Dual involvement of a Medicago truncatula NAC transcription factor in root abiotic stress response and symbiotic nodule senescence

Authors

  • Axel de Zélicourt,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    2. Université Paris Diderot Paris 7, Les Grands Moulins, 16 rue Marguerite Duras, 75205 Paris Cedex 13, France
    Search for more papers by this author
  • Anouck Diet,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    2. Université Paris Diderot Paris 7, Les Grands Moulins, 16 rue Marguerite Duras, 75205 Paris Cedex 13, France
    Search for more papers by this author
  • Jessica Marion,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    Search for more papers by this author
  • Carole Laffont,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    Search for more papers by this author
  • Federico Ariel,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    Search for more papers by this author
  • Michaël Moison,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    Search for more papers by this author
  • Ons Zahaf,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    Search for more papers by this author
  • Martin Crespi,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    Search for more papers by this author
  • Véronique Gruber,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
    2. Université Paris Diderot Paris 7, Les Grands Moulins, 16 rue Marguerite Duras, 75205 Paris Cedex 13, France
    Search for more papers by this author
  • Florian Frugier

    Corresponding author
    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique, 91198 Gif sur Yvette Cedex, France
      (fax +00 33 1 69 82 36 95; e-mail frugier@isv.cnrs-gif.fr).
    Search for more papers by this author

(fax +00 33 1 69 82 36 95; e-mail frugier@isv.cnrs-gif.fr).

Summary

Legume crops related to the model plant Medicago truncatula can adapt their root architecture to environmental conditions, both by branching and by establishing a symbiosis with rhizobial bacteria to form nitrogen-fixing nodules. Soil salinity is a major abiotic stress affecting plant yield and root growth. Previous transcriptomic analyses identified several transcription factors linked to the M. truncatula response to salt stress in roots, including NAC (NAM/ATAF/CUC)-encoding genes. Over-expression of one of these transcription factors, MtNAC969, induced formation of a shorter and less-branched root system, whereas RNAi-mediated MtNAC969 inactivation promoted lateral root formation. The altered root system of over-expressing plants was able to maintain its growth under high salinity, and roots in which MtNAC969 was down-regulated showed improved growth under salt stress. Accordingly, expression of salt stress markers was decreased or induced in MtNAC969 over-expressing or RNAi roots, respectively, suggesting a repressive function for this transcription factor in the salt-stress response. Expression of MtNAC969 in central symbiotic nodule tissues was induced by nitrate treatment, and antagonistically affected by salt in roots and nodules, similarly to senescence markers. MtNAC969 RNAi nodules accumulated amyloplasts in the nitrogen-fixing zone, and were prematurely senescent. Therefore, the MtNAC969 transcription factor, which is differentially affected by environmental cues in root and nodules, participates in several pathways controlling adaptation of the M. truncatula root system to the environment.

Introduction

Plants can adapt their root system architecture to changing soil environmental conditions through formation of new lateral organs. In legumes, in addition to root branching, nitrogen-fixing nodules are formed. Both root lateral organs develop post-embryonically in front of protoxylem poles by re-activation of differentiated cells of the parental root, mainly the pericycle (lateral roots) or cortex (root nodules) (Gonzalez-Rizzo et al., 2009). Formation of lateral roots depends on nutrient availability (Lopez-Bucio et al., 2003), while nodules develop under nitrogen-starvation conditions in the presence of a specific bacterial microsymbiont: Sinorhizobium meliloti in the case of the model legume Medicago truncatula. The interaction relies on recognition by legume roots of bacterial Nod factors, which elicit root hair deformations and lead to formation of infection threads, allowing bacteria to progress towards the cortex (Oldroyd and Downie, 2008). Simultaneously, nodule organogenesis proceeds in the inner root cell layers, leading to formation of a primordium that is colonized by infection threads and subsequently differentiates into a mature organ (Crespi and Frugier, 2008). The M. truncatula nodule has indeterminate growth, and four developmental zones can be defined: a persistent apical meristem (zone I), a rhizobial infection region in which cell differentiation is marked by accumulation of amyloplasts (zone II), a functional zone in which bacteria that have differentiated into bacteroids fix atmospheric nitrogen (zone III), and a senescence zone in which bacteria and plant cells are degraded (zone IV) (Vasse et al., 1990). Study of several plant mutants that are unable to form nodules (reviewed by Oldroyd and Downie, 2008; Kouchi et al., 2010) led to identification of transcription factors (TFs) involved in nodulation: two GRAS TFs (NSP1 and NSP2: Nodulation Signaling Pathway 1 and 2; Kalo et al., 2005; Smit et al., 2005; Murakami et al., 2006; Heckmann et al., 2006), an AP2/ERF TF (MtERN1: ERF Required for Nodulation), corresponding to the bit1 (branched infection threads 1) mutant (Middleton et al., 2007), a putative NIN (nodule inception) TF (Schauser et al., 1998; Marsh et al., 2007), and a CCAAT TF (HAP2-1/NF-YA/CBF-B, which affects nodule growth and differentiation; Combier et al., 2006). In addition, certain mutants that form more nodules than wild-type plants show alterations in TFs: the ASTRAY bZIP TF, which controls nodule number depending on environmental conditions (Nishimura et al., 2002), and the AP2/ERF TF encoded by efd (ERF required for nodule formation), which affects nodule number and differentiation (Vernie et al., 2008). Finally, the Zpt2-1 C2H2 zinc-finger TF regulates nitrogen-fixing zone differentiation (Frugier et al., 2000) and the M. truncatula root response to salt stress (Merchan et al., 2007).

Abiotic stresses such as salinity are major agricultural constraints affecting crop yield. Plants have developed various strategies to cope with ionic and osmotic stresses, including stress avoidance by modifying their root system architecture to minimize exposure to salt (Malamy, 2005). Transcriptomic profiling has been widely used in various plants, including M. truncatula, to identify genes whose expression levels change in response to saline conditions (Gruber et al., 2009; Li et al., 2009). Among regulatory genes, several TFs linked to abiotic stress responses have been identified, but, to our knowledge, only four have been functionally analyzed in a homologous legume context. These include the zinc-finger proteins Alfin1 in Medicago sativa (Winicov and Bastola, 1999) the M. truncatula TF Zpt2-1 mentioned above, the related MtZpt2-2 protein (de Lorenzo et al., 2007), and the MtHB1 HD-Zip I TF (Ariel et al., 2010).

Among TFs found to be up-regulated by salt in M. truncatula roots using an extensive real-time RT-PCR profiling approach, several NAC genes (family named based on its founding members NAM, ATAF1/2 and CUC2) were identified (Gruber et al., 2009). In non-legumes, certain NAC TFs are linked to developmental processes, such as those encoded by the CUC genes (Olsen et al., 2005), whereas others are related to abiotic stress responses. In Arabidopsis thaliana, over-expression of ATAF1 (ANAC002) conferred drought tolerance as well as salt and oxidative stress hypersensitivity, but the corresponding mutant did not show an altered phenotype, suggesting functional redundancy of NAC genes in response to abiotic stresses (Wu et al., 2009). Over-expression of the salt-induced AtNAC2/AtNAC6/ANAC092/ORE (Oresara1) gene enhanced lateral root formation under various conditions, including mild salt stress (4 days of 75 mm NaCl), but no root phenotype was observed in the corresponding mutant (He et al., 2005). However, the atnac2 mutant showed a salt-tolerant germination phenotype, whereas plants over-expressing this TF were salt-hypersensitive (Balazadeh et al., 2010). Related NAC genes have also been functionally linked to abiotic stresses in rice (Oryza sativa), as over-expression of the OsNAC6 gene (closely related to ANAC002/ATAF1 and ANAC072/RD26 in Arabidopsis) conferred salt tolerance. Interestingly, up- or down-regulation of OsNAC6 increased or decreased root length, respectively (Nakashima et al., 2007; Chung et al., 2009). In addition, over-expression of OsNAC10 (closely related to AtNAP/ANAC029) induced drought tolerance, associated with an enlarged root diameter under control conditions (Jeong et al., 2010). In legumes, information regarding functional interactions between NACs, abiotic stresses and the development of roots or symbiotic nodules is very scarce. As in non-legume species, expression of several NAC genes is up-regulated in response to various abiotic stresses (Pinheiro et al., 2009), but only two NAC TFs have been genetically characterized in M. truncatula: the nst1 mutant (NAC secondary wall thickening promoting factor 1), which disrupts lignin and polysaccharide biosynthesis pathways and show cell-wall disorganization (Zhao et al., 2010), and the nac1 mutant, which, in contrast to mutants of its closest Arabidopsis homolog, did not show any root phenotype (d’Haeseleer et al., 2011). Overall, no phenotype linked to roots, nodules or any specific stress was reported for these mutants affecting legume NAC TFs.

In this study, we determined that a NAC TF that is rapidly up-regulated by salt stress (Gruber et al., 2009), MtNAC969, regulates M. truncatula root architecture and symbiotic nodulation. In addition, expression of MtNAC969 was differentially regulated by salt in roots and nodules, and behaved similarly to previously reported senescence nodulation markers. Our results suggest that this NAC TF participates in various pathways required for adaptive root responses to salt stress and establishment of a functional symbiotic nodule.

Results

MtNAC969 negatively regulates the salt-stress root response

By altering the expression of previously identified salt-regulated TFs (Gruber et al., 2009), we developed a screen to identify salt-response phenotypes in M. truncatula roots (V.G., F.F. and M.C., unpublished results). Among the constructs tested, an RNAi down-regulating the MtNAC969 TF (either with or without salt stress; Figure S1A) allowed maintenance of better root growth under salt stress (as shown by measuring either root length or root dry weight; Figure 1a–c). Expression of previously reported Medicago salt-stress markers encoding a CORA homolog (cold-regulated A; Merchan et al., 2007), a germin and an osmotin (Gruber et al., 2009) was analyzed in MtNAC969 RNAi roots. Consistently increased expression was found for two of these markers: CORA and germin (Figure 1d). When MtNAC969 was over-expressed, no significant change in root length or dry weight was detected between salt-treated and non-treated plants (Figure 2a–c). Under non-treated conditions, MtNAC969 over-expressing roots showed an altered development, leading to reduced growth and dry weight. This phenotype may minimize the root surface exposed to the stress and explain the lower salt sensitivity of these roots compared to the wild-type. MtNAC969 over-expression was confirmed by real-time RT-PCR in the presence or absence of salt (Figure S1B), and expression of salt-stress markers was reduced in these roots (Figure 2d), in contrast to the RNAi roots.

Figure 1.

 Root phenotype in response to salt stress of composite plants in which MtNAC969 had been silenced.
(a) Representative images of 3-week-old GUS RNAi and MtNAC969 RNAi composite plants 7 days after treatment with or without salt (100 mm NaCl).
(b,c) Root length (b) and dry weight (c) of composite plants expressing an MtNAC969 or GUS RNAi construct. Three-week-old composite plants were treated with salt (100 mm NaCl) or not for 1 week, as shown in (a), and root dry weights and lengths were measured. A representative example of two biological experiments is shown. Error bars represent the confidence interval (α = 0.05), and asterisks indicate significant differences for a specific condition (treated or not) based on a Mann–Whitney test (α < 0.05; n > 15).
(d) Expression of CORA, germin and osmotin salt marker genes in GUS RNAi and MtNAC969 RNAi roots under control conditions. Bars show quantification of specific PCR amplification products for each gene, normalized using the mean value for three constitutive genes (as described in Experimental procedures). GUS RNAi plant 1 was used to define fold changes (indicated by the dashed line). Plants from a representative example of two biological replicates are shown, and error bars represent the confidence interval (α = 0.05) of two technical replicates.

Figure 2.

 Root phenotype in response to salt stress of composite plants over-expressing MtNAC969.
(a) Representative images of 3-week-old 35S::GUS and 35S::MtNAC969 composite plants 7 days after treatment with or without salt (100 mm NaCl).
(b,c) Root length (b) and dry weight (c) of composite plants over-expressing MtNAC969 or a GUS control grown under salt stress for 1 week, as shown in (a). A representative example of two biological experiments is shown. Error bars represent the confidence interval (α = 0.05), and asterisks indicate significant differences between genotypes for a specific condition (treated or not) based on a Mann–Whitney test (*α < 0.05; **α < 0.01; n > 15).
(d) Expression of CORA, germin and osmotin salt marker genes in 35S::GUS and 35S::MtNAC969 composite plants under control conditions. Bars show quantification of specific PCR amplification products for each gene, normalized using the mean value for three constitutive genes (as described in Experimental procedures). 35S::GUS plant 1 was used to define fold changes (indicated by the dashed line). Plants from a representative example of two biological replicates are shown, and error bars represent the confidence interval (α = 0.05) of two technical replicates.

The Arabidopsis protein to which MtNAC969 is most closely related (54% homology) is the AtNAP protein (‘NAC-like activated by APETALA3/PISTILLATA’; Figure S2), which has been linked to flower development (Sablowski and Meyerowitz, 1998). However, MtNAC969 shows a stronger expression in Medicago roots rather than flowers (Figure S3A), in which it is up-regulated mainly by salt among a variety of abiotic stresses (Figure S3B,C; Gruber et al., 2009). Data available in the Medicago truncatula gene expression atlas (MtGEA; http://mtgea.noble.org/) indicate that this gene is also induced in response to a biotic stress (Phymatotricum sp. infection; Figure S3C).

Overall, these results indicate that MtNAC969 negatively regulates M. truncatula root growth and molecular responses under salt stress.

MtNAC969 regulates lateral root formation but do not affect nodule number

As root growth was affected by over-expression of MtNAC969 under non-saline conditions, we analyzed the potential role of MtNAC969 in lateral organ formation. Using a lateral root-inducing medium (Gonzalez-Rizzo et al., 2006), we found opposite phenotypes for MtNAC969 over-expressing and silenced roots, i.e. a decreased or increased density of emerged lateral roots, respectively (Figure 3a), suggesting that MtNAC969 negatively regulates lateral root formation. We then tested whether hormones affecting root architecture can regulate MtNAC969 expression. This TF is rapidly induced in roots by cytokinins, ethylene and abscisic acid (ABA) (Figure 4a), revealing complex phytohormonal regulations. As MtNAC969 is expressed throughout the various stages of nodule development (Figure 4b and Figure S3C), we assessed the nodulation capacity of MtNAC969 over-expressing and RNAi-silenced roots. In contrast to the lateral root phenotypes previously identified, no significant difference in nodule number was observed (Figure 3b).

Figure 3.

 Root lateral organogenesis of composite plants in which MtNAC969 was over-expressed or silenced.
(a) Lateral root density (number of lateral roots per cm parental root) for composite plants in which MtNAC969 was over-expressed or silenced (normalized relative to 35S::GUS or GUS RNAi constructs, respectively, as indicated by the dashed line) measured 5 days after transfer to lateral root-inducing medium.
(b) Nodule number in composite plants in which MtNAC969 was over-expressed or silenced (normalized relative to 35S::GUS or GUS RNAi constructs, respectively, as indicated by the dashed line) determined 21 days post-inoculation with S. meliloti Sm1021. In both cases, a representative example of two biological experiments is shown (> 20). Error bars represent the confidence interval (α = 0.05), and asterisks indicate significant differences relative to the GUS control based on a Mann–Whitney test (*α < 0.05; **α < 0.01; > 20).

Figure 4.

 Regulation of MtNAC969 expression by phytohormones and in symbiotic nodules.
(a) MtNAC969 expression in roots after short-term (1 or 3 h) hormonal treatments: auxin (10−7 m IAA), cytokinin (10−7 m BAP), ethylene (10−7 m ACC) and abscisic acid (10−7 m ABA).
(b) Expression of MtNAC969 during a nodulation kinetic (from 0 to 40 days post-inoculation with S. meliloti Sm1021). The non-inoculated root (0 dpi) was used to define fold changes (indicated by the dashed line).
In (a) and (b), bars show quantification of specific PCR amplification products for each gene, normalized using the mean value for three constitutive genes (as described in Experimental procedures). Fold changes are indicated by the dashed line relative to the non-treated condition. A representative example of two biological replicates is shown, and error bars represent the confidence interval (α = 0.05) of two technical replicates.
(c) Analysis of MtNAC969 spatial expression using in situ hybridization in nodules treated with 10 mm KNO3 or not for 48 h. Left: MtNAC969 sense probe (negative control); right and middle: MtNAC969 antisense probe in nodules treated with KNO3 or not, respectively. Scale bars = 500 μm.

MtNAC969 is differentially regulated by salt between roots and nodules and behaves as a nodule senescence marker

Data from the Medicago truncatula gene expression atlas (Figure S3C) revealed that MtNAC969 expression is strongly induced in nodules treated with nitrate for 2 days. Indeed, a 48 h treatment with 10 mm KNO3 resulted in accumulation of MtNAC969 transcripts (Figure 5a). In situ hybridization also indicated that MtNAC969 expression in the central tissues of the nodule was enhanced after nitrate treatment (Figure 4c).

Figure 5.

 Expression of MtNAC969 in roots or nodules treated with salt or nitrate.
(a) Expression of MtNAC969 (left) and salt response marker genes (CORA, germin and osmotin; right) in roots treated with salt (100 mm NaCl) or nitrate (10 mm KNO3) for 48 h.
(b) Expression of MtNAC969 (left), salt response marker genes (CORA, germin and osmotin; middle) and senescence marker genes encoding a putative receptor kinase and cysteine proteases (MtCP2 and MtCP3) (right) in nodules treated with salt (100 mm NaCl) or nitrate (10 mm KNO3) for 48 h.
In (a) and (b), bars show quantification of specific PCR amplification products for each gene, normalized using the mean value for three constitutive genes (as described in Experimental procedures). Fold changes are indicated by the dashed line relative to non-treated conditions. A representative example of two biological experiments is shown, and error bars represent the confidence interval (α = 0.05).

To obtain a more comprehensive picture of MtNAC969 regulation by environmental cues in the two organs, we comparatively analyzed the effect of a 48 h salt (100 mm NaCl) or nitrate (10 mm KNO3) treatment in roots and nodules (Figure 5). Nitrate induced MtNAC969 expression in both organs (Figure 5a,b, left), but to a lower extent in roots than in nodules. In addition, MtNAC969 expression was induced by salt in roots, as expected, but was surprisingly repressed by salt in nodules (Figure 5a,b, left).

As nitrate treatment induces nodule senescence (Puppo et al., 2005), we attempted to correlate the MtNAC969 expression pattern with known salt and nodule senescence markers. The abiotic stress marker genes CORA, osmotin and germin were similarly induced by salt stress in roots and nodules (Figure 5a, right, and Figure 5b, middle), in contrast to MtNAC969 (Figure 5a,b, left). We then analyzed senescence nodulation markers that are either up-regulated during both aged and induced senescence (e.g. a putative S-receptor kinase) or continuously up-regulated during aged-senescence and transiently induced during dark-induced senescence (e.g. cysteine proteases; Perez Guerra et al., 2010). Expression of the gene encoding a putative S-receptor kinase is, as expected, strongly enhanced by nitrate treatment compared to markers of aged-senescence (e.g. cysteine proteases; Figure 5b, right). This further supports results showing that exogenous nitrate treatment induces nodule senescence. In contrast to MtNAC969, expression of the senescence markers was not detectable in roots, with or without salt or nitrate treatment (data not shown). Interestingly, salt treatment in nodules inhibited expression of these senescence markers, as observed for MtNAC969 (Figure 5b, left and right).

Altogether, these results show that this TF acts in different environmentally dependent regulatory pathways depending on the organ studied. Indeed, the differential expression of MtNAC969 between roots and nodules defines a new pattern, similar to that of salt stress-responsive markers in roots and senescence-related markers in nodules.

MtNAC969 RNAi nodules show premature senescence

To further assess the potential role for NAC969 in nodule senescence, we characterized the histology of differentiated MtNAC969 RNAi nodules. This revealed the presence of infection threads as well as infected cells, with elongated bacteroids radiating from a central vacuole as observed in control nodules (Figure 6a,b). However, cells located in the nitrogen-fixing zone of the MtNAC969 RNAi nodules accumulated a higher number of amyloplasts than those from wild-type nodules, as shown by lugol staining (Figure 6c,d) or electron microscopy (Figure 6e,f), suggesting an altered nodule metabolism. We then analyzed expression of the various senescence nodulation markers previously used (Figure 7). In accordance with the cellular phenotype, two cysteine protease-encoding markers of nodule aged-senescence accumulate in MtNAC969 RNAi nodules, indicating the premature senescence of these nodules. These results therefore indicate that the MtNAC969 TF is required for establishment of a functional nitrogen-fixing nodule, and may negatively regulate nodule senescence.

Figure 6.

 Comparison of the ultrastructural organization of GUS RNAi and MtNAC969 RNAi nodules.
(a,b) Toluidine blue-stained sections of GUS RNAi (a) and MtNAC969 RNAi (b) nodules 21 days post-inoculation. Zones I (meristem), II (infection/differentiation) and III (nitrogen-fixing) are defined as described by Vasse et al. (1990). Scale bars = 500 μm.
(c,d) Lugol staining of starch granules in GUS RNAi (c) and MtNAC969 RNAi (d) nodules 21 days post-inoculation. Starch accumulation, used as a marker for nodule differentiation, is enhanced in the nitrogen-fixing zone of MtNAC969 RNAi nodules. Scale bars = 500 μm.
(e,f) Electron micrographs of longitudinal ultrathin sections of GUS RNAi (e) and MtNAC969 RNAi (f) nodule cells from the nitrogen-fixing zone, as outlined in red in (a) and (b), 21 days post-inoculation. A high number of amyloplasts (asterisks) are found in MtNAC969 RNAi cells. Scale bars = 5 μm.

Figure 7.

 MtNAC969 RNAi nodules accumulate senescence markers. Expression of senescence marker genes encoding a putative receptor kinase and cysteine proteases (MtCP2 and MtCP3) in GUS RNAi and MtNAC969 RNAi roots under control conditions. The GUS RNAi nodules sample 1 (Nod 1) was used to define fold changes (indicated by the dashed line). Bars show quantification of specific PCR amplification products for each gene, normalized using the mean value for three constitutive genes (as described in Experimental procedures). A representative example of two biological experiments is shown, and error bars represent the confidence interval (α = 0.05).

Discussion

In this study, we characterized a NAC transcription factor that is rapidly induced by salt stress and various plant hormones and plays a role in the regulation of root architecture and symbiotic nodulation in Medicago truncatula. MtNAC969 ectopic expression and inactivation by a RNAi approach affected lateral root formation, as well as salt stress response. Accordingly, expression of salt-stress markers was regulated in opposite ways in these roots. In addition, MtNAC969 down-regulation led to formation of nodules with an altered development, characterized by aberrant amyloplast accumulation and de-regulation of senescence markers. These results reveal the dual involvement of this TF in root system adaptation to abiotic stress and in symbiotic nodule senescence. Nodules and roots may then use related TFs for various adaptive developmental responses depending on environmental conditions.

Ectopic or down-regulation of MtNAC969 expression led to changes in root system architecture. MtNAC969 over-expression reduced lateral root numbers, but RNAi roots showed increased lateral root density (Figure 3a). In addition to the root growth defect (Figure 2b), this finding probably explains the reduced dry weight of roots over-expressing MtNAC969 compared to control plants (Figure 2c). MtNAC969 thus acts as a negative regulator of lateral root formation. Interestingly, MtNAC969 expression in the roots is rapidly affected by various phytohormone treatments, being induced by cytokinins, ethylene and ABA (Figure 4a). In Arabidopsis, these hormones are predicted to negatively regulate lateral root formation (Fukaki and Tasaka, 2009; Perilli et al., 2010), in contrast to auxin, which is linked to the promotion of root growth and does not up-regulate MtNAC969 expression. Hormonal regulation of the MtNAC969 expression is therefore consistent with the root phenotypes observed when the gene is down-regulated. However, hormones that induce MtNAC969 expression play either negative or positive roles in nodule initiation (Gonzalez-Rizzo et al., 2009), and interestingly no effect on nodule number was observed in MtNAC969 RNAi roots. MtNAC969 appears to function at the crossroads of several hormonal pathways controlling root growth and architecture.

The closest homolog of MtNAC969 in Arabidopsis is AtNAP/ANAC029. This gene was initially linked to floral and stamen development that is dependent on APETALA3 and PISTILLATA TFs (Sablowski and Meyerowitz, 1998). AtNAP was later shown to be induced by salt and slightly regulated by other abiotic stresses (Jensen et al., 2010), as is the case of MtNAC969. With regard to hormonal induction, AtNAP is induced by ethylene and ABA (Guo and Gan, 2006), and the latter hormone also regulates many other NAC genes (Jensen et al., 2010). However, no NAP function in root development or any abiotic stress response has been identified in Arabidopsis. Over-expression of the closest rice MtNAC969 homolog, OsNAC10, induced drought tolerance and a modified root system under control conditions (Jeong et al., 2010). Similarly, over-expression of MtNAC969 modified the M. truncatula root system architecture, which may allow plants to maintain a higher root mass under salt stress. Additionally, roots over-expressing or down-regulating MtNAC969 showed decreased or increased expression of salt-stress markers, respectively. This suggests that MtNAC969 represses root responses to salt. In Arabidopsis, a few TFs transcriptionally induced by ABA and abiotic stresses (such as EARLY RESPONSIVE TO DEHYDRATION 15/ERD15) have been described as acting as negative regulators of the stress response (Kariola et al., 2006). These TFs may act in a stress attenuation response, preventing or modulating the stress response, or in a negative feedback mechanism to adapt root responses (and their ABA sensitivity) to the intensity or duration of the stress.

By comparing the expression of MtNAC969 in roots and nodules under salt stress, we identified an unexpected differential behavior of this TF depending on the organ studied (Figure 5). This gene thus behaves as a stress-responsive marker in roots and as an environmentally induced senescence marker in nodules, revealing interactions between nodule senescence and root abiotic stress responses. Several genes have already been described as stress-responsive and senescence-related, suggesting certain common pathways (Puppo et al., 2005). However, recent results obtained in M. truncatula indicate that developmentally programmed nodule senescence and the senescence induced in response to environmental signals, such as dark or nitrate, are more divergent than anticipated, as approximately 50% of the genes associated with these processes are differentially regulated (Perez Guerra et al., 2010). Comparable results were obtained for Arabidopsis leaf senescence, one of the best documented senescence processes (Lim et al., 2007), in which genes regulated during developmental or dark-induced senescence only partly overlap (Buchanan-Wollaston et al., 2005). NACs were the largest TF family regulated during leaf senescence (one-fifth of the family, i.e. 20 members; Guo et al., 2004). This includes AtNAP, and expression of this TF induced premature leaf senescence and activated the senescence-associated marker genes SAG12 and SAG13 (Guo and Gan, 2006). Conversely, atnap mutants show delayed developmental leaf senescence. A similar phenotype was recently reported for the atnac2 mutants, and delayed dark-induced senescence was observed in detached leaves (Balazadeh et al., 2010). The AtNAC2/ORE1/ANAC092 TF is salt-induced, and promotes stress-induced senescence and activates a large number of SAG genes (i.e. approximately one-third of the 170 reported SAG genes; Balazadeh et al., 2010). In addition, using the Leaf Senescence Database (http://www.eplantsenescence.org/), expression of 30 NAC genes was found to be associated with leaf senescence, including AtNAM/anac018/At1g52880, ANAC055/At3g15500, ANAC019/At1g52890 and AtATAF1/anac002/At1g01720, and these are included in the phylogenetic tree shown in Figure S2A. Interestingly, of the 827 and 445 genes involved in Arabidopsis and Medicago leaf senescence, respectively, MtNAC969 and its closest homolog AtNAP are among 150 genes showing an evolutionarily conserved senescence expression pattern (De Michele et al., 2009). Surprisingly, however, only one NAC-encoding gene was identified in transcriptomic studies that aimed to characterize nodule senescence in M. truncatula (Van de Velde et al., 2006). This TF shows only 35% homology with MtNAC969, and its closest Arabidopsis homolog FEZ/ANAC009/At1g26870 has not been linked to senescence (Willemsen et al., 2008). However, this low representation of NAC TFs in nodule senescence may be due to the reduced size of the transcriptome assayed in these studies (approximately 15 000 gene tags defined using cDNA-AFLP [amplified fragment length polymorphism]; Van de Velde et al., 2006).

In MtNAC969 RNAi nodules, a late symbiotic nodulation phenotype was identified, consisting of aberrant accumulation of amyloplasts and premature expression of senescence markers (Figures 6 and 7). Interestingly, several studies using bacteria defective in nitrogen fixation previously linked accumulation of amyloplasts to premature senescence (Lodwig et al., 2003; Harrison et al., 2005). Functional characterization of genes acting in nodule senescence remains very limited. Antisense plants with disrupted expression of MtZPT2-1, which encodes a C2H2 zinc finger TF, developed non-functional nodules that accumulated amyloplasts (Frugier et al., 2000). This gene is regulated by salt stress in Medicago roots in a similar way to MtNAC969, and roots over-expressing this TF showed better growth under salt stress (de Lorenzo et al., 2007). The MtZPT2-1 pathway therefore also links stress responses to differentiation and functionality of symbiotic nodules.

Overall, this study shows that the MtNAC969 TF characterized in this study regulates both root architecture and symbiotic nodulation, and links root abiotic stress responses to symbiotic nodule senescence. The divergent transcriptional regulation of MtNAC969 by abiotic signals in roots versus nodules further highlights the complexity of environmental responses in the legume below-ground system.

Experimental procedures

Plant material and treatments

Medicago truncatula Jemalong A17 seeds were sterilized as described by Gonzalez-Rizzo et al. (2006), and stratified at 4°C for 3 days to ensure uniform germination.

For hormonal treatments, 15 germinated seedlings were placed in a Magenta box with 30 mL of low-nitrogen ‘i’ medium and grown in a shaking incubator (125 rpm) at 24°C under long-day conditions (16 h light/8 h dark) as described by Gonzalez-Rizzo et al. (2006). After 5 days, seedlings were treated with 10−7 m indole-3-acetic acid (IAA), 10−7 m 1-aminocyclopropane-1-carboxylic acid (ACC), 10−7 m benzylaminopurine (BAP) or 10−7 m abscisic acid (ABA) (all obtained from Sigma-Aldrich, http://www.sigmaaldrich.com/), and maintained under the same growth conditions for various incubation times (0, 1 and 3 h). Bioactive hormonal conditions were defined based on previous studies (Gonzalez-Rizzo et al., 2006; Ariel et al., 2010; Plet et al., 2011), and M. truncatula root physiological and molecular responses were tested for a range of concentrations and time points.

For abiotic stresses analysis, plants were similarly grown and then treated for 1 h with 100 mm NaCl or 200 mm mannitol (i.e. an osmolarity of 200 mOsmol in each case) or kept for 1 h at 4 or 37°C.

For nodulation experiments, plants were transferred to the greenhouse (16 h light/8 h dark, 22°C, 60–70% relative humidity) in pots containing perlite/sand (3:1 v/v), and irrigated with ‘i’ medium (Gonzalez-Rizzo et al., 2006). After 5 days, roots were inoculated with 150 mL of Sinorhizobium meliloti Sm1021 suspension (OD600 nm = 0.05) per pot. Twenty-one days post-inoculation (dpi), plants were treated with 100 mm NaCl or 10 mm KNO3 for 48 h (based on results available in the M. truncatula gene expression atlas; Benedito et al., 2008). Uninoculated plants were submitted to the same treatments. The early nodulation stages (i.e. 1, 3 and 8 dpi) of the nodulation kinetics were determined in vitro as described by Laporte et al. (2010).

In all cases, two or three biological replicates were performed per treatment.

Agrobacterium rhizogenes root transformation

To knockdown MtNAC969 expression using RNA interference, a specific region was amplified by PCR using the primers NAC969-RNAi-F (5′-CACCTTTCATCCAACTGATGAGGAAC-3′) and NAC969-RNAi-R (5′-CTTTGATCCACTTTTGATTGCC-3′), and cloned into the pFRN destination vector (derived from pFGC5941; NCBI accession number AY310901) using Gateway technology (Invitrogen, http://www.invitrogen.com/). A pFRN vector that includes the GUS reporter gene was used as a control (Gonzalez-Rizzo et al., 2006).

To ectopically express MtNAC969, a full-length cDNA was amplified from salt-treated M. truncatula roots using primers NAC969-FL-F (5′-GCTCTAGAGAAGGAGATAGAACCATGGAGAGTAGTGCAAG-3′) and NAC969-FL-R (5′-GGGGTACCTATGCTAAATAATTATGCATTTGTTTCAC-3′). The MtNAC969 sequence was submitted to the National Center for Biotechnology Information database (accession number JN833713). After digestion with XbaI and KpnI (New England Biolabs, http://www.neb.com/; restriction sites underlined in primers described above), fragments were cloned into the binary vector pMF2 (containing a 35S CaMV promoter; Merchan et al., 2007). An empty pMF2 vector expressing the GUS reporter was used as control (35S:GUS).

All constructs were introduced into Agrobacterium rhizogenes ARqua1 strain (an SmR-derivative strain of A4T; Quandt et al., 1993) and used for M. truncatula root transformation as described by Boisson-Dernier et al. (2001). The ‘composite plants’ obtained consist of wild-type shoots and transgenic roots, and each plant therefore corresponds to at least one independent transformation event. The efficiency of silencing and specificity of the RNAi were checked for each construct using real-time RT-PCR on several representative individual clones (n > 3).

Treatment of Agrobacterium rhizogenes composite plants

For salt treatment, composite plants were transferred to the greenhouse (16 h light/8 h dark, 22°C, 60–70% relative humidity) in pots containing perlite/sand (3:1 v/v), and irrigated with ‘i’ medium for 1 week. Plants were then supplemented with or without 100 mm NaCl for 1 week. Root length was measured using ImageJ software (http://rsbweb.nih.gov/ij/), and aerial parts and roots were then individually dried at 60°C to determine their dry weight. Three independent biological replicates were performed (n > 12 plants for each construct and condition).

To determine lateral root number, composite plants were transferred in vitro onto growth papers (Mega International, http://www.duclosinternational.eu/) on SN/2 medium (Soluplant 18.6.26, Duclos International) with 0.9% Kalys agar (Kalys HP696, http://en.kalys.com/) for 3 days. Plants were then transferred to lateral root-inducing medium to induce synchronously lateral root development (Gonzalez-Rizzo et al., 2006). The position of root tips was marked (with a pencil) at the time of transfer. Primary root length and lateral root number were measured from this mark after 5 days using ImageJ software, and lateral root density was then calculated (i.e. lateral root number/cm main root length). Two biological experiments were performed, and a minimum of 20 independent plants per construct and per condition was measured.

For nodulation assays, transgenic roots were transferred to the greenhouse (16 h light/8 h dark, 22°C, 60–70% relative humidity) in pots containing perlite/sand (3:1 v/v) and irrigated with ‘i’ medium for 5 days. Transgenic roots were inoculated with 150 mL of Sinorhizobium meliloti Sm1021 suspension (OD600 nm = 0.05). Nodulation was evaluated at 21 dpi, and nodules were collected for histological analysis or harvested in liquid nitrogen for molecular assays.

Gene expression analysis

Total RNAs were isolated from belowground organs by phenol/chloroform extraction of samples ground to a fine powder and allowed to thaw into five volumes of a buffer containing 0.1 m NaCl, 2% SDS, 50 mm Tris/HCl, pH 9.0, 10 mm EDTA and 20 mmβ-mercaptoethanol. Total RNA (1 μg) was then subjected to DNase I treatment (Fermentas, http://www.fermentas.com/), and first-strand cDNAs were synthesized using the Superscript II first-strand synthesis system (Invitrogen). Primers were designed using Primer3 software (http://frodo.wi.mit.edu/primer3/). Primer combinations showing a minimum amplification efficiency of 85% were used in real-time RT-PCR experiments (Table S1), and specificity was checked using a dissociation curve. Real-time RT-CRs were performed on a Roche LightCycler480 using Roche reagents (http://www.roche.com). Two independent biological experiments and technical replicates (based on two independent cDNA syntheses derived from the same RNA sample) were performed in all cases. Three genes were defined as reference genes using GeNorm software (http://medgen.ugent.be/~jvdesomp/genorm/): MtACT (Actin 12), MtH3L (Histone 3-Like) and MtRBP1 (RNA Binding Protein 1). Gene fold changes were calculated using the mean expression of the three genes calculated using LightCycler480 software. The experimental control conditions were used to determine expression fold changes.

In situ hybridization

In situ hybridizations were performed as described by Boualem et al. (2008) on 21 dpi nodules using an Intavis InsituPro automat (http://www.intavis.com/en/). RNA probes were marked by digoxygenin (DIG) and revealed using an antibody coupled to alkaline phosphatase. An antisense RNA probe corresponding to a carbonic anhydrase gene (MtCA1) was included as a positive control (Boualem et al., 2008). An antisense probe designed to target specifically MtNAC969 transcripts (i.e. with <70% identity with other known Medicago genes) was designed in addition to a corresponding sense probe as a negative control. Primers used to generate MtNAC969 probes are listed in Table S1.

Bioinformatic and phylogenetic analyses

The amino acid sequences of Medicago NAC genes were obtained based on the M. truncatula Gene Index database (MtGI version 8, http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/gimain.pl?gudb=medicago). To construct the phylogenetic tree, the three NAC TFs previously identified as salt-induced in roots (Gruber et al., 2009) [MtNAC969 (TC96130), MtNAC1081 (TC80936) and MtNAC1126 (NP7267109)] were included, as well as TC108627, the Medicago sequence most closely related to MtNAC969 (based on a nucleotide BLAST search on MtGI version 8). Correspondence between various Medicago resources (mainly MtGI, MtGEA, 16K+ arrays and the Medicago genome) was established using the Legoo ‘nickname’ tool (http://www.legoo.org/). Several A. thaliana NAC proteins functionally characterized in relation to abiotic stresses or most closely related to MtNAC969 were also included: AtNAC2 (or ORE1, Oresara1/anac092/At5g39610), AtNAP (NAC-like activated by APETALA3/PISTILLATA, anac029/At1g69490), AtATAF1 (anac002/At1g01720), ATAF2 (anac081/At5g08790), AtNAM (No Apical Meristem, anac018/At1g52880), TIP (TCV-interacting protein; anac091/At5g24590), ANAC055 (At3g15500), CUC1 (CUP-SHAPED COTYLEDON1; anac054/At3g15170), CUC2 (anac098/At5g53950), CUC3 (anac031/At1g76420), NAC1 (anac022/At1g56010) and NAC2 (anac078/At5g04410). Two rice NAC peptide sequences functionally characterized in relation to abiotic stress were also used for this phylogenetic tree: OsNAC6 (onac048/AK068392) and OsNAC10 (AK069257).

Full-length amino acid sequences of selected NAC proteins were aligned using ClustalW (http://clustalw.genome.ad.jp/) with default parameter values, and analyzed using MEGA4 software version 4.0.2 (http://www.megasoftware.net/). The evolutionary history was inferred using the neighbor-joining method. The tree was drawn to scale, with branch lengths in the same units as those for the evolutionary distances used to create the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method, and are in the units of the number of amino acid substitutions per site. All positions containing gaps and missing data were eliminated from the dataset.

Histological analysis of nodules

For amyloplast detection, 21-day-old nodules were collected from composite plants silenced for MtNAC969 or GUS genes and embedded in 6% agarose (Eurobio, http://www.eurobio.fr/). Sections (100 μm) of nodules cut using a Vibratome VT1200S (Leica, http://www.leica.com/) were directly stained with lugol (Fluka, http://www.sigmaaldrich.com/) and observed using a Reichert Polyvar microscope equipped (http://www.reichert.com/) with a Nikon digital DXM1200 camera (http://www.nikon.com/).

For transmission electronic microscopy, 14 dpi nodules were processed as described by Laporte et al. (2010) using conventional methods described by Hawes and Satiat-Jeunemaitre (2001), except that the osmium post-fixation step was replaced by a mixture of 1% osmium and 1.5% potassium ferrocyanide (Aubert et al., 2011). Specimens were embedded in epoxy resin (agar low-viscosity premix kit medium, Oxford Instruments, http://www.oxinst.com/) and polymerized for 16 h at 60°C. Thick sections (300 nm), obtained using a Leica Ultracut UC6 ultramicrotome, were collected on a glass slide and stained with toluidine blue for 1 min. Samples were observed using a Nikon AZ100 macroscope. Ultrathin sections (70–90 nm) were counterstained using aqueous 2% uranyl acetate and lead citrate (Hawes and Satiat-Jeunemaitre, 2001). Grids were examined using a JEOL (http://www.jeol.com/) 1400 transmission electronic microscope operating at 120 kV. Images were acquired using a post-column high-resolution (11 megapixels) camera (SC1000 Orius, Gatan Inc., http://www.gatan.com/).

Acknowledgments

We thank Julie Plet and Philippe Laporte for providing cDNAs, Alexis Canette for his technical assistance in electron microscopy, Emmanuelle Dheilly for help with real-time RT-PCR experiments, and Fabien Marcel for help with composite plant experiments. This work used the imaging facilities of the Imagif platform (Cell Biology Unit, Centre de Recherche de Gif sur Yvette, France; http://www.imagif.cnrs.fr), supported by the Conseil Général de l’Essonne. O.Z. was supported by a grant from the Tunisian Government (Tunis). This work was partly supported by LEGUROOT, a French Agence Nationale de la Recherche project.

Accession number: The NCBI accession number for the MtNAC969 sequence is JN833713.

Ancillary