Introduction

Embryogenesis in higher plants is characterized by a set of unique events which shape the basic pattern of the plant. An important phase of embryogenesis is the development of the zygote into an embryo, composed of shoot and root poles, an embryonic axis, and cotyledons (reviewed by West and Harada 1993). It is during this transition that the shoot apical meristem (SAM) forms. Through a tightly controlled balance between cell proliferation in the central region and differentiation in the peripheral domains, the SAM is responsible for the reiterative generation of above-ground lateral organs during post-embryonic development (reviewed by Bowman and Eshed 2000). Formation and maintenance of the SAM are therefore paramount for normal growth and studies have revealed complex genetic networks governing both processes. In Arabidopsis, SAM homeostasis is maintained by a regulatory feedback loop involving the transcription factor WUSCHEL (WUS) and the CLAVATA (CLV) complex (Mayer et al. 1998; Fletcher et al. 1999; Brand et al. 2001), as well as essential pathways regulated by SHOOTMERISTEMLESS (STM) which also affect hormone signalling (Hay and Tsiantis 2009). SHOOTMERISTEMLESS belongs to the class I of KNOX homeodomain transcription factors (TFs) which in Arabidopsis also include KNAT1, KNAT2, and KNAT6 that share overlapping functions in the control of SAM function (Scofield and Murray 2006). Members of this class share distinct similarities in protein structure and conserved domains; the HOMEODOMAIN is required for DNA binding, the ELK domain encodes a nuclear localization signal, while the KNOX1 and KNOX2 domains are involved in transcriptional repression of target genes and in dimer formation and trans-activation, respectively (Scofield and Murray 2006). Through the suppression of differentiation and the maintenance of indeterminate cell fate STM controls homeostasis, the rate of cell divisions and differentiation, in the SAM. While induced STM over-expression increases cell proliferation and results in the generation of ectopic meristems from vegetative tissues (Brand et al. 2001; Gallois et al. 2002; Lenhard et al. 2002), a down-regulation of STM perturbs the functionality of the SAM. Strong stm loss-of function alleles produce terminated SAMs while weak stm alleles fail to maintain a stable SAM during post-embryonic growth as a result of the incorporation of meristematic cells into organ primordia (Barton and Poethig 1993; Clark et al. 1996; Endrizzi et al. 1996). Similar perturbations were also observed if the expression of STM homologs were altered in other species (Hay and Tsiantis 2006; Vollbrecht et al. 2000).

Formation of the SAM is a critical event during the development of in vitro embryos produced via somatic or gametophytic embryogenesis. Both systems are largely used due to the potential to produce a large number of embryos which are not embedded in the maternal tissue and therefore easily accessible for physiological and molecular studies. While somatic embryogenesis exploits the ability of somatic cells to produce embryos, gametophytic embryogenesis uses male or female gametophytes as the initial explants. The latter system is more desirable to capture genetic variations through the recovery of diploid homozygous embryos (Yao et al. 1997). The most commonly used gametophytic-embryogenic system is androgenesis, which involves the production of microspore-derived embryos (MDEs) from immature pollen grains (microspores). This system is often integrated in breeding programmes and has been applied to many species, including Brassica napus, which offers tremendous advantages to investigate plant embryogenesis. Besides the absence of a hormonal requirement for the initiation of the embryogenic pathway, the morphogenic events occurring during Brassica MDE development are very similar to those observed in zygotic embryos (Yeung et al. 1996). Furthermore, Brassica shares a lot of genetic similarities with Arabidopsis, thus facilitating molecular and genetic studies. Therefore, the B. napus MDE culture system has been largely utilized to examine not only the events associated with the initial gametophytic-embryogenic transition (Joosen et al. 2007; Malik et al. 2007), but also the developmental phases leading to the formation of fully mature MDEs (Yeung 2002; Stasolla et al. 2008; Belmonte et al. 2010; Elhiti et al. 2010). While the ontogeny of the SAM in B. napus MDEs is similar to that observed in zygotic embryos (Yeung et al. 1996), abnormalities in SAM architecture are recurrent during the middle-late phases of MDE development precluding successful post-embryonic growth (Belmonte et al. 2006, 2010). Sub-optimal functionality of the SAM has also been observed in other systems (Kong and Yeung 1992). Yeung (2002) proposed that failure to establish a functional embryonic SAM is one of the major factors limiting more extensive applications of in vitro propagation practices. It is therefore apparent that enhancing proper SAM formation in vitro and understanding molecular and physiological differences between functional and defective SAMs are crucial for expanding our knowledge on meristems, and improving culture conditions to develop efficient in vitro propagation systems.

In previous work we have isolated B. napus (Bn) STM and shown that when ectopically expressed in Arabidopsis this gene enhances the number of somatic embryos through changes in hormone signalling (Elhiti et al. 2010). A similar effect was also reported in Brassica where the over-expression of BnSTM increases the number of MDEs, while its down-regulation represses embryogenesis (Elhiti et al. 2010). However, these studies provide no information on whether, besides increasing embryo yield, BnSTM also affects the quality of the embryos, i.e., their ability to convert and regenerate viable plants, and the structure and functionality of the meristems. The objective of the present work is to evaluate whether manipulations in BnSTM levels influence (1) embryo quality and SAM architecture and functionality; and (2) gene expression patterns in laser microdissected SAMs. Our data reveal a novel function of BnSTM in enhancing SAM functionality through transcriptional alterations of specific regulatory pathways ranging from hormone signalling to cell wall metabolism. Findings from this work suggest that BnSTM can be used as a potential target to improve meristem formation and plant regeneration in culture.

Materials and methods

Plant material

Three Brassica napus cv. Topas (DH4079) lines were utilized for this experiment. A wild-type (WT) line, a line down-regulating BnSTM via antisense-mediated transformation (A5), and a line ectopically expressing BnSTM (S101). These lines were produced and characterized in previous studies (Elhiti et al. 2010). Compared to WT, a fourfold induction of BnSTM was observed in vegetative and reproductive tissues of the S101 line, whereas the expression of BnSTM was halved in the A5 line (see Supplementary Fig. 2 in Elhiti et al. 2010). Production of MDEs in the three lines was carried out as described by Belmonte et al. (2006). Briefly, plants with young buds were grown at 12/7 °C day/night temperature until bud collection. Buds (2–3 mm in length) were then collected and cultured in NLN medium (Lichter 1982) with 13 % sucrose (pH 5.8). Embryo development was triggered after an initial heat shock treatment at 32 °C for 72 h. Meristem quality was estimated by measuring the frequency of embryo conversion, i.e., the ability to regenerate viable shoot and root systems at germination, exactly as described by Belmonte et al. (2006). Treatments of MDEs with dl-buthionine sulfoximine (BSO) or 2,3,5-triiodobenzoic acid (TIBA) were carried out exactly as described in Stasolla et al. (2008) and Ramesar-Fortner and Yeung (2006), respectively. RNA in situ hybridization studies in the treated MDEs were performed as previously reported (Elhiti et al. 2012).

Arabidopsis plants with inducible STM function (35S::STM-GR construct) were kindly provided by Prof. Sablowski (John Innes Centre, Norwich, UK), and used to identify putative target genes directly activated by STM. Plant treatments and applications of dexamethasone (DEX) which induces translocation of the STM–GR fusion protein in the nucleus, and cycloheximide (CHX) which blocks protein translation, were performed as described in Spinelli et al. (2011). Transcript profiles of Arabidopsis orthologs of 12 Brassica genes which were differentially expressed in the microarray experiment, were measured in 14-day-old 35S::STM-GR plants. Expression studies were performed in three biological replicates. Primers used are listed in Supplemental Table S1.

Microscopy

For histological examination under light microscopy, developing MDEs were fixed in 2.5 % glutaraldehyde and 1.6 % paraformaldehyde buffered with 0.05 M phosphate buffer, pH 6.9, dehydrated with methyl cellosolve followed by two changes in absolute ethanol, and then infiltrated and embedded in Historesin (Leica, Concord, Ontario, Canada). Serial sections (3 μm) were stained with periodic acid-Schiff (PAS) reagent to detect carbohydrates and counterstained with toluidine blue (TBO) according to the methods of Yeung (1999). For transmission electron microscopy MDEs (21 days in culture) were fixed in 3 % glutaraldehyde in 25 mM cacodylate buffer supplemented with 5 mM CaCl2 (pH 7.0) for 12 h at 4 °C. Tissues were then post-fixed in a solution containing 2 % OsO4 and 0.8 % KFe(CN)6 overnight at 4 °C, rinsed six times with distilled water, stained overnight with 0.5 % uranyl acetate, and finally dehydrated using ethanol. Ethanol was replaced through a graded propylene oxide series and gradually infiltrated with Spurr resin (Spurr 1969). Ultrathin sections (50 nm) were examined using a Hitachi H 7000 transmission electron microscope at 75 V equipped with a digital camera. Cell wall thickness was measured using 20 longitudinal sections per line.

Laser capture microdissection of SAMs

Microspore-derived embryos (MDEs) from the three lines (WT, S101, and A5) were harvested at day 21 in culture, fixed in ice-cold 3:1 ethanol (95 %):acetic acid (v/v) and vacuum-infiltrated for 20 min before being fixed at 4 °C overnight. Embryos were rinsed three times with 70 % ethanol and dehydrated in a graded ethanol series followed by gradual infiltration with xylene (1:3, 1:1, 3:1 xylene:ethanol, followed by two washes in 100 % xylene). Paraffin chips were added to the vials and allowed to gradually infiltrate for 24 h at room temperature on a rotator. Samples were then allowed to incubate at 42 °C for 1 h before being placed in a histology oven set at 60 °C. Embryos were sectioned at a thickness of 7 μm using a Leica RM2125RT rotary microtome (Leica Microsystems).

The first three layers of apical cells (two layers of the tunica and one layer of the corpus) enclosed by the emerging cotyledons and displaying meristematic features in median sections, i.e. small size and densely stained cytoplasm (Yeung et al. 1996), were dissected using a Leica LMD6000 Laser Microdissection System (Leica Microsystems). Two biological replicates were captured for each line and each replicate consisted of more than 70 median sections of the SAM. All cells were captured within 30 days of fixation to avoid RNA degradation.

RNA isolation and amplification

Sections collected from the SAM of MDEs were harvested in RNA extraction buffer (RNAqueous Micro, Ambion). RNA was then extracted according to the manufacturer’s instructions with the addition of an on-column DNAase treatment (1:4 dilution of DNAse I in RDD buffer; Qiagen). Purified RNA was quantified using the Ribogreen method with a ND-3330 fluorospectrometer (Nano-drop). The total RNA yield per biological replicate was between 8.5 and 56.9 ng. Total RNA was then analysed for integrity using the RNA 6000 Pico Chip on an Agilent BioAnalyzer (Agilent Technologies, Waldbronn, Germany). Two rounds of RNA amplifications were performed. The first round using a RNA ampULSe:Amplification and Laelling kit (Kreatech Biotechnology, Amsterdam, The Netherlands) following the manufacturer’s instructions, and the second round with the MessageAmp™ aRNA Amplification Kit (Ambion, Cat. No. 1750). Cy5 dye molecules (GE Healthcare) were coupled to the amplified RNA, and the dye-labelled RNA was fragmented before hybridization.

Array hybridization, scanning and data processing

The high-density CombiMatrix 90K Brassica oligonucleotide array (v1.1) was produced by Plant Biotechnology Institute (Saskatoon, Canada). A total of 90,564 probes (35–40 mer) were in situ synthesized using the CustomArray Synthesizer (CombiMatrix, Irvine, CA, USA). Array hybridization, stripping and re-hybridization were performed following the CustomArray™ 90K Microarray: Hybridization and Imaging Protocol (PTL020) and Stripping and Preparation of CombiMatrix 90K Microarrays for Re-hybridization protocols, as recommended by CombiMatrix (PTL025) (http://www.combimatrix.com). The arrays were first rehydrated at 65 °C for 10 min then prehybridized at 45 °C for 30 min before use. Fragmented Cy5-aRNAs (5 μg) were mixed with hybridization buffer (6×SSPE, 0.05 % Tween-20, 20 mM EDTA, 25 % deionized formamide, 0.1 mg/mL sheared salmon sperm DNA and 0.04 % SDS) and hybridized to the Combimatrix Brassica 90K arrays at 45 °C overnight in the dark. After hybridization, stringent washes were performed according to the manufacturer’s instructions, and the slides were immediately coated with an imaging solution (CombiMatrix) prior to scanning with a GenePix 4000B scanner (Axon Instruments, Foster City, CA, USA) using 400 PMT, 5 μm resolution and 100 % laser power. After scanning, the hybridized dye-labelled aRNA targets were stripped from the arrays using the CustomArray Stripping kit (CombiMatrix) and re-hybridized.

Image analysis was performed using GenePix Pro version 6.0 software (Axon Instruments). After automatic alignment, feature indicators were manually moved, resized and fine-tuned. The local background-corrected spot fluorescence intensities for the Cy5 channel were saved as GPR files.

GPR files were loaded in GeneSpring GX 11.0 (Agilent Technologies Inc. and Strand Life Sciences Pvt. Ltd) with percentile shift normalization and baseline to median of all samples. Signal intensity cut-off was set to 200. One-way ANOVA analysis was carried out with the asymptotic P value computation method. Multiple testing correction method was performed using the Benjamini–Hochberg FDR (Benjamini and Hochberg 1995). Only gene probes having an adjusted P value (FDR) <0.05 and an absolute difference in expression fold-change higher than 2 were selected. Clustering by K-means based on Eucledean distance was utilized to group gene probes in categories according to their expression profiles.

Enrichment of gene ontology terms and hierarchical clustering

Gene ontology (GO) terms were assigned based on the annotation corresponding to the Arabidopsis gene that best aligns (BLASTX) to the B. napus cDNA from which each microarray probe was designed. Terms considered statistically enriched had P < 0.001 when using WeGO (Ye et al. 2006). Data was log10 transformed and a heat map was generated using TMeV (Saeed et al. 2006).

Expression studies and validation of microarray experiments

Quantitative (q)RT-PCR was used to validate selected genes of microarray data (Supplemental Fig. S1; Supplemental Table S1) and to measure gene expression levels (Stasolla et al. 2008). The relative level of gene expression was analysed with the 2−∆∆CT method described by Livak and Schmittgen (2001). Primers are listed in Supplemental Table S1.

Results

Effects of BnSTM expression on MDEs morphology

Well organized apical poles of WT MDEs collected 21 days in culture consisted of dome-shaped SAMs composed of densely cytoplasmic cells (Fig. 1a). Over-expression of BnSTM in MDEs of line S101 increased the size of the SAM (Fig. 1b) and in some instances resulted in the formation of two SAMs originating from the apical domain (Fig. 1c). Meristem formation was precluded in the BnSTM down-regulating MDEs of line A5 (Fig. 1d). Many of the BnSTM down-regulating embryos showed structural abnormalities such as hypocotyl expansion (Fig. 1e), irregular outgrowths of the epidermal layer (Fig. 1f, g), and production of secondary SAMs from the base of the cotyledons (Fig. 1h). Termination of the SAM (Fig. 1i) and formation of adventitious shoots (Fig. 1j) were sometimes observed in germinating MDEs down-regulating BnSTM. These two phenotypes mimicked closely those observed in Arabidopsis stm seedlings (Barton and Poethig 1993).

Fig. 1
figure 1

Effects of altered BnSTM expression on meristem development in Brassica MDEs. a Shoot apical meristem (SAM) of WT MDEs 21 days in culture. The SAM is dome-shaped and composed of several layers of densely cytoplasmic cells (arrow). Inset external morphology the embryo. b Enlarged SAM of MDEs over-expressing BnSTM (line S101). c Formation of two SAMs (arrows) in the apical pole of MDEs produced by the S101 line. d Poorly developed SAM of MDEs down-regulating BnSTM (line A5). The cotyledons are almost juxtaposed at their base. e Expansion of the hypocotyl observed in MDEs produced by the A5 line. f Irregular outgrowths along the epidermal layer of the hypocotyl of MDEs down-regulating BnSTM. g Heavy accumulation of storage products (mainly starch) in the clusters of cells emerging from the epidermal layer of embryos with reduced BnSTM expression. h In some instances adventitious SAMs (arrow) originated from the base of the cotyledons in developing A5 embryos. Inset: external morphology of an embryo with an adventitious SAM. i Termination of the SAM (arrow) with the formation of a single leaf in germinating MDEs down-regulating BnSTM (line A5). Many of these seedlings were not able to grow further. j Formation of adventitious shoots which sometimes originated from the terminated SAM of germinating A5 MDE. Scale bars 4 μm (a–d, f, g), 8 μm (e, h), 1 cm (i, j), 1 mm (insets a and h)

Full size image

The frequency of MDEs able to convert and regenerate viable plants at germination, used as an indirect estimation of meristem quality, was affected by the expression levels of BnSTM. A significant reduction in conversion occurred in the BnSTM down-regulating embryos produced by the A5 line. Ectopic expression of BnSTM encouraged post-embryonic growth (Fig. 2).

Fig. 2
figure 2

Effects of altered BnSTM expression on the ability of the MDEs to convert and regenerate viable plants at germination. Embryos of the WT line and lines down-regulating BnSTM (A5) or up-regulating BnSTM (S101) were harvested after 21 days in culture and transferred onto germination medium. Values ± SE are means of three independent biological replicates. Each biological replicate included 120 MDEs. The asterisk indicates values that are statistically different (P < 0.05) from WT

Full size image

Microscopic studies of cells of the SAMs in wild type and transformed plants

At day 21 in culture, the apical and sub-apical layers of the SAMs in WT MDEs were composed of densely cytoplasmic cells (Fig. 3a) separated by thin tangential cell walls (Fig. 3b). Small starch-containing amyloplasts were observed throughout the WT SAMs (Fig. 3c). These characteristics were also retained by the SAMs of S101 embryos over-expressing BnSTM (Fig. 3d, f). Apical and sub-apical cells of BnSTM-down-regulating SAMs contained many vacuoles and starch granules (Fig. 3g), and were separated by thicker cell walls (Fig. 3h). These cells also contained larger granules of starch (Fig. 3i) and many differentiating chloroplasts (Fig. 3j). Tangential cell wall width was measured in the three lines (Fig. 3k). Cell wall width increased more than 33 % in embryos down-regulating BnSTM.

Fig. 3
figure 3

Ultrastructural features of the SAMs of MDEs cultured for 21 days. a Apical and sub-apical cells of a wild-type SAM characterized by a dense cytoplasm and few starch grains. b Cell wall separating the apical and sub-apical cells of a WT SAM. c Starch granules observed in the apical pole of WT MDEs. d–f The morphology of meristematic cells (d), cell wall (e), and granules of starch (f) in SAMs of MDEs over-expressing BnSTM (line S101) was similar to those observed in WT embryos. g Prominent accumulation of starch granules occurred in the SAMs of MDEs down-regulating BnSTM (line A5). h Apical and sub-apical cells of the A5 SAMs were separated by a thick cell wall. i, j Large granules of starch (i) and differentiated chloroplasts (j) were visible in the apical pole of A5 MDEs. k Cell wall width in the SAMs of the transformed lines. Scale bar = 2 μm (a, d, g), 500 nm (b, e, h, i), 2 μm (c, f), 100 nm (j). The asterisk indicates values that are statistically different (P < 0.05) from WT

Full size image

Laser capture microdissection (LCM) of the SAMs and global transcriptional analysis

Shoot apical meristems were dissected from WT, S101, and A5 embryos at day 21, which was the first day in culture in which the structure of the SAM became distinct and easily recognizable. The captured sections included the first three apical layers of cells enclosed by the two cotyledons (Fig. 4a, e), which show clear and prominent meristematic characteristics (Yeung et al. 1996). This dissected region will be referred hereafter as “captured” SAM (cSAM). The size of the cSAM (Fig. 4f) in the different lines was proportional to the abundance of BnSTM transcripts measured in this region (Fig. 4g). As a further assessment of meristem functionality, we measured the expression levels of the B. napus homologs to the Arabidopsis WUS, a key regulator defining stem cell identity in the meristem (Mayer et al. 1998). The expression of BnWUS was induced in cSAMs of embryos over-expressing BnSTM and repressed in those of embryos down-regulating BnSTM (Fig. 4h).

Fig. 4
figure 4

a–h Laser capture microdissection of SAMs after 21 days in culture. a The SAM of WT MDEs before dissection. b Cluster of meristematic cells selected for dissection in the WT SAM. c The SAM of WT MDEs after dissection. d Sample section of meristematic cells in the SAM of MDEs over-expressing BnSTM (line S101). Pink box indicates the dissected area. e Sample section of meristematic cells in the SAM of MDEs down-regulating BnSMT (line A5). The microdissected or “captured” SAMs are referred to cSAMs. f Size of the cSAM in the three lines (WT, A5, and S101). Values are means of two biological replicates ± SE. Each replicate consisted of at least 30 sections. The asterisk indicates values that are statistically different (P < 0.05) from WT. g Expression level (by qRT-PCR) of BnSTM in the cSAMs of the three lines. Values ± SE are means of three independent biological replicates. The asterisk indicates values that are statistically different (P < 0.05) from WT. Values were normalized to the value of WT set at 1. h Expression level (by qRT-PCR) of BnWUS in the cSAMs of the three lines. Values ± SE are means of three independent biological replicates. The asterisk indicates values that are statistically different (P < 0.05) from WT. Values were normalized to the value of WT set at 1. i Experimental design used to compare gene expression in the cSAMs of the three lines by microarray analysis

Full size image

Amplified RNA from the cSAMs of the three lines (WT, A5, and S101) was utilized for microarray experiments in order to measure global changes of transcript levels associated with altered BnSTM expression. A complete loop design (Fig. 4i) using four biological replicates for each of the experimental materials allowed for a direct expression comparison among the cSAMs of different lines. A total of 2,213 probes on the array were differentially expressed (Benjamini–Hochberg FDR <0.05, fold change >2) between the cSAM of A5 MDEs down-regulating BnSTM and those dissected from WT MDEs. Of these, 652 were down-regulated and 1,561 up-regulated in the cSAM of A5 embryos (Supplemental Table S2). A lower numbers of probes (1,373) were differentially expressed between the cSAMs of S101 MDEs over-expressing BnSTM and those of WT embryos. The induction of BnSTM resulted in the down-regulation of 473 gene and the up-regulation of 900 gene (Supplemental Table S3).

To facilitate the analysis we identified several gene ontology functional categories and performed enrichment analyses on the differentially expressed genes. Down-regulation of BnSTM (line A5) resulted in a selective induction of genes belonging to the categories of cell wall, plasma membrane, glycolysis, chloroplast stroma and thylakoids, as well as response to light and cadmium (Fig. 5). These profiles were generally different from those evoked by the over-expression of BnSTM. Compared to WT, cSAMs of the S101 MDEs exhibited a down-regulation of genes belonging to the categories of cell wall, nucleolus, nucleosome assembly, and polyamine biosynthesis. A direct comparative analysis between cSAMs down-regulating BnSTM (A5) and over-expressing BnSTM (S101), i.e. A5 versus S101, was also performed to better outline differences in enriched categories evoked by altered levels of the transgene (Fig. 5).

Fig. 5
figure 5

Enrichment analysis of gene ontology (GO) terms in the cSAMs of the three lines. A GO term is considered to be enriched with a log10 p value of -3.0 when compared to the whole Arabidopsis genome. GO terms were compared between lists of genes that showed differential expression (Benjamini–Hochberg FDR < 0.05, fold change > 2). Comparisons include S101 versus WT, A5 versus WT, and A5 versus S101 in both up and down-regulated gene lists. A more intense red colour signifies a more enriched GO term

Full size image

Gene expression analysis in the cSAMs down-regulating BnSTM

Genes comprising predicted functional categories enriched by the down-regulation of BnSTM were analysed in greater detail. Within the cell wall and membrane categories several arabinogalactan proteins (AGP1, 2, 12, and 22), cellulose synthases (CESA3, 6, and 8), enzymes participating in wall modification, as well as different types of transporters were induced in cSAMs down-regulating BnSTM (Table 1). A similar induction occurred for key enzymes of glycolytic processes, such as phosphofructokinase and pyruvate kinase, and many structural constituents of chloroplast compartments including components of the light harvesting complexes, photosystem subunits and oxygen evolving complex. Repression of BnSTM also induced factors with a direct involvement in hormone signalling and response. These included ABA DEFICIENT1 and ABA HYPERSENSITIVE1 for abscisic acid, two ethylene-responsive elements (ATERF-4 and -7), ARR14 for cytokinin, and the regulator of gibberellin biosynthesis GA20OX4. Characterized genes participating in SAM function (TPR2, and WSIP1) and organ formation (AS1, LBD40 and 41) were also up-regulated in A5 cSAMs (Table 1). Many TFs were repressed in the A5 cSAMs, and they included well characterized elements of auxin and ethylene signalling, such as MONOPETEROS, IAA1, and ETHYLENE RESPONSE FACTOR2, as well as other TFs implicated in developmental processes such as FUSCA3, STERILE APETALA, and MERISTEM LAYER1 (Fig. 6).

Table 1 List of representative genes induced in the cSAMs of MDEs down-1 regulating BnSTM 2 (line A5) compared to cSAMs of WT MDEs
Full size table
Fig. 6
figure 6

Profiles of selected genes (Benjamini–Hochberg FDR < 0.05, fold change > 2) encoding transcription factors (TFs) which are repressed in the cSAMs of MDEs down-regulating BnSTM (A5 line), compared to WT

Full size image

Gene expression analysis in the cSAMs over-expressing BnSTM

The up-regulation of BnSTM in cSAMs induced a preponderance of TFs such as MYB-containing TFs (MYB4R1, 77, and 98), CUP-SHAPED COTYLEDON1 (CUC1), WOX13, KNAT4, SEUSS, and two SCARECROW-LIKE genes (Fig. 7). Many members of categories participating in cell cycle, oxidative stress and translation processes were also up-regulated. These included several cyclins, anaphase promoting factors, and the cell cycle regulatory WEE1 kinase protein. Within the oxidative stress category several glutathione transferases and scavenging enzymes, such as superoxide dismutase and ascorbate peroxidase 1, were also induced in cSAMs of S101 MDEs. A similar activation occurred for genes participating in ethylene synthesis and response such as ACC SYNTHASE10, ETHYLENE INSENSITIVE3, and ETHYLENE RESPONSE FACTOR5, and 7 (Fig. 7).

Fig. 7
figure 7

Profiles of selected genes (Benjamini–Hochberg FDR < 0.05, fold change > 2) induced in the cSAMs of MDEs over-expressing BnSTM (S101 line) compared to WT. Genes were grouped in functional categories (see “Materials and methods”). TF, transcription factors

Full size image

Many genes belonging to the enriched categories of cell wall, nucleosome, nucleus and polyamine synthesis were down-regulated in BnSTM over-expressing cSAMs (Table 2). These included XYLOGLUCAN ENDOTRANSGLYCOSYLASE 6 and 9, EXTENSIN A5, and two PECTINESTERASE genes belonging to the cell wall category, several HISTONE genes involved in nucleosome assembly, structural components of the nucleolus, and polyamine biosynthetic enzymes such as AGMATINE IMINOHYDROLASE and ARGININE DECARBOXYLASE2. The introduction of BnSTM also repressed the expression of several components participating in proton transport (Table 2).

Table 2 List of representative genes repressed in the cSAM of MDEs over-expressing BnSTM (line 101) compared to WT cSAM
Full size table

Clustering of genes with similar expression patterns

K-means analysis was performed to identify clusters of genes with similar expression patterns (Fig. 8a). Of the 13 clusters selected 2 (#7 and 13, Fig. 8a and Supplemental Table S4) were examined in detail as they included genes whose expression was closely regulated by the level of BnSTM. Cluster #7 was composed of 278 elements which are repressed in the cSAMs over-expressing BnSTM and induced in cSAMs down-regulating BnSTM. Many of the genes represented by these probes participate in wall synthesis and modification processes: GLUCAN SYNTHASE3, EXTENSIN3, POLYGALACTURONASE4, CELL WALL INVERTASE2, ANKYRIN-CONTAINING PROTEIN2, and the lignin biosynthetic gene CAFFEOYL-COA METHYLTRANSFERASE. Other members of cluster #7 included membrane-associated transporters, and constituents participating in oxygen evolving complexes, photosystem reaction centres and light harvesting units (Fig. 8b).

Fig. 8
figure 8

a Selected gene clusters (K-means) showing expressing profiles of probes (Benjamini–Hochberg FDR < 0.05, fold change > 2) with different behaviour in the cSAMs of MDEs generated by the WT line and lines down-regulating BnSTM (A5) or up-regulating BnSTM (S101). The number of probes in each cluster is represented in brackets. b Selected genes belonging to cluster #7. c Selected genes belonging to cluster #13

Full size image

Probes with an opposite behaviour, i.e. induced in BnSTM-over-expressing cSAMs and repressed in BnSTM-down-regulating SAMs, were grouped into cluster #13. This group included many transcriptional regulators such as RGL3 and RGA1, encoding DELLA proteins which repress gibberellin signalling, ETHYLENE RESPONSE FACTOR 6, STERILE APETALA and MERISTEM LAYER 1 (Fig. 8c). Other members representative of cluster #13 are translation factors, the CYCLIN-DEPENDENT KINASE A2.3, and three genes involved in detoxification processes: SUPEROXIDE DISMUTASE1, ASCORBATE PEROXIDASE1, and GLUTATHIONE REDUCTASE (Fig. 8c).

Cell wall and lignin biosynthetic pathways in cSAMs of the transformed lines

The opposite expression behaviour of several genes belonging to the cell wall category (Tables 1, 2; Fig. 8), and differences in meristematic cell wall characteristics (Fig. 3) observed between SAMs over-expressing or down-regulating BnSTM, prompted us to examine the regulatory network of cell wall and lignin in cSAMs (Fig. 9). All genes analysed by qRT-PCR, including the NAC domain transcription factors which activate secondary wall formation in vessel elements (VND7) and fibres (SND1 and NST1), MYC-transcription factors involved in the regulation of cell wall (MYB43) or lignin (MYB58, 63) and their downstream targets were induced in cSAMs down-regulating BnSTM. This induction was very pronounced for genes encoding lignin biosynthetic enzymes: PHENYLALANINE AMMONIA LYASE2 (PAL2), CINNAMATE 4-4HYDROXYLASE (C4H), and CINNAMYL ALCOHOL DEHYDROGENASE (CAD). With the exclusion of NST1, CESA3, and MYB63, a mild transcriptional repression of the cell wall and lignin biosynthetic pathways occurred in the BnSTM over-expressing cSAMs (Fig. 9).

Fig. 9
figure 9

Expression studies by qRT-PCR of genes involved in the regulatory network of cell wall and lignin biosynthesis in the cSAMs of MDEs generated by the WT line and lines down-regulating BnSTM (A5) or up-regulating BnSTM (S101). Values ± SE are means of three independent biological replicates. The asterisk indicates values that are statistically different (P < 0.05) from WT. Values were normalized to the value of WT set at 1. VND7, VASCULAR-RELATED NAC-DOMAIN; NST1, NAC SECONDARY WALL THICKENING PROMOTING FACTOR1; SND1, SECONDARY WALL-ASSOCIATED NAC DOMAIN1; CESA, CELLULOSE SYNTHASE; PAL, PHENYLALANINE AMMONIA LYASE; C4H, CINNAMATE 4-4HYDROXYLASE; CAD, CINNAMYL ALCOHOL DEHYDROGENASE; CCR2, CINNAMOYL COA REDUCTASE2

Full size image

To further establish a relationship between BnSTM, SAM, and cell wall and lignin biosynthesis we measured the expression level of cell wall and lignin biosynthetic genes in MDEs cultured under conditions affecting BnSTM expression. Previous studies showed that the quality of the SAMs in Brassica MDEs can be altered experimentally by simple manipulations of the culture medium. While inclusions of 0.1 mM dl-buthionine (BSO) enhance the structure and functionality of the shoot meristem (Belmonte et al. 2006), applications of the auxin inhibitor 2,3,5-triiodobenzoic acid (TIBA) at a concentration of 2 μM compromise the proper formation of the SAM (Ramesar-Fortner and Yeung 2006). The SAM of TIBA-treated MDE does not form properly and fails to produce primordia at germination (Ramesar-Fortner and Yeung 2006). Using these treatments, we showed that the opposite effects exercised by BSO and TIBA on meristem formation correlate to early changes in BnSTM expression. The up-regulation of BnSTM with BSO and its down-regulation with TIBA, were observed at day 7 in culture, 2 days after the applications of the respective compounds (day 5, Supplemental Fig. S2a) and preceded structural changes of the meristem [day 21 for both TIBA (Ramesar-Fortner and Yeung 2006) and BSO (Belmonte et al. 2006)]. Localization studies showed that compared to untreated (control) MDEs, applications of BSO enlarge the expression domain of BnSTM, which encompasses the whole apical region (Supplemental Fig. S2b). This was in contrast to TIBA-treated MDEs, where BnSTM mRNAs are restricted to the central cells of the SAM. At day 21, no BnSTM expression is detected in the TIBA-treated SAMs, which display structural abnormalities, i.e., fusion of cotyledons, similar to those observed in MDEs down-regulating BnSTM (compare Fig. 1d and Supplemental Fig. S2b). The TIBA-suppression of BnSTM correlates to an overall increase in expression of many cell wall and biosynthetic genes (Supplemental Fig. S3). This increase is apparent at day 14, prior to any visible structural abnormality in the SAM (Ramesar-Fortner and Yeung 2006), thus suggesting a causative role of this increased gene expression pattern on the improper formation of the SAM. A mild transcriptional repression of PAL2, CAD, and CCR2 was detected in BSO-treated MDEs, which show an increased expression of BnSTM (Supplemental Fig. S3).

Gene expression studies in Arabidopsis line with inducible STM function

In an effort to ascertain whether key genes differentially expressed in the microarray study by alterations in BnSTM levels are direct target of STM, we employed an Arabidopsis-inducible system (35S::STM-GR). In this system the combined applications of DEX and CHX ensure the respective migration of STM-GR in the nucleus and the repression of translation of primary targets RNAs, thus identifying those transcripts directly affected by STM (Gallois et al. 2002; Spinelli et al. 2011). Of the transcripts investigated only CUP-SHAPED COTYLEDON1 (CUC1) was directly regulated by STM, as also reported by Spinelli et al. (2011). The other genes, many of which are involved in cell wall and lignin synthesis, were found to be not direct targets of STM (Supplemental Fig. S4).

Discussion

The use of laser microdissection permits the accurate isolation of specific cells/tissues and therefore enables transcriptional analyses without the background noise of adjacent domains. This technique is especially critical to address fundamental questions about the regulatory mechanism of organs composed of distinct structural and functional regions, such as the SAMs. Transcriptional studies of microdissected meristems have revealed complex regulatory networks maintaining the precise equilibrium between cell division and differentiation, a unique feature of functional SAMs (Ohtsu et al. 2007; Brooks et al. 2009; Yadav et al. 2009). A key regulatory element participating in meristem homeostasis is STM, a KNOX-1 member which suppresses differentiation thus maintaining an undetermined cell fate within the SAM (Barton and Poethig 1993; Clark et al. 1996; Endrizzi et al. 1996). Our previous studies showed that altered levels of the Brassica napus (Bn) STM cause profound post-embryonic developmental abnormalities which phenocopy Arabidopsis STM over-expressers and mutants. These include the presence of lobed leaves in BnSTM over-expressers, and the premature termination of the shoot meristem in seedling down-regulating BnSTM (see Supplementary Fig. 1B of Elhiti et al. 2012). We also assessed the effects of altered expression of BnSTM during embryogenesis and showed an enhanced production of B. napus MDEs in lines over-expressing BnSTM and a reduction in MDE number in lines down-regulating BnSTM (Elhiti et al. 2010). As an extension of that work, the present study provides novel evidence that besides influencing embryo yield, transcriptional alterations of BnSTM also affect the quality of the MDEs and the functionality of their SAMs through profound changes in morphology and gene expression. During MDE development the down-regulation of BnSTM causes the termination of the SAM (Fig. 1d), a phenotype reminiscent of Arabidopsis stm mutants (Barton and Poethig 1993), as well as secondary embryonic SAMs (Fig. 1h) and outgrowth along the embryo axis (Fig. 1e–g). These latter developmental abnormalities, not observed in stm Arabidopsis mutants, might be indirect effects resulting from the inhibition of cell proliferation which under normal circumstances is provided by a functional SAM. The up-regulation of BnSTM enhances the structure and quality of the SAM (Fig. 1b) resulting in improved regeneration frequency at germination (Fig. 2). This observation is novel, and promising for enhancing regeneration in in vitro propagation systems. Successful propagation via somatic or gametophytic embryogenesis often relies on the quality of the SAMs and their ability to convert at germination through the production of vigorous shoots. These prerequisites are often lacking due to structural and physiological abnormalities arising in the SAMs of developing somatic or microspore-derived embryos (Kong and Yeung 1992; Belmonte et al. 2006, 2010). The improved conversion frequency of BnSTM-over-expressing MDEs may be partially due to the enlarged SAMs containing a higher number of cells displaying meristematic characteristics (Fig. 1b) and altered expression of discrete classes of genes. Examination of gene expression profiles revealed remarkable differences in the expression of TFs, with an overall repression in the cSAMs down-regulating BnSTM (Fig. 6), and an induction in the cSAMs over-expressing BnSTM (Fig. 7). A SAM-specific TF responsible for the formation of the organizing centre whose activity establishes a cluster of stem cells is WUS (Laux et al. 1996; Mayer et al. 1998). In Brassica, the expression of BnWUS shows a similar expression profile to that exhibited by BnSTM (Fig. 4g, h). While elevated levels of WUS transcripts might be the direct cause of enlarged SAMs (Fig. 1b; Schoof et al. 2000), its down-regulation is associated with the termination of the SAM (Fig. 1d; Mayer et al. 1998). Modulation of WUS activity is exerted through the proper balance of WUS-interacting proteins which act as transcriptional co-repressors of target genes (Kieffer et al. 2006). Perturbation of this balance is likely to occur in BnSTM down-regulating cSAMs, where the level of two TOPLESS-related proteins interactors of WUS (WSIP1 and TPR2) and repressor of downstream genes is increased (Table 1). Altered expression of TOPLESS-related proteins has been associated with abnormalities in embryonic meristem development (Long et al. 2006).

Other relevant meristem-related TFs induced by BnSTM include CUC1, ML1, SEUSS, INCURVATA 4, and COV1 (Figs 7, 8). In developing Arabidopsis embryos CUC1 is localized in cells predicting the embryonic SAM and its expression is required for defining meristem boundary (Aida et al. 1997). Its contribution to the enlarged SAMs of BnSTM-over-expressing MDEs cannot be excluded, given its role in maintaining undifferentiated cells during meristem formation (Takada et al. 2001). Recent studies (Spinelli et al. 2011) suggest a mechanistic link between STM and CUC1 whereby STM activates CUC1 which in turns induces STM expression through a feedback loop. It is likely that similar regulatory programmes operate in closely related Brassica, and these would contribute to the enlarged pool of cells displaying meristematic features in the SAM over-expressing BnSTM. Proper meristem formation during embryogenesis also requires controlled stratification, i.e. layer separation, resulting in the development of functional domains. In Arabidopsis stratification is regulated by a set of genes including ATML1 (Lu et al. 1996). This gene is specifically expressed in the layer 1 from the very early stages of meristem patterning and encodes a homeodomain-containing TF (Lu et al. 1996). Our studies suggest that the expression of the B. napus homolog of ATML1 might be regulated by BnSTM level (Fig. 8c), and its repression in cSAMs down-regulating BnSTM may possibly reflect defects in layer organization. Two unexpected TFs induced in BnSTM over-expressing cSAMs are INCURVATA4 and COV1, both conferring positional information for vascular patterning (Parker et al. 2003; Ochado et al. 2008). Without discounting an improbable contamination of procambial tissue during the microdissection of the SAM, it can be speculated that the expression of the two genes in meristematic cells might be the result of altered levels of auxin, which is known to activate INCURVATA4 and COV1 and promote vascularization (Parker et al. 2003; Ochado et al. 2008). Our previous studies suggest a possible role of BnSTM in altering auxin responsiveness (Elhiti et al. 2010), and the altered expression of two regulators of auxin signalling, MONOPTEROS and IAA1 in cSAMs down-regulating BnSTM (Fig. 6) supports this notion.

Another TF, regulator of SAM properties and with a potential involvement in auxin response is SEUSS, a transcriptional adaptor regulating the expression of target genes through the formation of a physical complex with LEUNIG (Bao et al. 2010). High transcript levels of SEUSS, which in our studies are induced by BnSTM (Fig. 7), are required for normal SAM formation (Bao et al. 2010).

Besides altering the expression of TFs with known functions in the SAM, BnSTM might regulate the expression of additional TFs required for other developmental processes, ZIM, STERILE APETALA, as well as hormone-related TFs participating in the signalling of cytokinin, ARR2, gibberellins, RGA1, RGL3, and ethylene ETHYLENE RESPONSE FACTOR 6 (Fig. 8c). Some of these changes are consistent with studies defining the regulatory role of STM in cytokinin and gibberellin pathways (Jasinski et al. 2005) and are suggestive of contrasting physiological conditions existing in SAMs with altered BnSTM levels.

Opposite expression patterns between cSAMs over-expressing or down-regulating BnSTM were also observed for genes predicted to function during oxidative stress, plastid development and cell wall synthesis and modification (Fig. 8). Accumulation of reactive oxygen species (ROS) produced continuously as by-products of several developmental pathways cause extensive cell damage with negative implications on tissue patterning and organ development (Foyer 2005). The extent of cellular ROS levels relies on the antioxidant systems composed by several enzymes including superoxide dismutase, glutathione reductase, and ascorbate peroxidase. The opposite expression profile (induction in cSAMs over-expressing BnSTM and repression in cSAMs down-regulating BnSTM) of several genes encoding for these enzymes (Fig. 8c) suggests a tight regulation of the antioxidant system mediated by BnSTM. While a reduced antioxidant response and subsequent ROS accumulation is known to disrupt developmental processes (Cairns et al. 2006) and possibly the formation of the SAM in BnSTM down-regulating MDEs, active antioxidant signalling might contribute to enhance meristem development in the BnSTM over-expressing MDEs. Furthermore, besides scavenging ROS, redox homeostasis modulated by several antioxidant enzymes including glutathione reductase and ascorbate peroxidase, acts as a signal for several morphogenic processes including in vitro meristem formation (Belmonte et al. 2006; Stasolla et al. 2008). In particular, the higher expression of glutathione reductase in the enlarged SAMs over-expressing BnSTM might be indicative of an enlarged pool of reduced glutathione, a promoter of cell division and meristematic growth (Sánchez-Fernández et al. 1997; Vernoux et al. 2000). Consistent with this notion we observed the induction of several cell cycle components in BnSTM over-expressing cSAMs (Fig. 7).

Plastid differentiation is an important event during embryogenesis which in Arabidopsis begins at the torpedo stage of development (Mansfield and Briarty 1991). The induction of many genes involved in chloroplast biogenesis and function observed in the cSAMs down-regulating BnSTM might be an indication of advanced cellular differentiation, while the repression of the same genes in cSAMs over-expressing BnSTM reflects an undifferentiated cellular state (Fig. 8b). It cannot be excluded, however, that plastid development may play a more active role during SAM formation, as shown by the requirement of plastid proteins during other developmental programmes. Through the analyses of publicly available datasets and mutant studies, Hsu et al. (2010) demonstrated that plastid activity is indispensable for Arabidopsis embryogenesis. Specifically while non-photosynthetic metabolic activity of plastids is needed for the transition from pre-globular to globular embryos, proteins participating in plastid gene expression become necessary during the later phases of morphogenesis. Thus, it would be worthwhile assessing if similar requirements also occur during the formation of the embryonic SAM.

The most striking finding of our data is the preponderance of genes involved in cell wall biology induced in cSAMs with low BnSTM levels and repressed in cSAMs over-expressing BnSTM (Tables 1, 2; Fig. 8b). A down-regulation of BnSTM results in the induction of a variety of genes some of which, CELLULOSE SYNTHASE (CESA 3, 6, and 8), are involved in cell wall synthesis while others, EXTENSIN 3 and PAL2, encode for wall modifying enzymes. This increase in transcription activity is associated with the formation of thicker walls in the meristematic cells (Fig. 3). To further examine these data, we measured the transcript levels of representative enzymes and TFs involved in known regulatory networks of cell wall and lignin biosynthetic pathways (Fig. 9). Besides confirming the microarray data these expression studies revealed two important concepts. Firstly, they showed that BnSTM affects the expression of a large number of genes, indicating that it may act early in the wall and lignin biosynthetic programmes. Secondly, the down-regulation of BnSTM results in a very pronounced induction of the lignin biosynthetic genes PAL2, C4H and CAD, whereas a moderate repression of the same genes occurs in cSAMs over-expressing BnSTM. It is therefore plausible to assume that the lignin pathway is one of the coordinate processes mediated by BnSTM and that this regulation is integral to the role fulfilled by the gene in maintaining cells in an indeterminate state. Since lignin formation is a feature of irreversible cell differentiation, it can be argued that one way in which BnSTM promotes indeterminate cell fate is by preventing the accumulation of lignin in the meristematic cells. The activation of the lignin biosynthetic pathway in the SAMs down-regulating BnSTM is suggestive of heavy synthesis and accumulation of lignin resulting in the differentiation of the meristematic cells and premature termination of the SAM. Increased expression of lignin biosynthetic genes resulting in improper meristem formation can also be induced in culture through manipulations of the culture environment, as shown by treatments with TIBA (Supplemental Fig. S3). The proposed BnSTM-mediated regulation of cell determinacy through modulation of cell wall components is a novel concept which might be a prerogative of other KNOX members. Mele et al. (2003) documented a similar control of cell wall enzymes by BREVIPEDICELLUS (BP), one of the seven KNOX genes in Arabidopsis. Consistent with our results the authors observed the induction of several lignin biosynthetic enzymes and increased lignin deposition in bp plants, and decreased lignification in plants over-expressing BP. These results were explained by the ability of BP to bind promoters of several genes of the lignin pathway (Mele et al. 2003).

Gene expression studies in the Arabidopsis-inducible system (35S::STM-GR), revealed that with the exclusion of CUC1, all the genes analysed are indirect targets of STM. These include several cell wall and lignin biosynthetic genes (CESA6, CAD, MYB43, MYB 58, PAL2, and CCR2), which were differentially expressed in the microarray experiment. Neomorphic effects unrelated to the function of STM cannot be excluded due to the approach used to up-regulate or suppress STM. The observed changes in transcript levels in the microarray experiments might not be necessarily the results of STM, but rather the results of other class1 KNOX genes whose expression might have been altered in the transgenic lines. This possibility is further substantiated by the fact that the transformed plants over-expressing or down-regulating STM do not exactly phenocopy the known STM over-expression and mutant phenotypes.

In conclusion, we show that altered expression of BnSTM has profound effects on in vitro SAM development. While repression of BnSTM levels in B. napus MDEs compromises the formation of the SAM and the post-embryonic performance of the embryos, its ectopic expression enlarges the size of the meristem and improves the ability of the embryos to regenerate. The regeneration frequency of BnSTM over-expressing MDEs more than doubles that of their WT counterparts. Transcriptional analyses of microdissected SAMs revealed profound changes in gene expression patterns. Over-expression of BnSTM alters the abundance of transcripts encoding TFs, some of which are directly involved in SAM formation, and transcripts of regulatory components of hormone signalling, antioxidant response, and cell wall formation and modification. Structural and expression studies confirm the role of BnSTM in mediating the regulatory network of cell wall and lignin synthesis, which may be a requirement for modulating fate determination in the meristematic cells. Overall, these novel findings position BnSTM as a valuable target for enhancing shoot meristem formation in vitro and improving regeneration of in vitro produced embryos.