•Legume roots develop two types of lateral organs, lateral roots and nodules. Nodules develop as a result of a symbiotic interaction with rhizobia and provide a niche for the bacteria to fix atmospheric nitrogen for the plant.
•The Arabidopsis NAC1 transcription factor is involved in lateral root formation, and is regulated post-transcriptionally by miRNA164 and by SINAT5-dependent ubiquitination. We analyzed in Medicago truncatula the role of the closest NAC1 homolog in lateral root formation and in nodulation.
•MtNAC1 shows a different expression pattern in response to auxin than its Arabidopsis homolog and no changes in lateral root number or nodulation were observed in plants affected in MtNAC1 expression. In addition, no interaction was found with SINA E3 ligases, suggesting that post-translational regulation of MtNAC1 does not occur in M. truncatula. Similar to what was found in Arabidopsis, a conserved miR164 target site was retrieved in MtNAC1, which reduced protein accumulation of a GFP-miR164 sensor. Furthermore, miR164 and MtNAC1 show an overlapping expression pattern in symbiotic nodules, and overexpression of this miRNA led to a reduction in nodule number.
•This work suggests that regulatory pathways controlling a conserved transcription factor are complex and divergent between M. truncatula and Arabidopsis.
Legume plants form two types of secondary root organs, lateral roots and nodules. Nodules develop as a result of a symbiotic interaction with rhizobia and provide the optimal environment for atmospheric nitrogen fixation. Nodule development is initiated by successful perception of the bacterial signal molecules, the Nod factors (NFs), through specific receptors in the epidermal root cells (Kouchi et al., 2010). Nodulation consist of two interlinked processes: bacterial infection and nodule primordium formation. Within most legumes, the bacteria enter the root hairs via infection threads (ITs) that grow towards deeper cortical cell layers in which a nodule primordium is formed. Functional nodules mature when the ITs reach the nodule primordium and bacteria are delivered to the primordium cells and differentiate into bacteroids to fix nitrogen. In the model legume Medicago truncatula, indeterminate nodules consist of different zones (Crespi & Frugier, 2008; Oldroyd & Downie, 2008): a persistent apical meristem, an infection zone characterized by IT growth, where bacterial uptake takes place, a fixation zone in which nitrogen fixation occurs, and a senescence zone, where bacteria and plant cells decay.
Inspired by the similarities between lateral root and nodule organogenesis, we selected the NAC1 gene, associated with lateral root formation in Arabidopsis thaliana for comparative studies in M. truncatula. NAC1 belongs to the NAC family (of which the founding members are NO APICAL MERISTEM (NAM), the Arabidopsis transcription factor 1 (ATAF1) and CUP-SHAPED COTYLEDON2 (CUC2)), that constitutes one of the largest groups of plant-specific transcription factors. NAC transcription factors are characterized by a highly conserved NAM DNA-binding domain in the N-terminal region (Olsen et al., 2005). With > 100 genes in the Arabidopsis genome and their occurrence in a wide range of land plants, including dicots and monocots (Ooka et al., 2003; Olsen et al., 2005), NAC transcription factors are involved in various plant developmental processes as well as in responses to diverse biotic and abiotic stresses (Olsen et al., 2005; Jensen et al., 2010). In Arabidopsis, auxin induces the expression of NAC1, predominantly in the root tip and lateral root initials and at a reduced level in leaves and stems (Xie et al., 2000; He et al., 2005; Wang et al., 2006). Plants ectopically expressing NAC1 have larger leaves, thicker stems and more lateral roots than the wild type. Conversely, knockdown of NAC1 results in reduced lateral root formation (Xie et al., 2000).
NAC1 transcript and protein levels are tightly regulated in time and space in Arabidopsis. First, a transcriptional control involving auxin-mediated repressor degradation induces NAC1 expression. Second, a dual post-transcriptional survey system controls the NAC1 protein accumulation, ensuring rapid signal suppression after auxin induction. In the nucleus, NAC1 is specifically targeted for proteasomal degradation through a Seven in Absentia homolog of A. thaliana 5 (SINAT5) E3 ligase, thereby attenuating NAC1 downstream responses (Xie et al., 2002). SINA E3 ligases were first found in Drosophila melanogaster and contain an N-terminally located RING finger domain, followed by the conserved SINA domain that is involved in substrate binding and dimerization (Hu & Fearon, 1999). Ectopic expression of AtSINAT5 resulted in a reduced number of lateral roots, whereas the number increased in transgenic plants expressing a dominant negative version of AtSINAT5. In M. truncatula, a small family of six SINA genes was identified and functional analysis revealed that SINA E3 ligases are implicated in root development and nodulation (Den Herder et al., 2008).
Here, MtNAC1, the closest Arabidopsis NAC1 homolog of the available Medicago genome, was isolated and its involvement in root and symbiotic nodule development was investigated. We also determined if post-transcriptional regulations involving SINAT-dependent ubiquitination and miR164 were conserved between MtNAC1 and the Arabidopsis NAC1 gene.
Materials and Methods
Plant material, bacterial strains, and growth conditions
Medicago truncatula Jemalong A17 was grown and inoculated as described in Mergaert et al. (2003). Root tips, shoot apical meristems, first leaves and cotyledons were harvested from in vitro grown 5-d-old plants. Stems, leaves and flowers were sampled from 1-month-old plants grown in nitrogen-rich soils. For auxin induction series, in vitro-grown 3-d-old plants were transferred to new plates containing 2 μM α-naphthalene acetic acid (NAA) (Duchefa Biochemie, Haarlem, the Netherlands) and, subsequently, the roots were harvested. In vitro growth was done in square Petri dishes on SOLi medium-containing Kalys agar (Kalys, Bernin, France) supplemented with 1 mM NH4NO3 (as described at http://www.isv.cnrs-gif.fr/embo01/index.html) Plants were grown at 25°C in a 16-h photoperiod with a light intensity of 70 μmol s−1 m−2.
Sinorhizobium meliloti 1021, 1021 pHC60-GFP (Cheng & Walker, 1998) and 1021 pQE81-dsRedT3 (Bevis & Glick, 2002) were grown at 28°C in yeast extract broth (Vervliet et al., 1975). The presence of green fluorescent protein (GFP) or Discosoma sp. red fluorescent protein (dsRED) in rhizobia and roots was screened with a stereomicroscope MZFLII (Leica, Wetzlar, Germany) equipped with a blue-light source and a Leica GFP Plus filter set (λex = 480/40; λem = 510 nm). To control the expression of MtNAC1 during nodulation by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) analysis, plants were infected with S. meliloti 1021 (pHC60-GFP) to visualize the bacteria by fluorescence stereomicroscopy. Similarly, at each time point, specific nodulation stages were harvested as indicated. The control sample was taken simultaneously with the sample 2 d post inoculation (dpi) and similar developmental zones of the root were cut from both samples. 35S:MtNAC1 and MtNAC1 RNAi transgenic roots were inoculated with Sm1021 pQE81-dsRedT3 to enable bacterial visualization.
The nac1-1 mutant was obtained via TILLING of an ethane methyl sulfonate (EMS)-mutagenized M. truncatula population (Le Signor et al., 2009) and carried a G-to-A transition at nucleotide position 449 introducing a ‘stop’ codon (Fig. 1a). The progeny of backcrossed lines was grown in germination pouches (Mega International, West St. Paul, MN, USA) supplemented with SOLi medium to measure root length and number of lateral roots. Nodulation in pouches was performed as previously described (Mortier et al., 2010). To genotype the backcrossed population, genomic DNA was prepared from each plant and subjected to a PCR amplifying a 400-bp MtNAC1 fragment covering the possible mutation with the primers 5′-CAGGGCACTTGCTCATATTC-3′ and 5′-TGAGGTGGTGGGATTTGTAA-3′.
Constructs and Agrobacterium rhizogenes-induced composite plants
The full-length open reading frame (ORF) of MtNAC1 was obtained by reverse transcription (RT)-PCR on nodule RNAs using the primers 5′-ATGAGCAACATAAGCATGG-3′ and 5′-TTAGAAATTGTTCCACATGTGGGGCATAC-3′, cloned into the Gateway vector pDONR221 (Invitrogen), and then transferred into the pK7WG2D destination vector (Karimi et al., 2002) to generate a 35S Cauliflower Mosaic Virus (CaMV): MtNAC1 construct. The 3′-untranslated region (UTR) fragment used for RNAi-mediated silencing was similarly obtained by RT-PCR on nodule RNAs with the primers 5′-AAAGGGTCATCACCAAGCT-3′ and 5′-GTGTTCGCCTTTTCATCG-3′, cloned into the Gateway vector pDONR221, and transferred into the pK7GWIWG2D destination vector (Karimi et al., 2002) to generate a RNAi hairpin construct. The resulting constructs were introduced into A. rhizogenes Arqua1 (Quandt et al., 1993) to obtain composite plants (i.e. bearing wild-type aerial parts; Boisson-Dernier et al., 2001) and inoculated as previously described (Mortier et al., 2010).
The MtMIR164c locus (miRBase v.16), located on the bacterial artificial clone BAC AC144541.12 (NCBI database), was amplified by PCR with Pfu (Promega) from A17 M. truncatula genomic DNA with the forward (Fdw) and reverse (Rev) primers: MtMIR164c-Fwd 5′-TTTCAAGAGCTTATTATTTATTG-3′ and MtMIR164c-Rev 5′-ATGTTGGAAAGTGGGTGTCC-3′ for the pre-MtMIR164c precursor (1450 bp), or MtMIR164c-HP-Fwd 5′-CCCTGTTTTTACTCCAAGCAATACG-3′ and MtMIR164c-HP-Rev 5′-TGCAACCCTCACCCTGAAG-3′ for the MIR164c hairpin region (475 bp). The PCR products were cloned into the pGreenII0029 binary vector (http://www.pgreen.ac.uk) allowing expression under a 2 × 35S-CaMV promoter (Laufs et al., 2004). The resulting constructs were introduced into A. rhizogenes Arqua1 together with the pSoup helper vector (http://www.pgreen.ac.uk) and used for M. truncatula root transformation. Composite plants were inoculated as described previously (Gonzalez-Rizzo et al., 2006). Biological experiments were done in triplicate based on at least 10 independent roots per construct.
For the miR164-GFP sensor constructs, a resistant version of the miR164 MtNAC1-binding site was generated by introducing point mutations with the Quickchange II Site-directed Mutagenesis kit (Agilent, Santa Clara, CA, USA) with the primer 5′-GAAAACATCGAAAAACACGGTACTTGTTCATATTC-3′ (underlined positions are the nucleotides that have been changed). The MtNAC1 wild-type and mutagenized miR164-binding sites were amplified with the primers 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTAATTTGACCAAGCTCAA-3′ and 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCCTACTTGCATTGTTATTG-3′ and introduced downstream of the enhanced (e)GFP reporter into the pK7WGF2 vector (Karimi et al., 2002) with the Gateway technology (Invitrogen). As another control, a pK7WG2 vector in which the ccdB gene had been replaced by the eGFP was used. Composite plants were generated as previously described and the GFP accumulation was analyzed after 2 wk of growth on Fahreus medium without nitrogen (Truchet et al., 1985) in the root region located 2 cm above the apices, with a MZFL III stereomicroscope equipped with a DC200 digital camera (Leica).
Subcellular localization of GFP fusion proteins in Nicotiana benthamiana leaf epidermal cells
A 35S:MtNAC1:eGFP construct was generated based on the pDONR221 MtNAC1 plasmid and the pK7FWG2 Gateway destination vector (Karimi et al., 2002). As controls for both nuclear and cytosolic localizations, 35S:eGFP and eGFP were fused to the protein Dimerization Partner b (AtDPb) (35S:AtDPb:eGFP) (Kosugi & Ohashi, 2002); for the specific nuclear localization, an eGFP fusion with the transcription factor E2Fc promoter-binding factor c (35S:AtE2Fc:eGFP) was used (del Pozo et al., 2002). Each construct was introduced into the A. tumefaciens LBA4404 strain by electroporation. A saturating culture resuspended at an OD600 nm =0.5 in an infiltration buffer (50 mM MES, 2 mM Na2HPO4, 0.5% glucose, pH 5.6) supplemented with 100 μM acetosyringone was injected into the lower epidermis of tobacco (N. benthamiana) leaves. The GFP signals were examined with a confocal microscope (Leica) 3–5 d after infection. Transient expression was analyzed in two different transformed leaves and in at least three independent biological experiments.
Transactivation assay in yeast
A fusion of the GAL4-DNA binding domain (BD) to the ORFs of MtNAC1 and AtE2Fa was generated with Gateway recombination between the pGBT9GW (Bartel et al., 1996) and the pDONR221 vectors (Invitrogen) containing the ORFs of the genes. The constructs and the empty vector were transformed in Saccharomyces cerevisiae MaV 203 (Invitrogen) by heat shock transformation. Transformants were grown in Minimal SD Base liquid medium (0.67% yeast nitrogen base and 2% glucose with appropriate amino acids; Clontech, Mountain View, CA, USA), diluted until OD600 nm = 0.1, and plated on histidine-free SD medium with 5, 10, 20, 40, 80 or 100 mM 3-amino-1,2,4-triazole (3-AT) to increase stringency of the selection. Yeast was grown on plates for 3 d at 30°C.
Protein interaction in yeast two-hybrid assays
The full-length coding sequence of MtNAC1, a shortened ORF containing only the first 570 bp (referred to as MtNAC*), and a full-length MtSINA1 ORF (Den Herder et al., 2008) were fused to the GAL4 activation domain (AD) of the pADGAL4 vector (Agilent), whereas the six MtSINA genes described in M. truncatula (Den Herder et al., 2008) were fused to the GAL4 DNA-binding domain (BD) of the pBDGAL4 vector that is used in yeast two-hybrid analysis, according to the manufacturer’s procedure. Cotransformed yeast was grown at equal density on a medium without Trp and Leu (TL medium) to select for the presence of both plasmids and on a medium with Trp, Leu and His (TLH medium) to investigate the protein interaction. The strength of the interactions was tested by addition of 0, 5, 10, 20, 40, or 80 mM of 3-AT to this medium. Yeast growth was analyzed after 3 d at 30°C.
RNA extraction, cDNA synthesis, and real time RT-PCR analysis
The RNA samples were isolated with the RNeasy Plant Mini Kit (Qiagen) according to the manufacturer’s instructions. The samples were DNase treated and purified through NH4-acetate (5 M) precipitation. Quality was controlled and quantified with a Nanodrop ND1000 spectrophotometer (Thermo Fischer Scientific, Wilmington, DE, USA). cDNAs were synthesized with the Superscript II reverse transcriptase (Invitrogen). Real time RT-PCR was done as described (Vlieghe et al., 2005) with a iCycler iQ apparatus (Bio-Rad) and the Platinum SYBR Green Supermix-UDG (Eurogentec, Seraing, Belgium). Relative expression was calculated as the ratio of the normalized gene expression against the constitutively expressed elongation factor 1 α (MtEFlα) with the primers 5′-ACTGTGCAGTAGTACTTGGTG-3′ and 5′-AAGCTAGGAGGTATTGACAAG-3′, according to the 2−ΔΔCT method (Livak & Schmittgen, 2001). Reactions were done in triplicate and averaged. Primers used to amplify MtNAC1 were 5′-GACACAGGCTCTTCATCACTTCC-3′ and 5′-ATTGTTGTAGATTGGGTTGGTTTGG-3′, and for MtENOD40 5′-CCCTCCATTTTCCTAAACAGTTTGC-3′ and 5′-ACTTGCCGGTTTGCCATGC-3′.
To determine expression levels in RNAi and overexpressing plants, cDNAs were prepared as described above from root RNAs and real time RT-PCRs were run on a LightCycler480 apparatus (Roche, Mannheim, Germany). Primers sequences used to amplify MtNAC1 were 5′-CCTCCAAGCATGGGTAGTTG-3′ and 5′-GGGTTGGTTTGGTTTTGAGA-3′, and for MtMIR164c 5′-TGCACCACCTTCTCATTTCTC-3′ and 5′-GGTGGGAAGATCAAGTTGGA-3′.
5′ Rapid amplification of cDNA ends (RACE)-PCR experiment and in situ hybridization
The 5′ RACE-PCRs were done on capped mRNAs according to the manufacturer’s instructions (SMART Race cDNA Amplification Kit, BD Biosciences, Palo Alto, CA, USA). For the primers used, see the Supporting Information, Fig. S1.
mRNA in situ hybridization was done as described in Boualem et al. (2008). A MtNAC1 antisense probe was synthesized by PCR amplification using a region showing less than 70% of overall identity with other M. truncatula genes, based on the 5′-GTTGCTACAAAACCTCCAAGC-3′ forward primer and the 5′-TGTAATACGACTCACTATAGGGCCATATTCATCTGTATGGAATTGAGC-3′ reverse primer associated to a T7 promoter extension. Digoxigenin (DIG)-labeled probes were generated with the SP6/T7 DIG-RNA labeling kit (Roche) and a locked nucleic acid (LNA) probe complementary to the miR164c was marked with the DIG Oligonucleotide 3′-end Labeling Kit (Roche) accordingly to the manufacturer’s protocol.
Prediction of miR164 targets in M. truncatula was based on miranda (version 3.0) software and the MIRMED database (http://medicago.toulouse.inra.fr/MIRMED). Stringent criteria were used to predict targets, that is, an alignment spanning at least 18 bp with a maximum penalty score of 2. Score calculation considered 0.5 point for each G:U wobble, one point for each non-G:U mismatch, and two points for each bulged nucleotide in either RNA strand (accordingly to Jones-Rhoades & Bartel, 2004). MIR164c hairpin folding was predicted with Mfold and default parameters (http://mfold.bioinfo.rpi.edu/). The region was used to generate the ‘MIR164c hairpin’ construct.
To estimate the genotype effects on root length, the linear mixed model (random terms underlined) y = μ+ Genotype + Experiment+ ε was fitted to the data (y represents the root length; μ, the overall mean; Genotype, the fixed genotype effect; Experiment, the random experimental effects; ε, the random error). Statistical significance of genotype effects was assessed by a Wald test. For nodulation, a generalized linear mixed model of the form y = μ+ Genotype + Experiment+ ε with a Poisson distribution and a logarithmic link, was fitted to the data. Again, the statistical significance of the genotype effects was assessed by a Wald test. All analyses were done with genstat (http://www.vsni.co.uk/software/genstat/).
To estimate the effect of the genotype on the repartition of transgenic roots over different classes (nod−, nod+, fix−) and to analyze whether data could be pooled, a univariate general mixed model was first fitted to the data (class of nodules represented the dependent variable; genotype, the fixed genotype effect; and experiment, the random experimental effects). Normality and homogeneity of variances were tested with a Kolmogorov–Smirnov and a Levene test, respectively. If both parametric conditions were not achieved, a Mann-Whitney U test was done, whereas, if they did, ANOVA was applied. SPSS 17 (http://spss.en.softonic.com/) was used for the analyses.
Characterization of the MtNAC1 protein
To identify M. truncatula NAC1 homologs, blast searches were performed against the available M. truncatula genomic data and M. truncatula Gene Index, MtGI8. One tentative consensus (TC) clone, TC95634, currently TC159990 (MtGI9), was closely related to NAC1, even outside the NAC domain. In silico expression analysis using the M. truncatula gene expression atlas (Benedito et al., 2008; MtGEA, http://mtgea.noble.org/v2/; probe set Mtr.12694.1.S1_at) indicated a slightly higher expression in developing nodules than in uninoculated roots (Fig. S1). A MtNAC1 cDNA clone was isolated from M. truncatula nodules, encoded a 306-amino-acid protein and contained an N-terminal NAC domain with a putative nuclear localization signal (NLS) (Fig. 1a; NCBI accession number HQ343415). A neighbor-joining tree of all NAM domain-containing proteins present in Arabidopsis and M. truncatula was constructed (Figs 1b, S2). MtNAC1 is most closely related to NAC1 and to a second Arabidopsis gene, At3g12977.1. No other M. truncatula sequence was present in that branch. Comparison of the amino acid sequences indicated that MtNAC1 showed 73% similarity with NAC1 from Arabidopsis and a 65% similarity with At3g12977.1.
In silico analysis revealed that MtNAC1 contains a DNA-binding NAM domain and a NLS. To confirm in vivo the putative nuclear localization (Fig. 2), an eGFP fusion was generated (MtNAC1:eGFP) and expressed from the 35S-CaMV promoter into tobacco epidermal cells. DPb:eGFP and E2Fc:eGFP fusions were used as controls. The MtNAC1:eGFP fusion was detected in the nucleus of tobacco cells and infrequently associated with a slight cytoplasmic signal considered as background (Fig. 2a). As previously shown, eGFP localized all over the cytoplasm and nucleus, whereas the DPb and E2Fc controls were located in the nucleus/cytoplasm and solely in the nucleus, respectively (Fig. 2b–d) (del Pozo et al., 2002; Kosugi & Ohashi, 2002).
To investigate whether MtNAC1 acts as a transcriptional activator, the MtNAC1 ORF was fused to the GAL4-BD and introduced into S. cerevisiae. As a positive control, a yeast strain expressing the transcriptional activator E2Fa (Kosugi & Ohashi, 2002) fused to the GAL4-BD was used. Expression of the MtNAC1 construct allowed yeast to grow on selective medium (up to 40 mM 3-AT), similarly as for the E2Fa-expressing strain (Fig. 2e), indicating that MtNAC1 can function as a transcriptional activator.
MtNAC1 expression pattern
Upon rhizobium inoculation, the amount of MtNAC1 transcript increased in the root region containing emerging root hairs with ITs (Fig. 3a). The expression level increased in young developing nodules (12 or 16 dpi) and started to decrease in mature nodules (22 dpi or 28 dpi), even though expression was still higher than that of uninoculated roots (Fig. 3a). MtENOD40 was used as a control for nodulation-related gene expression (Fig. 3b). The MtNAC1 expression was also analyzed in several M. truncatula organs (Fig. 3c), revealing that it occurred mainly in root tips, leaves and cotyledons. Overall, these data are in agreement with the expression reported on the Affymetrix chip (Fig. S1).
As MtNAC1 is a potential ortholog of NAC1, we explored whether the gene is upregulated by exogenous auxin in roots. No upregulation of the transcript levels could be detected after application of 2 μM NAA, and conversely, MtNAC1 was repressed after 4 h (Fig. 3d).
These experiments show that MtNAC1 is upregulated early during nodulation, when rhizobial infection and nodule primordium development occur, and that a high transcript level is maintained in functional nitrogen-fixing nodules. In addition, unlike the Arabidopsis NAC1, no upregulation of MtNAC1 expression was detected upon auxin treatment of M. truncatula roots.
Effect of MtNAC1 overexpression, MtNAC1 RNAi, and a MtNAC1 loss-of-function mutation
To obtain insight into the function of MtNAC1 in nodulation, the effect of modulating MtNAC1 transcript levels was studied. Therefore, 35S:MtNAC1 and RNAi-MtNAC1 constructs were generated and introduced into transgenic roots to investigate their effects on nodulation. No significant difference in the average nodule numbers was seen between control (2.34 ± 0.69 nodules per plant) and 35S:MtNAC1 roots (3.21 ± 0.69 nodules per plant) (P = 0.281; for details of the statistical tests used, see the Materials and Methods section). Nodules were scored for the presence of a pink color, suggesting the occurrence of leghemoglobin and nitrogen fixation. Plants were divided into three classes depending on the nodulation phenotype: non-nodulating roots (nod−), roots carrying only white nodules (fix−), and roots with pink nodules (fix+; Fig. 4a). The distribution of these classes between control and 35S:MtNAC1 lines were comparable (P =0.443). Overexpression of MtNAC1 was tested by real time RT-PCR analysis, confirming that the ectopic expression of MtNAC1 was efficient (Fig. 4b).
The nodulation was similarly analyzed with MtNAC1 RNAi constructs. The average nodule number on RNAi (3.41 ± 0.48 nodules per plant) and control (2.63 ± 0.48 nodules per plant) roots was comparable (P = 0.290). Plants were divided in classes similar to those for the overexpression analysis. Again, no difference in the number of plants per nodulation phenotype class was found (Fig. 4c); neither did the remaining MtNAC1 transcript level and the nodulation phenotype correlate (Fig. 4d).
An EMS-mutagenized M. truncatula population was screened for MtNAC1 mutants with a TILLING approach. One allele, nac1-1, caused a premature stop codon downstream of the DNA-binding NAC domain (Fig. 1a). Progenies of two independent backcrossed lines were tested for their nodulation phenotypes. In two biological replicates, the average nodule numbers were similar independently of the NAC1 genotype (P >0.05; Fig. 5a). Nodules were also analyzed for the presence of a pink color, but no differences were observed. To assess whether the root architecture was affected in the nac1-1 mutant, we controlled root length and number of emerged lateral roots (Fig. 5b,c). Comparison of wild-type plants with nac1-1 homozygous mutants did not reveal any significant difference. In conclusion, under the conditions tested, no nodulation or root phenotype could be detected by modulating the MtNAC1 expression level by overexpression, RNAi or in the nac1-1 mutant.
Post-transcriptional and post-translational regulation of the MtNAC1 expression
In Arabidopsis, NAC1 is subjected to controlled protein degradation because of the interaction with the AtSINAT5 E3 ligase and the regulation of the transcript accumulation by a miRNA-dependent mechanism (Xie et al., 2002; Guo et al., 2005). Therefore, we investigated whether the regulation of the closest MtNAC1 homolog was affected.
Six SINA genes have been found in the available M. truncatula genomic sequence (Den Herder et al., 2008). The MtNAC1 interaction with the six known MtSINA proteins was tested in a yeast two-hybrid assay. Because of its autoactivation, of MtNAC1 was fused with the GAL4-AD. Cotransformation of the MtNAC1:GAL4-AD clone with either of the six MtSINA:GAL4-BD clones did not result in yeast growth, even on medium without 3-AT (Fig. 6). As a positive control, the interaction between the MtSINAT1:GAL4-AD and MtSINAT1:GAL4-BD fusion constructs was used (Den Herder et al., 2008) (Fig. 6). Next, analogously to the method used to show the interaction of NAC1 with SINAT5 in Arabidopsis (Xie et al., 2002), the interaction between all MtSINA proteins and a shortened protein (MtNAC*) was tested. In contrast to Arabidopsis, a strong autoactivation activity was observed when MtNAC* was combined with the GAL4-BD. In addition, no growth was seen on the selective medium after cotransformation of yeast with a MtNAC*:GAL4-AD clone and any of the MtSINA:GAL-BD proteins (Fig. 6). Thus, no interaction between MtNAC1 and any of the six available MtSINA proteins could be detected.
In Arabidopsis, miR164 modulates the NAC1 expression (Xie et al., 2002; Guo et al., 2005). To assess whether the MtNAC1 expression was also regulated by this evolutionarily conserved miRNA, we analyzed the MtNAC1-coding sequence and found a miR164-binding site (Fig. 1a). In situ hybridization revealed the spatial expression patterns of MtNAC1 and miR164 in nodules (Fig. 7a–d). The mature miR164 (Fig. 7c,d) was detected in the meristematic and infection zones (zones I and II, respectively), as well as in the distal nitrogen-fixing zone (zone III). In parallel, the spatial expression of the MtNAC1 target on consecutive sections indicated a maximum of transcripts in the infection zone (II), and, to a smaller extent, in the distal nitrogen-fixing zone (III; Fig. 7a,b). Thus, the maximal level of miR164 and MtNAC1 transcript accumulation is found in adjacent but overlapping zones in the apical region of the symbiotic nodule.
To test whether MtNAC1 transcript accumulation was regulated by this miRNA, a 5′ RACE-PCR was done with different primer sets (see the Materials and Methods section). In three independent experiments, only sequences located downstream of the miR164-binding site were retrieved, suggesting that a 3′ degradation of the MtNAC1 transcript occurred (Fig. S3). Based on miRBase v.15 and MIRMED databases (http://www.mirbase.org/; http://medicago.toulouse.inra.fr/MIRMED), no other putative M. truncatula miRNA targeting MtNAC1 could be identified (Lelandais-Brière et al., 2009; MIRMED consortium unpublished results). Alternatively, we generated a miRNA sensor construct (Parizotto et al., 2004) by adding the MtNAC1 miR164-binding site to the GFP reporter transcript. Overexpression of GFP transcripts with or without the MtNAC1 miR164-binding site and of a control containing a modified version of the site predicted to be resistant to the miRNA action, revealed that the GFP protein levels decreased significantly only when a native miR164-binding site was present (Fig. 7e).
We analyzed whether overexpression of the MIR164 locus affected nodulation. The MIR164 locus encoding the isoform with the best score for targeting MtNAC1 (Lelandais-Brière et al., 2009), referred to as MIR164c (miRBase v.15), was cloned behind a 2 × 35SCaMV promoter. In addition, a construct corresponding to the hairpin region was generated. In roots overexpressing one or the other construct, the number of nodules was lower than in control roots (expressing the GUS reporter, Fig. 8). This reduced number of nodules was observed at 7 dpi as well as at 15 dpi, suggesting that a developmental delay had been induced by miR164 overexpression. In addition, the nodules formed were pink, indicating that they are probably functional and able to fix nitrogen.
Because the functional analysis of MtNAC1 did not reveal any nodulation phenotype, these results might indicate the existence of other miR164 targets controlling nodulation. Using the miranda software and stringent criteria (see the Materials and Methods section; Fig. S4), NAC1 and another NAC TF (homologous to AtNAC2/ORE1 (Oresara 1); Fig. 1b, TC108847) were retrieved as miR164 targets, as well as four unrelated genes (Fig. S4). Expression data of three of these potential targets was available on the MtGEA database. All were expressed in roots and nodules suggesting that miR164 regulation of symbiotic nodulation may involve several target genes in addition to MtNAC1.
In conclusion, the complex regulation of NAC1 described in Arabidopsis is only partly conserved in M. truncatula. Whereas no interaction with the SINAT proteins was detected, MtNAC1 expression seem to be subjected to miR164-mediated regulation.
In plants, NAC transcription factors are ubiquitous and involved in various aspects of development and in response to environmental stresses. In M. truncatula, the only NAC transcription factor recently functionally analyzed has been linked to cell wall development (Zhao et al., 2010). Here, we describe another member of the M. truncatula NAC transcription factor family that, based on a neighbor-joining tree of known NAC proteins from Arabidopsis and M. truncatula, has been classified in a small subgroup together with two Arabidopsis genes, NAC1 and At3g12977.1 (TAIR database). Therefore, MtNAC1 is the closest Arabidopsis NAC1 homolog in the currently known Medicago genome. MtNAC1 is a functional transcription factor because it is localized in the nucleus, contains a predicted DNA-binding domain, and has a transcriptional activation capacity in a yeast one-hybrid analysis. As the Arabidopsis NAC1 gene was involved in auxin-dependent lateral root development (Xie et al., 2000), and as lateral root and symbiotic nodule development share common regulatory pathways, we investigated whether MtNAC1 might have a function during the development of both root lateral organs in legumes.
The MtNAC1 expression pattern resembled that of the NAC1 in Arabidopsis (Xie et al., 2000), with the highest expression levels in root tips, leaves, and cotyledons. However, in contrast to Arabidopsis, the MtNAC1 expression was not induced upon auxin treatment of roots; neither did root length and lateral root number of a nac1-1 mutant, carrying a premature stop codon, differ significantly from wild-type plants. As antisense expression of the Arabidopsis NAC1 reduced lateral root initiation (Xie et al., 2000), our results suggest that MtNAC1 exerts a different function in the regulation of the root architecture in M. truncatula. Alternatively, as the predicted DNA-binding region of MtNAC1 might be produced in the mutant, we cannot exclude that this partial MtNAC1 protein may still be active. In addition, functional redundancy with another NAC transcription factor cannot be ruled out, nor that the antisense construct used in Arabidopsis (Xie et al., 2000) might have inactivated another related gene in addition to NAC1.
To assess the function of MtNAC1, its involvement in nodulation was investigated. A real-time RT-PCR analysis revealed that the MtNAC1 expression was upregulated during nodule formation, whereas in situ hybridization showed that it was associated with infection and young nitrogen-fixing zones of mature nodules. Nevertheless, no nodulation phenotype could be detected either by overexpression, silencing, or mutation of MtNAC1.
Expression of several NAC transcription factors is subjected to post-transcriptional regulations. In Arabidopsis, NAC1 is regulated by miR164 at the transcript level and the protein is subjected to SINAT5-dependent proteolysis (Xie et al., 2002; Axtell & Bartel, 2005; Guo et al., 2005). The E3 ligase SINAT5 physically interacts with NAC1 to induce its ubiquitination and degradation (Xie et al., 2002). Although MtNAC1 is a close homolog of NAC1, the stability of the protein did not seem to be subjected to SINA-dependent proteolysis, because MtNAC1 and none of the six previously identified M. truncatula SINA proteins interacted (Den Herder et al., 2008). In Arabidopsis, the root phenotypes obtained by overexpressing SINAT5 or a dominant-negative SINAT5 variant were explained by the interaction of SINAT5 with NAC1 (Xie et al., 2002). Our results show that the root phenotypes obtained in M. truncatula by overexpressing SINAT5 or the dominant-negative SINAT5 variant, which were similar to those of Arabidopsis (Den Herder et al., 2008), must be linked to targets other than MtNAC1. Medicago truncatula roots ectopically expressing the SINAT5 dominant-negative variant were also impaired in symbiotic nodulation (Den Herder et al., 2008). Hence, during nodulation also, the SINA proteins must regulate proteins other than MtNAC1 to provoke this phenotype.
In addition to ubiquitin-dependent proteolysis, NAC1 is also regulated by miR164 in Arabidopsis (Ooka et al., 2003; Guo et al., 2005). MtNAC1 contained a predicted miR164-binding site, raising the possibility that expression of this gene is post-transcriptionally controlled. Indeed, miR164 is expressed in roots and nodules (Boualem et al., 2008). In situ hybridization revealed an expression in consecutive and overlapping zones of M. truncatula nodules: miR164 transcripts occurred in the nodule meristem, infection zone and young nitrogen fixation zones whereas MtNAC1 transcripts were detected in infection and young nitrogen fixation zones. These spatial expression patterns of miR164 and MtNAC1 support a potential interaction between the miRNA and its predicted target. However, miR164-dependent mRNA cleavage was impossible to demonstrate because 5′ RACE-PCR indicated that MtNAC1 transcripts were highly unstable, resulting in short 3′-fragments that did not reach the miR164-binding site. As an alternative strategy, a miRNA-GFP sensor construct was used (Parizotto et al., 2004) and revealed that the presence of the MtNAC1 miR164-binding site reduced GFP accumulation. Together, these data indicate that MtNAC1 carries a functional miR164-binding site and that miR164 may preferentially regulate MtNAC1 at the translational level in M. truncatula roots. Indeed, several cases of Arabidopsis miRNA affecting translational efficiency have been reported (Chen, 2004; Brodersen et al., 2008; Lanet et al., 2009), even though this effect, to our knowledge, had not been observed for NAC genes.
Overexpression of miR164 resulted in a decrease in nodule number at 7 dpi as well as at 15 dpi, hinting at a provoked delay in developmental response. Further experiments are required to determine whether Rhizobium infection and/or nodule organogenesis (i.e. cortical cell divisions and primordia formation) depend on the miR164 action. No nodule differentiation or senescence phenotype were detected, but promoters able to drive expression of the miRNA more efficiently at these late nodulation stages would be needed to confirm this observation. The effect of miR164 in nodulation is probably not caused by inactivation of MtNAC1 because neither the RNAi lines nor the mutant had a similar phenotype. Although functional redundancy might explain the absence of phenotypes in plants downregulating MtNAC1, it is also possible that other miR164 targets are regulated during nodulation. Interestingly, another MtNAC transcription factor-encoding gene, designated MtNAC2, contains a miR164-binding site and is expressed in roots and nodules. Its closest homolog in Arabidopsis, the auxin-inducible AtNAC2/ORE1 gene, is involved in lateral root development and salt stress response and has been shown to contain a functional miR164-binding site (He et al., 2005; Kim et al., 2009). In addition, at least two other non-NAC miR164 targets expressed in roots and nodules are predicted in M. truncatula. Further work would be needed to analyze which of these genes are indeed regulated by miR164 and leads to nodulation phenotypes when misexpressed.
In summary, our data do not support a role for MtNAC1 in M. truncatula root or nodule development, although it is, to date, the closest homolog of the Arabidopsis NAC1 transcription factor. Therefore, different functions might have evolved for these related genes. Similarly, transcriptional and post-transcriptional regulations of NAC1 expression are also only partly evolutionarily conserved, because MtNAC1 did not interact with the MtSINA E3 ligases. By contrast, Medicago and Arabidopsis NAC1 homologs carry a functional miR164-binding site and expression data as well as the use of a miRNA sensor support an in vivo interaction in Medicago roots. However, miRNAs can target several transcripts, and the nodulation phenotype induced by miR164 overexpression may hint at additional targets, related, or unrelated, to the NAC family, that must be important for nodulation.
We thank David Baker and Jonathan Clarke (JIC-Norwich, UK) as well as Christine Le Signor and Richard Thompson (INRA-Dijon, France) for performing the TILLING screen of the nac1-1 mutant, Adnane Boualem for generating the miR164-overexpressing construct, Fanny Calenge for help with some 5′ RACE-PCR experiments. Marnik Vuylsteke for assistance with statistical analysis, Joanna Boruc for help with confocal imaging and nuclear localization assays, Christa Verplancke for technical assistance. Giel Van Noorden for helpful comments, and Martine De Cock for help preparing the manuscript. This work was supported in part by the Transnational (Germany, France, Spain and Portugal) Cooperation (PLANT-KBBE Initiative) ‘ROOT project’, including funding from Federal Ministry of Education and Research (BMBF, Germany), Agence Nationale de la Recherche (ANR, France), Ministerio de Ciencia e Innovación (MICINN, Spain), as well as by funding from the Ministerie van de Vlaamse Gemeenschap (grant no. CLO/IWT/020714), the Research Foundation-Flanders (grant no. G0066.07N), and the European Community Framework 6 Integrated project (GRAIN LEGUMES, contract FOOD-CT-2004-506623). KD was a Research fellow of the Research Foundation-Flanders, and JP was the recipient of a doctoral grant from the Ministère de la Recherche et de la Technologie (France).