic plants and p35s::AtGID1a transgenic plants, using NucleoZOL® (Macherey–Nagel) according to the manufacturer’s instructions. The total RNA was incubated with the Ambion TURBO DNA-free DNase (Invitrogen) and then reverse transcribed using the ImProm-II™ reverse transcription system (Promega). Oligonucleotide pairs were designed using Primer 3.0 program for AtGID1a forward (GCTGCGAGCGATGAAGTT) and reverse (GATTGGCGTTTGCAGTGA), UBIQUITIN10 forward (CTGTTCACGGAACCCAATTC) and reverse (GGAAAAAGGTCTGACCGACA), and ACTIN2-8 forward (CTCAGGTATTGCAGACCGTATGAG) and reverse (CTGGACCTGCTTCATCATACTCTG). UBIQUITIN10 and ACTIN2-8 genes were used as references and gene expression analysis was performed using the iQ5 Multicolor real-time PCR detection system (Bio-Rad) with SYBR Green PCR Master Mix (Bio-Rad). The assay was conducted in three technical replicates of each of the three different biological replicates for each sample.

Expression analysis by in situ hybridization

The RNA probe was synthesized using the DIG RNA labelling kit (Roche) according to the manufacturer’s protocol. The same PCR fragment of 300 bp from BbrizGID1 located in the 3′ region, flanked by positions 1334–1633 of the nucleotide sequence, was used as a probe. The BbrizGID1 fragment was cloned into pGEM-T Easy Vector System I (Invitrogen) and used as a template for in vitro transcription with SP6 and T7 polymerases, used as sense and antisense probes, respectively. In situ hybridization was performed in semi-thin sections of 3.5 μm of ovaries at megasporogenesis of sexual and apomictic plants. The sample preparation and in situ hybridization were carried out as previously described (Alves et al. 2007; Silveira et al. 2012). Hybridized sections were observed with a Zeiss Axiophot light microscope.

Construction of vectors and transformation of Arabidopsis plants

The full-length Arabidopsis AtGID1a gene was cloned using the Gateway system (Invitrogen) and the primers ATGGCTGCGAGCGATGAA and TTAACATTCCGCGTTTACAAAC. The promoters of the SEEDSTICK gene, pSTK (Kooiker et al. 2005), to drive the expression in ovules, and the 35S constitutive promoter of the Cauliflower mosaic virus (p35S) (Benfey et al. 1990) were used. Plants of Arabidopsis ecotype Columbia were transformed with p35S::AtGID1a and pSTK::AtGID1a constructs by floral dip method (Clough and Bent 1998). The pSTK::AtGID1a transformant lines from the ecotype Columbia were selected using BASTA, and p35S::AtGID1a transformant lines from the ecotype Columbia were selected using kanamycin. The pKNU:nlsYFP Arabidopsis line (Tucker et al. 2012) provided by Dr. Tucker was also transformed with p35S::AtGID1a by floral dip method (Clough and Bent 1998). The transformant lines were selected using kanamycin.

GUS expression analysis

The pAtGID1a::AtGID1a-GUS, pAtGIDb::AtGID1b-GUS and pAtGID1c::AtGID1c-GUS lines (Gallego-Giraldo et al. 2014; Suzuki et al. 2009) were analyzed at different stages of the ovule development. β-Glucuronidase (GUS) assays were performed on inflorescences overnight at 37 °C, as previously described (Liljegren et al. 2000). Samples were incubated in a clearing solution of chloral hydrate:glycerol:water (8:1:2; w/v/v), dissected and observed by DIC microscopy.

Morphological analysis of Arabidopsis flowers by DIC microscopy

Flowers at different developmental stages were collected, fixed with 9:1 ethanol:acetic acid and cleared overnight using clearing solution. After clearing, ovules were dissected from premature pistils on a slide using a stereomicroscope, mounted with a cover slip, and subsequently observed using a Zeiss Axiophot D1 microscope equipped with DIC optics. Images were captured on an Axiocam MRc5 camera (Zeiss) using the Axiovision program (version 4.1). Three different plants of wild-type (ecotype Columbia) and three independent events of each transformant line p35S::AtGID1a, pSTK::AtGID1a and the second generation of pKNU:nlsYFP p35S::AtGID1a line were analyzed. Three independent events of pKNU:nlsYFP p35S::AtGID1a line were observed with a Zeiss Axiophot microscope using a filter with 475/40 nm laser line for excitation. Emissions were detected between 530/40 nm. Images were captured on an Axiocam MRc5 camera (Zeiss) using the Axiovision program (version 4.1).

Results

BbrizGID1 encodes a protein with similarity to the GID1 gibberellin receptor

Previous RNA-Seq analysis pointed to a sequence of 606 nucleotides similar to the GID1-like gene from different plants. This sequence was present in both genomic libraries and was named BbrizGID1. Genomic libraries of sexual and apomictic B. brizantha (unpublished results) were screened and a full-length DNA sequence, constituted of 1646 nucleotides (Supplementary Fig. S1) and coding 352 amino acids (Fig. 1), was found in this analysis. The BbrizGID1 consists of two exons and one intron, similar to the GID1 structure from rice (Ueguchi-Tanaka et al. 2005). The predicted amino acid sequences of BbrizGID1 from sexual and apomictic plants have the same number of amino acids, differing in two amino acids at positions 152 and 217. In position 152, an isoleucine (I) found in sexual plants is replaced by a valine (V) in apomicts. In position 217, a lysine (K) from sexual plants is replaced by an arginine (R) in apomicts (Supplementary Fig. S2).

Fig. 1
figure 1

Multiple sequence alignment by ClustalW of BbrizGID1 sequence, from the apomictic plants, showing the Arabidopsis thaliana and the most similar sequences found: Setaria italica (Accession Number XP_004962116.1), Sorghum bicolor (XP_021303311.1), Saccharum officinarum (CAP64326.1), Zea mays (CAP64327.1), Brachypodium distachyon (XP_003568469.1), Oryza sativa japonica group (XP_015639961.1), Hordeum vulgare (CAO98733.1), AtGID1a (At3g05120), AtGID1b (At3g63010) and AtGID1c (At5g27320). Gray highlight shows the amino acid residue conserved amongst the sequences. Squares represent regions identified as conserved within the hormone-sensitive lipase (HSL) family, HGG and GXSXG motifs. Black arrows show the three catalytic centers, serine (S), aspartic acid (D), and isoleucine (I), which replace histidine (H), characteristic of the HSL family in all the GID1 sequences of the Poaceae plants. Numbers indicate the position from the start codon. White arrows indicate glycine (G) and arginine (R) residues that are essentials for maintaining the gibberellin binding activity in OsGID1 in rice (Ueguchi-Tanaka et al. 2005)

Full size image

To analyze the similarity of BbrizGID1 with GID1 sequences from other plants, an NCBI database search was performed. The BbrizGID1 predicted amino acid sequence showed similarity to the sequences of GID1-like gibberellin receptors of other plants of the Poaceae family (Fig. 1). The most similar sequences were from the following plants (with respective percentage identity): Setaria italica (92%), Sorghum bicolor (91%), Saccharum officinarum (90%), Zea mays (87%), Brachypodium distachyon (86%), Oryza sativa Japonica Group (84%) and Hordeum vulgare (83%). The similarity of BbrizGID1 to the three Arabidopsis genes was very close: AtGID1c (58%), AtGID1a (57%) and AtGID1b (55%). Alignment also showed the presence of the consensus sequence of the HSL family, including the conserved HSL motifs HGG and GXSXG in BbrizGID1 (Fig. 1). The BbrizGID1 deduced amino acid sequence showed two conserved catalytic centers with the HSL sequence, serine (S) and aspartic acid (D), whereas another catalytic center, histidine (H), was replaced by isoleucine (I) (Supplementary Fig. S3).

BbrizGID1 is present in the sexual and apomictic genotypes of B. brizantha

In the apomictic genotype of B. brizantha, the Southern blot displayed three fragments after EcoRI DNA digestion, four after HindIII and two after XbaI. In the sexual genotype, one fragment hybridized after EcoRI DNA digestion, three after HindIII and two after XbaI (Fig. 2). Restriction map analysis of genomic DNA from both genotypes (Supplementary Fig. S1) showed that there is no site for EcoRI and XbaI and a site for HindIII is found at the intron region, position 444. The probe hybridization site is between positions 1334 and 1633 from the exon region of the nucleotide sequence. Genomic hybridization confirmed that BbrizGID1 is present in both sexual and apomictic genotypes in a different restriction pattern.

Fig. 2
figure 2

Genomic DNA analysis of apomictic (APO) and sexual (SEX) Brachiaria brizantha digested with EcoRI (E); HindIII (H) and XbaI (X). a Electrophoresis in 0.8% agarose gel in b fragments hybridized in Southern blot analysis with a BbrizGID1 32P-labeled probe. In APO three fragments were displayed after EcoRI DNA digestion, four after HindIII digestion and two after XbaI digestion. In SEX, one fragment was hybridized after EcoRI digestion, three after HindIII and two fragments after XbaI digestion. Image was generated on a phosphor imaging plate. C as control, the same fragment used to synthesize the probe

Full size image

BbrizGID1 expression in ovaries of sexual and apomictic plants is higher during megasporogenesis

Analysis of BbrizGID1 expression profile from the RNA-Seq data of ovaries of sexuals and apomicts suggested higher expression of BbrizGID1 in ovaries at megasporogenesis stage than in ovaries at megagametogenesis stage (data not shown). To confirm these results, the transcript levels of BbrizGID1 were compared quantitatively in ovaries of sexual and apomictic plants (Fig. 3), using pistils of B. brizantha in four different stages of development as previously described (Araujo et al. 2000; Rodrigues et al. 2003). Ovaries from pistils in stage I, corresponding to the onset of megasporogenesis, showed higher expression of BbrizGID1, whereas BbrizGID1 expression decreased in the later developmental stages. At the 5% significance level, analysis of deviance (ANODEV) test showed that sexual and apomictic reproduction modes do not interfere with BbrizGID1 expression (F = 0.903, GL = 19, GL = 1, p = 0.353), but there are significant differences between stages I, II and III of ovule development in both plants (F = 49.66, GL = 19, GL = 3, p = 0.0001). Comparison of BbrizGID1 expression in stages III and IV, however, did not show a significant difference.

Fig. 3
figure 3

Boxplot showing the relative expression levels of the BbrizGID1 by RT-qPCR in Brachiaria brizantha. Analysis in developing ovaries before anthesis at megasporogenesis (stages I, II) and megagametogenesis (stages III and IV) of apomictic (APO) and sexual (SEX) plants. Bars represent standard deviation based on the mean of three biological samples. The mean quartiles, median and ANODEV were calculated in the R Core Team (2016) program. The significant differences between stages of ovary development are marked with asterisks (*p < 0.05)

Full size image

BbrizGID1 is expressed in MMC of sexual and apomictic B. brizantha and in nucellar cells of apomicts

In situ hybridization of BbrizGID1 was performed in the early stages of ovule development of sexual and apomictic plants to understand the cellular localization of BbrizGID1 within the initial events of ovule development (Fig. 4). The antisense probe of BbrizGID1 in the early stages of megasporogenesis resulted in high signal of hybridization in MMC of sexual (Fig. 4a, b) and apomictic plants (Fig. 4d, e). Nucellar cells in sexual ovules showed weak signal in comparison to the strong expression of BbrizGID1 in the nucellar cells of apomicts (Fig. 4d, e). In sexual and apomictic plants, a weak signal was detected in inner and outer integuments (Fig. 4a, d). The sense control probe did not show any hybridization signal in sexual and apomictic ovules, as expected (Fig. 4c, f).

Fig. 4
figure 4

Localization of BbrizGID1 expression by in situ hybridization in semi-thin sections of ovaries of Brachiaria brizantha sexual (a–c) and apomictic (d–f) plants. a Ovule with labeling in the megaspore mother cell (arrow) with weaker signal in the nucellar cells (nu), inner (ii) and outer (oi) integuments. b Detail of the megaspore mother cell (arrow) with strong signal of hybridization. c Control with sense probe showing no signal. d Detail of the ovule with labeling in the nucellar cells (nu) and the megaspore mother cell (arrow) with weaker signal in the inner (ii) and outer (oi) integuments. e Next section of the anterior ovule showing the cytoplasm of the megaspore mother cell (arrow). f Control with sense probe showing no signal. Scale bars a 12 μm, b 10 μm, c 20 μm and d–f 10 μm

Full size image

GID1 proteins show specific spatial expression patterns in Arabidopsis

To study the localization of GID1 (a, b, c) proteins during the first stages of ovule development, transgenic Arabidopsis lines carrying pAtGID1::AtGID1-GUS translational fusions (Gallego-Giraldo et al. 2014; Suzuki et al. 2009) were analyzed (Fig. 5). As visualized in the GUS assay, GID1a is present in the whole pistil during early development, before anthesis (Fig. 5a). The GUS signal was strongly detected in the inner and outer integuments of the ovules, at 2-III stage, in which MMC is in meiosis, and the inner and outer integuments already developing (Schneitz et al. 1995), of pAtGID1a::AtGID1a-GUS line. No signal was detected in MMCs and nucellar cells (Fig. 5b). The GUS signal was present throughout the mature ovule, including the ES and the integuments (Fig. 5c). In the pAtGID1b::AtGID1b-GUS line GUS signal was localized in the inner and outer integuments at 2-III-2-V stage (Schneitz et al. 1995) and no signal was detected in MMCs and nucellar cells (Fig. 5d, e). In mature ovules, at the FG7 stage in which the ES is formed, the GUS signal was localized mainly in the chalaza zone, at the base of the ES (Fig. 5f). AtGID1c was expressed specifically in the septum, but was completely absent in the ovules at 2-III stage and funiculi of pAtGID1c::AtGID1c-GUS line (Fig. 5g, h). GUS expression was not observed in ovules.

Fig. 5
figure 5

Histochemical GUS analysis in inflorescences and ovules of Arabidopsis thaliana pAtGID1::AtGID1–GUS transgenic lines before anthesis. a–c pAtGID1a::AtGID1a–GUS line; d–f pAtGID1b::AtGID1b–GUS line and g, h pAtGID1c::AtGID1c–GUS line. a GUS staining can be observed in stem (S), ovules (arrows) and anthers (arrow heads). b At 2-III stage of the ovule development GUS signal was localized in the inner (ii) and the outer (oi) integuments, the expression was not detected in MMC in meiosis (arrow) and nucellar cells. c In the mature ovules the GUS signal was localized throughout the embryo sac and the integuments. d In the 2-III stage and e first stage of gametogenesis, GUS signal was localized in the inner (ii) and outer (oi) integuments of the ovules, the expression was not detected in MMC in meiosis (arrow) and nucellar cells. f In the mature embryo sac the GUS signal was localized mainly in the chalaza (Ch) in comparison with the micropyle region (MP). g GUS signal was localized specifically in the septum (sp), but was completely absent in the ovules (ov) and funiculi. h GUS signal was undetectable in ovules, arrows showing the MMC in meiosis and the nucellar cells. MMC megaspore mother cell; Scale bars a 1 mm; b–f 20 μm and g, h 50 μm

Full size image

Phenotypic characteristics of Arabidopsis overexpressed AtGID1a lines are distinct from wild-type

To study whether ectopic AtGID1a was sufficient to determine the formation of MMC-like cell we have transformed Arabidopsis plants with p35S::AtGID1a and pSTK::AtGID1a. pSTK is specifically active during ovule development starting from stage I (Brambilla et al. 2008). We analyzed three transgenic plants containing p35S::AtGID1a and three containing pSTK::AtGID1a by RT-qPCR. Results showed that AtGID1a expression was higher in the inflorescences of plants from both groups of transgenic lines compared to wild-type plants (Supplementary Fig. S4). The early stages of ovule development of these transgenics were analyzed, especially the finger-like (2-I) stage, in which MMC is formed (Schneitz et al. 1995) and additional MMC-like cells were observed (Fig. 6). The pSTK::AtGID1a and p35S::AtGID1a plants showed asynchrony in development of the inner and outer integuments as regards megasporogenesis events if compared with wild-type (Fig. 6). These results suggest that ectopic expression of AtGID1a could interfere with proper integument development. The additional MMC-like cells of overexpressed AtGID1a plants were observed only in ovules at the finger-like stage (2-I). In later stages of development, only one functional megaspore was observed, suggesting that only one MMC completed the meiotic division and entered the gametogenesis process, as twin ES were not observed during the analysis of mature ovules. Results of the phenotype analysis of wild-type Arabidopsis and three lines of pSTK::AtGID1a, p35S::AtGID1a were evaluated for each plant (Supplementary Table S1). In ovules of wild-type Arabidopsis, additional MMC-like cells were also observed but in a reduced number, 17% of the total number of ovules analyzed. In transgenic plants, additional MMC-like cells were observed in 33 and 32% of the total number of ovules analyzed using pSTK or p35S promoter, respectively (Fig. 7). Statistical analysis (t test) showed that there is a significant difference between the number of additional MMC-like cells in the overexpressed AtGID1a lines compared to wild-type, independent of the promoter used for AtGID1a expression, pSTK (p = 0.009) or p35S (p = 0.013). In pSTK::AtGID1a and p35S::AtGID1a plants the number of seeds per silique increased relatively to wild-type. The transgenic plants from both genotypes showed an average of 55 seeds per silique in the 30 siliques counted with 1.5% of aborted seeds, while wild-type plants, under the same environmental conditions, presented an average of 46 seeds per silique in the 32 siliques counted with 2.5% of abortion rate. Statistical analysis (t test) showed a significant difference between the number of seeds of the transgenic plants ectopically expressing AtGID1a, independent of the promoter, pSTK (p = 0.0009) or p35S (p = 0.001) in respect of the wild-type plants (Fig. 8, Supplementary Table S1).

Fig. 6
figure 6

Ovules of wild-type and AtGID1a overexpression lines of Arabidopsis thaliana. a–d Wild-type showing normal female gametophyte development. a Finger-like stage (2-I) showing MMC-like cells (arrows). b FG1 stage displaying MMC (arrow) integument development. c FG3 with degenerated megaspores and FG. d FG6 stage, arrow indicates the central cell. e–h p35S::AtGID1a line. e Finger-like stage (2-I), showing MMC-like cells (arrows). f, g FG1 stage showing well-developed integuments. h FG6 stage, arrow indicates the central cell. i–n pSTK::AtGID1a line. i Finger-like stage (2-I), viewing MMC-like cells (arrows). j showing MMC (arrow) and inner (ii) and outer (oi) integuments already growing. k Tetrad (arrowhead) and growing inner (ii) and outer (oi) integuments. l FG1 stage showing functional megaspore (arrowhead), degenerated megaspores and well-developed integuments m FG6 stage with antipodal cells (arrow) and three nuclei at the micropylar pole (arrowhead). ii Inner integuments, oi outer integuments, MMC megaspore mother cell, asterisk degenerated megaspores, FG female gametophyte. Scale bars a 20 µm; b 10 µm; c 20 µm; d 40 µm; e 20 µm; f, g 50 µm; h 100 µm; i 50 µm; j–l 10 µm and m 20 µm

Full size image
Fig. 7
figure 7

Comparison among the number of ovules with additional MMC-like cells of wild-type and transgenic lines pSTK::AtGID1a and p35S::AtGID1a of Arabidopsis thaliana. Different letters indicate statistically significant differences between the number of additional MMC-like cells of transgenic plants and wild-type, independent of the promoter used for AtGID1a overexpression, pSTK (p = 0.009) or p35S (p = 0.013). WT wild-type, N number of ovules analyzed, MMC megaspore mother cell

Full size image
Fig. 8
figure 8

Comparison among the number of seeds per silique in pSTK::AtGID1a and p35S::AtGID1a Arabidopsis thaliana relatively to wild-type (WT) plants. Different letters indicate statistically significant differences between the number of seeds of the wild-type and the transgenic plants ectopically expressing AtGID1a, independent of the promoter used, pSTK (p = 0.0009) or p35S (p = 0.001)

Full size image

To investigate if the additional MMC-like cells had MMC identity, a pKNU:nlsYFP line (Tucker et al. 2012) was transformed with p35S::AtGID1a. The second generation of transgenic lines overexpressing AtGID1a showing 28% of additional MMC-like cells was analyzed (Supplementary Table S1). Plants pKNU:nlsYFP p35S::AtGID1a showed only one cell expressing the YFP signal, indicating that one out of the two MMC-like cells acquired the proper MMC identity (Fig. 9).

Fig. 9
figure 9

Arabidopsis thaliana ovules of pKNU:nlsYFP line with the MMC YFP-labeled overexpressing AtGID1a under the control of p35ScaMV. a Ovules at the 2-III stage with two MMC-like cells (arrows) and b at the same moment using UV light showed signal of fluorescence only in the MMC (arrow), not in the additional enlarged cells of the overexpressed plants. c Ovules finger-like (2-I) stage, showing yellow florescence in the MMC cells. MMC megaspore mother cell; scale bars a–c 10 μm

Full size image

Discussion

GID1 acts in the activation of gibberellin via degradation of DELLA proteins (Davière and Achard 2013; Ueguchi-Tanaka et al. 2007). The detection of GID1 in the transcriptome database of ovaries of sexual and apomictic B. brizantha led us to investigate a putative role of this gene in the differential ovule development of sexual and apomictic plants. Alignment of the BbrizGID1 sequence in databases showed high similarity with sequences from other species of the Poaceae family and the conserved motifs, HGG and GXSXG, from the HSL family. It was suggested that three conserved amino acids of the HSL family form the catalytic triad center: serine (S), aspartic acid (D) and histidine (H) (Osterlund 2001). In all the GID1 sequences of the Poaceae plants, in which similarity to BbrizGID1 was detected, replacements of residues from this HSL catalytic center were verified. In rice, a replacement of the H by valine (V) did not affect the binding activity with gibberellin, although an absence of enzymatic activity was detected (Ueguchi-Tanaka et al. 2005). In B. brizantha, this same H of the catalytic triad is replaced by isoleucine (I) but S and D are conserved. The residues glicina (G) e arginina (R) are essential to maintain binding between the receptor and the bioactive gibberellin in rice (Nakajima et al. 2006; Ueguchi-Tanaka et al. 2005). Comparing the GID1 sequence of B. brizantha with Arabidopsis revealed around 60% similarity of the BbrizGID1 sequence with the three genes AtGID1a, AtGID1b and AtGID1c. In Arabidopsis, analysis of mutant combinations of the three genes showed an overlap of function among them (Griffiths et al. 2006). Complementation with each AtGID1 gene rescued the dwarf phenotype of the rice gibberellin-insensitive mutant, showing conservation of function among them (Nakajima et al. 2006). BbrizGID1 is present in the sexual and apomictic genotypes of B. brizantha. The restriction pattern of DNA from these plants, observed after HindIII digestion and genomic hybridization with BbrizGID1 probe, suggested a possible different number of copies of the gene in sexual and apomictic plants.

Quantitative analysis of BbrizGID1 expression showed the highest expression level in ovaries at early megasporogenesis, decreasing in the subsequent stages, independent of the reproductive mode of the plants. At this stage the formation of the MMC in sexual and apomictic genotypes was previously observed, but aposporous initial cells were still not detected in apomicts (Araujo et al. 2000). Analysis of these results suggested that BbrizGID1 might participate in the early development of ovaries of sexual and apomictic B. brizantha, when growing integuments and the differentiation of MMC from nucellar cells occur. Localization of transcripts of BbrizGID1 in ovules of B. brizantha from both reproductive modes revealed that MMC is expressing this gene, while in the surrounding nucellar cells the expression occurs only in the apomicts. These results suggested association of BbrizGID1 expression not only with the development of MMC but also with the nucellar cells, previous to the differentiation of aposporous initials.

The overexpression of the AtGID1a gene in Arabidopsis significantly increased the number of additional MMC-like cells of the wild-type Columbia ecotype, suggesting that this gene might be associated with the differentiation of an additional MMC-like cell in the nucellus. Analysis of the second generation of pKNU:nlsYFP p35S::AtGID1a suggested that the trait was stably inherited. Different ratios of additional MMC-like cells were previously reported in Arabidopsis ecotypes (Rodríguez-Leal et al. 2015). Analysis of ago9 mutants showed a percentage of abnormally MMC-like cells varying from 37.16 to 47.7%, with identity to gametic cells that do not undergo meiosis, a phenotype that resembles apomictic reproduction by apospory (Olmedo-Monfil et al. 2010). In our work, the identity of MMC was investigated by the expression of YFP marker genes under the control of the KNUCKLES promoter, which drives expression to the MMC (Tucker et al. 2012). Arabidopsis pKNU:nlsYFP line, overexpressing AtGID1a, showed only one cell with MMC identity in each ovule, comparable to what happens in plants of the pKNU:nlsYFP line, suggesting the possibility of identity other than with MMC to the MMC-like cells. It is important to note that the KNUCKLES promoter was not previously tested for identity of MMC-like cells.

Gallego-Giraldo et al. (2014) previously verified the differential expression of AtGID1a, AtGID1b and AtGID1c in ovules, at anthesis, using GUS staining. Expression of AtGID1a was detected in all tissues of the ovules, with higher expression observed in the surrounding of the ES, while AtGID1b expression was observed mainly in the chalaza zone, at the base of the ES, and AtGID1c was undetectable in the ovules. In our study, we detected AtGID1a, AtGID1b and AtGID1c expression reported by GUS assays during the early stages of ovule development, before anthesis. AtGID1a and AtGID1b were localized in the inner and outer integuments, not in MMCs and the surrounding nucellar cells. Expression signal of AtGID1c was not detected in the ovules, before anthesis. Our results suggested that AtGID1a and AtGID1b expression is associated with integument development in Arabidopsis. The association of the gibberellin and integument development was previously described (Gomez et al. 2016). In plants overexpressing AtGID1a, in comparison to the wild-type plants, an asynchrony between integument development and the stage of gametogenesis suggests that the ectopic expression of AtGID1a resulted in either a small delay in megasporogenesis or a faster growth of integuments in the transgenic plants. Consistent with the importance of gibberellin in the correct seed development to form seed-filled siliques, overexpressed AtGID1a plants were fertile without any sign of ovule or seed abortion and their siliques were seed-filled.

We demonstrated GID1 expression during megasporogenesis in ovules of sexual and apomictic B. brizantha, especially in the MMC. Specifically in ovules of the apomictic plants, BbrizGID1 is also expressed in the nucellar cells, previously to aposporous initial cell differentiation. In Arabidopsis, overexpression of AtGID1a triggered the differentiation of MMC-like cells in the nucellar region. These data suggest the involvement of GID1 in the early events of ovule and ES development.

Author contribution statement

LGF, DMAD, ASTI, LC and VTCC designed the project; LGF, DMAD, ASTI, ACMMG, MAM and VTCC performed experimental procedures; LGF, DMAD, MAM, LC and VTCC, participated in the research, analyzed the data and wrote the paper. All authors read and approved the final manuscript.