Expression of PaNAC01, a Picea abies CUP-SHAPED COTYLEDON orthologue, is regulated by polar auxin transport and associated with differentiation of the shoot apical meristem and formation of separated cotyledons

Abstract

Background and Aims

During embryo development in most gymnosperms, the establishment of the shoot apical meristem (SAM) occurs concomitantly with the formation of a crown of cotyledons surrounding the SAM. It has previously been shown that the differentiation of cotyledons in somatic embryos of Picea abies is dependent on polar auxin transport (PAT). In the angiosperm model plant, Arabidopsis thaliana, the establishment of cotyledonary boundaries and the embryonal SAM is dependent on PAT and the expression of the CUP-SHAPED COTYLEDON (CUC) genes, which belong to the large NAC gene family. The aim of this study was to characterize CUC-like genes in a gymnosperm, and to elucidate their expression during SAM and cotyledon differentiation, and in response to PAT.

Methods

Sixteen Picea glauca NAC sequences were identified in GenBank and deployed to different clades within the NAC gene family using maximum parsimony analysis and Bayesian inference. Motifs conserved between angiosperms and gymnosperms were analysed using the motif discovery tool MEME. Expression profiles during embryo development were produced using quantitative real-time PCR. Protein conservation was analysed by introducing a P. abies CUC orthologue into the A. thaliana cuc1cuc2 double mutant.

Key Results

Two full-length CUC-like cDNAs denoted PaNAC01 and PaNAC02 were cloned from P. abies. PaNAC01, but not PaNAC02, harbours previously characterized functional motifs in CUC1 and CUC2. The expression profile of PaNAC01 showed that the gene is PAT regulated and associated with SAM differentiation and cotyledon formation. Furthermore, PaNAC01 could functionally substitute for CUC2 in the A. thaliana cuc1cuc2 double mutant.

Conclusions

The results show that CUC-like genes with distinct signature motifs existed before the separation of angiosperms and gymnosperms approx. 300 million years ago, and suggest a conserved function between PaNAC01 and CUC1/CUC2.

INTRODUCTION

Post-embryonic growth and development in plants depend on the establishment and subsequent activity of the shoot and the root meristems. Meristem formation has been extensively studied in the model angiosperm Arabidopsis thaliana, but to a much lesser extent in gymnosperms. Angiosperms and gymnosperms separated approx. 300 million years ago (Smith et al., 2010) and, apart from the distinctive feature of having enclosed or open (naked) seeds, there are many developmental differences between the two groups. For example, during embryogenesis of most conifers, the cotyledons develop as a crown surrounding the incipient shoot apical meristem (SAM). This contrasts with the A. thaliana embryo, in which a symmetry-breaking event (from radial to bilateral symmetry) is associated with the emergence of the SAM and the two opposing cotyledons (Chandler, 2008). Pattern formation in conifer embryos has been difficult to study as many conifers set cones irregularly, and the embryo development within one cone is highly unsynchronized. However, we have established a protocol for somatic embryogenesis in Picea abies, which allows us to regenerate a large number of somatic embryos at specific developmental stages at specific time points. This protocol has proven to be most valuable for studying pattern formation in conifer embryos (von Arnold and Clapham, 2008).

The NAC family constitutes one of the largest plant-specific families of transcription factors (Riechmann et al., 2000). Members of this family have a DNA-binding domain in the N-terminal region known as the NAC domain, an acronym derived from the first characterized NAC genes, NO APICAL MERISTEM (NAM) in Petunia, and ATAF1, ATAF2 and CUP-SHAPED COTYLEDON2 (CUC2) in A. thaliana (Aida et al., 1997). The function and the expression pattern of reported NAC genes are highly diverse, and include roles in SAM formation and organ separation, cell division, hormone signalling, senescence, growth of floral organs, lateral root formation, stress responses and wood formation (reviewed by Olsen et al., 2005; Shen et al., 2009). Despite the importance of NAC proteins in fundamental processes, to our knowledge, nothing is presently known about the role of NAC genes in conifers.

The first NAC gene to be characterized was NAM, which was discovered in a seedling-lethal mutant with no SAM in Petunia (Souer et al., 1996). A similar phenotype was later discovered in cuc1cuc2 double mutants of A. thaliana (Aida et al., 1997). To date, three CUC genes (CUC1, CUC2 and CUC3) have been characterized in A. thaliana (Aida et al., 1997; Takada et al., 2001; Vroemen et al., 2003). They act together with the KNOTTED1-like homeobox (KNOXI) gene SHOOT MERISTEMLESS (STM) (Barton and Poethig, 1993; Aida et al., 1999) to establish the embryonal SAM and to promote formation of two separated cotyledons. The CUC genes are expressed in the intercotyledonary regions, where the gene activity presumably prevents cell proliferation and cotyledon outgrowth (Breuil-Broyer et al., 2004). The restriction of CUC gene expression is regulated by PINFORMED1 (PIN1), PINOID (PID) and MONOPTEROS (MP), which are genes involved in auxin transport and response (Aida et al., 2002; Furutani et al., 2004). PIN-mediated polar auxin transport (PAT) is important for pattern formation, and studies of auxin-responsive gene expression have revealed that cotyledon outgrowth occurs at positions of PAT-mediated auxin response maxima (Benkova et al., 2003). Furthermore, blocked PAT leads to the development of seedlings with fused cotyledons (Okada et al., 1991; Liu et al., 1993; Hadfi et al., 1998). In addition to being regulated by PAT, the abundance of CUC1 and CUC2 mRNA is post-transcriptionally regulated by the microRNA miR164 (Laufs et al., 2004; Mallory et al., 2004).

We have previously shown that PAT is crucial for correct pattern formation in somatic embryos of P. abies (Larsson et al., 2008). Blocked PAT resulted in increased endogenous indole 3-acetic acid (IAA) levels, abnormal cell divisions, decreased programmed cell death, fused cotyledons and aborted SAM. Furthermore, we recently showed that the expression of two KNOXI genes (HBK2 and HBK4) was affected by blocked PAT during embryo development in P. abies (Larsson et al., 2012). As a first step to characterize the NAC gene family in conifers, and specifically the role of CUC genes during embryogenesis, we have here identified 16 P. glauca NAC genes in GenBank and allocated them to different clades within the NAC gene family. Furthermore, two P. abies NAC genes, PaNAC01 and PaNAC02, highly similar to the A. thaliana CUC genes, were characterized. The expression profile of PaNAC01 showed that the gene is regulated by PAT, and associated with SAM differentiation and cotyledon formation. Together the results show that CUC-like genes existed before the separation between angiosperms and gymnosperms, and suggest a conserved function between PaNAC01 and CUC1/CUC2.

MATERIALS AND METHODS

Bioinformatics

NAC domains from Picea glauca were retrieved by a tblastn search in the nucleotide collection at NCBI (http://blast.ncbi.nlm.nih.gov/Blast; 30 May 2011), using the NAC domains of representative Arabidopsis thaliana proteins as query (i.e. CUC2, NAC1, ATAF1, RD26 and TIP1). The obtained sequences were further analysed to ensure that only sequences representing full-length NAC domains were included. To put these sequences into a phylogenetic context, they were compared with 1–6 NAC domain-encoding nucleotide sequences from A. thaliana, Medicago truncatula and Physcomitrella patens (where applicable) from each monophyletic clade presented by Shen et al. (2009). The A. thaliana sequences were retrieved based on accession numbers from NCBI, sequences from P. patens were retrieved from the Pfam protein families database (Finn et al., 2010), and sequences from M. truncatula were retrieved from the DFCI Medicago gene index (http://compbio.dfci.harvard.edu/cgi-bin/tgi/gimain.pl?gudb=medicago), and through a tblastn search in the nucleotide collection at NCBI. When characterized proteins were available, these were prioritized before uncharacterized sequences. Upstream sequences (3000 bp) for all A. thaliana genes within the CUC clade were retrieved from The A. thaliana Information Resource (TAIR; www.arabidopsis.org).

Sequences were aligned using translational alignment available in the software package Geneious 4·8·5 (alignment available upon request). Maximum parsimony trees were generated using PAUP* 4·0 (Swofford, 2002). A heuristic search of replicates with 1000 random stepwise taxon additions was performed. All equally parsimonious trees were retained for analysis, and the Tree Bisection–Reconnection (TBR) algorithm was used for branch swapping. Gaps were treated as missing data. Bootstrap support for nodes (Felsenstein, 1985) were estimated with 1000 heuristic search replicates using the same settings as the original search, with 100 random stepwise additions for each bootstrap replicate. Bayesian analysis was performed using Metropolis-coupled Markov chain Monte Carlo methods as implemented in MrBayes V. 3·1·1 (Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck, 2003). The chain was run for 20 million generations without enforcing a molecular clock with a random starting tree. The chain was sampled every 5000 generations for a total of 8002 trees, and used to calculate posterior probabilities for clades (clade credibility values; CVs). The first 1000 trees were discarded as ‘burn-in’ (trees generated before likelihood values reached stationary). The General Time Reversible model with gamma distributed rates across sites was used, and the number of discrete categories was set to 4. Four chains were run, with one chain heated at 0·4.

The motif discovery tool, MEME (Bailey and Elkan, 1994), was used together with manual analysis in order to find conserved motifs in the C-terminal regions of the two P. abies sequences. For this, all A. thaliana and available M. truncatula sequences, that according to Shen et al. (2009) cluster with the CUC genes, were collected. In addition, dicot sequences characterized by Blein et al. (2008) and Souer et al. (1996), and monocot sequences characterized by Zimmermann and Werr (2005) were included in the analysis for comparison. No P. patens sequences were included in the analysis since most of these sequences lack a C-terminal domain. All accession numbers, common names and references are presented in Supplementary Data Table S1. MEME was also used to find conserved motifs in the regulatory sequences of genes within the CUC clade from P. abies and A. thaliana.

Plant materials

Embryogenic cell lines of P. abies were established from mature zygotic embryos according to von Arnold and Clapham (2008). Cell lines 28:05 and 88:22 were used throughout this study. The cell lines were stored in liquid nitrogen and thawed approx. 6 months before the start of the experiments. After thawing, the cell cultures were treated as described previously (von Arnold and Clapham 2008). Briefly, proembryogenic masses (PEMs) were maintained in liquid proliferation medium containing the plant growth regulators (PGRs) 2,4-dichlorophenoxyacetic acid (2,4-D) and N⁶-benzyladenine (BA) at 9·0 and 4·4 µm, respectively. To stimulate differentiation of early somatic embryos from PEMs, the cultures were transferred to pre-maturation medium lacking PGRs for 1 week. For development of late somatic embryos and maturation, the cultures were plated on solidified maturation medium containing 30 µm abscisic acid (ABA).

To study the effects of PAT on gene expression during embryo development, embryogenic cultures were treated with 20 µm 1-N-naphthylphtalamic acid (NPA) from pre-maturation until maturation, as previously described (Larsson et al., 2008). Cultures grown in media lacking additives were used as controls, since we have previously shown that the NPA solvent dimethylsulfoxide (DMSO) does not affect the cotyledon differentiation and SAM formation (Larsson et al., 2008). Samples for gene expression studies were collected from eight sequential stages (Fig. 5), from proliferating PEMs until fully mature cotyledonary embryos. The developmental stages of NPA-treated embryos were largely defined according to the sizes of phenotypically normal control embryos. Whole mount of tissue was sampled from proliferating PEMs (stage 1) and early embryos (stage 2). From stage 3 and onwards, the embryos were picked individually and sorted according to developmental stage to increase the signal to noise ratio in subsequent gene expression studies. The samples were stored at –80 °C until used.

Fig. 5.

Developmental stages of control (A–H) and NPA-treated (I–L) somatic embryos of Picea abies. The stages of NPA-treated embryos were largely defined according to the sizes of control embryos rather than according to age; however, the average exposure time to ABA for reaching a certain stage is presented. (A) Stage 1, proliferating proembryonic masses in the presence of the plant growth regulators (PGRs) auxin and cytokinin. (B) Stage 2, early embryo 1 week after withdrawal of PGRs. (C) Stage 3, developing late embryo after 1–2 weeks exposure to ABA. (D) Stage 4, late embryo after 2–3 weeks exposure to ABA. (E, I) Stage 5, early maturing embryos after 3–4 weeks exposure to ABA. Note that the shoot meristem and cotyledons start to become visible in the normal (E) but not in the NPA-treated (I) embryo. (F, J) Stage 6, maturing embryos after 4–5 weeks (F) or 5–6 weeks (J) exposure to ABA. (G, K) Stage 7, almost fully matured control embryo after 5–6 weeks exposure to ABA (G), almost fully matured embryo with doughnut-shaped apical part after 6–7 weeks exposure to ABA and NPA (K). (H, L) Stage 8, fully matured cotyledonary embryo after 6–7 weeks exposure to ABA (H), and fully matured embryo lacking separated cotyledons after 7–8 weeks exposure to ABA and NPA (L). co, cotyledon; cp, cotyledon primordia; dsc, doughnut-shaped cotyledon; em, embryonal mass; fc, fused cotyledons; s, suspensor; SAM, shoot apical meristem; sp, shoot apical meristem primordium; tc, tube cells. Scale bars = 100 µm (A–G, I–K) and 250 µm (H, L).

DNA and RNA isolation and cDNA synthesis

Genomic DNA was extracted using the cetyltrimethylammonium bromide (CTAB) method. Briefly, samples snap-frozen in liquid nitrogen were disrupted in tissue lyser II (Retsch, Haan, Germany). Extraction buffer [2 % CTAB, 2 % polyvinylpyrrolidone-40, 100 mm Tris–HCl (pH 8), 20 mm EDTA, 2·5 m NaCl, 20 mm β-mercaptoethanol and 10 µg mL⁻¹ RNase A] was added to each sample, which were then incubated at 65 °C for 10 min. Cell debris was removed by centrifugation, and DNA was isolated using chloroform/isoamylalcohol (24:1) and ethanol precipitation.

Total RNA was extracted according to a modified protocol from Azevedo et al. (2003). Briefly, samples frozen in liquid nitrogen were disrupted in tissue lyser II (Retsch). Extraction buffer was added to each sample, which again was frozen in liquid nitrogen, thawed at 42 °C, disrupted and finally incubated at 42 °C for 90 min, followed by subsequent RNA isolation. For each sample, 10 µg was treated with DNase using the DNA-free protocol by Ambion (Ambion Inc., Austin, TX, USA), and 1 µg of the DNA-free RNA was used to synthesize cDNA using the qScript™ cDNA Synthesis Kit (Quanta BioSciences Inc., Gaithersburg, MD, USA) according to the protocol provided by the manufacturer.

Cloning of PaNAC01 and PaNAC02

Full-length PaNAC01 cDNA was isolated from somatic embryos by 3′ rapid amplification of cDNA ends (RACE) using primers based on a P. glauca expressed sequence tag (EST) sequence (GenBank accession no. DV994679) and the GeneRacer™ kit (Invitrogen, Carlsbad, CA, USA). A putative translational start codon was identified and primers were designed to bind to the putative 5'-untranslated region (UTR) in order to obtain a full-length sequence. The full-length cDNA of PaNAC02 was isolated from somatic embryos by reverse transcription–PCR (RT–PCR) using primers designed from a P. glauca sequence (GenBank accession no. BT116398·1). Upstream sequences of both genes were amplified using the GenomeWalker™ Universal Kit (Clontech Laboratories, Mountain View, CA, USA). All primers are presented in Supplementary Data Table S2. Products were separated on a 1 % agarose gel and purified using a QIAquick^® PCR purification kit (QIAGEN, Hilden, Germany). The PaNAC01 cDNA sequence was inserted into the pCR^®4-TOPO^® cloning vector and cloned using the TOPO TA Cloning^® Kit for Sequencing (Invitrogen), while all other sequences were cloned into the pJET1·2/blunt cloning vector using the CloneJET™ PCR Cloning Kit (Fermentas, Helsingborg, Sweden). The clones were sequenced on both strands by Eurofins MWG Operon (Ebersberg, Germany). The cDNA sequences and genomic sequences were submitted to GenBank (accession nos HM68414 and HM638415 for PaNAC01 and PaNAC02, respectively).

Quantitative real-time PCR

Quantitative real-time PCR (qRT-PCR) was performed using the DyNAmo™ Flash SYBR^® Green qPCR Kit (Finnzymes, Espoo, Finland) in a BIO-RAD iQ™5 Multicolor Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). Primers used to quantify expression levels are presented in Supplementary Data Table S2. The expression data were normalized against the expression of the reference genes CELL DIVISION CONTROL2 (CDC2), ELONGATION FACTOR-1 ALPHA (EF1-α) and PHOSPHOGLUCOMUTASE, previously selected based on stability (Vestman et al., 2011) using the geNorm software (Vandesompele et al., 2002). PCR cycling conditions were as advised by the manufacturer in the DyNAmo™ Flash SYBR^® Green qPCR Kit (Finnzymes) with annealing and extension at 60 °C for 30 s. The reactions were run for 40 cycles, and at the end of each run a melting curve was generated to ensure product uniformity. All samples were added in triplicate to each plate and each gene was assessed in three independent biological replicates. All analyses were performed in the iQ5 software (Bio-Rad Laboratories).

Statistical analyses of the results from the qRT-PCR were performed using the SAS (2008) GLM procedure. Effects of treatment and sample on the expression were analysed using general linear models with treatment and sample as factors. The different genes were assessed separately.

Vector construction and transformation of A. thaliana

To study the effect of the identified conserved promoter motif on gene expression, the sequences flanking the motif in the 3 kb upstream region of A. thaliana CUC2 (proAtCUC2) was amplified from the Pro_CUC2:GUS reporter vector (Nikovics et al., 2006) kindly provided by Dr Patrick Laufs (Institut National de la Recherche Agronomique, Versailles, France), and ligated using T4 DNA ligase (Fermentas). The ligation product, thus lacking the motif, was amplified using primers with attB adaptors and inserted into the Gateway vector pGWB3 (Nakagawa et al., 2007) using the Gateway® technology according to the manufacturer's protocol (Invitrogen). The resulting vector was introduced into Agrobacterium tumefaciens, strain C58:C1 through freeze–thawing. Arabidopsis thaliana plants, ecotype Colombia, were transformed using the floral dip method (Bechtold et al., 1993; Desfeux et al., 2000). Transformed plants, selected on kanamycin-containing plates, were analysed histochemically for β-glucuronidase (GUS) activity. Samples were fixed in ice-cold 90 % acetone for 30 min and subsequently vacuum infiltrated with GUS buffer (50 mm NaH₂PO₄, 0·25 % Triton X-100, 1 mm Fe^+III/Fe^+IICN, 1 mm Na₂EDTA) containing 1 mm 5-bromo-4-chloro-3-indole β-d-glucuronic acid in the dark at 37 °C for 20 h. Pigments were removed by treatment with 70 % ethanol for at least 2 h.

In order to complement the A. thaliana cuc1cuc2 double mutant, the A. thaliana CUC2 promoter (proAtCUC2) was amplified from the Pro_CUC2:GUS reporter vector (Nikovics et al., 2006) and full-length cDNA of PaNAC01 was amplified from the pCR^®4-TOPO^® cloning vector. The products were ligated using T4 DNA ligase (Fermentas), and then amplified using primers with attB adaptors. The amplified fragment was inserted into the empty Gateway vector pGWB1 (Nakagawa et al., 2007). The resulting vector was electroporated into A. tumefaciens, strain C58:2260. Arabidopsis thaliana plants were transformed using the floral dip method. cuc1cuc2 seeds were obtained from The European Stock Centre (NASC). As plants homozygous for both mutations are seedling lethal, cuc1-1/cuc1-1 cuc2/ + plants had to be used for the transformation. Transformed plants (T₁) were selected on kanamycin-containing plates and scored for phenotype frequencies. The cuc2 mutation was detected by PCR as described by Takada et al. (2001). Three independently rescued lines and one transformed wild-type control line were taken further to the T₂ generation (also selected on kanamycin) for assessment of rescued phenotypes. Kanamycin selection did not interfere with the separation of cotyledons, as shown with the transformed control. Transgenes were detected by PCR (data not shown).

All primers used for the vector constructions are presented in Supplementary Data Table S2.

RESULTS

Phylogenetic analysis of the NAC gene family

Using the NAC domain of distinct A. thaliana NAC proteins as query, a blast search against P. glauca sequences in GenBank identified 16 P. glauca mRNA clones with full-length NAC domains. The clones were subsequently annotated as PgNAC01–PgNAC16 (Supplementary Data Table S1). Phylogenetic analyses of the NAC gene family have been carried out before (Ooka et al., 2003; Fang et al., 2008; Shen et al., 2009; Jensen et al., 2010). Shen et al. (2009) presented a thorough analysis covering 11 plant species, encompassing dicots, monocots and mosses, although no gymnosperms were included. To put the P. glauca NAC sequences into a phylogenetic context, and aiming at the identification of CUC orthologues in spruce, the P. glauca mRNA clones were compared to a sub-set of NAC domains in genes from A. thaliana, M. truncatula and P. patens (Supplementary Data Table S1), representing all clades (a–h) that had been identified by Shen et al. (2009). Our analysis showed support for most of these clades, except for clades a and h (Fig. 1). However, four P. glauca sequences clustered with PpNAC09, which according to Shen et al. (2009) belongs to clade a, and one P. glauca clone was supported to be an orthologue to the stress-related gene ATAF1 (Lu et al., 2007), also belonging to clade a in the analysis carried out by Shen et al. (2009). Six P. glauca sequences clustered with clade b together with the Turnip crinkle virus resistance-associated gene TIP (Ren et al., 2000). Clade c consisted of genes such as NAC SECONDARY WALL THICKENING PROMOTING FACTOR2 (NST2) and VASCULAR-RELATED NAC-DOMAIN (VND5) (Kubo et al., 2005; Mitsuda et al., 2005) as well as two P. glauca sequences. The CUC genes clustered with clade d (from hereon called the CUC clade) together with two P. glauca sequences. One P. glauca clone clustered with clade e, which also contained the root cap regulating gene FEZ (Willemsen et al., 2008). Clade f and g did not harbour any sequences from P. glauca.

Fig. 1.

Bayesian estimate of phylogenetic relationships among Arabidopsis thaliana (green), Medicago truncatula (blue), Physcomitrella patens (pink) and Picea glauca (bold, purple) NAC genes. Numbers above the branches indicate Bayesian posterior probability values and numbers below the branches (italics) represent percentage parsimony bootstrap support. Major clades previously identified by Shen et al. (2009) are indicated (a–h).

Identification and characterization of CUC orthologues in P. abies

Two P. abies cDNAs, orthologous to the P. glauca sequences that clustered with the CUC clade, were cloned by RT–PCR. One clone, showing 99 % sequence identity to PgNAC01, was denoted PaNAC01 (GenBank accession no. HM638414). The cDNA insert was 1663 bp and covered the coding region plus 193 bp of the 5′-UTR and 108 bp of the 3′-UTR. The deduced amino acid sequence comprised 454 amino acids and harboured a 153 amino acid NAC domain in the N-terminal region. Another cDNA fragment shared 98 % sequence identity with PgNAC02, and was therefore denoted PaNAC02 (GenBank, accession no. HM638415). The cDNA was 1005 bp and covered the coding region plus 9 bp of the 5'-UTR and 30 bp of the 3'-UTR. The deduced amino acid sequence comprised 322 amino acids and harboured a 153 amino acid NAC domain in the N-terminal region. Genome walking identified a further 2560 bp upstream sequence of PaNAC01 and 1247 bp upstream sequence of PaNAC02.

In order to relate the two PaNAC genes to other sequences within the CUC clade, a motif-based sequence analysis of the C-terminal domains of putative amino acid sequences was performed. In addition to three previously described L, V and W motifs (Takada et al., 2001; Taoka et al. 2004), one new motif was found to be conserved between angiosperms and gymnosperms (Fig. 2A). This motif was named the K-motif, since two lysines (K) were found in the beginning of the motif in all but one sequence. Based on the composition of the motifs, the analysed sequences were divided into four groups (Fig. 2A). The sequences in group I harboured all four motifs. These were CUC2, NAM and most of their orthologues in other dicots, but also ANAC100 and ANAC79/80, which have not yet been characterized. The sequences in group II lacked the K-motif. Some of these sequences, such as CUC1 and the monocot sequences ZmNAM1 and ZmNAM2, as well as the P. abies sequence PaNAC01 were highly similar to the sequences in group I. In addition, NAC1, which is involved in auxin signalling and lateral root development (Xie et al., 2000), ORE, which regulates age-dependent programmed cell death in leaves (Kim et al., 2009), as well as not yet characterized A. thaliana proteins such as ANAC059, and AT3G12977 joined this group. All 21 sequences in groups I and II, including PaNAC01, harboured the V-motif, which has also been assigned as a microRNA (miR164) recognition site (Rhoades et al., 2002). Six of these 21 genes have been shown to be post-transcriptionally regulated by miR164 (Sieber et al., 2007) and, based on sequence similarity, it seems likely that other members of groups I and II are also recognized by miR164 (Fig. 2B). PaNAC02 formed group III together with seven sequences from M. truncatula and StCUC3. All these sequences harboured the K-motif but lacked the V-motif and at least one of the other two motifs. Group IV consisted of sequences lacking both the K-motif and the V-motif. Here, most CUC3 sequences clustered together with four uncharacterized proteins from A. thaliana. The phylogenetic analysis together with the motif analysis demonstrate that sequence elements of functional importance in CUC1 and CUC2 from A. thaliana are conserved in PaNAC01, and that corresponding sequence elements have diverged or are lost in PaNAC02. This suggests that PaNAC01 might have a function similar to that of CUC1 and CUC2, while the function of PaNAC02 might be more diverse.

Fig. 2.

Motif analysis of sequences belonging to the CUC clade. The analysis is based on sequences from Medicago truncatula, Arabidopsis thaliana and Picea abies, as well as previously characterized CUC/NAM sequences. (A) The motif discovery tool, MEME, identified all three motifs (V, L and W) previously defined by Taoka et al. (2004) as well as a new motif (K) to be conserved between angiosperms and gymnosperms. Based on the composition of the motifs, the sequences were divided into four groups (I–IV). (B) The nucleotide sequences of all V-motifs were manually analysed for identifying mismatch with miR164 from Arabidopsis. Lower case letters indicate divergence to the sequence of Arabidopsis miR164. Sequences already known to be regulated by miR164 (Sieber et al., 2007) are marked with x.

The upstream sequences of PaNAC01 and PaNAC02 were compared with the 3000 bp upstream sequence of all A. thaliana genes within the CUC clade. Interestingly, a 9 bp motif positioned 300–1000 bp upstream of ATG was completely conserved between all CUC genes in A. thaliana as well as PaNAC01 and PaNAC02 (Fig. 3), but was absent in the upstream region of all other A. thaliana genes within the CUC clade. Further analysis revealed that altogether 21 bp (except for one nucleotide) around this motif was conserved between PaNAC01 and CUC2, while 18 of these nucleotides were also found in the upstream region of PaNAC02. To test the relevance of this motif for gene expression, the complete 21 bp motif was deleted from the CUC2 promoter, which was subsequently fused to the GUS reporter, and introduced into A. thaliana. In transformed seedlings harbouring the intact promoter–GUS construct, GUS expression was detected in the boundaries between the cotyledons, the first true leaves, and the SAM (Fig. 4A). In contrast, the GUS expression became extended into the leaves and petioles in transformed seedlings harbouring a promoter–GUS construct with the 21 bp motif deleted (Fig. 4B). This suggests that the conserved 21 bp promoter motif is important for restricting the expression of the CUC genes.

Fig. 3.

Schematic comparison of the promoter sequences of PaNAC01 and PaNAC02 from Picea abies with those of CUC1, CUC2 and CUC3 from Arabidopsis thaliana. All sequences harbour the 21 bp conserved promoter element located 300-1000 bp upstream of the translation start site (TSS).

Fig. 4.

CUC2–GUS expression in Arabidopsis thaliana seedlings. (A) GUS staining of Arabidopsis seedlings harbouring an intact 3 kb CUC2 promoter fused to the GUS reporter gene. Note the strong staining in the boundaries between leaves, cotyledons and SAM. (B) GUS staining of Arabidopsis seedlings harbouring the CUC2 promoter construct with a deletion of the conserved 21 bp motif shown in Fig. 3. Note the more widespread GUS staining in the first true leaves.

Expression of PaNAC01 coincides with the differentiation of the shoot apical meristem and formation of separated cotyledons

We have previously shown that the establishment of the SAM and differentiation of separated cotyledons is dependent on PAT (Larsson et al., 2008). To investigate the expression of PaNAC01 and PaNAC02 during SAM and cotyledon differentiation, we compared their expression level under normal development with that in embryos in which PAT was blocked. Both control somatic embryos and somatic embryos treated with NPA were divided into eight consecutive developmental stages (Fig. 5). Stage 1 (Fig. 5A) represented proliferating PEMs, and stage 8 (Fig. 5H) represented fully mature cotyledonary embryos. Early embryos started to differentiate at stage 2 (Fig. 5B), and developed further at stages 3 and 4 (Fig. 5C, D). At stage 5 (Fig. 5E), the SAM primordium became visible, and cotyledon primordia could be seen as small protuberances encircling the SAM. During stages 6–8 (Fig. 5F–H), the embryos continued to increase in size, the cotyledons expanded further, and finally fully spread out at stage 8 (Fig. 5H). When the developing embryos were treated with NPA, the development was disturbed as described previously (Larsson et al., 2008, 2012). Briefly, NPA-treated embryos developed more slowly. However, no distinct morphological differences between control embryos and NPA-treated embryos could be seen before the cotyledon primordia were clearly visible at stage 5 (cf. Fig. 5E, I). The small cotyledon protuberances that could be seen in control embryos were instead represented by a ring-shaped structure in NPA-treated embryos. As the embryos matured it became evident that the cotyledons were fused, giving the apex a doughnut appearance when observed from above. At stage 8, when the cotyledons of control embryos burst, the doughnut thinned out at the edges and there was a deep cavity in the centre (Fig. 5L).

A low accumulation of PaNAC01 mRNA was detected in proliferating PEMs (Fig. 6A). The relative expression of PaNAC01 increased dramatically as early embryos started to differentiate, and remained at a steady level until stage 6, when the separated cotyledons had emerged. NPA treatment of embryos led to a statistically significant lower increase in PaNAC01 expression during stages 2–6. At stages 7 and 8, when the cotyledons were clearly visible (separated or fused), there was no significant difference in the accumulation of PaNAC01 mRNA between NPA-treated and control embryos.

Fig. 6.

Quantitative real-time PCR analysis of the expression of (A) PaNAC01 and (B) PaNAC02 during development of control and NPA-treated somatic embryos of Picea abies, as indicated in the key. Definitions of the developmental stages 1–8 are given in Fig. 5. The embryos from stage 3 and later were harvested individually. Expression values are relative to the expression of each gene at stage 1, and normalized against three reference genes. The presented expression levels are mean values of three biological replicates analysed in triplicate. Error bars indicate the SEM of biological replicates. Asterisks indicate a significant difference in the expression level between control embryos at stage 1, and later stages (*P < 0·05; **P < 0·01). † indicates a significant difference (P < 0·01) in expression levels between control and NPA-treated embryos at the developmental stage indicated.

The relative expression level of PaNAC02 was high in PEMs and increased slightly as early embryos started to differentiate (Fig. 6B). However, from stage 4, when the embryos developed further and matured, the relative expression of PaNAC02 decreased and was about 100 times lower in mature embryos compared with in PEMs. There was no statistically significant difference between the expression levels of PaNAC02 in NPA-treated embryos compared with control embryos.

PaNAC01 expression can complement the A. thaliana cuc1cuc2 double mutant

Since PaNAC01 harbours conserved motifs previously defined in CUC proteins, and its expression was dependent on PAT (similar to that of CUC1 and CUC2), it was of interest also to analyse a functional similarity between PaNAC01 and CUC1/CUC2. Thus, the PaNAC01 coding region was fused to 3 kb of the AtCUC2 promoter, and transformed into A. thaliana cuc1cuc2 double mutants. As plants homozygous for both mutations are seedling lethal (Aida et al., 1997), cuc1/cuc1 cuc2/ + plants had to be used as host line for the transgene.

In the self-fertilized progeny of untransformed plants, about 70 % showed a wild-type phenotype (Fig. 7A), 28 % had cup-shaped cotyledons (Fig. 7B) and about 2 % of the seedlings had cotyledons that were fused along one of the edges, leading to a heart-shaped phenotype (Fig. 7C), well in accordance with findings by Aida et al. (1997).

Fig. 7.

Phenotypes of 5-day-old Arabidopsis thaliana seedlings from self-fertilized cuc1/cuc1 cuc2/ + plants (A–C) and self-fertilized cuc1/cuc1 cuc2/ + plants transformed with the pAtCUC2::PaNAC01 construct (D). (A) Wild-type, (B) cup-shaped cotyledon phenotype, (C) heart-shaped cotyledon phenotype and (D) completely rescued mutant seedling. Scale bars = 1 mm.

The cuc2 genotype of T₁ seedlings transformed with the proAtCUC2::PaNAC01 construct was verified by PCR analysis, and three independent lines with a homozygous cuc1cuc2 genotype were taken to the T₂ generation. In contrast to the self-fertilized M₂ progeny of host plants, these T₂ seedlings were all viable and did not display a cup-shaped cotyledon phenotype (Table 1). Rather, seedlings with either a partially fused cotyledon or a wild-type phenotype were predominant. No wild-type plants transformed with the construct showed a partially fused cotyledon phenotype (Table 1), excluding the possibility that the increase in this category among the complemented host lines was an artefact from the transformation process. The frequency of T₂ seedlings with a wild-type phenotype ranged between 28 % (line 386) and 91 % (line 396) (Table 1, Fig. 7D). All seedlings in the progeny of transformed plants were homozygous for the cuc1cuc2 mutations, and would thus be expected to show a cup-shaped phenotype as described by Aida et al. (1997). Therefore, the results collectively suggest an efficient complementation of the homozygous cuc1cuc2 mutant by the proAtCUC2::PaNAC01 construct.

Table 1.

Frequency of seedling phenotypes of A. thaliana cuc1/cuc1 cuc2/ + mutants transformed with a proAtCUC2::PaNAC01 construct

Genotype	Total no. of seedlings	Wild-type	Partially fused cotyledons	Cup-shaped cotyledons
18	59	85	15	0
386	58	28	72	0
396	224	91	9	0
Transformed wild-type	127	100	0	0
Untransformed cuc1/cuc1 cuc2/ +	266	70	2	28

The seedling phenotype of three independent T₂ lines from a transformation of A. thaliana cuc1/cuc1 cuc2/ + mutants with a pAtCUC2::PaNAC01 construct was determined at 5 d from sowing on selective medium. The frequencies of seedlings having cotyledons scored as either wild-type, partially fused or cup-shaped were calculated. Transformed wild-type plants, and the progeny of cuc1/cuc1 cuc2/ + untransformed mutants, are shown for comparison. The general appearance of representative seedlings is depicted in Fig. 7.

With regard to seedlings with partially fused cotyledons, a majority of the seedlings had cotyledons that were fused along one of the edges, breaking the bilateral symmetry and creating the heart-shaped phenotype shown in Fig. 7C, although the degree of fusion varied. Most of the seedlings with partially fused cotyledons formed a shoot and could germinate. However, the cotyledon fusions in line 386 were more severe than in the other two lines, and fewer of the seedlings developed a shoot. The adult plants of this line also gave fewer seeds. Some of the abnormal seedlings retained a bilateral symmetry with two cotyledons, although the petioles of the cotyledons were fused, forming a hollow cylinder with no visible SAM.

DISCUSSION

Phylogenetic diversity of NAC domain-containing proteins in Picea

Since NAM in Petunia was the first NAC gene to be characterized (Souer et al., 1996), several genes encoding NAC domain proteins in various plant species have been annotated NAM, merely based on the similarity to the NAC domain. To avoid misleading interpretations, we suggest that all new NAC domain-encoding genes should be annotated as nnNACxx, where nn is the species initials and xx is a number, until further functional characterization has been made. Thus, all P. glauca and P. abies sequences in the present study were annotated accordingly.

Phylogenetic analysis of NAC domain-encoding genes encompassing sequences from angiosperms, lycophytes and bryophytes has been carried out before (Shen et al., 2009). Here, we extended the analysis also to include sequences from a gymnosperm. The P. glauca sequences used in this study were retrieved from GenBank (30 May 2011) by searching with different NAC genes from A. thaliana representing the sequence variation in the entire NAC gene family. The same set of 16 mRNA clones was obtained repeatedly, independently of the query sequence used, which indicates that these 16 sequences constitute all NAC genes in P. glauca presently available in the databases. The established genome sequences from the green algae Ostreococcus lucimarinus and O. tauri were also analysed for the presence of NAC genes, although no such genes were found. It is therefore likely that the ancestral NAC gene or genes arose at the time of land plant colonization of Earth, dated to approx. 480 million years ago (Smith et al., 2010). However, it cannot be excluded that an ancestral NAC gene has evolved, and today shows such a sequence divergence that it cannot be recognized as a NAC gene with the search settings used in this study.

All NAC sequences from P. glauca could be deployed to different clades (Fig. 1), previously characterized by Shen et al. (2009). However, clades f and g did not harbour any P. glauca genes and there was no support for clade h. Since one P. patens sequence clustered with clade g, we assume that there exists at least one, as yet unidentified, P. glauca gene in this clade. On the other hand, the lack of both P. glauca and P. patens sequences in clades f and h suggests that these clades might be more recent than the other six clades.

It has been shown that NAC-encoding genes that are evolutionarily closely related often exert similar functions (Shen et al., 2009; Jensen et al., 2010). In clade c, PgNAC14 is an orthologue to VND5, which is one of seven VND genes believed to be important for xylem vessel element formation (Kubo et al., 2005). In addition, PgNAC15 clusters with clade c, which also harbours NST2, a gene involved in secondary wall thickening (Mitsuda et al., 2005). These results suggest that PgNAC14 and PgNAC15 are interesting candidates for studying wood formation in economically important conifers. The phylogenetic analysis also indicated that PgNAC07 might have similar functions to the drought-inducible ATAF1 (Lu et al., 2007), which could also be of interest for future studies. PgNAC01 and PgNAC02 cluster with the CUC genes. Hence, provided that the phylogenetic relationship is indicative of functional similarity, PgNAC01 and PgNAC02 are likely to be involved in the establishment of cotyledonary boundaries and embryonal SAM.

The two cDNA sequences isolated from P. abies somatic embryos showed high similarity to PgNAC01 and PgNAC02. PaNAC01 has conserved motifs that suggests a function similar to that of CUC1 and CUC2, while PaNAC02 is more reminiscent of other NAC sequences within the clade. An important difference between PaNAC01 and PaNAC02 is the presence of the W-motif in PaNAC01, but its absence in PaNAC02 (Fig. 2A). Similarly, CUC1 and CUC2, but not CUC3, carry the W-motif. This motif seems to be important for the function of at least the CUC1 protein, since a deletion of five amino acids around this motif in combination with the cuc2 mutation results in cup-shaped cotyledons (Takada et al., 2001). The new K-motif presented in this study is spread among the genes of the CUC clade. It is specific for the CUC2 orthologues, while it has been lost in the CUC1 lineage. Interestingly, the presence of the conserved 21 bp promoter motif in PaNAC01, PaNAC02 and in all three A. thaliana CUC genes (Fig. 3) suggests a common regulation for all five genes. The increased GUS expression in A. thaliana plants transformed with a promoter–GUS fusion where the 21 bp motif has been deleted (Fig. 4) indicates that the motif mediates a negative regulation.

CUC1 and CUC2 but not CUC3 are post-transcriptionally downregulated by miR164 cleavage (Laufs et al., 2004; Mallory et al., 2004). Interestingly, PaNAC01 harbours the miR164 recognition site, and miR164 has been identified in a microRNA screen of P. abies (U. Lagercrantz, pers. commun.). However, it remains to be investigated whether this microRNA cleaves the PaNAC01 mRNA. The miR164 recognition site could not be found in any NAC gene in P. patens or Selaginella moellendorffi, and large-scale sequencing of small RNAs has not identified miR164 in these species (Axtell et al., 2007). This indicates that miR164-mediated post-transcriptional regulation of the CUC genes was established after the origin of seed plants but before the separation between angiosperms and gymnosperms.

Expression of PaNAC01 coincides with the differentiation of the shoot apical meristem and formation of separated cotyledons

We have previously shown that PAT is crucial for the correct patterning of P. abies somatic embryos (Larsson et al., 2008). Mature embryos that have been treated with the PAT inhibitor NPA develop an abnormal apical part with fused cotyledons and aborted SAM (Larsson et al., 2008, 2012; Fig. 5). The aberrant morphologies of NPA-treated P. abies embryos are comparable with several auxin response and transport mutants in A. thaliana such as mp and pin1, which also show abnormal expression patterns of CUC1 and CUC2 (Aida et al., 2002).

The relative expression of PaNAC01 increases as early embryos differentiate, and remains at a steady level until the separated cotyledons are clearly visible (Fig. 6A). However, the upregulation of PaNAC01 is reduced in embryos that form fused cotyledons and lack a functional SAM after being treated with NPA. In A. thaliana, the expression of CUC2 is reduced, while CUC1 has a higher and broader expression when PAT is hampered (Aida et al., 2002). With regards to this, the auxin-induced regulation of PaNAC01 seems to be most similar to the regulation of CUC2. In accordance with this, Vialette-Guiraud et al. (2011) suggested that the CUC1 lineage within the Brassicales has evolved more rapidly through positive selection than the CUC2 lineage. In A. thaliana, the cotyledons initiate at sites of auxin response maxima, presumably generated by the localization of PIN1 in the epidermal cell layer (Benková et al., 2003). The CUC genes are first expressed in a region across the apical part of the early globular embryo, but as the embryo attains a bilateral symmetry, their expression is restricted to the marginal regions (Bowman and Floyd, 2008). Although we have not been able to localize where PaNAC01 is expressed, our results show that the temporal expression coincides with the differentiation of separated cotyledons and formation of the SAM.

The relative expression of PaNAC02 is high in PEMs and early differentiating embryos but decreases as the embryos develop and mature. Interestingly, the downregulation occurs at about the same time as the cotyledons are delineated. However, the lack of previously described motifs and the similarity to as yet uncharacterized genes make it difficult to assign a function to PaNAC02. The expression of PaNAC02 is not dependent on PAT. However, this does not rule out that PaNAC02 can act redundantly with PaNAC01, just like CUC3 acts redundantly with CUC1 and CUC2, despite its lack of functional motifs (Vroemen et al., 2003).

PaNAC01 can induce cotyledon separation

Similar to what has been shown for CUC1 and CUC2, PaNAC01 harbours all motifs that have been proven to be functionally important, and its expression is dependent on PAT. Hence, it was interesting to investigate if PaNAC01 functionally could complement the A. thaliana cuc1cuc2 double mutant. Indeed, it was possible to restore the wild-type phenotype of these plants by expressing PaNAC01 from the A. thaliana CUC2 promoter. A fraction of the mutant seedlings expressing PaNAC01 showed a heart-shaped phenotype. This phenotype has been considered as a weaker phenotype of cuc (Aida et al., 1997), and is particularly abundant in offspring of plants carrying multiple heterozygote mutations in two or more of the CUC genes (Aida et al., 1997; Vroemen et al., 2003). Expressing a conifer gene in an angiosperm does not necessarily provide functional information relevant for conifers. However, a functional substitution is indicative of an evolutionary conservation of a regulatory pathway.

We have shown here that the temporal expression of PaNAC01, and its regulation by PAT during somatic embryo development, is similar to that of CUC1 and CUC2 in A. thaliana. Our phylogenetic reconstruction and motif analysis, together with the fact that PaNAC01 can functionally substitute for CUC2, indicate that PaNAC01 and CUC1 and CUC2 act in an evolutionarily conserved pathway. Central parts of the regulatory network for SAM and cotyledon formation can thus be regarded as conserved between angiosperms and gymnosperms, indicative of a common ancestral function present already 300 million years ago.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Table S1: gene accession numbers, names and references for sequences used in phylogenetic analysis and motif analysis. Table S2: primers used for amplification of cDNA and promoter sequences.