The receptor-like cytoplasmic kinase AeRLCK2 mediates Nod-independent rhizobial symbiosis in Aeschynomene legumes

Abstract

Many plants interact symbiotically with arbuscular mycorrhizal fungi to enhance inorganic phosphorus uptake, and legumes also develop a nodule symbiosis with rhizobia for nitrogen acquisition. The establishment and functioning of both symbioses rely on a common plant signaling pathway activated by structurally related Myc and Nod factors. Recently, a SPARK receptor-like kinase (RLK)/receptor-like cytoplasmic kinase (RLCK) complex was shown to be essential for arbuscular mycorrhiza formation in both monocot and dicot plants. Here, we show that in Aeschynomene legumes, the RLCK component of this receptor complex has undergone a gene duplication event and mediates a unique nodule symbiosis that is independent of rhizobial Nod factors. In Aeschynomene evenia, AeRLCK2 is crucial for nodule initiation but not for arbuscular mycorrhiza symbiosis. Additionally, AeRLCK2 physically interacts with and is phosphorylated by the cysteine-rich RLK, AeCRK, which is also required for nodulation. This finding uncovers an important molecular mechanism that controls the establishment of nodulation and is associated with Nod-independent symbiosis.

The discovery of the receptor-like cytoplasmic kinase AeRLCK2 as being important for nodulation in Aeschynomene evenia sheds light on the molecular mechanisms of the Nod-independent symbiosis.

Introduction

Plants have evolved a range of mutualistic partnerships with soil-dwelling microorganisms to enhance their nutrient uptake. The oldest and most widespread symbiosis is the association with Glomeromycotina fungi, referred to as arbuscular mycorrhizal (AM) fungi. AM fungi develop extensive hyphal networks to take up inorganic phosphorus (Pi) and other nutrients from the soil. They also colonize plant roots and form intracellular branched structures called arbuscules, through which they provide these nutrients to plants (Rich et al. 2021). Plant species within the nitrogen (N)-fixing clade are able to develop an additional symbiosis with diazotrophic bacteria that are hosted in root nodules (van Velzen et al. 2019). This interaction allows the plants to access the abundant atmospheric N by converting it into ammonium (Roy et al. 2020). By providing Pi and N to the plants, these symbioses are central to the functioning of natural ecosystems and the productivity of agro-systems.

Genetic studies of Oryza sativa (rice) and model legumes (such as Lotus japonicus and Medicago truncatula), establishing a symbiosis with AM fungi and/or rhizobia, have demonstrated that the evolution of nitrogen-fixing symbiosis has co-opted perception, signaling, and infection mechanisms essential for the establishment of AM (Radhakrishnan et al. 2020). Chitooligosaccharides (COs) and lipo-chitosaccharides (LCOs) produced by AM fungi and LCOs produced by rhizobia (known as Nod factors) are perceived by distinct plasma membrane receptor-like kinases of the LysM-RLK subfamily (Buendia et al. 2018; Feng et al. 2019; Ding et al. 2024, 2025; Zhang et al. 2024). These signals are then transduced by SymRK (Symbiosis Receptor Kinase), a RLK belonging to the LRR-RLK subfamily, which initiates a common symbiosis signaling pathway leading to transcriptional reprograming (Gobbato 2015; Roy et al. 2020).

This transcriptional reprogramming drives infection by both AM fungi and rhizobia, culminating in their intracellular accomodation either into the plant root or into nodules (Roy et al. 2020). Another RLK belonging to the SPARK-RLK subfamily, KIN3 (KINASE 3), is crucial for the arbuscule formation during AM (Irving et al. 2022; Leng et al. 2023). Recently, KIN3 has been shown to interact with 2 receptor-like cytoplasmic kinase (RLCK) paralogs, ARBUSCULAR MYCORRHIZA-INDUCED KINASES 8 and 24 (AMK8 and AMK24), which also play a key role in arbuscule formation in L. japonicus (Leng et al. 2023). This RLK/RLCK complex was found to have a conserved role in AM in rice (Bravo et al. 2016; Roth et al. 2018; Montero et al. 2021; Leng et al. 2023). Interestingly, KIN3 orthologs are only found in plants able to form AM and KIN3 is dispensable for rhizobial symbiosis in M. truncatula (Irving et al. 2022). This questions whether the KIN3-interacting RLCKs are required or not for nodulation.

We challenged this view through the genetic analysis of nodulation in Aeschynomene evenia. This legume species has emerged as a model of choice for the study of a unique N-fixing symbiosis with photosynthetic Bradyrhizobium strains that do not produce Nod factors (Arrighi et al. 2012; Chaintreuil et al. 2016). In this process, bacteria intensely colonize crowns of axillary root hairs (ARHs) located at lateral root bases that are the initial sites for bacterial entry. From there, they progress intercellularly towards the root cortex, where a nodule primordium finally forms (Horta Araújo et al. 2024). While the nature of the bradyrhizobial signals that activate this so-called Nod-independent symbiosis remain to be identified, progress has been made on the plant side in recent years, thanks to the availability of a reference genome and a collection of ethyl methane sulfonate (EMS) nodulation mutants for A. evenia (Quilbé et al. 2021, 2022). It is now known that many components of the common symbiotic signaling pathway are conserved in A. evenia but that this pathway is likely activated by other type of receptor proteins. Notably, a Cysteine-rich RLK, AeCRK, was recently shown to be required to establish the Nod-independent symbiosis (Quilbé et al. 2021, 2022).

In this work, we identified AeRLCK2, a RLCK homolog to L. japonicus AMK8, as required for the Nod-independent symbiosis. Through the genetic, phenotypic and molecular characterization of allelic mutants, we show that AeRLCK2 is essential for nodulation but dispensable for AM. Furthermore, we provide evidence that AeRLCK2 originated from a recent tandem duplication event and offer clues to explain how it has evolved for a role in the Nod-independent symbiosis. Finally, we demonstrate that AeRLCK2 physically interacts with and serves as a phosphorylation substrate for AeCRK. These findings shed light on the evolution of RLCKs in plant symbioses and further elucidate the mechanisms by which the Nod-independent nodulation process is triggered in Aeschynomene legumes.

Results

Mutant-based identification of AeRLCK2, a symbiosis receptor-like cytoplasmic kinase

To identify genes involved in establishment of the Nod-independent symbiosis, we analyzed a set of 12 uncharacterized nodulation mutants obtained from screening in greenhouse conditions of an EMS-mutagenized population of A. evenia CIAT22838 (Supplementary Table S1) (Quilbé et al. 2021). The common characteristic of these mutants is that most plants in each mutant line had a Nod- phenotype similar to ccamk-2 mutant plants (Quilbé et al. 2022), while a few plants formed 1 or few enlarged nodules, when inoculated with the photosynthetic Bradyrhizobium strain sp. ORS278, (Fig. 1A). This type of nodulation was qualified as a Big Nodule (BN) phenotype. For these mutants, Nod- plants displayed typical nitrogen starvation symptoms (reduced plant growth and yellow leaves), whereas BN plants were better developed with green leaves, indicating that the formed BN nodules had nitrogen-fixing activity (Fig. 1B).

Figure 1.

Mutant-based identification of AeRLCK2 as required for Nod-independent symbiosis. A) Root phenotypes of Wilde Type (WT), ccamk-2 and rlck2-11 plants at 28 dpi with Bradyrhizobium ORS278 strain and grown in greenhouse conditions. Nod⁺: presence of WT nodules, Nod⁻: Absence of nodules, BN: Big Nodule (white arrowhead). Scale bar: 1 cm. B) Aerial phenotypes of the same plants grown for 15 additional days in greenhouse conditions after analysis of their root nodulation. Scale bar: 10 cm. C) Frequency of EMS-induced mutant alleles in pools of Nod⁻ backcrossed F2 plants derived from the rlck2-11 mutant using mapping-by-sequencing. The SNP corresponding to the putative causal mutation in rlck2-11 is marked with a black arrowhead. D)  AeRLCK2 gene and protein structure. Upper panel: genomic region of chromosome Ae01 containing the Ae01g26600 locus. Filled arrows indicate RLCK genes. Middle panel: gene structure of AeRLCK2. Blue boxes represent exons and red lines indicate the positions of the EMS mutations in the rlck2 mutants. Bottom panel: domain structure of the predicted AeRLCK2 protein. White boxes indicate the positions of the predicted domains: TM for transmembrane domain and KD for kinase domain. E) Functional complementation of A. evenia rlck2-11. Hairy roots of rlck2-11 transformed with either the empty vector (left images) or containing the AeRLCK2 CDS under the control of pLjUb (right images) at 14 dpi with Bradyrhizobium ORS278. GFP (Green Fluorescent Protein) was used as a plant transformation marker. Scale bar: 500 µm.

Phenotypic analysis of F2 progeny generated from crosses between the WT line and the 12 nodulation mutants revealed that they segregated in a 3:1 ratio of plants with a WT-like nodulation phenotype (with numerous and normal sized pink nodules) to plants with either a Nod- or BN phenotype, respectively (Supplementary Table S2). These data confirmed the dual mutant phenotype and demonstrated the monogenic and recessive nature of each mutation. To identify the gene(s) responsible for this dual nodulation phenotype, we performed a mapping-by-sequencing approach on bulked F2 mutant plants for 8 mutants. Linkage mapping repeatedly identified the same locus near the end of the Ae01 chromosome, with all mutations located in the Ae01g26600 gene (Fig. 1C, Supplementary Fig. S1 and Supplementary Table S2). Given these results, we amplified and sequenced the Ae01g26600 gene in the remaining 4 mutants and all had mutations (Supplementary Table S2). Further allelism tests between the 12 mutants showed that they belonged to the same complementation group, clearly establishing the Ae01g26600 mutations as responsible for the dual Nod-/BN phenotype (Supplementary Table S3).

Blast analysis revealed that Ae01g26600 codes for a RLCK (receptor-like cytoplasmic kinase), a subtype of RLK that lacks lack extracellular ligand-binding domains (Lin et al. 2013; Liang and Zhou 2018). However, we observed that the coding sequence (CDS) of Ae01g26600 was twice as long as that of other RLCK genes. Consistent with this, A. evenia RNAseq datasets showed that there are actually 2 different RLCK genes at the Ae01g26600 locus (Chaintreuil et al. 2016; Quilbé et al. 2021). This suggested that the Ae01g26600 gene is misannotated in the current A. evenia reference genome. We corrected this by manually delineating the 2 genes organized in tandem and named them AeRLCK1 and AeRLCK2. As all 12 identified mutations were in the AeRLCK2 gene, we numbered the corresponding allelic mutants rlck2-1 to rlck2-12 (Fig. 1D and Supplementary Table S2).

To validate our curated annotation, the AeRLCK2 CDS was amplified from WT A. evenia cDNAs and cloned downstream a L. japonicus ubiquitin promoter (pLjUb). Using Agrobacterium rhizogenes-mediated hairy root transformation, the construct was introduced into the roots of the strong allele mutant rlck2-11, characterized by a nonsense mutation. WT-like nodulation was restored in rlck2-11 (Fig. 1E and Supplementary Table S4). Functional annotation of the 403 amino acid AeRLCK2 predicted the presence of a transmembrane domain followed by a serine/threonine kinase domain, where all mutations were identified (Fig. 1D). We speculated that AeRLCK2 encodes a symbiosis plasma membrane-localized RLCK whose kinase domain integrity is essential to mediate signal transduction.

AeRLCK2 perfoms a key function in rhizobial symbiosis

To characterize in more detail the role of AeRLCK2 in nodulation, we performed in vitro growth chamber nodulation assays with Bradyrhizobium ORS278 on 4 mutant lines, rlck2-1, rlck2-5, rlck2-10 and rlck2-11, using the WT line and ccamk-2 mutant as controls. Nodulation kinetics revealed that in the WT line, nodules were pink at 10 dpi while in the rlck2 mutants the first BN nodules emerged at 14 dpi (Supplementary Fig. S2). At 21 dpi, the nodulation frequency of the plants was 100% for WT, with each plant containing numerous nodules, and 0% for the ccamk-2 mutant plants. In contrast, 30% to 80% of the rlck2 mutant plants were devoid of nodules, while the others contained 1 or a few BNs (see below) (Fig. 2A). This points to an important role of AeRLCK2 in nodule formation.

Figure 2.

Bradyrhizobium infection and symbiotic signaling in rlck2 mutants. A) Frequency of nodule occurrence at 21 dpi in Wild Type (WT), ccamk-2, rlck2-1, rlck2-5, rlck2-10 and rlck2-11 plants. Error bars represent SD (n = 3 biological replicates from independent experiments, with at least 15 plants per line in each replicate). B) Comparison of root nodulation phenotypes in WT, ccamk-2 and rlck2-11, under noninoculated (NI) or Bradyrhizobium ORS278-inoculated (I) plants at 21 dpi. Note the presence of either a Nod⁻ or an BN phenotype in rlck2-11 inoculated roots. Scale bar: 1 mm. C) ARH diameter in WT, ccamk-2 and rlck2-11 at different time-points in noninoculated (NI) and inoculated (I) plants with Bradyrhizobium ORS278. T0: time 0. T14: time 14 d after inoculation or not. Dots represent individual measurements. D) ARH colonization of WT, ccamk-2 and rlck2-11 plants at 21 dpi with GUS-tagged ORS278, observed on whole roots (upper and middle panels) and root sections (lower panels). Scale bars: 1 mm (upper panels) and 0.5 mm (middle and lower panels). E) Expression of nodulation-induced gene in WT, ccamk-1 and rlck2-11 plants. Relative expression levels (Rel. exp. level) of AeNIN (NODULE INCEPTION), AeSymREM1 (Symbiotic REMORIN 1), AeENOD40 (EARLY NODULIN 40), AeSBT (SUBTILASE), AeVPY (VAPYRIN) and AeCRK (Cystein-rich Receptor-like Kinase) were measured by RT-qPCR in plant roots at 0, 2, 4 and 7 dpi. The results were normalized against AeEF1a and Ubiquitin housekeeping genes. Data presented in boxplots correspond to 4 biological replicates, each derived from independent experiments, with at least 5 plants per line in each replicate. Different letters indicate significant differences between conditions as determined by analysis of variance (Kruskal-Wallis) and post-hoc analysis (Dunn's test), P < 0.05. Box plots showing the median (bold segment), interquartile range (box from Q1 to Q3), minimum and maximum (whiskers), and outliers (individual points).

Interestingly, inoculated roots of both Nod- and BN rlck2 plants displayed well-developped crowns of ARHs at lateral root bases, similar to noninoculated plants and the ccamk-2 mutant (Nod- strict), whereas on inoculated WT roots these ARHs were small (Fig. 2B). In A. evenia, these ARHs are the first colonization sites of bradyrhizobia and their development is tightly controlled by the nitrogen status of the plant (Quilbé et al. 2022). Kinetics of ARH development showed that they develop over time in uninoculated WT plants, whereas their development is suppressed in inoculated WT plants (Fig. 2C). No such Bradyrhizobium-induced repression of ARH development was observed in rlck2 and ccamk-2 mutants. This observation led us to analyse at which stage infection is blocked in rlck2 mutants. X-Gluc staining of ORS278-GUS inoculated rlck2-11 mutant roots at 21 dpi revealed intense blue staining on the surface and between the ARHs, but no penetration into the inner root cortex was observed (except for the BN), similar to the ccamk-2 mutant (Fig. 2D). In contrast, crk mutants, which develop small bumps containing infection pockets after inoculation (Quilbé et al. 2022), showed reduced ARH development in the presence of ORS278 (Supplementary Fig. S3). RT-qPCR analysis of 6 symbiosis-induced marker genes, NODULE INCEPTION (AeNIN), VAPYRIN (AeVPY), Symbiotic REMORIN 1 (AeSymREM1), EARLY NODULIN 40 (AeENOD40), SUBTILASE (AeSBT) and AeCRK (Quilbé et al. 2022), showed no induction of their expression in rlck2-11 and ccamk-2 inoculated with the ORS278 strain (Fig. 2E). Taken together, these results indicate that the mutations in AeRLCK2 block early symbiosis responses.

Additionally, rlck2 mutant plants showed a drastic decrease in nodule numbers. WT plants contained an average of 35 nodules per root at 21 dpi, while roots of most nodulated rlck2 mutant plants exhibited only 1 or 2 BN nodules (Fig. 3A). These BNs had an average diameter that was twice as large as that of WT nodules (Fig. 3B). We interpret this as a compensatory mechanism for their very few numbers in rlck2 mutants. Roots of rlck2 mutant plants containing BN nodules had nitrogenase enzyme activity, as measured by the acetylene reduction assay (ARA), and accordingly these plants carried green leaves (Fig. 3C, Supplementary Fig. S4). Light microscopy analysis of BN nodule sections from rlck2-11 plants showed that they had a similar structure to WT nodules, with a central tissue infected by bacteria and a peripheral uninfected nodule cortex containing vascular bundles (Fig. 3D). However, very often these BN nodules also contained small necrotic zones and brown spots. These areas had an intense yellow/green fluorescent appearance when observed with a FITC filter (pseudocolored in magenta for visualization purposes), suggesting the presence of polyphenolic compounds. Confocal microscopic analysis of the same nodule sections showed that, in contrast to WT nodules, BN nodules contained unevenly infected plant cells. In general, the infected plant cells of BN nodules contained spherical bacteroids, but in some cases, rod-shaped undifferentiated bacteria were also observed (Fig. 3E). These observations highlight an important role of AeRLCK2 in nodule infection and bacterial differentiation.

Figure 3.

Nodule development and colonization by Bradyrhizobium in rlck2 mutants. A) Number of pink nodules formed on nodulated plants in Wild Type (WT), rlck2-1, 5, 10 and 11 plants at 21 dpi with Bradyrhizobium ORS278. Numbers below the boxplots indicate the number of nodulated plants relative to the total number of inoculated plants. B) Nodule diameter and C) nitrogenase enzyme activity measured by ARA (n ≥ 3 nodulated roots per line and biological replicate) from the same plants as in A. Data in A to C correspond to 3 biological replicates from independent experiments. Dots represent individual plants. Letters indicate significant differences between conditions, as determined by analysis of variance (Kruskal-Wallis) and post-hoc analysis (Dunn's test), P < 0.05. Box plots showing the median (bold segment), interquartile range (box from Q1 to Q3), minimum and maximum (whiskers), and outliers (individual points). D) Cross-sections of WT and rlck2-11 nodules observed under brightfield (top) or FITC filter (bottom). Yellow/green fluorescence was pseudocolored in magenta, and red fluorescence was pseudocolored in yellow for visualization purposes. White arrows indicate the occurrence of defense-like responses within the nodule. Scale bar: 500 μm. E) Cytological analysis of nodule cross-sections from WT and rlck2-11 plants using a confocal microscope after staining with SYTO9 (live bacteria), propidium iodide (infected plant nuclei and dead bacteria or bacteria with a compromised membrane) and calcofluor (plant cell wall). For visualization purposes, SYTO 9 (originally green) was pseudocolored in magenta, propidium iodide (originally red) in yellow, and calcofluor (originally blue) in cyan. White arrows show elongated bacteria. Scale bars: 500 μm (top), 50 μm (bottom).

AeRLCK2 physically interacts with and is phosphorylated by AeCRK

Recently, we showed that a Cysteine-rich RLK-coding gene, AeCRK, is essential for the establishment of the N-fixing symbiosis in A. evenia (Quilbé et al. 2021, 2022). Since RLCKs are known to interact with RLKs to mediate downstream signaling (Lin et al. 2013; Liang and Zhou 2018), we investigated the hypothesis that AeRLCK2 and AeCRK form a plasma membrane-bound complex. For this, we first examined the subcellular localization of AeCRK and AeRLCK2 in Nicotiana benthamiana leaves, by generating a translational fusion with the Yellow Fluorescent Protein (YFP). Transient expression of AeCRK-YFP induced cell death in N. benthamiana leaves 5 d after Agrobacterium tumefaciens-mediated transformation. In contrast, an engineered kinase-dead version (AeCRKG359E) with a mutation in the glycine-rich loop did not trigger cell death and was therefore used (Supplementary Fig. S5). The YFP fusion constructs were transiently expressed in N. benthamiana leaves in combination with the plasma membrane marker MtLYK3 (Klaus-Heisen et al. 2011) fused with Cyan Fluorescent Protein (CFP) (Fig. 4A). AeCRKG359E and AeRLCK2 co-localized with MtLYK3, confirming their targeting to the plasma membrane. Since, AeRLCK2 is atypical in harboring a predicted transmembrane domain (TM), we also tested an N-terminal truncated version of AeRLCK2 (AeRLCK2ΔTM-YFP). For this latter, the signal was observed in the nucleus and cytoplasmic threads, indicating that the predicted TM is important for the protein anchoring to the plasma membrane (Fig. 4A).

Figure 4.

AeRLCK2-AeCRK interaction and kinase assays. A) Confocal microscopy observations of Nicotiana benthamiana leaf cells showing plasma membrane localization of AeCRK^G359E-YFP, AeRLCK2-YFP, and nucleo-cytoplasmic distribution of the truncated transmembrane version of RLCK2 (AeRLCK2^ΔTM). MtLYK3-CFP (Cyan FP) was used as a plasma membrane marker. Scale bar: 20 µm. B) Co-immunoprecipitation assay showing interaction of AeCRK^G359E-mCherry with AeCRK^G359E-YFP and AeRLCK2-YFP. Proteins were immunoprecipitated (IP) with αRFP magnetic agarose beads and co-purified proteins were detected with αGFP (Green FP) antibodies (upper panel). Input (middle panel) and band intensities were calculated and normalized to the negative control MtLYK3 (bottom panel, ranging from 2 to 4 biological replicates from independent experiments). Error bars represent SD, dots show biological replicates. Ponceau staining was used as loading control. C) Split-luciferase assay showing the interaction of AeCRK^G359E-CLuc with AeCRK^G359E-NLuc or AeRLCK2-NLuc (N/C-terminal part of the Luciferase). Boxplots represent bioluminescence intensity from 7 biological independent replicates. Dots show individual measurements. Expression levels of 3Flag-CLuc and 3HA-NLuc fusions were assessed by western blot (Supplementary Fig. S6). Bioluminescence intensities were normalized to protein expression and data were Log-transformed (Log₁₀). Letters indicate significant differences between samples, as determined by analysis of variance (Kruskal-Wallins) and post-hoc analysis (Dunn's test), P < 0.05. Box plots showing the median (bold segment), interquartile range (box from Q1 to Q3), minimum and maximum (whiskers), and outliers (individual points). RLU, Relative luminescence unit. D) Kinase activity assay showing transphosphorylation of AeRLCK2 by AeCRK in Nicotiana benthamiana leaf cells. Full-length YFP-tagged proteins were immunoprecipitated with αGFP magnetic agarose beads. Phosphorylation status was analyzed after SDS-PAGE and detected with anti-S, -T and -Y antibodies. Asterisks indicate the phosphorylation status of AeRLCK2-YFP (top). Input (bottom).

Next, we tested AeCRK-AeRLCK2 interactions by co-immunopurification (IP) assays. The mCherry-tagged AeCRKG359E was co-expressed with YFP-tagged AeCRKG359E, RLCK2, RLCK2ΔTM and the negative control MtLYK3-CFP in N. benthamiana leaves. After IP of AeCRKG359E, the co-purified proteins were detected with αGFP antibodies (Fig. 4B). AeCRKG359E-YFP and AeRLCK2-YFP were enriched 7.37-fold and 2.9-fold, respectively, compared with the negative control MtLYK3-CFP (Fig. 4B). Conversely, no significant enrichment was observed for AeRLCK2ΔTM-YFP (Fig. 4B). To assess these pairwise interactions, we performed split-luciferase assays by fusing AeCRKG359E with the C-terminal part of the luciferase (3Flag-CLuc) and the potential interactors with the N-terminal part of the luciferase (3HA-NLuc). Combination of -NLuc and -CLuc fusion proteins were co-expressed in N. benthamiana leaves. The corresponding bioluminescence was measured following luciferin infiltration, and the data were normalized by the expression level of the -NLuc fusion proteins (Supplementary Fig. S6). Co-expression of AeCRKG359E-CLuc with AeCRKG359E-NLuc or AeRLCK2-NLuc, but not with RLCK2ΔTM-NLuc, resulted in significantly higher bioluminescence intensities compared with MtLYK3-NLuc (Fig. 4C and Supplementary Fig. S6). Thus, these findings consistently showed that AeCRK can form homodimers and physically interact with AeRLCK2. The N-terminal domain of AeRLCK2, which confers plasma membrane localization, is essential for this interaction.

AeCRK and AeRLCK2 have typical Ser/Thr kinase domains, suggesting that their interaction may involve phosphorylation events. We therefore investigated their kinase activities. The kinase domains (KD) of AeCRK and AeRLCK2 were translationally fused to a GST-tag and expressed in Escherichia coli. After purification, their autophosphorylation activity was studied in vitro using radiolabelled ATP (32P-ATP). Autoradiography revealed a robust autophosphorylation activity for AeCRK^KD, whereas AeRLCK2^KD was much less efficient (Supplementary Fig. S7). The AeRLCK2^KD-G110E mutant, containing a mutation in the glycine-rich loop, found in rlck2-6, failed to incorporate 32P-ATP indicating a lack of kinase activity. Transphosphorylation studies showed that AeCRK^KD phosphorylated the kinase-dead AeRLCK2^KD-G110E but not free GST (Supplementary Fig. S7). These results were confirmed by in planta experiments with the full-length YFP-tagged proteins transiently expressed in N. benthamiana leaves. After IP, the phosphorylation status of the proteins was assessed using an antibody that recognizes phosphorylated serine, threonine, and tyrosine residues (Fig. 4D). AeCRK was highly phosphorylated whereas AeCRK^G359E showed either no or low levels of phosphorylation. Surprisingly, no phosphorylation was observed for AeRLCK2 in planta. To investigate the possibility of trans-phosphorylation, AeCRK-YFP was co-expressed with AeRLCK2-YFP or AeRLCK2^G110E-YFP. In both cases, AeRLCK2 was phosphorylated, whereas this was not observed when using the AeCRK^G359E version (Fig. 4D). Finally, the phosphorylation sites of AeRLCK2 targeted by the kinase activity of AeCRK were searched by comparative LC-MS/MS analysis of in planta immunopurified AeRLCK2 produced alone or together with AeCRK or AeCRK^G359E. Five phosphorylation sites, corresponding to 3 threonines (T66, T90 and T106) and 2 serines (S133 and S321) were identified in the cytoplasmic region of AeRLCK2 specifically in the presence of AeCRK (Supplementary Figs. S8 and S9). Consistently, AeCRK transphosphorylation of AeRLCK2 was also found using antibodies that recognize only phosphorylated threonine residues (Supplementary Fig. S8). To determine whether the AeCRK-mediated phosphorylation of AeRLCK2 is important for its function in nodulation, a phospho-silent version (all 5 sites were mutated to Ala—5A) was generated. YFP-tagged AeRLCK2 and AeRLCK2-5A were expressed under the control of pLjUb into the rlck2-11 mutant via A. rhizogenes. Following inoculation with Bradyrhizobium ORS278, the rlck2-11 plants expressing AeRLCK2-5A showed a partially stunt plant development and reduced nodule formation compared with plants expressing WT AeRLCK2, although the 2 AeRLCK2 forms showed similar levels of expression (Supplementary Fig. S10). Taken together, these results demonstrated that AeCRK and AeRLCK2 have distinct kinase activities, that AeCRK transphosphorylate AeRLCK2 on specific residues, and that these phosphosites contribute to AeRLCK2 function in nodulation.

AeRLCK2 arose from a duplication of a mycorrhiza-conserved gene in Aeschynomene

Since AeRLCK2 is tandemly organized with AeRLCK1 in the A. evenia genome, we investigated whether this RLCK gene tandem is present in other legumes. Synteny analysis based on genome sequence comparisons revealed the presence of a single RLCK homolog at the same locus in the analyzed legume species (Supplementary Fig. S11). To specify the relationships between AeRLCK1, AeRLCK2 and RLCK homologs, we analysed the genome sequences of 15 legume species (13 Papilionoideae and 2 Caesalpinoideae) and 3 nonlegume species (Supplementary Data Set 1). We also included in this analysis RNAseq data from Aeschynomene afraspera, a close relative of A. evenia that uses a Nod-dependent symbiosis (Bonaldi et al. 2011) (Supplementary Table S5). This search retrieved RLCK homologs for each analysed plant species except Arabidopsis thaliana and lupin sp, 2 species unable to form AM, consistently with previous phylogenomic studies that predicted their conservation for AM (Supplementary Table S6) (Bravo et al. 2016). In rice and L. japonicus, the homologous genes OsRLCK171, LjAMK8 and LjAMK24 were demonstrated to be essential for AM (Leng et al. 2023). Phylogenetic reconstruction, based on protein sequences, revealed that the RLCK homologs present in Papilionoideae legume species were distributed in 2 sister clades, 1 containing LjAMK8 and the other 1 LjAMK24 (Fig. 5A). These 2 clades most probably originated from the ancient whole genome duplication (WGD) in the Papilionoideae subfamily, indicating that the rice gene OsRLCK171 is pro-ortholog of the 2 Papilionoid paralogs, LjAMK8 and LjAMK24. In A. afraspera, AaRLCK_O and AaRLCK_P also corresponds to the paralogous Papilionoid gene pair. In contrast, the 2 A. evenia RLCK genes, AeRLCK1 and AeRLCK2, clustered together in the clade harboring LjAMK8 and AaRLCK_O, whereas the expected paralog was missing (Fig. 5A). This suggests a recent AeRLCK1-AeRLCK2 duplication accompanied by the loss of the paralogous gene in A. evenia.

Figure 5.

Phylogeny of legume RLCK genes and evolution in Aeschynomene species. A) Maximum likelihood (ML) phylogenetic reconstruction of the orthogroup containing AeRLCK2. Color coding indicates nonpapilionoid RLCKs (green), the 2 papilionoid RLCK clades (purple and yellow) putatively originating from the 58-MA WGD event (green dot), and the 2 RLCK copies present in A. evenia (red), which are derived from a recent tandem duplication (red dot). B) Detection of different RLCK gene versions in Aeschynomene species and the closely related species Soemmeringia semperflorens. The ML phylogenetic tree was constructed using concatenated ITS and matK sequences. Green stars indicate a Nod-dependent symbiosis and red stars indicate a Nod-independent symbiosis. The RLCK_O, RLCK1 and/or RLCK2 copies were identified in available RNAseq data (orange square) and by PCR amplification on genomic DNA (blue square). A and B support values were determined using 100,000 iterations of the ultrafast bootstraps approximation (UFboot). Tree scale: mutations per site. C) Domain structure of AaRLCK_O, AeRLCK1 and AeRLCK2 and sequence similarities between the proteins. White bars indicate predicted domains. TM, transmembrane domain; KD, kinase domain; AA, amino acids. Intensities of blue shaded backgrounds delineate zones with different level of sequence identity. All domains are to scale.

To clarify when these gene changes occurred in Aeschynomene, we searched for RLCK orthologs among RNAseq data previously generated from roots and nodules for 11 Aeschynomene species in the Nod-independent clade (Quilbé et al. 2021). We found an AeRLCK2 ortholog for each of these species and an AeRLCK1 ortholog only for A. scabra, but no putative RLCK paralog was detected (Fig. 5B, Supplementary Data Set 2). We completed this analysis by experimental investigation of their presence in Aeschynomene species (Brottier et al. 2018). To this end, we designed primers matching conserved or specific regions to AaRLCK_O, AeRLCK1 or AeRLCK2 copies and screened by PCR amplification followed by amplicon sequencing in Aeschynomene species and the allied species, Soemmeringia semperflorens. Sequences similar to AeRLCK1 and AeRLCK2 were identified in the 11 Nod-independent Aeschynomene species as for A. evenia, whereas a single RLCK sequence was recovered in the 5 Nod-dependent species as in A. afraspera (Fig. 5B). Thus, there is a perfect correlation between the Nod-independent symbiosis and the RLCK gene duplication in Aeschynomene legumes. Interestingly, no additional gene tandems or clusters specific to the Nod-independent Aeschynomene lineage could be evidenced in the set of 138 genes predicted to be required for AM (Supplementary Table S6) (Bravo et al. 2016).

To substantiate the changes associated with AeRLCK2, we compared the type of RLCK genes present in Aeschynomene species. Protein sequence alignment and 3D modeling showed that they have the same general structure and share a highly conserved sequence (AeRLCK2 has 86% and 80% amino acid identity with AeRLCK1 and AaRLCK_O, respectively) (Fig. 5C and Supplementary Fig. S12). However, RLCK2 proteins are shorter at both the N- and C-terminus compared with the other RLCK proteins. We next assessed the extent of gene structural variation using AaRLCK_O cDNA and AeRLCK1/AeRLCK2 genomic sequences. Significant differences were observed in their 5′- and 3′-UTR regions as well as in their flanking exons (Supplementary Fig. S13). We also identified a sequence in the promoter region of AeRLCK2 that is highly similar to a downstream gene, Ae01g26580 (Supplementary Fig. S14). Our interpretation is that both the ancestral RLCK and the downstream genes underwent a gene tandem duplication in the ancestor of the Nod-independent Aeschynomene species. Subsequently, complex rearrangements occurred in the promoter region and gene extremities of the RLCK2 copy (Supplementary Fig. S14). These data indicate that the evolution by duplication of AeRLCK2 correlates with the evolution of the Nod-independent symbiosis.

AeRLCK2 is not essential for arbuscular mycorrhiza

To determine whether AeRLCK2 is important for AM, as its homologs in rice (OsRLCK171) and L. japonicus (LjAMK8 and LjAMK28) (Leng et al. 2023), we tested 4 rlck2 mutants together with WT plants and the ccamk-2 mutant line. In contrast to the completely mycorrhiza-free ccamk-2 mutant, roots of both WT and the 4 rlck2 mutants contained fungal hyphae, arbuscules and vesicles, 6 wk after inoculation with Rhizophagus irregularis spores (Fig. 6A). Quantitative assessment of AM levels using the Trouvelot method further showed that the rlck2 mutants were colonized similarly to the WT plants (Fig. 6B, Supplementary Fig. S15).

Figure 6.

Arbuscular mycorrhizal (AM) root colonization in rlck2 mutants. A) Microscopy images of R. irregularis colonization of Wild Type (WT), ccamk-2 and rlck2 mutants at 6 wk post-inoculation (wpi), stained with Sheaffer skrip ink. Scale bars: 50 µm. B) Box plots show the colonization frequency and intensity, both expressed as percentages, in 6 wpi WT, ccamk-2 and rlck2-11 plants. C) Analysis of AM-induced gene expression in WT, ccamk-2 and rlck2-11 plants. Relative expression levels (Rel. exp. level) of plant AeRAM1 (Reduced Arbuscular Mycorrhization 1), AeVPY (VAPYRIN), AeSTR (STUNTED ARBUSCULE), AeSBTM1 (subtilase gene induced during mycorrhization) and fungal RiLSU (large ribosomal subunit), RiGADPH (glyceraldehyde 3-phosphate dehydrogenase) genes were measured by RT-qPCR in roots of 6 wpi plants. The results were normalized against AeEF1a and Ubiquitin. Data presented in boxplots correspond to 4 biological replicates from independent experiments, with 5 plants per line in each replicate. Box plots show the median (bold segment), interquartile range (box from Q1 to Q3), minimum and maximum (whiskers), and outliers (individual points). Dots show individual measurements. Letters indicate significant differences between lines, as determined by analysis of variance (Kruskal-Wallis) and post-hoc analysis (Dunn's test), P < 0.05.

To deepen the analysis, we focused on rlck2-11. At 6 wpi, there was again no difference in both the frequency and intensity of colonization between the rlck2-11 mutant and the WT line (Fig. 6B). In parallel, quantification of the fungal RiLSU (large risobomal subunit) and RiGADPH (glyceraldehyde 3-phosphate dehydrogenase) gene expressions, as markers of fungal biomass in the root tissues was performed by RT-qPCR analysis. Expression levels of these fungal genes in rlck2-11 roots were equivalent to those in WT roots, indicating that R. irregularis colonization in A. evenia roots is not affected by mutation in AeRLCK2 (Fig. 6C). We also determined the expression level of plant AM-induced genes, Reduced Arbuscular Mycorrhization 1 (AeRAM1), AeVPY, STUNTED ARBUSCULE (AeSTR) and AeSBTM1 (Subtilase gene induced during mycorrhization) (Quilbé et al. 2022) by RT-qPCR analysis in rlck2-11. In this mutant, the induction levels of all genes tested were similar to those in WT plants (Fig. 6C). Therefore, the rlck2 mutants appeared to develop functional AM.

The lack of any detectable mycorrhizal phenotype could either reflect an absence of involvement in AM of AeRLCK2, as recently evidenced for AeCRK (Quilbé et al. 2022), or a potential functional redundancy with AeRLCK1 in AM. To address these 2 possibilities, we first considered the fact that LjAMK8, LjAMK24 and the functionally related LjKIN3 are AM-induced genes in L. japonicus (Leng et al. 2023) to investigate whether such an upregulation of expression occurs for A. evenia homologs. To enable fine expression analysis, we generated RNAseq data for WT A. evenia inoculated or not with R. irregularis (Supplementary Table S7). As previously observed (Quilbé et al. 2022), the AM-marker genes AeRAM1, AeVPY, AeSTR and AeSBTM1 were well induced during AM (Supplementary Fig. S16). We observed a strong induction during AM for AeKIN3 (Ae06g09820), the putative ortholog of LjKIN3, whereas the induction level of AeRLCK1 was weaker and expression of AeRLCK2 itself appeared to be unaffected (Supplementary Fig. S16). Since measuring expression at the root level does not always allow determining cell-specific expression, we also used the AeRLCK2 upstream region (∼2.5 kb including the 5′-UTR) to examine the AeRLCK2 expression pattern during mycorrhization. Interestingly, the transgenic hairy roots harboring the pAeRLCK2-GUS construct showed a clear staining in mycorrhizal cells as evidenced for LjAMK8, LjAMK24 (Supplementary Fig. S17) (Leng et al. 2023). Considering that AeRLCK2 results from a recent duplication event, this AM-related expression might either represent a reminiscent transcriptional response to AM or, alternatively, functional redundantly with AeRLCK1.

AeRLCK2 shows adaptations to the Nod-independent symbiosis

To question how the Nod-independent Aeschynomene-specific RLCK gene duplication may have led to the involvement of AeRLCK2 in nodulation, we compared the functionality of AeRLCK2 with AeRLCK1, the duplicated gene in A. evenia, and with AaRLCK_O, the corresponding single copy gene in A. afraspera. First, the 3 RLCK proteins fused to YFP were expressed in N. benthamiana leaves. In contrast to AeRLCK2, cell death was observed in leaves expressing AaRLCK_O and AeRLCK1, at 8 dpi (Supplementary Fig. S18A). Confocal microscopy analysis performed at 3 dpi, i.e. before the onset of cell death, suggested that both AaRLCK_O and AeRLCK1 proteins reside at the plasma membrane, akin to AeRLCK2 (Supplementary Fig. S18B). As already evidenced for AeRLCK2, AaRLCK_O and AeRLCK1 showed no autophosphorylation activity in planta (Supplementary Fig. S18C). In contrast, in vitro kinase assays showed autophosphorylation activity for the kinase domain of the 3 RLCKs but at different levels, according to the following gradient: AaRLCK_O^KD > AeRLCK1^KD > AeRLCK2^KD (Supplementary Fig. S18D).

To further test whether AaRLCK_O, AeRLCK1 and AeRLCK2 are functionally equivalent, we performed cross-complementation studies. We used the AeRLCK2 promoter to drive the expression of the AaRLCK_O, AeRLCK1 and AeRLCK2 CDS in the rlck2-11 mutant line. Using A. rhizogenes root transformation and Bradyrizobium ORS278 inoculation, we found full complementation of the rlck2-11 mutant phenotype with AeRLCK2 at 3 wpi, both in terms of aerial plant development and nodule number, validating the functionality of the AeRLCK2 promoter for nodulation (Fig. 7, A and B, Supplementary Table S8). Unexpectedly, in contrast to AeRLCK1, AaRLCK_O was also able to rescue the rlck2-11 mutant phenotype (Fig. 7, A and B, Supplementary Table S6). Microscopy analysis of AaRLCK_O and AeRLCK2 complemented roots revealed WT sized nodules with cells well-filled with bacteria (Fig. 7C). Similar results were also obtained when expressing the 3 RLCK genes under the constitutive Ubiquitin promoter in rlck2-11 (Supplementary Fig. S19, Supplementary Table S8).

Figure 7.

A. evenia rlck2 mutant cross-complementation of root nodulation. Hairy roots of A. evenia rlck2-11 plants were transformed with the empty vector (EV) containing the DsRed marker, or the same vector containing pAeRLCK2:RLCK0, pAeRLCK2:RLCK1 or pAeRLCK2:RLCK2 and their nodulation phenotype was evaluated 21 dpi with Bradyrhizobium ORS278. Observations were made on 2 biological replicates from independent experiments. Representative root nodulation phenotypes are shown here and detailed in Supplementary Table S8. A) Plant aerial phenotype. B) Number of pink and white nodules formed on plants expressing the indicated constructs. Dots represent individual plants. Red numbers below the boxplots indicate the number of plants with pink nodules, relative to the total number of transformed plants. Letters indicate significant differences between constructs, as determined by analysis of variance (Kruskal-Wallis) and post-hoc analysis (Dunn's test), P < 0.05. Box plots showing the median (bold segment), interquartile range (box from Q1 to Q3), minimum and maximum (whiskers), and outliers (individual points). C) Nodule analysis on rlck2-11 roots transformed with the indicated constructs. Top and middle panels: microscopy observations of whole nodules under brightfield and red fluorescence using a DsRed filter, respectively. Bottom panels: cross-sections of nodules stained with SYTO 9, propidium iodide and calcofluor, and observed with a confocal microscope. For visualization purposes, SYTO 9 (originally green) was pseudocolored in magenta, propidium iodide (originally red) in yellow, and calcofluor (originally blue) in cyan. Scale bars: 1 mm (top and middle panels), 0.5 mm (bottom panel).

We next investigated whether AeRLCK2 behaves similarly to other related RLCK genes at the transcriptional level. Based on RNAseq data during nodulation that are available for A. evenia (Gully et al. 2018; Quilbé et al. 2021) and obtained in the present study for A. afraspera (Supplementary Table S4). AeRLCK1 and AeRLCK2 were found to be expressed in roots and nodules but AeRLCK2 was at least 10-folds more expressed that AeRLCK1 in both organs (Fig. 8A). In contrast, AaRLCK_O and AaRLCK_P showed clear induction of their expression level during nodulation (Fig. 8A). This behavior is similar to that described in L. japonicus for their respective orthologs, LjAMK8 and LjAMK24 (Leng et al. 2023). We also analyzed the expression levels of AeCRK and its ortholog AaCRK, identified by BLAST search in the A. afraspera transcriptome. Expression of both CRK genes was found to be induced during nodulation (Fig. 8A). To better understand these contrasting expression behaviors, we monitored the spatio-temporal expression profile of AeRLCK2 and AeCRK in WT roots transformed by A. rhizogenes with promoter-GUS fusions (Fig. 8, B and C). For AeRLCK2, a weak GUS staining was detected at the base of lateral roots before inoculation. After inoculation with Bradyrhizobium ORS278, increased GUS staining was observed at the base of lateral roots and in nodule primordia (2 and 4 dpi). When nodules emerged from the lateral root base (7 dpi), GUS staining was predominant at the nodule base and the vascular bundles of the adjacent lateral root. Finally, in mature nodules (14 dpi), GUS staining persisted at the nodule base and in the cell layers surrounding the central nitrogen-fixation zone. For AeCRK, no expression was detected before inoculation. At early stages of the interaction AeCRK expression mimicked that of AeRLCK2 in nodule primordia (4 dpi). But then, the expression of AeCRK was observed in the central infected tissue of mature nodules (7 and 14 dpi). It is noteworthy that in L. japonicus, LjAMK8 and LjAMK24, are expressed in the central infected tissue of mature nodules (Leng et al. 2023). These observations support the distinctness of the AeRLCK2 expression pattern observed in A. evenia.

Figure 8.

Expression pattern of AeRLCK2 and comparison with other Aeschynomene RLCK and CRK genes. A) RNAseq-based gene expression levels of AeRLCK1, AeRLCK2, AeCRK, AaRLCK_O, AaRLCK_P and AaCRK in roots of A. evenia and A. afraspera, noninoculated (NI) and inoculated (I) with compatible Bradyrhizobium strains. Data correspond to 3 biological replicates from independent experiments. Error bars indicate standard deviation (SD). B and C) Histochemical localization of GUS activity in hairy roots of WT A. evenia transformed with B  pRLCK2:GUS and C  pCRK:GUS during nodulation with Bradyrhizobium ORS278. NI: noninoculated, dpi: days post-inoculation. Top panels: whole roots observed under a light stereomicroscope. Bottom panels: sections of roots and nodules observed by microscope. Scale bars: 1 mm (upper panels), 0.1 mm (bottom panels).

Discussion

A. evenia shares with a few other Aeschynomene species a nitrogen-fixing symbiosis with photosynthetic Bradyrhizobium strains that is unique among legumes in that its initiation does not depend on the perception of rhizobial Nod factors (Giraud et al. 2007). The molecular processes underlying this Nod-independent symbiosis are still largely unknown. Recently, a forward genetic approach in A. evenia identified signaling components that are conserved in other legumes and led to the discovery of AeCRK, a Cysteine-rich RLK (Quilbé et al. 2021, 2022). Here, using our mutant-based approach, we have identified a second symbiosis actor, AeRLCK2, which corresponds to a receptor-like cytoplasmic kinase. This is another step forward in understanding the Nod-independent symbiosis signaling pathway in A. evenia. The 12 rlck2 mutants show a dual Nod-/BN phenotype, indicating a drastic reduction in the ability to initiate nodules. These mutants also lack early responses to rhizobial inoculation such as repression of ARH development, which is the first site of Bradyrhizobium colonization, and induction of symbiosis gene expression. In contrast to genes of the conserved symbiosis signaling pathway, for which most mutants are completely Nod-, all rlck2 mutants occasionally develop few enlarged nodules. This singular phenotype appears to be inherent to mutations in AeRLCK2 but it remains to be clarified whether the presence of BN nodules indicates a nontotal genetic penetrance for AeRLCK2 [as for NOOT in M. truncatula (Couzigou et al. 2012)], the existence of a partial functional redundancy, but not with AeRLCK1 because this latter does not complement a rlck2 mutant, or a function that is distinct from the conserved symbiosis signaling pathway [e.g. the infection receptor gene EPR3 in L. japonicus (Kawaharada et al. 2015)].

RLCKs lack extracellular ligand-binding domains, but they often functionally and physically associate with plasma membrane-localized RLKs to transduce intracellular signals (Lin et al. 2013; Liang and Zhou 2018). The AeRLCK2 is unusual among RLCKs in that it has a TM, but this property is conserved among its close homologs, including OsRLCK171, LjAMK8 and LjAMK24 (Vij et al. 2008; Leng et al. 2023). This TM is essential for its localization at the plasma membrane. Although its cytoplasmic domain corresponds to a typical Ser/Thr kinase, this activity was weak under in vitro conditions and not detected in planta. It cannot be ruled out that AeRLCK2 has a kinase activity in planta that was not detectable with the antibodies used. But it is also possible that some specific conditions [e.g. the RLCK BIK1 is activated by phosphorylation when bacterial flg22 binds to the FLS2-BAK1 complex in A. thaliana (Lee et al. 2017)] or the presence of interacting partners [e.g. several RLCKs have been shown to be strongly and specifically activated by Rop GTPases in A. thaliana and M. truncatula (Molendijk et al. 2008; Dorjgotov et al. 2009)] may be required for AeRLCK2 kinase activity. Protein-protein interaction assays revealed the association of AeRLCK2 and AeCRK in vitro and in vivo. In contrast to AeRLCK2, AeCRK showed a strong kinase activity and was able to trans-phosphorylate AeRLCK2 both in vitro and in vivo. This interaction is reminiscent of that between CRK36 and the RLCK BIK1, which is part of the FLS2-BAK1 receptor complex that perceive bacterial flg22 in A. thaliana (Lee et al. 2017). When activated, CRK36 increases BIK1 phosphorylation, leading to increased flg22 signaling and immunity. The AeRLCK2 residues phosphorylated by AeCRK were identified and shown to be important for AeRLCK2 function in nodulation. Additionally, the biological relevance of the AeCRK-RLCK interaction is supported by the expression of both AeCRK and AeRLCK2 in nodule primordia infected by Bradyrhizobium, as observed using promoter-driven GUS reporters. Nevertheless, their tissular expression patterns are distinct under nonsymbiotic conditions and in mature nodules. Furthermore, the nodulation phenotypes of the crk and rlck2 mutants both include early blocks in symbiosis establishment, although these blocks are different (Quilbé et al. 2021, 2022). A likely explanation is that AeCRK and AeRLCK2 have overlapping but not identical functions during symbiosis. Therefore, we hypothesize that they have other interacting partners to form 1 or more receptor complexes that mediate the Nod-independent symbiosis.

Many RLCKs have been characterized for their involvement in plant development, abiotic stress or immune responses (Lin et al. 2013; Liang and Zhou 2018). However, this analysis is very A. thaliana-centered, leaving out RLCKs that have no equivalent in this model plant (Vij et al. 2008). This is the case for the OsRLCK171/LjAMK8/LjAMK24 orthogroup to which AeRLCK1/AeRLCK2 belongs and for which a role in AM has only recently been uncovered (Leng et al. 2023). In L. japonicus, LjAMK8 and LjAMK24 interact with the RLK KIN3 and their counterparts in rice, OsRLCK171 and OsARK1, form a similar receptor complex, suggesting that this receptor complex has been evolutionarily conserved in plants for AM (Leng et al. 2023). Additionally, the expression of LjAMK8 and LjAMK24 in nodules suggests that they also play a role in the rhizobial symbiosis, but this remains to be confirmed (Leng et al. 2023). In A. evenia, what is clear is that, on 1 hand, AeRLCK2 is expressed during AM while not essential and, on the other hand, this gene is crucial for nodulation. Additionally, AeRLCK2 results from a tandem gene duplication event that is specific to the Nod-independent lineage within the genus Aeschynomene. In the duplicate RLCK genes, AeRLCK1 is structurally conserved and AeRLCK2 is more divergent, the latter having an unusual promoter sequence. This gene duplication and divergence may have facilitated the acquisition of the Nod-independent signaling. AeRLCK1 failed to complement a rlck2 mutant in A. evenia, indicating that it is functionally divergent from AeRLCK2. However, AaRLCK_O, the RLCK homolog in the Nod-dependent A. afraspera, was able to rescue the nodulation phenotype of rlck2 mutant plants. This suggests that AaRLCK_O can still functionally replace AeRLCK2. Based on the available data, the expression pattern of AeRLCK2 appears to differ from the RLCK homologs in A. afraspera and L. japonicus (Leng et al. 2023). But the lack of a genome sequence for A. afraspera currently precludes the analysis of tissular gene expression in this species. So far, it seems that the functional specialization of AeRLCK2 is based on the evolution of promotor specificity and on divergence of protein function with AeRLCK1. Further investigations may sort out whether AeRLCK2 represents a case of subfunctionalization or neofunctionalization, relative to AaRLCK_O and AeRLCK1.

From our work and most recent studies (Leng et al. 2023), we propose a model with proven RLCK involvements in AM through interaction of LjAMK8 and LjAMK24 with LjKIN3 in L. japonicus and in the Nod-independent rhizobial symbiosis through interaction of AeRLCK2 with AeCRK in A. evenia as well as potential roles for LjAMK8 and LjAMK24 in nodulation, and of AeRLCK2 in AM, as inferred from their tissular expression pattens (Fig. 9). A more comprehensive view of the function of RLCK genes in Aeschynomene species and the search for other occurrences of Nod-independent specific gene duplications should help us elucidate how the Nod-independent symbiosis evolved. The present advances also pave the way for the identification of additional molecular players that could be involved in the formation of receptor complex(es) with AeCRK and/or AeRLCK2 and mediate the Nod-independent symbiosis pathway in Aeschynomene legumes.

Figure 9.

Model of RLCK functions in arbuscular mycorrhiza (AM) and Nod-independent symbiosis in legumes. During AM in L. japonicus, the paralogs AMK8 and AMK24 (ARBUSCULAR MYCORRHIZA-INDUCED KINASES) interact with KIN3 (KINASE 3) at the periarbuscular membrane. Autophosphorylation and transphosphorylation events in this RLCK-RLK complex are linked to mediate downstream AM responses. In contrast to LjKIN3, LjAMK8 and LjAMK24 are also expressed during nodulation; however, their putative role in the rhizobial symbiosis remains unknown. In A. evenia, the LjAMK24 counterpart is absent, while 2 proteins, AeRLCK1 and AeRLCK2, are closely related to LjAMK8. AeRLCK2 is expressed during AM but does not appear to be essential for this process. While the symbiotic role of AeRLCK1 is currently unknown, AeRLCK2 is central in mediating Nod-independent symbiosis with photosynthetic bradyrhizobia. One of its functions is to interact with and be phosphorylated by AeCRK at the plasma membrane, which is important for RLCK2 function in nodule initiation. The upstream signal and downstream signaling components remain to be elucidated.

Materials and methods

Plant material and growth conditions

The A. evenia lines studied here include the CIAT22838 reference line, mutants derived from this reference line as obtained from a nodulation screen of an EMS-mutagenized population (Quilbé et al. 2021, 2022), and the other WT accession PI225551 (Chaintreuil et al. 2018) (Supplementary Tables S1 and S2). A selection of Aeschynomene species, which use either a Nod-dependent or -independent symbiosis process was also selected (Brottier et al. 2018) (Supplementary Table S1). Seeds were scarified with 96% v/v sulfuric acid for 25 to 40 min with agitation, and rinsed with distilled water. Scarfied A. evenia seeds were incubated overnight with 0.01% (v/v) ethrel (BAYER) to induce germination (Chaintreuil et al. 2016). Plant growth during in vitro and in greenhouse conditions according to the protocols established for Aeschynomene sp. (Chaintreuil et al. 2016).

Genetic characterization and sequencing of nodulation mutants

Genetic analyses, consisting of genetic determinism and allelism tests, were performed on rlck2 mutants following the methodology described previously (Quilbé et al. 2021). Without a priori gene identification by mapping-by-sequencing was performed on F2 mutant plants obtained from mutant × WT crosses (Quilbé et al. 2021). Illumina sequencing of the F2 mutant DNA pools was performed by the Norwegian Sequencing Center (CEES, Oslo, Norway) and the GeT-PlaGe platorm (INRAE, Toulouse, France). A targeted search for mutations in AeRLCK2 was performed by PCR amplification and followed by sequencing for the de novo mutation identification in mutant lines or co-segregation analysis in F2 mutant plants (Quilbé et al. 2021). The genetic characteristics of the mutants are listed in Supplementary Table S2 and the primers used for AeRLCK2 sequencing are listed in Supplementary Table S9.

Plant nodulation

Nodulation assays on A. evenia WT CIAT22838 and nodulation mutants were performed using Bradyrhizobium ORS278 as inoculum. To analyse the infection process, plants were inoculated with the derivative strains ORS278-GUS and ORS278-GFP (Giraud et al. 2007; Bonaldi et al. 2011).

For the in vitro nodulation test, 1-d-old seedlings were transferred to 0.8% agar-water plates at 37 °C for 24 h to achieve at least 1 cm of radicle growth. They were then transferred to covered glass tubes containing liquid buffered nodulation medium supplemented with 0.5 mm KNO3-, as described in detail (Arrighi et al. 2012). Seven days after transfer in tubes, plants were inoculated with 1 mL of bacterial culture per plant, adjusted to an OD600 = 1 using a spectrophotometer (Varian, UV-visible spectrophotometer Cary 50 scan). At 7, 10, 14, 17 and 21 dpi, the number of red and white nodules was assessed and counted using a binocular loupe. Other plant analyses were carried out on samples collected at 21 d post-inoculation (dpi). Nodule and ARH diameters were measured using ImageJ (version 2.14.0/1.54f, http://imageJ.nih.gov/ij), while the number of lateral roots per plant was evaluated using the Optimas software (version 6.1, Media Cybernitics, Silverspring, MD, USA). Nitrogenase enzyme activity is assessed by analyzing the reduction of acetylene to ethylene ARA on plants with nodules, as described (Arrighi et al. 2012).

For greenhouse experiments, scarified seeds were left overnight under gentle agitation (80 rpm) to induce radicle emergence. The next day they were transplanted into plastic trays containing attapulgite (Dry Oil, US Sorbix Special). Plants were inoculated at transplantating with 150 mL of ORS278 culture (OD600 = 1) per plastic tray (50 × 40 cm, 70 to 150 plants per tray) and grown for 4 wk before root observation.

Nodulation kinetic experiments with A. evenia PI225551 and A. afraspera LSTM1 were carried out under standard in vitro culture conditions, by inoculating plants with Bradyrhizobium ORS285 (Giraud et al. 2007) and collecting plant material at 0, 4 and 8 dpi.

Arbuscular mycorrhiza

Mycorrhizal phenotype studies were performed by inoculating 5-d-old A. evenia seedlings with spores of Rhizophagus irregularis DAOM197198 (Agronutrition, Carbonne, France) and growing them for 6 wk as previously described (Nouwen et al. 2024). Roots were stained with Sheaffer skrip ink, and fungal colonization was assessed on 20 root fragments per plant, with 6 plants per line, using the Myco-Calc method as described (Quilbé et al. 2022). AM was analysed using a Nikon AZ100 stereomicroscope (Champigny-sur-Marne, France), and images were taken using the Nikon Advanced software.

For analysis of GUS activity in A. evenia hairy roots, Plantago lanceolata was inoculated with Rhizophagus irregularis DAOM197198 spores and grown for 4 wk in pots containing zeolite and watered with 1× modified Long Ashton medium (15 µM NaH₂PO₄). Plantago shoots were cut and A. evenia chimeric plants were transplanted into the pots and grown for 3 more weeks. GUS staining and WGA staining were performed as described in (Girardin et al. 2019).

RNA sequencing analysis and real-time quantitative PCR

RNA was extracted from root material using the RNeasy Plant Mini Kit (Qiagen), treated with DNAse I (RNAse-Free DNAse set, Qiagen) and purified using the RNeasy Min Elute Cleanup Kit (Qiagen), according to the supplier's protocol. For RNAseq analysis, RNA material was prepared in biological triplicates of 5 plants/replicate for A. afraspera at 0, 4 and 8 dpi with Bradyrhizobium ORS285, and A. evenia CIAT22838 inoculated or not with R. irregularis DAOM197198 at 6 wpi (Supplementary Tables S5 and S7; Supplementary Data Set 3). Sequencing libraries were prepared using the TruSeq Stranded mRNA Kit and Illumina sequences generated on the MGX platform (Montpellier Genomix, Institut de Genomique Fonctionnelle, Montpellier France) and the GeT-PlaGe platorm (INRAE, Toulouse, France). A. afraspera Illumina RNA-seq datasets were de novo assembled using DRAP (Cabau et al. 2017), and those of A. evenia were mapped to the A. evenia reference genome using nf-core/rnaseq pipeline (Patel et al. 2024). Gene expression levels were normalized using the Diane pipeline (Cassan et al. 2021). For expression analysis, root material was generated in 4 biological replicates for the A. evenia CIAT22838 WT line and nodulation mutants at 0, 2, 4, and 7-dpi with Bradyrhizobium ORS278 (3 plants/line/replicate) and at 6 wpi with R. irregularis DAOM197198 (5 plants/line/replicate).

RT-qPCR was performed using the Takyon SYBRMaster Mix dTTP Blue kit (Eurogentec) in a 96-well plate format and the Stratagene MX3005P thermocycler (Agilent Technologies). The amplification protocol consisted of the following cycle: 3 min at 95 °C + 40 cycles of (10 s at 95 °C + 30 s at 60 °C + 60 s at 95 °C + 30 s at 60 °C) + 30 s at 95 °C. MXPro software was used to analyze the results based on the cycle threshold (CT) value. The gene expression level was obtained using the formula N = 10 ((CT-b)/a)—where a and b vary according to the efficiency of each primer pair. The housekeeping genes AeEF1α and AeUbi were used for subsequent normalization of expression levels. Primers used for quantification of gene expression are listed in Supplementary Table S10.

Sequence collection and in silico gene analysis

The Ae01g26600 gene is misannotated in the A. evenia genome v1. Based on A. evenia RNAseq data, it was manually curated to delineate AeRLCK1 and AeRLCK2 (Supplementary File 1). Microsynteny analysis was performed using the Legume Information System with the Genome Context Viewer (https://legumeinfo.org/lis_context_viewer) to visualize the gene collinearity in syntenic regions. RLCK protein domains were identified and annotated using InterProscan (http://www.ebi.ac.uk/interpro/) and DeepTMHMM (https://dtu.biolib.com/DeepTMHMM).

AeRLCK2 homologs were identified in legume species by mining the orthogroup database generated with OrthoFinder during the previous A. evenia genome project (Quilbé et al. 2021). RLCK sequences were also obtained by BLASTP searches in lupin genomes where RLCK genes are present but not annotated and, in the A. afraspera transcriptome generated in this study. The dataset was completed by searching for additional RLCK proteins in A. thaliana, O. sativa and P. persica in the Arabidopsis Information Resource (https://www.arabidopsis.org), the Rice Genome Annotation Project (http://rice.uga.edu) and the Phytozome (https://phytozome-next.jgi.doe.gov/) databases, respectively. A total of 32 RLCK protein sequences were retrieved from a set of 18 plant species and used for phylogenetic reconstruction (Supplementary Data Set 1; Supplementary File 1). Identified homologous proteins were aligned using MAFFT v7 (Katoh et al. 2019) with the auto strategy, allowing for gapped regions. To optimize the alignments and select the most appropriate approach for Maximum Likelihood analysis, trimAL v1.4.1 was used with the automated-1 option (Capella-Gutiérrez et al. 2009). The resulting alignments were used for phylogenetic analysis using IQ-tree v2.2.0.3 (Minh et al. 2020) with the recommended best-fit model from ModelFinder (Kalyaanamoorthy et al. 2017). Support values were determined with 100,000 iterations of ultrafast bootstrap approximation (UFboot) (Hoang et al. 2018). Tree visualization and annotation was performed using iTOL v6 (Letunic and Bork 2024). The tree was rooted with the most distant O. sativa RLCK homolog, which served as outgroup.

For the analysis of RLCK copy number in the genus Aeschynomene, additional protein sequences were retrieved in the OrthoFinder-derived RLCK orthogroup for species with available transcriptomes (Quilbé et al. 2021). For an extended set of Aeschynomene species (Supplementary Data Set 2), DNA was extracted using the CTAB method and served as matrix for PCR amplication using different pairs of primers designed as general or copy-specific to amplify an RLCK gene fragment in Aeschynomene species (Supplementary Table S9). The amplicons were amplified by Sanger technology. Transcriptome and PCR-derived sequences were translated into protein sequences and aligned to AaRLCK_O, AeRLCK1 and AeRLCK2 in Multalin (http://multalin.toulouse.inra.fr/multalin/multalin.html) for comparison (Supplementary Data Set 2; Supplementary File 2). To reconstruct the phylogeny of Aeschynomene species, previously published nuclear ITS (Internal Transcribed Spacer) and chloroplast matK sequences (Brottier et al. 2018) were concatenated and processed using the same methods described above. The symbiosis type and the presence of the different RLCK gene copies were added to the species tree.

Sequence alignments and tree files are provided in Supplementary Data Set 4.

Cloning and plasmid construction

For the initial complementation assay of the rlck2-11 mutant, the 1212 nucleotide AeRLCK2 CDS was PCR-amplified from A. evenia cDNAs, and cloned into the CR8/GW/TOPO entry vector, to generate pCR8-AeRLCK2. It was then transferred into the pUB-GW-GFP vector via the LR reaction (Invitrogen) to generate the pUbi-AeRLCK2-GFP construct, in which the GFP gene is used as a fluorescent marker for plant transformation.

To analyse AeRLCK2 and AeCRK expression in hairy roots, the AeRLCK2 promoter (2,537 pb upstream of the start codon) and the AeCRK promoter (1,381 pb upstream of the start codon) were synthesized and cloned into the Puc57-BSAI-free plasmid by GeneCust (www.genecust.com) (Supplementary File 3). The cloned promoters were subsequently fused to the GUS gene by GoldenGate cloning, using the pCambia2200-DsRed vector (Fliegmann et al. 2016).

For complementation tests of the rlck2-11 mutant with the phospho-silent version of AeRLCK2 (AeRLCK2^5A), the corresponding CDS was synthesized and cloned into the Puc57-BSAI-free plasmid by GeneCust (Supplementary File 3). The AeRLCK2 and AeRLCK2^5A were subsequently cloned in translational fusion to the YFP gene downstream of the pLjUb promoter by GoldenGate cloning, using the pCambia2200-DsRed vector (Fliegmann et al. 2016).

For cross-complementation tests on the rlck2-11 mutant, the AaRLCK_O, AeRLCK1 and AeRLCK2 CDS were PCR-amplified from A. afraspera and A. evenia cDNAs, respectively, and cloned in the pMiniT 2.0 vector (NEB PCR Cloning Kit, New England Biolabs). Using the GoldenGate cloning method (Fliegmann et al. 2016), they were placed downstream of both the pLjUb and the pAeRLCK2 promoters in the pCambia2200-DsRed vector, where the DsRed gene is used as a fluorescent marker for plant transformation.

For gene expression in N. benthamiana leaves and subsequent protein production, the AaRLCK_O, AeRLCK1, AeRLCK2 and AeCRK CDS without ending stop codon were cloned into pGEMT plasmids. An AeRLCK2ΔTM version, corresponding to the AeRLCK2 protein without the N-ter TM was also produced. Golden gate assembly was performed in the pCambia2200 vector as described (Fliegmann et al. 2016).

For the split-luciferase assay, the N-terminal and C-terminal parts of the luciferase were fused with a triple hemaglutin3HA or a triple Flag (3Flag) tag, respectively, and flanked by compatible Golden gate extensions. The luciferase modules were synthetized by Azenta (www.azenta.com), and assembled with the CDS to be tested as targets and placed under the control of the pLjUb promoter in a modified pCambia2200 vector by GoldenGate cloning (Fliegmann et al. 2016).

For in vitro assays, the predicted kinase domain of AeRLCK2 (G74—S403) was amplified by PCR and cloned into a modified pCDFDuet-1 vector (Novagen), as described for the cloning of AeCRK kinase (Quilbé et al. 2022). Site-directed mutagenesis was performed to generate the AeCRK^KD-G359E (inactive kinase mutation) and AeRLCK2^KD-G110E (rlck2-6 mutant allele mutation) variants using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs). The kinase domains of AaRLCK_O (E99 to Q447) and AeRLCK1 (E83 to F428) were amplified by PCR and cloned into pET-41a (Novagen) plasmid by restriction-ligation.

All PCR amplifications were performed using high-fidelity DNA polymerase Taq Phusion (New England Biolabs) or PrimeSTAR Max DNA polymerase (Takara). All constructs were verified by restriction enzyme digestion followed by sequencing with the Sanger technology. Different E. coli strains, Dh10b (ThermoFisher Scientific), TOP10 (ThermoFisher Scientific), XL10-Gold (Agilent) and Rosetta/DE3 (Novagen Sigma-Aldrich) were used for molecular cloning or protein expression. Final constructs were electroporated into Agrobacterium rhizogenes Arqua1 cells for transformation of A. evenia hairy roots (Quilbé et al. 2021) or into Agrobacterium tumefaciens LBA4404 VirGN54D for transient expression in Nicotiana benthamiana leaves (Voinnet et al. 2003). All primers are listed in Supplementary Table S9.

Analysis of promoter-GUS and complementation of A. evenia transformed hairy roots

Agrobacterium rhizogenes Arqua1 strains carrying the indicated constructs were used to transform roots of WT A. evenia CIAT22838 line and rlck2-11 mutant. Transformation of hairy roots was carried out as previously described (Quilbé et al. 2021). Briefly, 2-d-old seedlings with freshly cut radicles were directly inoculated with A. rhizogenes Arqua1 carrying the desired plasmid. They were grown on solid MS (Murashige and Skoog basal salt mixture) at 20 °C in the dark for 5 d and then transferred to solid MS medium containing 300 µg/mL cefotaxime. Plants bearing transgenic hairy roots were transferred to covered glass tubes containing liquid buffered nodulation media supplemented with 1 mm KNO3⁻. Seven days after transfer, plants were cultivated and inoculated with Bradyrhizobium ORS278 according standard procedures. The A. evenia WT hairy roots transformed with pAeRLCK2:GUS and pAeCRK:GUS were stained with X-Gluc to vizualize gene expression at the indicated time-points. For the A. evenia rlck2-11 complementation assays, nodule formation was monitored 21 d after inoculation.

Macroscopic and microscopic observations

All root samples from noninoculated and inoculated roots were visually inspected using a Nikon AZ100 stereomicroscope (Champigny-sur-Marne, France) and imaged using the Nikon Advanced software. Where needed, fresh 42 micron thick section of roots and nodules were made using a Leica VT1000s vibratome. For bacterial infection analysis (ORS278-GUS) and promoter activity studies, Whole plant roots or sectioned samples were stained with X-Gluc (Fabre et al. 2015) and then analysed using a Nikon macroscope or microscope. Gene expression was observed at different time-points in both young and older noninoculated plants, and at 2, 4, 7, 10, 14 and 21 dpi for plants inoculated with ORS278. To investigate bacterial infection with ORS278, freshly sectioned nodules were incubated with the live/dead reagent (Syto9/propidium iodide) and then stained with calcofluor white, as previously described (Nouwen et al. 2024). Samples were analysed using a confocal laser-scanning microscope (Carl Zeiss LSM 700, Jena, Germany). Calcofluor was excited at 405 nm and emitted light collected between within 405 to 470 nm, while SYTO 9 and propidium iodide were excited at 488 and 555 nm, respectively, with emissions collected between 490 to 522 nm and 555 to 700 nm. Images were acquired using the ZEN 2008 software (Zeiss, Oberkochen, Germany).

Subcellular localization in Nicotiana benthamiana leaves

N. benthamiana plants were grown in a controlled environment chamber under the following conditions: 19 to 21 °C with a 16 h light/8 h dark photoperiod. Four-week-old plants were used for A. tumefaciens-mediated transformation to achieve transient protein expression. A. tumefaciens LBA4404 VirGN54D strains containing the desired constructs were grown overnight in liquid LB medium, centrifuged at 7,000 g for 3 min, and washed twice with agroinfiltration buffer (10 mm MES-KOH pH 5.6, 10 mm MgCl2 and 150 µM acetosyringone). The optical density (OD600) was measured and adjusted to OD600 = 0.5. A. tumefaciens expressing the P19 protein (RNA silencing suppressor) was added to the A. tumefaciens solutions (OD600 = 0.2) to enhance protein expression (Voinnet et al. 2003). N. benthamiana leaves were agroinfiltrated with a needleless syringe and leaves were harvested 3 d later. Subcellular localization was assessed 72 h after infiltration with a ×25 water immersion objective lens (confocal microscope, SPE8 Leica). MtLYK3-CFP was used as a plasma membrane marker (Klaus-Heisen et al. 2011) and co-expressed with YFP fusion proteins. The excitation/emission filter sets for CFP and YFP were 458 nm/463 to 512 and 514 nm/525 to 580 nm, respectively.

Protein-protein interaction assays

Different construct combinations (MtLYK3-CFP, AeCRK^G359E-YFP, AeRLCK2-YFP, AeRLCK2ΔTM-YFP and AeCRK^G359E-mCherry) were agroinfiltrated into N. benthamiana leaves together with p19. Leaf material was collected 3 d after infiltration, and proteins were extracted for co-immunoprecipitation assays as previously published (Ding et al. 2024). YFP fusion bands were quantified using Image Lab 6.0 (volume function). Enrichment ratios (signal IP αGFP/signal input αGFP) were normalized using the negative control MtLYK3-CFP (signal αRFP).

The split-luciferase assay was performed as previously described (Landry et al. 2023). Briefly, A. tumefaciens LBA4404 VirGN54D strains containing the indicated NLuc and CLuc plasmids were co-infiltrated into 4-week-old N. benthamiana leaves. After 72 h, 4 mm leaf discs were placed in a 96-well plate, washed twice with water, and incubated with 1 mm luciferin substrate (Xenolight, PerkinElmer). Light emission was then quantified using a luminometer (VICTOR Nivo, PerkinElmer). Protein expression levels were assessed from 7 mm leaf discs, which were previously used to quantify luciferase activity. After protein extraction and Western blotting, 3HA-NLuc fusion bands were quantified using the volume function in Image Lab 6.0. The corresponding data were used to calculate a ratio normalized to the negative control AeCRK^G359E-Cluc/MtLYK3-Nluc. Raw data were normalized using the previously calculated ratio to account for variation in protein expression.

For Western blotting, leaf samples were ground in liquid nitrogen using a Retsch mixer mill (MM400). Proteins were solubilized in Laemmli buffer, heat-denatured at 95 °C for 5 min and separated by SDS-PAGE on home-made gels or 4% to 15% precast polyacrylamide gel (Bio-rad).

Proteins were transferred to a nitrocellulose membrane using the Transblot-Turbo system (Bio-rad) according to the manufacturer's instructions. The nitrocellulose membrane was blocked for 1 h at RT (or overnight at 4 °C) with 5% nonfat milk or 3% BSA solution in Tris-saline buffer (TBS) supplemented with 0.1% Tween-20 (TBS-T). As a loading control, Ponceau S staining solution was used to visualize Rubisco protein. The membrane was then incubated with the appropriate antibodies for 1 h at RT (or overnight at 4 °C). The following antibodies were used for protein biochemistry experiments: αHA-Hrp (12013819001, Roche, 1/5000), αFlag-Hrp (A8592, Sigma-Aldrich, 1/5000), αRabbit-Hrp (12–348, Millipore, 1/20000), αGFP (11814460001, Roche, 1/3000), goat αMouse-Hrp (1706516, Bio-Rad, 1/10000), rabbit αRFP (1/5000) (Lefebvre et al. 2012). Hrp bioluminescence was detected using Clarity Western ECL substrate (Bio-rad) and observed using the ChemiDoc imager (Bio-rad).

In vitro and in planta phosphorylation assays

For in vitro assays, sequences coding AeRLCK2^KD and its mutant form AeRLCK2^KD-G110E, AeCRK^KD and its mutated form AeCRK^KD-G359E, AaRLCK_O^KD and AeRLCK1^KD, were expressed in E. coli as GST fusion proteins at 16 °C. Proteins were purified using Glutathione-agarose beads (Amersham Biosciences) as described (Klaus-Heisen et al. 2011). AeCRK^KD was released from the resin using PreScission Protease (GE27-0843-01, Sigma-Aldrich, Germany). For radiolabeled kinase assays, proteins were incubated for 30 min at 30 °C in 10 mm HEPES-HCl pH 7.4 containing 5 mm MgCl₂, 5 mm MnCl₂, 20 µM ATP and 5 mCi 32P-ATP. Reactions were analysed by SDS-PAGE, followed by Coomassie staining and Phosphor imaging. Alternatively, proteins were incubated in the same buffer containing 40 µm ATP for 30 min at 30 °C. Reactions were analysed by SDS-PAGE, transferred to a nitrocellulose membrane and αPhospho-ser-thr-tyr (61-8300, Invitrogen, 1/333), followed by αRabbit-Hrp.

For in planta phosphorylation assays, proteins were extracted (w/v, 0.2/1) with protein extraction buffer [50 mm Tris-HCl 7.5, 150 mm NaCl, 10 mm EDTA, Triton X-100 1%, DTT 2 mm, supplemented with protease inhibitor cocktail (Sigma) and phosphatase inhibitor cocktail 3 (Sigma)]. Proteins were solubilized for 30 min at 4 °C and then centrifuged at 20,000 g for 5 min at 4 °C. The supernatant was filtered through Miracloth and incubated for 2 h at 4 °C with αGFP magnetic agarose beads (Chromotek). The beads were washed 3 times with protein extraction buffer. Proteins were solubilized in Laemmli 2× buffer, heat denaturated at 95 °C and subsequently separated by SDS-PAGE. The phosphorylation status was assessed using αPhospho-ser-thr-tyr or αPhospho-thr (1/2000, Zymed), followed by αRabbit-Hrp. Identification of phosphorylated sites by LC-MS/MS analysis was performed as described in detail in Supplementary Note S1.

Statistical analysis and graphs

Data analysis and visualization were performed using R with the ggplot2 package (Wickham 2016; R Core team et al. 2020). The Kruskal-Wallis test, followed by Dunn's post-hoc test for multiple comparisons, was used for all statistical analysis (Kruskal and Wallis 1952; Dunn 1964). All source data and detailed statistical results are provided in Supplementary Data Set 5.

Accession numbers

Supplementary sequence information used in this article for Aeschynomene evenia can be found in AeschynomeneBase (https://aeschynomenebase.fr/). Gene accession numbers are AeCCamK (Ae08g13330), AeCRK (Ae05g12380), AeEF1a (Ae09g20140), AeKIN3 (Ae06g09820), AeNIN (Ae07g00100), AeRAM1 (Ae06g18380), AeRLCK2 (Ae01g26600), AeSBT (Ae05g09230), AeSBTM1 (Ae05g09240), AeSTR (Ae05g35200), AeSYMREM1 (Ae03g30480), AeUbi (Ae10g10900), AeVPY (Ae05g16930).

Author contributions

J.-F.A. and B.L. conceived the whole project and supervised data analyses. J.Q. performed the genetic and molecular analysis of the rlck2 mutants, to which J.-F.A. contributed. N.H.A conducted the phenotypic characterization of the rlck2 mutants and RT-qPCR analyses relative to the nodulation and AM tests, to which J.Q. and M.P. contributed. N.H.A., D.L., J.Q. and C.G. conducted phylogenetic and evolutionary analyses. N.H.A., J.Q. and M.R. generated molecular constructs and conducted functional experiments on AeRLCK2. D.L. performed the biochemical characterization of AeRLCK2 and AeCRK, to which J.C. and C.P. contributed. M.P., F.G., and D.G. produced plant material and RNA material. L.B. screened the A. evenia mutagenized population to isolate rlck2 mutants. C.K. and M.R. analyzed the sequence data for the Mapping-by-Sequencing analysis of the rlck2 mutants as well as for the RNAseq data. M.P., F.G., C.C., N.N. and E.G. contributed to different experiments and provided their assistance for the achievement of the project. N.H.A., D.L., C.V., V.G., and B.L. conducted additional in vitro kinase assays, nodulation and AM tests for the revision of the manuscript. J.F.A., N.H.A., D.L. and B.L. wrote the manuscript. N.H.A produced the figures, to which D.L. contributed. All authors critically commented on and approved the manuscript.

Supplementary data

The following materials are available in the online version of this article.

Supplementary Figure S1. Identification of A. evenia rlck2 mutant alleles via mapping-by-sequencing.

Supplementary Figure S2. Nodulation kinetics of WT and rlck2 mutants.

Supplementary Figure S3. ARH development in the crk-1 mutant.

Supplementary Figure S4. Aerial phenotype of WT and rlck2 mutants grown in in vitro chambers.

Supplementary Figure S5. AeCRK induces cell death in Nicotiana benthamiana leaves.

Supplementary Figure S6. Expression of 3HA-NLuc fusion proteins.

Supplementary Figure S7. AeCRK kinase activity and AeRLCK2 transphosphorylation in vitro.

Supplementary Figure S8. The phosphorylation sites of AeRLCK2 specifically targeted by AeCRK.

Supplementary Figure S9. MS/MS fragmentation spectra supporting Supplementary Fig. S8.

Supplementary Figure S10.  Aeschynomene evenia rlck2-11 mutant complementation of root nodulation using a phospho-silent version.

Supplementary Figure S11. Microsynteny analysis of the AeRLCK2 locus.

Supplementary Figure S12. Alignment and structure of Aeschynomene RLKC proteins.

Supplementary Figure S13. Comparison of gene structure between AaRLCK_O, AeRLCK1 and AeRLCK2.

Supplementary Figure S14. Model of gene duplication at the AeRLCK2 locus.

Supplementary Figure S15. Mycorrhizal phenotype of the rlck2 mutants.

Supplementary Figure S16. Gene expression levels during AM in Aeschynomene evenia.

Supplementary Figure S17. Expression of AeRLCK2 in arbuscule-containing cells.

Supplementary Figure S18. Comparative analysis of AeRLCK1 and AaRLCK_O.

Supplementary Figure S19.  Aeschynomene evenia rlck2 mutant trans-complementation of root nodulation using pLjUb.

Supplementary Table S1.  Aeschynomene species used in this study.

Supplementary Table S2. Phenotypic, genetic and molecular data on the Aeschynomene evenia nodulation mutants.

Supplementary Table S3. Allelism analysis of the Aeschynomene evenia rlck2 mutants.

Supplementary Table S4. Functional complementation test for nodulation with AeRLCK2.

Supplementary Table S5. Summary of Illumina transcriptome sequencing and assembly for Aeschynomene afraspera.

Supplementary Table S6. Summary of Illumina transcriptome sequencing for Aeschynomene evenia CIAT22838.

Supplementary Table S7. Analysis of the list of 138 AMS conserved genes.

Supplementary Table S8. Nodulation data for the cross-complementation tests of rlck2 mutant.

Supplementary Table S9. Primer sequences used for PCR amplification, cloning and site-directed mutagenesis.

Supplementary Table S10. List of genes with the primers used for RT-qPCR analysis.

Supplementary Data Set 1. Annotation resources for plant genomes used in this study.

Supplementary Data Set 2. Annotation resources for Aeschynomene species used in this study.

Supplementary Data Set 3. Sequence data for Aeschynomene spp.

Supplementary Data Set 4. Sequence alignments and tree files used in Fig. 5, A and B.

Supplementary Data Set 5. Source data and statistical results.

Supplementary File 1.  RLCK sequences used for the phylogeny presented in Fig. 5A.

Supplementary File 2.  RLCK sequences used for the analysis presented in Fig. 5B.

Supplementary File 3. Sequences synthesized for AeRLCK2 and AeCRK.

Supplementary Note S1. Identification of AeRLCK2 phosphorylation sites.

Funding

This study was supported by 4 grants from the French National Research Agency (ANR-SymWay-21-CE20-0011-01, ANR-DUALITY-20-CE20-0017, ANR-AeschyNod-14-CE19-0005-01 and ANR-BugsInaCell-13-BSV7-0013-02).

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• CRK Gramene: AT4G01710

• CRK Araport: AT4G01710

• BIK1 Gramene: AT2G39660

• BIK1 Araport: AT2G39660

• proteins CHEBI: CHEBI:36080