Local and Systemic Regulation of Plant Root System Architecture and Symbiotic Nodulation by a Receptor-Like Kinase

In plants, root system architecture is determined by the activity of root apical meristems, which control the root growth rate, and by the formation of lateral roots. In legumes, an additional root lateral organ can develop: the symbiotic nitrogen-fixing nodule. We identified in Medicago truncatula ten allelic mutants showing a compact root architecture phenotype (cra2) independent of any major shoot phenotype, and that consisted of shorter roots, an increased number of lateral roots, and a reduced number of nodules. The CRA2 gene encodes a Leucine-Rich Repeat Receptor-Like Kinase (LRR-RLK) that primarily negatively regulates lateral root formation and positively regulates symbiotic nodulation. Grafting experiments revealed that CRA2 acts through different pathways to regulate these lateral organs originating from the roots, locally controlling the lateral root development and nodule formation systemically from the shoots. The CRA2 LRR-RLK therefore integrates short- and long-distance regulations to control root system architecture under non-symbiotic and symbiotic conditions.


Introduction
Plant growth requires the continuous development of the root system and its adaptation to changing environmental soil conditions. Mechanisms controlling root system architecture at the whole-plant level, including the systemic coordination of shoot and root development, are key breeding targets for maintaining crop productivity under adverse stress conditions but remain poorly understood [1]. Root system architecture is a consequence of the sustained activity of root apical meristems, leading to indeterminate root growth as well as the de novo formation of lateral organs. In legume (Fabaceae) plants, the root system can form two types of lateral organs depending on the environmental conditions: lateral roots and symbiotic nitrogen-fixing nodules [2][3][4]. Lateral root initiation, emergence and growth depend on water and nutrient availability and are regulated by a combination of local and systemic pathways [5]. Symbiotic nodules are formed under nitrogen-deprived conditions when the specific Rhizobium spp. soil bacteria are present in the rhizosphere [6][7]. In both types of lateral organogeneses, cell divisions are activated in specific tissues (pericycle, endodermis and cortex) above the growing root tip [2][3][4]8]. Root tissues contributing to primordium formation are, however, different depending on the plants and organs: in the Medicago truncatula model legume, lateral root primordia mainly develop from pericycle cell divisions, whereas the nodule primordia that are induced by Sinorhizobium meliloti mainly derive from the inner cortex. Both types of primordia will then subsequently emerge from the parental root and establish a meristematic stem cell niche ensuring their indeterminate growth.
To control meristematic activity, cell differentiation, and lateral organ initiation, non-cell autonomous cues are essential to carry positional information, which can be informed either by mobile phytohormones, small RNAs or peptides [9][10][11]. Among peptides, several CLAVATA 3/EMBRYO-SURROUNDING REGION (like) peptides (so called CLE peptides; [10]) that are perceived by Leucine-Rich Repeats -Receptor-Like Kinases (LRR-RLKs) are involved in local and long-distance (systemic) pathways controlling the development of different plant organs. First, several CLE peptide/LRR-RLK receptor modules carry positional information across a few cell layers to control the cell fate in different Arabidopsis thaliana meristems. The founding example is the CLAVATA3 (CLV3) peptide, which is perceived by the CLV1 receptor to control the shoot apical meristem stem cell niche [12,13] as well as columella cell differentiation in the Root Apical Meristem (RAM; [14][15][16]). A second example is the Tracheary element Differentiation Inhibitory Factor (TDIF) peptide, which is perceived by the Phloem Intercalated with Xylem (PXY) receptor, controlling stem cell proliferation/differentiation transition in the cambium meristem and therefore vasculature differentiation and organ thickening [17][18][19].
In legume plants, an additional CLE/LRR-RLK module controlling root lateral organs number was identified through grafting experiments as performing a long-distance systemic function from the shoots [7]. Mutants affecting orthologous LRR-RLKs that are closely related to CLV1 in Arabidopsis (SUNN, Super Numeric Nodules in M. truncatula; NARK, Nodule Autoregulation receptor Kinase in soybean; and HAR1, Hypernodulation and Aberrant Root 1 in Lotus japonicus; [20][21][22][23]) form an increased number of symbiotic nitrogen-fixing nodules depending on the receptor function in the shoots. In addition, the Lotus KLAVIER (KLV) LRR-RLK, which is closely related to the Arabidopsis RECEPTOR-LIKE PRO-TEIN KINASE 2 (RPK2)/TOADSTOOL 2 (TOAD2) that is functionally linked to anther and embryo development, is involved in the same systemic AON pathway as is HAR1 [24]. In M. truncatula and L. japonicus, CLE peptides that are specifically produced in nodulated roots can negatively regulate the nodule number depending on these LRR-RLK receptors [25][26][27]. This CLE peptide/LRR-RLK module therefore participates in the systemic ''Autoregulation of Nodulation'' (AON; [22]) pathway, which may involve a direct root-to-shoot CLE peptide transport and receptor binding in the shoots, as recently proposed [28]. Interestingly, an increased number of emerged lateral roots was reported in the Lotus har1 mutant under both symbiotic and non-symbiotic conditions [29]. Similarly, KLV also has a non-symbiotic function in the local regulation of SAM maintenance [24]. Collectively, these results indicate that CLE peptide/LRR-RLK signaling modules regulate the development of various organs using either local and/or systemic pathways.

Results
The COMPACT ROOT ARCHITECTURE 2 (CRA2) gene negatively regulates lateral root development Forward genetic screens that were performed on a M. truncatula Tnt1 insertional mutant collection [30][31] identified seven mutant lines with a wild-type shoot development and a ''compact root system architecture'' (so called ''cra''; Fig. 1A-C and S1A Fig.). Segregation analyses of this root phenotype revealed a 3:1 WT:mutant ratio (chi2 test, p,0.05, n = 30), suggestive of a single locus recessive mutation. Allelism tests (S1B Fig.) indicated that all of the mutants were affected in the same locus but differed from the previously identified cra1 mutant showing partially similar phenotypes [32] and were therefore named cra2. Detailed quantitative in vitro analyses revealed that the cra2 phenotype consists, compared to the WT, of shorter roots with an increased number of emerged lateral roots (Fig. 1D-E). Accordingly, lateral roots were observed three days post germination (dpg; Fig. 1E) in contrast to WT plants. This root phenotype was observed independently of the growth and nutrient conditions that were used (greenhouse versus in vitro, Fig. 1A, E; with or without nitrogen or carbon sources, Fig. 1E). In addition to the faster emergence of lateral roots, we observed a reduction in the primary root growth, which prompted us to analyze the structure of the RAM ( Fig. 2A-C). In contrast to the A. thaliana model, M. truncatula roots have an open meristem, and the transition between cell proliferation and differentiation is more progressive. In addition, cells from diverse files elongate at a slightly different distance from the root apex, leading to a coneshaped transition zone ( Fig. 2A, detail of the transition zone in S2 Fig.). When cra2 RAMs were observed three dpg, both the cell proliferation and elongation zones were reduced, and a lower number of cells was observed in the two zones ( Fig. 2 A-C). Root patterning, however, evaluated based on both longitudinal and transversal sections, seemed unaffected ( Fig. 2A-B and D). In addition, amyloplast accumulation in differentiated root cap cells and the expression of the RAM stem cell niche marker WOX5 (WUSCHEL-related homeobox 5; [33][34][35]) were both detected in cra2 (S3 Fig.), suggesting the maintenance of RAM cell identity. Among several hypotheses that could explain the cra2 root system architecture phenotype, we tested whether a RAM activity defect could indirectly lead to increased branching as compensation or if the reduced meristematic activity could be a consequence of the enhanced formation of lateral roots. An analysis of root apices at one dpg, i.e., before any lateral root primordium could be detected in the cra2 mutant, revealed no significant difference in the size of the cell proliferation and elongation zones between WT and mutant roots (Fig. 2E, F). Compared to previous observations of roots at three dpg ( Fig. 2A, B), this result indicates that a RAM defect does not precede the occurrence of the lateral root phenotype. As an independent approach, we experimentally removed the RAM at one or three dpg and followed the kinetics of lateral root formation (Fig. 2G, H). The cra2 mutant showed an increased ability to form lateral roots whether RAM excision occurred before or after lateral root initiation. Collectively, these results suggest that the cra2 RAM phenotype can be disconnected from its enhanced ability to form lateral roots.
The CRA2 gene positively regulates symbiotic nodule development In addition to root branching, legume roots can adapt to environmental conditions by developing another root lateral organ, the nitrogen-fixing symbiotic nodule. An analysis of cra2 roots under symbiotic conditions revealed that a similar ''compact

Author Summary
Despite the essential functions of roots in plant access to water and nutrients, root system architecture has not been directly considered for crop breeding improvement, but it is now considered key for a ''second green revolution.'' In this study, we aimed to decipher integrated molecular mechanisms coordinating lateral organ development in legume roots: lateral roots and nitrogen-fixing symbiotic nodules. The compact root architecture 2 (cra2) mutant form an increased number of lateral roots and a reduced number of symbiotic nitrogen-fixing nodules. This mutant is affected in a CLAVATA1-like Leucine-Rich Repeat Receptor-Like Kinase (LRR-RLK) that has not previously been linked to root development. Grafting experiments showed that CRA2 negatively controls lateral root formation and positively controls nodule development through local and systemic pathways, respectively. Overall, our results can be integrated in the framework of regulatory pathways controlling the symbiotic nodule number, the so-called ''Autoregulation of Nodulation'' (AON), involving another LRR-RLK that also acts systemically from the shoots, SUNN (Super Numeric Nodules). A coordinated function of the CRA2 and SUNN LRR-RLKs may thereby permit the dynamic fine tuning of the nodule number depending on the environmental conditions. root architecture'' phenotype was observed (Fig. 3A) and that mutant shoot growth was similar with or without Rhizobium inoculation (Fig. 3B, C). cra2 plants, however, developed a strongly reduced number of symbiotic nodules (Fig. 3 D, E). This low nodulation phenotype could be either linked to a direct CRA2 function in regulating nodulation or may reflect that the strongly altered ''compact root architecture'' phenotype (e.g., Fig. 3A) indirectly hampers nodule formation. Using a symbiotic infection kinetic analysis (1 to 14 dpg), we showed that Rhizobium inoculation as early as 1 or 3 dpg led to reduced nodulation (Fig. 3E), indicating that the cra2 nodulation phenotype was independent of the strength of the lateral root phenotype. The few symbiotic nodules that formed on cra2 roots were elongated (S4A The error bars represent confidence intervals (a = 5%). A Kruskal and Wallis test was used to determine the significant differences (indicated by letters, a,5%; n = 10). D, Quantification of the root length (upper graph), lateral root number (middle graph) and lateral root density (lower graph) of the WT, cra2-1 and cra2-2 plants that were grown in vitro for 10 days post-germination (dpg) on an N-deprived ''i'' medium [42]. The error bars represent confidence intervals (a = 1%). A Kruskal and Wallis test was used to determine the significant differences (indicated by letters, a,1%; n.25). E, Representative examples of the WT and cra2-1 plants that were grown in vitro for three dpg on an Ndeprived ''i'' medium [42] or for 14 dpg on the same medium (-N), on an N-rich medium (+N, Fahraeus with NH 4 NO 3 10 mM; [43]), or on an Nand C-rich medium (+N+C, ''Lateral Root Inducing Medium; [42]). Note that in cra2, the lateral roots were already emerged at three dpg (arrowhead) and that the ''compact root system architecture'' phenotype was detectable independently of the growth medium. Bars = 0,5 cm. doi:10.1371/journal.pgen.1004891.g001  Fig.) between the cell proliferation zone (CPZ) and the cell elongation zone (CEZ). Bar = 100 mm. B, F, Quantification in the WT and cra2-1 roots of CPZ and CEZ length at three dpg (B) or one dpg (F). C, Quantification in the WT and cra2-1 roots of the cells number in the same zones at three dpg. In B, C, and F, the error bars represent standard deviation, and a Mann-Whitney test was used to determine the significant differences between genotypes (*, a,5%, n = 10). D, Transversal sections of three dpg WT and cra2-1 roots showing a similar radial organization of cell layers. Bar = 100 mm. G, Representative examples of the WT and cra2-1 root system architecture seven days post excision (dpe) of the RAM. The excision was performed either one day after germination (dpg, upper pictures) or three dpg (lower pictures). Bar = 1 cm. H, Quantification of the lateral roots at two, four and seven days post excision (dpe) of the RAM in WT or cra2-1 plants. As shown in (G), the meristems were excised at either one dpg (upper graph) or three dpg (lower graph). The error bars represent the confidence interval (a = 5%), and a Mann-Whitney test was used to determine significant differences between the genotypes for each time point (*, a,5%; n.25). doi:10.1371/journal.pgen.1004891.g002 Fig.), revealing that a functional nodule meristem was formed. In addition, an analysis of bacterial nitrogenase activity using an Acetylene Reduction Assay (ARA) showed that despite cra2 plants have a strongly reduced ability to fix atmospheric nitrogen due to their lower number of symbiotic organs (S4B Fig.), cra2 nodules have a similar specific ARA activity compared to that of WT nodules (S4C Fig.). Overall, the detailed analysis of the mutant phenotypes indicates that, independently of a potential indirect effect of the increased lateral root formation on nodule formation, CRA2 antagonistically regulates lateral root and nodule formation. CRA2 regulates root system architecture depending on systemic and local pathways As root system architecture is controlled by both systemic and local pathways, we next determined using graftings experiments whether the cra2 phenotypes depend on roots and/or shoots. Under non-symbiotic conditions, the ''compact root architecture'' phenotype was recovered specifically when the graft combination included cra2 mutant roots (Fig. 4A-B) but was independent of the shoot genotype, indicating that CRA2 expression in roots is sufficient to negatively regulate lateral root formation. Under symbiotic conditions, we surprisingly observed a disconnect between the lateral root and nodulation phenotypes ( Fig. 4C-D). Indeed, similar to non-symbiotic conditions, the increased density of the lateral roots was associated with cra2 mutant roots, but the low nodulation phenotype was observed only in grafting combinations that included cra2 mutant shoots and was independent of the root genotype. This result indicates that the systemic activity of CRA2 in the shoots positively regulates symbiotic nodule formation in the roots. Strikingly, the cra2/WT grafts had a WT root system architecture but developed more than 10 times fewer nodules (Fig. 4C-D). In addition, when the nodule numbers were related to the root dry weight to counterbalance the strong differences existing between WT/WT and WT/cra2 root systems, the nodulation efficiency became strictly equivalent between these two grafting combinations (Fig. 4E). These results therefore unambiguously demonstrate that the low-nodulation phenotype is not an indirect consequence of an inhibitory signal that is produced in the numerous lateral roots in cra2. In addition, this result indicates that the root-dependent regulation of the lateral root number is independent of the shoot-regulation of the nodule number.  [42]) and then inoculated by Sinorhizobium meliloti 1021 for 14 days. Bar = 1 cm. C, Quantification of the shoot dry weights of the WT, cra2-1 and cra2-2 plants that are shown in (A). The error bars represent confidence intervals (a = 5%). A Kruskal and Wallis test was used to determine the significant differences (indicated by letters, a,5%; n = 30). D, Representative examples of the WT and cra2-1 plants that were grown for one week in vitro on an N-deprived medium (''i'', [42]) and then inoculated by S. meliloti 1021 for 14 days. The arrowheads indicate the position of the nodules. Bar = 1 cm. E, Quantification of the nodule number 14 days post inoculation with S. meliloti in the roots of the WT and cra2-1 plants that were grown in vitro. The inoculation was performed between one and 14 days post germination (dpg). The error bars represent confidence intervals (a = 5%). Significant differences were determined using a Mann-Whitney test (a,5%; n.20). doi:10.1371/journal.pgen.1004891.g003 The CRA2 gene encodes a Leucine-Rich Repeat -Receptor-Like Kinase that has not previously been linked to root or symbiotic nodule development To identify the gene that is affected in cra2 mutants, Tnt1 Flanking Sequence Tags (FSTs) were generated in the different alleles (cra2-1 to cra2-7) to detect Tnt1 insertions affecting a common genomic sequence in several of these mutant lines (cra2-1, cra2-4 to cra2-6; Fig. 5A and S5A-B Fig.). Interestingly, the three other alleles also showed genetic lesions in the same locus, consisting, respectively, in the presence of other M. truncatula endogenous insertional elements for cra2-3 and cra2-7 and in a point mutation for cra2-2 leading to a frameshift (S5C-E Fig.). Based on the FSTs that were available from the Noble foundation (http://bioinfo4.noble.org/mutant/), we could identify three additional lines with a Tnt1 insertion in the same open reading frame (cra2-8 to cra2-10) that showed a ''compact root architecture'' phenotype ( Fig. 5A and S1A Fig.), further confirming that mutation at this locus is causal for the phenotype. The genomic region that was affected in the 10 available cra2 alleles encodes an LRR-RLK belonging to subfamily XI (Fig. 5B), which, in agreement with the results of the grafting experiment, is expressed in the shoots, roots and nodules (S6 Fig.). Interestingly, this LRR-RLK subfamily contains several other receptors that were identified as regulating plant development via local or systemic regulation [10,36]. An analysis of CRA2 homology with other LRR-RLKs that have been functionally characterized in legumes showed that CRA2 was not closely related to SUNN/ HAR1 or KLV and was most closely related to A. thaliana XIP1 (Xylem Intermixed in Phloem 1; [37] ; Fig. 5B). An Arabidopsis xip1 mutant was recently described as showing defects in vasculature patterning specifically in the shoots but not in the roots, and no root developmental phenotype was reported [37]. To determine whether XIP1 and CRA2 may nevertheless have related functions, we analyzed the vascular patterning in cra2 mutant roots and shoots. No significant defect could be detected in the shoot or root cambium structure; in the vascular bundle differentiation, such as the misspecification of xylem and phloem bundles; or in the root or stele diameter (S7 Fig.). These results suggest that XIP1 and CRA2 are not functional homologs. In addition, these results indicate that the cra2 root and nodulation In B, D and E, the error bars represent confidence intervals (a = 5%), and the significant differences were determined using a Kruskal and Wallis test (indicated by letters, a,5%; n.10). doi:10.1371/journal.pgen.1004891.g004 phenotypes (Fig. 1, 2) are not linked to any detectable defect in vascular bundle patterning. We then analyzed the CRA2 spatial expression pattern under non-symbiotic and symbiotic conditions using either a transcriptional fusion between an ,2 kb CRA2 promoter region and the GUS reporter (Fig. 6A-D and I-N) or in situ hybridization (Fig. 6 E-H). Both approaches revealed an expression that was associated with the root stele and vascular bundles (Fig. 6A-B and E-F). Similarly, CRA2 expression was associated with vascular bundles in the shoots (Fig. 6G-H). This result agrees with the expression pattern of other LRR-RLKs regulating lateral root and nodule numbers in different legumes (HAR1/NARK/SUNN or KLV; 24, [38][39][40]). CRA2 was additionally expressed in the Cell Proliferation Zone (CPZ) of the open RAM ( Fig. 6A and C-D). In addition, CRA2 was expressed in the lateral root initiation site as soon as the pericycle divisions could be identified (Fig. 6I-J) and later in whole lateral root primordia during the initiation and emergence stages (Fig. 6K-L). Under symbiotic nodulation conditions, CRA2 expression was also detected in nodule primordia (Fig. 6M) as well as in mature nodules in relation to peripheral vascular bundles and the apical meristem (Fig. 6N).

Discussion
CRA2 functions in the regulation of legume root system architecture suggest that this LRR-RLK acts positively in the Autoregulation of Nodulation (AON) pathway. In the current model, the systemic SUNN-dependent AON pathway represses the number of nodules that are formed on the roots depending on the shoot metabolic status, e.g., the amount of carbon skeletons that are provided through photosynthesis and that are required for the subsequent assimilation of fixed nitrogen in the nodules. The SUNN pathway therefore limits the formation of extra nodules depending on environmental cues. When initiated, however, a negative feedback mechanism would be necessary to reset this negative regulation and further permit new symbiotic infection events. The CRA2 LRR-RLK may then participate in the systemic dynamic fine tuning of nodule formation. As both CRA2 and SUNN LRR-RLKs are expressed in the shoot vasculature, it remains to be determined whether and how these proteins act together or independently. As LRR-RLKs have been proposed to form large complexes comprising different members of the family [36] or even other RLKs (e.g., CLV1 with ACR4 in the RAM; [16]), CRA2 and SUNN may interact within a single complex.
The Lotus HAR1-dependent AON pathway was additionally shown to control negatively lateral root formation [20][21][22][23][24][25][26][27][28][29]. In the sunn Medicago mutant, however, the root length is reduced, but no specific function in the regulation of lateral root number has been identified [23]. CRA2 therefore negatively affects lateral roots similarly to HAR1 in Lotus but dissimilarly to SUNN in M. truncatula. Interestingly, a shorter root length and a decreased number of lateral roots occurred in the Lotus klv mutant, whereas additive effects of klv and har1 were identified for nodulation, indicating their related function in a single AON pathway [40]. In addition, both har1 and klv antagonistic lateral root phenotypes were observed under non-symbiotic conditions [29,40], similar to the phenotype of CRA2 in M. truncatula. This result indicates that these peptide/LRR-RLK regulatory modules acting under non-symbiotic conditions may also control lateral root formation in non-nodulating (non-legume) plants. In Arabidopsis, the ACR4 CRINKLY-RLK regulates lateral root initiation [41], but no LRR-RLK has been yet associated with this developmental process. More generally, despite the close relationships between HAR1, SUNN and CLV1, their mutant phenotypes are not conserved between plant species: the sunn and har1 mutants have divergent lateral root phenotypes, whereas no shoot fasciation phenotype was detected in the har1 or sunn mutant in contrast to clv1 in Arabidopsis. Similarly, despite CRA2 and XIP1 having closely related sequences, no vasculature patterning phenotype was identified in cra2 compared to xip1, and no altered root system architecture phenotype was reported in xip1. This result suggests that various sets of LRR-RLKs differentially regulate the ability of root systems to form lateral roots depending on species, even inside the legume family. Alternatively and non-exclusively, different patterns of LRR-RLK gene duplication may have occurred in the different plant genomes, generating functionally divergent or redundant pathways. These scenarios may explain the apparent phenotypic diversification that is observed.
Overall, this study demonstrates that a single LRR-RLK acts locally in roots and systemically in shoots to control root lateral organ development, thereby coordinating at the whole-plant level the plastic development of the root system depending on the changing environmental conditions. To elucidate the opposite effects of the CRA2 pathway on lateral root and nodule formation, the identification of downstream targets will be essential. Among other candidate pathways, cytokinins were previously reported to antagonistically control lateral root and symbiotic nodule formation [42]. Data on crosstalk between signaling peptides and hormones is just emerging, mainly with auxin and cytokinins [10]. Therefore, a remaining challenge will be to integrate the different peptide/receptor modules that are known to regulate lateral root and nodule formation, including the CRA2 pathway, into the framework of hormonal regulation.

Plant and bacteria material
The Medicago truncatula (Gaertn.) plants used in this study were of the R108-4 genotype. The Tnt1 insertional mutants were generated and screened at the Noble Foundation (USA; lines named NFxxx; [30]) or produced at the ''Institut des Sciences du Végétal'' (ISV, CNRS, Gif sur Yvette, France) and screened at the ''Agroécologie'' institute (INRA, Dijon, France; lines with other names than NFxxx). The seeds were scarified on sandpaper, sterilized for 20 min in bleach (12% [v/v] sodium hypochlorite), and thoroughly rinsed in sterile water. The seeds were then stratified at 4uC for one day and then germinated at 25uC in the dark on inverted water agar plates. The seedlings were grown in vitro in a growth chamber at 25uC with a 16 h light period and a 150 mE intensity on N-deprived media (''i'', [42]; or Fahraeus, [43]), on an N-rich medium (Fahraeus with NH 4 NO 3 10 mM), or on a N-and C-rich medium (''Lateral-Root-Inducing Medium'', LRIM; [42]) depending on the experiment. Alternatively, the plants were grown in a greenhouse (25uC, 16 h 150 mE light period, 60%-70% humidity) in soil or in pots containing a perlite:sand mixture (3:1) and watered every two days with ''i'' or ''SN/2'' media [42], depending on the experiment. The nodulation experiments used the Sinorhizobium meliloti 1021 strain (OD 600nm = 0.05) as described in [42].

Transposon display, genotyping, RT-PCR and cloning
Plant genomic DNA was extracted from the leaves as described by [44], and Flanking Sequence Tags (FSTs) that were linked to Tnt1 insertions were identified using the transposon display PCR method as described in [45] using the EcoR1 or Ase1 restriction enzymes. The transposon display, genotyping PCRs, and the sequencing of the different mutant alleles were performed using the primers that are listed in S1 Table. The RT-PCRs and real-time RT-PCRs were performed as described in [42] with the primers that are indicated in S1 Table. To generate the MtCRA2 reporter transcriptional fusion, an 1800-bp sequence upstream of the CRA2 start codon was identified, corresponding to the upstream region of the Medtr3g110840.1 open reading frame (M. truncatula genome Mt4.0v1, http://www.jcvi.org/medicago/index.php). This region was amplified by PCR using a high-fidelity polymerase (Phusion, Thermo Scientific) and primers pCRA2-F and pCRA2-R (S1 Table). The promoter was cloned using Gateway technology initially into the pTOPO-Entry vector (Invitrogen) and then into the pkGWFS7 vector (https://gateway.psb.ugent.be/search) carrying a Green Fluorescent Protein (GFP) -b Glucuronidase (GUS) fusion downstream of the cloning recombination site.  ). A, Structure of the CRA2 protein indicating the 10 mutant alleles (arrowheads) that were identified by forward and reverse genetic screens (the indicated position is related to the predicted ATG) and functional domains. The vertical black bars indicate the predicted transmembrane domains; in grey are the Leucine-Rich Repeats; and the hatched region represents the kinase domain. The black arrowheads represent alleles that are linked to a Tnt1 retro-element insertion; the grey arrowheads represent another insertional element; and the white arrowhead represents a nucleotide deletion causing a translational frameshift. Bar = 50 residues. B, Phylogenetic tree of selected LRR-RLKs that are related to CRA2 from Arabidopsis (subfamily XI) or are functionally characterized in legumes. The sequences were aligned using Muscle, and the regions that were conserved between all of the sequences were defined with Gblocks. The phylogenetic relationships were determined using a maximum likelihood analysis (PhyML), and statistical support for each node was estimated by approximate likelihood ratio tests. The Chlamydomonas reinhardtii XP001698687 protein was used to root the tree. doi:10.1371/journal.pgen.1004891.g005 Grafting, root apex excision, and composite plants The graftings were performed as described in the ''cuttings and grafts'' chapter of the Medicago handbook (http://www.noble. org/medicagohandbook/). The grafts, which were initially generated in vitro, were transferred after three weeks to pots containing a perlite:sand mixture (3:1) that was imbibed with the ''i'' solution in a growth chamber (same conditions as described above). After three weeks, the root length and root dry weight (60uC for 48 h) were measured.
For root apex excision, the roots were sectioned at one centimeter from the root tips one or three days post germination and placed in the ''i'' medium in a growth chamber (same conditions as described above). The number of lateral roots was quantified one to seven days post root tip excision using the ImageJ software (http://imagej.nih.gov/ij/).
To generate composite plants, constructs were introduced into the Agrobacterium rhizogenes ARqua1 strain and used for M. truncatula root transformation as described in [46]. Transgenic roots were selected on kanamycin (25 mg/L) for two weeks, and composite plants were then transferred onto growth papers (Mega International, http://www.megainternational.com/) on ''i'' medium for four to six days and, depending on the experiment, were inoculated by S. meliloti as described in [42].

Microscopy analyses, histological stainings and in situ hybridization
The roots or stems were cut into 5-mm segments and immediately embedded in 3% agarose (Euromedex). A vibratome (VT1200S, Leica) was used to generate 100 mm-thick cross-sections. Toluidine blue and aniline blue (Sigma) stainings were performed by incubating sections for 3 to 5 min in 1% toluidine blue or in 0.005% aniline blue and subsequently washing the sections twice with distilled water. The sections were viewed under bright-field illumination or under UV excitation (365 nm), respectively, for toluidine blue or for aniline blue stainings using a Leica DMI 6000B inverted microscope. The phloroglucinol-HCl reagent was prepared by mixing 2 volumes of 2% (w/v) phloroglucinol (Sigma) in 95% ethanol with one volume of concentrated HCl and observation under bright-field illumination (DMI 6000B, Leica). Callose fluorescence was visualized using a 405 nm excitation, and emission was collected between 480 and 515 nm (DMI 6000B, Leica). The amyloplasts were detected using Lugol (Sigma) staining and visualization under bright field illumination. To examine the detailed cellular organization of M. truncatula root apices, the roots were stained using a modified Pseudo Schiff-Propidium Iodide (PS-PI) staining protocol as described by [47]. Longitudinal optical sections were obtained using a Leica TCS SP2 confocal laser-scanning microscope using a 488 nm excitation, and emission was collected between 520 and 720 nm. The root meristem size was estimated as the number of cells in the outer cortex from the location of the tip to the first elongating cell using the ImageJ software. The length of the elongated cortical cells in the mature root region was also measured.
The GUS activity was revealed as described by (48] and observed under bright-field illumination (DMI 6000B, Leica). In addition, using the AOBS reflection mode of a Leica TCS SP8 spectral confocal laser-scanning microscope, the GUS precipitate reflectance was analyzed as described in [47] with a 488 nm excitation and a collection of the reflection signal between 485 and 491 nm on PS-PI counterstained samples (detected as described above).
In situ hybridization was performed as described in [48] with the probe as indicated in S1 Table. Nitrogen fixation activity To measure nitrogenase activity of symbiotic nodules, an Acetylene Reduction Assay (ARA) was performed on individual plants with a protocol that was derived from [49]. One month after inoculation with Rhizobium, the plants were placed into 10ml glass vials that were sealed with rubber septa. Acetylene was injected into each vial, and after 1 h of incubation at room temperature, the produced ethylene was measured using Gas Chromatography (7820A, Agilent technologies).

Phylogenetic and statistical analyses
Phylogenetic analyses were performed using the SeaView package (v.4.4.0; [50]). The full-length protein sequences that were retrieved from the NCBI database were aligned using Muscle and optimized with the Gblocks software. Phylogenetic relationships were defined using a maximum likelihood approach. The tree was built with the PhyML software using the LG substitution model and four substitution rate categories. Support for each node was gained by approximate likelihood ratio tests (aLRT SH-like; [51]). The XP_001698687.1 RLK from Chlamydomonas patens, which was identified by a BLAST search of NCBI (E-value = 1e-40), was used as an outgroup to anchor the tree.
The statistical analyses were performed with non-parametric tests (Mann-Whitney when n = 2 independent samples and Kruskal and Wallis when n.2 independent samples). Representative examples of wildtype (WT) and cra2 (alleles 1 to 10) root systems that were grown in vitro for two weeks on an N-deprived ''i'' medium [42]. The blue arrowheads are alleles that are tagged by the Tnt1 retro-element insertion; the green arrowheads are alleles that are tagged by another insertional element; and the yellow arrowhead is an allele containing a deletion of one nucleotide. Bars = 0,5 cm. B. Allelism test between different cra2 alleles as well as with the previously described cra1 mutant [32]. ''yes'' means that both of the mutants are allelic. (PDF) S2 Figure Detail of the cone-shaped transition zone in Medicago truncatula root apical meristems. Detail of the wild-type (WT) apical meristem transition zone (B) of the root that is shown in Fig. 2A (A). The roots were stained with Propidium Iodide to visualize the cell walls. The arrowhead indicates the apical position of the ''cone-shaped'' transition zone between the cell proliferation zone (CPZ) and the cell elongation zone (CEZ). Bars = 100 mm. (PDF) S3 Figure Expression of the root apical meristem markers. A. Amyloplast accumulation in Wild-Type (WT) and cra2-1 root apical meristems revealed by Lugol staining. B. Expression of the P WOX5 :GUS transcriptional fusion in the WT and cra2-1 root apical meristems. C. Real-time RT-PCR analysis of WOX5 expression in the WT and cra2-1 or cra2-2 roots. ACTIN11, RBP1 and H3L genes were used as references [48]. The expression was normalized relative to that of the WT, and the error bars represent standard deviations (n = 3). (PDF) S4 Figure cra2 nodules are elongated and fix nitrogen. A. Picture of a representative elongated nodule from a Wild-Type (WT) or a cra2-1 plant. Bar = 500 mm. B. Nitrogen fixation activity of the WT and cra2 plants (cra2-1 and cra2-2 alleles) six weeks postinoculation with Rhizobium was determined using an Acetylene Reduction Assay (ARA). C. Specific nitrogen-fixation activity of WT and cra2 nodules (cra2-1 and cra2-2 alleles) from plants shown in (B), corresponding to the ARA activity per milligram of nodule. In B and C, a Kruskal and Wallis test was performed (a,5%; n = 10), and the letters indicate significant differences. (PDF) S5 Figure CRA2 gene structure and mutant allele location. A, Predicted CRA2 gene model (FGenesh) and localization of the 10 mutant alleles that were identified by forward or reverse genetic screens. The blue arrowheads are alleles that are tagged by the Tnt1 retro-element insertion; the green arrowheads are alleles that are tagged by another insertional element; and the yellow arrowhead is an allele containing a deletion of one nucleotide. Bar = 250 nucleotides; TSS = predicted Transcription Start Site; polyA: predicted polyadenylation site. B, Nucleotide sequence of the CRA2 genomic region (from the predicted initial ATG start codon to the stop codon) locating the 10 mutant alleles (arrowheads; position related to the predicted ATG). C, Prediction (FGenesh) of a splicing site variant mutation in the cra2-2 allele carrying a single-nucleotide deletion. Red box (1): WT Exon 1; Grey box (2): new exon that was predicted from the new splicing site. The arrows represent the primers that were used for the RT-PCR as shown in (D). D, RT-PCR analysis of the region containing the predicted splicing site in the cra2-2 allele. No differential splicing was detected including after sequencing of the PCR product. E, Sequence of the CRA2 protein. The arrowhead shows the truncated protein that was generated by a frameshift in the cra2-2 allele carrying a single-nucleotide deletion. (PDF) S6 Figure CRA2 expression in various plant organs and growth conditions. The Mtr.38398.1.S1_at probe corresponding to the CRA2 gene on the M. truncatula Affymetrix arrays is shown for the selected organs (including both above-and below-ground organs) and experimental conditions that are available from the MtGEA (Medicago truncatula Gene Expression Atlas database). (PDF) S7 Figure cra2 roots and shoots do not present any detectable defect in vascular bundle patterning. A-G, Representative examples of stem (A-G) or root (D-F) transversal sections of wild-type (WT) and cra2-1 plants that were grown for two months and observed after different stainings: A and D, phloroglucinol staining lignin in red and sclerenchyma in white; B and E, aniline blue staining callose in blue under UV illumination; and C, F and G, toluidine blue staining xylem (Xy) and phloem (Phl) in blue and sclerenchyma (scl) in violet (the detail of a stem vascular bundle is shown in G). Bars = 150 mm in A and B; 50 mm in C-G. H, Quantification of the diameter of the roots and root steles based on transversal sections at one cm above the root apex in the WT and cra2-1 plants that were grown in a greenhouse on a perlite-sand mixture for one month. The error bars represent standard deviations, and a Kruskal and Wallis test was used to determine the significant differences (indicated by letters, a,5%, n = 15). (PDF) S1 Table Primers list. In italics: T7 promoter. (PDF)