miR396 affects mycorrhization and root meristem activity in the legume Medicago truncatula

Authors

  • Jérémie Bazin,

    1. Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    2. Université Paris Diderot, Sorbonne Paris Cité, Paris 13, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Ghazanfar Abbas Khan,

    1. Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Jean-Philippe Combier,

    1. Université de Toulouse, Université Paul Sabatier, UMR5546, Laboratoire de Recherche en Sciences Végétales, Castanet-Tolosan, France
    2. Centre National de la Recherche Scientifique, UMR5546, Castanet-Tolosan, France
    Search for more papers by this author
  • Pilar Bustos-Sanmamed,

    1. Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    Search for more papers by this author
  • Juan Manuel Debernardi,

    1. Instituto de Biología Molecular y Celular de Rosario, Ocampo y Esmeralda, Universidad nacional de Rosario, Rosario, Argentina
    Search for more papers by this author
  • Ramiro Rodriguez,

    1. Instituto de Biología Molecular y Celular de Rosario, Ocampo y Esmeralda, Universidad nacional de Rosario, Rosario, Argentina
    Search for more papers by this author
  • Céline Sorin,

    1. Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    2. Université Paris Diderot, Sorbonne Paris Cité, Paris 13, France
    Search for more papers by this author
  • Javier Palatnik,

    1. Instituto de Biología Molecular y Celular de Rosario, Ocampo y Esmeralda, Universidad nacional de Rosario, Rosario, Argentina
    Search for more papers by this author
  • Caroline Hartmann,

    1. Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    2. Université Paris Diderot, Sorbonne Paris Cité, Paris 13, France
    Search for more papers by this author
  • Martin Crespi,

    Corresponding author
    • Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    Search for more papers by this author
  • Christine Lelandais-Brière

    1. Institut des Sciences du Végétal, Centre National de la Recherche Scientifique, Gif-sur-Yvette, France
    2. Université Paris Diderot, Sorbonne Paris Cité, Paris 13, France
    Search for more papers by this author

For correspondence (email crespi@isv.cnrs-gif.fr).

Summary

The root system is crucial for acquisition of resources from the soil. In legumes, the efficiency of mineral and water uptake by the roots may be reinforced due to establishment of symbiotic relationships with mycorrhizal fungi and interactions with soil rhizobia. Here, we investigated the role of miR396 in regulating the architecture of the root system and in symbiotic interactions in the model legume Medicago truncatula. Analyses with promoter–GUS fusions suggested that the mtr-miR396a and miR396b genes are highly expressed in root tips, preferentially in the transition zone, and display distinct expression profiles during lateral root and nodule development. Transgenic roots of composite plants that over-express the miR396b precursor showed lower expression of six growth-regulating factor genes (MtGRF) and two bHLH79-like target genes, as well as reduced growth and mycorrhizal associations. miR396 inactivation by mimicry caused contrasting tendencies, with increased target expression, higher root biomass and more efficient colonization by arbuscular mycorrhizal fungi. In contrast to MtbHLH79, repression of three GRF targets by RNA interference severely impaired root growth. Early activation of mtr-miR396b, concomitant with post-transcriptional repression of MtGRF5 expression, was also observed in response to exogenous brassinosteroids. Growth limitation in miR396 over-expressing roots correlated with a reduction in cell-cycle gene expression and the number of dividing cells in the root apical meristem. These results link the miR396 network to the regulation of root growth and mycorrhizal associations in plants.

Introduction

In plants, root growth is controlled via coordinated cell division and expansion at the root apex, where three regions are observed: (i) the meristematic zone, which is called the root apical meristem (RAM) and is protected by the root cap and characterized by cell proliferation, (ii) the elongation zone, and (iii) the differentiation zone, where cells differentiate into a radial pattern. The RAM contains a small group of cells, called the quiescent centre (QC), that are less mitotically active and maintain the surrounding initial cells in an undifferentiated state. Initial cells and their daughter cells actively divide to give rise to each of the root cell files (Bennett and Scheres, 2010). Root development is regulated by complex interactions of hormone signalling pathways (Benková and Hejátko, 2009; Péret et al., 2009), in which auxin plays a central role (Overvoorde et al., 2010). However, other hormones, such as cytokinins, ethylene, gibberellins, brassinosteroids (BR) and abscisic acid, also participate in regulation of the RAM (Galinha et al., 2009; Benková and Hejátko, 2009; Bennett and Scheres, 2010; Perilli et al., 2012). In recent years, microRNAs (miRNAs) have emerged as important regulators of the architecture of the root system (Simon et al., 2009; Khan et al., 2011). For example, in Arabidopsis thaliana, miR165/166 governs the radial patterning of vascular and pericycle tissues (Carlsbecker et al., 2010) via quantitative repression of HD-ZIP III transcription factors (TFs). A set of auxin-related miRNAs, including miR160, also regulate lateral root initiation and/or root growth (Wang et al., 2005; Khan et al., 2011). However, few miRNAs have been reported to play roles in RAM function (Wang et al., 2011a).

Under nutrient starvation conditions, legumes may establish symbiotic interactions with soil arbuscular mycorrhizal (AM) fungi or rhizobial bacteria. Both interactions assist the plant host to cope with adverse conditions. Beneficial associations with rhizobia lead to production of nitrogen-fixing nodules (Crespi and Frugier, 2008; Desbrosses and Stougaard, 2011). During mycorrhizal associations, both the internal arbuscules formed inside the roots and external hyphae improve exploration of the soil environment and the uptake of nutrients by the roots (Parniske, 2008). The nodulation process appears to have evolved from both lateral root organogenesis and fungal mycorrhization. Indeed, recent studies have revealed that components of the NOD factor signalling pathway are required for nodulation and mycorrhization (Gough and Cullimore, 2011). Although large sets of mycorrhization or nodulation-responsive miRNAs have been identified in legumes (Li et al., 2008; Subramanian et al., 2008; Lelandais-Brière et al., 2009; Devers et al., 2011), few have been functionally linked to these processes. In the model legume Medicago truncatula, miR166 and miR164 participate in the control of lateral root and/or nodule organogenesis by repressing HD-ZIP III and NAC1 TFs, respectively (Boualem et al., 2008; D'haeseleer et al., 2011). In soybean (Glycine max), over-expression of gma-miR482, gma-miR1512 and gma-miR1515 increased nodule density (Li et al., 2010). mtr-miR169 has been reported to repress the nodulation-responsive HAP2 TF in M. truncatula (Combier et al., 2006), and lja-miR397 has been reported to repress laccases, which play roles in copper homeostasis, in Lotus japonicus nodules (De Luis et al., 2012). Recently, mtr-miR171h, which targets the NSP2 GRAS TF, was shown to control both the AM fungal colonization of M. truncatula roots (Lauressergues et al., 2012) and rhizobial symbiosis in L. japonicus (De Luis et al., 2012).

To identify miRNAs involved in root architecture and symbiotic associations in M. truncatula, we analysed small RNAs from root apexes and nodules (Lelandais-Brière et al., 2009). Interestingly, miR396 accumulated to a very high level in root tips. In A. thaliana, this miRNA, which is encoded by two loci, post-transcriptionally represses several members of the growth-regulating factor (GRF) family. These plant-specific TFs control the growth and development of leaves and stems (Van der Knaap et al., 2000; Kim et al., 2003). In A. thaliana, where single mutants have no obvious phenotype, the triple null mutant grf1 grf2 grf3 develops smaller leaves and cotyledons. Recently, it has been shown that miR396 plays a critical role in leaf-cell proliferation, by controlling the switch between the mitotic cycle and the endocycle, in addition to adaxial–abaxial polarity and shoot meristem size in A. thaliana (Rodriguez et al., 2010; Wang et al., 2011b; Mecchia et al., 2013). Furthermore, miR396 may play contrasting roles in tolerance to abiotic stress in different species (Liu et al., 2009; Gao et al., 2010). Although the regulation of GRFs by miR396 is conserved in angiosperms and gymnosperms, miR396 may target other genes in certain species (Debernardi et al., 2012; Chorostecki et al., 2012). Thus, miR396 may have evolved to play roles in various regulatory networks in plants.

In this paper, we describe the roles of the miR396 regulatory network in M. truncatula roots. Over-expression or inactivation of miR396 in transgenic roots de-regulates not only the expression of MtGRF genes but also of two targets encoding homologues of AtbHLH79 TFs. Concomitantly, mycorrhizal associations, root growth, the number of replicating cells and the expression of cell-cycle genes were affected, all further linking miR396 to RAM activity.

Results

Mtr-miR396a and mtr-miR396b genes show different tissue-specific expression profiles in Medicago truncatula roots

In M. truncatula, two miR396 genes (Mt genome version 3.5.1: http://medicagohapmap.org/; mtr-miR396a: chromosome 4, nucleotides 39 789 612–39 789 704; mtr-miR396b: chromosome 4, nucleotides 39 798 253–39 798 347) produce pre-miRNAs of 93 and 95 nt, respectively (miRBAse version 18, November 2011). These precursors give rise to various miRNA variants or isomiRs that differ by one nucleotide at their 3' termini (Figure 1). An mtr-miR396c gene was registered in miRBAse version 19 (August 2012), but we did not consider it due to the high divergence of its mature miRNA sequence from the other two isomiRs. Quantitative RT-PCR experiments showed that the mtr-miR396a and mtr-miR396b genes were expressed in all organs tested, and the highest expression levels were found in leaves. mtr-miR396b was found to be more abundant than mtr-miR396a in roots, nodules, leaves and pods, but not in flowers (Figure S1). In addition, miR396a was enriched in root tips compared with whole roots (Figure 1). To analyse their tissue-specific expression in more detail, we fused the uidA (GUS) gene to the upstream genomic regions of miR396a and miR396b precursors (1947 and 1938 bp, respectively), and introduced these constructs into M. truncatula roots (Boisson-Dernier et al., 2001). Both promoters drove GUS expression in the stele of mature roots, with higher expression observed for mtr-miR396b (Figure 2a,f). Transverse sections revealed that mtr-miR396a was expressed primarily in the pericycle and the protoxylem surrounding cells (Figure 2c), whereas prom-miR396b:GUS staining was more diffuse and was found in the parenchymal cells that surround vascular tissues (Figure 2h). In the root tips, both promoters showed highest activity in the transition zone. However, mtr-miR396a expression increased progressively from the proximal meristem to the transition zone (Figure 2b), whereas mtr-miR396b promoter activity was restricted to the transition and elongation zones and the root cap columella (Figure 2g). Unlike miR396a (Figure 2d), prom-miR396b:GUS activity was detected in lateral root primordia (Figure 2i). At later stages of lateral root organogenesis, expression of both miRNAs returned to patterns similar to those of primary roots (Figure 2e,j).

Figure 1.

In M. truncatula, six MtGRF and two bHLH79 genes have miR396 binding sites. Schematic representations of the MtGRF and MtbHLH79 genes, showing their respective conserved domains in boxes and the position of the miR396 recognition sites, as well as an alignment of mtr-miR396a and miR396b isoforms with their targets or a non-target transcript (i.e. MtbHLH74). Penalty scores were calculated as described Jones-Rhoades and Bartel (2004) with scores of 2 per gap (hyphen), 1 per mismatch (asterisk) and 0.5 per GU pair (circle).

Figure 2.

The two mtr-miR396 genes show distinct expression patterns in roots. Expression of mtr-miR396-a (a–e) and mtr-miR396b (f–j) was analysed in 3-week-old transgenic roots using promoter–GUS analyses. (b,g) Magnified images of the roots shown in (a) and (f). Arrows indicate the position of the transition zone. (c,h) Root cross-sections, showing pericycle cells (Pc), xylem (Xy), phloem (Ph), endodermis (Ed) and cortex (Co). (d,i) Lateral root primordia. (e,j) Later stages of lateral root organogenesis. (k) Longitudinal optical section of a typical root apex stained using a modified pseudo-Schiff propidium iodide protocol, showing the position of the elongation/differentiation zone (EDZ), transition zone (TZ) and meristematic zone (MZ) divided into the basal (BM) and proximal (PM) meristems, quiescent centre (QC) and root cap (RC). Scale bars = 1 mm (a–f), 100 μm (b,g,k), 20 μm (c,h) and 50 μm (d,e,i,j).

In M. truncatula roots, miR396 regulates six GRF and two bHLH79 transcription factors

The post-transcriptional repression of GRF expression by miR396 is a conserved mechanism in A. thaliana and rice (Rodriguez et al., 2010; Gao et al., 2010). In M. truncatula, Devers et al. (2011) performed degradome sequencing, a technology that enables detection of cleavage products of miRNA targets, in roots infected or not infected with the AM fungus Rhizophagus irregularis. Among these data, we found reads corresponding to the miR396-related cleavage products of six GRF mRNAs (MtGRF1, MtGi10-TC183867; MtGRF2, MtGi10-TC183494; MtGRF3, MtGi10-BG454006; MtGRF4, Medtr5g027250; MtGRF5, Medtr8g020560; MtGRF6, Medtr7g126820). These transcripts possess a putative miR396 binding site (Figure 1), and the encoded proteins all contain the conserved domains of GRFs (Figure S2) (Van der Knaap et al., 2000; Kim et al., 2003). Mapping of the degradome reads from Devers et al. (2011) on MtGRF transcripts confirmed perfect alignment at the predicted miR396 cleavage position (Figure S3). No additional GRF genes were found in the M. truncatula genome version 3.5.1 (http://medicagohapmap.org/ and http://www.legoo.org/) or EST libraries (MtGi-11, http://compbio.dfci.harvard.edu/tgi/).

We also observed abundant degradome reads corresponding to specific cleavage of two non-GRF transcripts at miR396 binding sites (Medtr5 g038250.1 and TC127247/Medtr8 g074580, Figure S3). Based on tBLASTX analysis, the 2 predicted proteins showed high similarity (43 and 46% respectively) to AtbHLH79/BIGPETAL (At1g59640), a TF involved in the control of petal growth in A. thaliana (Szécsi et al., 2006). Recently, Debernardi et al. (2012) reported AtbHLH74 (At1g10120) as a non-conserved target of miR396 in Brassicales. However, the M. truncatula bHLH74 homologue (Medtr8g065740) had no complementary sequence for miR396 (Figure 1). When we searched for conservation of the miR396/bHLH79 node among land plants (tBLASTX, http://blast.ncbi.nlm.nih.gov/BLAST.cgi; Table S1), complementary sites for miR396 were present systematically in species of the Fabidae taxon, including Fabaceae (rosid I), but not in other dicots (Malvidae and asterids taxa) or in monocots.

Characterization of the miR396 regulatory network in M. truncatula

To investigate miR396-mediated gene regulation in planta, we over-expressed mtr-miR396a and mtr-miR396b precursors in M. truncatula roots (miR396-OE). Quantitative RT-PCR, performed on pools of 20 independent transgenic roots, confirmed over-accumulation of both precursors and mature miRNAs (Figure 3a). Concomitantly, MtGRF mRNA levels were two- to five-fold reduced, suggesting that miR396 may repress these genes under our experimental conditions (Figure 3b). MtbHLH79a and b transcript levels were also decreased in miR396-OE roots, with changes of 2- and 2.5-fold, respectively (Figure 3b). In contrast, transcripts levels of six additional in silico-predicted targets of miR396 remained unchanged (Figure S4). To investigate the specificity of miR396 isoforms toward MtGRF and MtbHLH79 transcripts, we also measured the expression of MtGRF and MtbHLH79 genes in transgenic roots over-expressing the mtr-miR396a isoform, which originates from a different precursor. All transcripts were repressed at levels similar to roots over-expressing mtr-miR396b (Figure 3b), suggesting that there was no target specificity between the two miR396 isoforms. To further validate their targets, we generated roots expressing a target mimicry construct that was successfully designed to hinder miR396 activity in A. thaliana (Debernardi et al., 2012). Mimicry (MIM) transcripts specifically trap members of an miRNA family, and thus prevent their activity on endogenous targets (Franco-Zorrilla et al., 2007; Todesco et al., 2010). In MIM396 roots, all MtGRF genes, except MtGRF6, and the two MtbHLH79 genes were up-regulated with changes from 1.5–8-fold (Figure 3c). As in the miR396-OE roots, no difference was observed for the six other predicted targets tested (Figure S4B). Quantitative RT-PCR experiments on organs (Figure S5A,B), together with transcriptomics data obtained using the Affymetrix Medicago GeneChip (MtGEA, http://mtgea.noble.org/v2/) (Figure S5C) revealed that MtGRF genes were robustly expressed in flowers, vegetative buds and root tips. The majority of MtGRF transcripts, in particular MtGRF2, were enriched in 3 mm root tips compared with 1 cm root tips or whole roots (Figure S5C). bHLH79 mRNA levels were highest in vegetative buds (Figure S5B) and varied in the other organs (Figure S5B,C). In addition, the Affymetrix data revealed that both bHLH79 transcripts were also enriched in 3 mm root tips compared with 1 cm root tips and whole roots (Figure S5C). In addition, we constructed promoter:GUS transcriptional fusions for MtGRF4, MtGRF5 and MtGRF6 genes (upstream regions of 1978, 2070 and 1936 bp, respectively). Genomic regions corresponding to other MtGRF genes are not present in the M. truncatula genome version 3.5.1 (http://medicagohapmap.org). For the three GRF genes, robust GUS staining was noted in the proximal meristem (Figure S6) where active cell proliferation occurs.

Figure 3.

mtr-miR396 represses six MtGRF genes and two MtbHLH79 genes. (a) Accumulation of pre-miR396a, pre-miR396b and mature miR396 was analysed by quantitative RT-PCR in 3-week-old roots expressing the 35S:pre-miR396a or miR396b constructs (miR396a-OE and miR396b-OE, respectively). (b,c) Levels of transcripts of miR396 targets were measured using quantitative RT-PCR in miR396-OE roots (b) and in roots inactivated for miR396 activity (MIM396) (c). Three pools of 20 independent transgenic roots per construct were used to prepare RNA, and two cDNA syntheses (technical repeats) were performed from each pool. Ct values from each pair of technical replicates were averaged, and error bars represent the standard deviation of three biological experiments (pools of transgenic roots). Control roots were generated using the appropriate empty binary vectors for transformation.

miR396 negatively affects root growth, RAM size and cell proliferation in the root apex

To investigate the function of miR396 in roots, we measured the length and dry weight of roots that either over-expressed the mtr-miR396b precursor (miR396-OE) or were inactivated for miR396 activity (MIM396). A significant reduction in primary root length and dry weight was observed in miR396-OE roots compared with control roots transformed using an empty vector (Figure 4a,b). As expected, a similar reduction in root length was observed in mtr-miR396a over-expressing roots, as the two variants regulate the same targets (Figure S7). In turn, MIM396 roots exhibited a slight increase in root length (mean of 15% longer than controls, Figure 4a) whereas root biomass increased significantly (2.5-fold higher than the corresponding controls, Figure 4b). Staining of miR396b-OE and MIM396 roots with propidium iodide using an adapted protocol (see 'Experimental procedures') revealed no differences in global RAM organization or tissue radial patterning (Figure 5a) or in the length of differentiated cortical cells (Figure 5c,e). The RAM length (estimated by the cell number of a given cortical file, from the quiescent centre to the first elongating cell) was significantly reduced (37.5%) in miR396-OE and slightly enhanced (mean 15%) in MIM396 roots when compared with their respective controls (Figure 5b,d). Flow cytometric analysis was performed on nuclei extracted from 5 mm tips of roots expressing the various constructs. No significant change in the level of endo-reduplication was observed when either the proportion of 2C to 16C cells (Figure 6a) or expression of MtCCS52B (Figure 6b), which encodes a mitotic inhibitor required for endo-reduplication (Cebolla et al., 1999), was examined. Thus, the switch between the mitotic cycle and the endocycle may not be regulated by miR396 in roots, at least in M. truncatula.

Figure 4.

Over-expression of miR396 restricts root length. (a,b) The mean primary root length (a) and the weight of the complete root system (b) of composite plants that over-accumulated mtr-miR396b (miR396-OE, black), were inactivated for miR396 (MIM396, grey) or the corresponding controls (white) were measured at 3 weeks after inoculation with A. rhizogenes of at least 22 independent transgenic roots per construct in three experiments. Error bars indicate the standard deviation of the three independent experiments. Statistically significant differences compared with controls were determined using a Kruskal–Wallis test (**P < 0.01). (c) Representative images of two miR396-OE or MIM396 composite plants and their respective controls (control roots express the appropriate empty vector).

Figure 5.

Root meristem size is reduced in miR396-OE composite plants. (a) miR396-OE and MIM396 roots and the corresponding controls were stained using a modified pseudo-Schiff propidium iodide staining protocol. Scale bars = 100 μm. The arrowheads indicate the position of the transition zone. A magnified transition zone is shown for each root apex. (b–e) The meristem height (cell number in a cortex cell file from the quiescent centre to the elongation zone) (b,d) and the differentiated cortex cell length were measured (c,e) in at least 30 roots per construct. Error bars indicate the standard deviation of meristematic cell numbers and differentiated cortex cell length of two independent experiments. Statistically significant differences compared with controls were determined using Student's t test (**P < 0.01; n = 60).

Figure 6.

miR396 restricts cell proliferation activity in the RAM. (a) The distribution of nuclear ploidy level was measured in 5 mm root tips of 2-week-old roots that expressed the mir396-OE construct, the MIM396 constructs or their respective controls. Following DAPI staining, nuclei were extracted and analysed using flow cytometry. The percentages of cells with different nuclear ploidy levels in 20 roots per construct in three independent experiments are shown. (b) The expression of mitotic cell cycle (CYCB1;1, CYCB1;3, CYCB2;1, CYCD3;1, histone H4, CDC16) or endocycle (CCS52B) markers was measured using quantitative RT-PCR in miR396-OE and control roots. The transcript levels were normalized to values obtained in control roots. Two pools of 20 independent transgenic roots per construct were used to prepare RNA, and two cDNA syntheses (technical repeats) were performed from each pool. Ct values from each pair of technical replicates were averaged, and error bars represent the standard deviation of two biological experiments (pools of transgenic roots). (c,d) Roots expressing the various constructs were incubated with 10 mm EdU for 4 h to label replicating cells. Nuclei from 5 mm root tips were then extracted, double-labelled with Alexa Fluor 647 azide and DAPI, and then analysed using flow cytometry. (c) Biparametric dot-plot analysis of Alexa Fluor 647 EdU intensity [EdU-Alexa647 intensity (log values)] versus DAPI intensity [DAPI-Area (area of DAPi-stained nuclei)]. Lines in the biparametric plots indicate the EdU intensity threshold used to determine the EdU-labelled population. (d) The percentage of cells replicating (i.e. EdU-positive nuclei) obtained from 20 root tips per construct in four independent experiments. Error bars indicate the standard deviation of the four independent experiments. Statistically significant differences compared with controls were determined using a Kruskal–Wallis test (*< 0.05; **P < 0.01).

The impact of miR396 de-regulation on cell proliferation was further investigated using two approaches. First, we determined the expression levels of genes homologous to several currently used markers of the cell cycle. Homologues of A. thaliana H4, CYCB1;1, CYCB1;3, CYCB2;1 and CYCD3;1 genes were identified in M. truncatula genomic and EST databases (Medtr7g106110.1, Medtr8g038460, Medtr7g106110, Medtr5g023790 and Favery et al., 2002; respectively). Quantitative RT-PCR experiments showed that expression of the cell-cycle markers was reduced by approximately two-fold in miR396-OE roots compared with the corresponding controls (Figure 6b). A similar reduction was observed for MtCDC16, a gene that encodes a core component of the anaphase promoting complex, which controls lateral root and nodule number and root length in M. truncatula (Kuppusamy et al., 2009). Second, we estimated the number of replicating cells in the RAM using 5-ethynyl-2'-deoxyuridine (EdU) staining combined with flow cytometry to determine quantitative changes in replicating cells from root apices. These experiments revealed that the percentage of replicating cells (EdU-positive) was significantly lower in miR396-OE roots (6.41%) and higher in MIM396 roots (18.96%) when compared with their appropriate controls (14.3% and 11.6% respectively, Figure 6c,d). Together, these data strongly suggest that miR396-dependent modulation of root growth may essentially be due to a restriction of cell division activity. The increase in expression of miR396 genes in root tips from the proximal meristem to the transition zone where cell division gradually stops is consistent with these results.

Silencing of MtGRF genes mimics the miR396-OE root growth phenotype

To study the role of the miR396 targets on root growth, we inactivated them using an RNAi strategy. First, we generated composite plants expressing three RNAi constructs targeting a conserved region of various GRF genes (Figure S8A, MtGRF2, MtGRF4 and MtGRF6). Quantitative RT-PCR analysis of these transgenic roots confirmed that MtGRF2, MtGRF4 and MtGRF6 mRNA levels were reduced by approximately tenfold by the corresponding RNAi construct (Figure S8C). We also observed additional, but lower, repression of other MtGRF genes, but expression of MtbHLH79 remained unaffected (Figure S8C). Similar to miR396 over-expression, a significant reduction in root length and weight was observed following expression of the three GRF RNAi constructs (Figure 7a,b). In addition, we prepared an RNAi vector targeting both MtbHLH79 genes (Figure S8B) and confirmed their efficient silencing, whereas expression of MtGRF genes remained unchanged (Figure S8C). These plants showed wild-type-like roots, suggesting that the phenotypes observed in miR396-OE and MIM396 roots are most likely due to de-regulation of GRF genes (Figure 7).

Figure 7.

Silencing of MtGRF genes mimics miR396 over-expression in roots. The length (a) and dry weight (b) of roots expressing RNAi constructs that target primarily MtGRF2 (iGRF2), MtGRF4 (iGRF4), MtGRF6 (iGRF6) or MtbHLH79a (ibHLH79) were measured 3 weeks after inoculation of A. rhizogenes in three independent experiments (n > 22 transgenic roots per construct in each experiment). Values were reported to roots transformed with the RNAi empty vector. Error bars indicate the standard deviation of all values obtained in three experiments. Statistically significant differences compared with controls were determined using a Kruskal–Wallis test (**P < 0.01; ***P < 0.001).

Hormonal control of miR396 expression

RAM activity is known to be regulated by plant hormones. Therefore, we explored their putative link to miR396. We treated wild-type plants for 1, 3 or 5 h with auxin, cytokinin, gibberellin, abscisic acid or brassinosteroids (BR), and measured pre-miR396 and target mRNA levels in roots. Gibberellin, cytokinin and auxin treatments did not affect expression of any of these genes (Figure S9A–C). Expression of pre-miR396b was transiently reinforced by exogenous abscisic acid, and peaked at 3.3 after 1 h of treatment. However, neither MtGRF nor MtbHLH79 transcript levels were affected by this hormone (Figure S9D). In contrast, pre-miR396b levels increased between 1 and 5 h on treatment with BR at a concentration of 1 nM (mean fold ratio between 2.2 and 3.4), whereas this effect was not observed on pre-miR396a or other non-related miRNA precursors (Figure 8a). However, quantitative RT-PCR experiments that did not discriminate between miR396a and miR396b isomiRs did not reveal any significant change in global accumulation of miR396. Down-regulation of MtGRF5 was observed (Figure 8a). Activation of mtr-miR396b promoter activity by BR was confirmed by quantifying GUS transcript levels in the prom-miR396b:GUS roots (Figure 8b), although no change in its spatial expression profile was detected in these roots. Interestingly, no significant reduction in GUS expression was observed in prom-GRF5:GUS roots treated with BR (Figure 8b), suggesting that repression of MtGRF5 by this hormone may be primarily post-transcriptional. These results suggest a potential link between BR and miR396, which should be explored in future experiments.

Figure 8.

Mtr-miR396b and MtGRF5 are regulated by exogenous brassinosteroids. (a) The expression of mtr-miR396a (396a), mtr-miR396b (396b) precursors, mature miR396 (miR396), target genes (MtGRF1–6 and MtbHLH79a and b) and other non-related miRNA precursors (miR167c, miR390a and miR2658) was analysed using quantitative RT-PCR in roots treated for 1, 3 or 5 h with epibrassinolide (BR) at a concentration of 1 nm. Expression was normalized to the values of untreated roots. Three references genes (defined using Genorm software, see 'Experimental procedures') were used for normalization. (b) Levels of GUS transcript in roots expressing the prom-miR396b:GUS construct (grey bars) or the prom-GRF5:GUS construct (black bars) treated or not treated with 1 nm BR for 3 h were measured using quantitative RT-PCR. For each construct, three pools of 20 independent transgenic roots were used to prepare the RNA, and two cDNA syntheses were performed from each pool (technical replicates). Ct values from each pair of technical replicates were averaged, and the error bars indicate the standard deviation of the three biological replicates (pools of transgenic roots).

miR396 limits mycorrhizal colonization but not nodulation

We next decided to study the potential role of miR396 in root symbioses. Transcriptomics data from the Affymetrix Medicago GeneChip revealed that the MtGRF2 and MtbHLH79b genes were down-regulated during the nodulation process (Figure S10). We thus analysed mtr-miR396a and mtr-miR396b expression profiles in roots inoculated with the symbiotic bacteria Sinorhizobium meliloti. mtr-miR396a promoter activity was primarily detected in nodule vascular tissues, although weak GUS staining was also observed in the nitrogen-fixing region (zone III, Figure S11A). GUS activity, driven by the mtr-miR396b promoter (Figure S11B), was high in vascular tissues, the infection zone (zone II) and the upper parts of the fixation zone (zone III). The promoter activities of MtGRF4, MtGRF5 and MtGRF6 were also analysed in mature nodules (Figure S12). MtGRF4 and MtGRF5 expression patterns partly overlapped with that of miR396 in nodule vascular tissues (Figure S12A,B). With respect to the three GRF genes, GUS staining was observed in the meristem. However, neither miR396 over-expression nor its inactivation by mimicry resulted in any major modification in nodule density, morphology or cellular organization at 5 weeks after inoculation with S. meliloti (Figure S11C–E).

We also analysed the expression of miR396 in response to Myc-LCOs, the lipochitooligosaccharide signal molecules produced by AM fungi. Quantitative RT-PCR analyses revealed that expression of mtr-miR396a decreased (twofold ratio) whereas MtGRF4 expression increased (2.8-fold change) upon treatment with Myc-LCOs (Figure S13) (Maillet et al., 2011). The involvement of miR396 during mycorrhization was then analysed at 5 weeks after colonization with R. irregularis (formerly Glomus intraradices) in miR396-OE and MIM396 roots. Interestingly, although arbuscular structures were not affected by mtr-miR396b over-expression or inactivation (Figure 9a,b), when using the Mycocalc method (http://www2.dijon.inra.fr/mychintec/Mycocalc-prg/download.html; Trouvelot et al., 1986) to evaluate overall root colonization by AM fungi, we observed that roots over-expressing miR396b were significantly less colonized than control roots (64% in controls versus 38% in test roots; Figure 9b). In contrast, roots expressing the MIM396 construct were significantly more colonized than control roots (64% in controls versus 80% in test roots; Figure 9b). In both cases, the roots showed normal arbuscule abundance and morphology (Figure 9a,b). Hence, the action of miR396 limits mycorrhizal colonization.

Figure 9.

miR396 affects fungal colonization during AM symbiosis. Mycorrhizal structures (a) and quantification of mycorrhization ability (b) were analysed at 5 weeks after inoculation with Rhizophagus irregularis spores in control (empty vector), miR 396-over-expressing (miR396-OE) and MIM396 transgenic roots as described by Trouvelot et al. (1986). ‘F’ indicates the frequency of colonization of the root system, ‘a’ indicates the arbuscule abundance (percentage) of the colonized root sections. For each construct, n = 30 independent transgenic roots were used; error bars indicate the standard error of the mean. Asterisks indicate statistically significant differences compared to controls based on Kruskal–Wallis tests (**P < 0.01). Scale bars = 100 μm.

Discussion

miR396 is one of the most conserved miRNAs among land plants, and has been reported to mediate post-transcriptional repression of genes encoding transcription factors of the GRF family. GRF targets with a miR396 binding site have been detected in all angiosperms and gymnosperms studied so far (Debernardi et al., 2012). In A. thaliana, GRFs are primarily involved in the control of cell proliferation during leaf development (Kim et al., 2003). Here, we showed that, in the model legume M. truncatula, miR396 represses not only MtGRF but also two MtbHLH79 TF-encoding genes, and that its de-regulation affects root growth and the ability to be colonized by mycorrhizal fungi.

In M. truncatula, we identified six GRF-encoding genes, all of which contained a miR396 binding site and for which cleavage was validated in roots using degradome analysis (Devers et al., 2011). We showed that over-expression of both miR396 isoforms in transgenic roots led to down-regulation of the MtGRF transcripts, although different efficiencies were observed depending on the target. A similar variability has already been reported in A. thaliana leaves (Liu et al., 2009). A mimicry construct designed to trap both miR396 variants (Debernardi et al., 2012) increased the accumulation of all MtGRF mRNAs in roots except MtGRF6. We cannot rule out the possibility that additional factors regulate MtGRF6, because, in A. thaliana, AtGRF6 mRNA levels were reduced following miR396 over-expression although it does not contain a miRNA target site (Rodriguez et al., 2010).

The miR396–MtGRF regulatory node thus appears more complex than expected, and it is likely that specific spatio-temporal expression of the various miRNAs and target genes underlies differential regulation. Reinforcing this idea, quantitative RT-PCR and promoter–GUS analyses revealed that not only the spatio-temporal expression but also the hormonal regulation of mtr-miR396a and mtr-miR396b genes differed. For example, unlike mtr-miR396a, mtr-miR396b was expressed in lateral root primordia. Moreover, mtr-miR396a was unaffected by hormonal treatments, whereas mtr-miR396b responded rapidly to both abscisic acid and BR. These data suggest that the miR396a variant may modulate the levels of GRF transcripts in root apexes to enable appropriate growth under normal conditions, whereas miR396b may ensure more rapid and dynamic target regulation during specific developmental stages or in response to the environment. In M. truncatula roots, exogenous application of BR at a concentration that limits root growth led to activation of the mtr-miR396b promoter and post-transcriptional repression of its MtGRF5 target. BRs have already been shown to control root growth (Yang et al., 2011; Hu et al., 2000; Kim et al., 2007; Park et al., 2010), and mutations in the Arabidopsis BR receptor gene BRI1 resulted in aberrant cell-cycle progression in the RAM (González-García et al., 2011 Hacham et al., 2011). However, both loss- and gain-of-function BR-related mutants showed reduced meristem size, indicating that BR signalling and homeostasis must be tightly controlled to enable optimal root growth. Identification of brassinosteroid-related mutants in M. truncatula will assist in addressing the putative link between miR396/GRF and the BR signalling pathway.

The recent demonstration that AtGRF2 and AtGRF7 regulate developmental responses to environmental stresses, such as nematode infection (Hewezi et al., 2012) and dehydration (Kim et al., 2012), highlighted the importance of dynamic and specific regulation of GRF genes. The interaction between miR396 and the GRF genes is unusual in plants because the miR/target duplex contains a 1-nt bulge in the 5′ region between positions 7 and 8 of the miRNA (Debernardi et al., 2012). This structure has been shown to confer sub-optimal regulation of GRF genes by miR396 in leaves, and maintain an inverse correlation of expression of the targets to the expression of the miRNA. In M. truncatula, the bulge between miR396 and the GRFs is conserved, which suggests a conserved biological role for formation of the expression gradient of the targets in the RAM.

Recently, Debernardi et al. (2012) identified AtbHLH74, a bHLH TF, as a non-conserved target of miR396 in A. thaliana. miR396-mediated repression of bHLH74 was required for correct leaf vein patterning. Interestingly, the miR396/bHLH74 regulatory node appeared to be specific to Brassicales. Target diversification within one miRNA family has been well documented for miR159/miR319 (Palatnik et al., 2007). Indeed, although miR159 and miR319 were first considered as members of a single family, a progressive specialization of the two variants occurred during evolution that led to differential target regulation. Due to their inverse expression in OE and mimicry roots, the present study reveals two MtbHLH79 genes as non-conserved miR396 targets in M. truncatula. These genes encode homologues of BIGPETAL (BPE)/bHLH79, a TF that is involved in the control of petal size and shape in A. thaliana (Szécsi et al., 2006). However, the lack of a root growth phenotype in roots expressing an RNAi construct targeted against MtbHLH79 genes suggests that these genes do not have a major effect on RAM function in M. truncatula. Interestingly, we found only putative miR396 binding sites in bHLH79/BPE homologues of the Fabidae species. Hence, the miR396/bHLH79 node may be more ancestral than miR396/bHLH74. These results argue in favour of the idea that miR396 has acquired novel targets inside the bHLH family throughout evolution to increase the complexity of the miR396-regulated network, although their biological relevance requires further investigation.

In Glycine max (soybean), Subramanian et al. (2008) observed a transient accumulation of miR396 during the first hours after inoculation with the symbiotic rhizobia Bradyrhizobium japonicum. Unlike mtr-miR396a, mtr-miR396b is expressed in the lateral root primordia of M. truncatula. This isoform may therefore play a general function in root lateral organogenesis rather than a specific role in nodule development. Furthermore, although miR396 and MtGRF targets are expressed in mature nodules, no obvious defect on nodule density, morphology or cellular organization was observed in roots with modified miR396 activity. Although a more precise analysis of MtGRF and MtbHLH79 expression and function must be completed, our results suggest that the miR396 network is unlikely to be crucial for nodule development.

In roots of tomato (Solanum lycopersicum) and M. truncatula, microarray profiling and small RNA deep sequencing identified several miRNAs that differentially accumulated during AM fungal colonization (Gu et al., 2010; Devers et al., 2011). However, to our knowledge, only miR171 has been functionally associated with the mycorrhization process (Lauressergues et al., 2012). We show here that, although the arbuscular structures remained apparently normal, the efficiency of mycorrhization was reduced following over-expression of miR396 and was promoted following its inactivation. This phenotype was similar to that of mutants affected in later stages of the sym pathway, such as some alleles of ipd3 (Horváth et al., 2011) or nsp2 (Maillet et al., 2011; Lauressergues et al., 2012), which lead to a defect in colonization frequency whereas arbuscule formation appears normal. In contrast, the enhanced colonization phenotype observed in MIM396 roots has not been observed frequently so far. A similar phenotype was recently observed when a miR171 h-resistant version of the NSP2 TF target was expressed (Lauressergues et al., 2012), suggesting potential roles of miR171 in regulation of a number of infection events. Moreover, over-expression of miR396 led to a mycorrhizal defect without affecting nodulation, which is also infrequent among the various previously described mutants because processes common to both symbioses are typically affected.

Finally, so far, few miRNAs (e.g. miR160 and miR166) have been associated with the control of RAM functions (Khan et al., 2011). A role for miR396/GRF in the control of cell proliferation was previously reported in A. thaliana leaves (Rodriguez et al., 2010). These authors showed that miR396 accumulated in the distal part of young developing leaves, thus restricting AtGRF2 expression to their proximal regions. Concomitantly, plants over-expressing miR396 developed smaller leaves, associated with a lower expression of cell proliferation markers, a global decrease in the number of dividing cells, and an increase in leaf-cell ploidy (Liu et al., 2009; Rodriguez et al., 2010). Here, we show that miR396 controls root growth, most likely via repression of MtGRF genes. Indeed, similar phenotypes were noted in roots, where MtGRF genes were down-regulated either by over-expression of miR396 or by RNAi. In addition, the opposite overlapping gradients observed for MtGRF and miR396 promoter activities in root tips strongly suggest that miR396 may restrict GRF activity to the active proliferation region of the RAM. Interestingly, the robust increase in MIM396 root dry weight without any significant effect on primary root length suggests an additional effect on root branching. This hypothesis may correlate with the detection of mtr-miR396b promoter activity in lateral root primordia. Further studies on stable transgenic lines in which the miR396/GRF regulatory node is modulated will enable more precise determination of the effect of miR396 on lateral root development.

Root growth is controlled by a balance between cell proliferation and elongation. In this study, miR396 over-expression led to changes in RAM size via modulation of cell proliferation rather than cell elongation. In A. thaliana, Wang et al. (2011b) reported that miR396 over-expression increased the percentage of leaf cells with high ploidy levels, indicating that ath-miR396 negatively regulated cell proliferation by controlling the switch between the mitotic cycle and the endocycle (Wang et al., 2011b). However, in M. truncatula root tips, we did not observe any significant change in the percentage of polyploid cells or the expression of MtCCS52B, which is a gene required for endo-reduplication in M. truncatula roots and nodules (Cebolla et al., 1999), even when expression of the homologue of CYCD3.1, which is a gene associated with endo-reduplication in A. thaliana (Menges et al., 2006), was perturbed. Thus, the repressive role of miR396 on root cell proliferation, measured by the repression of marker genes and a reduction in incorporation of EdU, appears to be associated with a global decrease in cell division activity in the RAM.

In conclusion, our findings identified targets of miR396 in plants and demonstrate involvement of the miR396/GRF regulatory node in the modulation of root growth and fungal mycorrhizal colonization of M. truncatula.

Experimental procedures

Plant material, growth conditions and hormonal treatments

For organ sampling, M. truncatula Jemalong A17 plants were grown in a greenhouse (16 h light/8 h dark, 22°C, 60-70% relative humidity) in a perlite:sand (4:1) mixed substrate, and irrigated with a nitrogen-poor solution (I medium) (Blondon, 1964). Mature nodules were harvested 5 weeks after plant inoculation with 20 ml of a suspension of the strain Sinorhizobium meliloti Sm2011 (OD600 = 0.1). Root apices versus whole roots were obtained as described by Gruber et al. (2009). For hormonal treatments, A17 seeds were sterilized as described by Gonzalez-Rizzo et al. (2006), and, after one night at 24°C in the dark, seedlings were transferred onto metal grids in Magenta vessels GA-7 (Sigma-Aldrich, http://www.sigmaaldrich.com) with roots submerged in I medium, and grown under rotation (125 rpm) at 24°C under long-day conditions (16 h light/8 h dark). Four- to five-day-old plants were treated with 0.1 μM 1-Naphthaleneacetic acid (NAA), 0.1 µM 6-Benzylaminopurine (BAP), 1 μM gibberellin (GA3), 1 nM epibrassinolide (BR) or 1 μM abscisic acid) for 1, 3 or 5 h. All phytohormones were purchased at Sigma-Aldrich.

Bioinformatic analyses

Sequences of the miR396 precursors and mature isoforms in M. truncatula and other plant species were retrieved from the miRBase database version 19 (November 2012, http://microrna.sanger.ac.uk). To identify M. truncatula homologues of A. thaliana GRF, bHLH74, bHLH79, CYCB1;3, CYCB2;1, CYCD3.1 and histone H4 proteins (Table S1), tBLASTx searches were performed for ESTs on the Institute for Genomic Research database, release 10.0 (http://compbio.dfci.harvard.edu/cgi-bin/tgi/Blast/index.cgi), and for available genomic sequences on the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/BLAST), Hapmap (http://www.medicagohapmap.org) and Legoo (Mt genome version 3.5.1,. http://www.legoo.org/) sites. MtbHLH79 homologues in land plants were researched using the tBLASTx program on the National Center for Biotechnology Information database. To evaluate the complementarity between miR396 and their putative targets, alignments of 21 bp between the mature miRNA variants and the target mRNAs were examined. Penalty scores were calculated as described by in Jones-Rhoades and Bartel (2004). μM. truncatula gene expression profiles obtained using the Affymetrix Medicago GeneChip® were retrieved from the Medicago truncatula Gene Expression Atlas database (MtGEA, http://mtgea.noble.org/v2/, Benedito et al., 2008; He et al., 2009).

Production of genetic material

To over-express miR396, the miR396a or miR396b precursors (93 and 95 bp, respectively) were amplified by PCR using Pfu DNA polymerase (Promega, www.promega.com) from A17 total DNA using specific primers (Table S2). The PCR products were first cloned into a pENTR/D-TOPO vector (Invitrogen, http://www.invitrogen.com), and then transferred into binary vector pK7GW2D (containing a 35S CaMV promoter, Karimi et al., 2002) using Gateway technology (Invitrogen). An empty pK7GW2D vector was used as a control. To inactivate miRNA activity, we used a mimicry approach (Franco-Zorrilla et al., 2007). The MIM396 construct used to inactivate miR396 has been described previously (Debernardi et al., 2012). For repression of MtGRF or MtbHLH79 expression by RNAi, fragments of MtGRF2, MtGRF4, MtGRF6 and MtbHLH79a corresponding to conserved regions of GRF and bHLH79 proteins, respectively (Figure S8A), were amplified by PCR using Pfu DNA polymerase (Promega) with specific primers (Table S2). The PCR products were first cloned into the pENTR/D-TOPO vector (Invitrogen) and then recombined into the pFRN destination vector (derived from pFGC5941; National Center for Biotechnology Information accession number AY310901) using Gateway technology (Invitrogen). An empty pFRN vector was used as a control (Gonzalez-Rizzo et al., 2006).

Agrobacterium rhizogenes root transformation and root growth measurements

Genetic constructs and the corresponding control empty vectors were introduced into the A. rhizogenes Arqua1 strain and then used for stable root transformation of M. truncatula. A17 seeds were surface-sterilized and germinated as previously described, and transgenic roots of composite plants (e.g. bearing wild-type aerial parts) were obtained as described by Boisson-Dernier et al. (2001). For each construct, 30 transgenic roots were obtained in at least three independent experiments. Root length was measured using ImageJ software (http://imagej.en.softonic.com/) 6 days after plant transfer to a non-selective medium, using the initial position of the root apex at the time of transfer as a baseline. At this point, roots were also dried over a period of 4 days at 70°C, and weighed to estimate the global biomass of the root system. Two biological replicates were completed (n > 22 for each construct in each experiment), and a Kruskal–Wallis test was performed to determine significant differences.

Meristem organization and cell proliferation/endo-reduplication analyses

To examine the cellular organization of the root tips in more detail, roots were stained using a modified pseudo-Schiff propidium iodide staining protocol as described by Truernit et al. (2008), and the root anatomy was analysed from longitudinal optical sections obtained using a Leica TCS SP2 confocal laser-scanning microscope (http://www.leica.com). Root meristem size was estimated as the number of cells in a given cell file of the future cortex from the quiescent centre to the first elongating cell. The length of cortical cells in fully mature regions of the root was also measured as an indicator of cell elongation.

For flow cytometric analysis of the ploidy level, 20 root apices, 5 mm long, were cut with a razor blade into 500 μl of Galbraith buffer (Galbraith et al., 1983), supplemented with 1% polyvinylpyrrolidone 10 000 (PVP-10, Sigma-Aldrich) 5 mm metabisulfite and 5 mg ml−1 RNase A. Propidium iodide (50 μg ml−1; Sigma-Aldrich) was added to the filtered (50 μm) supernatants. Endo-reduplication levels of 5000–10 000 stained nuclei were examined using a Cyflow SL flow cytometer (Partec, http://www.partec.com) with 532 nm solid-state laser excitation, and emission was collected using a 590 nm long-pass filter. The EdU cell proliferation assay was performed as described by Kotogány et al. (2010) with several modifications. Roots of composite plants expressing the various constructs were incubated with 10 mm of 5-ethynyl-2'-deoxyuridine (EdU, Invitrogen) in I medium over 4 h to label replicating cells. Nuclei were extracted as described above, and fixed with 1% formaldehyde for 30 min at 4°C. Nuclei were washed once during 5 min with 500 μl PBS, and the nuclear pellet was re-suspended in 200 μl EdU detection cocktail (2 μl ml−1 Alexa Fluor 647 azide (Invitrogen), 4 mm CuSO4, 40 mm sodium ascorbate in PBS), and then incubated for 30 min at room temperature in the dark. Nuclei were washed once during 5 min using 500 μl PBS, and the nuclear pellet was re-suspended in PBS containing 1 μm 4',6'-diamidino-2-phenylindole (DAPI) (Invitrogen). Isolated nuclei (5000–10 000) were analysed on a MoFlo XDP cytometer (Beckman Coulter, https://www.beckmancoulter.com). Alexa Fluor 647 was excited using a 640 nm solid-state laser, and emission was detected using a 670/30 band-pass filter. DAPI was excited using a 405 nm solid-state laser, and emission was detected using a 457/50 band-pass filter.

Nodulation and mycorrhization assays of transgenic roots

For the nodulation assays, after kanamycin selection, composite plants were transferred onto Fahraeus medium without nitrogen for 4 days, and were inoculated with a suspension of the strain S. meliloti Sm1021 (OD600 = 0.05; 10 ml per 10 cm2 Petri dish) for 1 h. After 1 week, nodules were counted, and composite plants grown in vitro were transferred to a greenhouse (16 h light/8 h dark, 22°C, 60–70% relative humidity) on a perlite:sand (4:1) mixed substrate and irrigated with nitrogen-free Fahraeus solution (0.132 g/L CaCl2, 0.12 g/L MgSO4.7H2O, 0.1 g/L KH2PO4, 0.075 g/L Na2HPO4.2H2O, 5 mg/L Fe-citrate, and 0.07 mg/L each of MnCl2.4H2O, CuSO4.5H2O, ZnCl2, H3BO3, and Na2MoO4.2H2O, pH 7.5). Total nodule number was examined at 21 days post-inoculation in three independent experiments (biological replicates) on at least 30 independent roots per construct. To estimate nodule density, this value was relative to the corresponding root dry weight, which was measured as described above. For mycorrhization assays, R. irregularis DAOM197198 spores were purchased from Agronutrition, www.agronutrition.fr). After root transformation (Boisson-Dernier et al., 2001), the plants were cultivated in 200 ml pots on Oil-Dri US-Special Substrate (Damolin, damolin.com) for 5–8 weeks in a growth chamber (temperature 22°C; relative humidity 75%; light intensity 200 μE m−2 sec−1; light and dark photoperiods, 16 and 8 h, respectively), and watered every 2 days using modified Long Ashton medium containing 7.5 μm phosphate (Lauressergues et al., 2012). For inoculation with R. irregularis, we used 400 spores per plant. The Myc-LCOs used in this study were an equimolar mix of the four Myc-LCOs LCO-IV (C16:0), LCO-IV (C16:0,S), LCO-IV (C18:1Δ9Z) and LCO-IV (C18:1Δ9Z,S) described by Maillet et al. (2011), used at a final concentration of 10 nm. The plants treated with Myc-LCOs (12 h) were cultivated on Fahraeus medium as described by Lauressergues et al. (2012). Phenotyping of mycorrhization was performed as described by Trouvelot et al. (1986): the frequency (F) of mycorrhiza in the root system and the arbuscule abundance (a) (both as percentages) were calculated in the colonized root sections using Mycocalc software (http://www2.dijon.inra.fr/mychintec/Mycocalc-prg/download.html; Trouvelot et al., 1986). Each mycorrhization experiment was performed twice using 15 independent transgenic roots per construct.

GUS histochemical and histological analyses

Genomic regions upstream of mtr-miR396a and miR396b precursors (1947 and 1938 bp, respectively) were amplified by PCR from genomic DNA (primers in Table S2) and cloned into the pENTR/D-TOPO vector (Invitrogen) before insertion into Gateway vector pKGWFS7 (Karimi et al., 2002) upstream of the GUS reporter gene. Similarly, promoter–GUS constructs were generated using the upstream regions of MtGRF4, MtGRF5 and MtGRF6 genes (1978, 2070 and 1936 bp, respectively). For each construct, transgenic roots were obtained as previously described, and GUS activity was revealed as described by De Lorenzo et al. (2009). For root transverse and nodule longitudinal sections, samples were embedded in 6% agarose, and sections were generated using a vibratome (100 μm; Leica VT1200S). For determination of the cellular organization of miR396-OE and MIM396 nodules, mature nodules were dissected from roots at 4 weeks after inoculation and fixed in 4% paraformaldehyde, dehydrated in a series of ethanol solutions, and embedded in Paraplast (Sigma-Aldrich). Longitudinal sections (10 μm) were generated using a microtome (Leica RM2245), and stained for 10 min in toluidine blue (0.015%). Stained roots and sections were observed using a microscope (Leica DMI6000B).

Quantitative real-time RT-PCR analysis

To monitor miR396 precursor and target mRNA accumulation, total RNAs were extracted using the Trizol method (Invitrogen), and 2 μg of total RNAs treated with DNAse I (Fermentas, http://www.thermoscientificbio.com/fermentas) were used for reverse transcription in a total volume of 20 μl, using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Real-time PCR was performed from 1 μl cDNA using a LightCycler 480 SYBR Green I master kit (Roche Diagnostics, www.roche-diagnostics.fr) as described by De Lorenzo et al. (2009). Oligonucleotides were designed using PRIMER3 version 0.4.0 software (Rozen and Skaletsky, 2000). The reference genes, MtACTIN11, MtRBP1 and MtH3L, were selected using Genorm software (http://medgen.ugent.be/~jvdesomp/genorm/; Vandesompele et al., 2002). Primers used for MtGRF genes, MtbHLH79 genes, MtCYCB1;3, MtCYCB2;1, MtCYCD3.1, MtH4, MtCCS52b, MtCDC16 and GUS are listed in Table S2. To assess mature miR396 accumulation in the over-expressing roots, the miScript reverse transcription kit and miScript SYBR Green PCR kit (Qiagen, www.qiagen.com) were used for cDNA synthesis and real-time PCR, respectively, using 500 ng of total RNA for the reverse transcription reaction in a total volume of 20 μl, and 1 of the cDNA solution for the PCR. miRNA quantification was performed using the mature mtr-miR396a and mtr-miR396b sequences as forward primers and a universal reverse primer provided by the kit manufacturer. Real-time PCR was performed at an annealing temperature of 55°C using the LightCycler 480 apparatus (Roche Diagnostics). As a reference, we used a U6 snoRNA-specific primer in addition to the universal primer. For RT-PCR experiments, two technical replicates and at least two independent biological replicates were performed for each experiment.

Acknowledgements

We thank C. Roux and F. Frugier for valuable discussions, and J. Gouzy and E. Sallet (LIPM, Toulouse, Centre national de la recherche scientifique, France) for the miR396 target search. Microscopy and flow cytometry experiments were performed using the Imagif platform (https://www.imagif.cnrs.fr/), with the kind assistance of Spencer Brown and Michael Bourge. This study was supported by grants from the MIRMED project (Genoscope, Centre National de la Recherche Scientifique) and the Saclay Plant Sciences program (ANR-10-LABX-40). P.B. was supported by a Marie Curie Postdoctoral Intra-European Fellowship (European Commission, MEDEPIMIR, number PIEF-GA-2010-273743). G.A.K. was supported by the Higher Education Commission (Pakistan). We thank Eric Samain, Sébastien Fort and Sylvain Cottaz (Centre de Recherches sur les Macromolécules Végétales, Grenoble, France) for providing the Myc-LCOs.

Ancillary