Introduction

The shoot-system architecture of vascular plants is largely dependent on the temporal and spatial pattern of meristem activity (Sussex and Kerk 2001). Meristems placed in leaf axils as well as on flanks of the inflorescence meristem produce side shoots and flowers, respectively (Grbic 2005; Barthélémy and Caraglio 2007; Barton 2010; Agusti and Greb 2013). The initiation and development of axillary meristems (AMs) in boundary zones are pivotal to establish plant architecture, and a fine integration of environmental factors and developmental pathways is required to control shoot branching (Wang and Li 2008; Leyser 2009; Žádníková and Simon 2014; Drummond et al. 2015; Rameau et al. 2015; Tian and Jiao 2015). Moreover, iterative branching is an important aspect of lifespan in plants and recent data support the hypothesis that stem cell arrangement and positioning of axillary meristems distribute somatic mutations around the main shoot, preventing their fixation and maximizing genetic heterogeneity (Burian et al. 2016).

Understanding how plant architecture is built represents an attractive topic to investigate plant domestication (Doebley et al. 2006). For instance, the TEOSINTE BRANCHED 1 (TB1) gene, which acts on branch size and inflorescence sex determination, represents the major contributor to the architectural conversion of the wild ancestor teosinte into maize (Doebley et al. 1997). TB1 is a member of the plant-specific transcription factor (TF) TCP family (Martín-Trillo and Cubas 2010; Uberti Monassero et al. 2013). Many other genes controlling AM initiation and outgrowth have been identified (reviewed in Bennett and Leyser 2006; Leyser 2009; Žádníková and Simon 2014). For example, in the GRAS gene family, mutations in Lateral suppressor (Ls) of tomato (Schumacher et al. 1999) and its homologous LAS of Arabidopsis (Greb et al. 2003) cause strong defects in AM formation, leading to loss of side shoots during vegetative development (Schumacher et al. 1999; Greb et al. 2003). Furthermore, the genetic defect of Ls homologous determines different plant architectures also in monocots. In fact, loss of MONOCULM 1 (MOC1) function in rice results in a lack of tillers as well as in a reduction of reproductive rachis branches and spikelets (Li et al. 2003), suggesting that the floral meristem is a modified AM (Fambrini et al. 2003; Bennett and Leyser 2006).

Additionally, within the largest MYB TF R2R3-MYB family of Arabidopsis, the genes REGULATOR OF AXILLARY MERISTEMS (RAX1), RAX2, and RAX3 are partially redundant in function and control early steps in AM initiation and development (Keller et al. 2006; Müller et al. 2006). In fact, the triple mutant rax1;rax2;rax3 fails to start AMs. In sunflower, the transcription of Ha-R2R3-MYB2, a RAX-like genes, was strongly restricted to a small domain within the boundary zone separating the shoot apical meristem (SAM) and the leaf primordia, also suggesting that Ha-R2R3-MYB2 could play a key role in establishing and (or) maintaining a leaf axil stem cell niche for AM formation (Pugliesi et al. 2013). In turn, RAX genes are homologues to BLIND, which also controls AM formation in tomato (Schmitz et al. 2002; Keller et al. 2006; Müller et al. 2006).

In rice, the lax panicle 1 (lax1) mutant displays a strong reduction in the number of panicle branches and spikelets (Komatsu et al. 2003) and fails to form AMs during vegetative development (Oikawa and Kyozuka 2009). Similarly, maize (Zea mays) barren stalk1 (ba1) mutant fails to initiate AMs during all stages of the life cycle (Gallavotti et al. 2004). LAX1 and ba1 transcripts accumulate at the adaxial boundary of developing lateral meristems and encode orthologous bHLH proteins (Gallavotti et al. 2004; Komatsu et al. 2003). In Arabidopsis, mutations in the gene REGULATOR OF AXILLARY MERISTEM FORMATION (ROX), which is the ortholog of LAX1 and BA1, compromised axillary bud formation during vegetative shoot development (Yang et al. 2012). ROX acts in concert to RAX1 and LAS to modulate AM formation (Yang et al. 2012).

The growth of AMs is inhibited in single-stemmed plants by the main shoot, a morphological process generally known as apical dominance (Cline 1991). The auxin indole 3-acetic acid (IAA) is essential for apical dominance. Auxin synthesized in the shoot apex moves basipetally in the polar auxin transport (PAT) stream and indirectly inhibits outgrowth of axillary buds (Snow 1929; Thimann and Skoog 1933; Okada et al. 1991; Reinhardt et al. 2000, 2003; Vernoux et al. 2000; Benková et al. 2003; Heisler et al. 2005; Bennett et al. 2006; Cheng et al. 2007; Cheng and Zhao 2007; Wu and McSteen 2007; Durbak et al. 2012; Taylor-Teeples et al. 2016). Recently, Mason et al. (2014) demonstrated that apical dominance strongly correlates with sugar availability, suggesting an important role about the relationship between hormones and sugars. In addition to auxin, a complex hormonal homeostasis involving citokinins, strigolactones, brassinosteroids, and gibberellins is required to specify the fate of cells placed in the boundaries region competent for AM initiation (Okada et al. 1991; Przemeck et al. 1996; Shimizu-Sato and Mori 2001; Beveridge and Kyozuka 2010; Domagalska and Leyser 2011; Cheng et al. 2013; Sassi and Vernoux 2013; Janssen et al. 2014; Wang et al., 2014a; 2016; Wang et al., 2014b; Taylor-Teeples et al. 2016; Yang and Jiao 2016).

Cultivated and wild sunflowers exhibit a number of morphological differences that trace back to the original domestication event, and five types of branching have been described (IBPGR 1985). For example, wild sunflower is characterized by a highly branched growth form with numerous small flowering heads. The origin of no branched sunflower, producing a single large head, has been a key event in the domestication process of this species (Heiser 1976; Burke et al. 2005; Baute et al. 2015). Although plastic in its response to environmental cues, including light, temperature, humidity, and nutrient availability (Reinhardt and Kuhlemeier 2002), the branching pattern is genetically determined. In sunflower, both dominant and recessive genetic controls of branching outgrowth are known (Putt 1964; Miller and Fick 1997; Fambrini et al. 2003) and four genes were proposed by Hockett and Knowles (1970), even if a more complex inheritance of branching was recognized (Sandu et al. 1999). As regards Helianthus genus, very few genes involved in AM differentiation have been isolated, examples being RAX- and GRAS-LIKE (SCL4/7) genes (Pugliesi et al. 2013; Fambrini et al. 2015).

Here, we isolate and characterize for the first time in sunflower the genes Ha-LS-LIKE (Ha-LSL) and Ha-ROX-LIKE (Ha-ROXL) putatively involved in AM initiation and/or outgrowth. The transcription analyses performed with real-time RT-PCR showed a low level of transcription of Ha-ROXL in all organs analyzed; differently a high mRNA steady state was detected for Ha-LSL in roots as well as vegetative and inflorescence shoots. In situ hybridization indicated a clear restricted boundary transcription for Ha-ROXL in vegetative and inflorescence shoots, while Ha-LSL expression pattern was not exclusive in boundary regions because signals were detected likewise in other cellular domains, in both vegetative and reproductive stages. The transcription of Ha-LSL was also extended in the early stage of development of lateral primordia suggesting a role of this gene in the initial outgrowth of lateral primordia.

Materials and methods

Plant material and growth conditions

Sunflower (Helianthus annuus L.) seeds of the inbred line MISSING FLOWERS (Fambrini et al. 2003) were germinated at 23 ± 1 °C in Petri dishes in the dark. Germinated seeds were transferred to 20-cm diameter pots containing a mixture of soil and sand. Plants were grown in a growth chamber at 23 ± 1 °C under a 16-h photoperiod (400 μmol photons m−2 s−1). Irradiance was provided by a mercury vapor lamps (Osram HQI-TS 250 W/NDN, Wembley, England). The plants were grown in three groups with 2-week sowing interval.

RNA extraction and isolation of Ha-LSL and Ha-ROXL cDNAs from sunflower

Total RNA was extracted from H. annuus samples with the TriPure Isolation Reagent, according to the manufacturer’s instructions (Roche Diagnostics GmbH). Total RNA (0.4 μg), from vegetative shoots (VS) of 20-day-old plants, was used with the iScript™ cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA), to produce the first strand cDNA in conditions recommended by manufacturer.

One expressed sequence tag (EST) of Helianthus ciliaris (GenBank accession number EE633872) with high homology (4e-64) with Ls of Solanum lycopersicum (GenBank accession number AJ303345) was identified to NCBI. The EST sequence was used to choose the primers for the PCR amplification of a Ha-LSL cDNA fragment in sunflower. One cDNA fragment, named Ha-LSL1, was obtained with the primers LSEXF and LSEXR2 (Table S1). The Ha-LSL1 cDNA fragment sequencing allowed choosing primers to use in the 5′-Rapid Amplification of cDNA End (RACE) approach, according to the manufacturer’s instructions (Invitrogen). Ha-LSL1-specific primers LSRACE1 and LSRACE2 (Table S1) were used in the first and second PCR, respectively. The first PCR conditions were as follows: 94 °C for 4 min, 30 cycles (30 s at 94 °C, 30 s at 68 °C, 1 min 30 s at 72 °C), and 72 °C for 7 min. The second PCR conditions were 94 °C for 4 min, 30 cycles (30 s at 94 °C, 30 s at 68 °C, 60 s at 72 °C), and 72 °C for 7 min. Full-length cDNA (Ha-LSL) was obtained with the following primers: LSIN and LSFIN (Table S1). The PCR conditions were 94 °C for 4 min, 30 cycles (30 s at 94 °C, 30 s at 62 °C, 1 min 30 s at 72 °C), and 72 °C for 7 min.

One cDNA fragment of Ha-ROXL, named Ha-ROXL1, was obtained with the primers ROXF and ROXR (Table S1), placed in conserved regions shared between ROX genes, of Arabidopsis thaliana, Theobroma cacao, and Nicotiana tomentosiformis (GenBank accession numbers NM_120209.1, XM_007026565.1, and XM_009611007.1, respectively). The Ha-ROXL1 cDNA fragment sequencing allowed choosing primers to use in the 5′-3′-RACE. A 3′-RACE was conducted using the specific primer ROX3RACE1 and the Universal Amplification Primer 9 (UAP9) (Table S1) with the following PCR conditions: 4 min for 94 °C, 35 cycles (30 s at 94 °C, 30 s at 62 °C, 40 s at 72 °C), and 72 °C for 7 min. For 5′-RACE, Ha-ROXL1-specific primers ROX5RACE1 and ROX5RACE2 (Table S1) were used in the first and second PCR, respectively. The first PCR conditions were 94 °C for 4 min, 30 cycles (30 s at 94 °C, 30 s at 62 °C, 30 s at 72 °C), and 72 °C for 7 min. The second PCR conditions were 94 °C for 4 min, 30 cycles (30 s at 94 °C, 30 s at 62 °C, 30 s at 72 °C), and 72 °C for 7 min. Full-length cDNA (Ha-ROXL) was obtained with the following primers: ROXIN and ROXFIN (Table S1). The PCR conditions were 94 °C for 4 min, 30 cycles (30 s at 94 °C, 30 s at 63 °C, 50 s at 72 °C), and 72 °C for 7 min.

DNA extraction and isolation of Ha-LSL and Ha-ROXL genes from sunflower

DNA was extracted from young leaves (2–3 cm long) of 20-day-old plants of H. annuus (inbred line MISSING FLOWERS) with the Nucleon DNA extraction and purification protocol for plant tissue, according to the manufacturer’s instructions (GE Healthcare, Bio-Sciences). To isolate the full-length introns/exons region of Ha-LSL and Ha-ROXL, the gene-specific primer combinations (LSIN and LSFIN for Ha-LSL; ROXIN and ROXFIN for Ha-ROXL) were used for PCRs with H. annuus genomic DNA (Table S1). The PCR conditions were the same used to isolate the full length of cDNAs. All the amplified products were purified using the Wizard® SV Gel and PCR Clean-UP System (Promega), ligated into the pGEM-T Vector (Promega) and transformed in Escherichia coli JM109 competent cells (Promega). Plasmid DNA was prepared using Wizard® Plus Minipreps DNA Purification Kit (Promega). Several clones were sequenced on both strands. Sequence data from this article have been deposited in GenBank under the accession numbers LT608344 and LT608345 for Ha-LSL and LT596651 for Ha-ROXL.

Database searches and phylogenetic analysis

Database searches were carried out using the BLAST program at the National Center for Biotechnology Information (NCBI) (Altschul et al. 1997); moreover, PROSITE and PFAM databases were searched to identify conserved domains (Bateman et al. 2002; Falquet et al. 2002). The disorder predictions for the deduced Ha-LSL amino acid sequence were carried out using a meta-predictor PONDR-FIT, which is one of the predictors with the highest prediction accuracy (Xue et al. 2010). The deduced amino acid sequence of Ha-LSL gene was compared to GRAS proteins from different species (Fig. S1). The deduced amino acid sequence of Ha-ROXL gene was compared to bHLH proteins of dicot and monocotyledon species (Fig. S2). Alignments were performed with the BLOSUM matrix using the ClustalW algorithm-based AlignX module from Mega4. Phylogenetic relationships were calculated using Unweighted Pair Group Method with Arithmetic Mean (UPGMA) (Sneath and Sokal 1973), and bootstrap values with 100 replicates according to Felsenstein (1985) were determined using the Mega4 package (Tamura et al. 2007).

Gene expression analysis by real-time RT-PCR (qPCR)

H. annuus total RNA was extracted from roots (R), internodal segments of apical stem (IST), blades of young leaves (2–3 cm long, YL), vegetative shoots (VS), and very young leaves [1–4 mm long proximal to VS, (VYL)] of 25–30-day-old plants; young inflorescence shoots (IS) of 40–45-day-old plants; and young ligulate flowers (1.5–2.5 mm long, YLF), young tubular (disc) flowers (1.0–0.5 mm long, YTF) of 60-day-old plants, and from anthers (ANT) near (3–4 days) to anthesis. Extraction was performed using the TriPure Isolation Reagent, according to the manufacturer’s instructions (Roche). The RNA integrity was controlled by gel electrophoresis and treated with RNase-Free DNase (Life technologies, Thermo Fisher Scientific Inc.) following the manufacturer’s instructions.

Real-time quantitative PCRs were performed using a Real-time Step One (Applied Biosystem, Thermo Fisher Scientific Inc. USA) and gene-specific primers for Ha-LSL and Ha-ROXL. Quantitative PCR was performed using 20 ng of cDNA and Power SYBR Green RNA-to-Ct 1 Step Kit (Applied Biosystem, Thermo Fisher Scientific Inc. USA, cat. num. 4389986), according to the manufacturer’s instructions. The thermal cycling conditions of RT-PCR were as follows: reverse transcription, 48 °C–30′; activation, 95 °C–10′; cycling, 40 cycles 95 °C–15′/60 °C–30′; and melt curve, 95 °C–15″/60 °C–15″/95 °C–15″. Relative quantification of specific mRNA levels was performed using the comparative 2−ΔΔCT method (Livak and Schmittgen 2001). Briefly, the C T values of the amplified regions in all samples were normalized with the C T values of the reference housekeeping gene [(Ha-ACT mRNA, encoding a β-actin (GenBank accession number AF282624.1)] to eliminate the variations caused by sample handling. ANT mRNAs were used as reference sample. For each gene, the normalized C T values in ANT were subtracted from the corresponding C T values in the analyzed organs and also in the same pollen. The derived values (ΔΔC T), 0 for the ANT, were inserted in the formula 2−ΔΔCT (according to Livak and Schmittgen 2001) that returns 1 for the reference sample and the final relative quantification for the examined genes in the analyzed organs. The data were the average of three biological replicates (sampled from plants grown at different moments), with each including three technical replicates. The software Real-time Step One v2.3, provided with the instrument by which we carried out the qPCR, was used. Ha-ACT was used as the reference gene based on preliminary data that revealed consistent expression levels regardless of the tissue type. The Ha-ACT was preferred after preliminary comparison with other putative housekeeping genes [i.e., phosphoglycerate kinase 2 (PGK2) and SAND, GenBank accession numbers HM490307 and GE516373, respectively]. The primers used to amplify Ha-ACT cDNA were ACTQF and ACTQR. Primers used to amplify Ha-LSL cDNA were LSQF and LSQR. Primers used to amplify Ha-ROXL cDNA were ROXQF and ROXQR (Table S1).

Statistical analysis

In each experiment, the values displayed on graphs are means (± SD) from three independent qPCR analyses with three different RNA replicates for each organ or for ANT. All qPCR analyses were performed with three different replicates in each run for all biological samples. Data were treated using analysis of variance (ANOVA test) between groups available at http://www.physics.csbsju.edu/stats/anova.html, and means were compared by Tukey’s test available at http://faculty.vassar.edu/lowry/hsd.html.

Construction of digoxigenin-RNA probes

For probes, a 482-bp Ha-LSL fragment was amplified with gene-specific primers LSSONF and LSSONR (Table S1), while a 470-bp Ha-ROXL fragment was amplified with gene-specific primers ROXSONF and ROXSONR (Table S1). In the PCR amplifications, cDNAs were used as template. The fragments obtained were cloned into pGEM®-T easy vector (Promega). Plasmid DNA was prepared using Wizard® Plus Minipreps DNA Purification Kit (Promega). Selected clones were automatically sequenced on both strands, and then they were used as template to synthesize digoxigenin (DIG)-labeled RNA sense and antisense probes, according to the DIG-RNA Labeling Kit protocol (Roche).

In situ hybridization

From the inbred line MISSING FLOWERS (Fambrini et al. 2003), VS of 25–30-day-old plants, AM of 35–40-day-old plants, IS of 40–45-day-old plants, and axillary inflorescence shoots of 60-day-old plants were excised and fixed in p-formaldehyde. The materials were dehydrated, paraffin embedded, cut into 8 μm sections, and hybridized at 45 °C overnight to DIG-labeled Ha-LSL and Ha-ROXL RNA probes according to Jackson (1991). For immunological detection, the slides were processed according to Fambrini et al. (2006). Accumulation of Ha-LSL and Ha-ROXL transcripts was visualized as a violet/brownish stain.

Results and discussion

Isolation and sequence analysis of Ha-LSL and Ha-ROXL

The LATERAL SUPPRESSOR-LIKE cDNA of sunflower (Ha-LS-LIKE, Ha-LSL) contained a complete open reading frame (ORF) of 1269 bp, flanked by 5′- and a partial 3′-untranslated regions (UTR) of 379 and 76 bp, respectively (Fig. S3). The putative peptide Ha-LSL is 422 amino acid long with a calculated molecular mass of 48.0 KDa and an isoelectric point of 6.39 (Fig. 1a). A search in GenBank database with the BLAST program confirms that Ha-LSL is a member of plant GRAS family. Ha-LSL shared a query coverage of 88–100 % and the highest amino acid identity (63–65 %) with the LS proteins of Nicotiana tabacum (Nt-Ls, GenBank accession number ACH47034), N icotiana sylvestris scarecrow-like protein 18 (GenBank accession number XP_009765072), and Solanum lycopersicum Ls (Sl-Ls, GenBank accession number NP_001234179).

Fig. 1
figure 1

Sequence analysis of a LATERAL SUPPRESSOR-LIKE gene from Helianthus annuus (Ha-LSL). a Predicted amino acid sequence of Ha-LSL. The leucine-rich motifs (LHRI and LHRII) are in orange-bold letters. The LXXLL motif is double underlined. The VHIID motif is in red-italic and bold letters. Within the VHIID motif, the highly conserved P-H-D-Q-L residues are in black-bold letters. The PFYRE motif is in blue-italic and bold letters and the specific PFYRE amino acid residues are in black-bold letters. The SAW motif is in green-italic and bold letters and the conserved amino acid residue pairs R–E, W–G, and W–W are in black-bold letters. The characteristic motifs of the Ha-GRASL sequence are designed according to Pysh et al. (1999) and Bolle (2004). b The deduced amino acid sequences of the H. annuus Ha-LSL gene were compared to annotated proteins of the GRAS transcription factor family of several species (Fig. S1). The consensus tree was inferred by Unweighted Pair Group Method with Arithmetic Mean (UPGMA) using Mega4 software (Tamura et al. 2007). The optimal tree with the sum of branch length = 5.65893720 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (100 replicates) is shown next to the branches (Felsenstein 1985). The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method (Zuckerkandl and Pauling 1965) and are in the units of the number of amino acid substitutions per site. All positions containing gaps and missing data were eliminated from the dataset (complete deletion option). There were a total of 258 positions in the final dataset. The arrow indicates Ha-LSL

Full size image

Accordingly, to most GRAS genes in Arabidopsis, rice, tomato, soybean, Populus trichocarpa, Prunus mume, Selaginella moellendorffii, and Physcomitrella patens (Schumacher et al. 1999; Liu and Widmer 2014; Lu et al. 2014; Song et al. 2014, Wu et al. 2014; Fambrini et al. 2015), the comparison of sequences obtained from amplified cDNA and genomic DNA indicated that the Ha-LSL gene is intronless. The reason why plant GRAS gene family is enriched with intronless genes is likely to be its prokaryotic origin occurred by horizontal gene transfer followed by extensive duplications in flowering plants (Zhang et al. 2012; Liu and Widmer 2014). In fact, a sequence, structure, and evolutionary analysis showed that the GRAS family emerged in bacteria and belongs to the Rossmann-fold, AdoMET (SAM)-dependent methyltransferase superfamily (Zhang et al. 2012). However, in plants, differently from bacterial, most GRAS proteins are likely to lack methyltransferase activity, though they are still able to bind similar substrates.

The GRAS gene products are characterized by a highly conserved C-terminal region and a variable N-terminal region (Pysh et al. 1999; Sun et al. 2011, 2012; Fambrini et al. 2015). The intrinsically disordered nature of the N-domains improves the functions of GRAS proteins and allows them to perform key roles in multiple developmental and environmental signals (Sun et al. 2012, 2013). The distribution of the intrinsic disorder propensities along the Ha-LSL amino acid sequence, generated by disorder predictor PONDR-FIT, is shown in Fig. S4. According to the characteristics of other GRAS proteins (Sun et al. 2011), the Ha-LSL N-domain is mainly located in the disordered zone, while the C-domain is located predominantly in the structured zone with few peaks representing local short and disordered/flexible fragments within the long ordered region (Fig. S4). Analogously to other GRAS proteins, homopolymeric stretches of certain amino acid residues such as S, T, P, Q, and H are present in the N-domains of Ha-LSL (Fig. 1a). In addition, the Ha-LSL protein contains all the sequence motifs described for members of the GRAS family (Pysh et al. 1999; Bolle 2004; Lee et al. 2008; Sun et al. 2011; Fambrini et al. 2015), except for the DELLA motif that was found in the N-domain of GAI, RGA, and a few other members of this gene family. In fact, the C-terminal region of the Ha-LSL includes two leucine-rich areas flanking a VHIID motif (name derived from the most prominent amino acid residues), the PFYRE, and SAW motifs (Fig. 1a). These conserved motifs can directly affect the function of the GRAS proteins; in fact, mutations in the SAW and PFYRE motifs of SLR1 and RGA proteins result in huge phenotypic variations in Arabidopsis (Silverstone et al. 1998; Itoh et al. 2002). In Ha-LSL, two leucine-rich areas (LHRI and LHRII) flank the VHIID motif (Fig. 1a). The presence of LHRI and II suggests that the Ha-LSL gene product may function as multimers. An LXXLL sequence occurs in Ha-LSL at the beginning of the first leucine-rich domain as well in other GRAS proteins. It has been suggested that LHRI, II, and LXLL motifs mediate protein–protein interaction (Pysh et al. 1999; Lee et al. 2008). The LXXLL motif has been shown to mediate the binding of transcriptional co-activators to nuclear receptors (Heery et al. 1997; Bolle 2004). Thus, it is possible that the Ha-LSL protein interacts with factors involved in the regulation of transcription to modulate the level of transcriptional activation. In tomato, Ls and GOBLET act in hierarchical order to regulate ectopic meristem formation at the base of leaflets (Rossmann et al. 2015). Within the VHII motif of Ha-LSL, the P-H-D-Q-L residues are revealed (Fig. 1a), although only the H and the D are absolutely conserved in all GRAS sequences (Bolle 2004). Several additional amino acid residues are invariant in most members of the GRAS protein family. These include the SAW motif that in Ha-LSL contains three pairs of conserved residues: R (x)4 E, W (x)7 G, and W (x)10 W (Fig. 1a), analogously to most GRAS proteins (Bolle 2004). Studies of DELLA proteins reveal that the N-domain is a GA signal perception domain while the GRAS domain is a repressor-domain regulating the repressive activity (Hirano et al. 2010).

We conducted a phylogenetic analysis on the Ha-LSL amino acid sequence using a data set, which included annotated proteins of the GRAS transcriptional regulator family of several species (Fig. S1). The analysis showed that Ha-LSL forms a subclade with LS-LIKE proteins of the dicots Daucus carota (Dc-LSLP), tomato (SlLS), tobacco (Nt-LS), and Arabidopsis (AtLS), within a cluster including a protein of a monocot, MOC1 of rice (Fig. 1b). Therefore, BLAST and phylogenetic analyses suggested that the GRAS Ha-LSL amino acid sequence belongs to LS TF subfamily. However, LS cluster and SCARECROW-LIKE4/7 proteins (i.e., Ha-GRASL of sunflower, RcGRAS of castor bean, and PtGRAS17 of P. trichocarpa) belong to the same major clade (Fig. 1b), corroborating recent results by Fambrini et al. (2015).

The REGULATOR OF AXILLARY MERISTEM FORMATION-LIKE cDNA of sunflower (Ha-ROX-LIKE, Ha-ROXL) contained a complete ORF of 576 bp, flanked by 5′- and 3′-UTR of 165 and 210 bp, respectively (Fig. S5). The putative peptide Ha-ROXL is 191 amino acid long with a calculated molecular mass of 22.4 KDa and an isoelectric point of 6.85 (Fig. 2a). A search in GenBank database with the BLAST program confirms that Ha-ROXL is a member of a plant bHLH TF family. The second largest class of plant TFs displayed a high residue conservation within the bHLH domain but large amino acid differences in both N- and C-terminal regions. Therefore, Ha-ROXL shared a very ample variation in the query coverage (33–85 %) with the highest amino acid identity (75 %) with the LAX PANICLE-LIKE protein of Fragaria vesca (GenBank accession number XP_011468442), 72 % with the bHLH140 protein of T. cacao (GenBank accession number XP_007026627.1), and 42.8 % with the A. thaliana ROX1 protein (GenBank accession number NM_120209.1).

Fig. 2
figure 2

Sequence analysis of a REGULATOR OF AXILLARY MERISTEM FORMATION-LIKE gene from Helianthus annuus (Ha-ROXL). a Predicted amino acid sequence of the bHLH transcription factor Ha-ROXL. The bHLH domain is in red-bold letters. The QAR residues are in black bold letters. The leucine (L) residue at position 23 is double underlined. b The deduced amino acid sequences of the H. annuus Ha-ROXL gene were compared to annotated bHLH transcription factors of several species (Fig. S2). The consensus tree was inferred by UPGMA using Mega4 software (Tamura et al. 2007). The analysis was performed as described for Fig. 1b. The arrow indicates Ha-ROXL

Full size image

The comparison of sequences obtained from amplified cDNA and genomic DNA indicated that the Ha-ROXL gene is intronless in accordance with most bHLH genes as well as BA1 of mays, LAX1 of rice, and ROX1 of Arabidopsis (Gallavotti et al. 2004; Oikawa and Kyozuga 2009; Woods et al. 2011; Yang et al. 2012). Analogously to BA1, LAX1, and ROX1, the Ha-ROXL protein is characterized by the presence of the QAR motif, which contains the amino acids glutamine (Q), alanine (A), and arginine (R) at positions 5, 9, and 13, respectively, within the bHLH domain (Heim et al. 2003). The highly conserved hydrophobic residues in helix 1 and 2 are believed to be necessary for dimerization (Heim et al. 2003). For example, the leucine (L) at position 23 (Fig. 2a), also present in bHLH proteins of Arabidopsis, is supposed fundamental for dimerization.

A phylogenetic analysis was performed with bHLH TFs of several species containing the QAR residues (Fig. S2). The analysis grouped Ha-ROXL and AtROX1, within a major clade containing LAX1 of both dicot and monocot as well as BA1 of Z. mays (ZmBA1) and LAX of Oryza sativa (OsLAX) and separate by other QAR bHLH proteins like HEC1, HEC2, and HEC3 (Fig. 2b). This evolutionary conservation suggests similar functions for Ha-ROXL, AtROX1, OsLAX, and ZmBA1 in AM initiation.

Expression profiles of Ha-LSL and Ha-ROXL

The expression profiles of Ha-LSL and Ha-ROXL in different sunflower organs were examined by qRT-PCR (Figs. 3 and 5) and in situ hybridization (Fig. 4). Transcripts of Ha-LSL were predominantly accumulated in young inflorescence shoots and lower mRNA levels were in roots and vegetative shoots, while the lowest amount of Ha-LSL transcripts has been observed in internodal stem and young leaves (Fig. 3a). Analogous results were obtained for Ls of tomato (Schumacher et al. 1999), LAS of Arabidopsis (Greb et al. 2003), and LS of cucumber (Yuan et al. 2010).

Fig. 3
figure 3

Steady-state levels of Ha-LSL (a) and Ha-ROXL (b) mRNA in different organs of sunflower (Helianthus annuus L). Relative transcript values were calculated using ANT mRNA as reference sample and normalized to those of β-actin gene. The graph shows the mean (± SD) of three biological replicates (n = 3). Same letters above the bars indicate no significant differences from each other (ANOVA P ≤ 0.05) according to Tukey’s test. R, roots; IST, apical internodal stem; YL, young leaves; VS, vegetative shoots; IS, young inflorescence shoots

Full size image
Fig. 4
figure 4

Patterns of transcript localization of Ha-LSL (a–e) and Ha-ROXL (h–m) in sunflower (Helianthus annuus L.) meristems. a–d Longitudinal sections of a shoot apical meristem (SAM, a), an axillary meristem (AM, b), an inflorescence shoot (IS, c), and an axillary inflorescence shoot (AIS, d) hybridized with the Ha-LSL antisense probe. e A transverse section of an IS hybridized with the Ha-LSL antisense probe. f, g Longitudinal sections of a SAM (f) and an IS (g) hybridized with the Ha-LSL sense probe. In a, Ha-LSL RNA accumulation was detected in a domain within the boundary zones separating the SAM and the leaf (black arrowhead), in an inner domain (white arrowhead), as well as in very young leaves (arrows). In b, Ha-LSL RNA accumulation was detected in a boundary domain (black arrowhead), in the inner region of SAM (white arrowhead), and in leaf primordia (arrows). In c, Ha-LSL transcripts were detected in tubular (disc) flower primordia (arrowheads) and along the inflorescence meristem (arrow). In d, Ha-LSL transcripts were detected in inflorescence meristem (arrowhead) and leaf primordium (arrow). In e, Ha-LSL transcripts were abundantly accumulated in corolla of disc flower primordia (we used arrowheads to mark disc flowers to show the Ha-LSL signal). f, g Longitudinal sections of SAM and IS hybridized with the Ha-LSL sense probe. No Ha-LSL transcripts were detected. h, i Longitudinal sections of a SAM (h) and AM (i) hybridized with the Ha-ROXL antisense probe. The Ha-ROXL mRNA was detected in small domains within the boundary zones separating the SAM and the leaf primordia (arrowheads). j A transverse section of SAM. Ha-ROXL RNA accumulation is detected in small domains in the adaxial zones of leaf primordia (arrowhead). k A longitudinal section of AIS. Ha-ROXL transcripts were accumulated in a small domain within the boundary zones separating the inflorescence meristem and a leaf primordium (we used single arrowheads to mark the Ha-ROXL transcripts to simplify the figure). l A longitudinal section of an IS. Ha-ROXL transcripts (arrowheads) were accumulated in disc flower primordia before the separation between floral bract and floral primordia (Fambrini et al. 2003). m A transverse section of an IS. Ha-ROXL mRNAs were accumulated as in l (we used a single arrowheads to mark the Ha-ROXL transcripts to simplify the figure). n, o Longitudinal sections of SAM and IS hybridized with the Ha-ROXL sense probe. No Ha-ROXL transcripts were detected. Scale bars represent 140 μm in a, 215 μm in b, 196 μm in c, 193 μm in d, 117 μm in e, 161 μm in f, 363 μm in g, 102 μm in h, 78 μm in i, 209 μm in j, 139 μm in k, 172 μm in l, 216 μm in m, 177 μm in n, and 241 μm in o

Full size image

In dicot, bHLH genes, supposed ortholog to BA1 of maize (Gallavotti et al. 2004) and LAX1 (Komatsu et al. 2003) of rice, have been characterized in few species (Woods et al. 2011; Yang et al. 2012). Transcription analyses by semiquantitative RT-PCR indicated an accumulation of mRNA in inflorescences of Brassica oleracea subsp. Italica, Medicago truncatula, Carica papaya, and Sorghum bicolor, while in leaves, transcript accumulation was observed only in some species (Woods et al. 2011). In sunflower, the highest expression levels of Ha-ROX were in vegetative shoots, in young inflorescence shoots, and in roots while a significant reduction of transcripts was observed in young leaves and internodal stem (Fig. 3b). Therefore, when the two transcription patterns were compared, a clear difference between Ha-LSL and Ha-ROX was recognized for young inflorescence shoots (compare Fig. 3a with Fig. 3b).

We conducted ISH experiments to identify the cellular localization of Ha-LSL and Ha-ROXL transcripts during both vegetative and reproductive development of sunflower plants. Our results demonstrated that these genes are expressed in all organs tested: shoot apical meristems (SAMs), axillary buds, primary inflorescence meristem, and secondary capitula, at very early stages of development (Fig. 4); nevertheless, the cellular domains of transcripts accumulation for Ha-LSL and Ha-ROXL displayed a different pattern.

In the model species A. thaliana, LAS is expressed especially in the boundary zones (Greb et al. 2003), but with respect to Ha-LSL, we observed a pattern of transcript distribution not restricted to these cellular domains. In fact, we detected an evident signal in boundary districts of primary (Fig. 4a; black arrowhead) and secondary shoot apex (Fig. 4b; black arrowhead) but also the presence of Ha-LSL mRNAs in cell layers of other inner regions of SAM and AMs (Fig. 4a, b; white arrowheads). In addition, the presence of Ha-LSL transcripts was observed in early stages of leaf development (Fig. 4a, b) and in flower primordia of immature inflorescence (Fig. 4c). Analogously, different cellular domains were characterized by a positive signal in secondary capitula before the initiation of floret primordia (Fig. 4d). In transverse sections of tubular (disc) flower primordia, we observed Ha-LSL transcripts mainly within the corolla tube (Fig. 4e).

An expanded cellular domain beyond axil zones has been identified in cucumber for the LAS-orthologous CLS gene (Yuan et al. 2010). Furthermore, in tomato flowers, the high expression of the Ls gene has been detected (Schumacher et al. 1999) and a specific role for this gene to control the development of the second whorl organs has been assumed (Szymkowiak and Sussex 1993). TaMOC1, the wheat ortholog of rice MOC1, is mainly expressed in roots, tiller buds, and leaf blades at the seedling stage, while at the late booting stage, the expression is high in roots, internodes, nodes, and leaf sheaths (Zhang et al. 2015). These results suggest that the extension of cellular domains characterized by LATERAL SUPPRESSOR transcript accumulation could be affected by the species analyzed.

To support the transcription pattern detected for Ha-LAS by ISH, a qRT-PCR analysis on very young leaves (VYL) proximal to SAM as well as young ligulate (YLF) and tubular (disc) flowers (YTF) was performed (Fig. 5, see also “Material and methods”). The findings corroborate the ISH results because a high Ha-LSL transcription was detected in VYL as well as in YTF (compare Fig. 3a with Fig. 5) suggesting a role of Ha-LSL in the early stage of lateral primordia development.

Fig. 5
figure 5

Steady-state levels of Ha-LSL in different organs of sunflower (Helianthus annuus L). Relative transcript values were calculated using ANT mRNA as reference sample and normalized to those of β-actin gene. The graph shows the mean (± SD) of three biological replicates (n = 3). Same letters above the bars indicate no significant differences from each other (ANOVA P ≤ 0.05) according to Tukey’s test. VYL, very young leaves; YLF, young ligulate flowers; YTF, young tubular (disc) flowers; IS, young inflorescence shoots

Full size image

Ha-ROXL expression in sampled longitudinal sections followed a more specific band-shaped pattern (Fig. 4h–k). Indeed, ISH results suggested that Ha-ROXL transcription was strongly restricted to a small domain within the boundary zone separating the SAM and the leaf primordia during vegetative growth (Fig. 4h–j); analogously, in reproductive phase, cellular domains with Ha-ROXL transcripts were observed at the axils between inflorescence bracts and inflorescence meristem (Fig. 4k). However, it is notable that at later stage of reproductive development, Ha-ROXL transcripts were localized in specific arcs of cells where florets primordia will initiate and in any case, before a clear separation between flower primordium and related bract (Fig. 4l). Therefore, extent of Ha-ROXL expression in sunflower capitula is probably very transient and soon undetectable once distinct floral primordia differentiated. These results are in accordance with the hypothesis that BA1/LAX1 expression in eudicots is required for the initiation and early maintenance of AMs but not during later stages of outgrowth (Woods et al. 2011). In situ results in transverse sections showed that Ha-ROXL transcripts accumulated within a circular domain between the shoot apical meristem and lateral young leaf primordia, in specific spots at boundary regions (Fig. 4j). A similar pattern has been also identified in A. thaliana (Yang et al. 2012). Moreover, we observed a positive signal in the stamen primordia of very young tubular florets sectioned with transverse orientation (Fig. 4m). Similar results have been obtained in B. oleracea using BA1/LAX1 co-orthologs (Woods et al. 2011).

Conclusion

Ha-ROXL is mainly active in boundary zones that separate leaf primordia from the SAM as well as in a restricted area of disc flower primordia. It is hypothesized that Ha-ROXL, likely with other boundary genes, acts to specify boundary zone identity. In particular, in sunflower inflorescences, the peripheral extremity of the flower primordia arise first as subtle bumps; then, each primordium assumes a parallelogram form (Fambrini et al. 2003). A bisecting crease divides the primordia in two distinct bumps that adopt different fates. The peripheral (abaxial) part of the primordium becomes the bilaterally symmetrical bract and the adaxial part becomes the flower. The activity of Ha-ROXL appears before the specification of the above distinct developmental features. Our results indicate that the activity of Ha-LAS is not restricted to the boundary zone but likely regulates AM initiation as well as early development outgrowth of lateral primordia. Further investigation of the role of Ha-LAS and molecular cloning of other lateral primordia regulators should allow additional insight into how developmental and environmental as well as hormonal signals are integrated to control AM and flower development.